1----------------------------------------------------------------------------- 2This file contains a concatenation of the PCRE2 man pages, converted to plain 3text format for ease of searching with a text editor, or for use on systems 4that do not have a man page processor. The small individual files that give 5synopses of each function in the library have not been included. Neither has 6the pcre2demo program. There are separate text files for the pcre2grep and 7pcre2test commands. 8----------------------------------------------------------------------------- 9 10 11PCRE2(3) Library Functions Manual PCRE2(3) 12 13 14 15NAME 16 PCRE2 - Perl-compatible regular expressions (revised API) 17 18INTRODUCTION 19 20 PCRE2 is the name used for a revised API for the PCRE library, which is 21 a set of functions, written in C, that implement regular expression 22 pattern matching using the same syntax and semantics as Perl, with just 23 a few differences. After nearly two decades, the limitations of the 24 original API were making development increasingly difficult. The new 25 API is more extensible, and it was simplified by abolishing the sepa- 26 rate "study" optimizing function; in PCRE2, patterns are automatically 27 optimized where possible. Since forking from PCRE1, the code has been 28 extensively refactored and new features introduced. The old library is 29 now obsolete and is no longer maintained. 30 31 As well as Perl-style regular expression patterns, some features that 32 appeared in Python and the original PCRE before they appeared in Perl 33 are available using the Python syntax. There is also some support for 34 one or two .NET and Oniguruma syntax items, and there are options for 35 requesting some minor changes that give better ECMAScript (aka Java- 36 Script) compatibility. 37 38 The source code for PCRE2 can be compiled to support strings of 8-bit, 39 16-bit, or 32-bit code units, which means that up to three separate li- 40 braries may be installed, one for each code unit size. The size of code 41 unit is not related to the bit size of the underlying hardware. In a 42 64-bit environment that also supports 32-bit applications, versions of 43 PCRE2 that are compiled in both 64-bit and 32-bit modes may be needed. 44 45 The original work to extend PCRE to 16-bit and 32-bit code units was 46 done by Zoltan Herczeg and Christian Persch, respectively. In all three 47 cases, strings can be interpreted either as one character per code 48 unit, or as UTF-encoded Unicode, with support for Unicode general cate- 49 gory properties. Unicode support is optional at build time (but is the 50 default). However, processing strings as UTF code units must be enabled 51 explicitly at run time. The version of Unicode in use can be discovered 52 by running 53 54 pcre2test -C 55 56 The three libraries contain identical sets of functions, with names 57 ending in _8, _16, or _32, respectively (for example, pcre2_com- 58 pile_8()). However, by defining PCRE2_CODE_UNIT_WIDTH to be 8, 16, or 59 32, a program that uses just one code unit width can be written using 60 generic names such as pcre2_compile(), and the documentation is written 61 assuming that this is the case. 62 63 In addition to the Perl-compatible matching function, PCRE2 contains an 64 alternative function that matches the same compiled patterns in a dif- 65 ferent way. In certain circumstances, the alternative function has some 66 advantages. For a discussion of the two matching algorithms, see the 67 pcre2matching page. 68 69 Details of exactly which Perl regular expression features are and are 70 not supported by PCRE2 are given in separate documents. See the 71 pcre2pattern and pcre2compat pages. There is a syntax summary in the 72 pcre2syntax page. 73 74 Some features of PCRE2 can be included, excluded, or changed when the 75 library is built. The pcre2_config() function makes it possible for a 76 client to discover which features are available. The features them- 77 selves are described in the pcre2build page. Documentation about build- 78 ing PCRE2 for various operating systems can be found in the README and 79 NON-AUTOTOOLS_BUILD files in the source distribution. 80 81 The libraries contains a number of undocumented internal functions and 82 data tables that are used by more than one of the exported external 83 functions, but which are not intended for use by external callers. 84 Their names all begin with "_pcre2", which hopefully will not provoke 85 any name clashes. In some environments, it is possible to control which 86 external symbols are exported when a shared library is built, and in 87 these cases the undocumented symbols are not exported. 88 89 90SECURITY CONSIDERATIONS 91 92 If you are using PCRE2 in a non-UTF application that permits users to 93 supply arbitrary patterns for compilation, you should be aware of a 94 feature that allows users to turn on UTF support from within a pattern. 95 For example, an 8-bit pattern that begins with "(*UTF)" turns on UTF-8 96 mode, which interprets patterns and subjects as strings of UTF-8 code 97 units instead of individual 8-bit characters. This causes both the pat- 98 tern and any data against which it is matched to be checked for UTF-8 99 validity. If the data string is very long, such a check might use suf- 100 ficiently many resources as to cause your application to lose perfor- 101 mance. 102 103 One way of guarding against this possibility is to use the pcre2_pat- 104 tern_info() function to check the compiled pattern's options for 105 PCRE2_UTF. Alternatively, you can set the PCRE2_NEVER_UTF option when 106 calling pcre2_compile(). This causes a compile time error if the pat- 107 tern contains a UTF-setting sequence. 108 109 The use of Unicode properties for character types such as \d can also 110 be enabled from within the pattern, by specifying "(*UCP)". This fea- 111 ture can be disallowed by setting the PCRE2_NEVER_UCP option. 112 113 If your application is one that supports UTF, be aware that validity 114 checking can take time. If the same data string is to be matched many 115 times, you can use the PCRE2_NO_UTF_CHECK option for the second and 116 subsequent matches to avoid running redundant checks. 117 118 The use of the \C escape sequence in a UTF-8 or UTF-16 pattern can lead 119 to problems, because it may leave the current matching point in the 120 middle of a multi-code-unit character. The PCRE2_NEVER_BACKSLASH_C op- 121 tion can be used by an application to lock out the use of \C, causing a 122 compile-time error if it is encountered. It is also possible to build 123 PCRE2 with the use of \C permanently disabled. 124 125 Another way that performance can be hit is by running a pattern that 126 has a very large search tree against a string that will never match. 127 Nested unlimited repeats in a pattern are a common example. PCRE2 pro- 128 vides some protection against this: see the pcre2_set_match_limit() 129 function in the pcre2api page. There is a similar function called 130 pcre2_set_depth_limit() that can be used to restrict the amount of mem- 131 ory that is used. 132 133 134USER DOCUMENTATION 135 136 The user documentation for PCRE2 comprises a number of different sec- 137 tions. In the "man" format, each of these is a separate "man page". In 138 the HTML format, each is a separate page, linked from the index page. 139 In the plain text format, the descriptions of the pcre2grep and 140 pcre2test programs are in files called pcre2grep.txt and pcre2test.txt, 141 respectively. The remaining sections, except for the pcre2demo section 142 (which is a program listing), and the short pages for individual func- 143 tions, are concatenated in pcre2.txt, for ease of searching. The sec- 144 tions are as follows: 145 146 pcre2 this document 147 pcre2-config show PCRE2 installation configuration information 148 pcre2api details of PCRE2's native C API 149 pcre2build building PCRE2 150 pcre2callout details of the pattern callout feature 151 pcre2compat discussion of Perl compatibility 152 pcre2convert details of pattern conversion functions 153 pcre2demo a demonstration C program that uses PCRE2 154 pcre2grep description of the pcre2grep command (8-bit only) 155 pcre2jit discussion of just-in-time optimization support 156 pcre2limits details of size and other limits 157 pcre2matching discussion of the two matching algorithms 158 pcre2partial details of the partial matching facility 159 pcre2pattern syntax and semantics of supported regular 160 expression patterns 161 pcre2perform discussion of performance issues 162 pcre2posix the POSIX-compatible C API for the 8-bit library 163 pcre2sample discussion of the pcre2demo program 164 pcre2serialize details of pattern serialization 165 pcre2syntax quick syntax reference 166 pcre2test description of the pcre2test command 167 pcre2unicode discussion of Unicode and UTF support 168 169 In the "man" and HTML formats, there is also a short page for each C 170 library function, listing its arguments and results. 171 172 173AUTHOR 174 175 Philip Hazel 176 Retired from University Computing Service 177 Cambridge, England. 178 179 Putting an actual email address here is a spam magnet. If you want to 180 email me, use my two names separated by a dot at gmail.com. 181 182 183REVISION 184 185 Last updated: 27 August 2021 186 Copyright (c) 1997-2021 University of Cambridge. 187------------------------------------------------------------------------------ 188 189 190PCRE2API(3) Library Functions Manual PCRE2API(3) 191 192 193 194NAME 195 PCRE2 - Perl-compatible regular expressions (revised API) 196 197 #include <pcre2.h> 198 199 PCRE2 is a new API for PCRE, starting at release 10.0. This document 200 contains a description of all its native functions. See the pcre2 docu- 201 ment for an overview of all the PCRE2 documentation. 202 203 204PCRE2 NATIVE API BASIC FUNCTIONS 205 206 pcre2_code *pcre2_compile(PCRE2_SPTR pattern, PCRE2_SIZE length, 207 uint32_t options, int *errorcode, PCRE2_SIZE *erroroffset, 208 pcre2_compile_context *ccontext); 209 210 void pcre2_code_free(pcre2_code *code); 211 212 pcre2_match_data *pcre2_match_data_create(uint32_t ovecsize, 213 pcre2_general_context *gcontext); 214 215 pcre2_match_data *pcre2_match_data_create_from_pattern( 216 const pcre2_code *code, pcre2_general_context *gcontext); 217 218 int pcre2_match(const pcre2_code *code, PCRE2_SPTR subject, 219 PCRE2_SIZE length, PCRE2_SIZE startoffset, 220 uint32_t options, pcre2_match_data *match_data, 221 pcre2_match_context *mcontext); 222 223 int pcre2_dfa_match(const pcre2_code *code, PCRE2_SPTR subject, 224 PCRE2_SIZE length, PCRE2_SIZE startoffset, 225 uint32_t options, pcre2_match_data *match_data, 226 pcre2_match_context *mcontext, 227 int *workspace, PCRE2_SIZE wscount); 228 229 void pcre2_match_data_free(pcre2_match_data *match_data); 230 231 232PCRE2 NATIVE API AUXILIARY MATCH FUNCTIONS 233 234 PCRE2_SPTR pcre2_get_mark(pcre2_match_data *match_data); 235 236 uint32_t pcre2_get_ovector_count(pcre2_match_data *match_data); 237 238 PCRE2_SIZE *pcre2_get_ovector_pointer(pcre2_match_data *match_data); 239 240 PCRE2_SIZE pcre2_get_startchar(pcre2_match_data *match_data); 241 242 243PCRE2 NATIVE API GENERAL CONTEXT FUNCTIONS 244 245 pcre2_general_context *pcre2_general_context_create( 246 void *(*private_malloc)(PCRE2_SIZE, void *), 247 void (*private_free)(void *, void *), void *memory_data); 248 249 pcre2_general_context *pcre2_general_context_copy( 250 pcre2_general_context *gcontext); 251 252 void pcre2_general_context_free(pcre2_general_context *gcontext); 253 254 255PCRE2 NATIVE API COMPILE CONTEXT FUNCTIONS 256 257 pcre2_compile_context *pcre2_compile_context_create( 258 pcre2_general_context *gcontext); 259 260 pcre2_compile_context *pcre2_compile_context_copy( 261 pcre2_compile_context *ccontext); 262 263 void pcre2_compile_context_free(pcre2_compile_context *ccontext); 264 265 int pcre2_set_bsr(pcre2_compile_context *ccontext, 266 uint32_t value); 267 268 int pcre2_set_character_tables(pcre2_compile_context *ccontext, 269 const uint8_t *tables); 270 271 int pcre2_set_compile_extra_options(pcre2_compile_context *ccontext, 272 uint32_t extra_options); 273 274 int pcre2_set_max_pattern_length(pcre2_compile_context *ccontext, 275 PCRE2_SIZE value); 276 277 int pcre2_set_newline(pcre2_compile_context *ccontext, 278 uint32_t value); 279 280 int pcre2_set_parens_nest_limit(pcre2_compile_context *ccontext, 281 uint32_t value); 282 283 int pcre2_set_compile_recursion_guard(pcre2_compile_context *ccontext, 284 int (*guard_function)(uint32_t, void *), void *user_data); 285 286 287PCRE2 NATIVE API MATCH CONTEXT FUNCTIONS 288 289 pcre2_match_context *pcre2_match_context_create( 290 pcre2_general_context *gcontext); 291 292 pcre2_match_context *pcre2_match_context_copy( 293 pcre2_match_context *mcontext); 294 295 void pcre2_match_context_free(pcre2_match_context *mcontext); 296 297 int pcre2_set_callout(pcre2_match_context *mcontext, 298 int (*callout_function)(pcre2_callout_block *, void *), 299 void *callout_data); 300 301 int pcre2_set_substitute_callout(pcre2_match_context *mcontext, 302 int (*callout_function)(pcre2_substitute_callout_block *, void *), 303 void *callout_data); 304 305 int pcre2_set_offset_limit(pcre2_match_context *mcontext, 306 PCRE2_SIZE value); 307 308 int pcre2_set_heap_limit(pcre2_match_context *mcontext, 309 uint32_t value); 310 311 int pcre2_set_match_limit(pcre2_match_context *mcontext, 312 uint32_t value); 313 314 int pcre2_set_depth_limit(pcre2_match_context *mcontext, 315 uint32_t value); 316 317 318PCRE2 NATIVE API STRING EXTRACTION FUNCTIONS 319 320 int pcre2_substring_copy_byname(pcre2_match_data *match_data, 321 PCRE2_SPTR name, PCRE2_UCHAR *buffer, PCRE2_SIZE *bufflen); 322 323 int pcre2_substring_copy_bynumber(pcre2_match_data *match_data, 324 uint32_t number, PCRE2_UCHAR *buffer, 325 PCRE2_SIZE *bufflen); 326 327 void pcre2_substring_free(PCRE2_UCHAR *buffer); 328 329 int pcre2_substring_get_byname(pcre2_match_data *match_data, 330 PCRE2_SPTR name, PCRE2_UCHAR **bufferptr, PCRE2_SIZE *bufflen); 331 332 int pcre2_substring_get_bynumber(pcre2_match_data *match_data, 333 uint32_t number, PCRE2_UCHAR **bufferptr, 334 PCRE2_SIZE *bufflen); 335 336 int pcre2_substring_length_byname(pcre2_match_data *match_data, 337 PCRE2_SPTR name, PCRE2_SIZE *length); 338 339 int pcre2_substring_length_bynumber(pcre2_match_data *match_data, 340 uint32_t number, PCRE2_SIZE *length); 341 342 int pcre2_substring_nametable_scan(const pcre2_code *code, 343 PCRE2_SPTR name, PCRE2_SPTR *first, PCRE2_SPTR *last); 344 345 int pcre2_substring_number_from_name(const pcre2_code *code, 346 PCRE2_SPTR name); 347 348 void pcre2_substring_list_free(PCRE2_SPTR *list); 349 350 int pcre2_substring_list_get(pcre2_match_data *match_data, 351 PCRE2_UCHAR ***listptr, PCRE2_SIZE **lengthsptr); 352 353 354PCRE2 NATIVE API STRING SUBSTITUTION FUNCTION 355 356 int pcre2_substitute(const pcre2_code *code, PCRE2_SPTR subject, 357 PCRE2_SIZE length, PCRE2_SIZE startoffset, 358 uint32_t options, pcre2_match_data *match_data, 359 pcre2_match_context *mcontext, PCRE2_SPTR replacementz, 360 PCRE2_SIZE rlength, PCRE2_UCHAR *outputbuffer, 361 PCRE2_SIZE *outlengthptr); 362 363 364PCRE2 NATIVE API JIT FUNCTIONS 365 366 int pcre2_jit_compile(pcre2_code *code, uint32_t options); 367 368 int pcre2_jit_match(const pcre2_code *code, PCRE2_SPTR subject, 369 PCRE2_SIZE length, PCRE2_SIZE startoffset, 370 uint32_t options, pcre2_match_data *match_data, 371 pcre2_match_context *mcontext); 372 373 void pcre2_jit_free_unused_memory(pcre2_general_context *gcontext); 374 375 pcre2_jit_stack *pcre2_jit_stack_create(PCRE2_SIZE startsize, 376 PCRE2_SIZE maxsize, pcre2_general_context *gcontext); 377 378 void pcre2_jit_stack_assign(pcre2_match_context *mcontext, 379 pcre2_jit_callback callback_function, void *callback_data); 380 381 void pcre2_jit_stack_free(pcre2_jit_stack *jit_stack); 382 383 384PCRE2 NATIVE API SERIALIZATION FUNCTIONS 385 386 int32_t pcre2_serialize_decode(pcre2_code **codes, 387 int32_t number_of_codes, const uint8_t *bytes, 388 pcre2_general_context *gcontext); 389 390 int32_t pcre2_serialize_encode(const pcre2_code **codes, 391 int32_t number_of_codes, uint8_t **serialized_bytes, 392 PCRE2_SIZE *serialized_size, pcre2_general_context *gcontext); 393 394 void pcre2_serialize_free(uint8_t *bytes); 395 396 int32_t pcre2_serialize_get_number_of_codes(const uint8_t *bytes); 397 398 399PCRE2 NATIVE API AUXILIARY FUNCTIONS 400 401 pcre2_code *pcre2_code_copy(const pcre2_code *code); 402 403 pcre2_code *pcre2_code_copy_with_tables(const pcre2_code *code); 404 405 int pcre2_get_error_message(int errorcode, PCRE2_UCHAR *buffer, 406 PCRE2_SIZE bufflen); 407 408 const uint8_t *pcre2_maketables(pcre2_general_context *gcontext); 409 410 void pcre2_maketables_free(pcre2_general_context *gcontext, 411 const uint8_t *tables); 412 413 int pcre2_pattern_info(const pcre2_code *code, uint32_t what, 414 void *where); 415 416 int pcre2_callout_enumerate(const pcre2_code *code, 417 int (*callback)(pcre2_callout_enumerate_block *, void *), 418 void *user_data); 419 420 int pcre2_config(uint32_t what, void *where); 421 422 423PCRE2 NATIVE API OBSOLETE FUNCTIONS 424 425 int pcre2_set_recursion_limit(pcre2_match_context *mcontext, 426 uint32_t value); 427 428 int pcre2_set_recursion_memory_management( 429 pcre2_match_context *mcontext, 430 void *(*private_malloc)(PCRE2_SIZE, void *), 431 void (*private_free)(void *, void *), void *memory_data); 432 433 These functions became obsolete at release 10.30 and are retained only 434 for backward compatibility. They should not be used in new code. The 435 first is replaced by pcre2_set_depth_limit(); the second is no longer 436 needed and has no effect (it always returns zero). 437 438 439PCRE2 EXPERIMENTAL PATTERN CONVERSION FUNCTIONS 440 441 pcre2_convert_context *pcre2_convert_context_create( 442 pcre2_general_context *gcontext); 443 444 pcre2_convert_context *pcre2_convert_context_copy( 445 pcre2_convert_context *cvcontext); 446 447 void pcre2_convert_context_free(pcre2_convert_context *cvcontext); 448 449 int pcre2_set_glob_escape(pcre2_convert_context *cvcontext, 450 uint32_t escape_char); 451 452 int pcre2_set_glob_separator(pcre2_convert_context *cvcontext, 453 uint32_t separator_char); 454 455 int pcre2_pattern_convert(PCRE2_SPTR pattern, PCRE2_SIZE length, 456 uint32_t options, PCRE2_UCHAR **buffer, 457 PCRE2_SIZE *blength, pcre2_convert_context *cvcontext); 458 459 void pcre2_converted_pattern_free(PCRE2_UCHAR *converted_pattern); 460 461 These functions provide a way of converting non-PCRE2 patterns into 462 patterns that can be processed by pcre2_compile(). This facility is ex- 463 perimental and may be changed in future releases. At present, "globs" 464 and POSIX basic and extended patterns can be converted. Details are 465 given in the pcre2convert documentation. 466 467 468PCRE2 8-BIT, 16-BIT, AND 32-BIT LIBRARIES 469 470 There are three PCRE2 libraries, supporting 8-bit, 16-bit, and 32-bit 471 code units, respectively. However, there is just one header file, 472 pcre2.h. This contains the function prototypes and other definitions 473 for all three libraries. One, two, or all three can be installed simul- 474 taneously. On Unix-like systems the libraries are called libpcre2-8, 475 libpcre2-16, and libpcre2-32, and they can also co-exist with the orig- 476 inal PCRE libraries. 477 478 Character strings are passed to and from a PCRE2 library as a sequence 479 of unsigned integers in code units of the appropriate width. Every 480 PCRE2 function comes in three different forms, one for each library, 481 for example: 482 483 pcre2_compile_8() 484 pcre2_compile_16() 485 pcre2_compile_32() 486 487 There are also three different sets of data types: 488 489 PCRE2_UCHAR8, PCRE2_UCHAR16, PCRE2_UCHAR32 490 PCRE2_SPTR8, PCRE2_SPTR16, PCRE2_SPTR32 491 492 The UCHAR types define unsigned code units of the appropriate widths. 493 For example, PCRE2_UCHAR16 is usually defined as `uint16_t'. The SPTR 494 types are constant pointers to the equivalent UCHAR types, that is, 495 they are pointers to vectors of unsigned code units. 496 497 Many applications use only one code unit width. For their convenience, 498 macros are defined whose names are the generic forms such as pcre2_com- 499 pile() and PCRE2_SPTR. These macros use the value of the macro 500 PCRE2_CODE_UNIT_WIDTH to generate the appropriate width-specific func- 501 tion and macro names. PCRE2_CODE_UNIT_WIDTH is not defined by default. 502 An application must define it to be 8, 16, or 32 before including 503 pcre2.h in order to make use of the generic names. 504 505 Applications that use more than one code unit width can be linked with 506 more than one PCRE2 library, but must define PCRE2_CODE_UNIT_WIDTH to 507 be 0 before including pcre2.h, and then use the real function names. 508 Any code that is to be included in an environment where the value of 509 PCRE2_CODE_UNIT_WIDTH is unknown should also use the real function 510 names. (Unfortunately, it is not possible in C code to save and restore 511 the value of a macro.) 512 513 If PCRE2_CODE_UNIT_WIDTH is not defined before including pcre2.h, a 514 compiler error occurs. 515 516 When using multiple libraries in an application, you must take care 517 when processing any particular pattern to use only functions from a 518 single library. For example, if you want to run a match using a pat- 519 tern that was compiled with pcre2_compile_16(), you must do so with 520 pcre2_match_16(), not pcre2_match_8() or pcre2_match_32(). 521 522 In the function summaries above, and in the rest of this document and 523 other PCRE2 documents, functions and data types are described using 524 their generic names, without the _8, _16, or _32 suffix. 525 526 527PCRE2 API OVERVIEW 528 529 PCRE2 has its own native API, which is described in this document. 530 There are also some wrapper functions for the 8-bit library that corre- 531 spond to the POSIX regular expression API, but they do not give access 532 to all the functionality of PCRE2. They are described in the pcre2posix 533 documentation. Both these APIs define a set of C function calls. 534 535 The native API C data types, function prototypes, option values, and 536 error codes are defined in the header file pcre2.h, which also contains 537 definitions of PCRE2_MAJOR and PCRE2_MINOR, the major and minor release 538 numbers for the library. Applications can use these to include support 539 for different releases of PCRE2. 540 541 In a Windows environment, if you want to statically link an application 542 program against a non-dll PCRE2 library, you must define PCRE2_STATIC 543 before including pcre2.h. 544 545 The functions pcre2_compile() and pcre2_match() are used for compiling 546 and matching regular expressions in a Perl-compatible manner. A sample 547 program that demonstrates the simplest way of using them is provided in 548 the file called pcre2demo.c in the PCRE2 source distribution. A listing 549 of this program is given in the pcre2demo documentation, and the 550 pcre2sample documentation describes how to compile and run it. 551 552 The compiling and matching functions recognize various options that are 553 passed as bits in an options argument. There are also some more compli- 554 cated parameters such as custom memory management functions and re- 555 source limits that are passed in "contexts" (which are just memory 556 blocks, described below). Simple applications do not need to make use 557 of contexts. 558 559 Just-in-time (JIT) compiler support is an optional feature of PCRE2 560 that can be built in appropriate hardware environments. It greatly 561 speeds up the matching performance of many patterns. Programs can re- 562 quest that it be used if available by calling pcre2_jit_compile() after 563 a pattern has been successfully compiled by pcre2_compile(). This does 564 nothing if JIT support is not available. 565 566 More complicated programs might need to make use of the specialist 567 functions pcre2_jit_stack_create(), pcre2_jit_stack_free(), and 568 pcre2_jit_stack_assign() in order to control the JIT code's memory us- 569 age. 570 571 JIT matching is automatically used by pcre2_match() if it is available, 572 unless the PCRE2_NO_JIT option is set. There is also a direct interface 573 for JIT matching, which gives improved performance at the expense of 574 less sanity checking. The JIT-specific functions are discussed in the 575 pcre2jit documentation. 576 577 A second matching function, pcre2_dfa_match(), which is not Perl-com- 578 patible, is also provided. This uses a different algorithm for the 579 matching. The alternative algorithm finds all possible matches (at a 580 given point in the subject), and scans the subject just once (unless 581 there are lookaround assertions). However, this algorithm does not re- 582 turn captured substrings. A description of the two matching algorithms 583 and their advantages and disadvantages is given in the pcre2matching 584 documentation. There is no JIT support for pcre2_dfa_match(). 585 586 In addition to the main compiling and matching functions, there are 587 convenience functions for extracting captured substrings from a subject 588 string that has been matched by pcre2_match(). They are: 589 590 pcre2_substring_copy_byname() 591 pcre2_substring_copy_bynumber() 592 pcre2_substring_get_byname() 593 pcre2_substring_get_bynumber() 594 pcre2_substring_list_get() 595 pcre2_substring_length_byname() 596 pcre2_substring_length_bynumber() 597 pcre2_substring_nametable_scan() 598 pcre2_substring_number_from_name() 599 600 pcre2_substring_free() and pcre2_substring_list_free() are also pro- 601 vided, to free memory used for extracted strings. If either of these 602 functions is called with a NULL argument, the function returns immedi- 603 ately without doing anything. 604 605 The function pcre2_substitute() can be called to match a pattern and 606 return a copy of the subject string with substitutions for parts that 607 were matched. 608 609 Functions whose names begin with pcre2_serialize_ are used for saving 610 compiled patterns on disc or elsewhere, and reloading them later. 611 612 Finally, there are functions for finding out information about a com- 613 piled pattern (pcre2_pattern_info()) and about the configuration with 614 which PCRE2 was built (pcre2_config()). 615 616 Functions with names ending with _free() are used for freeing memory 617 blocks of various sorts. In all cases, if one of these functions is 618 called with a NULL argument, it does nothing. 619 620 621STRING LENGTHS AND OFFSETS 622 623 The PCRE2 API uses string lengths and offsets into strings of code 624 units in several places. These values are always of type PCRE2_SIZE, 625 which is an unsigned integer type, currently always defined as size_t. 626 The largest value that can be stored in such a type (that is 627 ~(PCRE2_SIZE)0) is reserved as a special indicator for zero-terminated 628 strings and unset offsets. Therefore, the longest string that can be 629 handled is one less than this maximum. 630 631 632NEWLINES 633 634 PCRE2 supports five different conventions for indicating line breaks in 635 strings: a single CR (carriage return) character, a single LF (line- 636 feed) character, the two-character sequence CRLF, any of the three pre- 637 ceding, or any Unicode newline sequence. The Unicode newline sequences 638 are the three just mentioned, plus the single characters VT (vertical 639 tab, U+000B), FF (form feed, U+000C), NEL (next line, U+0085), LS (line 640 separator, U+2028), and PS (paragraph separator, U+2029). 641 642 Each of the first three conventions is used by at least one operating 643 system as its standard newline sequence. When PCRE2 is built, a default 644 can be specified. If it is not, the default is set to LF, which is the 645 Unix standard. However, the newline convention can be changed by an ap- 646 plication when calling pcre2_compile(), or it can be specified by spe- 647 cial text at the start of the pattern itself; this overrides any other 648 settings. See the pcre2pattern page for details of the special charac- 649 ter sequences. 650 651 In the PCRE2 documentation the word "newline" is used to mean "the 652 character or pair of characters that indicate a line break". The choice 653 of newline convention affects the handling of the dot, circumflex, and 654 dollar metacharacters, the handling of #-comments in /x mode, and, when 655 CRLF is a recognized line ending sequence, the match position advance- 656 ment for a non-anchored pattern. There is more detail about this in the 657 section on pcre2_match() options below. 658 659 The choice of newline convention does not affect the interpretation of 660 the \n or \r escape sequences, nor does it affect what \R matches; this 661 has its own separate convention. 662 663 664MULTITHREADING 665 666 In a multithreaded application it is important to keep thread-specific 667 data separate from data that can be shared between threads. The PCRE2 668 library code itself is thread-safe: it contains no static or global 669 variables. The API is designed to be fairly simple for non-threaded ap- 670 plications while at the same time ensuring that multithreaded applica- 671 tions can use it. 672 673 There are several different blocks of data that are used to pass infor- 674 mation between the application and the PCRE2 libraries. 675 676 The compiled pattern 677 678 A pointer to the compiled form of a pattern is returned to the user 679 when pcre2_compile() is successful. The data in the compiled pattern is 680 fixed, and does not change when the pattern is matched. Therefore, it 681 is thread-safe, that is, the same compiled pattern can be used by more 682 than one thread simultaneously. For example, an application can compile 683 all its patterns at the start, before forking off multiple threads that 684 use them. However, if the just-in-time (JIT) optimization feature is 685 being used, it needs separate memory stack areas for each thread. See 686 the pcre2jit documentation for more details. 687 688 In a more complicated situation, where patterns are compiled only when 689 they are first needed, but are still shared between threads, pointers 690 to compiled patterns must be protected from simultaneous writing by 691 multiple threads. This is somewhat tricky to do correctly. If you know 692 that writing to a pointer is atomic in your environment, you can use 693 logic like this: 694 695 Get a read-only (shared) lock (mutex) for pointer 696 if (pointer == NULL) 697 { 698 Get a write (unique) lock for pointer 699 if (pointer == NULL) pointer = pcre2_compile(... 700 } 701 Release the lock 702 Use pointer in pcre2_match() 703 704 Of course, testing for compilation errors should also be included in 705 the code. 706 707 The reason for checking the pointer a second time is as follows: Sev- 708 eral threads may have acquired the shared lock and tested the pointer 709 for being NULL, but only one of them will be given the write lock, with 710 the rest kept waiting. The winning thread will compile the pattern and 711 store the result. After this thread releases the write lock, another 712 thread will get it, and if it does not retest pointer for being NULL, 713 will recompile the pattern and overwrite the pointer, creating a memory 714 leak and possibly causing other issues. 715 716 In an environment where writing to a pointer may not be atomic, the 717 above logic is not sufficient. The thread that is doing the compiling 718 may be descheduled after writing only part of the pointer, which could 719 cause other threads to use an invalid value. Instead of checking the 720 pointer itself, a separate "pointer is valid" flag (that can be updated 721 atomically) must be used: 722 723 Get a read-only (shared) lock (mutex) for pointer 724 if (!pointer_is_valid) 725 { 726 Get a write (unique) lock for pointer 727 if (!pointer_is_valid) 728 { 729 pointer = pcre2_compile(... 730 pointer_is_valid = TRUE 731 } 732 } 733 Release the lock 734 Use pointer in pcre2_match() 735 736 If JIT is being used, but the JIT compilation is not being done immedi- 737 ately (perhaps waiting to see if the pattern is used often enough), 738 similar logic is required. JIT compilation updates a value within the 739 compiled code block, so a thread must gain unique write access to the 740 pointer before calling pcre2_jit_compile(). Alternatively, 741 pcre2_code_copy() or pcre2_code_copy_with_tables() can be used to ob- 742 tain a private copy of the compiled code before calling the JIT com- 743 piler. 744 745 Context blocks 746 747 The next main section below introduces the idea of "contexts" in which 748 PCRE2 functions are called. A context is nothing more than a collection 749 of parameters that control the way PCRE2 operates. Grouping a number of 750 parameters together in a context is a convenient way of passing them to 751 a PCRE2 function without using lots of arguments. The parameters that 752 are stored in contexts are in some sense "advanced features" of the 753 API. Many straightforward applications will not need to use contexts. 754 755 In a multithreaded application, if the parameters in a context are val- 756 ues that are never changed, the same context can be used by all the 757 threads. However, if any thread needs to change any value in a context, 758 it must make its own thread-specific copy. 759 760 Match blocks 761 762 The matching functions need a block of memory for storing the results 763 of a match. This includes details of what was matched, as well as addi- 764 tional information such as the name of a (*MARK) setting. Each thread 765 must provide its own copy of this memory. 766 767 768PCRE2 CONTEXTS 769 770 Some PCRE2 functions have a lot of parameters, many of which are used 771 only by specialist applications, for example, those that use custom 772 memory management or non-standard character tables. To keep function 773 argument lists at a reasonable size, and at the same time to keep the 774 API extensible, "uncommon" parameters are passed to certain functions 775 in a context instead of directly. A context is just a block of memory 776 that holds the parameter values. Applications that do not need to ad- 777 just any of the context parameters can pass NULL when a context pointer 778 is required. 779 780 There are three different types of context: a general context that is 781 relevant for several PCRE2 operations, a compile-time context, and a 782 match-time context. 783 784 The general context 785 786 At present, this context just contains pointers to (and data for) ex- 787 ternal memory management functions that are called from several places 788 in the PCRE2 library. The context is named `general' rather than 789 specifically `memory' because in future other fields may be added. If 790 you do not want to supply your own custom memory management functions, 791 you do not need to bother with a general context. A general context is 792 created by: 793 794 pcre2_general_context *pcre2_general_context_create( 795 void *(*private_malloc)(PCRE2_SIZE, void *), 796 void (*private_free)(void *, void *), void *memory_data); 797 798 The two function pointers specify custom memory management functions, 799 whose prototypes are: 800 801 void *private_malloc(PCRE2_SIZE, void *); 802 void private_free(void *, void *); 803 804 Whenever code in PCRE2 calls these functions, the final argument is the 805 value of memory_data. Either of the first two arguments of the creation 806 function may be NULL, in which case the system memory management func- 807 tions malloc() and free() are used. (This is not currently useful, as 808 there are no other fields in a general context, but in future there 809 might be.) The private_malloc() function is used (if supplied) to ob- 810 tain memory for storing the context, and all three values are saved as 811 part of the context. 812 813 Whenever PCRE2 creates a data block of any kind, the block contains a 814 pointer to the free() function that matches the malloc() function that 815 was used. When the time comes to free the block, this function is 816 called. 817 818 A general context can be copied by calling: 819 820 pcre2_general_context *pcre2_general_context_copy( 821 pcre2_general_context *gcontext); 822 823 The memory used for a general context should be freed by calling: 824 825 void pcre2_general_context_free(pcre2_general_context *gcontext); 826 827 If this function is passed a NULL argument, it returns immediately 828 without doing anything. 829 830 The compile context 831 832 A compile context is required if you want to provide an external func- 833 tion for stack checking during compilation or to change the default 834 values of any of the following compile-time parameters: 835 836 What \R matches (Unicode newlines or CR, LF, CRLF only) 837 PCRE2's character tables 838 The newline character sequence 839 The compile time nested parentheses limit 840 The maximum length of the pattern string 841 The extra options bits (none set by default) 842 843 A compile context is also required if you are using custom memory man- 844 agement. If none of these apply, just pass NULL as the context argu- 845 ment of pcre2_compile(). 846 847 A compile context is created, copied, and freed by the following func- 848 tions: 849 850 pcre2_compile_context *pcre2_compile_context_create( 851 pcre2_general_context *gcontext); 852 853 pcre2_compile_context *pcre2_compile_context_copy( 854 pcre2_compile_context *ccontext); 855 856 void pcre2_compile_context_free(pcre2_compile_context *ccontext); 857 858 A compile context is created with default values for its parameters. 859 These can be changed by calling the following functions, which return 0 860 on success, or PCRE2_ERROR_BADDATA if invalid data is detected. 861 862 int pcre2_set_bsr(pcre2_compile_context *ccontext, 863 uint32_t value); 864 865 The value must be PCRE2_BSR_ANYCRLF, to specify that \R matches only 866 CR, LF, or CRLF, or PCRE2_BSR_UNICODE, to specify that \R matches any 867 Unicode line ending sequence. The value is used by the JIT compiler and 868 by the two interpreted matching functions, pcre2_match() and 869 pcre2_dfa_match(). 870 871 int pcre2_set_character_tables(pcre2_compile_context *ccontext, 872 const uint8_t *tables); 873 874 The value must be the result of a call to pcre2_maketables(), whose 875 only argument is a general context. This function builds a set of char- 876 acter tables in the current locale. 877 878 int pcre2_set_compile_extra_options(pcre2_compile_context *ccontext, 879 uint32_t extra_options); 880 881 As PCRE2 has developed, almost all the 32 option bits that are avail- 882 able in the options argument of pcre2_compile() have been used up. To 883 avoid running out, the compile context contains a set of extra option 884 bits which are used for some newer, assumed rarer, options. This func- 885 tion sets those bits. It always sets all the bits (either on or off). 886 It does not modify any existing setting. The available options are de- 887 fined in the section entitled "Extra compile options" below. 888 889 int pcre2_set_max_pattern_length(pcre2_compile_context *ccontext, 890 PCRE2_SIZE value); 891 892 This sets a maximum length, in code units, for any pattern string that 893 is compiled with this context. If the pattern is longer, an error is 894 generated. This facility is provided so that applications that accept 895 patterns from external sources can limit their size. The default is the 896 largest number that a PCRE2_SIZE variable can hold, which is effec- 897 tively unlimited. 898 899 int pcre2_set_newline(pcre2_compile_context *ccontext, 900 uint32_t value); 901 902 This specifies which characters or character sequences are to be recog- 903 nized as newlines. The value must be one of PCRE2_NEWLINE_CR (carriage 904 return only), PCRE2_NEWLINE_LF (linefeed only), PCRE2_NEWLINE_CRLF (the 905 two-character sequence CR followed by LF), PCRE2_NEWLINE_ANYCRLF (any 906 of the above), PCRE2_NEWLINE_ANY (any Unicode newline sequence), or 907 PCRE2_NEWLINE_NUL (the NUL character, that is a binary zero). 908 909 A pattern can override the value set in the compile context by starting 910 with a sequence such as (*CRLF). See the pcre2pattern page for details. 911 912 When a pattern is compiled with the PCRE2_EXTENDED or PCRE2_EX- 913 TENDED_MORE option, the newline convention affects the recognition of 914 the end of internal comments starting with #. The value is saved with 915 the compiled pattern for subsequent use by the JIT compiler and by the 916 two interpreted matching functions, pcre2_match() and 917 pcre2_dfa_match(). 918 919 int pcre2_set_parens_nest_limit(pcre2_compile_context *ccontext, 920 uint32_t value); 921 922 This parameter adjusts the limit, set when PCRE2 is built (default 923 250), on the depth of parenthesis nesting in a pattern. This limit 924 stops rogue patterns using up too much system stack when being com- 925 piled. The limit applies to parentheses of all kinds, not just captur- 926 ing parentheses. 927 928 int pcre2_set_compile_recursion_guard(pcre2_compile_context *ccontext, 929 int (*guard_function)(uint32_t, void *), void *user_data); 930 931 There is at least one application that runs PCRE2 in threads with very 932 limited system stack, where running out of stack is to be avoided at 933 all costs. The parenthesis limit above cannot take account of how much 934 stack is actually available during compilation. For a finer control, 935 you can supply a function that is called whenever pcre2_compile() 936 starts to compile a parenthesized part of a pattern. This function can 937 check the actual stack size (or anything else that it wants to, of 938 course). 939 940 The first argument to the callout function gives the current depth of 941 nesting, and the second is user data that is set up by the last argu- 942 ment of pcre2_set_compile_recursion_guard(). The callout function 943 should return zero if all is well, or non-zero to force an error. 944 945 The match context 946 947 A match context is required if you want to: 948 949 Set up a callout function 950 Set an offset limit for matching an unanchored pattern 951 Change the limit on the amount of heap used when matching 952 Change the backtracking match limit 953 Change the backtracking depth limit 954 Set custom memory management specifically for the match 955 956 If none of these apply, just pass NULL as the context argument of 957 pcre2_match(), pcre2_dfa_match(), or pcre2_jit_match(). 958 959 A match context is created, copied, and freed by the following func- 960 tions: 961 962 pcre2_match_context *pcre2_match_context_create( 963 pcre2_general_context *gcontext); 964 965 pcre2_match_context *pcre2_match_context_copy( 966 pcre2_match_context *mcontext); 967 968 void pcre2_match_context_free(pcre2_match_context *mcontext); 969 970 A match context is created with default values for its parameters. 971 These can be changed by calling the following functions, which return 0 972 on success, or PCRE2_ERROR_BADDATA if invalid data is detected. 973 974 int pcre2_set_callout(pcre2_match_context *mcontext, 975 int (*callout_function)(pcre2_callout_block *, void *), 976 void *callout_data); 977 978 This sets up a callout function for PCRE2 to call at specified points 979 during a matching operation. Details are given in the pcre2callout doc- 980 umentation. 981 982 int pcre2_set_substitute_callout(pcre2_match_context *mcontext, 983 int (*callout_function)(pcre2_substitute_callout_block *, void *), 984 void *callout_data); 985 986 This sets up a callout function for PCRE2 to call after each substitu- 987 tion made by pcre2_substitute(). Details are given in the section enti- 988 tled "Creating a new string with substitutions" below. 989 990 int pcre2_set_offset_limit(pcre2_match_context *mcontext, 991 PCRE2_SIZE value); 992 993 The offset_limit parameter limits how far an unanchored search can ad- 994 vance in the subject string. The default value is PCRE2_UNSET. The 995 pcre2_match() and pcre2_dfa_match() functions return PCRE2_ERROR_NO- 996 MATCH if a match with a starting point before or at the given offset is 997 not found. The pcre2_substitute() function makes no more substitutions. 998 999 For example, if the pattern /abc/ is matched against "123abc" with an 1000 offset limit less than 3, the result is PCRE2_ERROR_NOMATCH. A match 1001 can never be found if the startoffset argument of pcre2_match(), 1002 pcre2_dfa_match(), or pcre2_substitute() is greater than the offset 1003 limit set in the match context. 1004 1005 When using this facility, you must set the PCRE2_USE_OFFSET_LIMIT op- 1006 tion when calling pcre2_compile() so that when JIT is in use, different 1007 code can be compiled. If a match is started with a non-default match 1008 limit when PCRE2_USE_OFFSET_LIMIT is not set, an error is generated. 1009 1010 The offset limit facility can be used to track progress when searching 1011 large subject strings or to limit the extent of global substitutions. 1012 See also the PCRE2_FIRSTLINE option, which requires a match to start 1013 before or at the first newline that follows the start of matching in 1014 the subject. If this is set with an offset limit, a match must occur in 1015 the first line and also within the offset limit. In other words, which- 1016 ever limit comes first is used. 1017 1018 int pcre2_set_heap_limit(pcre2_match_context *mcontext, 1019 uint32_t value); 1020 1021 The heap_limit parameter specifies, in units of kibibytes (1024 bytes), 1022 the maximum amount of heap memory that pcre2_match() may use to hold 1023 backtracking information when running an interpretive match. This limit 1024 also applies to pcre2_dfa_match(), which may use the heap when process- 1025 ing patterns with a lot of nested pattern recursion or lookarounds or 1026 atomic groups. This limit does not apply to matching with the JIT opti- 1027 mization, which has its own memory control arrangements (see the 1028 pcre2jit documentation for more details). If the limit is reached, the 1029 negative error code PCRE2_ERROR_HEAPLIMIT is returned. The default 1030 limit can be set when PCRE2 is built; if it is not, the default is set 1031 very large and is essentially "unlimited". 1032 1033 A value for the heap limit may also be supplied by an item at the start 1034 of a pattern of the form 1035 1036 (*LIMIT_HEAP=ddd) 1037 1038 where ddd is a decimal number. However, such a setting is ignored un- 1039 less ddd is less than the limit set by the caller of pcre2_match() or, 1040 if no such limit is set, less than the default. 1041 1042 The pcre2_match() function starts out using a 20KiB vector on the sys- 1043 tem stack for recording backtracking points. The more nested backtrack- 1044 ing points there are (that is, the deeper the search tree), the more 1045 memory is needed. Heap memory is used only if the initial vector is 1046 too small. If the heap limit is set to a value less than 21 (in partic- 1047 ular, zero) no heap memory will be used. In this case, only patterns 1048 that do not have a lot of nested backtracking can be successfully pro- 1049 cessed. 1050 1051 Similarly, for pcre2_dfa_match(), a vector on the system stack is used 1052 when processing pattern recursions, lookarounds, or atomic groups, and 1053 only if this is not big enough is heap memory used. In this case, too, 1054 setting a value of zero disables the use of the heap. 1055 1056 int pcre2_set_match_limit(pcre2_match_context *mcontext, 1057 uint32_t value); 1058 1059 The match_limit parameter provides a means of preventing PCRE2 from us- 1060 ing up too many computing resources when processing patterns that are 1061 not going to match, but which have a very large number of possibilities 1062 in their search trees. The classic example is a pattern that uses 1063 nested unlimited repeats. 1064 1065 There is an internal counter in pcre2_match() that is incremented each 1066 time round its main matching loop. If this value reaches the match 1067 limit, pcre2_match() returns the negative value PCRE2_ERROR_MATCHLIMIT. 1068 This has the effect of limiting the amount of backtracking that can 1069 take place. For patterns that are not anchored, the count restarts from 1070 zero for each position in the subject string. This limit also applies 1071 to pcre2_dfa_match(), though the counting is done in a different way. 1072 1073 When pcre2_match() is called with a pattern that was successfully pro- 1074 cessed by pcre2_jit_compile(), the way in which matching is executed is 1075 entirely different. However, there is still the possibility of runaway 1076 matching that goes on for a very long time, and so the match_limit 1077 value is also used in this case (but in a different way) to limit how 1078 long the matching can continue. 1079 1080 The default value for the limit can be set when PCRE2 is built; the de- 1081 fault default is 10 million, which handles all but the most extreme 1082 cases. A value for the match limit may also be supplied by an item at 1083 the start of a pattern of the form 1084 1085 (*LIMIT_MATCH=ddd) 1086 1087 where ddd is a decimal number. However, such a setting is ignored un- 1088 less ddd is less than the limit set by the caller of pcre2_match() or 1089 pcre2_dfa_match() or, if no such limit is set, less than the default. 1090 1091 int pcre2_set_depth_limit(pcre2_match_context *mcontext, 1092 uint32_t value); 1093 1094 This parameter limits the depth of nested backtracking in 1095 pcre2_match(). Each time a nested backtracking point is passed, a new 1096 memory "frame" is used to remember the state of matching at that point. 1097 Thus, this parameter indirectly limits the amount of memory that is 1098 used in a match. However, because the size of each memory "frame" de- 1099 pends on the number of capturing parentheses, the actual memory limit 1100 varies from pattern to pattern. This limit was more useful in versions 1101 before 10.30, where function recursion was used for backtracking. 1102 1103 The depth limit is not relevant, and is ignored, when matching is done 1104 using JIT compiled code. However, it is supported by pcre2_dfa_match(), 1105 which uses it to limit the depth of nested internal recursive function 1106 calls that implement atomic groups, lookaround assertions, and pattern 1107 recursions. This limits, indirectly, the amount of system stack that is 1108 used. It was more useful in versions before 10.32, when stack memory 1109 was used for local workspace vectors for recursive function calls. From 1110 version 10.32, only local variables are allocated on the stack and as 1111 each call uses only a few hundred bytes, even a small stack can support 1112 quite a lot of recursion. 1113 1114 If the depth of internal recursive function calls is great enough, lo- 1115 cal workspace vectors are allocated on the heap from version 10.32 on- 1116 wards, so the depth limit also indirectly limits the amount of heap 1117 memory that is used. A recursive pattern such as /(.(?2))((?1)|)/, when 1118 matched to a very long string using pcre2_dfa_match(), can use a great 1119 deal of memory. However, it is probably better to limit heap usage di- 1120 rectly by calling pcre2_set_heap_limit(). 1121 1122 The default value for the depth limit can be set when PCRE2 is built; 1123 if it is not, the default is set to the same value as the default for 1124 the match limit. If the limit is exceeded, pcre2_match() or 1125 pcre2_dfa_match() returns PCRE2_ERROR_DEPTHLIMIT. A value for the depth 1126 limit may also be supplied by an item at the start of a pattern of the 1127 form 1128 1129 (*LIMIT_DEPTH=ddd) 1130 1131 where ddd is a decimal number. However, such a setting is ignored un- 1132 less ddd is less than the limit set by the caller of pcre2_match() or 1133 pcre2_dfa_match() or, if no such limit is set, less than the default. 1134 1135 1136CHECKING BUILD-TIME OPTIONS 1137 1138 int pcre2_config(uint32_t what, void *where); 1139 1140 The function pcre2_config() makes it possible for a PCRE2 client to 1141 find the value of certain configuration parameters and to discover 1142 which optional features have been compiled into the PCRE2 library. The 1143 pcre2build documentation has more details about these features. 1144 1145 The first argument for pcre2_config() specifies which information is 1146 required. The second argument is a pointer to memory into which the in- 1147 formation is placed. If NULL is passed, the function returns the amount 1148 of memory that is needed for the requested information. For calls that 1149 return numerical values, the value is in bytes; when requesting these 1150 values, where should point to appropriately aligned memory. For calls 1151 that return strings, the required length is given in code units, not 1152 counting the terminating zero. 1153 1154 When requesting information, the returned value from pcre2_config() is 1155 non-negative on success, or the negative error code PCRE2_ERROR_BADOP- 1156 TION if the value in the first argument is not recognized. The follow- 1157 ing information is available: 1158 1159 PCRE2_CONFIG_BSR 1160 1161 The output is a uint32_t integer whose value indicates what character 1162 sequences the \R escape sequence matches by default. A value of 1163 PCRE2_BSR_UNICODE means that \R matches any Unicode line ending se- 1164 quence; a value of PCRE2_BSR_ANYCRLF means that \R matches only CR, LF, 1165 or CRLF. The default can be overridden when a pattern is compiled. 1166 1167 PCRE2_CONFIG_COMPILED_WIDTHS 1168 1169 The output is a uint32_t integer whose lower bits indicate which code 1170 unit widths were selected when PCRE2 was built. The 1-bit indicates 1171 8-bit support, and the 2-bit and 4-bit indicate 16-bit and 32-bit sup- 1172 port, respectively. 1173 1174 PCRE2_CONFIG_DEPTHLIMIT 1175 1176 The output is a uint32_t integer that gives the default limit for the 1177 depth of nested backtracking in pcre2_match() or the depth of nested 1178 recursions, lookarounds, and atomic groups in pcre2_dfa_match(). Fur- 1179 ther details are given with pcre2_set_depth_limit() above. 1180 1181 PCRE2_CONFIG_HEAPLIMIT 1182 1183 The output is a uint32_t integer that gives, in kibibytes, the default 1184 limit for the amount of heap memory used by pcre2_match() or 1185 pcre2_dfa_match(). Further details are given with 1186 pcre2_set_heap_limit() above. 1187 1188 PCRE2_CONFIG_JIT 1189 1190 The output is a uint32_t integer that is set to one if support for 1191 just-in-time compiling is available; otherwise it is set to zero. 1192 1193 PCRE2_CONFIG_JITTARGET 1194 1195 The where argument should point to a buffer that is at least 48 code 1196 units long. (The exact length required can be found by calling 1197 pcre2_config() with where set to NULL.) The buffer is filled with a 1198 string that contains the name of the architecture for which the JIT 1199 compiler is configured, for example "x86 32bit (little endian + un- 1200 aligned)". If JIT support is not available, PCRE2_ERROR_BADOPTION is 1201 returned, otherwise the number of code units used is returned. This is 1202 the length of the string, plus one unit for the terminating zero. 1203 1204 PCRE2_CONFIG_LINKSIZE 1205 1206 The output is a uint32_t integer that contains the number of bytes used 1207 for internal linkage in compiled regular expressions. When PCRE2 is 1208 configured, the value can be set to 2, 3, or 4, with the default being 1209 2. This is the value that is returned by pcre2_config(). However, when 1210 the 16-bit library is compiled, a value of 3 is rounded up to 4, and 1211 when the 32-bit library is compiled, internal linkages always use 4 1212 bytes, so the configured value is not relevant. 1213 1214 The default value of 2 for the 8-bit and 16-bit libraries is sufficient 1215 for all but the most massive patterns, since it allows the size of the 1216 compiled pattern to be up to 65535 code units. Larger values allow 1217 larger regular expressions to be compiled by those two libraries, but 1218 at the expense of slower matching. 1219 1220 PCRE2_CONFIG_MATCHLIMIT 1221 1222 The output is a uint32_t integer that gives the default match limit for 1223 pcre2_match(). Further details are given with pcre2_set_match_limit() 1224 above. 1225 1226 PCRE2_CONFIG_NEWLINE 1227 1228 The output is a uint32_t integer whose value specifies the default 1229 character sequence that is recognized as meaning "newline". The values 1230 are: 1231 1232 PCRE2_NEWLINE_CR Carriage return (CR) 1233 PCRE2_NEWLINE_LF Linefeed (LF) 1234 PCRE2_NEWLINE_CRLF Carriage return, linefeed (CRLF) 1235 PCRE2_NEWLINE_ANY Any Unicode line ending 1236 PCRE2_NEWLINE_ANYCRLF Any of CR, LF, or CRLF 1237 PCRE2_NEWLINE_NUL The NUL character (binary zero) 1238 1239 The default should normally correspond to the standard sequence for 1240 your operating system. 1241 1242 PCRE2_CONFIG_NEVER_BACKSLASH_C 1243 1244 The output is a uint32_t integer that is set to one if the use of \C 1245 was permanently disabled when PCRE2 was built; otherwise it is set to 1246 zero. 1247 1248 PCRE2_CONFIG_PARENSLIMIT 1249 1250 The output is a uint32_t integer that gives the maximum depth of nest- 1251 ing of parentheses (of any kind) in a pattern. This limit is imposed to 1252 cap the amount of system stack used when a pattern is compiled. It is 1253 specified when PCRE2 is built; the default is 250. This limit does not 1254 take into account the stack that may already be used by the calling ap- 1255 plication. For finer control over compilation stack usage, see 1256 pcre2_set_compile_recursion_guard(). 1257 1258 PCRE2_CONFIG_STACKRECURSE 1259 1260 This parameter is obsolete and should not be used in new code. The out- 1261 put is a uint32_t integer that is always set to zero. 1262 1263 PCRE2_CONFIG_TABLES_LENGTH 1264 1265 The output is a uint32_t integer that gives the length of PCRE2's char- 1266 acter processing tables in bytes. For details of these tables see the 1267 section on locale support below. 1268 1269 PCRE2_CONFIG_UNICODE_VERSION 1270 1271 The where argument should point to a buffer that is at least 24 code 1272 units long. (The exact length required can be found by calling 1273 pcre2_config() with where set to NULL.) If PCRE2 has been compiled 1274 without Unicode support, the buffer is filled with the text "Unicode 1275 not supported". Otherwise, the Unicode version string (for example, 1276 "8.0.0") is inserted. The number of code units used is returned. This 1277 is the length of the string plus one unit for the terminating zero. 1278 1279 PCRE2_CONFIG_UNICODE 1280 1281 The output is a uint32_t integer that is set to one if Unicode support 1282 is available; otherwise it is set to zero. Unicode support implies UTF 1283 support. 1284 1285 PCRE2_CONFIG_VERSION 1286 1287 The where argument should point to a buffer that is at least 24 code 1288 units long. (The exact length required can be found by calling 1289 pcre2_config() with where set to NULL.) The buffer is filled with the 1290 PCRE2 version string, zero-terminated. The number of code units used is 1291 returned. This is the length of the string plus one unit for the termi- 1292 nating zero. 1293 1294 1295COMPILING A PATTERN 1296 1297 pcre2_code *pcre2_compile(PCRE2_SPTR pattern, PCRE2_SIZE length, 1298 uint32_t options, int *errorcode, PCRE2_SIZE *erroroffset, 1299 pcre2_compile_context *ccontext); 1300 1301 void pcre2_code_free(pcre2_code *code); 1302 1303 pcre2_code *pcre2_code_copy(const pcre2_code *code); 1304 1305 pcre2_code *pcre2_code_copy_with_tables(const pcre2_code *code); 1306 1307 The pcre2_compile() function compiles a pattern into an internal form. 1308 The pattern is defined by a pointer to a string of code units and a 1309 length (in code units). If the pattern is zero-terminated, the length 1310 can be specified as PCRE2_ZERO_TERMINATED. The function returns a 1311 pointer to a block of memory that contains the compiled pattern and re- 1312 lated data, or NULL if an error occurred. 1313 1314 If the compile context argument ccontext is NULL, memory for the com- 1315 piled pattern is obtained by calling malloc(). Otherwise, it is ob- 1316 tained from the same memory function that was used for the compile con- 1317 text. The caller must free the memory by calling pcre2_code_free() when 1318 it is no longer needed. If pcre2_code_free() is called with a NULL ar- 1319 gument, it returns immediately, without doing anything. 1320 1321 The function pcre2_code_copy() makes a copy of the compiled code in new 1322 memory, using the same memory allocator as was used for the original. 1323 However, if the code has been processed by the JIT compiler (see be- 1324 low), the JIT information cannot be copied (because it is position-de- 1325 pendent). The new copy can initially be used only for non-JIT match- 1326 ing, though it can be passed to pcre2_jit_compile() if required. If 1327 pcre2_code_copy() is called with a NULL argument, it returns NULL. 1328 1329 The pcre2_code_copy() function provides a way for individual threads in 1330 a multithreaded application to acquire a private copy of shared com- 1331 piled code. However, it does not make a copy of the character tables 1332 used by the compiled pattern; the new pattern code points to the same 1333 tables as the original code. (See "Locale Support" below for details 1334 of these character tables.) In many applications the same tables are 1335 used throughout, so this behaviour is appropriate. Nevertheless, there 1336 are occasions when a copy of a compiled pattern and the relevant tables 1337 are needed. The pcre2_code_copy_with_tables() provides this facility. 1338 Copies of both the code and the tables are made, with the new code 1339 pointing to the new tables. The memory for the new tables is automati- 1340 cally freed when pcre2_code_free() is called for the new copy of the 1341 compiled code. If pcre2_code_copy_with_tables() is called with a NULL 1342 argument, it returns NULL. 1343 1344 NOTE: When one of the matching functions is called, pointers to the 1345 compiled pattern and the subject string are set in the match data block 1346 so that they can be referenced by the substring extraction functions 1347 after a successful match. After running a match, you must not free a 1348 compiled pattern or a subject string until after all operations on the 1349 match data block have taken place, unless, in the case of the subject 1350 string, you have used the PCRE2_COPY_MATCHED_SUBJECT option, which is 1351 described in the section entitled "Option bits for pcre2_match()" be- 1352 low. 1353 1354 The options argument for pcre2_compile() contains various bit settings 1355 that affect the compilation. It should be zero if none of them are re- 1356 quired. The available options are described below. Some of them (in 1357 particular, those that are compatible with Perl, but some others as 1358 well) can also be set and unset from within the pattern (see the de- 1359 tailed description in the pcre2pattern documentation). 1360 1361 For those options that can be different in different parts of the pat- 1362 tern, the contents of the options argument specifies their settings at 1363 the start of compilation. The PCRE2_ANCHORED, PCRE2_ENDANCHORED, and 1364 PCRE2_NO_UTF_CHECK options can be set at the time of matching as well 1365 as at compile time. 1366 1367 Some additional options and less frequently required compile-time pa- 1368 rameters (for example, the newline setting) can be provided in a com- 1369 pile context (as described above). 1370 1371 If errorcode or erroroffset is NULL, pcre2_compile() returns NULL imme- 1372 diately. Otherwise, the variables to which these point are set to an 1373 error code and an offset (number of code units) within the pattern, re- 1374 spectively, when pcre2_compile() returns NULL because a compilation er- 1375 ror has occurred. The values are not defined when compilation is suc- 1376 cessful and pcre2_compile() returns a non-NULL value. 1377 1378 There are nearly 100 positive error codes that pcre2_compile() may re- 1379 turn if it finds an error in the pattern. There are also some negative 1380 error codes that are used for invalid UTF strings when validity check- 1381 ing is in force. These are the same as given by pcre2_match() and 1382 pcre2_dfa_match(), and are described in the pcre2unicode documentation. 1383 There is no separate documentation for the positive error codes, be- 1384 cause the textual error messages that are obtained by calling the 1385 pcre2_get_error_message() function (see "Obtaining a textual error mes- 1386 sage" below) should be self-explanatory. Macro names starting with 1387 PCRE2_ERROR_ are defined for both positive and negative error codes in 1388 pcre2.h. 1389 1390 The value returned in erroroffset is an indication of where in the pat- 1391 tern the error occurred. It is not necessarily the furthest point in 1392 the pattern that was read. For example, after the error "lookbehind as- 1393 sertion is not fixed length", the error offset points to the start of 1394 the failing assertion. For an invalid UTF-8 or UTF-16 string, the off- 1395 set is that of the first code unit of the failing character. 1396 1397 Some errors are not detected until the whole pattern has been scanned; 1398 in these cases, the offset passed back is the length of the pattern. 1399 Note that the offset is in code units, not characters, even in a UTF 1400 mode. It may sometimes point into the middle of a UTF-8 or UTF-16 char- 1401 acter. 1402 1403 This code fragment shows a typical straightforward call to pcre2_com- 1404 pile(): 1405 1406 pcre2_code *re; 1407 PCRE2_SIZE erroffset; 1408 int errorcode; 1409 re = pcre2_compile( 1410 "^A.*Z", /* the pattern */ 1411 PCRE2_ZERO_TERMINATED, /* the pattern is zero-terminated */ 1412 0, /* default options */ 1413 &errorcode, /* for error code */ 1414 &erroffset, /* for error offset */ 1415 NULL); /* no compile context */ 1416 1417 1418 Main compile options 1419 1420 The following names for option bits are defined in the pcre2.h header 1421 file: 1422 1423 PCRE2_ANCHORED 1424 1425 If this bit is set, the pattern is forced to be "anchored", that is, it 1426 is constrained to match only at the first matching point in the string 1427 that is being searched (the "subject string"). This effect can also be 1428 achieved by appropriate constructs in the pattern itself, which is the 1429 only way to do it in Perl. 1430 1431 PCRE2_ALLOW_EMPTY_CLASS 1432 1433 By default, for compatibility with Perl, a closing square bracket that 1434 immediately follows an opening one is treated as a data character for 1435 the class. When PCRE2_ALLOW_EMPTY_CLASS is set, it terminates the 1436 class, which therefore contains no characters and so can never match. 1437 1438 PCRE2_ALT_BSUX 1439 1440 This option request alternative handling of three escape sequences, 1441 which makes PCRE2's behaviour more like ECMAscript (aka JavaScript). 1442 When it is set: 1443 1444 (1) \U matches an upper case "U" character; by default \U causes a com- 1445 pile time error (Perl uses \U to upper case subsequent characters). 1446 1447 (2) \u matches a lower case "u" character unless it is followed by four 1448 hexadecimal digits, in which case the hexadecimal number defines the 1449 code point to match. By default, \u causes a compile time error (Perl 1450 uses it to upper case the following character). 1451 1452 (3) \x matches a lower case "x" character unless it is followed by two 1453 hexadecimal digits, in which case the hexadecimal number defines the 1454 code point to match. By default, as in Perl, a hexadecimal number is 1455 always expected after \x, but it may have zero, one, or two digits (so, 1456 for example, \xz matches a binary zero character followed by z). 1457 1458 ECMAscript 6 added additional functionality to \u. This can be accessed 1459 using the PCRE2_EXTRA_ALT_BSUX extra option (see "Extra compile op- 1460 tions" below). Note that this alternative escape handling applies only 1461 to patterns. Neither of these options affects the processing of re- 1462 placement strings passed to pcre2_substitute(). 1463 1464 PCRE2_ALT_CIRCUMFLEX 1465 1466 In multiline mode (when PCRE2_MULTILINE is set), the circumflex 1467 metacharacter matches at the start of the subject (unless PCRE2_NOTBOL 1468 is set), and also after any internal newline. However, it does not 1469 match after a newline at the end of the subject, for compatibility with 1470 Perl. If you want a multiline circumflex also to match after a termi- 1471 nating newline, you must set PCRE2_ALT_CIRCUMFLEX. 1472 1473 PCRE2_ALT_VERBNAMES 1474 1475 By default, for compatibility with Perl, the name in any verb sequence 1476 such as (*MARK:NAME) is any sequence of characters that does not in- 1477 clude a closing parenthesis. The name is not processed in any way, and 1478 it is not possible to include a closing parenthesis in the name. How- 1479 ever, if the PCRE2_ALT_VERBNAMES option is set, normal backslash pro- 1480 cessing is applied to verb names and only an unescaped closing paren- 1481 thesis terminates the name. A closing parenthesis can be included in a 1482 name either as \) or between \Q and \E. If the PCRE2_EXTENDED or 1483 PCRE2_EXTENDED_MORE option is set with PCRE2_ALT_VERBNAMES, unescaped 1484 whitespace in verb names is skipped and #-comments are recognized, ex- 1485 actly as in the rest of the pattern. 1486 1487 PCRE2_AUTO_CALLOUT 1488 1489 If this bit is set, pcre2_compile() automatically inserts callout 1490 items, all with number 255, before each pattern item, except immedi- 1491 ately before or after an explicit callout in the pattern. For discus- 1492 sion of the callout facility, see the pcre2callout documentation. 1493 1494 PCRE2_CASELESS 1495 1496 If this bit is set, letters in the pattern match both upper and lower 1497 case letters in the subject. It is equivalent to Perl's /i option, and 1498 it can be changed within a pattern by a (?i) option setting. If either 1499 PCRE2_UTF or PCRE2_UCP is set, Unicode properties are used for all 1500 characters with more than one other case, and for all characters whose 1501 code points are greater than U+007F. Note that there are two ASCII 1502 characters, K and S, that, in addition to their lower case ASCII equiv- 1503 alents, are case-equivalent with U+212A (Kelvin sign) and U+017F (long 1504 S) respectively. For lower valued characters with only one other case, 1505 a lookup table is used for speed. When neither PCRE2_UTF nor PCRE2_UCP 1506 is set, a lookup table is used for all code points less than 256, and 1507 higher code points (available only in 16-bit or 32-bit mode) are 1508 treated as not having another case. 1509 1510 PCRE2_DOLLAR_ENDONLY 1511 1512 If this bit is set, a dollar metacharacter in the pattern matches only 1513 at the end of the subject string. Without this option, a dollar also 1514 matches immediately before a newline at the end of the string (but not 1515 before any other newlines). The PCRE2_DOLLAR_ENDONLY option is ignored 1516 if PCRE2_MULTILINE is set. There is no equivalent to this option in 1517 Perl, and no way to set it within a pattern. 1518 1519 PCRE2_DOTALL 1520 1521 If this bit is set, a dot metacharacter in the pattern matches any 1522 character, including one that indicates a newline. However, it only 1523 ever matches one character, even if newlines are coded as CRLF. Without 1524 this option, a dot does not match when the current position in the sub- 1525 ject is at a newline. This option is equivalent to Perl's /s option, 1526 and it can be changed within a pattern by a (?s) option setting. A neg- 1527 ative class such as [^a] always matches newline characters, and the \N 1528 escape sequence always matches a non-newline character, independent of 1529 the setting of PCRE2_DOTALL. 1530 1531 PCRE2_DUPNAMES 1532 1533 If this bit is set, names used to identify capture groups need not be 1534 unique. This can be helpful for certain types of pattern when it is 1535 known that only one instance of the named group can ever be matched. 1536 There are more details of named capture groups below; see also the 1537 pcre2pattern documentation. 1538 1539 PCRE2_ENDANCHORED 1540 1541 If this bit is set, the end of any pattern match must be right at the 1542 end of the string being searched (the "subject string"). If the pattern 1543 match succeeds by reaching (*ACCEPT), but does not reach the end of the 1544 subject, the match fails at the current starting point. For unanchored 1545 patterns, a new match is then tried at the next starting point. How- 1546 ever, if the match succeeds by reaching the end of the pattern, but not 1547 the end of the subject, backtracking occurs and an alternative match 1548 may be found. Consider these two patterns: 1549 1550 .(*ACCEPT)|.. 1551 .|.. 1552 1553 If matched against "abc" with PCRE2_ENDANCHORED set, the first matches 1554 "c" whereas the second matches "bc". The effect of PCRE2_ENDANCHORED 1555 can also be achieved by appropriate constructs in the pattern itself, 1556 which is the only way to do it in Perl. 1557 1558 For DFA matching with pcre2_dfa_match(), PCRE2_ENDANCHORED applies only 1559 to the first (that is, the longest) matched string. Other parallel 1560 matches, which are necessarily substrings of the first one, must obvi- 1561 ously end before the end of the subject. 1562 1563 PCRE2_EXTENDED 1564 1565 If this bit is set, most white space characters in the pattern are to- 1566 tally ignored except when escaped or inside a character class. However, 1567 white space is not allowed within sequences such as (?> that introduce 1568 various parenthesized groups, nor within numerical quantifiers such as 1569 {1,3}. Ignorable white space is permitted between an item and a follow- 1570 ing quantifier and between a quantifier and a following + that indi- 1571 cates possessiveness. PCRE2_EXTENDED is equivalent to Perl's /x option, 1572 and it can be changed within a pattern by a (?x) option setting. 1573 1574 When PCRE2 is compiled without Unicode support, PCRE2_EXTENDED recog- 1575 nizes as white space only those characters with code points less than 1576 256 that are flagged as white space in its low-character table. The ta- 1577 ble is normally created by pcre2_maketables(), which uses the isspace() 1578 function to identify space characters. In most ASCII environments, the 1579 relevant characters are those with code points 0x0009 (tab), 0x000A 1580 (linefeed), 0x000B (vertical tab), 0x000C (formfeed), 0x000D (carriage 1581 return), and 0x0020 (space). 1582 1583 When PCRE2 is compiled with Unicode support, in addition to these char- 1584 acters, five more Unicode "Pattern White Space" characters are recog- 1585 nized by PCRE2_EXTENDED. These are U+0085 (next line), U+200E (left-to- 1586 right mark), U+200F (right-to-left mark), U+2028 (line separator), and 1587 U+2029 (paragraph separator). This set of characters is the same as 1588 recognized by Perl's /x option. Note that the horizontal and vertical 1589 space characters that are matched by the \h and \v escapes in patterns 1590 are a much bigger set. 1591 1592 As well as ignoring most white space, PCRE2_EXTENDED also causes char- 1593 acters between an unescaped # outside a character class and the next 1594 newline, inclusive, to be ignored, which makes it possible to include 1595 comments inside complicated patterns. Note that the end of this type of 1596 comment is a literal newline sequence in the pattern; escape sequences 1597 that happen to represent a newline do not count. 1598 1599 Which characters are interpreted as newlines can be specified by a set- 1600 ting in the compile context that is passed to pcre2_compile() or by a 1601 special sequence at the start of the pattern, as described in the sec- 1602 tion entitled "Newline conventions" in the pcre2pattern documentation. 1603 A default is defined when PCRE2 is built. 1604 1605 PCRE2_EXTENDED_MORE 1606 1607 This option has the effect of PCRE2_EXTENDED, but, in addition, un- 1608 escaped space and horizontal tab characters are ignored inside a char- 1609 acter class. Note: only these two characters are ignored, not the full 1610 set of pattern white space characters that are ignored outside a char- 1611 acter class. PCRE2_EXTENDED_MORE is equivalent to Perl's /xx option, 1612 and it can be changed within a pattern by a (?xx) option setting. 1613 1614 PCRE2_FIRSTLINE 1615 1616 If this option is set, the start of an unanchored pattern match must be 1617 before or at the first newline in the subject string following the 1618 start of matching, though the matched text may continue over the new- 1619 line. If startoffset is non-zero, the limiting newline is not necessar- 1620 ily the first newline in the subject. For example, if the subject 1621 string is "abc\nxyz" (where \n represents a single-character newline) a 1622 pattern match for "yz" succeeds with PCRE2_FIRSTLINE if startoffset is 1623 greater than 3. See also PCRE2_USE_OFFSET_LIMIT, which provides a more 1624 general limiting facility. If PCRE2_FIRSTLINE is set with an offset 1625 limit, a match must occur in the first line and also within the offset 1626 limit. In other words, whichever limit comes first is used. 1627 1628 PCRE2_LITERAL 1629 1630 If this option is set, all meta-characters in the pattern are disabled, 1631 and it is treated as a literal string. Matching literal strings with a 1632 regular expression engine is not the most efficient way of doing it. If 1633 you are doing a lot of literal matching and are worried about effi- 1634 ciency, you should consider using other approaches. The only other main 1635 options that are allowed with PCRE2_LITERAL are: PCRE2_ANCHORED, 1636 PCRE2_ENDANCHORED, PCRE2_AUTO_CALLOUT, PCRE2_CASELESS, PCRE2_FIRSTLINE, 1637 PCRE2_MATCH_INVALID_UTF, PCRE2_NO_START_OPTIMIZE, PCRE2_NO_UTF_CHECK, 1638 PCRE2_UTF, and PCRE2_USE_OFFSET_LIMIT. The extra options PCRE2_EX- 1639 TRA_MATCH_LINE and PCRE2_EXTRA_MATCH_WORD are also supported. Any other 1640 options cause an error. 1641 1642 PCRE2_MATCH_INVALID_UTF 1643 1644 This option forces PCRE2_UTF (see below) and also enables support for 1645 matching by pcre2_match() in subject strings that contain invalid UTF 1646 sequences. This facility is not supported for DFA matching. For de- 1647 tails, see the pcre2unicode documentation. 1648 1649 PCRE2_MATCH_UNSET_BACKREF 1650 1651 If this option is set, a backreference to an unset capture group 1652 matches an empty string (by default this causes the current matching 1653 alternative to fail). A pattern such as (\1)(a) succeeds when this op- 1654 tion is set (assuming it can find an "a" in the subject), whereas it 1655 fails by default, for Perl compatibility. Setting this option makes 1656 PCRE2 behave more like ECMAscript (aka JavaScript). 1657 1658 PCRE2_MULTILINE 1659 1660 By default, for the purposes of matching "start of line" and "end of 1661 line", PCRE2 treats the subject string as consisting of a single line 1662 of characters, even if it actually contains newlines. The "start of 1663 line" metacharacter (^) matches only at the start of the string, and 1664 the "end of line" metacharacter ($) matches only at the end of the 1665 string, or before a terminating newline (except when PCRE2_DOLLAR_EN- 1666 DONLY is set). Note, however, that unless PCRE2_DOTALL is set, the "any 1667 character" metacharacter (.) does not match at a newline. This behav- 1668 iour (for ^, $, and dot) is the same as Perl. 1669 1670 When PCRE2_MULTILINE it is set, the "start of line" and "end of line" 1671 constructs match immediately following or immediately before internal 1672 newlines in the subject string, respectively, as well as at the very 1673 start and end. This is equivalent to Perl's /m option, and it can be 1674 changed within a pattern by a (?m) option setting. Note that the "start 1675 of line" metacharacter does not match after a newline at the end of the 1676 subject, for compatibility with Perl. However, you can change this by 1677 setting the PCRE2_ALT_CIRCUMFLEX option. If there are no newlines in a 1678 subject string, or no occurrences of ^ or $ in a pattern, setting 1679 PCRE2_MULTILINE has no effect. 1680 1681 PCRE2_NEVER_BACKSLASH_C 1682 1683 This option locks out the use of \C in the pattern that is being com- 1684 piled. This escape can cause unpredictable behaviour in UTF-8 or 1685 UTF-16 modes, because it may leave the current matching point in the 1686 middle of a multi-code-unit character. This option may be useful in ap- 1687 plications that process patterns from external sources. Note that there 1688 is also a build-time option that permanently locks out the use of \C. 1689 1690 PCRE2_NEVER_UCP 1691 1692 This option locks out the use of Unicode properties for handling \B, 1693 \b, \D, \d, \S, \s, \W, \w, and some of the POSIX character classes, as 1694 described for the PCRE2_UCP option below. In particular, it prevents 1695 the creator of the pattern from enabling this facility by starting the 1696 pattern with (*UCP). This option may be useful in applications that 1697 process patterns from external sources. The option combination PCRE_UCP 1698 and PCRE_NEVER_UCP causes an error. 1699 1700 PCRE2_NEVER_UTF 1701 1702 This option locks out interpretation of the pattern as UTF-8, UTF-16, 1703 or UTF-32, depending on which library is in use. In particular, it pre- 1704 vents the creator of the pattern from switching to UTF interpretation 1705 by starting the pattern with (*UTF). This option may be useful in ap- 1706 plications that process patterns from external sources. The combination 1707 of PCRE2_UTF and PCRE2_NEVER_UTF causes an error. 1708 1709 PCRE2_NO_AUTO_CAPTURE 1710 1711 If this option is set, it disables the use of numbered capturing paren- 1712 theses in the pattern. Any opening parenthesis that is not followed by 1713 ? behaves as if it were followed by ?: but named parentheses can still 1714 be used for capturing (and they acquire numbers in the usual way). This 1715 is the same as Perl's /n option. Note that, when this option is set, 1716 references to capture groups (backreferences or recursion/subroutine 1717 calls) may only refer to named groups, though the reference can be by 1718 name or by number. 1719 1720 PCRE2_NO_AUTO_POSSESS 1721 1722 If this option is set, it disables "auto-possessification", which is an 1723 optimization that, for example, turns a+b into a++b in order to avoid 1724 backtracks into a+ that can never be successful. However, if callouts 1725 are in use, auto-possessification means that some callouts are never 1726 taken. You can set this option if you want the matching functions to do 1727 a full unoptimized search and run all the callouts, but it is mainly 1728 provided for testing purposes. 1729 1730 PCRE2_NO_DOTSTAR_ANCHOR 1731 1732 If this option is set, it disables an optimization that is applied when 1733 .* is the first significant item in a top-level branch of a pattern, 1734 and all the other branches also start with .* or with \A or \G or ^. 1735 The optimization is automatically disabled for .* if it is inside an 1736 atomic group or a capture group that is the subject of a backreference, 1737 or if the pattern contains (*PRUNE) or (*SKIP). When the optimization 1738 is not disabled, such a pattern is automatically anchored if 1739 PCRE2_DOTALL is set for all the .* items and PCRE2_MULTILINE is not set 1740 for any ^ items. Otherwise, the fact that any match must start either 1741 at the start of the subject or following a newline is remembered. Like 1742 other optimizations, this can cause callouts to be skipped. 1743 1744 PCRE2_NO_START_OPTIMIZE 1745 1746 This is an option whose main effect is at matching time. It does not 1747 change what pcre2_compile() generates, but it does affect the output of 1748 the JIT compiler. 1749 1750 There are a number of optimizations that may occur at the start of a 1751 match, in order to speed up the process. For example, if it is known 1752 that an unanchored match must start with a specific code unit value, 1753 the matching code searches the subject for that value, and fails imme- 1754 diately if it cannot find it, without actually running the main match- 1755 ing function. This means that a special item such as (*COMMIT) at the 1756 start of a pattern is not considered until after a suitable starting 1757 point for the match has been found. Also, when callouts or (*MARK) 1758 items are in use, these "start-up" optimizations can cause them to be 1759 skipped if the pattern is never actually used. The start-up optimiza- 1760 tions are in effect a pre-scan of the subject that takes place before 1761 the pattern is run. 1762 1763 The PCRE2_NO_START_OPTIMIZE option disables the start-up optimizations, 1764 possibly causing performance to suffer, but ensuring that in cases 1765 where the result is "no match", the callouts do occur, and that items 1766 such as (*COMMIT) and (*MARK) are considered at every possible starting 1767 position in the subject string. 1768 1769 Setting PCRE2_NO_START_OPTIMIZE may change the outcome of a matching 1770 operation. Consider the pattern 1771 1772 (*COMMIT)ABC 1773 1774 When this is compiled, PCRE2 records the fact that a match must start 1775 with the character "A". Suppose the subject string is "DEFABC". The 1776 start-up optimization scans along the subject, finds "A" and runs the 1777 first match attempt from there. The (*COMMIT) item means that the pat- 1778 tern must match the current starting position, which in this case, it 1779 does. However, if the same match is run with PCRE2_NO_START_OPTIMIZE 1780 set, the initial scan along the subject string does not happen. The 1781 first match attempt is run starting from "D" and when this fails, 1782 (*COMMIT) prevents any further matches being tried, so the overall re- 1783 sult is "no match". 1784 1785 As another start-up optimization makes use of a minimum length for a 1786 matching subject, which is recorded when possible. Consider the pattern 1787 1788 (*MARK:1)B(*MARK:2)(X|Y) 1789 1790 The minimum length for a match is two characters. If the subject is 1791 "XXBB", the "starting character" optimization skips "XX", then tries to 1792 match "BB", which is long enough. In the process, (*MARK:2) is encoun- 1793 tered and remembered. When the match attempt fails, the next "B" is 1794 found, but there is only one character left, so there are no more at- 1795 tempts, and "no match" is returned with the "last mark seen" set to 1796 "2". If NO_START_OPTIMIZE is set, however, matches are tried at every 1797 possible starting position, including at the end of the subject, where 1798 (*MARK:1) is encountered, but there is no "B", so the "last mark seen" 1799 that is returned is "1". In this case, the optimizations do not affect 1800 the overall match result, which is still "no match", but they do affect 1801 the auxiliary information that is returned. 1802 1803 PCRE2_NO_UTF_CHECK 1804 1805 When PCRE2_UTF is set, the validity of the pattern as a UTF string is 1806 automatically checked. There are discussions about the validity of 1807 UTF-8 strings, UTF-16 strings, and UTF-32 strings in the pcre2unicode 1808 document. If an invalid UTF sequence is found, pcre2_compile() returns 1809 a negative error code. 1810 1811 If you know that your pattern is a valid UTF string, and you want to 1812 skip this check for performance reasons, you can set the 1813 PCRE2_NO_UTF_CHECK option. When it is set, the effect of passing an in- 1814 valid UTF string as a pattern is undefined. It may cause your program 1815 to crash or loop. 1816 1817 Note that this option can also be passed to pcre2_match() and 1818 pcre2_dfa_match(), to suppress UTF validity checking of the subject 1819 string. 1820 1821 Note also that setting PCRE2_NO_UTF_CHECK at compile time does not dis- 1822 able the error that is given if an escape sequence for an invalid Uni- 1823 code code point is encountered in the pattern. In particular, the so- 1824 called "surrogate" code points (0xd800 to 0xdfff) are invalid. If you 1825 want to allow escape sequences such as \x{d800} you can set the 1826 PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES extra option, as described in the 1827 section entitled "Extra compile options" below. However, this is pos- 1828 sible only in UTF-8 and UTF-32 modes, because these values are not rep- 1829 resentable in UTF-16. 1830 1831 PCRE2_UCP 1832 1833 This option has two effects. Firstly, it change the way PCRE2 processes 1834 \B, \b, \D, \d, \S, \s, \W, \w, and some of the POSIX character 1835 classes. By default, only ASCII characters are recognized, but if 1836 PCRE2_UCP is set, Unicode properties are used instead to classify char- 1837 acters. More details are given in the section on generic character 1838 types in the pcre2pattern page. If you set PCRE2_UCP, matching one of 1839 the items it affects takes much longer. 1840 1841 The second effect of PCRE2_UCP is to force the use of Unicode proper- 1842 ties for upper/lower casing operations on characters with code points 1843 greater than 127, even when PCRE2_UTF is not set. This makes it possi- 1844 ble, for example, to process strings in the 16-bit UCS-2 code. This op- 1845 tion is available only if PCRE2 has been compiled with Unicode support 1846 (which is the default). 1847 1848 PCRE2_UNGREEDY 1849 1850 This option inverts the "greediness" of the quantifiers so that they 1851 are not greedy by default, but become greedy if followed by "?". It is 1852 not compatible with Perl. It can also be set by a (?U) option setting 1853 within the pattern. 1854 1855 PCRE2_USE_OFFSET_LIMIT 1856 1857 This option must be set for pcre2_compile() if pcre2_set_offset_limit() 1858 is going to be used to set a non-default offset limit in a match con- 1859 text for matches that use this pattern. An error is generated if an 1860 offset limit is set without this option. For more details, see the de- 1861 scription of pcre2_set_offset_limit() in the section that describes 1862 match contexts. See also the PCRE2_FIRSTLINE option above. 1863 1864 PCRE2_UTF 1865 1866 This option causes PCRE2 to regard both the pattern and the subject 1867 strings that are subsequently processed as strings of UTF characters 1868 instead of single-code-unit strings. It is available when PCRE2 is 1869 built to include Unicode support (which is the default). If Unicode 1870 support is not available, the use of this option provokes an error. De- 1871 tails of how PCRE2_UTF changes the behaviour of PCRE2 are given in the 1872 pcre2unicode page. In particular, note that it changes the way 1873 PCRE2_CASELESS handles characters with code points greater than 127. 1874 1875 Extra compile options 1876 1877 The option bits that can be set in a compile context by calling the 1878 pcre2_set_compile_extra_options() function are as follows: 1879 1880 PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK 1881 1882 Since release 10.38 PCRE2 has forbidden the use of \K within lookaround 1883 assertions, following Perl's lead. This option is provided to re-enable 1884 the previous behaviour (act in positive lookarounds, ignore in negative 1885 ones) in case anybody is relying on it. 1886 1887 PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES 1888 1889 This option applies when compiling a pattern in UTF-8 or UTF-32 mode. 1890 It is forbidden in UTF-16 mode, and ignored in non-UTF modes. Unicode 1891 "surrogate" code points in the range 0xd800 to 0xdfff are used in pairs 1892 in UTF-16 to encode code points with values in the range 0x10000 to 1893 0x10ffff. The surrogates cannot therefore be represented in UTF-16. 1894 They can be represented in UTF-8 and UTF-32, but are defined as invalid 1895 code points, and cause errors if encountered in a UTF-8 or UTF-32 1896 string that is being checked for validity by PCRE2. 1897 1898 These values also cause errors if encountered in escape sequences such 1899 as \x{d912} within a pattern. However, it seems that some applications, 1900 when using PCRE2 to check for unwanted characters in UTF-8 strings, ex- 1901 plicitly test for the surrogates using escape sequences. The 1902 PCRE2_NO_UTF_CHECK option does not disable the error that occurs, be- 1903 cause it applies only to the testing of input strings for UTF validity. 1904 1905 If the extra option PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES is set, surro- 1906 gate code point values in UTF-8 and UTF-32 patterns no longer provoke 1907 errors and are incorporated in the compiled pattern. However, they can 1908 only match subject characters if the matching function is called with 1909 PCRE2_NO_UTF_CHECK set. 1910 1911 PCRE2_EXTRA_ALT_BSUX 1912 1913 The original option PCRE2_ALT_BSUX causes PCRE2 to process \U, \u, and 1914 \x in the way that ECMAscript (aka JavaScript) does. Additional func- 1915 tionality was defined by ECMAscript 6; setting PCRE2_EXTRA_ALT_BSUX has 1916 the effect of PCRE2_ALT_BSUX, but in addition it recognizes \u{hhh..} 1917 as a hexadecimal character code, where hhh.. is any number of hexadeci- 1918 mal digits. 1919 1920 PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL 1921 1922 This is a dangerous option. Use with care. By default, an unrecognized 1923 escape such as \j or a malformed one such as \x{2z} causes a compile- 1924 time error when detected by pcre2_compile(). Perl is somewhat inconsis- 1925 tent in handling such items: for example, \j is treated as a literal 1926 "j", and non-hexadecimal digits in \x{} are just ignored, though warn- 1927 ings are given in both cases if Perl's warning switch is enabled. How- 1928 ever, a malformed octal number after \o{ always causes an error in 1929 Perl. 1930 1931 If the PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL extra option is passed to 1932 pcre2_compile(), all unrecognized or malformed escape sequences are 1933 treated as single-character escapes. For example, \j is a literal "j" 1934 and \x{2z} is treated as the literal string "x{2z}". Setting this op- 1935 tion means that typos in patterns may go undetected and have unexpected 1936 results. Also note that a sequence such as [\N{] is interpreted as a 1937 malformed attempt at [\N{...}] and so is treated as [N{] whereas [\N] 1938 gives an error because an unqualified \N is a valid escape sequence but 1939 is not supported in a character class. To reiterate: this is a danger- 1940 ous option. Use with great care. 1941 1942 PCRE2_EXTRA_ESCAPED_CR_IS_LF 1943 1944 There are some legacy applications where the escape sequence \r in a 1945 pattern is expected to match a newline. If this option is set, \r in a 1946 pattern is converted to \n so that it matches a LF (linefeed) instead 1947 of a CR (carriage return) character. The option does not affect a lit- 1948 eral CR in the pattern, nor does it affect CR specified as an explicit 1949 code point such as \x{0D}. 1950 1951 PCRE2_EXTRA_MATCH_LINE 1952 1953 This option is provided for use by the -x option of pcre2grep. It 1954 causes the pattern only to match complete lines. This is achieved by 1955 automatically inserting the code for "^(?:" at the start of the com- 1956 piled pattern and ")$" at the end. Thus, when PCRE2_MULTILINE is set, 1957 the matched line may be in the middle of the subject string. This op- 1958 tion can be used with PCRE2_LITERAL. 1959 1960 PCRE2_EXTRA_MATCH_WORD 1961 1962 This option is provided for use by the -w option of pcre2grep. It 1963 causes the pattern only to match strings that have a word boundary at 1964 the start and the end. This is achieved by automatically inserting the 1965 code for "\b(?:" at the start of the compiled pattern and ")\b" at the 1966 end. The option may be used with PCRE2_LITERAL. However, it is ignored 1967 if PCRE2_EXTRA_MATCH_LINE is also set. 1968 1969 1970JUST-IN-TIME (JIT) COMPILATION 1971 1972 int pcre2_jit_compile(pcre2_code *code, uint32_t options); 1973 1974 int pcre2_jit_match(const pcre2_code *code, PCRE2_SPTR subject, 1975 PCRE2_SIZE length, PCRE2_SIZE startoffset, 1976 uint32_t options, pcre2_match_data *match_data, 1977 pcre2_match_context *mcontext); 1978 1979 void pcre2_jit_free_unused_memory(pcre2_general_context *gcontext); 1980 1981 pcre2_jit_stack *pcre2_jit_stack_create(PCRE2_SIZE startsize, 1982 PCRE2_SIZE maxsize, pcre2_general_context *gcontext); 1983 1984 void pcre2_jit_stack_assign(pcre2_match_context *mcontext, 1985 pcre2_jit_callback callback_function, void *callback_data); 1986 1987 void pcre2_jit_stack_free(pcre2_jit_stack *jit_stack); 1988 1989 These functions provide support for JIT compilation, which, if the 1990 just-in-time compiler is available, further processes a compiled pat- 1991 tern into machine code that executes much faster than the pcre2_match() 1992 interpretive matching function. Full details are given in the pcre2jit 1993 documentation. 1994 1995 JIT compilation is a heavyweight optimization. It can take some time 1996 for patterns to be analyzed, and for one-off matches and simple pat- 1997 terns the benefit of faster execution might be offset by a much slower 1998 compilation time. Most (but not all) patterns can be optimized by the 1999 JIT compiler. 2000 2001 2002LOCALE SUPPORT 2003 2004 const uint8_t *pcre2_maketables(pcre2_general_context *gcontext); 2005 2006 void pcre2_maketables_free(pcre2_general_context *gcontext, 2007 const uint8_t *tables); 2008 2009 PCRE2 handles caseless matching, and determines whether characters are 2010 letters, digits, or whatever, by reference to a set of tables, indexed 2011 by character code point. However, this applies only to characters whose 2012 code points are less than 256. By default, higher-valued code points 2013 never match escapes such as \w or \d. 2014 2015 When PCRE2 is built with Unicode support (the default), certain Unicode 2016 character properties can be tested with \p and \P, or, alternatively, 2017 the PCRE2_UCP option can be set when a pattern is compiled; this causes 2018 \w and friends to use Unicode property support instead of the built-in 2019 tables. PCRE2_UCP also causes upper/lower casing operations on charac- 2020 ters with code points greater than 127 to use Unicode properties. These 2021 effects apply even when PCRE2_UTF is not set. 2022 2023 The use of locales with Unicode is discouraged. If you are handling 2024 characters with code points greater than 127, you should either use 2025 Unicode support, or use locales, but not try to mix the two. 2026 2027 PCRE2 contains a built-in set of character tables that are used by de- 2028 fault. These are sufficient for many applications. Normally, the in- 2029 ternal tables recognize only ASCII characters. However, when PCRE2 is 2030 built, it is possible to cause the internal tables to be rebuilt in the 2031 default "C" locale of the local system, which may cause them to be dif- 2032 ferent. 2033 2034 The built-in tables can be overridden by tables supplied by the appli- 2035 cation that calls PCRE2. These may be created in a different locale 2036 from the default. As more and more applications change to using Uni- 2037 code, the need for this locale support is expected to die away. 2038 2039 External tables are built by calling the pcre2_maketables() function, 2040 in the relevant locale. The only argument to this function is a general 2041 context, which can be used to pass a custom memory allocator. If the 2042 argument is NULL, the system malloc() is used. The result can be passed 2043 to pcre2_compile() as often as necessary, by creating a compile context 2044 and calling pcre2_set_character_tables() to set the tables pointer 2045 therein. 2046 2047 For example, to build and use tables that are appropriate for the 2048 French locale (where accented characters with values greater than 127 2049 are treated as letters), the following code could be used: 2050 2051 setlocale(LC_CTYPE, "fr_FR"); 2052 tables = pcre2_maketables(NULL); 2053 ccontext = pcre2_compile_context_create(NULL); 2054 pcre2_set_character_tables(ccontext, tables); 2055 re = pcre2_compile(..., ccontext); 2056 2057 The locale name "fr_FR" is used on Linux and other Unix-like systems; 2058 if you are using Windows, the name for the French locale is "french". 2059 2060 The pointer that is passed (via the compile context) to pcre2_compile() 2061 is saved with the compiled pattern, and the same tables are used by the 2062 matching functions. Thus, for any single pattern, compilation and 2063 matching both happen in the same locale, but different patterns can be 2064 processed in different locales. 2065 2066 It is the caller's responsibility to ensure that the memory containing 2067 the tables remains available while they are still in use. When they are 2068 no longer needed, you can discard them using pcre2_maketables_free(), 2069 which should pass as its first parameter the same global context that 2070 was used to create the tables. 2071 2072 Saving locale tables 2073 2074 The tables described above are just a sequence of binary bytes, which 2075 makes them independent of hardware characteristics such as endianness 2076 or whether the processor is 32-bit or 64-bit. A copy of the result of 2077 pcre2_maketables() can therefore be saved in a file or elsewhere and 2078 re-used later, even in a different program or on another computer. The 2079 size of the tables (number of bytes) must be obtained by calling 2080 pcre2_config() with the PCRE2_CONFIG_TABLES_LENGTH option because 2081 pcre2_maketables() does not return this value. Note that the 2082 pcre2_dftables program, which is part of the PCRE2 build system, can be 2083 used stand-alone to create a file that contains a set of binary tables. 2084 See the pcre2build documentation for details. 2085 2086 2087INFORMATION ABOUT A COMPILED PATTERN 2088 2089 int pcre2_pattern_info(const pcre2 *code, uint32_t what, void *where); 2090 2091 The pcre2_pattern_info() function returns general information about a 2092 compiled pattern. For information about callouts, see the next section. 2093 The first argument for pcre2_pattern_info() is a pointer to the com- 2094 piled pattern. The second argument specifies which piece of information 2095 is required, and the third argument is a pointer to a variable to re- 2096 ceive the data. If the third argument is NULL, the first argument is 2097 ignored, and the function returns the size in bytes of the variable 2098 that is required for the information requested. Otherwise, the yield of 2099 the function is zero for success, or one of the following negative num- 2100 bers: 2101 2102 PCRE2_ERROR_NULL the argument code was NULL 2103 PCRE2_ERROR_BADMAGIC the "magic number" was not found 2104 PCRE2_ERROR_BADOPTION the value of what was invalid 2105 PCRE2_ERROR_UNSET the requested field is not set 2106 2107 The "magic number" is placed at the start of each compiled pattern as a 2108 simple check against passing an arbitrary memory pointer. Here is a 2109 typical call of pcre2_pattern_info(), to obtain the length of the com- 2110 piled pattern: 2111 2112 int rc; 2113 size_t length; 2114 rc = pcre2_pattern_info( 2115 re, /* result of pcre2_compile() */ 2116 PCRE2_INFO_SIZE, /* what is required */ 2117 &length); /* where to put the data */ 2118 2119 The possible values for the second argument are defined in pcre2.h, and 2120 are as follows: 2121 2122 PCRE2_INFO_ALLOPTIONS 2123 PCRE2_INFO_ARGOPTIONS 2124 PCRE2_INFO_EXTRAOPTIONS 2125 2126 Return copies of the pattern's options. The third argument should point 2127 to a uint32_t variable. PCRE2_INFO_ARGOPTIONS returns exactly the op- 2128 tions that were passed to pcre2_compile(), whereas PCRE2_INFO_ALLOP- 2129 TIONS returns the compile options as modified by any top-level (*XXX) 2130 option settings such as (*UTF) at the start of the pattern itself. 2131 PCRE2_INFO_EXTRAOPTIONS returns the extra options that were set in the 2132 compile context by calling the pcre2_set_compile_extra_options() func- 2133 tion. 2134 2135 For example, if the pattern /(*UTF)abc/ is compiled with the PCRE2_EX- 2136 TENDED option, the result for PCRE2_INFO_ALLOPTIONS is PCRE2_EXTENDED 2137 and PCRE2_UTF. Option settings such as (?i) that can change within a 2138 pattern do not affect the result of PCRE2_INFO_ALLOPTIONS, even if they 2139 appear right at the start of the pattern. (This was different in some 2140 earlier releases.) 2141 2142 A pattern compiled without PCRE2_ANCHORED is automatically anchored by 2143 PCRE2 if the first significant item in every top-level branch is one of 2144 the following: 2145 2146 ^ unless PCRE2_MULTILINE is set 2147 \A always 2148 \G always 2149 .* sometimes - see below 2150 2151 When .* is the first significant item, anchoring is possible only when 2152 all the following are true: 2153 2154 .* is not in an atomic group 2155 .* is not in a capture group that is the subject 2156 of a backreference 2157 PCRE2_DOTALL is in force for .* 2158 Neither (*PRUNE) nor (*SKIP) appears in the pattern 2159 PCRE2_NO_DOTSTAR_ANCHOR is not set 2160 2161 For patterns that are auto-anchored, the PCRE2_ANCHORED bit is set in 2162 the options returned for PCRE2_INFO_ALLOPTIONS. 2163 2164 PCRE2_INFO_BACKREFMAX 2165 2166 Return the number of the highest backreference in the pattern. The 2167 third argument should point to a uint32_t variable. Named capture 2168 groups acquire numbers as well as names, and these count towards the 2169 highest backreference. Backreferences such as \4 or \g{12} match the 2170 captured characters of the given group, but in addition, the check that 2171 a capture group is set in a conditional group such as (?(3)a|b) is also 2172 a backreference. Zero is returned if there are no backreferences. 2173 2174 PCRE2_INFO_BSR 2175 2176 The output is a uint32_t integer whose value indicates what character 2177 sequences the \R escape sequence matches. A value of PCRE2_BSR_UNICODE 2178 means that \R matches any Unicode line ending sequence; a value of 2179 PCRE2_BSR_ANYCRLF means that \R matches only CR, LF, or CRLF. 2180 2181 PCRE2_INFO_CAPTURECOUNT 2182 2183 Return the highest capture group number in the pattern. In patterns 2184 where (?| is not used, this is also the total number of capture groups. 2185 The third argument should point to a uint32_t variable. 2186 2187 PCRE2_INFO_DEPTHLIMIT 2188 2189 If the pattern set a backtracking depth limit by including an item of 2190 the form (*LIMIT_DEPTH=nnnn) at the start, the value is returned. The 2191 third argument should point to a uint32_t integer. If no such value has 2192 been set, the call to pcre2_pattern_info() returns the error PCRE2_ER- 2193 ROR_UNSET. Note that this limit will only be used during matching if it 2194 is less than the limit set or defaulted by the caller of the match 2195 function. 2196 2197 PCRE2_INFO_FIRSTBITMAP 2198 2199 In the absence of a single first code unit for a non-anchored pattern, 2200 pcre2_compile() may construct a 256-bit table that defines a fixed set 2201 of values for the first code unit in any match. For example, a pattern 2202 that starts with [abc] results in a table with three bits set. When 2203 code unit values greater than 255 are supported, the flag bit for 255 2204 means "any code unit of value 255 or above". If such a table was con- 2205 structed, a pointer to it is returned. Otherwise NULL is returned. The 2206 third argument should point to a const uint8_t * variable. 2207 2208 PCRE2_INFO_FIRSTCODETYPE 2209 2210 Return information about the first code unit of any matched string, for 2211 a non-anchored pattern. The third argument should point to a uint32_t 2212 variable. If there is a fixed first value, for example, the letter "c" 2213 from a pattern such as (cat|cow|coyote), 1 is returned, and the value 2214 can be retrieved using PCRE2_INFO_FIRSTCODEUNIT. If there is no fixed 2215 first value, but it is known that a match can occur only at the start 2216 of the subject or following a newline in the subject, 2 is returned. 2217 Otherwise, and for anchored patterns, 0 is returned. 2218 2219 PCRE2_INFO_FIRSTCODEUNIT 2220 2221 Return the value of the first code unit of any matched string for a 2222 pattern where PCRE2_INFO_FIRSTCODETYPE returns 1; otherwise return 0. 2223 The third argument should point to a uint32_t variable. In the 8-bit 2224 library, the value is always less than 256. In the 16-bit library the 2225 value can be up to 0xffff. In the 32-bit library in UTF-32 mode the 2226 value can be up to 0x10ffff, and up to 0xffffffff when not using UTF-32 2227 mode. 2228 2229 PCRE2_INFO_FRAMESIZE 2230 2231 Return the size (in bytes) of the data frames that are used to remember 2232 backtracking positions when the pattern is processed by pcre2_match() 2233 without the use of JIT. The third argument should point to a size_t 2234 variable. The frame size depends on the number of capturing parentheses 2235 in the pattern. Each additional capture group adds two PCRE2_SIZE vari- 2236 ables. 2237 2238 PCRE2_INFO_HASBACKSLASHC 2239 2240 Return 1 if the pattern contains any instances of \C, otherwise 0. The 2241 third argument should point to a uint32_t variable. 2242 2243 PCRE2_INFO_HASCRORLF 2244 2245 Return 1 if the pattern contains any explicit matches for CR or LF 2246 characters, otherwise 0. The third argument should point to a uint32_t 2247 variable. An explicit match is either a literal CR or LF character, or 2248 \r or \n or one of the equivalent hexadecimal or octal escape se- 2249 quences. 2250 2251 PCRE2_INFO_HEAPLIMIT 2252 2253 If the pattern set a heap memory limit by including an item of the form 2254 (*LIMIT_HEAP=nnnn) at the start, the value is returned. The third argu- 2255 ment should point to a uint32_t integer. If no such value has been set, 2256 the call to pcre2_pattern_info() returns the error PCRE2_ERROR_UNSET. 2257 Note that this limit will only be used during matching if it is less 2258 than the limit set or defaulted by the caller of the match function. 2259 2260 PCRE2_INFO_JCHANGED 2261 2262 Return 1 if the (?J) or (?-J) option setting is used in the pattern, 2263 otherwise 0. The third argument should point to a uint32_t variable. 2264 (?J) and (?-J) set and unset the local PCRE2_DUPNAMES option, respec- 2265 tively. 2266 2267 PCRE2_INFO_JITSIZE 2268 2269 If the compiled pattern was successfully processed by pcre2_jit_com- 2270 pile(), return the size of the JIT compiled code, otherwise return 2271 zero. The third argument should point to a size_t variable. 2272 2273 PCRE2_INFO_LASTCODETYPE 2274 2275 Returns 1 if there is a rightmost literal code unit that must exist in 2276 any matched string, other than at its start. The third argument should 2277 point to a uint32_t variable. If there is no such value, 0 is returned. 2278 When 1 is returned, the code unit value itself can be retrieved using 2279 PCRE2_INFO_LASTCODEUNIT. For anchored patterns, a last literal value is 2280 recorded only if it follows something of variable length. For example, 2281 for the pattern /^a\d+z\d+/ the returned value is 1 (with "z" returned 2282 from PCRE2_INFO_LASTCODEUNIT), but for /^a\dz\d/ the returned value is 2283 0. 2284 2285 PCRE2_INFO_LASTCODEUNIT 2286 2287 Return the value of the rightmost literal code unit that must exist in 2288 any matched string, other than at its start, for a pattern where 2289 PCRE2_INFO_LASTCODETYPE returns 1. Otherwise, return 0. The third argu- 2290 ment should point to a uint32_t variable. 2291 2292 PCRE2_INFO_MATCHEMPTY 2293 2294 Return 1 if the pattern might match an empty string, otherwise 0. The 2295 third argument should point to a uint32_t variable. When a pattern con- 2296 tains recursive subroutine calls it is not always possible to determine 2297 whether or not it can match an empty string. PCRE2 takes a cautious ap- 2298 proach and returns 1 in such cases. 2299 2300 PCRE2_INFO_MATCHLIMIT 2301 2302 If the pattern set a match limit by including an item of the form 2303 (*LIMIT_MATCH=nnnn) at the start, the value is returned. The third ar- 2304 gument should point to a uint32_t integer. If no such value has been 2305 set, the call to pcre2_pattern_info() returns the error PCRE2_ERROR_UN- 2306 SET. Note that this limit will only be used during matching if it is 2307 less than the limit set or defaulted by the caller of the match func- 2308 tion. 2309 2310 PCRE2_INFO_MAXLOOKBEHIND 2311 2312 A lookbehind assertion moves back a certain number of characters (not 2313 code units) when it starts to process each of its branches. This re- 2314 quest returns the largest of these backward moves. The third argument 2315 should point to a uint32_t integer. The simple assertions \b and \B re- 2316 quire a one-character lookbehind and cause PCRE2_INFO_MAXLOOKBEHIND to 2317 return 1 in the absence of anything longer. \A also registers a one- 2318 character lookbehind, though it does not actually inspect the previous 2319 character. 2320 2321 Note that this information is useful for multi-segment matching only if 2322 the pattern contains no nested lookbehinds. For example, the pattern 2323 (?<=a(?<=ba)c) returns a maximum lookbehind of 2, but when it is pro- 2324 cessed, the first lookbehind moves back by two characters, matches one 2325 character, then the nested lookbehind also moves back by two charac- 2326 ters. This puts the matching point three characters earlier than it was 2327 at the start. PCRE2_INFO_MAXLOOKBEHIND is really only useful as a de- 2328 bugging tool. See the pcre2partial documentation for a discussion of 2329 multi-segment matching. 2330 2331 PCRE2_INFO_MINLENGTH 2332 2333 If a minimum length for matching subject strings was computed, its 2334 value is returned. Otherwise the returned value is 0. This value is not 2335 computed when PCRE2_NO_START_OPTIMIZE is set. The value is a number of 2336 characters, which in UTF mode may be different from the number of code 2337 units. The third argument should point to a uint32_t variable. The 2338 value is a lower bound to the length of any matching string. There may 2339 not be any strings of that length that do actually match, but every 2340 string that does match is at least that long. 2341 2342 PCRE2_INFO_NAMECOUNT 2343 PCRE2_INFO_NAMEENTRYSIZE 2344 PCRE2_INFO_NAMETABLE 2345 2346 PCRE2 supports the use of named as well as numbered capturing parenthe- 2347 ses. The names are just an additional way of identifying the parenthe- 2348 ses, which still acquire numbers. Several convenience functions such as 2349 pcre2_substring_get_byname() are provided for extracting captured sub- 2350 strings by name. It is also possible to extract the data directly, by 2351 first converting the name to a number in order to access the correct 2352 pointers in the output vector (described with pcre2_match() below). To 2353 do the conversion, you need to use the name-to-number map, which is de- 2354 scribed by these three values. 2355 2356 The map consists of a number of fixed-size entries. PCRE2_INFO_NAME- 2357 COUNT gives the number of entries, and PCRE2_INFO_NAMEENTRYSIZE gives 2358 the size of each entry in code units; both of these return a uint32_t 2359 value. The entry size depends on the length of the longest name. 2360 2361 PCRE2_INFO_NAMETABLE returns a pointer to the first entry of the table. 2362 This is a PCRE2_SPTR pointer to a block of code units. In the 8-bit li- 2363 brary, the first two bytes of each entry are the number of the captur- 2364 ing parenthesis, most significant byte first. In the 16-bit library, 2365 the pointer points to 16-bit code units, the first of which contains 2366 the parenthesis number. In the 32-bit library, the pointer points to 2367 32-bit code units, the first of which contains the parenthesis number. 2368 The rest of the entry is the corresponding name, zero terminated. 2369 2370 The names are in alphabetical order. If (?| is used to create multiple 2371 capture groups with the same number, as described in the section on du- 2372 plicate group numbers in the pcre2pattern page, the groups may be given 2373 the same name, but there is only one entry in the table. Different 2374 names for groups of the same number are not permitted. 2375 2376 Duplicate names for capture groups with different numbers are permit- 2377 ted, but only if PCRE2_DUPNAMES is set. They appear in the table in the 2378 order in which they were found in the pattern. In the absence of (?| 2379 this is the order of increasing number; when (?| is used this is not 2380 necessarily the case because later capture groups may have lower num- 2381 bers. 2382 2383 As a simple example of the name/number table, consider the following 2384 pattern after compilation by the 8-bit library (assume PCRE2_EXTENDED 2385 is set, so white space - including newlines - is ignored): 2386 2387 (?<date> (?<year>(\d\d)?\d\d) - 2388 (?<month>\d\d) - (?<day>\d\d) ) 2389 2390 There are four named capture groups, so the table has four entries, and 2391 each entry in the table is eight bytes long. The table is as follows, 2392 with non-printing bytes shows in hexadecimal, and undefined bytes shown 2393 as ??: 2394 2395 00 01 d a t e 00 ?? 2396 00 05 d a y 00 ?? ?? 2397 00 04 m o n t h 00 2398 00 02 y e a r 00 ?? 2399 2400 When writing code to extract data from named capture groups using the 2401 name-to-number map, remember that the length of the entries is likely 2402 to be different for each compiled pattern. 2403 2404 PCRE2_INFO_NEWLINE 2405 2406 The output is one of the following uint32_t values: 2407 2408 PCRE2_NEWLINE_CR Carriage return (CR) 2409 PCRE2_NEWLINE_LF Linefeed (LF) 2410 PCRE2_NEWLINE_CRLF Carriage return, linefeed (CRLF) 2411 PCRE2_NEWLINE_ANY Any Unicode line ending 2412 PCRE2_NEWLINE_ANYCRLF Any of CR, LF, or CRLF 2413 PCRE2_NEWLINE_NUL The NUL character (binary zero) 2414 2415 This identifies the character sequence that will be recognized as mean- 2416 ing "newline" while matching. 2417 2418 PCRE2_INFO_SIZE 2419 2420 Return the size of the compiled pattern in bytes (for all three li- 2421 braries). The third argument should point to a size_t variable. This 2422 value includes the size of the general data block that precedes the 2423 code units of the compiled pattern itself. The value that is used when 2424 pcre2_compile() is getting memory in which to place the compiled pat- 2425 tern may be slightly larger than the value returned by this option, be- 2426 cause there are cases where the code that calculates the size has to 2427 over-estimate. Processing a pattern with the JIT compiler does not al- 2428 ter the value returned by this option. 2429 2430 2431INFORMATION ABOUT A PATTERN'S CALLOUTS 2432 2433 int pcre2_callout_enumerate(const pcre2_code *code, 2434 int (*callback)(pcre2_callout_enumerate_block *, void *), 2435 void *user_data); 2436 2437 A script language that supports the use of string arguments in callouts 2438 might like to scan all the callouts in a pattern before running the 2439 match. This can be done by calling pcre2_callout_enumerate(). The first 2440 argument is a pointer to a compiled pattern, the second points to a 2441 callback function, and the third is arbitrary user data. The callback 2442 function is called for every callout in the pattern in the order in 2443 which they appear. Its first argument is a pointer to a callout enumer- 2444 ation block, and its second argument is the user_data value that was 2445 passed to pcre2_callout_enumerate(). The contents of the callout enu- 2446 meration block are described in the pcre2callout documentation, which 2447 also gives further details about callouts. 2448 2449 2450SERIALIZATION AND PRECOMPILING 2451 2452 It is possible to save compiled patterns on disc or elsewhere, and 2453 reload them later, subject to a number of restrictions. The host on 2454 which the patterns are reloaded must be running the same version of 2455 PCRE2, with the same code unit width, and must also have the same endi- 2456 anness, pointer width, and PCRE2_SIZE type. Before compiled patterns 2457 can be saved, they must be converted to a "serialized" form, which in 2458 the case of PCRE2 is really just a bytecode dump. The functions whose 2459 names begin with pcre2_serialize_ are used for converting to and from 2460 the serialized form. They are described in the pcre2serialize documen- 2461 tation. Note that PCRE2 serialization does not convert compiled pat- 2462 terns to an abstract format like Java or .NET serialization. 2463 2464 2465THE MATCH DATA BLOCK 2466 2467 pcre2_match_data *pcre2_match_data_create(uint32_t ovecsize, 2468 pcre2_general_context *gcontext); 2469 2470 pcre2_match_data *pcre2_match_data_create_from_pattern( 2471 const pcre2_code *code, pcre2_general_context *gcontext); 2472 2473 void pcre2_match_data_free(pcre2_match_data *match_data); 2474 2475 Information about a successful or unsuccessful match is placed in a 2476 match data block, which is an opaque structure that is accessed by 2477 function calls. In particular, the match data block contains a vector 2478 of offsets into the subject string that define the matched parts of the 2479 subject. This is known as the ovector. 2480 2481 Before calling pcre2_match(), pcre2_dfa_match(), or pcre2_jit_match() 2482 you must create a match data block by calling one of the creation func- 2483 tions above. For pcre2_match_data_create(), the first argument is the 2484 number of pairs of offsets in the ovector. 2485 2486 When using pcre2_match(), one pair of offsets is required to identify 2487 the string that matched the whole pattern, with an additional pair for 2488 each captured substring. For example, a value of 4 creates enough space 2489 to record the matched portion of the subject plus three captured sub- 2490 strings. 2491 2492 When using pcre2_dfa_match() there may be multiple matched substrings 2493 of different lengths at the same point in the subject. The ovector 2494 should be made large enough to hold as many as are expected. 2495 2496 A minimum of at least 1 pair is imposed by pcre2_match_data_create(), 2497 so it is always possible to return the overall matched string in the 2498 case of pcre2_match() or the longest match in the case of 2499 pcre2_dfa_match(). 2500 2501 The second argument of pcre2_match_data_create() is a pointer to a gen- 2502 eral context, which can specify custom memory management for obtaining 2503 the memory for the match data block. If you are not using custom memory 2504 management, pass NULL, which causes malloc() to be used. 2505 2506 For pcre2_match_data_create_from_pattern(), the first argument is a 2507 pointer to a compiled pattern. The ovector is created to be exactly the 2508 right size to hold all the substrings a pattern might capture when 2509 matched using pcre2_match(). You should not use this call when matching 2510 with pcre2_dfa_match(). The second argument is again a pointer to a 2511 general context, but in this case if NULL is passed, the memory is ob- 2512 tained using the same allocator that was used for the compiled pattern 2513 (custom or default). 2514 2515 A match data block can be used many times, with the same or different 2516 compiled patterns. You can extract information from a match data block 2517 after a match operation has finished, using functions that are de- 2518 scribed in the sections on matched strings and other match data below. 2519 2520 When a call of pcre2_match() fails, valid data is available in the 2521 match block only when the error is PCRE2_ERROR_NOMATCH, PCRE2_ER- 2522 ROR_PARTIAL, or one of the error codes for an invalid UTF string. Ex- 2523 actly what is available depends on the error, and is detailed below. 2524 2525 When one of the matching functions is called, pointers to the compiled 2526 pattern and the subject string are set in the match data block so that 2527 they can be referenced by the extraction functions after a successful 2528 match. After running a match, you must not free a compiled pattern or a 2529 subject string until after all operations on the match data block (for 2530 that match) have taken place, unless, in the case of the subject 2531 string, you have used the PCRE2_COPY_MATCHED_SUBJECT option, which is 2532 described in the section entitled "Option bits for pcre2_match()" be- 2533 low. 2534 2535 When a match data block itself is no longer needed, it should be freed 2536 by calling pcre2_match_data_free(). If this function is called with a 2537 NULL argument, it returns immediately, without doing anything. 2538 2539 2540MATCHING A PATTERN: THE TRADITIONAL FUNCTION 2541 2542 int pcre2_match(const pcre2_code *code, PCRE2_SPTR subject, 2543 PCRE2_SIZE length, PCRE2_SIZE startoffset, 2544 uint32_t options, pcre2_match_data *match_data, 2545 pcre2_match_context *mcontext); 2546 2547 The function pcre2_match() is called to match a subject string against 2548 a compiled pattern, which is passed in the code argument. You can call 2549 pcre2_match() with the same code argument as many times as you like, in 2550 order to find multiple matches in the subject string or to match dif- 2551 ferent subject strings with the same pattern. 2552 2553 This function is the main matching facility of the library, and it op- 2554 erates in a Perl-like manner. For specialist use there is also an al- 2555 ternative matching function, which is described below in the section 2556 about the pcre2_dfa_match() function. 2557 2558 Here is an example of a simple call to pcre2_match(): 2559 2560 pcre2_match_data *md = pcre2_match_data_create(4, NULL); 2561 int rc = pcre2_match( 2562 re, /* result of pcre2_compile() */ 2563 "some string", /* the subject string */ 2564 11, /* the length of the subject string */ 2565 0, /* start at offset 0 in the subject */ 2566 0, /* default options */ 2567 md, /* the match data block */ 2568 NULL); /* a match context; NULL means use defaults */ 2569 2570 If the subject string is zero-terminated, the length can be given as 2571 PCRE2_ZERO_TERMINATED. A match context must be provided if certain less 2572 common matching parameters are to be changed. For details, see the sec- 2573 tion on the match context above. 2574 2575 The string to be matched by pcre2_match() 2576 2577 The subject string is passed to pcre2_match() as a pointer in subject, 2578 a length in length, and a starting offset in startoffset. The length 2579 and offset are in code units, not characters. That is, they are in 2580 bytes for the 8-bit library, 16-bit code units for the 16-bit library, 2581 and 32-bit code units for the 32-bit library, whether or not UTF pro- 2582 cessing is enabled. As a special case, if subject is NULL and length is 2583 zero, the subject is assumed to be an empty string. If length is non- 2584 zero, an error occurs if subject is NULL. 2585 2586 If startoffset is greater than the length of the subject, pcre2_match() 2587 returns PCRE2_ERROR_BADOFFSET. When the starting offset is zero, the 2588 search for a match starts at the beginning of the subject, and this is 2589 by far the most common case. In UTF-8 or UTF-16 mode, the starting off- 2590 set must point to the start of a character, or to the end of the sub- 2591 ject (in UTF-32 mode, one code unit equals one character, so all off- 2592 sets are valid). Like the pattern string, the subject may contain bi- 2593 nary zeros. 2594 2595 A non-zero starting offset is useful when searching for another match 2596 in the same subject by calling pcre2_match() again after a previous 2597 success. Setting startoffset differs from passing over a shortened 2598 string and setting PCRE2_NOTBOL in the case of a pattern that begins 2599 with any kind of lookbehind. For example, consider the pattern 2600 2601 \Biss\B 2602 2603 which finds occurrences of "iss" in the middle of words. (\B matches 2604 only if the current position in the subject is not a word boundary.) 2605 When applied to the string "Mississippi" the first call to 2606 pcre2_match() finds the first occurrence. If pcre2_match() is called 2607 again with just the remainder of the subject, namely "issippi", it does 2608 not match, because \B is always false at the start of the subject, 2609 which is deemed to be a word boundary. However, if pcre2_match() is 2610 passed the entire string again, but with startoffset set to 4, it finds 2611 the second occurrence of "iss" because it is able to look behind the 2612 starting point to discover that it is preceded by a letter. 2613 2614 Finding all the matches in a subject is tricky when the pattern can 2615 match an empty string. It is possible to emulate Perl's /g behaviour by 2616 first trying the match again at the same offset, with the 2617 PCRE2_NOTEMPTY_ATSTART and PCRE2_ANCHORED options, and then if that 2618 fails, advancing the starting offset and trying an ordinary match 2619 again. There is some code that demonstrates how to do this in the 2620 pcre2demo sample program. In the most general case, you have to check 2621 to see if the newline convention recognizes CRLF as a newline, and if 2622 so, and the current character is CR followed by LF, advance the start- 2623 ing offset by two characters instead of one. 2624 2625 If a non-zero starting offset is passed when the pattern is anchored, a 2626 single attempt to match at the given offset is made. This can only suc- 2627 ceed if the pattern does not require the match to be at the start of 2628 the subject. In other words, the anchoring must be the result of set- 2629 ting the PCRE2_ANCHORED option or the use of .* with PCRE2_DOTALL, not 2630 by starting the pattern with ^ or \A. 2631 2632 Option bits for pcre2_match() 2633 2634 The unused bits of the options argument for pcre2_match() must be zero. 2635 The only bits that may be set are PCRE2_ANCHORED, 2636 PCRE2_COPY_MATCHED_SUBJECT, PCRE2_ENDANCHORED, PCRE2_NOTBOL, PCRE2_NO- 2637 TEOL, PCRE2_NOTEMPTY, PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_JIT, 2638 PCRE2_NO_UTF_CHECK, PCRE2_PARTIAL_HARD, and PCRE2_PARTIAL_SOFT. Their 2639 action is described below. 2640 2641 Setting PCRE2_ANCHORED or PCRE2_ENDANCHORED at match time is not sup- 2642 ported by the just-in-time (JIT) compiler. If it is set, JIT matching 2643 is disabled and the interpretive code in pcre2_match() is run. Apart 2644 from PCRE2_NO_JIT (obviously), the remaining options are supported for 2645 JIT matching. 2646 2647 PCRE2_ANCHORED 2648 2649 The PCRE2_ANCHORED option limits pcre2_match() to matching at the first 2650 matching position. If a pattern was compiled with PCRE2_ANCHORED, or 2651 turned out to be anchored by virtue of its contents, it cannot be made 2652 unachored at matching time. Note that setting the option at match time 2653 disables JIT matching. 2654 2655 PCRE2_COPY_MATCHED_SUBJECT 2656 2657 By default, a pointer to the subject is remembered in the match data 2658 block so that, after a successful match, it can be referenced by the 2659 substring extraction functions. This means that the subject's memory 2660 must not be freed until all such operations are complete. For some ap- 2661 plications where the lifetime of the subject string is not guaranteed, 2662 it may be necessary to make a copy of the subject string, but it is 2663 wasteful to do this unless the match is successful. After a successful 2664 match, if PCRE2_COPY_MATCHED_SUBJECT is set, the subject is copied and 2665 the new pointer is remembered in the match data block instead of the 2666 original subject pointer. The memory allocator that was used for the 2667 match block itself is used. The copy is automatically freed when 2668 pcre2_match_data_free() is called to free the match data block. It is 2669 also automatically freed if the match data block is re-used for another 2670 match operation. 2671 2672 PCRE2_ENDANCHORED 2673 2674 If the PCRE2_ENDANCHORED option is set, any string that pcre2_match() 2675 matches must be right at the end of the subject string. Note that set- 2676 ting the option at match time disables JIT matching. 2677 2678 PCRE2_NOTBOL 2679 2680 This option specifies that first character of the subject string is not 2681 the beginning of a line, so the circumflex metacharacter should not 2682 match before it. Setting this without having set PCRE2_MULTILINE at 2683 compile time causes circumflex never to match. This option affects only 2684 the behaviour of the circumflex metacharacter. It does not affect \A. 2685 2686 PCRE2_NOTEOL 2687 2688 This option specifies that the end of the subject string is not the end 2689 of a line, so the dollar metacharacter should not match it nor (except 2690 in multiline mode) a newline immediately before it. Setting this with- 2691 out having set PCRE2_MULTILINE at compile time causes dollar never to 2692 match. This option affects only the behaviour of the dollar metacharac- 2693 ter. It does not affect \Z or \z. 2694 2695 PCRE2_NOTEMPTY 2696 2697 An empty string is not considered to be a valid match if this option is 2698 set. If there are alternatives in the pattern, they are tried. If all 2699 the alternatives match the empty string, the entire match fails. For 2700 example, if the pattern 2701 2702 a?b? 2703 2704 is applied to a string not beginning with "a" or "b", it matches an 2705 empty string at the start of the subject. With PCRE2_NOTEMPTY set, this 2706 match is not valid, so pcre2_match() searches further into the string 2707 for occurrences of "a" or "b". 2708 2709 PCRE2_NOTEMPTY_ATSTART 2710 2711 This is like PCRE2_NOTEMPTY, except that it locks out an empty string 2712 match only at the first matching position, that is, at the start of the 2713 subject plus the starting offset. An empty string match later in the 2714 subject is permitted. If the pattern is anchored, such a match can oc- 2715 cur only if the pattern contains \K. 2716 2717 PCRE2_NO_JIT 2718 2719 By default, if a pattern has been successfully processed by 2720 pcre2_jit_compile(), JIT is automatically used when pcre2_match() is 2721 called with options that JIT supports. Setting PCRE2_NO_JIT disables 2722 the use of JIT; it forces matching to be done by the interpreter. 2723 2724 PCRE2_NO_UTF_CHECK 2725 2726 When PCRE2_UTF is set at compile time, the validity of the subject as a 2727 UTF string is checked unless PCRE2_NO_UTF_CHECK is passed to 2728 pcre2_match() or PCRE2_MATCH_INVALID_UTF was passed to pcre2_compile(). 2729 The latter special case is discussed in detail in the pcre2unicode doc- 2730 umentation. 2731 2732 In the default case, if a non-zero starting offset is given, the check 2733 is applied only to that part of the subject that could be inspected 2734 during matching, and there is a check that the starting offset points 2735 to the first code unit of a character or to the end of the subject. If 2736 there are no lookbehind assertions in the pattern, the check starts at 2737 the starting offset. Otherwise, it starts at the length of the longest 2738 lookbehind before the starting offset, or at the start of the subject 2739 if there are not that many characters before the starting offset. Note 2740 that the sequences \b and \B are one-character lookbehinds. 2741 2742 The check is carried out before any other processing takes place, and a 2743 negative error code is returned if the check fails. There are several 2744 UTF error codes for each code unit width, corresponding to different 2745 problems with the code unit sequence. There are discussions about the 2746 validity of UTF-8 strings, UTF-16 strings, and UTF-32 strings in the 2747 pcre2unicode documentation. 2748 2749 If you know that your subject is valid, and you want to skip this check 2750 for performance reasons, you can set the PCRE2_NO_UTF_CHECK option when 2751 calling pcre2_match(). You might want to do this for the second and 2752 subsequent calls to pcre2_match() if you are making repeated calls to 2753 find multiple matches in the same subject string. 2754 2755 Warning: Unless PCRE2_MATCH_INVALID_UTF was set at compile time, when 2756 PCRE2_NO_UTF_CHECK is set at match time the effect of passing an in- 2757 valid string as a subject, or an invalid value of startoffset, is unde- 2758 fined. Your program may crash or loop indefinitely or give wrong re- 2759 sults. 2760 2761 PCRE2_PARTIAL_HARD 2762 PCRE2_PARTIAL_SOFT 2763 2764 These options turn on the partial matching feature. A partial match oc- 2765 curs if the end of the subject string is reached successfully, but 2766 there are not enough subject characters to complete the match. In addi- 2767 tion, either at least one character must have been inspected or the 2768 pattern must contain a lookbehind, or the pattern must be one that 2769 could match an empty string. 2770 2771 If this situation arises when PCRE2_PARTIAL_SOFT (but not PCRE2_PAR- 2772 TIAL_HARD) is set, matching continues by testing any remaining alterna- 2773 tives. Only if no complete match can be found is PCRE2_ERROR_PARTIAL 2774 returned instead of PCRE2_ERROR_NOMATCH. In other words, PCRE2_PAR- 2775 TIAL_SOFT specifies that the caller is prepared to handle a partial 2776 match, but only if no complete match can be found. 2777 2778 If PCRE2_PARTIAL_HARD is set, it overrides PCRE2_PARTIAL_SOFT. In this 2779 case, if a partial match is found, pcre2_match() immediately returns 2780 PCRE2_ERROR_PARTIAL, without considering any other alternatives. In 2781 other words, when PCRE2_PARTIAL_HARD is set, a partial match is consid- 2782 ered to be more important that an alternative complete match. 2783 2784 There is a more detailed discussion of partial and multi-segment match- 2785 ing, with examples, in the pcre2partial documentation. 2786 2787 2788NEWLINE HANDLING WHEN MATCHING 2789 2790 When PCRE2 is built, a default newline convention is set; this is usu- 2791 ally the standard convention for the operating system. The default can 2792 be overridden in a compile context by calling pcre2_set_newline(). It 2793 can also be overridden by starting a pattern string with, for example, 2794 (*CRLF), as described in the section on newline conventions in the 2795 pcre2pattern page. During matching, the newline choice affects the be- 2796 haviour of the dot, circumflex, and dollar metacharacters. It may also 2797 alter the way the match starting position is advanced after a match 2798 failure for an unanchored pattern. 2799 2800 When PCRE2_NEWLINE_CRLF, PCRE2_NEWLINE_ANYCRLF, or PCRE2_NEWLINE_ANY is 2801 set as the newline convention, and a match attempt for an unanchored 2802 pattern fails when the current starting position is at a CRLF sequence, 2803 and the pattern contains no explicit matches for CR or LF characters, 2804 the match position is advanced by two characters instead of one, in 2805 other words, to after the CRLF. 2806 2807 The above rule is a compromise that makes the most common cases work as 2808 expected. For example, if the pattern is .+A (and the PCRE2_DOTALL op- 2809 tion is not set), it does not match the string "\r\nA" because, after 2810 failing at the start, it skips both the CR and the LF before retrying. 2811 However, the pattern [\r\n]A does match that string, because it con- 2812 tains an explicit CR or LF reference, and so advances only by one char- 2813 acter after the first failure. 2814 2815 An explicit match for CR of LF is either a literal appearance of one of 2816 those characters in the pattern, or one of the \r or \n or equivalent 2817 octal or hexadecimal escape sequences. Implicit matches such as [^X] do 2818 not count, nor does \s, even though it includes CR and LF in the char- 2819 acters that it matches. 2820 2821 Notwithstanding the above, anomalous effects may still occur when CRLF 2822 is a valid newline sequence and explicit \r or \n escapes appear in the 2823 pattern. 2824 2825 2826HOW PCRE2_MATCH() RETURNS A STRING AND CAPTURED SUBSTRINGS 2827 2828 uint32_t pcre2_get_ovector_count(pcre2_match_data *match_data); 2829 2830 PCRE2_SIZE *pcre2_get_ovector_pointer(pcre2_match_data *match_data); 2831 2832 In general, a pattern matches a certain portion of the subject, and in 2833 addition, further substrings from the subject may be picked out by 2834 parenthesized parts of the pattern. Following the usage in Jeffrey 2835 Friedl's book, this is called "capturing" in what follows, and the 2836 phrase "capture group" (Perl terminology) is used for a fragment of a 2837 pattern that picks out a substring. PCRE2 supports several other kinds 2838 of parenthesized group that do not cause substrings to be captured. The 2839 pcre2_pattern_info() function can be used to find out how many capture 2840 groups there are in a compiled pattern. 2841 2842 You can use auxiliary functions for accessing captured substrings by 2843 number or by name, as described in sections below. 2844 2845 Alternatively, you can make direct use of the vector of PCRE2_SIZE val- 2846 ues, called the ovector, which contains the offsets of captured 2847 strings. It is part of the match data block. The function 2848 pcre2_get_ovector_pointer() returns the address of the ovector, and 2849 pcre2_get_ovector_count() returns the number of pairs of values it con- 2850 tains. 2851 2852 Within the ovector, the first in each pair of values is set to the off- 2853 set of the first code unit of a substring, and the second is set to the 2854 offset of the first code unit after the end of a substring. These val- 2855 ues are always code unit offsets, not character offsets. That is, they 2856 are byte offsets in the 8-bit library, 16-bit offsets in the 16-bit li- 2857 brary, and 32-bit offsets in the 32-bit library. 2858 2859 After a partial match (error return PCRE2_ERROR_PARTIAL), only the 2860 first pair of offsets (that is, ovector[0] and ovector[1]) are set. 2861 They identify the part of the subject that was partially matched. See 2862 the pcre2partial documentation for details of partial matching. 2863 2864 After a fully successful match, the first pair of offsets identifies 2865 the portion of the subject string that was matched by the entire pat- 2866 tern. The next pair is used for the first captured substring, and so 2867 on. The value returned by pcre2_match() is one more than the highest 2868 numbered pair that has been set. For example, if two substrings have 2869 been captured, the returned value is 3. If there are no captured sub- 2870 strings, the return value from a successful match is 1, indicating that 2871 just the first pair of offsets has been set. 2872 2873 If a pattern uses the \K escape sequence within a positive assertion, 2874 the reported start of a successful match can be greater than the end of 2875 the match. For example, if the pattern (?=ab\K) is matched against 2876 "ab", the start and end offset values for the match are 2 and 0. 2877 2878 If a capture group is matched repeatedly within a single match opera- 2879 tion, it is the last portion of the subject that it matched that is re- 2880 turned. 2881 2882 If the ovector is too small to hold all the captured substring offsets, 2883 as much as possible is filled in, and the function returns a value of 2884 zero. If captured substrings are not of interest, pcre2_match() may be 2885 called with a match data block whose ovector is of minimum length (that 2886 is, one pair). 2887 2888 It is possible for capture group number n+1 to match some part of the 2889 subject when group n has not been used at all. For example, if the 2890 string "abc" is matched against the pattern (a|(z))(bc) the return from 2891 the function is 4, and groups 1 and 3 are matched, but 2 is not. When 2892 this happens, both values in the offset pairs corresponding to unused 2893 groups are set to PCRE2_UNSET. 2894 2895 Offset values that correspond to unused groups at the end of the ex- 2896 pression are also set to PCRE2_UNSET. For example, if the string "abc" 2897 is matched against the pattern (abc)(x(yz)?)? groups 2 and 3 are not 2898 matched. The return from the function is 2, because the highest used 2899 capture group number is 1. The offsets for for the second and third 2900 capture groupss (assuming the vector is large enough, of course) are 2901 set to PCRE2_UNSET. 2902 2903 Elements in the ovector that do not correspond to capturing parentheses 2904 in the pattern are never changed. That is, if a pattern contains n cap- 2905 turing parentheses, no more than ovector[0] to ovector[2n+1] are set by 2906 pcre2_match(). The other elements retain whatever values they previ- 2907 ously had. After a failed match attempt, the contents of the ovector 2908 are unchanged. 2909 2910 2911OTHER INFORMATION ABOUT A MATCH 2912 2913 PCRE2_SPTR pcre2_get_mark(pcre2_match_data *match_data); 2914 2915 PCRE2_SIZE pcre2_get_startchar(pcre2_match_data *match_data); 2916 2917 As well as the offsets in the ovector, other information about a match 2918 is retained in the match data block and can be retrieved by the above 2919 functions in appropriate circumstances. If they are called at other 2920 times, the result is undefined. 2921 2922 After a successful match, a partial match (PCRE2_ERROR_PARTIAL), or a 2923 failure to match (PCRE2_ERROR_NOMATCH), a mark name may be available. 2924 The function pcre2_get_mark() can be called to access this name, which 2925 can be specified in the pattern by any of the backtracking control 2926 verbs, not just (*MARK). The same function applies to all the verbs. It 2927 returns a pointer to the zero-terminated name, which is within the com- 2928 piled pattern. If no name is available, NULL is returned. The length of 2929 the name (excluding the terminating zero) is stored in the code unit 2930 that precedes the name. You should use this length instead of relying 2931 on the terminating zero if the name might contain a binary zero. 2932 2933 After a successful match, the name that is returned is the last mark 2934 name encountered on the matching path through the pattern. Instances of 2935 backtracking verbs without names do not count. Thus, for example, if 2936 the matching path contains (*MARK:A)(*PRUNE), the name "A" is returned. 2937 After a "no match" or a partial match, the last encountered name is re- 2938 turned. For example, consider this pattern: 2939 2940 ^(*MARK:A)((*MARK:B)a|b)c 2941 2942 When it matches "bc", the returned name is A. The B mark is "seen" in 2943 the first branch of the group, but it is not on the matching path. On 2944 the other hand, when this pattern fails to match "bx", the returned 2945 name is B. 2946 2947 Warning: By default, certain start-of-match optimizations are used to 2948 give a fast "no match" result in some situations. For example, if the 2949 anchoring is removed from the pattern above, there is an initial check 2950 for the presence of "c" in the subject before running the matching en- 2951 gine. This check fails for "bx", causing a match failure without seeing 2952 any marks. You can disable the start-of-match optimizations by setting 2953 the PCRE2_NO_START_OPTIMIZE option for pcre2_compile() or by starting 2954 the pattern with (*NO_START_OPT). 2955 2956 After a successful match, a partial match, or one of the invalid UTF 2957 errors (for example, PCRE2_ERROR_UTF8_ERR5), pcre2_get_startchar() can 2958 be called. After a successful or partial match it returns the code unit 2959 offset of the character at which the match started. For a non-partial 2960 match, this can be different to the value of ovector[0] if the pattern 2961 contains the \K escape sequence. After a partial match, however, this 2962 value is always the same as ovector[0] because \K does not affect the 2963 result of a partial match. 2964 2965 After a UTF check failure, pcre2_get_startchar() can be used to obtain 2966 the code unit offset of the invalid UTF character. Details are given in 2967 the pcre2unicode page. 2968 2969 2970ERROR RETURNS FROM pcre2_match() 2971 2972 If pcre2_match() fails, it returns a negative number. This can be con- 2973 verted to a text string by calling the pcre2_get_error_message() func- 2974 tion (see "Obtaining a textual error message" below). Negative error 2975 codes are also returned by other functions, and are documented with 2976 them. The codes are given names in the header file. If UTF checking is 2977 in force and an invalid UTF subject string is detected, one of a number 2978 of UTF-specific negative error codes is returned. Details are given in 2979 the pcre2unicode page. The following are the other errors that may be 2980 returned by pcre2_match(): 2981 2982 PCRE2_ERROR_NOMATCH 2983 2984 The subject string did not match the pattern. 2985 2986 PCRE2_ERROR_PARTIAL 2987 2988 The subject string did not match, but it did match partially. See the 2989 pcre2partial documentation for details of partial matching. 2990 2991 PCRE2_ERROR_BADMAGIC 2992 2993 PCRE2 stores a 4-byte "magic number" at the start of the compiled code, 2994 to catch the case when it is passed a junk pointer. This is the error 2995 that is returned when the magic number is not present. 2996 2997 PCRE2_ERROR_BADMODE 2998 2999 This error is given when a compiled pattern is passed to a function in 3000 a library of a different code unit width, for example, a pattern com- 3001 piled by the 8-bit library is passed to a 16-bit or 32-bit library 3002 function. 3003 3004 PCRE2_ERROR_BADOFFSET 3005 3006 The value of startoffset was greater than the length of the subject. 3007 3008 PCRE2_ERROR_BADOPTION 3009 3010 An unrecognized bit was set in the options argument. 3011 3012 PCRE2_ERROR_BADUTFOFFSET 3013 3014 The UTF code unit sequence that was passed as a subject was checked and 3015 found to be valid (the PCRE2_NO_UTF_CHECK option was not set), but the 3016 value of startoffset did not point to the beginning of a UTF character 3017 or the end of the subject. 3018 3019 PCRE2_ERROR_CALLOUT 3020 3021 This error is never generated by pcre2_match() itself. It is provided 3022 for use by callout functions that want to cause pcre2_match() or 3023 pcre2_callout_enumerate() to return a distinctive error code. See the 3024 pcre2callout documentation for details. 3025 3026 PCRE2_ERROR_DEPTHLIMIT 3027 3028 The nested backtracking depth limit was reached. 3029 3030 PCRE2_ERROR_HEAPLIMIT 3031 3032 The heap limit was reached. 3033 3034 PCRE2_ERROR_INTERNAL 3035 3036 An unexpected internal error has occurred. This error could be caused 3037 by a bug in PCRE2 or by overwriting of the compiled pattern. 3038 3039 PCRE2_ERROR_JIT_STACKLIMIT 3040 3041 This error is returned when a pattern that was successfully studied us- 3042 ing JIT is being matched, but the memory available for the just-in-time 3043 processing stack is not large enough. See the pcre2jit documentation 3044 for more details. 3045 3046 PCRE2_ERROR_MATCHLIMIT 3047 3048 The backtracking match limit was reached. 3049 3050 PCRE2_ERROR_NOMEMORY 3051 3052 If a pattern contains many nested backtracking points, heap memory is 3053 used to remember them. This error is given when the memory allocation 3054 function (default or custom) fails. Note that a different error, 3055 PCRE2_ERROR_HEAPLIMIT, is given if the amount of memory needed exceeds 3056 the heap limit. PCRE2_ERROR_NOMEMORY is also returned if 3057 PCRE2_COPY_MATCHED_SUBJECT is set and memory allocation fails. 3058 3059 PCRE2_ERROR_NULL 3060 3061 Either the code, subject, or match_data argument was passed as NULL. 3062 3063 PCRE2_ERROR_RECURSELOOP 3064 3065 This error is returned when pcre2_match() detects a recursion loop 3066 within the pattern. Specifically, it means that either the whole pat- 3067 tern or a capture group has been called recursively for the second time 3068 at the same position in the subject string. Some simple patterns that 3069 might do this are detected and faulted at compile time, but more com- 3070 plicated cases, in particular mutual recursions between two different 3071 groups, cannot be detected until matching is attempted. 3072 3073 3074OBTAINING A TEXTUAL ERROR MESSAGE 3075 3076 int pcre2_get_error_message(int errorcode, PCRE2_UCHAR *buffer, 3077 PCRE2_SIZE bufflen); 3078 3079 A text message for an error code from any PCRE2 function (compile, 3080 match, or auxiliary) can be obtained by calling pcre2_get_error_mes- 3081 sage(). The code is passed as the first argument, with the remaining 3082 two arguments specifying a code unit buffer and its length in code 3083 units, into which the text message is placed. The message is returned 3084 in code units of the appropriate width for the library that is being 3085 used. 3086 3087 The returned message is terminated with a trailing zero, and the func- 3088 tion returns the number of code units used, excluding the trailing 3089 zero. If the error number is unknown, the negative error code PCRE2_ER- 3090 ROR_BADDATA is returned. If the buffer is too small, the message is 3091 truncated (but still with a trailing zero), and the negative error code 3092 PCRE2_ERROR_NOMEMORY is returned. None of the messages are very long; 3093 a buffer size of 120 code units is ample. 3094 3095 3096EXTRACTING CAPTURED SUBSTRINGS BY NUMBER 3097 3098 int pcre2_substring_length_bynumber(pcre2_match_data *match_data, 3099 uint32_t number, PCRE2_SIZE *length); 3100 3101 int pcre2_substring_copy_bynumber(pcre2_match_data *match_data, 3102 uint32_t number, PCRE2_UCHAR *buffer, 3103 PCRE2_SIZE *bufflen); 3104 3105 int pcre2_substring_get_bynumber(pcre2_match_data *match_data, 3106 uint32_t number, PCRE2_UCHAR **bufferptr, 3107 PCRE2_SIZE *bufflen); 3108 3109 void pcre2_substring_free(PCRE2_UCHAR *buffer); 3110 3111 Captured substrings can be accessed directly by using the ovector as 3112 described above. For convenience, auxiliary functions are provided for 3113 extracting captured substrings as new, separate, zero-terminated 3114 strings. A substring that contains a binary zero is correctly extracted 3115 and has a further zero added on the end, but the result is not, of 3116 course, a C string. 3117 3118 The functions in this section identify substrings by number. The number 3119 zero refers to the entire matched substring, with higher numbers refer- 3120 ring to substrings captured by parenthesized groups. After a partial 3121 match, only substring zero is available. An attempt to extract any 3122 other substring gives the error PCRE2_ERROR_PARTIAL. The next section 3123 describes similar functions for extracting captured substrings by name. 3124 3125 If a pattern uses the \K escape sequence within a positive assertion, 3126 the reported start of a successful match can be greater than the end of 3127 the match. For example, if the pattern (?=ab\K) is matched against 3128 "ab", the start and end offset values for the match are 2 and 0. In 3129 this situation, calling these functions with a zero substring number 3130 extracts a zero-length empty string. 3131 3132 You can find the length in code units of a captured substring without 3133 extracting it by calling pcre2_substring_length_bynumber(). The first 3134 argument is a pointer to the match data block, the second is the group 3135 number, and the third is a pointer to a variable into which the length 3136 is placed. If you just want to know whether or not the substring has 3137 been captured, you can pass the third argument as NULL. 3138 3139 The pcre2_substring_copy_bynumber() function copies a captured sub- 3140 string into a supplied buffer, whereas pcre2_substring_get_bynumber() 3141 copies it into new memory, obtained using the same memory allocation 3142 function that was used for the match data block. The first two argu- 3143 ments of these functions are a pointer to the match data block and a 3144 capture group number. 3145 3146 The final arguments of pcre2_substring_copy_bynumber() are a pointer to 3147 the buffer and a pointer to a variable that contains its length in code 3148 units. This is updated to contain the actual number of code units used 3149 for the extracted substring, excluding the terminating zero. 3150 3151 For pcre2_substring_get_bynumber() the third and fourth arguments point 3152 to variables that are updated with a pointer to the new memory and the 3153 number of code units that comprise the substring, again excluding the 3154 terminating zero. When the substring is no longer needed, the memory 3155 should be freed by calling pcre2_substring_free(). 3156 3157 The return value from all these functions is zero for success, or a 3158 negative error code. If the pattern match failed, the match failure 3159 code is returned. If a substring number greater than zero is used af- 3160 ter a partial match, PCRE2_ERROR_PARTIAL is returned. Other possible 3161 error codes are: 3162 3163 PCRE2_ERROR_NOMEMORY 3164 3165 The buffer was too small for pcre2_substring_copy_bynumber(), or the 3166 attempt to get memory failed for pcre2_substring_get_bynumber(). 3167 3168 PCRE2_ERROR_NOSUBSTRING 3169 3170 There is no substring with that number in the pattern, that is, the 3171 number is greater than the number of capturing parentheses. 3172 3173 PCRE2_ERROR_UNAVAILABLE 3174 3175 The substring number, though not greater than the number of captures in 3176 the pattern, is greater than the number of slots in the ovector, so the 3177 substring could not be captured. 3178 3179 PCRE2_ERROR_UNSET 3180 3181 The substring did not participate in the match. For example, if the 3182 pattern is (abc)|(def) and the subject is "def", and the ovector con- 3183 tains at least two capturing slots, substring number 1 is unset. 3184 3185 3186EXTRACTING A LIST OF ALL CAPTURED SUBSTRINGS 3187 3188 int pcre2_substring_list_get(pcre2_match_data *match_data, 3189 PCRE2_UCHAR ***listptr, PCRE2_SIZE **lengthsptr); 3190 3191 void pcre2_substring_list_free(PCRE2_SPTR *list); 3192 3193 The pcre2_substring_list_get() function extracts all available sub- 3194 strings and builds a list of pointers to them. It also (optionally) 3195 builds a second list that contains their lengths (in code units), ex- 3196 cluding a terminating zero that is added to each of them. All this is 3197 done in a single block of memory that is obtained using the same memory 3198 allocation function that was used to get the match data block. 3199 3200 This function must be called only after a successful match. If called 3201 after a partial match, the error code PCRE2_ERROR_PARTIAL is returned. 3202 3203 The address of the memory block is returned via listptr, which is also 3204 the start of the list of string pointers. The end of the list is marked 3205 by a NULL pointer. The address of the list of lengths is returned via 3206 lengthsptr. If your strings do not contain binary zeros and you do not 3207 therefore need the lengths, you may supply NULL as the lengthsptr argu- 3208 ment to disable the creation of a list of lengths. The yield of the 3209 function is zero if all went well, or PCRE2_ERROR_NOMEMORY if the mem- 3210 ory block could not be obtained. When the list is no longer needed, it 3211 should be freed by calling pcre2_substring_list_free(). 3212 3213 If this function encounters a substring that is unset, which can happen 3214 when capture group number n+1 matches some part of the subject, but 3215 group n has not been used at all, it returns an empty string. This can 3216 be distinguished from a genuine zero-length substring by inspecting the 3217 appropriate offset in the ovector, which contain PCRE2_UNSET for unset 3218 substrings, or by calling pcre2_substring_length_bynumber(). 3219 3220 3221EXTRACTING CAPTURED SUBSTRINGS BY NAME 3222 3223 int pcre2_substring_number_from_name(const pcre2_code *code, 3224 PCRE2_SPTR name); 3225 3226 int pcre2_substring_length_byname(pcre2_match_data *match_data, 3227 PCRE2_SPTR name, PCRE2_SIZE *length); 3228 3229 int pcre2_substring_copy_byname(pcre2_match_data *match_data, 3230 PCRE2_SPTR name, PCRE2_UCHAR *buffer, PCRE2_SIZE *bufflen); 3231 3232 int pcre2_substring_get_byname(pcre2_match_data *match_data, 3233 PCRE2_SPTR name, PCRE2_UCHAR **bufferptr, PCRE2_SIZE *bufflen); 3234 3235 void pcre2_substring_free(PCRE2_UCHAR *buffer); 3236 3237 To extract a substring by name, you first have to find associated num- 3238 ber. For example, for this pattern: 3239 3240 (a+)b(?<xxx>\d+)... 3241 3242 the number of the capture group called "xxx" is 2. If the name is known 3243 to be unique (PCRE2_DUPNAMES was not set), you can find the number from 3244 the name by calling pcre2_substring_number_from_name(). The first argu- 3245 ment is the compiled pattern, and the second is the name. The yield of 3246 the function is the group number, PCRE2_ERROR_NOSUBSTRING if there is 3247 no group with that name, or PCRE2_ERROR_NOUNIQUESUBSTRING if there is 3248 more than one group with that name. Given the number, you can extract 3249 the substring directly from the ovector, or use one of the "bynumber" 3250 functions described above. 3251 3252 For convenience, there are also "byname" functions that correspond to 3253 the "bynumber" functions, the only difference being that the second ar- 3254 gument is a name instead of a number. If PCRE2_DUPNAMES is set and 3255 there are duplicate names, these functions scan all the groups with the 3256 given name, and return the captured substring from the first named 3257 group that is set. 3258 3259 If there are no groups with the given name, PCRE2_ERROR_NOSUBSTRING is 3260 returned. If all groups with the name have numbers that are greater 3261 than the number of slots in the ovector, PCRE2_ERROR_UNAVAILABLE is re- 3262 turned. If there is at least one group with a slot in the ovector, but 3263 no group is found to be set, PCRE2_ERROR_UNSET is returned. 3264 3265 Warning: If the pattern uses the (?| feature to set up multiple capture 3266 groups with the same number, as described in the section on duplicate 3267 group numbers in the pcre2pattern page, you cannot use names to distin- 3268 guish the different capture groups, because names are not included in 3269 the compiled code. The matching process uses only numbers. For this 3270 reason, the use of different names for groups with the same number 3271 causes an error at compile time. 3272 3273 3274CREATING A NEW STRING WITH SUBSTITUTIONS 3275 3276 int pcre2_substitute(const pcre2_code *code, PCRE2_SPTR subject, 3277 PCRE2_SIZE length, PCRE2_SIZE startoffset, 3278 uint32_t options, pcre2_match_data *match_data, 3279 pcre2_match_context *mcontext, PCRE2_SPTR replacement, 3280 PCRE2_SIZE rlength, PCRE2_UCHAR *outputbuffer, 3281 PCRE2_SIZE *outlengthptr); 3282 3283 This function optionally calls pcre2_match() and then makes a copy of 3284 the subject string in outputbuffer, replacing parts that were matched 3285 with the replacement string, whose length is supplied in rlength, which 3286 can be given as PCRE2_ZERO_TERMINATED for a zero-terminated string. As 3287 a special case, if replacement is NULL and rlength is zero, the re- 3288 placement is assumed to be an empty string. If rlength is non-zero, an 3289 error occurs if replacement is NULL. 3290 3291 There is an option (see PCRE2_SUBSTITUTE_REPLACEMENT_ONLY below) to re- 3292 turn just the replacement string(s). The default action is to perform 3293 just one replacement if the pattern matches, but there is an option 3294 that requests multiple replacements (see PCRE2_SUBSTITUTE_GLOBAL be- 3295 low). 3296 3297 If successful, pcre2_substitute() returns the number of substitutions 3298 that were carried out. This may be zero if no match was found, and is 3299 never greater than one unless PCRE2_SUBSTITUTE_GLOBAL is set. A nega- 3300 tive value is returned if an error is detected. 3301 3302 Matches in which a \K item in a lookahead in the pattern causes the 3303 match to end before it starts are not supported, and give rise to an 3304 error return. For global replacements, matches in which \K in a lookbe- 3305 hind causes the match to start earlier than the point that was reached 3306 in the previous iteration are also not supported. 3307 3308 The first seven arguments of pcre2_substitute() are the same as for 3309 pcre2_match(), except that the partial matching options are not permit- 3310 ted, and match_data may be passed as NULL, in which case a match data 3311 block is obtained and freed within this function, using memory manage- 3312 ment functions from the match context, if provided, or else those that 3313 were used to allocate memory for the compiled code. 3314 3315 If match_data is not NULL and PCRE2_SUBSTITUTE_MATCHED is not set, the 3316 provided block is used for all calls to pcre2_match(), and its contents 3317 afterwards are the result of the final call. For global changes, this 3318 will always be a no-match error. The contents of the ovector within the 3319 match data block may or may not have been changed. 3320 3321 As well as the usual options for pcre2_match(), a number of additional 3322 options can be set in the options argument of pcre2_substitute(). One 3323 such option is PCRE2_SUBSTITUTE_MATCHED. When this is set, an external 3324 match_data block must be provided, and it must have already been used 3325 for an external call to pcre2_match() with the same pattern and subject 3326 arguments. The data in the match_data block (return code, offset vec- 3327 tor) is then used for the first substitution instead of calling 3328 pcre2_match() from within pcre2_substitute(). This allows an applica- 3329 tion to check for a match before choosing to substitute, without having 3330 to repeat the match. 3331 3332 The contents of the externally supplied match data block are not 3333 changed when PCRE2_SUBSTITUTE_MATCHED is set. If PCRE2_SUBSTI- 3334 TUTE_GLOBAL is also set, pcre2_match() is called after the first sub- 3335 stitution to check for further matches, but this is done using an in- 3336 ternally obtained match data block, thus always leaving the external 3337 block unchanged. 3338 3339 The code argument is not used for matching before the first substitu- 3340 tion when PCRE2_SUBSTITUTE_MATCHED is set, but it must be provided, 3341 even when PCRE2_SUBSTITUTE_GLOBAL is not set, because it contains in- 3342 formation such as the UTF setting and the number of capturing parenthe- 3343 ses in the pattern. 3344 3345 The default action of pcre2_substitute() is to return a copy of the 3346 subject string with matched substrings replaced. However, if PCRE2_SUB- 3347 STITUTE_REPLACEMENT_ONLY is set, only the replacement substrings are 3348 returned. In the global case, multiple replacements are concatenated in 3349 the output buffer. Substitution callouts (see below) can be used to 3350 separate them if necessary. 3351 3352 The outlengthptr argument of pcre2_substitute() must point to a vari- 3353 able that contains the length, in code units, of the output buffer. If 3354 the function is successful, the value is updated to contain the length 3355 in code units of the new string, excluding the trailing zero that is 3356 automatically added. 3357 3358 If the function is not successful, the value set via outlengthptr de- 3359 pends on the type of error. For syntax errors in the replacement 3360 string, the value is the offset in the replacement string where the er- 3361 ror was detected. For other errors, the value is PCRE2_UNSET by de- 3362 fault. This includes the case of the output buffer being too small, un- 3363 less PCRE2_SUBSTITUTE_OVERFLOW_LENGTH is set. 3364 3365 PCRE2_SUBSTITUTE_OVERFLOW_LENGTH changes what happens when the output 3366 buffer is too small. The default action is to return PCRE2_ERROR_NOMEM- 3367 ORY immediately. If this option is set, however, pcre2_substitute() 3368 continues to go through the motions of matching and substituting (with- 3369 out, of course, writing anything) in order to compute the size of buf- 3370 fer that is needed. This value is passed back via the outlengthptr 3371 variable, with the result of the function still being PCRE2_ER- 3372 ROR_NOMEMORY. 3373 3374 Passing a buffer size of zero is a permitted way of finding out how 3375 much memory is needed for given substitution. However, this does mean 3376 that the entire operation is carried out twice. Depending on the appli- 3377 cation, it may be more efficient to allocate a large buffer and free 3378 the excess afterwards, instead of using PCRE2_SUBSTITUTE_OVER- 3379 FLOW_LENGTH. 3380 3381 The replacement string, which is interpreted as a UTF string in UTF 3382 mode, is checked for UTF validity unless PCRE2_NO_UTF_CHECK is set. An 3383 invalid UTF replacement string causes an immediate return with the rel- 3384 evant UTF error code. 3385 3386 If PCRE2_SUBSTITUTE_LITERAL is set, the replacement string is not in- 3387 terpreted in any way. By default, however, a dollar character is an es- 3388 cape character that can specify the insertion of characters from cap- 3389 ture groups and names from (*MARK) or other control verbs in the pat- 3390 tern. The following forms are always recognized: 3391 3392 $$ insert a dollar character 3393 $<n> or ${<n>} insert the contents of group <n> 3394 $*MARK or ${*MARK} insert a control verb name 3395 3396 Either a group number or a group name can be given for <n>. Curly 3397 brackets are required only if the following character would be inter- 3398 preted as part of the number or name. The number may be zero to include 3399 the entire matched string. For example, if the pattern a(b)c is 3400 matched with "=abc=" and the replacement string "+$1$0$1+", the result 3401 is "=+babcb+=". 3402 3403 $*MARK inserts the name from the last encountered backtracking control 3404 verb on the matching path that has a name. (*MARK) must always include 3405 a name, but the other verbs need not. For example, in the case of 3406 (*MARK:A)(*PRUNE) the name inserted is "A", but for (*MARK:A)(*PRUNE:B) 3407 the relevant name is "B". This facility can be used to perform simple 3408 simultaneous substitutions, as this pcre2test example shows: 3409 3410 /(*MARK:pear)apple|(*MARK:orange)lemon/g,replace=${*MARK} 3411 apple lemon 3412 2: pear orange 3413 3414 PCRE2_SUBSTITUTE_GLOBAL causes the function to iterate over the subject 3415 string, replacing every matching substring. If this option is not set, 3416 only the first matching substring is replaced. The search for matches 3417 takes place in the original subject string (that is, previous replace- 3418 ments do not affect it). Iteration is implemented by advancing the 3419 startoffset value for each search, which is always passed the entire 3420 subject string. If an offset limit is set in the match context, search- 3421 ing stops when that limit is reached. 3422 3423 You can restrict the effect of a global substitution to a portion of 3424 the subject string by setting either or both of startoffset and an off- 3425 set limit. Here is a pcre2test example: 3426 3427 /B/g,replace=!,use_offset_limit 3428 ABC ABC ABC ABC\=offset=3,offset_limit=12 3429 2: ABC A!C A!C ABC 3430 3431 When continuing with global substitutions after matching a substring 3432 with zero length, an attempt to find a non-empty match at the same off- 3433 set is performed. If this is not successful, the offset is advanced by 3434 one character except when CRLF is a valid newline sequence and the next 3435 two characters are CR, LF. In this case, the offset is advanced by two 3436 characters. 3437 3438 PCRE2_SUBSTITUTE_UNKNOWN_UNSET causes references to capture groups that 3439 do not appear in the pattern to be treated as unset groups. This option 3440 should be used with care, because it means that a typo in a group name 3441 or number no longer causes the PCRE2_ERROR_NOSUBSTRING error. 3442 3443 PCRE2_SUBSTITUTE_UNSET_EMPTY causes unset capture groups (including un- 3444 known groups when PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set) to be treated 3445 as empty strings when inserted as described above. If this option is 3446 not set, an attempt to insert an unset group causes the PCRE2_ERROR_UN- 3447 SET error. This option does not influence the extended substitution 3448 syntax described below. 3449 3450 PCRE2_SUBSTITUTE_EXTENDED causes extra processing to be applied to the 3451 replacement string. Without this option, only the dollar character is 3452 special, and only the group insertion forms listed above are valid. 3453 When PCRE2_SUBSTITUTE_EXTENDED is set, two things change: 3454 3455 Firstly, backslash in a replacement string is interpreted as an escape 3456 character. The usual forms such as \n or \x{ddd} can be used to specify 3457 particular character codes, and backslash followed by any non-alphanu- 3458 meric character quotes that character. Extended quoting can be coded 3459 using \Q...\E, exactly as in pattern strings. 3460 3461 There are also four escape sequences for forcing the case of inserted 3462 letters. The insertion mechanism has three states: no case forcing, 3463 force upper case, and force lower case. The escape sequences change the 3464 current state: \U and \L change to upper or lower case forcing, respec- 3465 tively, and \E (when not terminating a \Q quoted sequence) reverts to 3466 no case forcing. The sequences \u and \l force the next character (if 3467 it is a letter) to upper or lower case, respectively, and then the 3468 state automatically reverts to no case forcing. Case forcing applies to 3469 all inserted characters, including those from capture groups and let- 3470 ters within \Q...\E quoted sequences. If either PCRE2_UTF or PCRE2_UCP 3471 was set when the pattern was compiled, Unicode properties are used for 3472 case forcing characters whose code points are greater than 127. 3473 3474 Note that case forcing sequences such as \U...\E do not nest. For exam- 3475 ple, the result of processing "\Uaa\LBB\Ecc\E" is "AAbbcc"; the final 3476 \E has no effect. Note also that the PCRE2_ALT_BSUX and PCRE2_EX- 3477 TRA_ALT_BSUX options do not apply to replacement strings. 3478 3479 The second effect of setting PCRE2_SUBSTITUTE_EXTENDED is to add more 3480 flexibility to capture group substitution. The syntax is similar to 3481 that used by Bash: 3482 3483 ${<n>:-<string>} 3484 ${<n>:+<string1>:<string2>} 3485 3486 As before, <n> may be a group number or a name. The first form speci- 3487 fies a default value. If group <n> is set, its value is inserted; if 3488 not, <string> is expanded and the result inserted. The second form 3489 specifies strings that are expanded and inserted when group <n> is set 3490 or unset, respectively. The first form is just a convenient shorthand 3491 for 3492 3493 ${<n>:+${<n>}:<string>} 3494 3495 Backslash can be used to escape colons and closing curly brackets in 3496 the replacement strings. A change of the case forcing state within a 3497 replacement string remains in force afterwards, as shown in this 3498 pcre2test example: 3499 3500 /(some)?(body)/substitute_extended,replace=${1:+\U:\L}HeLLo 3501 body 3502 1: hello 3503 somebody 3504 1: HELLO 3505 3506 The PCRE2_SUBSTITUTE_UNSET_EMPTY option does not affect these extended 3507 substitutions. However, PCRE2_SUBSTITUTE_UNKNOWN_UNSET does cause un- 3508 known groups in the extended syntax forms to be treated as unset. 3509 3510 If PCRE2_SUBSTITUTE_LITERAL is set, PCRE2_SUBSTITUTE_UNKNOWN_UNSET, 3511 PCRE2_SUBSTITUTE_UNSET_EMPTY, and PCRE2_SUBSTITUTE_EXTENDED are irrele- 3512 vant and are ignored. 3513 3514 Substitution errors 3515 3516 In the event of an error, pcre2_substitute() returns a negative error 3517 code. Except for PCRE2_ERROR_NOMATCH (which is never returned), errors 3518 from pcre2_match() are passed straight back. 3519 3520 PCRE2_ERROR_NOSUBSTRING is returned for a non-existent substring inser- 3521 tion, unless PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set. 3522 3523 PCRE2_ERROR_UNSET is returned for an unset substring insertion (includ- 3524 ing an unknown substring when PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set) 3525 when the simple (non-extended) syntax is used and PCRE2_SUBSTITUTE_UN- 3526 SET_EMPTY is not set. 3527 3528 PCRE2_ERROR_NOMEMORY is returned if the output buffer is not big 3529 enough. If the PCRE2_SUBSTITUTE_OVERFLOW_LENGTH option is set, the size 3530 of buffer that is needed is returned via outlengthptr. Note that this 3531 does not happen by default. 3532 3533 PCRE2_ERROR_NULL is returned if PCRE2_SUBSTITUTE_MATCHED is set but the 3534 match_data argument is NULL or if the subject or replacement arguments 3535 are NULL. For backward compatibility reasons an exception is made for 3536 the replacement argument if the rlength argument is also 0. 3537 3538 PCRE2_ERROR_BADREPLACEMENT is used for miscellaneous syntax errors in 3539 the replacement string, with more particular errors being PCRE2_ER- 3540 ROR_BADREPESCAPE (invalid escape sequence), PCRE2_ERROR_REPMISSINGBRACE 3541 (closing curly bracket not found), PCRE2_ERROR_BADSUBSTITUTION (syntax 3542 error in extended group substitution), and PCRE2_ERROR_BADSUBSPATTERN 3543 (the pattern match ended before it started or the match started earlier 3544 than the current position in the subject, which can happen if \K is 3545 used in an assertion). 3546 3547 As for all PCRE2 errors, a text message that describes the error can be 3548 obtained by calling the pcre2_get_error_message() function (see "Ob- 3549 taining a textual error message" above). 3550 3551 Substitution callouts 3552 3553 int pcre2_set_substitute_callout(pcre2_match_context *mcontext, 3554 int (*callout_function)(pcre2_substitute_callout_block *, void *), 3555 void *callout_data); 3556 3557 The pcre2_set_substitution_callout() function can be used to specify a 3558 callout function for pcre2_substitute(). This information is passed in 3559 a match context. The callout function is called after each substitution 3560 has been processed, but it can cause the replacement not to happen. The 3561 callout function is not called for simulated substitutions that happen 3562 as a result of the PCRE2_SUBSTITUTE_OVERFLOW_LENGTH option. 3563 3564 The first argument of the callout function is a pointer to a substitute 3565 callout block structure, which contains the following fields, not nec- 3566 essarily in this order: 3567 3568 uint32_t version; 3569 uint32_t subscount; 3570 PCRE2_SPTR input; 3571 PCRE2_SPTR output; 3572 PCRE2_SIZE *ovector; 3573 uint32_t oveccount; 3574 PCRE2_SIZE output_offsets[2]; 3575 3576 The version field contains the version number of the block format. The 3577 current version is 0. The version number will increase in future if 3578 more fields are added, but the intention is never to remove any of the 3579 existing fields. 3580 3581 The subscount field is the number of the current match. It is 1 for the 3582 first callout, 2 for the second, and so on. The input and output point- 3583 ers are copies of the values passed to pcre2_substitute(). 3584 3585 The ovector field points to the ovector, which contains the result of 3586 the most recent match. The oveccount field contains the number of pairs 3587 that are set in the ovector, and is always greater than zero. 3588 3589 The output_offsets vector contains the offsets of the replacement in 3590 the output string. This has already been processed for dollar and (if 3591 requested) backslash substitutions as described above. 3592 3593 The second argument of the callout function is the value passed as 3594 callout_data when the function was registered. The value returned by 3595 the callout function is interpreted as follows: 3596 3597 If the value is zero, the replacement is accepted, and, if PCRE2_SUB- 3598 STITUTE_GLOBAL is set, processing continues with a search for the next 3599 match. If the value is not zero, the current replacement is not ac- 3600 cepted. If the value is greater than zero, processing continues when 3601 PCRE2_SUBSTITUTE_GLOBAL is set. Otherwise (the value is less than zero 3602 or PCRE2_SUBSTITUTE_GLOBAL is not set), the the rest of the input is 3603 copied to the output and the call to pcre2_substitute() exits, return- 3604 ing the number of matches so far. 3605 3606 3607DUPLICATE CAPTURE GROUP NAMES 3608 3609 int pcre2_substring_nametable_scan(const pcre2_code *code, 3610 PCRE2_SPTR name, PCRE2_SPTR *first, PCRE2_SPTR *last); 3611 3612 When a pattern is compiled with the PCRE2_DUPNAMES option, names for 3613 capture groups are not required to be unique. Duplicate names are al- 3614 ways allowed for groups with the same number, created by using the (?| 3615 feature. Indeed, if such groups are named, they are required to use the 3616 same names. 3617 3618 Normally, patterns that use duplicate names are such that in any one 3619 match, only one of each set of identically-named groups participates. 3620 An example is shown in the pcre2pattern documentation. 3621 3622 When duplicates are present, pcre2_substring_copy_byname() and 3623 pcre2_substring_get_byname() return the first substring corresponding 3624 to the given name that is set. Only if none are set is PCRE2_ERROR_UN- 3625 SET is returned. The pcre2_substring_number_from_name() function re- 3626 turns the error PCRE2_ERROR_NOUNIQUESUBSTRING when there are duplicate 3627 names. 3628 3629 If you want to get full details of all captured substrings for a given 3630 name, you must use the pcre2_substring_nametable_scan() function. The 3631 first argument is the compiled pattern, and the second is the name. If 3632 the third and fourth arguments are NULL, the function returns a group 3633 number for a unique name, or PCRE2_ERROR_NOUNIQUESUBSTRING otherwise. 3634 3635 When the third and fourth arguments are not NULL, they must be pointers 3636 to variables that are updated by the function. After it has run, they 3637 point to the first and last entries in the name-to-number table for the 3638 given name, and the function returns the length of each entry in code 3639 units. In both cases, PCRE2_ERROR_NOSUBSTRING is returned if there are 3640 no entries for the given name. 3641 3642 The format of the name table is described above in the section entitled 3643 Information about a pattern. Given all the relevant entries for the 3644 name, you can extract each of their numbers, and hence the captured 3645 data. 3646 3647 3648FINDING ALL POSSIBLE MATCHES AT ONE POSITION 3649 3650 The traditional matching function uses a similar algorithm to Perl, 3651 which stops when it finds the first match at a given point in the sub- 3652 ject. If you want to find all possible matches, or the longest possible 3653 match at a given position, consider using the alternative matching 3654 function (see below) instead. If you cannot use the alternative func- 3655 tion, you can kludge it up by making use of the callout facility, which 3656 is described in the pcre2callout documentation. 3657 3658 What you have to do is to insert a callout right at the end of the pat- 3659 tern. When your callout function is called, extract and save the cur- 3660 rent matched substring. Then return 1, which forces pcre2_match() to 3661 backtrack and try other alternatives. Ultimately, when it runs out of 3662 matches, pcre2_match() will yield PCRE2_ERROR_NOMATCH. 3663 3664 3665MATCHING A PATTERN: THE ALTERNATIVE FUNCTION 3666 3667 int pcre2_dfa_match(const pcre2_code *code, PCRE2_SPTR subject, 3668 PCRE2_SIZE length, PCRE2_SIZE startoffset, 3669 uint32_t options, pcre2_match_data *match_data, 3670 pcre2_match_context *mcontext, 3671 int *workspace, PCRE2_SIZE wscount); 3672 3673 The function pcre2_dfa_match() is called to match a subject string 3674 against a compiled pattern, using a matching algorithm that scans the 3675 subject string just once (not counting lookaround assertions), and does 3676 not backtrack (except when processing lookaround assertions). This has 3677 different characteristics to the normal algorithm, and is not compati- 3678 ble with Perl. Some of the features of PCRE2 patterns are not sup- 3679 ported. Nevertheless, there are times when this kind of matching can be 3680 useful. For a discussion of the two matching algorithms, and a list of 3681 features that pcre2_dfa_match() does not support, see the pcre2matching 3682 documentation. 3683 3684 The arguments for the pcre2_dfa_match() function are the same as for 3685 pcre2_match(), plus two extras. The ovector within the match data block 3686 is used in a different way, and this is described below. The other com- 3687 mon arguments are used in the same way as for pcre2_match(), so their 3688 description is not repeated here. 3689 3690 The two additional arguments provide workspace for the function. The 3691 workspace vector should contain at least 20 elements. It is used for 3692 keeping track of multiple paths through the pattern tree. More 3693 workspace is needed for patterns and subjects where there are a lot of 3694 potential matches. 3695 3696 Here is an example of a simple call to pcre2_dfa_match(): 3697 3698 int wspace[20]; 3699 pcre2_match_data *md = pcre2_match_data_create(4, NULL); 3700 int rc = pcre2_dfa_match( 3701 re, /* result of pcre2_compile() */ 3702 "some string", /* the subject string */ 3703 11, /* the length of the subject string */ 3704 0, /* start at offset 0 in the subject */ 3705 0, /* default options */ 3706 md, /* the match data block */ 3707 NULL, /* a match context; NULL means use defaults */ 3708 wspace, /* working space vector */ 3709 20); /* number of elements (NOT size in bytes) */ 3710 3711 Option bits for pcre2_dfa_match() 3712 3713 The unused bits of the options argument for pcre2_dfa_match() must be 3714 zero. The only bits that may be set are PCRE2_ANCHORED, 3715 PCRE2_COPY_MATCHED_SUBJECT, PCRE2_ENDANCHORED, PCRE2_NOTBOL, PCRE2_NO- 3716 TEOL, PCRE2_NOTEMPTY, PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_UTF_CHECK, 3717 PCRE2_PARTIAL_HARD, PCRE2_PARTIAL_SOFT, PCRE2_DFA_SHORTEST, and 3718 PCRE2_DFA_RESTART. All but the last four of these are exactly the same 3719 as for pcre2_match(), so their description is not repeated here. 3720 3721 PCRE2_PARTIAL_HARD 3722 PCRE2_PARTIAL_SOFT 3723 3724 These have the same general effect as they do for pcre2_match(), but 3725 the details are slightly different. When PCRE2_PARTIAL_HARD is set for 3726 pcre2_dfa_match(), it returns PCRE2_ERROR_PARTIAL if the end of the 3727 subject is reached and there is still at least one matching possibility 3728 that requires additional characters. This happens even if some complete 3729 matches have already been found. When PCRE2_PARTIAL_SOFT is set, the 3730 return code PCRE2_ERROR_NOMATCH is converted into PCRE2_ERROR_PARTIAL 3731 if the end of the subject is reached, there have been no complete 3732 matches, but there is still at least one matching possibility. The por- 3733 tion of the string that was inspected when the longest partial match 3734 was found is set as the first matching string in both cases. There is a 3735 more detailed discussion of partial and multi-segment matching, with 3736 examples, in the pcre2partial documentation. 3737 3738 PCRE2_DFA_SHORTEST 3739 3740 Setting the PCRE2_DFA_SHORTEST option causes the matching algorithm to 3741 stop as soon as it has found one match. Because of the way the alterna- 3742 tive algorithm works, this is necessarily the shortest possible match 3743 at the first possible matching point in the subject string. 3744 3745 PCRE2_DFA_RESTART 3746 3747 When pcre2_dfa_match() returns a partial match, it is possible to call 3748 it again, with additional subject characters, and have it continue with 3749 the same match. The PCRE2_DFA_RESTART option requests this action; when 3750 it is set, the workspace and wscount options must reference the same 3751 vector as before because data about the match so far is left in them 3752 after a partial match. There is more discussion of this facility in the 3753 pcre2partial documentation. 3754 3755 Successful returns from pcre2_dfa_match() 3756 3757 When pcre2_dfa_match() succeeds, it may have matched more than one sub- 3758 string in the subject. Note, however, that all the matches from one run 3759 of the function start at the same point in the subject. The shorter 3760 matches are all initial substrings of the longer matches. For example, 3761 if the pattern 3762 3763 <.*> 3764 3765 is matched against the string 3766 3767 This is <something> <something else> <something further> no more 3768 3769 the three matched strings are 3770 3771 <something> <something else> <something further> 3772 <something> <something else> 3773 <something> 3774 3775 On success, the yield of the function is a number greater than zero, 3776 which is the number of matched substrings. The offsets of the sub- 3777 strings are returned in the ovector, and can be extracted by number in 3778 the same way as for pcre2_match(), but the numbers bear no relation to 3779 any capture groups that may exist in the pattern, because DFA matching 3780 does not support capturing. 3781 3782 Calls to the convenience functions that extract substrings by name re- 3783 turn the error PCRE2_ERROR_DFA_UFUNC (unsupported function) if used af- 3784 ter a DFA match. The convenience functions that extract substrings by 3785 number never return PCRE2_ERROR_NOSUBSTRING. 3786 3787 The matched strings are stored in the ovector in reverse order of 3788 length; that is, the longest matching string is first. If there were 3789 too many matches to fit into the ovector, the yield of the function is 3790 zero, and the vector is filled with the longest matches. 3791 3792 NOTE: PCRE2's "auto-possessification" optimization usually applies to 3793 character repeats at the end of a pattern (as well as internally). For 3794 example, the pattern "a\d+" is compiled as if it were "a\d++". For DFA 3795 matching, this means that only one possible match is found. If you re- 3796 ally do want multiple matches in such cases, either use an ungreedy re- 3797 peat such as "a\d+?" or set the PCRE2_NO_AUTO_POSSESS option when com- 3798 piling. 3799 3800 Error returns from pcre2_dfa_match() 3801 3802 The pcre2_dfa_match() function returns a negative number when it fails. 3803 Many of the errors are the same as for pcre2_match(), as described 3804 above. There are in addition the following errors that are specific to 3805 pcre2_dfa_match(): 3806 3807 PCRE2_ERROR_DFA_UITEM 3808 3809 This return is given if pcre2_dfa_match() encounters an item in the 3810 pattern that it does not support, for instance, the use of \C in a UTF 3811 mode or a backreference. 3812 3813 PCRE2_ERROR_DFA_UCOND 3814 3815 This return is given if pcre2_dfa_match() encounters a condition item 3816 that uses a backreference for the condition, or a test for recursion in 3817 a specific capture group. These are not supported. 3818 3819 PCRE2_ERROR_DFA_UINVALID_UTF 3820 3821 This return is given if pcre2_dfa_match() is called for a pattern that 3822 was compiled with PCRE2_MATCH_INVALID_UTF. This is not supported for 3823 DFA matching. 3824 3825 PCRE2_ERROR_DFA_WSSIZE 3826 3827 This return is given if pcre2_dfa_match() runs out of space in the 3828 workspace vector. 3829 3830 PCRE2_ERROR_DFA_RECURSE 3831 3832 When a recursion or subroutine call is processed, the matching function 3833 calls itself recursively, using private memory for the ovector and 3834 workspace. This error is given if the internal ovector is not large 3835 enough. This should be extremely rare, as a vector of size 1000 is 3836 used. 3837 3838 PCRE2_ERROR_DFA_BADRESTART 3839 3840 When pcre2_dfa_match() is called with the PCRE2_DFA_RESTART option, 3841 some plausibility checks are made on the contents of the workspace, 3842 which should contain data about the previous partial match. If any of 3843 these checks fail, this error is given. 3844 3845 3846SEE ALSO 3847 3848 pcre2build(3), pcre2callout(3), pcre2demo(3), pcre2matching(3), 3849 pcre2partial(3), pcre2posix(3), pcre2sample(3), pcre2unicode(3). 3850 3851 3852AUTHOR 3853 3854 Philip Hazel 3855 Retired from University Computing Service 3856 Cambridge, England. 3857 3858 3859REVISION 3860 3861 Last updated: 14 December 2021 3862 Copyright (c) 1997-2021 University of Cambridge. 3863------------------------------------------------------------------------------ 3864 3865 3866PCRE2BUILD(3) Library Functions Manual PCRE2BUILD(3) 3867 3868 3869 3870NAME 3871 PCRE2 - Perl-compatible regular expressions (revised API) 3872 3873BUILDING PCRE2 3874 3875 PCRE2 is distributed with a configure script that can be used to build 3876 the library in Unix-like environments using the applications known as 3877 Autotools. Also in the distribution are files to support building using 3878 CMake instead of configure. The text file README contains general in- 3879 formation about building with Autotools (some of which is repeated be- 3880 low), and also has some comments about building on various operating 3881 systems. There is a lot more information about building PCRE2 without 3882 using Autotools (including information about using CMake and building 3883 "by hand") in the text file called NON-AUTOTOOLS-BUILD. You should 3884 consult this file as well as the README file if you are building in a 3885 non-Unix-like environment. 3886 3887 3888PCRE2 BUILD-TIME OPTIONS 3889 3890 The rest of this document describes the optional features of PCRE2 that 3891 can be selected when the library is compiled. It assumes use of the 3892 configure script, where the optional features are selected or dese- 3893 lected by providing options to configure before running the make com- 3894 mand. However, the same options can be selected in both Unix-like and 3895 non-Unix-like environments if you are using CMake instead of configure 3896 to build PCRE2. 3897 3898 If you are not using Autotools or CMake, option selection can be done 3899 by editing the config.h file, or by passing parameter settings to the 3900 compiler, as described in NON-AUTOTOOLS-BUILD. 3901 3902 The complete list of options for configure (which includes the standard 3903 ones such as the selection of the installation directory) can be ob- 3904 tained by running 3905 3906 ./configure --help 3907 3908 The following sections include descriptions of "on/off" options whose 3909 names begin with --enable or --disable. Because of the way that config- 3910 ure works, --enable and --disable always come in pairs, so the comple- 3911 mentary option always exists as well, but as it specifies the default, 3912 it is not described. Options that specify values have names that start 3913 with --with. At the end of a configure run, a summary of the configura- 3914 tion is output. 3915 3916 3917BUILDING 8-BIT, 16-BIT AND 32-BIT LIBRARIES 3918 3919 By default, a library called libpcre2-8 is built, containing functions 3920 that take string arguments contained in arrays of bytes, interpreted 3921 either as single-byte characters, or UTF-8 strings. You can also build 3922 two other libraries, called libpcre2-16 and libpcre2-32, which process 3923 strings that are contained in arrays of 16-bit and 32-bit code units, 3924 respectively. These can be interpreted either as single-unit characters 3925 or UTF-16/UTF-32 strings. To build these additional libraries, add one 3926 or both of the following to the configure command: 3927 3928 --enable-pcre2-16 3929 --enable-pcre2-32 3930 3931 If you do not want the 8-bit library, add 3932 3933 --disable-pcre2-8 3934 3935 as well. At least one of the three libraries must be built. Note that 3936 the POSIX wrapper is for the 8-bit library only, and that pcre2grep is 3937 an 8-bit program. Neither of these are built if you select only the 3938 16-bit or 32-bit libraries. 3939 3940 3941BUILDING SHARED AND STATIC LIBRARIES 3942 3943 The Autotools PCRE2 building process uses libtool to build both shared 3944 and static libraries by default. You can suppress an unwanted library 3945 by adding one of 3946 3947 --disable-shared 3948 --disable-static 3949 3950 to the configure command. 3951 3952 3953UNICODE AND UTF SUPPORT 3954 3955 By default, PCRE2 is built with support for Unicode and UTF character 3956 strings. To build it without Unicode support, add 3957 3958 --disable-unicode 3959 3960 to the configure command. This setting applies to all three libraries. 3961 It is not possible to build one library with Unicode support and an- 3962 other without in the same configuration. 3963 3964 Of itself, Unicode support does not make PCRE2 treat strings as UTF-8, 3965 UTF-16 or UTF-32. To do that, applications that use the library can set 3966 the PCRE2_UTF option when they call pcre2_compile() to compile a pat- 3967 tern. Alternatively, patterns may be started with (*UTF) unless the 3968 application has locked this out by setting PCRE2_NEVER_UTF. 3969 3970 UTF support allows the libraries to process character code points up to 3971 0x10ffff in the strings that they handle. Unicode support also gives 3972 access to the Unicode properties of characters, using pattern escapes 3973 such as \P, \p, and \X. Only the general category properties such as Lu 3974 and Nd, script names, and some bi-directional properties are supported. 3975 Details are given in the pcre2pattern documentation. 3976 3977 Pattern escapes such as \d and \w do not by default make use of Unicode 3978 properties. The application can request that they do by setting the 3979 PCRE2_UCP option. Unless the application has set PCRE2_NEVER_UCP, a 3980 pattern may also request this by starting with (*UCP). 3981 3982 3983DISABLING THE USE OF \C 3984 3985 The \C escape sequence, which matches a single code unit, even in a UTF 3986 mode, can cause unpredictable behaviour because it may leave the cur- 3987 rent matching point in the middle of a multi-code-unit character. The 3988 application can lock it out by setting the PCRE2_NEVER_BACKSLASH_C op- 3989 tion when calling pcre2_compile(). There is also a build-time option 3990 3991 --enable-never-backslash-C 3992 3993 (note the upper case C) which locks out the use of \C entirely. 3994 3995 3996JUST-IN-TIME COMPILER SUPPORT 3997 3998 Just-in-time (JIT) compiler support is included in the build by speci- 3999 fying 4000 4001 --enable-jit 4002 4003 This support is available only for certain hardware architectures. If 4004 this option is set for an unsupported architecture, a building error 4005 occurs. If in doubt, use 4006 4007 --enable-jit=auto 4008 4009 which enables JIT only if the current hardware is supported. You can 4010 check if JIT is enabled in the configuration summary that is output at 4011 the end of a configure run. If you are enabling JIT under SELinux you 4012 may also want to add 4013 4014 --enable-jit-sealloc 4015 4016 which enables the use of an execmem allocator in JIT that is compatible 4017 with SELinux. This has no effect if JIT is not enabled. See the 4018 pcre2jit documentation for a discussion of JIT usage. When JIT support 4019 is enabled, pcre2grep automatically makes use of it, unless you add 4020 4021 --disable-pcre2grep-jit 4022 4023 to the configure command. 4024 4025 4026NEWLINE RECOGNITION 4027 4028 By default, PCRE2 interprets the linefeed (LF) character as indicating 4029 the end of a line. This is the normal newline character on Unix-like 4030 systems. You can compile PCRE2 to use carriage return (CR) instead, by 4031 adding 4032 4033 --enable-newline-is-cr 4034 4035 to the configure command. There is also an --enable-newline-is-lf op- 4036 tion, which explicitly specifies linefeed as the newline character. 4037 4038 Alternatively, you can specify that line endings are to be indicated by 4039 the two-character sequence CRLF (CR immediately followed by LF). If you 4040 want this, add 4041 4042 --enable-newline-is-crlf 4043 4044 to the configure command. There is a fourth option, specified by 4045 4046 --enable-newline-is-anycrlf 4047 4048 which causes PCRE2 to recognize any of the three sequences CR, LF, or 4049 CRLF as indicating a line ending. A fifth option, specified by 4050 4051 --enable-newline-is-any 4052 4053 causes PCRE2 to recognize any Unicode newline sequence. The Unicode 4054 newline sequences are the three just mentioned, plus the single charac- 4055 ters VT (vertical tab, U+000B), FF (form feed, U+000C), NEL (next line, 4056 U+0085), LS (line separator, U+2028), and PS (paragraph separator, 4057 U+2029). The final option is 4058 4059 --enable-newline-is-nul 4060 4061 which causes NUL (binary zero) to be set as the default line-ending 4062 character. 4063 4064 Whatever default line ending convention is selected when PCRE2 is built 4065 can be overridden by applications that use the library. At build time 4066 it is recommended to use the standard for your operating system. 4067 4068 4069WHAT \R MATCHES 4070 4071 By default, the sequence \R in a pattern matches any Unicode newline 4072 sequence, independently of what has been selected as the line ending 4073 sequence. If you specify 4074 4075 --enable-bsr-anycrlf 4076 4077 the default is changed so that \R matches only CR, LF, or CRLF. What- 4078 ever is selected when PCRE2 is built can be overridden by applications 4079 that use the library. 4080 4081 4082HANDLING VERY LARGE PATTERNS 4083 4084 Within a compiled pattern, offset values are used to point from one 4085 part to another (for example, from an opening parenthesis to an alter- 4086 nation metacharacter). By default, in the 8-bit and 16-bit libraries, 4087 two-byte values are used for these offsets, leading to a maximum size 4088 for a compiled pattern of around 64 thousand code units. This is suffi- 4089 cient to handle all but the most gigantic patterns. Nevertheless, some 4090 people do want to process truly enormous patterns, so it is possible to 4091 compile PCRE2 to use three-byte or four-byte offsets by adding a set- 4092 ting such as 4093 4094 --with-link-size=3 4095 4096 to the configure command. The value given must be 2, 3, or 4. For the 4097 16-bit library, a value of 3 is rounded up to 4. In these libraries, 4098 using longer offsets slows down the operation of PCRE2 because it has 4099 to load additional data when handling them. For the 32-bit library the 4100 value is always 4 and cannot be overridden; the value of --with-link- 4101 size is ignored. 4102 4103 4104LIMITING PCRE2 RESOURCE USAGE 4105 4106 The pcre2_match() function increments a counter each time it goes round 4107 its main loop. Putting a limit on this counter controls the amount of 4108 computing resource used by a single call to pcre2_match(). The limit 4109 can be changed at run time, as described in the pcre2api documentation. 4110 The default is 10 million, but this can be changed by adding a setting 4111 such as 4112 4113 --with-match-limit=500000 4114 4115 to the configure command. This setting also applies to the 4116 pcre2_dfa_match() matching function, and to JIT matching (though the 4117 counting is done differently). 4118 4119 The pcre2_match() function starts out using a 20KiB vector on the sys- 4120 tem stack to record backtracking points. The more nested backtracking 4121 points there are (that is, the deeper the search tree), the more memory 4122 is needed. If the initial vector is not large enough, heap memory is 4123 used, up to a certain limit, which is specified in kibibytes (units of 4124 1024 bytes). The limit can be changed at run time, as described in the 4125 pcre2api documentation. The default limit (in effect unlimited) is 20 4126 million. You can change this by a setting such as 4127 4128 --with-heap-limit=500 4129 4130 which limits the amount of heap to 500 KiB. This limit applies only to 4131 interpretive matching in pcre2_match() and pcre2_dfa_match(), which may 4132 also use the heap for internal workspace when processing complicated 4133 patterns. This limit does not apply when JIT (which has its own memory 4134 arrangements) is used. 4135 4136 You can also explicitly limit the depth of nested backtracking in the 4137 pcre2_match() interpreter. This limit defaults to the value that is set 4138 for --with-match-limit. You can set a lower default limit by adding, 4139 for example, 4140 4141 --with-match-limit-depth=10000 4142 4143 to the configure command. This value can be overridden at run time. 4144 This depth limit indirectly limits the amount of heap memory that is 4145 used, but because the size of each backtracking "frame" depends on the 4146 number of capturing parentheses in a pattern, the amount of heap that 4147 is used before the limit is reached varies from pattern to pattern. 4148 This limit was more useful in versions before 10.30, where function re- 4149 cursion was used for backtracking. 4150 4151 As well as applying to pcre2_match(), the depth limit also controls the 4152 depth of recursive function calls in pcre2_dfa_match(). These are used 4153 for lookaround assertions, atomic groups, and recursion within pat- 4154 terns. The limit does not apply to JIT matching. 4155 4156 4157CREATING CHARACTER TABLES AT BUILD TIME 4158 4159 PCRE2 uses fixed tables for processing characters whose code points are 4160 less than 256. By default, PCRE2 is built with a set of tables that are 4161 distributed in the file src/pcre2_chartables.c.dist. These tables are 4162 for ASCII codes only. If you add 4163 4164 --enable-rebuild-chartables 4165 4166 to the configure command, the distributed tables are no longer used. 4167 Instead, a program called pcre2_dftables is compiled and run. This out- 4168 puts the source for new set of tables, created in the default locale of 4169 your C run-time system. This method of replacing the tables does not 4170 work if you are cross compiling, because pcre2_dftables needs to be run 4171 on the local host and therefore not compiled with the cross compiler. 4172 4173 If you need to create alternative tables when cross compiling, you will 4174 have to do so "by hand". There may also be other reasons for creating 4175 tables manually. To cause pcre2_dftables to be built on the local 4176 host, run a normal compiling command, and then run the program with the 4177 output file as its argument, for example: 4178 4179 cc src/pcre2_dftables.c -o pcre2_dftables 4180 ./pcre2_dftables src/pcre2_chartables.c 4181 4182 This builds the tables in the default locale of the local host. If you 4183 want to specify a locale, you must use the -L option: 4184 4185 LC_ALL=fr_FR ./pcre2_dftables -L src/pcre2_chartables.c 4186 4187 You can also specify -b (with or without -L). This causes the tables to 4188 be written in binary instead of as source code. A set of binary tables 4189 can be loaded into memory by an application and passed to pcre2_com- 4190 pile() in the same way as tables created by calling pcre2_maketables(). 4191 The tables are just a string of bytes, independent of hardware charac- 4192 teristics such as endianness. This means they can be bundled with an 4193 application that runs in different environments, to ensure consistent 4194 behaviour. 4195 4196 4197USING EBCDIC CODE 4198 4199 PCRE2 assumes by default that it will run in an environment where the 4200 character code is ASCII or Unicode, which is a superset of ASCII. This 4201 is the case for most computer operating systems. PCRE2 can, however, be 4202 compiled to run in an 8-bit EBCDIC environment by adding 4203 4204 --enable-ebcdic --disable-unicode 4205 4206 to the configure command. This setting implies --enable-rebuild-charta- 4207 bles. You should only use it if you know that you are in an EBCDIC en- 4208 vironment (for example, an IBM mainframe operating system). 4209 4210 It is not possible to support both EBCDIC and UTF-8 codes in the same 4211 version of the library. Consequently, --enable-unicode and --enable- 4212 ebcdic are mutually exclusive. 4213 4214 The EBCDIC character that corresponds to an ASCII LF is assumed to have 4215 the value 0x15 by default. However, in some EBCDIC environments, 0x25 4216 is used. In such an environment you should use 4217 4218 --enable-ebcdic-nl25 4219 4220 as well as, or instead of, --enable-ebcdic. The EBCDIC character for CR 4221 has the same value as in ASCII, namely, 0x0d. Whichever of 0x15 and 4222 0x25 is not chosen as LF is made to correspond to the Unicode NEL char- 4223 acter (which, in Unicode, is 0x85). 4224 4225 The options that select newline behaviour, such as --enable-newline-is- 4226 cr, and equivalent run-time options, refer to these character values in 4227 an EBCDIC environment. 4228 4229 4230PCRE2GREP SUPPORT FOR EXTERNAL SCRIPTS 4231 4232 By default pcre2grep supports the use of callouts with string arguments 4233 within the patterns it is matching. There are two kinds: one that gen- 4234 erates output using local code, and another that calls an external pro- 4235 gram or script. If --disable-pcre2grep-callout-fork is added to the 4236 configure command, only the first kind of callout is supported; if 4237 --disable-pcre2grep-callout is used, all callouts are completely ig- 4238 nored. For more details of pcre2grep callouts, see the pcre2grep docu- 4239 mentation. 4240 4241 4242PCRE2GREP OPTIONS FOR COMPRESSED FILE SUPPORT 4243 4244 By default, pcre2grep reads all files as plain text. You can build it 4245 so that it recognizes files whose names end in .gz or .bz2, and reads 4246 them with libz or libbz2, respectively, by adding one or both of 4247 4248 --enable-pcre2grep-libz 4249 --enable-pcre2grep-libbz2 4250 4251 to the configure command. These options naturally require that the rel- 4252 evant libraries are installed on your system. Configuration will fail 4253 if they are not. 4254 4255 4256PCRE2GREP BUFFER SIZE 4257 4258 pcre2grep uses an internal buffer to hold a "window" on the file it is 4259 scanning, in order to be able to output "before" and "after" lines when 4260 it finds a match. The default starting size of the buffer is 20KiB. The 4261 buffer itself is three times this size, but because of the way it is 4262 used for holding "before" lines, the longest line that is guaranteed to 4263 be processable is the notional buffer size. If a longer line is encoun- 4264 tered, pcre2grep automatically expands the buffer, up to a specified 4265 maximum size, whose default is 1MiB or the starting size, whichever is 4266 the larger. You can change the default parameter values by adding, for 4267 example, 4268 4269 --with-pcre2grep-bufsize=51200 4270 --with-pcre2grep-max-bufsize=2097152 4271 4272 to the configure command. The caller of pcre2grep can override these 4273 values by using --buffer-size and --max-buffer-size on the command 4274 line. 4275 4276 4277PCRE2TEST OPTION FOR LIBREADLINE SUPPORT 4278 4279 If you add one of 4280 4281 --enable-pcre2test-libreadline 4282 --enable-pcre2test-libedit 4283 4284 to the configure command, pcre2test is linked with the libreadline or- 4285 libedit library, respectively, and when its input is from a terminal, 4286 it reads it using the readline() function. This provides line-editing 4287 and history facilities. Note that libreadline is GPL-licensed, so if 4288 you distribute a binary of pcre2test linked in this way, there may be 4289 licensing issues. These can be avoided by linking instead with libedit, 4290 which has a BSD licence. 4291 4292 Setting --enable-pcre2test-libreadline causes the -lreadline option to 4293 be added to the pcre2test build. In many operating environments with a 4294 sytem-installed readline library this is sufficient. However, in some 4295 environments (e.g. if an unmodified distribution version of readline is 4296 in use), some extra configuration may be necessary. The INSTALL file 4297 for libreadline says this: 4298 4299 "Readline uses the termcap functions, but does not link with 4300 the termcap or curses library itself, allowing applications 4301 which link with readline the to choose an appropriate library." 4302 4303 If your environment has not been set up so that an appropriate library 4304 is automatically included, you may need to add something like 4305 4306 LIBS="-ncurses" 4307 4308 immediately before the configure command. 4309 4310 4311INCLUDING DEBUGGING CODE 4312 4313 If you add 4314 4315 --enable-debug 4316 4317 to the configure command, additional debugging code is included in the 4318 build. This feature is intended for use by the PCRE2 maintainers. 4319 4320 4321DEBUGGING WITH VALGRIND SUPPORT 4322 4323 If you add 4324 4325 --enable-valgrind 4326 4327 to the configure command, PCRE2 will use valgrind annotations to mark 4328 certain memory regions as unaddressable. This allows it to detect in- 4329 valid memory accesses, and is mostly useful for debugging PCRE2 itself. 4330 4331 4332CODE COVERAGE REPORTING 4333 4334 If your C compiler is gcc, you can build a version of PCRE2 that can 4335 generate a code coverage report for its test suite. To enable this, you 4336 must install lcov version 1.6 or above. Then specify 4337 4338 --enable-coverage 4339 4340 to the configure command and build PCRE2 in the usual way. 4341 4342 Note that using ccache (a caching C compiler) is incompatible with code 4343 coverage reporting. If you have configured ccache to run automatically 4344 on your system, you must set the environment variable 4345 4346 CCACHE_DISABLE=1 4347 4348 before running make to build PCRE2, so that ccache is not used. 4349 4350 When --enable-coverage is used, the following addition targets are 4351 added to the Makefile: 4352 4353 make coverage 4354 4355 This creates a fresh coverage report for the PCRE2 test suite. It is 4356 equivalent to running "make coverage-reset", "make coverage-baseline", 4357 "make check", and then "make coverage-report". 4358 4359 make coverage-reset 4360 4361 This zeroes the coverage counters, but does nothing else. 4362 4363 make coverage-baseline 4364 4365 This captures baseline coverage information. 4366 4367 make coverage-report 4368 4369 This creates the coverage report. 4370 4371 make coverage-clean-report 4372 4373 This removes the generated coverage report without cleaning the cover- 4374 age data itself. 4375 4376 make coverage-clean-data 4377 4378 This removes the captured coverage data without removing the coverage 4379 files created at compile time (*.gcno). 4380 4381 make coverage-clean 4382 4383 This cleans all coverage data including the generated coverage report. 4384 For more information about code coverage, see the gcov and lcov docu- 4385 mentation. 4386 4387 4388DISABLING THE Z AND T FORMATTING MODIFIERS 4389 4390 The C99 standard defines formatting modifiers z and t for size_t and 4391 ptrdiff_t values, respectively. By default, PCRE2 uses these modifiers 4392 in environments other than old versions of Microsoft Visual Studio when 4393 __STDC_VERSION__ is defined and has a value greater than or equal to 4394 199901L (indicating support for C99). However, there is at least one 4395 environment that claims to be C99 but does not support these modifiers. 4396 If 4397 4398 --disable-percent-zt 4399 4400 is specified, no use is made of the z or t modifiers. Instead of %td or 4401 %zu, a suitable format is used depending in the size of long for the 4402 platform. 4403 4404 4405SUPPORT FOR FUZZERS 4406 4407 There is a special option for use by people who want to run fuzzing 4408 tests on PCRE2: 4409 4410 --enable-fuzz-support 4411 4412 At present this applies only to the 8-bit library. If set, it causes an 4413 extra library called libpcre2-fuzzsupport.a to be built, but not in- 4414 stalled. This contains a single function called LLVMFuzzerTestOneIn- 4415 put() whose arguments are a pointer to a string and the length of the 4416 string. When called, this function tries to compile the string as a 4417 pattern, and if that succeeds, to match it. This is done both with no 4418 options and with some random options bits that are generated from the 4419 string. 4420 4421 Setting --enable-fuzz-support also causes a binary called pcre2fuz- 4422 zcheck to be created. This is normally run under valgrind or used when 4423 PCRE2 is compiled with address sanitizing enabled. It calls the fuzzing 4424 function and outputs information about what it is doing. The input 4425 strings are specified by arguments: if an argument starts with "=" the 4426 rest of it is a literal input string. Otherwise, it is assumed to be a 4427 file name, and the contents of the file are the test string. 4428 4429 4430OBSOLETE OPTION 4431 4432 In versions of PCRE2 prior to 10.30, there were two ways of handling 4433 backtracking in the pcre2_match() function. The default was to use the 4434 system stack, but if 4435 4436 --disable-stack-for-recursion 4437 4438 was set, memory on the heap was used. From release 10.30 onwards this 4439 has changed (the stack is no longer used) and this option now does 4440 nothing except give a warning. 4441 4442 4443SEE ALSO 4444 4445 pcre2api(3), pcre2-config(3). 4446 4447 4448AUTHOR 4449 4450 Philip Hazel 4451 University Computing Service 4452 Cambridge, England. 4453 4454 4455REVISION 4456 4457 Last updated: 08 December 2021 4458 Copyright (c) 1997-2021 University of Cambridge. 4459------------------------------------------------------------------------------ 4460 4461 4462PCRE2CALLOUT(3) Library Functions Manual PCRE2CALLOUT(3) 4463 4464 4465 4466NAME 4467 PCRE2 - Perl-compatible regular expressions (revised API) 4468 4469SYNOPSIS 4470 4471 #include <pcre2.h> 4472 4473 int (*pcre2_callout)(pcre2_callout_block *, void *); 4474 4475 int pcre2_callout_enumerate(const pcre2_code *code, 4476 int (*callback)(pcre2_callout_enumerate_block *, void *), 4477 void *user_data); 4478 4479 4480DESCRIPTION 4481 4482 PCRE2 provides a feature called "callout", which is a means of tempo- 4483 rarily passing control to the caller of PCRE2 in the middle of pattern 4484 matching. The caller of PCRE2 provides an external function by putting 4485 its entry point in a match context (see pcre2_set_callout() in the 4486 pcre2api documentation). 4487 4488 When using the pcre2_substitute() function, an additional callout fea- 4489 ture is available. This does a callout after each change to the subject 4490 string and is described in the pcre2api documentation; the rest of this 4491 document is concerned with callouts during pattern matching. 4492 4493 Within a regular expression, (?C<arg>) indicates a point at which the 4494 external function is to be called. Different callout points can be 4495 identified by putting a number less than 256 after the letter C. The 4496 default value is zero. Alternatively, the argument may be a delimited 4497 string. The starting delimiter must be one of ` ' " ^ % # $ { and the 4498 ending delimiter is the same as the start, except for {, where the end- 4499 ing delimiter is }. If the ending delimiter is needed within the 4500 string, it must be doubled. For example, this pattern has two callout 4501 points: 4502 4503 (?C1)abc(?C"some ""arbitrary"" text")def 4504 4505 If the PCRE2_AUTO_CALLOUT option bit is set when a pattern is compiled, 4506 PCRE2 automatically inserts callouts, all with number 255, before each 4507 item in the pattern except for immediately before or after an explicit 4508 callout. For example, if PCRE2_AUTO_CALLOUT is used with the pattern 4509 4510 A(?C3)B 4511 4512 it is processed as if it were 4513 4514 (?C255)A(?C3)B(?C255) 4515 4516 Here is a more complicated example: 4517 4518 A(\d{2}|--) 4519 4520 With PCRE2_AUTO_CALLOUT, this pattern is processed as if it were 4521 4522 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255) 4523 4524 Notice that there is a callout before and after each parenthesis and 4525 alternation bar. If the pattern contains a conditional group whose con- 4526 dition is an assertion, an automatic callout is inserted immediately 4527 before the condition. Such a callout may also be inserted explicitly, 4528 for example: 4529 4530 (?(?C9)(?=a)ab|de) (?(?C%text%)(?!=d)ab|de) 4531 4532 This applies only to assertion conditions (because they are themselves 4533 independent groups). 4534 4535 Callouts can be useful for tracking the progress of pattern matching. 4536 The pcre2test program has a pattern qualifier (/auto_callout) that sets 4537 automatic callouts. When any callouts are present, the output from 4538 pcre2test indicates how the pattern is being matched. This is useful 4539 information when you are trying to optimize the performance of a par- 4540 ticular pattern. 4541 4542 4543MISSING CALLOUTS 4544 4545 You should be aware that, because of optimizations in the way PCRE2 4546 compiles and matches patterns, callouts sometimes do not happen exactly 4547 as you might expect. 4548 4549 Auto-possessification 4550 4551 At compile time, PCRE2 "auto-possessifies" repeated items when it knows 4552 that what follows cannot be part of the repeat. For example, a+[bc] is 4553 compiled as if it were a++[bc]. The pcre2test output when this pattern 4554 is compiled with PCRE2_ANCHORED and PCRE2_AUTO_CALLOUT and then applied 4555 to the string "aaaa" is: 4556 4557 --->aaaa 4558 +0 ^ a+ 4559 +2 ^ ^ [bc] 4560 No match 4561 4562 This indicates that when matching [bc] fails, there is no backtracking 4563 into a+ (because it is being treated as a++) and therefore the callouts 4564 that would be taken for the backtracks do not occur. You can disable 4565 the auto-possessify feature by passing PCRE2_NO_AUTO_POSSESS to 4566 pcre2_compile(), or starting the pattern with (*NO_AUTO_POSSESS). In 4567 this case, the output changes to this: 4568 4569 --->aaaa 4570 +0 ^ a+ 4571 +2 ^ ^ [bc] 4572 +2 ^ ^ [bc] 4573 +2 ^ ^ [bc] 4574 +2 ^^ [bc] 4575 No match 4576 4577 This time, when matching [bc] fails, the matcher backtracks into a+ and 4578 tries again, repeatedly, until a+ itself fails. 4579 4580 Automatic .* anchoring 4581 4582 By default, an optimization is applied when .* is the first significant 4583 item in a pattern. If PCRE2_DOTALL is set, so that the dot can match 4584 any character, the pattern is automatically anchored. If PCRE2_DOTALL 4585 is not set, a match can start only after an internal newline or at the 4586 beginning of the subject, and pcre2_compile() remembers this. If a pat- 4587 tern has more than one top-level branch, automatic anchoring occurs if 4588 all branches are anchorable. 4589 4590 This optimization is disabled, however, if .* is in an atomic group or 4591 if there is a backreference to the capture group in which it appears. 4592 It is also disabled if the pattern contains (*PRUNE) or (*SKIP). How- 4593 ever, the presence of callouts does not affect it. 4594 4595 For example, if the pattern .*\d is compiled with PCRE2_AUTO_CALLOUT 4596 and applied to the string "aa", the pcre2test output is: 4597 4598 --->aa 4599 +0 ^ .* 4600 +2 ^ ^ \d 4601 +2 ^^ \d 4602 +2 ^ \d 4603 No match 4604 4605 This shows that all match attempts start at the beginning of the sub- 4606 ject. In other words, the pattern is anchored. You can disable this op- 4607 timization by passing PCRE2_NO_DOTSTAR_ANCHOR to pcre2_compile(), or 4608 starting the pattern with (*NO_DOTSTAR_ANCHOR). In this case, the out- 4609 put changes to: 4610 4611 --->aa 4612 +0 ^ .* 4613 +2 ^ ^ \d 4614 +2 ^^ \d 4615 +2 ^ \d 4616 +0 ^ .* 4617 +2 ^^ \d 4618 +2 ^ \d 4619 No match 4620 4621 This shows more match attempts, starting at the second subject charac- 4622 ter. Another optimization, described in the next section, means that 4623 there is no subsequent attempt to match with an empty subject. 4624 4625 Other optimizations 4626 4627 Other optimizations that provide fast "no match" results also affect 4628 callouts. For example, if the pattern is 4629 4630 ab(?C4)cd 4631 4632 PCRE2 knows that any matching string must contain the letter "d". If 4633 the subject string is "abyz", the lack of "d" means that matching 4634 doesn't ever start, and the callout is never reached. However, with 4635 "abyd", though the result is still no match, the callout is obeyed. 4636 4637 For most patterns PCRE2 also knows the minimum length of a matching 4638 string, and will immediately give a "no match" return without actually 4639 running a match if the subject is not long enough, or, for unanchored 4640 patterns, if it has been scanned far enough. 4641 4642 You can disable these optimizations by passing the PCRE2_NO_START_OPTI- 4643 MIZE option to pcre2_compile(), or by starting the pattern with 4644 (*NO_START_OPT). This slows down the matching process, but does ensure 4645 that callouts such as the example above are obeyed. 4646 4647 4648THE CALLOUT INTERFACE 4649 4650 During matching, when PCRE2 reaches a callout point, if an external 4651 function is provided in the match context, it is called. This applies 4652 to both normal, DFA, and JIT matching. The first argument to the call- 4653 out function is a pointer to a pcre2_callout block. The second argument 4654 is the void * callout data that was supplied when the callout was set 4655 up by calling pcre2_set_callout() (see the pcre2api documentation). The 4656 callout block structure contains the following fields, not necessarily 4657 in this order: 4658 4659 uint32_t version; 4660 uint32_t callout_number; 4661 uint32_t capture_top; 4662 uint32_t capture_last; 4663 uint32_t callout_flags; 4664 PCRE2_SIZE *offset_vector; 4665 PCRE2_SPTR mark; 4666 PCRE2_SPTR subject; 4667 PCRE2_SIZE subject_length; 4668 PCRE2_SIZE start_match; 4669 PCRE2_SIZE current_position; 4670 PCRE2_SIZE pattern_position; 4671 PCRE2_SIZE next_item_length; 4672 PCRE2_SIZE callout_string_offset; 4673 PCRE2_SIZE callout_string_length; 4674 PCRE2_SPTR callout_string; 4675 4676 The version field contains the version number of the block format. The 4677 current version is 2; the three callout string fields were added for 4678 version 1, and the callout_flags field for version 2. If you are writ- 4679 ing an application that might use an earlier release of PCRE2, you 4680 should check the version number before accessing any of these fields. 4681 The version number will increase in future if more fields are added, 4682 but the intention is never to remove any of the existing fields. 4683 4684 Fields for numerical callouts 4685 4686 For a numerical callout, callout_string is NULL, and callout_number 4687 contains the number of the callout, in the range 0-255. This is the 4688 number that follows (?C for callouts that part of the pattern; it is 4689 255 for automatically generated callouts. 4690 4691 Fields for string callouts 4692 4693 For callouts with string arguments, callout_number is always zero, and 4694 callout_string points to the string that is contained within the com- 4695 piled pattern. Its length is given by callout_string_length. Duplicated 4696 ending delimiters that were present in the original pattern string have 4697 been turned into single characters, but there is no other processing of 4698 the callout string argument. An additional code unit containing binary 4699 zero is present after the string, but is not included in the length. 4700 The delimiter that was used to start the string is also stored within 4701 the pattern, immediately before the string itself. You can access this 4702 delimiter as callout_string[-1] if you need it. 4703 4704 The callout_string_offset field is the code unit offset to the start of 4705 the callout argument string within the original pattern string. This is 4706 provided for the benefit of applications such as script languages that 4707 might need to report errors in the callout string within the pattern. 4708 4709 Fields for all callouts 4710 4711 The remaining fields in the callout block are the same for both kinds 4712 of callout. 4713 4714 The offset_vector field is a pointer to a vector of capturing offsets 4715 (the "ovector"). You may read the elements in this vector, but you must 4716 not change any of them. 4717 4718 For calls to pcre2_match(), the offset_vector field is not (since re- 4719 lease 10.30) a pointer to the actual ovector that was passed to the 4720 matching function in the match data block. Instead it points to an in- 4721 ternal ovector of a size large enough to hold all possible captured 4722 substrings in the pattern. Note that whenever a recursion or subroutine 4723 call within a pattern completes, the capturing state is reset to what 4724 it was before. 4725 4726 The capture_last field contains the number of the most recently cap- 4727 tured substring, and the capture_top field contains one more than the 4728 number of the highest numbered captured substring so far. If no sub- 4729 strings have yet been captured, the value of capture_last is 0 and the 4730 value of capture_top is 1. The values of these fields do not always 4731 differ by one; for example, when the callout in the pattern 4732 ((a)(b))(?C2) is taken, capture_last is 1 but capture_top is 4. 4733 4734 The contents of ovector[2] to ovector[<capture_top>*2-1] can be in- 4735 spected in order to extract substrings that have been matched so far, 4736 in the same way as extracting substrings after a match has completed. 4737 The values in ovector[0] and ovector[1] are always PCRE2_UNSET because 4738 the match is by definition not complete. Substrings that have not been 4739 captured but whose numbers are less than capture_top also have both of 4740 their ovector slots set to PCRE2_UNSET. 4741 4742 For DFA matching, the offset_vector field points to the ovector that 4743 was passed to the matching function in the match data block for call- 4744 outs at the top level, but to an internal ovector during the processing 4745 of pattern recursions, lookarounds, and atomic groups. However, these 4746 ovectors hold no useful information because pcre2_dfa_match() does not 4747 support substring capturing. The value of capture_top is always 1 and 4748 the value of capture_last is always 0 for DFA matching. 4749 4750 The subject and subject_length fields contain copies of the values that 4751 were passed to the matching function. 4752 4753 The start_match field normally contains the offset within the subject 4754 at which the current match attempt started. However, if the escape se- 4755 quence \K has been encountered, this value is changed to reflect the 4756 modified starting point. If the pattern is not anchored, the callout 4757 function may be called several times from the same point in the pattern 4758 for different starting points in the subject. 4759 4760 The current_position field contains the offset within the subject of 4761 the current match pointer. 4762 4763 The pattern_position field contains the offset in the pattern string to 4764 the next item to be matched. 4765 4766 The next_item_length field contains the length of the next item to be 4767 processed in the pattern string. When the callout is at the end of the 4768 pattern, the length is zero. When the callout precedes an opening 4769 parenthesis, the length includes meta characters that follow the paren- 4770 thesis. For example, in a callout before an assertion such as (?=ab) 4771 the length is 3. For an an alternation bar or a closing parenthesis, 4772 the length is one, unless a closing parenthesis is followed by a quan- 4773 tifier, in which case its length is included. (This changed in release 4774 10.23. In earlier releases, before an opening parenthesis the length 4775 was that of the entire group, and before an alternation bar or a clos- 4776 ing parenthesis the length was zero.) 4777 4778 The pattern_position and next_item_length fields are intended to help 4779 in distinguishing between different automatic callouts, which all have 4780 the same callout number. However, they are set for all callouts, and 4781 are used by pcre2test to show the next item to be matched when display- 4782 ing callout information. 4783 4784 In callouts from pcre2_match() the mark field contains a pointer to the 4785 zero-terminated name of the most recently passed (*MARK), (*PRUNE), or 4786 (*THEN) item in the match, or NULL if no such items have been passed. 4787 Instances of (*PRUNE) or (*THEN) without a name do not obliterate a 4788 previous (*MARK). In callouts from the DFA matching function this field 4789 always contains NULL. 4790 4791 The callout_flags field is always zero in callouts from 4792 pcre2_dfa_match() or when JIT is being used. When pcre2_match() without 4793 JIT is used, the following bits may be set: 4794 4795 PCRE2_CALLOUT_STARTMATCH 4796 4797 This is set for the first callout after the start of matching for each 4798 new starting position in the subject. 4799 4800 PCRE2_CALLOUT_BACKTRACK 4801 4802 This is set if there has been a matching backtrack since the previous 4803 callout, or since the start of matching if this is the first callout 4804 from a pcre2_match() run. 4805 4806 Both bits are set when a backtrack has caused a "bumpalong" to a new 4807 starting position in the subject. Output from pcre2test does not indi- 4808 cate the presence of these bits unless the callout_extra modifier is 4809 set. 4810 4811 The information in the callout_flags field is provided so that applica- 4812 tions can track and tell their users how matching with backtracking is 4813 done. This can be useful when trying to optimize patterns, or just to 4814 understand how PCRE2 works. There is no support in pcre2_dfa_match() 4815 because there is no backtracking in DFA matching, and there is no sup- 4816 port in JIT because JIT is all about maximimizing matching performance. 4817 In both these cases the callout_flags field is always zero. 4818 4819 4820RETURN VALUES FROM CALLOUTS 4821 4822 The external callout function returns an integer to PCRE2. If the value 4823 is zero, matching proceeds as normal. If the value is greater than 4824 zero, matching fails at the current point, but the testing of other 4825 matching possibilities goes ahead, just as if a lookahead assertion had 4826 failed. If the value is less than zero, the match is abandoned, and the 4827 matching function returns the negative value. 4828 4829 Negative values should normally be chosen from the set of PCRE2_ER- 4830 ROR_xxx values. In particular, PCRE2_ERROR_NOMATCH forces a standard 4831 "no match" failure. The error number PCRE2_ERROR_CALLOUT is reserved 4832 for use by callout functions; it will never be used by PCRE2 itself. 4833 4834 4835CALLOUT ENUMERATION 4836 4837 int pcre2_callout_enumerate(const pcre2_code *code, 4838 int (*callback)(pcre2_callout_enumerate_block *, void *), 4839 void *user_data); 4840 4841 A script language that supports the use of string arguments in callouts 4842 might like to scan all the callouts in a pattern before running the 4843 match. This can be done by calling pcre2_callout_enumerate(). The first 4844 argument is a pointer to a compiled pattern, the second points to a 4845 callback function, and the third is arbitrary user data. The callback 4846 function is called for every callout in the pattern in the order in 4847 which they appear. Its first argument is a pointer to a callout enumer- 4848 ation block, and its second argument is the user_data value that was 4849 passed to pcre2_callout_enumerate(). The data block contains the fol- 4850 lowing fields: 4851 4852 version Block version number 4853 pattern_position Offset to next item in pattern 4854 next_item_length Length of next item in pattern 4855 callout_number Number for numbered callouts 4856 callout_string_offset Offset to string within pattern 4857 callout_string_length Length of callout string 4858 callout_string Points to callout string or is NULL 4859 4860 The version number is currently 0. It will increase if new fields are 4861 ever added to the block. The remaining fields are the same as their 4862 namesakes in the pcre2_callout block that is used for callouts during 4863 matching, as described above. 4864 4865 Note that the value of pattern_position is unique for each callout. 4866 However, if a callout occurs inside a group that is quantified with a 4867 non-zero minimum or a fixed maximum, the group is replicated inside the 4868 compiled pattern. For example, a pattern such as /(a){2}/ is compiled 4869 as if it were /(a)(a)/. This means that the callout will be enumerated 4870 more than once, but with the same value for pattern_position in each 4871 case. 4872 4873 The callback function should normally return zero. If it returns a non- 4874 zero value, scanning the pattern stops, and that value is returned from 4875 pcre2_callout_enumerate(). 4876 4877 4878AUTHOR 4879 4880 Philip Hazel 4881 University Computing Service 4882 Cambridge, England. 4883 4884 4885REVISION 4886 4887 Last updated: 03 February 2019 4888 Copyright (c) 1997-2019 University of Cambridge. 4889------------------------------------------------------------------------------ 4890 4891 4892PCRE2COMPAT(3) Library Functions Manual PCRE2COMPAT(3) 4893 4894 4895 4896NAME 4897 PCRE2 - Perl-compatible regular expressions (revised API) 4898 4899DIFFERENCES BETWEEN PCRE2 AND PERL 4900 4901 This document describes some of the differences in the ways that PCRE2 4902 and Perl handle regular expressions. The differences described here are 4903 with respect to Perl version 5.34.0, but as both Perl and PCRE2 are 4904 continually changing, the information may at times be out of date. 4905 4906 1. When PCRE2_DOTALL (equivalent to Perl's /s qualifier) is not set, 4907 the behaviour of the '.' metacharacter differs from Perl. In PCRE2, '.' 4908 matches the next character unless it is the start of a newline se- 4909 quence. This means that, if the newline setting is CR, CRLF, or NUL, 4910 '.' will match the code point LF (0x0A) in ASCII/Unicode environments, 4911 and NL (either 0x15 or 0x25) when using EBCDIC. In Perl, '.' appears 4912 never to match LF, even when 0x0A is not a newline indicator. 4913 4914 2. PCRE2 has only a subset of Perl's Unicode support. Details of what 4915 it does have are given in the pcre2unicode page. 4916 4917 3. Like Perl, PCRE2 allows repeat quantifiers on parenthesized asser- 4918 tions, but they do not mean what you might think. For example, (?!a){3} 4919 does not assert that the next three characters are not "a". It just as- 4920 serts that the next character is not "a" three times (in principle; 4921 PCRE2 optimizes this to run the assertion just once). Perl allows some 4922 repeat quantifiers on other assertions, for example, \b* , but these do 4923 not seem to have any use. PCRE2 does not allow any kind of quantifier 4924 on non-lookaround assertions. 4925 4926 4. Capture groups that occur inside negative lookaround assertions are 4927 counted, but their entries in the offsets vector are set only when a 4928 negative assertion is a condition that has a matching branch (that is, 4929 the condition is false). Perl may set such capture groups in other 4930 circumstances. 4931 4932 5. The following Perl escape sequences are not supported: \F, \l, \L, 4933 \u, \U, and \N when followed by a character name. \N on its own, match- 4934 ing a non-newline character, and \N{U+dd..}, matching a Unicode code 4935 point, are supported. The escapes that modify the case of following 4936 letters are implemented by Perl's general string-handling and are not 4937 part of its pattern matching engine. If any of these are encountered by 4938 PCRE2, an error is generated by default. However, if either of the 4939 PCRE2_ALT_BSUX or PCRE2_EXTRA_ALT_BSUX options is set, \U and \u are 4940 interpreted as ECMAScript interprets them. 4941 4942 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE2 4943 is built with Unicode support (the default). The properties that can be 4944 tested with \p and \P are limited to the general category properties 4945 such as Lu and Nd, script names such as Greek or Han, Bidi_Class, 4946 Bidi_Control, and the derived properties Any and LC (synonym L&). Both 4947 PCRE2 and Perl support the Cs (surrogate) property, but in PCRE2 its 4948 use is limited. See the pcre2pattern documentation for details. The 4949 long synonyms for property names that Perl supports (such as \p{Let- 4950 ter}) are not supported by PCRE2, nor is it permitted to prefix any of 4951 these properties with "Is". 4952 4953 7. PCRE2 supports the \Q...\E escape for quoting substrings. Characters 4954 in between are treated as literals. However, this is slightly different 4955 from Perl in that $ and @ are also handled as literals inside the 4956 quotes. In Perl, they cause variable interpolation (PCRE2 does not have 4957 variables). Also, Perl does "double-quotish backslash interpolation" on 4958 any backslashes between \Q and \E which, its documentation says, "may 4959 lead to confusing results". PCRE2 treats a backslash between \Q and \E 4960 just like any other character. Note the following examples: 4961 4962 Pattern PCRE2 matches Perl matches 4963 4964 \Qabc$xyz\E abc$xyz abc followed by the 4965 contents of $xyz 4966 \Qabc\$xyz\E abc\$xyz abc\$xyz 4967 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz 4968 \QA\B\E A\B A\B 4969 \Q\\E \ \\E 4970 4971 The \Q...\E sequence is recognized both inside and outside character 4972 classes by both PCRE2 and Perl. 4973 4974 8. Fairly obviously, PCRE2 does not support the (?{code}) and 4975 (??{code}) constructions. However, PCRE2 does have a "callout" feature, 4976 which allows an external function to be called during pattern matching. 4977 See the pcre2callout documentation for details. 4978 4979 9. Subroutine calls (whether recursive or not) were treated as atomic 4980 groups up to PCRE2 release 10.23, but from release 10.30 this changed, 4981 and backtracking into subroutine calls is now supported, as in Perl. 4982 4983 10. In PCRE2, if any of the backtracking control verbs are used in a 4984 group that is called as a subroutine (whether or not recursively), 4985 their effect is confined to that group; it does not extend to the sur- 4986 rounding pattern. This is not always the case in Perl. In particular, 4987 if (*THEN) is present in a group that is called as a subroutine, its 4988 action is limited to that group, even if the group does not contain any 4989 | characters. Note that such groups are processed as anchored at the 4990 point where they are tested. 4991 4992 11. If a pattern contains more than one backtracking control verb, the 4993 first one that is backtracked onto acts. For example, in the pattern 4994 A(*COMMIT)B(*PRUNE)C a failure in B triggers (*COMMIT), but a failure 4995 in C triggers (*PRUNE). Perl's behaviour is more complex; in many cases 4996 it is the same as PCRE2, but there are cases where it differs. 4997 4998 12. There are some differences that are concerned with the settings of 4999 captured strings when part of a pattern is repeated. For example, 5000 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2 un- 5001 set, but in PCRE2 it is set to "b". 5002 5003 13. PCRE2's handling of duplicate capture group numbers and names is 5004 not as general as Perl's. This is a consequence of the fact the PCRE2 5005 works internally just with numbers, using an external table to trans- 5006 late between numbers and names. In particular, a pattern such as 5007 (?|(?<a>A)|(?<b>B)), where the two capture groups have the same number 5008 but different names, is not supported, and causes an error at compile 5009 time. If it were allowed, it would not be possible to distinguish which 5010 group matched, because both names map to capture group number 1. To 5011 avoid this confusing situation, an error is given at compile time. 5012 5013 14. Perl used to recognize comments in some places that PCRE2 does not, 5014 for example, between the ( and ? at the start of a group. If the /x 5015 modifier is set, Perl allowed white space between ( and ? though the 5016 latest Perls give an error (for a while it was just deprecated). There 5017 may still be some cases where Perl behaves differently. 5018 5019 15. Perl, when in warning mode, gives warnings for character classes 5020 such as [A-\d] or [a-[:digit:]]. It then treats the hyphens as liter- 5021 als. PCRE2 has no warning features, so it gives an error in these cases 5022 because they are almost certainly user mistakes. 5023 5024 16. In PCRE2, the upper/lower case character properties Lu and Ll are 5025 not affected when case-independent matching is specified. For example, 5026 \p{Lu} always matches an upper case letter. I think Perl has changed in 5027 this respect; in the release at the time of writing (5.34), \p{Lu} and 5028 \p{Ll} match all letters, regardless of case, when case independence is 5029 specified. 5030 5031 17. From release 5.32.0, Perl locks out the use of \K in lookaround as- 5032 sertions. From release 10.38 PCRE2 does the same by default. However, 5033 there is an option for re-enabling the previous behaviour. When this 5034 option is set, \K is acted on when it occurs in positive assertions, 5035 but is ignored in negative assertions. 5036 5037 18. PCRE2 provides some extensions to the Perl regular expression fa- 5038 cilities. Perl 5.10 included new features that were not in earlier 5039 versions of Perl, some of which (such as named parentheses) were in 5040 PCRE2 for some time before. This list is with respect to Perl 5.34: 5041 5042 (a) Although lookbehind assertions in PCRE2 must match fixed length 5043 strings, each alternative toplevel branch of a lookbehind assertion can 5044 match a different length of string. Perl used to require them all to 5045 have the same length, but the latest version has some variable length 5046 support. 5047 5048 (b) From PCRE2 10.23, backreferences to groups of fixed length are sup- 5049 ported in lookbehinds, provided that there is no possibility of refer- 5050 encing a non-unique number or name. Perl does not support backrefer- 5051 ences in lookbehinds. 5052 5053 (c) If PCRE2_DOLLAR_ENDONLY is set and PCRE2_MULTILINE is not set, the 5054 $ meta-character matches only at the very end of the string. 5055 5056 (d) A backslash followed by a letter with no special meaning is 5057 faulted. (Perl can be made to issue a warning.) 5058 5059 (e) If PCRE2_UNGREEDY is set, the greediness of the repetition quanti- 5060 fiers is inverted, that is, by default they are not greedy, but if fol- 5061 lowed by a question mark they are. 5062 5063 (f) PCRE2_ANCHORED can be used at matching time to force a pattern to 5064 be tried only at the first matching position in the subject string. 5065 5066 (g) The PCRE2_NOTBOL, PCRE2_NOTEOL, PCRE2_NOTEMPTY and 5067 PCRE2_NOTEMPTY_ATSTART options have no Perl equivalents. 5068 5069 (h) The \R escape sequence can be restricted to match only CR, LF, or 5070 CRLF by the PCRE2_BSR_ANYCRLF option. 5071 5072 (i) The callout facility is PCRE2-specific. Perl supports codeblocks 5073 and variable interpolation, but not general hooks on every match. 5074 5075 (j) The partial matching facility is PCRE2-specific. 5076 5077 (k) The alternative matching function (pcre2_dfa_match() matches in a 5078 different way and is not Perl-compatible. 5079 5080 (l) PCRE2 recognizes some special sequences such as (*CR) or (*NO_JIT) 5081 at the start of a pattern. These set overall options that cannot be 5082 changed within the pattern. 5083 5084 (m) PCRE2 supports non-atomic positive lookaround assertions. This is 5085 an extension to the lookaround facilities. The default, Perl-compatible 5086 lookarounds are atomic. 5087 5088 19. The Perl /a modifier restricts /d numbers to pure ascii, and the 5089 /aa modifier restricts /i case-insensitive matching to pure ascii, ig- 5090 noring Unicode rules. This separation cannot be represented with 5091 PCRE2_UCP. 5092 5093 20. Perl has different limits than PCRE2. See the pcre2limit documenta- 5094 tion for details. Perl went with 5.10 from recursion to iteration keep- 5095 ing the intermediate matches on the heap, which is ~10% slower but does 5096 not fall into any stack-overflow limit. PCRE2 made a similar change at 5097 release 10.30, and also has many build-time and run-time customizable 5098 limits. 5099 5100 5101AUTHOR 5102 5103 Philip Hazel 5104 Retired from University Computing Service 5105 Cambridge, England. 5106 5107 5108REVISION 5109 5110 Last updated: 08 December 2021 5111 Copyright (c) 1997-2021 University of Cambridge. 5112------------------------------------------------------------------------------ 5113 5114 5115PCRE2JIT(3) Library Functions Manual PCRE2JIT(3) 5116 5117 5118 5119NAME 5120 PCRE2 - Perl-compatible regular expressions (revised API) 5121 5122PCRE2 JUST-IN-TIME COMPILER SUPPORT 5123 5124 Just-in-time compiling is a heavyweight optimization that can greatly 5125 speed up pattern matching. However, it comes at the cost of extra pro- 5126 cessing before the match is performed, so it is of most benefit when 5127 the same pattern is going to be matched many times. This does not nec- 5128 essarily mean many calls of a matching function; if the pattern is not 5129 anchored, matching attempts may take place many times at various posi- 5130 tions in the subject, even for a single call. Therefore, if the subject 5131 string is very long, it may still pay to use JIT even for one-off 5132 matches. JIT support is available for all of the 8-bit, 16-bit and 5133 32-bit PCRE2 libraries. 5134 5135 JIT support applies only to the traditional Perl-compatible matching 5136 function. It does not apply when the DFA matching function is being 5137 used. The code for this support was written by Zoltan Herczeg. 5138 5139 5140AVAILABILITY OF JIT SUPPORT 5141 5142 JIT support is an optional feature of PCRE2. The "configure" option 5143 --enable-jit (or equivalent CMake option) must be set when PCRE2 is 5144 built if you want to use JIT. The support is limited to the following 5145 hardware platforms: 5146 5147 ARM 32-bit (v5, v7, and Thumb2) 5148 ARM 64-bit 5149 IBM s390x 64 bit 5150 Intel x86 32-bit and 64-bit 5151 MIPS 32-bit and 64-bit 5152 Power PC 32-bit and 64-bit 5153 SPARC 32-bit 5154 5155 If --enable-jit is set on an unsupported platform, compilation fails. 5156 5157 A program can tell if JIT support is available by calling pcre2_con- 5158 fig() with the PCRE2_CONFIG_JIT option. The result is 1 when JIT is 5159 available, and 0 otherwise. However, a simple program does not need to 5160 check this in order to use JIT. The API is implemented in a way that 5161 falls back to the interpretive code if JIT is not available. For pro- 5162 grams that need the best possible performance, there is also a "fast 5163 path" API that is JIT-specific. 5164 5165 5166SIMPLE USE OF JIT 5167 5168 To make use of the JIT support in the simplest way, all you have to do 5169 is to call pcre2_jit_compile() after successfully compiling a pattern 5170 with pcre2_compile(). This function has two arguments: the first is the 5171 compiled pattern pointer that was returned by pcre2_compile(), and the 5172 second is zero or more of the following option bits: PCRE2_JIT_COM- 5173 PLETE, PCRE2_JIT_PARTIAL_HARD, or PCRE2_JIT_PARTIAL_SOFT. 5174 5175 If JIT support is not available, a call to pcre2_jit_compile() does 5176 nothing and returns PCRE2_ERROR_JIT_BADOPTION. Otherwise, the compiled 5177 pattern is passed to the JIT compiler, which turns it into machine code 5178 that executes much faster than the normal interpretive code, but yields 5179 exactly the same results. The returned value from pcre2_jit_compile() 5180 is zero on success, or a negative error code. 5181 5182 There is a limit to the size of pattern that JIT supports, imposed by 5183 the size of machine stack that it uses. The exact rules are not docu- 5184 mented because they may change at any time, in particular, when new op- 5185 timizations are introduced. If a pattern is too big, a call to 5186 pcre2_jit_compile() returns PCRE2_ERROR_NOMEMORY. 5187 5188 PCRE2_JIT_COMPLETE requests the JIT compiler to generate code for com- 5189 plete matches. If you want to run partial matches using the PCRE2_PAR- 5190 TIAL_HARD or PCRE2_PARTIAL_SOFT options of pcre2_match(), you should 5191 set one or both of the other options as well as, or instead of 5192 PCRE2_JIT_COMPLETE. The JIT compiler generates different optimized code 5193 for each of the three modes (normal, soft partial, hard partial). When 5194 pcre2_match() is called, the appropriate code is run if it is avail- 5195 able. Otherwise, the pattern is matched using interpretive code. 5196 5197 You can call pcre2_jit_compile() multiple times for the same compiled 5198 pattern. It does nothing if it has previously compiled code for any of 5199 the option bits. For example, you can call it once with PCRE2_JIT_COM- 5200 PLETE and (perhaps later, when you find you need partial matching) 5201 again with PCRE2_JIT_COMPLETE and PCRE2_JIT_PARTIAL_HARD. This time it 5202 will ignore PCRE2_JIT_COMPLETE and just compile code for partial match- 5203 ing. If pcre2_jit_compile() is called with no option bits set, it imme- 5204 diately returns zero. This is an alternative way of testing whether JIT 5205 is available. 5206 5207 At present, it is not possible to free JIT compiled code except when 5208 the entire compiled pattern is freed by calling pcre2_code_free(). 5209 5210 In some circumstances you may need to call additional functions. These 5211 are described in the section entitled "Controlling the JIT stack" be- 5212 low. 5213 5214 There are some pcre2_match() options that are not supported by JIT, and 5215 there are also some pattern items that JIT cannot handle. Details are 5216 given below. In both cases, matching automatically falls back to the 5217 interpretive code. If you want to know whether JIT was actually used 5218 for a particular match, you should arrange for a JIT callback function 5219 to be set up as described in the section entitled "Controlling the JIT 5220 stack" below, even if you do not need to supply a non-default JIT 5221 stack. Such a callback function is called whenever JIT code is about to 5222 be obeyed. If the match-time options are not right for JIT execution, 5223 the callback function is not obeyed. 5224 5225 If the JIT compiler finds an unsupported item, no JIT data is gener- 5226 ated. You can find out if JIT matching is available after compiling a 5227 pattern by calling pcre2_pattern_info() with the PCRE2_INFO_JITSIZE op- 5228 tion. A non-zero result means that JIT compilation was successful. A 5229 result of 0 means that JIT support is not available, or the pattern was 5230 not processed by pcre2_jit_compile(), or the JIT compiler was not able 5231 to handle the pattern. 5232 5233 5234MATCHING SUBJECTS CONTAINING INVALID UTF 5235 5236 When a pattern is compiled with the PCRE2_UTF option, subject strings 5237 are normally expected to be a valid sequence of UTF code units. By de- 5238 fault, this is checked at the start of matching and an error is gener- 5239 ated if invalid UTF is detected. The PCRE2_NO_UTF_CHECK option can be 5240 passed to pcre2_match() to skip the check (for improved performance) if 5241 you are sure that a subject string is valid. If this option is used 5242 with an invalid string, the result is undefined. 5243 5244 However, a way of running matches on strings that may contain invalid 5245 UTF sequences is available. Calling pcre2_compile() with the 5246 PCRE2_MATCH_INVALID_UTF option has two effects: it tells the inter- 5247 preter in pcre2_match() to support invalid UTF, and, if pcre2_jit_com- 5248 pile() is called, the compiled JIT code also supports invalid UTF. De- 5249 tails of how this support works, in both the JIT and the interpretive 5250 cases, is given in the pcre2unicode documentation. 5251 5252 There is also an obsolete option for pcre2_jit_compile() called 5253 PCRE2_JIT_INVALID_UTF, which currently exists only for backward compat- 5254 ibility. It is superseded by the pcre2_compile() option 5255 PCRE2_MATCH_INVALID_UTF and should no longer be used. It may be removed 5256 in future. 5257 5258 5259UNSUPPORTED OPTIONS AND PATTERN ITEMS 5260 5261 The pcre2_match() options that are supported for JIT matching are 5262 PCRE2_COPY_MATCHED_SUBJECT, PCRE2_NOTBOL, PCRE2_NOTEOL, PCRE2_NOTEMPTY, 5263 PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_UTF_CHECK, PCRE2_PARTIAL_HARD, and 5264 PCRE2_PARTIAL_SOFT. The PCRE2_ANCHORED and PCRE2_ENDANCHORED options 5265 are not supported at match time. 5266 5267 If the PCRE2_NO_JIT option is passed to pcre2_match() it disables the 5268 use of JIT, forcing matching by the interpreter code. 5269 5270 The only unsupported pattern items are \C (match a single data unit) 5271 when running in a UTF mode, and a callout immediately before an asser- 5272 tion condition in a conditional group. 5273 5274 5275RETURN VALUES FROM JIT MATCHING 5276 5277 When a pattern is matched using JIT matching, the return values are the 5278 same as those given by the interpretive pcre2_match() code, with the 5279 addition of one new error code: PCRE2_ERROR_JIT_STACKLIMIT. This means 5280 that the memory used for the JIT stack was insufficient. See "Control- 5281 ling the JIT stack" below for a discussion of JIT stack usage. 5282 5283 The error code PCRE2_ERROR_MATCHLIMIT is returned by the JIT code if 5284 searching a very large pattern tree goes on for too long, as it is in 5285 the same circumstance when JIT is not used, but the details of exactly 5286 what is counted are not the same. The PCRE2_ERROR_DEPTHLIMIT error code 5287 is never returned when JIT matching is used. 5288 5289 5290CONTROLLING THE JIT STACK 5291 5292 When the compiled JIT code runs, it needs a block of memory to use as a 5293 stack. By default, it uses 32KiB on the machine stack. However, some 5294 large or complicated patterns need more than this. The error PCRE2_ER- 5295 ROR_JIT_STACKLIMIT is given when there is not enough stack. Three func- 5296 tions are provided for managing blocks of memory for use as JIT stacks. 5297 There is further discussion about the use of JIT stacks in the section 5298 entitled "JIT stack FAQ" below. 5299 5300 The pcre2_jit_stack_create() function creates a JIT stack. Its argu- 5301 ments are a starting size, a maximum size, and a general context (for 5302 memory allocation functions, or NULL for standard memory allocation). 5303 It returns a pointer to an opaque structure of type pcre2_jit_stack, or 5304 NULL if there is an error. The pcre2_jit_stack_free() function is used 5305 to free a stack that is no longer needed. If its argument is NULL, this 5306 function returns immediately, without doing anything. (For the techni- 5307 cally minded: the address space is allocated by mmap or VirtualAlloc.) 5308 A maximum stack size of 512KiB to 1MiB should be more than enough for 5309 any pattern. 5310 5311 The pcre2_jit_stack_assign() function specifies which stack JIT code 5312 should use. Its arguments are as follows: 5313 5314 pcre2_match_context *mcontext 5315 pcre2_jit_callback callback 5316 void *data 5317 5318 The first argument is a pointer to a match context. When this is subse- 5319 quently passed to a matching function, its information determines which 5320 JIT stack is used. If this argument is NULL, the function returns imme- 5321 diately, without doing anything. There are three cases for the values 5322 of the other two options: 5323 5324 (1) If callback is NULL and data is NULL, an internal 32KiB block 5325 on the machine stack is used. This is the default when a match 5326 context is created. 5327 5328 (2) If callback is NULL and data is not NULL, data must be 5329 a pointer to a valid JIT stack, the result of calling 5330 pcre2_jit_stack_create(). 5331 5332 (3) If callback is not NULL, it must point to a function that is 5333 called with data as an argument at the start of matching, in 5334 order to set up a JIT stack. If the return from the callback 5335 function is NULL, the internal 32KiB stack is used; otherwise the 5336 return value must be a valid JIT stack, the result of calling 5337 pcre2_jit_stack_create(). 5338 5339 A callback function is obeyed whenever JIT code is about to be run; it 5340 is not obeyed when pcre2_match() is called with options that are incom- 5341 patible for JIT matching. A callback function can therefore be used to 5342 determine whether a match operation was executed by JIT or by the in- 5343 terpreter. 5344 5345 You may safely use the same JIT stack for more than one pattern (either 5346 by assigning directly or by callback), as long as the patterns are 5347 matched sequentially in the same thread. Currently, the only way to set 5348 up non-sequential matches in one thread is to use callouts: if a call- 5349 out function starts another match, that match must use a different JIT 5350 stack to the one used for currently suspended match(es). 5351 5352 In a multithread application, if you do not specify a JIT stack, or if 5353 you assign or pass back NULL from a callback, that is thread-safe, be- 5354 cause each thread has its own machine stack. However, if you assign or 5355 pass back a non-NULL JIT stack, this must be a different stack for each 5356 thread so that the application is thread-safe. 5357 5358 Strictly speaking, even more is allowed. You can assign the same non- 5359 NULL stack to a match context that is used by any number of patterns, 5360 as long as they are not used for matching by multiple threads at the 5361 same time. For example, you could use the same stack in all compiled 5362 patterns, with a global mutex in the callback to wait until the stack 5363 is available for use. However, this is an inefficient solution, and not 5364 recommended. 5365 5366 This is a suggestion for how a multithreaded program that needs to set 5367 up non-default JIT stacks might operate: 5368 5369 During thread initialization 5370 thread_local_var = pcre2_jit_stack_create(...) 5371 5372 During thread exit 5373 pcre2_jit_stack_free(thread_local_var) 5374 5375 Use a one-line callback function 5376 return thread_local_var 5377 5378 All the functions described in this section do nothing if JIT is not 5379 available. 5380 5381 5382JIT STACK FAQ 5383 5384 (1) Why do we need JIT stacks? 5385 5386 PCRE2 (and JIT) is a recursive, depth-first engine, so it needs a stack 5387 where the local data of the current node is pushed before checking its 5388 child nodes. Allocating real machine stack on some platforms is diffi- 5389 cult. For example, the stack chain needs to be updated every time if we 5390 extend the stack on PowerPC. Although it is possible, its updating 5391 time overhead decreases performance. So we do the recursion in memory. 5392 5393 (2) Why don't we simply allocate blocks of memory with malloc()? 5394 5395 Modern operating systems have a nice feature: they can reserve an ad- 5396 dress space instead of allocating memory. We can safely allocate memory 5397 pages inside this address space, so the stack could grow without moving 5398 memory data (this is important because of pointers). Thus we can allo- 5399 cate 1MiB address space, and use only a single memory page (usually 5400 4KiB) if that is enough. However, we can still grow up to 1MiB anytime 5401 if needed. 5402 5403 (3) Who "owns" a JIT stack? 5404 5405 The owner of the stack is the user program, not the JIT studied pattern 5406 or anything else. The user program must ensure that if a stack is being 5407 used by pcre2_match(), (that is, it is assigned to a match context that 5408 is passed to the pattern currently running), that stack must not be 5409 used by any other threads (to avoid overwriting the same memory area). 5410 The best practice for multithreaded programs is to allocate a stack for 5411 each thread, and return this stack through the JIT callback function. 5412 5413 (4) When should a JIT stack be freed? 5414 5415 You can free a JIT stack at any time, as long as it will not be used by 5416 pcre2_match() again. When you assign the stack to a match context, only 5417 a pointer is set. There is no reference counting or any other magic. 5418 You can free compiled patterns, contexts, and stacks in any order, any- 5419 time. Just do not call pcre2_match() with a match context pointing to 5420 an already freed stack, as that will cause SEGFAULT. (Also, do not free 5421 a stack currently used by pcre2_match() in another thread). You can 5422 also replace the stack in a context at any time when it is not in use. 5423 You should free the previous stack before assigning a replacement. 5424 5425 (5) Should I allocate/free a stack every time before/after calling 5426 pcre2_match()? 5427 5428 No, because this is too costly in terms of resources. However, you 5429 could implement some clever idea which release the stack if it is not 5430 used in let's say two minutes. The JIT callback can help to achieve 5431 this without keeping a list of patterns. 5432 5433 (6) OK, the stack is for long term memory allocation. But what happens 5434 if a pattern causes stack overflow with a stack of 1MiB? Is that 1MiB 5435 kept until the stack is freed? 5436 5437 Especially on embedded sytems, it might be a good idea to release mem- 5438 ory sometimes without freeing the stack. There is no API for this at 5439 the moment. Probably a function call which returns with the currently 5440 allocated memory for any stack and another which allows releasing mem- 5441 ory (shrinking the stack) would be a good idea if someone needs this. 5442 5443 (7) This is too much of a headache. Isn't there any better solution for 5444 JIT stack handling? 5445 5446 No, thanks to Windows. If POSIX threads were used everywhere, we could 5447 throw out this complicated API. 5448 5449 5450FREEING JIT SPECULATIVE MEMORY 5451 5452 void pcre2_jit_free_unused_memory(pcre2_general_context *gcontext); 5453 5454 The JIT executable allocator does not free all memory when it is possi- 5455 ble. It expects new allocations, and keeps some free memory around to 5456 improve allocation speed. However, in low memory conditions, it might 5457 be better to free all possible memory. You can cause this to happen by 5458 calling pcre2_jit_free_unused_memory(). Its argument is a general con- 5459 text, for custom memory management, or NULL for standard memory manage- 5460 ment. 5461 5462 5463EXAMPLE CODE 5464 5465 This is a single-threaded example that specifies a JIT stack without 5466 using a callback. A real program should include error checking after 5467 all the function calls. 5468 5469 int rc; 5470 pcre2_code *re; 5471 pcre2_match_data *match_data; 5472 pcre2_match_context *mcontext; 5473 pcre2_jit_stack *jit_stack; 5474 5475 re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0, 5476 &errornumber, &erroffset, NULL); 5477 rc = pcre2_jit_compile(re, PCRE2_JIT_COMPLETE); 5478 mcontext = pcre2_match_context_create(NULL); 5479 jit_stack = pcre2_jit_stack_create(32*1024, 512*1024, NULL); 5480 pcre2_jit_stack_assign(mcontext, NULL, jit_stack); 5481 match_data = pcre2_match_data_create(re, 10); 5482 rc = pcre2_match(re, subject, length, 0, 0, match_data, mcontext); 5483 /* Process result */ 5484 5485 pcre2_code_free(re); 5486 pcre2_match_data_free(match_data); 5487 pcre2_match_context_free(mcontext); 5488 pcre2_jit_stack_free(jit_stack); 5489 5490 5491JIT FAST PATH API 5492 5493 Because the API described above falls back to interpreted matching when 5494 JIT is not available, it is convenient for programs that are written 5495 for general use in many environments. However, calling JIT via 5496 pcre2_match() does have a performance impact. Programs that are written 5497 for use where JIT is known to be available, and which need the best 5498 possible performance, can instead use a "fast path" API to call JIT 5499 matching directly instead of calling pcre2_match() (obviously only for 5500 patterns that have been successfully processed by pcre2_jit_compile()). 5501 5502 The fast path function is called pcre2_jit_match(), and it takes ex- 5503 actly the same arguments as pcre2_match(). However, the subject string 5504 must be specified with a length; PCRE2_ZERO_TERMINATED is not sup- 5505 ported. Unsupported option bits (for example, PCRE2_ANCHORED, PCRE2_EN- 5506 DANCHORED and PCRE2_COPY_MATCHED_SUBJECT) are ignored, as is the 5507 PCRE2_NO_JIT option. The return values are also the same as for 5508 pcre2_match(), plus PCRE2_ERROR_JIT_BADOPTION if a matching mode (par- 5509 tial or complete) is requested that was not compiled. 5510 5511 When you call pcre2_match(), as well as testing for invalid options, a 5512 number of other sanity checks are performed on the arguments. For exam- 5513 ple, if the subject pointer is NULL but the length is non-zero, an im- 5514 mediate error is given. Also, unless PCRE2_NO_UTF_CHECK is set, a UTF 5515 subject string is tested for validity. In the interests of speed, these 5516 checks do not happen on the JIT fast path, and if invalid data is 5517 passed, the result is undefined. 5518 5519 Bypassing the sanity checks and the pcre2_match() wrapping can give 5520 speedups of more than 10%. 5521 5522 5523SEE ALSO 5524 5525 pcre2api(3) 5526 5527 5528AUTHOR 5529 5530 Philip Hazel (FAQ by Zoltan Herczeg) 5531 University Computing Service 5532 Cambridge, England. 5533 5534 5535REVISION 5536 5537 Last updated: 30 November 2021 5538 Copyright (c) 1997-2021 University of Cambridge. 5539------------------------------------------------------------------------------ 5540 5541 5542PCRE2LIMITS(3) Library Functions Manual PCRE2LIMITS(3) 5543 5544 5545 5546NAME 5547 PCRE2 - Perl-compatible regular expressions (revised API) 5548 5549SIZE AND OTHER LIMITATIONS 5550 5551 There are some size limitations in PCRE2 but it is hoped that they will 5552 never in practice be relevant. 5553 5554 The maximum size of a compiled pattern is approximately 64 thousand 5555 code units for the 8-bit and 16-bit libraries if PCRE2 is compiled with 5556 the default internal linkage size, which is 2 bytes for these li- 5557 braries. If you want to process regular expressions that are truly 5558 enormous, you can compile PCRE2 with an internal linkage size of 3 or 4 5559 (when building the 16-bit library, 3 is rounded up to 4). See the 5560 README file in the source distribution and the pcre2build documentation 5561 for details. In these cases the limit is substantially larger. How- 5562 ever, the speed of execution is slower. In the 32-bit library, the in- 5563 ternal linkage size is always 4. 5564 5565 The maximum length of a source pattern string is essentially unlimited; 5566 it is the largest number a PCRE2_SIZE variable can hold. However, the 5567 program that calls pcre2_compile() can specify a smaller limit. 5568 5569 The maximum length (in code units) of a subject string is one less than 5570 the largest number a PCRE2_SIZE variable can hold. PCRE2_SIZE is an un- 5571 signed integer type, usually defined as size_t. Its maximum value (that 5572 is ~(PCRE2_SIZE)0) is reserved as a special indicator for zero-termi- 5573 nated strings and unset offsets. 5574 5575 All values in repeating quantifiers must be less than 65536. 5576 5577 The maximum length of a lookbehind assertion is 65535 characters. 5578 5579 There is no limit to the number of parenthesized groups, but there can 5580 be no more than 65535 capture groups, and there is a limit to the depth 5581 of nesting of parenthesized subpatterns of all kinds. This is imposed 5582 in order to limit the amount of system stack used at compile time. The 5583 default limit can be specified when PCRE2 is built; if not, the default 5584 is set to 250. An application can change this limit by calling 5585 pcre2_set_parens_nest_limit() to set the limit in a compile context. 5586 5587 The maximum length of name for a named capture group is 32 code units, 5588 and the maximum number of such groups is 10000. 5589 5590 The maximum length of a name in a (*MARK), (*PRUNE), (*SKIP), or 5591 (*THEN) verb is 255 code units for the 8-bit library and 65535 code 5592 units for the 16-bit and 32-bit libraries. 5593 5594 The maximum length of a string argument to a callout is the largest 5595 number a 32-bit unsigned integer can hold. 5596 5597 5598AUTHOR 5599 5600 Philip Hazel 5601 University Computing Service 5602 Cambridge, England. 5603 5604 5605REVISION 5606 5607 Last updated: 02 February 2019 5608 Copyright (c) 1997-2019 University of Cambridge. 5609------------------------------------------------------------------------------ 5610 5611 5612PCRE2MATCHING(3) Library Functions Manual PCRE2MATCHING(3) 5613 5614 5615 5616NAME 5617 PCRE2 - Perl-compatible regular expressions (revised API) 5618 5619PCRE2 MATCHING ALGORITHMS 5620 5621 This document describes the two different algorithms that are available 5622 in PCRE2 for matching a compiled regular expression against a given 5623 subject string. The "standard" algorithm is the one provided by the 5624 pcre2_match() function. This works in the same as as Perl's matching 5625 function, and provide a Perl-compatible matching operation. The just- 5626 in-time (JIT) optimization that is described in the pcre2jit documenta- 5627 tion is compatible with this function. 5628 5629 An alternative algorithm is provided by the pcre2_dfa_match() function; 5630 it operates in a different way, and is not Perl-compatible. This alter- 5631 native has advantages and disadvantages compared with the standard al- 5632 gorithm, and these are described below. 5633 5634 When there is only one possible way in which a given subject string can 5635 match a pattern, the two algorithms give the same answer. A difference 5636 arises, however, when there are multiple possibilities. For example, if 5637 the pattern 5638 5639 ^<.*> 5640 5641 is matched against the string 5642 5643 <something> <something else> <something further> 5644 5645 there are three possible answers. The standard algorithm finds only one 5646 of them, whereas the alternative algorithm finds all three. 5647 5648 5649REGULAR EXPRESSIONS AS TREES 5650 5651 The set of strings that are matched by a regular expression can be rep- 5652 resented as a tree structure. An unlimited repetition in the pattern 5653 makes the tree of infinite size, but it is still a tree. Matching the 5654 pattern to a given subject string (from a given starting point) can be 5655 thought of as a search of the tree. There are two ways to search a 5656 tree: depth-first and breadth-first, and these correspond to the two 5657 matching algorithms provided by PCRE2. 5658 5659 5660THE STANDARD MATCHING ALGORITHM 5661 5662 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres- 5663 sions", the standard algorithm is an "NFA algorithm". It conducts a 5664 depth-first search of the pattern tree. That is, it proceeds along a 5665 single path through the tree, checking that the subject matches what is 5666 required. When there is a mismatch, the algorithm tries any alterna- 5667 tives at the current point, and if they all fail, it backs up to the 5668 previous branch point in the tree, and tries the next alternative 5669 branch at that level. This often involves backing up (moving to the 5670 left) in the subject string as well. The order in which repetition 5671 branches are tried is controlled by the greedy or ungreedy nature of 5672 the quantifier. 5673 5674 If a leaf node is reached, a matching string has been found, and at 5675 that point the algorithm stops. Thus, if there is more than one possi- 5676 ble match, this algorithm returns the first one that it finds. Whether 5677 this is the shortest, the longest, or some intermediate length depends 5678 on the way the alternations and the greedy or ungreedy repetition quan- 5679 tifiers are specified in the pattern. 5680 5681 Because it ends up with a single path through the tree, it is rela- 5682 tively straightforward for this algorithm to keep track of the sub- 5683 strings that are matched by portions of the pattern in parentheses. 5684 This provides support for capturing parentheses and backreferences. 5685 5686 5687THE ALTERNATIVE MATCHING ALGORITHM 5688 5689 This algorithm conducts a breadth-first search of the tree. Starting 5690 from the first matching point in the subject, it scans the subject 5691 string from left to right, once, character by character, and as it does 5692 this, it remembers all the paths through the tree that represent valid 5693 matches. In Friedl's terminology, this is a kind of "DFA algorithm", 5694 though it is not implemented as a traditional finite state machine (it 5695 keeps multiple states active simultaneously). 5696 5697 Although the general principle of this matching algorithm is that it 5698 scans the subject string only once, without backtracking, there is one 5699 exception: when a lookaround assertion is encountered, the characters 5700 following or preceding the current point have to be independently in- 5701 spected. 5702 5703 The scan continues until either the end of the subject is reached, or 5704 there are no more unterminated paths. At this point, terminated paths 5705 represent the different matching possibilities (if there are none, the 5706 match has failed). Thus, if there is more than one possible match, 5707 this algorithm finds all of them, and in particular, it finds the long- 5708 est. The matches are returned in the output vector in decreasing order 5709 of length. There is an option to stop the algorithm after the first 5710 match (which is necessarily the shortest) is found. 5711 5712 Note that the size of vector needed to contain all the results depends 5713 on the number of simultaneous matches, not on the number of parentheses 5714 in the pattern. Using pcre2_match_data_create_from_pattern() to create 5715 the match data block is therefore not advisable when doing DFA match- 5716 ing. 5717 5718 Note also that all the matches that are found start at the same point 5719 in the subject. If the pattern 5720 5721 cat(er(pillar)?)? 5722 5723 is matched against the string "the caterpillar catchment", the result 5724 is the three strings "caterpillar", "cater", and "cat" that start at 5725 the fifth character of the subject. The algorithm does not automati- 5726 cally move on to find matches that start at later positions. 5727 5728 PCRE2's "auto-possessification" optimization usually applies to charac- 5729 ter repeats at the end of a pattern (as well as internally). For exam- 5730 ple, the pattern "a\d+" is compiled as if it were "a\d++" because there 5731 is no point even considering the possibility of backtracking into the 5732 repeated digits. For DFA matching, this means that only one possible 5733 match is found. If you really do want multiple matches in such cases, 5734 either use an ungreedy repeat ("a\d+?") or set the PCRE2_NO_AUTO_POS- 5735 SESS option when compiling. 5736 5737 There are a number of features of PCRE2 regular expressions that are 5738 not supported or behave differently in the alternative matching func- 5739 tion. Those that are not supported cause an error if encountered. 5740 5741 1. Because the algorithm finds all possible matches, the greedy or un- 5742 greedy nature of repetition quantifiers is not relevant (though it may 5743 affect auto-possessification, as just described). During matching, 5744 greedy and ungreedy quantifiers are treated in exactly the same way. 5745 However, possessive quantifiers can make a difference when what follows 5746 could also match what is quantified, for example in a pattern like 5747 this: 5748 5749 ^a++\w! 5750 5751 This pattern matches "aaab!" but not "aaa!", which would be matched by 5752 a non-possessive quantifier. Similarly, if an atomic group is present, 5753 it is matched as if it were a standalone pattern at the current point, 5754 and the longest match is then "locked in" for the rest of the overall 5755 pattern. 5756 5757 2. When dealing with multiple paths through the tree simultaneously, it 5758 is not straightforward to keep track of captured substrings for the 5759 different matching possibilities, and PCRE2's implementation of this 5760 algorithm does not attempt to do this. This means that no captured sub- 5761 strings are available. 5762 5763 3. Because no substrings are captured, backreferences within the pat- 5764 tern are not supported. 5765 5766 4. For the same reason, conditional expressions that use a backrefer- 5767 ence as the condition or test for a specific group recursion are not 5768 supported. 5769 5770 5. Again for the same reason, script runs are not supported. 5771 5772 6. Because many paths through the tree may be active, the \K escape se- 5773 quence, which resets the start of the match when encountered (but may 5774 be on some paths and not on others), is not supported. 5775 5776 7. Callouts are supported, but the value of the capture_top field is 5777 always 1, and the value of the capture_last field is always 0. 5778 5779 8. The \C escape sequence, which (in the standard algorithm) always 5780 matches a single code unit, even in a UTF mode, is not supported in 5781 these modes, because the alternative algorithm moves through the sub- 5782 ject string one character (not code unit) at a time, for all active 5783 paths through the tree. 5784 5785 9. Except for (*FAIL), the backtracking control verbs such as (*PRUNE) 5786 are not supported. (*FAIL) is supported, and behaves like a failing 5787 negative assertion. 5788 5789 10. The PCRE2_MATCH_INVALID_UTF option for pcre2_compile() is not sup- 5790 ported by pcre2_dfa_match(). 5791 5792 5793ADVANTAGES OF THE ALTERNATIVE ALGORITHM 5794 5795 The main advantage of the alternative algorithm is that all possible 5796 matches (at a single point in the subject) are automatically found, and 5797 in particular, the longest match is found. To find more than one match 5798 at the same point using the standard algorithm, you have to do kludgy 5799 things with callouts. 5800 5801 Partial matching is possible with this algorithm, though it has some 5802 limitations. The pcre2partial documentation gives details of partial 5803 matching and discusses multi-segment matching. 5804 5805 5806DISADVANTAGES OF THE ALTERNATIVE ALGORITHM 5807 5808 The alternative algorithm suffers from a number of disadvantages: 5809 5810 1. It is substantially slower than the standard algorithm. This is 5811 partly because it has to search for all possible matches, but is also 5812 because it is less susceptible to optimization. 5813 5814 2. Capturing parentheses, backreferences, script runs, and matching 5815 within invalid UTF string are not supported. 5816 5817 3. Although atomic groups are supported, their use does not provide the 5818 performance advantage that it does for the standard algorithm. 5819 5820 4. JIT optimization is not supported. 5821 5822 5823AUTHOR 5824 5825 Philip Hazel 5826 Retired from University Computing Service 5827 Cambridge, England. 5828 5829 5830REVISION 5831 5832 Last updated: 28 August 2021 5833 Copyright (c) 1997-2021 University of Cambridge. 5834------------------------------------------------------------------------------ 5835 5836 5837PCRE2PARTIAL(3) Library Functions Manual PCRE2PARTIAL(3) 5838 5839 5840 5841NAME 5842 PCRE2 - Perl-compatible regular expressions 5843 5844PARTIAL MATCHING IN PCRE2 5845 5846 In normal use of PCRE2, if there is a match up to the end of a subject 5847 string, but more characters are needed to match the entire pattern, 5848 PCRE2_ERROR_NOMATCH is returned, just like any other failing match. 5849 There are circumstances where it might be helpful to distinguish this 5850 "partial match" case. 5851 5852 One example is an application where the subject string is very long, 5853 and not all available at once. The requirement here is to be able to do 5854 the matching segment by segment, but special action is needed when a 5855 matched substring spans the boundary between two segments. 5856 5857 Another example is checking a user input string as it is typed, to en- 5858 sure that it conforms to a required format. Invalid characters can be 5859 immediately diagnosed and rejected, giving instant feedback. 5860 5861 Partial matching is a PCRE2-specific feature; it is not Perl-compati- 5862 ble. It is requested by setting one of the PCRE2_PARTIAL_HARD or 5863 PCRE2_PARTIAL_SOFT options when calling a matching function. The dif- 5864 ference between the two options is whether or not a partial match is 5865 preferred to an alternative complete match, though the details differ 5866 between the two types of matching function. If both options are set, 5867 PCRE2_PARTIAL_HARD takes precedence. 5868 5869 If you want to use partial matching with just-in-time optimized code, 5870 as well as setting a partial match option for the matching function, 5871 you must also call pcre2_jit_compile() with one or both of these op- 5872 tions: 5873 5874 PCRE2_JIT_PARTIAL_HARD 5875 PCRE2_JIT_PARTIAL_SOFT 5876 5877 PCRE2_JIT_COMPLETE should also be set if you are going to run non-par- 5878 tial matches on the same pattern. Separate code is compiled for each 5879 mode. If the appropriate JIT mode has not been compiled, interpretive 5880 matching code is used. 5881 5882 Setting a partial matching option disables two of PCRE2's standard op- 5883 timization hints. PCRE2 remembers the last literal code unit in a pat- 5884 tern, and abandons matching immediately if it is not present in the 5885 subject string. This optimization cannot be used for a subject string 5886 that might match only partially. PCRE2 also remembers a minimum length 5887 of a matching string, and does not bother to run the matching function 5888 on shorter strings. This optimization is also disabled for partial 5889 matching. 5890 5891 5892REQUIREMENTS FOR A PARTIAL MATCH 5893 5894 A possible partial match occurs during matching when the end of the 5895 subject string is reached successfully, but either more characters are 5896 needed to complete the match, or the addition of more characters might 5897 change what is matched. 5898 5899 Example 1: if the pattern is /abc/ and the subject is "ab", more char- 5900 acters are definitely needed to complete a match. In this case both 5901 hard and soft matching options yield a partial match. 5902 5903 Example 2: if the pattern is /ab+/ and the subject is "ab", a complete 5904 match can be found, but the addition of more characters might change 5905 what is matched. In this case, only PCRE2_PARTIAL_HARD returns a par- 5906 tial match; PCRE2_PARTIAL_SOFT returns the complete match. 5907 5908 On reaching the end of the subject, when PCRE2_PARTIAL_HARD is set, if 5909 the next pattern item is \z, \Z, \b, \B, or $ there is always a partial 5910 match. Otherwise, for both options, the next pattern item must be one 5911 that inspects a character, and at least one of the following must be 5912 true: 5913 5914 (1) At least one character has already been inspected. An inspected 5915 character need not form part of the final matched string; lookbehind 5916 assertions and the \K escape sequence provide ways of inspecting char- 5917 acters before the start of a matched string. 5918 5919 (2) The pattern contains one or more lookbehind assertions. This condi- 5920 tion exists in case there is a lookbehind that inspects characters be- 5921 fore the start of the match. 5922 5923 (3) There is a special case when the whole pattern can match an empty 5924 string. When the starting point is at the end of the subject, the 5925 empty string match is a possibility, and if PCRE2_PARTIAL_SOFT is set 5926 and neither of the above conditions is true, it is returned. However, 5927 because adding more characters might result in a non-empty match, 5928 PCRE2_PARTIAL_HARD returns a partial match, which in this case means 5929 "there is going to be a match at this point, but until some more char- 5930 acters are added, we do not know if it will be an empty string or some- 5931 thing longer". 5932 5933 5934PARTIAL MATCHING USING pcre2_match() 5935 5936 When a partial matching option is set, the result of calling 5937 pcre2_match() can be one of the following: 5938 5939 A successful match 5940 A complete match has been found, starting and ending within this sub- 5941 ject. 5942 5943 PCRE2_ERROR_NOMATCH 5944 No match can start anywhere in this subject. 5945 5946 PCRE2_ERROR_PARTIAL 5947 Adding more characters may result in a complete match that uses one 5948 or more characters from the end of this subject. 5949 5950 When a partial match is returned, the first two elements in the ovector 5951 point to the portion of the subject that was matched, but the values in 5952 the rest of the ovector are undefined. The appearance of \K in the pat- 5953 tern has no effect for a partial match. Consider this pattern: 5954 5955 /abc\K123/ 5956 5957 If it is matched against "456abc123xyz" the result is a complete match, 5958 and the ovector defines the matched string as "123", because \K resets 5959 the "start of match" point. However, if a partial match is requested 5960 and the subject string is "456abc12", a partial match is found for the 5961 string "abc12", because all these characters are needed for a subse- 5962 quent re-match with additional characters. 5963 5964 If there is more than one partial match, the first one that was found 5965 provides the data that is returned. Consider this pattern: 5966 5967 /123\w+X|dogY/ 5968 5969 If this is matched against the subject string "abc123dog", both alter- 5970 natives fail to match, but the end of the subject is reached during 5971 matching, so PCRE2_ERROR_PARTIAL is returned. The offsets are set to 3 5972 and 9, identifying "123dog" as the first partial match. (In this exam- 5973 ple, there are two partial matches, because "dog" on its own partially 5974 matches the second alternative.) 5975 5976 How a partial match is processed by pcre2_match() 5977 5978 What happens when a partial match is identified depends on which of the 5979 two partial matching options is set. 5980 5981 If PCRE2_PARTIAL_HARD is set, PCRE2_ERROR_PARTIAL is returned as soon 5982 as a partial match is found, without continuing to search for possible 5983 complete matches. This option is "hard" because it prefers an earlier 5984 partial match over a later complete match. For this reason, the assump- 5985 tion is made that the end of the supplied subject string is not the 5986 true end of the available data, which is why \z, \Z, \b, \B, and $ al- 5987 ways give a partial match. 5988 5989 If PCRE2_PARTIAL_SOFT is set, the partial match is remembered, but 5990 matching continues as normal, and other alternatives in the pattern are 5991 tried. If no complete match can be found, PCRE2_ERROR_PARTIAL is re- 5992 turned instead of PCRE2_ERROR_NOMATCH. This option is "soft" because it 5993 prefers a complete match over a partial match. All the various matching 5994 items in a pattern behave as if the subject string is potentially com- 5995 plete; \z, \Z, and $ match at the end of the subject, as normal, and 5996 for \b and \B the end of the subject is treated as a non-alphanumeric. 5997 5998 The difference between the two partial matching options can be illus- 5999 trated by a pattern such as: 6000 6001 /dog(sbody)?/ 6002 6003 This matches either "dog" or "dogsbody", greedily (that is, it prefers 6004 the longer string if possible). If it is matched against the string 6005 "dog" with PCRE2_PARTIAL_SOFT, it yields a complete match for "dog". 6006 However, if PCRE2_PARTIAL_HARD is set, the result is PCRE2_ERROR_PAR- 6007 TIAL. On the other hand, if the pattern is made ungreedy the result is 6008 different: 6009 6010 /dog(sbody)??/ 6011 6012 In this case the result is always a complete match because that is 6013 found first, and matching never continues after finding a complete 6014 match. It might be easier to follow this explanation by thinking of the 6015 two patterns like this: 6016 6017 /dog(sbody)?/ is the same as /dogsbody|dog/ 6018 /dog(sbody)??/ is the same as /dog|dogsbody/ 6019 6020 The second pattern will never match "dogsbody", because it will always 6021 find the shorter match first. 6022 6023 Example of partial matching using pcre2test 6024 6025 The pcre2test data modifiers partial_hard (or ph) and partial_soft (or 6026 ps) set PCRE2_PARTIAL_HARD and PCRE2_PARTIAL_SOFT, respectively, when 6027 calling pcre2_match(). Here is a run of pcre2test using a pattern that 6028 matches the whole subject in the form of a date: 6029 6030 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/ 6031 data> 25dec3\=ph 6032 Partial match: 23dec3 6033 data> 3ju\=ph 6034 Partial match: 3ju 6035 data> 3juj\=ph 6036 No match 6037 6038 This example gives the same results for both hard and soft partial 6039 matching options. Here is an example where there is a difference: 6040 6041 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/ 6042 data> 25jun04\=ps 6043 0: 25jun04 6044 1: jun 6045 data> 25jun04\=ph 6046 Partial match: 25jun04 6047 6048 With PCRE2_PARTIAL_SOFT, the subject is matched completely. For 6049 PCRE2_PARTIAL_HARD, however, the subject is assumed not to be complete, 6050 so there is only a partial match. 6051 6052 6053MULTI-SEGMENT MATCHING WITH pcre2_match() 6054 6055 PCRE was not originally designed with multi-segment matching in mind. 6056 However, over time, features (including partial matching) that make 6057 multi-segment matching possible have been added. A very long string can 6058 be searched segment by segment by calling pcre2_match() repeatedly, 6059 with the aim of achieving the same results that would happen if the en- 6060 tire string was available for searching all the time. Normally, the 6061 strings that are being sought are much shorter than each individual 6062 segment, and are in the middle of very long strings, so the pattern is 6063 normally not anchored. 6064 6065 Special logic must be implemented to handle a matched substring that 6066 spans a segment boundary. PCRE2_PARTIAL_HARD should be used, because it 6067 returns a partial match at the end of a segment whenever there is the 6068 possibility of changing the match by adding more characters. The 6069 PCRE2_NOTBOL option should also be set for all but the first segment. 6070 6071 When a partial match occurs, the next segment must be added to the cur- 6072 rent subject and the match re-run, using the startoffset argument of 6073 pcre2_match() to begin at the point where the partial match started. 6074 For example: 6075 6076 re> /\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d/ 6077 data> ...the date is 23ja\=ph 6078 Partial match: 23ja 6079 data> ...the date is 23jan19 and on that day...\=offset=15 6080 0: 23jan19 6081 1: jan 6082 6083 Note the use of the offset modifier to start the new match where the 6084 partial match was found. In this example, the next segment was added to 6085 the one in which the partial match was found. This is the most 6086 straightforward approach, typically using a memory buffer that is twice 6087 the size of each segment. After a partial match, the first half of the 6088 buffer is discarded, the second half is moved to the start of the buf- 6089 fer, and a new segment is added before repeating the match as in the 6090 example above. After a no match, the entire buffer can be discarded. 6091 6092 If there are memory constraints, you may want to discard text that pre- 6093 cedes a partial match before adding the next segment. Unfortunately, 6094 this is not at present straightforward. In cases such as the above, 6095 where the pattern does not contain any lookbehinds, it is sufficient to 6096 retain only the partially matched substring. However, if the pattern 6097 contains a lookbehind assertion, characters that precede the start of 6098 the partial match may have been inspected during the matching process. 6099 When pcre2test displays a partial match, it indicates these characters 6100 with '<' if the allusedtext modifier is set: 6101 6102 re> "(?<=123)abc" 6103 data> xx123ab\=ph,allusedtext 6104 Partial match: 123ab 6105 <<< 6106 6107 However, the allusedtext modifier is not available for JIT matching, 6108 because JIT matching does not record the first (or last) consulted 6109 characters. For this reason, this information is not available via the 6110 API. It is therefore not possible in general to obtain the exact number 6111 of characters that must be retained in order to get the right match re- 6112 sult. If you cannot retain the entire segment, you must find some 6113 heuristic way of choosing. 6114 6115 If you know the approximate length of the matching substrings, you can 6116 use that to decide how much text to retain. The only lookbehind infor- 6117 mation that is currently available via the API is the length of the 6118 longest individual lookbehind in a pattern, but this can be misleading 6119 if there are nested lookbehinds. The value returned by calling 6120 pcre2_pattern_info() with the PCRE2_INFO_MAXLOOKBEHIND option is the 6121 maximum number of characters (not code units) that any individual look- 6122 behind moves back when it is processed. A pattern such as 6123 "(?<=(?<!b)a)" has a maximum lookbehind value of one, but inspects two 6124 characters before its starting point. 6125 6126 In a non-UTF or a 32-bit case, moving back is just a subtraction, but 6127 in UTF-8 or UTF-16 you have to count characters while moving back 6128 through the code units. 6129 6130 6131PARTIAL MATCHING USING pcre2_dfa_match() 6132 6133 The DFA function moves along the subject string character by character, 6134 without backtracking, searching for all possible matches simultane- 6135 ously. If the end of the subject is reached before the end of the pat- 6136 tern, there is the possibility of a partial match. 6137 6138 When PCRE2_PARTIAL_SOFT is set, PCRE2_ERROR_PARTIAL is returned only if 6139 there have been no complete matches. Otherwise, the complete matches 6140 are returned. If PCRE2_PARTIAL_HARD is set, a partial match takes 6141 precedence over any complete matches. The portion of the string that 6142 was matched when the longest partial match was found is set as the 6143 first matching string. 6144 6145 Because the DFA function always searches for all possible matches, and 6146 there is no difference between greedy and ungreedy repetition, its be- 6147 haviour is different from the pcre2_match(). Consider the string "dog" 6148 matched against this ungreedy pattern: 6149 6150 /dog(sbody)??/ 6151 6152 Whereas the standard function stops as soon as it finds the complete 6153 match for "dog", the DFA function also finds the partial match for 6154 "dogsbody", and so returns that when PCRE2_PARTIAL_HARD is set. 6155 6156 6157MULTI-SEGMENT MATCHING WITH pcre2_dfa_match() 6158 6159 When a partial match has been found using the DFA matching function, it 6160 is possible to continue the match by providing additional subject data 6161 and calling the function again with the same compiled regular expres- 6162 sion, this time setting the PCRE2_DFA_RESTART option. You must pass the 6163 same working space as before, because this is where details of the pre- 6164 vious partial match are stored. You can set the PCRE2_PARTIAL_SOFT or 6165 PCRE2_PARTIAL_HARD options with PCRE2_DFA_RESTART to continue partial 6166 matching over multiple segments. Here is an example using pcre2test: 6167 6168 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/ 6169 data> 23ja\=dfa,ps 6170 Partial match: 23ja 6171 data> n05\=dfa,dfa_restart 6172 0: n05 6173 6174 The first call has "23ja" as the subject, and requests partial match- 6175 ing; the second call has "n05" as the subject for the continued 6176 (restarted) match. Notice that when the match is complete, only the 6177 last part is shown; PCRE2 does not retain the previously partially- 6178 matched string. It is up to the calling program to do that if it needs 6179 to. This means that, for an unanchored pattern, if a continued match 6180 fails, it is not possible to try again at a new starting point. All 6181 this facility is capable of doing is continuing with the previous match 6182 attempt. For example, consider this pattern: 6183 6184 1234|3789 6185 6186 If the first part of the subject is "ABC123", a partial match of the 6187 first alternative is found at offset 3. There is no partial match for 6188 the second alternative, because such a match does not start at the same 6189 point in the subject string. Attempting to continue with the string 6190 "7890" does not yield a match because only those alternatives that 6191 match at one point in the subject are remembered. Depending on the ap- 6192 plication, this may or may not be what you want. 6193 6194 If you do want to allow for starting again at the next character, one 6195 way of doing it is to retain some or all of the segment and try a new 6196 complete match, as described for pcre2_match() above. Another possibil- 6197 ity is to work with two buffers. If a partial match at offset n in the 6198 first buffer is followed by "no match" when PCRE2_DFA_RESTART is used 6199 on the second buffer, you can then try a new match starting at offset 6200 n+1 in the first buffer. 6201 6202 6203AUTHOR 6204 6205 Philip Hazel 6206 University Computing Service 6207 Cambridge, England. 6208 6209 6210REVISION 6211 6212 Last updated: 04 September 2019 6213 Copyright (c) 1997-2019 University of Cambridge. 6214------------------------------------------------------------------------------ 6215 6216 6217PCRE2PATTERN(3) Library Functions Manual PCRE2PATTERN(3) 6218 6219 6220 6221NAME 6222 PCRE2 - Perl-compatible regular expressions (revised API) 6223 6224PCRE2 REGULAR EXPRESSION DETAILS 6225 6226 The syntax and semantics of the regular expressions that are supported 6227 by PCRE2 are described in detail below. There is a quick-reference syn- 6228 tax summary in the pcre2syntax page. PCRE2 tries to match Perl syntax 6229 and semantics as closely as it can. PCRE2 also supports some alterna- 6230 tive regular expression syntax (which does not conflict with the Perl 6231 syntax) in order to provide some compatibility with regular expressions 6232 in Python, .NET, and Oniguruma. 6233 6234 Perl's regular expressions are described in its own documentation, and 6235 regular expressions in general are covered in a number of books, some 6236 of which have copious examples. Jeffrey Friedl's "Mastering Regular Ex- 6237 pressions", published by O'Reilly, covers regular expressions in great 6238 detail. This description of PCRE2's regular expressions is intended as 6239 reference material. 6240 6241 This document discusses the regular expression patterns that are sup- 6242 ported by PCRE2 when its main matching function, pcre2_match(), is 6243 used. PCRE2 also has an alternative matching function, 6244 pcre2_dfa_match(), which matches using a different algorithm that is 6245 not Perl-compatible. Some of the features discussed below are not 6246 available when DFA matching is used. The advantages and disadvantages 6247 of the alternative function, and how it differs from the normal func- 6248 tion, are discussed in the pcre2matching page. 6249 6250 6251SPECIAL START-OF-PATTERN ITEMS 6252 6253 A number of options that can be passed to pcre2_compile() can also be 6254 set by special items at the start of a pattern. These are not Perl-com- 6255 patible, but are provided to make these options accessible to pattern 6256 writers who are not able to change the program that processes the pat- 6257 tern. Any number of these items may appear, but they must all be to- 6258 gether right at the start of the pattern string, and the letters must 6259 be in upper case. 6260 6261 UTF support 6262 6263 In the 8-bit and 16-bit PCRE2 libraries, characters may be coded either 6264 as single code units, or as multiple UTF-8 or UTF-16 code units. UTF-32 6265 can be specified for the 32-bit library, in which case it constrains 6266 the character values to valid Unicode code points. To process UTF 6267 strings, PCRE2 must be built to include Unicode support (which is the 6268 default). When using UTF strings you must either call the compiling 6269 function with one or both of the PCRE2_UTF or PCRE2_MATCH_INVALID_UTF 6270 options, or the pattern must start with the special sequence (*UTF), 6271 which is equivalent to setting the relevant PCRE2_UTF. How setting a 6272 UTF mode affects pattern matching is mentioned in several places below. 6273 There is also a summary of features in the pcre2unicode page. 6274 6275 Some applications that allow their users to supply patterns may wish to 6276 restrict them to non-UTF data for security reasons. If the 6277 PCRE2_NEVER_UTF option is passed to pcre2_compile(), (*UTF) is not al- 6278 lowed, and its appearance in a pattern causes an error. 6279 6280 Unicode property support 6281 6282 Another special sequence that may appear at the start of a pattern is 6283 (*UCP). This has the same effect as setting the PCRE2_UCP option: it 6284 causes sequences such as \d and \w to use Unicode properties to deter- 6285 mine character types, instead of recognizing only characters with codes 6286 less than 256 via a lookup table. If also causes upper/lower casing op- 6287 erations to use Unicode properties for characters with code points 6288 greater than 127, even when UTF is not set. 6289 6290 Some applications that allow their users to supply patterns may wish to 6291 restrict them for security reasons. If the PCRE2_NEVER_UCP option is 6292 passed to pcre2_compile(), (*UCP) is not allowed, and its appearance in 6293 a pattern causes an error. 6294 6295 Locking out empty string matching 6296 6297 Starting a pattern with (*NOTEMPTY) or (*NOTEMPTY_ATSTART) has the same 6298 effect as passing the PCRE2_NOTEMPTY or PCRE2_NOTEMPTY_ATSTART option 6299 to whichever matching function is subsequently called to match the pat- 6300 tern. These options lock out the matching of empty strings, either en- 6301 tirely, or only at the start of the subject. 6302 6303 Disabling auto-possessification 6304 6305 If a pattern starts with (*NO_AUTO_POSSESS), it has the same effect as 6306 setting the PCRE2_NO_AUTO_POSSESS option. This stops PCRE2 from making 6307 quantifiers possessive when what follows cannot match the repeated 6308 item. For example, by default a+b is treated as a++b. For more details, 6309 see the pcre2api documentation. 6310 6311 Disabling start-up optimizations 6312 6313 If a pattern starts with (*NO_START_OPT), it has the same effect as 6314 setting the PCRE2_NO_START_OPTIMIZE option. This disables several opti- 6315 mizations for quickly reaching "no match" results. For more details, 6316 see the pcre2api documentation. 6317 6318 Disabling automatic anchoring 6319 6320 If a pattern starts with (*NO_DOTSTAR_ANCHOR), it has the same effect 6321 as setting the PCRE2_NO_DOTSTAR_ANCHOR option. This disables optimiza- 6322 tions that apply to patterns whose top-level branches all start with .* 6323 (match any number of arbitrary characters). For more details, see the 6324 pcre2api documentation. 6325 6326 Disabling JIT compilation 6327 6328 If a pattern that starts with (*NO_JIT) is successfully compiled, an 6329 attempt by the application to apply the JIT optimization by calling 6330 pcre2_jit_compile() is ignored. 6331 6332 Setting match resource limits 6333 6334 The pcre2_match() function contains a counter that is incremented every 6335 time it goes round its main loop. The caller of pcre2_match() can set a 6336 limit on this counter, which therefore limits the amount of computing 6337 resource used for a match. The maximum depth of nested backtracking can 6338 also be limited; this indirectly restricts the amount of heap memory 6339 that is used, but there is also an explicit memory limit that can be 6340 set. 6341 6342 These facilities are provided to catch runaway matches that are pro- 6343 voked by patterns with huge matching trees. A common example is a pat- 6344 tern with nested unlimited repeats applied to a long string that does 6345 not match. When one of these limits is reached, pcre2_match() gives an 6346 error return. The limits can also be set by items at the start of the 6347 pattern of the form 6348 6349 (*LIMIT_HEAP=d) 6350 (*LIMIT_MATCH=d) 6351 (*LIMIT_DEPTH=d) 6352 6353 where d is any number of decimal digits. However, the value of the set- 6354 ting must be less than the value set (or defaulted) by the caller of 6355 pcre2_match() for it to have any effect. In other words, the pattern 6356 writer can lower the limits set by the programmer, but not raise them. 6357 If there is more than one setting of one of these limits, the lower 6358 value is used. The heap limit is specified in kibibytes (units of 1024 6359 bytes). 6360 6361 Prior to release 10.30, LIMIT_DEPTH was called LIMIT_RECURSION. This 6362 name is still recognized for backwards compatibility. 6363 6364 The heap limit applies only when the pcre2_match() or pcre2_dfa_match() 6365 interpreters are used for matching. It does not apply to JIT. The match 6366 limit is used (but in a different way) when JIT is being used, or when 6367 pcre2_dfa_match() is called, to limit computing resource usage by those 6368 matching functions. The depth limit is ignored by JIT but is relevant 6369 for DFA matching, which uses function recursion for recursions within 6370 the pattern and for lookaround assertions and atomic groups. In this 6371 case, the depth limit controls the depth of such recursion. 6372 6373 Newline conventions 6374 6375 PCRE2 supports six different conventions for indicating line breaks in 6376 strings: a single CR (carriage return) character, a single LF (line- 6377 feed) character, the two-character sequence CRLF, any of the three pre- 6378 ceding, any Unicode newline sequence, or the NUL character (binary 6379 zero). The pcre2api page has further discussion about newlines, and 6380 shows how to set the newline convention when calling pcre2_compile(). 6381 6382 It is also possible to specify a newline convention by starting a pat- 6383 tern string with one of the following sequences: 6384 6385 (*CR) carriage return 6386 (*LF) linefeed 6387 (*CRLF) carriage return, followed by linefeed 6388 (*ANYCRLF) any of the three above 6389 (*ANY) all Unicode newline sequences 6390 (*NUL) the NUL character (binary zero) 6391 6392 These override the default and the options given to the compiling func- 6393 tion. For example, on a Unix system where LF is the default newline se- 6394 quence, the pattern 6395 6396 (*CR)a.b 6397 6398 changes the convention to CR. That pattern matches "a\nb" because LF is 6399 no longer a newline. If more than one of these settings is present, the 6400 last one is used. 6401 6402 The newline convention affects where the circumflex and dollar asser- 6403 tions are true. It also affects the interpretation of the dot metachar- 6404 acter when PCRE2_DOTALL is not set, and the behaviour of \N when not 6405 followed by an opening brace. However, it does not affect what the \R 6406 escape sequence matches. By default, this is any Unicode newline se- 6407 quence, for Perl compatibility. However, this can be changed; see the 6408 next section and the description of \R in the section entitled "Newline 6409 sequences" below. A change of \R setting can be combined with a change 6410 of newline convention. 6411 6412 Specifying what \R matches 6413 6414 It is possible to restrict \R to match only CR, LF, or CRLF (instead of 6415 the complete set of Unicode line endings) by setting the option 6416 PCRE2_BSR_ANYCRLF at compile time. This effect can also be achieved by 6417 starting a pattern with (*BSR_ANYCRLF). For completeness, (*BSR_UNI- 6418 CODE) is also recognized, corresponding to PCRE2_BSR_UNICODE. 6419 6420 6421EBCDIC CHARACTER CODES 6422 6423 PCRE2 can be compiled to run in an environment that uses EBCDIC as its 6424 character code instead of ASCII or Unicode (typically a mainframe sys- 6425 tem). In the sections below, character code values are ASCII or Uni- 6426 code; in an EBCDIC environment these characters may have different code 6427 values, and there are no code points greater than 255. 6428 6429 6430CHARACTERS AND METACHARACTERS 6431 6432 A regular expression is a pattern that is matched against a subject 6433 string from left to right. Most characters stand for themselves in a 6434 pattern, and match the corresponding characters in the subject. As a 6435 trivial example, the pattern 6436 6437 The quick brown fox 6438 6439 matches a portion of a subject string that is identical to itself. When 6440 caseless matching is specified (the PCRE2_CASELESS option or (?i) 6441 within the pattern), letters are matched independently of case. Note 6442 that there are two ASCII characters, K and S, that, in addition to 6443 their lower case ASCII equivalents, are case-equivalent with Unicode 6444 U+212A (Kelvin sign) and U+017F (long S) respectively when either 6445 PCRE2_UTF or PCRE2_UCP is set. 6446 6447 The power of regular expressions comes from the ability to include wild 6448 cards, character classes, alternatives, and repetitions in the pattern. 6449 These are encoded in the pattern by the use of metacharacters, which do 6450 not stand for themselves but instead are interpreted in some special 6451 way. 6452 6453 There are two different sets of metacharacters: those that are recog- 6454 nized anywhere in the pattern except within square brackets, and those 6455 that are recognized within square brackets. Outside square brackets, 6456 the metacharacters are as follows: 6457 6458 \ general escape character with several uses 6459 ^ assert start of string (or line, in multiline mode) 6460 $ assert end of string (or line, in multiline mode) 6461 . match any character except newline (by default) 6462 [ start character class definition 6463 | start of alternative branch 6464 ( start group or control verb 6465 ) end group or control verb 6466 * 0 or more quantifier 6467 + 1 or more quantifier; also "possessive quantifier" 6468 ? 0 or 1 quantifier; also quantifier minimizer 6469 { start min/max quantifier 6470 6471 Part of a pattern that is in square brackets is called a "character 6472 class". In a character class the only metacharacters are: 6473 6474 \ general escape character 6475 ^ negate the class, but only if the first character 6476 - indicates character range 6477 [ POSIX character class (if followed by POSIX syntax) 6478 ] terminates the character class 6479 6480 If a pattern is compiled with the PCRE2_EXTENDED option, most white 6481 space in the pattern, other than in a character class, and characters 6482 between a # outside a character class and the next newline, inclusive, 6483 are ignored. An escaping backslash can be used to include a white space 6484 or a # character as part of the pattern. If the PCRE2_EXTENDED_MORE op- 6485 tion is set, the same applies, but in addition unescaped space and hor- 6486 izontal tab characters are ignored inside a character class. Note: only 6487 these two characters are ignored, not the full set of pattern white 6488 space characters that are ignored outside a character class. Option 6489 settings can be changed within a pattern; see the section entitled "In- 6490 ternal Option Setting" below. 6491 6492 The following sections describe the use of each of the metacharacters. 6493 6494 6495BACKSLASH 6496 6497 The backslash character has several uses. Firstly, if it is followed by 6498 a character that is not a digit or a letter, it takes away any special 6499 meaning that character may have. This use of backslash as an escape 6500 character applies both inside and outside character classes. 6501 6502 For example, if you want to match a * character, you must write \* in 6503 the pattern. This escaping action applies whether or not the following 6504 character would otherwise be interpreted as a metacharacter, so it is 6505 always safe to precede a non-alphanumeric with backslash to specify 6506 that it stands for itself. In particular, if you want to match a back- 6507 slash, you write \\. 6508 6509 Only ASCII digits and letters have any special meaning after a back- 6510 slash. All other characters (in particular, those whose code points are 6511 greater than 127) are treated as literals. 6512 6513 If you want to treat all characters in a sequence as literals, you can 6514 do so by putting them between \Q and \E. This is different from Perl in 6515 that $ and @ are handled as literals in \Q...\E sequences in PCRE2, 6516 whereas in Perl, $ and @ cause variable interpolation. Also, Perl does 6517 "double-quotish backslash interpolation" on any backslashes between \Q 6518 and \E which, its documentation says, "may lead to confusing results". 6519 PCRE2 treats a backslash between \Q and \E just like any other charac- 6520 ter. Note the following examples: 6521 6522 Pattern PCRE2 matches Perl matches 6523 6524 \Qabc$xyz\E abc$xyz abc followed by the 6525 contents of $xyz 6526 \Qabc\$xyz\E abc\$xyz abc\$xyz 6527 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz 6528 \QA\B\E A\B A\B 6529 \Q\\E \ \\E 6530 6531 The \Q...\E sequence is recognized both inside and outside character 6532 classes. An isolated \E that is not preceded by \Q is ignored. If \Q 6533 is not followed by \E later in the pattern, the literal interpretation 6534 continues to the end of the pattern (that is, \E is assumed at the 6535 end). If the isolated \Q is inside a character class, this causes an 6536 error, because the character class is not terminated by a closing 6537 square bracket. 6538 6539 Non-printing characters 6540 6541 A second use of backslash provides a way of encoding non-printing char- 6542 acters in patterns in a visible manner. There is no restriction on the 6543 appearance of non-printing characters in a pattern, but when a pattern 6544 is being prepared by text editing, it is often easier to use one of the 6545 following escape sequences instead of the binary character it repre- 6546 sents. In an ASCII or Unicode environment, these escapes are as fol- 6547 lows: 6548 6549 \a alarm, that is, the BEL character (hex 07) 6550 \cx "control-x", where x is any printable ASCII character 6551 \e escape (hex 1B) 6552 \f form feed (hex 0C) 6553 \n linefeed (hex 0A) 6554 \r carriage return (hex 0D) (but see below) 6555 \t tab (hex 09) 6556 \0dd character with octal code 0dd 6557 \ddd character with octal code ddd, or backreference 6558 \o{ddd..} character with octal code ddd.. 6559 \xhh character with hex code hh 6560 \x{hhh..} character with hex code hhh.. 6561 \N{U+hhh..} character with Unicode hex code point hhh.. 6562 6563 By default, after \x that is not followed by {, from zero to two hexa- 6564 decimal digits are read (letters can be in upper or lower case). Any 6565 number of hexadecimal digits may appear between \x{ and }. If a charac- 6566 ter other than a hexadecimal digit appears between \x{ and }, or if 6567 there is no terminating }, an error occurs. 6568 6569 Characters whose code points are less than 256 can be defined by either 6570 of the two syntaxes for \x or by an octal sequence. There is no differ- 6571 ence in the way they are handled. For example, \xdc is exactly the same 6572 as \x{dc} or \334. However, using the braced versions does make such 6573 sequences easier to read. 6574 6575 Support is available for some ECMAScript (aka JavaScript) escape se- 6576 quences via two compile-time options. If PCRE2_ALT_BSUX is set, the se- 6577 quence \x followed by { is not recognized. Only if \x is followed by 6578 two hexadecimal digits is it recognized as a character escape. Other- 6579 wise it is interpreted as a literal "x" character. In this mode, sup- 6580 port for code points greater than 256 is provided by \u, which must be 6581 followed by four hexadecimal digits; otherwise it is interpreted as a 6582 literal "u" character. 6583 6584 PCRE2_EXTRA_ALT_BSUX has the same effect as PCRE2_ALT_BSUX and, in ad- 6585 dition, \u{hhh..} is recognized as the character specified by hexadeci- 6586 mal code point. There may be any number of hexadecimal digits. This 6587 syntax is from ECMAScript 6. 6588 6589 The \N{U+hhh..} escape sequence is recognized only when PCRE2 is oper- 6590 ating in UTF mode. Perl also uses \N{name} to specify characters by 6591 Unicode name; PCRE2 does not support this. Note that when \N is not 6592 followed by an opening brace (curly bracket) it has an entirely differ- 6593 ent meaning, matching any character that is not a newline. 6594 6595 There are some legacy applications where the escape sequence \r is ex- 6596 pected to match a newline. If the PCRE2_EXTRA_ESCAPED_CR_IS_LF option 6597 is set, \r in a pattern is converted to \n so that it matches a LF 6598 (linefeed) instead of a CR (carriage return) character. 6599 6600 The precise effect of \cx on ASCII characters is as follows: if x is a 6601 lower case letter, it is converted to upper case. Then bit 6 of the 6602 character (hex 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A 6603 (A is 41, Z is 5A), but \c{ becomes hex 3B ({ is 7B), and \c; becomes 6604 hex 7B (; is 3B). If the code unit following \c has a value less than 6605 32 or greater than 126, a compile-time error occurs. 6606 6607 When PCRE2 is compiled in EBCDIC mode, \N{U+hhh..} is not supported. 6608 \a, \e, \f, \n, \r, and \t generate the appropriate EBCDIC code values. 6609 The \c escape is processed as specified for Perl in the perlebcdic doc- 6610 ument. The only characters that are allowed after \c are A-Z, a-z, or 6611 one of @, [, \, ], ^, _, or ?. Any other character provokes a compile- 6612 time error. The sequence \c@ encodes character code 0; after \c the 6613 letters (in either case) encode characters 1-26 (hex 01 to hex 1A); [, 6614 \, ], ^, and _ encode characters 27-31 (hex 1B to hex 1F), and \c? be- 6615 comes either 255 (hex FF) or 95 (hex 5F). 6616 6617 Thus, apart from \c?, these escapes generate the same character code 6618 values as they do in an ASCII environment, though the meanings of the 6619 values mostly differ. For example, \cG always generates code value 7, 6620 which is BEL in ASCII but DEL in EBCDIC. 6621 6622 The sequence \c? generates DEL (127, hex 7F) in an ASCII environment, 6623 but because 127 is not a control character in EBCDIC, Perl makes it 6624 generate the APC character. Unfortunately, there are several variants 6625 of EBCDIC. In most of them the APC character has the value 255 (hex 6626 FF), but in the one Perl calls POSIX-BC its value is 95 (hex 5F). If 6627 certain other characters have POSIX-BC values, PCRE2 makes \c? generate 6628 95; otherwise it generates 255. 6629 6630 After \0 up to two further octal digits are read. If there are fewer 6631 than two digits, just those that are present are used. Thus the se- 6632 quence \0\x\015 specifies two binary zeros followed by a CR character 6633 (code value 13). Make sure you supply two digits after the initial zero 6634 if the pattern character that follows is itself an octal digit. 6635 6636 The escape \o must be followed by a sequence of octal digits, enclosed 6637 in braces. An error occurs if this is not the case. This escape is a 6638 recent addition to Perl; it provides way of specifying character code 6639 points as octal numbers greater than 0777, and it also allows octal 6640 numbers and backreferences to be unambiguously specified. 6641 6642 For greater clarity and unambiguity, it is best to avoid following \ by 6643 a digit greater than zero. Instead, use \o{} or \x{} to specify numeri- 6644 cal character code points, and \g{} to specify backreferences. The fol- 6645 lowing paragraphs describe the old, ambiguous syntax. 6646 6647 The handling of a backslash followed by a digit other than 0 is compli- 6648 cated, and Perl has changed over time, causing PCRE2 also to change. 6649 6650 Outside a character class, PCRE2 reads the digit and any following dig- 6651 its as a decimal number. If the number is less than 10, begins with the 6652 digit 8 or 9, or if there are at least that many previous capture 6653 groups in the expression, the entire sequence is taken as a backrefer- 6654 ence. A description of how this works is given later, following the 6655 discussion of parenthesized groups. Otherwise, up to three octal dig- 6656 its are read to form a character code. 6657 6658 Inside a character class, PCRE2 handles \8 and \9 as the literal char- 6659 acters "8" and "9", and otherwise reads up to three octal digits fol- 6660 lowing the backslash, using them to generate a data character. Any sub- 6661 sequent digits stand for themselves. For example, outside a character 6662 class: 6663 6664 \040 is another way of writing an ASCII space 6665 \40 is the same, provided there are fewer than 40 6666 previous capture groups 6667 \7 is always a backreference 6668 \11 might be a backreference, or another way of 6669 writing a tab 6670 \011 is always a tab 6671 \0113 is a tab followed by the character "3" 6672 \113 might be a backreference, otherwise the 6673 character with octal code 113 6674 \377 might be a backreference, otherwise 6675 the value 255 (decimal) 6676 \81 is always a backreference 6677 6678 Note that octal values of 100 or greater that are specified using this 6679 syntax must not be introduced by a leading zero, because no more than 6680 three octal digits are ever read. 6681 6682 Constraints on character values 6683 6684 Characters that are specified using octal or hexadecimal numbers are 6685 limited to certain values, as follows: 6686 6687 8-bit non-UTF mode no greater than 0xff 6688 16-bit non-UTF mode no greater than 0xffff 6689 32-bit non-UTF mode no greater than 0xffffffff 6690 All UTF modes no greater than 0x10ffff and a valid code point 6691 6692 Invalid Unicode code points are all those in the range 0xd800 to 0xdfff 6693 (the so-called "surrogate" code points). The check for these can be 6694 disabled by the caller of pcre2_compile() by setting the option 6695 PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES. However, this is possible only in 6696 UTF-8 and UTF-32 modes, because these values are not representable in 6697 UTF-16. 6698 6699 Escape sequences in character classes 6700 6701 All the sequences that define a single character value can be used both 6702 inside and outside character classes. In addition, inside a character 6703 class, \b is interpreted as the backspace character (hex 08). 6704 6705 When not followed by an opening brace, \N is not allowed in a character 6706 class. \B, \R, and \X are not special inside a character class. Like 6707 other unrecognized alphabetic escape sequences, they cause an error. 6708 Outside a character class, these sequences have different meanings. 6709 6710 Unsupported escape sequences 6711 6712 In Perl, the sequences \F, \l, \L, \u, and \U are recognized by its 6713 string handler and used to modify the case of following characters. By 6714 default, PCRE2 does not support these escape sequences in patterns. 6715 However, if either of the PCRE2_ALT_BSUX or PCRE2_EXTRA_ALT_BSUX op- 6716 tions is set, \U matches a "U" character, and \u can be used to define 6717 a character by code point, as described above. 6718 6719 Absolute and relative backreferences 6720 6721 The sequence \g followed by a signed or unsigned number, optionally en- 6722 closed in braces, is an absolute or relative backreference. A named 6723 backreference can be coded as \g{name}. Backreferences are discussed 6724 later, following the discussion of parenthesized groups. 6725 6726 Absolute and relative subroutine calls 6727 6728 For compatibility with Oniguruma, the non-Perl syntax \g followed by a 6729 name or a number enclosed either in angle brackets or single quotes, is 6730 an alternative syntax for referencing a capture group as a subroutine. 6731 Details are discussed later. Note that \g{...} (Perl syntax) and 6732 \g<...> (Oniguruma syntax) are not synonymous. The former is a backref- 6733 erence; the latter is a subroutine call. 6734 6735 Generic character types 6736 6737 Another use of backslash is for specifying generic character types: 6738 6739 \d any decimal digit 6740 \D any character that is not a decimal digit 6741 \h any horizontal white space character 6742 \H any character that is not a horizontal white space character 6743 \N any character that is not a newline 6744 \s any white space character 6745 \S any character that is not a white space character 6746 \v any vertical white space character 6747 \V any character that is not a vertical white space character 6748 \w any "word" character 6749 \W any "non-word" character 6750 6751 The \N escape sequence has the same meaning as the "." metacharacter 6752 when PCRE2_DOTALL is not set, but setting PCRE2_DOTALL does not change 6753 the meaning of \N. Note that when \N is followed by an opening brace it 6754 has a different meaning. See the section entitled "Non-printing charac- 6755 ters" above for details. Perl also uses \N{name} to specify characters 6756 by Unicode name; PCRE2 does not support this. 6757 6758 Each pair of lower and upper case escape sequences partitions the com- 6759 plete set of characters into two disjoint sets. Any given character 6760 matches one, and only one, of each pair. The sequences can appear both 6761 inside and outside character classes. They each match one character of 6762 the appropriate type. If the current matching point is at the end of 6763 the subject string, all of them fail, because there is no character to 6764 match. 6765 6766 The default \s characters are HT (9), LF (10), VT (11), FF (12), CR 6767 (13), and space (32), which are defined as white space in the "C" lo- 6768 cale. This list may vary if locale-specific matching is taking place. 6769 For example, in some locales the "non-breaking space" character (\xA0) 6770 is recognized as white space, and in others the VT character is not. 6771 6772 A "word" character is an underscore or any character that is a letter 6773 or digit. By default, the definition of letters and digits is con- 6774 trolled by PCRE2's low-valued character tables, and may vary if locale- 6775 specific matching is taking place (see "Locale support" in the pcre2api 6776 page). For example, in a French locale such as "fr_FR" in Unix-like 6777 systems, or "french" in Windows, some character codes greater than 127 6778 are used for accented letters, and these are then matched by \w. The 6779 use of locales with Unicode is discouraged. 6780 6781 By default, characters whose code points are greater than 127 never 6782 match \d, \s, or \w, and always match \D, \S, and \W, although this may 6783 be different for characters in the range 128-255 when locale-specific 6784 matching is happening. These escape sequences retain their original 6785 meanings from before Unicode support was available, mainly for effi- 6786 ciency reasons. If the PCRE2_UCP option is set, the behaviour is 6787 changed so that Unicode properties are used to determine character 6788 types, as follows: 6789 6790 \d any character that matches \p{Nd} (decimal digit) 6791 \s any character that matches \p{Z} or \h or \v 6792 \w any character that matches \p{L} or \p{N}, plus underscore 6793 6794 The upper case escapes match the inverse sets of characters. Note that 6795 \d matches only decimal digits, whereas \w matches any Unicode digit, 6796 as well as any Unicode letter, and underscore. Note also that PCRE2_UCP 6797 affects \b, and \B because they are defined in terms of \w and \W. 6798 Matching these sequences is noticeably slower when PCRE2_UCP is set. 6799 6800 The sequences \h, \H, \v, and \V, in contrast to the other sequences, 6801 which match only ASCII characters by default, always match a specific 6802 list of code points, whether or not PCRE2_UCP is set. The horizontal 6803 space characters are: 6804 6805 U+0009 Horizontal tab (HT) 6806 U+0020 Space 6807 U+00A0 Non-break space 6808 U+1680 Ogham space mark 6809 U+180E Mongolian vowel separator 6810 U+2000 En quad 6811 U+2001 Em quad 6812 U+2002 En space 6813 U+2003 Em space 6814 U+2004 Three-per-em space 6815 U+2005 Four-per-em space 6816 U+2006 Six-per-em space 6817 U+2007 Figure space 6818 U+2008 Punctuation space 6819 U+2009 Thin space 6820 U+200A Hair space 6821 U+202F Narrow no-break space 6822 U+205F Medium mathematical space 6823 U+3000 Ideographic space 6824 6825 The vertical space characters are: 6826 6827 U+000A Linefeed (LF) 6828 U+000B Vertical tab (VT) 6829 U+000C Form feed (FF) 6830 U+000D Carriage return (CR) 6831 U+0085 Next line (NEL) 6832 U+2028 Line separator 6833 U+2029 Paragraph separator 6834 6835 In 8-bit, non-UTF-8 mode, only the characters with code points less 6836 than 256 are relevant. 6837 6838 Newline sequences 6839 6840 Outside a character class, by default, the escape sequence \R matches 6841 any Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent 6842 to the following: 6843 6844 (?>\r\n|\n|\x0b|\f|\r|\x85) 6845 6846 This is an example of an "atomic group", details of which are given be- 6847 low. This particular group matches either the two-character sequence 6848 CR followed by LF, or one of the single characters LF (linefeed, 6849 U+000A), VT (vertical tab, U+000B), FF (form feed, U+000C), CR (car- 6850 riage return, U+000D), or NEL (next line, U+0085). Because this is an 6851 atomic group, the two-character sequence is treated as a single unit 6852 that cannot be split. 6853 6854 In other modes, two additional characters whose code points are greater 6855 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa- 6856 rator, U+2029). Unicode support is not needed for these characters to 6857 be recognized. 6858 6859 It is possible to restrict \R to match only CR, LF, or CRLF (instead of 6860 the complete set of Unicode line endings) by setting the option 6861 PCRE2_BSR_ANYCRLF at compile time. (BSR is an abbreviation for "back- 6862 slash R".) This can be made the default when PCRE2 is built; if this is 6863 the case, the other behaviour can be requested via the PCRE2_BSR_UNI- 6864 CODE option. It is also possible to specify these settings by starting 6865 a pattern string with one of the following sequences: 6866 6867 (*BSR_ANYCRLF) CR, LF, or CRLF only 6868 (*BSR_UNICODE) any Unicode newline sequence 6869 6870 These override the default and the options given to the compiling func- 6871 tion. Note that these special settings, which are not Perl-compatible, 6872 are recognized only at the very start of a pattern, and that they must 6873 be in upper case. If more than one of them is present, the last one is 6874 used. They can be combined with a change of newline convention; for ex- 6875 ample, a pattern can start with: 6876 6877 (*ANY)(*BSR_ANYCRLF) 6878 6879 They can also be combined with the (*UTF) or (*UCP) special sequences. 6880 Inside a character class, \R is treated as an unrecognized escape se- 6881 quence, and causes an error. 6882 6883 Unicode character properties 6884 6885 When PCRE2 is built with Unicode support (the default), three addi- 6886 tional escape sequences that match characters with specific properties 6887 are available. They can be used in any mode, though in 8-bit and 16-bit 6888 non-UTF modes these sequences are of course limited to testing charac- 6889 ters whose code points are less than U+0100 and U+10000, respectively. 6890 In 32-bit non-UTF mode, code points greater than 0x10ffff (the Unicode 6891 limit) may be encountered. These are all treated as being in the Un- 6892 known script and with an unassigned type. 6893 6894 Matching characters by Unicode property is not fast, because PCRE2 has 6895 to do a multistage table lookup in order to find a character's prop- 6896 erty. That is why the traditional escape sequences such as \d and \w do 6897 not use Unicode properties in PCRE2 by default, though you can make 6898 them do so by setting the PCRE2_UCP option or by starting the pattern 6899 with (*UCP). 6900 6901 The extra escape sequences that provide property support are: 6902 6903 \p{xx} a character with the xx property 6904 \P{xx} a character without the xx property 6905 \X a Unicode extended grapheme cluster 6906 6907 The property names represented by xx above are not case-sensitive, and 6908 in accordance with Unicode's "loose matching" rules, spaces, hyphens, 6909 and underscores are ignored. There is support for Unicode script names, 6910 Unicode general category properties, "Any", which matches any character 6911 (including newline), Bidi_Class, a number of binary (yes/no) proper- 6912 ties, and some special PCRE2 properties (described below). Certain 6913 other Perl properties such as "InMusicalSymbols" are not supported by 6914 PCRE2. Note that \P{Any} does not match any characters, so always 6915 causes a match failure. 6916 6917 Script properties for \p and \P 6918 6919 There are three different syntax forms for matching a script. Each Uni- 6920 code character has a basic script and, optionally, a list of other 6921 scripts ("Script Extensions") with which it is commonly used. Using the 6922 Adlam script as an example, \p{sc:Adlam} matches characters whose basic 6923 script is Adlam, whereas \p{scx:Adlam} matches, in addition, characters 6924 that have Adlam in their extensions list. The full names "script" and 6925 "script extensions" for the property types are recognized, and a equals 6926 sign is an alternative to the colon. If a script name is given without 6927 a property type, for example, \p{Adlam}, it is treated as \p{scx:Ad- 6928 lam}. Perl changed to this interpretation at release 5.26 and PCRE2 6929 changed at release 10.40. 6930 6931 Unassigned characters (and in non-UTF 32-bit mode, characters with code 6932 points greater than 0x10FFFF) are assigned the "Unknown" script. Others 6933 that are not part of an identified script are lumped together as "Com- 6934 mon". The current list of recognized script names and their 4-character 6935 abbreviations can be obtained by running this command: 6936 6937 pcre2test -LS 6938 6939 6940 The general category property for \p and \P 6941 6942 Each character has exactly one Unicode general category property, spec- 6943 ified by a two-letter abbreviation. For compatibility with Perl, nega- 6944 tion can be specified by including a circumflex between the opening 6945 brace and the property name. For example, \p{^Lu} is the same as 6946 \P{Lu}. 6947 6948 If only one letter is specified with \p or \P, it includes all the gen- 6949 eral category properties that start with that letter. In this case, in 6950 the absence of negation, the curly brackets in the escape sequence are 6951 optional; these two examples have the same effect: 6952 6953 \p{L} 6954 \pL 6955 6956 The following general category property codes are supported: 6957 6958 C Other 6959 Cc Control 6960 Cf Format 6961 Cn Unassigned 6962 Co Private use 6963 Cs Surrogate 6964 6965 L Letter 6966 Ll Lower case letter 6967 Lm Modifier letter 6968 Lo Other letter 6969 Lt Title case letter 6970 Lu Upper case letter 6971 6972 M Mark 6973 Mc Spacing mark 6974 Me Enclosing mark 6975 Mn Non-spacing mark 6976 6977 N Number 6978 Nd Decimal number 6979 Nl Letter number 6980 No Other number 6981 6982 P Punctuation 6983 Pc Connector punctuation 6984 Pd Dash punctuation 6985 Pe Close punctuation 6986 Pf Final punctuation 6987 Pi Initial punctuation 6988 Po Other punctuation 6989 Ps Open punctuation 6990 6991 S Symbol 6992 Sc Currency symbol 6993 Sk Modifier symbol 6994 Sm Mathematical symbol 6995 So Other symbol 6996 6997 Z Separator 6998 Zl Line separator 6999 Zp Paragraph separator 7000 Zs Space separator 7001 7002 The special property LC, which has the synonym L&, is also supported: 7003 it matches a character that has the Lu, Ll, or Lt property, in other 7004 words, a letter that is not classified as a modifier or "other". 7005 7006 The Cs (Surrogate) property applies only to characters whose code 7007 points are in the range U+D800 to U+DFFF. These characters are no dif- 7008 ferent to any other character when PCRE2 is not in UTF mode (using the 7009 16-bit or 32-bit library). However, they are not valid in Unicode 7010 strings and so cannot be tested by PCRE2 in UTF mode, unless UTF valid- 7011 ity checking has been turned off (see the discussion of 7012 PCRE2_NO_UTF_CHECK in the pcre2api page). 7013 7014 The long synonyms for property names that Perl supports (such as 7015 \p{Letter}) are not supported by PCRE2, nor is it permitted to prefix 7016 any of these properties with "Is". 7017 7018 No character that is in the Unicode table has the Cn (unassigned) prop- 7019 erty. Instead, this property is assumed for any code point that is not 7020 in the Unicode table. 7021 7022 Specifying caseless matching does not affect these escape sequences. 7023 For example, \p{Lu} always matches only upper case letters. This is 7024 different from the behaviour of current versions of Perl. 7025 7026 Binary (yes/no) properties for \p and \P 7027 7028 Unicode defines a number of binary properties, that is, properties 7029 whose only values are true or false. You can obtain a list of those 7030 that are recognized by \p and \P, along with their abbreviations, by 7031 running this command: 7032 7033 pcre2test -LP 7034 7035 7036 The Bidi_Class property for \p and \P 7037 7038 \p{Bidi_Class:<class>} matches a character with the given class 7039 \p{BC:<class>} matches a character with the given class 7040 7041 The recognized classes are: 7042 7043 AL Arabic letter 7044 AN Arabic number 7045 B paragraph separator 7046 BN boundary neutral 7047 CS common separator 7048 EN European number 7049 ES European separator 7050 ET European terminator 7051 FSI first strong isolate 7052 L left-to-right 7053 LRE left-to-right embedding 7054 LRI left-to-right isolate 7055 LRO left-to-right override 7056 NSM non-spacing mark 7057 ON other neutral 7058 PDF pop directional format 7059 PDI pop directional isolate 7060 R right-to-left 7061 RLE right-to-left embedding 7062 RLI right-to-left isolate 7063 RLO right-to-left override 7064 S segment separator 7065 WS which space 7066 7067 An equals sign may be used instead of a colon. The class names are 7068 case-insensitive; only the short names listed above are recognized. 7069 7070 Extended grapheme clusters 7071 7072 The \X escape matches any number of Unicode characters that form an 7073 "extended grapheme cluster", and treats the sequence as an atomic group 7074 (see below). Unicode supports various kinds of composite character by 7075 giving each character a grapheme breaking property, and having rules 7076 that use these properties to define the boundaries of extended grapheme 7077 clusters. The rules are defined in Unicode Standard Annex 29, "Unicode 7078 Text Segmentation". Unicode 11.0.0 abandoned the use of some previous 7079 properties that had been used for emojis. Instead it introduced vari- 7080 ous emoji-specific properties. PCRE2 uses only the Extended Picto- 7081 graphic property. 7082 7083 \X always matches at least one character. Then it decides whether to 7084 add additional characters according to the following rules for ending a 7085 cluster: 7086 7087 1. End at the end of the subject string. 7088 7089 2. Do not end between CR and LF; otherwise end after any control char- 7090 acter. 7091 7092 3. Do not break Hangul (a Korean script) syllable sequences. Hangul 7093 characters are of five types: L, V, T, LV, and LVT. An L character may 7094 be followed by an L, V, LV, or LVT character; an LV or V character may 7095 be followed by a V or T character; an LVT or T character may be fol- 7096 lowed only by a T character. 7097 7098 4. Do not end before extending characters or spacing marks or the 7099 "zero-width joiner" character. Characters with the "mark" property al- 7100 ways have the "extend" grapheme breaking property. 7101 7102 5. Do not end after prepend characters. 7103 7104 6. Do not break within emoji modifier sequences or emoji zwj sequences. 7105 That is, do not break between characters with the Extended_Pictographic 7106 property. Extend and ZWJ characters are allowed between the charac- 7107 ters. 7108 7109 7. Do not break within emoji flag sequences. That is, do not break be- 7110 tween regional indicator (RI) characters if there are an odd number of 7111 RI characters before the break point. 7112 7113 8. Otherwise, end the cluster. 7114 7115 PCRE2's additional properties 7116 7117 As well as the standard Unicode properties described above, PCRE2 sup- 7118 ports four more that make it possible to convert traditional escape se- 7119 quences such as \w and \s to use Unicode properties. PCRE2 uses these 7120 non-standard, non-Perl properties internally when PCRE2_UCP is set. 7121 However, they may also be used explicitly. These properties are: 7122 7123 Xan Any alphanumeric character 7124 Xps Any POSIX space character 7125 Xsp Any Perl space character 7126 Xwd Any Perl "word" character 7127 7128 Xan matches characters that have either the L (letter) or the N (num- 7129 ber) property. Xps matches the characters tab, linefeed, vertical tab, 7130 form feed, or carriage return, and any other character that has the Z 7131 (separator) property. Xsp is the same as Xps; in PCRE1 it used to ex- 7132 clude vertical tab, for Perl compatibility, but Perl changed. Xwd 7133 matches the same characters as Xan, plus underscore. 7134 7135 There is another non-standard property, Xuc, which matches any charac- 7136 ter that can be represented by a Universal Character Name in C++ and 7137 other programming languages. These are the characters $, @, ` (grave 7138 accent), and all characters with Unicode code points greater than or 7139 equal to U+00A0, except for the surrogates U+D800 to U+DFFF. Note that 7140 most base (ASCII) characters are excluded. (Universal Character Names 7141 are of the form \uHHHH or \UHHHHHHHH where H is a hexadecimal digit. 7142 Note that the Xuc property does not match these sequences but the char- 7143 acters that they represent.) 7144 7145 Resetting the match start 7146 7147 In normal use, the escape sequence \K causes any previously matched 7148 characters not to be included in the final matched sequence that is re- 7149 turned. For example, the pattern: 7150 7151 foo\Kbar 7152 7153 matches "foobar", but reports that it has matched "bar". \K does not 7154 interact with anchoring in any way. The pattern: 7155 7156 ^foo\Kbar 7157 7158 matches only when the subject begins with "foobar" (in single line 7159 mode), though it again reports the matched string as "bar". This fea- 7160 ture is similar to a lookbehind assertion (described below). However, 7161 in this case, the part of the subject before the real match does not 7162 have to be of fixed length, as lookbehind assertions do. The use of \K 7163 does not interfere with the setting of captured substrings. For exam- 7164 ple, when the pattern 7165 7166 (foo)\Kbar 7167 7168 matches "foobar", the first substring is still set to "foo". 7169 7170 From version 5.32.0 Perl forbids the use of \K in lookaround asser- 7171 tions. From release 10.38 PCRE2 also forbids this by default. However, 7172 the PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK option can be used when calling 7173 pcre2_compile() to re-enable the previous behaviour. When this option 7174 is set, \K is acted upon when it occurs inside positive assertions, but 7175 is ignored in negative assertions. Note that when a pattern such as 7176 (?=ab\K) matches, the reported start of the match can be greater than 7177 the end of the match. Using \K in a lookbehind assertion at the start 7178 of a pattern can also lead to odd effects. For example, consider this 7179 pattern: 7180 7181 (?<=\Kfoo)bar 7182 7183 If the subject is "foobar", a call to pcre2_match() with a starting 7184 offset of 3 succeeds and reports the matching string as "foobar", that 7185 is, the start of the reported match is earlier than where the match 7186 started. 7187 7188 Simple assertions 7189 7190 The final use of backslash is for certain simple assertions. An asser- 7191 tion specifies a condition that has to be met at a particular point in 7192 a match, without consuming any characters from the subject string. The 7193 use of groups for more complicated assertions is described below. The 7194 backslashed assertions are: 7195 7196 \b matches at a word boundary 7197 \B matches when not at a word boundary 7198 \A matches at the start of the subject 7199 \Z matches at the end of the subject 7200 also matches before a newline at the end of the subject 7201 \z matches only at the end of the subject 7202 \G matches at the first matching position in the subject 7203 7204 Inside a character class, \b has a different meaning; it matches the 7205 backspace character. If any other of these assertions appears in a 7206 character class, an "invalid escape sequence" error is generated. 7207 7208 A word boundary is a position in the subject string where the current 7209 character and the previous character do not both match \w or \W (i.e. 7210 one matches \w and the other matches \W), or the start or end of the 7211 string if the first or last character matches \w, respectively. When 7212 PCRE2 is built with Unicode support, the meanings of \w and \W can be 7213 changed by setting the PCRE2_UCP option. When this is done, it also af- 7214 fects \b and \B. Neither PCRE2 nor Perl has a separate "start of word" 7215 or "end of word" metasequence. However, whatever follows \b normally 7216 determines which it is. For example, the fragment \ba matches "a" at 7217 the start of a word. 7218 7219 The \A, \Z, and \z assertions differ from the traditional circumflex 7220 and dollar (described in the next section) in that they only ever match 7221 at the very start and end of the subject string, whatever options are 7222 set. Thus, they are independent of multiline mode. These three asser- 7223 tions are not affected by the PCRE2_NOTBOL or PCRE2_NOTEOL options, 7224 which affect only the behaviour of the circumflex and dollar metachar- 7225 acters. However, if the startoffset argument of pcre2_match() is non- 7226 zero, indicating that matching is to start at a point other than the 7227 beginning of the subject, \A can never match. The difference between 7228 \Z and \z is that \Z matches before a newline at the end of the string 7229 as well as at the very end, whereas \z matches only at the end. 7230 7231 The \G assertion is true only when the current matching position is at 7232 the start point of the matching process, as specified by the startoff- 7233 set argument of pcre2_match(). It differs from \A when the value of 7234 startoffset is non-zero. By calling pcre2_match() multiple times with 7235 appropriate arguments, you can mimic Perl's /g option, and it is in 7236 this kind of implementation where \G can be useful. 7237 7238 Note, however, that PCRE2's implementation of \G, being true at the 7239 starting character of the matching process, is subtly different from 7240 Perl's, which defines it as true at the end of the previous match. In 7241 Perl, these can be different when the previously matched string was 7242 empty. Because PCRE2 does just one match at a time, it cannot reproduce 7243 this behaviour. 7244 7245 If all the alternatives of a pattern begin with \G, the expression is 7246 anchored to the starting match position, and the "anchored" flag is set 7247 in the compiled regular expression. 7248 7249 7250CIRCUMFLEX AND DOLLAR 7251 7252 The circumflex and dollar metacharacters are zero-width assertions. 7253 That is, they test for a particular condition being true without con- 7254 suming any characters from the subject string. These two metacharacters 7255 are concerned with matching the starts and ends of lines. If the new- 7256 line convention is set so that only the two-character sequence CRLF is 7257 recognized as a newline, isolated CR and LF characters are treated as 7258 ordinary data characters, and are not recognized as newlines. 7259 7260 Outside a character class, in the default matching mode, the circumflex 7261 character is an assertion that is true only if the current matching 7262 point is at the start of the subject string. If the startoffset argu- 7263 ment of pcre2_match() is non-zero, or if PCRE2_NOTBOL is set, circum- 7264 flex can never match if the PCRE2_MULTILINE option is unset. Inside a 7265 character class, circumflex has an entirely different meaning (see be- 7266 low). 7267 7268 Circumflex need not be the first character of the pattern if a number 7269 of alternatives are involved, but it should be the first thing in each 7270 alternative in which it appears if the pattern is ever to match that 7271 branch. If all possible alternatives start with a circumflex, that is, 7272 if the pattern is constrained to match only at the start of the sub- 7273 ject, it is said to be an "anchored" pattern. (There are also other 7274 constructs that can cause a pattern to be anchored.) 7275 7276 The dollar character is an assertion that is true only if the current 7277 matching point is at the end of the subject string, or immediately be- 7278 fore a newline at the end of the string (by default), unless PCRE2_NO- 7279 TEOL is set. Note, however, that it does not actually match the new- 7280 line. Dollar need not be the last character of the pattern if a number 7281 of alternatives are involved, but it should be the last item in any 7282 branch in which it appears. Dollar has no special meaning in a charac- 7283 ter class. 7284 7285 The meaning of dollar can be changed so that it matches only at the 7286 very end of the string, by setting the PCRE2_DOLLAR_ENDONLY option at 7287 compile time. This does not affect the \Z assertion. 7288 7289 The meanings of the circumflex and dollar metacharacters are changed if 7290 the PCRE2_MULTILINE option is set. When this is the case, a dollar 7291 character matches before any newlines in the string, as well as at the 7292 very end, and a circumflex matches immediately after internal newlines 7293 as well as at the start of the subject string. It does not match after 7294 a newline that ends the string, for compatibility with Perl. However, 7295 this can be changed by setting the PCRE2_ALT_CIRCUMFLEX option. 7296 7297 For example, the pattern /^abc$/ matches the subject string "def\nabc" 7298 (where \n represents a newline) in multiline mode, but not otherwise. 7299 Consequently, patterns that are anchored in single line mode because 7300 all branches start with ^ are not anchored in multiline mode, and a 7301 match for circumflex is possible when the startoffset argument of 7302 pcre2_match() is non-zero. The PCRE2_DOLLAR_ENDONLY option is ignored 7303 if PCRE2_MULTILINE is set. 7304 7305 When the newline convention (see "Newline conventions" below) recog- 7306 nizes the two-character sequence CRLF as a newline, this is preferred, 7307 even if the single characters CR and LF are also recognized as new- 7308 lines. For example, if the newline convention is "any", a multiline 7309 mode circumflex matches before "xyz" in the string "abc\r\nxyz" rather 7310 than after CR, even though CR on its own is a valid newline. (It also 7311 matches at the very start of the string, of course.) 7312 7313 Note that the sequences \A, \Z, and \z can be used to match the start 7314 and end of the subject in both modes, and if all branches of a pattern 7315 start with \A it is always anchored, whether or not PCRE2_MULTILINE is 7316 set. 7317 7318 7319FULL STOP (PERIOD, DOT) AND \N 7320 7321 Outside a character class, a dot in the pattern matches any one charac- 7322 ter in the subject string except (by default) a character that signi- 7323 fies the end of a line. One or more characters may be specified as line 7324 terminators (see "Newline conventions" above). 7325 7326 Dot never matches a single line-ending character. When the two-charac- 7327 ter sequence CRLF is the only line ending, dot does not match CR if it 7328 is immediately followed by LF, but otherwise it matches all characters 7329 (including isolated CRs and LFs). When ANYCRLF is selected for line 7330 endings, no occurences of CR of LF match dot. When all Unicode line 7331 endings are being recognized, dot does not match CR or LF or any of the 7332 other line ending characters. 7333 7334 The behaviour of dot with regard to newlines can be changed. If the 7335 PCRE2_DOTALL option is set, a dot matches any one character, without 7336 exception. If the two-character sequence CRLF is present in the sub- 7337 ject string, it takes two dots to match it. 7338 7339 The handling of dot is entirely independent of the handling of circum- 7340 flex and dollar, the only relationship being that they both involve 7341 newlines. Dot has no special meaning in a character class. 7342 7343 The escape sequence \N when not followed by an opening brace behaves 7344 like a dot, except that it is not affected by the PCRE2_DOTALL option. 7345 In other words, it matches any character except one that signifies the 7346 end of a line. 7347 7348 When \N is followed by an opening brace it has a different meaning. See 7349 the section entitled "Non-printing characters" above for details. Perl 7350 also uses \N{name} to specify characters by Unicode name; PCRE2 does 7351 not support this. 7352 7353 7354MATCHING A SINGLE CODE UNIT 7355 7356 Outside a character class, the escape sequence \C matches any one code 7357 unit, whether or not a UTF mode is set. In the 8-bit library, one code 7358 unit is one byte; in the 16-bit library it is a 16-bit unit; in the 7359 32-bit library it is a 32-bit unit. Unlike a dot, \C always matches 7360 line-ending characters. The feature is provided in Perl in order to 7361 match individual bytes in UTF-8 mode, but it is unclear how it can use- 7362 fully be used. 7363 7364 Because \C breaks up characters into individual code units, matching 7365 one unit with \C in UTF-8 or UTF-16 mode means that the rest of the 7366 string may start with a malformed UTF character. This has undefined re- 7367 sults, because PCRE2 assumes that it is matching character by character 7368 in a valid UTF string (by default it checks the subject string's valid- 7369 ity at the start of processing unless the PCRE2_NO_UTF_CHECK or 7370 PCRE2_MATCH_INVALID_UTF option is used). 7371 7372 An application can lock out the use of \C by setting the 7373 PCRE2_NEVER_BACKSLASH_C option when compiling a pattern. It is also 7374 possible to build PCRE2 with the use of \C permanently disabled. 7375 7376 PCRE2 does not allow \C to appear in lookbehind assertions (described 7377 below) in UTF-8 or UTF-16 modes, because this would make it impossible 7378 to calculate the length of the lookbehind. Neither the alternative 7379 matching function pcre2_dfa_match() nor the JIT optimizer support \C in 7380 these UTF modes. The former gives a match-time error; the latter fails 7381 to optimize and so the match is always run using the interpreter. 7382 7383 In the 32-bit library, however, \C is always supported (when not ex- 7384 plicitly locked out) because it always matches a single code unit, 7385 whether or not UTF-32 is specified. 7386 7387 In general, the \C escape sequence is best avoided. However, one way of 7388 using it that avoids the problem of malformed UTF-8 or UTF-16 charac- 7389 ters is to use a lookahead to check the length of the next character, 7390 as in this pattern, which could be used with a UTF-8 string (ignore 7391 white space and line breaks): 7392 7393 (?| (?=[\x00-\x7f])(\C) | 7394 (?=[\x80-\x{7ff}])(\C)(\C) | 7395 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) | 7396 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C)) 7397 7398 In this example, a group that starts with (?| resets the capturing 7399 parentheses numbers in each alternative (see "Duplicate Group Numbers" 7400 below). The assertions at the start of each branch check the next UTF-8 7401 character for values whose encoding uses 1, 2, 3, or 4 bytes, respec- 7402 tively. The character's individual bytes are then captured by the ap- 7403 propriate number of \C groups. 7404 7405 7406SQUARE BRACKETS AND CHARACTER CLASSES 7407 7408 An opening square bracket introduces a character class, terminated by a 7409 closing square bracket. A closing square bracket on its own is not spe- 7410 cial by default. If a closing square bracket is required as a member 7411 of the class, it should be the first data character in the class (after 7412 an initial circumflex, if present) or escaped with a backslash. This 7413 means that, by default, an empty class cannot be defined. However, if 7414 the PCRE2_ALLOW_EMPTY_CLASS option is set, a closing square bracket at 7415 the start does end the (empty) class. 7416 7417 A character class matches a single character in the subject. A matched 7418 character must be in the set of characters defined by the class, unless 7419 the first character in the class definition is a circumflex, in which 7420 case the subject character must not be in the set defined by the class. 7421 If a circumflex is actually required as a member of the class, ensure 7422 it is not the first character, or escape it with a backslash. 7423 7424 For example, the character class [aeiou] matches any lower case vowel, 7425 while [^aeiou] matches any character that is not a lower case vowel. 7426 Note that a circumflex is just a convenient notation for specifying the 7427 characters that are in the class by enumerating those that are not. A 7428 class that starts with a circumflex is not an assertion; it still con- 7429 sumes a character from the subject string, and therefore it fails if 7430 the current pointer is at the end of the string. 7431 7432 Characters in a class may be specified by their code points using \o, 7433 \x, or \N{U+hh..} in the usual way. When caseless matching is set, any 7434 letters in a class represent both their upper case and lower case ver- 7435 sions, so for example, a caseless [aeiou] matches "A" as well as "a", 7436 and a caseless [^aeiou] does not match "A", whereas a caseful version 7437 would. Note that there are two ASCII characters, K and S, that, in ad- 7438 dition to their lower case ASCII equivalents, are case-equivalent with 7439 Unicode U+212A (Kelvin sign) and U+017F (long S) respectively when ei- 7440 ther PCRE2_UTF or PCRE2_UCP is set. 7441 7442 Characters that might indicate line breaks are never treated in any 7443 special way when matching character classes, whatever line-ending se- 7444 quence is in use, and whatever setting of the PCRE2_DOTALL and 7445 PCRE2_MULTILINE options is used. A class such as [^a] always matches 7446 one of these characters. 7447 7448 The generic character type escape sequences \d, \D, \h, \H, \p, \P, \s, 7449 \S, \v, \V, \w, and \W may appear in a character class, and add the 7450 characters that they match to the class. For example, [\dABCDEF] 7451 matches any hexadecimal digit. In UTF modes, the PCRE2_UCP option af- 7452 fects the meanings of \d, \s, \w and their upper case partners, just as 7453 it does when they appear outside a character class, as described in the 7454 section entitled "Generic character types" above. The escape sequence 7455 \b has a different meaning inside a character class; it matches the 7456 backspace character. The sequences \B, \R, and \X are not special in- 7457 side a character class. Like any other unrecognized escape sequences, 7458 they cause an error. The same is true for \N when not followed by an 7459 opening brace. 7460 7461 The minus (hyphen) character can be used to specify a range of charac- 7462 ters in a character class. For example, [d-m] matches any letter be- 7463 tween d and m, inclusive. If a minus character is required in a class, 7464 it must be escaped with a backslash or appear in a position where it 7465 cannot be interpreted as indicating a range, typically as the first or 7466 last character in the class, or immediately after a range. For example, 7467 [b-d-z] matches letters in the range b to d, a hyphen character, or z. 7468 7469 Perl treats a hyphen as a literal if it appears before or after a POSIX 7470 class (see below) or before or after a character type escape such as as 7471 \d or \H. However, unless the hyphen is the last character in the 7472 class, Perl outputs a warning in its warning mode, as this is most 7473 likely a user error. As PCRE2 has no facility for warning, an error is 7474 given in these cases. 7475 7476 It is not possible to have the literal character "]" as the end charac- 7477 ter of a range. A pattern such as [W-]46] is interpreted as a class of 7478 two characters ("W" and "-") followed by a literal string "46]", so it 7479 would match "W46]" or "-46]". However, if the "]" is escaped with a 7480 backslash it is interpreted as the end of range, so [W-\]46] is inter- 7481 preted as a class containing a range followed by two other characters. 7482 The octal or hexadecimal representation of "]" can also be used to end 7483 a range. 7484 7485 Ranges normally include all code points between the start and end char- 7486 acters, inclusive. They can also be used for code points specified nu- 7487 merically, for example [\000-\037]. Ranges can include any characters 7488 that are valid for the current mode. In any UTF mode, the so-called 7489 "surrogate" characters (those whose code points lie between 0xd800 and 7490 0xdfff inclusive) may not be specified explicitly by default (the 7491 PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES option disables this check). How- 7492 ever, ranges such as [\x{d7ff}-\x{e000}], which include the surrogates, 7493 are always permitted. 7494 7495 There is a special case in EBCDIC environments for ranges whose end 7496 points are both specified as literal letters in the same case. For com- 7497 patibility with Perl, EBCDIC code points within the range that are not 7498 letters are omitted. For example, [h-k] matches only four characters, 7499 even though the codes for h and k are 0x88 and 0x92, a range of 11 code 7500 points. However, if the range is specified numerically, for example, 7501 [\x88-\x92] or [h-\x92], all code points are included. 7502 7503 If a range that includes letters is used when caseless matching is set, 7504 it matches the letters in either case. For example, [W-c] is equivalent 7505 to [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if 7506 character tables for a French locale are in use, [\xc8-\xcb] matches 7507 accented E characters in both cases. 7508 7509 A circumflex can conveniently be used with the upper case character 7510 types to specify a more restricted set of characters than the matching 7511 lower case type. For example, the class [^\W_] matches any letter or 7512 digit, but not underscore, whereas [\w] includes underscore. A positive 7513 character class should be read as "something OR something OR ..." and a 7514 negative class as "NOT something AND NOT something AND NOT ...". 7515 7516 The only metacharacters that are recognized in character classes are 7517 backslash, hyphen (only where it can be interpreted as specifying a 7518 range), circumflex (only at the start), opening square bracket (only 7519 when it can be interpreted as introducing a POSIX class name, or for a 7520 special compatibility feature - see the next two sections), and the 7521 terminating closing square bracket. However, escaping other non-al- 7522 phanumeric characters does no harm. 7523 7524 7525POSIX CHARACTER CLASSES 7526 7527 Perl supports the POSIX notation for character classes. This uses names 7528 enclosed by [: and :] within the enclosing square brackets. PCRE2 also 7529 supports this notation. For example, 7530 7531 [01[:alpha:]%] 7532 7533 matches "0", "1", any alphabetic character, or "%". The supported class 7534 names are: 7535 7536 alnum letters and digits 7537 alpha letters 7538 ascii character codes 0 - 127 7539 blank space or tab only 7540 cntrl control characters 7541 digit decimal digits (same as \d) 7542 graph printing characters, excluding space 7543 lower lower case letters 7544 print printing characters, including space 7545 punct printing characters, excluding letters and digits and space 7546 space white space (the same as \s from PCRE2 8.34) 7547 upper upper case letters 7548 word "word" characters (same as \w) 7549 xdigit hexadecimal digits 7550 7551 The default "space" characters are HT (9), LF (10), VT (11), FF (12), 7552 CR (13), and space (32). If locale-specific matching is taking place, 7553 the list of space characters may be different; there may be fewer or 7554 more of them. "Space" and \s match the same set of characters. 7555 7556 The name "word" is a Perl extension, and "blank" is a GNU extension 7557 from Perl 5.8. Another Perl extension is negation, which is indicated 7558 by a ^ character after the colon. For example, 7559 7560 [12[:^digit:]] 7561 7562 matches "1", "2", or any non-digit. PCRE2 (and Perl) also recognize the 7563 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but 7564 these are not supported, and an error is given if they are encountered. 7565 7566 By default, characters with values greater than 127 do not match any of 7567 the POSIX character classes, although this may be different for charac- 7568 ters in the range 128-255 when locale-specific matching is happening. 7569 However, if the PCRE2_UCP option is passed to pcre2_compile(), some of 7570 the classes are changed so that Unicode character properties are used. 7571 This is achieved by replacing certain POSIX classes with other se- 7572 quences, as follows: 7573 7574 [:alnum:] becomes \p{Xan} 7575 [:alpha:] becomes \p{L} 7576 [:blank:] becomes \h 7577 [:cntrl:] becomes \p{Cc} 7578 [:digit:] becomes \p{Nd} 7579 [:lower:] becomes \p{Ll} 7580 [:space:] becomes \p{Xps} 7581 [:upper:] becomes \p{Lu} 7582 [:word:] becomes \p{Xwd} 7583 7584 Negated versions, such as [:^alpha:] use \P instead of \p. Three other 7585 POSIX classes are handled specially in UCP mode: 7586 7587 [:graph:] This matches characters that have glyphs that mark the page 7588 when printed. In Unicode property terms, it matches all char- 7589 acters with the L, M, N, P, S, or Cf properties, except for: 7590 7591 U+061C Arabic Letter Mark 7592 U+180E Mongolian Vowel Separator 7593 U+2066 - U+2069 Various "isolate"s 7594 7595 7596 [:print:] This matches the same characters as [:graph:] plus space 7597 characters that are not controls, that is, characters with 7598 the Zs property. 7599 7600 [:punct:] This matches all characters that have the Unicode P (punctua- 7601 tion) property, plus those characters with code points less 7602 than 256 that have the S (Symbol) property. 7603 7604 The other POSIX classes are unchanged, and match only characters with 7605 code points less than 256. 7606 7607 7608COMPATIBILITY FEATURE FOR WORD BOUNDARIES 7609 7610 In the POSIX.2 compliant library that was included in 4.4BSD Unix, the 7611 ugly syntax [[:<:]] and [[:>:]] is used for matching "start of word" 7612 and "end of word". PCRE2 treats these items as follows: 7613 7614 [[:<:]] is converted to \b(?=\w) 7615 [[:>:]] is converted to \b(?<=\w) 7616 7617 Only these exact character sequences are recognized. A sequence such as 7618 [a[:<:]b] provokes error for an unrecognized POSIX class name. This 7619 support is not compatible with Perl. It is provided to help migrations 7620 from other environments, and is best not used in any new patterns. Note 7621 that \b matches at the start and the end of a word (see "Simple asser- 7622 tions" above), and in a Perl-style pattern the preceding or following 7623 character normally shows which is wanted, without the need for the as- 7624 sertions that are used above in order to give exactly the POSIX behav- 7625 iour. 7626 7627 7628VERTICAL BAR 7629 7630 Vertical bar characters are used to separate alternative patterns. For 7631 example, the pattern 7632 7633 gilbert|sullivan 7634 7635 matches either "gilbert" or "sullivan". Any number of alternatives may 7636 appear, and an empty alternative is permitted (matching the empty 7637 string). The matching process tries each alternative in turn, from left 7638 to right, and the first one that succeeds is used. If the alternatives 7639 are within a group (defined below), "succeeds" means matching the rest 7640 of the main pattern as well as the alternative in the group. 7641 7642 7643INTERNAL OPTION SETTING 7644 7645 The settings of the PCRE2_CASELESS, PCRE2_MULTILINE, PCRE2_DOTALL, 7646 PCRE2_EXTENDED, PCRE2_EXTENDED_MORE, and PCRE2_NO_AUTO_CAPTURE options 7647 can be changed from within the pattern by a sequence of letters en- 7648 closed between "(?" and ")". These options are Perl-compatible, and 7649 are described in detail in the pcre2api documentation. The option let- 7650 ters are: 7651 7652 i for PCRE2_CASELESS 7653 m for PCRE2_MULTILINE 7654 n for PCRE2_NO_AUTO_CAPTURE 7655 s for PCRE2_DOTALL 7656 x for PCRE2_EXTENDED 7657 xx for PCRE2_EXTENDED_MORE 7658 7659 For example, (?im) sets caseless, multiline matching. It is also possi- 7660 ble to unset these options by preceding the relevant letters with a hy- 7661 phen, for example (?-im). The two "extended" options are not indepen- 7662 dent; unsetting either one cancels the effects of both of them. 7663 7664 A combined setting and unsetting such as (?im-sx), which sets 7665 PCRE2_CASELESS and PCRE2_MULTILINE while unsetting PCRE2_DOTALL and 7666 PCRE2_EXTENDED, is also permitted. Only one hyphen may appear in the 7667 options string. If a letter appears both before and after the hyphen, 7668 the option is unset. An empty options setting "(?)" is allowed. Need- 7669 less to say, it has no effect. 7670 7671 If the first character following (? is a circumflex, it causes all of 7672 the above options to be unset. Thus, (?^) is equivalent to (?-imnsx). 7673 Letters may follow the circumflex to cause some options to be re-in- 7674 stated, but a hyphen may not appear. 7675 7676 The PCRE2-specific options PCRE2_DUPNAMES and PCRE2_UNGREEDY can be 7677 changed in the same way as the Perl-compatible options by using the 7678 characters J and U respectively. However, these are not unset by (?^). 7679 7680 When one of these option changes occurs at top level (that is, not in- 7681 side group parentheses), the change applies to the remainder of the 7682 pattern that follows. An option change within a group (see below for a 7683 description of groups) affects only that part of the group that follows 7684 it, so 7685 7686 (a(?i)b)c 7687 7688 matches abc and aBc and no other strings (assuming PCRE2_CASELESS is 7689 not used). By this means, options can be made to have different set- 7690 tings in different parts of the pattern. Any changes made in one alter- 7691 native do carry on into subsequent branches within the same group. For 7692 example, 7693 7694 (a(?i)b|c) 7695 7696 matches "ab", "aB", "c", and "C", even though when matching "C" the 7697 first branch is abandoned before the option setting. This is because 7698 the effects of option settings happen at compile time. There would be 7699 some very weird behaviour otherwise. 7700 7701 As a convenient shorthand, if any option settings are required at the 7702 start of a non-capturing group (see the next section), the option let- 7703 ters may appear between the "?" and the ":". Thus the two patterns 7704 7705 (?i:saturday|sunday) 7706 (?:(?i)saturday|sunday) 7707 7708 match exactly the same set of strings. 7709 7710 Note: There are other PCRE2-specific options, applying to the whole 7711 pattern, which can be set by the application when the compiling func- 7712 tion is called. In addition, the pattern can contain special leading 7713 sequences such as (*CRLF) to override what the application has set or 7714 what has been defaulted. Details are given in the section entitled 7715 "Newline sequences" above. There are also the (*UTF) and (*UCP) leading 7716 sequences that can be used to set UTF and Unicode property modes; they 7717 are equivalent to setting the PCRE2_UTF and PCRE2_UCP options, respec- 7718 tively. However, the application can set the PCRE2_NEVER_UTF and 7719 PCRE2_NEVER_UCP options, which lock out the use of the (*UTF) and 7720 (*UCP) sequences. 7721 7722 7723GROUPS 7724 7725 Groups are delimited by parentheses (round brackets), which can be 7726 nested. Turning part of a pattern into a group does two things: 7727 7728 1. It localizes a set of alternatives. For example, the pattern 7729 7730 cat(aract|erpillar|) 7731 7732 matches "cataract", "caterpillar", or "cat". Without the parentheses, 7733 it would match "cataract", "erpillar" or an empty string. 7734 7735 2. It creates a "capture group". This means that, when the whole pat- 7736 tern matches, the portion of the subject string that matched the group 7737 is passed back to the caller, separately from the portion that matched 7738 the whole pattern. (This applies only to the traditional matching 7739 function; the DFA matching function does not support capturing.) 7740 7741 Opening parentheses are counted from left to right (starting from 1) to 7742 obtain numbers for capture groups. For example, if the string "the red 7743 king" is matched against the pattern 7744 7745 the ((red|white) (king|queen)) 7746 7747 the captured substrings are "red king", "red", and "king", and are num- 7748 bered 1, 2, and 3, respectively. 7749 7750 The fact that plain parentheses fulfil two functions is not always 7751 helpful. There are often times when grouping is required without cap- 7752 turing. If an opening parenthesis is followed by a question mark and a 7753 colon, the group does not do any capturing, and is not counted when 7754 computing the number of any subsequent capture groups. For example, if 7755 the string "the white queen" is matched against the pattern 7756 7757 the ((?:red|white) (king|queen)) 7758 7759 the captured substrings are "white queen" and "queen", and are numbered 7760 1 and 2. The maximum number of capture groups is 65535. 7761 7762 As a convenient shorthand, if any option settings are required at the 7763 start of a non-capturing group, the option letters may appear between 7764 the "?" and the ":". Thus the two patterns 7765 7766 (?i:saturday|sunday) 7767 (?:(?i)saturday|sunday) 7768 7769 match exactly the same set of strings. Because alternative branches are 7770 tried from left to right, and options are not reset until the end of 7771 the group is reached, an option setting in one branch does affect sub- 7772 sequent branches, so the above patterns match "SUNDAY" as well as "Sat- 7773 urday". 7774 7775 7776DUPLICATE GROUP NUMBERS 7777 7778 Perl 5.10 introduced a feature whereby each alternative in a group uses 7779 the same numbers for its capturing parentheses. Such a group starts 7780 with (?| and is itself a non-capturing group. For example, consider 7781 this pattern: 7782 7783 (?|(Sat)ur|(Sun))day 7784 7785 Because the two alternatives are inside a (?| group, both sets of cap- 7786 turing parentheses are numbered one. Thus, when the pattern matches, 7787 you can look at captured substring number one, whichever alternative 7788 matched. This construct is useful when you want to capture part, but 7789 not all, of one of a number of alternatives. Inside a (?| group, paren- 7790 theses are numbered as usual, but the number is reset at the start of 7791 each branch. The numbers of any capturing parentheses that follow the 7792 whole group start after the highest number used in any branch. The fol- 7793 lowing example is taken from the Perl documentation. The numbers under- 7794 neath show in which buffer the captured content will be stored. 7795 7796 # before ---------------branch-reset----------- after 7797 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x 7798 # 1 2 2 3 2 3 4 7799 7800 A backreference to a capture group uses the most recent value that is 7801 set for the group. The following pattern matches "abcabc" or "defdef": 7802 7803 /(?|(abc)|(def))\1/ 7804 7805 In contrast, a subroutine call to a capture group always refers to the 7806 first one in the pattern with the given number. The following pattern 7807 matches "abcabc" or "defabc": 7808 7809 /(?|(abc)|(def))(?1)/ 7810 7811 A relative reference such as (?-1) is no different: it is just a conve- 7812 nient way of computing an absolute group number. 7813 7814 If a condition test for a group's having matched refers to a non-unique 7815 number, the test is true if any group with that number has matched. 7816 7817 An alternative approach to using this "branch reset" feature is to use 7818 duplicate named groups, as described in the next section. 7819 7820 7821NAMED CAPTURE GROUPS 7822 7823 Identifying capture groups by number is simple, but it can be very hard 7824 to keep track of the numbers in complicated patterns. Furthermore, if 7825 an expression is modified, the numbers may change. To help with this 7826 difficulty, PCRE2 supports the naming of capture groups. This feature 7827 was not added to Perl until release 5.10. Python had the feature ear- 7828 lier, and PCRE1 introduced it at release 4.0, using the Python syntax. 7829 PCRE2 supports both the Perl and the Python syntax. 7830 7831 In PCRE2, a capture group can be named in one of three ways: 7832 (?<name>...) or (?'name'...) as in Perl, or (?P<name>...) as in Python. 7833 Names may be up to 32 code units long. When PCRE2_UTF is not set, they 7834 may contain only ASCII alphanumeric characters and underscores, but 7835 must start with a non-digit. When PCRE2_UTF is set, the syntax of group 7836 names is extended to allow any Unicode letter or Unicode decimal digit. 7837 In other words, group names must match one of these patterns: 7838 7839 ^[_A-Za-z][_A-Za-z0-9]*\z when PCRE2_UTF is not set 7840 ^[_\p{L}][_\p{L}\p{Nd}]*\z when PCRE2_UTF is set 7841 7842 References to capture groups from other parts of the pattern, such as 7843 backreferences, recursion, and conditions, can all be made by name as 7844 well as by number. 7845 7846 Named capture groups are allocated numbers as well as names, exactly as 7847 if the names were not present. In both PCRE2 and Perl, capture groups 7848 are primarily identified by numbers; any names are just aliases for 7849 these numbers. The PCRE2 API provides function calls for extracting the 7850 complete name-to-number translation table from a compiled pattern, as 7851 well as convenience functions for extracting captured substrings by 7852 name. 7853 7854 Warning: When more than one capture group has the same number, as de- 7855 scribed in the previous section, a name given to one of them applies to 7856 all of them. Perl allows identically numbered groups to have different 7857 names. Consider this pattern, where there are two capture groups, both 7858 numbered 1: 7859 7860 (?|(?<AA>aa)|(?<BB>bb)) 7861 7862 Perl allows this, with both names AA and BB as aliases of group 1. 7863 Thus, after a successful match, both names yield the same value (either 7864 "aa" or "bb"). 7865 7866 In an attempt to reduce confusion, PCRE2 does not allow the same group 7867 number to be associated with more than one name. The example above pro- 7868 vokes a compile-time error. However, there is still scope for confu- 7869 sion. Consider this pattern: 7870 7871 (?|(?<AA>aa)|(bb)) 7872 7873 Although the second group number 1 is not explicitly named, the name AA 7874 is still an alias for any group 1. Whether the pattern matches "aa" or 7875 "bb", a reference by name to group AA yields the matched string. 7876 7877 By default, a name must be unique within a pattern, except that dupli- 7878 cate names are permitted for groups with the same number, for example: 7879 7880 (?|(?<AA>aa)|(?<AA>bb)) 7881 7882 The duplicate name constraint can be disabled by setting the PCRE2_DUP- 7883 NAMES option at compile time, or by the use of (?J) within the pattern, 7884 as described in the section entitled "Internal Option Setting" above. 7885 7886 Duplicate names can be useful for patterns where only one instance of 7887 the named capture group can match. Suppose you want to match the name 7888 of a weekday, either as a 3-letter abbreviation or as the full name, 7889 and in both cases you want to extract the abbreviation. This pattern 7890 (ignoring the line breaks) does the job: 7891 7892 (?J) 7893 (?<DN>Mon|Fri|Sun)(?:day)?| 7894 (?<DN>Tue)(?:sday)?| 7895 (?<DN>Wed)(?:nesday)?| 7896 (?<DN>Thu)(?:rsday)?| 7897 (?<DN>Sat)(?:urday)? 7898 7899 There are five capture groups, but only one is ever set after a match. 7900 The convenience functions for extracting the data by name returns the 7901 substring for the first (and in this example, the only) group of that 7902 name that matched. This saves searching to find which numbered group it 7903 was. (An alternative way of solving this problem is to use a "branch 7904 reset" group, as described in the previous section.) 7905 7906 If you make a backreference to a non-unique named group from elsewhere 7907 in the pattern, the groups to which the name refers are checked in the 7908 order in which they appear in the overall pattern. The first one that 7909 is set is used for the reference. For example, this pattern matches 7910 both "foofoo" and "barbar" but not "foobar" or "barfoo": 7911 7912 (?J)(?:(?<n>foo)|(?<n>bar))\k<n> 7913 7914 7915 If you make a subroutine call to a non-unique named group, the one that 7916 corresponds to the first occurrence of the name is used. In the absence 7917 of duplicate numbers this is the one with the lowest number. 7918 7919 If you use a named reference in a condition test (see the section about 7920 conditions below), either to check whether a capture group has matched, 7921 or to check for recursion, all groups with the same name are tested. If 7922 the condition is true for any one of them, the overall condition is 7923 true. This is the same behaviour as testing by number. For further de- 7924 tails of the interfaces for handling named capture groups, see the 7925 pcre2api documentation. 7926 7927 7928REPETITION 7929 7930 Repetition is specified by quantifiers, which can follow any of the 7931 following items: 7932 7933 a literal data character 7934 the dot metacharacter 7935 the \C escape sequence 7936 the \R escape sequence 7937 the \X escape sequence 7938 an escape such as \d or \pL that matches a single character 7939 a character class 7940 a backreference 7941 a parenthesized group (including lookaround assertions) 7942 a subroutine call (recursive or otherwise) 7943 7944 The general repetition quantifier specifies a minimum and maximum num- 7945 ber of permitted matches, by giving the two numbers in curly brackets 7946 (braces), separated by a comma. The numbers must be less than 65536, 7947 and the first must be less than or equal to the second. For example, 7948 7949 z{2,4} 7950 7951 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a 7952 special character. If the second number is omitted, but the comma is 7953 present, there is no upper limit; if the second number and the comma 7954 are both omitted, the quantifier specifies an exact number of required 7955 matches. Thus 7956 7957 [aeiou]{3,} 7958 7959 matches at least 3 successive vowels, but may match many more, whereas 7960 7961 \d{8} 7962 7963 matches exactly 8 digits. An opening curly bracket that appears in a 7964 position where a quantifier is not allowed, or one that does not match 7965 the syntax of a quantifier, is taken as a literal character. For exam- 7966 ple, {,6} is not a quantifier, but a literal string of four characters. 7967 7968 In UTF modes, quantifiers apply to characters rather than to individual 7969 code units. Thus, for example, \x{100}{2} matches two characters, each 7970 of which is represented by a two-byte sequence in a UTF-8 string. Simi- 7971 larly, \X{3} matches three Unicode extended grapheme clusters, each of 7972 which may be several code units long (and they may be of different 7973 lengths). 7974 7975 The quantifier {0} is permitted, causing the expression to behave as if 7976 the previous item and the quantifier were not present. This may be use- 7977 ful for capture groups that are referenced as subroutines from else- 7978 where in the pattern (but see also the section entitled "Defining cap- 7979 ture groups for use by reference only" below). Except for parenthesized 7980 groups, items that have a {0} quantifier are omitted from the compiled 7981 pattern. 7982 7983 For convenience, the three most common quantifiers have single-charac- 7984 ter abbreviations: 7985 7986 * is equivalent to {0,} 7987 + is equivalent to {1,} 7988 ? is equivalent to {0,1} 7989 7990 It is possible to construct infinite loops by following a group that 7991 can match no characters with a quantifier that has no upper limit, for 7992 example: 7993 7994 (a?)* 7995 7996 Earlier versions of Perl and PCRE1 used to give an error at compile 7997 time for such patterns. However, because there are cases where this can 7998 be useful, such patterns are now accepted, but whenever an iteration of 7999 such a group matches no characters, matching moves on to the next item 8000 in the pattern instead of repeatedly matching an empty string. This 8001 does not prevent backtracking into any of the iterations if a subse- 8002 quent item fails to match. 8003 8004 By default, quantifiers are "greedy", that is, they match as much as 8005 possible (up to the maximum number of permitted times), without causing 8006 the rest of the pattern to fail. The classic example of where this 8007 gives problems is in trying to match comments in C programs. These ap- 8008 pear between /* and */ and within the comment, individual * and / char- 8009 acters may appear. An attempt to match C comments by applying the pat- 8010 tern 8011 8012 /\*.*\*/ 8013 8014 to the string 8015 8016 /* first comment */ not comment /* second comment */ 8017 8018 fails, because it matches the entire string owing to the greediness of 8019 the .* item. However, if a quantifier is followed by a question mark, 8020 it ceases to be greedy, and instead matches the minimum number of times 8021 possible, so the pattern 8022 8023 /\*.*?\*/ 8024 8025 does the right thing with the C comments. The meaning of the various 8026 quantifiers is not otherwise changed, just the preferred number of 8027 matches. Do not confuse this use of question mark with its use as a 8028 quantifier in its own right. Because it has two uses, it can sometimes 8029 appear doubled, as in 8030 8031 \d??\d 8032 8033 which matches one digit by preference, but can match two if that is the 8034 only way the rest of the pattern matches. 8035 8036 If the PCRE2_UNGREEDY option is set (an option that is not available in 8037 Perl), the quantifiers are not greedy by default, but individual ones 8038 can be made greedy by following them with a question mark. In other 8039 words, it inverts the default behaviour. 8040 8041 When a parenthesized group is quantified with a minimum repeat count 8042 that is greater than 1 or with a limited maximum, more memory is re- 8043 quired for the compiled pattern, in proportion to the size of the mini- 8044 mum or maximum. 8045 8046 If a pattern starts with .* or .{0,} and the PCRE2_DOTALL option 8047 (equivalent to Perl's /s) is set, thus allowing the dot to match new- 8048 lines, the pattern is implicitly anchored, because whatever follows 8049 will be tried against every character position in the subject string, 8050 so there is no point in retrying the overall match at any position af- 8051 ter the first. PCRE2 normally treats such a pattern as though it were 8052 preceded by \A. 8053 8054 In cases where it is known that the subject string contains no new- 8055 lines, it is worth setting PCRE2_DOTALL in order to obtain this opti- 8056 mization, or alternatively, using ^ to indicate anchoring explicitly. 8057 8058 However, there are some cases where the optimization cannot be used. 8059 When .* is inside capturing parentheses that are the subject of a 8060 backreference elsewhere in the pattern, a match at the start may fail 8061 where a later one succeeds. Consider, for example: 8062 8063 (.*)abc\1 8064 8065 If the subject is "xyz123abc123" the match point is the fourth charac- 8066 ter. For this reason, such a pattern is not implicitly anchored. 8067 8068 Another case where implicit anchoring is not applied is when the lead- 8069 ing .* is inside an atomic group. Once again, a match at the start may 8070 fail where a later one succeeds. Consider this pattern: 8071 8072 (?>.*?a)b 8073 8074 It matches "ab" in the subject "aab". The use of the backtracking con- 8075 trol verbs (*PRUNE) and (*SKIP) also disable this optimization, and 8076 there is an option, PCRE2_NO_DOTSTAR_ANCHOR, to do so explicitly. 8077 8078 When a capture group is repeated, the value captured is the substring 8079 that matched the final iteration. For example, after 8080 8081 (tweedle[dume]{3}\s*)+ 8082 8083 has matched "tweedledum tweedledee" the value of the captured substring 8084 is "tweedledee". However, if there are nested capture groups, the cor- 8085 responding captured values may have been set in previous iterations. 8086 For example, after 8087 8088 (a|(b))+ 8089 8090 matches "aba" the value of the second captured substring is "b". 8091 8092 8093ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS 8094 8095 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy") 8096 repetition, failure of what follows normally causes the repeated item 8097 to be re-evaluated to see if a different number of repeats allows the 8098 rest of the pattern to match. Sometimes it is useful to prevent this, 8099 either to change the nature of the match, or to cause it fail earlier 8100 than it otherwise might, when the author of the pattern knows there is 8101 no point in carrying on. 8102 8103 Consider, for example, the pattern \d+foo when applied to the subject 8104 line 8105 8106 123456bar 8107 8108 After matching all 6 digits and then failing to match "foo", the normal 8109 action of the matcher is to try again with only 5 digits matching the 8110 \d+ item, and then with 4, and so on, before ultimately failing. 8111 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides 8112 the means for specifying that once a group has matched, it is not to be 8113 re-evaluated in this way. 8114 8115 If we use atomic grouping for the previous example, the matcher gives 8116 up immediately on failing to match "foo" the first time. The notation 8117 is a kind of special parenthesis, starting with (?> as in this example: 8118 8119 (?>\d+)foo 8120 8121 Perl 5.28 introduced an experimental alphabetic form starting with (* 8122 which may be easier to remember: 8123 8124 (*atomic:\d+)foo 8125 8126 This kind of parenthesized group "locks up" the part of the pattern it 8127 contains once it has matched, and a failure further into the pattern is 8128 prevented from backtracking into it. Backtracking past it to previous 8129 items, however, works as normal. 8130 8131 An alternative description is that a group of this type matches exactly 8132 the string of characters that an identical standalone pattern would 8133 match, if anchored at the current point in the subject string. 8134 8135 Atomic groups are not capture groups. Simple cases such as the above 8136 example can be thought of as a maximizing repeat that must swallow ev- 8137 erything it can. So, while both \d+ and \d+? are prepared to adjust 8138 the number of digits they match in order to make the rest of the pat- 8139 tern match, (?>\d+) can only match an entire sequence of digits. 8140 8141 Atomic groups in general can of course contain arbitrarily complicated 8142 expressions, and can be nested. However, when the contents of an atomic 8143 group is just a single repeated item, as in the example above, a sim- 8144 pler notation, called a "possessive quantifier" can be used. This con- 8145 sists of an additional + character following a quantifier. Using this 8146 notation, the previous example can be rewritten as 8147 8148 \d++foo 8149 8150 Note that a possessive quantifier can be used with an entire group, for 8151 example: 8152 8153 (abc|xyz){2,3}+ 8154 8155 Possessive quantifiers are always greedy; the setting of the PCRE2_UN- 8156 GREEDY option is ignored. They are a convenient notation for the sim- 8157 pler forms of atomic group. However, there is no difference in the 8158 meaning of a possessive quantifier and the equivalent atomic group, 8159 though there may be a performance difference; possessive quantifiers 8160 should be slightly faster. 8161 8162 The possessive quantifier syntax is an extension to the Perl 5.8 syn- 8163 tax. Jeffrey Friedl originated the idea (and the name) in the first 8164 edition of his book. Mike McCloskey liked it, so implemented it when he 8165 built Sun's Java package, and PCRE1 copied it from there. It found its 8166 way into Perl at release 5.10. 8167 8168 PCRE2 has an optimization that automatically "possessifies" certain 8169 simple pattern constructs. For example, the sequence A+B is treated as 8170 A++B because there is no point in backtracking into a sequence of A's 8171 when B must follow. This feature can be disabled by the PCRE2_NO_AUTO- 8172 POSSESS option, or starting the pattern with (*NO_AUTO_POSSESS). 8173 8174 When a pattern contains an unlimited repeat inside a group that can it- 8175 self be repeated an unlimited number of times, the use of an atomic 8176 group is the only way to avoid some failing matches taking a very long 8177 time indeed. The pattern 8178 8179 (\D+|<\d+>)*[!?] 8180 8181 matches an unlimited number of substrings that either consist of non- 8182 digits, or digits enclosed in <>, followed by either ! or ?. When it 8183 matches, it runs quickly. However, if it is applied to 8184 8185 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa 8186 8187 it takes a long time before reporting failure. This is because the 8188 string can be divided between the internal \D+ repeat and the external 8189 * repeat in a large number of ways, and all have to be tried. (The ex- 8190 ample uses [!?] rather than a single character at the end, because both 8191 PCRE2 and Perl have an optimization that allows for fast failure when a 8192 single character is used. They remember the last single character that 8193 is required for a match, and fail early if it is not present in the 8194 string.) If the pattern is changed so that it uses an atomic group, 8195 like this: 8196 8197 ((?>\D+)|<\d+>)*[!?] 8198 8199 sequences of non-digits cannot be broken, and failure happens quickly. 8200 8201 8202BACKREFERENCES 8203 8204 Outside a character class, a backslash followed by a digit greater than 8205 0 (and possibly further digits) is a backreference to a capture group 8206 earlier (that is, to its left) in the pattern, provided there have been 8207 that many previous capture groups. 8208 8209 However, if the decimal number following the backslash is less than 8, 8210 it is always taken as a backreference, and causes an error only if 8211 there are not that many capture groups in the entire pattern. In other 8212 words, the group that is referenced need not be to the left of the ref- 8213 erence for numbers less than 8. A "forward backreference" of this type 8214 can make sense when a repetition is involved and the group to the right 8215 has participated in an earlier iteration. 8216 8217 It is not possible to have a numerical "forward backreference" to a 8218 group whose number is 8 or more using this syntax because a sequence 8219 such as \50 is interpreted as a character defined in octal. See the 8220 subsection entitled "Non-printing characters" above for further details 8221 of the handling of digits following a backslash. Other forms of back- 8222 referencing do not suffer from this restriction. In particular, there 8223 is no problem when named capture groups are used (see below). 8224 8225 Another way of avoiding the ambiguity inherent in the use of digits 8226 following a backslash is to use the \g escape sequence. This escape 8227 must be followed by a signed or unsigned number, optionally enclosed in 8228 braces. These examples are all identical: 8229 8230 (ring), \1 8231 (ring), \g1 8232 (ring), \g{1} 8233 8234 An unsigned number specifies an absolute reference without the ambigu- 8235 ity that is present in the older syntax. It is also useful when literal 8236 digits follow the reference. A signed number is a relative reference. 8237 Consider this example: 8238 8239 (abc(def)ghi)\g{-1} 8240 8241 The sequence \g{-1} is a reference to the most recently started capture 8242 group before \g, that is, is it equivalent to \2 in this example. Simi- 8243 larly, \g{-2} would be equivalent to \1. The use of relative references 8244 can be helpful in long patterns, and also in patterns that are created 8245 by joining together fragments that contain references within them- 8246 selves. 8247 8248 The sequence \g{+1} is a reference to the next capture group. This kind 8249 of forward reference can be useful in patterns that repeat. Perl does 8250 not support the use of + in this way. 8251 8252 A backreference matches whatever actually most recently matched the 8253 capture group in the current subject string, rather than anything at 8254 all that matches the group (see "Groups as subroutines" below for a way 8255 of doing that). So the pattern 8256 8257 (sens|respons)e and \1ibility 8258 8259 matches "sense and sensibility" and "response and responsibility", but 8260 not "sense and responsibility". If caseful matching is in force at the 8261 time of the backreference, the case of letters is relevant. For exam- 8262 ple, 8263 8264 ((?i)rah)\s+\1 8265 8266 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the 8267 original capture group is matched caselessly. 8268 8269 There are several different ways of writing backreferences to named 8270 capture groups. The .NET syntax \k{name} and the Perl syntax \k<name> 8271 or \k'name' are supported, as is the Python syntax (?P=name). Perl 8272 5.10's unified backreference syntax, in which \g can be used for both 8273 numeric and named references, is also supported. We could rewrite the 8274 above example in any of the following ways: 8275 8276 (?<p1>(?i)rah)\s+\k<p1> 8277 (?'p1'(?i)rah)\s+\k{p1} 8278 (?P<p1>(?i)rah)\s+(?P=p1) 8279 (?<p1>(?i)rah)\s+\g{p1} 8280 8281 A capture group that is referenced by name may appear in the pattern 8282 before or after the reference. 8283 8284 There may be more than one backreference to the same group. If a group 8285 has not actually been used in a particular match, backreferences to it 8286 always fail by default. For example, the pattern 8287 8288 (a|(bc))\2 8289 8290 always fails if it starts to match "a" rather than "bc". However, if 8291 the PCRE2_MATCH_UNSET_BACKREF option is set at compile time, a backref- 8292 erence to an unset value matches an empty string. 8293 8294 Because there may be many capture groups in a pattern, all digits fol- 8295 lowing a backslash are taken as part of a potential backreference num- 8296 ber. If the pattern continues with a digit character, some delimiter 8297 must be used to terminate the backreference. If the PCRE2_EXTENDED or 8298 PCRE2_EXTENDED_MORE option is set, this can be white space. Otherwise, 8299 the \g{} syntax or an empty comment (see "Comments" below) can be used. 8300 8301 Recursive backreferences 8302 8303 A backreference that occurs inside the group to which it refers fails 8304 when the group is first used, so, for example, (a\1) never matches. 8305 However, such references can be useful inside repeated groups. For ex- 8306 ample, the pattern 8307 8308 (a|b\1)+ 8309 8310 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter- 8311 ation of the group, the backreference matches the character string cor- 8312 responding to the previous iteration. In order for this to work, the 8313 pattern must be such that the first iteration does not need to match 8314 the backreference. This can be done using alternation, as in the exam- 8315 ple above, or by a quantifier with a minimum of zero. 8316 8317 For versions of PCRE2 less than 10.25, backreferences of this type used 8318 to cause the group that they reference to be treated as an atomic 8319 group. This restriction no longer applies, and backtracking into such 8320 groups can occur as normal. 8321 8322 8323ASSERTIONS 8324 8325 An assertion is a test on the characters following or preceding the 8326 current matching point that does not consume any characters. The simple 8327 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described 8328 above. 8329 8330 More complicated assertions are coded as parenthesized groups. There 8331 are two kinds: those that look ahead of the current position in the 8332 subject string, and those that look behind it, and in each case an as- 8333 sertion may be positive (must match for the assertion to be true) or 8334 negative (must not match for the assertion to be true). An assertion 8335 group is matched in the normal way, and if it is true, matching contin- 8336 ues after it, but with the matching position in the subject string re- 8337 set to what it was before the assertion was processed. 8338 8339 The Perl-compatible lookaround assertions are atomic. If an assertion 8340 is true, but there is a subsequent matching failure, there is no back- 8341 tracking into the assertion. However, there are some cases where non- 8342 atomic assertions can be useful. PCRE2 has some support for these, de- 8343 scribed in the section entitled "Non-atomic assertions" below, but they 8344 are not Perl-compatible. 8345 8346 A lookaround assertion may appear as the condition in a conditional 8347 group (see below). In this case, the result of matching the assertion 8348 determines which branch of the condition is followed. 8349 8350 Assertion groups are not capture groups. If an assertion contains cap- 8351 ture groups within it, these are counted for the purposes of numbering 8352 the capture groups in the whole pattern. Within each branch of an as- 8353 sertion, locally captured substrings may be referenced in the usual 8354 way. For example, a sequence such as (.)\g{-1} can be used to check 8355 that two adjacent characters are the same. 8356 8357 When a branch within an assertion fails to match, any substrings that 8358 were captured are discarded (as happens with any pattern branch that 8359 fails to match). A negative assertion is true only when all its 8360 branches fail to match; this means that no captured substrings are ever 8361 retained after a successful negative assertion. When an assertion con- 8362 tains a matching branch, what happens depends on the type of assertion. 8363 8364 For a positive assertion, internally captured substrings in the suc- 8365 cessful branch are retained, and matching continues with the next pat- 8366 tern item after the assertion. For a negative assertion, a matching 8367 branch means that the assertion is not true. If such an assertion is 8368 being used as a condition in a conditional group (see below), captured 8369 substrings are retained, because matching continues with the "no" 8370 branch of the condition. For other failing negative assertions, control 8371 passes to the previous backtracking point, thus discarding any captured 8372 strings within the assertion. 8373 8374 Most assertion groups may be repeated; though it makes no sense to as- 8375 sert the same thing several times, the side effect of capturing in pos- 8376 itive assertions may occasionally be useful. However, an assertion that 8377 forms the condition for a conditional group may not be quantified. 8378 PCRE2 used to restrict the repetition of assertions, but from release 8379 10.35 the only restriction is that an unlimited maximum repetition is 8380 changed to be one more than the minimum. For example, {3,} is treated 8381 as {3,4}. 8382 8383 Alphabetic assertion names 8384 8385 Traditionally, symbolic sequences such as (?= and (?<= have been used 8386 to specify lookaround assertions. Perl 5.28 introduced some experimen- 8387 tal alphabetic alternatives which might be easier to remember. They all 8388 start with (* instead of (? and must be written using lower case let- 8389 ters. PCRE2 supports the following synonyms: 8390 8391 (*positive_lookahead: or (*pla: is the same as (?= 8392 (*negative_lookahead: or (*nla: is the same as (?! 8393 (*positive_lookbehind: or (*plb: is the same as (?<= 8394 (*negative_lookbehind: or (*nlb: is the same as (?<! 8395 8396 For example, (*pla:foo) is the same assertion as (?=foo). In the fol- 8397 lowing sections, the various assertions are described using the origi- 8398 nal symbolic forms. 8399 8400 Lookahead assertions 8401 8402 Lookahead assertions start with (?= for positive assertions and (?! for 8403 negative assertions. For example, 8404 8405 \w+(?=;) 8406 8407 matches a word followed by a semicolon, but does not include the semi- 8408 colon in the match, and 8409 8410 foo(?!bar) 8411 8412 matches any occurrence of "foo" that is not followed by "bar". Note 8413 that the apparently similar pattern 8414 8415 (?!foo)bar 8416 8417 does not find an occurrence of "bar" that is preceded by something 8418 other than "foo"; it finds any occurrence of "bar" whatsoever, because 8419 the assertion (?!foo) is always true when the next three characters are 8420 "bar". A lookbehind assertion is needed to achieve the other effect. 8421 8422 If you want to force a matching failure at some point in a pattern, the 8423 most convenient way to do it is with (?!) because an empty string al- 8424 ways matches, so an assertion that requires there not to be an empty 8425 string must always fail. The backtracking control verb (*FAIL) or (*F) 8426 is a synonym for (?!). 8427 8428 Lookbehind assertions 8429 8430 Lookbehind assertions start with (?<= for positive assertions and (?<! 8431 for negative assertions. For example, 8432 8433 (?<!foo)bar 8434 8435 does find an occurrence of "bar" that is not preceded by "foo". The 8436 contents of a lookbehind assertion are restricted such that all the 8437 strings it matches must have a fixed length. However, if there are sev- 8438 eral top-level alternatives, they do not all have to have the same 8439 fixed length. Thus 8440 8441 (?<=bullock|donkey) 8442 8443 is permitted, but 8444 8445 (?<!dogs?|cats?) 8446 8447 causes an error at compile time. Branches that match different length 8448 strings are permitted only at the top level of a lookbehind assertion. 8449 This is an extension compared with Perl, which requires all branches to 8450 match the same length of string. An assertion such as 8451 8452 (?<=ab(c|de)) 8453 8454 is not permitted, because its single top-level branch can match two 8455 different lengths, but it is acceptable to PCRE2 if rewritten to use 8456 two top-level branches: 8457 8458 (?<=abc|abde) 8459 8460 In some cases, the escape sequence \K (see above) can be used instead 8461 of a lookbehind assertion to get round the fixed-length restriction. 8462 8463 The implementation of lookbehind assertions is, for each alternative, 8464 to temporarily move the current position back by the fixed length and 8465 then try to match. If there are insufficient characters before the cur- 8466 rent position, the assertion fails. 8467 8468 In UTF-8 and UTF-16 modes, PCRE2 does not allow the \C escape (which 8469 matches a single code unit even in a UTF mode) to appear in lookbehind 8470 assertions, because it makes it impossible to calculate the length of 8471 the lookbehind. The \X and \R escapes, which can match different num- 8472 bers of code units, are never permitted in lookbehinds. 8473 8474 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in 8475 lookbehinds, as long as the called capture group matches a fixed-length 8476 string. However, recursion, that is, a "subroutine" call into a group 8477 that is already active, is not supported. 8478 8479 Perl does not support backreferences in lookbehinds. PCRE2 does support 8480 them, but only if certain conditions are met. The PCRE2_MATCH_UN- 8481 SET_BACKREF option must not be set, there must be no use of (?| in the 8482 pattern (it creates duplicate group numbers), and if the backreference 8483 is by name, the name must be unique. Of course, the referenced group 8484 must itself match a fixed length substring. The following pattern 8485 matches words containing at least two characters that begin and end 8486 with the same character: 8487 8488 \b(\w)\w++(?<=\1) 8489 8490 Possessive quantifiers can be used in conjunction with lookbehind as- 8491 sertions to specify efficient matching of fixed-length strings at the 8492 end of subject strings. Consider a simple pattern such as 8493 8494 abcd$ 8495 8496 when applied to a long string that does not match. Because matching 8497 proceeds from left to right, PCRE2 will look for each "a" in the sub- 8498 ject and then see if what follows matches the rest of the pattern. If 8499 the pattern is specified as 8500 8501 ^.*abcd$ 8502 8503 the initial .* matches the entire string at first, but when this fails 8504 (because there is no following "a"), it backtracks to match all but the 8505 last character, then all but the last two characters, and so on. Once 8506 again the search for "a" covers the entire string, from right to left, 8507 so we are no better off. However, if the pattern is written as 8508 8509 ^.*+(?<=abcd) 8510 8511 there can be no backtracking for the .*+ item because of the possessive 8512 quantifier; it can match only the entire string. The subsequent lookbe- 8513 hind assertion does a single test on the last four characters. If it 8514 fails, the match fails immediately. For long strings, this approach 8515 makes a significant difference to the processing time. 8516 8517 Using multiple assertions 8518 8519 Several assertions (of any sort) may occur in succession. For example, 8520 8521 (?<=\d{3})(?<!999)foo 8522 8523 matches "foo" preceded by three digits that are not "999". Notice that 8524 each of the assertions is applied independently at the same point in 8525 the subject string. First there is a check that the previous three 8526 characters are all digits, and then there is a check that the same 8527 three characters are not "999". This pattern does not match "foo" pre- 8528 ceded by six characters, the first of which are digits and the last 8529 three of which are not "999". For example, it doesn't match "123abc- 8530 foo". A pattern to do that is 8531 8532 (?<=\d{3}...)(?<!999)foo 8533 8534 This time the first assertion looks at the preceding six characters, 8535 checking that the first three are digits, and then the second assertion 8536 checks that the preceding three characters are not "999". 8537 8538 Assertions can be nested in any combination. For example, 8539 8540 (?<=(?<!foo)bar)baz 8541 8542 matches an occurrence of "baz" that is preceded by "bar" which in turn 8543 is not preceded by "foo", while 8544 8545 (?<=\d{3}(?!999)...)foo 8546 8547 is another pattern that matches "foo" preceded by three digits and any 8548 three characters that are not "999". 8549 8550 8551NON-ATOMIC ASSERTIONS 8552 8553 The traditional Perl-compatible lookaround assertions are atomic. That 8554 is, if an assertion is true, but there is a subsequent matching fail- 8555 ure, there is no backtracking into the assertion. However, there are 8556 some cases where non-atomic positive assertions can be useful. PCRE2 8557 provides these using the following syntax: 8558 8559 (*non_atomic_positive_lookahead: or (*napla: or (?* 8560 (*non_atomic_positive_lookbehind: or (*naplb: or (?<* 8561 8562 Consider the problem of finding the right-most word in a string that 8563 also appears earlier in the string, that is, it must appear at least 8564 twice in total. This pattern returns the required result as captured 8565 substring 1: 8566 8567 ^(?x)(*napla: .* \b(\w++)) (?> .*? \b\1\b ){2} 8568 8569 For a subject such as "word1 word2 word3 word2 word3 word4" the result 8570 is "word3". How does it work? At the start, ^(?x) anchors the pattern 8571 and sets the "x" option, which causes white space (introduced for read- 8572 ability) to be ignored. Inside the assertion, the greedy .* at first 8573 consumes the entire string, but then has to backtrack until the rest of 8574 the assertion can match a word, which is captured by group 1. In other 8575 words, when the assertion first succeeds, it captures the right-most 8576 word in the string. 8577 8578 The current matching point is then reset to the start of the subject, 8579 and the rest of the pattern match checks for two occurrences of the 8580 captured word, using an ungreedy .*? to scan from the left. If this 8581 succeeds, we are done, but if the last word in the string does not oc- 8582 cur twice, this part of the pattern fails. If a traditional atomic 8583 lookhead (?= or (*pla: had been used, the assertion could not be re-en- 8584 tered, and the whole match would fail. The pattern would succeed only 8585 if the very last word in the subject was found twice. 8586 8587 Using a non-atomic lookahead, however, means that when the last word 8588 does not occur twice in the string, the lookahead can backtrack and 8589 find the second-last word, and so on, until either the match succeeds, 8590 or all words have been tested. 8591 8592 Two conditions must be met for a non-atomic assertion to be useful: the 8593 contents of one or more capturing groups must change after a backtrack 8594 into the assertion, and there must be a backreference to a changed 8595 group later in the pattern. If this is not the case, the rest of the 8596 pattern match fails exactly as before because nothing has changed, so 8597 using a non-atomic assertion just wastes resources. 8598 8599 There is one exception to backtracking into a non-atomic assertion. If 8600 an (*ACCEPT) control verb is triggered, the assertion succeeds atomi- 8601 cally. That is, a subsequent match failure cannot backtrack into the 8602 assertion. 8603 8604 Non-atomic assertions are not supported by the alternative matching 8605 function pcre2_dfa_match(). They are supported by JIT, but only if they 8606 do not contain any control verbs such as (*ACCEPT). (This may change in 8607 future). Note that assertions that appear as conditions for conditional 8608 groups (see below) must be atomic. 8609 8610 8611SCRIPT RUNS 8612 8613 In concept, a script run is a sequence of characters that are all from 8614 the same Unicode script such as Latin or Greek. However, because some 8615 scripts are commonly used together, and because some diacritical and 8616 other marks are used with multiple scripts, it is not that simple. 8617 There is a full description of the rules that PCRE2 uses in the section 8618 entitled "Script Runs" in the pcre2unicode documentation. 8619 8620 If part of a pattern is enclosed between (*script_run: or (*sr: and a 8621 closing parenthesis, it fails if the sequence of characters that it 8622 matches are not a script run. After a failure, normal backtracking oc- 8623 curs. Script runs can be used to detect spoofing attacks using charac- 8624 ters that look the same, but are from different scripts. The string 8625 "paypal.com" is an infamous example, where the letters could be a mix- 8626 ture of Latin and Cyrillic. This pattern ensures that the matched char- 8627 acters in a sequence of non-spaces that follow white space are a script 8628 run: 8629 8630 \s+(*sr:\S+) 8631 8632 To be sure that they are all from the Latin script (for example), a 8633 lookahead can be used: 8634 8635 \s+(?=\p{Latin})(*sr:\S+) 8636 8637 This works as long as the first character is expected to be a character 8638 in that script, and not (for example) punctuation, which is allowed 8639 with any script. If this is not the case, a more creative lookahead is 8640 needed. For example, if digits, underscore, and dots are permitted at 8641 the start: 8642 8643 \s+(?=[0-9_.]*\p{Latin})(*sr:\S+) 8644 8645 8646 In many cases, backtracking into a script run pattern fragment is not 8647 desirable. The script run can employ an atomic group to prevent this. 8648 Because this is a common requirement, a shorthand notation is provided 8649 by (*atomic_script_run: or (*asr: 8650 8651 (*asr:...) is the same as (*sr:(?>...)) 8652 8653 Note that the atomic group is inside the script run. Putting it outside 8654 would not prevent backtracking into the script run pattern. 8655 8656 Support for script runs is not available if PCRE2 is compiled without 8657 Unicode support. A compile-time error is given if any of the above con- 8658 structs is encountered. Script runs are not supported by the alternate 8659 matching function, pcre2_dfa_match() because they use the same mecha- 8660 nism as capturing parentheses. 8661 8662 Warning: The (*ACCEPT) control verb (see below) should not be used 8663 within a script run group, because it causes an immediate exit from the 8664 group, bypassing the script run checking. 8665 8666 8667CONDITIONAL GROUPS 8668 8669 It is possible to cause the matching process to obey a pattern fragment 8670 conditionally or to choose between two alternative fragments, depending 8671 on the result of an assertion, or whether a specific capture group has 8672 already been matched. The two possible forms of conditional group are: 8673 8674 (?(condition)yes-pattern) 8675 (?(condition)yes-pattern|no-pattern) 8676 8677 If the condition is satisfied, the yes-pattern is used; otherwise the 8678 no-pattern (if present) is used. An absent no-pattern is equivalent to 8679 an empty string (it always matches). If there are more than two alter- 8680 natives in the group, a compile-time error occurs. Each of the two al- 8681 ternatives may itself contain nested groups of any form, including con- 8682 ditional groups; the restriction to two alternatives applies only at 8683 the level of the condition itself. This pattern fragment is an example 8684 where the alternatives are complex: 8685 8686 (?(1) (A|B|C) | (D | (?(2)E|F) | E) ) 8687 8688 8689 There are five kinds of condition: references to capture groups, refer- 8690 ences to recursion, two pseudo-conditions called DEFINE and VERSION, 8691 and assertions. 8692 8693 Checking for a used capture group by number 8694 8695 If the text between the parentheses consists of a sequence of digits, 8696 the condition is true if a capture group of that number has previously 8697 matched. If there is more than one capture group with the same number 8698 (see the earlier section about duplicate group numbers), the condition 8699 is true if any of them have matched. An alternative notation is to pre- 8700 cede the digits with a plus or minus sign. In this case, the group num- 8701 ber is relative rather than absolute. The most recently opened capture 8702 group can be referenced by (?(-1), the next most recent by (?(-2), and 8703 so on. Inside loops it can also make sense to refer to subsequent 8704 groups. The next capture group can be referenced as (?(+1), and so on. 8705 (The value zero in any of these forms is not used; it provokes a com- 8706 pile-time error.) 8707 8708 Consider the following pattern, which contains non-significant white 8709 space to make it more readable (assume the PCRE2_EXTENDED option) and 8710 to divide it into three parts for ease of discussion: 8711 8712 ( \( )? [^()]+ (?(1) \) ) 8713 8714 The first part matches an optional opening parenthesis, and if that 8715 character is present, sets it as the first captured substring. The sec- 8716 ond part matches one or more characters that are not parentheses. The 8717 third part is a conditional group that tests whether or not the first 8718 capture group matched. If it did, that is, if subject started with an 8719 opening parenthesis, the condition is true, and so the yes-pattern is 8720 executed and a closing parenthesis is required. Otherwise, since no- 8721 pattern is not present, the conditional group matches nothing. In other 8722 words, this pattern matches a sequence of non-parentheses, optionally 8723 enclosed in parentheses. 8724 8725 If you were embedding this pattern in a larger one, you could use a 8726 relative reference: 8727 8728 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ... 8729 8730 This makes the fragment independent of the parentheses in the larger 8731 pattern. 8732 8733 Checking for a used capture group by name 8734 8735 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a 8736 used capture group by name. For compatibility with earlier versions of 8737 PCRE1, which had this facility before Perl, the syntax (?(name)...) is 8738 also recognized. Note, however, that undelimited names consisting of 8739 the letter R followed by digits are ambiguous (see the following sec- 8740 tion). Rewriting the above example to use a named group gives this: 8741 8742 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) ) 8743 8744 If the name used in a condition of this kind is a duplicate, the test 8745 is applied to all groups of the same name, and is true if any one of 8746 them has matched. 8747 8748 Checking for pattern recursion 8749 8750 "Recursion" in this sense refers to any subroutine-like call from one 8751 part of the pattern to another, whether or not it is actually recur- 8752 sive. See the sections entitled "Recursive patterns" and "Groups as 8753 subroutines" below for details of recursion and subroutine calls. 8754 8755 If a condition is the string (R), and there is no capture group with 8756 the name R, the condition is true if matching is currently in a recur- 8757 sion or subroutine call to the whole pattern or any capture group. If 8758 digits follow the letter R, and there is no group with that name, the 8759 condition is true if the most recent call is into a group with the 8760 given number, which must exist somewhere in the overall pattern. This 8761 is a contrived example that is equivalent to a+b: 8762 8763 ((?(R1)a+|(?1)b)) 8764 8765 However, in both cases, if there is a capture group with a matching 8766 name, the condition tests for its being set, as described in the sec- 8767 tion above, instead of testing for recursion. For example, creating a 8768 group with the name R1 by adding (?<R1>) to the above pattern com- 8769 pletely changes its meaning. 8770 8771 If a name preceded by ampersand follows the letter R, for example: 8772 8773 (?(R&name)...) 8774 8775 the condition is true if the most recent recursion is into a group of 8776 that name (which must exist within the pattern). 8777 8778 This condition does not check the entire recursion stack. It tests only 8779 the current level. If the name used in a condition of this kind is a 8780 duplicate, the test is applied to all groups of the same name, and is 8781 true if any one of them is the most recent recursion. 8782 8783 At "top level", all these recursion test conditions are false. 8784 8785 Defining capture groups for use by reference only 8786 8787 If the condition is the string (DEFINE), the condition is always false, 8788 even if there is a group with the name DEFINE. In this case, there may 8789 be only one alternative in the rest of the conditional group. It is al- 8790 ways skipped if control reaches this point in the pattern; the idea of 8791 DEFINE is that it can be used to define subroutines that can be refer- 8792 enced from elsewhere. (The use of subroutines is described below.) For 8793 example, a pattern to match an IPv4 address such as "192.168.23.245" 8794 could be written like this (ignore white space and line breaks): 8795 8796 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) ) 8797 \b (?&byte) (\.(?&byte)){3} \b 8798 8799 The first part of the pattern is a DEFINE group inside which another 8800 group named "byte" is defined. This matches an individual component of 8801 an IPv4 address (a number less than 256). When matching takes place, 8802 this part of the pattern is skipped because DEFINE acts like a false 8803 condition. The rest of the pattern uses references to the named group 8804 to match the four dot-separated components of an IPv4 address, insist- 8805 ing on a word boundary at each end. 8806 8807 Checking the PCRE2 version 8808 8809 Programs that link with a PCRE2 library can check the version by call- 8810 ing pcre2_config() with appropriate arguments. Users of applications 8811 that do not have access to the underlying code cannot do this. A spe- 8812 cial "condition" called VERSION exists to allow such users to discover 8813 which version of PCRE2 they are dealing with by using this condition to 8814 match a string such as "yesno". VERSION must be followed either by "=" 8815 or ">=" and a version number. For example: 8816 8817 (?(VERSION>=10.4)yes|no) 8818 8819 This pattern matches "yes" if the PCRE2 version is greater or equal to 8820 10.4, or "no" otherwise. The fractional part of the version number may 8821 not contain more than two digits. 8822 8823 Assertion conditions 8824 8825 If the condition is not in any of the above formats, it must be a 8826 parenthesized assertion. This may be a positive or negative lookahead 8827 or lookbehind assertion. However, it must be a traditional atomic as- 8828 sertion, not one of the PCRE2-specific non-atomic assertions. 8829 8830 Consider this pattern, again containing non-significant white space, 8831 and with the two alternatives on the second line: 8832 8833 (?(?=[^a-z]*[a-z]) 8834 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} ) 8835 8836 The condition is a positive lookahead assertion that matches an op- 8837 tional sequence of non-letters followed by a letter. In other words, it 8838 tests for the presence of at least one letter in the subject. If a let- 8839 ter is found, the subject is matched against the first alternative; 8840 otherwise it is matched against the second. This pattern matches 8841 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are 8842 letters and dd are digits. 8843 8844 When an assertion that is a condition contains capture groups, any cap- 8845 turing that occurs in a matching branch is retained afterwards, for 8846 both positive and negative assertions, because matching always contin- 8847 ues after the assertion, whether it succeeds or fails. (Compare non- 8848 conditional assertions, for which captures are retained only for posi- 8849 tive assertions that succeed.) 8850 8851 8852COMMENTS 8853 8854 There are two ways of including comments in patterns that are processed 8855 by PCRE2. In both cases, the start of the comment must not be in a 8856 character class, nor in the middle of any other sequence of related 8857 characters such as (?: or a group name or number. The characters that 8858 make up a comment play no part in the pattern matching. 8859 8860 The sequence (?# marks the start of a comment that continues up to the 8861 next closing parenthesis. Nested parentheses are not permitted. If the 8862 PCRE2_EXTENDED or PCRE2_EXTENDED_MORE option is set, an unescaped # 8863 character also introduces a comment, which in this case continues to 8864 immediately after the next newline character or character sequence in 8865 the pattern. Which characters are interpreted as newlines is controlled 8866 by an option passed to the compiling function or by a special sequence 8867 at the start of the pattern, as described in the section entitled "New- 8868 line conventions" above. Note that the end of this type of comment is a 8869 literal newline sequence in the pattern; escape sequences that happen 8870 to represent a newline do not count. For example, consider this pattern 8871 when PCRE2_EXTENDED is set, and the default newline convention (a sin- 8872 gle linefeed character) is in force: 8873 8874 abc #comment \n still comment 8875 8876 On encountering the # character, pcre2_compile() skips along, looking 8877 for a newline in the pattern. The sequence \n is still literal at this 8878 stage, so it does not terminate the comment. Only an actual character 8879 with the code value 0x0a (the default newline) does so. 8880 8881 8882RECURSIVE PATTERNS 8883 8884 Consider the problem of matching a string in parentheses, allowing for 8885 unlimited nested parentheses. Without the use of recursion, the best 8886 that can be done is to use a pattern that matches up to some fixed 8887 depth of nesting. It is not possible to handle an arbitrary nesting 8888 depth. 8889 8890 For some time, Perl has provided a facility that allows regular expres- 8891 sions to recurse (amongst other things). It does this by interpolating 8892 Perl code in the expression at run time, and the code can refer to the 8893 expression itself. A Perl pattern using code interpolation to solve the 8894 parentheses problem can be created like this: 8895 8896 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x; 8897 8898 The (?p{...}) item interpolates Perl code at run time, and in this case 8899 refers recursively to the pattern in which it appears. 8900 8901 Obviously, PCRE2 cannot support the interpolation of Perl code. In- 8902 stead, it supports special syntax for recursion of the entire pattern, 8903 and also for individual capture group recursion. After its introduction 8904 in PCRE1 and Python, this kind of recursion was subsequently introduced 8905 into Perl at release 5.10. 8906 8907 A special item that consists of (? followed by a number greater than 8908 zero and a closing parenthesis is a recursive subroutine call of the 8909 capture group of the given number, provided that it occurs inside that 8910 group. (If not, it is a non-recursive subroutine call, which is de- 8911 scribed in the next section.) The special item (?R) or (?0) is a recur- 8912 sive call of the entire regular expression. 8913 8914 This PCRE2 pattern solves the nested parentheses problem (assume the 8915 PCRE2_EXTENDED option is set so that white space is ignored): 8916 8917 \( ( [^()]++ | (?R) )* \) 8918 8919 First it matches an opening parenthesis. Then it matches any number of 8920 substrings which can either be a sequence of non-parentheses, or a re- 8921 cursive match of the pattern itself (that is, a correctly parenthesized 8922 substring). Finally there is a closing parenthesis. Note the use of a 8923 possessive quantifier to avoid backtracking into sequences of non- 8924 parentheses. 8925 8926 If this were part of a larger pattern, you would not want to recurse 8927 the entire pattern, so instead you could use this: 8928 8929 ( \( ( [^()]++ | (?1) )* \) ) 8930 8931 We have put the pattern into parentheses, and caused the recursion to 8932 refer to them instead of the whole pattern. 8933 8934 In a larger pattern, keeping track of parenthesis numbers can be 8935 tricky. This is made easier by the use of relative references. Instead 8936 of (?1) in the pattern above you can write (?-2) to refer to the second 8937 most recently opened parentheses preceding the recursion. In other 8938 words, a negative number counts capturing parentheses leftwards from 8939 the point at which it is encountered. 8940 8941 Be aware however, that if duplicate capture group numbers are in use, 8942 relative references refer to the earliest group with the appropriate 8943 number. Consider, for example: 8944 8945 (?|(a)|(b)) (c) (?-2) 8946 8947 The first two capture groups (a) and (b) are both numbered 1, and group 8948 (c) is number 2. When the reference (?-2) is encountered, the second 8949 most recently opened parentheses has the number 1, but it is the first 8950 such group (the (a) group) to which the recursion refers. This would be 8951 the same if an absolute reference (?1) was used. In other words, rela- 8952 tive references are just a shorthand for computing a group number. 8953 8954 It is also possible to refer to subsequent capture groups, by writing 8955 references such as (?+2). However, these cannot be recursive because 8956 the reference is not inside the parentheses that are referenced. They 8957 are always non-recursive subroutine calls, as described in the next 8958 section. 8959 8960 An alternative approach is to use named parentheses. The Perl syntax 8961 for this is (?&name); PCRE1's earlier syntax (?P>name) is also sup- 8962 ported. We could rewrite the above example as follows: 8963 8964 (?<pn> \( ( [^()]++ | (?&pn) )* \) ) 8965 8966 If there is more than one group with the same name, the earliest one is 8967 used. 8968 8969 The example pattern that we have been looking at contains nested unlim- 8970 ited repeats, and so the use of a possessive quantifier for matching 8971 strings of non-parentheses is important when applying the pattern to 8972 strings that do not match. For example, when this pattern is applied to 8973 8974 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa() 8975 8976 it yields "no match" quickly. However, if a possessive quantifier is 8977 not used, the match runs for a very long time indeed because there are 8978 so many different ways the + and * repeats can carve up the subject, 8979 and all have to be tested before failure can be reported. 8980 8981 At the end of a match, the values of capturing parentheses are those 8982 from the outermost level. If you want to obtain intermediate values, a 8983 callout function can be used (see below and the pcre2callout documenta- 8984 tion). If the pattern above is matched against 8985 8986 (ab(cd)ef) 8987 8988 the value for the inner capturing parentheses (numbered 2) is "ef", 8989 which is the last value taken on at the top level. If a capture group 8990 is not matched at the top level, its final captured value is unset, 8991 even if it was (temporarily) set at a deeper level during the matching 8992 process. 8993 8994 Do not confuse the (?R) item with the condition (R), which tests for 8995 recursion. Consider this pattern, which matches text in angle brack- 8996 ets, allowing for arbitrary nesting. Only digits are allowed in nested 8997 brackets (that is, when recursing), whereas any characters are permit- 8998 ted at the outer level. 8999 9000 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * > 9001 9002 In this pattern, (?(R) is the start of a conditional group, with two 9003 different alternatives for the recursive and non-recursive cases. The 9004 (?R) item is the actual recursive call. 9005 9006 Differences in recursion processing between PCRE2 and Perl 9007 9008 Some former differences between PCRE2 and Perl no longer exist. 9009 9010 Before release 10.30, recursion processing in PCRE2 differed from Perl 9011 in that a recursive subroutine call was always treated as an atomic 9012 group. That is, once it had matched some of the subject string, it was 9013 never re-entered, even if it contained untried alternatives and there 9014 was a subsequent matching failure. (Historical note: PCRE implemented 9015 recursion before Perl did.) 9016 9017 Starting with release 10.30, recursive subroutine calls are no longer 9018 treated as atomic. That is, they can be re-entered to try unused alter- 9019 natives if there is a matching failure later in the pattern. This is 9020 now compatible with the way Perl works. If you want a subroutine call 9021 to be atomic, you must explicitly enclose it in an atomic group. 9022 9023 Supporting backtracking into recursions simplifies certain types of re- 9024 cursive pattern. For example, this pattern matches palindromic strings: 9025 9026 ^((.)(?1)\2|.?)$ 9027 9028 The second branch in the group matches a single central character in 9029 the palindrome when there are an odd number of characters, or nothing 9030 when there are an even number of characters, but in order to work it 9031 has to be able to try the second case when the rest of the pattern 9032 match fails. If you want to match typical palindromic phrases, the pat- 9033 tern has to ignore all non-word characters, which can be done like 9034 this: 9035 9036 ^\W*+((.)\W*+(?1)\W*+\2|\W*+.?)\W*+$ 9037 9038 If run with the PCRE2_CASELESS option, this pattern matches phrases 9039 such as "A man, a plan, a canal: Panama!". Note the use of the posses- 9040 sive quantifier *+ to avoid backtracking into sequences of non-word 9041 characters. Without this, PCRE2 takes a great deal longer (ten times or 9042 more) to match typical phrases, and Perl takes so long that you think 9043 it has gone into a loop. 9044 9045 Another way in which PCRE2 and Perl used to differ in their recursion 9046 processing is in the handling of captured values. Formerly in Perl, 9047 when a group was called recursively or as a subroutine (see the next 9048 section), it had no access to any values that were captured outside the 9049 recursion, whereas in PCRE2 these values can be referenced. Consider 9050 this pattern: 9051 9052 ^(.)(\1|a(?2)) 9053 9054 This pattern matches "bab". The first capturing parentheses match "b", 9055 then in the second group, when the backreference \1 fails to match "b", 9056 the second alternative matches "a" and then recurses. In the recursion, 9057 \1 does now match "b" and so the whole match succeeds. This match used 9058 to fail in Perl, but in later versions (I tried 5.024) it now works. 9059 9060 9061GROUPS AS SUBROUTINES 9062 9063 If the syntax for a recursive group call (either by number or by name) 9064 is used outside the parentheses to which it refers, it operates a bit 9065 like a subroutine in a programming language. More accurately, PCRE2 9066 treats the referenced group as an independent subpattern which it tries 9067 to match at the current matching position. The called group may be de- 9068 fined before or after the reference. A numbered reference can be abso- 9069 lute or relative, as in these examples: 9070 9071 (...(absolute)...)...(?2)... 9072 (...(relative)...)...(?-1)... 9073 (...(?+1)...(relative)... 9074 9075 An earlier example pointed out that the pattern 9076 9077 (sens|respons)e and \1ibility 9078 9079 matches "sense and sensibility" and "response and responsibility", but 9080 not "sense and responsibility". If instead the pattern 9081 9082 (sens|respons)e and (?1)ibility 9083 9084 is used, it does match "sense and responsibility" as well as the other 9085 two strings. Another example is given in the discussion of DEFINE 9086 above. 9087 9088 Like recursions, subroutine calls used to be treated as atomic, but 9089 this changed at PCRE2 release 10.30, so backtracking into subroutine 9090 calls can now occur. However, any capturing parentheses that are set 9091 during the subroutine call revert to their previous values afterwards. 9092 9093 Processing options such as case-independence are fixed when a group is 9094 defined, so if it is used as a subroutine, such options cannot be 9095 changed for different calls. For example, consider this pattern: 9096 9097 (abc)(?i:(?-1)) 9098 9099 It matches "abcabc". It does not match "abcABC" because the change of 9100 processing option does not affect the called group. 9101 9102 The behaviour of backtracking control verbs in groups when called as 9103 subroutines is described in the section entitled "Backtracking verbs in 9104 subroutines" below. 9105 9106 9107ONIGURUMA SUBROUTINE SYNTAX 9108 9109 For compatibility with Oniguruma, the non-Perl syntax \g followed by a 9110 name or a number enclosed either in angle brackets or single quotes, is 9111 an alternative syntax for calling a group as a subroutine, possibly re- 9112 cursively. Here are two of the examples used above, rewritten using 9113 this syntax: 9114 9115 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) ) 9116 (sens|respons)e and \g'1'ibility 9117 9118 PCRE2 supports an extension to Oniguruma: if a number is preceded by a 9119 plus or a minus sign it is taken as a relative reference. For example: 9120 9121 (abc)(?i:\g<-1>) 9122 9123 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not 9124 synonymous. The former is a backreference; the latter is a subroutine 9125 call. 9126 9127 9128CALLOUTS 9129 9130 Perl has a feature whereby using the sequence (?{...}) causes arbitrary 9131 Perl code to be obeyed in the middle of matching a regular expression. 9132 This makes it possible, amongst other things, to extract different sub- 9133 strings that match the same pair of parentheses when there is a repeti- 9134 tion. 9135 9136 PCRE2 provides a similar feature, but of course it cannot obey arbi- 9137 trary Perl code. The feature is called "callout". The caller of PCRE2 9138 provides an external function by putting its entry point in a match 9139 context using the function pcre2_set_callout(), and then passing that 9140 context to pcre2_match() or pcre2_dfa_match(). If no match context is 9141 passed, or if the callout entry point is set to NULL, callouts are dis- 9142 abled. 9143 9144 Within a regular expression, (?C<arg>) indicates a point at which the 9145 external function is to be called. There are two kinds of callout: 9146 those with a numerical argument and those with a string argument. (?C) 9147 on its own with no argument is treated as (?C0). A numerical argument 9148 allows the application to distinguish between different callouts. 9149 String arguments were added for release 10.20 to make it possible for 9150 script languages that use PCRE2 to embed short scripts within patterns 9151 in a similar way to Perl. 9152 9153 During matching, when PCRE2 reaches a callout point, the external func- 9154 tion is called. It is provided with the number or string argument of 9155 the callout, the position in the pattern, and one item of data that is 9156 also set in the match block. The callout function may cause matching to 9157 proceed, to backtrack, or to fail. 9158 9159 By default, PCRE2 implements a number of optimizations at matching 9160 time, and one side-effect is that sometimes callouts are skipped. If 9161 you need all possible callouts to happen, you need to set options that 9162 disable the relevant optimizations. More details, including a complete 9163 description of the programming interface to the callout function, are 9164 given in the pcre2callout documentation. 9165 9166 Callouts with numerical arguments 9167 9168 If you just want to have a means of identifying different callout 9169 points, put a number less than 256 after the letter C. For example, 9170 this pattern has two callout points: 9171 9172 (?C1)abc(?C2)def 9173 9174 If the PCRE2_AUTO_CALLOUT flag is passed to pcre2_compile(), numerical 9175 callouts are automatically installed before each item in the pattern. 9176 They are all numbered 255. If there is a conditional group in the pat- 9177 tern whose condition is an assertion, an additional callout is inserted 9178 just before the condition. An explicit callout may also be set at this 9179 position, as in this example: 9180 9181 (?(?C9)(?=a)abc|def) 9182 9183 Note that this applies only to assertion conditions, not to other types 9184 of condition. 9185 9186 Callouts with string arguments 9187 9188 A delimited string may be used instead of a number as a callout argu- 9189 ment. The starting delimiter must be one of ` ' " ^ % # $ { and the 9190 ending delimiter is the same as the start, except for {, where the end- 9191 ing delimiter is }. If the ending delimiter is needed within the 9192 string, it must be doubled. For example: 9193 9194 (?C'ab ''c'' d')xyz(?C{any text})pqr 9195 9196 The doubling is removed before the string is passed to the callout 9197 function. 9198 9199 9200BACKTRACKING CONTROL 9201 9202 There are a number of special "Backtracking Control Verbs" (to use 9203 Perl's terminology) that modify the behaviour of backtracking during 9204 matching. They are generally of the form (*VERB) or (*VERB:NAME). Some 9205 verbs take either form, and may behave differently depending on whether 9206 or not a name argument is present. The names are not required to be 9207 unique within the pattern. 9208 9209 By default, for compatibility with Perl, a name is any sequence of 9210 characters that does not include a closing parenthesis. The name is not 9211 processed in any way, and it is not possible to include a closing 9212 parenthesis in the name. This can be changed by setting the 9213 PCRE2_ALT_VERBNAMES option, but the result is no longer Perl-compati- 9214 ble. 9215 9216 When PCRE2_ALT_VERBNAMES is set, backslash processing is applied to 9217 verb names and only an unescaped closing parenthesis terminates the 9218 name. However, the only backslash items that are permitted are \Q, \E, 9219 and sequences such as \x{100} that define character code points. Char- 9220 acter type escapes such as \d are faulted. 9221 9222 A closing parenthesis can be included in a name either as \) or between 9223 \Q and \E. In addition to backslash processing, if the PCRE2_EXTENDED 9224 or PCRE2_EXTENDED_MORE option is also set, unescaped whitespace in verb 9225 names is skipped, and #-comments are recognized, exactly as in the rest 9226 of the pattern. PCRE2_EXTENDED and PCRE2_EXTENDED_MORE do not affect 9227 verb names unless PCRE2_ALT_VERBNAMES is also set. 9228 9229 The maximum length of a name is 255 in the 8-bit library and 65535 in 9230 the 16-bit and 32-bit libraries. If the name is empty, that is, if the 9231 closing parenthesis immediately follows the colon, the effect is as if 9232 the colon were not there. Any number of these verbs may occur in a pat- 9233 tern. Except for (*ACCEPT), they may not be quantified. 9234 9235 Since these verbs are specifically related to backtracking, most of 9236 them can be used only when the pattern is to be matched using the tra- 9237 ditional matching function, because that uses a backtracking algorithm. 9238 With the exception of (*FAIL), which behaves like a failing negative 9239 assertion, the backtracking control verbs cause an error if encountered 9240 by the DFA matching function. 9241 9242 The behaviour of these verbs in repeated groups, assertions, and in 9243 capture groups called as subroutines (whether or not recursively) is 9244 documented below. 9245 9246 Optimizations that affect backtracking verbs 9247 9248 PCRE2 contains some optimizations that are used to speed up matching by 9249 running some checks at the start of each match attempt. For example, it 9250 may know the minimum length of matching subject, or that a particular 9251 character must be present. When one of these optimizations bypasses the 9252 running of a match, any included backtracking verbs will not, of 9253 course, be processed. You can suppress the start-of-match optimizations 9254 by setting the PCRE2_NO_START_OPTIMIZE option when calling pcre2_com- 9255 pile(), or by starting the pattern with (*NO_START_OPT). There is more 9256 discussion of this option in the section entitled "Compiling a pattern" 9257 in the pcre2api documentation. 9258 9259 Experiments with Perl suggest that it too has similar optimizations, 9260 and like PCRE2, turning them off can change the result of a match. 9261 9262 Verbs that act immediately 9263 9264 The following verbs act as soon as they are encountered. 9265 9266 (*ACCEPT) or (*ACCEPT:NAME) 9267 9268 This verb causes the match to end successfully, skipping the remainder 9269 of the pattern. However, when it is inside a capture group that is 9270 called as a subroutine, only that group is ended successfully. Matching 9271 then continues at the outer level. If (*ACCEPT) in triggered in a posi- 9272 tive assertion, the assertion succeeds; in a negative assertion, the 9273 assertion fails. 9274 9275 If (*ACCEPT) is inside capturing parentheses, the data so far is cap- 9276 tured. For example: 9277 9278 A((?:A|B(*ACCEPT)|C)D) 9279 9280 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap- 9281 tured by the outer parentheses. 9282 9283 (*ACCEPT) is the only backtracking verb that is allowed to be quanti- 9284 fied because an ungreedy quantification with a minimum of zero acts 9285 only when a backtrack happens. Consider, for example, 9286 9287 (A(*ACCEPT)??B)C 9288 9289 where A, B, and C may be complex expressions. After matching "A", the 9290 matcher processes "BC"; if that fails, causing a backtrack, (*ACCEPT) 9291 is triggered and the match succeeds. In both cases, all but C is cap- 9292 tured. Whereas (*COMMIT) (see below) means "fail on backtrack", a re- 9293 peated (*ACCEPT) of this type means "succeed on backtrack". 9294 9295 Warning: (*ACCEPT) should not be used within a script run group, be- 9296 cause it causes an immediate exit from the group, bypassing the script 9297 run checking. 9298 9299 (*FAIL) or (*FAIL:NAME) 9300 9301 This verb causes a matching failure, forcing backtracking to occur. It 9302 may be abbreviated to (*F). It is equivalent to (?!) but easier to 9303 read. The Perl documentation notes that it is probably useful only when 9304 combined with (?{}) or (??{}). Those are, of course, Perl features that 9305 are not present in PCRE2. The nearest equivalent is the callout fea- 9306 ture, as for example in this pattern: 9307 9308 a+(?C)(*FAIL) 9309 9310 A match with the string "aaaa" always fails, but the callout is taken 9311 before each backtrack happens (in this example, 10 times). 9312 9313 (*ACCEPT:NAME) and (*FAIL:NAME) behave the same as (*MARK:NAME)(*AC- 9314 CEPT) and (*MARK:NAME)(*FAIL), respectively, that is, a (*MARK) is 9315 recorded just before the verb acts. 9316 9317 Recording which path was taken 9318 9319 There is one verb whose main purpose is to track how a match was ar- 9320 rived at, though it also has a secondary use in conjunction with ad- 9321 vancing the match starting point (see (*SKIP) below). 9322 9323 (*MARK:NAME) or (*:NAME) 9324 9325 A name is always required with this verb. For all the other backtrack- 9326 ing control verbs, a NAME argument is optional. 9327 9328 When a match succeeds, the name of the last-encountered mark name on 9329 the matching path is passed back to the caller as described in the sec- 9330 tion entitled "Other information about the match" in the pcre2api docu- 9331 mentation. This applies to all instances of (*MARK) and other verbs, 9332 including those inside assertions and atomic groups. However, there are 9333 differences in those cases when (*MARK) is used in conjunction with 9334 (*SKIP) as described below. 9335 9336 The mark name that was last encountered on the matching path is passed 9337 back. A verb without a NAME argument is ignored for this purpose. Here 9338 is an example of pcre2test output, where the "mark" modifier requests 9339 the retrieval and outputting of (*MARK) data: 9340 9341 re> /X(*MARK:A)Y|X(*MARK:B)Z/mark 9342 data> XY 9343 0: XY 9344 MK: A 9345 XZ 9346 0: XZ 9347 MK: B 9348 9349 The (*MARK) name is tagged with "MK:" in this output, and in this exam- 9350 ple it indicates which of the two alternatives matched. This is a more 9351 efficient way of obtaining this information than putting each alterna- 9352 tive in its own capturing parentheses. 9353 9354 If a verb with a name is encountered in a positive assertion that is 9355 true, the name is recorded and passed back if it is the last-encoun- 9356 tered. This does not happen for negative assertions or failing positive 9357 assertions. 9358 9359 After a partial match or a failed match, the last encountered name in 9360 the entire match process is returned. For example: 9361 9362 re> /X(*MARK:A)Y|X(*MARK:B)Z/mark 9363 data> XP 9364 No match, mark = B 9365 9366 Note that in this unanchored example the mark is retained from the 9367 match attempt that started at the letter "X" in the subject. Subsequent 9368 match attempts starting at "P" and then with an empty string do not get 9369 as far as the (*MARK) item, but nevertheless do not reset it. 9370 9371 If you are interested in (*MARK) values after failed matches, you 9372 should probably set the PCRE2_NO_START_OPTIMIZE option (see above) to 9373 ensure that the match is always attempted. 9374 9375 Verbs that act after backtracking 9376 9377 The following verbs do nothing when they are encountered. Matching con- 9378 tinues with what follows, but if there is a subsequent match failure, 9379 causing a backtrack to the verb, a failure is forced. That is, back- 9380 tracking cannot pass to the left of the verb. However, when one of 9381 these verbs appears inside an atomic group or in a lookaround assertion 9382 that is true, its effect is confined to that group, because once the 9383 group has been matched, there is never any backtracking into it. Back- 9384 tracking from beyond an assertion or an atomic group ignores the entire 9385 group, and seeks a preceding backtracking point. 9386 9387 These verbs differ in exactly what kind of failure occurs when back- 9388 tracking reaches them. The behaviour described below is what happens 9389 when the verb is not in a subroutine or an assertion. Subsequent sec- 9390 tions cover these special cases. 9391 9392 (*COMMIT) or (*COMMIT:NAME) 9393 9394 This verb causes the whole match to fail outright if there is a later 9395 matching failure that causes backtracking to reach it. Even if the pat- 9396 tern is unanchored, no further attempts to find a match by advancing 9397 the starting point take place. If (*COMMIT) is the only backtracking 9398 verb that is encountered, once it has been passed pcre2_match() is com- 9399 mitted to finding a match at the current starting point, or not at all. 9400 For example: 9401 9402 a+(*COMMIT)b 9403 9404 This matches "xxaab" but not "aacaab". It can be thought of as a kind 9405 of dynamic anchor, or "I've started, so I must finish." 9406 9407 The behaviour of (*COMMIT:NAME) is not the same as (*MARK:NAME)(*COM- 9408 MIT). It is like (*MARK:NAME) in that the name is remembered for pass- 9409 ing back to the caller. However, (*SKIP:NAME) searches only for names 9410 that are set with (*MARK), ignoring those set by any of the other back- 9411 tracking verbs. 9412 9413 If there is more than one backtracking verb in a pattern, a different 9414 one that follows (*COMMIT) may be triggered first, so merely passing 9415 (*COMMIT) during a match does not always guarantee that a match must be 9416 at this starting point. 9417 9418 Note that (*COMMIT) at the start of a pattern is not the same as an an- 9419 chor, unless PCRE2's start-of-match optimizations are turned off, as 9420 shown in this output from pcre2test: 9421 9422 re> /(*COMMIT)abc/ 9423 data> xyzabc 9424 0: abc 9425 data> 9426 re> /(*COMMIT)abc/no_start_optimize 9427 data> xyzabc 9428 No match 9429 9430 For the first pattern, PCRE2 knows that any match must start with "a", 9431 so the optimization skips along the subject to "a" before applying the 9432 pattern to the first set of data. The match attempt then succeeds. The 9433 second pattern disables the optimization that skips along to the first 9434 character. The pattern is now applied starting at "x", and so the 9435 (*COMMIT) causes the match to fail without trying any other starting 9436 points. 9437 9438 (*PRUNE) or (*PRUNE:NAME) 9439 9440 This verb causes the match to fail at the current starting position in 9441 the subject if there is a later matching failure that causes backtrack- 9442 ing to reach it. If the pattern is unanchored, the normal "bumpalong" 9443 advance to the next starting character then happens. Backtracking can 9444 occur as usual to the left of (*PRUNE), before it is reached, or when 9445 matching to the right of (*PRUNE), but if there is no match to the 9446 right, backtracking cannot cross (*PRUNE). In simple cases, the use of 9447 (*PRUNE) is just an alternative to an atomic group or possessive quan- 9448 tifier, but there are some uses of (*PRUNE) that cannot be expressed in 9449 any other way. In an anchored pattern (*PRUNE) has the same effect as 9450 (*COMMIT). 9451 9452 The behaviour of (*PRUNE:NAME) is not the same as (*MARK:NAME)(*PRUNE). 9453 It is like (*MARK:NAME) in that the name is remembered for passing back 9454 to the caller. However, (*SKIP:NAME) searches only for names set with 9455 (*MARK), ignoring those set by other backtracking verbs. 9456 9457 (*SKIP) 9458 9459 This verb, when given without a name, is like (*PRUNE), except that if 9460 the pattern is unanchored, the "bumpalong" advance is not to the next 9461 character, but to the position in the subject where (*SKIP) was encoun- 9462 tered. (*SKIP) signifies that whatever text was matched leading up to 9463 it cannot be part of a successful match if there is a later mismatch. 9464 Consider: 9465 9466 a+(*SKIP)b 9467 9468 If the subject is "aaaac...", after the first match attempt fails 9469 (starting at the first character in the string), the starting point 9470 skips on to start the next attempt at "c". Note that a possessive quan- 9471 tifier does not have the same effect as this example; although it would 9472 suppress backtracking during the first match attempt, the second at- 9473 tempt would start at the second character instead of skipping on to 9474 "c". 9475 9476 If (*SKIP) is used to specify a new starting position that is the same 9477 as the starting position of the current match, or (by being inside a 9478 lookbehind) earlier, the position specified by (*SKIP) is ignored, and 9479 instead the normal "bumpalong" occurs. 9480 9481 (*SKIP:NAME) 9482 9483 When (*SKIP) has an associated name, its behaviour is modified. When 9484 such a (*SKIP) is triggered, the previous path through the pattern is 9485 searched for the most recent (*MARK) that has the same name. If one is 9486 found, the "bumpalong" advance is to the subject position that corre- 9487 sponds to that (*MARK) instead of to where (*SKIP) was encountered. If 9488 no (*MARK) with a matching name is found, the (*SKIP) is ignored. 9489 9490 The search for a (*MARK) name uses the normal backtracking mechanism, 9491 which means that it does not see (*MARK) settings that are inside 9492 atomic groups or assertions, because they are never re-entered by back- 9493 tracking. Compare the following pcre2test examples: 9494 9495 re> /a(?>(*MARK:X))(*SKIP:X)(*F)|(.)/ 9496 data: abc 9497 0: a 9498 1: a 9499 data: 9500 re> /a(?:(*MARK:X))(*SKIP:X)(*F)|(.)/ 9501 data: abc 9502 0: b 9503 1: b 9504 9505 In the first example, the (*MARK) setting is in an atomic group, so it 9506 is not seen when (*SKIP:X) triggers, causing the (*SKIP) to be ignored. 9507 This allows the second branch of the pattern to be tried at the first 9508 character position. In the second example, the (*MARK) setting is not 9509 in an atomic group. This allows (*SKIP:X) to find the (*MARK) when it 9510 backtracks, and this causes a new matching attempt to start at the sec- 9511 ond character. This time, the (*MARK) is never seen because "a" does 9512 not match "b", so the matcher immediately jumps to the second branch of 9513 the pattern. 9514 9515 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It 9516 ignores names that are set by other backtracking verbs. 9517 9518 (*THEN) or (*THEN:NAME) 9519 9520 This verb causes a skip to the next innermost alternative when back- 9521 tracking reaches it. That is, it cancels any further backtracking 9522 within the current alternative. Its name comes from the observation 9523 that it can be used for a pattern-based if-then-else block: 9524 9525 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ... 9526 9527 If the COND1 pattern matches, FOO is tried (and possibly further items 9528 after the end of the group if FOO succeeds); on failure, the matcher 9529 skips to the second alternative and tries COND2, without backtracking 9530 into COND1. If that succeeds and BAR fails, COND3 is tried. If subse- 9531 quently BAZ fails, there are no more alternatives, so there is a back- 9532 track to whatever came before the entire group. If (*THEN) is not in- 9533 side an alternation, it acts like (*PRUNE). 9534 9535 The behaviour of (*THEN:NAME) is not the same as (*MARK:NAME)(*THEN). 9536 It is like (*MARK:NAME) in that the name is remembered for passing back 9537 to the caller. However, (*SKIP:NAME) searches only for names set with 9538 (*MARK), ignoring those set by other backtracking verbs. 9539 9540 A group that does not contain a | character is just a part of the en- 9541 closing alternative; it is not a nested alternation with only one al- 9542 ternative. The effect of (*THEN) extends beyond such a group to the en- 9543 closing alternative. Consider this pattern, where A, B, etc. are com- 9544 plex pattern fragments that do not contain any | characters at this 9545 level: 9546 9547 A (B(*THEN)C) | D 9548 9549 If A and B are matched, but there is a failure in C, matching does not 9550 backtrack into A; instead it moves to the next alternative, that is, D. 9551 However, if the group containing (*THEN) is given an alternative, it 9552 behaves differently: 9553 9554 A (B(*THEN)C | (*FAIL)) | D 9555 9556 The effect of (*THEN) is now confined to the inner group. After a fail- 9557 ure in C, matching moves to (*FAIL), which causes the whole group to 9558 fail because there are no more alternatives to try. In this case, 9559 matching does backtrack into A. 9560 9561 Note that a conditional group is not considered as having two alterna- 9562 tives, because only one is ever used. In other words, the | character 9563 in a conditional group has a different meaning. Ignoring white space, 9564 consider: 9565 9566 ^.*? (?(?=a) a | b(*THEN)c ) 9567 9568 If the subject is "ba", this pattern does not match. Because .*? is un- 9569 greedy, it initially matches zero characters. The condition (?=a) then 9570 fails, the character "b" is matched, but "c" is not. At this point, 9571 matching does not backtrack to .*? as might perhaps be expected from 9572 the presence of the | character. The conditional group is part of the 9573 single alternative that comprises the whole pattern, and so the match 9574 fails. (If there was a backtrack into .*?, allowing it to match "b", 9575 the match would succeed.) 9576 9577 The verbs just described provide four different "strengths" of control 9578 when subsequent matching fails. (*THEN) is the weakest, carrying on the 9579 match at the next alternative. (*PRUNE) comes next, failing the match 9580 at the current starting position, but allowing an advance to the next 9581 character (for an unanchored pattern). (*SKIP) is similar, except that 9582 the advance may be more than one character. (*COMMIT) is the strongest, 9583 causing the entire match to fail. 9584 9585 More than one backtracking verb 9586 9587 If more than one backtracking verb is present in a pattern, the one 9588 that is backtracked onto first acts. For example, consider this pat- 9589 tern, where A, B, etc. are complex pattern fragments: 9590 9591 (A(*COMMIT)B(*THEN)C|ABD) 9592 9593 If A matches but B fails, the backtrack to (*COMMIT) causes the entire 9594 match to fail. However, if A and B match, but C fails, the backtrack to 9595 (*THEN) causes the next alternative (ABD) to be tried. This behaviour 9596 is consistent, but is not always the same as Perl's. It means that if 9597 two or more backtracking verbs appear in succession, all the the last 9598 of them has no effect. Consider this example: 9599 9600 ...(*COMMIT)(*PRUNE)... 9601 9602 If there is a matching failure to the right, backtracking onto (*PRUNE) 9603 causes it to be triggered, and its action is taken. There can never be 9604 a backtrack onto (*COMMIT). 9605 9606 Backtracking verbs in repeated groups 9607 9608 PCRE2 sometimes differs from Perl in its handling of backtracking verbs 9609 in repeated groups. For example, consider: 9610 9611 /(a(*COMMIT)b)+ac/ 9612 9613 If the subject is "abac", Perl matches unless its optimizations are 9614 disabled, but PCRE2 always fails because the (*COMMIT) in the second 9615 repeat of the group acts. 9616 9617 Backtracking verbs in assertions 9618 9619 (*FAIL) in any assertion has its normal effect: it forces an immediate 9620 backtrack. The behaviour of the other backtracking verbs depends on 9621 whether or not the assertion is standalone or acting as the condition 9622 in a conditional group. 9623 9624 (*ACCEPT) in a standalone positive assertion causes the assertion to 9625 succeed without any further processing; captured strings and a mark 9626 name (if set) are retained. In a standalone negative assertion, (*AC- 9627 CEPT) causes the assertion to fail without any further processing; cap- 9628 tured substrings and any mark name are discarded. 9629 9630 If the assertion is a condition, (*ACCEPT) causes the condition to be 9631 true for a positive assertion and false for a negative one; captured 9632 substrings are retained in both cases. 9633 9634 The remaining verbs act only when a later failure causes a backtrack to 9635 reach them. This means that, for the Perl-compatible assertions, their 9636 effect is confined to the assertion, because Perl lookaround assertions 9637 are atomic. A backtrack that occurs after such an assertion is complete 9638 does not jump back into the assertion. Note in particular that a 9639 (*MARK) name that is set in an assertion is not "seen" by an instance 9640 of (*SKIP:NAME) later in the pattern. 9641 9642 PCRE2 now supports non-atomic positive assertions, as described in the 9643 section entitled "Non-atomic assertions" above. These assertions must 9644 be standalone (not used as conditions). They are not Perl-compatible. 9645 For these assertions, a later backtrack does jump back into the asser- 9646 tion, and therefore verbs such as (*COMMIT) can be triggered by back- 9647 tracks from later in the pattern. 9648 9649 The effect of (*THEN) is not allowed to escape beyond an assertion. If 9650 there are no more branches to try, (*THEN) causes a positive assertion 9651 to be false, and a negative assertion to be true. 9652 9653 The other backtracking verbs are not treated specially if they appear 9654 in a standalone positive assertion. In a conditional positive asser- 9655 tion, backtracking (from within the assertion) into (*COMMIT), (*SKIP), 9656 or (*PRUNE) causes the condition to be false. However, for both stand- 9657 alone and conditional negative assertions, backtracking into (*COMMIT), 9658 (*SKIP), or (*PRUNE) causes the assertion to be true, without consider- 9659 ing any further alternative branches. 9660 9661 Backtracking verbs in subroutines 9662 9663 These behaviours occur whether or not the group is called recursively. 9664 9665 (*ACCEPT) in a group called as a subroutine causes the subroutine match 9666 to succeed without any further processing. Matching then continues af- 9667 ter the subroutine call. Perl documents this behaviour. Perl's treat- 9668 ment of the other verbs in subroutines is different in some cases. 9669 9670 (*FAIL) in a group called as a subroutine has its normal effect: it 9671 forces an immediate backtrack. 9672 9673 (*COMMIT), (*SKIP), and (*PRUNE) cause the subroutine match to fail 9674 when triggered by being backtracked to in a group called as a subrou- 9675 tine. There is then a backtrack at the outer level. 9676 9677 (*THEN), when triggered, skips to the next alternative in the innermost 9678 enclosing group that has alternatives (its normal behaviour). However, 9679 if there is no such group within the subroutine's group, the subroutine 9680 match fails and there is a backtrack at the outer level. 9681 9682 9683SEE ALSO 9684 9685 pcre2api(3), pcre2callout(3), pcre2matching(3), pcre2syntax(3), 9686 pcre2(3). 9687 9688 9689AUTHOR 9690 9691 Philip Hazel 9692 Retired from University Computing Service 9693 Cambridge, England. 9694 9695 9696REVISION 9697 9698 Last updated: 12 January 2022 9699 Copyright (c) 1997-2022 University of Cambridge. 9700------------------------------------------------------------------------------ 9701 9702 9703PCRE2PERFORM(3) Library Functions Manual PCRE2PERFORM(3) 9704 9705 9706 9707NAME 9708 PCRE2 - Perl-compatible regular expressions (revised API) 9709 9710PCRE2 PERFORMANCE 9711 9712 Two aspects of performance are discussed below: memory usage and pro- 9713 cessing time. The way you express your pattern as a regular expression 9714 can affect both of them. 9715 9716 9717COMPILED PATTERN MEMORY USAGE 9718 9719 Patterns are compiled by PCRE2 into a reasonably efficient interpretive 9720 code, so that most simple patterns do not use much memory for storing 9721 the compiled version. However, there is one case where the memory usage 9722 of a compiled pattern can be unexpectedly large. If a parenthesized 9723 group has a quantifier with a minimum greater than 1 and/or a limited 9724 maximum, the whole group is repeated in the compiled code. For example, 9725 the pattern 9726 9727 (abc|def){2,4} 9728 9729 is compiled as if it were 9730 9731 (abc|def)(abc|def)((abc|def)(abc|def)?)? 9732 9733 (Technical aside: It is done this way so that backtrack points within 9734 each of the repetitions can be independently maintained.) 9735 9736 For regular expressions whose quantifiers use only small numbers, this 9737 is not usually a problem. However, if the numbers are large, and par- 9738 ticularly if such repetitions are nested, the memory usage can become 9739 an embarrassment. For example, the very simple pattern 9740 9741 ((ab){1,1000}c){1,3} 9742 9743 uses over 50KiB when compiled using the 8-bit library. When PCRE2 is 9744 compiled with its default internal pointer size of two bytes, the size 9745 limit on a compiled pattern is 65535 code units in the 8-bit and 16-bit 9746 libraries, and this is reached with the above pattern if the outer rep- 9747 etition is increased from 3 to 4. PCRE2 can be compiled to use larger 9748 internal pointers and thus handle larger compiled patterns, but it is 9749 better to try to rewrite your pattern to use less memory if you can. 9750 9751 One way of reducing the memory usage for such patterns is to make use 9752 of PCRE2's "subroutine" facility. Re-writing the above pattern as 9753 9754 ((ab)(?2){0,999}c)(?1){0,2} 9755 9756 reduces the memory requirements to around 16KiB, and indeed it remains 9757 under 20KiB even with the outer repetition increased to 100. However, 9758 this kind of pattern is not always exactly equivalent, because any cap- 9759 tures within subroutine calls are lost when the subroutine completes. 9760 If this is not a problem, this kind of rewriting will allow you to 9761 process patterns that PCRE2 cannot otherwise handle. The matching per- 9762 formance of the two different versions of the pattern are roughly the 9763 same. (This applies from release 10.30 - things were different in ear- 9764 lier releases.) 9765 9766 9767STACK AND HEAP USAGE AT RUN TIME 9768 9769 From release 10.30, the interpretive (non-JIT) version of pcre2_match() 9770 uses very little system stack at run time. In earlier releases recur- 9771 sive function calls could use a great deal of stack, and this could 9772 cause problems, but this usage has been eliminated. Backtracking posi- 9773 tions are now explicitly remembered in memory frames controlled by the 9774 code. An initial 20KiB vector of frames is allocated on the system 9775 stack (enough for about 100 frames for small patterns), but if this is 9776 insufficient, heap memory is used. The amount of heap memory can be 9777 limited; if the limit is set to zero, only the initial stack vector is 9778 used. Rewriting patterns to be time-efficient, as described below, may 9779 also reduce the memory requirements. 9780 9781 In contrast to pcre2_match(), pcre2_dfa_match() does use recursive 9782 function calls, but only for processing atomic groups, lookaround as- 9783 sertions, and recursion within the pattern. The original version of the 9784 code used to allocate quite large internal workspace vectors on the 9785 stack, which caused some problems for some patterns in environments 9786 with small stacks. From release 10.32 the code for pcre2_dfa_match() 9787 has been re-factored to use heap memory when necessary for internal 9788 workspace when recursing, though recursive function calls are still 9789 used. 9790 9791 The "match depth" parameter can be used to limit the depth of function 9792 recursion, and the "match heap" parameter to limit heap memory in 9793 pcre2_dfa_match(). 9794 9795 9796PROCESSING TIME 9797 9798 Certain items in regular expression patterns are processed more effi- 9799 ciently than others. It is more efficient to use a character class like 9800 [aeiou] than a set of single-character alternatives such as 9801 (a|e|i|o|u). In general, the simplest construction that provides the 9802 required behaviour is usually the most efficient. Jeffrey Friedl's book 9803 contains a lot of useful general discussion about optimizing regular 9804 expressions for efficient performance. This document contains a few ob- 9805 servations about PCRE2. 9806 9807 Using Unicode character properties (the \p, \P, and \X escapes) is 9808 slow, because PCRE2 has to use a multi-stage table lookup whenever it 9809 needs a character's property. If you can find an alternative pattern 9810 that does not use character properties, it will probably be faster. 9811 9812 By default, the escape sequences \b, \d, \s, and \w, and the POSIX 9813 character classes such as [:alpha:] do not use Unicode properties, 9814 partly for backwards compatibility, and partly for performance reasons. 9815 However, you can set the PCRE2_UCP option or start the pattern with 9816 (*UCP) if you want Unicode character properties to be used. This can 9817 double the matching time for items such as \d, when matched with 9818 pcre2_match(); the performance loss is less with a DFA matching func- 9819 tion, and in both cases there is not much difference for \b. 9820 9821 When a pattern begins with .* not in atomic parentheses, nor in paren- 9822 theses that are the subject of a backreference, and the PCRE2_DOTALL 9823 option is set, the pattern is implicitly anchored by PCRE2, since it 9824 can match only at the start of a subject string. If the pattern has 9825 multiple top-level branches, they must all be anchorable. The optimiza- 9826 tion can be disabled by the PCRE2_NO_DOTSTAR_ANCHOR option, and is au- 9827 tomatically disabled if the pattern contains (*PRUNE) or (*SKIP). 9828 9829 If PCRE2_DOTALL is not set, PCRE2 cannot make this optimization, be- 9830 cause the dot metacharacter does not then match a newline, and if the 9831 subject string contains newlines, the pattern may match from the char- 9832 acter immediately following one of them instead of from the very start. 9833 For example, the pattern 9834 9835 .*second 9836 9837 matches the subject "first\nand second" (where \n stands for a newline 9838 character), with the match starting at the seventh character. In order 9839 to do this, PCRE2 has to retry the match starting after every newline 9840 in the subject. 9841 9842 If you are using such a pattern with subject strings that do not con- 9843 tain newlines, the best performance is obtained by setting 9844 PCRE2_DOTALL, or starting the pattern with ^.* or ^.*? to indicate ex- 9845 plicit anchoring. That saves PCRE2 from having to scan along the sub- 9846 ject looking for a newline to restart at. 9847 9848 Beware of patterns that contain nested indefinite repeats. These can 9849 take a long time to run when applied to a string that does not match. 9850 Consider the pattern fragment 9851 9852 ^(a+)* 9853 9854 This can match "aaaa" in 16 different ways, and this number increases 9855 very rapidly as the string gets longer. (The * repeat can match 0, 1, 9856 2, 3, or 4 times, and for each of those cases other than 0 or 4, the + 9857 repeats can match different numbers of times.) When the remainder of 9858 the pattern is such that the entire match is going to fail, PCRE2 has 9859 in principle to try every possible variation, and this can take an ex- 9860 tremely long time, even for relatively short strings. 9861 9862 An optimization catches some of the more simple cases such as 9863 9864 (a+)*b 9865 9866 where a literal character follows. Before embarking on the standard 9867 matching procedure, PCRE2 checks that there is a "b" later in the sub- 9868 ject string, and if there is not, it fails the match immediately. How- 9869 ever, when there is no following literal this optimization cannot be 9870 used. You can see the difference by comparing the behaviour of 9871 9872 (a+)*\d 9873 9874 with the pattern above. The former gives a failure almost instantly 9875 when applied to a whole line of "a" characters, whereas the latter 9876 takes an appreciable time with strings longer than about 20 characters. 9877 9878 In many cases, the solution to this kind of performance issue is to use 9879 an atomic group or a possessive quantifier. This can often reduce mem- 9880 ory requirements as well. As another example, consider this pattern: 9881 9882 ([^<]|<(?!inet))+ 9883 9884 It matches from wherever it starts until it encounters "<inet" or the 9885 end of the data, and is the kind of pattern that might be used when 9886 processing an XML file. Each iteration of the outer parentheses matches 9887 either one character that is not "<" or a "<" that is not followed by 9888 "inet". However, each time a parenthesis is processed, a backtracking 9889 position is passed, so this formulation uses a memory frame for each 9890 matched character. For a long string, a lot of memory is required. Con- 9891 sider now this rewritten pattern, which matches exactly the same 9892 strings: 9893 9894 ([^<]++|<(?!inet))+ 9895 9896 This runs much faster, because sequences of characters that do not con- 9897 tain "<" are "swallowed" in one item inside the parentheses, and a pos- 9898 sessive quantifier is used to stop any backtracking into the runs of 9899 non-"<" characters. This version also uses a lot less memory because 9900 entry to a new set of parentheses happens only when a "<" character 9901 that is not followed by "inet" is encountered (and we assume this is 9902 relatively rare). 9903 9904 This example shows that one way of optimizing performance when matching 9905 long subject strings is to write repeated parenthesized subpatterns to 9906 match more than one character whenever possible. 9907 9908 SETTING RESOURCE LIMITS 9909 9910 You can set limits on the amount of processing that takes place when 9911 matching, and on the amount of heap memory that is used. The default 9912 values of the limits are very large, and unlikely ever to operate. They 9913 can be changed when PCRE2 is built, and they can also be set when 9914 pcre2_match() or pcre2_dfa_match() is called. For details of these in- 9915 terfaces, see the pcre2build documentation and the section entitled 9916 "The match context" in the pcre2api documentation. 9917 9918 The pcre2test test program has a modifier called "find_limits" which, 9919 if applied to a subject line, causes it to find the smallest limits 9920 that allow a pattern to match. This is done by repeatedly matching with 9921 different limits. 9922 9923 9924AUTHOR 9925 9926 Philip Hazel 9927 University Computing Service 9928 Cambridge, England. 9929 9930 9931REVISION 9932 9933 Last updated: 03 February 2019 9934 Copyright (c) 1997-2019 University of Cambridge. 9935------------------------------------------------------------------------------ 9936 9937 9938PCRE2POSIX(3) Library Functions Manual PCRE2POSIX(3) 9939 9940 9941 9942NAME 9943 PCRE2 - Perl-compatible regular expressions (revised API) 9944 9945SYNOPSIS 9946 9947 #include <pcre2posix.h> 9948 9949 int pcre2_regcomp(regex_t *preg, const char *pattern, 9950 int cflags); 9951 9952 int pcre2_regexec(const regex_t *preg, const char *string, 9953 size_t nmatch, regmatch_t pmatch[], int eflags); 9954 9955 size_t pcre2_regerror(int errcode, const regex_t *preg, 9956 char *errbuf, size_t errbuf_size); 9957 9958 void pcre2_regfree(regex_t *preg); 9959 9960 9961DESCRIPTION 9962 9963 This set of functions provides a POSIX-style API for the PCRE2 regular 9964 expression 8-bit library. There are no POSIX-style wrappers for PCRE2's 9965 16-bit and 32-bit libraries. See the pcre2api documentation for a de- 9966 scription of PCRE2's native API, which contains much additional func- 9967 tionality. 9968 9969 The functions described here are wrapper functions that ultimately call 9970 the PCRE2 native API. Their prototypes are defined in the pcre2posix.h 9971 header file, and they all have unique names starting with pcre2_. How- 9972 ever, the pcre2posix.h header also contains macro definitions that con- 9973 vert the standard POSIX names such regcomp() into pcre2_regcomp() etc. 9974 This means that a program can use the usual POSIX names without running 9975 the risk of accidentally linking with POSIX functions from a different 9976 library. 9977 9978 On Unix-like systems the PCRE2 POSIX library is called libpcre2-posix, 9979 so can be accessed by adding -lpcre2-posix to the command for linking 9980 an application. Because the POSIX functions call the native ones, it is 9981 also necessary to add -lpcre2-8. 9982 9983 Although they were not defined as protypes in pcre2posix.h, releases 9984 10.33 to 10.36 of the library contained functions with the POSIX names 9985 regcomp() etc. These simply passed their arguments to the PCRE2 func- 9986 tions. These functions were provided for backwards compatibility with 9987 earlier versions of PCRE2, which had only POSIX names. However, this 9988 has proved troublesome in situations where a program links with several 9989 libraries, some of which use PCRE2's POSIX interface while others use 9990 the real POSIX functions. For this reason, the POSIX names have been 9991 removed since release 10.37. 9992 9993 Calling the header file pcre2posix.h avoids any conflict with other 9994 POSIX libraries. It can, of course, be renamed or aliased as regex.h, 9995 which is the "correct" name, if there is no clash. It provides two 9996 structure types, regex_t for compiled internal forms, and regmatch_t 9997 for returning captured substrings. It also defines some constants whose 9998 names start with "REG_"; these are used for setting options and identi- 9999 fying error codes. 10000 10001 10002USING THE POSIX FUNCTIONS 10003 10004 Those POSIX option bits that can reasonably be mapped to PCRE2 native 10005 options have been implemented. In addition, the option REG_EXTENDED is 10006 defined with the value zero. This has no effect, but since programs 10007 that are written to the POSIX interface often use it, this makes it 10008 easier to slot in PCRE2 as a replacement library. Other POSIX options 10009 are not even defined. 10010 10011 There are also some options that are not defined by POSIX. These have 10012 been added at the request of users who want to make use of certain 10013 PCRE2-specific features via the POSIX calling interface or to add BSD 10014 or GNU functionality. 10015 10016 When PCRE2 is called via these functions, it is only the API that is 10017 POSIX-like in style. The syntax and semantics of the regular expres- 10018 sions themselves are still those of Perl, subject to the setting of 10019 various PCRE2 options, as described below. "POSIX-like in style" means 10020 that the API approximates to the POSIX definition; it is not fully 10021 POSIX-compatible, and in multi-unit encoding domains it is probably 10022 even less compatible. 10023 10024 The descriptions below use the actual names of the functions, but, as 10025 described above, the standard POSIX names (without the pcre2_ prefix) 10026 may also be used. 10027 10028 10029COMPILING A PATTERN 10030 10031 The function pcre2_regcomp() is called to compile a pattern into an in- 10032 ternal form. By default, the pattern is a C string terminated by a bi- 10033 nary zero (but see REG_PEND below). The preg argument is a pointer to a 10034 regex_t structure that is used as a base for storing information about 10035 the compiled regular expression. (It is also used for input when 10036 REG_PEND is set.) 10037 10038 The argument cflags is either zero, or contains one or more of the bits 10039 defined by the following macros: 10040 10041 REG_DOTALL 10042 10043 The PCRE2_DOTALL option is set when the regular expression is passed 10044 for compilation to the native function. Note that REG_DOTALL is not 10045 part of the POSIX standard. 10046 10047 REG_ICASE 10048 10049 The PCRE2_CASELESS option is set when the regular expression is passed 10050 for compilation to the native function. 10051 10052 REG_NEWLINE 10053 10054 The PCRE2_MULTILINE option is set when the regular expression is passed 10055 for compilation to the native function. Note that this does not mimic 10056 the defined POSIX behaviour for REG_NEWLINE (see the following sec- 10057 tion). 10058 10059 REG_NOSPEC 10060 10061 The PCRE2_LITERAL option is set when the regular expression is passed 10062 for compilation to the native function. This disables all meta charac- 10063 ters in the pattern, causing it to be treated as a literal string. The 10064 only other options that are allowed with REG_NOSPEC are REG_ICASE, 10065 REG_NOSUB, REG_PEND, and REG_UTF. Note that REG_NOSPEC is not part of 10066 the POSIX standard. 10067 10068 REG_NOSUB 10069 10070 When a pattern that is compiled with this flag is passed to 10071 pcre2_regexec() for matching, the nmatch and pmatch arguments are ig- 10072 nored, and no captured strings are returned. Versions of the PCRE li- 10073 brary prior to 10.22 used to set the PCRE2_NO_AUTO_CAPTURE compile op- 10074 tion, but this no longer happens because it disables the use of back- 10075 references. 10076 10077 REG_PEND 10078 10079 If this option is set, the reg_endp field in the preg structure (which 10080 has the type const char *) must be set to point to the character beyond 10081 the end of the pattern before calling pcre2_regcomp(). The pattern it- 10082 self may now contain binary zeros, which are treated as data charac- 10083 ters. Without REG_PEND, a binary zero terminates the pattern and the 10084 re_endp field is ignored. This is a GNU extension to the POSIX standard 10085 and should be used with caution in software intended to be portable to 10086 other systems. 10087 10088 REG_UCP 10089 10090 The PCRE2_UCP option is set when the regular expression is passed for 10091 compilation to the native function. This causes PCRE2 to use Unicode 10092 properties when matchine \d, \w, etc., instead of just recognizing 10093 ASCII values. Note that REG_UCP is not part of the POSIX standard. 10094 10095 REG_UNGREEDY 10096 10097 The PCRE2_UNGREEDY option is set when the regular expression is passed 10098 for compilation to the native function. Note that REG_UNGREEDY is not 10099 part of the POSIX standard. 10100 10101 REG_UTF 10102 10103 The PCRE2_UTF option is set when the regular expression is passed for 10104 compilation to the native function. This causes the pattern itself and 10105 all data strings used for matching it to be treated as UTF-8 strings. 10106 Note that REG_UTF is not part of the POSIX standard. 10107 10108 In the absence of these flags, no options are passed to the native 10109 function. This means the the regex is compiled with PCRE2 default se- 10110 mantics. In particular, the way it handles newline characters in the 10111 subject string is the Perl way, not the POSIX way. Note that setting 10112 PCRE2_MULTILINE has only some of the effects specified for REG_NEWLINE. 10113 It does not affect the way newlines are matched by the dot metacharac- 10114 ter (they are not) or by a negative class such as [^a] (they are). 10115 10116 The yield of pcre2_regcomp() is zero on success, and non-zero other- 10117 wise. The preg structure is filled in on success, and one other member 10118 of the structure (as well as re_endp) is public: re_nsub contains the 10119 number of capturing subpatterns in the regular expression. Various er- 10120 ror codes are defined in the header file. 10121 10122 NOTE: If the yield of pcre2_regcomp() is non-zero, you must not attempt 10123 to use the contents of the preg structure. If, for example, you pass it 10124 to pcre2_regexec(), the result is undefined and your program is likely 10125 to crash. 10126 10127 10128MATCHING NEWLINE CHARACTERS 10129 10130 This area is not simple, because POSIX and Perl take different views of 10131 things. It is not possible to get PCRE2 to obey POSIX semantics, but 10132 then PCRE2 was never intended to be a POSIX engine. The following table 10133 lists the different possibilities for matching newline characters in 10134 Perl and PCRE2: 10135 10136 Default Change with 10137 10138 . matches newline no PCRE2_DOTALL 10139 newline matches [^a] yes not changeable 10140 $ matches \n at end yes PCRE2_DOLLAR_ENDONLY 10141 $ matches \n in middle no PCRE2_MULTILINE 10142 ^ matches \n in middle no PCRE2_MULTILINE 10143 10144 This is the equivalent table for a POSIX-compatible pattern matcher: 10145 10146 Default Change with 10147 10148 . matches newline yes REG_NEWLINE 10149 newline matches [^a] yes REG_NEWLINE 10150 $ matches \n at end no REG_NEWLINE 10151 $ matches \n in middle no REG_NEWLINE 10152 ^ matches \n in middle no REG_NEWLINE 10153 10154 This behaviour is not what happens when PCRE2 is called via its POSIX 10155 API. By default, PCRE2's behaviour is the same as Perl's, except that 10156 there is no equivalent for PCRE2_DOLLAR_ENDONLY in Perl. In both PCRE2 10157 and Perl, there is no way to stop newline from matching [^a]. 10158 10159 Default POSIX newline handling can be obtained by setting PCRE2_DOTALL 10160 and PCRE2_DOLLAR_ENDONLY when calling pcre2_compile() directly, but 10161 there is no way to make PCRE2 behave exactly as for the REG_NEWLINE ac- 10162 tion. When using the POSIX API, passing REG_NEWLINE to PCRE2's 10163 pcre2_regcomp() function causes PCRE2_MULTILINE to be passed to 10164 pcre2_compile(), and REG_DOTALL passes PCRE2_DOTALL. There is no way to 10165 pass PCRE2_DOLLAR_ENDONLY. 10166 10167 10168MATCHING A PATTERN 10169 10170 The function pcre2_regexec() is called to match a compiled pattern preg 10171 against a given string, which is by default terminated by a zero byte 10172 (but see REG_STARTEND below), subject to the options in eflags. These 10173 can be: 10174 10175 REG_NOTBOL 10176 10177 The PCRE2_NOTBOL option is set when calling the underlying PCRE2 match- 10178 ing function. 10179 10180 REG_NOTEMPTY 10181 10182 The PCRE2_NOTEMPTY option is set when calling the underlying PCRE2 10183 matching function. Note that REG_NOTEMPTY is not part of the POSIX 10184 standard. However, setting this option can give more POSIX-like behav- 10185 iour in some situations. 10186 10187 REG_NOTEOL 10188 10189 The PCRE2_NOTEOL option is set when calling the underlying PCRE2 match- 10190 ing function. 10191 10192 REG_STARTEND 10193 10194 When this option is set, the subject string starts at string + 10195 pmatch[0].rm_so and ends at string + pmatch[0].rm_eo, which should 10196 point to the first character beyond the string. There may be binary ze- 10197 ros within the subject string, and indeed, using REG_STARTEND is the 10198 only way to pass a subject string that contains a binary zero. 10199 10200 Whatever the value of pmatch[0].rm_so, the offsets of the matched 10201 string and any captured substrings are still given relative to the 10202 start of string itself. (Before PCRE2 release 10.30 these were given 10203 relative to string + pmatch[0].rm_so, but this differs from other im- 10204 plementations.) 10205 10206 This is a BSD extension, compatible with but not specified by IEEE 10207 Standard 1003.2 (POSIX.2), and should be used with caution in software 10208 intended to be portable to other systems. Note that a non-zero rm_so 10209 does not imply REG_NOTBOL; REG_STARTEND affects only the location and 10210 length of the string, not how it is matched. Setting REG_STARTEND and 10211 passing pmatch as NULL are mutually exclusive; the error REG_INVARG is 10212 returned. 10213 10214 If the pattern was compiled with the REG_NOSUB flag, no data about any 10215 matched strings is returned. The nmatch and pmatch arguments of 10216 pcre2_regexec() are ignored (except possibly as input for REG_STAR- 10217 TEND). 10218 10219 The value of nmatch may be zero, and the value pmatch may be NULL (un- 10220 less REG_STARTEND is set); in both these cases no data about any 10221 matched strings is returned. 10222 10223 Otherwise, the portion of the string that was matched, and also any 10224 captured substrings, are returned via the pmatch argument, which points 10225 to an array of nmatch structures of type regmatch_t, containing the 10226 members rm_so and rm_eo. These contain the byte offset to the first 10227 character of each substring and the offset to the first character after 10228 the end of each substring, respectively. The 0th element of the vector 10229 relates to the entire portion of string that was matched; subsequent 10230 elements relate to the capturing subpatterns of the regular expression. 10231 Unused entries in the array have both structure members set to -1. 10232 10233 A successful match yields a zero return; various error codes are de- 10234 fined in the header file, of which REG_NOMATCH is the "expected" fail- 10235 ure code. 10236 10237 10238ERROR MESSAGES 10239 10240 The pcre2_regerror() function maps a non-zero errorcode from either 10241 pcre2_regcomp() or pcre2_regexec() to a printable message. If preg is 10242 not NULL, the error should have arisen from the use of that structure. 10243 A message terminated by a binary zero is placed in errbuf. If the buf- 10244 fer is too short, only the first errbuf_size - 1 characters of the er- 10245 ror message are used. The yield of the function is the size of buffer 10246 needed to hold the whole message, including the terminating zero. This 10247 value is greater than errbuf_size if the message was truncated. 10248 10249 10250MEMORY USAGE 10251 10252 Compiling a regular expression causes memory to be allocated and asso- 10253 ciated with the preg structure. The function pcre2_regfree() frees all 10254 such memory, after which preg may no longer be used as a compiled ex- 10255 pression. 10256 10257 10258AUTHOR 10259 10260 Philip Hazel 10261 University Computing Service 10262 Cambridge, England. 10263 10264 10265REVISION 10266 10267 Last updated: 26 April 2021 10268 Copyright (c) 1997-2021 University of Cambridge. 10269------------------------------------------------------------------------------ 10270 10271 10272PCRE2SAMPLE(3) Library Functions Manual PCRE2SAMPLE(3) 10273 10274 10275 10276NAME 10277 PCRE2 - Perl-compatible regular expressions (revised API) 10278 10279PCRE2 SAMPLE PROGRAM 10280 10281 A simple, complete demonstration program to get you started with using 10282 PCRE2 is supplied in the file pcre2demo.c in the src directory in the 10283 PCRE2 distribution. A listing of this program is given in the pcre2demo 10284 documentation. If you do not have a copy of the PCRE2 distribution, you 10285 can save this listing to re-create the contents of pcre2demo.c. 10286 10287 The demonstration program compiles the regular expression that is its 10288 first argument, and matches it against the subject string in its second 10289 argument. No PCRE2 options are set, and default character tables are 10290 used. If matching succeeds, the program outputs the portion of the sub- 10291 ject that matched, together with the contents of any captured sub- 10292 strings. 10293 10294 If the -g option is given on the command line, the program then goes on 10295 to check for further matches of the same regular expression in the same 10296 subject string. The logic is a little bit tricky because of the possi- 10297 bility of matching an empty string. Comments in the code explain what 10298 is going on. 10299 10300 The code in pcre2demo.c is an 8-bit program that uses the PCRE2 8-bit 10301 library. It handles strings and characters that are stored in 8-bit 10302 code units. By default, one character corresponds to one code unit, 10303 but if the pattern starts with "(*UTF)", both it and the subject are 10304 treated as UTF-8 strings, where characters may occupy multiple code 10305 units. 10306 10307 If PCRE2 is installed in the standard include and library directories 10308 for your operating system, you should be able to compile the demonstra- 10309 tion program using a command like this: 10310 10311 cc -o pcre2demo pcre2demo.c -lpcre2-8 10312 10313 If PCRE2 is installed elsewhere, you may need to add additional options 10314 to the command line. For example, on a Unix-like system that has PCRE2 10315 installed in /usr/local, you can compile the demonstration program us- 10316 ing a command like this: 10317 10318 cc -o pcre2demo -I/usr/local/include pcre2demo.c \ 10319 -L/usr/local/lib -lpcre2-8 10320 10321 Once you have built the demonstration program, you can run simple tests 10322 like this: 10323 10324 ./pcre2demo 'cat|dog' 'the cat sat on the mat' 10325 ./pcre2demo -g 'cat|dog' 'the dog sat on the cat' 10326 10327 Note that there is a much more comprehensive test program, called 10328 pcre2test, which supports many more facilities for testing regular ex- 10329 pressions using all three PCRE2 libraries (8-bit, 16-bit, and 32-bit, 10330 though not all three need be installed). The pcre2demo program is pro- 10331 vided as a relatively simple coding example. 10332 10333 If you try to run pcre2demo when PCRE2 is not installed in the standard 10334 library directory, you may get an error like this on some operating 10335 systems (e.g. Solaris): 10336 10337 ld.so.1: pcre2demo: fatal: libpcre2-8.so.0: open failed: No such file 10338 or directory 10339 10340 This is caused by the way shared library support works on those sys- 10341 tems. You need to add 10342 10343 -R/usr/local/lib 10344 10345 (for example) to the compile command to get round this problem. 10346 10347 10348AUTHOR 10349 10350 Philip Hazel 10351 University Computing Service 10352 Cambridge, England. 10353 10354 10355REVISION 10356 10357 Last updated: 02 February 2016 10358 Copyright (c) 1997-2016 University of Cambridge. 10359------------------------------------------------------------------------------ 10360PCRE2SERIALIZE(3) Library Functions Manual PCRE2SERIALIZE(3) 10361 10362 10363 10364NAME 10365 PCRE2 - Perl-compatible regular expressions (revised API) 10366 10367SAVING AND RE-USING PRECOMPILED PCRE2 PATTERNS 10368 10369 int32_t pcre2_serialize_decode(pcre2_code **codes, 10370 int32_t number_of_codes, const uint8_t *bytes, 10371 pcre2_general_context *gcontext); 10372 10373 int32_t pcre2_serialize_encode(const pcre2_code **codes, 10374 int32_t number_of_codes, uint8_t **serialized_bytes, 10375 PCRE2_SIZE *serialized_size, pcre2_general_context *gcontext); 10376 10377 void pcre2_serialize_free(uint8_t *bytes); 10378 10379 int32_t pcre2_serialize_get_number_of_codes(const uint8_t *bytes); 10380 10381 If you are running an application that uses a large number of regular 10382 expression patterns, it may be useful to store them in a precompiled 10383 form instead of having to compile them every time the application is 10384 run. However, if you are using the just-in-time optimization feature, 10385 it is not possible to save and reload the JIT data, because it is posi- 10386 tion-dependent. The host on which the patterns are reloaded must be 10387 running the same version of PCRE2, with the same code unit width, and 10388 must also have the same endianness, pointer width and PCRE2_SIZE type. 10389 For example, patterns compiled on a 32-bit system using PCRE2's 16-bit 10390 library cannot be reloaded on a 64-bit system, nor can they be reloaded 10391 using the 8-bit library. 10392 10393 Note that "serialization" in PCRE2 does not convert compiled patterns 10394 to an abstract format like Java or .NET serialization. The serialized 10395 output is really just a bytecode dump, which is why it can only be 10396 reloaded in the same environment as the one that created it. Hence the 10397 restrictions mentioned above. Applications that are not statically 10398 linked with a fixed version of PCRE2 must be prepared to recompile pat- 10399 terns from their sources, in order to be immune to PCRE2 upgrades. 10400 10401 10402SECURITY CONCERNS 10403 10404 The facility for saving and restoring compiled patterns is intended for 10405 use within individual applications. As such, the data supplied to 10406 pcre2_serialize_decode() is expected to be trusted data, not data from 10407 arbitrary external sources. There is only some simple consistency 10408 checking, not complete validation of what is being re-loaded. Corrupted 10409 data may cause undefined results. For example, if the length field of a 10410 pattern in the serialized data is corrupted, the deserializing code may 10411 read beyond the end of the byte stream that is passed to it. 10412 10413 10414SAVING COMPILED PATTERNS 10415 10416 Before compiled patterns can be saved they must be serialized, which in 10417 PCRE2 means converting the pattern to a stream of bytes. A single byte 10418 stream may contain any number of compiled patterns, but they must all 10419 use the same character tables. A single copy of the tables is included 10420 in the byte stream (its size is 1088 bytes). For more details of char- 10421 acter tables, see the section on locale support in the pcre2api docu- 10422 mentation. 10423 10424 The function pcre2_serialize_encode() creates a serialized byte stream 10425 from a list of compiled patterns. Its first two arguments specify the 10426 list, being a pointer to a vector of pointers to compiled patterns, and 10427 the length of the vector. The third and fourth arguments point to vari- 10428 ables which are set to point to the created byte stream and its length, 10429 respectively. The final argument is a pointer to a general context, 10430 which can be used to specify custom memory mangagement functions. If 10431 this argument is NULL, malloc() is used to obtain memory for the byte 10432 stream. The yield of the function is the number of serialized patterns, 10433 or one of the following negative error codes: 10434 10435 PCRE2_ERROR_BADDATA the number of patterns is zero or less 10436 PCRE2_ERROR_BADMAGIC mismatch of id bytes in one of the patterns 10437 PCRE2_ERROR_MEMORY memory allocation failed 10438 PCRE2_ERROR_MIXEDTABLES the patterns do not all use the same tables 10439 PCRE2_ERROR_NULL the 1st, 3rd, or 4th argument is NULL 10440 10441 PCRE2_ERROR_BADMAGIC means either that a pattern's code has been cor- 10442 rupted, or that a slot in the vector does not point to a compiled pat- 10443 tern. 10444 10445 Once a set of patterns has been serialized you can save the data in any 10446 appropriate manner. Here is sample code that compiles two patterns and 10447 writes them to a file. It assumes that the variable fd refers to a file 10448 that is open for output. The error checking that should be present in a 10449 real application has been omitted for simplicity. 10450 10451 int errorcode; 10452 uint8_t *bytes; 10453 PCRE2_SIZE erroroffset; 10454 PCRE2_SIZE bytescount; 10455 pcre2_code *list_of_codes[2]; 10456 list_of_codes[0] = pcre2_compile("first pattern", 10457 PCRE2_ZERO_TERMINATED, 0, &errorcode, &erroroffset, NULL); 10458 list_of_codes[1] = pcre2_compile("second pattern", 10459 PCRE2_ZERO_TERMINATED, 0, &errorcode, &erroroffset, NULL); 10460 errorcode = pcre2_serialize_encode(list_of_codes, 2, &bytes, 10461 &bytescount, NULL); 10462 errorcode = fwrite(bytes, 1, bytescount, fd); 10463 10464 Note that the serialized data is binary data that may contain any of 10465 the 256 possible byte values. On systems that make a distinction be- 10466 tween binary and non-binary data, be sure that the file is opened for 10467 binary output. 10468 10469 Serializing a set of patterns leaves the original data untouched, so 10470 they can still be used for matching. Their memory must eventually be 10471 freed in the usual way by calling pcre2_code_free(). When you have fin- 10472 ished with the byte stream, it too must be freed by calling pcre2_seri- 10473 alize_free(). If this function is called with a NULL argument, it re- 10474 turns immediately without doing anything. 10475 10476 10477RE-USING PRECOMPILED PATTERNS 10478 10479 In order to re-use a set of saved patterns you must first make the se- 10480 rialized byte stream available in main memory (for example, by reading 10481 from a file). The management of this memory block is up to the applica- 10482 tion. You can use the pcre2_serialize_get_number_of_codes() function to 10483 find out how many compiled patterns are in the serialized data without 10484 actually decoding the patterns: 10485 10486 uint8_t *bytes = <serialized data>; 10487 int32_t number_of_codes = pcre2_serialize_get_number_of_codes(bytes); 10488 10489 The pcre2_serialize_decode() function reads a byte stream and recreates 10490 the compiled patterns in new memory blocks, setting pointers to them in 10491 a vector. The first two arguments are a pointer to a suitable vector 10492 and its length, and the third argument points to a byte stream. The fi- 10493 nal argument is a pointer to a general context, which can be used to 10494 specify custom memory mangagement functions for the decoded patterns. 10495 If this argument is NULL, malloc() and free() are used. After deserial- 10496 ization, the byte stream is no longer needed and can be discarded. 10497 10498 pcre2_code *list_of_codes[2]; 10499 uint8_t *bytes = <serialized data>; 10500 int32_t number_of_codes = 10501 pcre2_serialize_decode(list_of_codes, 2, bytes, NULL); 10502 10503 If the vector is not large enough for all the patterns in the byte 10504 stream, it is filled with those that fit, and the remainder are ig- 10505 nored. The yield of the function is the number of decoded patterns, or 10506 one of the following negative error codes: 10507 10508 PCRE2_ERROR_BADDATA second argument is zero or less 10509 PCRE2_ERROR_BADMAGIC mismatch of id bytes in the data 10510 PCRE2_ERROR_BADMODE mismatch of code unit size or PCRE2 version 10511 PCRE2_ERROR_BADSERIALIZEDDATA other sanity check failure 10512 PCRE2_ERROR_MEMORY memory allocation failed 10513 PCRE2_ERROR_NULL first or third argument is NULL 10514 10515 PCRE2_ERROR_BADMAGIC may mean that the data is corrupt, or that it was 10516 compiled on a system with different endianness. 10517 10518 Decoded patterns can be used for matching in the usual way, and must be 10519 freed by calling pcre2_code_free(). However, be aware that there is a 10520 potential race issue if you are using multiple patterns that were de- 10521 coded from a single byte stream in a multithreaded application. A sin- 10522 gle copy of the character tables is used by all the decoded patterns 10523 and a reference count is used to arrange for its memory to be automati- 10524 cally freed when the last pattern is freed, but there is no locking on 10525 this reference count. Therefore, if you want to call pcre2_code_free() 10526 for these patterns in different threads, you must arrange your own 10527 locking, and ensure that pcre2_code_free() cannot be called by two 10528 threads at the same time. 10529 10530 If a pattern was processed by pcre2_jit_compile() before being serial- 10531 ized, the JIT data is discarded and so is no longer available after a 10532 save/restore cycle. You can, however, process a restored pattern with 10533 pcre2_jit_compile() if you wish. 10534 10535 10536AUTHOR 10537 10538 Philip Hazel 10539 University Computing Service 10540 Cambridge, England. 10541 10542 10543REVISION 10544 10545 Last updated: 27 June 2018 10546 Copyright (c) 1997-2018 University of Cambridge. 10547------------------------------------------------------------------------------ 10548 10549 10550PCRE2SYNTAX(3) Library Functions Manual PCRE2SYNTAX(3) 10551 10552 10553 10554NAME 10555 PCRE2 - Perl-compatible regular expressions (revised API) 10556 10557PCRE2 REGULAR EXPRESSION SYNTAX SUMMARY 10558 10559 The full syntax and semantics of the regular expressions that are sup- 10560 ported by PCRE2 are described in the pcre2pattern documentation. This 10561 document contains a quick-reference summary of the syntax. 10562 10563 10564QUOTING 10565 10566 \x where x is non-alphanumeric is a literal x 10567 \Q...\E treat enclosed characters as literal 10568 10569 10570ESCAPED CHARACTERS 10571 10572 This table applies to ASCII and Unicode environments. An unrecognized 10573 escape sequence causes an error. 10574 10575 \a alarm, that is, the BEL character (hex 07) 10576 \cx "control-x", where x is any ASCII printing character 10577 \e escape (hex 1B) 10578 \f form feed (hex 0C) 10579 \n newline (hex 0A) 10580 \r carriage return (hex 0D) 10581 \t tab (hex 09) 10582 \0dd character with octal code 0dd 10583 \ddd character with octal code ddd, or backreference 10584 \o{ddd..} character with octal code ddd.. 10585 \N{U+hh..} character with Unicode code point hh.. (Unicode mode only) 10586 \xhh character with hex code hh 10587 \x{hh..} character with hex code hh.. 10588 10589 If PCRE2_ALT_BSUX or PCRE2_EXTRA_ALT_BSUX is set ("ALT_BSUX mode"), the 10590 following are also recognized: 10591 10592 \U the character "U" 10593 \uhhhh character with hex code hhhh 10594 \u{hh..} character with hex code hh.. but only for EXTRA_ALT_BSUX 10595 10596 When \x is not followed by {, from zero to two hexadecimal digits are 10597 read, but in ALT_BSUX mode \x must be followed by two hexadecimal dig- 10598 its to be recognized as a hexadecimal escape; otherwise it matches a 10599 literal "x". Likewise, if \u (in ALT_BSUX mode) is not followed by 10600 four hexadecimal digits or (in EXTRA_ALT_BSUX mode) a sequence of hex 10601 digits in curly brackets, it matches a literal "u". 10602 10603 Note that \0dd is always an octal code. The treatment of backslash fol- 10604 lowed by a non-zero digit is complicated; for details see the section 10605 "Non-printing characters" in the pcre2pattern documentation, where de- 10606 tails of escape processing in EBCDIC environments are also given. 10607 \N{U+hh..} is synonymous with \x{hh..} in PCRE2 but is not supported in 10608 EBCDIC environments. Note that \N not followed by an opening curly 10609 bracket has a different meaning (see below). 10610 10611 10612CHARACTER TYPES 10613 10614 . any character except newline; 10615 in dotall mode, any character whatsoever 10616 \C one code unit, even in UTF mode (best avoided) 10617 \d a decimal digit 10618 \D a character that is not a decimal digit 10619 \h a horizontal white space character 10620 \H a character that is not a horizontal white space character 10621 \N a character that is not a newline 10622 \p{xx} a character with the xx property 10623 \P{xx} a character without the xx property 10624 \R a newline sequence 10625 \s a white space character 10626 \S a character that is not a white space character 10627 \v a vertical white space character 10628 \V a character that is not a vertical white space character 10629 \w a "word" character 10630 \W a "non-word" character 10631 \X a Unicode extended grapheme cluster 10632 10633 \C is dangerous because it may leave the current matching point in the 10634 middle of a UTF-8 or UTF-16 character. The application can lock out the 10635 use of \C by setting the PCRE2_NEVER_BACKSLASH_C option. It is also 10636 possible to build PCRE2 with the use of \C permanently disabled. 10637 10638 By default, \d, \s, and \w match only ASCII characters, even in UTF-8 10639 mode or in the 16-bit and 32-bit libraries. However, if locale-specific 10640 matching is happening, \s and \w may also match characters with code 10641 points in the range 128-255. If the PCRE2_UCP option is set, the behav- 10642 iour of these escape sequences is changed to use Unicode properties and 10643 they match many more characters. 10644 10645 Property descriptions in \p and \P are matched caselessly; hyphens, un- 10646 derscores, and white space are ignored, in accordance with Unicode's 10647 "loose matching" rules. 10648 10649 10650GENERAL CATEGORY PROPERTIES FOR \p and \P 10651 10652 C Other 10653 Cc Control 10654 Cf Format 10655 Cn Unassigned 10656 Co Private use 10657 Cs Surrogate 10658 10659 L Letter 10660 Ll Lower case letter 10661 Lm Modifier letter 10662 Lo Other letter 10663 Lt Title case letter 10664 Lu Upper case letter 10665 Lc Ll, Lu, or Lt 10666 L& Ll, Lu, or Lt 10667 10668 M Mark 10669 Mc Spacing mark 10670 Me Enclosing mark 10671 Mn Non-spacing mark 10672 10673 N Number 10674 Nd Decimal number 10675 Nl Letter number 10676 No Other number 10677 10678 P Punctuation 10679 Pc Connector punctuation 10680 Pd Dash punctuation 10681 Pe Close punctuation 10682 Pf Final punctuation 10683 Pi Initial punctuation 10684 Po Other punctuation 10685 Ps Open punctuation 10686 10687 S Symbol 10688 Sc Currency symbol 10689 Sk Modifier symbol 10690 Sm Mathematical symbol 10691 So Other symbol 10692 10693 Z Separator 10694 Zl Line separator 10695 Zp Paragraph separator 10696 Zs Space separator 10697 10698 10699PCRE2 SPECIAL CATEGORY PROPERTIES FOR \p and \P 10700 10701 Xan Alphanumeric: union of properties L and N 10702 Xps POSIX space: property Z or tab, NL, VT, FF, CR 10703 Xsp Perl space: property Z or tab, NL, VT, FF, CR 10704 Xuc Univerally-named character: one that can be 10705 represented by a Universal Character Name 10706 Xwd Perl word: property Xan or underscore 10707 10708 Perl and POSIX space are now the same. Perl added VT to its space char- 10709 acter set at release 5.18. 10710 10711 10712BINARY PROPERTIES FOR \p AND \P 10713 10714 Unicode defines a number of binary properties, that is, properties 10715 whose only values are true or false. You can obtain a list of those 10716 that are recognized by \p and \P, along with their abbreviations, by 10717 running this command: 10718 10719 pcre2test -LP 10720 10721 10722SCRIPT MATCHING WITH \p AND \P 10723 10724 Many script names and their 4-letter abbreviations are recognized in 10725 \p{sc:...} or \p{scx:...} items, or on their own with \p (and also \P 10726 of course). You can obtain a list of these scripts by running this com- 10727 mand: 10728 10729 pcre2test -LS 10730 10731 10732THE BIDI_CLASS PROPERTY FOR \p AND \P 10733 10734 \p{Bidi_Class:<class>} matches a character with the given class 10735 \p{BC:<class>} matches a character with the given class 10736 10737 The recognized classes are: 10738 10739 AL Arabic letter 10740 AN Arabic number 10741 B paragraph separator 10742 BN boundary neutral 10743 CS common separator 10744 EN European number 10745 ES European separator 10746 ET European terminator 10747 FSI first strong isolate 10748 L left-to-right 10749 LRE left-to-right embedding 10750 LRI left-to-right isolate 10751 LRO left-to-right override 10752 NSM non-spacing mark 10753 ON other neutral 10754 PDF pop directional format 10755 PDI pop directional isolate 10756 R right-to-left 10757 RLE right-to-left embedding 10758 RLI right-to-left isolate 10759 RLO right-to-left override 10760 S segment separator 10761 WS which space 10762 10763 10764CHARACTER CLASSES 10765 10766 [...] positive character class 10767 [^...] negative character class 10768 [x-y] range (can be used for hex characters) 10769 [[:xxx:]] positive POSIX named set 10770 [[:^xxx:]] negative POSIX named set 10771 10772 alnum alphanumeric 10773 alpha alphabetic 10774 ascii 0-127 10775 blank space or tab 10776 cntrl control character 10777 digit decimal digit 10778 graph printing, excluding space 10779 lower lower case letter 10780 print printing, including space 10781 punct printing, excluding alphanumeric 10782 space white space 10783 upper upper case letter 10784 word same as \w 10785 xdigit hexadecimal digit 10786 10787 In PCRE2, POSIX character set names recognize only ASCII characters by 10788 default, but some of them use Unicode properties if PCRE2_UCP is set. 10789 You can use \Q...\E inside a character class. 10790 10791 10792QUANTIFIERS 10793 10794 ? 0 or 1, greedy 10795 ?+ 0 or 1, possessive 10796 ?? 0 or 1, lazy 10797 * 0 or more, greedy 10798 *+ 0 or more, possessive 10799 *? 0 or more, lazy 10800 + 1 or more, greedy 10801 ++ 1 or more, possessive 10802 +? 1 or more, lazy 10803 {n} exactly n 10804 {n,m} at least n, no more than m, greedy 10805 {n,m}+ at least n, no more than m, possessive 10806 {n,m}? at least n, no more than m, lazy 10807 {n,} n or more, greedy 10808 {n,}+ n or more, possessive 10809 {n,}? n or more, lazy 10810 10811 10812ANCHORS AND SIMPLE ASSERTIONS 10813 10814 \b word boundary 10815 \B not a word boundary 10816 ^ start of subject 10817 also after an internal newline in multiline mode 10818 (after any newline if PCRE2_ALT_CIRCUMFLEX is set) 10819 \A start of subject 10820 $ end of subject 10821 also before newline at end of subject 10822 also before internal newline in multiline mode 10823 \Z end of subject 10824 also before newline at end of subject 10825 \z end of subject 10826 \G first matching position in subject 10827 10828 10829REPORTED MATCH POINT SETTING 10830 10831 \K set reported start of match 10832 10833 From release 10.38 \K is not permitted by default in lookaround asser- 10834 tions, for compatibility with Perl. However, if the PCRE2_EXTRA_AL- 10835 LOW_LOOKAROUND_BSK option is set, the previous behaviour is re-enabled. 10836 When this option is set, \K is honoured in positive assertions, but ig- 10837 nored in negative ones. 10838 10839 10840ALTERNATION 10841 10842 expr|expr|expr... 10843 10844 10845CAPTURING 10846 10847 (...) capture group 10848 (?<name>...) named capture group (Perl) 10849 (?'name'...) named capture group (Perl) 10850 (?P<name>...) named capture group (Python) 10851 (?:...) non-capture group 10852 (?|...) non-capture group; reset group numbers for 10853 capture groups in each alternative 10854 10855 In non-UTF modes, names may contain underscores and ASCII letters and 10856 digits; in UTF modes, any Unicode letters and Unicode decimal digits 10857 are permitted. In both cases, a name must not start with a digit. 10858 10859 10860ATOMIC GROUPS 10861 10862 (?>...) atomic non-capture group 10863 (*atomic:...) atomic non-capture group 10864 10865 10866COMMENT 10867 10868 (?#....) comment (not nestable) 10869 10870 10871OPTION SETTING 10872 Changes of these options within a group are automatically cancelled at 10873 the end of the group. 10874 10875 (?i) caseless 10876 (?J) allow duplicate named groups 10877 (?m) multiline 10878 (?n) no auto capture 10879 (?s) single line (dotall) 10880 (?U) default ungreedy (lazy) 10881 (?x) extended: ignore white space except in classes 10882 (?xx) as (?x) but also ignore space and tab in classes 10883 (?-...) unset option(s) 10884 (?^) unset imnsx options 10885 10886 Unsetting x or xx unsets both. Several options may be set at once, and 10887 a mixture of setting and unsetting such as (?i-x) is allowed, but there 10888 may be only one hyphen. Setting (but no unsetting) is allowed after (?^ 10889 for example (?^in). An option setting may appear at the start of a non- 10890 capture group, for example (?i:...). 10891 10892 The following are recognized only at the very start of a pattern or af- 10893 ter one of the newline or \R options with similar syntax. More than one 10894 of them may appear. For the first three, d is a decimal number. 10895 10896 (*LIMIT_DEPTH=d) set the backtracking limit to d 10897 (*LIMIT_HEAP=d) set the heap size limit to d * 1024 bytes 10898 (*LIMIT_MATCH=d) set the match limit to d 10899 (*NOTEMPTY) set PCRE2_NOTEMPTY when matching 10900 (*NOTEMPTY_ATSTART) set PCRE2_NOTEMPTY_ATSTART when matching 10901 (*NO_AUTO_POSSESS) no auto-possessification (PCRE2_NO_AUTO_POSSESS) 10902 (*NO_DOTSTAR_ANCHOR) no .* anchoring (PCRE2_NO_DOTSTAR_ANCHOR) 10903 (*NO_JIT) disable JIT optimization 10904 (*NO_START_OPT) no start-match optimization (PCRE2_NO_START_OPTIMIZE) 10905 (*UTF) set appropriate UTF mode for the library in use 10906 (*UCP) set PCRE2_UCP (use Unicode properties for \d etc) 10907 10908 Note that LIMIT_DEPTH, LIMIT_HEAP, and LIMIT_MATCH can only reduce the 10909 value of the limits set by the caller of pcre2_match() or 10910 pcre2_dfa_match(), not increase them. LIMIT_RECURSION is an obsolete 10911 synonym for LIMIT_DEPTH. The application can lock out the use of (*UTF) 10912 and (*UCP) by setting the PCRE2_NEVER_UTF or PCRE2_NEVER_UCP options, 10913 respectively, at compile time. 10914 10915 10916NEWLINE CONVENTION 10917 10918 These are recognized only at the very start of the pattern or after op- 10919 tion settings with a similar syntax. 10920 10921 (*CR) carriage return only 10922 (*LF) linefeed only 10923 (*CRLF) carriage return followed by linefeed 10924 (*ANYCRLF) all three of the above 10925 (*ANY) any Unicode newline sequence 10926 (*NUL) the NUL character (binary zero) 10927 10928 10929WHAT \R MATCHES 10930 10931 These are recognized only at the very start of the pattern or after op- 10932 tion setting with a similar syntax. 10933 10934 (*BSR_ANYCRLF) CR, LF, or CRLF 10935 (*BSR_UNICODE) any Unicode newline sequence 10936 10937 10938LOOKAHEAD AND LOOKBEHIND ASSERTIONS 10939 10940 (?=...) ) 10941 (*pla:...) ) positive lookahead 10942 (*positive_lookahead:...) ) 10943 10944 (?!...) ) 10945 (*nla:...) ) negative lookahead 10946 (*negative_lookahead:...) ) 10947 10948 (?<=...) ) 10949 (*plb:...) ) positive lookbehind 10950 (*positive_lookbehind:...) ) 10951 10952 (?<!...) ) 10953 (*nlb:...) ) negative lookbehind 10954 (*negative_lookbehind:...) ) 10955 10956 Each top-level branch of a lookbehind must be of a fixed length. 10957 10958 10959NON-ATOMIC LOOKAROUND ASSERTIONS 10960 10961 These assertions are specific to PCRE2 and are not Perl-compatible. 10962 10963 (?*...) ) 10964 (*napla:...) ) synonyms 10965 (*non_atomic_positive_lookahead:...) ) 10966 10967 (?<*...) ) 10968 (*naplb:...) ) synonyms 10969 (*non_atomic_positive_lookbehind:...) ) 10970 10971 10972SCRIPT RUNS 10973 10974 (*script_run:...) ) script run, can be backtracked into 10975 (*sr:...) ) 10976 10977 (*atomic_script_run:...) ) atomic script run 10978 (*asr:...) ) 10979 10980 10981BACKREFERENCES 10982 10983 \n reference by number (can be ambiguous) 10984 \gn reference by number 10985 \g{n} reference by number 10986 \g+n relative reference by number (PCRE2 extension) 10987 \g-n relative reference by number 10988 \g{+n} relative reference by number (PCRE2 extension) 10989 \g{-n} relative reference by number 10990 \k<name> reference by name (Perl) 10991 \k'name' reference by name (Perl) 10992 \g{name} reference by name (Perl) 10993 \k{name} reference by name (.NET) 10994 (?P=name) reference by name (Python) 10995 10996 10997SUBROUTINE REFERENCES (POSSIBLY RECURSIVE) 10998 10999 (?R) recurse whole pattern 11000 (?n) call subroutine by absolute number 11001 (?+n) call subroutine by relative number 11002 (?-n) call subroutine by relative number 11003 (?&name) call subroutine by name (Perl) 11004 (?P>name) call subroutine by name (Python) 11005 \g<name> call subroutine by name (Oniguruma) 11006 \g'name' call subroutine by name (Oniguruma) 11007 \g<n> call subroutine by absolute number (Oniguruma) 11008 \g'n' call subroutine by absolute number (Oniguruma) 11009 \g<+n> call subroutine by relative number (PCRE2 extension) 11010 \g'+n' call subroutine by relative number (PCRE2 extension) 11011 \g<-n> call subroutine by relative number (PCRE2 extension) 11012 \g'-n' call subroutine by relative number (PCRE2 extension) 11013 11014 11015CONDITIONAL PATTERNS 11016 11017 (?(condition)yes-pattern) 11018 (?(condition)yes-pattern|no-pattern) 11019 11020 (?(n) absolute reference condition 11021 (?(+n) relative reference condition 11022 (?(-n) relative reference condition 11023 (?(<name>) named reference condition (Perl) 11024 (?('name') named reference condition (Perl) 11025 (?(name) named reference condition (PCRE2, deprecated) 11026 (?(R) overall recursion condition 11027 (?(Rn) specific numbered group recursion condition 11028 (?(R&name) specific named group recursion condition 11029 (?(DEFINE) define groups for reference 11030 (?(VERSION[>]=n.m) test PCRE2 version 11031 (?(assert) assertion condition 11032 11033 Note the ambiguity of (?(R) and (?(Rn) which might be named reference 11034 conditions or recursion tests. Such a condition is interpreted as a 11035 reference condition if the relevant named group exists. 11036 11037 11038BACKTRACKING CONTROL 11039 11040 All backtracking control verbs may be in the form (*VERB:NAME). For 11041 (*MARK) the name is mandatory, for the others it is optional. (*SKIP) 11042 changes its behaviour if :NAME is present. The others just set a name 11043 for passing back to the caller, but this is not a name that (*SKIP) can 11044 see. The following act immediately they are reached: 11045 11046 (*ACCEPT) force successful match 11047 (*FAIL) force backtrack; synonym (*F) 11048 (*MARK:NAME) set name to be passed back; synonym (*:NAME) 11049 11050 The following act only when a subsequent match failure causes a back- 11051 track to reach them. They all force a match failure, but they differ in 11052 what happens afterwards. Those that advance the start-of-match point do 11053 so only if the pattern is not anchored. 11054 11055 (*COMMIT) overall failure, no advance of starting point 11056 (*PRUNE) advance to next starting character 11057 (*SKIP) advance to current matching position 11058 (*SKIP:NAME) advance to position corresponding to an earlier 11059 (*MARK:NAME); if not found, the (*SKIP) is ignored 11060 (*THEN) local failure, backtrack to next alternation 11061 11062 The effect of one of these verbs in a group called as a subroutine is 11063 confined to the subroutine call. 11064 11065 11066CALLOUTS 11067 11068 (?C) callout (assumed number 0) 11069 (?Cn) callout with numerical data n 11070 (?C"text") callout with string data 11071 11072 The allowed string delimiters are ` ' " ^ % # $ (which are the same for 11073 the start and the end), and the starting delimiter { matched with the 11074 ending delimiter }. To encode the ending delimiter within the string, 11075 double it. 11076 11077 11078SEE ALSO 11079 11080 pcre2pattern(3), pcre2api(3), pcre2callout(3), pcre2matching(3), 11081 pcre2(3). 11082 11083 11084AUTHOR 11085 11086 Philip Hazel 11087 Retired from University Computing Service 11088 Cambridge, England. 11089 11090 11091REVISION 11092 11093 Last updated: 12 January 2022 11094 Copyright (c) 1997-2022 University of Cambridge. 11095------------------------------------------------------------------------------ 11096 11097 11098PCRE2UNICODE(3) Library Functions Manual PCRE2UNICODE(3) 11099 11100 11101 11102NAME 11103 PCRE - Perl-compatible regular expressions (revised API) 11104 11105UNICODE AND UTF SUPPORT 11106 11107 PCRE2 is normally built with Unicode support, though if you do not need 11108 it, you can build it without, in which case the library will be 11109 smaller. With Unicode support, PCRE2 has knowledge of Unicode character 11110 properties and can process strings of text in UTF-8, UTF-16, and UTF-32 11111 format (depending on the code unit width), but this is not the default. 11112 Unless specifically requested, PCRE2 treats each code unit in a string 11113 as one character. 11114 11115 There are two ways of telling PCRE2 to switch to UTF mode, where char- 11116 acters may consist of more than one code unit and the range of values 11117 is constrained. The program can call pcre2_compile() with the PCRE2_UTF 11118 option, or the pattern may start with the sequence (*UTF). However, 11119 the latter facility can be locked out by the PCRE2_NEVER_UTF option. 11120 That is, the programmer can prevent the supplier of the pattern from 11121 switching to UTF mode. 11122 11123 Note that the PCRE2_MATCH_INVALID_UTF option (see below) forces 11124 PCRE2_UTF to be set. 11125 11126 In UTF mode, both the pattern and any subject strings that are matched 11127 against it are treated as UTF strings instead of strings of individual 11128 one-code-unit characters. There are also some other changes to the way 11129 characters are handled, as documented below. 11130 11131 11132UNICODE PROPERTY SUPPORT 11133 11134 When PCRE2 is built with Unicode support, the escape sequences \p{..}, 11135 \P{..}, and \X can be used. This is not dependent on the PCRE2_UTF set- 11136 ting. The Unicode properties that can be tested are a subset of those 11137 that Perl supports. Currently they are limited to the general category 11138 properties such as Lu for an upper case letter or Nd for a decimal num- 11139 ber, the Unicode script names such as Arabic or Han, Bidi_Class, 11140 Bidi_Control, and the derived properties Any and LC (synonym L&). Full 11141 lists are given in the pcre2pattern and pcre2syntax documentation. In 11142 general, only the short names for properties are supported. For exam- 11143 ple, \p{L} matches a letter. Its longer synonym, \p{Letter}, is not 11144 supported. Furthermore, in Perl, many properties may optionally be pre- 11145 fixed by "Is", for compatibility with Perl 5.6. PCRE2 does not support 11146 this. 11147 11148 11149WIDE CHARACTERS AND UTF MODES 11150 11151 Code points less than 256 can be specified in patterns by either braced 11152 or unbraced hexadecimal escape sequences (for example, \x{b3} or \xb3). 11153 Larger values have to use braced sequences. Unbraced octal code points 11154 up to \777 are also recognized; larger ones can be coded using \o{...}. 11155 11156 The escape sequence \N{U+<hex digits>} is recognized as another way of 11157 specifying a Unicode character by code point in a UTF mode. It is not 11158 allowed in non-UTF mode. 11159 11160 In UTF mode, repeat quantifiers apply to complete UTF characters, not 11161 to individual code units. 11162 11163 In UTF mode, the dot metacharacter matches one UTF character instead of 11164 a single code unit. 11165 11166 In UTF mode, capture group names are not restricted to ASCII, and may 11167 contain any Unicode letters and decimal digits, as well as underscore. 11168 11169 The escape sequence \C can be used to match a single code unit in UTF 11170 mode, but its use can lead to some strange effects because it breaks up 11171 multi-unit characters (see the description of \C in the pcre2pattern 11172 documentation). For this reason, there is a build-time option that dis- 11173 ables support for \C completely. There is also a less draconian com- 11174 pile-time option for locking out the use of \C when a pattern is com- 11175 piled. 11176 11177 The use of \C is not supported by the alternative matching function 11178 pcre2_dfa_match() when in UTF-8 or UTF-16 mode, that is, when a charac- 11179 ter may consist of more than one code unit. The use of \C in these 11180 modes provokes a match-time error. Also, the JIT optimization does not 11181 support \C in these modes. If JIT optimization is requested for a UTF-8 11182 or UTF-16 pattern that contains \C, it will not succeed, and so when 11183 pcre2_match() is called, the matching will be carried out by the inter- 11184 pretive function. 11185 11186 The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly test 11187 characters of any code value, but, by default, the characters that 11188 PCRE2 recognizes as digits, spaces, or word characters remain the same 11189 set as in non-UTF mode, all with code points less than 256. This re- 11190 mains true even when PCRE2 is built to include Unicode support, because 11191 to do otherwise would slow down matching in many common cases. Note 11192 that this also applies to \b and \B, because they are defined in terms 11193 of \w and \W. If you want to test for a wider sense of, say, "digit", 11194 you can use explicit Unicode property tests such as \p{Nd}. Alterna- 11195 tively, if you set the PCRE2_UCP option, the way that the character es- 11196 capes work is changed so that Unicode properties are used to determine 11197 which characters match. There are more details in the section on 11198 generic character types in the pcre2pattern documentation. 11199 11200 Similarly, characters that match the POSIX named character classes are 11201 all low-valued characters, unless the PCRE2_UCP option is set. 11202 11203 However, the special horizontal and vertical white space matching es- 11204 capes (\h, \H, \v, and \V) do match all the appropriate Unicode charac- 11205 ters, whether or not PCRE2_UCP is set. 11206 11207 11208UNICODE CASE-EQUIVALENCE 11209 11210 If either PCRE2_UTF or PCRE2_UCP is set, upper/lower case processing 11211 makes use of Unicode properties except for characters whose code points 11212 are less than 128 and that have at most two case-equivalent values. For 11213 these, a direct table lookup is used for speed. A few Unicode charac- 11214 ters such as Greek sigma have more than two code points that are case- 11215 equivalent, and these are treated specially. Setting PCRE2_UCP without 11216 PCRE2_UTF allows Unicode-style case processing for non-UTF character 11217 encodings such as UCS-2. 11218 11219 11220SCRIPT RUNS 11221 11222 The pattern constructs (*script_run:...) and (*atomic_script_run:...), 11223 with synonyms (*sr:...) and (*asr:...), verify that the string matched 11224 within the parentheses is a script run. In concept, a script run is a 11225 sequence of characters that are all from the same Unicode script. How- 11226 ever, because some scripts are commonly used together, and because some 11227 diacritical and other marks are used with multiple scripts, it is not 11228 that simple. 11229 11230 Every Unicode character has a Script property, mostly with a value cor- 11231 responding to the name of a script, such as Latin, Greek, or Cyrillic. 11232 There are also three special values: 11233 11234 "Unknown" is used for code points that have not been assigned, and also 11235 for the surrogate code points. In the PCRE2 32-bit library, characters 11236 whose code points are greater than the Unicode maximum (U+10FFFF), 11237 which are accessible only in non-UTF mode, are assigned the Unknown 11238 script. 11239 11240 "Common" is used for characters that are used with many scripts. These 11241 include punctuation, emoji, mathematical, musical, and currency sym- 11242 bols, and the ASCII digits 0 to 9. 11243 11244 "Inherited" is used for characters such as diacritical marks that mod- 11245 ify a previous character. These are considered to take on the script of 11246 the character that they modify. 11247 11248 Some Inherited characters are used with many scripts, but many of them 11249 are only normally used with a small number of scripts. For example, 11250 U+102E0 (Coptic Epact thousands mark) is used only with Arabic and Cop- 11251 tic. In order to make it possible to check this, a Unicode property 11252 called Script Extension exists. Its value is a list of scripts that ap- 11253 ply to the character. For the majority of characters, the list contains 11254 just one script, the same one as the Script property. However, for 11255 characters such as U+102E0 more than one Script is listed. There are 11256 also some Common characters that have a single, non-Common script in 11257 their Script Extension list. 11258 11259 The next section describes the basic rules for deciding whether a given 11260 string of characters is a script run. Note, however, that there are 11261 some special cases involving the Chinese Han script, and an additional 11262 constraint for decimal digits. These are covered in subsequent sec- 11263 tions. 11264 11265 Basic script run rules 11266 11267 A string that is less than two characters long is a script run. This is 11268 the only case in which an Unknown character can be part of a script 11269 run. Longer strings are checked using only the Script Extensions prop- 11270 erty, not the basic Script property. 11271 11272 If a character's Script Extension property is the single value "Inher- 11273 ited", it is always accepted as part of a script run. This is also true 11274 for the property "Common", subject to the checking of decimal digits 11275 described below. All the remaining characters in a script run must have 11276 at least one script in common in their Script Extension lists. In set- 11277 theoretic terminology, the intersection of all the sets of scripts must 11278 not be empty. 11279 11280 A simple example is an Internet name such as "google.com". The letters 11281 are all in the Latin script, and the dot is Common, so this string is a 11282 script run. However, the Cyrillic letter "o" looks exactly the same as 11283 the Latin "o"; a string that looks the same, but with Cyrillic "o"s is 11284 not a script run. 11285 11286 More interesting examples involve characters with more than one script 11287 in their Script Extension. Consider the following characters: 11288 11289 U+060C Arabic comma 11290 U+06D4 Arabic full stop 11291 11292 The first has the Script Extension list Arabic, Hanifi Rohingya, Syr- 11293 iac, and Thaana; the second has just Arabic and Hanifi Rohingya. Both 11294 of them could appear in script runs of either Arabic or Hanifi Ro- 11295 hingya. The first could also appear in Syriac or Thaana script runs, 11296 but the second could not. 11297 11298 The Chinese Han script 11299 11300 The Chinese Han script is commonly used in conjunction with other 11301 scripts for writing certain languages. Japanese uses the Hiragana and 11302 Katakana scripts together with Han; Korean uses Hangul and Han; Tai- 11303 wanese Mandarin uses Bopomofo and Han. These three combinations are 11304 treated as special cases when checking script runs and are, in effect, 11305 "virtual scripts". Thus, a script run may contain a mixture of Hira- 11306 gana, Katakana, and Han, or a mixture of Hangul and Han, or a mixture 11307 of Bopomofo and Han, but not, for example, a mixture of Hangul and 11308 Bopomofo and Han. PCRE2 (like Perl) follows Unicode's Technical Stan- 11309 dard 39 ("Unicode Security Mechanisms", http://unicode.org/re- 11310 ports/tr39/) in allowing such mixtures. 11311 11312 Decimal digits 11313 11314 Unicode contains many sets of 10 decimal digits in different scripts, 11315 and some scripts (including the Common script) contain more than one 11316 set. Some of these decimal digits them are visually indistinguishable 11317 from the common ASCII digits. In addition to the script checking de- 11318 scribed above, if a script run contains any decimal digits, they must 11319 all come from the same set of 10 adjacent characters. 11320 11321 11322VALIDITY OF UTF STRINGS 11323 11324 When the PCRE2_UTF option is set, the strings passed as patterns and 11325 subjects are (by default) checked for validity on entry to the relevant 11326 functions. If an invalid UTF string is passed, a negative error code is 11327 returned. The code unit offset to the offending character can be ex- 11328 tracted from the match data block by calling pcre2_get_startchar(), 11329 which is used for this purpose after a UTF error. 11330 11331 In some situations, you may already know that your strings are valid, 11332 and therefore want to skip these checks in order to improve perfor- 11333 mance, for example in the case of a long subject string that is being 11334 scanned repeatedly. If you set the PCRE2_NO_UTF_CHECK option at com- 11335 pile time or at match time, PCRE2 assumes that the pattern or subject 11336 it is given (respectively) contains only valid UTF code unit sequences. 11337 11338 If you pass an invalid UTF string when PCRE2_NO_UTF_CHECK is set, the 11339 result is undefined and your program may crash or loop indefinitely or 11340 give incorrect results. There is, however, one mode of matching that 11341 can handle invalid UTF subject strings. This is enabled by passing 11342 PCRE2_MATCH_INVALID_UTF to pcre2_compile() and is discussed below in 11343 the next section. The rest of this section covers the case when 11344 PCRE2_MATCH_INVALID_UTF is not set. 11345 11346 Passing PCRE2_NO_UTF_CHECK to pcre2_compile() just disables the UTF 11347 check for the pattern; it does not also apply to subject strings. If 11348 you want to disable the check for a subject string you must pass this 11349 same option to pcre2_match() or pcre2_dfa_match(). 11350 11351 UTF-16 and UTF-32 strings can indicate their endianness by special code 11352 knows as a byte-order mark (BOM). The PCRE2 functions do not handle 11353 this, expecting strings to be in host byte order. 11354 11355 Unless PCRE2_NO_UTF_CHECK is set, a UTF string is checked before any 11356 other processing takes place. In the case of pcre2_match() and 11357 pcre2_dfa_match() calls with a non-zero starting offset, the check is 11358 applied only to that part of the subject that could be inspected during 11359 matching, and there is a check that the starting offset points to the 11360 first code unit of a character or to the end of the subject. If there 11361 are no lookbehind assertions in the pattern, the check starts at the 11362 starting offset. Otherwise, it starts at the length of the longest 11363 lookbehind before the starting offset, or at the start of the subject 11364 if there are not that many characters before the starting offset. Note 11365 that the sequences \b and \B are one-character lookbehinds. 11366 11367 In addition to checking the format of the string, there is a check to 11368 ensure that all code points lie in the range U+0 to U+10FFFF, excluding 11369 the surrogate area. The so-called "non-character" code points are not 11370 excluded because Unicode corrigendum #9 makes it clear that they should 11371 not be. 11372 11373 Characters in the "Surrogate Area" of Unicode are reserved for use by 11374 UTF-16, where they are used in pairs to encode code points with values 11375 greater than 0xFFFF. The code points that are encoded by UTF-16 pairs 11376 are available independently in the UTF-8 and UTF-32 encodings. (In 11377 other words, the whole surrogate thing is a fudge for UTF-16 which un- 11378 fortunately messes up UTF-8 and UTF-32.) 11379 11380 Setting PCRE2_NO_UTF_CHECK at compile time does not disable the error 11381 that is given if an escape sequence for an invalid Unicode code point 11382 is encountered in the pattern. If you want to allow escape sequences 11383 such as \x{d800} (a surrogate code point) you can set the PCRE2_EX- 11384 TRA_ALLOW_SURROGATE_ESCAPES extra option. However, this is possible 11385 only in UTF-8 and UTF-32 modes, because these values are not repre- 11386 sentable in UTF-16. 11387 11388 Errors in UTF-8 strings 11389 11390 The following negative error codes are given for invalid UTF-8 strings: 11391 11392 PCRE2_ERROR_UTF8_ERR1 11393 PCRE2_ERROR_UTF8_ERR2 11394 PCRE2_ERROR_UTF8_ERR3 11395 PCRE2_ERROR_UTF8_ERR4 11396 PCRE2_ERROR_UTF8_ERR5 11397 11398 The string ends with a truncated UTF-8 character; the code specifies 11399 how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8 11400 characters to be no longer than 4 bytes, the encoding scheme (origi- 11401 nally defined by RFC 2279) allows for up to 6 bytes, and this is 11402 checked first; hence the possibility of 4 or 5 missing bytes. 11403 11404 PCRE2_ERROR_UTF8_ERR6 11405 PCRE2_ERROR_UTF8_ERR7 11406 PCRE2_ERROR_UTF8_ERR8 11407 PCRE2_ERROR_UTF8_ERR9 11408 PCRE2_ERROR_UTF8_ERR10 11409 11410 The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of 11411 the character do not have the binary value 0b10 (that is, either the 11412 most significant bit is 0, or the next bit is 1). 11413 11414 PCRE2_ERROR_UTF8_ERR11 11415 PCRE2_ERROR_UTF8_ERR12 11416 11417 A character that is valid by the RFC 2279 rules is either 5 or 6 bytes 11418 long; these code points are excluded by RFC 3629. 11419 11420 PCRE2_ERROR_UTF8_ERR13 11421 11422 A 4-byte character has a value greater than 0x10ffff; these code points 11423 are excluded by RFC 3629. 11424 11425 PCRE2_ERROR_UTF8_ERR14 11426 11427 A 3-byte character has a value in the range 0xd800 to 0xdfff; this 11428 range of code points are reserved by RFC 3629 for use with UTF-16, and 11429 so are excluded from UTF-8. 11430 11431 PCRE2_ERROR_UTF8_ERR15 11432 PCRE2_ERROR_UTF8_ERR16 11433 PCRE2_ERROR_UTF8_ERR17 11434 PCRE2_ERROR_UTF8_ERR18 11435 PCRE2_ERROR_UTF8_ERR19 11436 11437 A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes 11438 for a value that can be represented by fewer bytes, which is invalid. 11439 For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor- 11440 rect coding uses just one byte. 11441 11442 PCRE2_ERROR_UTF8_ERR20 11443 11444 The two most significant bits of the first byte of a character have the 11445 binary value 0b10 (that is, the most significant bit is 1 and the sec- 11446 ond is 0). Such a byte can only validly occur as the second or subse- 11447 quent byte of a multi-byte character. 11448 11449 PCRE2_ERROR_UTF8_ERR21 11450 11451 The first byte of a character has the value 0xfe or 0xff. These values 11452 can never occur in a valid UTF-8 string. 11453 11454 Errors in UTF-16 strings 11455 11456 The following negative error codes are given for invalid UTF-16 11457 strings: 11458 11459 PCRE2_ERROR_UTF16_ERR1 Missing low surrogate at end of string 11460 PCRE2_ERROR_UTF16_ERR2 Invalid low surrogate follows high surrogate 11461 PCRE2_ERROR_UTF16_ERR3 Isolated low surrogate 11462 11463 11464 Errors in UTF-32 strings 11465 11466 The following negative error codes are given for invalid UTF-32 11467 strings: 11468 11469 PCRE2_ERROR_UTF32_ERR1 Surrogate character (0xd800 to 0xdfff) 11470 PCRE2_ERROR_UTF32_ERR2 Code point is greater than 0x10ffff 11471 11472 11473MATCHING IN INVALID UTF STRINGS 11474 11475 You can run pattern matches on subject strings that may contain invalid 11476 UTF sequences if you call pcre2_compile() with the PCRE2_MATCH_IN- 11477 VALID_UTF option. This is supported by pcre2_match(), including JIT 11478 matching, but not by pcre2_dfa_match(). When PCRE2_MATCH_INVALID_UTF is 11479 set, it forces PCRE2_UTF to be set as well. Note, however, that the 11480 pattern itself must be a valid UTF string. 11481 11482 Setting PCRE2_MATCH_INVALID_UTF does not affect what pcre2_compile() 11483 generates, but if pcre2_jit_compile() is subsequently called, it does 11484 generate different code. If JIT is not used, the option affects the be- 11485 haviour of the interpretive code in pcre2_match(). When PCRE2_MATCH_IN- 11486 VALID_UTF is set at compile time, PCRE2_NO_UTF_CHECK is ignored at 11487 match time. 11488 11489 In this mode, an invalid code unit sequence in the subject never 11490 matches any pattern item. It does not match dot, it does not match 11491 \p{Any}, it does not even match negative items such as [^X]. A lookbe- 11492 hind assertion fails if it encounters an invalid sequence while moving 11493 the current point backwards. In other words, an invalid UTF code unit 11494 sequence acts as a barrier which no match can cross. 11495 11496 You can also think of this as the subject being split up into fragments 11497 of valid UTF, delimited internally by invalid code unit sequences. The 11498 pattern is matched fragment by fragment. The result of a successful 11499 match, however, is given as code unit offsets in the entire subject 11500 string in the usual way. There are a few points to consider: 11501 11502 The internal boundaries are not interpreted as the beginnings or ends 11503 of lines and so do not match circumflex or dollar characters in the 11504 pattern. 11505 11506 If pcre2_match() is called with an offset that points to an invalid 11507 UTF-sequence, that sequence is skipped, and the match starts at the 11508 next valid UTF character, or the end of the subject. 11509 11510 At internal fragment boundaries, \b and \B behave in the same way as at 11511 the beginning and end of the subject. For example, a sequence such as 11512 \bWORD\b would match an instance of WORD that is surrounded by invalid 11513 UTF code units. 11514 11515 Using PCRE2_MATCH_INVALID_UTF, an application can run matches on arbi- 11516 trary data, knowing that any matched strings that are returned are 11517 valid UTF. This can be useful when searching for UTF text in executable 11518 or other binary files. 11519 11520 11521AUTHOR 11522 11523 Philip Hazel 11524 Retired from University Computing Service 11525 Cambridge, England. 11526 11527 11528REVISION 11529 11530 Last updated: 22 December 2021 11531 Copyright (c) 1997-2021 University of Cambridge. 11532------------------------------------------------------------------------------ 11533 11534 11535