1<html> 2<head> 3<title>pcretest specification</title> 4</head> 5<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB"> 6<h1>pcretest man page</h1> 7<p> 8Return to the <a href="index.html">PCRE index page</a>. 9</p> 10<p> 11This page is part of the PCRE HTML documentation. It was generated automatically 12from the original man page. If there is any nonsense in it, please consult the 13man page, in case the conversion went wrong. 14<br> 15<ul> 16<li><a name="TOC1" href="#SEC1">SYNOPSIS</a> 17<li><a name="TOC2" href="#SEC2">INPUT DATA FORMAT</a> 18<li><a name="TOC3" href="#SEC3">PCRE's 8-BIT, 16-BIT AND 32-BIT LIBRARIES</a> 19<li><a name="TOC4" href="#SEC4">COMMAND LINE OPTIONS</a> 20<li><a name="TOC5" href="#SEC5">DESCRIPTION</a> 21<li><a name="TOC6" href="#SEC6">PATTERN MODIFIERS</a> 22<li><a name="TOC7" href="#SEC7">DATA LINES</a> 23<li><a name="TOC8" href="#SEC8">THE ALTERNATIVE MATCHING FUNCTION</a> 24<li><a name="TOC9" href="#SEC9">DEFAULT OUTPUT FROM PCRETEST</a> 25<li><a name="TOC10" href="#SEC10">OUTPUT FROM THE ALTERNATIVE MATCHING FUNCTION</a> 26<li><a name="TOC11" href="#SEC11">RESTARTING AFTER A PARTIAL MATCH</a> 27<li><a name="TOC12" href="#SEC12">CALLOUTS</a> 28<li><a name="TOC13" href="#SEC13">NON-PRINTING CHARACTERS</a> 29<li><a name="TOC14" href="#SEC14">SAVING AND RELOADING COMPILED PATTERNS</a> 30<li><a name="TOC15" href="#SEC15">SEE ALSO</a> 31<li><a name="TOC16" href="#SEC16">AUTHOR</a> 32<li><a name="TOC17" href="#SEC17">REVISION</a> 33</ul> 34<br><a name="SEC1" href="#TOC1">SYNOPSIS</a><br> 35<P> 36<b>pcretest [options] [input file [output file]]</b> 37<br> 38<br> 39<b>pcretest</b> was written as a test program for the PCRE regular expression 40library itself, but it can also be used for experimenting with regular 41expressions. This document describes the features of the test program; for 42details of the regular expressions themselves, see the 43<a href="pcrepattern.html"><b>pcrepattern</b></a> 44documentation. For details of the PCRE library function calls and their 45options, see the 46<a href="pcreapi.html"><b>pcreapi</b></a> 47, 48<a href="pcre16.html"><b>pcre16</b></a> 49and 50<a href="pcre32.html"><b>pcre32</b></a> 51documentation. 52</P> 53<P> 54The input for <b>pcretest</b> is a sequence of regular expression patterns and 55strings to be matched, as described below. The output shows the result of each 56match. Options on the command line and the patterns control PCRE options and 57exactly what is output. 58</P> 59<P> 60As PCRE has evolved, it has acquired many different features, and as a result, 61<b>pcretest</b> now has rather a lot of obscure options for testing every 62possible feature. Some of these options are specifically designed for use in 63conjunction with the test script and data files that are distributed as part of 64PCRE, and are unlikely to be of use otherwise. They are all documented here, 65but without much justification. 66</P> 67<br><a name="SEC2" href="#TOC1">INPUT DATA FORMAT</a><br> 68<P> 69Input to <b>pcretest</b> is processed line by line, either by calling the C 70library's <b>fgets()</b> function, or via the <b>libreadline</b> library (see 71below). In Unix-like environments, <b>fgets()</b> treats any bytes other than 72newline as data characters. However, in some Windows environments character 26 73(hex 1A) causes an immediate end of file, and no further data is read. For 74maximum portability, therefore, it is safest to use only ASCII characters in 75<b>pcretest</b> input files. 76</P> 77<br><a name="SEC3" href="#TOC1">PCRE's 8-BIT, 16-BIT AND 32-BIT LIBRARIES</a><br> 78<P> 79From release 8.30, two separate PCRE libraries can be built. The original one 80supports 8-bit character strings, whereas the newer 16-bit library supports 81character strings encoded in 16-bit units. From release 8.32, a third library 82can be built, supporting character strings encoded in 32-bit units. The 83<b>pcretest</b> program can be used to test all three libraries. However, it is 84itself still an 8-bit program, reading 8-bit input and writing 8-bit output. 85When testing the 16-bit or 32-bit library, the patterns and data strings are 86converted to 16- or 32-bit format before being passed to the PCRE library 87functions. Results are converted to 8-bit for output. 88</P> 89<P> 90References to functions and structures of the form <b>pcre[16|32]_xx</b> below 91mean "<b>pcre_xx</b> when using the 8-bit library, <b>pcre16_xx</b> when using 92the 16-bit library, or <b>pcre32_xx</b> when using the 32-bit library". 93</P> 94<br><a name="SEC4" href="#TOC1">COMMAND LINE OPTIONS</a><br> 95<P> 96<b>-8</b> 97If both the 8-bit library has been built, this option causes the 8-bit library 98to be used (which is the default); if the 8-bit library has not been built, 99this option causes an error. 100</P> 101<P> 102<b>-16</b> 103If both the 8-bit or the 32-bit, and the 16-bit libraries have been built, this 104option causes the 16-bit library to be used. If only the 16-bit library has been 105built, this is the default (so has no effect). If only the 8-bit or the 32-bit 106library has been built, this option causes an error. 107</P> 108<P> 109<b>-32</b> 110If both the 8-bit or the 16-bit, and the 32-bit libraries have been built, this 111option causes the 32-bit library to be used. If only the 32-bit library has been 112built, this is the default (so has no effect). If only the 8-bit or the 16-bit 113library has been built, this option causes an error. 114</P> 115<P> 116<b>-b</b> 117Behave as if each pattern has the <b>/B</b> (show byte code) modifier; the 118internal form is output after compilation. 119</P> 120<P> 121<b>-C</b> 122Output the version number of the PCRE library, and all available information 123about the optional features that are included, and then exit with zero exit 124code. All other options are ignored. 125</P> 126<P> 127<b>-C</b> <i>option</i> 128Output information about a specific build-time option, then exit. This 129functionality is intended for use in scripts such as <b>RunTest</b>. The 130following options output the value and set the exit code as indicated: 131<pre> 132 ebcdic-nl the code for LF (= NL) in an EBCDIC environment: 133 0x15 or 0x25 134 0 if used in an ASCII environment 135 exit code is always 0 136 linksize the configured internal link size (2, 3, or 4) 137 exit code is set to the link size 138 newline the default newline setting: 139 CR, LF, CRLF, ANYCRLF, or ANY 140 exit code is always 0 141 bsr the default setting for what \R matches: 142 ANYCRLF or ANY 143 exit code is always 0 144</pre> 145The following options output 1 for true or 0 for false, and set the exit code 146to the same value: 147<pre> 148 ebcdic compiled for an EBCDIC environment 149 jit just-in-time support is available 150 pcre16 the 16-bit library was built 151 pcre32 the 32-bit library was built 152 pcre8 the 8-bit library was built 153 ucp Unicode property support is available 154 utf UTF-8 and/or UTF-16 and/or UTF-32 support 155 is available 156</pre> 157If an unknown option is given, an error message is output; the exit code is 0. 158</P> 159<P> 160<b>-d</b> 161Behave as if each pattern has the <b>/D</b> (debug) modifier; the internal 162form and information about the compiled pattern is output after compilation; 163<b>-d</b> is equivalent to <b>-b -i</b>. 164</P> 165<P> 166<b>-dfa</b> 167Behave as if each data line contains the \D escape sequence; this causes the 168alternative matching function, <b>pcre[16|32]_dfa_exec()</b>, to be used instead 169of the standard <b>pcre[16|32]_exec()</b> function (more detail is given below). 170</P> 171<P> 172<b>-help</b> 173Output a brief summary these options and then exit. 174</P> 175<P> 176<b>-i</b> 177Behave as if each pattern has the <b>/I</b> modifier; information about the 178compiled pattern is given after compilation. 179</P> 180<P> 181<b>-M</b> 182Behave as if each data line contains the \M escape sequence; this causes 183PCRE to discover the minimum MATCH_LIMIT and MATCH_LIMIT_RECURSION settings by 184calling <b>pcre[16|32]_exec()</b> repeatedly with different limits. 185</P> 186<P> 187<b>-m</b> 188Output the size of each compiled pattern after it has been compiled. This is 189equivalent to adding <b>/M</b> to each regular expression. The size is given in 190bytes for both libraries. 191</P> 192<P> 193<b>-O</b> 194Behave as if each pattern has the <b>/O</b> modifier, that is disable 195auto-possessification for all patterns. 196</P> 197<P> 198<b>-o</b> <i>osize</i> 199Set the number of elements in the output vector that is used when calling 200<b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> to be <i>osize</i>. The 201default value is 45, which is enough for 14 capturing subexpressions for 202<b>pcre[16|32]_exec()</b> or 22 different matches for 203<b>pcre[16|32]_dfa_exec()</b>. 204The vector size can be changed for individual matching calls by including \O 205in the data line (see below). 206</P> 207<P> 208<b>-p</b> 209Behave as if each pattern has the <b>/P</b> modifier; the POSIX wrapper API is 210used to call PCRE. None of the other options has any effect when <b>-p</b> is 211set. This option can be used only with the 8-bit library. 212</P> 213<P> 214<b>-q</b> 215Do not output the version number of <b>pcretest</b> at the start of execution. 216</P> 217<P> 218<b>-S</b> <i>size</i> 219On Unix-like systems, set the size of the run-time stack to <i>size</i> 220megabytes. 221</P> 222<P> 223<b>-s</b> or <b>-s+</b> 224Behave as if each pattern has the <b>/S</b> modifier; in other words, force each 225pattern to be studied. If <b>-s+</b> is used, all the JIT compile options are 226passed to <b>pcre[16|32]_study()</b>, causing just-in-time optimization to be set 227up if it is available, for both full and partial matching. Specific JIT compile 228options can be selected by following <b>-s+</b> with a digit in the range 1 to 2297, which selects the JIT compile modes as follows: 230<pre> 231 1 normal match only 232 2 soft partial match only 233 3 normal match and soft partial match 234 4 hard partial match only 235 6 soft and hard partial match 236 7 all three modes (default) 237</pre> 238If <b>-s++</b> is used instead of <b>-s+</b> (with or without a following digit), 239the text "(JIT)" is added to the first output line after a match or no match 240when JIT-compiled code was actually used. 241<br> 242<br> 243Note that there are pattern options that can override <b>-s</b>, either 244specifying no studying at all, or suppressing JIT compilation. 245<br> 246<br> 247If the <b>/I</b> or <b>/D</b> option is present on a pattern (requesting output 248about the compiled pattern), information about the result of studying is not 249included when studying is caused only by <b>-s</b> and neither <b>-i</b> nor 250<b>-d</b> is present on the command line. This behaviour means that the output 251from tests that are run with and without <b>-s</b> should be identical, except 252when options that output information about the actual running of a match are 253set. 254<br> 255<br> 256The <b>-M</b>, <b>-t</b>, and <b>-tm</b> options, which give information about 257resources used, are likely to produce different output with and without 258<b>-s</b>. Output may also differ if the <b>/C</b> option is present on an 259individual pattern. This uses callouts to trace the the matching process, and 260this may be different between studied and non-studied patterns. If the pattern 261contains (*MARK) items there may also be differences, for the same reason. The 262<b>-s</b> command line option can be overridden for specific patterns that 263should never be studied (see the <b>/S</b> pattern modifier below). 264</P> 265<P> 266<b>-t</b> 267Run each compile, study, and match many times with a timer, and output the 268resulting times per compile, study, or match (in milliseconds). Do not set 269<b>-m</b> with <b>-t</b>, because you will then get the size output a zillion 270times, and the timing will be distorted. You can control the number of 271iterations that are used for timing by following <b>-t</b> with a number (as a 272separate item on the command line). For example, "-t 1000" iterates 1000 times. 273The default is to iterate 500000 times. 274</P> 275<P> 276<b>-tm</b> 277This is like <b>-t</b> except that it times only the matching phase, not the 278compile or study phases. 279</P> 280<P> 281<b>-T</b> <b>-TM</b> 282These behave like <b>-t</b> and <b>-tm</b>, but in addition, at the end of a run, 283the total times for all compiles, studies, and matches are output. 284</P> 285<br><a name="SEC5" href="#TOC1">DESCRIPTION</a><br> 286<P> 287If <b>pcretest</b> is given two filename arguments, it reads from the first and 288writes to the second. If it is given only one filename argument, it reads from 289that file and writes to stdout. Otherwise, it reads from stdin and writes to 290stdout, and prompts for each line of input, using "re>" to prompt for regular 291expressions, and "data>" to prompt for data lines. 292</P> 293<P> 294When <b>pcretest</b> is built, a configuration option can specify that it should 295be linked with the <b>libreadline</b> library. When this is done, if the input 296is from a terminal, it is read using the <b>readline()</b> function. This 297provides line-editing and history facilities. The output from the <b>-help</b> 298option states whether or not <b>readline()</b> will be used. 299</P> 300<P> 301The program handles any number of sets of input on a single input file. Each 302set starts with a regular expression, and continues with any number of data 303lines to be matched against that pattern. 304</P> 305<P> 306Each data line is matched separately and independently. If you want to do 307multi-line matches, you have to use the \n escape sequence (or \r or \r\n, 308etc., depending on the newline setting) in a single line of input to encode the 309newline sequences. There is no limit on the length of data lines; the input 310buffer is automatically extended if it is too small. 311</P> 312<P> 313An empty line signals the end of the data lines, at which point a new regular 314expression is read. The regular expressions are given enclosed in any 315non-alphanumeric delimiters other than backslash, for example: 316<pre> 317 /(a|bc)x+yz/ 318</pre> 319White space before the initial delimiter is ignored. A regular expression may 320be continued over several input lines, in which case the newline characters are 321included within it. It is possible to include the delimiter within the pattern 322by escaping it, for example 323<pre> 324 /abc\/def/ 325</pre> 326If you do so, the escape and the delimiter form part of the pattern, but since 327delimiters are always non-alphanumeric, this does not affect its interpretation. 328If the terminating delimiter is immediately followed by a backslash, for 329example, 330<pre> 331 /abc/\ 332</pre> 333then a backslash is added to the end of the pattern. This is done to provide a 334way of testing the error condition that arises if a pattern finishes with a 335backslash, because 336<pre> 337 /abc\/ 338</pre> 339is interpreted as the first line of a pattern that starts with "abc/", causing 340pcretest to read the next line as a continuation of the regular expression. 341</P> 342<br><a name="SEC6" href="#TOC1">PATTERN MODIFIERS</a><br> 343<P> 344A pattern may be followed by any number of modifiers, which are mostly single 345characters, though some of these can be qualified by further characters. 346Following Perl usage, these are referred to below as, for example, "the 347<b>/i</b> modifier", even though the delimiter of the pattern need not always be 348a slash, and no slash is used when writing modifiers. White space may appear 349between the final pattern delimiter and the first modifier, and between the 350modifiers themselves. For reference, here is a complete list of modifiers. They 351fall into several groups that are described in detail in the following 352sections. 353<pre> 354 <b>/8</b> set UTF mode 355 <b>/9</b> set PCRE_NEVER_UTF (locks out UTF mode) 356 <b>/?</b> disable UTF validity check 357 <b>/+</b> show remainder of subject after match 358 <b>/=</b> show all captures (not just those that are set) 359 360 <b>/A</b> set PCRE_ANCHORED 361 <b>/B</b> show compiled code 362 <b>/C</b> set PCRE_AUTO_CALLOUT 363 <b>/D</b> same as <b>/B</b> plus <b>/I</b> 364 <b>/E</b> set PCRE_DOLLAR_ENDONLY 365 <b>/F</b> flip byte order in compiled pattern 366 <b>/f</b> set PCRE_FIRSTLINE 367 <b>/G</b> find all matches (shorten string) 368 <b>/g</b> find all matches (use startoffset) 369 <b>/I</b> show information about pattern 370 <b>/i</b> set PCRE_CASELESS 371 <b>/J</b> set PCRE_DUPNAMES 372 <b>/K</b> show backtracking control names 373 <b>/L</b> set locale 374 <b>/M</b> show compiled memory size 375 <b>/m</b> set PCRE_MULTILINE 376 <b>/N</b> set PCRE_NO_AUTO_CAPTURE 377 <b>/O</b> set PCRE_NO_AUTO_POSSESS 378 <b>/P</b> use the POSIX wrapper 379 <b>/Q</b> test external stack check function 380 <b>/S</b> study the pattern after compilation 381 <b>/s</b> set PCRE_DOTALL 382 <b>/T</b> select character tables 383 <b>/U</b> set PCRE_UNGREEDY 384 <b>/W</b> set PCRE_UCP 385 <b>/X</b> set PCRE_EXTRA 386 <b>/x</b> set PCRE_EXTENDED 387 <b>/Y</b> set PCRE_NO_START_OPTIMIZE 388 <b>/Z</b> don't show lengths in <b>/B</b> output 389 390 <b>/<any></b> set PCRE_NEWLINE_ANY 391 <b>/<anycrlf></b> set PCRE_NEWLINE_ANYCRLF 392 <b>/<cr></b> set PCRE_NEWLINE_CR 393 <b>/<crlf></b> set PCRE_NEWLINE_CRLF 394 <b>/<lf></b> set PCRE_NEWLINE_LF 395 <b>/<bsr_anycrlf></b> set PCRE_BSR_ANYCRLF 396 <b>/<bsr_unicode></b> set PCRE_BSR_UNICODE 397 <b>/<JS></b> set PCRE_JAVASCRIPT_COMPAT 398 399</PRE> 400</P> 401<br><b> 402Perl-compatible modifiers 403</b><br> 404<P> 405The <b>/i</b>, <b>/m</b>, <b>/s</b>, and <b>/x</b> modifiers set the PCRE_CASELESS, 406PCRE_MULTILINE, PCRE_DOTALL, or PCRE_EXTENDED options, respectively, when 407<b>pcre[16|32]_compile()</b> is called. These four modifier letters have the same 408effect as they do in Perl. For example: 409<pre> 410 /caseless/i 411 412</PRE> 413</P> 414<br><b> 415Modifiers for other PCRE options 416</b><br> 417<P> 418The following table shows additional modifiers for setting PCRE compile-time 419options that do not correspond to anything in Perl: 420<pre> 421 <b>/8</b> PCRE_UTF8 ) when using the 8-bit 422 <b>/?</b> PCRE_NO_UTF8_CHECK ) library 423 424 <b>/8</b> PCRE_UTF16 ) when using the 16-bit 425 <b>/?</b> PCRE_NO_UTF16_CHECK ) library 426 427 <b>/8</b> PCRE_UTF32 ) when using the 32-bit 428 <b>/?</b> PCRE_NO_UTF32_CHECK ) library 429 430 <b>/9</b> PCRE_NEVER_UTF 431 <b>/A</b> PCRE_ANCHORED 432 <b>/C</b> PCRE_AUTO_CALLOUT 433 <b>/E</b> PCRE_DOLLAR_ENDONLY 434 <b>/f</b> PCRE_FIRSTLINE 435 <b>/J</b> PCRE_DUPNAMES 436 <b>/N</b> PCRE_NO_AUTO_CAPTURE 437 <b>/O</b> PCRE_NO_AUTO_POSSESS 438 <b>/U</b> PCRE_UNGREEDY 439 <b>/W</b> PCRE_UCP 440 <b>/X</b> PCRE_EXTRA 441 <b>/Y</b> PCRE_NO_START_OPTIMIZE 442 <b>/<any></b> PCRE_NEWLINE_ANY 443 <b>/<anycrlf></b> PCRE_NEWLINE_ANYCRLF 444 <b>/<cr></b> PCRE_NEWLINE_CR 445 <b>/<crlf></b> PCRE_NEWLINE_CRLF 446 <b>/<lf></b> PCRE_NEWLINE_LF 447 <b>/<bsr_anycrlf></b> PCRE_BSR_ANYCRLF 448 <b>/<bsr_unicode></b> PCRE_BSR_UNICODE 449 <b>/<JS></b> PCRE_JAVASCRIPT_COMPAT 450</pre> 451The modifiers that are enclosed in angle brackets are literal strings as shown, 452including the angle brackets, but the letters within can be in either case. 453This example sets multiline matching with CRLF as the line ending sequence: 454<pre> 455 /^abc/m<CRLF> 456</pre> 457As well as turning on the PCRE_UTF8/16/32 option, the <b>/8</b> modifier causes 458all non-printing characters in output strings to be printed using the 459\x{hh...} notation. Otherwise, those less than 0x100 are output in hex without 460the curly brackets. 461</P> 462<P> 463Full details of the PCRE options are given in the 464<a href="pcreapi.html"><b>pcreapi</b></a> 465documentation. 466</P> 467<br><b> 468Finding all matches in a string 469</b><br> 470<P> 471Searching for all possible matches within each subject string can be requested 472by the <b>/g</b> or <b>/G</b> modifier. After finding a match, PCRE is called 473again to search the remainder of the subject string. The difference between 474<b>/g</b> and <b>/G</b> is that the former uses the <i>startoffset</i> argument to 475<b>pcre[16|32]_exec()</b> to start searching at a new point within the entire 476string (which is in effect what Perl does), whereas the latter passes over a 477shortened substring. This makes a difference to the matching process if the 478pattern begins with a lookbehind assertion (including \b or \B). 479</P> 480<P> 481If any call to <b>pcre[16|32]_exec()</b> in a <b>/g</b> or <b>/G</b> sequence matches 482an empty string, the next call is done with the PCRE_NOTEMPTY_ATSTART and 483PCRE_ANCHORED flags set in order to search for another, non-empty, match at the 484same point. If this second match fails, the start offset is advanced, and the 485normal match is retried. This imitates the way Perl handles such cases when 486using the <b>/g</b> modifier or the <b>split()</b> function. Normally, the start 487offset is advanced by one character, but if the newline convention recognizes 488CRLF as a newline, and the current character is CR followed by LF, an advance 489of two is used. 490</P> 491<br><b> 492Other modifiers 493</b><br> 494<P> 495There are yet more modifiers for controlling the way <b>pcretest</b> 496operates. 497</P> 498<P> 499The <b>/+</b> modifier requests that as well as outputting the substring that 500matched the entire pattern, <b>pcretest</b> should in addition output the 501remainder of the subject string. This is useful for tests where the subject 502contains multiple copies of the same substring. If the <b>+</b> modifier appears 503twice, the same action is taken for captured substrings. In each case the 504remainder is output on the following line with a plus character following the 505capture number. Note that this modifier must not immediately follow the /S 506modifier because /S+ and /S++ have other meanings. 507</P> 508<P> 509The <b>/=</b> modifier requests that the values of all potential captured 510parentheses be output after a match. By default, only those up to the highest 511one actually used in the match are output (corresponding to the return code 512from <b>pcre[16|32]_exec()</b>). Values in the offsets vector corresponding to 513higher numbers should be set to -1, and these are output as "<unset>". This 514modifier gives a way of checking that this is happening. 515</P> 516<P> 517The <b>/B</b> modifier is a debugging feature. It requests that <b>pcretest</b> 518output a representation of the compiled code after compilation. Normally this 519information contains length and offset values; however, if <b>/Z</b> is also 520present, this data is replaced by spaces. This is a special feature for use in 521the automatic test scripts; it ensures that the same output is generated for 522different internal link sizes. 523</P> 524<P> 525The <b>/D</b> modifier is a PCRE debugging feature, and is equivalent to 526<b>/BI</b>, that is, both the <b>/B</b> and the <b>/I</b> modifiers. 527</P> 528<P> 529The <b>/F</b> modifier causes <b>pcretest</b> to flip the byte order of the 5302-byte and 4-byte fields in the compiled pattern. This facility is for testing 531the feature in PCRE that allows it to execute patterns that were compiled on a 532host with a different endianness. This feature is not available when the POSIX 533interface to PCRE is being used, that is, when the <b>/P</b> pattern modifier is 534specified. See also the section about saving and reloading compiled patterns 535below. 536</P> 537<P> 538The <b>/I</b> modifier requests that <b>pcretest</b> output information about the 539compiled pattern (whether it is anchored, has a fixed first character, and 540so on). It does this by calling <b>pcre[16|32]_fullinfo()</b> after compiling a 541pattern. If the pattern is studied, the results of that are also output. In 542this output, the word "char" means a non-UTF character, that is, the value of a 543single data item (8-bit, 16-bit, or 32-bit, depending on the library that is 544being tested). 545</P> 546<P> 547The <b>/K</b> modifier requests <b>pcretest</b> to show names from backtracking 548control verbs that are returned from calls to <b>pcre[16|32]_exec()</b>. It causes 549<b>pcretest</b> to create a <b>pcre[16|32]_extra</b> block if one has not already 550been created by a call to <b>pcre[16|32]_study()</b>, and to set the 551PCRE_EXTRA_MARK flag and the <b>mark</b> field within it, every time that 552<b>pcre[16|32]_exec()</b> is called. If the variable that the <b>mark</b> field 553points to is non-NULL for a match, non-match, or partial match, <b>pcretest</b> 554prints the string to which it points. For a match, this is shown on a line by 555itself, tagged with "MK:". For a non-match it is added to the message. 556</P> 557<P> 558The <b>/L</b> modifier must be followed directly by the name of a locale, for 559example, 560<pre> 561 /pattern/Lfr_FR 562</pre> 563For this reason, it must be the last modifier. The given locale is set, 564<b>pcre[16|32]_maketables()</b> is called to build a set of character tables for 565the locale, and this is then passed to <b>pcre[16|32]_compile()</b> when compiling 566the regular expression. Without an <b>/L</b> (or <b>/T</b>) modifier, NULL is 567passed as the tables pointer; that is, <b>/L</b> applies only to the expression 568on which it appears. 569</P> 570<P> 571The <b>/M</b> modifier causes the size in bytes of the memory block used to hold 572the compiled pattern to be output. This does not include the size of the 573<b>pcre[16|32]</b> block; it is just the actual compiled data. If the pattern is 574successfully studied with the PCRE_STUDY_JIT_COMPILE option, the size of the 575JIT compiled code is also output. 576</P> 577<P> 578The <b>/Q</b> modifier is used to test the use of <b>pcre_stack_guard</b>. It 579must be followed by '0' or '1', specifying the return code to be given from an 580external function that is passed to PCRE and used for stack checking during 581compilation (see the 582<a href="pcreapi.html"><b>pcreapi</b></a> 583documentation for details). 584</P> 585<P> 586The <b>/S</b> modifier causes <b>pcre[16|32]_study()</b> to be called after the 587expression has been compiled, and the results used when the expression is 588matched. There are a number of qualifying characters that may follow <b>/S</b>. 589They may appear in any order. 590</P> 591<P> 592If <b>/S</b> is followed by an exclamation mark, <b>pcre[16|32]_study()</b> is 593called with the PCRE_STUDY_EXTRA_NEEDED option, causing it always to return a 594<b>pcre_extra</b> block, even when studying discovers no useful information. 595</P> 596<P> 597If <b>/S</b> is followed by a second S character, it suppresses studying, even 598if it was requested externally by the <b>-s</b> command line option. This makes 599it possible to specify that certain patterns are always studied, and others are 600never studied, independently of <b>-s</b>. This feature is used in the test 601files in a few cases where the output is different when the pattern is studied. 602</P> 603<P> 604If the <b>/S</b> modifier is followed by a + character, the call to 605<b>pcre[16|32]_study()</b> is made with all the JIT study options, requesting 606just-in-time optimization support if it is available, for both normal and 607partial matching. If you want to restrict the JIT compiling modes, you can 608follow <b>/S+</b> with a digit in the range 1 to 7: 609<pre> 610 1 normal match only 611 2 soft partial match only 612 3 normal match and soft partial match 613 4 hard partial match only 614 6 soft and hard partial match 615 7 all three modes (default) 616</pre> 617If <b>/S++</b> is used instead of <b>/S+</b> (with or without a following digit), 618the text "(JIT)" is added to the first output line after a match or no match 619when JIT-compiled code was actually used. 620</P> 621<P> 622Note that there is also an independent <b>/+</b> modifier; it must not be given 623immediately after <b>/S</b> or <b>/S+</b> because this will be misinterpreted. 624</P> 625<P> 626If JIT studying is successful, the compiled JIT code will automatically be used 627when <b>pcre[16|32]_exec()</b> is run, except when incompatible run-time options 628are specified. For more details, see the 629<a href="pcrejit.html"><b>pcrejit</b></a> 630documentation. See also the <b>\J</b> escape sequence below for a way of 631setting the size of the JIT stack. 632</P> 633<P> 634Finally, if <b>/S</b> is followed by a minus character, JIT compilation is 635suppressed, even if it was requested externally by the <b>-s</b> command line 636option. This makes it possible to specify that JIT is never to be used for 637certain patterns. 638</P> 639<P> 640The <b>/T</b> modifier must be followed by a single digit. It causes a specific 641set of built-in character tables to be passed to <b>pcre[16|32]_compile()</b>. It 642is used in the standard PCRE tests to check behaviour with different character 643tables. The digit specifies the tables as follows: 644<pre> 645 0 the default ASCII tables, as distributed in 646 pcre_chartables.c.dist 647 1 a set of tables defining ISO 8859 characters 648</pre> 649In table 1, some characters whose codes are greater than 128 are identified as 650letters, digits, spaces, etc. 651</P> 652<br><b> 653Using the POSIX wrapper API 654</b><br> 655<P> 656The <b>/P</b> modifier causes <b>pcretest</b> to call PCRE via the POSIX wrapper 657API rather than its native API. This supports only the 8-bit library. When 658<b>/P</b> is set, the following modifiers set options for the <b>regcomp()</b> 659function: 660<pre> 661 /i REG_ICASE 662 /m REG_NEWLINE 663 /N REG_NOSUB 664 /s REG_DOTALL ) 665 /U REG_UNGREEDY ) These options are not part of 666 /W REG_UCP ) the POSIX standard 667 /8 REG_UTF8 ) 668</pre> 669The <b>/+</b> modifier works as described above. All other modifiers are 670ignored. 671</P> 672<br><b> 673Locking out certain modifiers 674</b><br> 675<P> 676PCRE can be compiled with or without support for certain features such as 677UTF-8/16/32 or Unicode properties. Accordingly, the standard tests are split up 678into a number of different files that are selected for running depending on 679which features are available. When updating the tests, it is all too easy to 680put a new test into the wrong file by mistake; for example, to put a test that 681requires UTF support into a file that is used when it is not available. To help 682detect such mistakes as early as possible, there is a facility for locking out 683specific modifiers. If an input line for <b>pcretest</b> starts with the string 684"< forbid " the following sequence of characters is taken as a list of 685forbidden modifiers. For example, in the test files that must not use UTF or 686Unicode property support, this line appears: 687<pre> 688 < forbid 8W 689</pre> 690This locks out the /8 and /W modifiers. An immediate error is given if they are 691subsequently encountered. If the character string contains < but not >, all the 692multi-character modifiers that begin with < are locked out. Otherwise, such 693modifiers must be explicitly listed, for example: 694<pre> 695 < forbid <JS><cr> 696</pre> 697There must be a single space between < and "forbid" for this feature to be 698recognised. If there is not, the line is interpreted either as a request to 699re-load a pre-compiled pattern (see "SAVING AND RELOADING COMPILED PATTERNS" 700below) or, if there is a another < character, as a pattern that uses < as its 701delimiter. 702</P> 703<br><a name="SEC7" href="#TOC1">DATA LINES</a><br> 704<P> 705Before each data line is passed to <b>pcre[16|32]_exec()</b>, leading and trailing 706white space is removed, and it is then scanned for \ escapes. Some of these 707are pretty esoteric features, intended for checking out some of the more 708complicated features of PCRE. If you are just testing "ordinary" regular 709expressions, you probably don't need any of these. The following escapes are 710recognized: 711<pre> 712 \a alarm (BEL, \x07) 713 \b backspace (\x08) 714 \e escape (\x27) 715 \f form feed (\x0c) 716 \n newline (\x0a) 717 \qdd set the PCRE_MATCH_LIMIT limit to dd (any number of digits) 718 \r carriage return (\x0d) 719 \t tab (\x09) 720 \v vertical tab (\x0b) 721 \nnn octal character (up to 3 octal digits); always 722 a byte unless > 255 in UTF-8 or 16-bit or 32-bit mode 723 \o{dd...} octal character (any number of octal digits} 724 \xhh hexadecimal byte (up to 2 hex digits) 725 \x{hh...} hexadecimal character (any number of hex digits) 726 \A pass the PCRE_ANCHORED option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 727 \B pass the PCRE_NOTBOL option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 728 \Cdd call pcre[16|32]_copy_substring() for substring dd after a successful match (number less than 32) 729 \Cname call pcre[16|32]_copy_named_substring() for substring "name" after a successful match (name termin- 730 ated by next non alphanumeric character) 731 \C+ show the current captured substrings at callout time 732 \C- do not supply a callout function 733 \C!n return 1 instead of 0 when callout number n is reached 734 \C!n!m return 1 instead of 0 when callout number n is reached for the nth time 735 \C*n pass the number n (may be negative) as callout data; this is used as the callout return value 736 \D use the <b>pcre[16|32]_dfa_exec()</b> match function 737 \F only shortest match for <b>pcre[16|32]_dfa_exec()</b> 738 \Gdd call pcre[16|32]_get_substring() for substring dd after a successful match (number less than 32) 739 \Gname call pcre[16|32]_get_named_substring() for substring "name" after a successful match (name termin- 740 ated by next non-alphanumeric character) 741 \Jdd set up a JIT stack of dd kilobytes maximum (any number of digits) 742 \L call pcre[16|32]_get_substringlist() after a successful match 743 \M discover the minimum MATCH_LIMIT and MATCH_LIMIT_RECURSION settings 744 \N pass the PCRE_NOTEMPTY option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b>; if used twice, pass the 745 PCRE_NOTEMPTY_ATSTART option 746 \Odd set the size of the output vector passed to <b>pcre[16|32]_exec()</b> to dd (any number of digits) 747 \P pass the PCRE_PARTIAL_SOFT option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b>; if used twice, pass the 748 PCRE_PARTIAL_HARD option 749 \Qdd set the PCRE_MATCH_LIMIT_RECURSION limit to dd (any number of digits) 750 \R pass the PCRE_DFA_RESTART option to <b>pcre[16|32]_dfa_exec()</b> 751 \S output details of memory get/free calls during matching 752 \Y pass the PCRE_NO_START_OPTIMIZE option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 753 \Z pass the PCRE_NOTEOL option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 754 \? pass the PCRE_NO_UTF[8|16|32]_CHECK option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 755 \>dd start the match at offset dd (optional "-"; then any number of digits); this sets the <i>startoffset</i> 756 argument for <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 757 \<cr> pass the PCRE_NEWLINE_CR option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 758 \<lf> pass the PCRE_NEWLINE_LF option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 759 \<crlf> pass the PCRE_NEWLINE_CRLF option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 760 \<anycrlf> pass the PCRE_NEWLINE_ANYCRLF option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 761 \<any> pass the PCRE_NEWLINE_ANY option to <b>pcre[16|32]_exec()</b> or <b>pcre[16|32]_dfa_exec()</b> 762</pre> 763The use of \x{hh...} is not dependent on the use of the <b>/8</b> modifier on 764the pattern. It is recognized always. There may be any number of hexadecimal 765digits inside the braces; invalid values provoke error messages. 766</P> 767<P> 768Note that \xhh specifies one byte rather than one character in UTF-8 mode; 769this makes it possible to construct invalid UTF-8 sequences for testing 770purposes. On the other hand, \x{hh} is interpreted as a UTF-8 character in 771UTF-8 mode, generating more than one byte if the value is greater than 127. 772When testing the 8-bit library not in UTF-8 mode, \x{hh} generates one byte 773for values less than 256, and causes an error for greater values. 774</P> 775<P> 776In UTF-16 mode, all 4-digit \x{hhhh} values are accepted. This makes it 777possible to construct invalid UTF-16 sequences for testing purposes. 778</P> 779<P> 780In UTF-32 mode, all 4- to 8-digit \x{...} values are accepted. This makes it 781possible to construct invalid UTF-32 sequences for testing purposes. 782</P> 783<P> 784The escapes that specify line ending sequences are literal strings, exactly as 785shown. No more than one newline setting should be present in any data line. 786</P> 787<P> 788A backslash followed by anything else just escapes the anything else. If 789the very last character is a backslash, it is ignored. This gives a way of 790passing an empty line as data, since a real empty line terminates the data 791input. 792</P> 793<P> 794The <b>\J</b> escape provides a way of setting the maximum stack size that is 795used by the just-in-time optimization code. It is ignored if JIT optimization 796is not being used. Providing a stack that is larger than the default 32K is 797necessary only for very complicated patterns. 798</P> 799<P> 800If \M is present, <b>pcretest</b> calls <b>pcre[16|32]_exec()</b> several times, 801with different values in the <i>match_limit</i> and <i>match_limit_recursion</i> 802fields of the <b>pcre[16|32]_extra</b> data structure, until it finds the minimum 803numbers for each parameter that allow <b>pcre[16|32]_exec()</b> to complete without 804error. Because this is testing a specific feature of the normal interpretive 805<b>pcre[16|32]_exec()</b> execution, the use of any JIT optimization that might 806have been set up by the <b>/S+</b> qualifier of <b>-s+</b> option is disabled. 807</P> 808<P> 809The <i>match_limit</i> number is a measure of the amount of backtracking 810that takes place, and checking it out can be instructive. For most simple 811matches, the number is quite small, but for patterns with very large numbers of 812matching possibilities, it can become large very quickly with increasing length 813of subject string. The <i>match_limit_recursion</i> number is a measure of how 814much stack (or, if PCRE is compiled with NO_RECURSE, how much heap) memory is 815needed to complete the match attempt. 816</P> 817<P> 818When \O is used, the value specified may be higher or lower than the size set 819by the <b>-O</b> command line option (or defaulted to 45); \O applies only to 820the call of <b>pcre[16|32]_exec()</b> for the line in which it appears. 821</P> 822<P> 823If the <b>/P</b> modifier was present on the pattern, causing the POSIX wrapper 824API to be used, the only option-setting sequences that have any effect are \B, 825\N, and \Z, causing REG_NOTBOL, REG_NOTEMPTY, and REG_NOTEOL, respectively, 826to be passed to <b>regexec()</b>. 827</P> 828<br><a name="SEC8" href="#TOC1">THE ALTERNATIVE MATCHING FUNCTION</a><br> 829<P> 830By default, <b>pcretest</b> uses the standard PCRE matching function, 831<b>pcre[16|32]_exec()</b> to match each data line. PCRE also supports an 832alternative matching function, <b>pcre[16|32]_dfa_test()</b>, which operates in a 833different way, and has some restrictions. The differences between the two 834functions are described in the 835<a href="pcrematching.html"><b>pcrematching</b></a> 836documentation. 837</P> 838<P> 839If a data line contains the \D escape sequence, or if the command line 840contains the <b>-dfa</b> option, the alternative matching function is used. 841This function finds all possible matches at a given point. If, however, the \F 842escape sequence is present in the data line, it stops after the first match is 843found. This is always the shortest possible match. 844</P> 845<br><a name="SEC9" href="#TOC1">DEFAULT OUTPUT FROM PCRETEST</a><br> 846<P> 847This section describes the output when the normal matching function, 848<b>pcre[16|32]_exec()</b>, is being used. 849</P> 850<P> 851When a match succeeds, <b>pcretest</b> outputs the list of captured substrings 852that <b>pcre[16|32]_exec()</b> returns, starting with number 0 for the string that 853matched the whole pattern. Otherwise, it outputs "No match" when the return is 854PCRE_ERROR_NOMATCH, and "Partial match:" followed by the partially matching 855substring when <b>pcre[16|32]_exec()</b> returns PCRE_ERROR_PARTIAL. (Note that 856this is the entire substring that was inspected during the partial match; it 857may include characters before the actual match start if a lookbehind assertion, 858\K, \b, or \B was involved.) For any other return, <b>pcretest</b> outputs 859the PCRE negative error number and a short descriptive phrase. If the error is 860a failed UTF string check, the offset of the start of the failing character and 861the reason code are also output, provided that the size of the output vector is 862at least two. Here is an example of an interactive <b>pcretest</b> run. 863<pre> 864 $ pcretest 865 PCRE version 8.13 2011-04-30 866 867 re> /^abc(\d+)/ 868 data> abc123 869 0: abc123 870 1: 123 871 data> xyz 872 No match 873</pre> 874Unset capturing substrings that are not followed by one that is set are not 875returned by <b>pcre[16|32]_exec()</b>, and are not shown by <b>pcretest</b>. In the 876following example, there are two capturing substrings, but when the first data 877line is matched, the second, unset substring is not shown. An "internal" unset 878substring is shown as "<unset>", as for the second data line. 879<pre> 880 re> /(a)|(b)/ 881 data> a 882 0: a 883 1: a 884 data> b 885 0: b 886 1: <unset> 887 2: b 888</pre> 889If the strings contain any non-printing characters, they are output as \xhh 890escapes if the value is less than 256 and UTF mode is not set. Otherwise they 891are output as \x{hh...} escapes. See below for the definition of non-printing 892characters. If the pattern has the <b>/+</b> modifier, the output for substring 8930 is followed by the the rest of the subject string, identified by "0+" like 894this: 895<pre> 896 re> /cat/+ 897 data> cataract 898 0: cat 899 0+ aract 900</pre> 901If the pattern has the <b>/g</b> or <b>/G</b> modifier, the results of successive 902matching attempts are output in sequence, like this: 903<pre> 904 re> /\Bi(\w\w)/g 905 data> Mississippi 906 0: iss 907 1: ss 908 0: iss 909 1: ss 910 0: ipp 911 1: pp 912</pre> 913"No match" is output only if the first match attempt fails. Here is an example 914of a failure message (the offset 4 that is specified by \>4 is past the end of 915the subject string): 916<pre> 917 re> /xyz/ 918 data> xyz\>4 919 Error -24 (bad offset value) 920</PRE> 921</P> 922<P> 923If any of the sequences <b>\C</b>, <b>\G</b>, or <b>\L</b> are present in a 924data line that is successfully matched, the substrings extracted by the 925convenience functions are output with C, G, or L after the string number 926instead of a colon. This is in addition to the normal full list. The string 927length (that is, the return from the extraction function) is given in 928parentheses after each string for <b>\C</b> and <b>\G</b>. 929</P> 930<P> 931Note that whereas patterns can be continued over several lines (a plain ">" 932prompt is used for continuations), data lines may not. However newlines can be 933included in data by means of the \n escape (or \r, \r\n, etc., depending on 934the newline sequence setting). 935</P> 936<br><a name="SEC10" href="#TOC1">OUTPUT FROM THE ALTERNATIVE MATCHING FUNCTION</a><br> 937<P> 938When the alternative matching function, <b>pcre[16|32]_dfa_exec()</b>, is used (by 939means of the \D escape sequence or the <b>-dfa</b> command line option), the 940output consists of a list of all the matches that start at the first point in 941the subject where there is at least one match. For example: 942<pre> 943 re> /(tang|tangerine|tan)/ 944 data> yellow tangerine\D 945 0: tangerine 946 1: tang 947 2: tan 948</pre> 949(Using the normal matching function on this data finds only "tang".) The 950longest matching string is always given first (and numbered zero). After a 951PCRE_ERROR_PARTIAL return, the output is "Partial match:", followed by the 952partially matching substring. (Note that this is the entire substring that was 953inspected during the partial match; it may include characters before the actual 954match start if a lookbehind assertion, \K, \b, or \B was involved.) 955</P> 956<P> 957If <b>/g</b> is present on the pattern, the search for further matches resumes 958at the end of the longest match. For example: 959<pre> 960 re> /(tang|tangerine|tan)/g 961 data> yellow tangerine and tangy sultana\D 962 0: tangerine 963 1: tang 964 2: tan 965 0: tang 966 1: tan 967 0: tan 968</pre> 969Since the matching function does not support substring capture, the escape 970sequences that are concerned with captured substrings are not relevant. 971</P> 972<br><a name="SEC11" href="#TOC1">RESTARTING AFTER A PARTIAL MATCH</a><br> 973<P> 974When the alternative matching function has given the PCRE_ERROR_PARTIAL return, 975indicating that the subject partially matched the pattern, you can restart the 976match with additional subject data by means of the \R escape sequence. For 977example: 978<pre> 979 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/ 980 data> 23ja\P\D 981 Partial match: 23ja 982 data> n05\R\D 983 0: n05 984</pre> 985For further information about partial matching, see the 986<a href="pcrepartial.html"><b>pcrepartial</b></a> 987documentation. 988</P> 989<br><a name="SEC12" href="#TOC1">CALLOUTS</a><br> 990<P> 991If the pattern contains any callout requests, <b>pcretest</b>'s callout function 992is called during matching. This works with both matching functions. By default, 993the called function displays the callout number, the start and current 994positions in the text at the callout time, and the next pattern item to be 995tested. For example: 996<pre> 997 --->pqrabcdef 998 0 ^ ^ \d 999</pre> 1000This output indicates that callout number 0 occurred for a match attempt 1001starting at the fourth character of the subject string, when the pointer was at 1002the seventh character of the data, and when the next pattern item was \d. Just 1003one circumflex is output if the start and current positions are the same. 1004</P> 1005<P> 1006Callouts numbered 255 are assumed to be automatic callouts, inserted as a 1007result of the <b>/C</b> pattern modifier. In this case, instead of showing the 1008callout number, the offset in the pattern, preceded by a plus, is output. For 1009example: 1010<pre> 1011 re> /\d?[A-E]\*/C 1012 data> E* 1013 --->E* 1014 +0 ^ \d? 1015 +3 ^ [A-E] 1016 +8 ^^ \* 1017 +10 ^ ^ 1018 0: E* 1019</pre> 1020If a pattern contains (*MARK) items, an additional line is output whenever 1021a change of latest mark is passed to the callout function. For example: 1022<pre> 1023 re> /a(*MARK:X)bc/C 1024 data> abc 1025 --->abc 1026 +0 ^ a 1027 +1 ^^ (*MARK:X) 1028 +10 ^^ b 1029 Latest Mark: X 1030 +11 ^ ^ c 1031 +12 ^ ^ 1032 0: abc 1033</pre> 1034The mark changes between matching "a" and "b", but stays the same for the rest 1035of the match, so nothing more is output. If, as a result of backtracking, the 1036mark reverts to being unset, the text "<unset>" is output. 1037</P> 1038<P> 1039The callout function in <b>pcretest</b> returns zero (carry on matching) by 1040default, but you can use a \C item in a data line (as described above) to 1041change this and other parameters of the callout. 1042</P> 1043<P> 1044Inserting callouts can be helpful when using <b>pcretest</b> to check 1045complicated regular expressions. For further information about callouts, see 1046the 1047<a href="pcrecallout.html"><b>pcrecallout</b></a> 1048documentation. 1049</P> 1050<br><a name="SEC13" href="#TOC1">NON-PRINTING CHARACTERS</a><br> 1051<P> 1052When <b>pcretest</b> is outputting text in the compiled version of a pattern, 1053bytes other than 32-126 are always treated as non-printing characters are are 1054therefore shown as hex escapes. 1055</P> 1056<P> 1057When <b>pcretest</b> is outputting text that is a matched part of a subject 1058string, it behaves in the same way, unless a different locale has been set for 1059the pattern (using the <b>/L</b> modifier). In this case, the <b>isprint()</b> 1060function to distinguish printing and non-printing characters. 1061</P> 1062<br><a name="SEC14" href="#TOC1">SAVING AND RELOADING COMPILED PATTERNS</a><br> 1063<P> 1064The facilities described in this section are not available when the POSIX 1065interface to PCRE is being used, that is, when the <b>/P</b> pattern modifier is 1066specified. 1067</P> 1068<P> 1069When the POSIX interface is not in use, you can cause <b>pcretest</b> to write a 1070compiled pattern to a file, by following the modifiers with > and a file name. 1071For example: 1072<pre> 1073 /pattern/im >/some/file 1074</pre> 1075See the 1076<a href="pcreprecompile.html"><b>pcreprecompile</b></a> 1077documentation for a discussion about saving and re-using compiled patterns. 1078Note that if the pattern was successfully studied with JIT optimization, the 1079JIT data cannot be saved. 1080</P> 1081<P> 1082The data that is written is binary. The first eight bytes are the length of the 1083compiled pattern data followed by the length of the optional study data, each 1084written as four bytes in big-endian order (most significant byte first). If 1085there is no study data (either the pattern was not studied, or studying did not 1086return any data), the second length is zero. The lengths are followed by an 1087exact copy of the compiled pattern. If there is additional study data, this 1088(excluding any JIT data) follows immediately after the compiled pattern. After 1089writing the file, <b>pcretest</b> expects to read a new pattern. 1090</P> 1091<P> 1092A saved pattern can be reloaded into <b>pcretest</b> by specifying < and a file 1093name instead of a pattern. There must be no space between < and the file name, 1094which must not contain a < character, as otherwise <b>pcretest</b> will 1095interpret the line as a pattern delimited by < characters. For example: 1096<pre> 1097 re> </some/file 1098 Compiled pattern loaded from /some/file 1099 No study data 1100</pre> 1101If the pattern was previously studied with the JIT optimization, the JIT 1102information cannot be saved and restored, and so is lost. When the pattern has 1103been loaded, <b>pcretest</b> proceeds to read data lines in the usual way. 1104</P> 1105<P> 1106You can copy a file written by <b>pcretest</b> to a different host and reload it 1107there, even if the new host has opposite endianness to the one on which the 1108pattern was compiled. For example, you can compile on an i86 machine and run on 1109a SPARC machine. When a pattern is reloaded on a host with different 1110endianness, the confirmation message is changed to: 1111<pre> 1112 Compiled pattern (byte-inverted) loaded from /some/file 1113</pre> 1114The test suite contains some saved pre-compiled patterns with different 1115endianness. These are reloaded using "<!" instead of just "<". This suppresses 1116the "(byte-inverted)" text so that the output is the same on all hosts. It also 1117forces debugging output once the pattern has been reloaded. 1118</P> 1119<P> 1120File names for saving and reloading can be absolute or relative, but note that 1121the shell facility of expanding a file name that starts with a tilde (~) is not 1122available. 1123</P> 1124<P> 1125The ability to save and reload files in <b>pcretest</b> is intended for testing 1126and experimentation. It is not intended for production use because only a 1127single pattern can be written to a file. Furthermore, there is no facility for 1128supplying custom character tables for use with a reloaded pattern. If the 1129original pattern was compiled with custom tables, an attempt to match a subject 1130string using a reloaded pattern is likely to cause <b>pcretest</b> to crash. 1131Finally, if you attempt to load a file that is not in the correct format, the 1132result is undefined. 1133</P> 1134<br><a name="SEC15" href="#TOC1">SEE ALSO</a><br> 1135<P> 1136<b>pcre</b>(3), <b>pcre16</b>(3), <b>pcre32</b>(3), <b>pcreapi</b>(3), 1137<b>pcrecallout</b>(3), 1138<b>pcrejit</b>, <b>pcrematching</b>(3), <b>pcrepartial</b>(d), 1139<b>pcrepattern</b>(3), <b>pcreprecompile</b>(3). 1140</P> 1141<br><a name="SEC16" href="#TOC1">AUTHOR</a><br> 1142<P> 1143Philip Hazel 1144<br> 1145University Computing Service 1146<br> 1147Cambridge CB2 3QH, England. 1148<br> 1149</P> 1150<br><a name="SEC17" href="#TOC1">REVISION</a><br> 1151<P> 1152Last updated: 09 February 2014 1153<br> 1154Copyright © 1997-2014 University of Cambridge. 1155<br> 1156<p> 1157Return to the <a href="index.html">PCRE index page</a>. 1158</p> 1159