• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1<html>
2<head>
3<title>pcre2api specification</title>
4</head>
5<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6<h1>pcre2api man page</h1>
7<p>
8Return to the <a href="index.html">PCRE2 index page</a>.
9</p>
10<p>
11This page is part of the PCRE2 HTML documentation. It was generated
12automatically from the original man page. If there is any nonsense in it,
13please consult the man page, in case the conversion went wrong.
14<br>
15<ul>
16<li><a name="TOC1" href="#SEC1">PCRE2 NATIVE API BASIC FUNCTIONS</a>
17<li><a name="TOC2" href="#SEC2">PCRE2 NATIVE API AUXILIARY MATCH FUNCTIONS</a>
18<li><a name="TOC3" href="#SEC3">PCRE2 NATIVE API GENERAL CONTEXT FUNCTIONS</a>
19<li><a name="TOC4" href="#SEC4">PCRE2 NATIVE API COMPILE CONTEXT FUNCTIONS</a>
20<li><a name="TOC5" href="#SEC5">PCRE2 NATIVE API MATCH CONTEXT FUNCTIONS</a>
21<li><a name="TOC6" href="#SEC6">PCRE2 NATIVE API STRING EXTRACTION FUNCTIONS</a>
22<li><a name="TOC7" href="#SEC7">PCRE2 NATIVE API STRING SUBSTITUTION FUNCTION</a>
23<li><a name="TOC8" href="#SEC8">PCRE2 NATIVE API JIT FUNCTIONS</a>
24<li><a name="TOC9" href="#SEC9">PCRE2 NATIVE API SERIALIZATION FUNCTIONS</a>
25<li><a name="TOC10" href="#SEC10">PCRE2 NATIVE API AUXILIARY FUNCTIONS</a>
26<li><a name="TOC11" href="#SEC11">PCRE2 NATIVE API OBSOLETE FUNCTIONS</a>
27<li><a name="TOC12" href="#SEC12">PCRE2 EXPERIMENTAL PATTERN CONVERSION FUNCTIONS</a>
28<li><a name="TOC13" href="#SEC13">PCRE2 8-BIT, 16-BIT, AND 32-BIT LIBRARIES</a>
29<li><a name="TOC14" href="#SEC14">PCRE2 API OVERVIEW</a>
30<li><a name="TOC15" href="#SEC15">STRING LENGTHS AND OFFSETS</a>
31<li><a name="TOC16" href="#SEC16">NEWLINES</a>
32<li><a name="TOC17" href="#SEC17">MULTITHREADING</a>
33<li><a name="TOC18" href="#SEC18">PCRE2 CONTEXTS</a>
34<li><a name="TOC19" href="#SEC19">CHECKING BUILD-TIME OPTIONS</a>
35<li><a name="TOC20" href="#SEC20">COMPILING A PATTERN</a>
36<li><a name="TOC21" href="#SEC21">JUST-IN-TIME (JIT) COMPILATION</a>
37<li><a name="TOC22" href="#SEC22">LOCALE SUPPORT</a>
38<li><a name="TOC23" href="#SEC23">INFORMATION ABOUT A COMPILED PATTERN</a>
39<li><a name="TOC24" href="#SEC24">INFORMATION ABOUT A PATTERN'S CALLOUTS</a>
40<li><a name="TOC25" href="#SEC25">SERIALIZATION AND PRECOMPILING</a>
41<li><a name="TOC26" href="#SEC26">THE MATCH DATA BLOCK</a>
42<li><a name="TOC27" href="#SEC27">MATCHING A PATTERN: THE TRADITIONAL FUNCTION</a>
43<li><a name="TOC28" href="#SEC28">NEWLINE HANDLING WHEN MATCHING</a>
44<li><a name="TOC29" href="#SEC29">HOW PCRE2_MATCH() RETURNS A STRING AND CAPTURED SUBSTRINGS</a>
45<li><a name="TOC30" href="#SEC30">OTHER INFORMATION ABOUT A MATCH</a>
46<li><a name="TOC31" href="#SEC31">ERROR RETURNS FROM <b>pcre2_match()</b></a>
47<li><a name="TOC32" href="#SEC32">OBTAINING A TEXTUAL ERROR MESSAGE</a>
48<li><a name="TOC33" href="#SEC33">EXTRACTING CAPTURED SUBSTRINGS BY NUMBER</a>
49<li><a name="TOC34" href="#SEC34">EXTRACTING A LIST OF ALL CAPTURED SUBSTRINGS</a>
50<li><a name="TOC35" href="#SEC35">EXTRACTING CAPTURED SUBSTRINGS BY NAME</a>
51<li><a name="TOC36" href="#SEC36">CREATING A NEW STRING WITH SUBSTITUTIONS</a>
52<li><a name="TOC37" href="#SEC37">DUPLICATE CAPTURE GROUP NAMES</a>
53<li><a name="TOC38" href="#SEC38">FINDING ALL POSSIBLE MATCHES AT ONE POSITION</a>
54<li><a name="TOC39" href="#SEC39">MATCHING A PATTERN: THE ALTERNATIVE FUNCTION</a>
55<li><a name="TOC40" href="#SEC40">SEE ALSO</a>
56<li><a name="TOC41" href="#SEC41">AUTHOR</a>
57<li><a name="TOC42" href="#SEC42">REVISION</a>
58</ul>
59<P>
60<b>#include &#60;pcre2.h&#62;</b>
61<br>
62<br>
63PCRE2 is a new API for PCRE, starting at release 10.0. This document contains a
64description of all its native functions. See the
65<a href="pcre2.html"><b>pcre2</b></a>
66document for an overview of all the PCRE2 documentation.
67</P>
68<br><a name="SEC1" href="#TOC1">PCRE2 NATIVE API BASIC FUNCTIONS</a><br>
69<P>
70<b>pcre2_code *pcre2_compile(PCRE2_SPTR <i>pattern</i>, PCRE2_SIZE <i>length</i>,</b>
71<b>  uint32_t <i>options</i>, int *<i>errorcode</i>, PCRE2_SIZE *<i>erroroffset,</i></b>
72<b>  pcre2_compile_context *<i>ccontext</i>);</b>
73<br>
74<br>
75<b>void pcre2_code_free(pcre2_code *<i>code</i>);</b>
76<br>
77<br>
78<b>pcre2_match_data *pcre2_match_data_create(uint32_t <i>ovecsize</i>,</b>
79<b>  pcre2_general_context *<i>gcontext</i>);</b>
80<br>
81<br>
82<b>pcre2_match_data *pcre2_match_data_create_from_pattern(</b>
83<b>  const pcre2_code *<i>code</i>, pcre2_general_context *<i>gcontext</i>);</b>
84<br>
85<br>
86<b>int pcre2_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
87<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
88<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
89<b>  pcre2_match_context *<i>mcontext</i>);</b>
90<br>
91<br>
92<b>int pcre2_dfa_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
93<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
94<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
95<b>  pcre2_match_context *<i>mcontext</i>,</b>
96<b>  int *<i>workspace</i>, PCRE2_SIZE <i>wscount</i>);</b>
97<br>
98<br>
99<b>void pcre2_match_data_free(pcre2_match_data *<i>match_data</i>);</b>
100</P>
101<br><a name="SEC2" href="#TOC1">PCRE2 NATIVE API AUXILIARY MATCH FUNCTIONS</a><br>
102<P>
103<b>PCRE2_SPTR pcre2_get_mark(pcre2_match_data *<i>match_data</i>);</b>
104<br>
105<br>
106<b>uint32_t pcre2_get_ovector_count(pcre2_match_data *<i>match_data</i>);</b>
107<br>
108<br>
109<b>PCRE2_SIZE *pcre2_get_ovector_pointer(pcre2_match_data *<i>match_data</i>);</b>
110<br>
111<br>
112<b>PCRE2_SIZE pcre2_get_startchar(pcre2_match_data *<i>match_data</i>);</b>
113</P>
114<br><a name="SEC3" href="#TOC1">PCRE2 NATIVE API GENERAL CONTEXT FUNCTIONS</a><br>
115<P>
116<b>pcre2_general_context *pcre2_general_context_create(</b>
117<b>  void *(*<i>private_malloc</i>)(PCRE2_SIZE, void *),</b>
118<b>  void (*<i>private_free</i>)(void *, void *), void *<i>memory_data</i>);</b>
119<br>
120<br>
121<b>pcre2_general_context *pcre2_general_context_copy(</b>
122<b>  pcre2_general_context *<i>gcontext</i>);</b>
123<br>
124<br>
125<b>void pcre2_general_context_free(pcre2_general_context *<i>gcontext</i>);</b>
126</P>
127<br><a name="SEC4" href="#TOC1">PCRE2 NATIVE API COMPILE CONTEXT FUNCTIONS</a><br>
128<P>
129<b>pcre2_compile_context *pcre2_compile_context_create(</b>
130<b>  pcre2_general_context *<i>gcontext</i>);</b>
131<br>
132<br>
133<b>pcre2_compile_context *pcre2_compile_context_copy(</b>
134<b>  pcre2_compile_context *<i>ccontext</i>);</b>
135<br>
136<br>
137<b>void pcre2_compile_context_free(pcre2_compile_context *<i>ccontext</i>);</b>
138<br>
139<br>
140<b>int pcre2_set_bsr(pcre2_compile_context *<i>ccontext</i>,</b>
141<b>  uint32_t <i>value</i>);</b>
142<br>
143<br>
144<b>int pcre2_set_character_tables(pcre2_compile_context *<i>ccontext</i>,</b>
145<b>  const uint8_t *<i>tables</i>);</b>
146<br>
147<br>
148<b>int pcre2_set_compile_extra_options(pcre2_compile_context *<i>ccontext</i>,</b>
149<b>  uint32_t <i>extra_options</i>);</b>
150<br>
151<br>
152<b>int pcre2_set_max_pattern_length(pcre2_compile_context *<i>ccontext</i>,</b>
153<b>  PCRE2_SIZE <i>value</i>);</b>
154<br>
155<br>
156<b>int pcre2_set_newline(pcre2_compile_context *<i>ccontext</i>,</b>
157<b>  uint32_t <i>value</i>);</b>
158<br>
159<br>
160<b>int pcre2_set_parens_nest_limit(pcre2_compile_context *<i>ccontext</i>,</b>
161<b>  uint32_t <i>value</i>);</b>
162<br>
163<br>
164<b>int pcre2_set_compile_recursion_guard(pcre2_compile_context *<i>ccontext</i>,</b>
165<b>  int (*<i>guard_function</i>)(uint32_t, void *), void *<i>user_data</i>);</b>
166</P>
167<br><a name="SEC5" href="#TOC1">PCRE2 NATIVE API MATCH CONTEXT FUNCTIONS</a><br>
168<P>
169<b>pcre2_match_context *pcre2_match_context_create(</b>
170<b>  pcre2_general_context *<i>gcontext</i>);</b>
171<br>
172<br>
173<b>pcre2_match_context *pcre2_match_context_copy(</b>
174<b>  pcre2_match_context *<i>mcontext</i>);</b>
175<br>
176<br>
177<b>void pcre2_match_context_free(pcre2_match_context *<i>mcontext</i>);</b>
178<br>
179<br>
180<b>int pcre2_set_callout(pcre2_match_context *<i>mcontext</i>,</b>
181<b>  int (*<i>callout_function</i>)(pcre2_callout_block *, void *),</b>
182<b>  void *<i>callout_data</i>);</b>
183<br>
184<br>
185<b>int pcre2_set_substitute_callout(pcre2_match_context *<i>mcontext</i>,</b>
186<b>  int (*<i>callout_function</i>)(pcre2_substitute_callout_block *, void *),</b>
187<b>  void *<i>callout_data</i>);</b>
188<br>
189<br>
190<b>int pcre2_set_offset_limit(pcre2_match_context *<i>mcontext</i>,</b>
191<b>  PCRE2_SIZE <i>value</i>);</b>
192<br>
193<br>
194<b>int pcre2_set_heap_limit(pcre2_match_context *<i>mcontext</i>,</b>
195<b>  uint32_t <i>value</i>);</b>
196<br>
197<br>
198<b>int pcre2_set_match_limit(pcre2_match_context *<i>mcontext</i>,</b>
199<b>  uint32_t <i>value</i>);</b>
200<br>
201<br>
202<b>int pcre2_set_depth_limit(pcre2_match_context *<i>mcontext</i>,</b>
203<b>  uint32_t <i>value</i>);</b>
204</P>
205<br><a name="SEC6" href="#TOC1">PCRE2 NATIVE API STRING EXTRACTION FUNCTIONS</a><br>
206<P>
207<b>int pcre2_substring_copy_byname(pcre2_match_data *<i>match_data</i>,</b>
208<b>  PCRE2_SPTR <i>name</i>, PCRE2_UCHAR *<i>buffer</i>, PCRE2_SIZE *<i>bufflen</i>);</b>
209<br>
210<br>
211<b>int pcre2_substring_copy_bynumber(pcre2_match_data *<i>match_data</i>,</b>
212<b>  uint32_t <i>number</i>, PCRE2_UCHAR *<i>buffer</i>,</b>
213<b>  PCRE2_SIZE *<i>bufflen</i>);</b>
214<br>
215<br>
216<b>void pcre2_substring_free(PCRE2_UCHAR *<i>buffer</i>);</b>
217<br>
218<br>
219<b>int pcre2_substring_get_byname(pcre2_match_data *<i>match_data</i>,</b>
220<b>  PCRE2_SPTR <i>name</i>, PCRE2_UCHAR **<i>bufferptr</i>, PCRE2_SIZE *<i>bufflen</i>);</b>
221<br>
222<br>
223<b>int pcre2_substring_get_bynumber(pcre2_match_data *<i>match_data</i>,</b>
224<b>  uint32_t <i>number</i>, PCRE2_UCHAR **<i>bufferptr</i>,</b>
225<b>  PCRE2_SIZE *<i>bufflen</i>);</b>
226<br>
227<br>
228<b>int pcre2_substring_length_byname(pcre2_match_data *<i>match_data</i>,</b>
229<b>  PCRE2_SPTR <i>name</i>, PCRE2_SIZE *<i>length</i>);</b>
230<br>
231<br>
232<b>int pcre2_substring_length_bynumber(pcre2_match_data *<i>match_data</i>,</b>
233<b>  uint32_t <i>number</i>, PCRE2_SIZE *<i>length</i>);</b>
234<br>
235<br>
236<b>int pcre2_substring_nametable_scan(const pcre2_code *<i>code</i>,</b>
237<b>  PCRE2_SPTR <i>name</i>, PCRE2_SPTR *<i>first</i>, PCRE2_SPTR *<i>last</i>);</b>
238<br>
239<br>
240<b>int pcre2_substring_number_from_name(const pcre2_code *<i>code</i>,</b>
241<b>  PCRE2_SPTR <i>name</i>);</b>
242<br>
243<br>
244<b>void pcre2_substring_list_free(PCRE2_SPTR *<i>list</i>);</b>
245<br>
246<br>
247<b>int pcre2_substring_list_get(pcre2_match_data *<i>match_data</i>,</b>
248<b>"  PCRE2_UCHAR ***<i>listptr</i>, PCRE2_SIZE **<i>lengthsptr</i>);</b>
249</P>
250<br><a name="SEC7" href="#TOC1">PCRE2 NATIVE API STRING SUBSTITUTION FUNCTION</a><br>
251<P>
252<b>int pcre2_substitute(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
253<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
254<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
255<b>  pcre2_match_context *<i>mcontext</i>, PCRE2_SPTR <i>replacementz</i>,</b>
256<b>  PCRE2_SIZE <i>rlength</i>, PCRE2_UCHAR *<i>outputbuffer</i>,</b>
257<b>  PCRE2_SIZE *<i>outlengthptr</i>);</b>
258</P>
259<br><a name="SEC8" href="#TOC1">PCRE2 NATIVE API JIT FUNCTIONS</a><br>
260<P>
261<b>int pcre2_jit_compile(pcre2_code *<i>code</i>, uint32_t <i>options</i>);</b>
262<br>
263<br>
264<b>int pcre2_jit_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
265<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
266<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
267<b>  pcre2_match_context *<i>mcontext</i>);</b>
268<br>
269<br>
270<b>void pcre2_jit_free_unused_memory(pcre2_general_context *<i>gcontext</i>);</b>
271<br>
272<br>
273<b>pcre2_jit_stack *pcre2_jit_stack_create(PCRE2_SIZE <i>startsize</i>,</b>
274<b>  PCRE2_SIZE <i>maxsize</i>, pcre2_general_context *<i>gcontext</i>);</b>
275<br>
276<br>
277<b>void pcre2_jit_stack_assign(pcre2_match_context *<i>mcontext</i>,</b>
278<b>  pcre2_jit_callback <i>callback_function</i>, void *<i>callback_data</i>);</b>
279<br>
280<br>
281<b>void pcre2_jit_stack_free(pcre2_jit_stack *<i>jit_stack</i>);</b>
282</P>
283<br><a name="SEC9" href="#TOC1">PCRE2 NATIVE API SERIALIZATION FUNCTIONS</a><br>
284<P>
285<b>int32_t pcre2_serialize_decode(pcre2_code **<i>codes</i>,</b>
286<b>  int32_t <i>number_of_codes</i>, const uint8_t *<i>bytes</i>,</b>
287<b>  pcre2_general_context *<i>gcontext</i>);</b>
288<br>
289<br>
290<b>int32_t pcre2_serialize_encode(const pcre2_code **<i>codes</i>,</b>
291<b>  int32_t <i>number_of_codes</i>, uint8_t **<i>serialized_bytes</i>,</b>
292<b>  PCRE2_SIZE *<i>serialized_size</i>, pcre2_general_context *<i>gcontext</i>);</b>
293<br>
294<br>
295<b>void pcre2_serialize_free(uint8_t *<i>bytes</i>);</b>
296<br>
297<br>
298<b>int32_t pcre2_serialize_get_number_of_codes(const uint8_t *<i>bytes</i>);</b>
299</P>
300<br><a name="SEC10" href="#TOC1">PCRE2 NATIVE API AUXILIARY FUNCTIONS</a><br>
301<P>
302<b>pcre2_code *pcre2_code_copy(const pcre2_code *<i>code</i>);</b>
303<br>
304<br>
305<b>pcre2_code *pcre2_code_copy_with_tables(const pcre2_code *<i>code</i>);</b>
306<br>
307<br>
308<b>int pcre2_get_error_message(int <i>errorcode</i>, PCRE2_UCHAR *<i>buffer</i>,</b>
309<b>  PCRE2_SIZE <i>bufflen</i>);</b>
310<br>
311<br>
312<b>const uint8_t *pcre2_maketables(pcre2_general_context *<i>gcontext</i>);</b>
313<br>
314<br>
315<b>void pcre2_maketables_free(pcre2_general_context *<i>gcontext</i>,</b>
316<b>  const uint8_t *<i>tables</i>);</b>
317<br>
318<br>
319<b>int pcre2_pattern_info(const pcre2_code *<i>code</i>, uint32_t <i>what</i>,</b>
320<b>  void *<i>where</i>);</b>
321<br>
322<br>
323<b>int pcre2_callout_enumerate(const pcre2_code *<i>code</i>,</b>
324<b>  int (*<i>callback</i>)(pcre2_callout_enumerate_block *, void *),</b>
325<b>  void *<i>user_data</i>);</b>
326<br>
327<br>
328<b>int pcre2_config(uint32_t <i>what</i>, void *<i>where</i>);</b>
329</P>
330<br><a name="SEC11" href="#TOC1">PCRE2 NATIVE API OBSOLETE FUNCTIONS</a><br>
331<P>
332<b>int pcre2_set_recursion_limit(pcre2_match_context *<i>mcontext</i>,</b>
333<b>  uint32_t <i>value</i>);</b>
334<br>
335<br>
336<b>int pcre2_set_recursion_memory_management(</b>
337<b>  pcre2_match_context *<i>mcontext</i>,</b>
338<b>  void *(*<i>private_malloc</i>)(PCRE2_SIZE, void *),</b>
339<b>  void (*<i>private_free</i>)(void *, void *), void *<i>memory_data</i>);</b>
340<br>
341<br>
342These functions became obsolete at release 10.30 and are retained only for
343backward compatibility. They should not be used in new code. The first is
344replaced by <b>pcre2_set_depth_limit()</b>; the second is no longer needed and
345has no effect (it always returns zero).
346</P>
347<br><a name="SEC12" href="#TOC1">PCRE2 EXPERIMENTAL PATTERN CONVERSION FUNCTIONS</a><br>
348<P>
349<b>pcre2_convert_context *pcre2_convert_context_create(</b>
350<b>  pcre2_general_context *<i>gcontext</i>);</b>
351<br>
352<br>
353<b>pcre2_convert_context *pcre2_convert_context_copy(</b>
354<b>  pcre2_convert_context *<i>cvcontext</i>);</b>
355<br>
356<br>
357<b>void pcre2_convert_context_free(pcre2_convert_context *<i>cvcontext</i>);</b>
358<br>
359<br>
360<b>int pcre2_set_glob_escape(pcre2_convert_context *<i>cvcontext</i>,</b>
361<b>  uint32_t <i>escape_char</i>);</b>
362<br>
363<br>
364<b>int pcre2_set_glob_separator(pcre2_convert_context *<i>cvcontext</i>,</b>
365<b>  uint32_t <i>separator_char</i>);</b>
366<br>
367<br>
368<b>int pcre2_pattern_convert(PCRE2_SPTR <i>pattern</i>, PCRE2_SIZE <i>length</i>,</b>
369<b>  uint32_t <i>options</i>, PCRE2_UCHAR **<i>buffer</i>,</b>
370<b>  PCRE2_SIZE *<i>blength</i>, pcre2_convert_context *<i>cvcontext</i>);</b>
371<br>
372<br>
373<b>void pcre2_converted_pattern_free(PCRE2_UCHAR *<i>converted_pattern</i>);</b>
374<br>
375<br>
376These functions provide a way of converting non-PCRE2 patterns into
377patterns that can be processed by <b>pcre2_compile()</b>. This facility is
378experimental and may be changed in future releases. At present, "globs" and
379POSIX basic and extended patterns can be converted. Details are given in the
380<a href="pcre2convert.html"><b>pcre2convert</b></a>
381documentation.
382</P>
383<br><a name="SEC13" href="#TOC1">PCRE2 8-BIT, 16-BIT, AND 32-BIT LIBRARIES</a><br>
384<P>
385There are three PCRE2 libraries, supporting 8-bit, 16-bit, and 32-bit code
386units, respectively. However, there is just one header file, <b>pcre2.h</b>.
387This contains the function prototypes and other definitions for all three
388libraries. One, two, or all three can be installed simultaneously. On Unix-like
389systems the libraries are called <b>libpcre2-8</b>, <b>libpcre2-16</b>, and
390<b>libpcre2-32</b>, and they can also co-exist with the original PCRE libraries.
391</P>
392<P>
393Character strings are passed to and from a PCRE2 library as a sequence of
394unsigned integers in code units of the appropriate width. Every PCRE2 function
395comes in three different forms, one for each library, for example:
396<pre>
397  <b>pcre2_compile_8()</b>
398  <b>pcre2_compile_16()</b>
399  <b>pcre2_compile_32()</b>
400</pre>
401There are also three different sets of data types:
402<pre>
403  <b>PCRE2_UCHAR8, PCRE2_UCHAR16, PCRE2_UCHAR32</b>
404  <b>PCRE2_SPTR8,  PCRE2_SPTR16,  PCRE2_SPTR32</b>
405</pre>
406The UCHAR types define unsigned code units of the appropriate widths. For
407example, PCRE2_UCHAR16 is usually defined as `uint16_t'. The SPTR types are
408constant pointers to the equivalent UCHAR types, that is, they are pointers to
409vectors of unsigned code units.
410</P>
411<P>
412Many applications use only one code unit width. For their convenience, macros
413are defined whose names are the generic forms such as <b>pcre2_compile()</b> and
414PCRE2_SPTR. These macros use the value of the macro PCRE2_CODE_UNIT_WIDTH to
415generate the appropriate width-specific function and macro names.
416PCRE2_CODE_UNIT_WIDTH is not defined by default. An application must define it
417to be 8, 16, or 32 before including <b>pcre2.h</b> in order to make use of the
418generic names.
419</P>
420<P>
421Applications that use more than one code unit width can be linked with more
422than one PCRE2 library, but must define PCRE2_CODE_UNIT_WIDTH to be 0 before
423including <b>pcre2.h</b>, and then use the real function names. Any code that is
424to be included in an environment where the value of PCRE2_CODE_UNIT_WIDTH is
425unknown should also use the real function names. (Unfortunately, it is not
426possible in C code to save and restore the value of a macro.)
427</P>
428<P>
429If PCRE2_CODE_UNIT_WIDTH is not defined before including <b>pcre2.h</b>, a
430compiler error occurs.
431</P>
432<P>
433When using multiple libraries in an application, you must take care when
434processing any particular pattern to use only functions from a single library.
435For example, if you want to run a match using a pattern that was compiled with
436<b>pcre2_compile_16()</b>, you must do so with <b>pcre2_match_16()</b>, not
437<b>pcre2_match_8()</b> or <b>pcre2_match_32()</b>.
438</P>
439<P>
440In the function summaries above, and in the rest of this document and other
441PCRE2 documents, functions and data types are described using their generic
442names, without the _8, _16, or _32 suffix.
443</P>
444<br><a name="SEC14" href="#TOC1">PCRE2 API OVERVIEW</a><br>
445<P>
446PCRE2 has its own native API, which is described in this document. There are
447also some wrapper functions for the 8-bit library that correspond to the
448POSIX regular expression API, but they do not give access to all the
449functionality of PCRE2. They are described in the
450<a href="pcre2posix.html"><b>pcre2posix</b></a>
451documentation. Both these APIs define a set of C function calls.
452</P>
453<P>
454The native API C data types, function prototypes, option values, and error
455codes are defined in the header file <b>pcre2.h</b>, which also contains
456definitions of PCRE2_MAJOR and PCRE2_MINOR, the major and minor release numbers
457for the library. Applications can use these to include support for different
458releases of PCRE2.
459</P>
460<P>
461In a Windows environment, if you want to statically link an application program
462against a non-dll PCRE2 library, you must define PCRE2_STATIC before including
463<b>pcre2.h</b>.
464</P>
465<P>
466The functions <b>pcre2_compile()</b> and <b>pcre2_match()</b> are used for
467compiling and matching regular expressions in a Perl-compatible manner. A
468sample program that demonstrates the simplest way of using them is provided in
469the file called <i>pcre2demo.c</i> in the PCRE2 source distribution. A listing
470of this program is given in the
471<a href="pcre2demo.html"><b>pcre2demo</b></a>
472documentation, and the
473<a href="pcre2sample.html"><b>pcre2sample</b></a>
474documentation describes how to compile and run it.
475</P>
476<P>
477The compiling and matching functions recognize various options that are passed
478as bits in an options argument. There are also some more complicated parameters
479such as custom memory management functions and resource limits that are passed
480in "contexts" (which are just memory blocks, described below). Simple
481applications do not need to make use of contexts.
482</P>
483<P>
484Just-in-time (JIT) compiler support is an optional feature of PCRE2 that can be
485built in appropriate hardware environments. It greatly speeds up the matching
486performance of many patterns. Programs can request that it be used if
487available by calling <b>pcre2_jit_compile()</b> after a pattern has been
488successfully compiled by <b>pcre2_compile()</b>. This does nothing if JIT
489support is not available.
490</P>
491<P>
492More complicated programs might need to make use of the specialist functions
493<b>pcre2_jit_stack_create()</b>, <b>pcre2_jit_stack_free()</b>, and
494<b>pcre2_jit_stack_assign()</b> in order to control the JIT code's memory usage.
495</P>
496<P>
497JIT matching is automatically used by <b>pcre2_match()</b> if it is available,
498unless the PCRE2_NO_JIT option is set. There is also a direct interface for JIT
499matching, which gives improved performance at the expense of less sanity
500checking. The JIT-specific functions are discussed in the
501<a href="pcre2jit.html"><b>pcre2jit</b></a>
502documentation.
503</P>
504<P>
505A second matching function, <b>pcre2_dfa_match()</b>, which is not
506Perl-compatible, is also provided. This uses a different algorithm for the
507matching. The alternative algorithm finds all possible matches (at a given
508point in the subject), and scans the subject just once (unless there are
509lookaround assertions). However, this algorithm does not return captured
510substrings. A description of the two matching algorithms and their advantages
511and disadvantages is given in the
512<a href="pcre2matching.html"><b>pcre2matching</b></a>
513documentation. There is no JIT support for <b>pcre2_dfa_match()</b>.
514</P>
515<P>
516In addition to the main compiling and matching functions, there are convenience
517functions for extracting captured substrings from a subject string that has
518been matched by <b>pcre2_match()</b>. They are:
519<pre>
520  <b>pcre2_substring_copy_byname()</b>
521  <b>pcre2_substring_copy_bynumber()</b>
522  <b>pcre2_substring_get_byname()</b>
523  <b>pcre2_substring_get_bynumber()</b>
524  <b>pcre2_substring_list_get()</b>
525  <b>pcre2_substring_length_byname()</b>
526  <b>pcre2_substring_length_bynumber()</b>
527  <b>pcre2_substring_nametable_scan()</b>
528  <b>pcre2_substring_number_from_name()</b>
529</pre>
530<b>pcre2_substring_free()</b> and <b>pcre2_substring_list_free()</b> are also
531provided, to free memory used for extracted strings. If either of these
532functions is called with a NULL argument, the function returns immediately
533without doing anything.
534</P>
535<P>
536The function <b>pcre2_substitute()</b> can be called to match a pattern and
537return a copy of the subject string with substitutions for parts that were
538matched.
539</P>
540<P>
541Functions whose names begin with <b>pcre2_serialize_</b> are used for saving
542compiled patterns on disc or elsewhere, and reloading them later.
543</P>
544<P>
545Finally, there are functions for finding out information about a compiled
546pattern (<b>pcre2_pattern_info()</b>) and about the configuration with which
547PCRE2 was built (<b>pcre2_config()</b>).
548</P>
549<P>
550Functions with names ending with <b>_free()</b> are used for freeing memory
551blocks of various sorts. In all cases, if one of these functions is called with
552a NULL argument, it does nothing.
553</P>
554<br><a name="SEC15" href="#TOC1">STRING LENGTHS AND OFFSETS</a><br>
555<P>
556The PCRE2 API uses string lengths and offsets into strings of code units in
557several places. These values are always of type PCRE2_SIZE, which is an
558unsigned integer type, currently always defined as <i>size_t</i>. The largest
559value that can be stored in such a type (that is ~(PCRE2_SIZE)0) is reserved
560as a special indicator for zero-terminated strings and unset offsets.
561Therefore, the longest string that can be handled is one less than this
562maximum.
563<a name="newlines"></a></P>
564<br><a name="SEC16" href="#TOC1">NEWLINES</a><br>
565<P>
566PCRE2 supports five different conventions for indicating line breaks in
567strings: a single CR (carriage return) character, a single LF (linefeed)
568character, the two-character sequence CRLF, any of the three preceding, or any
569Unicode newline sequence. The Unicode newline sequences are the three just
570mentioned, plus the single characters VT (vertical tab, U+000B), FF (form feed,
571U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
572(paragraph separator, U+2029).
573</P>
574<P>
575Each of the first three conventions is used by at least one operating system as
576its standard newline sequence. When PCRE2 is built, a default can be specified.
577If it is not, the default is set to LF, which is the Unix standard. However,
578the newline convention can be changed by an application when calling
579<b>pcre2_compile()</b>, or it can be specified by special text at the start of
580the pattern itself; this overrides any other settings. See the
581<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
582page for details of the special character sequences.
583</P>
584<P>
585In the PCRE2 documentation the word "newline" is used to mean "the character or
586pair of characters that indicate a line break". The choice of newline
587convention affects the handling of the dot, circumflex, and dollar
588metacharacters, the handling of #-comments in /x mode, and, when CRLF is a
589recognized line ending sequence, the match position advancement for a
590non-anchored pattern. There is more detail about this in the
591<a href="#matchoptions">section on <b>pcre2_match()</b> options</a>
592below.
593</P>
594<P>
595The choice of newline convention does not affect the interpretation of
596the \n or \r escape sequences, nor does it affect what \R matches; this has
597its own separate convention.
598</P>
599<br><a name="SEC17" href="#TOC1">MULTITHREADING</a><br>
600<P>
601In a multithreaded application it is important to keep thread-specific data
602separate from data that can be shared between threads. The PCRE2 library code
603itself is thread-safe: it contains no static or global variables. The API is
604designed to be fairly simple for non-threaded applications while at the same
605time ensuring that multithreaded applications can use it.
606</P>
607<P>
608There are several different blocks of data that are used to pass information
609between the application and the PCRE2 libraries.
610</P>
611<br><b>
612The compiled pattern
613</b><br>
614<P>
615A pointer to the compiled form of a pattern is returned to the user when
616<b>pcre2_compile()</b> is successful. The data in the compiled pattern is fixed,
617and does not change when the pattern is matched. Therefore, it is thread-safe,
618that is, the same compiled pattern can be used by more than one thread
619simultaneously. For example, an application can compile all its patterns at the
620start, before forking off multiple threads that use them. However, if the
621just-in-time (JIT) optimization feature is being used, it needs separate memory
622stack areas for each thread. See the
623<a href="pcre2jit.html"><b>pcre2jit</b></a>
624documentation for more details.
625</P>
626<P>
627In a more complicated situation, where patterns are compiled only when they are
628first needed, but are still shared between threads, pointers to compiled
629patterns must be protected from simultaneous writing by multiple threads. This
630is somewhat tricky to do correctly. If you know that writing to a pointer is
631atomic in your environment, you can use logic like this:
632<pre>
633  Get a read-only (shared) lock (mutex) for pointer
634  if (pointer == NULL)
635    {
636    Get a write (unique) lock for pointer
637    if (pointer == NULL) pointer = pcre2_compile(...
638    }
639  Release the lock
640  Use pointer in pcre2_match()
641</pre>
642Of course, testing for compilation errors should also be included in the code.
643</P>
644<P>
645The reason for checking the pointer a second time is as follows: Several
646threads may have acquired the shared lock and tested the pointer for being
647NULL, but only one of them will be given the write lock, with the rest kept
648waiting. The winning thread will compile the pattern and store the result.
649After this thread releases the write lock, another thread will get it, and if
650it does not retest pointer for being NULL, will recompile the pattern and
651overwrite the pointer, creating a memory leak and possibly causing other
652issues.
653</P>
654<P>
655In an environment where writing to a pointer may not be atomic, the above logic
656is not sufficient. The thread that is doing the compiling may be descheduled
657after writing only part of the pointer, which could cause other threads to use
658an invalid value. Instead of checking the pointer itself, a separate "pointer
659is valid" flag (that can be updated atomically) must be used:
660<pre>
661  Get a read-only (shared) lock (mutex) for pointer
662  if (!pointer_is_valid)
663    {
664    Get a write (unique) lock for pointer
665    if (!pointer_is_valid)
666      {
667      pointer = pcre2_compile(...
668      pointer_is_valid = TRUE
669      }
670    }
671  Release the lock
672  Use pointer in pcre2_match()
673</pre>
674If JIT is being used, but the JIT compilation is not being done immediately
675(perhaps waiting to see if the pattern is used often enough), similar logic is
676required. JIT compilation updates a value within the compiled code block, so a
677thread must gain unique write access to the pointer before calling
678<b>pcre2_jit_compile()</b>. Alternatively, <b>pcre2_code_copy()</b> or
679<b>pcre2_code_copy_with_tables()</b> can be used to obtain a private copy of the
680compiled code before calling the JIT compiler.
681</P>
682<br><b>
683Context blocks
684</b><br>
685<P>
686The next main section below introduces the idea of "contexts" in which PCRE2
687functions are called. A context is nothing more than a collection of parameters
688that control the way PCRE2 operates. Grouping a number of parameters together
689in a context is a convenient way of passing them to a PCRE2 function without
690using lots of arguments. The parameters that are stored in contexts are in some
691sense "advanced features" of the API. Many straightforward applications will
692not need to use contexts.
693</P>
694<P>
695In a multithreaded application, if the parameters in a context are values that
696are never changed, the same context can be used by all the threads. However, if
697any thread needs to change any value in a context, it must make its own
698thread-specific copy.
699</P>
700<br><b>
701Match blocks
702</b><br>
703<P>
704The matching functions need a block of memory for storing the results of a
705match. This includes details of what was matched, as well as additional
706information such as the name of a (*MARK) setting. Each thread must provide its
707own copy of this memory.
708</P>
709<br><a name="SEC18" href="#TOC1">PCRE2 CONTEXTS</a><br>
710<P>
711Some PCRE2 functions have a lot of parameters, many of which are used only by
712specialist applications, for example, those that use custom memory management
713or non-standard character tables. To keep function argument lists at a
714reasonable size, and at the same time to keep the API extensible, "uncommon"
715parameters are passed to certain functions in a <b>context</b> instead of
716directly. A context is just a block of memory that holds the parameter values.
717Applications that do not need to adjust any of the context parameters can pass
718NULL when a context pointer is required.
719</P>
720<P>
721There are three different types of context: a general context that is relevant
722for several PCRE2 operations, a compile-time context, and a match-time context.
723</P>
724<br><b>
725The general context
726</b><br>
727<P>
728At present, this context just contains pointers to (and data for) external
729memory management functions that are called from several places in the PCRE2
730library. The context is named `general' rather than specifically `memory'
731because in future other fields may be added. If you do not want to supply your
732own custom memory management functions, you do not need to bother with a
733general context. A general context is created by:
734<br>
735<br>
736<b>pcre2_general_context *pcre2_general_context_create(</b>
737<b>  void *(*<i>private_malloc</i>)(PCRE2_SIZE, void *),</b>
738<b>  void (*<i>private_free</i>)(void *, void *), void *<i>memory_data</i>);</b>
739<br>
740<br>
741The two function pointers specify custom memory management functions, whose
742prototypes are:
743<pre>
744  <b>void *private_malloc(PCRE2_SIZE, void *);</b>
745  <b>void  private_free(void *, void *);</b>
746</pre>
747Whenever code in PCRE2 calls these functions, the final argument is the value
748of <i>memory_data</i>. Either of the first two arguments of the creation
749function may be NULL, in which case the system memory management functions
750<i>malloc()</i> and <i>free()</i> are used. (This is not currently useful, as
751there are no other fields in a general context, but in future there might be.)
752The <i>private_malloc()</i> function is used (if supplied) to obtain memory for
753storing the context, and all three values are saved as part of the context.
754</P>
755<P>
756Whenever PCRE2 creates a data block of any kind, the block contains a pointer
757to the <i>free()</i> function that matches the <i>malloc()</i> function that was
758used. When the time comes to free the block, this function is called.
759</P>
760<P>
761A general context can be copied by calling:
762<br>
763<br>
764<b>pcre2_general_context *pcre2_general_context_copy(</b>
765<b>  pcre2_general_context *<i>gcontext</i>);</b>
766<br>
767<br>
768The memory used for a general context should be freed by calling:
769<br>
770<br>
771<b>void pcre2_general_context_free(pcre2_general_context *<i>gcontext</i>);</b>
772<br>
773<br>
774If this function is passed a NULL argument, it returns immediately without
775doing anything.
776<a name="compilecontext"></a></P>
777<br><b>
778The compile context
779</b><br>
780<P>
781A compile context is required if you want to provide an external function for
782stack checking during compilation or to change the default values of any of the
783following compile-time parameters:
784<pre>
785  What \R matches (Unicode newlines or CR, LF, CRLF only)
786  PCRE2's character tables
787  The newline character sequence
788  The compile time nested parentheses limit
789  The maximum length of the pattern string
790  The extra options bits (none set by default)
791</pre>
792A compile context is also required if you are using custom memory management.
793If none of these apply, just pass NULL as the context argument of
794<i>pcre2_compile()</i>.
795</P>
796<P>
797A compile context is created, copied, and freed by the following functions:
798<br>
799<br>
800<b>pcre2_compile_context *pcre2_compile_context_create(</b>
801<b>  pcre2_general_context *<i>gcontext</i>);</b>
802<br>
803<br>
804<b>pcre2_compile_context *pcre2_compile_context_copy(</b>
805<b>  pcre2_compile_context *<i>ccontext</i>);</b>
806<br>
807<br>
808<b>void pcre2_compile_context_free(pcre2_compile_context *<i>ccontext</i>);</b>
809<br>
810<br>
811A compile context is created with default values for its parameters. These can
812be changed by calling the following functions, which return 0 on success, or
813PCRE2_ERROR_BADDATA if invalid data is detected.
814<br>
815<br>
816<b>int pcre2_set_bsr(pcre2_compile_context *<i>ccontext</i>,</b>
817<b>  uint32_t <i>value</i>);</b>
818<br>
819<br>
820The value must be PCRE2_BSR_ANYCRLF, to specify that \R matches only CR, LF,
821or CRLF, or PCRE2_BSR_UNICODE, to specify that \R matches any Unicode line
822ending sequence. The value is used by the JIT compiler and by the two
823interpreted matching functions, <i>pcre2_match()</i> and
824<i>pcre2_dfa_match()</i>.
825<br>
826<br>
827<b>int pcre2_set_character_tables(pcre2_compile_context *<i>ccontext</i>,</b>
828<b>  const uint8_t *<i>tables</i>);</b>
829<br>
830<br>
831The value must be the result of a call to <b>pcre2_maketables()</b>, whose only
832argument is a general context. This function builds a set of character tables
833in the current locale.
834<br>
835<br>
836<b>int pcre2_set_compile_extra_options(pcre2_compile_context *<i>ccontext</i>,</b>
837<b>  uint32_t <i>extra_options</i>);</b>
838<br>
839<br>
840As PCRE2 has developed, almost all the 32 option bits that are available in
841the <i>options</i> argument of <b>pcre2_compile()</b> have been used up. To avoid
842running out, the compile context contains a set of extra option bits which are
843used for some newer, assumed rarer, options. This function sets those bits. It
844always sets all the bits (either on or off). It does not modify any existing
845setting. The available options are defined in the section entitled "Extra
846compile options"
847<a href="#extracompileoptions">below.</a>
848<br>
849<br>
850<b>int pcre2_set_max_pattern_length(pcre2_compile_context *<i>ccontext</i>,</b>
851<b>  PCRE2_SIZE <i>value</i>);</b>
852<br>
853<br>
854This sets a maximum length, in code units, for any pattern string that is
855compiled with this context. If the pattern is longer, an error is generated.
856This facility is provided so that applications that accept patterns from
857external sources can limit their size. The default is the largest number that a
858PCRE2_SIZE variable can hold, which is effectively unlimited.
859<br>
860<br>
861<b>int pcre2_set_newline(pcre2_compile_context *<i>ccontext</i>,</b>
862<b>  uint32_t <i>value</i>);</b>
863<br>
864<br>
865This specifies which characters or character sequences are to be recognized as
866newlines. The value must be one of PCRE2_NEWLINE_CR (carriage return only),
867PCRE2_NEWLINE_LF (linefeed only), PCRE2_NEWLINE_CRLF (the two-character
868sequence CR followed by LF), PCRE2_NEWLINE_ANYCRLF (any of the above),
869PCRE2_NEWLINE_ANY (any Unicode newline sequence), or PCRE2_NEWLINE_NUL (the
870NUL character, that is a binary zero).
871</P>
872<P>
873A pattern can override the value set in the compile context by starting with a
874sequence such as (*CRLF). See the
875<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
876page for details.
877</P>
878<P>
879When a pattern is compiled with the PCRE2_EXTENDED or PCRE2_EXTENDED_MORE
880option, the newline convention affects the recognition of the end of internal
881comments starting with #. The value is saved with the compiled pattern for
882subsequent use by the JIT compiler and by the two interpreted matching
883functions, <i>pcre2_match()</i> and <i>pcre2_dfa_match()</i>.
884<br>
885<br>
886<b>int pcre2_set_parens_nest_limit(pcre2_compile_context *<i>ccontext</i>,</b>
887<b>  uint32_t <i>value</i>);</b>
888<br>
889<br>
890This parameter adjusts the limit, set when PCRE2 is built (default 250), on the
891depth of parenthesis nesting in a pattern. This limit stops rogue patterns
892using up too much system stack when being compiled. The limit applies to
893parentheses of all kinds, not just capturing parentheses.
894<br>
895<br>
896<b>int pcre2_set_compile_recursion_guard(pcre2_compile_context *<i>ccontext</i>,</b>
897<b>  int (*<i>guard_function</i>)(uint32_t, void *), void *<i>user_data</i>);</b>
898<br>
899<br>
900There is at least one application that runs PCRE2 in threads with very limited
901system stack, where running out of stack is to be avoided at all costs. The
902parenthesis limit above cannot take account of how much stack is actually
903available during compilation. For a finer control, you can supply a function
904that is called whenever <b>pcre2_compile()</b> starts to compile a parenthesized
905part of a pattern. This function can check the actual stack size (or anything
906else that it wants to, of course).
907</P>
908<P>
909The first argument to the callout function gives the current depth of
910nesting, and the second is user data that is set up by the last argument of
911<b>pcre2_set_compile_recursion_guard()</b>. The callout function should return
912zero if all is well, or non-zero to force an error.
913<a name="matchcontext"></a></P>
914<br><b>
915The match context
916</b><br>
917<P>
918A match context is required if you want to:
919<pre>
920  Set up a callout function
921  Set an offset limit for matching an unanchored pattern
922  Change the limit on the amount of heap used when matching
923  Change the backtracking match limit
924  Change the backtracking depth limit
925  Set custom memory management specifically for the match
926</pre>
927If none of these apply, just pass NULL as the context argument of
928<b>pcre2_match()</b>, <b>pcre2_dfa_match()</b>, or <b>pcre2_jit_match()</b>.
929</P>
930<P>
931A match context is created, copied, and freed by the following functions:
932<br>
933<br>
934<b>pcre2_match_context *pcre2_match_context_create(</b>
935<b>  pcre2_general_context *<i>gcontext</i>);</b>
936<br>
937<br>
938<b>pcre2_match_context *pcre2_match_context_copy(</b>
939<b>  pcre2_match_context *<i>mcontext</i>);</b>
940<br>
941<br>
942<b>void pcre2_match_context_free(pcre2_match_context *<i>mcontext</i>);</b>
943<br>
944<br>
945A match context is created with default values for its parameters. These can
946be changed by calling the following functions, which return 0 on success, or
947PCRE2_ERROR_BADDATA if invalid data is detected.
948<br>
949<br>
950<b>int pcre2_set_callout(pcre2_match_context *<i>mcontext</i>,</b>
951<b>  int (*<i>callout_function</i>)(pcre2_callout_block *, void *),</b>
952<b>  void *<i>callout_data</i>);</b>
953<br>
954<br>
955This sets up a callout function for PCRE2 to call at specified points
956during a matching operation. Details are given in the
957<a href="pcre2callout.html"><b>pcre2callout</b></a>
958documentation.
959<br>
960<br>
961<b>int pcre2_set_substitute_callout(pcre2_match_context *<i>mcontext</i>,</b>
962<b>  int (*<i>callout_function</i>)(pcre2_substitute_callout_block *, void *),</b>
963<b>  void *<i>callout_data</i>);</b>
964<br>
965<br>
966This sets up a callout function for PCRE2 to call after each substitution
967made by <b>pcre2_substitute()</b>. Details are given in the section entitled
968"Creating a new string with substitutions"
969<a href="#substitutions">below.</a>
970<br>
971<br>
972<b>int pcre2_set_offset_limit(pcre2_match_context *<i>mcontext</i>,</b>
973<b>  PCRE2_SIZE <i>value</i>);</b>
974<br>
975<br>
976The <i>offset_limit</i> parameter limits how far an unanchored search can
977advance in the subject string. The default value is PCRE2_UNSET. The
978<b>pcre2_match()</b> and <b>pcre2_dfa_match()</b> functions return
979PCRE2_ERROR_NOMATCH if a match with a starting point before or at the given
980offset is not found. The <b>pcre2_substitute()</b> function makes no more
981substitutions.
982</P>
983<P>
984For example, if the pattern /abc/ is matched against "123abc" with an offset
985limit less than 3, the result is PCRE2_ERROR_NOMATCH. A match can never be
986found if the <i>startoffset</i> argument of <b>pcre2_match()</b>,
987<b>pcre2_dfa_match()</b>, or <b>pcre2_substitute()</b> is greater than the offset
988limit set in the match context.
989</P>
990<P>
991When using this facility, you must set the PCRE2_USE_OFFSET_LIMIT option when
992calling <b>pcre2_compile()</b> so that when JIT is in use, different code can be
993compiled. If a match is started with a non-default match limit when
994PCRE2_USE_OFFSET_LIMIT is not set, an error is generated.
995</P>
996<P>
997The offset limit facility can be used to track progress when searching large
998subject strings or to limit the extent of global substitutions. See also the
999PCRE2_FIRSTLINE option, which requires a match to start before or at the first
1000newline that follows the start of matching in the subject. If this is set with
1001an offset limit, a match must occur in the first line and also within the
1002offset limit. In other words, whichever limit comes first is used.
1003<br>
1004<br>
1005<b>int pcre2_set_heap_limit(pcre2_match_context *<i>mcontext</i>,</b>
1006<b>  uint32_t <i>value</i>);</b>
1007<br>
1008<br>
1009The <i>heap_limit</i> parameter specifies, in units of kibibytes (1024 bytes),
1010the maximum amount of heap memory that <b>pcre2_match()</b> may use to hold
1011backtracking information when running an interpretive match. This limit also
1012applies to <b>pcre2_dfa_match()</b>, which may use the heap when processing
1013patterns with a lot of nested pattern recursion or lookarounds or atomic
1014groups. This limit does not apply to matching with the JIT optimization, which
1015has its own memory control arrangements (see the
1016<a href="pcre2jit.html"><b>pcre2jit</b></a>
1017documentation for more details). If the limit is reached, the negative error
1018code PCRE2_ERROR_HEAPLIMIT is returned. The default limit can be set when PCRE2
1019is built; if it is not, the default is set very large and is essentially
1020"unlimited".
1021</P>
1022<P>
1023A value for the heap limit may also be supplied by an item at the start of a
1024pattern of the form
1025<pre>
1026  (*LIMIT_HEAP=ddd)
1027</pre>
1028where ddd is a decimal number. However, such a setting is ignored unless ddd is
1029less than the limit set by the caller of <b>pcre2_match()</b> or, if no such
1030limit is set, less than the default.
1031</P>
1032<P>
1033The <b>pcre2_match()</b> function starts out using a 20KiB vector on the system
1034stack for recording backtracking points. The more nested backtracking points
1035there are (that is, the deeper the search tree), the more memory is needed.
1036Heap memory is used only if the initial vector is too small. If the heap limit
1037is set to a value less than 21 (in particular, zero) no heap memory will be
1038used. In this case, only patterns that do not have a lot of nested backtracking
1039can be successfully processed.
1040</P>
1041<P>
1042Similarly, for <b>pcre2_dfa_match()</b>, a vector on the system stack is used
1043when processing pattern recursions, lookarounds, or atomic groups, and only if
1044this is not big enough is heap memory used. In this case, too, setting a value
1045of zero disables the use of the heap.
1046<br>
1047<br>
1048<b>int pcre2_set_match_limit(pcre2_match_context *<i>mcontext</i>,</b>
1049<b>  uint32_t <i>value</i>);</b>
1050<br>
1051<br>
1052The <i>match_limit</i> parameter provides a means of preventing PCRE2 from using
1053up too many computing resources when processing patterns that are not going to
1054match, but which have a very large number of possibilities in their search
1055trees. The classic example is a pattern that uses nested unlimited repeats.
1056</P>
1057<P>
1058There is an internal counter in <b>pcre2_match()</b> that is incremented each
1059time round its main matching loop. If this value reaches the match limit,
1060<b>pcre2_match()</b> returns the negative value PCRE2_ERROR_MATCHLIMIT. This has
1061the effect of limiting the amount of backtracking that can take place. For
1062patterns that are not anchored, the count restarts from zero for each position
1063in the subject string. This limit also applies to <b>pcre2_dfa_match()</b>,
1064though the counting is done in a different way.
1065</P>
1066<P>
1067When <b>pcre2_match()</b> is called with a pattern that was successfully
1068processed by <b>pcre2_jit_compile()</b>, the way in which matching is executed
1069is entirely different. However, there is still the possibility of runaway
1070matching that goes on for a very long time, and so the <i>match_limit</i> value
1071is also used in this case (but in a different way) to limit how long the
1072matching can continue.
1073</P>
1074<P>
1075The default value for the limit can be set when PCRE2 is built; the default
1076default is 10 million, which handles all but the most extreme cases. A value
1077for the match limit may also be supplied by an item at the start of a pattern
1078of the form
1079<pre>
1080  (*LIMIT_MATCH=ddd)
1081</pre>
1082where ddd is a decimal number. However, such a setting is ignored unless ddd is
1083less than the limit set by the caller of <b>pcre2_match()</b> or
1084<b>pcre2_dfa_match()</b> or, if no such limit is set, less than the default.
1085<br>
1086<br>
1087<b>int pcre2_set_depth_limit(pcre2_match_context *<i>mcontext</i>,</b>
1088<b>  uint32_t <i>value</i>);</b>
1089<br>
1090<br>
1091This parameter limits the depth of nested backtracking in <b>pcre2_match()</b>.
1092Each time a nested backtracking point is passed, a new memory "frame" is used
1093to remember the state of matching at that point. Thus, this parameter
1094indirectly limits the amount of memory that is used in a match. However,
1095because the size of each memory "frame" depends on the number of capturing
1096parentheses, the actual memory limit varies from pattern to pattern. This limit
1097was more useful in versions before 10.30, where function recursion was used for
1098backtracking.
1099</P>
1100<P>
1101The depth limit is not relevant, and is ignored, when matching is done using
1102JIT compiled code. However, it is supported by <b>pcre2_dfa_match()</b>, which
1103uses it to limit the depth of nested internal recursive function calls that
1104implement atomic groups, lookaround assertions, and pattern recursions. This
1105limits, indirectly, the amount of system stack that is used. It was more useful
1106in versions before 10.32, when stack memory was used for local workspace
1107vectors for recursive function calls. From version 10.32, only local variables
1108are allocated on the stack and as each call uses only a few hundred bytes, even
1109a small stack can support quite a lot of recursion.
1110</P>
1111<P>
1112If the depth of internal recursive function calls is great enough, local
1113workspace vectors are allocated on the heap from version 10.32 onwards, so the
1114depth limit also indirectly limits the amount of heap memory that is used. A
1115recursive pattern such as /(.(?2))((?1)|)/, when matched to a very long string
1116using <b>pcre2_dfa_match()</b>, can use a great deal of memory. However, it is
1117probably better to limit heap usage directly by calling
1118<b>pcre2_set_heap_limit()</b>.
1119</P>
1120<P>
1121The default value for the depth limit can be set when PCRE2 is built; if it is
1122not, the default is set to the same value as the default for the match limit.
1123If the limit is exceeded, <b>pcre2_match()</b> or <b>pcre2_dfa_match()</b>
1124returns PCRE2_ERROR_DEPTHLIMIT. A value for the depth limit may also be
1125supplied by an item at the start of a pattern of the form
1126<pre>
1127  (*LIMIT_DEPTH=ddd)
1128</pre>
1129where ddd is a decimal number. However, such a setting is ignored unless ddd is
1130less than the limit set by the caller of <b>pcre2_match()</b> or
1131<b>pcre2_dfa_match()</b> or, if no such limit is set, less than the default.
1132</P>
1133<br><a name="SEC19" href="#TOC1">CHECKING BUILD-TIME OPTIONS</a><br>
1134<P>
1135<b>int pcre2_config(uint32_t <i>what</i>, void *<i>where</i>);</b>
1136</P>
1137<P>
1138The function <b>pcre2_config()</b> makes it possible for a PCRE2 client to find
1139the value of certain configuration parameters and to discover which optional
1140features have been compiled into the PCRE2 library. The
1141<a href="pcre2build.html"><b>pcre2build</b></a>
1142documentation has more details about these features.
1143</P>
1144<P>
1145The first argument for <b>pcre2_config()</b> specifies which information is
1146required. The second argument is a pointer to memory into which the information
1147is placed. If NULL is passed, the function returns the amount of memory that is
1148needed for the requested information. For calls that return numerical values,
1149the value is in bytes; when requesting these values, <i>where</i> should point
1150to appropriately aligned memory. For calls that return strings, the required
1151length is given in code units, not counting the terminating zero.
1152</P>
1153<P>
1154When requesting information, the returned value from <b>pcre2_config()</b> is
1155non-negative on success, or the negative error code PCRE2_ERROR_BADOPTION if
1156the value in the first argument is not recognized. The following information is
1157available:
1158<pre>
1159  PCRE2_CONFIG_BSR
1160</pre>
1161The output is a uint32_t integer whose value indicates what character
1162sequences the \R escape sequence matches by default. A value of
1163PCRE2_BSR_UNICODE means that \R matches any Unicode line ending sequence; a
1164value of PCRE2_BSR_ANYCRLF means that \R matches only CR, LF, or CRLF. The
1165default can be overridden when a pattern is compiled.
1166<pre>
1167  PCRE2_CONFIG_COMPILED_WIDTHS
1168</pre>
1169The output is a uint32_t integer whose lower bits indicate which code unit
1170widths were selected when PCRE2 was built. The 1-bit indicates 8-bit support,
1171and the 2-bit and 4-bit indicate 16-bit and 32-bit support, respectively.
1172<pre>
1173  PCRE2_CONFIG_DEPTHLIMIT
1174</pre>
1175The output is a uint32_t integer that gives the default limit for the depth of
1176nested backtracking in <b>pcre2_match()</b> or the depth of nested recursions,
1177lookarounds, and atomic groups in <b>pcre2_dfa_match()</b>. Further details are
1178given with <b>pcre2_set_depth_limit()</b> above.
1179<pre>
1180  PCRE2_CONFIG_HEAPLIMIT
1181</pre>
1182The output is a uint32_t integer that gives, in kibibytes, the default limit
1183for the amount of heap memory used by <b>pcre2_match()</b> or
1184<b>pcre2_dfa_match()</b>. Further details are given with
1185<b>pcre2_set_heap_limit()</b> above.
1186<pre>
1187  PCRE2_CONFIG_JIT
1188</pre>
1189The output is a uint32_t integer that is set to one if support for just-in-time
1190compiling is available; otherwise it is set to zero.
1191<pre>
1192  PCRE2_CONFIG_JITTARGET
1193</pre>
1194The <i>where</i> argument should point to a buffer that is at least 48 code
1195units long. (The exact length required can be found by calling
1196<b>pcre2_config()</b> with <b>where</b> set to NULL.) The buffer is filled with a
1197string that contains the name of the architecture for which the JIT compiler is
1198configured, for example "x86 32bit (little endian + unaligned)". If JIT support
1199is not available, PCRE2_ERROR_BADOPTION is returned, otherwise the number of
1200code units used is returned. This is the length of the string, plus one unit
1201for the terminating zero.
1202<pre>
1203  PCRE2_CONFIG_LINKSIZE
1204</pre>
1205The output is a uint32_t integer that contains the number of bytes used for
1206internal linkage in compiled regular expressions. When PCRE2 is configured, the
1207value can be set to 2, 3, or 4, with the default being 2. This is the value
1208that is returned by <b>pcre2_config()</b>. However, when the 16-bit library is
1209compiled, a value of 3 is rounded up to 4, and when the 32-bit library is
1210compiled, internal linkages always use 4 bytes, so the configured value is not
1211relevant.
1212</P>
1213<P>
1214The default value of 2 for the 8-bit and 16-bit libraries is sufficient for all
1215but the most massive patterns, since it allows the size of the compiled pattern
1216to be up to 65535 code units. Larger values allow larger regular expressions to
1217be compiled by those two libraries, but at the expense of slower matching.
1218<pre>
1219  PCRE2_CONFIG_MATCHLIMIT
1220</pre>
1221The output is a uint32_t integer that gives the default match limit for
1222<b>pcre2_match()</b>. Further details are given with
1223<b>pcre2_set_match_limit()</b> above.
1224<pre>
1225  PCRE2_CONFIG_NEWLINE
1226</pre>
1227The output is a uint32_t integer whose value specifies the default character
1228sequence that is recognized as meaning "newline". The values are:
1229<pre>
1230  PCRE2_NEWLINE_CR       Carriage return (CR)
1231  PCRE2_NEWLINE_LF       Linefeed (LF)
1232  PCRE2_NEWLINE_CRLF     Carriage return, linefeed (CRLF)
1233  PCRE2_NEWLINE_ANY      Any Unicode line ending
1234  PCRE2_NEWLINE_ANYCRLF  Any of CR, LF, or CRLF
1235  PCRE2_NEWLINE_NUL      The NUL character (binary zero)
1236</pre>
1237The default should normally correspond to the standard sequence for your
1238operating system.
1239<pre>
1240  PCRE2_CONFIG_NEVER_BACKSLASH_C
1241</pre>
1242The output is a uint32_t integer that is set to one if the use of \C was
1243permanently disabled when PCRE2 was built; otherwise it is set to zero.
1244<pre>
1245  PCRE2_CONFIG_PARENSLIMIT
1246</pre>
1247The output is a uint32_t integer that gives the maximum depth of nesting
1248of parentheses (of any kind) in a pattern. This limit is imposed to cap the
1249amount of system stack used when a pattern is compiled. It is specified when
1250PCRE2 is built; the default is 250. This limit does not take into account the
1251stack that may already be used by the calling application. For finer control
1252over compilation stack usage, see <b>pcre2_set_compile_recursion_guard()</b>.
1253<pre>
1254  PCRE2_CONFIG_STACKRECURSE
1255</pre>
1256This parameter is obsolete and should not be used in new code. The output is a
1257uint32_t integer that is always set to zero.
1258<pre>
1259  PCRE2_CONFIG_TABLES_LENGTH
1260</pre>
1261The output is a uint32_t integer that gives the length of PCRE2's character
1262processing tables in bytes. For details of these tables see the
1263<a href="#localesupport">section on locale support</a>
1264below.
1265<pre>
1266  PCRE2_CONFIG_UNICODE_VERSION
1267</pre>
1268The <i>where</i> argument should point to a buffer that is at least 24 code
1269units long. (The exact length required can be found by calling
1270<b>pcre2_config()</b> with <b>where</b> set to NULL.) If PCRE2 has been compiled
1271without Unicode support, the buffer is filled with the text "Unicode not
1272supported". Otherwise, the Unicode version string (for example, "8.0.0") is
1273inserted. The number of code units used is returned. This is the length of the
1274string plus one unit for the terminating zero.
1275<pre>
1276  PCRE2_CONFIG_UNICODE
1277</pre>
1278The output is a uint32_t integer that is set to one if Unicode support is
1279available; otherwise it is set to zero. Unicode support implies UTF support.
1280<pre>
1281  PCRE2_CONFIG_VERSION
1282</pre>
1283The <i>where</i> argument should point to a buffer that is at least 24 code
1284units long. (The exact length required can be found by calling
1285<b>pcre2_config()</b> with <b>where</b> set to NULL.) The buffer is filled with
1286the PCRE2 version string, zero-terminated. The number of code units used is
1287returned. This is the length of the string plus one unit for the terminating
1288zero.
1289<a name="compiling"></a></P>
1290<br><a name="SEC20" href="#TOC1">COMPILING A PATTERN</a><br>
1291<P>
1292<b>pcre2_code *pcre2_compile(PCRE2_SPTR <i>pattern</i>, PCRE2_SIZE <i>length</i>,</b>
1293<b>  uint32_t <i>options</i>, int *<i>errorcode</i>, PCRE2_SIZE *<i>erroroffset,</i></b>
1294<b>  pcre2_compile_context *<i>ccontext</i>);</b>
1295<br>
1296<br>
1297<b>void pcre2_code_free(pcre2_code *<i>code</i>);</b>
1298<br>
1299<br>
1300<b>pcre2_code *pcre2_code_copy(const pcre2_code *<i>code</i>);</b>
1301<br>
1302<br>
1303<b>pcre2_code *pcre2_code_copy_with_tables(const pcre2_code *<i>code</i>);</b>
1304</P>
1305<P>
1306The <b>pcre2_compile()</b> function compiles a pattern into an internal form.
1307The pattern is defined by a pointer to a string of code units and a length (in
1308code units). If the pattern is zero-terminated, the length can be specified as
1309PCRE2_ZERO_TERMINATED. The function returns a pointer to a block of memory that
1310contains the compiled pattern and related data, or NULL if an error occurred.
1311</P>
1312<P>
1313If the compile context argument <i>ccontext</i> is NULL, memory for the compiled
1314pattern is obtained by calling <b>malloc()</b>. Otherwise, it is obtained from
1315the same memory function that was used for the compile context. The caller must
1316free the memory by calling <b>pcre2_code_free()</b> when it is no longer needed.
1317If <b>pcre2_code_free()</b> is called with a NULL argument, it returns
1318immediately, without doing anything.
1319</P>
1320<P>
1321The function <b>pcre2_code_copy()</b> makes a copy of the compiled code in new
1322memory, using the same memory allocator as was used for the original. However,
1323if the code has been processed by the JIT compiler (see
1324<a href="#jitcompiling">below),</a>
1325the JIT information cannot be copied (because it is position-dependent).
1326The new copy can initially be used only for non-JIT matching, though it can be
1327passed to <b>pcre2_jit_compile()</b> if required. If <b>pcre2_code_copy()</b> is
1328called with a NULL argument, it returns NULL.
1329</P>
1330<P>
1331The <b>pcre2_code_copy()</b> function provides a way for individual threads in a
1332multithreaded application to acquire a private copy of shared compiled code.
1333However, it does not make a copy of the character tables used by the compiled
1334pattern; the new pattern code points to the same tables as the original code.
1335(See
1336<a href="#jitcompiling">"Locale Support"</a>
1337below for details of these character tables.) In many applications the same
1338tables are used throughout, so this behaviour is appropriate. Nevertheless,
1339there are occasions when a copy of a compiled pattern and the relevant tables
1340are needed. The <b>pcre2_code_copy_with_tables()</b> provides this facility.
1341Copies of both the code and the tables are made, with the new code pointing to
1342the new tables. The memory for the new tables is automatically freed when
1343<b>pcre2_code_free()</b> is called for the new copy of the compiled code. If
1344<b>pcre2_code_copy_with_tables()</b> is called with a NULL argument, it returns
1345NULL.
1346</P>
1347<P>
1348NOTE: When one of the matching functions is called, pointers to the compiled
1349pattern and the subject string are set in the match data block so that they can
1350be referenced by the substring extraction functions after a successful match.
1351After running a match, you must not free a compiled pattern or a subject string
1352until after all operations on the
1353<a href="#matchdatablock">match data block</a>
1354have taken place, unless, in the case of the subject string, you have used the
1355PCRE2_COPY_MATCHED_SUBJECT option, which is described in the section entitled
1356"Option bits for <b>pcre2_match()</b>"
1357<a href="#matchoptions>">below.</a>
1358</P>
1359<P>
1360The <i>options</i> argument for <b>pcre2_compile()</b> contains various bit
1361settings that affect the compilation. It should be zero if none of them are
1362required. The available options are described below. Some of them (in
1363particular, those that are compatible with Perl, but some others as well) can
1364also be set and unset from within the pattern (see the detailed description in
1365the
1366<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
1367documentation).
1368</P>
1369<P>
1370For those options that can be different in different parts of the pattern, the
1371contents of the <i>options</i> argument specifies their settings at the start of
1372compilation. The PCRE2_ANCHORED, PCRE2_ENDANCHORED, and PCRE2_NO_UTF_CHECK
1373options can be set at the time of matching as well as at compile time.
1374</P>
1375<P>
1376Some additional options and less frequently required compile-time parameters
1377(for example, the newline setting) can be provided in a compile context (as
1378described
1379<a href="#compilecontext">above).</a>
1380</P>
1381<P>
1382If <i>errorcode</i> or <i>erroroffset</i> is NULL, <b>pcre2_compile()</b> returns
1383NULL immediately. Otherwise, the variables to which these point are set to an
1384error code and an offset (number of code units) within the pattern,
1385respectively, when <b>pcre2_compile()</b> returns NULL because a compilation
1386error has occurred. The values are not defined when compilation is successful
1387and <b>pcre2_compile()</b> returns a non-NULL value.
1388</P>
1389<P>
1390There are nearly 100 positive error codes that <b>pcre2_compile()</b> may return
1391if it finds an error in the pattern. There are also some negative error codes
1392that are used for invalid UTF strings when validity checking is in force. These
1393are the same as given by <b>pcre2_match()</b> and <b>pcre2_dfa_match()</b>, and
1394are described in the
1395<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
1396documentation. There is no separate documentation for the positive error codes,
1397because the textual error messages that are obtained by calling the
1398<b>pcre2_get_error_message()</b> function (see "Obtaining a textual error
1399message"
1400<a href="#geterrormessage">below)</a>
1401should be self-explanatory. Macro names starting with PCRE2_ERROR_ are defined
1402for both positive and negative error codes in <b>pcre2.h</b>.
1403</P>
1404<P>
1405The value returned in <i>erroroffset</i> is an indication of where in the
1406pattern the error occurred. It is not necessarily the furthest point in the
1407pattern that was read. For example, after the error "lookbehind assertion is
1408not fixed length", the error offset points to the start of the failing
1409assertion. For an invalid UTF-8 or UTF-16 string, the offset is that of the
1410first code unit of the failing character.
1411</P>
1412<P>
1413Some errors are not detected until the whole pattern has been scanned; in these
1414cases, the offset passed back is the length of the pattern. Note that the
1415offset is in code units, not characters, even in a UTF mode. It may sometimes
1416point into the middle of a UTF-8 or UTF-16 character.
1417</P>
1418<P>
1419This code fragment shows a typical straightforward call to
1420<b>pcre2_compile()</b>:
1421<pre>
1422  pcre2_code *re;
1423  PCRE2_SIZE erroffset;
1424  int errorcode;
1425  re = pcre2_compile(
1426    "^A.*Z",                /* the pattern */
1427    PCRE2_ZERO_TERMINATED,  /* the pattern is zero-terminated */
1428    0,                      /* default options */
1429    &errorcode,             /* for error code */
1430    &erroffset,             /* for error offset */
1431    NULL);                  /* no compile context */
1432
1433</PRE>
1434</P>
1435<br><b>
1436Main compile options
1437</b><br>
1438<P>
1439The following names for option bits are defined in the <b>pcre2.h</b> header
1440file:
1441<pre>
1442  PCRE2_ANCHORED
1443</pre>
1444If this bit is set, the pattern is forced to be "anchored", that is, it is
1445constrained to match only at the first matching point in the string that is
1446being searched (the "subject string"). This effect can also be achieved by
1447appropriate constructs in the pattern itself, which is the only way to do it in
1448Perl.
1449<pre>
1450  PCRE2_ALLOW_EMPTY_CLASS
1451</pre>
1452By default, for compatibility with Perl, a closing square bracket that
1453immediately follows an opening one is treated as a data character for the
1454class. When PCRE2_ALLOW_EMPTY_CLASS is set, it terminates the class, which
1455therefore contains no characters and so can never match.
1456<pre>
1457  PCRE2_ALT_BSUX
1458</pre>
1459This option request alternative handling of three escape sequences, which
1460makes PCRE2's behaviour more like ECMAscript (aka JavaScript). When it is set:
1461</P>
1462<P>
1463(1) \U matches an upper case "U" character; by default \U causes a compile
1464time error (Perl uses \U to upper case subsequent characters).
1465</P>
1466<P>
1467(2) \u matches a lower case "u" character unless it is followed by four
1468hexadecimal digits, in which case the hexadecimal number defines the code point
1469to match. By default, \u causes a compile time error (Perl uses it to upper
1470case the following character).
1471</P>
1472<P>
1473(3) \x matches a lower case "x" character unless it is followed by two
1474hexadecimal digits, in which case the hexadecimal number defines the code point
1475to match. By default, as in Perl, a hexadecimal number is always expected after
1476\x, but it may have zero, one, or two digits (so, for example, \xz matches a
1477binary zero character followed by z).
1478</P>
1479<P>
1480ECMAscript 6 added additional functionality to \u. This can be accessed using
1481the PCRE2_EXTRA_ALT_BSUX extra option (see "Extra compile options"
1482<a href="#extracompileoptions">below).</a>
1483Note that this alternative escape handling applies only to patterns. Neither of
1484these options affects the processing of replacement strings passed to
1485<b>pcre2_substitute()</b>.
1486<pre>
1487  PCRE2_ALT_CIRCUMFLEX
1488</pre>
1489In multiline mode (when PCRE2_MULTILINE is set), the circumflex metacharacter
1490matches at the start of the subject (unless PCRE2_NOTBOL is set), and also
1491after any internal newline. However, it does not match after a newline at the
1492end of the subject, for compatibility with Perl. If you want a multiline
1493circumflex also to match after a terminating newline, you must set
1494PCRE2_ALT_CIRCUMFLEX.
1495<pre>
1496  PCRE2_ALT_VERBNAMES
1497</pre>
1498By default, for compatibility with Perl, the name in any verb sequence such as
1499(*MARK:NAME) is any sequence of characters that does not include a closing
1500parenthesis. The name is not processed in any way, and it is not possible to
1501include a closing parenthesis in the name. However, if the PCRE2_ALT_VERBNAMES
1502option is set, normal backslash processing is applied to verb names and only an
1503unescaped closing parenthesis terminates the name. A closing parenthesis can be
1504included in a name either as \) or between \Q and \E. If the PCRE2_EXTENDED
1505or PCRE2_EXTENDED_MORE option is set with PCRE2_ALT_VERBNAMES, unescaped
1506whitespace in verb names is skipped and #-comments are recognized, exactly as
1507in the rest of the pattern.
1508<pre>
1509  PCRE2_AUTO_CALLOUT
1510</pre>
1511If this bit is set, <b>pcre2_compile()</b> automatically inserts callout items,
1512all with number 255, before each pattern item, except immediately before or
1513after an explicit callout in the pattern. For discussion of the callout
1514facility, see the
1515<a href="pcre2callout.html"><b>pcre2callout</b></a>
1516documentation.
1517<pre>
1518  PCRE2_CASELESS
1519</pre>
1520If this bit is set, letters in the pattern match both upper and lower case
1521letters in the subject. It is equivalent to Perl's /i option, and it can be
1522changed within a pattern by a (?i) option setting. If either PCRE2_UTF or
1523PCRE2_UCP is set, Unicode properties are used for all characters with more than
1524one other case, and for all characters whose code points are greater than
1525U+007F. Note that there are two ASCII characters, K and S, that, in addition to
1526their lower case ASCII equivalents, are case-equivalent with U+212A (Kelvin
1527sign) and U+017F (long S) respectively. For lower valued characters with only
1528one other case, a lookup table is used for speed. When neither PCRE2_UTF nor
1529PCRE2_UCP is set, a lookup table is used for all code points less than 256, and
1530higher code points (available only in 16-bit or 32-bit mode) are treated as not
1531having another case.
1532<pre>
1533  PCRE2_DOLLAR_ENDONLY
1534</pre>
1535If this bit is set, a dollar metacharacter in the pattern matches only at the
1536end of the subject string. Without this option, a dollar also matches
1537immediately before a newline at the end of the string (but not before any other
1538newlines). The PCRE2_DOLLAR_ENDONLY option is ignored if PCRE2_MULTILINE is
1539set. There is no equivalent to this option in Perl, and no way to set it within
1540a pattern.
1541<pre>
1542  PCRE2_DOTALL
1543</pre>
1544If this bit is set, a dot metacharacter in the pattern matches any character,
1545including one that indicates a newline. However, it only ever matches one
1546character, even if newlines are coded as CRLF. Without this option, a dot does
1547not match when the current position in the subject is at a newline. This option
1548is equivalent to Perl's /s option, and it can be changed within a pattern by a
1549(?s) option setting. A negative class such as [^a] always matches newline
1550characters, and the \N escape sequence always matches a non-newline character,
1551independent of the setting of PCRE2_DOTALL.
1552<pre>
1553  PCRE2_DUPNAMES
1554</pre>
1555If this bit is set, names used to identify capture groups need not be unique.
1556This can be helpful for certain types of pattern when it is known that only one
1557instance of the named group can ever be matched. There are more details of
1558named capture groups below; see also the
1559<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
1560documentation.
1561<pre>
1562  PCRE2_ENDANCHORED
1563</pre>
1564If this bit is set, the end of any pattern match must be right at the end of
1565the string being searched (the "subject string"). If the pattern match
1566succeeds by reaching (*ACCEPT), but does not reach the end of the subject, the
1567match fails at the current starting point. For unanchored patterns, a new match
1568is then tried at the next starting point. However, if the match succeeds by
1569reaching the end of the pattern, but not the end of the subject, backtracking
1570occurs and an alternative match may be found. Consider these two patterns:
1571<pre>
1572  .(*ACCEPT)|..
1573  .|..
1574</pre>
1575If matched against "abc" with PCRE2_ENDANCHORED set, the first matches "c"
1576whereas the second matches "bc". The effect of PCRE2_ENDANCHORED can also be
1577achieved by appropriate constructs in the pattern itself, which is the only way
1578to do it in Perl.
1579</P>
1580<P>
1581For DFA matching with <b>pcre2_dfa_match()</b>, PCRE2_ENDANCHORED applies only
1582to the first (that is, the longest) matched string. Other parallel matches,
1583which are necessarily substrings of the first one, must obviously end before
1584the end of the subject.
1585<pre>
1586  PCRE2_EXTENDED
1587</pre>
1588If this bit is set, most white space characters in the pattern are totally
1589ignored except when escaped or inside a character class. However, white space
1590is not allowed within sequences such as (?&#62; that introduce various
1591parenthesized groups, nor within numerical quantifiers such as {1,3}. Ignorable
1592white space is permitted between an item and a following quantifier and between
1593a quantifier and a following + that indicates possessiveness. PCRE2_EXTENDED is
1594equivalent to Perl's /x option, and it can be changed within a pattern by a
1595(?x) option setting.
1596</P>
1597<P>
1598When PCRE2 is compiled without Unicode support, PCRE2_EXTENDED recognizes as
1599white space only those characters with code points less than 256 that are
1600flagged as white space in its low-character table. The table is normally
1601created by
1602<a href="pcre2_maketables.html"><b>pcre2_maketables()</b>,</a>
1603which uses the <b>isspace()</b> function to identify space characters. In most
1604ASCII environments, the relevant characters are those with code points 0x0009
1605(tab), 0x000A (linefeed), 0x000B (vertical tab), 0x000C (formfeed), 0x000D
1606(carriage return), and 0x0020 (space).
1607</P>
1608<P>
1609When PCRE2 is compiled with Unicode support, in addition to these characters,
1610five more Unicode "Pattern White Space" characters are recognized by
1611PCRE2_EXTENDED. These are U+0085 (next line), U+200E (left-to-right mark),
1612U+200F (right-to-left mark), U+2028 (line separator), and U+2029 (paragraph
1613separator). This set of characters is the same as recognized by Perl's /x
1614option. Note that the horizontal and vertical space characters that are matched
1615by the \h and \v escapes in patterns are a much bigger set.
1616</P>
1617<P>
1618As well as ignoring most white space, PCRE2_EXTENDED also causes characters
1619between an unescaped # outside a character class and the next newline,
1620inclusive, to be ignored, which makes it possible to include comments inside
1621complicated patterns. Note that the end of this type of comment is a literal
1622newline sequence in the pattern; escape sequences that happen to represent a
1623newline do not count.
1624</P>
1625<P>
1626Which characters are interpreted as newlines can be specified by a setting in
1627the compile context that is passed to <b>pcre2_compile()</b> or by a special
1628sequence at the start of the pattern, as described in the section entitled
1629<a href="pcre2pattern.html#newlines">"Newline conventions"</a>
1630in the <b>pcre2pattern</b> documentation. A default is defined when PCRE2 is
1631built.
1632<pre>
1633  PCRE2_EXTENDED_MORE
1634</pre>
1635This option has the effect of PCRE2_EXTENDED, but, in addition, unescaped space
1636and horizontal tab characters are ignored inside a character class. Note: only
1637these two characters are ignored, not the full set of pattern white space
1638characters that are ignored outside a character class. PCRE2_EXTENDED_MORE is
1639equivalent to Perl's /xx option, and it can be changed within a pattern by a
1640(?xx) option setting.
1641<pre>
1642  PCRE2_FIRSTLINE
1643</pre>
1644If this option is set, the start of an unanchored pattern match must be before
1645or at the first newline in the subject string following the start of matching,
1646though the matched text may continue over the newline. If <i>startoffset</i> is
1647non-zero, the limiting newline is not necessarily the first newline in the
1648subject. For example, if the subject string is "abc\nxyz" (where \n
1649represents a single-character newline) a pattern match for "yz" succeeds with
1650PCRE2_FIRSTLINE if <i>startoffset</i> is greater than 3. See also
1651PCRE2_USE_OFFSET_LIMIT, which provides a more general limiting facility. If
1652PCRE2_FIRSTLINE is set with an offset limit, a match must occur in the first
1653line and also within the offset limit. In other words, whichever limit comes
1654first is used.
1655<pre>
1656  PCRE2_LITERAL
1657</pre>
1658If this option is set, all meta-characters in the pattern are disabled, and it
1659is treated as a literal string. Matching literal strings with a regular
1660expression engine is not the most efficient way of doing it. If you are doing a
1661lot of literal matching and are worried about efficiency, you should consider
1662using other approaches. The only other main options that are allowed with
1663PCRE2_LITERAL are: PCRE2_ANCHORED, PCRE2_ENDANCHORED, PCRE2_AUTO_CALLOUT,
1664PCRE2_CASELESS, PCRE2_FIRSTLINE, PCRE2_MATCH_INVALID_UTF,
1665PCRE2_NO_START_OPTIMIZE, PCRE2_NO_UTF_CHECK, PCRE2_UTF, and
1666PCRE2_USE_OFFSET_LIMIT. The extra options PCRE2_EXTRA_MATCH_LINE and
1667PCRE2_EXTRA_MATCH_WORD are also supported. Any other options cause an error.
1668<pre>
1669  PCRE2_MATCH_INVALID_UTF
1670</pre>
1671This option forces PCRE2_UTF (see below) and also enables support for matching
1672by <b>pcre2_match()</b> in subject strings that contain invalid UTF sequences.
1673This facility is not supported for DFA matching. For details, see the
1674<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
1675documentation.
1676<pre>
1677  PCRE2_MATCH_UNSET_BACKREF
1678</pre>
1679If this option is set, a backreference to an unset capture group matches an
1680empty string (by default this causes the current matching alternative to fail).
1681A pattern such as (\1)(a) succeeds when this option is set (assuming it can
1682find an "a" in the subject), whereas it fails by default, for Perl
1683compatibility. Setting this option makes PCRE2 behave more like ECMAscript (aka
1684JavaScript).
1685<pre>
1686  PCRE2_MULTILINE
1687</pre>
1688By default, for the purposes of matching "start of line" and "end of line",
1689PCRE2 treats the subject string as consisting of a single line of characters,
1690even if it actually contains newlines. The "start of line" metacharacter (^)
1691matches only at the start of the string, and the "end of line" metacharacter
1692($) matches only at the end of the string, or before a terminating newline
1693(except when PCRE2_DOLLAR_ENDONLY is set). Note, however, that unless
1694PCRE2_DOTALL is set, the "any character" metacharacter (.) does not match at a
1695newline. This behaviour (for ^, $, and dot) is the same as Perl.
1696</P>
1697<P>
1698When PCRE2_MULTILINE it is set, the "start of line" and "end of line"
1699constructs match immediately following or immediately before internal newlines
1700in the subject string, respectively, as well as at the very start and end. This
1701is equivalent to Perl's /m option, and it can be changed within a pattern by a
1702(?m) option setting. Note that the "start of line" metacharacter does not match
1703after a newline at the end of the subject, for compatibility with Perl.
1704However, you can change this by setting the PCRE2_ALT_CIRCUMFLEX option. If
1705there are no newlines in a subject string, or no occurrences of ^ or $ in a
1706pattern, setting PCRE2_MULTILINE has no effect.
1707<pre>
1708  PCRE2_NEVER_BACKSLASH_C
1709</pre>
1710This option locks out the use of \C in the pattern that is being compiled.
1711This escape can cause unpredictable behaviour in UTF-8 or UTF-16 modes, because
1712it may leave the current matching point in the middle of a multi-code-unit
1713character. This option may be useful in applications that process patterns from
1714external sources. Note that there is also a build-time option that permanently
1715locks out the use of \C.
1716<pre>
1717  PCRE2_NEVER_UCP
1718</pre>
1719This option locks out the use of Unicode properties for handling \B, \b, \D,
1720\d, \S, \s, \W, \w, and some of the POSIX character classes, as described
1721for the PCRE2_UCP option below. In particular, it prevents the creator of the
1722pattern from enabling this facility by starting the pattern with (*UCP). This
1723option may be useful in applications that process patterns from external
1724sources. The option combination PCRE_UCP and PCRE_NEVER_UCP causes an error.
1725<pre>
1726  PCRE2_NEVER_UTF
1727</pre>
1728This option locks out interpretation of the pattern as UTF-8, UTF-16, or
1729UTF-32, depending on which library is in use. In particular, it prevents the
1730creator of the pattern from switching to UTF interpretation by starting the
1731pattern with (*UTF). This option may be useful in applications that process
1732patterns from external sources. The combination of PCRE2_UTF and
1733PCRE2_NEVER_UTF causes an error.
1734<pre>
1735  PCRE2_NO_AUTO_CAPTURE
1736</pre>
1737If this option is set, it disables the use of numbered capturing parentheses in
1738the pattern. Any opening parenthesis that is not followed by ? behaves as if it
1739were followed by ?: but named parentheses can still be used for capturing (and
1740they acquire numbers in the usual way). This is the same as Perl's /n option.
1741Note that, when this option is set, references to capture groups
1742(backreferences or recursion/subroutine calls) may only refer to named groups,
1743though the reference can be by name or by number.
1744<pre>
1745  PCRE2_NO_AUTO_POSSESS
1746</pre>
1747If this option is set, it disables "auto-possessification", which is an
1748optimization that, for example, turns a+b into a++b in order to avoid
1749backtracks into a+ that can never be successful. However, if callouts are in
1750use, auto-possessification means that some callouts are never taken. You can
1751set this option if you want the matching functions to do a full unoptimized
1752search and run all the callouts, but it is mainly provided for testing
1753purposes.
1754<pre>
1755  PCRE2_NO_DOTSTAR_ANCHOR
1756</pre>
1757If this option is set, it disables an optimization that is applied when .* is
1758the first significant item in a top-level branch of a pattern, and all the
1759other branches also start with .* or with \A or \G or ^. The optimization is
1760automatically disabled for .* if it is inside an atomic group or a capture
1761group that is the subject of a backreference, or if the pattern contains
1762(*PRUNE) or (*SKIP). When the optimization is not disabled, such a pattern is
1763automatically anchored if PCRE2_DOTALL is set for all the .* items and
1764PCRE2_MULTILINE is not set for any ^ items. Otherwise, the fact that any match
1765must start either at the start of the subject or following a newline is
1766remembered. Like other optimizations, this can cause callouts to be skipped.
1767<pre>
1768  PCRE2_NO_START_OPTIMIZE
1769</pre>
1770This is an option whose main effect is at matching time. It does not change
1771what <b>pcre2_compile()</b> generates, but it does affect the output of the JIT
1772compiler.
1773</P>
1774<P>
1775There are a number of optimizations that may occur at the start of a match, in
1776order to speed up the process. For example, if it is known that an unanchored
1777match must start with a specific code unit value, the matching code searches
1778the subject for that value, and fails immediately if it cannot find it, without
1779actually running the main matching function. This means that a special item
1780such as (*COMMIT) at the start of a pattern is not considered until after a
1781suitable starting point for the match has been found. Also, when callouts or
1782(*MARK) items are in use, these "start-up" optimizations can cause them to be
1783skipped if the pattern is never actually used. The start-up optimizations are
1784in effect a pre-scan of the subject that takes place before the pattern is run.
1785</P>
1786<P>
1787The PCRE2_NO_START_OPTIMIZE option disables the start-up optimizations,
1788possibly causing performance to suffer, but ensuring that in cases where the
1789result is "no match", the callouts do occur, and that items such as (*COMMIT)
1790and (*MARK) are considered at every possible starting position in the subject
1791string.
1792</P>
1793<P>
1794Setting PCRE2_NO_START_OPTIMIZE may change the outcome of a matching operation.
1795Consider the pattern
1796<pre>
1797  (*COMMIT)ABC
1798</pre>
1799When this is compiled, PCRE2 records the fact that a match must start with the
1800character "A". Suppose the subject string is "DEFABC". The start-up
1801optimization scans along the subject, finds "A" and runs the first match
1802attempt from there. The (*COMMIT) item means that the pattern must match the
1803current starting position, which in this case, it does. However, if the same
1804match is run with PCRE2_NO_START_OPTIMIZE set, the initial scan along the
1805subject string does not happen. The first match attempt is run starting from
1806"D" and when this fails, (*COMMIT) prevents any further matches being tried, so
1807the overall result is "no match".
1808</P>
1809<P>
1810As another start-up optimization makes use of a minimum length for a matching
1811subject, which is recorded when possible. Consider the pattern
1812<pre>
1813  (*MARK:1)B(*MARK:2)(X|Y)
1814</pre>
1815The minimum length for a match is two characters. If the subject is "XXBB", the
1816"starting character" optimization skips "XX", then tries to match "BB", which
1817is long enough. In the process, (*MARK:2) is encountered and remembered. When
1818the match attempt fails, the next "B" is found, but there is only one character
1819left, so there are no more attempts, and "no match" is returned with the "last
1820mark seen" set to "2". If NO_START_OPTIMIZE is set, however, matches are tried
1821at every possible starting position, including at the end of the subject, where
1822(*MARK:1) is encountered, but there is no "B", so the "last mark seen" that is
1823returned is "1". In this case, the optimizations do not affect the overall
1824match result, which is still "no match", but they do affect the auxiliary
1825information that is returned.
1826<pre>
1827  PCRE2_NO_UTF_CHECK
1828</pre>
1829When PCRE2_UTF is set, the validity of the pattern as a UTF string is
1830automatically checked. There are discussions about the validity of
1831<a href="pcre2unicode.html#utf8strings">UTF-8 strings,</a>
1832<a href="pcre2unicode.html#utf16strings">UTF-16 strings,</a>
1833and
1834<a href="pcre2unicode.html#utf32strings">UTF-32 strings</a>
1835in the
1836<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
1837document. If an invalid UTF sequence is found, <b>pcre2_compile()</b> returns a
1838negative error code.
1839</P>
1840<P>
1841If you know that your pattern is a valid UTF string, and you want to skip this
1842check for performance reasons, you can set the PCRE2_NO_UTF_CHECK option. When
1843it is set, the effect of passing an invalid UTF string as a pattern is
1844undefined. It may cause your program to crash or loop.
1845</P>
1846<P>
1847Note that this option can also be passed to <b>pcre2_match()</b> and
1848<b>pcre2_dfa_match()</b>, to suppress UTF validity checking of the subject
1849string.
1850</P>
1851<P>
1852Note also that setting PCRE2_NO_UTF_CHECK at compile time does not disable the
1853error that is given if an escape sequence for an invalid Unicode code point is
1854encountered in the pattern. In particular, the so-called "surrogate" code
1855points (0xd800 to 0xdfff) are invalid. If you want to allow escape sequences
1856such as \x{d800} you can set the PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES extra
1857option, as described in the section entitled "Extra compile options"
1858<a href="#extracompileoptions">below.</a>
1859However, this is possible only in UTF-8 and UTF-32 modes, because these values
1860are not representable in UTF-16.
1861<pre>
1862  PCRE2_UCP
1863</pre>
1864This option has two effects. Firstly, it change the way PCRE2 processes \B,
1865\b, \D, \d, \S, \s, \W, \w, and some of the POSIX character classes. By
1866default, only ASCII characters are recognized, but if PCRE2_UCP is set, Unicode
1867properties are used instead to classify characters. More details are given in
1868the section on
1869<a href="pcre2pattern.html#genericchartypes">generic character types</a>
1870in the
1871<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
1872page. If you set PCRE2_UCP, matching one of the items it affects takes much
1873longer.
1874</P>
1875<P>
1876The second effect of PCRE2_UCP is to force the use of Unicode properties for
1877upper/lower casing operations on characters with code points greater than 127,
1878even when PCRE2_UTF is not set. This makes it possible, for example, to process
1879strings in the 16-bit UCS-2 code. This option is available only if PCRE2 has
1880been compiled with Unicode support (which is the default).
1881<pre>
1882  PCRE2_UNGREEDY
1883</pre>
1884This option inverts the "greediness" of the quantifiers so that they are not
1885greedy by default, but become greedy if followed by "?". It is not compatible
1886with Perl. It can also be set by a (?U) option setting within the pattern.
1887<pre>
1888  PCRE2_USE_OFFSET_LIMIT
1889</pre>
1890This option must be set for <b>pcre2_compile()</b> if
1891<b>pcre2_set_offset_limit()</b> is going to be used to set a non-default offset
1892limit in a match context for matches that use this pattern. An error is
1893generated if an offset limit is set without this option. For more details, see
1894the description of <b>pcre2_set_offset_limit()</b> in the
1895<a href="#matchcontext">section</a>
1896that describes match contexts. See also the PCRE2_FIRSTLINE
1897option above.
1898<pre>
1899  PCRE2_UTF
1900</pre>
1901This option causes PCRE2 to regard both the pattern and the subject strings
1902that are subsequently processed as strings of UTF characters instead of
1903single-code-unit strings. It is available when PCRE2 is built to include
1904Unicode support (which is the default). If Unicode support is not available,
1905the use of this option provokes an error. Details of how PCRE2_UTF changes the
1906behaviour of PCRE2 are given in the
1907<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
1908page. In particular, note that it changes the way PCRE2_CASELESS handles
1909characters with code points greater than 127.
1910<a name="extracompileoptions"></a></P>
1911<br><b>
1912Extra compile options
1913</b><br>
1914<P>
1915The option bits that can be set in a compile context by calling the
1916<b>pcre2_set_compile_extra_options()</b> function are as follows:
1917<pre>
1918  PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK
1919</pre>
1920Since release 10.38 PCRE2 has forbidden the use of \K within lookaround
1921assertions, following Perl's lead. This option is provided to re-enable the
1922previous behaviour (act in positive lookarounds, ignore in negative ones) in
1923case anybody is relying on it.
1924<pre>
1925  PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES
1926</pre>
1927This option applies when compiling a pattern in UTF-8 or UTF-32 mode. It is
1928forbidden in UTF-16 mode, and ignored in non-UTF modes. Unicode "surrogate"
1929code points in the range 0xd800 to 0xdfff are used in pairs in UTF-16 to encode
1930code points with values in the range 0x10000 to 0x10ffff. The surrogates cannot
1931therefore be represented in UTF-16. They can be represented in UTF-8 and
1932UTF-32, but are defined as invalid code points, and cause errors if encountered
1933in a UTF-8 or UTF-32 string that is being checked for validity by PCRE2.
1934</P>
1935<P>
1936These values also cause errors if encountered in escape sequences such as
1937\x{d912} within a pattern. However, it seems that some applications, when
1938using PCRE2 to check for unwanted characters in UTF-8 strings, explicitly test
1939for the surrogates using escape sequences. The PCRE2_NO_UTF_CHECK option does
1940not disable the error that occurs, because it applies only to the testing of
1941input strings for UTF validity.
1942</P>
1943<P>
1944If the extra option PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES is set, surrogate code
1945point values in UTF-8 and UTF-32 patterns no longer provoke errors and are
1946incorporated in the compiled pattern. However, they can only match subject
1947characters if the matching function is called with PCRE2_NO_UTF_CHECK set.
1948<pre>
1949  PCRE2_EXTRA_ALT_BSUX
1950</pre>
1951The original option PCRE2_ALT_BSUX causes PCRE2 to process \U, \u, and \x in
1952the way that ECMAscript (aka JavaScript) does. Additional functionality was
1953defined by ECMAscript 6; setting PCRE2_EXTRA_ALT_BSUX has the effect of
1954PCRE2_ALT_BSUX, but in addition it recognizes \u{hhh..} as a hexadecimal
1955character code, where hhh.. is any number of hexadecimal digits.
1956<pre>
1957  PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL
1958</pre>
1959This is a dangerous option. Use with care. By default, an unrecognized escape
1960such as \j or a malformed one such as \x{2z} causes a compile-time error when
1961detected by <b>pcre2_compile()</b>. Perl is somewhat inconsistent in handling
1962such items: for example, \j is treated as a literal "j", and non-hexadecimal
1963digits in \x{} are just ignored, though warnings are given in both cases if
1964Perl's warning switch is enabled. However, a malformed octal number after \o{
1965always causes an error in Perl.
1966</P>
1967<P>
1968If the PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL extra option is passed to
1969<b>pcre2_compile()</b>, all unrecognized or malformed escape sequences are
1970treated as single-character escapes. For example, \j is a literal "j" and
1971\x{2z} is treated as the literal string "x{2z}". Setting this option means
1972that typos in patterns may go undetected and have unexpected results. Also note
1973that a sequence such as [\N{] is interpreted as a malformed attempt at
1974[\N{...}] and so is treated as [N{] whereas [\N] gives an error because an
1975unqualified \N is a valid escape sequence but is not supported in a character
1976class. To reiterate: this is a dangerous option. Use with great care.
1977<pre>
1978  PCRE2_EXTRA_ESCAPED_CR_IS_LF
1979</pre>
1980There are some legacy applications where the escape sequence \r in a pattern
1981is expected to match a newline. If this option is set, \r in a pattern is
1982converted to \n so that it matches a LF (linefeed) instead of a CR (carriage
1983return) character. The option does not affect a literal CR in the pattern, nor
1984does it affect CR specified as an explicit code point such as \x{0D}.
1985<pre>
1986  PCRE2_EXTRA_MATCH_LINE
1987</pre>
1988This option is provided for use by the <b>-x</b> option of <b>pcre2grep</b>. It
1989causes the pattern only to match complete lines. This is achieved by
1990automatically inserting the code for "^(?:" at the start of the compiled
1991pattern and ")$" at the end. Thus, when PCRE2_MULTILINE is set, the matched
1992line may be in the middle of the subject string. This option can be used with
1993PCRE2_LITERAL.
1994<pre>
1995  PCRE2_EXTRA_MATCH_WORD
1996</pre>
1997This option is provided for use by the <b>-w</b> option of <b>pcre2grep</b>. It
1998causes the pattern only to match strings that have a word boundary at the start
1999and the end. This is achieved by automatically inserting the code for "\b(?:"
2000at the start of the compiled pattern and ")\b" at the end. The option may be
2001used with PCRE2_LITERAL. However, it is ignored if PCRE2_EXTRA_MATCH_LINE is
2002also set.
2003<a name="jitcompiling"></a></P>
2004<br><a name="SEC21" href="#TOC1">JUST-IN-TIME (JIT) COMPILATION</a><br>
2005<P>
2006<b>int pcre2_jit_compile(pcre2_code *<i>code</i>, uint32_t <i>options</i>);</b>
2007<br>
2008<br>
2009<b>int pcre2_jit_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
2010<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
2011<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
2012<b>  pcre2_match_context *<i>mcontext</i>);</b>
2013<br>
2014<br>
2015<b>void pcre2_jit_free_unused_memory(pcre2_general_context *<i>gcontext</i>);</b>
2016<br>
2017<br>
2018<b>pcre2_jit_stack *pcre2_jit_stack_create(PCRE2_SIZE <i>startsize</i>,</b>
2019<b>  PCRE2_SIZE <i>maxsize</i>, pcre2_general_context *<i>gcontext</i>);</b>
2020<br>
2021<br>
2022<b>void pcre2_jit_stack_assign(pcre2_match_context *<i>mcontext</i>,</b>
2023<b>  pcre2_jit_callback <i>callback_function</i>, void *<i>callback_data</i>);</b>
2024<br>
2025<br>
2026<b>void pcre2_jit_stack_free(pcre2_jit_stack *<i>jit_stack</i>);</b>
2027</P>
2028<P>
2029These functions provide support for JIT compilation, which, if the just-in-time
2030compiler is available, further processes a compiled pattern into machine code
2031that executes much faster than the <b>pcre2_match()</b> interpretive matching
2032function. Full details are given in the
2033<a href="pcre2jit.html"><b>pcre2jit</b></a>
2034documentation.
2035</P>
2036<P>
2037JIT compilation is a heavyweight optimization. It can take some time for
2038patterns to be analyzed, and for one-off matches and simple patterns the
2039benefit of faster execution might be offset by a much slower compilation time.
2040Most (but not all) patterns can be optimized by the JIT compiler.
2041<a name="localesupport"></a></P>
2042<br><a name="SEC22" href="#TOC1">LOCALE SUPPORT</a><br>
2043<P>
2044<b>const uint8_t *pcre2_maketables(pcre2_general_context *<i>gcontext</i>);</b>
2045<br>
2046<br>
2047<b>void pcre2_maketables_free(pcre2_general_context *<i>gcontext</i>,</b>
2048<b>  const uint8_t *<i>tables</i>);</b>
2049</P>
2050<P>
2051PCRE2 handles caseless matching, and determines whether characters are letters,
2052digits, or whatever, by reference to a set of tables, indexed by character code
2053point. However, this applies only to characters whose code points are less than
2054256. By default, higher-valued code points never match escapes such as \w or
2055\d.
2056</P>
2057<P>
2058When PCRE2 is built with Unicode support (the default), certain Unicode
2059character properties can be tested with \p and \P, or, alternatively, the
2060PCRE2_UCP option can be set when a pattern is compiled; this causes \w and
2061friends to use Unicode property support instead of the built-in tables.
2062PCRE2_UCP also causes upper/lower casing operations on characters with code
2063points greater than 127 to use Unicode properties. These effects apply even
2064when PCRE2_UTF is not set.
2065</P>
2066<P>
2067The use of locales with Unicode is discouraged. If you are handling characters
2068with code points greater than 127, you should either use Unicode support, or
2069use locales, but not try to mix the two.
2070</P>
2071<P>
2072PCRE2 contains a built-in set of character tables that are used by default.
2073These are sufficient for many applications. Normally, the internal tables
2074recognize only ASCII characters. However, when PCRE2 is built, it is possible
2075to cause the internal tables to be rebuilt in the default "C" locale of the
2076local system, which may cause them to be different.
2077</P>
2078<P>
2079The built-in tables can be overridden by tables supplied by the application
2080that calls PCRE2. These may be created in a different locale from the default.
2081As more and more applications change to using Unicode, the need for this locale
2082support is expected to die away.
2083</P>
2084<P>
2085External tables are built by calling the <b>pcre2_maketables()</b> function, in
2086the relevant locale. The only argument to this function is a general context,
2087which can be used to pass a custom memory allocator. If the argument is NULL,
2088the system <b>malloc()</b> is used. The result can be passed to
2089<b>pcre2_compile()</b> as often as necessary, by creating a compile context and
2090calling <b>pcre2_set_character_tables()</b> to set the tables pointer therein.
2091</P>
2092<P>
2093For example, to build and use tables that are appropriate for the French locale
2094(where accented characters with values greater than 127 are treated as
2095letters), the following code could be used:
2096<pre>
2097  setlocale(LC_CTYPE, "fr_FR");
2098  tables = pcre2_maketables(NULL);
2099  ccontext = pcre2_compile_context_create(NULL);
2100  pcre2_set_character_tables(ccontext, tables);
2101  re = pcre2_compile(..., ccontext);
2102</pre>
2103The locale name "fr_FR" is used on Linux and other Unix-like systems; if you
2104are using Windows, the name for the French locale is "french".
2105</P>
2106<P>
2107The pointer that is passed (via the compile context) to <b>pcre2_compile()</b>
2108is saved with the compiled pattern, and the same tables are used by the
2109matching functions. Thus, for any single pattern, compilation and matching both
2110happen in the same locale, but different patterns can be processed in different
2111locales.
2112</P>
2113<P>
2114It is the caller's responsibility to ensure that the memory containing the
2115tables remains available while they are still in use. When they are no longer
2116needed, you can discard them using <b>pcre2_maketables_free()</b>, which should
2117pass as its first parameter the same global context that was used to create the
2118tables.
2119</P>
2120<br><b>
2121Saving locale tables
2122</b><br>
2123<P>
2124The tables described above are just a sequence of binary bytes, which makes
2125them independent of hardware characteristics such as endianness or whether the
2126processor is 32-bit or 64-bit. A copy of the result of <b>pcre2_maketables()</b>
2127can therefore be saved in a file or elsewhere and re-used later, even in a
2128different program or on another computer. The size of the tables (number of
2129bytes) must be obtained by calling <b>pcre2_config()</b> with the
2130PCRE2_CONFIG_TABLES_LENGTH option because <b>pcre2_maketables()</b> does not
2131return this value. Note that the <b>pcre2_dftables</b> program, which is part of
2132the PCRE2 build system, can be used stand-alone to create a file that contains
2133a set of binary tables. See the
2134<a href="pcre2build.html#createtables"><b>pcre2build</b></a>
2135documentation for details.
2136<a name="infoaboutpattern"></a></P>
2137<br><a name="SEC23" href="#TOC1">INFORMATION ABOUT A COMPILED PATTERN</a><br>
2138<P>
2139<b>int pcre2_pattern_info(const pcre2 *<i>code</i>, uint32_t <i>what</i>, void *<i>where</i>);</b>
2140</P>
2141<P>
2142The <b>pcre2_pattern_info()</b> function returns general information about a
2143compiled pattern. For information about callouts, see the
2144<a href="#infoaboutcallouts">next section.</a>
2145The first argument for <b>pcre2_pattern_info()</b> is a pointer to the compiled
2146pattern. The second argument specifies which piece of information is required,
2147and the third argument is a pointer to a variable to receive the data. If the
2148third argument is NULL, the first argument is ignored, and the function returns
2149the size in bytes of the variable that is required for the information
2150requested. Otherwise, the yield of the function is zero for success, or one of
2151the following negative numbers:
2152<pre>
2153  PCRE2_ERROR_NULL           the argument <i>code</i> was NULL
2154  PCRE2_ERROR_BADMAGIC       the "magic number" was not found
2155  PCRE2_ERROR_BADOPTION      the value of <i>what</i> was invalid
2156  PCRE2_ERROR_UNSET          the requested field is not set
2157</pre>
2158The "magic number" is placed at the start of each compiled pattern as a simple
2159check against passing an arbitrary memory pointer. Here is a typical call of
2160<b>pcre2_pattern_info()</b>, to obtain the length of the compiled pattern:
2161<pre>
2162  int rc;
2163  size_t length;
2164  rc = pcre2_pattern_info(
2165    re,               /* result of pcre2_compile() */
2166    PCRE2_INFO_SIZE,  /* what is required */
2167    &length);         /* where to put the data */
2168</pre>
2169The possible values for the second argument are defined in <b>pcre2.h</b>, and
2170are as follows:
2171<pre>
2172  PCRE2_INFO_ALLOPTIONS
2173  PCRE2_INFO_ARGOPTIONS
2174  PCRE2_INFO_EXTRAOPTIONS
2175</pre>
2176Return copies of the pattern's options. The third argument should point to a
2177<b>uint32_t</b> variable. PCRE2_INFO_ARGOPTIONS returns exactly the options that
2178were passed to <b>pcre2_compile()</b>, whereas PCRE2_INFO_ALLOPTIONS returns
2179the compile options as modified by any top-level (*XXX) option settings such as
2180(*UTF) at the start of the pattern itself. PCRE2_INFO_EXTRAOPTIONS returns the
2181extra options that were set in the compile context by calling the
2182pcre2_set_compile_extra_options() function.
2183</P>
2184<P>
2185For example, if the pattern /(*UTF)abc/ is compiled with the PCRE2_EXTENDED
2186option, the result for PCRE2_INFO_ALLOPTIONS is PCRE2_EXTENDED and PCRE2_UTF.
2187Option settings such as (?i) that can change within a pattern do not affect the
2188result of PCRE2_INFO_ALLOPTIONS, even if they appear right at the start of the
2189pattern. (This was different in some earlier releases.)
2190</P>
2191<P>
2192A pattern compiled without PCRE2_ANCHORED is automatically anchored by PCRE2 if
2193the first significant item in every top-level branch is one of the following:
2194<pre>
2195  ^     unless PCRE2_MULTILINE is set
2196  \A    always
2197  \G    always
2198  .*    sometimes - see below
2199</pre>
2200When .* is the first significant item, anchoring is possible only when all the
2201following are true:
2202<pre>
2203  .* is not in an atomic group
2204  .* is not in a capture group that is the subject of a backreference
2205  PCRE2_DOTALL is in force for .*
2206  Neither (*PRUNE) nor (*SKIP) appears in the pattern
2207  PCRE2_NO_DOTSTAR_ANCHOR is not set
2208</pre>
2209For patterns that are auto-anchored, the PCRE2_ANCHORED bit is set in the
2210options returned for PCRE2_INFO_ALLOPTIONS.
2211<pre>
2212  PCRE2_INFO_BACKREFMAX
2213</pre>
2214Return the number of the highest backreference in the pattern. The third
2215argument should point to a <b>uint32_t</b> variable. Named capture groups
2216acquire numbers as well as names, and these count towards the highest
2217backreference. Backreferences such as \4 or \g{12} match the captured
2218characters of the given group, but in addition, the check that a capture
2219group is set in a conditional group such as (?(3)a|b) is also a backreference.
2220Zero is returned if there are no backreferences.
2221<pre>
2222  PCRE2_INFO_BSR
2223</pre>
2224The output is a uint32_t integer whose value indicates what character sequences
2225the \R escape sequence matches. A value of PCRE2_BSR_UNICODE means that \R
2226matches any Unicode line ending sequence; a value of PCRE2_BSR_ANYCRLF means
2227that \R matches only CR, LF, or CRLF.
2228<pre>
2229  PCRE2_INFO_CAPTURECOUNT
2230</pre>
2231Return the highest capture group number in the pattern. In patterns where (?|
2232is not used, this is also the total number of capture groups. The third
2233argument should point to a <b>uint32_t</b> variable.
2234<pre>
2235  PCRE2_INFO_DEPTHLIMIT
2236</pre>
2237If the pattern set a backtracking depth limit by including an item of the form
2238(*LIMIT_DEPTH=nnnn) at the start, the value is returned. The third argument
2239should point to a uint32_t integer. If no such value has been set, the call to
2240<b>pcre2_pattern_info()</b> returns the error PCRE2_ERROR_UNSET. Note that this
2241limit will only be used during matching if it is less than the limit set or
2242defaulted by the caller of the match function.
2243<pre>
2244  PCRE2_INFO_FIRSTBITMAP
2245</pre>
2246In the absence of a single first code unit for a non-anchored pattern,
2247<b>pcre2_compile()</b> may construct a 256-bit table that defines a fixed set of
2248values for the first code unit in any match. For example, a pattern that starts
2249with [abc] results in a table with three bits set. When code unit values
2250greater than 255 are supported, the flag bit for 255 means "any code unit of
2251value 255 or above". If such a table was constructed, a pointer to it is
2252returned. Otherwise NULL is returned. The third argument should point to a
2253<b>const uint8_t *</b> variable.
2254<pre>
2255  PCRE2_INFO_FIRSTCODETYPE
2256</pre>
2257Return information about the first code unit of any matched string, for a
2258non-anchored pattern. The third argument should point to a <b>uint32_t</b>
2259variable. If there is a fixed first value, for example, the letter "c" from a
2260pattern such as (cat|cow|coyote), 1 is returned, and the value can be retrieved
2261using PCRE2_INFO_FIRSTCODEUNIT. If there is no fixed first value, but it is
2262known that a match can occur only at the start of the subject or following a
2263newline in the subject, 2 is returned. Otherwise, and for anchored patterns, 0
2264is returned.
2265<pre>
2266  PCRE2_INFO_FIRSTCODEUNIT
2267</pre>
2268Return the value of the first code unit of any matched string for a pattern
2269where PCRE2_INFO_FIRSTCODETYPE returns 1; otherwise return 0. The third
2270argument should point to a <b>uint32_t</b> variable. In the 8-bit library, the
2271value is always less than 256. In the 16-bit library the value can be up to
22720xffff. In the 32-bit library in UTF-32 mode the value can be up to 0x10ffff,
2273and up to 0xffffffff when not using UTF-32 mode.
2274<pre>
2275  PCRE2_INFO_FRAMESIZE
2276</pre>
2277Return the size (in bytes) of the data frames that are used to remember
2278backtracking positions when the pattern is processed by <b>pcre2_match()</b>
2279without the use of JIT. The third argument should point to a <b>size_t</b>
2280variable. The frame size depends on the number of capturing parentheses in the
2281pattern. Each additional capture group adds two PCRE2_SIZE variables.
2282<pre>
2283  PCRE2_INFO_HASBACKSLASHC
2284</pre>
2285Return 1 if the pattern contains any instances of \C, otherwise 0. The third
2286argument should point to a <b>uint32_t</b> variable.
2287<pre>
2288  PCRE2_INFO_HASCRORLF
2289</pre>
2290Return 1 if the pattern contains any explicit matches for CR or LF characters,
2291otherwise 0. The third argument should point to a <b>uint32_t</b> variable. An
2292explicit match is either a literal CR or LF character, or \r or \n or one of
2293the equivalent hexadecimal or octal escape sequences.
2294<pre>
2295  PCRE2_INFO_HEAPLIMIT
2296</pre>
2297If the pattern set a heap memory limit by including an item of the form
2298(*LIMIT_HEAP=nnnn) at the start, the value is returned. The third argument
2299should point to a uint32_t integer. If no such value has been set, the call to
2300<b>pcre2_pattern_info()</b> returns the error PCRE2_ERROR_UNSET. Note that this
2301limit will only be used during matching if it is less than the limit set or
2302defaulted by the caller of the match function.
2303<pre>
2304  PCRE2_INFO_JCHANGED
2305</pre>
2306Return 1 if the (?J) or (?-J) option setting is used in the pattern, otherwise
23070. The third argument should point to a <b>uint32_t</b> variable. (?J) and
2308(?-J) set and unset the local PCRE2_DUPNAMES option, respectively.
2309<pre>
2310  PCRE2_INFO_JITSIZE
2311</pre>
2312If the compiled pattern was successfully processed by
2313<b>pcre2_jit_compile()</b>, return the size of the JIT compiled code, otherwise
2314return zero. The third argument should point to a <b>size_t</b> variable.
2315<pre>
2316  PCRE2_INFO_LASTCODETYPE
2317</pre>
2318Returns 1 if there is a rightmost literal code unit that must exist in any
2319matched string, other than at its start. The third argument should point to a
2320<b>uint32_t</b> variable. If there is no such value, 0 is returned. When 1 is
2321returned, the code unit value itself can be retrieved using
2322PCRE2_INFO_LASTCODEUNIT. For anchored patterns, a last literal value is
2323recorded only if it follows something of variable length. For example, for the
2324pattern /^a\d+z\d+/ the returned value is 1 (with "z" returned from
2325PCRE2_INFO_LASTCODEUNIT), but for /^a\dz\d/ the returned value is 0.
2326<pre>
2327  PCRE2_INFO_LASTCODEUNIT
2328</pre>
2329Return the value of the rightmost literal code unit that must exist in any
2330matched string, other than at its start, for a pattern where
2331PCRE2_INFO_LASTCODETYPE returns 1. Otherwise, return 0. The third argument
2332should point to a <b>uint32_t</b> variable.
2333<pre>
2334  PCRE2_INFO_MATCHEMPTY
2335</pre>
2336Return 1 if the pattern might match an empty string, otherwise 0. The third
2337argument should point to a <b>uint32_t</b> variable. When a pattern contains
2338recursive subroutine calls it is not always possible to determine whether or
2339not it can match an empty string. PCRE2 takes a cautious approach and returns 1
2340in such cases.
2341<pre>
2342  PCRE2_INFO_MATCHLIMIT
2343</pre>
2344If the pattern set a match limit by including an item of the form
2345(*LIMIT_MATCH=nnnn) at the start, the value is returned. The third argument
2346should point to a uint32_t integer. If no such value has been set, the call to
2347<b>pcre2_pattern_info()</b> returns the error PCRE2_ERROR_UNSET. Note that this
2348limit will only be used during matching if it is less than the limit set or
2349defaulted by the caller of the match function.
2350<pre>
2351  PCRE2_INFO_MAXLOOKBEHIND
2352</pre>
2353A lookbehind assertion moves back a certain number of characters (not code
2354units) when it starts to process each of its branches. This request returns the
2355largest of these backward moves. The third argument should point to a uint32_t
2356integer. The simple assertions \b and \B require a one-character lookbehind
2357and cause PCRE2_INFO_MAXLOOKBEHIND to return 1 in the absence of anything
2358longer. \A also registers a one-character lookbehind, though it does not
2359actually inspect the previous character.
2360</P>
2361<P>
2362Note that this information is useful for multi-segment matching only
2363if the pattern contains no nested lookbehinds. For example, the pattern
2364(?&#60;=a(?&#60;=ba)c) returns a maximum lookbehind of 2, but when it is processed, the
2365first lookbehind moves back by two characters, matches one character, then the
2366nested lookbehind also moves back by two characters. This puts the matching
2367point three characters earlier than it was at the start.
2368PCRE2_INFO_MAXLOOKBEHIND is really only useful as a debugging tool. See the
2369<a href="pcre2partial.html"><b>pcre2partial</b></a>
2370documentation for a discussion of multi-segment matching.
2371<pre>
2372  PCRE2_INFO_MINLENGTH
2373</pre>
2374If a minimum length for matching subject strings was computed, its value is
2375returned. Otherwise the returned value is 0. This value is not computed when
2376PCRE2_NO_START_OPTIMIZE is set. The value is a number of characters, which in
2377UTF mode may be different from the number of code units. The third argument
2378should point to a <b>uint32_t</b> variable. The value is a lower bound to the
2379length of any matching string. There may not be any strings of that length that
2380do actually match, but every string that does match is at least that long.
2381<pre>
2382  PCRE2_INFO_NAMECOUNT
2383  PCRE2_INFO_NAMEENTRYSIZE
2384  PCRE2_INFO_NAMETABLE
2385</pre>
2386PCRE2 supports the use of named as well as numbered capturing parentheses. The
2387names are just an additional way of identifying the parentheses, which still
2388acquire numbers. Several convenience functions such as
2389<b>pcre2_substring_get_byname()</b> are provided for extracting captured
2390substrings by name. It is also possible to extract the data directly, by first
2391converting the name to a number in order to access the correct pointers in the
2392output vector (described with <b>pcre2_match()</b> below). To do the conversion,
2393you need to use the name-to-number map, which is described by these three
2394values.
2395</P>
2396<P>
2397The map consists of a number of fixed-size entries. PCRE2_INFO_NAMECOUNT gives
2398the number of entries, and PCRE2_INFO_NAMEENTRYSIZE gives the size of each
2399entry in code units; both of these return a <b>uint32_t</b> value. The entry
2400size depends on the length of the longest name.
2401</P>
2402<P>
2403PCRE2_INFO_NAMETABLE returns a pointer to the first entry of the table. This is
2404a PCRE2_SPTR pointer to a block of code units. In the 8-bit library, the first
2405two bytes of each entry are the number of the capturing parenthesis, most
2406significant byte first. In the 16-bit library, the pointer points to 16-bit
2407code units, the first of which contains the parenthesis number. In the 32-bit
2408library, the pointer points to 32-bit code units, the first of which contains
2409the parenthesis number. The rest of the entry is the corresponding name, zero
2410terminated.
2411</P>
2412<P>
2413The names are in alphabetical order. If (?| is used to create multiple capture
2414groups with the same number, as described in the
2415<a href="pcre2pattern.html#dupgroupnumber">section on duplicate group numbers</a>
2416in the
2417<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
2418page, the groups may be given the same name, but there is only one entry in the
2419table. Different names for groups of the same number are not permitted.
2420</P>
2421<P>
2422Duplicate names for capture groups with different numbers are permitted, but
2423only if PCRE2_DUPNAMES is set. They appear in the table in the order in which
2424they were found in the pattern. In the absence of (?| this is the order of
2425increasing number; when (?| is used this is not necessarily the case because
2426later capture groups may have lower numbers.
2427</P>
2428<P>
2429As a simple example of the name/number table, consider the following pattern
2430after compilation by the 8-bit library (assume PCRE2_EXTENDED is set, so white
2431space - including newlines - is ignored):
2432<pre>
2433  (?&#60;date&#62; (?&#60;year&#62;(\d\d)?\d\d) - (?&#60;month&#62;\d\d) - (?&#60;day&#62;\d\d) )
2434</pre>
2435There are four named capture groups, so the table has four entries, and each
2436entry in the table is eight bytes long. The table is as follows, with
2437non-printing bytes shows in hexadecimal, and undefined bytes shown as ??:
2438<pre>
2439  00 01 d  a  t  e  00 ??
2440  00 05 d  a  y  00 ?? ??
2441  00 04 m  o  n  t  h  00
2442  00 02 y  e  a  r  00 ??
2443</pre>
2444When writing code to extract data from named capture groups using the
2445name-to-number map, remember that the length of the entries is likely to be
2446different for each compiled pattern.
2447<pre>
2448  PCRE2_INFO_NEWLINE
2449</pre>
2450The output is one of the following <b>uint32_t</b> values:
2451<pre>
2452  PCRE2_NEWLINE_CR       Carriage return (CR)
2453  PCRE2_NEWLINE_LF       Linefeed (LF)
2454  PCRE2_NEWLINE_CRLF     Carriage return, linefeed (CRLF)
2455  PCRE2_NEWLINE_ANY      Any Unicode line ending
2456  PCRE2_NEWLINE_ANYCRLF  Any of CR, LF, or CRLF
2457  PCRE2_NEWLINE_NUL      The NUL character (binary zero)
2458</pre>
2459This identifies the character sequence that will be recognized as meaning
2460"newline" while matching.
2461<pre>
2462  PCRE2_INFO_SIZE
2463</pre>
2464Return the size of the compiled pattern in bytes (for all three libraries). The
2465third argument should point to a <b>size_t</b> variable. This value includes the
2466size of the general data block that precedes the code units of the compiled
2467pattern itself. The value that is used when <b>pcre2_compile()</b> is getting
2468memory in which to place the compiled pattern may be slightly larger than the
2469value returned by this option, because there are cases where the code that
2470calculates the size has to over-estimate. Processing a pattern with the JIT
2471compiler does not alter the value returned by this option.
2472<a name="infoaboutcallouts"></a></P>
2473<br><a name="SEC24" href="#TOC1">INFORMATION ABOUT A PATTERN'S CALLOUTS</a><br>
2474<P>
2475<b>int pcre2_callout_enumerate(const pcre2_code *<i>code</i>,</b>
2476<b>  int (*<i>callback</i>)(pcre2_callout_enumerate_block *, void *),</b>
2477<b>  void *<i>user_data</i>);</b>
2478<br>
2479<br>
2480A script language that supports the use of string arguments in callouts might
2481like to scan all the callouts in a pattern before running the match. This can
2482be done by calling <b>pcre2_callout_enumerate()</b>. The first argument is a
2483pointer to a compiled pattern, the second points to a callback function, and
2484the third is arbitrary user data. The callback function is called for every
2485callout in the pattern in the order in which they appear. Its first argument is
2486a pointer to a callout enumeration block, and its second argument is the
2487<i>user_data</i> value that was passed to <b>pcre2_callout_enumerate()</b>. The
2488contents of the callout enumeration block are described in the
2489<a href="pcre2callout.html"><b>pcre2callout</b></a>
2490documentation, which also gives further details about callouts.
2491</P>
2492<br><a name="SEC25" href="#TOC1">SERIALIZATION AND PRECOMPILING</a><br>
2493<P>
2494It is possible to save compiled patterns on disc or elsewhere, and reload them
2495later, subject to a number of restrictions. The host on which the patterns are
2496reloaded must be running the same version of PCRE2, with the same code unit
2497width, and must also have the same endianness, pointer width, and PCRE2_SIZE
2498type. Before compiled patterns can be saved, they must be converted to a
2499"serialized" form, which in the case of PCRE2 is really just a bytecode dump.
2500The functions whose names begin with <b>pcre2_serialize_</b> are used for
2501converting to and from the serialized form. They are described in the
2502<a href="pcre2serialize.html"><b>pcre2serialize</b></a>
2503documentation. Note that PCRE2 serialization does not convert compiled patterns
2504to an abstract format like Java or .NET serialization.
2505<a name="matchdatablock"></a></P>
2506<br><a name="SEC26" href="#TOC1">THE MATCH DATA BLOCK</a><br>
2507<P>
2508<b>pcre2_match_data *pcre2_match_data_create(uint32_t <i>ovecsize</i>,</b>
2509<b>  pcre2_general_context *<i>gcontext</i>);</b>
2510<br>
2511<br>
2512<b>pcre2_match_data *pcre2_match_data_create_from_pattern(</b>
2513<b>  const pcre2_code *<i>code</i>, pcre2_general_context *<i>gcontext</i>);</b>
2514<br>
2515<br>
2516<b>void pcre2_match_data_free(pcre2_match_data *<i>match_data</i>);</b>
2517</P>
2518<P>
2519Information about a successful or unsuccessful match is placed in a match
2520data block, which is an opaque structure that is accessed by function calls. In
2521particular, the match data block contains a vector of offsets into the subject
2522string that define the matched parts of the subject. This is known as the
2523<i>ovector</i>.
2524</P>
2525<P>
2526Before calling <b>pcre2_match()</b>, <b>pcre2_dfa_match()</b>, or
2527<b>pcre2_jit_match()</b> you must create a match data block by calling one of
2528the creation functions above. For <b>pcre2_match_data_create()</b>, the first
2529argument is the number of pairs of offsets in the <i>ovector</i>.
2530</P>
2531<P>
2532When using <b>pcre2_match()</b>, one pair of offsets is required to identify the
2533string that matched the whole pattern, with an additional pair for each
2534captured substring. For example, a value of 4 creates enough space to record
2535the matched portion of the subject plus three captured substrings.
2536</P>
2537<P>
2538When using <b>pcre2_dfa_match()</b> there may be multiple matched substrings of
2539different lengths at the same point in the subject. The ovector should be made
2540large enough to hold as many as are expected.
2541</P>
2542<P>
2543A minimum of at least 1 pair is imposed by <b>pcre2_match_data_create()</b>, so
2544it is always possible to return the overall matched string in the case of
2545<b>pcre2_match()</b> or the longest match in the case of
2546<b>pcre2_dfa_match()</b>.
2547</P>
2548<P>
2549The second argument of <b>pcre2_match_data_create()</b> is a pointer to a
2550general context, which can specify custom memory management for obtaining the
2551memory for the match data block. If you are not using custom memory management,
2552pass NULL, which causes <b>malloc()</b> to be used.
2553</P>
2554<P>
2555For <b>pcre2_match_data_create_from_pattern()</b>, the first argument is a
2556pointer to a compiled pattern. The ovector is created to be exactly the right
2557size to hold all the substrings a pattern might capture when matched using
2558<b>pcre2_match()</b>. You should not use this call when matching with
2559<b>pcre2_dfa_match()</b>. The second argument is again a pointer to a general
2560context, but in this case if NULL is passed, the memory is obtained using the
2561same allocator that was used for the compiled pattern (custom or default).
2562</P>
2563<P>
2564A match data block can be used many times, with the same or different compiled
2565patterns. You can extract information from a match data block after a match
2566operation has finished, using functions that are described in the sections on
2567<a href="#matchedstrings">matched strings</a>
2568and
2569<a href="#matchotherdata">other match data</a>
2570below.
2571</P>
2572<P>
2573When a call of <b>pcre2_match()</b> fails, valid data is available in the match
2574block only when the error is PCRE2_ERROR_NOMATCH, PCRE2_ERROR_PARTIAL, or one
2575of the error codes for an invalid UTF string. Exactly what is available depends
2576on the error, and is detailed below.
2577</P>
2578<P>
2579When one of the matching functions is called, pointers to the compiled pattern
2580and the subject string are set in the match data block so that they can be
2581referenced by the extraction functions after a successful match. After running
2582a match, you must not free a compiled pattern or a subject string until after
2583all operations on the match data block (for that match) have taken place,
2584unless, in the case of the subject string, you have used the
2585PCRE2_COPY_MATCHED_SUBJECT option, which is described in the section entitled
2586"Option bits for <b>pcre2_match()</b>"
2587<a href="#matchoptions>">below.</a>
2588</P>
2589<P>
2590When a match data block itself is no longer needed, it should be freed by
2591calling <b>pcre2_match_data_free()</b>. If this function is called with a NULL
2592argument, it returns immediately, without doing anything.
2593</P>
2594<br><a name="SEC27" href="#TOC1">MATCHING A PATTERN: THE TRADITIONAL FUNCTION</a><br>
2595<P>
2596<b>int pcre2_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
2597<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
2598<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
2599<b>  pcre2_match_context *<i>mcontext</i>);</b>
2600</P>
2601<P>
2602The function <b>pcre2_match()</b> is called to match a subject string against a
2603compiled pattern, which is passed in the <i>code</i> argument. You can call
2604<b>pcre2_match()</b> with the same <i>code</i> argument as many times as you
2605like, in order to find multiple matches in the subject string or to match
2606different subject strings with the same pattern.
2607</P>
2608<P>
2609This function is the main matching facility of the library, and it operates in
2610a Perl-like manner. For specialist use there is also an alternative matching
2611function, which is described
2612<a href="#dfamatch">below</a>
2613in the section about the <b>pcre2_dfa_match()</b> function.
2614</P>
2615<P>
2616Here is an example of a simple call to <b>pcre2_match()</b>:
2617<pre>
2618  pcre2_match_data *md = pcre2_match_data_create(4, NULL);
2619  int rc = pcre2_match(
2620    re,             /* result of pcre2_compile() */
2621    "some string",  /* the subject string */
2622    11,             /* the length of the subject string */
2623    0,              /* start at offset 0 in the subject */
2624    0,              /* default options */
2625    md,             /* the match data block */
2626    NULL);          /* a match context; NULL means use defaults */
2627</pre>
2628If the subject string is zero-terminated, the length can be given as
2629PCRE2_ZERO_TERMINATED. A match context must be provided if certain less common
2630matching parameters are to be changed. For details, see the section on
2631<a href="#matchcontext">the match context</a>
2632above.
2633</P>
2634<br><b>
2635The string to be matched by <b>pcre2_match()</b>
2636</b><br>
2637<P>
2638The subject string is passed to <b>pcre2_match()</b> as a pointer in
2639<i>subject</i>, a length in <i>length</i>, and a starting offset in
2640<i>startoffset</i>. The length and offset are in code units, not characters.
2641That is, they are in bytes for the 8-bit library, 16-bit code units for the
264216-bit library, and 32-bit code units for the 32-bit library, whether or not
2643UTF processing is enabled. As a special case, if <i>subject</i> is NULL and
2644<i>length</i> is zero, the subject is assumed to be an empty string. If
2645<i>length</i> is non-zero, an error occurs if <i>subject</i> is NULL.
2646</P>
2647<P>
2648If <i>startoffset</i> is greater than the length of the subject,
2649<b>pcre2_match()</b> returns PCRE2_ERROR_BADOFFSET. When the starting offset is
2650zero, the search for a match starts at the beginning of the subject, and this
2651is by far the most common case. In UTF-8 or UTF-16 mode, the starting offset
2652must point to the start of a character, or to the end of the subject (in UTF-32
2653mode, one code unit equals one character, so all offsets are valid). Like the
2654pattern string, the subject may contain binary zeros.
2655</P>
2656<P>
2657A non-zero starting offset is useful when searching for another match in the
2658same subject by calling <b>pcre2_match()</b> again after a previous success.
2659Setting <i>startoffset</i> differs from passing over a shortened string and
2660setting PCRE2_NOTBOL in the case of a pattern that begins with any kind of
2661lookbehind. For example, consider the pattern
2662<pre>
2663  \Biss\B
2664</pre>
2665which finds occurrences of "iss" in the middle of words. (\B matches only if
2666the current position in the subject is not a word boundary.) When applied to
2667the string "Mississippi" the first call to <b>pcre2_match()</b> finds the first
2668occurrence. If <b>pcre2_match()</b> is called again with just the remainder of
2669the subject, namely "issippi", it does not match, because \B is always false
2670at the start of the subject, which is deemed to be a word boundary. However, if
2671<b>pcre2_match()</b> is passed the entire string again, but with
2672<i>startoffset</i> set to 4, it finds the second occurrence of "iss" because it
2673is able to look behind the starting point to discover that it is preceded by a
2674letter.
2675</P>
2676<P>
2677Finding all the matches in a subject is tricky when the pattern can match an
2678empty string. It is possible to emulate Perl's /g behaviour by first trying the
2679match again at the same offset, with the PCRE2_NOTEMPTY_ATSTART and
2680PCRE2_ANCHORED options, and then if that fails, advancing the starting offset
2681and trying an ordinary match again. There is some code that demonstrates how to
2682do this in the
2683<a href="pcre2demo.html"><b>pcre2demo</b></a>
2684sample program. In the most general case, you have to check to see if the
2685newline convention recognizes CRLF as a newline, and if so, and the current
2686character is CR followed by LF, advance the starting offset by two characters
2687instead of one.
2688</P>
2689<P>
2690If a non-zero starting offset is passed when the pattern is anchored, a single
2691attempt to match at the given offset is made. This can only succeed if the
2692pattern does not require the match to be at the start of the subject. In other
2693words, the anchoring must be the result of setting the PCRE2_ANCHORED option or
2694the use of .* with PCRE2_DOTALL, not by starting the pattern with ^ or \A.
2695<a name="matchoptions"></a></P>
2696<br><b>
2697Option bits for <b>pcre2_match()</b>
2698</b><br>
2699<P>
2700The unused bits of the <i>options</i> argument for <b>pcre2_match()</b> must be
2701zero. The only bits that may be set are PCRE2_ANCHORED,
2702PCRE2_COPY_MATCHED_SUBJECT, PCRE2_ENDANCHORED, PCRE2_NOTBOL, PCRE2_NOTEOL,
2703PCRE2_NOTEMPTY, PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_JIT, PCRE2_NO_UTF_CHECK,
2704PCRE2_PARTIAL_HARD, and PCRE2_PARTIAL_SOFT. Their action is described below.
2705</P>
2706<P>
2707Setting PCRE2_ANCHORED or PCRE2_ENDANCHORED at match time is not supported by
2708the just-in-time (JIT) compiler. If it is set, JIT matching is disabled and the
2709interpretive code in <b>pcre2_match()</b> is run. Apart from PCRE2_NO_JIT
2710(obviously), the remaining options are supported for JIT matching.
2711<pre>
2712  PCRE2_ANCHORED
2713</pre>
2714The PCRE2_ANCHORED option limits <b>pcre2_match()</b> to matching at the first
2715matching position. If a pattern was compiled with PCRE2_ANCHORED, or turned out
2716to be anchored by virtue of its contents, it cannot be made unachored at
2717matching time. Note that setting the option at match time disables JIT
2718matching.
2719<pre>
2720  PCRE2_COPY_MATCHED_SUBJECT
2721</pre>
2722By default, a pointer to the subject is remembered in the match data block so
2723that, after a successful match, it can be referenced by the substring
2724extraction functions. This means that the subject's memory must not be freed
2725until all such operations are complete. For some applications where the
2726lifetime of the subject string is not guaranteed, it may be necessary to make a
2727copy of the subject string, but it is wasteful to do this unless the match is
2728successful. After a successful match, if PCRE2_COPY_MATCHED_SUBJECT is set, the
2729subject is copied and the new pointer is remembered in the match data block
2730instead of the original subject pointer. The memory allocator that was used for
2731the match block itself is used. The copy is automatically freed when
2732<b>pcre2_match_data_free()</b> is called to free the match data block. It is also
2733automatically freed if the match data block is re-used for another match
2734operation.
2735<pre>
2736  PCRE2_ENDANCHORED
2737</pre>
2738If the PCRE2_ENDANCHORED option is set, any string that <b>pcre2_match()</b>
2739matches must be right at the end of the subject string. Note that setting the
2740option at match time disables JIT matching.
2741<pre>
2742  PCRE2_NOTBOL
2743</pre>
2744This option specifies that first character of the subject string is not the
2745beginning of a line, so the circumflex metacharacter should not match before
2746it. Setting this without having set PCRE2_MULTILINE at compile time causes
2747circumflex never to match. This option affects only the behaviour of the
2748circumflex metacharacter. It does not affect \A.
2749<pre>
2750  PCRE2_NOTEOL
2751</pre>
2752This option specifies that the end of the subject string is not the end of a
2753line, so the dollar metacharacter should not match it nor (except in multiline
2754mode) a newline immediately before it. Setting this without having set
2755PCRE2_MULTILINE at compile time causes dollar never to match. This option
2756affects only the behaviour of the dollar metacharacter. It does not affect \Z
2757or \z.
2758<pre>
2759  PCRE2_NOTEMPTY
2760</pre>
2761An empty string is not considered to be a valid match if this option is set. If
2762there are alternatives in the pattern, they are tried. If all the alternatives
2763match the empty string, the entire match fails. For example, if the pattern
2764<pre>
2765  a?b?
2766</pre>
2767is applied to a string not beginning with "a" or "b", it matches an empty
2768string at the start of the subject. With PCRE2_NOTEMPTY set, this match is not
2769valid, so <b>pcre2_match()</b> searches further into the string for occurrences
2770of "a" or "b".
2771<pre>
2772  PCRE2_NOTEMPTY_ATSTART
2773</pre>
2774This is like PCRE2_NOTEMPTY, except that it locks out an empty string match
2775only at the first matching position, that is, at the start of the subject plus
2776the starting offset. An empty string match later in the subject is permitted.
2777If the pattern is anchored, such a match can occur only if the pattern contains
2778\K.
2779<pre>
2780  PCRE2_NO_JIT
2781</pre>
2782By default, if a pattern has been successfully processed by
2783<b>pcre2_jit_compile()</b>, JIT is automatically used when <b>pcre2_match()</b>
2784is called with options that JIT supports. Setting PCRE2_NO_JIT disables the use
2785of JIT; it forces matching to be done by the interpreter.
2786<pre>
2787  PCRE2_NO_UTF_CHECK
2788</pre>
2789When PCRE2_UTF is set at compile time, the validity of the subject as a UTF
2790string is checked unless PCRE2_NO_UTF_CHECK is passed to <b>pcre2_match()</b> or
2791PCRE2_MATCH_INVALID_UTF was passed to <b>pcre2_compile()</b>. The latter special
2792case is discussed in detail in the
2793<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
2794documentation.
2795</P>
2796<P>
2797In the default case, if a non-zero starting offset is given, the check is
2798applied only to that part of the subject that could be inspected during
2799matching, and there is a check that the starting offset points to the first
2800code unit of a character or to the end of the subject. If there are no
2801lookbehind assertions in the pattern, the check starts at the starting offset.
2802Otherwise, it starts at the length of the longest lookbehind before the
2803starting offset, or at the start of the subject if there are not that many
2804characters before the starting offset. Note that the sequences \b and \B are
2805one-character lookbehinds.
2806</P>
2807<P>
2808The check is carried out before any other processing takes place, and a
2809negative error code is returned if the check fails. There are several UTF error
2810codes for each code unit width, corresponding to different problems with the
2811code unit sequence. There are discussions about the validity of
2812<a href="pcre2unicode.html#utf8strings">UTF-8 strings,</a>
2813<a href="pcre2unicode.html#utf16strings">UTF-16 strings,</a>
2814and
2815<a href="pcre2unicode.html#utf32strings">UTF-32 strings</a>
2816in the
2817<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
2818documentation.
2819</P>
2820<P>
2821If you know that your subject is valid, and you want to skip this check for
2822performance reasons, you can set the PCRE2_NO_UTF_CHECK option when calling
2823<b>pcre2_match()</b>. You might want to do this for the second and subsequent
2824calls to <b>pcre2_match()</b> if you are making repeated calls to find multiple
2825matches in the same subject string.
2826</P>
2827<P>
2828<b>Warning:</b> Unless PCRE2_MATCH_INVALID_UTF was set at compile time, when
2829PCRE2_NO_UTF_CHECK is set at match time the effect of passing an invalid
2830string as a subject, or an invalid value of <i>startoffset</i>, is undefined.
2831Your program may crash or loop indefinitely or give wrong results.
2832<pre>
2833  PCRE2_PARTIAL_HARD
2834  PCRE2_PARTIAL_SOFT
2835</pre>
2836These options turn on the partial matching feature. A partial match occurs if
2837the end of the subject string is reached successfully, but there are not enough
2838subject characters to complete the match. In addition, either at least one
2839character must have been inspected or the pattern must contain a lookbehind, or
2840the pattern must be one that could match an empty string.
2841</P>
2842<P>
2843If this situation arises when PCRE2_PARTIAL_SOFT (but not PCRE2_PARTIAL_HARD)
2844is set, matching continues by testing any remaining alternatives. Only if no
2845complete match can be found is PCRE2_ERROR_PARTIAL returned instead of
2846PCRE2_ERROR_NOMATCH. In other words, PCRE2_PARTIAL_SOFT specifies that the
2847caller is prepared to handle a partial match, but only if no complete match can
2848be found.
2849</P>
2850<P>
2851If PCRE2_PARTIAL_HARD is set, it overrides PCRE2_PARTIAL_SOFT. In this case, if
2852a partial match is found, <b>pcre2_match()</b> immediately returns
2853PCRE2_ERROR_PARTIAL, without considering any other alternatives. In other
2854words, when PCRE2_PARTIAL_HARD is set, a partial match is considered to be more
2855important that an alternative complete match.
2856</P>
2857<P>
2858There is a more detailed discussion of partial and multi-segment matching, with
2859examples, in the
2860<a href="pcre2partial.html"><b>pcre2partial</b></a>
2861documentation.
2862</P>
2863<br><a name="SEC28" href="#TOC1">NEWLINE HANDLING WHEN MATCHING</a><br>
2864<P>
2865When PCRE2 is built, a default newline convention is set; this is usually the
2866standard convention for the operating system. The default can be overridden in
2867a
2868<a href="#compilecontext">compile context</a>
2869by calling <b>pcre2_set_newline()</b>. It can also be overridden by starting a
2870pattern string with, for example, (*CRLF), as described in the
2871<a href="pcre2pattern.html#newlines">section on newline conventions</a>
2872in the
2873<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
2874page. During matching, the newline choice affects the behaviour of the dot,
2875circumflex, and dollar metacharacters. It may also alter the way the match
2876starting position is advanced after a match failure for an unanchored pattern.
2877</P>
2878<P>
2879When PCRE2_NEWLINE_CRLF, PCRE2_NEWLINE_ANYCRLF, or PCRE2_NEWLINE_ANY is set as
2880the newline convention, and a match attempt for an unanchored pattern fails
2881when the current starting position is at a CRLF sequence, and the pattern
2882contains no explicit matches for CR or LF characters, the match position is
2883advanced by two characters instead of one, in other words, to after the CRLF.
2884</P>
2885<P>
2886The above rule is a compromise that makes the most common cases work as
2887expected. For example, if the pattern is .+A (and the PCRE2_DOTALL option is
2888not set), it does not match the string "\r\nA" because, after failing at the
2889start, it skips both the CR and the LF before retrying. However, the pattern
2890[\r\n]A does match that string, because it contains an explicit CR or LF
2891reference, and so advances only by one character after the first failure.
2892</P>
2893<P>
2894An explicit match for CR of LF is either a literal appearance of one of those
2895characters in the pattern, or one of the \r or \n or equivalent octal or
2896hexadecimal escape sequences. Implicit matches such as [^X] do not count, nor
2897does \s, even though it includes CR and LF in the characters that it matches.
2898</P>
2899<P>
2900Notwithstanding the above, anomalous effects may still occur when CRLF is a
2901valid newline sequence and explicit \r or \n escapes appear in the pattern.
2902<a name="matchedstrings"></a></P>
2903<br><a name="SEC29" href="#TOC1">HOW PCRE2_MATCH() RETURNS A STRING AND CAPTURED SUBSTRINGS</a><br>
2904<P>
2905<b>uint32_t pcre2_get_ovector_count(pcre2_match_data *<i>match_data</i>);</b>
2906<br>
2907<br>
2908<b>PCRE2_SIZE *pcre2_get_ovector_pointer(pcre2_match_data *<i>match_data</i>);</b>
2909</P>
2910<P>
2911In general, a pattern matches a certain portion of the subject, and in
2912addition, further substrings from the subject may be picked out by
2913parenthesized parts of the pattern. Following the usage in Jeffrey Friedl's
2914book, this is called "capturing" in what follows, and the phrase "capture
2915group" (Perl terminology) is used for a fragment of a pattern that picks out a
2916substring. PCRE2 supports several other kinds of parenthesized group that do
2917not cause substrings to be captured. The <b>pcre2_pattern_info()</b> function
2918can be used to find out how many capture groups there are in a compiled
2919pattern.
2920</P>
2921<P>
2922You can use auxiliary functions for accessing captured substrings
2923<a href="#extractbynumber">by number</a>
2924or
2925<a href="#extractbyname">by name,</a>
2926as described in sections below.
2927</P>
2928<P>
2929Alternatively, you can make direct use of the vector of PCRE2_SIZE values,
2930called the <b>ovector</b>, which contains the offsets of captured strings. It is
2931part of the
2932<a href="#matchdatablock">match data block.</a>
2933The function <b>pcre2_get_ovector_pointer()</b> returns the address of the
2934ovector, and <b>pcre2_get_ovector_count()</b> returns the number of pairs of
2935values it contains.
2936</P>
2937<P>
2938Within the ovector, the first in each pair of values is set to the offset of
2939the first code unit of a substring, and the second is set to the offset of the
2940first code unit after the end of a substring. These values are always code unit
2941offsets, not character offsets. That is, they are byte offsets in the 8-bit
2942library, 16-bit offsets in the 16-bit library, and 32-bit offsets in the 32-bit
2943library.
2944</P>
2945<P>
2946After a partial match (error return PCRE2_ERROR_PARTIAL), only the first pair
2947of offsets (that is, <i>ovector[0]</i> and <i>ovector[1]</i>) are set. They
2948identify the part of the subject that was partially matched. See the
2949<a href="pcre2partial.html"><b>pcre2partial</b></a>
2950documentation for details of partial matching.
2951</P>
2952<P>
2953After a fully successful match, the first pair of offsets identifies the
2954portion of the subject string that was matched by the entire pattern. The next
2955pair is used for the first captured substring, and so on. The value returned by
2956<b>pcre2_match()</b> is one more than the highest numbered pair that has been
2957set. For example, if two substrings have been captured, the returned value is
29583. If there are no captured substrings, the return value from a successful
2959match is 1, indicating that just the first pair of offsets has been set.
2960</P>
2961<P>
2962If a pattern uses the \K escape sequence within a positive assertion, the
2963reported start of a successful match can be greater than the end of the match.
2964For example, if the pattern (?=ab\K) is matched against "ab", the start and
2965end offset values for the match are 2 and 0.
2966</P>
2967<P>
2968If a capture group is matched repeatedly within a single match operation, it is
2969the last portion of the subject that it matched that is returned.
2970</P>
2971<P>
2972If the ovector is too small to hold all the captured substring offsets, as much
2973as possible is filled in, and the function returns a value of zero. If captured
2974substrings are not of interest, <b>pcre2_match()</b> may be called with a match
2975data block whose ovector is of minimum length (that is, one pair).
2976</P>
2977<P>
2978It is possible for capture group number <i>n+1</i> to match some part of the
2979subject when group <i>n</i> has not been used at all. For example, if the string
2980"abc" is matched against the pattern (a|(z))(bc) the return from the function
2981is 4, and groups 1 and 3 are matched, but 2 is not. When this happens, both
2982values in the offset pairs corresponding to unused groups are set to
2983PCRE2_UNSET.
2984</P>
2985<P>
2986Offset values that correspond to unused groups at the end of the expression are
2987also set to PCRE2_UNSET. For example, if the string "abc" is matched against
2988the pattern (abc)(x(yz)?)? groups 2 and 3 are not matched. The return from the
2989function is 2, because the highest used capture group number is 1. The offsets
2990for for the second and third capture groupss (assuming the vector is large
2991enough, of course) are set to PCRE2_UNSET.
2992</P>
2993<P>
2994Elements in the ovector that do not correspond to capturing parentheses in the
2995pattern are never changed. That is, if a pattern contains <i>n</i> capturing
2996parentheses, no more than <i>ovector[0]</i> to <i>ovector[2n+1]</i> are set by
2997<b>pcre2_match()</b>. The other elements retain whatever values they previously
2998had. After a failed match attempt, the contents of the ovector are unchanged.
2999<a name="matchotherdata"></a></P>
3000<br><a name="SEC30" href="#TOC1">OTHER INFORMATION ABOUT A MATCH</a><br>
3001<P>
3002<b>PCRE2_SPTR pcre2_get_mark(pcre2_match_data *<i>match_data</i>);</b>
3003<br>
3004<br>
3005<b>PCRE2_SIZE pcre2_get_startchar(pcre2_match_data *<i>match_data</i>);</b>
3006</P>
3007<P>
3008As well as the offsets in the ovector, other information about a match is
3009retained in the match data block and can be retrieved by the above functions in
3010appropriate circumstances. If they are called at other times, the result is
3011undefined.
3012</P>
3013<P>
3014After a successful match, a partial match (PCRE2_ERROR_PARTIAL), or a failure
3015to match (PCRE2_ERROR_NOMATCH), a mark name may be available. The function
3016<b>pcre2_get_mark()</b> can be called to access this name, which can be
3017specified in the pattern by any of the backtracking control verbs, not just
3018(*MARK). The same function applies to all the verbs. It returns a pointer to
3019the zero-terminated name, which is within the compiled pattern. If no name is
3020available, NULL is returned. The length of the name (excluding the terminating
3021zero) is stored in the code unit that precedes the name. You should use this
3022length instead of relying on the terminating zero if the name might contain a
3023binary zero.
3024</P>
3025<P>
3026After a successful match, the name that is returned is the last mark name
3027encountered on the matching path through the pattern. Instances of backtracking
3028verbs without names do not count. Thus, for example, if the matching path
3029contains (*MARK:A)(*PRUNE), the name "A" is returned. After a "no match" or a
3030partial match, the last encountered name is returned. For example, consider
3031this pattern:
3032<pre>
3033  ^(*MARK:A)((*MARK:B)a|b)c
3034</pre>
3035When it matches "bc", the returned name is A. The B mark is "seen" in the first
3036branch of the group, but it is not on the matching path. On the other hand,
3037when this pattern fails to match "bx", the returned name is B.
3038</P>
3039<P>
3040<b>Warning:</b> By default, certain start-of-match optimizations are used to
3041give a fast "no match" result in some situations. For example, if the anchoring
3042is removed from the pattern above, there is an initial check for the presence
3043of "c" in the subject before running the matching engine. This check fails for
3044"bx", causing a match failure without seeing any marks. You can disable the
3045start-of-match optimizations by setting the PCRE2_NO_START_OPTIMIZE option for
3046<b>pcre2_compile()</b> or by starting the pattern with (*NO_START_OPT).
3047</P>
3048<P>
3049After a successful match, a partial match, or one of the invalid UTF errors
3050(for example, PCRE2_ERROR_UTF8_ERR5), <b>pcre2_get_startchar()</b> can be
3051called. After a successful or partial match it returns the code unit offset of
3052the character at which the match started. For a non-partial match, this can be
3053different to the value of <i>ovector[0]</i> if the pattern contains the \K
3054escape sequence. After a partial match, however, this value is always the same
3055as <i>ovector[0]</i> because \K does not affect the result of a partial match.
3056</P>
3057<P>
3058After a UTF check failure, <b>pcre2_get_startchar()</b> can be used to obtain
3059the code unit offset of the invalid UTF character. Details are given in the
3060<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
3061page.
3062<a name="errorlist"></a></P>
3063<br><a name="SEC31" href="#TOC1">ERROR RETURNS FROM <b>pcre2_match()</b></a><br>
3064<P>
3065If <b>pcre2_match()</b> fails, it returns a negative number. This can be
3066converted to a text string by calling the <b>pcre2_get_error_message()</b>
3067function (see "Obtaining a textual error message"
3068<a href="#geterrormessage">below).</a>
3069Negative error codes are also returned by other functions, and are documented
3070with them. The codes are given names in the header file. If UTF checking is in
3071force and an invalid UTF subject string is detected, one of a number of
3072UTF-specific negative error codes is returned. Details are given in the
3073<a href="pcre2unicode.html"><b>pcre2unicode</b></a>
3074page. The following are the other errors that may be returned by
3075<b>pcre2_match()</b>:
3076<pre>
3077  PCRE2_ERROR_NOMATCH
3078</pre>
3079The subject string did not match the pattern.
3080<pre>
3081  PCRE2_ERROR_PARTIAL
3082</pre>
3083The subject string did not match, but it did match partially. See the
3084<a href="pcre2partial.html"><b>pcre2partial</b></a>
3085documentation for details of partial matching.
3086<pre>
3087  PCRE2_ERROR_BADMAGIC
3088</pre>
3089PCRE2 stores a 4-byte "magic number" at the start of the compiled code, to
3090catch the case when it is passed a junk pointer. This is the error that is
3091returned when the magic number is not present.
3092<pre>
3093  PCRE2_ERROR_BADMODE
3094</pre>
3095This error is given when a compiled pattern is passed to a function in a
3096library of a different code unit width, for example, a pattern compiled by
3097the 8-bit library is passed to a 16-bit or 32-bit library function.
3098<pre>
3099  PCRE2_ERROR_BADOFFSET
3100</pre>
3101The value of <i>startoffset</i> was greater than the length of the subject.
3102<pre>
3103  PCRE2_ERROR_BADOPTION
3104</pre>
3105An unrecognized bit was set in the <i>options</i> argument.
3106<pre>
3107  PCRE2_ERROR_BADUTFOFFSET
3108</pre>
3109The UTF code unit sequence that was passed as a subject was checked and found
3110to be valid (the PCRE2_NO_UTF_CHECK option was not set), but the value of
3111<i>startoffset</i> did not point to the beginning of a UTF character or the end
3112of the subject.
3113<pre>
3114  PCRE2_ERROR_CALLOUT
3115</pre>
3116This error is never generated by <b>pcre2_match()</b> itself. It is provided for
3117use by callout functions that want to cause <b>pcre2_match()</b> or
3118<b>pcre2_callout_enumerate()</b> to return a distinctive error code. See the
3119<a href="pcre2callout.html"><b>pcre2callout</b></a>
3120documentation for details.
3121<pre>
3122  PCRE2_ERROR_DEPTHLIMIT
3123</pre>
3124The nested backtracking depth limit was reached.
3125<pre>
3126  PCRE2_ERROR_HEAPLIMIT
3127</pre>
3128The heap limit was reached.
3129<pre>
3130  PCRE2_ERROR_INTERNAL
3131</pre>
3132An unexpected internal error has occurred. This error could be caused by a bug
3133in PCRE2 or by overwriting of the compiled pattern.
3134<pre>
3135  PCRE2_ERROR_JIT_STACKLIMIT
3136</pre>
3137This error is returned when a pattern that was successfully studied using JIT
3138is being matched, but the memory available for the just-in-time processing
3139stack is not large enough. See the
3140<a href="pcre2jit.html"><b>pcre2jit</b></a>
3141documentation for more details.
3142<pre>
3143  PCRE2_ERROR_MATCHLIMIT
3144</pre>
3145The backtracking match limit was reached.
3146<pre>
3147  PCRE2_ERROR_NOMEMORY
3148</pre>
3149If a pattern contains many nested backtracking points, heap memory is used to
3150remember them. This error is given when the memory allocation function (default
3151or custom) fails. Note that a different error, PCRE2_ERROR_HEAPLIMIT, is given
3152if the amount of memory needed exceeds the heap limit. PCRE2_ERROR_NOMEMORY is
3153also returned if PCRE2_COPY_MATCHED_SUBJECT is set and memory allocation fails.
3154<pre>
3155  PCRE2_ERROR_NULL
3156</pre>
3157Either the <i>code</i>, <i>subject</i>, or <i>match_data</i> argument was passed
3158as NULL.
3159<pre>
3160  PCRE2_ERROR_RECURSELOOP
3161</pre>
3162This error is returned when <b>pcre2_match()</b> detects a recursion loop within
3163the pattern. Specifically, it means that either the whole pattern or a
3164capture group has been called recursively for the second time at the same
3165position in the subject string. Some simple patterns that might do this are
3166detected and faulted at compile time, but more complicated cases, in particular
3167mutual recursions between two different groups, cannot be detected until
3168matching is attempted.
3169<a name="geterrormessage"></a></P>
3170<br><a name="SEC32" href="#TOC1">OBTAINING A TEXTUAL ERROR MESSAGE</a><br>
3171<P>
3172<b>int pcre2_get_error_message(int <i>errorcode</i>, PCRE2_UCHAR *<i>buffer</i>,</b>
3173<b>  PCRE2_SIZE <i>bufflen</i>);</b>
3174</P>
3175<P>
3176A text message for an error code from any PCRE2 function (compile, match, or
3177auxiliary) can be obtained by calling <b>pcre2_get_error_message()</b>. The code
3178is passed as the first argument, with the remaining two arguments specifying a
3179code unit buffer and its length in code units, into which the text message is
3180placed. The message is returned in code units of the appropriate width for the
3181library that is being used.
3182</P>
3183<P>
3184The returned message is terminated with a trailing zero, and the function
3185returns the number of code units used, excluding the trailing zero. If the
3186error number is unknown, the negative error code PCRE2_ERROR_BADDATA is
3187returned. If the buffer is too small, the message is truncated (but still with
3188a trailing zero), and the negative error code PCRE2_ERROR_NOMEMORY is returned.
3189None of the messages are very long; a buffer size of 120 code units is ample.
3190<a name="extractbynumber"></a></P>
3191<br><a name="SEC33" href="#TOC1">EXTRACTING CAPTURED SUBSTRINGS BY NUMBER</a><br>
3192<P>
3193<b>int pcre2_substring_length_bynumber(pcre2_match_data *<i>match_data</i>,</b>
3194<b>  uint32_t <i>number</i>, PCRE2_SIZE *<i>length</i>);</b>
3195<br>
3196<br>
3197<b>int pcre2_substring_copy_bynumber(pcre2_match_data *<i>match_data</i>,</b>
3198<b>  uint32_t <i>number</i>, PCRE2_UCHAR *<i>buffer</i>,</b>
3199<b>  PCRE2_SIZE *<i>bufflen</i>);</b>
3200<br>
3201<br>
3202<b>int pcre2_substring_get_bynumber(pcre2_match_data *<i>match_data</i>,</b>
3203<b>  uint32_t <i>number</i>, PCRE2_UCHAR **<i>bufferptr</i>,</b>
3204<b>  PCRE2_SIZE *<i>bufflen</i>);</b>
3205<br>
3206<br>
3207<b>void pcre2_substring_free(PCRE2_UCHAR *<i>buffer</i>);</b>
3208</P>
3209<P>
3210Captured substrings can be accessed directly by using the ovector as described
3211<a href="#matchedstrings">above.</a>
3212For convenience, auxiliary functions are provided for extracting captured
3213substrings as new, separate, zero-terminated strings. A substring that contains
3214a binary zero is correctly extracted and has a further zero added on the end,
3215but the result is not, of course, a C string.
3216</P>
3217<P>
3218The functions in this section identify substrings by number. The number zero
3219refers to the entire matched substring, with higher numbers referring to
3220substrings captured by parenthesized groups. After a partial match, only
3221substring zero is available. An attempt to extract any other substring gives
3222the error PCRE2_ERROR_PARTIAL. The next section describes similar functions for
3223extracting captured substrings by name.
3224</P>
3225<P>
3226If a pattern uses the \K escape sequence within a positive assertion, the
3227reported start of a successful match can be greater than the end of the match.
3228For example, if the pattern (?=ab\K) is matched against "ab", the start and
3229end offset values for the match are 2 and 0. In this situation, calling these
3230functions with a zero substring number extracts a zero-length empty string.
3231</P>
3232<P>
3233You can find the length in code units of a captured substring without
3234extracting it by calling <b>pcre2_substring_length_bynumber()</b>. The first
3235argument is a pointer to the match data block, the second is the group number,
3236and the third is a pointer to a variable into which the length is placed. If
3237you just want to know whether or not the substring has been captured, you can
3238pass the third argument as NULL.
3239</P>
3240<P>
3241The <b>pcre2_substring_copy_bynumber()</b> function copies a captured substring
3242into a supplied buffer, whereas <b>pcre2_substring_get_bynumber()</b> copies it
3243into new memory, obtained using the same memory allocation function that was
3244used for the match data block. The first two arguments of these functions are a
3245pointer to the match data block and a capture group number.
3246</P>
3247<P>
3248The final arguments of <b>pcre2_substring_copy_bynumber()</b> are a pointer to
3249the buffer and a pointer to a variable that contains its length in code units.
3250This is updated to contain the actual number of code units used for the
3251extracted substring, excluding the terminating zero.
3252</P>
3253<P>
3254For <b>pcre2_substring_get_bynumber()</b> the third and fourth arguments point
3255to variables that are updated with a pointer to the new memory and the number
3256of code units that comprise the substring, again excluding the terminating
3257zero. When the substring is no longer needed, the memory should be freed by
3258calling <b>pcre2_substring_free()</b>.
3259</P>
3260<P>
3261The return value from all these functions is zero for success, or a negative
3262error code. If the pattern match failed, the match failure code is returned.
3263If a substring number greater than zero is used after a partial match,
3264PCRE2_ERROR_PARTIAL is returned. Other possible error codes are:
3265<pre>
3266  PCRE2_ERROR_NOMEMORY
3267</pre>
3268The buffer was too small for <b>pcre2_substring_copy_bynumber()</b>, or the
3269attempt to get memory failed for <b>pcre2_substring_get_bynumber()</b>.
3270<pre>
3271  PCRE2_ERROR_NOSUBSTRING
3272</pre>
3273There is no substring with that number in the pattern, that is, the number is
3274greater than the number of capturing parentheses.
3275<pre>
3276  PCRE2_ERROR_UNAVAILABLE
3277</pre>
3278The substring number, though not greater than the number of captures in the
3279pattern, is greater than the number of slots in the ovector, so the substring
3280could not be captured.
3281<pre>
3282  PCRE2_ERROR_UNSET
3283</pre>
3284The substring did not participate in the match. For example, if the pattern is
3285(abc)|(def) and the subject is "def", and the ovector contains at least two
3286capturing slots, substring number 1 is unset.
3287</P>
3288<br><a name="SEC34" href="#TOC1">EXTRACTING A LIST OF ALL CAPTURED SUBSTRINGS</a><br>
3289<P>
3290<b>int pcre2_substring_list_get(pcre2_match_data *<i>match_data</i>,</b>
3291<b>"  PCRE2_UCHAR ***<i>listptr</i>, PCRE2_SIZE **<i>lengthsptr</i>);</b>
3292<br>
3293<br>
3294<b>void pcre2_substring_list_free(PCRE2_SPTR *<i>list</i>);</b>
3295</P>
3296<P>
3297The <b>pcre2_substring_list_get()</b> function extracts all available substrings
3298and builds a list of pointers to them. It also (optionally) builds a second
3299list that contains their lengths (in code units), excluding a terminating zero
3300that is added to each of them. All this is done in a single block of memory
3301that is obtained using the same memory allocation function that was used to get
3302the match data block.
3303</P>
3304<P>
3305This function must be called only after a successful match. If called after a
3306partial match, the error code PCRE2_ERROR_PARTIAL is returned.
3307</P>
3308<P>
3309The address of the memory block is returned via <i>listptr</i>, which is also
3310the start of the list of string pointers. The end of the list is marked by a
3311NULL pointer. The address of the list of lengths is returned via
3312<i>lengthsptr</i>. If your strings do not contain binary zeros and you do not
3313therefore need the lengths, you may supply NULL as the <b>lengthsptr</b>
3314argument to disable the creation of a list of lengths. The yield of the
3315function is zero if all went well, or PCRE2_ERROR_NOMEMORY if the memory block
3316could not be obtained. When the list is no longer needed, it should be freed by
3317calling <b>pcre2_substring_list_free()</b>.
3318</P>
3319<P>
3320If this function encounters a substring that is unset, which can happen when
3321capture group number <i>n+1</i> matches some part of the subject, but group
3322<i>n</i> has not been used at all, it returns an empty string. This can be
3323distinguished from a genuine zero-length substring by inspecting the
3324appropriate offset in the ovector, which contain PCRE2_UNSET for unset
3325substrings, or by calling <b>pcre2_substring_length_bynumber()</b>.
3326<a name="extractbyname"></a></P>
3327<br><a name="SEC35" href="#TOC1">EXTRACTING CAPTURED SUBSTRINGS BY NAME</a><br>
3328<P>
3329<b>int pcre2_substring_number_from_name(const pcre2_code *<i>code</i>,</b>
3330<b>  PCRE2_SPTR <i>name</i>);</b>
3331<br>
3332<br>
3333<b>int pcre2_substring_length_byname(pcre2_match_data *<i>match_data</i>,</b>
3334<b>  PCRE2_SPTR <i>name</i>, PCRE2_SIZE *<i>length</i>);</b>
3335<br>
3336<br>
3337<b>int pcre2_substring_copy_byname(pcre2_match_data *<i>match_data</i>,</b>
3338<b>  PCRE2_SPTR <i>name</i>, PCRE2_UCHAR *<i>buffer</i>, PCRE2_SIZE *<i>bufflen</i>);</b>
3339<br>
3340<br>
3341<b>int pcre2_substring_get_byname(pcre2_match_data *<i>match_data</i>,</b>
3342<b>  PCRE2_SPTR <i>name</i>, PCRE2_UCHAR **<i>bufferptr</i>, PCRE2_SIZE *<i>bufflen</i>);</b>
3343<br>
3344<br>
3345<b>void pcre2_substring_free(PCRE2_UCHAR *<i>buffer</i>);</b>
3346</P>
3347<P>
3348To extract a substring by name, you first have to find associated number.
3349For example, for this pattern:
3350<pre>
3351  (a+)b(?&#60;xxx&#62;\d+)...
3352</pre>
3353the number of the capture group called "xxx" is 2. If the name is known to be
3354unique (PCRE2_DUPNAMES was not set), you can find the number from the name by
3355calling <b>pcre2_substring_number_from_name()</b>. The first argument is the
3356compiled pattern, and the second is the name. The yield of the function is the
3357group number, PCRE2_ERROR_NOSUBSTRING if there is no group with that name, or
3358PCRE2_ERROR_NOUNIQUESUBSTRING if there is more than one group with that name.
3359Given the number, you can extract the substring directly from the ovector, or
3360use one of the "bynumber" functions described above.
3361</P>
3362<P>
3363For convenience, there are also "byname" functions that correspond to the
3364"bynumber" functions, the only difference being that the second argument is a
3365name instead of a number. If PCRE2_DUPNAMES is set and there are duplicate
3366names, these functions scan all the groups with the given name, and return the
3367captured substring from the first named group that is set.
3368</P>
3369<P>
3370If there are no groups with the given name, PCRE2_ERROR_NOSUBSTRING is
3371returned. If all groups with the name have numbers that are greater than the
3372number of slots in the ovector, PCRE2_ERROR_UNAVAILABLE is returned. If there
3373is at least one group with a slot in the ovector, but no group is found to be
3374set, PCRE2_ERROR_UNSET is returned.
3375</P>
3376<P>
3377<b>Warning:</b> If the pattern uses the (?| feature to set up multiple
3378capture groups with the same number, as described in the
3379<a href="pcre2pattern.html#dupgroupnumber">section on duplicate group numbers</a>
3380in the
3381<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
3382page, you cannot use names to distinguish the different capture groups, because
3383names are not included in the compiled code. The matching process uses only
3384numbers. For this reason, the use of different names for groups with the
3385same number causes an error at compile time.
3386<a name="substitutions"></a></P>
3387<br><a name="SEC36" href="#TOC1">CREATING A NEW STRING WITH SUBSTITUTIONS</a><br>
3388<P>
3389<b>int pcre2_substitute(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
3390<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
3391<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
3392<b>  pcre2_match_context *<i>mcontext</i>, PCRE2_SPTR <i>replacement</i>,</b>
3393<b>  PCRE2_SIZE <i>rlength</i>, PCRE2_UCHAR *<i>outputbuffer</i>,</b>
3394<b>  PCRE2_SIZE *<i>outlengthptr</i>);</b>
3395</P>
3396<P>
3397This function optionally calls <b>pcre2_match()</b> and then makes a copy of the
3398subject string in <i>outputbuffer</i>, replacing parts that were matched with
3399the <i>replacement</i> string, whose length is supplied in <b>rlength</b>, which
3400can be given as PCRE2_ZERO_TERMINATED for a zero-terminated string. As a
3401special case, if <i>replacement</i> is NULL and <i>rlength</i> is zero, the
3402replacement is assumed to be an empty string. If <i>rlength</i> is non-zero, an
3403error occurs if <i>replacement</i> is NULL.
3404</P>
3405<P>
3406There is an option (see PCRE2_SUBSTITUTE_REPLACEMENT_ONLY below) to return just
3407the replacement string(s). The default action is to perform just one
3408replacement if the pattern matches, but there is an option that requests
3409multiple replacements (see PCRE2_SUBSTITUTE_GLOBAL below).
3410</P>
3411<P>
3412If successful, <b>pcre2_substitute()</b> returns the number of substitutions
3413that were carried out. This may be zero if no match was found, and is never
3414greater than one unless PCRE2_SUBSTITUTE_GLOBAL is set. A negative value is
3415returned if an error is detected.
3416</P>
3417<P>
3418Matches in which a \K item in a lookahead in the pattern causes the match to
3419end before it starts are not supported, and give rise to an error return. For
3420global replacements, matches in which \K in a lookbehind causes the match to
3421start earlier than the point that was reached in the previous iteration are
3422also not supported.
3423</P>
3424<P>
3425The first seven arguments of <b>pcre2_substitute()</b> are the same as for
3426<b>pcre2_match()</b>, except that the partial matching options are not
3427permitted, and <i>match_data</i> may be passed as NULL, in which case a match
3428data block is obtained and freed within this function, using memory management
3429functions from the match context, if provided, or else those that were used to
3430allocate memory for the compiled code.
3431</P>
3432<P>
3433If <i>match_data</i> is not NULL and PCRE2_SUBSTITUTE_MATCHED is not set, the
3434provided block is used for all calls to <b>pcre2_match()</b>, and its contents
3435afterwards are the result of the final call. For global changes, this will
3436always be a no-match error. The contents of the ovector within the match data
3437block may or may not have been changed.
3438</P>
3439<P>
3440As well as the usual options for <b>pcre2_match()</b>, a number of additional
3441options can be set in the <i>options</i> argument of <b>pcre2_substitute()</b>.
3442One such option is PCRE2_SUBSTITUTE_MATCHED. When this is set, an external
3443<i>match_data</i> block must be provided, and it must have already been used for
3444an external call to <b>pcre2_match()</b> with the same pattern and subject
3445arguments. The data in the <i>match_data</i> block (return code, offset vector)
3446is then used for the first substitution instead of calling <b>pcre2_match()</b>
3447from within <b>pcre2_substitute()</b>. This allows an application to check for a
3448match before choosing to substitute, without having to repeat the match.
3449</P>
3450<P>
3451The contents of the externally supplied match data block are not changed when
3452PCRE2_SUBSTITUTE_MATCHED is set. If PCRE2_SUBSTITUTE_GLOBAL is also set,
3453<b>pcre2_match()</b> is called after the first substitution to check for further
3454matches, but this is done using an internally obtained match data block, thus
3455always leaving the external block unchanged.
3456</P>
3457<P>
3458The <i>code</i> argument is not used for matching before the first substitution
3459when PCRE2_SUBSTITUTE_MATCHED is set, but it must be provided, even when
3460PCRE2_SUBSTITUTE_GLOBAL is not set, because it contains information such as the
3461UTF setting and the number of capturing parentheses in the pattern.
3462</P>
3463<P>
3464The default action of <b>pcre2_substitute()</b> is to return a copy of the
3465subject string with matched substrings replaced. However, if
3466PCRE2_SUBSTITUTE_REPLACEMENT_ONLY is set, only the replacement substrings are
3467returned. In the global case, multiple replacements are concatenated in the
3468output buffer. Substitution callouts (see
3469<a href="#subcallouts">below)</a>
3470can be used to separate them if necessary.
3471</P>
3472<P>
3473The <i>outlengthptr</i> argument of <b>pcre2_substitute()</b> must point to a
3474variable that contains the length, in code units, of the output buffer. If the
3475function is successful, the value is updated to contain the length in code
3476units of the new string, excluding the trailing zero that is automatically
3477added.
3478</P>
3479<P>
3480If the function is not successful, the value set via <i>outlengthptr</i> depends
3481on the type of error. For syntax errors in the replacement string, the value is
3482the offset in the replacement string where the error was detected. For other
3483errors, the value is PCRE2_UNSET by default. This includes the case of the
3484output buffer being too small, unless PCRE2_SUBSTITUTE_OVERFLOW_LENGTH is set.
3485</P>
3486<P>
3487PCRE2_SUBSTITUTE_OVERFLOW_LENGTH changes what happens when the output buffer is
3488too small. The default action is to return PCRE2_ERROR_NOMEMORY immediately. If
3489this option is set, however, <b>pcre2_substitute()</b> continues to go through
3490the motions of matching and substituting (without, of course, writing anything)
3491in order to compute the size of buffer that is needed. This value is passed
3492back via the <i>outlengthptr</i> variable, with the result of the function still
3493being PCRE2_ERROR_NOMEMORY.
3494</P>
3495<P>
3496Passing a buffer size of zero is a permitted way of finding out how much memory
3497is needed for given substitution. However, this does mean that the entire
3498operation is carried out twice. Depending on the application, it may be more
3499efficient to allocate a large buffer and free the excess afterwards, instead of
3500using PCRE2_SUBSTITUTE_OVERFLOW_LENGTH.
3501</P>
3502<P>
3503The replacement string, which is interpreted as a UTF string in UTF mode, is
3504checked for UTF validity unless PCRE2_NO_UTF_CHECK is set. An invalid UTF
3505replacement string causes an immediate return with the relevant UTF error code.
3506</P>
3507<P>
3508If PCRE2_SUBSTITUTE_LITERAL is set, the replacement string is not interpreted
3509in any way. By default, however, a dollar character is an escape character that
3510can specify the insertion of characters from capture groups and names from
3511(*MARK) or other control verbs in the pattern. The following forms are always
3512recognized:
3513<pre>
3514  $$                  insert a dollar character
3515  $&#60;n&#62; or ${&#60;n&#62;}      insert the contents of group &#60;n&#62;
3516  $*MARK or ${*MARK}  insert a control verb name
3517</pre>
3518Either a group number or a group name can be given for &#60;n&#62;. Curly brackets are
3519required only if the following character would be interpreted as part of the
3520number or name. The number may be zero to include the entire matched string.
3521For example, if the pattern a(b)c is matched with "=abc=" and the replacement
3522string "+$1$0$1+", the result is "=+babcb+=".
3523</P>
3524<P>
3525$*MARK inserts the name from the last encountered backtracking control verb on
3526the matching path that has a name. (*MARK) must always include a name, but the
3527other verbs need not. For example, in the case of (*MARK:A)(*PRUNE) the name
3528inserted is "A", but for (*MARK:A)(*PRUNE:B) the relevant name is "B". This
3529facility can be used to perform simple simultaneous substitutions, as this
3530<b>pcre2test</b> example shows:
3531<pre>
3532  /(*MARK:pear)apple|(*MARK:orange)lemon/g,replace=${*MARK}
3533      apple lemon
3534   2: pear orange
3535</pre>
3536PCRE2_SUBSTITUTE_GLOBAL causes the function to iterate over the subject string,
3537replacing every matching substring. If this option is not set, only the first
3538matching substring is replaced. The search for matches takes place in the
3539original subject string (that is, previous replacements do not affect it).
3540Iteration is implemented by advancing the <i>startoffset</i> value for each
3541search, which is always passed the entire subject string. If an offset limit is
3542set in the match context, searching stops when that limit is reached.
3543</P>
3544<P>
3545You can restrict the effect of a global substitution to a portion of the
3546subject string by setting either or both of <i>startoffset</i> and an offset
3547limit. Here is a <b>pcre2test</b> example:
3548<pre>
3549  /B/g,replace=!,use_offset_limit
3550  ABC ABC ABC ABC\=offset=3,offset_limit=12
3551   2: ABC A!C A!C ABC
3552</pre>
3553When continuing with global substitutions after matching a substring with zero
3554length, an attempt to find a non-empty match at the same offset is performed.
3555If this is not successful, the offset is advanced by one character except when
3556CRLF is a valid newline sequence and the next two characters are CR, LF. In
3557this case, the offset is advanced by two characters.
3558</P>
3559<P>
3560PCRE2_SUBSTITUTE_UNKNOWN_UNSET causes references to capture groups that do
3561not appear in the pattern to be treated as unset groups. This option should be
3562used with care, because it means that a typo in a group name or number no
3563longer causes the PCRE2_ERROR_NOSUBSTRING error.
3564</P>
3565<P>
3566PCRE2_SUBSTITUTE_UNSET_EMPTY causes unset capture groups (including unknown
3567groups when PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set) to be treated as empty
3568strings when inserted as described above. If this option is not set, an attempt
3569to insert an unset group causes the PCRE2_ERROR_UNSET error. This option does
3570not influence the extended substitution syntax described below.
3571</P>
3572<P>
3573PCRE2_SUBSTITUTE_EXTENDED causes extra processing to be applied to the
3574replacement string. Without this option, only the dollar character is special,
3575and only the group insertion forms listed above are valid. When
3576PCRE2_SUBSTITUTE_EXTENDED is set, two things change:
3577</P>
3578<P>
3579Firstly, backslash in a replacement string is interpreted as an escape
3580character. The usual forms such as \n or \x{ddd} can be used to specify
3581particular character codes, and backslash followed by any non-alphanumeric
3582character quotes that character. Extended quoting can be coded using \Q...\E,
3583exactly as in pattern strings.
3584</P>
3585<P>
3586There are also four escape sequences for forcing the case of inserted letters.
3587The insertion mechanism has three states: no case forcing, force upper case,
3588and force lower case. The escape sequences change the current state: \U and
3589\L change to upper or lower case forcing, respectively, and \E (when not
3590terminating a \Q quoted sequence) reverts to no case forcing. The sequences
3591\u and \l force the next character (if it is a letter) to upper or lower
3592case, respectively, and then the state automatically reverts to no case
3593forcing. Case forcing applies to all inserted characters, including those from
3594capture groups and letters within \Q...\E quoted sequences. If either
3595PCRE2_UTF or PCRE2_UCP was set when the pattern was compiled, Unicode
3596properties are used for case forcing characters whose code points are greater
3597than 127.
3598</P>
3599<P>
3600Note that case forcing sequences such as \U...\E do not nest. For example,
3601the result of processing "\Uaa\LBB\Ecc\E" is "AAbbcc"; the final \E has no
3602effect. Note also that the PCRE2_ALT_BSUX and PCRE2_EXTRA_ALT_BSUX options do
3603not apply to replacement strings.
3604</P>
3605<P>
3606The second effect of setting PCRE2_SUBSTITUTE_EXTENDED is to add more
3607flexibility to capture group substitution. The syntax is similar to that used
3608by Bash:
3609<pre>
3610  ${&#60;n&#62;:-&#60;string&#62;}
3611  ${&#60;n&#62;:+&#60;string1&#62;:&#60;string2&#62;}
3612</pre>
3613As before, &#60;n&#62; may be a group number or a name. The first form specifies a
3614default value. If group &#60;n&#62; is set, its value is inserted; if not, &#60;string&#62; is
3615expanded and the result inserted. The second form specifies strings that are
3616expanded and inserted when group &#60;n&#62; is set or unset, respectively. The first
3617form is just a convenient shorthand for
3618<pre>
3619  ${&#60;n&#62;:+${&#60;n&#62;}:&#60;string&#62;}
3620</pre>
3621Backslash can be used to escape colons and closing curly brackets in the
3622replacement strings. A change of the case forcing state within a replacement
3623string remains in force afterwards, as shown in this <b>pcre2test</b> example:
3624<pre>
3625  /(some)?(body)/substitute_extended,replace=${1:+\U:\L}HeLLo
3626      body
3627   1: hello
3628      somebody
3629   1: HELLO
3630</pre>
3631The PCRE2_SUBSTITUTE_UNSET_EMPTY option does not affect these extended
3632substitutions. However, PCRE2_SUBSTITUTE_UNKNOWN_UNSET does cause unknown
3633groups in the extended syntax forms to be treated as unset.
3634</P>
3635<P>
3636If PCRE2_SUBSTITUTE_LITERAL is set, PCRE2_SUBSTITUTE_UNKNOWN_UNSET,
3637PCRE2_SUBSTITUTE_UNSET_EMPTY, and PCRE2_SUBSTITUTE_EXTENDED are irrelevant and
3638are ignored.
3639</P>
3640<br><b>
3641Substitution errors
3642</b><br>
3643<P>
3644In the event of an error, <b>pcre2_substitute()</b> returns a negative error
3645code. Except for PCRE2_ERROR_NOMATCH (which is never returned), errors from
3646<b>pcre2_match()</b> are passed straight back.
3647</P>
3648<P>
3649PCRE2_ERROR_NOSUBSTRING is returned for a non-existent substring insertion,
3650unless PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set.
3651</P>
3652<P>
3653PCRE2_ERROR_UNSET is returned for an unset substring insertion (including an
3654unknown substring when PCRE2_SUBSTITUTE_UNKNOWN_UNSET is set) when the simple
3655(non-extended) syntax is used and PCRE2_SUBSTITUTE_UNSET_EMPTY is not set.
3656</P>
3657<P>
3658PCRE2_ERROR_NOMEMORY is returned if the output buffer is not big enough. If the
3659PCRE2_SUBSTITUTE_OVERFLOW_LENGTH option is set, the size of buffer that is
3660needed is returned via <i>outlengthptr</i>. Note that this does not happen by
3661default.
3662</P>
3663<P>
3664PCRE2_ERROR_NULL is returned if PCRE2_SUBSTITUTE_MATCHED is set but the
3665<i>match_data</i> argument is NULL or if the <i>subject</i> or <i>replacement</i>
3666arguments are NULL. For backward compatibility reasons an exception is made for
3667the <i>replacement</i> argument if the <i>rlength</i> argument is also 0.
3668</P>
3669<P>
3670PCRE2_ERROR_BADREPLACEMENT is used for miscellaneous syntax errors in the
3671replacement string, with more particular errors being PCRE2_ERROR_BADREPESCAPE
3672(invalid escape sequence), PCRE2_ERROR_REPMISSINGBRACE (closing curly bracket
3673not found), PCRE2_ERROR_BADSUBSTITUTION (syntax error in extended group
3674substitution), and PCRE2_ERROR_BADSUBSPATTERN (the pattern match ended before
3675it started or the match started earlier than the current position in the
3676subject, which can happen if \K is used in an assertion).
3677</P>
3678<P>
3679As for all PCRE2 errors, a text message that describes the error can be
3680obtained by calling the <b>pcre2_get_error_message()</b> function (see
3681"Obtaining a textual error message"
3682<a href="#geterrormessage">above).</a>
3683<a name="subcallouts"></a></P>
3684<br><b>
3685Substitution callouts
3686</b><br>
3687<P>
3688<b>int pcre2_set_substitute_callout(pcre2_match_context *<i>mcontext</i>,</b>
3689<b>  int (*<i>callout_function</i>)(pcre2_substitute_callout_block *, void *),</b>
3690<b>  void *<i>callout_data</i>);</b>
3691<br>
3692<br>
3693The <b>pcre2_set_substitution_callout()</b> function can be used to specify a
3694callout function for <b>pcre2_substitute()</b>. This information is passed in
3695a match context. The callout function is called after each substitution has
3696been processed, but it can cause the replacement not to happen. The callout
3697function is not called for simulated substitutions that happen as a result of
3698the PCRE2_SUBSTITUTE_OVERFLOW_LENGTH option.
3699</P>
3700<P>
3701The first argument of the callout function is a pointer to a substitute callout
3702block structure, which contains the following fields, not necessarily in this
3703order:
3704<pre>
3705  uint32_t    <i>version</i>;
3706  uint32_t    <i>subscount</i>;
3707  PCRE2_SPTR  <i>input</i>;
3708  PCRE2_SPTR  <i>output</i>;
3709  PCRE2_SIZE <i>*ovector</i>;
3710  uint32_t    <i>oveccount</i>;
3711  PCRE2_SIZE  <i>output_offsets[2]</i>;
3712</pre>
3713The <i>version</i> field contains the version number of the block format. The
3714current version is 0. The version number will increase in future if more fields
3715are added, but the intention is never to remove any of the existing fields.
3716</P>
3717<P>
3718The <i>subscount</i> field is the number of the current match. It is 1 for the
3719first callout, 2 for the second, and so on. The <i>input</i> and <i>output</i>
3720pointers are copies of the values passed to <b>pcre2_substitute()</b>.
3721</P>
3722<P>
3723The <i>ovector</i> field points to the ovector, which contains the result of the
3724most recent match. The <i>oveccount</i> field contains the number of pairs that
3725are set in the ovector, and is always greater than zero.
3726</P>
3727<P>
3728The <i>output_offsets</i> vector contains the offsets of the replacement in the
3729output string. This has already been processed for dollar and (if requested)
3730backslash substitutions as described above.
3731</P>
3732<P>
3733The second argument of the callout function is the value passed as
3734<i>callout_data</i> when the function was registered. The value returned by the
3735callout function is interpreted as follows:
3736</P>
3737<P>
3738If the value is zero, the replacement is accepted, and, if
3739PCRE2_SUBSTITUTE_GLOBAL is set, processing continues with a search for the next
3740match. If the value is not zero, the current replacement is not accepted. If
3741the value is greater than zero, processing continues when
3742PCRE2_SUBSTITUTE_GLOBAL is set. Otherwise (the value is less than zero or
3743PCRE2_SUBSTITUTE_GLOBAL is not set), the the rest of the input is copied to the
3744output and the call to <b>pcre2_substitute()</b> exits, returning the number of
3745matches so far.
3746</P>
3747<br><a name="SEC37" href="#TOC1">DUPLICATE CAPTURE GROUP NAMES</a><br>
3748<P>
3749<b>int pcre2_substring_nametable_scan(const pcre2_code *<i>code</i>,</b>
3750<b>  PCRE2_SPTR <i>name</i>, PCRE2_SPTR *<i>first</i>, PCRE2_SPTR *<i>last</i>);</b>
3751</P>
3752<P>
3753When a pattern is compiled with the PCRE2_DUPNAMES option, names for capture
3754groups are not required to be unique. Duplicate names are always allowed for
3755groups with the same number, created by using the (?| feature. Indeed, if such
3756groups are named, they are required to use the same names.
3757</P>
3758<P>
3759Normally, patterns that use duplicate names are such that in any one match,
3760only one of each set of identically-named groups participates. An example is
3761shown in the
3762<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
3763documentation.
3764</P>
3765<P>
3766When duplicates are present, <b>pcre2_substring_copy_byname()</b> and
3767<b>pcre2_substring_get_byname()</b> return the first substring corresponding to
3768the given name that is set. Only if none are set is PCRE2_ERROR_UNSET is
3769returned. The <b>pcre2_substring_number_from_name()</b> function returns the
3770error PCRE2_ERROR_NOUNIQUESUBSTRING when there are duplicate names.
3771</P>
3772<P>
3773If you want to get full details of all captured substrings for a given name,
3774you must use the <b>pcre2_substring_nametable_scan()</b> function. The first
3775argument is the compiled pattern, and the second is the name. If the third and
3776fourth arguments are NULL, the function returns a group number for a unique
3777name, or PCRE2_ERROR_NOUNIQUESUBSTRING otherwise.
3778</P>
3779<P>
3780When the third and fourth arguments are not NULL, they must be pointers to
3781variables that are updated by the function. After it has run, they point to the
3782first and last entries in the name-to-number table for the given name, and the
3783function returns the length of each entry in code units. In both cases,
3784PCRE2_ERROR_NOSUBSTRING is returned if there are no entries for the given name.
3785</P>
3786<P>
3787The format of the name table is described
3788<a href="#infoaboutpattern">above</a>
3789in the section entitled <i>Information about a pattern</i>. Given all the
3790relevant entries for the name, you can extract each of their numbers, and hence
3791the captured data.
3792</P>
3793<br><a name="SEC38" href="#TOC1">FINDING ALL POSSIBLE MATCHES AT ONE POSITION</a><br>
3794<P>
3795The traditional matching function uses a similar algorithm to Perl, which stops
3796when it finds the first match at a given point in the subject. If you want to
3797find all possible matches, or the longest possible match at a given position,
3798consider using the alternative matching function (see below) instead. If you
3799cannot use the alternative function, you can kludge it up by making use of the
3800callout facility, which is described in the
3801<a href="pcre2callout.html"><b>pcre2callout</b></a>
3802documentation.
3803</P>
3804<P>
3805What you have to do is to insert a callout right at the end of the pattern.
3806When your callout function is called, extract and save the current matched
3807substring. Then return 1, which forces <b>pcre2_match()</b> to backtrack and try
3808other alternatives. Ultimately, when it runs out of matches,
3809<b>pcre2_match()</b> will yield PCRE2_ERROR_NOMATCH.
3810<a name="dfamatch"></a></P>
3811<br><a name="SEC39" href="#TOC1">MATCHING A PATTERN: THE ALTERNATIVE FUNCTION</a><br>
3812<P>
3813<b>int pcre2_dfa_match(const pcre2_code *<i>code</i>, PCRE2_SPTR <i>subject</i>,</b>
3814<b>  PCRE2_SIZE <i>length</i>, PCRE2_SIZE <i>startoffset</i>,</b>
3815<b>  uint32_t <i>options</i>, pcre2_match_data *<i>match_data</i>,</b>
3816<b>  pcre2_match_context *<i>mcontext</i>,</b>
3817<b>  int *<i>workspace</i>, PCRE2_SIZE <i>wscount</i>);</b>
3818</P>
3819<P>
3820The function <b>pcre2_dfa_match()</b> is called to match a subject string
3821against a compiled pattern, using a matching algorithm that scans the subject
3822string just once (not counting lookaround assertions), and does not backtrack
3823(except when processing lookaround assertions). This has different
3824characteristics to the normal algorithm, and is not compatible with Perl. Some
3825of the features of PCRE2 patterns are not supported. Nevertheless, there are
3826times when this kind of matching can be useful. For a discussion of the two
3827matching algorithms, and a list of features that <b>pcre2_dfa_match()</b> does
3828not support, see the
3829<a href="pcre2matching.html"><b>pcre2matching</b></a>
3830documentation.
3831</P>
3832<P>
3833The arguments for the <b>pcre2_dfa_match()</b> function are the same as for
3834<b>pcre2_match()</b>, plus two extras. The ovector within the match data block
3835is used in a different way, and this is described below. The other common
3836arguments are used in the same way as for <b>pcre2_match()</b>, so their
3837description is not repeated here.
3838</P>
3839<P>
3840The two additional arguments provide workspace for the function. The workspace
3841vector should contain at least 20 elements. It is used for keeping track of
3842multiple paths through the pattern tree. More workspace is needed for patterns
3843and subjects where there are a lot of potential matches.
3844</P>
3845<P>
3846Here is an example of a simple call to <b>pcre2_dfa_match()</b>:
3847<pre>
3848  int wspace[20];
3849  pcre2_match_data *md = pcre2_match_data_create(4, NULL);
3850  int rc = pcre2_dfa_match(
3851    re,             /* result of pcre2_compile() */
3852    "some string",  /* the subject string */
3853    11,             /* the length of the subject string */
3854    0,              /* start at offset 0 in the subject */
3855    0,              /* default options */
3856    md,             /* the match data block */
3857    NULL,           /* a match context; NULL means use defaults */
3858    wspace,         /* working space vector */
3859    20);            /* number of elements (NOT size in bytes) */
3860</PRE>
3861</P>
3862<br><b>
3863Option bits for <b>pcre2_dfa_match()</b>
3864</b><br>
3865<P>
3866The unused bits of the <i>options</i> argument for <b>pcre2_dfa_match()</b> must
3867be zero. The only bits that may be set are PCRE2_ANCHORED,
3868PCRE2_COPY_MATCHED_SUBJECT, PCRE2_ENDANCHORED, PCRE2_NOTBOL, PCRE2_NOTEOL,
3869PCRE2_NOTEMPTY, PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_UTF_CHECK, PCRE2_PARTIAL_HARD,
3870PCRE2_PARTIAL_SOFT, PCRE2_DFA_SHORTEST, and PCRE2_DFA_RESTART. All but the last
3871four of these are exactly the same as for <b>pcre2_match()</b>, so their
3872description is not repeated here.
3873<pre>
3874  PCRE2_PARTIAL_HARD
3875  PCRE2_PARTIAL_SOFT
3876</pre>
3877These have the same general effect as they do for <b>pcre2_match()</b>, but the
3878details are slightly different. When PCRE2_PARTIAL_HARD is set for
3879<b>pcre2_dfa_match()</b>, it returns PCRE2_ERROR_PARTIAL if the end of the
3880subject is reached and there is still at least one matching possibility that
3881requires additional characters. This happens even if some complete matches have
3882already been found. When PCRE2_PARTIAL_SOFT is set, the return code
3883PCRE2_ERROR_NOMATCH is converted into PCRE2_ERROR_PARTIAL if the end of the
3884subject is reached, there have been no complete matches, but there is still at
3885least one matching possibility. The portion of the string that was inspected
3886when the longest partial match was found is set as the first matching string in
3887both cases. There is a more detailed discussion of partial and multi-segment
3888matching, with examples, in the
3889<a href="pcre2partial.html"><b>pcre2partial</b></a>
3890documentation.
3891<pre>
3892  PCRE2_DFA_SHORTEST
3893</pre>
3894Setting the PCRE2_DFA_SHORTEST option causes the matching algorithm to stop as
3895soon as it has found one match. Because of the way the alternative algorithm
3896works, this is necessarily the shortest possible match at the first possible
3897matching point in the subject string.
3898<pre>
3899  PCRE2_DFA_RESTART
3900</pre>
3901When <b>pcre2_dfa_match()</b> returns a partial match, it is possible to call it
3902again, with additional subject characters, and have it continue with the same
3903match. The PCRE2_DFA_RESTART option requests this action; when it is set, the
3904<i>workspace</i> and <i>wscount</i> options must reference the same vector as
3905before because data about the match so far is left in them after a partial
3906match. There is more discussion of this facility in the
3907<a href="pcre2partial.html"><b>pcre2partial</b></a>
3908documentation.
3909</P>
3910<br><b>
3911Successful returns from <b>pcre2_dfa_match()</b>
3912</b><br>
3913<P>
3914When <b>pcre2_dfa_match()</b> succeeds, it may have matched more than one
3915substring in the subject. Note, however, that all the matches from one run of
3916the function start at the same point in the subject. The shorter matches are
3917all initial substrings of the longer matches. For example, if the pattern
3918<pre>
3919  &#60;.*&#62;
3920</pre>
3921is matched against the string
3922<pre>
3923  This is &#60;something&#62; &#60;something else&#62; &#60;something further&#62; no more
3924</pre>
3925the three matched strings are
3926<pre>
3927  &#60;something&#62; &#60;something else&#62; &#60;something further&#62;
3928  &#60;something&#62; &#60;something else&#62;
3929  &#60;something&#62;
3930</pre>
3931On success, the yield of the function is a number greater than zero, which is
3932the number of matched substrings. The offsets of the substrings are returned in
3933the ovector, and can be extracted by number in the same way as for
3934<b>pcre2_match()</b>, but the numbers bear no relation to any capture groups
3935that may exist in the pattern, because DFA matching does not support capturing.
3936</P>
3937<P>
3938Calls to the convenience functions that extract substrings by name
3939return the error PCRE2_ERROR_DFA_UFUNC (unsupported function) if used after a
3940DFA match. The convenience functions that extract substrings by number never
3941return PCRE2_ERROR_NOSUBSTRING.
3942</P>
3943<P>
3944The matched strings are stored in the ovector in reverse order of length; that
3945is, the longest matching string is first. If there were too many matches to fit
3946into the ovector, the yield of the function is zero, and the vector is filled
3947with the longest matches.
3948</P>
3949<P>
3950NOTE: PCRE2's "auto-possessification" optimization usually applies to character
3951repeats at the end of a pattern (as well as internally). For example, the
3952pattern "a\d+" is compiled as if it were "a\d++". For DFA matching, this
3953means that only one possible match is found. If you really do want multiple
3954matches in such cases, either use an ungreedy repeat such as "a\d+?" or set
3955the PCRE2_NO_AUTO_POSSESS option when compiling.
3956</P>
3957<br><b>
3958Error returns from <b>pcre2_dfa_match()</b>
3959</b><br>
3960<P>
3961The <b>pcre2_dfa_match()</b> function returns a negative number when it fails.
3962Many of the errors are the same as for <b>pcre2_match()</b>, as described
3963<a href="#errorlist">above.</a>
3964There are in addition the following errors that are specific to
3965<b>pcre2_dfa_match()</b>:
3966<pre>
3967  PCRE2_ERROR_DFA_UITEM
3968</pre>
3969This return is given if <b>pcre2_dfa_match()</b> encounters an item in the
3970pattern that it does not support, for instance, the use of \C in a UTF mode or
3971a backreference.
3972<pre>
3973  PCRE2_ERROR_DFA_UCOND
3974</pre>
3975This return is given if <b>pcre2_dfa_match()</b> encounters a condition item
3976that uses a backreference for the condition, or a test for recursion in a
3977specific capture group. These are not supported.
3978<pre>
3979  PCRE2_ERROR_DFA_UINVALID_UTF
3980</pre>
3981This return is given if <b>pcre2_dfa_match()</b> is called for a pattern that
3982was compiled with PCRE2_MATCH_INVALID_UTF. This is not supported for DFA
3983matching.
3984<pre>
3985  PCRE2_ERROR_DFA_WSSIZE
3986</pre>
3987This return is given if <b>pcre2_dfa_match()</b> runs out of space in the
3988<i>workspace</i> vector.
3989<pre>
3990  PCRE2_ERROR_DFA_RECURSE
3991</pre>
3992When a recursion or subroutine call is processed, the matching function calls
3993itself recursively, using private memory for the ovector and <i>workspace</i>.
3994This error is given if the internal ovector is not large enough. This should be
3995extremely rare, as a vector of size 1000 is used.
3996<pre>
3997  PCRE2_ERROR_DFA_BADRESTART
3998</pre>
3999When <b>pcre2_dfa_match()</b> is called with the <b>PCRE2_DFA_RESTART</b> option,
4000some plausibility checks are made on the contents of the workspace, which
4001should contain data about the previous partial match. If any of these checks
4002fail, this error is given.
4003</P>
4004<br><a name="SEC40" href="#TOC1">SEE ALSO</a><br>
4005<P>
4006<b>pcre2build</b>(3), <b>pcre2callout</b>(3), <b>pcre2demo(3)</b>,
4007<b>pcre2matching</b>(3), <b>pcre2partial</b>(3), <b>pcre2posix</b>(3),
4008<b>pcre2sample</b>(3), <b>pcre2unicode</b>(3).
4009</P>
4010<br><a name="SEC41" href="#TOC1">AUTHOR</a><br>
4011<P>
4012Philip Hazel
4013<br>
4014Retired from University Computing Service
4015<br>
4016Cambridge, England.
4017<br>
4018</P>
4019<br><a name="SEC42" href="#TOC1">REVISION</a><br>
4020<P>
4021Last updated: 14 December 2021
4022<br>
4023Copyright &copy; 1997-2021 University of Cambridge.
4024<br>
4025<p>
4026Return to the <a href="index.html">PCRE2 index page</a>.
4027</p>
4028