/[pcre]/code/trunk/doc/pcre.txt
ViewVC logotype

Contents of /code/trunk/doc/pcre.txt

Parent Directory Parent Directory | Revision Log Revision Log


Revision 91 - (show annotations) (download)
Sat Feb 24 21:41:34 2007 UTC (7 years, 8 months ago) by nigel
File MIME type: text/plain
File size: 247843 byte(s)
Load pcre-6.7 into code/trunk.

1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. The current implementation of PCRE (release
22 6.x) corresponds approximately with Perl 5.8, including support for
23 UTF-8 encoded strings and Unicode general category properties. However,
24 this support has to be explicitly enabled; it is not the default.
25
26 In addition to the Perl-compatible matching function, PCRE also con-
27 tains an alternative matching function that matches the same compiled
28 patterns in a different way. In certain circumstances, the alternative
29 function has some advantages. For a discussion of the two matching
30 algorithms, see the pcrematching page.
31
32 PCRE is written in C and released as a C library. A number of people
33 have written wrappers and interfaces of various kinds. In particular,
34 Google Inc. have provided a comprehensive C++ wrapper. This is now
35 included as part of the PCRE distribution. The pcrecpp page has details
36 of this interface. Other people's contributions can be found in the
37 Contrib directory at the primary FTP site, which is:
38
39 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
40
41 Details of exactly which Perl regular expression features are and are
42 not supported by PCRE are given in separate documents. See the pcrepat-
43 tern and pcrecompat pages.
44
45 Some features of PCRE can be included, excluded, or changed when the
46 library is built. The pcre_config() function makes it possible for a
47 client to discover which features are available. The features them-
48 selves are described in the pcrebuild page. Documentation about build-
49 ing PCRE for various operating systems can be found in the README file
50 in the source distribution.
51
52 The library contains a number of undocumented internal functions and
53 data tables that are used by more than one of the exported external
54 functions, but which are not intended for use by external callers.
55 Their names all begin with "_pcre_", which hopefully will not provoke
56 any name clashes. In some environments, it is possible to control which
57 external symbols are exported when a shared library is built, and in
58 these cases the undocumented symbols are not exported.
59
60
61 USER DOCUMENTATION
62
63 The user documentation for PCRE comprises a number of different sec-
64 tions. In the "man" format, each of these is a separate "man page". In
65 the HTML format, each is a separate page, linked from the index page.
66 In the plain text format, all the sections are concatenated, for ease
67 of searching. The sections are as follows:
68
69 pcre this document
70 pcreapi details of PCRE's native C API
71 pcrebuild options for building PCRE
72 pcrecallout details of the callout feature
73 pcrecompat discussion of Perl compatibility
74 pcrecpp details of the C++ wrapper
75 pcregrep description of the pcregrep command
76 pcrematching discussion of the two matching algorithms
77 pcrepartial details of the partial matching facility
78 pcrepattern syntax and semantics of supported
79 regular expressions
80 pcreperform discussion of performance issues
81 pcreposix the POSIX-compatible C API
82 pcreprecompile details of saving and re-using precompiled patterns
83 pcresample discussion of the sample program
84 pcrestack discussion of stack usage
85 pcretest description of the pcretest testing command
86
87 In addition, in the "man" and HTML formats, there is a short page for
88 each C library function, listing its arguments and results.
89
90
91 LIMITATIONS
92
93 There are some size limitations in PCRE but it is hoped that they will
94 never in practice be relevant.
95
96 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
97 is compiled with the default internal linkage size of 2. If you want to
98 process regular expressions that are truly enormous, you can compile
99 PCRE with an internal linkage size of 3 or 4 (see the README file in
100 the source distribution and the pcrebuild documentation for details).
101 In these cases the limit is substantially larger. However, the speed
102 of execution will be slower.
103
104 All values in repeating quantifiers must be less than 65536. The maxi-
105 mum compiled length of subpattern with an explicit repeat count is
106 30000 bytes. The maximum number of capturing subpatterns is 65535.
107
108 There is no limit to the number of non-capturing subpatterns, but the
109 maximum depth of nesting of all kinds of parenthesized subpattern,
110 including capturing subpatterns, assertions, and other types of subpat-
111 tern, is 200.
112
113 The maximum length of name for a named subpattern is 32, and the maxi-
114 mum number of named subpatterns is 10000.
115
116 The maximum length of a subject string is the largest positive number
117 that an integer variable can hold. However, when using the traditional
118 matching function, PCRE uses recursion to handle subpatterns and indef-
119 inite repetition. This means that the available stack space may limit
120 the size of a subject string that can be processed by certain patterns.
121 For a discussion of stack issues, see the pcrestack documentation.
122
123
124 UTF-8 AND UNICODE PROPERTY SUPPORT
125
126 From release 3.3, PCRE has had some support for character strings
127 encoded in the UTF-8 format. For release 4.0 this was greatly extended
128 to cover most common requirements, and in release 5.0 additional sup-
129 port for Unicode general category properties was added.
130
131 In order process UTF-8 strings, you must build PCRE to include UTF-8
132 support in the code, and, in addition, you must call pcre_compile()
133 with the PCRE_UTF8 option flag. When you do this, both the pattern and
134 any subject strings that are matched against it are treated as UTF-8
135 strings instead of just strings of bytes.
136
137 If you compile PCRE with UTF-8 support, but do not use it at run time,
138 the library will be a bit bigger, but the additional run time overhead
139 is limited to testing the PCRE_UTF8 flag in several places, so should
140 not be very large.
141
142 If PCRE is built with Unicode character property support (which implies
143 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
144 ported. The available properties that can be tested are limited to the
145 general category properties such as Lu for an upper case letter or Nd
146 for a decimal number, the Unicode script names such as Arabic or Han,
147 and the derived properties Any and L&. A full list is given in the
148 pcrepattern documentation. Only the short names for properties are sup-
149 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
150 ter}, is not supported. Furthermore, in Perl, many properties may
151 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
152 does not support this.
153
154 The following comments apply when PCRE is running in UTF-8 mode:
155
156 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
157 subjects are checked for validity on entry to the relevant functions.
158 If an invalid UTF-8 string is passed, an error return is given. In some
159 situations, you may already know that your strings are valid, and
160 therefore want to skip these checks in order to improve performance. If
161 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
162 PCRE assumes that the pattern or subject it is given (respectively)
163 contains only valid UTF-8 codes. In this case, it does not diagnose an
164 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
165 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
166 crash.
167
168 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
169 two-byte UTF-8 character if the value is greater than 127.
170
171 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
172 characters for values greater than \177.
173
174 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
175 vidual bytes, for example: \x{100}{3}.
176
177 5. The dot metacharacter matches one UTF-8 character instead of a sin-
178 gle byte.
179
180 6. The escape sequence \C can be used to match a single byte in UTF-8
181 mode, but its use can lead to some strange effects. This facility is
182 not available in the alternative matching function, pcre_dfa_exec().
183
184 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
185 test characters of any code value, but the characters that PCRE recog-
186 nizes as digits, spaces, or word characters remain the same set as
187 before, all with values less than 256. This remains true even when PCRE
188 includes Unicode property support, because to do otherwise would slow
189 down PCRE in many common cases. If you really want to test for a wider
190 sense of, say, "digit", you must use Unicode property tests such as
191 \p{Nd}.
192
193 8. Similarly, characters that match the POSIX named character classes
194 are all low-valued characters.
195
196 9. Case-insensitive matching applies only to characters whose values
197 are less than 128, unless PCRE is built with Unicode property support.
198 Even when Unicode property support is available, PCRE still uses its
199 own character tables when checking the case of low-valued characters,
200 so as not to degrade performance. The Unicode property information is
201 used only for characters with higher values. Even when Unicode property
202 support is available, PCRE supports case-insensitive matching only when
203 there is a one-to-one mapping between a letter's cases. There are a
204 small number of many-to-one mappings in Unicode; these are not sup-
205 ported by PCRE.
206
207
208 AUTHOR
209
210 Philip Hazel
211 University Computing Service,
212 Cambridge CB2 3QG, England.
213
214 Putting an actual email address here seems to have been a spam magnet,
215 so I've taken it away. If you want to email me, use my initial and sur-
216 name, separated by a dot, at the domain ucs.cam.ac.uk.
217
218 Last updated: 05 June 2006
219 Copyright (c) 1997-2006 University of Cambridge.
220 ------------------------------------------------------------------------------
221
222
223 PCREBUILD(3) PCREBUILD(3)
224
225
226 NAME
227 PCRE - Perl-compatible regular expressions
228
229
230 PCRE BUILD-TIME OPTIONS
231
232 This document describes the optional features of PCRE that can be
233 selected when the library is compiled. They are all selected, or dese-
234 lected, by providing options to the configure script that is run before
235 the make command. The complete list of options for configure (which
236 includes the standard ones such as the selection of the installation
237 directory) can be obtained by running
238
239 ./configure --help
240
241 The following sections describe certain options whose names begin with
242 --enable or --disable. These settings specify changes to the defaults
243 for the configure command. Because of the way that configure works,
244 --enable and --disable always come in pairs, so the complementary
245 option always exists as well, but as it specifies the default, it is
246 not described.
247
248
249 C++ SUPPORT
250
251 By default, the configure script will search for a C++ compiler and C++
252 header files. If it finds them, it automatically builds the C++ wrapper
253 library for PCRE. You can disable this by adding
254
255 --disable-cpp
256
257 to the configure command.
258
259
260 UTF-8 SUPPORT
261
262 To build PCRE with support for UTF-8 character strings, add
263
264 --enable-utf8
265
266 to the configure command. Of itself, this does not make PCRE treat
267 strings as UTF-8. As well as compiling PCRE with this option, you also
268 have have to set the PCRE_UTF8 option when you call the pcre_compile()
269 function.
270
271
272 UNICODE CHARACTER PROPERTY SUPPORT
273
274 UTF-8 support allows PCRE to process character values greater than 255
275 in the strings that it handles. On its own, however, it does not pro-
276 vide any facilities for accessing the properties of such characters. If
277 you want to be able to use the pattern escapes \P, \p, and \X, which
278 refer to Unicode character properties, you must add
279
280 --enable-unicode-properties
281
282 to the configure command. This implies UTF-8 support, even if you have
283 not explicitly requested it.
284
285 Including Unicode property support adds around 90K of tables to the
286 PCRE library, approximately doubling its size. Only the general cate-
287 gory properties such as Lu and Nd are supported. Details are given in
288 the pcrepattern documentation.
289
290
291 CODE VALUE OF NEWLINE
292
293 By default, PCRE interprets character 10 (linefeed, LF) as indicating
294 the end of a line. This is the normal newline character on Unix-like
295 systems. You can compile PCRE to use character 13 (carriage return, CR)
296 instead, by adding
297
298 --enable-newline-is-cr
299
300 to the configure command. There is also a --enable-newline-is-lf
301 option, which explicitly specifies linefeed as the newline character.
302
303 Alternatively, you can specify that line endings are to be indicated by
304 the two character sequence CRLF. If you want this, add
305
306 --enable-newline-is-crlf
307
308 to the configure command. Whatever line ending convention is selected
309 when PCRE is built can be overridden when the library functions are
310 called. At build time it is conventional to use the standard for your
311 operating system.
312
313
314 BUILDING SHARED AND STATIC LIBRARIES
315
316 The PCRE building process uses libtool to build both shared and static
317 Unix libraries by default. You can suppress one of these by adding one
318 of
319
320 --disable-shared
321 --disable-static
322
323 to the configure command, as required.
324
325
326 POSIX MALLOC USAGE
327
328 When PCRE is called through the POSIX interface (see the pcreposix doc-
329 umentation), additional working storage is required for holding the
330 pointers to capturing substrings, because PCRE requires three integers
331 per substring, whereas the POSIX interface provides only two. If the
332 number of expected substrings is small, the wrapper function uses space
333 on the stack, because this is faster than using malloc() for each call.
334 The default threshold above which the stack is no longer used is 10; it
335 can be changed by adding a setting such as
336
337 --with-posix-malloc-threshold=20
338
339 to the configure command.
340
341
342 HANDLING VERY LARGE PATTERNS
343
344 Within a compiled pattern, offset values are used to point from one
345 part to another (for example, from an opening parenthesis to an alter-
346 nation metacharacter). By default, two-byte values are used for these
347 offsets, leading to a maximum size for a compiled pattern of around
348 64K. This is sufficient to handle all but the most gigantic patterns.
349 Nevertheless, some people do want to process enormous patterns, so it
350 is possible to compile PCRE to use three-byte or four-byte offsets by
351 adding a setting such as
352
353 --with-link-size=3
354
355 to the configure command. The value given must be 2, 3, or 4. Using
356 longer offsets slows down the operation of PCRE because it has to load
357 additional bytes when handling them.
358
359 If you build PCRE with an increased link size, test 2 (and test 5 if
360 you are using UTF-8) will fail. Part of the output of these tests is a
361 representation of the compiled pattern, and this changes with the link
362 size.
363
364
365 AVOIDING EXCESSIVE STACK USAGE
366
367 When matching with the pcre_exec() function, PCRE implements backtrack-
368 ing by making recursive calls to an internal function called match().
369 In environments where the size of the stack is limited, this can se-
370 verely limit PCRE's operation. (The Unix environment does not usually
371 suffer from this problem, but it may sometimes be necessary to increase
372 the maximum stack size. There is a discussion in the pcrestack docu-
373 mentation.) An alternative approach to recursion that uses memory from
374 the heap to remember data, instead of using recursive function calls,
375 has been implemented to work round the problem of limited stack size.
376 If you want to build a version of PCRE that works this way, add
377
378 --disable-stack-for-recursion
379
380 to the configure command. With this configuration, PCRE will use the
381 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
382 ment functions. Separate functions are provided because the usage is
383 very predictable: the block sizes requested are always the same, and
384 the blocks are always freed in reverse order. A calling program might
385 be able to implement optimized functions that perform better than the
386 standard malloc() and free() functions. PCRE runs noticeably more
387 slowly when built in this way. This option affects only the pcre_exec()
388 function; it is not relevant for the the pcre_dfa_exec() function.
389
390
391 LIMITING PCRE RESOURCE USAGE
392
393 Internally, PCRE has a function called match(), which it calls repeat-
394 edly (sometimes recursively) when matching a pattern with the
395 pcre_exec() function. By controlling the maximum number of times this
396 function may be called during a single matching operation, a limit can
397 be placed on the resources used by a single call to pcre_exec(). The
398 limit can be changed at run time, as described in the pcreapi documen-
399 tation. The default is 10 million, but this can be changed by adding a
400 setting such as
401
402 --with-match-limit=500000
403
404 to the configure command. This setting has no effect on the
405 pcre_dfa_exec() matching function.
406
407 In some environments it is desirable to limit the depth of recursive
408 calls of match() more strictly than the total number of calls, in order
409 to restrict the maximum amount of stack (or heap, if --disable-stack-
410 for-recursion is specified) that is used. A second limit controls this;
411 it defaults to the value that is set for --with-match-limit, which
412 imposes no additional constraints. However, you can set a lower limit
413 by adding, for example,
414
415 --with-match-limit-recursion=10000
416
417 to the configure command. This value can also be overridden at run
418 time.
419
420
421 USING EBCDIC CODE
422
423 PCRE assumes by default that it will run in an environment where the
424 character code is ASCII (or Unicode, which is a superset of ASCII).
425 PCRE can, however, be compiled to run in an EBCDIC environment by
426 adding
427
428 --enable-ebcdic
429
430 to the configure command.
431
432 Last updated: 06 June 2006
433 Copyright (c) 1997-2006 University of Cambridge.
434 ------------------------------------------------------------------------------
435
436
437 PCREMATCHING(3) PCREMATCHING(3)
438
439
440 NAME
441 PCRE - Perl-compatible regular expressions
442
443
444 PCRE MATCHING ALGORITHMS
445
446 This document describes the two different algorithms that are available
447 in PCRE for matching a compiled regular expression against a given sub-
448 ject string. The "standard" algorithm is the one provided by the
449 pcre_exec() function. This works in the same was as Perl's matching
450 function, and provides a Perl-compatible matching operation.
451
452 An alternative algorithm is provided by the pcre_dfa_exec() function;
453 this operates in a different way, and is not Perl-compatible. It has
454 advantages and disadvantages compared with the standard algorithm, and
455 these are described below.
456
457 When there is only one possible way in which a given subject string can
458 match a pattern, the two algorithms give the same answer. A difference
459 arises, however, when there are multiple possibilities. For example, if
460 the pattern
461
462 ^<.*>
463
464 is matched against the string
465
466 <something> <something else> <something further>
467
468 there are three possible answers. The standard algorithm finds only one
469 of them, whereas the DFA algorithm finds all three.
470
471
472 REGULAR EXPRESSIONS AS TREES
473
474 The set of strings that are matched by a regular expression can be rep-
475 resented as a tree structure. An unlimited repetition in the pattern
476 makes the tree of infinite size, but it is still a tree. Matching the
477 pattern to a given subject string (from a given starting point) can be
478 thought of as a search of the tree. There are two ways to search a
479 tree: depth-first and breadth-first, and these correspond to the two
480 matching algorithms provided by PCRE.
481
482
483 THE STANDARD MATCHING ALGORITHM
484
485 In the terminology of Jeffrey Friedl's book Mastering Regular Expres-
486 sions, the standard algorithm is an "NFA algorithm". It conducts a
487 depth-first search of the pattern tree. That is, it proceeds along a
488 single path through the tree, checking that the subject matches what is
489 required. When there is a mismatch, the algorithm tries any alterna-
490 tives at the current point, and if they all fail, it backs up to the
491 previous branch point in the tree, and tries the next alternative
492 branch at that level. This often involves backing up (moving to the
493 left) in the subject string as well. The order in which repetition
494 branches are tried is controlled by the greedy or ungreedy nature of
495 the quantifier.
496
497 If a leaf node is reached, a matching string has been found, and at
498 that point the algorithm stops. Thus, if there is more than one possi-
499 ble match, this algorithm returns the first one that it finds. Whether
500 this is the shortest, the longest, or some intermediate length depends
501 on the way the greedy and ungreedy repetition quantifiers are specified
502 in the pattern.
503
504 Because it ends up with a single path through the tree, it is rela-
505 tively straightforward for this algorithm to keep track of the sub-
506 strings that are matched by portions of the pattern in parentheses.
507 This provides support for capturing parentheses and back references.
508
509
510 THE DFA MATCHING ALGORITHM
511
512 DFA stands for "deterministic finite automaton", but you do not need to
513 understand the origins of that name. This algorithm conducts a breadth-
514 first search of the tree. Starting from the first matching point in the
515 subject, it scans the subject string from left to right, once, charac-
516 ter by character, and as it does this, it remembers all the paths
517 through the tree that represent valid matches.
518
519 The scan continues until either the end of the subject is reached, or
520 there are no more unterminated paths. At this point, terminated paths
521 represent the different matching possibilities (if there are none, the
522 match has failed). Thus, if there is more than one possible match,
523 this algorithm finds all of them, and in particular, it finds the long-
524 est. In PCRE, there is an option to stop the algorithm after the first
525 match (which is necessarily the shortest) has been found.
526
527 Note that all the matches that are found start at the same point in the
528 subject. If the pattern
529
530 cat(er(pillar)?)
531
532 is matched against the string "the caterpillar catchment", the result
533 will be the three strings "cat", "cater", and "caterpillar" that start
534 at the fourth character of the subject. The algorithm does not automat-
535 ically move on to find matches that start at later positions.
536
537 There are a number of features of PCRE regular expressions that are not
538 supported by the DFA matching algorithm. They are as follows:
539
540 1. Because the algorithm finds all possible matches, the greedy or
541 ungreedy nature of repetition quantifiers is not relevant. Greedy and
542 ungreedy quantifiers are treated in exactly the same way.
543
544 2. When dealing with multiple paths through the tree simultaneously, it
545 is not straightforward to keep track of captured substrings for the
546 different matching possibilities, and PCRE's implementation of this
547 algorithm does not attempt to do this. This means that no captured sub-
548 strings are available.
549
550 3. Because no substrings are captured, back references within the pat-
551 tern are not supported, and cause errors if encountered.
552
553 4. For the same reason, conditional expressions that use a backrefer-
554 ence as the condition are not supported.
555
556 5. Callouts are supported, but the value of the capture_top field is
557 always 1, and the value of the capture_last field is always -1.
558
559 6. The \C escape sequence, which (in the standard algorithm) matches a
560 single byte, even in UTF-8 mode, is not supported because the DFA algo-
561 rithm moves through the subject string one character at a time, for all
562 active paths through the tree.
563
564
565 ADVANTAGES OF THE DFA ALGORITHM
566
567 Using the DFA matching algorithm provides the following advantages:
568
569 1. All possible matches (at a single point in the subject) are automat-
570 ically found, and in particular, the longest match is found. To find
571 more than one match using the standard algorithm, you have to do kludgy
572 things with callouts.
573
574 2. There is much better support for partial matching. The restrictions
575 on the content of the pattern that apply when using the standard algo-
576 rithm for partial matching do not apply to the DFA algorithm. For non-
577 anchored patterns, the starting position of a partial match is avail-
578 able.
579
580 3. Because the DFA algorithm scans the subject string just once, and
581 never needs to backtrack, it is possible to pass very long subject
582 strings to the matching function in several pieces, checking for par-
583 tial matching each time.
584
585
586 DISADVANTAGES OF THE DFA ALGORITHM
587
588 The DFA algorithm suffers from a number of disadvantages:
589
590 1. It is substantially slower than the standard algorithm. This is
591 partly because it has to search for all possible matches, but is also
592 because it is less susceptible to optimization.
593
594 2. Capturing parentheses and back references are not supported.
595
596 3. The "atomic group" feature of PCRE regular expressions is supported,
597 but does not provide the advantage that it does for the standard algo-
598 rithm.
599
600 Last updated: 06 June 2006
601 Copyright (c) 1997-2006 University of Cambridge.
602 ------------------------------------------------------------------------------
603
604
605 PCREAPI(3) PCREAPI(3)
606
607
608 NAME
609 PCRE - Perl-compatible regular expressions
610
611
612 PCRE NATIVE API
613
614 #include <pcre.h>
615
616 pcre *pcre_compile(const char *pattern, int options,
617 const char **errptr, int *erroffset,
618 const unsigned char *tableptr);
619
620 pcre *pcre_compile2(const char *pattern, int options,
621 int *errorcodeptr,
622 const char **errptr, int *erroffset,
623 const unsigned char *tableptr);
624
625 pcre_extra *pcre_study(const pcre *code, int options,
626 const char **errptr);
627
628 int pcre_exec(const pcre *code, const pcre_extra *extra,
629 const char *subject, int length, int startoffset,
630 int options, int *ovector, int ovecsize);
631
632 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
633 const char *subject, int length, int startoffset,
634 int options, int *ovector, int ovecsize,
635 int *workspace, int wscount);
636
637 int pcre_copy_named_substring(const pcre *code,
638 const char *subject, int *ovector,
639 int stringcount, const char *stringname,
640 char *buffer, int buffersize);
641
642 int pcre_copy_substring(const char *subject, int *ovector,
643 int stringcount, int stringnumber, char *buffer,
644 int buffersize);
645
646 int pcre_get_named_substring(const pcre *code,
647 const char *subject, int *ovector,
648 int stringcount, const char *stringname,
649 const char **stringptr);
650
651 int pcre_get_stringnumber(const pcre *code,
652 const char *name);
653
654 int pcre_get_stringtable_entries(const pcre *code,
655 const char *name, char **first, char **last);
656
657 int pcre_get_substring(const char *subject, int *ovector,
658 int stringcount, int stringnumber,
659 const char **stringptr);
660
661 int pcre_get_substring_list(const char *subject,
662 int *ovector, int stringcount, const char ***listptr);
663
664 void pcre_free_substring(const char *stringptr);
665
666 void pcre_free_substring_list(const char **stringptr);
667
668 const unsigned char *pcre_maketables(void);
669
670 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
671 int what, void *where);
672
673 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
674
675 int pcre_refcount(pcre *code, int adjust);
676
677 int pcre_config(int what, void *where);
678
679 char *pcre_version(void);
680
681 void *(*pcre_malloc)(size_t);
682
683 void (*pcre_free)(void *);
684
685 void *(*pcre_stack_malloc)(size_t);
686
687 void (*pcre_stack_free)(void *);
688
689 int (*pcre_callout)(pcre_callout_block *);
690
691
692 PCRE API OVERVIEW
693
694 PCRE has its own native API, which is described in this document. There
695 is also a set of wrapper functions that correspond to the POSIX regular
696 expression API. These are described in the pcreposix documentation.
697 Both of these APIs define a set of C function calls. A C++ wrapper is
698 distributed with PCRE. It is documented in the pcrecpp page.
699
700 The native API C function prototypes are defined in the header file
701 pcre.h, and on Unix systems the library itself is called libpcre. It
702 can normally be accessed by adding -lpcre to the command for linking an
703 application that uses PCRE. The header file defines the macros
704 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
705 bers for the library. Applications can use these to include support
706 for different releases of PCRE.
707
708 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
709 pcre_exec() are used for compiling and matching regular expressions in
710 a Perl-compatible manner. A sample program that demonstrates the sim-
711 plest way of using them is provided in the file called pcredemo.c in
712 the source distribution. The pcresample documentation describes how to
713 run it.
714
715 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
716 ble, is also provided. This uses a different algorithm for the match-
717 ing. The alternative algorithm finds all possible matches (at a given
718 point in the subject). However, this algorithm does not return captured
719 substrings. A description of the two matching algorithms and their
720 advantages and disadvantages is given in the pcrematching documenta-
721 tion.
722
723 In addition to the main compiling and matching functions, there are
724 convenience functions for extracting captured substrings from a subject
725 string that is matched by pcre_exec(). They are:
726
727 pcre_copy_substring()
728 pcre_copy_named_substring()
729 pcre_get_substring()
730 pcre_get_named_substring()
731 pcre_get_substring_list()
732 pcre_get_stringnumber()
733 pcre_get_stringtable_entries()
734
735 pcre_free_substring() and pcre_free_substring_list() are also provided,
736 to free the memory used for extracted strings.
737
738 The function pcre_maketables() is used to build a set of character
739 tables in the current locale for passing to pcre_compile(),
740 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
741 provided for specialist use. Most commonly, no special tables are
742 passed, in which case internal tables that are generated when PCRE is
743 built are used.
744
745 The function pcre_fullinfo() is used to find out information about a
746 compiled pattern; pcre_info() is an obsolete version that returns only
747 some of the available information, but is retained for backwards com-
748 patibility. The function pcre_version() returns a pointer to a string
749 containing the version of PCRE and its date of release.
750
751 The function pcre_refcount() maintains a reference count in a data
752 block containing a compiled pattern. This is provided for the benefit
753 of object-oriented applications.
754
755 The global variables pcre_malloc and pcre_free initially contain the
756 entry points of the standard malloc() and free() functions, respec-
757 tively. PCRE calls the memory management functions via these variables,
758 so a calling program can replace them if it wishes to intercept the
759 calls. This should be done before calling any PCRE functions.
760
761 The global variables pcre_stack_malloc and pcre_stack_free are also
762 indirections to memory management functions. These special functions
763 are used only when PCRE is compiled to use the heap for remembering
764 data, instead of recursive function calls, when running the pcre_exec()
765 function. See the pcrebuild documentation for details of how to do
766 this. It is a non-standard way of building PCRE, for use in environ-
767 ments that have limited stacks. Because of the greater use of memory
768 management, it runs more slowly. Separate functions are provided so
769 that special-purpose external code can be used for this case. When
770 used, these functions are always called in a stack-like manner (last
771 obtained, first freed), and always for memory blocks of the same size.
772 There is a discussion about PCRE's stack usage in the pcrestack docu-
773 mentation.
774
775 The global variable pcre_callout initially contains NULL. It can be set
776 by the caller to a "callout" function, which PCRE will then call at
777 specified points during a matching operation. Details are given in the
778 pcrecallout documentation.
779
780
781 NEWLINES
782 PCRE supports three different conventions for indicating line breaks in
783 strings: a single CR character, a single LF character, or the two-char-
784 acter sequence CRLF. All three are used as "standard" by different
785 operating systems. When PCRE is built, a default can be specified. The
786 default default is LF, which is the Unix standard. When PCRE is run,
787 the default can be overridden, either when a pattern is compiled, or
788 when it is matched.
789
790 In the PCRE documentation the word "newline" is used to mean "the char-
791 acter or pair of characters that indicate a line break".
792
793
794 MULTITHREADING
795
796 The PCRE functions can be used in multi-threading applications, with
797 the proviso that the memory management functions pointed to by
798 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
799 callout function pointed to by pcre_callout, are shared by all threads.
800
801 The compiled form of a regular expression is not altered during match-
802 ing, so the same compiled pattern can safely be used by several threads
803 at once.
804
805
806 SAVING PRECOMPILED PATTERNS FOR LATER USE
807
808 The compiled form of a regular expression can be saved and re-used at a
809 later time, possibly by a different program, and even on a host other
810 than the one on which it was compiled. Details are given in the
811 pcreprecompile documentation.
812
813
814 CHECKING BUILD-TIME OPTIONS
815
816 int pcre_config(int what, void *where);
817
818 The function pcre_config() makes it possible for a PCRE client to dis-
819 cover which optional features have been compiled into the PCRE library.
820 The pcrebuild documentation has more details about these optional fea-
821 tures.
822
823 The first argument for pcre_config() is an integer, specifying which
824 information is required; the second argument is a pointer to a variable
825 into which the information is placed. The following information is
826 available:
827
828 PCRE_CONFIG_UTF8
829
830 The output is an integer that is set to one if UTF-8 support is avail-
831 able; otherwise it is set to zero.
832
833 PCRE_CONFIG_UNICODE_PROPERTIES
834
835 The output is an integer that is set to one if support for Unicode
836 character properties is available; otherwise it is set to zero.
837
838 PCRE_CONFIG_NEWLINE
839
840 The output is an integer whose value specifies the default character
841 sequence that is recognized as meaning "newline". The three values that
842 are supported are: 10 for LF, 13 for CR, and 3338 for CRLF. The default
843 should normally be the standard sequence for your operating system.
844
845 PCRE_CONFIG_LINK_SIZE
846
847 The output is an integer that contains the number of bytes used for
848 internal linkage in compiled regular expressions. The value is 2, 3, or
849 4. Larger values allow larger regular expressions to be compiled, at
850 the expense of slower matching. The default value of 2 is sufficient
851 for all but the most massive patterns, since it allows the compiled
852 pattern to be up to 64K in size.
853
854 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
855
856 The output is an integer that contains the threshold above which the
857 POSIX interface uses malloc() for output vectors. Further details are
858 given in the pcreposix documentation.
859
860 PCRE_CONFIG_MATCH_LIMIT
861
862 The output is an integer that gives the default limit for the number of
863 internal matching function calls in a pcre_exec() execution. Further
864 details are given with pcre_exec() below.
865
866 PCRE_CONFIG_MATCH_LIMIT_RECURSION
867
868 The output is an integer that gives the default limit for the depth of
869 recursion when calling the internal matching function in a pcre_exec()
870 execution. Further details are given with pcre_exec() below.
871
872 PCRE_CONFIG_STACKRECURSE
873
874 The output is an integer that is set to one if internal recursion when
875 running pcre_exec() is implemented by recursive function calls that use
876 the stack to remember their state. This is the usual way that PCRE is
877 compiled. The output is zero if PCRE was compiled to use blocks of data
878 on the heap instead of recursive function calls. In this case,
879 pcre_stack_malloc and pcre_stack_free are called to manage memory
880 blocks on the heap, thus avoiding the use of the stack.
881
882
883 COMPILING A PATTERN
884
885 pcre *pcre_compile(const char *pattern, int options,
886 const char **errptr, int *erroffset,
887 const unsigned char *tableptr);
888
889 pcre *pcre_compile2(const char *pattern, int options,
890 int *errorcodeptr,
891 const char **errptr, int *erroffset,
892 const unsigned char *tableptr);
893
894 Either of the functions pcre_compile() or pcre_compile2() can be called
895 to compile a pattern into an internal form. The only difference between
896 the two interfaces is that pcre_compile2() has an additional argument,
897 errorcodeptr, via which a numerical error code can be returned.
898
899 The pattern is a C string terminated by a binary zero, and is passed in
900 the pattern argument. A pointer to a single block of memory that is
901 obtained via pcre_malloc is returned. This contains the compiled code
902 and related data. The pcre type is defined for the returned block; this
903 is a typedef for a structure whose contents are not externally defined.
904 It is up to the caller to free the memory (via pcre_free) when it is no
905 longer required.
906
907 Although the compiled code of a PCRE regex is relocatable, that is, it
908 does not depend on memory location, the complete pcre data block is not
909 fully relocatable, because it may contain a copy of the tableptr argu-
910 ment, which is an address (see below).
911
912 The options argument contains independent bits that affect the compila-
913 tion. It should be zero if no options are required. The available
914 options are described below. Some of them, in particular, those that
915 are compatible with Perl, can also be set and unset from within the
916 pattern (see the detailed description in the pcrepattern documenta-
917 tion). For these options, the contents of the options argument speci-
918 fies their initial settings at the start of compilation and execution.
919 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
920 of matching as well as at compile time.
921
922 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
923 if compilation of a pattern fails, pcre_compile() returns NULL, and
924 sets the variable pointed to by errptr to point to a textual error mes-
925 sage. This is a static string that is part of the library. You must not
926 try to free it. The offset from the start of the pattern to the charac-
927 ter where the error was discovered is placed in the variable pointed to
928 by erroffset, which must not be NULL. If it is, an immediate error is
929 given.
930
931 If pcre_compile2() is used instead of pcre_compile(), and the error-
932 codeptr argument is not NULL, a non-zero error code number is returned
933 via this argument in the event of an error. This is in addition to the
934 textual error message. Error codes and messages are listed below.
935
936 If the final argument, tableptr, is NULL, PCRE uses a default set of
937 character tables that are built when PCRE is compiled, using the
938 default C locale. Otherwise, tableptr must be an address that is the
939 result of a call to pcre_maketables(). This value is stored with the
940 compiled pattern, and used again by pcre_exec(), unless another table
941 pointer is passed to it. For more discussion, see the section on locale
942 support below.
943
944 This code fragment shows a typical straightforward call to pcre_com-
945 pile():
946
947 pcre *re;
948 const char *error;
949 int erroffset;
950 re = pcre_compile(
951 "^A.*Z", /* the pattern */
952 0, /* default options */
953 &error, /* for error message */
954 &erroffset, /* for error offset */
955 NULL); /* use default character tables */
956
957 The following names for option bits are defined in the pcre.h header
958 file:
959
960 PCRE_ANCHORED
961
962 If this bit is set, the pattern is forced to be "anchored", that is, it
963 is constrained to match only at the first matching point in the string
964 that is being searched (the "subject string"). This effect can also be
965 achieved by appropriate constructs in the pattern itself, which is the
966 only way to do it in Perl.
967
968 PCRE_AUTO_CALLOUT
969
970 If this bit is set, pcre_compile() automatically inserts callout items,
971 all with number 255, before each pattern item. For discussion of the
972 callout facility, see the pcrecallout documentation.
973
974 PCRE_CASELESS
975
976 If this bit is set, letters in the pattern match both upper and lower
977 case letters. It is equivalent to Perl's /i option, and it can be
978 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
979 always understands the concept of case for characters whose values are
980 less than 128, so caseless matching is always possible. For characters
981 with higher values, the concept of case is supported if PCRE is com-
982 piled with Unicode property support, but not otherwise. If you want to
983 use caseless matching for characters 128 and above, you must ensure
984 that PCRE is compiled with Unicode property support as well as with
985 UTF-8 support.
986
987 PCRE_DOLLAR_ENDONLY
988
989 If this bit is set, a dollar metacharacter in the pattern matches only
990 at the end of the subject string. Without this option, a dollar also
991 matches immediately before a newline at the end of the string (but not
992 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
993 if PCRE_MULTILINE is set. There is no equivalent to this option in
994 Perl, and no way to set it within a pattern.
995
996 PCRE_DOTALL
997
998 If this bit is set, a dot metacharater in the pattern matches all char-
999 acters, including those that indicate newline. Without it, a dot does
1000 not match when the current position is at a newline. This option is
1001 equivalent to Perl's /s option, and it can be changed within a pattern
1002 by a (?s) option setting. A negative class such as [^a] always matches
1003 newlines, independent of the setting of this option.
1004
1005 PCRE_DUPNAMES
1006
1007 If this bit is set, names used to identify capturing subpatterns need
1008 not be unique. This can be helpful for certain types of pattern when it
1009 is known that only one instance of the named subpattern can ever be
1010 matched. There are more details of named subpatterns below; see also
1011 the pcrepattern documentation.
1012
1013 PCRE_EXTENDED
1014
1015 If this bit is set, whitespace data characters in the pattern are
1016 totally ignored except when escaped or inside a character class. White-
1017 space does not include the VT character (code 11). In addition, charac-
1018 ters between an unescaped # outside a character class and the next new-
1019 line, inclusive, are also ignored. This is equivalent to Perl's /x
1020 option, and it can be changed within a pattern by a (?x) option set-
1021 ting.
1022
1023 This option makes it possible to include comments inside complicated
1024 patterns. Note, however, that this applies only to data characters.
1025 Whitespace characters may never appear within special character
1026 sequences in a pattern, for example within the sequence (?( which
1027 introduces a conditional subpattern.
1028
1029 PCRE_EXTRA
1030
1031 This option was invented in order to turn on additional functionality
1032 of PCRE that is incompatible with Perl, but it is currently of very
1033 little use. When set, any backslash in a pattern that is followed by a
1034 letter that has no special meaning causes an error, thus reserving
1035 these combinations for future expansion. By default, as in Perl, a
1036 backslash followed by a letter with no special meaning is treated as a
1037 literal. (Perl can, however, be persuaded to give a warning for this.)
1038 There are at present no other features controlled by this option. It
1039 can also be set by a (?X) option setting within a pattern.
1040
1041 PCRE_FIRSTLINE
1042
1043 If this option is set, an unanchored pattern is required to match
1044 before or at the first newline in the subject string, though the
1045 matched text may continue over the newline.
1046
1047 PCRE_MULTILINE
1048
1049 By default, PCRE treats the subject string as consisting of a single
1050 line of characters (even if it actually contains newlines). The "start
1051 of line" metacharacter (^) matches only at the start of the string,
1052 while the "end of line" metacharacter ($) matches only at the end of
1053 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1054 is set). This is the same as Perl.
1055
1056 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1057 constructs match immediately following or immediately before internal
1058 newlines in the subject string, respectively, as well as at the very
1059 start and end. This is equivalent to Perl's /m option, and it can be
1060 changed within a pattern by a (?m) option setting. If there are no new-
1061 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1062 setting PCRE_MULTILINE has no effect.
1063
1064 PCRE_NEWLINE_CR
1065 PCRE_NEWLINE_LF
1066 PCRE_NEWLINE_CRLF
1067
1068 These options override the default newline definition that was chosen
1069 when PCRE was built. Setting the first or the second specifies that a
1070 newline is indicated by a single character (CR or LF, respectively).
1071 Setting both of them specifies that a newline is indicated by the two-
1072 character CRLF sequence. For convenience, PCRE_NEWLINE_CRLF is defined
1073 to contain both bits. The only time that a line break is relevant when
1074 compiling a pattern is if PCRE_EXTENDED is set, and an unescaped # out-
1075 side a character class is encountered. This indicates a comment that
1076 lasts until after the next newline.
1077
1078 The newline option set at compile time becomes the default that is used
1079 for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1080
1081 PCRE_NO_AUTO_CAPTURE
1082
1083 If this option is set, it disables the use of numbered capturing paren-
1084 theses in the pattern. Any opening parenthesis that is not followed by
1085 ? behaves as if it were followed by ?: but named parentheses can still
1086 be used for capturing (and they acquire numbers in the usual way).
1087 There is no equivalent of this option in Perl.
1088
1089 PCRE_UNGREEDY
1090
1091 This option inverts the "greediness" of the quantifiers so that they
1092 are not greedy by default, but become greedy if followed by "?". It is
1093 not compatible with Perl. It can also be set by a (?U) option setting
1094 within the pattern.
1095
1096 PCRE_UTF8
1097
1098 This option causes PCRE to regard both the pattern and the subject as
1099 strings of UTF-8 characters instead of single-byte character strings.
1100 However, it is available only when PCRE is built to include UTF-8 sup-
1101 port. If not, the use of this option provokes an error. Details of how
1102 this option changes the behaviour of PCRE are given in the section on
1103 UTF-8 support in the main pcre page.
1104
1105 PCRE_NO_UTF8_CHECK
1106
1107 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1108 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1109 pcre_compile() returns an error. If you already know that your pattern
1110 is valid, and you want to skip this check for performance reasons, you
1111 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1112 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1113 your program to crash. Note that this option can also be passed to
1114 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1115 ing of subject strings.
1116
1117
1118 COMPILATION ERROR CODES
1119
1120 The following table lists the error codes than may be returned by
1121 pcre_compile2(), along with the error messages that may be returned by
1122 both compiling functions.
1123
1124 0 no error
1125 1 \ at end of pattern
1126 2 \c at end of pattern
1127 3 unrecognized character follows \
1128 4 numbers out of order in {} quantifier
1129 5 number too big in {} quantifier
1130 6 missing terminating ] for character class
1131 7 invalid escape sequence in character class
1132 8 range out of order in character class
1133 9 nothing to repeat
1134 10 operand of unlimited repeat could match the empty string
1135 11 internal error: unexpected repeat
1136 12 unrecognized character after (?
1137 13 POSIX named classes are supported only within a class
1138 14 missing )
1139 15 reference to non-existent subpattern
1140 16 erroffset passed as NULL
1141 17 unknown option bit(s) set
1142 18 missing ) after comment
1143 19 parentheses nested too deeply
1144 20 regular expression too large
1145 21 failed to get memory
1146 22 unmatched parentheses
1147 23 internal error: code overflow
1148 24 unrecognized character after (?<
1149 25 lookbehind assertion is not fixed length
1150 26 malformed number or name after (?(
1151 27 conditional group contains more than two branches
1152 28 assertion expected after (?(
1153 29 (?R or (?digits must be followed by )
1154 30 unknown POSIX class name
1155 31 POSIX collating elements are not supported
1156 32 this version of PCRE is not compiled with PCRE_UTF8 support
1157 33 spare error
1158 34 character value in \x{...} sequence is too large
1159 35 invalid condition (?(0)
1160 36 \C not allowed in lookbehind assertion
1161 37 PCRE does not support \L, \l, \N, \U, or \u
1162 38 number after (?C is > 255
1163 39 closing ) for (?C expected
1164 40 recursive call could loop indefinitely
1165 41 unrecognized character after (?P
1166 42 syntax error after (?P
1167 43 two named subpatterns have the same name
1168 44 invalid UTF-8 string
1169 45 support for \P, \p, and \X has not been compiled
1170 46 malformed \P or \p sequence
1171 47 unknown property name after \P or \p
1172 48 subpattern name is too long (maximum 32 characters)
1173 49 too many named subpatterns (maximum 10,000)
1174 50 repeated subpattern is too long
1175 51 octal value is greater than \377 (not in UTF-8 mode)
1176
1177
1178 STUDYING A PATTERN
1179
1180 pcre_extra *pcre_study(const pcre *code, int options
1181 const char **errptr);
1182
1183 If a compiled pattern is going to be used several times, it is worth
1184 spending more time analyzing it in order to speed up the time taken for
1185 matching. The function pcre_study() takes a pointer to a compiled pat-
1186 tern as its first argument. If studying the pattern produces additional
1187 information that will help speed up matching, pcre_study() returns a
1188 pointer to a pcre_extra block, in which the study_data field points to
1189 the results of the study.
1190
1191 The returned value from pcre_study() can be passed directly to
1192 pcre_exec(). However, a pcre_extra block also contains other fields
1193 that can be set by the caller before the block is passed; these are
1194 described below in the section on matching a pattern.
1195
1196 If studying the pattern does not produce any additional information
1197 pcre_study() returns NULL. In that circumstance, if the calling program
1198 wants to pass any of the other fields to pcre_exec(), it must set up
1199 its own pcre_extra block.
1200
1201 The second argument of pcre_study() contains option bits. At present,
1202 no options are defined, and this argument should always be zero.
1203
1204 The third argument for pcre_study() is a pointer for an error message.
1205 If studying succeeds (even if no data is returned), the variable it
1206 points to is set to NULL. Otherwise it is set to point to a textual
1207 error message. This is a static string that is part of the library. You
1208 must not try to free it. You should test the error pointer for NULL
1209 after calling pcre_study(), to be sure that it has run successfully.
1210
1211 This is a typical call to pcre_study():
1212
1213 pcre_extra *pe;
1214 pe = pcre_study(
1215 re, /* result of pcre_compile() */
1216 0, /* no options exist */
1217 &error); /* set to NULL or points to a message */
1218
1219 At present, studying a pattern is useful only for non-anchored patterns
1220 that do not have a single fixed starting character. A bitmap of possi-
1221 ble starting bytes is created.
1222
1223
1224 LOCALE SUPPORT
1225
1226 PCRE handles caseless matching, and determines whether characters are
1227 letters digits, or whatever, by reference to a set of tables, indexed
1228 by character value. When running in UTF-8 mode, this applies only to
1229 characters with codes less than 128. Higher-valued codes never match
1230 escapes such as \w or \d, but can be tested with \p if PCRE is built
1231 with Unicode character property support. The use of locales with Uni-
1232 code is discouraged.
1233
1234 An internal set of tables is created in the default C locale when PCRE
1235 is built. This is used when the final argument of pcre_compile() is
1236 NULL, and is sufficient for many applications. An alternative set of
1237 tables can, however, be supplied. These may be created in a different
1238 locale from the default. As more and more applications change to using
1239 Unicode, the need for this locale support is expected to die away.
1240
1241 External tables are built by calling the pcre_maketables() function,
1242 which has no arguments, in the relevant locale. The result can then be
1243 passed to pcre_compile() or pcre_exec() as often as necessary. For
1244 example, to build and use tables that are appropriate for the French
1245 locale (where accented characters with values greater than 128 are
1246 treated as letters), the following code could be used:
1247
1248 setlocale(LC_CTYPE, "fr_FR");
1249 tables = pcre_maketables();
1250 re = pcre_compile(..., tables);
1251
1252 When pcre_maketables() runs, the tables are built in memory that is
1253 obtained via pcre_malloc. It is the caller's responsibility to ensure
1254 that the memory containing the tables remains available for as long as
1255 it is needed.
1256
1257 The pointer that is passed to pcre_compile() is saved with the compiled
1258 pattern, and the same tables are used via this pointer by pcre_study()
1259 and normally also by pcre_exec(). Thus, by default, for any single pat-
1260 tern, compilation, studying and matching all happen in the same locale,
1261 but different patterns can be compiled in different locales.
1262
1263 It is possible to pass a table pointer or NULL (indicating the use of
1264 the internal tables) to pcre_exec(). Although not intended for this
1265 purpose, this facility could be used to match a pattern in a different
1266 locale from the one in which it was compiled. Passing table pointers at
1267 run time is discussed below in the section on matching a pattern.
1268
1269
1270 INFORMATION ABOUT A PATTERN
1271
1272 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1273 int what, void *where);
1274
1275 The pcre_fullinfo() function returns information about a compiled pat-
1276 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1277 less retained for backwards compability (and is documented below).
1278
1279 The first argument for pcre_fullinfo() is a pointer to the compiled
1280 pattern. The second argument is the result of pcre_study(), or NULL if
1281 the pattern was not studied. The third argument specifies which piece
1282 of information is required, and the fourth argument is a pointer to a
1283 variable to receive the data. The yield of the function is zero for
1284 success, or one of the following negative numbers:
1285
1286 PCRE_ERROR_NULL the argument code was NULL
1287 the argument where was NULL
1288 PCRE_ERROR_BADMAGIC the "magic number" was not found
1289 PCRE_ERROR_BADOPTION the value of what was invalid
1290
1291 The "magic number" is placed at the start of each compiled pattern as
1292 an simple check against passing an arbitrary memory pointer. Here is a
1293 typical call of pcre_fullinfo(), to obtain the length of the compiled
1294 pattern:
1295
1296 int rc;
1297 size_t length;
1298 rc = pcre_fullinfo(
1299 re, /* result of pcre_compile() */
1300 pe, /* result of pcre_study(), or NULL */
1301 PCRE_INFO_SIZE, /* what is required */
1302 &length); /* where to put the data */
1303
1304 The possible values for the third argument are defined in pcre.h, and
1305 are as follows:
1306
1307 PCRE_INFO_BACKREFMAX
1308
1309 Return the number of the highest back reference in the pattern. The
1310 fourth argument should point to an int variable. Zero is returned if
1311 there are no back references.
1312
1313 PCRE_INFO_CAPTURECOUNT
1314
1315 Return the number of capturing subpatterns in the pattern. The fourth
1316 argument should point to an int variable.
1317
1318 PCRE_INFO_DEFAULT_TABLES
1319
1320 Return a pointer to the internal default character tables within PCRE.
1321 The fourth argument should point to an unsigned char * variable. This
1322 information call is provided for internal use by the pcre_study() func-
1323 tion. External callers can cause PCRE to use its internal tables by
1324 passing a NULL table pointer.
1325
1326 PCRE_INFO_FIRSTBYTE
1327
1328 Return information about the first byte of any matched string, for a
1329 non-anchored pattern. The fourth argument should point to an int vari-
1330 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1331 is still recognized for backwards compatibility.)
1332
1333 If there is a fixed first byte, for example, from a pattern such as
1334 (cat|cow|coyote). Otherwise, if either
1335
1336 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1337 branch starts with "^", or
1338
1339 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1340 set (if it were set, the pattern would be anchored),
1341
1342 -1 is returned, indicating that the pattern matches only at the start
1343 of a subject string or after any newline within the string. Otherwise
1344 -2 is returned. For anchored patterns, -2 is returned.
1345
1346 PCRE_INFO_FIRSTTABLE
1347
1348 If the pattern was studied, and this resulted in the construction of a
1349 256-bit table indicating a fixed set of bytes for the first byte in any
1350 matching string, a pointer to the table is returned. Otherwise NULL is
1351 returned. The fourth argument should point to an unsigned char * vari-
1352 able.
1353
1354 PCRE_INFO_LASTLITERAL
1355
1356 Return the value of the rightmost literal byte that must exist in any
1357 matched string, other than at its start, if such a byte has been
1358 recorded. The fourth argument should point to an int variable. If there
1359 is no such byte, -1 is returned. For anchored patterns, a last literal
1360 byte is recorded only if it follows something of variable length. For
1361 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1362 /^a\dz\d/ the returned value is -1.
1363
1364 PCRE_INFO_NAMECOUNT
1365 PCRE_INFO_NAMEENTRYSIZE
1366 PCRE_INFO_NAMETABLE
1367
1368 PCRE supports the use of named as well as numbered capturing parenthe-
1369 ses. The names are just an additional way of identifying the parenthe-
1370 ses, which still acquire numbers. Several convenience functions such as
1371 pcre_get_named_substring() are provided for extracting captured sub-
1372 strings by name. It is also possible to extract the data directly, by
1373 first converting the name to a number in order to access the correct
1374 pointers in the output vector (described with pcre_exec() below). To do
1375 the conversion, you need to use the name-to-number map, which is
1376 described by these three values.
1377
1378 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1379 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1380 of each entry; both of these return an int value. The entry size
1381 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1382 a pointer to the first entry of the table (a pointer to char). The
1383 first two bytes of each entry are the number of the capturing parenthe-
1384 sis, most significant byte first. The rest of the entry is the corre-
1385 sponding name, zero terminated. The names are in alphabetical order.
1386 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1387 theses numbers. For example, consider the following pattern (assume
1388 PCRE_EXTENDED is set, so white space - including newlines - is
1389 ignored):
1390
1391 (?P<date> (?P<year>(\d\d)?\d\d) -
1392 (?P<month>\d\d) - (?P<day>\d\d) )
1393
1394 There are four named subpatterns, so the table has four entries, and
1395 each entry in the table is eight bytes long. The table is as follows,
1396 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1397 as ??:
1398
1399 00 01 d a t e 00 ??
1400 00 05 d a y 00 ?? ??
1401 00 04 m o n t h 00
1402 00 02 y e a r 00 ??
1403
1404 When writing code to extract data from named subpatterns using the
1405 name-to-number map, remember that the length of the entries is likely
1406 to be different for each compiled pattern.
1407
1408 PCRE_INFO_OPTIONS
1409
1410 Return a copy of the options with which the pattern was compiled. The
1411 fourth argument should point to an unsigned long int variable. These
1412 option bits are those specified in the call to pcre_compile(), modified
1413 by any top-level option settings within the pattern itself.
1414
1415 A pattern is automatically anchored by PCRE if all of its top-level
1416 alternatives begin with one of the following:
1417
1418 ^ unless PCRE_MULTILINE is set
1419 \A always
1420 \G always
1421 .* if PCRE_DOTALL is set and there are no back
1422 references to the subpattern in which .* appears
1423
1424 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1425 by pcre_fullinfo().
1426
1427 PCRE_INFO_SIZE
1428
1429 Return the size of the compiled pattern, that is, the value that was
1430 passed as the argument to pcre_malloc() when PCRE was getting memory in
1431 which to place the compiled data. The fourth argument should point to a
1432 size_t variable.
1433
1434 PCRE_INFO_STUDYSIZE
1435
1436 Return the size of the data block pointed to by the study_data field in
1437 a pcre_extra block. That is, it is the value that was passed to
1438 pcre_malloc() when PCRE was getting memory into which to place the data
1439 created by pcre_study(). The fourth argument should point to a size_t
1440 variable.
1441
1442
1443 OBSOLETE INFO FUNCTION
1444
1445 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1446
1447 The pcre_info() function is now obsolete because its interface is too
1448 restrictive to return all the available data about a compiled pattern.
1449 New programs should use pcre_fullinfo() instead. The yield of
1450 pcre_info() is the number of capturing subpatterns, or one of the fol-
1451 lowing negative numbers:
1452
1453 PCRE_ERROR_NULL the argument code was NULL
1454 PCRE_ERROR_BADMAGIC the "magic number" was not found
1455
1456 If the optptr argument is not NULL, a copy of the options with which
1457 the pattern was compiled is placed in the integer it points to (see
1458 PCRE_INFO_OPTIONS above).
1459
1460 If the pattern is not anchored and the firstcharptr argument is not
1461 NULL, it is used to pass back information about the first character of
1462 any matched string (see PCRE_INFO_FIRSTBYTE above).
1463
1464
1465 REFERENCE COUNTS
1466
1467 int pcre_refcount(pcre *code, int adjust);
1468
1469 The pcre_refcount() function is used to maintain a reference count in
1470 the data block that contains a compiled pattern. It is provided for the
1471 benefit of applications that operate in an object-oriented manner,
1472 where different parts of the application may be using the same compiled
1473 pattern, but you want to free the block when they are all done.
1474
1475 When a pattern is compiled, the reference count field is initialized to
1476 zero. It is changed only by calling this function, whose action is to
1477 add the adjust value (which may be positive or negative) to it. The
1478 yield of the function is the new value. However, the value of the count
1479 is constrained to lie between 0 and 65535, inclusive. If the new value
1480 is outside these limits, it is forced to the appropriate limit value.
1481
1482 Except when it is zero, the reference count is not correctly preserved
1483 if a pattern is compiled on one host and then transferred to a host
1484 whose byte-order is different. (This seems a highly unlikely scenario.)
1485
1486
1487 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1488
1489 int pcre_exec(const pcre *code, const pcre_extra *extra,
1490 const char *subject, int length, int startoffset,
1491 int options, int *ovector, int ovecsize);
1492
1493 The function pcre_exec() is called to match a subject string against a
1494 compiled pattern, which is passed in the code argument. If the pattern
1495 has been studied, the result of the study should be passed in the extra
1496 argument. This function is the main matching facility of the library,
1497 and it operates in a Perl-like manner. For specialist use there is also
1498 an alternative matching function, which is described below in the sec-
1499 tion about the pcre_dfa_exec() function.
1500
1501 In most applications, the pattern will have been compiled (and option-
1502 ally studied) in the same process that calls pcre_exec(). However, it
1503 is possible to save compiled patterns and study data, and then use them
1504 later in different processes, possibly even on different hosts. For a
1505 discussion about this, see the pcreprecompile documentation.
1506
1507 Here is an example of a simple call to pcre_exec():
1508
1509 int rc;
1510 int ovector[30];
1511 rc = pcre_exec(
1512 re, /* result of pcre_compile() */
1513 NULL, /* we didn't study the pattern */
1514 "some string", /* the subject string */
1515 11, /* the length of the subject string */
1516 0, /* start at offset 0 in the subject */
1517 0, /* default options */
1518 ovector, /* vector of integers for substring information */
1519 30); /* number of elements (NOT size in bytes) */
1520
1521 Extra data for pcre_exec()
1522
1523 If the extra argument is not NULL, it must point to a pcre_extra data
1524 block. The pcre_study() function returns such a block (when it doesn't
1525 return NULL), but you can also create one for yourself, and pass addi-
1526 tional information in it. The pcre_extra block contains the following
1527 fields (not necessarily in this order):
1528
1529 unsigned long int flags;
1530 void *study_data;
1531 unsigned long int match_limit;
1532 unsigned long int match_limit_recursion;
1533 void *callout_data;
1534 const unsigned char *tables;
1535
1536 The flags field is a bitmap that specifies which of the other fields
1537 are set. The flag bits are:
1538
1539 PCRE_EXTRA_STUDY_DATA
1540 PCRE_EXTRA_MATCH_LIMIT
1541 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1542 PCRE_EXTRA_CALLOUT_DATA
1543 PCRE_EXTRA_TABLES
1544
1545 Other flag bits should be set to zero. The study_data field is set in
1546 the pcre_extra block that is returned by pcre_study(), together with
1547 the appropriate flag bit. You should not set this yourself, but you may
1548 add to the block by setting the other fields and their corresponding
1549 flag bits.
1550
1551 The match_limit field provides a means of preventing PCRE from using up
1552 a vast amount of resources when running patterns that are not going to
1553 match, but which have a very large number of possibilities in their
1554 search trees. The classic example is the use of nested unlimited
1555 repeats.
1556
1557 Internally, PCRE uses a function called match() which it calls repeat-
1558 edly (sometimes recursively). The limit set by match_limit is imposed
1559 on the number of times this function is called during a match, which
1560 has the effect of limiting the amount of backtracking that can take
1561 place. For patterns that are not anchored, the count restarts from zero
1562 for each position in the subject string.
1563
1564 The default value for the limit can be set when PCRE is built; the
1565 default default is 10 million, which handles all but the most extreme
1566 cases. You can override the default by suppling pcre_exec() with a
1567 pcre_extra block in which match_limit is set, and
1568 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1569 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1570
1571 The match_limit_recursion field is similar to match_limit, but instead
1572 of limiting the total number of times that match() is called, it limits
1573 the depth of recursion. The recursion depth is a smaller number than
1574 the total number of calls, because not all calls to match() are recur-
1575 sive. This limit is of use only if it is set smaller than match_limit.
1576
1577 Limiting the recursion depth limits the amount of stack that can be
1578 used, or, when PCRE has been compiled to use memory on the heap instead
1579 of the stack, the amount of heap memory that can be used.
1580
1581 The default value for match_limit_recursion can be set when PCRE is
1582 built; the default default is the same value as the default for
1583 match_limit. You can override the default by suppling pcre_exec() with
1584 a pcre_extra block in which match_limit_recursion is set, and
1585 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1586 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1587
1588 The pcre_callout field is used in conjunction with the "callout" fea-
1589 ture, which is described in the pcrecallout documentation.
1590
1591 The tables field is used to pass a character tables pointer to
1592 pcre_exec(); this overrides the value that is stored with the compiled
1593 pattern. A non-NULL value is stored with the compiled pattern only if
1594 custom tables were supplied to pcre_compile() via its tableptr argu-
1595 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1596 PCRE's internal tables to be used. This facility is helpful when re-
1597 using patterns that have been saved after compiling with an external
1598 set of tables, because the external tables might be at a different
1599 address when pcre_exec() is called. See the pcreprecompile documenta-
1600 tion for a discussion of saving compiled patterns for later use.
1601
1602 Option bits for pcre_exec()
1603
1604 The unused bits of the options argument for pcre_exec() must be zero.
1605 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1606 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1607 PCRE_PARTIAL.
1608
1609 PCRE_ANCHORED
1610
1611 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1612 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1613 turned out to be anchored by virtue of its contents, it cannot be made
1614 unachored at matching time.
1615
1616 PCRE_NEWLINE_CR
1617 PCRE_NEWLINE_LF
1618 PCRE_NEWLINE_CRLF
1619
1620 These options override the newline definition that was chosen or
1621 defaulted when the pattern was compiled. For details, see the descrip-
1622 tion pcre_compile() above. During matching, the newline choice affects
1623 the behaviour of the dot, circumflex, and dollar metacharacters.
1624
1625 PCRE_NOTBOL
1626
1627 This option specifies that first character of the subject string is not
1628 the beginning of a line, so the circumflex metacharacter should not
1629 match before it. Setting this without PCRE_MULTILINE (at compile time)
1630 causes circumflex never to match. This option affects only the behav-
1631 iour of the circumflex metacharacter. It does not affect \A.
1632
1633 PCRE_NOTEOL
1634
1635 This option specifies that the end of the subject string is not the end
1636 of a line, so the dollar metacharacter should not match it nor (except
1637 in multiline mode) a newline immediately before it. Setting this with-
1638 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1639 option affects only the behaviour of the dollar metacharacter. It does
1640 not affect \Z or \z.
1641
1642 PCRE_NOTEMPTY
1643
1644 An empty string is not considered to be a valid match if this option is
1645 set. If there are alternatives in the pattern, they are tried. If all
1646 the alternatives match the empty string, the entire match fails. For
1647 example, if the pattern
1648
1649 a?b?
1650
1651 is applied to a string not beginning with "a" or "b", it matches the
1652 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1653 match is not valid, so PCRE searches further into the string for occur-
1654 rences of "a" or "b".
1655
1656 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1657 cial case of a pattern match of the empty string within its split()
1658 function, and when using the /g modifier. It is possible to emulate
1659 Perl's behaviour after matching a null string by first trying the match
1660 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1661 if that fails by advancing the starting offset (see below) and trying
1662 an ordinary match again. There is some code that demonstrates how to do
1663 this in the pcredemo.c sample program.
1664
1665 PCRE_NO_UTF8_CHECK
1666
1667 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1668 UTF-8 string is automatically checked when pcre_exec() is subsequently
1669 called. The value of startoffset is also checked to ensure that it
1670 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1671 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1672 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1673 returned.
1674
1675 If you already know that your subject is valid, and you want to skip
1676 these checks for performance reasons, you can set the
1677 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1678 do this for the second and subsequent calls to pcre_exec() if you are
1679 making repeated calls to find all the matches in a single subject
1680 string. However, you should be sure that the value of startoffset
1681 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1682 set, the effect of passing an invalid UTF-8 string as a subject, or a
1683 value of startoffset that does not point to the start of a UTF-8 char-
1684 acter, is undefined. Your program may crash.
1685
1686 PCRE_PARTIAL
1687
1688 This option turns on the partial matching feature. If the subject
1689 string fails to match the pattern, but at some point during the match-
1690 ing process the end of the subject was reached (that is, the subject
1691 partially matches the pattern and the failure to match occurred only
1692 because there were not enough subject characters), pcre_exec() returns
1693 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1694 used, there are restrictions on what may appear in the pattern. These
1695 are discussed in the pcrepartial documentation.
1696
1697 The string to be matched by pcre_exec()
1698
1699 The subject string is passed to pcre_exec() as a pointer in subject, a
1700 length in length, and a starting byte offset in startoffset. In UTF-8
1701 mode, the byte offset must point to the start of a UTF-8 character.
1702 Unlike the pattern string, the subject may contain binary zero bytes.
1703 When the starting offset is zero, the search for a match starts at the
1704 beginning of the subject, and this is by far the most common case.
1705
1706 A non-zero starting offset is useful when searching for another match
1707 in the same subject by calling pcre_exec() again after a previous suc-
1708 cess. Setting startoffset differs from just passing over a shortened
1709 string and setting PCRE_NOTBOL in the case of a pattern that begins
1710 with any kind of lookbehind. For example, consider the pattern
1711
1712 \Biss\B
1713
1714 which finds occurrences of "iss" in the middle of words. (\B matches
1715 only if the current position in the subject is not a word boundary.)
1716 When applied to the string "Mississipi" the first call to pcre_exec()
1717 finds the first occurrence. If pcre_exec() is called again with just
1718 the remainder of the subject, namely "issipi", it does not match,
1719 because \B is always false at the start of the subject, which is deemed
1720 to be a word boundary. However, if pcre_exec() is passed the entire
1721 string again, but with startoffset set to 4, it finds the second occur-
1722 rence of "iss" because it is able to look behind the starting point to
1723 discover that it is preceded by a letter.
1724
1725 If a non-zero starting offset is passed when the pattern is anchored,
1726 one attempt to match at the given offset is made. This can only succeed
1727 if the pattern does not require the match to be at the start of the
1728 subject.
1729
1730 How pcre_exec() returns captured substrings
1731
1732 In general, a pattern matches a certain portion of the subject, and in
1733 addition, further substrings from the subject may be picked out by
1734 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1735 this is called "capturing" in what follows, and the phrase "capturing
1736 subpattern" is used for a fragment of a pattern that picks out a sub-
1737 string. PCRE supports several other kinds of parenthesized subpattern
1738 that do not cause substrings to be captured.
1739
1740 Captured substrings are returned to the caller via a vector of integer
1741 offsets whose address is passed in ovector. The number of elements in
1742 the vector is passed in ovecsize, which must be a non-negative number.
1743 Note: this argument is NOT the size of ovector in bytes.
1744
1745 The first two-thirds of the vector is used to pass back captured sub-
1746 strings, each substring using a pair of integers. The remaining third
1747 of the vector is used as workspace by pcre_exec() while matching cap-
1748 turing subpatterns, and is not available for passing back information.
1749 The length passed in ovecsize should always be a multiple of three. If
1750 it is not, it is rounded down.
1751
1752 When a match is successful, information about captured substrings is
1753 returned in pairs of integers, starting at the beginning of ovector,
1754 and continuing up to two-thirds of its length at the most. The first
1755 element of a pair is set to the offset of the first character in a sub-
1756 string, and the second is set to the offset of the first character
1757 after the end of a substring. The first pair, ovector[0] and ovec-
1758 tor[1], identify the portion of the subject string matched by the
1759 entire pattern. The next pair is used for the first capturing subpat-
1760 tern, and so on. The value returned by pcre_exec() is one more than the
1761 highest numbered pair that has been set. For example, if two substrings
1762 have been captured, the returned value is 3. If there are no capturing
1763 subpatterns, the return value from a successful match is 1, indicating
1764 that just the first pair of offsets has been set.
1765
1766 If a capturing subpattern is matched repeatedly, it is the last portion
1767 of the string that it matched that is returned.
1768
1769 If the vector is too small to hold all the captured substring offsets,
1770 it is used as far as possible (up to two-thirds of its length), and the
1771 function returns a value of zero. In particular, if the substring off-
1772 sets are not of interest, pcre_exec() may be called with ovector passed
1773 as NULL and ovecsize as zero. However, if the pattern contains back
1774 references and the ovector is not big enough to remember the related
1775 substrings, PCRE has to get additional memory for use during matching.
1776 Thus it is usually advisable to supply an ovector.
1777
1778 The pcre_info() function can be used to find out how many capturing
1779 subpatterns there are in a compiled pattern. The smallest size for
1780 ovector that will allow for n captured substrings, in addition to the
1781 offsets of the substring matched by the whole pattern, is (n+1)*3.
1782
1783 It is possible for capturing subpattern number n+1 to match some part
1784 of the subject when subpattern n has not been used at all. For example,
1785 if the string "abc" is matched against the pattern (a|(z))(bc) the
1786 return from the function is 4, and subpatterns 1 and 3 are matched, but
1787 2 is not. When this happens, both values in the offset pairs corre-
1788 sponding to unused subpatterns are set to -1.
1789
1790 Offset values that correspond to unused subpatterns at the end of the
1791 expression are also set to -1. For example, if the string "abc" is
1792 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1793 matched. The return from the function is 2, because the highest used
1794 capturing subpattern number is 1. However, you can refer to the offsets
1795 for the second and third capturing subpatterns if you wish (assuming
1796 the vector is large enough, of course).
1797
1798 Some convenience functions are provided for extracting the captured
1799 substrings as separate strings. These are described below.
1800
1801 Error return values from pcre_exec()
1802
1803 If pcre_exec() fails, it returns a negative number. The following are
1804 defined in the header file:
1805
1806 PCRE_ERROR_NOMATCH (-1)
1807
1808 The subject string did not match the pattern.
1809
1810 PCRE_ERROR_NULL (-2)
1811
1812 Either code or subject was passed as NULL, or ovector was NULL and
1813 ovecsize was not zero.
1814
1815 PCRE_ERROR_BADOPTION (-3)
1816
1817 An unrecognized bit was set in the options argument.
1818
1819 PCRE_ERROR_BADMAGIC (-4)
1820
1821 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1822 to catch the case when it is passed a junk pointer and to detect when a
1823 pattern that was compiled in an environment of one endianness is run in
1824 an environment with the other endianness. This is the error that PCRE
1825 gives when the magic number is not present.
1826
1827 PCRE_ERROR_UNKNOWN_NODE (-5)
1828
1829 While running the pattern match, an unknown item was encountered in the
1830 compiled pattern. This error could be caused by a bug in PCRE or by
1831 overwriting of the compiled pattern.
1832
1833 PCRE_ERROR_NOMEMORY (-6)
1834
1835 If a pattern contains back references, but the ovector that is passed
1836 to pcre_exec() is not big enough to remember the referenced substrings,
1837 PCRE gets a block of memory at the start of matching to use for this
1838 purpose. If the call via pcre_malloc() fails, this error is given. The
1839 memory is automatically freed at the end of matching.
1840
1841 PCRE_ERROR_NOSUBSTRING (-7)
1842
1843 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1844 and pcre_get_substring_list() functions (see below). It is never
1845 returned by pcre_exec().
1846
1847 PCRE_ERROR_MATCHLIMIT (-8)
1848
1849 The backtracking limit, as specified by the match_limit field in a
1850 pcre_extra structure (or defaulted) was reached. See the description
1851 above.
1852
1853 PCRE_ERROR_RECURSIONLIMIT (-21)
1854
1855 The internal recursion limit, as specified by the match_limit_recursion
1856 field in a pcre_extra structure (or defaulted) was reached. See the
1857 description above.
1858
1859 PCRE_ERROR_CALLOUT (-9)
1860
1861 This error is never generated by pcre_exec() itself. It is provided for
1862 use by callout functions that want to yield a distinctive error code.
1863 See the pcrecallout documentation for details.
1864
1865 PCRE_ERROR_BADUTF8 (-10)
1866
1867 A string that contains an invalid UTF-8 byte sequence was passed as a
1868 subject.
1869
1870 PCRE_ERROR_BADUTF8_OFFSET (-11)
1871
1872 The UTF-8 byte sequence that was passed as a subject was valid, but the
1873 value of startoffset did not point to the beginning of a UTF-8 charac-
1874 ter.
1875
1876 PCRE_ERROR_PARTIAL (-12)
1877
1878 The subject string did not match, but it did match partially. See the
1879 pcrepartial documentation for details of partial matching.
1880
1881 PCRE_ERROR_BADPARTIAL (-13)
1882
1883 The PCRE_PARTIAL option was used with a compiled pattern containing
1884 items that are not supported for partial matching. See the pcrepartial
1885 documentation for details of partial matching.
1886
1887 PCRE_ERROR_INTERNAL (-14)
1888
1889 An unexpected internal error has occurred. This error could be caused
1890 by a bug in PCRE or by overwriting of the compiled pattern.
1891
1892 PCRE_ERROR_BADCOUNT (-15)
1893
1894 This error is given if the value of the ovecsize argument is negative.
1895
1896
1897 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
1898
1899 int pcre_copy_substring(const char *subject, int *ovector,
1900 int stringcount, int stringnumber, char *buffer,
1901 int buffersize);
1902
1903 int pcre_get_substring(const char *subject, int *ovector,
1904 int stringcount, int stringnumber,
1905 const char **stringptr);
1906
1907 int pcre_get_substring_list(const char *subject,
1908 int *ovector, int stringcount, const char ***listptr);
1909
1910 Captured substrings can be accessed directly by using the offsets
1911 returned by pcre_exec() in ovector. For convenience, the functions
1912 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
1913 string_list() are provided for extracting captured substrings as new,
1914 separate, zero-terminated strings. These functions identify substrings
1915 by number. The next section describes functions for extracting named
1916 substrings.
1917
1918 A substring that contains a binary zero is correctly extracted and has
1919 a further zero added on the end, but the result is not, of course, a C
1920 string. However, you can process such a string by referring to the
1921 length that is returned by pcre_copy_substring() and pcre_get_sub-
1922 string(). Unfortunately, the interface to pcre_get_substring_list() is
1923 not adequate for handling strings containing binary zeros, because the
1924 end of the final string is not independently indicated.
1925
1926 The first three arguments are the same for all three of these func-
1927 tions: subject is the subject string that has just been successfully
1928 matched, ovector is a pointer to the vector of integer offsets that was
1929 passed to pcre_exec(), and stringcount is the number of substrings that
1930 were captured by the match, including the substring that matched the
1931 entire regular expression. This is the value returned by pcre_exec() if
1932 it is greater than zero. If pcre_exec() returned zero, indicating that
1933 it ran out of space in ovector, the value passed as stringcount should
1934 be the number of elements in the vector divided by three.
1935
1936 The functions pcre_copy_substring() and pcre_get_substring() extract a
1937 single substring, whose number is given as stringnumber. A value of
1938 zero extracts the substring that matched the entire pattern, whereas
1939 higher values extract the captured substrings. For pcre_copy_sub-
1940 string(), the string is placed in buffer, whose length is given by
1941 buffersize, while for pcre_get_substring() a new block of memory is
1942 obtained via pcre_malloc, and its address is returned via stringptr.
1943 The yield of the function is the length of the string, not including
1944 the terminating zero, or one of
1945
1946 PCRE_ERROR_NOMEMORY (-6)
1947
1948 The buffer was too small for pcre_copy_substring(), or the attempt to
1949 get memory failed for pcre_get_substring().
1950
1951 PCRE_ERROR_NOSUBSTRING (-7)
1952
1953 There is no substring whose number is stringnumber.
1954
1955 The pcre_get_substring_list() function extracts all available sub-
1956 strings and builds a list of pointers to them. All this is done in a
1957 single block of memory that is obtained via pcre_malloc. The address of
1958 the memory block is returned via listptr, which is also the start of
1959 the list of string pointers. The end of the list is marked by a NULL
1960 pointer. The yield of the function is zero if all went well, or
1961
1962 PCRE_ERROR_NOMEMORY (-6)
1963
1964 if the attempt to get the memory block failed.
1965
1966 When any of these functions encounter a substring that is unset, which
1967 can happen when capturing subpattern number n+1 matches some part of
1968 the subject, but subpattern n has not been used at all, they return an
1969 empty string. This can be distinguished from a genuine zero-length sub-
1970 string by inspecting the appropriate offset in ovector, which is nega-
1971 tive for unset substrings.
1972
1973 The two convenience functions pcre_free_substring() and pcre_free_sub-
1974 string_list() can be used to free the memory returned by a previous
1975 call of pcre_get_substring() or pcre_get_substring_list(), respec-
1976 tively. They do nothing more than call the function pointed to by
1977 pcre_free, which of course could be called directly from a C program.
1978 However, PCRE is used in some situations where it is linked via a spe-
1979 cial interface to another programming language that cannot use
1980 pcre_free directly; it is for these cases that the functions are pro-
1981 vided.
1982
1983
1984 EXTRACTING CAPTURED SUBSTRINGS BY NAME
1985
1986 int pcre_get_stringnumber(const pcre *code,
1987 const char *name);
1988
1989 int pcre_copy_named_substring(const pcre *code,
1990 const char *subject, int *ovector,
1991 int stringcount, const char *stringname,
1992 char *buffer, int buffersize);
1993
1994 int pcre_get_named_substring(const pcre *code,
1995 const char *subject, int *ovector,
1996 int stringcount, const char *stringname,
1997 const char **stringptr);
1998
1999 To extract a substring by name, you first have to find associated num-
2000 ber. For example, for this pattern
2001
2002 (a+)b(?P<xxx>\d+)...
2003
2004 the number of the subpattern called "xxx" is 2. If the name is known to
2005 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2006 name by calling pcre_get_stringnumber(). The first argument is the com-
2007 piled pattern, and the second is the name. The yield of the function is
2008 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2009 subpattern of that name.
2010
2011 Given the number, you can extract the substring directly, or use one of
2012 the functions described in the previous section. For convenience, there
2013 are also two functions that do the whole job.
2014
2015 Most of the arguments of pcre_copy_named_substring() and
2016 pcre_get_named_substring() are the same as those for the similarly
2017 named functions that extract by number. As these are described in the
2018 previous section, they are not re-described here. There are just two
2019 differences:
2020
2021 First, instead of a substring number, a substring name is given. Sec-
2022 ond, there is an extra argument, given at the start, which is a pointer
2023 to the compiled pattern. This is needed in order to gain access to the
2024 name-to-number translation table.
2025
2026 These functions call pcre_get_stringnumber(), and if it succeeds, they
2027 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2028 ate.
2029
2030
2031 DUPLICATE SUBPATTERN NAMES
2032
2033 int pcre_get_stringtable_entries(const pcre *code,
2034 const char *name, char **first, char **last);
2035
2036 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2037 subpatterns are not required to be unique. Normally, patterns with
2038 duplicate names are such that in any one match, only one of the named
2039 subpatterns participates. An example is shown in the pcrepattern docu-
2040 mentation. When duplicates are present, pcre_copy_named_substring() and
2041 pcre_get_named_substring() return the first substring corresponding to
2042 the given name that is set. If none are set, an empty string is
2043 returned. The pcre_get_stringnumber() function returns one of the num-
2044 bers that are associated with the name, but it is not defined which it
2045 is.
2046
2047 If you want to get full details of all captured substrings for a given
2048 name, you must use the pcre_get_stringtable_entries() function. The
2049 first argument is the compiled pattern, and the second is the name. The
2050 third and fourth are pointers to variables which are updated by the
2051 function. After it has run, they point to the first and last entries in
2052 the name-to-number table for the given name. The function itself
2053 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING if there
2054 are none. The format of the table is described above in the section
2055 entitled Information about a pattern. Given all the relevant entries
2056 for the name, you can extract each of their numbers, and hence the cap-
2057 tured data, if any.
2058
2059
2060 FINDING ALL POSSIBLE MATCHES
2061
2062 The traditional matching function uses a similar algorithm to Perl,
2063 which stops when it finds the first match, starting at a given point in
2064 the subject. If you want to find all possible matches, or the longest
2065 possible match, consider using the alternative matching function (see
2066 below) instead. If you cannot use the alternative function, but still
2067 need to find all possible matches, you can kludge it up by making use
2068 of the callout facility, which is described in the pcrecallout documen-
2069 tation.
2070
2071 What you have to do is to insert a callout right at the end of the pat-
2072 tern. When your callout function is called, extract and save the cur-
2073 rent matched substring. Then return 1, which forces pcre_exec() to
2074 backtrack and try other alternatives. Ultimately, when it runs out of
2075 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2076
2077
2078 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2079
2080 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2081 const char *subject, int length, int startoffset,
2082 int options, int *ovector, int ovecsize,
2083 int *workspace, int wscount);
2084
2085 The function pcre_dfa_exec() is called to match a subject string
2086 against a compiled pattern, using a "DFA" matching algorithm. This has
2087 different characteristics to the normal algorithm, and is not compati-
2088 ble with Perl. Some of the features of PCRE patterns are not supported.
2089 Nevertheless, there are times when this kind of matching can be useful.
2090 For a discussion of the two matching algorithms, see the pcrematching
2091 documentation.
2092
2093 The arguments for the pcre_dfa_exec() function are the same as for
2094 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2095 ent way, and this is described below. The other common arguments are
2096 used in the same way as for pcre_exec(), so their description is not
2097 repeated here.
2098
2099 The two additional arguments provide workspace for the function. The
2100 workspace vector should contain at least 20 elements. It is used for
2101 keeping track of multiple paths through the pattern tree. More
2102 workspace will be needed for patterns and subjects where there are a
2103 lot of potential matches.
2104
2105 Here is an example of a simple call to pcre_dfa_exec():
2106
2107 int rc;
2108 int ovector[10];
2109 int wspace[20];
2110 rc = pcre_dfa_exec(
2111 re, /* result of pcre_compile() */
2112 NULL, /* we didn't study the pattern */
2113 "some string", /* the subject string */
2114 11, /* the length of the subject string */
2115 0, /* start at offset 0 in the subject */
2116 0, /* default options */
2117 ovector, /* vector of integers for substring information */
2118 10, /* number of elements (NOT size in bytes) */
2119 wspace, /* working space vector */
2120 20); /* number of elements (NOT size in bytes) */
2121
2122 Option bits for pcre_dfa_exec()
2123
2124 The unused bits of the options argument for pcre_dfa_exec() must be
2125 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2126 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2127 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2128 three of these are the same as for pcre_exec(), so their description is
2129 not repeated here.
2130
2131 PCRE_PARTIAL
2132
2133 This has the same general effect as it does for pcre_exec(), but the
2134 details are slightly different. When PCRE_PARTIAL is set for
2135 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2136 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2137 been no complete matches, but there is still at least one matching pos-
2138 sibility. The portion of the string that provided the partial match is
2139 set as the first matching string.
2140
2141 PCRE_DFA_SHORTEST
2142
2143 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2144 stop as soon as it has found one match. Because of the way the DFA
2145 algorithm works, this is necessarily the shortest possible match at the
2146 first possible matching point in the subject string.
2147
2148 PCRE_DFA_RESTART
2149
2150 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2151 returns a partial match, it is possible to call it again, with addi-
2152 tional subject characters, and have it continue with the same match.
2153 The PCRE_DFA_RESTART option requests this action; when it is set, the
2154 workspace and wscount options must reference the same vector as before
2155 because data about the match so far is left in them after a partial
2156 match. There is more discussion of this facility in the pcrepartial
2157 documentation.
2158
2159 Successful returns from pcre_dfa_exec()
2160
2161 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2162 string in the subject. Note, however, that all the matches from one run
2163 of the function start at the same point in the subject. The shorter
2164 matches are all initial substrings of the longer matches. For example,
2165 if the pattern
2166
2167 <.*>
2168
2169 is matched against the string
2170
2171 This is <something> <something else> <something further> no more
2172
2173 the three matched strings are
2174
2175 <something>
2176 <something> <something else>
2177 <something> <something else> <something further>
2178
2179 On success, the yield of the function is a number greater than zero,
2180 which is the number of matched substrings. The substrings themselves
2181 are returned in ovector. Each string uses two elements; the first is
2182 the offset to the start, and the second is the offset to the end. All
2183 the strings have the same start offset. (Space could have been saved by
2184 giving this only once, but it was decided to retain some compatibility
2185 with the way pcre_exec() returns data, even though the meaning of the
2186 strings is different.)
2187
2188 The strings are returned in reverse order of length; that is, the long-
2189 est matching string is given first. If there were too many matches to
2190 fit into ovector, the yield of the function is zero, and the vector is
2191 filled with the longest matches.
2192
2193 Error returns from pcre_dfa_exec()
2194
2195 The pcre_dfa_exec() function returns a negative number when it fails.
2196 Many of the errors are the same as for pcre_exec(), and these are
2197 described above. There are in addition the following errors that are
2198 specific to pcre_dfa_exec():
2199
2200 PCRE_ERROR_DFA_UITEM (-16)
2201
2202 This return is given if pcre_dfa_exec() encounters an item in the pat-
2203 tern that it does not support, for instance, the use of \C or a back
2204 reference.
2205
2206 PCRE_ERROR_DFA_UCOND (-17)
2207
2208 This return is given if pcre_dfa_exec() encounters a condition item in
2209 a pattern that uses a back reference for the condition. This is not
2210 supported.
2211
2212 PCRE_ERROR_DFA_UMLIMIT (-18)
2213
2214 This return is given if pcre_dfa_exec() is called with an extra block
2215 that contains a setting of the match_limit field. This is not supported
2216 (it is meaningless).
2217
2218 PCRE_ERROR_DFA_WSSIZE (-19)
2219
2220 This return is given if pcre_dfa_exec() runs out of space in the
2221 workspace vector.
2222
2223 PCRE_ERROR_DFA_RECURSE (-20)
2224
2225 When a recursive subpattern is processed, the matching function calls
2226 itself recursively, using private vectors for ovector and workspace.
2227 This error is given if the output vector is not large enough. This
2228 should be extremely rare, as a vector of size 1000 is used.
2229
2230 Last updated: 08 June 2006
2231 Copyright (c) 1997-2006 University of Cambridge.
2232 ------------------------------------------------------------------------------
2233
2234
2235 PCRECALLOUT(3) PCRECALLOUT(3)
2236
2237
2238 NAME
2239 PCRE - Perl-compatible regular expressions
2240
2241
2242 PCRE CALLOUTS
2243
2244 int (*pcre_callout)(pcre_callout_block *);
2245
2246 PCRE provides a feature called "callout", which is a means of temporar-
2247 ily passing control to the caller of PCRE in the middle of pattern
2248 matching. The caller of PCRE provides an external function by putting
2249 its entry point in the global variable pcre_callout. By default, this
2250 variable contains NULL, which disables all calling out.
2251
2252 Within a regular expression, (?C) indicates the points at which the
2253 external function is to be called. Different callout points can be
2254 identified by putting a number less than 256 after the letter C. The
2255 default value is zero. For example, this pattern has two callout
2256 points:
2257
2258 (?C1)eabc(?C2)def
2259
2260 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2261 called, PCRE automatically inserts callouts, all with number 255,
2262 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2263 used with the pattern
2264
2265 A(\d{2}|--)
2266
2267 it is processed as if it were
2268
2269 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2270
2271 Notice that there is a callout before and after each parenthesis and
2272 alternation bar. Automatic callouts can be used for tracking the
2273 progress of pattern matching. The pcretest command has an option that
2274 sets automatic callouts; when it is used, the output indicates how the
2275 pattern is matched. This is useful information when you are trying to
2276 optimize the performance of a particular pattern.
2277
2278
2279 MISSING CALLOUTS
2280
2281 You should be aware that, because of optimizations in the way PCRE
2282 matches patterns, callouts sometimes do not happen. For example, if the
2283 pattern is
2284
2285 ab(?C4)cd
2286
2287 PCRE knows that any matching string must contain the letter "d". If the
2288 subject string is "abyz", the lack of "d" means that matching doesn't
2289 ever start, and the callout is never reached. However, with "abyd",
2290 though the result is still no match, the callout is obeyed.
2291
2292
2293 THE CALLOUT INTERFACE
2294
2295 During matching, when PCRE reaches a callout point, the external func-
2296 tion defined by pcre_callout is called (if it is set). This applies to
2297 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2298 only argument to the callout function is a pointer to a pcre_callout
2299 block. This structure contains the following fields:
2300
2301 int version;
2302 int callout_number;
2303 int *offset_vector;
2304 const char *subject;
2305 int subject_length;
2306 int start_match;
2307 int current_position;
2308 int capture_top;
2309 int capture_last;
2310 void *callout_data;
2311 int pattern_position;
2312 int next_item_length;
2313
2314 The version field is an integer containing the version number of the
2315 block format. The initial version was 0; the current version is 1. The
2316 version number will change again in future if additional fields are
2317 added, but the intention is never to remove any of the existing fields.
2318
2319 The callout_number field contains the number of the callout, as com-
2320 piled into the pattern (that is, the number after ?C for manual call-
2321 outs, and 255 for automatically generated callouts).
2322
2323 The offset_vector field is a pointer to the vector of offsets that was
2324 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2325 pcre_exec() is used, the contents can be inspected in order to extract
2326 substrings that have been matched so far, in the same way as for
2327 extracting substrings after a match has completed. For pcre_dfa_exec()
2328 this field is not useful.
2329
2330 The subject and subject_length fields contain copies of the values that
2331 were passed to pcre_exec().
2332
2333 The start_match field contains the offset within the subject at which
2334 the current match attempt started. If the pattern is not anchored, the
2335 callout function may be called several times from the same point in the
2336 pattern for different starting points in the subject.
2337
2338 The current_position field contains the offset within the subject of
2339 the current match pointer.
2340
2341 When the pcre_exec() function is used, the capture_top field contains
2342 one more than the number of the highest numbered captured substring so
2343 far. If no substrings have been captured, the value of capture_top is
2344 one. This is always the case when pcre_dfa_exec() is used, because it
2345 does not support captured substrings.
2346
2347 The capture_last field contains the number of the most recently cap-
2348 tured substring. If no substrings have been captured, its value is -1.
2349 This is always the case when pcre_dfa_exec() is used.
2350
2351 The callout_data field contains a value that is passed to pcre_exec()
2352 or pcre_dfa_exec() specifically so that it can be passed back in call-
2353 outs. It is passed in the pcre_callout field of the pcre_extra data
2354 structure. If no such data was passed, the value of callout_data in a
2355 pcre_callout block is NULL. There is a description of the pcre_extra
2356 structure in the pcreapi documentation.
2357
2358 The pattern_position field is present from version 1 of the pcre_call-
2359 out structure. It contains the offset to the next item to be matched in
2360 the pattern string.
2361
2362 The next_item_length field is present from version 1 of the pcre_call-
2363 out structure. It contains the length of the next item to be matched in
2364 the pattern string. When the callout immediately precedes an alterna-
2365 tion bar, a closing parenthesis, or the end of the pattern, the length
2366 is zero. When the callout precedes an opening parenthesis, the length
2367 is that of the entire subpattern.
2368
2369 The pattern_position and next_item_length fields are intended to help
2370 in distinguishing between different automatic callouts, which all have
2371 the same callout number. However, they are set for all callouts.
2372
2373
2374 RETURN VALUES
2375
2376 The external callout function returns an integer to PCRE. If the value
2377 is zero, matching proceeds as normal. If the value is greater than
2378 zero, matching fails at the current point, but the testing of other
2379 matching possibilities goes ahead, just as if a lookahead assertion had
2380 failed. If the value is less than zero, the match is abandoned, and
2381 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2382
2383 Negative values should normally be chosen from the set of
2384 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2385 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2386 reserved for use by callout functions; it will never be used by PCRE
2387 itself.
2388
2389 Last updated: 28 February 2005
2390 Copyright (c) 1997-2005 University of Cambridge.
2391 ------------------------------------------------------------------------------
2392
2393
2394 PCRECOMPAT(3) PCRECOMPAT(3)
2395
2396
2397 NAME
2398 PCRE - Perl-compatible regular expressions
2399
2400
2401 DIFFERENCES BETWEEN PCRE AND PERL
2402
2403 This document describes the differences in the ways that PCRE and Perl
2404 handle regular expressions. The differences described here are with
2405 respect to Perl 5.8.
2406
2407 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2408 of what it does have are given in the section on UTF-8 support in the
2409 main pcre page.
2410
2411 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2412 permits them, but they do not mean what you might think. For example,
2413 (?!a){3} does not assert that the next three characters are not "a". It
2414 just asserts that the next character is not "a" three times.
2415
2416 3. Capturing subpatterns that occur inside negative lookahead asser-
2417 tions are counted, but their entries in the offsets vector are never
2418 set. Perl sets its numerical variables from any such patterns that are
2419 matched before the assertion fails to match something (thereby succeed-
2420 ing), but only if the negative lookahead assertion contains just one
2421 branch.
2422
2423 4. Though binary zero characters are supported in the subject string,
2424 they are not allowed in a pattern string because it is passed as a nor-
2425 mal C string, terminated by zero. The escape sequence \0 can be used in
2426 the pattern to represent a binary zero.
2427
2428 5. The following Perl escape sequences are not supported: \l, \u, \L,
2429 \U, and \N. In fact these are implemented by Perl's general string-han-
2430 dling and are not part of its pattern matching engine. If any of these
2431 are encountered by PCRE, an error is generated.
2432
2433 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2434 is built with Unicode character property support. The properties that
2435 can be tested with \p and \P are limited to the general category prop-
2436 erties such as Lu and Nd, script names such as Greek or Han, and the
2437 derived properties Any and L&.
2438
2439 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2440 ters in between are treated as literals. This is slightly different
2441 from Perl in that $ and @ are also handled as literals inside the
2442 quotes. In Perl, they cause variable interpolation (but of course PCRE
2443 does not have variables). Note the following examples:
2444
2445 Pattern PCRE matches Perl matches
2446
2447 \Qabc$xyz\E abc$xyz abc followed by the
2448 contents of $xyz
2449 \Qabc\$xyz\E abc\$xyz abc\$xyz
2450 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2451
2452 The \Q...\E sequence is recognized both inside and outside character
2453 classes.
2454
2455 8. Fairly obviously, PCRE does not support the (?{code}) and (?p{code})
2456 constructions. However, there is support for recursive patterns using
2457 the non-Perl items (?R), (?number), and (?P>name). Also, the PCRE
2458 "callout" feature allows an external function to be called during pat-
2459 tern matching. See the pcrecallout documentation for details.
2460
2461 9. There are some differences that are concerned with the settings of
2462 captured strings when part of a pattern is repeated. For example,
2463 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2464 unset, but in PCRE it is set to "b".
2465
2466 10. PCRE provides some extensions to the Perl regular expression facil-
2467 ities:
2468
2469 (a) Although lookbehind assertions must match fixed length strings,
2470 each alternative branch of a lookbehind assertion can match a different
2471 length of string. Perl requires them all to have the same length.
2472
2473 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2474 meta-character matches only at the very end of the string.
2475
2476 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2477 cial meaning is faulted. Otherwise, like Perl, the backslash is
2478 ignored. (Perl can be made to issue a warning.)
2479
2480 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2481 fiers is inverted, that is, by default they are not greedy, but if fol-
2482 lowed by a question mark they are.
2483
2484 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2485 tried only at the first matching position in the subject string.
2486
2487 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2488 TURE options for pcre_exec() have no Perl equivalents.
2489
2490 (g) The (?R), (?number), and (?P>name) constructs allows for recursive
2491 pattern matching (Perl can do this using the (?p{code}) construct,
2492 which PCRE cannot support.)
2493
2494 (h) PCRE supports named capturing substrings, using the Python syntax.
2495
2496 (i) PCRE supports the possessive quantifier "++" syntax, taken from
2497 Sun's Java package.
2498
2499 (j) The (R) condition, for testing recursion, is a PCRE extension.
2500
2501 (k) The callout facility is PCRE-specific.
2502
2503 (l) The partial matching facility is PCRE-specific.
2504
2505 (m) Patterns compiled by PCRE can be saved and re-used at a later time,
2506 even on different hosts that have the other endianness.
2507
2508 (n) The alternative matching function (pcre_dfa_exec()) matches in a
2509 different way and is not Perl-compatible.
2510
2511 Last updated: 06 June 2006
2512 Copyright (c) 1997-2006 University of Cambridge.
2513 ------------------------------------------------------------------------------
2514
2515
2516 PCREPATTERN(3) PCREPATTERN(3)
2517
2518
2519 NAME
2520 PCRE - Perl-compatible regular expressions
2521
2522
2523 PCRE REGULAR EXPRESSION DETAILS
2524
2525 The syntax and semantics of the regular expressions supported by PCRE
2526 are described below. Regular expressions are also described in the Perl
2527 documentation and in a number of books, some of which have copious
2528 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2529 by O'Reilly, covers regular expressions in great detail. This descrip-
2530 tion of PCRE's regular expressions is intended as reference material.
2531
2532 The original operation of PCRE was on strings of one-byte characters.
2533 However, there is now also support for UTF-8 character strings. To use
2534 this, you must build PCRE to include UTF-8 support, and then call
2535 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2536 matching is mentioned in several places below. There is also a summary
2537 of UTF-8 features in the section on UTF-8 support in the main pcre
2538 page.
2539
2540 The remainder of this document discusses the patterns that are sup-
2541 ported by PCRE when its main matching function, pcre_exec(), is used.
2542 From release 6.0, PCRE offers a second matching function,
2543 pcre_dfa_exec(), which matches using a different algorithm that is not
2544 Perl-compatible. The advantages and disadvantages of the alternative
2545 function, and how it differs from the normal function, are discussed in
2546 the pcrematching page.
2547
2548 A regular expression is a pattern that is matched against a subject
2549 string from left to right. Most characters stand for themselves in a
2550 pattern, and match the corresponding characters in the subject. As a
2551 trivial example, the pattern
2552
2553 The quick brown fox
2554
2555 matches a portion of a subject string that is identical to itself. When
2556 caseless matching is specified (the PCRE_CASELESS option), letters are
2557 matched independently of case. In UTF-8 mode, PCRE always understands
2558 the concept of case for characters whose values are less than 128, so
2559 caseless matching is always possible. For characters with higher val-
2560 ues, the concept of case is supported if PCRE is compiled with Unicode
2561 property support, but not otherwise. If you want to use caseless
2562 matching for characters 128 and above, you must ensure that PCRE is
2563 compiled with Unicode property support as well as with UTF-8 support.
2564
2565 The power of regular expressions comes from the ability to include
2566 alternatives and repetitions in the pattern. These are encoded in the
2567 pattern by the use of metacharacters, which do not stand for themselves
2568 but instead are interpreted in some special way.
2569
2570 There are two different sets of metacharacters: those that are recog-
2571 nized anywhere in the pattern except within square brackets, and those
2572 that are recognized in square brackets. Outside square brackets, the
2573 metacharacters are as follows:
2574
2575 \ general escape character with several uses
2576 ^ assert start of string (or line, in multiline mode)
2577 $ assert end of string (or line, in multiline mode)
2578 . match any character except newline (by default)
2579 [ start character class definition
2580 | start of alternative branch
2581 ( start subpattern
2582 ) end subpattern
2583 ? extends the meaning of (
2584 also 0 or 1 quantifier
2585 also quantifier minimizer
2586 * 0 or more quantifier
2587 + 1 or more quantifier
2588 also "possessive quantifier"
2589 { start min/max quantifier
2590
2591 Part of a pattern that is in square brackets is called a "character
2592 class". In a character class the only metacharacters are:
2593
2594 \ general escape character
2595 ^ negate the class, but only if the first character
2596 - indicates character range
2597 [ POSIX character class (only if followed by POSIX
2598 syntax)
2599 ] terminates the character class
2600
2601 The following sections describe the use of each of the metacharacters.
2602
2603
2604 BACKSLASH
2605
2606 The backslash character has several uses. Firstly, if it is followed by
2607 a non-alphanumeric character, it takes away any special meaning that
2608 character may have. This use of backslash as an escape character
2609 applies both inside and outside character classes.
2610
2611 For example, if you want to match a * character, you write \* in the
2612 pattern. This escaping action applies whether or not the following
2613 character would otherwise be interpreted as a metacharacter, so it is
2614 always safe to precede a non-alphanumeric with backslash to specify
2615 that it stands for itself. In particular, if you want to match a back-
2616 slash, you write \\.
2617
2618 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2619 the pattern (other than in a character class) and characters between a
2620 # outside a character class and the next newline are ignored. An escap-
2621 ing backslash can be used to include a whitespace or # character as
2622 part of the pattern.
2623
2624 If you want to remove the special meaning from a sequence of charac-
2625 ters, you can do so by putting them between \Q and \E. This is differ-
2626 ent from Perl in that $ and @ are handled as literals in \Q...\E
2627 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2628 tion. Note the following examples:
2629
2630 Pattern PCRE matches Perl matches
2631
2632 \Qabc$xyz\E abc$xyz abc followed by the
2633 contents of $xyz
2634 \Qabc\$xyz\E abc\$xyz abc\$xyz
2635 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2636
2637 The \Q...\E sequence is recognized both inside and outside character
2638 classes.
2639
2640 Non-printing characters
2641
2642 A second use of backslash provides a way of encoding non-printing char-
2643 acters in patterns in a visible manner. There is no restriction on the
2644 appearance of non-printing characters, apart from the binary zero that
2645 terminates a pattern, but when a pattern is being prepared by text
2646 editing, it is usually easier to use one of the following escape
2647 sequences than the binary character it represents:
2648
2649 \a alarm, that is, the BEL character (hex 07)
2650 \cx "control-x", where x is any character
2651 \e escape (hex 1B)
2652 \f formfeed (hex 0C)
2653 \n newline (hex 0A)
2654 \r carriage return (hex 0D)
2655 \t tab (hex 09)
2656 \ddd character with octal code ddd, or backreference
2657 \xhh character with hex code hh
2658 \x{hhh..} character with hex code hhh..
2659
2660 The precise effect of \cx is as follows: if x is a lower case letter,
2661 it is converted to upper case. Then bit 6 of the character (hex 40) is
2662 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2663 becomes hex 7B.
2664
2665 After \x, from zero to two hexadecimal digits are read (letters can be
2666 in upper or lower case). Any number of hexadecimal digits may appear
2667 between \x{ and }, but the value of the character code must be less
2668 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2669 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2670 hexadecimal digits appear between \x{ and }, or if there is no termi-
2671 nating }, this form of escape is not recognized. Instead, the initial
2672 \x will be interpreted as a basic hexadecimal escape, with no following
2673 digits, giving a character whose value is zero.
2674
2675 Characters whose value is less than 256 can be defined by either of the
2676 two syntaxes for \x. There is no difference in the way they are han-
2677 dled. For example, \xdc is exactly the same as \x{dc}.
2678
2679 After \0 up to two further octal digits are read. If there are fewer
2680 than two digits, just those that are present are used. Thus the
2681 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2682 (code value 7). Make sure you supply two digits after the initial zero
2683 if the pattern character that follows is itself an octal digit.
2684
2685 The handling of a backslash followed by a digit other than 0 is compli-
2686 cated. Outside a character class, PCRE reads it and any following dig-
2687 its as a decimal number. If the number is less than 10, or if there
2688 have been at least that many previous capturing left parentheses in the
2689 expression, the entire sequence is taken as a back reference. A
2690 description of how this works is given later, following the discussion
2691 of parenthesized subpatterns.
2692
2693 Inside a character class, or if the decimal number is greater than 9
2694 and there have not been that many capturing subpatterns, PCRE re-reads
2695 up to three octal digits following the backslash, ane uses them to gen-
2696 erate a data character. Any subsequent digits stand for themselves. In
2697 non-UTF-8 mode, the value of a character specified in octal must be
2698 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2699 example:
2700
2701 \040 is another way of writing a space
2702 \40 is the same, provided there are fewer than 40
2703 previous capturing subpatterns
2704 \7 is always a back reference
2705 \11 might be a back reference, or another way of
2706 writing a tab
2707 \011 is always a tab
2708 \0113 is a tab followed by the character "3"
2709 \113 might be a back reference, otherwise the
2710 character with octal code 113
2711 \377 might be a back reference, otherwise
2712 the byte consisting entirely of 1 bits
2713 \81 is either a back reference, or a binary zero
2714 followed by the two characters "8" and "1"
2715
2716 Note that octal values of 100 or greater must not be introduced by a
2717 leading zero, because no more than three octal digits are ever read.
2718
2719 All the sequences that define a single character value can be used both
2720 inside and outside character classes. In addition, inside a character
2721 class, the sequence \b is interpreted as the backspace character (hex
2722 08), and the sequence \X is interpreted as the character "X". Outside a
2723 character class, these sequences have different meanings (see below).
2724
2725 Generic character types
2726
2727 The third use of backslash is for specifying generic character types.
2728 The following are always recognized:
2729
2730 \d any decimal digit
2731 \D any character that is not a decimal digit
2732 \s any whitespace character
2733 \S any character that is not a whitespace character
2734 \w any "word" character
2735 \W any "non-word" character
2736
2737 Each pair of escape sequences partitions the complete set of characters
2738 into two disjoint sets. Any given character matches one, and only one,
2739 of each pair.
2740
2741 These character type sequences can appear both inside and outside char-
2742 acter classes. They each match one character of the appropriate type.
2743 If the current matching point is at the end of the subject string, all
2744 of them fail, since there is no character to match.
2745
2746 For compatibility with Perl, \s does not match the VT character (code
2747 11). This makes it different from the the POSIX "space" class. The \s
2748 characters are HT (9), LF (10), FF (12), CR (13), and space (32). (If
2749 "use locale;" is included in a Perl script, \s may match the VT charac-
2750 ter. In PCRE, it never does.)
2751
2752 A "word" character is an underscore or any character less than 256 that
2753 is a letter or digit. The definition of letters and digits is con-
2754 trolled by PCRE's low-valued character tables, and may vary if locale-
2755 specific matching is taking place (see "Locale support" in the pcreapi
2756 page). For example, in the "fr_FR" (French) locale, some character
2757 codes greater than 128 are used for accented letters, and these are
2758 matched by \w.
2759
2760 In UTF-8 mode, characters with values greater than 128 never match \d,
2761 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2762 code character property support is available. The use of locales with
2763 Unicode is discouraged.
2764
2765 Unicode character properties
2766
2767 When PCRE is built with Unicode character property support, three addi-
2768 tional escape sequences to match character properties are available
2769 when UTF-8 mode is selected. They are:
2770
2771 \p{xx} a character with the xx property
2772 \P{xx} a character without the xx property
2773 \X an extended Unicode sequence
2774
2775 The property names represented by xx above are limited to the Unicode
2776 script names, the general category properties, and "Any", which matches
2777 any character (including newline). Other properties such as "InMusical-
2778 Symbols" are not currently supported by PCRE. Note that \P{Any} does
2779 not match any characters, so always causes a match failure.
2780
2781 Sets of Unicode characters are defined as belonging to certain scripts.
2782 A character from one of these sets can be matched using a script name.
2783 For example:
2784
2785 \p{Greek}
2786 \P{Han}
2787
2788 Those that are not part of an identified script are lumped together as
2789 "Common". The current list of scripts is:
2790
2791 Arabic, Armenian, Bengali, Bopomofo, Braille, Buginese, Buhid, Cana-
2792 dian_Aboriginal, Cherokee, Common, Coptic, Cypriot, Cyrillic, Deseret,
2793 Devanagari, Ethiopic, Georgian, Glagolitic, Gothic, Greek, Gujarati,
2794 Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Inherited, Kannada,
2795 Katakana, Kharoshthi, Khmer, Lao, Latin, Limbu, Linear_B, Malayalam,
2796 Mongolian, Myanmar, New_Tai_Lue, Ogham, Old_Italic, Old_Persian, Oriya,
2797 Osmanya, Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tag-
2798 banwa, Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
2799 Ugaritic, Yi.
2800
2801 Each character has exactly one general category property, specified by
2802 a two-letter abbreviation. For compatibility with Perl, negation can be
2803 specified by including a circumflex between the opening brace and the
2804 property name. For example, \p{^Lu} is the same as \P{Lu}.
2805
2806 If only one letter is specified with \p or \P, it includes all the gen-
2807 eral category properties that start with that letter. In this case, in
2808 the absence of negation, the curly brackets in the escape sequence are
2809 optional; these two examples have the same effect:
2810
2811 \p{L}
2812 \pL
2813
2814 The following general category property codes are supported:
2815
2816 C Other
2817 Cc Control
2818 Cf Format
2819 Cn Unassigned
2820 Co Private use
2821 Cs Surrogate
2822
2823 L Letter
2824 Ll Lower case letter
2825 Lm Modifier letter
2826 Lo Other letter
2827 Lt Title case letter
2828 Lu Upper case letter
2829
2830 M Mark
2831 Mc Spacing mark
2832 Me Enclosing mark
2833 Mn Non-spacing mark
2834
2835 N Number
2836 Nd Decimal number
2837 Nl Letter number
2838 No Other number
2839
2840 P Punctuation
2841 Pc Connector punctuation
2842 Pd Dash punctuation
2843 Pe Close punctuation
2844 Pf Final punctuation
2845 Pi Initial punctuation
2846 Po Other punctuation
2847 Ps Open punctuation
2848
2849 S Symbol
2850 Sc Currency symbol
2851 Sk Modifier symbol
2852 Sm Mathematical symbol
2853 So Other symbol
2854
2855 Z Separator
2856 Zl Line separator
2857 Zp Paragraph separator
2858 Zs Space separator
2859
2860 The special property L& is also supported: it matches a character that
2861 has the Lu, Ll, or Lt property, in other words, a letter that is not
2862 classified as a modifier or "other".
2863
2864 The long synonyms for these properties that Perl supports (such as
2865 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
2866 any of these properties with "Is".
2867
2868 No character that is in the Unicode table has the Cn (unassigned) prop-
2869 erty. Instead, this property is assumed for any code point that is not
2870 in the Unicode table.
2871
2872 Specifying caseless matching does not affect these escape sequences.
2873 For example, \p{Lu} always matches only upper case letters.
2874
2875 The \X escape matches any number of Unicode characters that form an
2876 extended Unicode sequence. \X is equivalent to
2877
2878 (?>\PM\pM*)
2879
2880 That is, it matches a character without the "mark" property, followed
2881 by zero or more characters with the "mark" property, and treats the
2882 sequence as an atomic group (see below). Characters with the "mark"
2883 property are typically accents that affect the preceding character.
2884
2885 Matching characters by Unicode property is not fast, because PCRE has
2886 to search a structure that contains data for over fifteen thousand
2887 characters. That is why the traditional escape sequences such as \d and
2888 \w do not use Unicode properties in PCRE.
2889
2890 Simple assertions
2891
2892 The fourth use of backslash is for certain simple assertions. An asser-
2893 tion specifies a condition that has to be met at a particular point in
2894 a match, without consuming any characters from the subject string. The
2895 use of subpatterns for more complicated assertions is described below.
2896 The backslashed assertions are:
2897
2898 \b matches at a word boundary
2899 \B matches when not at a word boundary
2900 \A matches at start of subject
2901 \Z matches at end of subject or before newline at end
2902 \z matches at end of subject
2903 \G matches at first matching position in subject
2904
2905 These assertions may not appear in character classes (but note that \b
2906 has a different meaning, namely the backspace character, inside a char-
2907 acter class).
2908
2909 A word boundary is a position in the subject string where the current
2910 character and the previous character do not both match \w or \W (i.e.
2911 one matches \w and the other matches \W), or the start or end of the
2912 string if the first or last character matches \w, respectively.
2913
2914 The \A, \Z, and \z assertions differ from the traditional circumflex
2915 and dollar (described in the next section) in that they only ever match
2916 at the very start and end of the subject string, whatever options are
2917 set. Thus, they are independent of multiline mode. These three asser-
2918 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
2919 affect only the behaviour of the circumflex and dollar metacharacters.
2920 However, if the startoffset argument of pcre_exec() is non-zero, indi-
2921 cating that matching is to start at a point other than the beginning of
2922 the subject, \A can never match. The difference between \Z and \z is
2923 that \Z matches before a newline at the end of the string as well as at
2924 the very end, whereas \z matches only at the end.
2925
2926 The \G assertion is true only when the current matching position is at
2927 the start point of the match, as specified by the startoffset argument
2928 of pcre_exec(). It differs from \A when the value of startoffset is
2929 non-zero. By calling pcre_exec() multiple times with appropriate argu-
2930 ments, you can mimic Perl's /g option, and it is in this kind of imple-
2931 mentation where \G can be useful.
2932
2933 Note, however, that PCRE's interpretation of \G, as the start of the
2934 current match, is subtly different from Perl's, which defines it as the
2935 end of the previous match. In Perl, these can be different when the
2936 previously matched string was empty. Because PCRE does just one match
2937 at a time, it cannot reproduce this behaviour.
2938
2939 If all the alternatives of a pattern begin with \G, the expression is
2940 anchored to the starting match position, and the "anchored" flag is set
2941 in the compiled regular expression.
2942
2943
2944 CIRCUMFLEX AND DOLLAR
2945
2946 Outside a character class, in the default matching mode, the circumflex
2947 character is an assertion that is true only if the current matching
2948 point is at the start of the subject string. If the startoffset argu-
2949 ment of pcre_exec() is non-zero, circumflex can never match if the
2950 PCRE_MULTILINE option is unset. Inside a character class, circumflex
2951 has an entirely different meaning (see below).
2952
2953 Circumflex need not be the first character of the pattern if a number
2954 of alternatives are involved, but it should be the first thing in each
2955 alternative in which it appears if the pattern is ever to match that
2956 branch. If all possible alternatives start with a circumflex, that is,
2957 if the pattern is constrained to match only at the start of the sub-
2958 ject, it is said to be an "anchored" pattern. (There are also other
2959 constructs that can cause a pattern to be anchored.)
2960
2961 A dollar character is an assertion that is true only if the current
2962 matching point is at the end of the subject string, or immediately
2963 before a newline at the end of the string (by default). Dollar need not
2964 be the last character of the pattern if a number of alternatives are
2965 involved, but it should be the last item in any branch in which it
2966 appears. Dollar has no special meaning in a character class.
2967
2968 The meaning of dollar can be changed so that it matches only at the
2969 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
2970 compile time. This does not affect the \Z assertion.
2971
2972 The meanings of the circumflex and dollar characters are changed if the
2973 PCRE_MULTILINE option is set. When this is the case, a circumflex
2974 matches immediately after internal newlines as well as at the start of
2975 the subject string. It does not match after a newline that ends the
2976 string. A dollar matches before any newlines in the string, as well as
2977 at the very end, when PCRE_MULTILINE is set. When newline is specified
2978 as the two-character sequence CRLF, isolated CR and LF characters do
2979 not indicate newlines.
2980
2981 For example, the pattern /^abc$/ matches the subject string "def\nabc"
2982 (where \n represents a newline) in multiline mode, but not otherwise.
2983 Consequently, patterns that are anchored in single line mode because
2984 all branches start with ^ are not anchored in multiline mode, and a
2985 match for circumflex is possible when the startoffset argument of
2986 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
2987 PCRE_MULTILINE is set.
2988
2989 Note that the sequences \A, \Z, and \z can be used to match the start
2990 and end of the subject in both modes, and if all branches of a pattern
2991 start with \A it is always anchored, whether or not PCRE_MULTILINE is
2992 set.
2993
2994
2995 FULL STOP (PERIOD, DOT)
2996
2997 Outside a character class, a dot in the pattern matches any one charac-
2998 ter in the subject string except (by default) a character that signi-
2999 fies the end of a line. In UTF-8 mode, the matched character may be
3000 more than one byte long. When a line ending is defined as a single
3001 character (CR or LF), dot never matches that character; when the two-
3002 character sequence CRLF is used, dot does not match CR if it is immedi-
3003 ately followed by LF, but otherwise it matches all characters (includ-
3004 ing isolated CRs and LFs).
3005
3006 The behaviour of dot with regard to newlines can be changed. If the
3007 PCRE_DOTALL option is set, a dot matches any one character, without
3008 exception. If newline is defined as the two-character sequence CRLF, it
3009 takes two dots to match it.
3010
3011 The handling of dot is entirely independent of the handling of circum-
3012 flex and dollar, the only relationship being that they both involve
3013 newlines. Dot has no special meaning in a character class.
3014
3015
3016 MATCHING A SINGLE BYTE
3017
3018 Outside a character class, the escape sequence \C matches any one byte,
3019 both in and out of UTF-8 mode. Unlike a dot, it always matches CR and
3020 LF. The feature is provided in Perl in order to match individual bytes
3021 in UTF-8 mode. Because it breaks up UTF-8 characters into individual
3022 bytes, what remains in the string may be a malformed UTF-8 string. For
3023 this reason, the \C escape sequence is best avoided.
3024
3025 PCRE does not allow \C to appear in lookbehind assertions (described
3026 below), because in UTF-8 mode this would make it impossible to calcu-
3027 late the length of the lookbehind.
3028
3029
3030 SQUARE BRACKETS AND CHARACTER CLASSES
3031
3032 An opening square bracket introduces a character class, terminated by a
3033 closing square bracket. A closing square bracket on its own is not spe-
3034 cial. If a closing square bracket is required as a member of the class,
3035 it should be the first data character in the class (after an initial
3036 circumflex, if present) or escaped with a backslash.
3037
3038 A character class matches a single character in the subject. In UTF-8
3039 mode, the character may occupy more than one byte. A matched character
3040 must be in the set of characters defined by the class, unless the first
3041 character in the class definition is a circumflex, in which case the
3042 subject character must not be in the set defined by the class. If a
3043 circumflex is actually required as a member of the class, ensure it is
3044 not the first character, or escape it with a backslash.
3045
3046 For example, the character class [aeiou] matches any lower case vowel,
3047 while [^aeiou] matches any character that is not a lower case vowel.
3048 Note that a circumflex is just a convenient notation for specifying the
3049 characters that are in the class by enumerating those that are not. A
3050 class that starts with a circumflex is not an assertion: it still con-
3051 sumes a character from the subject string, and therefore it fails if
3052 the current pointer is at the end of the string.
3053
3054 In UTF-8 mode, characters with values greater than 255 can be included
3055 in a class as a literal string of bytes, or by using the \x{ escaping
3056 mechanism.
3057
3058 When caseless matching is set, any letters in a class represent both
3059 their upper case and lower case versions, so for example, a caseless
3060 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3061 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3062 understands the concept of case for characters whose values are less
3063 than 128, so caseless matching is always possible. For characters with
3064 higher values, the concept of case is supported if PCRE is compiled
3065 with Unicode property support, but not otherwise. If you want to use
3066 caseless matching for characters 128 and above, you must ensure that
3067 PCRE is compiled with Unicode property support as well as with UTF-8
3068 support.
3069
3070 Characters that might indicate line breaks (CR and LF) are never
3071 treated in any special way when matching character classes, whatever
3072 line-ending sequence is in use, and whatever setting of the PCRE_DOTALL
3073 and PCRE_MULTILINE options is used. A class such as [^a] always matches
3074 one of these characters.
3075
3076 The minus (hyphen) character can be used to specify a range of charac-
3077 ters in a character class. For example, [d-m] matches any letter
3078 between d and m, inclusive. If a minus character is required in a
3079 class, it must be escaped with a backslash or appear in a position
3080 where it cannot be interpreted as indicating a range, typically as the
3081 first or last character in the class.
3082
3083 It is not possible to have the literal character "]" as the end charac-
3084 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3085 two characters ("W" and "-") followed by a literal string "46]", so it
3086 would match "W46]" or "-46]". However, if the "]" is escaped with a
3087 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3088 preted as a class containing a range followed by two other characters.
3089 The octal or hexadecimal representation of "]" can also be used to end
3090 a range.
3091
3092 Ranges operate in the collating sequence of character values. They can
3093 also be used for characters specified numerically, for example
3094 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3095 are greater than 255, for example [\x{100}-\x{2ff}].
3096
3097 If a range that includes letters is used when caseless matching is set,
3098 it matches the letters in either case. For example, [W-c] is equivalent
3099 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3100 character tables for the "fr_FR" locale are in use, [\xc8-\xcb] matches
3101 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3102 concept of case for characters with values greater than 128 only when
3103 it is compiled with Unicode property support.
3104
3105 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3106 in a character class, and add the characters that they match to the
3107 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3108 flex can conveniently be used with the upper case character types to
3109 specify a more restricted set of characters than the matching lower
3110 case type. For example, the class [^\W_] matches any letter or digit,
3111 but not underscore.
3112
3113 The only metacharacters that are recognized in character classes are
3114 backslash, hyphen (only where it can be interpreted as specifying a
3115 range), circumflex (only at the start), opening square bracket (only
3116 when it can be interpreted as introducing a POSIX class name - see the
3117 next section), and the terminating closing square bracket. However,
3118 escaping other non-alphanumeric characters does no harm.
3119
3120
3121 POSIX CHARACTER CLASSES
3122
3123 Perl supports the POSIX notation for character classes. This uses names
3124 enclosed by [: and :] within the enclosing square brackets. PCRE also
3125 supports this notation. For example,
3126
3127 [01[:alpha:]%]
3128
3129 matches "0", "1", any alphabetic character, or "%". The supported class
3130 names are
3131
3132 alnum letters and digits
3133 alpha letters
3134 ascii character codes 0 - 127
3135 blank space or tab only
3136 cntrl control characters
3137 digit decimal digits (same as \d)
3138 graph printing characters, excluding space
3139 lower lower case letters
3140 print printing characters, including space
3141 punct printing characters, excluding letters and digits
3142 space white space (not quite the same as \s)
3143 upper upper case letters
3144 word "word" characters (same as \w)
3145 xdigit hexadecimal digits
3146
3147 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3148 and space (32). Notice that this list includes the VT character (code
3149 11). This makes "space" different to \s, which does not include VT (for
3150 Perl compatibility).
3151
3152 The name "word" is a Perl extension, and "blank" is a GNU extension
3153 from Perl 5.8. Another Perl extension is negation, which is indicated
3154 by a ^ character after the colon. For example,
3155
3156 [12[:^digit:]]
3157
3158 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3159 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3160 these are not supported, and an error is given if they are encountered.
3161
3162 In UTF-8 mode, characters with values greater than 128 do not match any
3163 of the POSIX character classes.
3164
3165
3166 VERTICAL BAR
3167
3168 Vertical bar characters are used to separate alternative patterns. For
3169 example, the pattern
3170
3171 gilbert|sullivan
3172
3173 matches either "gilbert" or "sullivan". Any number of alternatives may
3174 appear, and an empty alternative is permitted (matching the empty
3175 string). The matching process tries each alternative in turn, from left
3176 to right, and the first one that succeeds is used. If the alternatives
3177 are within a subpattern (defined below), "succeeds" means matching the
3178 rest of the main pattern as well as the alternative in the subpattern.
3179
3180
3181 INTERNAL OPTION SETTING
3182
3183 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3184 PCRE_EXTENDED options can be changed from within the pattern by a
3185 sequence of Perl option letters enclosed between "(?" and ")". The
3186 option letters are
3187
3188 i for PCRE_CASELESS
3189 m for PCRE_MULTILINE
3190 s for PCRE_DOTALL
3191 x for PCRE_EXTENDED
3192
3193 For example, (?im) sets caseless, multiline matching. It is also possi-
3194 ble to unset these options by preceding the letter with a hyphen, and a
3195 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3196 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3197 is also permitted. If a letter appears both before and after the
3198 hyphen, the option is unset.
3199
3200 When an option change occurs at top level (that is, not inside subpat-
3201 tern parentheses), the change applies to the remainder of the pattern
3202 that follows. If the change is placed right at the start of a pattern,
3203 PCRE extracts it into the global options (and it will therefore show up
3204 in data extracted by the pcre_fullinfo() function).
3205
3206 An option change within a subpattern affects only that part of the cur-
3207 rent pattern that follows it, so
3208
3209 (a(?i)b)c
3210
3211 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3212 used). By this means, options can be made to have different settings
3213 in different parts of the pattern. Any changes made in one alternative
3214 do carry on into subsequent branches within the same subpattern. For
3215 example,
3216
3217 (a(?i)b|c)
3218
3219 matches "ab", "aB", "c", and "C", even though when matching "C" the
3220 first branch is abandoned before the option setting. This is because
3221 the effects of option settings happen at compile time. There would be
3222 some very weird behaviour otherwise.
3223
3224 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3225 can be changed in the same way as the Perl-compatible options by using
3226 the characters J, U and X respectively.
3227
3228
3229 SUBPATTERNS
3230
3231 Subpatterns are delimited by parentheses (round brackets), which can be
3232 nested. Turning part of a pattern into a subpattern does two things:
3233
3234 1. It localizes a set of alternatives. For example, the pattern
3235
3236 cat(aract|erpillar|)
3237
3238 matches one of the words "cat", "cataract", or "caterpillar". Without
3239 the parentheses, it would match "cataract", "erpillar" or the empty
3240 string.
3241
3242 2. It sets up the subpattern as a capturing subpattern. This means
3243 that, when the whole pattern matches, that portion of the subject
3244 string that matched the subpattern is passed back to the caller via the
3245 ovector argument of pcre_exec(). Opening parentheses are counted from
3246 left to right (starting from 1) to obtain numbers for the capturing
3247 subpatterns.
3248
3249 For example, if the string "the red king" is matched against the pat-
3250 tern
3251
3252 the ((red|white) (king|queen))
3253
3254 the captured substrings are "red king", "red", and "king", and are num-
3255 bered 1, 2, and 3, respectively.
3256
3257 The fact that plain parentheses fulfil two functions is not always
3258 helpful. There are often times when a grouping subpattern is required
3259 without a capturing requirement. If an opening parenthesis is followed
3260 by a question mark and a colon, the subpattern does not do any captur-
3261 ing, and is not counted when computing the number of any subsequent
3262 capturing subpatterns. For example, if the string "the white queen" is
3263 matched against the pattern
3264
3265 the ((?:red|white) (king|queen))
3266
3267 the captured substrings are "white queen" and "queen", and are numbered
3268 1 and 2. The maximum number of capturing subpatterns is 65535, and the
3269 maximum depth of nesting of all subpatterns, both capturing and non-
3270 capturing, is 200.
3271
3272 As a convenient shorthand, if any option settings are required at the
3273 start of a non-capturing subpattern, the option letters may appear
3274 between the "?" and the ":". Thus the two patterns
3275
3276 (?i:saturday|sunday)
3277 (?:(?i)saturday|sunday)
3278
3279 match exactly the same set of strings. Because alternative branches are
3280 tried from left to right, and options are not reset until the end of
3281 the subpattern is reached, an option setting in one branch does affect
3282 subsequent branches, so the above patterns match "SUNDAY" as well as
3283 "Saturday".
3284
3285
3286 NAMED SUBPATTERNS
3287
3288 Identifying capturing parentheses by number is simple, but it can be
3289 very hard to keep track of the numbers in complicated regular expres-
3290 sions. Furthermore, if an expression is modified, the numbers may
3291 change. To help with this difficulty, PCRE supports the naming of sub-
3292 patterns, something that Perl does not provide. The Python syntax
3293 (?P<name>...) is used. References to capturing parentheses from other
3294 parts of the pattern, such as backreferences, recursion, and condi-
3295 tions, can be made by name as well as by number.
3296
3297 Names consist of up to 32 alphanumeric characters and underscores.
3298 Named capturing parentheses are still allocated numbers as well as
3299 names. The PCRE API provides function calls for extracting the name-to-
3300 number translation table from a compiled pattern. There is also a con-
3301 venience function for extracting a captured substring by name.
3302
3303 By default, a name must be unique within a pattern, but it is possible
3304 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3305 time. This can be useful for patterns where only one instance of the
3306 named parentheses can match. Suppose you want to match the name of a
3307 weekday, either as a 3-letter abbreviation or as the full name, and in
3308 both cases you want to extract the abbreviation. This pattern (ignoring
3309 the line breaks) does the job:
3310
3311 (?P<DN>Mon|Fri|Sun)(?:day)?|
3312 (?P<DN>Tue)(?:sday)?|
3313 (?P<DN>Wed)(?:nesday)?|
3314 (?P<DN>Thu)(?:rsday)?|
3315 (?P<DN>Sat)(?:urday)?
3316
3317 There are five capturing substrings, but only one is ever set after a
3318 match. The convenience function for extracting the data by name
3319 returns the substring for the first, and in this example, the only,
3320 subpattern of that name that matched. This saves searching to find
3321 which numbered subpattern it was. If you make a reference to a non-
3322 unique named subpattern from elsewhere in the pattern, the one that
3323 corresponds to the lowest number is used. For further details of the
3324 interfaces for handling named subpatterns, see the pcreapi documenta-
3325 tion.
3326
3327
3328 REPETITION
3329
3330 Repetition is specified by quantifiers, which can follow any of the
3331 following items:
3332
3333 a literal data character
3334 the . metacharacter
3335 the \C escape sequence
3336 the \X escape sequence (in UTF-8 mode with Unicode properties)
3337 an escape such as \d that matches a single character
3338 a character class
3339 a back reference (see next section)
3340 a parenthesized subpattern (unless it is an assertion)
3341
3342 The general repetition quantifier specifies a minimum and maximum num-
3343 ber of permitted matches, by giving the two numbers in curly brackets
3344 (braces), separated by a comma. The numbers must be less than 65536,
3345 and the first must be less than or equal to the second. For example:
3346
3347 z{2,4}
3348
3349 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3350 special character. If the second number is omitted, but the comma is
3351 present, there is no upper limit; if the second number and the comma
3352 are both omitted, the quantifier specifies an exact number of required
3353 matches. Thus
3354
3355 [aeiou]{3,}
3356
3357 matches at least 3 successive vowels, but may match many more, while
3358
3359 \d{8}
3360
3361 matches exactly 8 digits. An opening curly bracket that appears in a
3362 position where a quantifier is not allowed, or one that does not match
3363 the syntax of a quantifier, is taken as a literal character. For exam-
3364 ple, {,6} is not a quantifier, but a literal string of four characters.
3365
3366 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3367 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3368 acters, each of which is represented by a two-byte sequence. Similarly,
3369 when Unicode property support is available, \X{3} matches three Unicode
3370 extended sequences, each of which may be several bytes long (and they
3371 may be of different lengths).
3372
3373 The quantifier {0} is permitted, causing the expression to behave as if
3374 the previous item and the quantifier were not present.
3375
3376 For convenience (and historical compatibility) the three most common
3377 quantifiers have single-character abbreviations:
3378
3379 * is equivalent to {0,}
3380 + is equivalent to {1,}
3381 ? is equivalent to {0,1}
3382
3383 It is possible to construct infinite loops by following a subpattern
3384 that can match no characters with a quantifier that has no upper limit,
3385 for example:
3386
3387 (a?)*
3388
3389 Earlier versions of Perl and PCRE used to give an error at compile time
3390 for such patterns. However, because there are cases where this can be
3391 useful, such patterns are now accepted, but if any repetition of the
3392 subpattern does in fact match no characters, the loop is forcibly bro-
3393 ken.
3394
3395 By default, the quantifiers are "greedy", that is, they match as much
3396 as possible (up to the maximum number of permitted times), without
3397 causing the rest of the pattern to fail. The classic example of where
3398 this gives problems is in trying to match comments in C programs. These
3399 appear between /* and */ and within the comment, individual * and /
3400 characters may appear. An attempt to match C comments by applying the
3401 pattern
3402
3403 /\*.*\*/
3404
3405 to the string
3406
3407 /* first comment */ not comment /* second comment */
3408
3409 fails, because it matches the entire string owing to the greediness of
3410 the .* item.
3411
3412 However, if a quantifier is followed by a question mark, it ceases to
3413 be greedy, and instead matches the minimum number of times possible, so
3414 the pattern
3415
3416 /\*.*?\*/
3417
3418 does the right thing with the C comments. The meaning of the various
3419 quantifiers is not otherwise changed, just the preferred number of
3420 matches. Do not confuse this use of question mark with its use as a
3421 quantifier in its own right. Because it has two uses, it can sometimes
3422 appear doubled, as in
3423
3424 \d??\d
3425
3426 which matches one digit by preference, but can match two if that is the
3427 only way the rest of the pattern matches.
3428
3429 If the PCRE_UNGREEDY option is set (an option which is not available in
3430 Perl), the quantifiers are not greedy by default, but individual ones
3431 can be made greedy by following them with a question mark. In other
3432 words, it inverts the default behaviour.
3433
3434 When a parenthesized subpattern is quantified with a minimum repeat
3435 count that is greater than 1 or with a limited maximum, more memory is
3436 required for the compiled pattern, in proportion to the size of the
3437 minimum or maximum.
3438
3439 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3440 alent to Perl's /s) is set, thus allowing the . to match newlines, the
3441 pattern is implicitly anchored, because whatever follows will be tried
3442 against every character position in the subject string, so there is no
3443 point in retrying the overall match at any position after the first.
3444 PCRE normally treats such a pattern as though it were preceded by \A.
3445
3446 In cases where it is known that the subject string contains no new-
3447 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3448 mization, or alternatively using ^ to indicate anchoring explicitly.
3449
3450 However, there is one situation where the optimization cannot be used.
3451 When .* is inside capturing parentheses that are the subject of a
3452 backreference elsewhere in the pattern, a match at the start may fail,
3453 and a later one succeed. Consider, for example:
3454
3455 (.*)abc\1
3456
3457 If the subject is "xyz123abc123" the match point is the fourth charac-
3458 ter. For this reason, such a pattern is not implicitly anchored.
3459
3460 When a capturing subpattern is repeated, the value captured is the sub-
3461 string that matched the final iteration. For example, after
3462
3463 (tweedle[dume]{3}\s*)+
3464
3465 has matched "tweedledum tweedledee" the value of the captured substring
3466 is "tweedledee". However, if there are nested capturing subpatterns,
3467 the corresponding captured values may have been set in previous itera-
3468 tions. For example, after
3469
3470 /(a|(b))+/
3471
3472 matches "aba" the value of the second captured substring is "b".
3473
3474
3475 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3476
3477 With both maximizing and minimizing repetition, failure of what follows
3478 normally causes the repeated item to be re-evaluated to see if a dif-
3479 ferent number of repeats allows the rest of the pattern to match. Some-
3480 times it is useful to prevent this, either to change the nature of the
3481 match, or to cause it fail earlier than it otherwise might, when the
3482 author of the pattern knows there is no point in carrying on.
3483
3484 Consider, for example, the pattern \d+foo when applied to the subject
3485 line
3486
3487 123456bar
3488
3489 After matching all 6 digits and then failing to match "foo", the normal
3490 action of the matcher is to try again with only 5 digits matching the
3491 \d+ item, and then with 4, and so on, before ultimately failing.
3492 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3493 the means for specifying that once a subpattern has matched, it is not
3494 to be re-evaluated in this way.
3495
3496 If we use atomic grouping for the previous example, the matcher would
3497 give up immediately on failing to match "foo" the first time. The nota-
3498 tion is a kind of special parenthesis, starting with (?> as in this
3499 example:
3500
3501 (?>\d+)foo
3502
3503 This kind of parenthesis "locks up" the part of the pattern it con-
3504 tains once it has matched, and a failure further into the pattern is
3505 prevented from backtracking into it. Backtracking past it to previous
3506 items, however, works as normal.
3507
3508 An alternative description is that a subpattern of this type matches
3509 the string of characters that an identical standalone pattern would
3510 match, if anchored at the current point in the subject string.
3511
3512 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3513 such as the above example can be thought of as a maximizing repeat that
3514 must swallow everything it can. So, while both \d+ and \d+? are pre-
3515 pared to adjust the number of digits they match in order to make the
3516 rest of the pattern match, (?>\d+) can only match an entire sequence of
3517 digits.
3518
3519 Atomic groups in general can of course contain arbitrarily complicated
3520 subpatterns, and can be nested. However, when the subpattern for an
3521 atomic group is just a single repeated item, as in the example above, a
3522 simpler notation, called a "possessive quantifier" can be used. This
3523 consists of an additional + character following a quantifier. Using
3524 this notation, the previous example can be rewritten as
3525
3526 \d++foo
3527
3528 Possessive quantifiers are always greedy; the setting of the
3529 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3530 simpler forms of atomic group. However, there is no difference in the
3531 meaning or processing of a possessive quantifier and the equivalent
3532 atomic group.
3533
3534 The possessive quantifier syntax is an extension to the Perl syntax.
3535 Jeffrey Friedl originated the idea (and the name) in the first edition
3536 of his book. Mike McCloskey liked it, so implemented it when he built
3537 Sun's Java package, and PCRE copied it from there.
3538
3539 When a pattern contains an unlimited repeat inside a subpattern that
3540 can itself be repeated an unlimited number of times, the use of an
3541 atomic group is the only way to avoid some failing matches taking a
3542 very long time indeed. The pattern
3543
3544 (\D+|<\d+>)*[!?]
3545
3546 matches an unlimited number of substrings that either consist of non-
3547 digits, or digits enclosed in <>, followed by either ! or ?. When it
3548 matches, it runs quickly. However, if it is applied to
3549
3550 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3551
3552 it takes a long time before reporting failure. This is because the
3553 string can be divided between the internal \D+ repeat and the external
3554 * repeat in a large number of ways, and all have to be tried. (The
3555 example uses [!?] rather than a single character at the end, because
3556 both PCRE and Perl have an optimization that allows for fast failure
3557 when a single character is used. They remember the last single charac-
3558 ter that is required for a match, and fail early if it is not present
3559 in the string.) If the pattern is changed so that it uses an atomic
3560 group, like this:
3561
3562 ((?>\D+)|<\d+>)*[!?]
3563
3564 sequences of non-digits cannot be broken, and failure happens quickly.
3565
3566
3567 BACK REFERENCES
3568
3569 Outside a character class, a backslash followed by a digit greater than
3570 0 (and possibly further digits) is a back reference to a capturing sub-
3571 pattern earlier (that is, to its left) in the pattern, provided there
3572 have been that many previous capturing left parentheses.
3573
3574 However, if the decimal number following the backslash is less than 10,
3575 it is always taken as a back reference, and causes an error only if
3576 there are not that many capturing left parentheses in the entire pat-
3577 tern. In other words, the parentheses that are referenced need not be
3578 to the left of the reference for numbers less than 10. A "forward back
3579 reference" of this type can make sense when a repetition is involved
3580 and the subpattern to the right has participated in an earlier itera-
3581 tion.
3582
3583 It is not possible to have a numerical "forward back reference" to sub-
3584 pattern whose number is 10 or more. However, a back reference to any
3585 subpattern is possible using named parentheses (see below). See also
3586 the subsection entitled "Non-printing characters" above for further
3587 details of the handling of digits following a backslash.
3588
3589 A back reference matches whatever actually matched the capturing sub-
3590 pattern in the current subject string, rather than anything matching
3591 the subpattern itself (see "Subpatterns as subroutines" below for a way
3592 of doing that). So the pattern
3593
3594 (sens|respons)e and \1ibility
3595
3596 matches "sense and sensibility" and "response and responsibility", but
3597 not "sense and responsibility". If caseful matching is in force at the
3598 time of the back reference, the case of letters is relevant. For exam-
3599 ple,
3600
3601 ((?i)rah)\s+\1
3602
3603 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3604 original capturing subpattern is matched caselessly.
3605
3606 Back references to named subpatterns use the Python syntax (?P=name).
3607 We could rewrite the above example as follows:
3608
3609 (?P<p1>(?i)rah)\s+(?P=p1)
3610
3611 A subpattern that is referenced by name may appear in the pattern
3612 before or after the reference.
3613
3614 There may be more than one back reference to the same subpattern. If a
3615 subpattern has not actually been used in a particular match, any back
3616 references to it always fail. For example, the pattern
3617
3618 (a|(bc))\2
3619
3620 always fails if it starts to match "a" rather than "bc". Because there
3621 may be many capturing parentheses in a pattern, all digits following
3622 the backslash are taken as part of a potential back reference number.
3623 If the pattern continues with a digit character, some delimiter must be
3624 used to terminate the back reference. If the PCRE_EXTENDED option is
3625 set, this can be whitespace. Otherwise an empty comment (see "Com-
3626 ments" below) can be used.
3627
3628 A back reference that occurs inside the parentheses to which it refers
3629 fails when the subpattern is first used, so, for example, (a\1) never
3630 matches. However, such references can be useful inside repeated sub-
3631 patterns. For example, the pattern
3632
3633 (a|b\1)+
3634
3635 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3636 ation of the subpattern, the back reference matches the character
3637 string corresponding to the previous iteration. In order for this to
3638 work, the pattern must be such that the first iteration does not need
3639 to match the back reference. This can be done using alternation, as in
3640 the example above, or by a quantifier with a minimum of zero.
3641
3642
3643 ASSERTIONS
3644
3645 An assertion is a test on the characters following or preceding the
3646 current matching point that does not actually consume any characters.
3647 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3648 described above.
3649
3650 More complicated assertions are coded as subpatterns. There are two
3651 kinds: those that look ahead of the current position in the subject
3652 string, and those that look behind it. An assertion subpattern is
3653 matched in the normal way, except that it does not cause the current
3654 matching position to be changed.
3655
3656 Assertion subpatterns are not capturing subpatterns, and may not be
3657 repeated, because it makes no sense to assert the same thing several
3658 times. If any kind of assertion contains capturing subpatterns within
3659 it, these are counted for the purposes of numbering the capturing sub-
3660 patterns in the whole pattern. However, substring capturing is carried
3661 out only for positive assertions, because it does not make sense for
3662 negative assertions.
3663
3664 Lookahead assertions
3665
3666 Lookahead assertions start with (?= for positive assertions and (?! for
3667 negative assertions. For example,
3668
3669 \w+(?=;)
3670
3671 matches a word followed by a semicolon, but does not include the semi-
3672 colon in the match, and
3673
3674 foo(?!bar)
3675
3676 matches any occurrence of "foo" that is not followed by "bar". Note
3677 that the apparently similar pattern
3678
3679 (?!foo)bar
3680
3681 does not find an occurrence of "bar" that is preceded by something
3682 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3683 the assertion (?!foo) is always true when the next three characters are
3684 "bar". A lookbehind assertion is needed to achieve the other effect.
3685
3686 If you want to force a matching failure at some point in a pattern, the
3687 most convenient way to do it is with (?!) because an empty string
3688 always matches, so an assertion that requires there not to be an empty
3689 string must always fail.
3690
3691 Lookbehind assertions
3692
3693 Lookbehind assertions start with (?<= for positive assertions and (?<!
3694 for negative assertions. For example,
3695
3696 (?<!foo)bar
3697
3698 does find an occurrence of "bar" that is not preceded by "foo". The
3699 contents of a lookbehind assertion are restricted such that all the
3700 strings it matches must have a fixed length. However, if there are sev-
3701 eral top-level alternatives, they do not all have to have the same
3702 fixed length. Thus
3703
3704 (?<=bullock|donkey)
3705
3706 is permitted, but
3707
3708 (?<!dogs?|cats?)
3709
3710 causes an error at compile time. Branches that match different length
3711 strings are permitted only at the top level of a lookbehind assertion.
3712 This is an extension compared with Perl (at least for 5.8), which
3713 requires all branches to match the same length of string. An assertion
3714 such as
3715
3716 (?<=ab(c|de))
3717
3718 is not permitted, because its single top-level branch can match two
3719 different lengths, but it is acceptable if rewritten to use two top-
3720 level branches:
3721
3722 (?<=abc|abde)
3723
3724 The implementation of lookbehind assertions is, for each alternative,
3725 to temporarily move the current position back by the fixed width and
3726 then try to match. If there are insufficient characters before the cur-
3727 rent position, the match is deemed to fail.
3728
3729 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3730 mode) to appear in lookbehind assertions, because it makes it impossi-
3731 ble to calculate the length of the lookbehind. The \X escape, which can
3732 match different numbers of bytes, is also not permitted.
3733
3734 Atomic groups can be used in conjunction with lookbehind assertions to
3735 specify efficient matching at the end of the subject string. Consider a
3736 simple pattern such as
3737
3738 abcd$
3739
3740 when applied to a long string that does not match. Because matching
3741 proceeds from left to right, PCRE will look for each "a" in the subject
3742 and then see if what follows matches the rest of the pattern. If the
3743 pattern is specified as
3744
3745 ^.*abcd$
3746
3747 the initial .* matches the entire string at first, but when this fails
3748 (because there is no following "a"), it backtracks to match all but the
3749 last character, then all but the last two characters, and so on. Once
3750 again the search for "a" covers the entire string, from right to left,
3751 so we are no better off. However, if the pattern is written as
3752
3753 ^(?>.*)(?<=abcd)
3754
3755 or, equivalently, using the possessive quantifier syntax,
3756
3757 ^.*+(?<=abcd)
3758
3759 there can be no backtracking for the .* item; it can match only the
3760 entire string. The subsequent lookbehind assertion does a single test
3761 on the last four characters. If it fails, the match fails immediately.
3762 For long strings, this approach makes a significant difference to the
3763 processing time.
3764
3765 Using multiple assertions
3766
3767 Several assertions (of any sort) may occur in succession. For example,
3768
3769 (?<=\d{3})(?<!999)foo
3770
3771 matches "foo" preceded by three digits that are not "999". Notice that
3772 each of the assertions is applied independently at the same point in
3773 the subject string. First there is a check that the previous three
3774 characters are all digits, and then there is a check that the same
3775 three characters are not "999". This pattern does not match "foo" pre-
3776 ceded by six characters, the first of which are digits and the last
3777 three of which are not "999". For example, it doesn't match "123abc-
3778 foo". A pattern to do that is
3779
3780 (?<=\d{3}...)(?<!999)foo
3781
3782 This time the first assertion looks at the preceding six characters,
3783 checking that the first three are digits, and then the second assertion
3784 checks that the preceding three characters are not "999".
3785
3786 Assertions can be nested in any combination. For example,
3787
3788 (?<=(?<!foo)bar)baz
3789
3790 matches an occurrence of "baz" that is preceded by "bar" which in turn
3791 is not preceded by "foo", while
3792
3793 (?<=\d{3}(?!999)...)foo
3794
3795 is another pattern that matches "foo" preceded by three digits and any
3796 three characters that are not "999".
3797
3798
3799 CONDITIONAL SUBPATTERNS
3800
3801 It is possible to cause the matching process to obey a subpattern con-
3802 ditionally or to choose between two alternative subpatterns, depending
3803 on the result of an assertion, or whether a previous capturing subpat-
3804 tern matched or not. The two possible forms of conditional subpattern
3805 are
3806
3807 (?(condition)yes-pattern)
3808 (?(condition)yes-pattern|no-pattern)
3809
3810 If the condition is satisfied, the yes-pattern is used; otherwise the
3811 no-pattern (if present) is used. If there are more than two alterna-
3812 tives in the subpattern, a compile-time error occurs.
3813
3814 There are three kinds of condition. If the text between the parentheses
3815 consists of a sequence of digits, or a sequence of alphanumeric charac-
3816 ters and underscores, the condition is satisfied if the capturing sub-
3817 pattern of that number or name has previously matched. There is a pos-
3818 sible ambiguity here, because subpattern names may consist entirely of
3819 digits. PCRE looks first for a named subpattern; if it cannot find one
3820 and the text consists entirely of digits, it looks for a subpattern of
3821 that number, which must be greater than zero. Using subpattern names
3822 that consist entirely of digits is not recommended.
3823
3824 Consider the following pattern, which contains non-significant white
3825 space to make it more readable (assume the PCRE_EXTENDED option) and to
3826 divide it into three parts for ease of discussion:
3827
3828 ( \( )? [^()]+ (?(1) \) )
3829
3830 The first part matches an optional opening parenthesis, and if that
3831 character is present, sets it as the first captured substring. The sec-
3832 ond part matches one or more characters that are not parentheses. The
3833 third part is a conditional subpattern that tests whether the first set
3834 of parentheses matched or not. If they did, that is, if subject started
3835 with an opening parenthesis, the condition is true, and so the yes-pat-
3836 tern is executed and a closing parenthesis is required. Otherwise,
3837 since no-pattern is not present, the subpattern matches nothing. In
3838 other words, this pattern matches a sequence of non-parentheses,
3839 optionally enclosed in parentheses. Rewriting it to use a named subpat-
3840 tern gives this:
3841
3842 (?P<OPEN> \( )? [^()]+ (?(OPEN) \) )
3843
3844 If the condition is the string (R), and there is no subpattern with the
3845 name R, the condition is satisfied if a recursive call to the pattern
3846 or subpattern has been made. At "top level", the condition is false.
3847 This is a PCRE extension. Recursive patterns are described in the next
3848 section.
3849
3850 If the condition is not a sequence of digits or (R), it must be an
3851 assertion. This may be a positive or negative lookahead or lookbehind
3852 assertion. Consider this pattern, again containing non-significant
3853 white space, and with the two alternatives on the second line:
3854
3855 (?(?=[^a-z]*[a-z])
3856 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
3857
3858 The condition is a positive lookahead assertion that matches an
3859 optional sequence of non-letters followed by a letter. In other words,
3860 it tests for the presence of at least one letter in the subject. If a
3861 letter is found, the subject is matched against the first alternative;
3862 otherwise it is matched against the second. This pattern matches
3863 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
3864 letters and dd are digits.
3865
3866
3867 COMMENTS
3868
3869 The sequence (?# marks the start of a comment that continues up to the
3870 next closing parenthesis. Nested parentheses are not permitted. The
3871 characters that make up a comment play no part in the pattern matching
3872 at all.
3873
3874 If the PCRE_EXTENDED option is set, an unescaped # character outside a
3875 character class introduces a comment that continues to immediately
3876 after the next newline in the pattern.
3877
3878
3879 RECURSIVE PATTERNS
3880
3881 Consider the problem of matching a string in parentheses, allowing for
3882 unlimited nested parentheses. Without the use of recursion, the best
3883 that can be done is to use a pattern that matches up to some fixed
3884 depth of nesting. It is not possible to handle an arbitrary nesting
3885 depth. Perl provides a facility that allows regular expressions to
3886 recurse (amongst other things). It does this by interpolating Perl code
3887 in the expression at run time, and the code can refer to the expression
3888 itself. A Perl pattern to solve the parentheses problem can be created
3889 like this:
3890
3891 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
3892
3893 The (?p{...}) item interpolates Perl code at run time, and in this case
3894 refers recursively to the pattern in which it appears. Obviously, PCRE
3895 cannot support the interpolation of Perl code. Instead, it supports
3896 some special syntax for recursion of the entire pattern, and also for
3897 individual subpattern recursion.
3898
3899 The special item that consists of (? followed by a number greater than
3900 zero and a closing parenthesis is a recursive call of the subpattern of
3901 the given number, provided that it occurs inside that subpattern. (If
3902 not, it is a "subroutine" call, which is described in the next sec-
3903 tion.) The special item (?R) is a recursive call of the entire regular
3904 expression.
3905
3906 A recursive subpattern call is always treated as an atomic group. That
3907 is, once it has matched some of the subject string, it is never re-
3908 entered, even if it contains untried alternatives and there is a subse-
3909 quent matching failure.
3910
3911 This PCRE pattern solves the nested parentheses problem (assume the
3912 PCRE_EXTENDED option is set so that white space is ignored):
3913
3914 \( ( (?>[^()]+) | (?R) )* \)
3915
3916 First it matches an opening parenthesis. Then it matches any number of
3917 substrings which can either be a sequence of non-parentheses, or a
3918 recursive match of the pattern itself (that is, a correctly parenthe-
3919 sized substring). Finally there is a closing parenthesis.
3920
3921 If this were part of a larger pattern, you would not want to recurse
3922 the entire pattern, so instead you could use this:
3923
3924 ( \( ( (?>[^()]+) | (?1) )* \) )
3925
3926 We have put the pattern into parentheses, and caused the recursion to
3927 refer to them instead of the whole pattern. In a larger pattern, keep-
3928 ing track of parenthesis numbers can be tricky. It may be more conve-
3929 nient to use named parentheses instead. For this, PCRE uses (?P>name),
3930 which is an extension to the Python syntax that PCRE uses for named
3931 parentheses (Perl does not provide named parentheses). We could rewrite
3932 the above example as follows:
3933
3934 (?P<pn> \( ( (?>[^()]+) | (?P>pn) )* \) )
3935
3936 This particular example pattern contains nested unlimited repeats, and
3937 so the use of atomic grouping for matching strings of non-parentheses
3938 is important when applying the pattern to strings that do not match.
3939 For example, when this pattern is applied to
3940
3941 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
3942
3943 it yields "no match" quickly. However, if atomic grouping is not used,
3944 the match runs for a very long time indeed because there are so many
3945 different ways the + and * repeats can carve up the subject, and all
3946 have to be tested before failure can be reported.
3947
3948 At the end of a match, the values set for any capturing subpatterns are
3949 those from the outermost level of the recursion at which the subpattern
3950 value is set. If you want to obtain intermediate values, a callout
3951 function can be used (see the next section and the pcrecallout documen-
3952 tation). If the pattern above is matched against
3953
3954 (ab(cd)ef)
3955
3956 the value for the capturing parentheses is "ef", which is the last
3957 value taken on at the top level. If additional parentheses are added,
3958 giving
3959
3960 \( ( ( (?>[^()]+) | (?R) )* ) \)
3961 ^ ^
3962 ^ ^
3963
3964 the string they capture is "ab(cd)ef", the contents of the top level
3965 parentheses. If there are more than 15 capturing parentheses in a pat-
3966 tern, PCRE has to obtain extra memory to store data during a recursion,
3967 which it does by using pcre_malloc, freeing it via pcre_free after-
3968 wards. If no memory can be obtained, the match fails with the
3969 PCRE_ERROR_NOMEMORY error.
3970
3971 Do not confuse the (?R) item with the condition (R), which tests for
3972 recursion. Consider this pattern, which matches text in angle brack-
3973 ets, allowing for arbitrary nesting. Only digits are allowed in nested
3974 brackets (that is, when recursing), whereas any characters are permit-
3975 ted at the outer level.
3976
3977 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
3978
3979 In this pattern, (?(R) is the start of a conditional subpattern, with
3980 two different alternatives for the recursive and non-recursive cases.
3981 The (?R) item is the actual recursive call.
3982
3983
3984 SUBPATTERNS AS SUBROUTINES
3985
3986 If the syntax for a recursive subpattern reference (either by number or
3987 by name) is used outside the parentheses to which it refers, it oper-
3988 ates like a subroutine in a programming language. An earlier example
3989 pointed out that the pattern
3990
3991 (sens|respons)e and \1ibility
3992
3993 matches "sense and sensibility" and "response and responsibility", but
3994 not "sense and responsibility". If instead the pattern
3995
3996 (sens|respons)e and (?1)ibility
3997
3998 is used, it does match "sense and responsibility" as well as the other
3999 two strings. Such references, if given numerically, must follow the
4000 subpattern to which they refer. However, named references can refer to
4001 later subpatterns.
4002
4003 Like recursive subpatterns, a "subroutine" call is always treated as an
4004 atomic group. That is, once it has matched some of the subject string,
4005 it is never re-entered, even if it contains untried alternatives and
4006 there is a subsequent matching failure.
4007
4008
4009 CALLOUTS
4010
4011 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4012 Perl code to be obeyed in the middle of matching a regular expression.
4013 This makes it possible, amongst other things, to extract different sub-
4014 strings that match the same pair of parentheses when there is a repeti-
4015 tion.
4016
4017 PCRE provides a similar feature, but of course it cannot obey arbitrary
4018 Perl code. The feature is called "callout". The caller of PCRE provides
4019 an external function by putting its entry point in the global variable
4020 pcre_callout. By default, this variable contains NULL, which disables
4021 all calling out.
4022
4023 Within a regular expression, (?C) indicates the points at which the
4024 external function is to be called. If you want to identify different
4025 callout points, you can put a number less than 256 after the letter C.
4026 The default value is zero. For example, this pattern has two callout
4027 points:
4028
4029 (?C1)abc(?C2)def
4030
4031 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4032 automatically installed before each item in the pattern. They are all
4033 numbered 255.
4034
4035 During matching, when PCRE reaches a callout point (and pcre_callout is
4036 set), the external function is called. It is provided with the number
4037 of the callout, the position in the pattern, and, optionally, one item
4038 of data originally supplied by the caller of pcre_exec(). The callout
4039 function may cause matching to proceed, to backtrack, or to fail alto-
4040 gether. A complete description of the interface to the callout function
4041 is given in the pcrecallout documentation.
4042
4043 Last updated: 06 June 2006
4044 Copyright (c) 1997-2006 University of Cambridge.
4045 ------------------------------------------------------------------------------
4046
4047
4048 PCREPARTIAL(3) PCREPARTIAL(3)
4049
4050
4051 NAME
4052 PCRE - Perl-compatible regular expressions
4053
4054
4055 PARTIAL MATCHING IN PCRE
4056
4057 In normal use of PCRE, if the subject string that is passed to
4058 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4059 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4060 There are circumstances where it might be helpful to distinguish this
4061 case from other cases in which there is no match.
4062
4063 Consider, for example, an application where a human is required to type
4064 in data for a field with specific formatting requirements. An example
4065 might be a date in the form ddmmmyy, defined by this pattern:
4066
4067 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4068
4069 If the application sees the user's keystrokes one by one, and can check
4070 that what has been typed so far is potentially valid, it is able to
4071 raise an error as soon as a mistake is made, possibly beeping and not
4072 reflecting the character that has been typed. This immediate feedback
4073 is likely to be a better user interface than a check that is delayed
4074 until the entire string has been entered.
4075
4076 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4077 TIAL option, which can be set when calling pcre_exec() or
4078 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4079 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4080 during the matching process the last part of the subject string matched
4081 part of the pattern. Unfortunately, for non-anchored matching, it is
4082 not possible to obtain the position of the start of the partial match.
4083 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4084
4085 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4086 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4087 the subject is reached, there have been no complete matches, but there
4088 is still at least one matching possibility. The portion of the string
4089 that provided the partial match is set as the first matching string.
4090
4091 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4092 the last literal byte in a pattern, and abandons matching immediately
4093 if such a byte is not present in the subject string. This optimization
4094 cannot be used for a subject string that might match only partially.
4095
4096
4097 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4098
4099 Because of the way certain internal optimizations are implemented in
4100 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4101 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4102 used. For pcre_exec(), repeated single characters such as
4103
4104 a{2,4}
4105
4106 and repeated single metasequences such as
4107
4108 \d+
4109
4110 are not permitted if the maximum number of occurrences is greater than
4111 one. Optional items such as \d? (where the maximum is one) are permit-
4112 ted. Quantifiers with any values are permitted after parentheses, so
4113 the invalid examples above can be coded thus:
4114
4115 (a){2,4}
4116 (\d)+
4117
4118 These constructions run more slowly, but for the kinds of application
4119 that are envisaged for this facility, this is not felt to be a major
4120 restriction.
4121
4122 If PCRE_PARTIAL is set for a pattern that does not conform to the
4123 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4124 (-13).
4125
4126
4127 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4128
4129 If the escape sequence \P is present in a pcretest data line, the
4130 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4131 uses the date example quoted above:
4132
4133 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4134 data> 25jun04\P
4135 0: 25jun04
4136 1: jun
4137 data> 25dec3\P
4138 Partial match
4139 data> 3ju\P
4140 Partial match
4141 data> 3juj\P
4142 No match
4143 data> j\P
4144 No match
4145
4146 The first data string is matched completely, so pcretest shows the
4147 matched substrings. The remaining four strings do not match the com-
4148 plete pattern, but the first two are partial matches. The same test,
4149 using DFA matching (by means of the \D escape sequence), produces the
4150 following output:
4151
4152 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4153 data> 25jun04\P\D
4154 0: 25jun04
4155 data> 23dec3\P\D
4156 Partial match: 23dec3
4157 data> 3ju\P\D
4158 Partial match: 3ju
4159 data> 3juj\P\D
4160 No match
4161 data> j\P\D
4162 No match
4163
4164 Notice that in this case the portion of the string that was matched is
4165 made available.
4166
4167
4168 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4169
4170 When a partial match has been found using pcre_dfa_exec(), it is possi-
4171 ble to continue the match by providing additional subject data and
4172 calling pcre_dfa_exec() again with the PCRE_DFA_RESTART option and the
4173 same working space (where details of the previous partial match are
4174 stored). Here is an example using pcretest, where the \R escape
4175 sequence sets the PCRE_DFA_RESTART option and the \D escape sequence
4176 requests the use of pcre_dfa_exec():
4177
4178 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4179 data> 23ja\P\D
4180 Partial match: 23ja
4181 data> n05\R\D
4182 0: n05
4183
4184 The first call has "23ja" as the subject, and requests partial match-
4185 ing; the second call has "n05" as the subject for the continued
4186 (restarted) match. Notice that when the match is complete, only the
4187 last part is shown; PCRE does not retain the previously partially-
4188 matched string. It is up to the calling program to do that if it needs
4189 to.
4190
4191 This facility can be used to pass very long subject strings to
4192 pcre_dfa_exec(). However, some care is needed for certain types of pat-
4193 tern.
4194
4195 1. If the pattern contains tests for the beginning or end of a line,
4196 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4197 ate, when the subject string for any call does not contain the begin-
4198 ning or end of a line.
4199
4200 2. If the pattern contains backward assertions (including \b or \B),
4201 you need to arrange for some overlap in the subject strings to allow
4202 for this. For example, you could pass the subject in chunks that were
4203 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4204 set to 200 and the previous 200 bytes at the start of the buffer.
4205
4206 3. Matching a subject string that is split into multiple segments does
4207 not always produce exactly the same result as matching over one single
4208 long string. The difference arises when there are multiple matching
4209 possibilities, because a partial match result is given only when there
4210 are no completed matches in a call to fBpcre_dfa_exec(). This means
4211 that as soon as the shortest match has been found, continuation to a
4212 new subject segment is no longer possible. Consider this pcretest
4213 example:
4214
4215 re> /dog(sbody)?/
4216 data> do\P\D
4217 Partial match: do
4218 data> gsb\R\P\D
4219 0: g
4220 data> dogsbody\D
4221 0: dogsbody
4222 1: dog
4223
4224 The pattern matches the words "dog" or "dogsbody". When the subject is
4225 presented in several parts ("do" and "gsb" being the first two) the
4226 match stops when "dog" has been found, and it is not possible to con-
4227 tinue. On the other hand, if "dogsbody" is presented as a single
4228 string, both matches are found.
4229
4230 Because of this phenomenon, it does not usually make sense to end a
4231 pattern that is going to be matched in this way with a variable repeat.
4232
4233 4. Patterns that contain alternatives at the top level which do not all
4234 start with the same pattern item may not work as expected. For example,
4235 consider this pattern:
4236
4237 1234|3789
4238
4239 If the first part of the subject is "ABC123", a partial match of the
4240 first alternative is found at offset 3. There is no partial match for
4241 the second alternative, because such a match does not start at the same
4242 point in the subject string. Attempting to continue with the string
4243 "789" does not yield a match because only those alternatives that match
4244 at one point in the subject are remembered. The problem arises because
4245 the start of the second alternative matches within the first alterna-
4246 tive. There is no problem with anchored patterns or patterns such as:
4247
4248 1234|ABCD
4249
4250 where no string can be a partial match for both alternatives.
4251
4252 Last updated: 16 January 2006
4253 Copyright (c) 1997-2006 University of Cambridge.
4254 ------------------------------------------------------------------------------
4255
4256
4257 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4258
4259
4260 NAME
4261 PCRE - Perl-compatible regular expressions
4262
4263
4264 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4265
4266 If you are running an application that uses a large number of regular
4267 expression patterns, it may be useful to store them in a precompiled
4268 form instead of having to compile them every time the application is
4269 run. If you are not using any private character tables (see the
4270 pcre_maketables() documentation), this is relatively straightforward.
4271 If you are using private tables, it is a little bit more complicated.
4272
4273 If you save compiled patterns to a file, you can copy them to a differ-
4274 ent host and run them there. This works even if the new host has the
4275 opposite endianness to the one on which the patterns were compiled.
4276 There may be a small performance penalty, but it should be insignifi-
4277 cant.
4278
4279
4280 SAVING A COMPILED PATTERN
4281 The value returned by pcre_compile() points to a single block of memory
4282 that holds the compiled pattern and associated data. You can find the
4283 length of this block in bytes by calling pcre_fullinfo() with an argu-
4284 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4285 manner. Here is sample code that compiles a pattern and writes it to a
4286 file. It assumes that the variable fd refers to a file that is open for
4287 output:
4288
4289 int erroroffset, rc, size;
4290 char *error;
4291 pcre *re;
4292
4293 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4294 if (re == NULL) { ... handle errors ... }
4295 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4296 if (rc < 0) { ... handle errors ... }
4297 rc = fwrite(re, 1, size, fd);
4298 if (rc != size) { ... handle errors ... }
4299
4300 In this example, the bytes that comprise the compiled pattern are
4301 copied exactly. Note that this is binary data that may contain any of
4302 the 256 possible byte values. On systems that make a distinction
4303 between binary and non-binary data, be sure that the file is opened for
4304 binary output.
4305
4306 If you want to write more than one pattern to a file, you will have to
4307 devise a way of separating them. For binary data, preceding each pat-
4308 tern with its length is probably the most straightforward approach.
4309 Another possibility is to write out the data in hexadecimal instead of
4310 binary, one pattern to a line.
4311
4312 Saving compiled patterns in a file is only one possible way of storing
4313 them for later use. They could equally well be saved in a database, or
4314 in the memory of some daemon process that passes them via sockets to
4315 the processes that want them.
4316
4317 If the pattern has been studied, it is also possible to save the study
4318 data in a similar way to the compiled pattern itself. When studying
4319 generates additional information, pcre_study() returns a pointer to a
4320 pcre_extra data block. Its format is defined in the section on matching
4321 a pattern in the pcreapi documentation. The study_data field points to
4322 the binary study data, and this is what you must save (not the
4323 pcre_extra block itself). The length of the study data can be obtained
4324 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4325 Remember to check that pcre_study() did return a non-NULL value before
4326 trying to save the study data.
4327
4328
4329 RE-USING A PRECOMPILED PATTERN
4330
4331 Re-using a precompiled pattern is straightforward. Having reloaded it
4332 into main memory, you pass its pointer to pcre_exec() or
4333 pcre_dfa_exec() in the usual way. This should work even on another
4334 host, and even if that host has the opposite endianness to the one
4335 where the pattern was compiled.
4336
4337 However, if you passed a pointer to custom character tables when the
4338 pattern was compiled (the tableptr argument of pcre_compile()), you
4339 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4340 because the value saved with the compiled pattern will obviously be
4341 nonsense. A field in a pcre_extra() block is used to pass this data, as
4342 described in the section on matching a pattern in the pcreapi documen-
4343 tation.
4344
4345 If you did not provide custom character tables when the pattern was
4346 compiled, the pointer in the compiled pattern is NULL, which causes
4347 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4348 take any special action at run time in this case.
4349
4350 If you saved study data with the compiled pattern, you need to create
4351 your own pcre_extra data block and set the study_data field to point to
4352 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4353 bit in the flags field to indicate that study data is present. Then
4354 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4355 usual way.
4356
4357
4358 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4359
4360 The layout of the control block that is at the start of the data that
4361 makes up a compiled pattern was changed for release 5.0. If you have
4362 any saved patterns that were compiled with previous releases (not a
4363 facility that was previously advertised), you will have to recompile
4364 them for release 5.0. However, from now on, it should be possible to
4365 make changes in a compatible manner.
4366
4367 Notwithstanding the above, if you have any saved patterns in UTF-8 mode
4368 that use \p or \P that were compiled with any release up to and includ-
4369 ing 6.4, you will have to recompile them for release 6.5 and above.
4370
4371 Last updated: 01 February 2006
4372 Copyright (c) 1997-2006 University of Cambridge.
4373 ------------------------------------------------------------------------------
4374
4375
4376 PCREPERFORM(3) PCREPERFORM(3)
4377
4378
4379 NAME
4380 PCRE - Perl-compatible regular expressions
4381
4382
4383 PCRE PERFORMANCE
4384
4385 Certain items that may appear in regular expression patterns are more
4386 efficient than others. It is more efficient to use a character class
4387 like [aeiou] than a set of alternatives such as (a|e|i|o|u). In gen-
4388 eral, the simplest construction that provides the required behaviour is
4389 usually the most efficient. Jeffrey Friedl's book contains a lot of
4390 useful general discussion about optimizing regular expressions for
4391 efficient performance. This document contains a few observations about
4392 PCRE.
4393
4394 Using Unicode character properties (the \p, \P, and \X escapes) is
4395 slow, because PCRE has to scan a structure that contains data for over
4396 fifteen thousand characters whenever it needs a character's property.
4397 If you can find an alternative pattern that does not use character
4398 properties, it will probably be faster.
4399
4400 When a pattern begins with .* not in parentheses, or in parentheses
4401 that are not the subject of a backreference, and the PCRE_DOTALL option
4402 is set, the pattern is implicitly anchored by PCRE, since it can match
4403 only at the start of a subject string. However, if PCRE_DOTALL is not
4404 set, PCRE cannot make this optimization, because the . metacharacter
4405 does not then match a newline, and if the subject string contains new-
4406 lines, the pattern may match from the character immediately following
4407 one of them instead of from the very start. For example, the pattern
4408
4409 .*second
4410
4411 matches the subject "first\nand second" (where \n stands for a newline
4412 character), with the match starting at the seventh character. In order
4413 to do this, PCRE has to retry the match starting after every newline in
4414 the subject.
4415
4416 If you are using such a pattern with subject strings that do not con-
4417 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4418 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4419 ing. That saves PCRE from having to scan along the subject looking for
4420 a newline to restart at.
4421
4422 Beware of patterns that contain nested indefinite repeats. These can
4423 take a long time to run when applied to a string that does not match.
4424 Consider the pattern fragment
4425
4426 (a+)*
4427
4428 This can match "aaaa" in 33 different ways, and this number increases
4429 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4430 2, 3, or 4 times, and for each of those cases other than 0, the +
4431 repeats can match different numbers of times.) When the remainder of
4432 the pattern is such that the entire match is going to fail, PCRE has in
4433 principle to try every possible variation, and this can take an
4434 extremely long time.
4435
4436 An optimization catches some of the more simple cases such as
4437
4438 (a+)*b
4439
4440 where a literal character follows. Before embarking on the standard
4441 matching procedure, PCRE checks that there is a "b" later in the sub-
4442 ject string, and if there is not, it fails the match immediately. How-
4443 ever, when there is no following literal this optimization cannot be
4444 used. You can see the difference by comparing the behaviour of
4445
4446 (a+)*\d
4447
4448 with the pattern above. The former gives a failure almost instantly
4449 when applied to a whole line of "a" characters, whereas the latter
4450 takes an appreciable time with strings longer than about 20 characters.
4451
4452 In many cases, the solution to this kind of performance issue is to use
4453 an atomic group or a possessive quantifier.
4454
4455 Last updated: 28 February 2005
4456 Copyright (c) 1997-2005 University of Cambridge.
4457 ------------------------------------------------------------------------------
4458
4459
4460 PCREPOSIX(3) PCREPOSIX(3)
4461
4462
4463 NAME
4464 PCRE - Perl-compatible regular expressions.
4465
4466
4467 SYNOPSIS OF POSIX API
4468
4469 #include <pcreposix.h>
4470
4471 int regcomp(regex_t *preg, const char *pattern,
4472 int cflags);
4473
4474 int regexec(regex_t *preg, const char *string,
4475 size_t nmatch, regmatch_t pmatch[], int eflags);
4476
4477 size_t regerror(int errcode, const regex_t *preg,
4478 char *errbuf, size_t errbuf_size);
4479
4480 void regfree(regex_t *preg);
4481
4482
4483 DESCRIPTION
4484
4485 This set of functions provides a POSIX-style API to the PCRE regular
4486 expression package. See the pcreapi documentation for a description of
4487 PCRE's native API, which contains much additional functionality.
4488
4489 The functions described here are just wrapper functions that ultimately
4490 call the PCRE native API. Their prototypes are defined in the
4491 pcreposix.h header file, and on Unix systems the library itself is
4492 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4493 command for linking an application that uses them. Because the POSIX
4494 functions call the native ones, it is also necessary to add -lpcre.
4495
4496 I have implemented only those option bits that can be reasonably mapped
4497 to PCRE native options. In addition, the option REG_EXTENDED is defined
4498 with the value zero. This has no effect, but since programs that are
4499 written to the POSIX interface often use it, this makes it easier to
4500 slot in PCRE as a replacement library. Other POSIX options are not even
4501 defined.
4502
4503 When PCRE is called via these functions, it is only the API that is
4504 POSIX-like in style. The syntax and semantics of the regular expres-
4505 sions themselves are still those of Perl, subject to the setting of
4506 various PCRE options, as described below. "POSIX-like in style" means
4507 that the API approximates to the POSIX definition; it is not fully
4508 POSIX-compatible, and in multi-byte encoding domains it is probably
4509 even less compatible.
4510
4511 The header for these functions is supplied as pcreposix.h to avoid any
4512 potential clash with other POSIX libraries. It can, of course, be
4513 renamed or aliased as regex.h, which is the "correct" name. It provides
4514 two structure types, regex_t for compiled internal forms, and reg-
4515 match_t for returning captured substrings. It also defines some con-
4516 stants whose names start with "REG_"; these are used for setting
4517 options and identifying error codes.
4518
4519
4520 COMPILING A PATTERN
4521
4522 The function regcomp() is called to compile a pattern into an internal
4523 form. The pattern is a C string terminated by a binary zero, and is
4524 passed in the argument pattern. The preg argument is a pointer to a
4525 regex_t structure that is used as a base for storing information about
4526 the compiled regular expression.
4527
4528 The argument cflags is either zero, or contains one or more of the bits
4529 defined by the following macros:
4530
4531 REG_DOTALL
4532
4533 The PCRE_DOTALL option is set when the regular expression is passed for
4534 compilation to the native function. Note that REG_DOTALL is not part of
4535 the POSIX standard.
4536
4537 REG_ICASE
4538
4539 The PCRE_CASELESS option is set when the regular expression is passed
4540 for compilation to the native function.
4541
4542 REG_NEWLINE
4543
4544 The PCRE_MULTILINE option is set when the regular expression is passed
4545 for compilation to the native function. Note that this does not mimic
4546 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
4547 tion).
4548
4549 REG_NOSUB
4550
4551 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
4552 passed for compilation to the native function. In addition, when a pat-
4553 tern that is compiled with this flag is passed to regexec() for match-
4554 ing, the nmatch and pmatch arguments are ignored, and no captured
4555 strings are returned.
4556
4557 REG_UTF8
4558
4559 The PCRE_UTF8 option is set when the regular expression is passed for
4560 compilation to the native function. This causes the pattern itself and
4561 all data strings used for matching it to be treated as UTF-8 strings.
4562 Note that REG_UTF8 is not part of the POSIX standard.
4563
4564 In the absence of these flags, no options are passed to the native
4565 function. This means the the regex is compiled with PCRE default
4566 semantics. In particular, the way it handles newline characters in the
4567 subject string is the Perl way, not the POSIX way. Note that setting
4568 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4569 It does not affect the way newlines are matched by . (they aren't) or
4570 by a negative class such as [^a] (they are).
4571
4572 The yield of regcomp() is zero on success, and non-zero otherwise. The
4573 preg structure is filled in on success, and one member of the structure
4574 is public: re_nsub contains the number of capturing subpatterns in the
4575 regular expression. Various error codes are defined in the header file.
4576
4577
4578 MATCHING NEWLINE CHARACTERS
4579
4580 This area is not simple, because POSIX and Perl take different views of
4581 things. It is not possible to get PCRE to obey POSIX semantics, but
4582 then PCRE was never intended to be a POSIX engine. The following table
4583 lists the different possibilities for matching newline characters in
4584 PCRE:
4585
4586 Default Change with
4587
4588 . matches newline no PCRE_DOTALL
4589 newline matches [^a] yes not changeable
4590 $ matches \n at end yes PCRE_DOLLARENDONLY
4591 $ matches \n in middle no PCRE_MULTILINE
4592 ^ matches \n in middle no PCRE_MULTILINE
4593
4594 This is the equivalent table for POSIX:
4595
4596 Default Change with
4597
4598 . matches newline yes REG_NEWLINE
4599 newline matches [^a] yes REG_NEWLINE
4600 $ matches \n at end no REG_NEWLINE
4601 $ matches \n in middle no REG_NEWLINE
4602 ^ matches \n in middle no REG_NEWLINE
4603
4604 PCRE's behaviour is the same as Perl's, except that there is no equiva-
4605 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
4606 no way to stop newline from matching [^a].
4607
4608 The default POSIX newline handling can be obtained by setting
4609 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
4610 behave exactly as for the REG_NEWLINE action.
4611
4612
4613 MATCHING A PATTERN
4614
4615 The function regexec() is called to match a compiled pattern preg
4616 against a given string, which is terminated by a zero byte, subject to
4617 the options in eflags. These can be:
4618
4619 REG_NOTBOL
4620
4621 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
4622 function.
4623
4624 REG_NOTEOL
4625
4626 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
4627 function.
4628
4629 If the pattern was compiled with the REG_NOSUB flag, no data about any
4630 matched strings is returned. The nmatch and pmatch arguments of
4631 regexec() are ignored.
4632
4633 Otherwise,the portion of the string that was matched, and also any cap-
4634 tured substrings, are returned via the pmatch argument, which points to
4635 an array of nmatch structures of type regmatch_t, containing the mem-
4636 bers rm_so and rm_eo. These contain the offset to the first character
4637 of each substring and the offset to the first character after the end
4638 of each substring, respectively. The 0th element of the vector relates
4639 to the entire portion of string that was matched; subsequent elements
4640 relate to the capturing subpatterns of the regular expression. Unused
4641 entries in the array have both structure members set to -1.
4642
4643 A successful match yields a zero return; various error codes are
4644 defined in the header file, of which REG_NOMATCH is the "expected"
4645 failure code.
4646
4647
4648 ERROR MESSAGES
4649
4650 The regerror() function maps a non-zero errorcode from either regcomp()
4651 or regexec() to a printable message. If preg is not NULL, the error
4652 should have arisen from the use of that structure. A message terminated
4653 by a binary zero is placed in errbuf. The length of the message,
4654 including the zero, is limited to errbuf_size. The yield of the func-
4655 tion is the size of buffer needed to hold the whole message.
4656
4657
4658 MEMORY USAGE
4659
4660 Compiling a regular expression causes memory to be allocated and asso-
4661 ciated with the preg structure. The function regfree() frees all such
4662 memory, after which preg may no longer be used as a compiled expres-
4663 sion.
4664
4665
4666 AUTHOR
4667
4668 Philip Hazel
4669 University Computing Service,
4670 Cambridge CB2 3QG, England.
4671
4672 Last updated: 16 January 2006
4673 Copyright (c) 1997-2006 University of Cambridge.
4674 ------------------------------------------------------------------------------
4675
4676
4677 PCRECPP(3) PCRECPP(3)
4678
4679
4680 NAME
4681 PCRE - Perl-compatible regular expressions.
4682
4683
4684 SYNOPSIS OF C++ WRAPPER
4685
4686 #include <pcrecpp.h>
4687
4688
4689 DESCRIPTION
4690
4691 The C++ wrapper for PCRE was provided by Google Inc. Some additional
4692 functionality was added by Giuseppe Maxia. This brief man page was con-
4693 structed from the notes in the pcrecpp.h file, which should be con-
4694 sulted for further details.
4695
4696
4697 MATCHING INTERFACE
4698
4699 The "FullMatch" operation checks that supplied text matches a supplied
4700 pattern exactly. If pointer arguments are supplied, it copies matched
4701 sub-strings that match sub-patterns into them.
4702
4703 Example: successful match
4704 pcrecpp::RE re("h.*o");
4705 re.FullMatch("hello");
4706
4707 Example: unsuccessful match (requires full match):
4708 pcrecpp::RE re("e");
4709 !re.FullMatch("hello");
4710
4711 Example: creating a temporary RE object:
4712 pcrecpp::RE("h.*o").FullMatch("hello");
4713
4714 You can pass in a "const char*" or a "string" for "text". The examples
4715 below tend to use a const char*. You can, as in the different examples
4716 above, store the RE object explicitly in a variable or use a temporary
4717 RE object. The examples below use one mode or the other arbitrarily.
4718 Either could correctly be used for any of these examples.
4719
4720 You must supply extra pointer arguments to extract matched subpieces.
4721
4722 Example: extracts "ruby" into "s" and 1234 into "i"
4723 int i;
4724 string s;
4725 pcrecpp::RE re("(\\w+):(\\d+)");
4726 re.FullMatch("ruby:1234", &s, &i);
4727
4728 Example: does not try to extract any extra sub-patterns
4729 re.FullMatch("ruby:1234", &s);
4730
4731 Example: does not try to extract into NULL
4732 re.FullMatch("ruby:1234", NULL, &i);
4733
4734 Example: integer overflow causes failure
4735 !re.FullMatch("ruby:1234567891234", NULL, &i);
4736
4737 Example: fails because there aren't enough sub-patterns:
4738 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
4739
4740 Example: fails because string cannot be stored in integer
4741 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
4742
4743 The provided pointer arguments can be pointers to any scalar numeric
4744 type, or one of:
4745
4746 string (matched piece is copied to string)
4747 StringPiece (StringPiece is mutated to point to matched piece)
4748 T (where "bool T::ParseFrom(const char*, int)" exists)
4749 NULL (the corresponding matched sub-pattern is not copied)
4750
4751 The function returns true iff all of the following conditions are sat-
4752 isfied:
4753
4754 a. "text" matches "pattern" exactly;
4755
4756 b. The number of matched sub-patterns is >= number of supplied
4757 pointers;
4758
4759 c. The "i"th argument has a suitable type for holding the
4760 string captured as the "i"th sub-pattern. If you pass in
4761 NULL for the "i"th argument, or pass fewer arguments than
4762 number of sub-patterns, "i"th captured sub-pattern is
4763 ignored.
4764
4765 The matching interface supports at most 16 arguments per call. If you
4766 need more, consider using the more general interface
4767 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
4768
4769
4770 PARTIAL MATCHES
4771
4772 You can use the "PartialMatch" operation when you want the pattern to
4773 match any substring of the text.
4774
4775 Example: simple search for a string:
4776 pcrecpp::RE("ell").PartialMatch("hello");
4777
4778 Example: find first number in a string:
4779 int number;
4780 pcrecpp::RE re("(\\d+)");
4781 re.PartialMatch("x*100 + 20", &number);
4782 assert(number == 100);
4783
4784
4785 UTF-8 AND THE MATCHING INTERFACE
4786
4787 By default, pattern and text are plain text, one byte per character.
4788 The UTF8 flag, passed to the constructor, causes both pattern and
4789 string to be treated as UTF-8 text, still a byte stream but potentially
4790 multiple bytes per character. In practice, the text is likelier to be
4791 UTF-8 than the pattern, but the match returned may depend on the UTF8
4792 flag, so always use it when matching UTF8 text. For example, "." will
4793 match one byte normally but with UTF8 set may match up to three bytes
4794 of a multi-byte character.
4795
4796 Example:
4797 pcrecpp::RE_Options options;
4798 options.set_utf8();
4799 pcrecpp::RE re(utf8_pattern, options);
4800 re.FullMatch(utf8_string);
4801
4802 Example: using the convenience function UTF8():
4803 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
4804 re.FullMatch(utf8_string);
4805
4806 NOTE: The UTF8 flag is ignored if pcre was not configured with the
4807 --enable-utf8 flag.
4808
4809
4810 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
4811
4812 PCRE defines some modifiers to change the behavior of the regular
4813 expression engine. The C++ wrapper defines an auxiliary class,
4814 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
4815 rently, the following modifiers are supported:
4816
4817 modifier description Perl corresponding
4818
4819 PCRE_CASELESS case insensitive match /i
4820 PCRE_MULTILINE multiple lines match /m
4821 PCRE_DOTALL dot matches newlines /s
4822 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
4823 PCRE_EXTRA strict escape parsing N/A
4824 PCRE_EXTENDED ignore whitespaces /x
4825 PCRE_UTF8 handles UTF8 chars built-in
4826 PCRE_UNGREEDY reverses * and *? N/A
4827 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
4828
4829 (*) Both Perl and PCRE allow non capturing parentheses by means of the
4830 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
4831 ture, while (ab|cd) does.
4832
4833 For a full account on how each modifier works, please check the PCRE
4834 API reference page.
4835
4836 For each modifier, there are two member functions whose name is made
4837 out of the modifier in lowercase, without the "PCRE_" prefix. For
4838 instance, PCRE_CASELESS is handled by
4839
4840 bool caseless()
4841
4842 which returns true if the modifier is set, and
4843
4844 RE_Options & set_caseless(bool)
4845
4846 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
4847 be accessed through the set_match_limit() and match_limit() member
4848 functions. Setting match_limit to a non-zero value will limit the exe-
4849 cution of pcre to keep it from doing bad things like blowing the stack
4850 or taking an eternity to return a result. A value of 5000 is good
4851 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
4852 to zero disables match limiting. Alternatively, you can call
4853 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
4854 limit how much PCRE recurses. match_limit() limits the number of
4855 matches PCRE does; match_limit_recursion() limits the depth of internal
4856 recursion, and therefore the amount of stack that is used.
4857
4858 Normally, to pass one or more modifiers to a RE class, you declare a
4859 RE_Options object, set the appropriate options, and pass this object to
4860 a RE constructor. Example:
4861
4862 RE_options opt;
4863 opt.set_caseless(true);
4864 if (RE("HELLO", opt).PartialMatch("hello world")) ...
4865
4866 RE_options has two constructors. The default constructor takes no argu-
4867 ments and creates a set of flags that are off by default. The optional
4868 parameter option_flags is to facilitate transfer of legacy code from C
4869 programs. This lets you do
4870
4871 RE(pattern,
4872 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
4873
4874 However, new code is better off doing
4875
4876 RE(pattern,
4877 RE_Options().set_caseless(true).set_multiline(true))
4878 .PartialMatch(str);
4879
4880 If you are going to pass one of the most used modifiers, there are some
4881 convenience functions that return a RE_Options class with the appropri-
4882 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
4883 and EXTENDED().
4884
4885 If you need to set several options at once, and you don't want to go
4886 through the pains of declaring a RE_Options object and setting several
4887 options, there is a parallel method that give you such ability on the
4888 fly. You can concatenate several set_xxxxx() member functions, since
4889 each of them returns a reference to its class object. For example, to
4890 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
4891 statement, you may write:
4892
4893 RE(" ^ xyz \\s+ .* blah$",
4894 RE_Options()
4895 .set_caseless(true)
4896 .set_extended(true)
4897 .set_multiline(true)).PartialMatch(sometext);
4898
4899
4900 SCANNING TEXT INCREMENTALLY
4901
4902 The "Consume" operation may be useful if you want to repeatedly match
4903 regular expressions at the front of a string and skip over them as they
4904 match. This requires use of the "StringPiece" type, which represents a
4905 sub-range of a real string. Like RE, StringPiece is defined in the
4906 pcrecpp namespace.
4907
4908 Example: read lines of the form "var = value" from a string.
4909 string contents = ...; // Fill string somehow
4910 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
4911
4912 string var;
4913 int value;
4914 pcrecpp::RE re("(\\w+) = (\\d+)\n");
4915 while (re.Consume(&input, &var, &value)) {
4916 ...;
4917 }
4918
4919 Each successful call to "Consume" will set "var/value", and also
4920 advance "input" so it points past the matched text.
4921
4922 The "FindAndConsume" operation is similar to "Consume" but does not
4923 anchor your match at the beginning of the string. For example, you
4924 could extract all words from a string by repeatedly calling
4925
4926 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
4927
4928
4929 PARSING HEX/OCTAL/C-RADIX NUMBERS
4930
4931 By default, if you pass a pointer to a numeric value, the corresponding
4932 text is interpreted as a base-10 number. You can instead wrap the
4933 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
4934 to interpret the text in another base. The CRadix operator interprets
4935 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
4936 base-10.
4937
4938 Example:
4939 int a, b, c, d;
4940 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
4941 re.FullMatch("100 40 0100 0x40",
4942 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
4943 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
4944
4945 will leave 64 in a, b, c, and d.
4946
4947
4948 REPLACING PARTS OF STRINGS
4949
4950 You can replace the first match of "pattern" in "str" with "rewrite".
4951 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
4952 insert text matching corresponding parenthesized group from the pat-
4953 tern. \0 in "rewrite" refers to the entire matching text. For example:
4954
4955 string s = "yabba dabba doo";
4956 pcrecpp::RE("b+").Replace("d", &s);
4957
4958 will leave "s" containing "yada dabba doo". The result is true if the
4959 pattern matches and a replacement occurs, false otherwise.
4960
4961 GlobalReplace is like Replace except that it replaces all occurrences
4962 of the pattern in the string with the rewrite. Replacements are not
4963 subject to re-matching. For example:
4964
4965 string s = "yabba dabba doo";
4966 pcrecpp::RE("b+").GlobalReplace("d", &s);
4967
4968 will leave "s" containing "yada dada doo". It returns the number of
4969 replacements made.
4970
4971 Extract is like Replace, except that if the pattern matches, "rewrite"
4972 is copied into "out" (an additional argument) with substitutions. The
4973 non-matching portions of "text" are ignored. Returns true iff a match
4974 occurred and the extraction happened successfully; if no match occurs,
4975 the string is left unaffected.
4976
4977
4978 AUTHOR
4979
4980 The C++ wrapper was contributed by Google Inc.
4981 Copyright (c) 2005 Google Inc.
4982 ------------------------------------------------------------------------------
4983
4984
4985 PCRESAMPLE(3) PCRESAMPLE(3)
4986
4987
4988 NAME
4989 PCRE - Perl-compatible regular expressions
4990
4991
4992 PCRE SAMPLE PROGRAM
4993
4994 A simple, complete demonstration program, to get you started with using
4995 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
4996
4997 The program compiles the regular expression that is its first argument,
4998 and matches it against the subject string in its second argument. No
4999 PCRE options are set, and default character tables are used. If match-
5000 ing succeeds, the program outputs the portion of the subject that
5001 matched, together with the contents of any captured substrings.
5002
5003 If the -g option is given on the command line, the program then goes on
5004 to check for further matches of the same regular expression in the same
5005 subject string. The logic is a little bit tricky because of the possi-
5006 bility of matching an empty string. Comments in the code explain what
5007 is going on.
5008
5009 If PCRE is installed in the standard include and library directories
5010 for your system, you should be able to compile the demonstration pro-
5011 gram using this command:
5012
5013 gcc -o pcredemo pcredemo.c -lpcre
5014
5015 If PCRE is installed elsewhere, you may need to add additional options
5016 to the command line. For example, on a Unix-like system that has PCRE
5017 installed in /usr/local, you can compile the demonstration program
5018 using a command like this:
5019
5020 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5021 -L/usr/local/lib -lpcre
5022
5023 Once you have compiled the demonstration program, you can run simple
5024 tests like this:
5025
5026 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5027 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5028
5029 Note that there is a much more comprehensive test program, called
5030 pcretest, which supports many more facilities for testing regular
5031 expressions and the PCRE library. The pcredemo program is provided as a
5032 simple coding example.
5033
5034 On some operating systems (e.g. Solaris), when PCRE is not installed in
5035 the standard library directory, you may get an error like this when you
5036 try to run pcredemo:
5037
5038 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5039 directory
5040
5041 This is caused by the way shared library support works on those sys-
5042 tems. You need to add
5043
5044 -R/usr/local/lib
5045
5046 (for example) to the compile command to get round this problem.
5047
5048 Last updated: 09 September 2004
5049 Copyright (c) 1997-2004 University of Cambridge.
5050 ------------------------------------------------------------------------------
5051 PCRESTACK(3) PCRESTACK(3)
5052
5053
5054 NAME
5055 PCRE - Perl-compatible regular expressions
5056
5057
5058 PCRE DISCUSSION OF STACK USAGE
5059
5060 When you call pcre_exec(), it makes use of an internal function called
5061 match(). This calls itself recursively at branch points in the pattern,
5062 in order to remember the state of the match so that it can back up and
5063 try a different alternative if the first one fails. As matching pro-
5064 ceeds deeper and deeper into the tree of possibilities, the recursion
5065 depth increases.
5066
5067 Not all calls of match() increase the recursion depth; for an item such
5068 as a* it may be called several times at the same level, after matching
5069 different numbers of a's. Furthermore, in a number of cases where the
5070 result of the recursive call would immediately be passed back as the
5071 result of the current call (a "tail recursion"), the function is just
5072 restarted instead.
5073
5074 The pcre_dfa_exec() function operates in an entirely different way, and
5075 hardly uses recursion at all. The limit on its complexity is the amount
5076 of workspace it is given. The comments that follow do NOT apply to
5077 pcre_dfa_exec(); they are relevant only for pcre_exec().
5078
5079 You can set limits on the number of times that match() is called, both
5080 in total and recursively. If the limit is exceeded, an error occurs.
5081 For details, see the section on extra data for pcre_exec() in the
5082 pcreapi documentation.
5083
5084 Each time that match() is actually called recursively, it uses memory
5085 from the process stack. For certain kinds of pattern and data, very
5086 large amounts of stack may be needed, despite the recognition of "tail
5087 recursion". You can often reduce the amount of recursion, and there-
5088 fore the amount of stack used, by modifying the pattern that is being
5089 matched. Consider, for example, this pattern:
5090
5091 ([^<]|<(?!inet))+
5092
5093 It matches from wherever it starts until it encounters "<inet" or the
5094 end of the data, and is the kind of pattern that might be used when
5095 processing an XML file. Each iteration of the outer parentheses matches
5096 either one character that is not "<" or a "<" that is not followed by
5097 "inet". However, each time a parenthesis is processed, a recursion
5098 occurs, so this formulation uses a stack frame for each matched charac-
5099 ter. For a long string, a lot of stack is required. Consider now this
5100 rewritten pattern, which matches exactly the same strings:
5101
5102 ([^<]++|<(?!inet))
5103
5104 This uses very much less stack, because runs of characters that do not
5105 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5106 sion happens only when a "<" character that is not followed by "inet"
5107 is encountered (and we assume this is relatively rare). A possessive
5108 quantifier is used to stop any backtracking into the runs of non-"<"
5109 characters, but that is not related to stack usage.
5110
5111 In environments where stack memory is constrained, you might want to
5112 compile PCRE to use heap memory instead of stack for remembering back-
5113 up points. This makes it run a lot more slowly, however. Details of how
5114 to do this are given in the pcrebuild documentation.
5115
5116 In Unix-like environments, there is not often a problem with the stack,
5117 though the default limit on stack size varies from system to system.
5118 Values from 8Mb to 64Mb are common. You can find your default limit by
5119 running the command:
5120
5121 ulimit -s
5122
5123 The effect of running out of stack is often SIGSEGV, though sometimes
5124 an error message is given. You can normally increase the limit on stack
5125 size by code such as this:
5126
5127 struct rlimit rlim;
5128 getrlimit(RLIMIT_STACK, &rlim);
5129 rlim.rlim_cur = 100*1024*1024;
5130 setrlimit(RLIMIT_STACK, &rlim);
5131
5132 This reads the current limits (soft and hard) using getrlimit(), then
5133 attempts to increase the soft limit to 100Mb using setrlimit(). You
5134 must do this before calling pcre_exec().
5135
5136 PCRE has an internal counter that can be used to limit the depth of
5137 recursion, and thus cause pcre_exec() to give an error code before it
5138 runs out of stack. By default, the limit is very large, and unlikely
5139 ever to operate. It can be changed when PCRE is built, and it can also
5140 be set when pcre_exec() is called. For details of these interfaces, see
5141 the pcrebuild and pcreapi documentation.
5142
5143 As a very rough rule of thumb, you should reckon on about 500 bytes per
5144 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5145 should set the limit at 16000 recursions. A 64Mb stack, on the other
5146 hand, can support around 128000 recursions. The pcretest test program
5147 has a command line option (-S) that can be used to increase its stack.
5148
5149 Last updated: 29 June 2006
5150 Copyright (c) 1997-2006 University of Cambridge.
5151 ------------------------------------------------------------------------------
5152
5153

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12