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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 150 - (show annotations) (download)
Tue Apr 17 08:22:40 2007 UTC (7 years, 7 months ago) by ph10
File MIME type: text/plain
File size: 266746 byte(s)
Update HTML documentation.

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. (Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages.
49
50 Some features of PCRE can be included, excluded, or changed when the
51 library is built. The pcre_config() function makes it possible for a
52 client to discover which features are available. The features them-
53 selves are described in the pcrebuild page. Documentation about build-
54 ing PCRE for various operating systems can be found in the README file
55 in the source distribution.
56
57 The library contains a number of undocumented internal functions and
58 data tables that are used by more than one of the exported external
59 functions, but which are not intended for use by external callers.
60 Their names all begin with "_pcre_", which hopefully will not provoke
61 any name clashes. In some environments, it is possible to control which
62 external symbols are exported when a shared library is built, and in
63 these cases the undocumented symbols are not exported.
64
65
66 USER DOCUMENTATION
67
68 The user documentation for PCRE comprises a number of different sec-
69 tions. In the "man" format, each of these is a separate "man page". In
70 the HTML format, each is a separate page, linked from the index page.
71 In the plain text format, all the sections are concatenated, for ease
72 of searching. The sections are as follows:
73
74 pcre this document
75 pcreapi details of PCRE's native C API
76 pcrebuild options for building PCRE
77 pcrecallout details of the callout feature
78 pcrecompat discussion of Perl compatibility
79 pcrecpp details of the C++ wrapper
80 pcregrep description of the pcregrep command
81 pcrematching discussion of the two matching algorithms
82 pcrepartial details of the partial matching facility
83 pcrepattern syntax and semantics of supported
84 regular expressions
85 pcreperform discussion of performance issues
86 pcreposix the POSIX-compatible C API
87 pcreprecompile details of saving and re-using precompiled patterns
88 pcresample discussion of the sample program
89 pcrestack discussion of stack usage
90 pcretest description of the pcretest testing command
91
92 In addition, in the "man" and HTML formats, there is a short page for
93 each C library function, listing its arguments and results.
94
95
96 LIMITATIONS
97
98 There are some size limitations in PCRE but it is hoped that they will
99 never in practice be relevant.
100
101 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
102 is compiled with the default internal linkage size of 2. If you want to
103 process regular expressions that are truly enormous, you can compile
104 PCRE with an internal linkage size of 3 or 4 (see the README file in
105 the source distribution and the pcrebuild documentation for details).
106 In these cases the limit is substantially larger. However, the speed
107 of execution is slower.
108
109 All values in repeating quantifiers must be less than 65536. The maxi-
110 mum compiled length of subpattern with an explicit repeat count is
111 30000 bytes. The maximum number of capturing subpatterns is 65535.
112
113 There is no limit to the number of parenthesized subpatterns, but there
114 can be no more than 65535 capturing subpatterns.
115
116 The maximum length of name for a named subpattern is 32 characters, and
117 the maximum number of named subpatterns is 10000.
118
119 The maximum length of a subject string is the largest positive number
120 that an integer variable can hold. However, when using the traditional
121 matching function, PCRE uses recursion to handle subpatterns and indef-
122 inite repetition. This means that the available stack space may limit
123 the size of a subject string that can be processed by certain patterns.
124 For a discussion of stack issues, see the pcrestack documentation.
125
126
127 UTF-8 AND UNICODE PROPERTY SUPPORT
128
129 From release 3.3, PCRE has had some support for character strings
130 encoded in the UTF-8 format. For release 4.0 this was greatly extended
131 to cover most common requirements, and in release 5.0 additional sup-
132 port for Unicode general category properties was added.
133
134 In order process UTF-8 strings, you must build PCRE to include UTF-8
135 support in the code, and, in addition, you must call pcre_compile()
136 with the PCRE_UTF8 option flag. When you do this, both the pattern and
137 any subject strings that are matched against it are treated as UTF-8
138 strings instead of just strings of bytes.
139
140 If you compile PCRE with UTF-8 support, but do not use it at run time,
141 the library will be a bit bigger, but the additional run time overhead
142 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
143 very big.
144
145 If PCRE is built with Unicode character property support (which implies
146 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
147 ported. The available properties that can be tested are limited to the
148 general category properties such as Lu for an upper case letter or Nd
149 for a decimal number, the Unicode script names such as Arabic or Han,
150 and the derived properties Any and L&. A full list is given in the
151 pcrepattern documentation. Only the short names for properties are sup-
152 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
153 ter}, is not supported. Furthermore, in Perl, many properties may
154 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
155 does not support this.
156
157 The following comments apply when PCRE is running in UTF-8 mode:
158
159 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
160 subjects are checked for validity on entry to the relevant functions.
161 If an invalid UTF-8 string is passed, an error return is given. In some
162 situations, you may already know that your strings are valid, and
163 therefore want to skip these checks in order to improve performance. If
164 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
165 PCRE assumes that the pattern or subject it is given (respectively)
166 contains only valid UTF-8 codes. In this case, it does not diagnose an
167 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
168 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
169 crash.
170
171 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
172 two-byte UTF-8 character if the value is greater than 127.
173
174 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
175 characters for values greater than \177.
176
177 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
178 vidual bytes, for example: \x{100}{3}.
179
180 5. The dot metacharacter matches one UTF-8 character instead of a sin-
181 gle byte.
182
183 6. The escape sequence \C can be used to match a single byte in UTF-8
184 mode, but its use can lead to some strange effects. This facility is
185 not available in the alternative matching function, pcre_dfa_exec().
186
187 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
188 test characters of any code value, but the characters that PCRE recog-
189 nizes as digits, spaces, or word characters remain the same set as
190 before, all with values less than 256. This remains true even when PCRE
191 includes Unicode property support, because to do otherwise would slow
192 down PCRE in many common cases. If you really want to test for a wider
193 sense of, say, "digit", you must use Unicode property tests such as
194 \p{Nd}.
195
196 8. Similarly, characters that match the POSIX named character classes
197 are all low-valued characters.
198
199 9. Case-insensitive matching applies only to characters whose values
200 are less than 128, unless PCRE is built with Unicode property support.
201 Even when Unicode property support is available, PCRE still uses its
202 own character tables when checking the case of low-valued characters,
203 so as not to degrade performance. The Unicode property information is
204 used only for characters with higher values. Even when Unicode property
205 support is available, PCRE supports case-insensitive matching only when
206 there is a one-to-one mapping between a letter's cases. There are a
207 small number of many-to-one mappings in Unicode; these are not sup-
208 ported by PCRE.
209
210
211 AUTHOR
212
213 Philip Hazel
214 University Computing Service
215 Cambridge CB2 3QH, England.
216
217 Putting an actual email address here seems to have been a spam magnet,
218 so I've taken it away. If you want to email me, use my initial and sur-
219 name, separated by a dot, at the domain ucs.cam.ac.uk.
220
221
222 REVISION
223
224 Last updated: 06 March 2007
225 Copyright (c) 1997-2007 University of Cambridge.
226 ------------------------------------------------------------------------------
227
228
229 PCREBUILD(3) PCREBUILD(3)
230
231
232 NAME
233 PCRE - Perl-compatible regular expressions
234
235
236 PCRE BUILD-TIME OPTIONS
237
238 This document describes the optional features of PCRE that can be
239 selected when the library is compiled. They are all selected, or dese-
240 lected, by providing options to the configure script that is run before
241 the make command. The complete list of options for configure (which
242 includes the standard ones such as the selection of the installation
243 directory) can be obtained by running
244
245 ./configure --help
246
247 The following sections include descriptions of options whose names
248 begin with --enable or --disable. These settings specify changes to the
249 defaults for the configure command. Because of the way that configure
250 works, --enable and --disable always come in pairs, so the complemen-
251 tary option always exists as well, but as it specifies the default, it
252 is not described.
253
254
255 C++ SUPPORT
256
257 By default, the configure script will search for a C++ compiler and C++
258 header files. If it finds them, it automatically builds the C++ wrapper
259 library for PCRE. You can disable this by adding
260
261 --disable-cpp
262
263 to the configure command.
264
265
266 UTF-8 SUPPORT
267
268 To build PCRE with support for UTF-8 character strings, add
269
270 --enable-utf8
271
272 to the configure command. Of itself, this does not make PCRE treat
273 strings as UTF-8. As well as compiling PCRE with this option, you also
274 have have to set the PCRE_UTF8 option when you call the pcre_compile()
275 function.
276
277
278 UNICODE CHARACTER PROPERTY SUPPORT
279
280 UTF-8 support allows PCRE to process character values greater than 255
281 in the strings that it handles. On its own, however, it does not pro-
282 vide any facilities for accessing the properties of such characters. If
283 you want to be able to use the pattern escapes \P, \p, and \X, which
284 refer to Unicode character properties, you must add
285
286 --enable-unicode-properties
287
288 to the configure command. This implies UTF-8 support, even if you have
289 not explicitly requested it.
290
291 Including Unicode property support adds around 30K of tables to the
292 PCRE library. Only the general category properties such as Lu and Nd
293 are supported. Details are given in the pcrepattern documentation.
294
295
296 CODE VALUE OF NEWLINE
297
298 By default, PCRE interprets character 10 (linefeed, LF) as indicating
299 the end of a line. This is the normal newline character on Unix-like
300 systems. You can compile PCRE to use character 13 (carriage return, CR)
301 instead, by adding
302
303 --enable-newline-is-cr
304
305 to the configure command. There is also a --enable-newline-is-lf
306 option, which explicitly specifies linefeed as the newline character.
307
308 Alternatively, you can specify that line endings are to be indicated by
309 the two character sequence CRLF. If you want this, add
310
311 --enable-newline-is-crlf
312
313 to the configure command. There is a fourth option, specified by
314
315 --enable-newline-is-anycrlf
316
317 which causes PCRE to recognize any of the three sequences CR, LF, or
318 CRLF as indicating a line ending. Finally, a fifth option, specified by
319
320 --enable-newline-is-any
321
322 causes PCRE to recognize any Unicode newline sequence.
323
324 Whatever line ending convention is selected when PCRE is built can be
325 overridden when the library functions are called. At build time it is
326 conventional to use the standard for your operating system.
327
328
329 BUILDING SHARED AND STATIC LIBRARIES
330
331 The PCRE building process uses libtool to build both shared and static
332 Unix libraries by default. You can suppress one of these by adding one
333 of
334
335 --disable-shared
336 --disable-static
337
338 to the configure command, as required.
339
340
341 POSIX MALLOC USAGE
342
343 When PCRE is called through the POSIX interface (see the pcreposix doc-
344 umentation), additional working storage is required for holding the
345 pointers to capturing substrings, because PCRE requires three integers
346 per substring, whereas the POSIX interface provides only two. If the
347 number of expected substrings is small, the wrapper function uses space
348 on the stack, because this is faster than using malloc() for each call.
349 The default threshold above which the stack is no longer used is 10; it
350 can be changed by adding a setting such as
351
352 --with-posix-malloc-threshold=20
353
354 to the configure command.
355
356
357 HANDLING VERY LARGE PATTERNS
358
359 Within a compiled pattern, offset values are used to point from one
360 part to another (for example, from an opening parenthesis to an alter-
361 nation metacharacter). By default, two-byte values are used for these
362 offsets, leading to a maximum size for a compiled pattern of around
363 64K. This is sufficient to handle all but the most gigantic patterns.
364 Nevertheless, some people do want to process enormous patterns, so it
365 is possible to compile PCRE to use three-byte or four-byte offsets by
366 adding a setting such as
367
368 --with-link-size=3
369
370 to the configure command. The value given must be 2, 3, or 4. Using
371 longer offsets slows down the operation of PCRE because it has to load
372 additional bytes when handling them.
373
374
375 AVOIDING EXCESSIVE STACK USAGE
376
377 When matching with the pcre_exec() function, PCRE implements backtrack-
378 ing by making recursive calls to an internal function called match().
379 In environments where the size of the stack is limited, this can se-
380 verely limit PCRE's operation. (The Unix environment does not usually
381 suffer from this problem, but it may sometimes be necessary to increase
382 the maximum stack size. There is a discussion in the pcrestack docu-
383 mentation.) An alternative approach to recursion that uses memory from
384 the heap to remember data, instead of using recursive function calls,
385 has been implemented to work round the problem of limited stack size.
386 If you want to build a version of PCRE that works this way, add
387
388 --disable-stack-for-recursion
389
390 to the configure command. With this configuration, PCRE will use the
391 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
392 ment functions. Separate functions are provided because the usage is
393 very predictable: the block sizes requested are always the same, and
394 the blocks are always freed in reverse order. A calling program might
395 be able to implement optimized functions that perform better than the
396 standard malloc() and free() functions. PCRE runs noticeably more
397 slowly when built in this way. This option affects only the pcre_exec()
398 function; it is not relevant for the the pcre_dfa_exec() function.
399
400
401 LIMITING PCRE RESOURCE USAGE
402
403 Internally, PCRE has a function called match(), which it calls repeat-
404 edly (sometimes recursively) when matching a pattern with the
405 pcre_exec() function. By controlling the maximum number of times this
406 function may be called during a single matching operation, a limit can
407 be placed on the resources used by a single call to pcre_exec(). The
408 limit can be changed at run time, as described in the pcreapi documen-
409 tation. The default is 10 million, but this can be changed by adding a
410 setting such as
411
412 --with-match-limit=500000
413
414 to the configure command. This setting has no effect on the
415 pcre_dfa_exec() matching function.
416
417 In some environments it is desirable to limit the depth of recursive
418 calls of match() more strictly than the total number of calls, in order
419 to restrict the maximum amount of stack (or heap, if --disable-stack-
420 for-recursion is specified) that is used. A second limit controls this;
421 it defaults to the value that is set for --with-match-limit, which
422 imposes no additional constraints. However, you can set a lower limit
423 by adding, for example,
424
425 --with-match-limit-recursion=10000
426
427 to the configure command. This value can also be overridden at run
428 time.
429
430
431 CREATING CHARACTER TABLES AT BUILD TIME
432
433 PCRE uses fixed tables for processing characters whose code values are
434 less than 256. By default, PCRE is built with a set of tables that are
435 distributed in the file pcre_chartables.c.dist. These tables are for
436 ASCII codes only. If you add
437
438 --enable-rebuild-chartables
439
440 to the configure command, the distributed tables are no longer used.
441 Instead, a program called dftables is compiled and run. This outputs
442 the source for new set of tables, created in the default locale of your
443 C runtime system. (This method of replacing the tables does not work if
444 you are cross compiling, because dftables is run on the local host. If
445 you need to create alternative tables when cross compiling, you will
446 have to do so "by hand".)
447
448
449 USING EBCDIC CODE
450
451 PCRE assumes by default that it will run in an environment where the
452 character code is ASCII (or Unicode, which is a superset of ASCII).
453 PCRE can, however, be compiled to run in an EBCDIC environment by
454 adding
455
456 --enable-ebcdic
457
458 to the configure command. This setting implies --enable-rebuild-charta-
459 bles.
460
461
462 SEE ALSO
463
464 pcreapi(3), pcre_config(3).
465
466
467 AUTHOR
468
469 Philip Hazel
470 University Computing Service
471 Cambridge CB2 3QH, England.
472
473
474 REVISION
475
476 Last updated: 16 April 2007
477 Copyright (c) 1997-2007 University of Cambridge.
478 ------------------------------------------------------------------------------
479
480
481 PCREMATCHING(3) PCREMATCHING(3)
482
483
484 NAME
485 PCRE - Perl-compatible regular expressions
486
487
488 PCRE MATCHING ALGORITHMS
489
490 This document describes the two different algorithms that are available
491 in PCRE for matching a compiled regular expression against a given sub-
492 ject string. The "standard" algorithm is the one provided by the
493 pcre_exec() function. This works in the same was as Perl's matching
494 function, and provides a Perl-compatible matching operation.
495
496 An alternative algorithm is provided by the pcre_dfa_exec() function;
497 this operates in a different way, and is not Perl-compatible. It has
498 advantages and disadvantages compared with the standard algorithm, and
499 these are described below.
500
501 When there is only one possible way in which a given subject string can
502 match a pattern, the two algorithms give the same answer. A difference
503 arises, however, when there are multiple possibilities. For example, if
504 the pattern
505
506 ^<.*>
507
508 is matched against the string
509
510 <something> <something else> <something further>
511
512 there are three possible answers. The standard algorithm finds only one
513 of them, whereas the alternative algorithm finds all three.
514
515
516 REGULAR EXPRESSIONS AS TREES
517
518 The set of strings that are matched by a regular expression can be rep-
519 resented as a tree structure. An unlimited repetition in the pattern
520 makes the tree of infinite size, but it is still a tree. Matching the
521 pattern to a given subject string (from a given starting point) can be
522 thought of as a search of the tree. There are two ways to search a
523 tree: depth-first and breadth-first, and these correspond to the two
524 matching algorithms provided by PCRE.
525
526
527 THE STANDARD MATCHING ALGORITHM
528
529 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
530 sions", the standard algorithm is an "NFA algorithm". It conducts a
531 depth-first search of the pattern tree. That is, it proceeds along a
532 single path through the tree, checking that the subject matches what is
533 required. When there is a mismatch, the algorithm tries any alterna-
534 tives at the current point, and if they all fail, it backs up to the
535 previous branch point in the tree, and tries the next alternative
536 branch at that level. This often involves backing up (moving to the
537 left) in the subject string as well. The order in which repetition
538 branches are tried is controlled by the greedy or ungreedy nature of
539 the quantifier.
540
541 If a leaf node is reached, a matching string has been found, and at
542 that point the algorithm stops. Thus, if there is more than one possi-
543 ble match, this algorithm returns the first one that it finds. Whether
544 this is the shortest, the longest, or some intermediate length depends
545 on the way the greedy and ungreedy repetition quantifiers are specified
546 in the pattern.
547
548 Because it ends up with a single path through the tree, it is rela-
549 tively straightforward for this algorithm to keep track of the sub-
550 strings that are matched by portions of the pattern in parentheses.
551 This provides support for capturing parentheses and back references.
552
553
554 THE ALTERNATIVE MATCHING ALGORITHM
555
556 This algorithm conducts a breadth-first search of the tree. Starting
557 from the first matching point in the subject, it scans the subject
558 string from left to right, once, character by character, and as it does
559 this, it remembers all the paths through the tree that represent valid
560 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
561 though it is not implemented as a traditional finite state machine (it
562 keeps multiple states active simultaneously).
563
564 The scan continues until either the end of the subject is reached, or
565 there are no more unterminated paths. At this point, terminated paths
566 represent the different matching possibilities (if there are none, the
567 match has failed). Thus, if there is more than one possible match,
568 this algorithm finds all of them, and in particular, it finds the long-
569 est. In PCRE, there is an option to stop the algorithm after the first
570 match (which is necessarily the shortest) has been found.
571
572 Note that all the matches that are found start at the same point in the
573 subject. If the pattern
574
575 cat(er(pillar)?)
576
577 is matched against the string "the caterpillar catchment", the result
578 will be the three strings "cat", "cater", and "caterpillar" that start
579 at the fourth character of the subject. The algorithm does not automat-
580 ically move on to find matches that start at later positions.
581
582 There are a number of features of PCRE regular expressions that are not
583 supported by the alternative matching algorithm. They are as follows:
584
585 1. Because the algorithm finds all possible matches, the greedy or
586 ungreedy nature of repetition quantifiers is not relevant. Greedy and
587 ungreedy quantifiers are treated in exactly the same way. However, pos-
588 sessive quantifiers can make a difference when what follows could also
589 match what is quantified, for example in a pattern like this:
590
591 ^a++\w!
592
593 This pattern matches "aaab!" but not "aaa!", which would be matched by
594 a non-possessive quantifier. Similarly, if an atomic group is present,
595 it is matched as if it were a standalone pattern at the current point,
596 and the longest match is then "locked in" for the rest of the overall
597 pattern.
598
599 2. When dealing with multiple paths through the tree simultaneously, it
600 is not straightforward to keep track of captured substrings for the
601 different matching possibilities, and PCRE's implementation of this
602 algorithm does not attempt to do this. This means that no captured sub-
603 strings are available.
604
605 3. Because no substrings are captured, back references within the pat-
606 tern are not supported, and cause errors if encountered.
607
608 4. For the same reason, conditional expressions that use a backrefer-
609 ence as the condition or test for a specific group recursion are not
610 supported.
611
612 5. Callouts are supported, but the value of the capture_top field is
613 always 1, and the value of the capture_last field is always -1.
614
615 6. The \C escape sequence, which (in the standard algorithm) matches a
616 single byte, even in UTF-8 mode, is not supported because the alterna-
617 tive algorithm moves through the subject string one character at a
618 time, for all active paths through the tree.
619
620
621 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
622
623 Using the alternative matching algorithm provides the following advan-
624 tages:
625
626 1. All possible matches (at a single point in the subject) are automat-
627 ically found, and in particular, the longest match is found. To find
628 more than one match using the standard algorithm, you have to do kludgy
629 things with callouts.
630
631 2. There is much better support for partial matching. The restrictions
632 on the content of the pattern that apply when using the standard algo-
633 rithm for partial matching do not apply to the alternative algorithm.
634 For non-anchored patterns, the starting position of a partial match is
635 available.
636
637 3. Because the alternative algorithm scans the subject string just
638 once, and never needs to backtrack, it is possible to pass very long
639 subject strings to the matching function in several pieces, checking
640 for partial matching each time.
641
642
643 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
644
645 The alternative algorithm suffers from a number of disadvantages:
646
647 1. It is substantially slower than the standard algorithm. This is
648 partly because it has to search for all possible matches, but is also
649 because it is less susceptible to optimization.
650
651 2. Capturing parentheses and back references are not supported.
652
653 3. Although atomic groups are supported, their use does not provide the
654 performance advantage that it does for the standard algorithm.
655
656
657 AUTHOR
658
659 Philip Hazel
660 University Computing Service
661 Cambridge CB2 3QH, England.
662
663
664 REVISION
665
666 Last updated: 06 March 2007
667 Copyright (c) 1997-2007 University of Cambridge.
668 ------------------------------------------------------------------------------
669
670
671 PCREAPI(3) PCREAPI(3)
672
673
674 NAME
675 PCRE - Perl-compatible regular expressions
676
677
678 PCRE NATIVE API
679
680 #include <pcre.h>
681
682 pcre *pcre_compile(const char *pattern, int options,
683 const char **errptr, int *erroffset,
684 const unsigned char *tableptr);
685
686 pcre *pcre_compile2(const char *pattern, int options,
687 int *errorcodeptr,
688 const char **errptr, int *erroffset,
689 const unsigned char *tableptr);
690
691 pcre_extra *pcre_study(const pcre *code, int options,
692 const char **errptr);
693
694 int pcre_exec(const pcre *code, const pcre_extra *extra,
695 const char *subject, int length, int startoffset,
696 int options, int *ovector, int ovecsize);
697
698 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
699 const char *subject, int length, int startoffset,
700 int options, int *ovector, int ovecsize,
701 int *workspace, int wscount);
702
703 int pcre_copy_named_substring(const pcre *code,
704 const char *subject, int *ovector,
705 int stringcount, const char *stringname,
706 char *buffer, int buffersize);
707
708 int pcre_copy_substring(const char *subject, int *ovector,
709 int stringcount, int stringnumber, char *buffer,
710 int buffersize);
711
712 int pcre_get_named_substring(const pcre *code,
713 const char *subject, int *ovector,
714 int stringcount, const char *stringname,
715 const char **stringptr);
716
717 int pcre_get_stringnumber(const pcre *code,
718 const char *name);
719
720 int pcre_get_stringtable_entries(const pcre *code,
721 const char *name, char **first, char **last);
722
723 int pcre_get_substring(const char *subject, int *ovector,
724 int stringcount, int stringnumber,
725 const char **stringptr);
726
727 int pcre_get_substring_list(const char *subject,
728 int *ovector, int stringcount, const char ***listptr);
729
730 void pcre_free_substring(const char *stringptr);
731
732 void pcre_free_substring_list(const char **stringptr);
733
734 const unsigned char *pcre_maketables(void);
735
736 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
737 int what, void *where);
738
739 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
740
741 int pcre_refcount(pcre *code, int adjust);
742
743 int pcre_config(int what, void *where);
744
745 char *pcre_version(void);
746
747 void *(*pcre_malloc)(size_t);
748
749 void (*pcre_free)(void *);
750
751 void *(*pcre_stack_malloc)(size_t);
752
753 void (*pcre_stack_free)(void *);
754
755 int (*pcre_callout)(pcre_callout_block *);
756
757
758 PCRE API OVERVIEW
759
760 PCRE has its own native API, which is described in this document. There
761 are also some wrapper functions that correspond to the POSIX regular
762 expression API. These are described in the pcreposix documentation.
763 Both of these APIs define a set of C function calls. A C++ wrapper is
764 distributed with PCRE. It is documented in the pcrecpp page.
765
766 The native API C function prototypes are defined in the header file
767 pcre.h, and on Unix systems the library itself is called libpcre. It
768 can normally be accessed by adding -lpcre to the command for linking an
769 application that uses PCRE. The header file defines the macros
770 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
771 bers for the library. Applications can use these to include support
772 for different releases of PCRE.
773
774 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
775 pcre_exec() are used for compiling and matching regular expressions in
776 a Perl-compatible manner. A sample program that demonstrates the sim-
777 plest way of using them is provided in the file called pcredemo.c in
778 the source distribution. The pcresample documentation describes how to
779 run it.
780
781 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
782 ble, is also provided. This uses a different algorithm for the match-
783 ing. The alternative algorithm finds all possible matches (at a given
784 point in the subject), and scans the subject just once. However, this
785 algorithm does not return captured substrings. A description of the two
786 matching algorithms and their advantages and disadvantages is given in
787 the pcrematching documentation.
788
789 In addition to the main compiling and matching functions, there are
790 convenience functions for extracting captured substrings from a subject
791 string that is matched by pcre_exec(). They are:
792
793 pcre_copy_substring()
794 pcre_copy_named_substring()
795 pcre_get_substring()
796 pcre_get_named_substring()
797 pcre_get_substring_list()
798 pcre_get_stringnumber()
799 pcre_get_stringtable_entries()
800
801 pcre_free_substring() and pcre_free_substring_list() are also provided,
802 to free the memory used for extracted strings.
803
804 The function pcre_maketables() is used to build a set of character
805 tables in the current locale for passing to pcre_compile(),
806 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
807 provided for specialist use. Most commonly, no special tables are
808 passed, in which case internal tables that are generated when PCRE is
809 built are used.
810
811 The function pcre_fullinfo() is used to find out information about a
812 compiled pattern; pcre_info() is an obsolete version that returns only
813 some of the available information, but is retained for backwards com-
814 patibility. The function pcre_version() returns a pointer to a string
815 containing the version of PCRE and its date of release.
816
817 The function pcre_refcount() maintains a reference count in a data
818 block containing a compiled pattern. This is provided for the benefit
819 of object-oriented applications.
820
821 The global variables pcre_malloc and pcre_free initially contain the
822 entry points of the standard malloc() and free() functions, respec-
823 tively. PCRE calls the memory management functions via these variables,
824 so a calling program can replace them if it wishes to intercept the
825 calls. This should be done before calling any PCRE functions.
826
827 The global variables pcre_stack_malloc and pcre_stack_free are also
828 indirections to memory management functions. These special functions
829 are used only when PCRE is compiled to use the heap for remembering
830 data, instead of recursive function calls, when running the pcre_exec()
831 function. See the pcrebuild documentation for details of how to do
832 this. It is a non-standard way of building PCRE, for use in environ-
833 ments that have limited stacks. Because of the greater use of memory
834 management, it runs more slowly. Separate functions are provided so
835 that special-purpose external code can be used for this case. When
836 used, these functions are always called in a stack-like manner (last
837 obtained, first freed), and always for memory blocks of the same size.
838 There is a discussion about PCRE's stack usage in the pcrestack docu-
839 mentation.
840
841 The global variable pcre_callout initially contains NULL. It can be set
842 by the caller to a "callout" function, which PCRE will then call at
843 specified points during a matching operation. Details are given in the
844 pcrecallout documentation.
845
846
847 NEWLINES
848
849 PCRE supports five different conventions for indicating line breaks in
850 strings: a single CR (carriage return) character, a single LF (line-
851 feed) character, the two-character sequence CRLF, any of the three pre-
852 ceding, or any Unicode newline sequence. The Unicode newline sequences
853 are the three just mentioned, plus the single characters VT (vertical
854 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
855 separator, U+2028), and PS (paragraph separator, U+2029).
856
857 Each of the first three conventions is used by at least one operating
858 system as its standard newline sequence. When PCRE is built, a default
859 can be specified. The default default is LF, which is the Unix stan-
860 dard. When PCRE is run, the default can be overridden, either when a
861 pattern is compiled, or when it is matched.
862
863 In the PCRE documentation the word "newline" is used to mean "the char-
864 acter or pair of characters that indicate a line break". The choice of
865 newline convention affects the handling of the dot, circumflex, and
866 dollar metacharacters, the handling of #-comments in /x mode, and, when
867 CRLF is a recognized line ending sequence, the match position advance-
868 ment for a non-anchored pattern. The choice of newline convention does
869 not affect the interpretation of the \n or \r escape sequences.
870
871
872 MULTITHREADING
873
874 The PCRE functions can be used in multi-threading applications, with
875 the proviso that the memory management functions pointed to by
876 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
877 callout function pointed to by pcre_callout, are shared by all threads.
878
879 The compiled form of a regular expression is not altered during match-
880 ing, so the same compiled pattern can safely be used by several threads
881 at once.
882
883
884 SAVING PRECOMPILED PATTERNS FOR LATER USE
885
886 The compiled form of a regular expression can be saved and re-used at a
887 later time, possibly by a different program, and even on a host other
888 than the one on which it was compiled. Details are given in the
889 pcreprecompile documentation.
890
891
892 CHECKING BUILD-TIME OPTIONS
893
894 int pcre_config(int what, void *where);
895
896 The function pcre_config() makes it possible for a PCRE client to dis-
897 cover which optional features have been compiled into the PCRE library.
898 The pcrebuild documentation has more details about these optional fea-
899 tures.
900
901 The first argument for pcre_config() is an integer, specifying which
902 information is required; the second argument is a pointer to a variable
903 into which the information is placed. The following information is
904 available:
905
906 PCRE_CONFIG_UTF8
907
908 The output is an integer that is set to one if UTF-8 support is avail-
909 able; otherwise it is set to zero.
910
911 PCRE_CONFIG_UNICODE_PROPERTIES
912
913 The output is an integer that is set to one if support for Unicode
914 character properties is available; otherwise it is set to zero.
915
916 PCRE_CONFIG_NEWLINE
917
918 The output is an integer whose value specifies the default character
919 sequence that is recognized as meaning "newline". The four values that
920 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
921 and -1 for ANY. The default should normally be the standard sequence
922 for your operating system.
923
924 PCRE_CONFIG_LINK_SIZE
925
926 The output is an integer that contains the number of bytes used for
927 internal linkage in compiled regular expressions. The value is 2, 3, or
928 4. Larger values allow larger regular expressions to be compiled, at
929 the expense of slower matching. The default value of 2 is sufficient
930 for all but the most massive patterns, since it allows the compiled
931 pattern to be up to 64K in size.
932
933 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
934
935 The output is an integer that contains the threshold above which the
936 POSIX interface uses malloc() for output vectors. Further details are
937 given in the pcreposix documentation.
938
939 PCRE_CONFIG_MATCH_LIMIT
940
941 The output is an integer that gives the default limit for the number of
942 internal matching function calls in a pcre_exec() execution. Further
943 details are given with pcre_exec() below.
944
945 PCRE_CONFIG_MATCH_LIMIT_RECURSION
946
947 The output is an integer that gives the default limit for the depth of
948 recursion when calling the internal matching function in a pcre_exec()
949 execution. Further details are given with pcre_exec() below.
950
951 PCRE_CONFIG_STACKRECURSE
952
953 The output is an integer that is set to one if internal recursion when
954 running pcre_exec() is implemented by recursive function calls that use
955 the stack to remember their state. This is the usual way that PCRE is
956 compiled. The output is zero if PCRE was compiled to use blocks of data
957 on the heap instead of recursive function calls. In this case,
958 pcre_stack_malloc and pcre_stack_free are called to manage memory
959 blocks on the heap, thus avoiding the use of the stack.
960
961
962 COMPILING A PATTERN
963
964 pcre *pcre_compile(const char *pattern, int options,
965 const char **errptr, int *erroffset,
966 const unsigned char *tableptr);
967
968 pcre *pcre_compile2(const char *pattern, int options,
969 int *errorcodeptr,
970 const char **errptr, int *erroffset,
971 const unsigned char *tableptr);
972
973 Either of the functions pcre_compile() or pcre_compile2() can be called
974 to compile a pattern into an internal form. The only difference between
975 the two interfaces is that pcre_compile2() has an additional argument,
976 errorcodeptr, via which a numerical error code can be returned.
977
978 The pattern is a C string terminated by a binary zero, and is passed in
979 the pattern argument. A pointer to a single block of memory that is
980 obtained via pcre_malloc is returned. This contains the compiled code
981 and related data. The pcre type is defined for the returned block; this
982 is a typedef for a structure whose contents are not externally defined.
983 It is up to the caller to free the memory (via pcre_free) when it is no
984 longer required.
985
986 Although the compiled code of a PCRE regex is relocatable, that is, it
987 does not depend on memory location, the complete pcre data block is not
988 fully relocatable, because it may contain a copy of the tableptr argu-
989 ment, which is an address (see below).
990
991 The options argument contains various bit settings that affect the com-
992 pilation. It should be zero if no options are required. The available
993 options are described below. Some of them, in particular, those that
994 are compatible with Perl, can also be set and unset from within the
995 pattern (see the detailed description in the pcrepattern documenta-
996 tion). For these options, the contents of the options argument speci-
997 fies their initial settings at the start of compilation and execution.
998 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
999 of matching as well as at compile time.
1000
1001 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1002 if compilation of a pattern fails, pcre_compile() returns NULL, and
1003 sets the variable pointed to by errptr to point to a textual error mes-
1004 sage. This is a static string that is part of the library. You must not
1005 try to free it. The offset from the start of the pattern to the charac-
1006 ter where the error was discovered is placed in the variable pointed to
1007 by erroffset, which must not be NULL. If it is, an immediate error is
1008 given.
1009
1010 If pcre_compile2() is used instead of pcre_compile(), and the error-
1011 codeptr argument is not NULL, a non-zero error code number is returned
1012 via this argument in the event of an error. This is in addition to the
1013 textual error message. Error codes and messages are listed below.
1014
1015 If the final argument, tableptr, is NULL, PCRE uses a default set of
1016 character tables that are built when PCRE is compiled, using the
1017 default C locale. Otherwise, tableptr must be an address that is the
1018 result of a call to pcre_maketables(). This value is stored with the
1019 compiled pattern, and used again by pcre_exec(), unless another table
1020 pointer is passed to it. For more discussion, see the section on locale
1021 support below.
1022
1023 This code fragment shows a typical straightforward call to pcre_com-
1024 pile():
1025
1026 pcre *re;
1027 const char *error;
1028 int erroffset;
1029 re = pcre_compile(
1030 "^A.*Z", /* the pattern */
1031 0, /* default options */
1032 &error, /* for error message */
1033 &erroffset, /* for error offset */
1034 NULL); /* use default character tables */
1035
1036 The following names for option bits are defined in the pcre.h header
1037 file:
1038
1039 PCRE_ANCHORED
1040
1041 If this bit is set, the pattern is forced to be "anchored", that is, it
1042 is constrained to match only at the first matching point in the string
1043 that is being searched (the "subject string"). This effect can also be
1044 achieved by appropriate constructs in the pattern itself, which is the
1045 only way to do it in Perl.
1046
1047 PCRE_AUTO_CALLOUT
1048
1049 If this bit is set, pcre_compile() automatically inserts callout items,
1050 all with number 255, before each pattern item. For discussion of the
1051 callout facility, see the pcrecallout documentation.
1052
1053 PCRE_CASELESS
1054
1055 If this bit is set, letters in the pattern match both upper and lower
1056 case letters. It is equivalent to Perl's /i option, and it can be
1057 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1058 always understands the concept of case for characters whose values are
1059 less than 128, so caseless matching is always possible. For characters
1060 with higher values, the concept of case is supported if PCRE is com-
1061 piled with Unicode property support, but not otherwise. If you want to
1062 use caseless matching for characters 128 and above, you must ensure
1063 that PCRE is compiled with Unicode property support as well as with
1064 UTF-8 support.
1065
1066 PCRE_DOLLAR_ENDONLY
1067
1068 If this bit is set, a dollar metacharacter in the pattern matches only
1069 at the end of the subject string. Without this option, a dollar also
1070 matches immediately before a newline at the end of the string (but not
1071 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1072 if PCRE_MULTILINE is set. There is no equivalent to this option in
1073 Perl, and no way to set it within a pattern.
1074
1075 PCRE_DOTALL
1076
1077 If this bit is set, a dot metacharater in the pattern matches all char-
1078 acters, including those that indicate newline. Without it, a dot does
1079 not match when the current position is at a newline. This option is
1080 equivalent to Perl's /s option, and it can be changed within a pattern
1081 by a (?s) option setting. A negative class such as [^a] always matches
1082 newline characters, independent of the setting of this option.
1083
1084 PCRE_DUPNAMES
1085
1086 If this bit is set, names used to identify capturing subpatterns need
1087 not be unique. This can be helpful for certain types of pattern when it
1088 is known that only one instance of the named subpattern can ever be
1089 matched. There are more details of named subpatterns below; see also
1090 the pcrepattern documentation.
1091
1092 PCRE_EXTENDED
1093
1094 If this bit is set, whitespace data characters in the pattern are
1095 totally ignored except when escaped or inside a character class. White-
1096 space does not include the VT character (code 11). In addition, charac-
1097 ters between an unescaped # outside a character class and the next new-
1098 line, inclusive, are also ignored. This is equivalent to Perl's /x
1099 option, and it can be changed within a pattern by a (?x) option set-
1100 ting.
1101
1102 This option makes it possible to include comments inside complicated
1103 patterns. Note, however, that this applies only to data characters.
1104 Whitespace characters may never appear within special character
1105 sequences in a pattern, for example within the sequence (?( which
1106 introduces a conditional subpattern.
1107
1108 PCRE_EXTRA
1109
1110 This option was invented in order to turn on additional functionality
1111 of PCRE that is incompatible with Perl, but it is currently of very
1112 little use. When set, any backslash in a pattern that is followed by a
1113 letter that has no special meaning causes an error, thus reserving
1114 these combinations for future expansion. By default, as in Perl, a
1115 backslash followed by a letter with no special meaning is treated as a
1116 literal. (Perl can, however, be persuaded to give a warning for this.)
1117 There are at present no other features controlled by this option. It
1118 can also be set by a (?X) option setting within a pattern.
1119
1120 PCRE_FIRSTLINE
1121
1122 If this option is set, an unanchored pattern is required to match
1123 before or at the first newline in the subject string, though the
1124 matched text may continue over the newline.
1125
1126 PCRE_MULTILINE
1127
1128 By default, PCRE treats the subject string as consisting of a single
1129 line of characters (even if it actually contains newlines). The "start
1130 of line" metacharacter (^) matches only at the start of the string,
1131 while the "end of line" metacharacter ($) matches only at the end of
1132 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1133 is set). This is the same as Perl.
1134
1135 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1136 constructs match immediately following or immediately before internal
1137 newlines in the subject string, respectively, as well as at the very
1138 start and end. This is equivalent to Perl's /m option, and it can be
1139 changed within a pattern by a (?m) option setting. If there are no new-
1140 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1141 setting PCRE_MULTILINE has no effect.
1142
1143 PCRE_NEWLINE_CR
1144 PCRE_NEWLINE_LF
1145 PCRE_NEWLINE_CRLF
1146 PCRE_NEWLINE_ANYCRLF
1147 PCRE_NEWLINE_ANY
1148
1149 These options override the default newline definition that was chosen
1150 when PCRE was built. Setting the first or the second specifies that a
1151 newline is indicated by a single character (CR or LF, respectively).
1152 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1153 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1154 that any of the three preceding sequences should be recognized. Setting
1155 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1156 recognized. The Unicode newline sequences are the three just mentioned,
1157 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1158 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1159 (paragraph separator, U+2029). The last two are recognized only in
1160 UTF-8 mode.
1161
1162 The newline setting in the options word uses three bits that are
1163 treated as a number, giving eight possibilities. Currently only six are
1164 used (default plus the five values above). This means that if you set
1165 more than one newline option, the combination may or may not be sensi-
1166 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1167 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1168 cause an error.
1169
1170 The only time that a line break is specially recognized when compiling
1171 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1172 character class is encountered. This indicates a comment that lasts
1173 until after the next line break sequence. In other circumstances, line
1174 break sequences are treated as literal data, except that in
1175 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1176 and are therefore ignored.
1177
1178 The newline option that is set at compile time becomes the default that
1179 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1180
1181 PCRE_NO_AUTO_CAPTURE
1182
1183 If this option is set, it disables the use of numbered capturing paren-
1184 theses in the pattern. Any opening parenthesis that is not followed by
1185 ? behaves as if it were followed by ?: but named parentheses can still
1186 be used for capturing (and they acquire numbers in the usual way).
1187 There is no equivalent of this option in Perl.
1188
1189 PCRE_UNGREEDY
1190
1191 This option inverts the "greediness" of the quantifiers so that they
1192 are not greedy by default, but become greedy if followed by "?". It is
1193 not compatible with Perl. It can also be set by a (?U) option setting
1194 within the pattern.
1195
1196 PCRE_UTF8
1197
1198 This option causes PCRE to regard both the pattern and the subject as
1199 strings of UTF-8 characters instead of single-byte character strings.
1200 However, it is available only when PCRE is built to include UTF-8 sup-
1201 port. If not, the use of this option provokes an error. Details of how
1202 this option changes the behaviour of PCRE are given in the section on
1203 UTF-8 support in the main pcre page.
1204
1205 PCRE_NO_UTF8_CHECK
1206
1207 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1208 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1209 pcre_compile() returns an error. If you already know that your pattern
1210 is valid, and you want to skip this check for performance reasons, you
1211 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1212 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1213 your program to crash. Note that this option can also be passed to
1214 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1215 ing of subject strings.
1216
1217
1218 COMPILATION ERROR CODES
1219
1220 The following table lists the error codes than may be returned by
1221 pcre_compile2(), along with the error messages that may be returned by
1222 both compiling functions. As PCRE has developed, some error codes have
1223 fallen out of use. To avoid confusion, they have not been re-used.
1224
1225 0 no error
1226 1 \ at end of pattern
1227 2 \c at end of pattern
1228 3 unrecognized character follows \
1229 4 numbers out of order in {} quantifier
1230 5 number too big in {} quantifier
1231 6 missing terminating ] for character class
1232 7 invalid escape sequence in character class
1233 8 range out of order in character class
1234 9 nothing to repeat
1235 10 [this code is not in use]
1236 11 internal error: unexpected repeat
1237 12 unrecognized character after (?
1238 13 POSIX named classes are supported only within a class
1239 14 missing )
1240 15 reference to non-existent subpattern
1241 16 erroffset passed as NULL
1242 17 unknown option bit(s) set
1243 18 missing ) after comment
1244 19 [this code is not in use]
1245 20 regular expression too large
1246 21 failed to get memory
1247 22 unmatched parentheses
1248 23 internal error: code overflow
1249 24 unrecognized character after (?<
1250 25 lookbehind assertion is not fixed length
1251 26 malformed number or name after (?(
1252 27 conditional group contains more than two branches
1253 28 assertion expected after (?(
1254 29 (?R or (?digits must be followed by )
1255 30 unknown POSIX class name
1256 31 POSIX collating elements are not supported
1257 32 this version of PCRE is not compiled with PCRE_UTF8 support
1258 33 [this code is not in use]
1259 34 character value in \x{...} sequence is too large
1260 35 invalid condition (?(0)
1261 36 \C not allowed in lookbehind assertion
1262 37 PCRE does not support \L, \l, \N, \U, or \u
1263 38 number after (?C is > 255
1264 39 closing ) for (?C expected
1265 40 recursive call could loop indefinitely
1266 41 unrecognized character after (?P
1267 42 syntax error in subpattern name (missing terminator)
1268 43 two named subpatterns have the same name
1269 44 invalid UTF-8 string
1270 45 support for \P, \p, and \X has not been compiled
1271 46 malformed \P or \p sequence
1272 47 unknown property name after \P or \p
1273 48 subpattern name is too long (maximum 32 characters)
1274 49 too many named subpatterns (maximum 10,000)
1275 50 repeated subpattern is too long
1276 51 octal value is greater than \377 (not in UTF-8 mode)
1277 52 internal error: overran compiling workspace
1278 53 internal error: previously-checked referenced subpattern not
1279 found
1280 54 DEFINE group contains more than one branch
1281 55 repeating a DEFINE group is not allowed
1282 56 inconsistent NEWLINE options"
1283
1284
1285 STUDYING A PATTERN
1286
1287 pcre_extra *pcre_study(const pcre *code, int options
1288 const char **errptr);
1289
1290 If a compiled pattern is going to be used several times, it is worth
1291 spending more time analyzing it in order to speed up the time taken for
1292 matching. The function pcre_study() takes a pointer to a compiled pat-
1293 tern as its first argument. If studying the pattern produces additional
1294 information that will help speed up matching, pcre_study() returns a
1295 pointer to a pcre_extra block, in which the study_data field points to
1296 the results of the study.
1297
1298 The returned value from pcre_study() can be passed directly to
1299 pcre_exec(). However, a pcre_extra block also contains other fields
1300 that can be set by the caller before the block is passed; these are
1301 described below in the section on matching a pattern.
1302
1303 If studying the pattern does not produce any additional information
1304 pcre_study() returns NULL. In that circumstance, if the calling program
1305 wants to pass any of the other fields to pcre_exec(), it must set up
1306 its own pcre_extra block.
1307
1308 The second argument of pcre_study() contains option bits. At present,
1309 no options are defined, and this argument should always be zero.
1310
1311 The third argument for pcre_study() is a pointer for an error message.
1312 If studying succeeds (even if no data is returned), the variable it
1313 points to is set to NULL. Otherwise it is set to point to a textual
1314 error message. This is a static string that is part of the library. You
1315 must not try to free it. You should test the error pointer for NULL
1316 after calling pcre_study(), to be sure that it has run successfully.
1317
1318 This is a typical call to pcre_study():
1319
1320 pcre_extra *pe;
1321 pe = pcre_study(
1322 re, /* result of pcre_compile() */
1323 0, /* no options exist */
1324 &error); /* set to NULL or points to a message */
1325
1326 At present, studying a pattern is useful only for non-anchored patterns
1327 that do not have a single fixed starting character. A bitmap of possi-
1328 ble starting bytes is created.
1329
1330
1331 LOCALE SUPPORT
1332
1333 PCRE handles caseless matching, and determines whether characters are
1334 letters, digits, or whatever, by reference to a set of tables, indexed
1335 by character value. When running in UTF-8 mode, this applies only to
1336 characters with codes less than 128. Higher-valued codes never match
1337 escapes such as \w or \d, but can be tested with \p if PCRE is built
1338 with Unicode character property support. The use of locales with Uni-
1339 code is discouraged. If you are handling characters with codes greater
1340 than 128, you should either use UTF-8 and Unicode, or use locales, but
1341 not try to mix the two.
1342
1343 PCRE contains an internal set of tables that are used when the final
1344 argument of pcre_compile() is NULL. These are sufficient for many
1345 applications. Normally, the internal tables recognize only ASCII char-
1346 acters. However, when PCRE is built, it is possible to cause the inter-
1347 nal tables to be rebuilt in the default "C" locale of the local system,
1348 which may cause them to be different.
1349
1350 The internal tables can always be overridden by tables supplied by the
1351 application that calls PCRE. These may be created in a different locale
1352 from the default. As more and more applications change to using Uni-
1353 code, the need for this locale support is expected to die away.
1354
1355 External tables are built by calling the pcre_maketables() function,
1356 which has no arguments, in the relevant locale. The result can then be
1357 passed to pcre_compile() or pcre_exec() as often as necessary. For
1358 example, to build and use tables that are appropriate for the French
1359 locale (where accented characters with values greater than 128 are
1360 treated as letters), the following code could be used:
1361
1362 setlocale(LC_CTYPE, "fr_FR");
1363 tables = pcre_maketables();
1364 re = pcre_compile(..., tables);
1365
1366 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1367 if you are using Windows, the name for the French locale is "french".
1368
1369 When pcre_maketables() runs, the tables are built in memory that is
1370 obtained via pcre_malloc. It is the caller's responsibility to ensure
1371 that the memory containing the tables remains available for as long as
1372 it is needed.
1373
1374 The pointer that is passed to pcre_compile() is saved with the compiled
1375 pattern, and the same tables are used via this pointer by pcre_study()
1376 and normally also by pcre_exec(). Thus, by default, for any single pat-
1377 tern, compilation, studying and matching all happen in the same locale,
1378 but different patterns can be compiled in different locales.
1379
1380 It is possible to pass a table pointer or NULL (indicating the use of
1381 the internal tables) to pcre_exec(). Although not intended for this
1382 purpose, this facility could be used to match a pattern in a different
1383 locale from the one in which it was compiled. Passing table pointers at
1384 run time is discussed below in the section on matching a pattern.
1385
1386
1387 INFORMATION ABOUT A PATTERN
1388
1389 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1390 int what, void *where);
1391
1392 The pcre_fullinfo() function returns information about a compiled pat-
1393 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1394 less retained for backwards compability (and is documented below).
1395
1396 The first argument for pcre_fullinfo() is a pointer to the compiled
1397 pattern. The second argument is the result of pcre_study(), or NULL if
1398 the pattern was not studied. The third argument specifies which piece
1399 of information is required, and the fourth argument is a pointer to a
1400 variable to receive the data. The yield of the function is zero for
1401 success, or one of the following negative numbers:
1402
1403 PCRE_ERROR_NULL the argument code was NULL
1404 the argument where was NULL
1405 PCRE_ERROR_BADMAGIC the "magic number" was not found
1406 PCRE_ERROR_BADOPTION the value of what was invalid
1407
1408 The "magic number" is placed at the start of each compiled pattern as
1409 an simple check against passing an arbitrary memory pointer. Here is a
1410 typical call of pcre_fullinfo(), to obtain the length of the compiled
1411 pattern:
1412
1413 int rc;
1414 size_t length;
1415 rc = pcre_fullinfo(
1416 re, /* result of pcre_compile() */
1417 pe, /* result of pcre_study(), or NULL */
1418 PCRE_INFO_SIZE, /* what is required */
1419 &length); /* where to put the data */
1420
1421 The possible values for the third argument are defined in pcre.h, and
1422 are as follows:
1423
1424 PCRE_INFO_BACKREFMAX
1425
1426 Return the number of the highest back reference in the pattern. The
1427 fourth argument should point to an int variable. Zero is returned if
1428 there are no back references.
1429
1430 PCRE_INFO_CAPTURECOUNT
1431
1432 Return the number of capturing subpatterns in the pattern. The fourth
1433 argument should point to an int variable.
1434
1435 PCRE_INFO_DEFAULT_TABLES
1436
1437 Return a pointer to the internal default character tables within PCRE.
1438 The fourth argument should point to an unsigned char * variable. This
1439 information call is provided for internal use by the pcre_study() func-
1440 tion. External callers can cause PCRE to use its internal tables by
1441 passing a NULL table pointer.
1442
1443 PCRE_INFO_FIRSTBYTE
1444
1445 Return information about the first byte of any matched string, for a
1446 non-anchored pattern. The fourth argument should point to an int vari-
1447 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1448 is still recognized for backwards compatibility.)
1449
1450 If there is a fixed first byte, for example, from a pattern such as
1451 (cat|cow|coyote), its value is returned. Otherwise, if either
1452
1453 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1454 branch starts with "^", or
1455
1456 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1457 set (if it were set, the pattern would be anchored),
1458
1459 -1 is returned, indicating that the pattern matches only at the start
1460 of a subject string or after any newline within the string. Otherwise
1461 -2 is returned. For anchored patterns, -2 is returned.
1462
1463 PCRE_INFO_FIRSTTABLE
1464
1465 If the pattern was studied, and this resulted in the construction of a
1466 256-bit table indicating a fixed set of bytes for the first byte in any
1467 matching string, a pointer to the table is returned. Otherwise NULL is
1468 returned. The fourth argument should point to an unsigned char * vari-
1469 able.
1470
1471 PCRE_INFO_LASTLITERAL
1472
1473 Return the value of the rightmost literal byte that must exist in any
1474 matched string, other than at its start, if such a byte has been
1475 recorded. The fourth argument should point to an int variable. If there
1476 is no such byte, -1 is returned. For anchored patterns, a last literal
1477 byte is recorded only if it follows something of variable length. For
1478 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1479 /^a\dz\d/ the returned value is -1.
1480
1481 PCRE_INFO_NAMECOUNT
1482 PCRE_INFO_NAMEENTRYSIZE
1483 PCRE_INFO_NAMETABLE
1484
1485 PCRE supports the use of named as well as numbered capturing parenthe-
1486 ses. The names are just an additional way of identifying the parenthe-
1487 ses, which still acquire numbers. Several convenience functions such as
1488 pcre_get_named_substring() are provided for extracting captured sub-
1489 strings by name. It is also possible to extract the data directly, by
1490 first converting the name to a number in order to access the correct
1491 pointers in the output vector (described with pcre_exec() below). To do
1492 the conversion, you need to use the name-to-number map, which is
1493 described by these three values.
1494
1495 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1496 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1497 of each entry; both of these return an int value. The entry size
1498 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1499 a pointer to the first entry of the table (a pointer to char). The
1500 first two bytes of each entry are the number of the capturing parenthe-
1501 sis, most significant byte first. The rest of the entry is the corre-
1502 sponding name, zero terminated. The names are in alphabetical order.
1503 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1504 theses numbers. For example, consider the following pattern (assume
1505 PCRE_EXTENDED is set, so white space - including newlines - is
1506 ignored):
1507
1508 (?<date> (?<year>(\d\d)?\d\d) -
1509 (?<month>\d\d) - (?<day>\d\d) )
1510
1511 There are four named subpatterns, so the table has four entries, and
1512 each entry in the table is eight bytes long. The table is as follows,
1513 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1514 as ??:
1515
1516 00 01 d a t e 00 ??
1517 00 05 d a y 00 ?? ??
1518 00 04 m o n t h 00
1519 00 02 y e a r 00 ??
1520
1521 When writing code to extract data from named subpatterns using the
1522 name-to-number map, remember that the length of the entries is likely
1523 to be different for each compiled pattern.
1524
1525 PCRE_INFO_OPTIONS
1526
1527 Return a copy of the options with which the pattern was compiled. The
1528 fourth argument should point to an unsigned long int variable. These
1529 option bits are those specified in the call to pcre_compile(), modified
1530 by any top-level option settings within the pattern itself.
1531
1532 A pattern is automatically anchored by PCRE if all of its top-level
1533 alternatives begin with one of the following:
1534
1535 ^ unless PCRE_MULTILINE is set
1536 \A always
1537 \G always
1538 .* if PCRE_DOTALL is set and there are no back
1539 references to the subpattern in which .* appears
1540
1541 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1542 by pcre_fullinfo().
1543
1544 PCRE_INFO_SIZE
1545
1546 Return the size of the compiled pattern, that is, the value that was
1547 passed as the argument to pcre_malloc() when PCRE was getting memory in
1548 which to place the compiled data. The fourth argument should point to a
1549 size_t variable.
1550
1551 PCRE_INFO_STUDYSIZE
1552
1553 Return the size of the data block pointed to by the study_data field in
1554 a pcre_extra block. That is, it is the value that was passed to
1555 pcre_malloc() when PCRE was getting memory into which to place the data
1556 created by pcre_study(). The fourth argument should point to a size_t
1557 variable.
1558
1559
1560 OBSOLETE INFO FUNCTION
1561
1562 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1563
1564 The pcre_info() function is now obsolete because its interface is too
1565 restrictive to return all the available data about a compiled pattern.
1566 New programs should use pcre_fullinfo() instead. The yield of
1567 pcre_info() is the number of capturing subpatterns, or one of the fol-
1568 lowing negative numbers:
1569
1570 PCRE_ERROR_NULL the argument code was NULL
1571 PCRE_ERROR_BADMAGIC the "magic number" was not found
1572
1573 If the optptr argument is not NULL, a copy of the options with which
1574 the pattern was compiled is placed in the integer it points to (see
1575 PCRE_INFO_OPTIONS above).
1576
1577 If the pattern is not anchored and the firstcharptr argument is not
1578 NULL, it is used to pass back information about the first character of
1579 any matched string (see PCRE_INFO_FIRSTBYTE above).
1580
1581
1582 REFERENCE COUNTS
1583
1584 int pcre_refcount(pcre *code, int adjust);
1585
1586 The pcre_refcount() function is used to maintain a reference count in
1587 the data block that contains a compiled pattern. It is provided for the
1588 benefit of applications that operate in an object-oriented manner,
1589 where different parts of the application may be using the same compiled
1590 pattern, but you want to free the block when they are all done.
1591
1592 When a pattern is compiled, the reference count field is initialized to
1593 zero. It is changed only by calling this function, whose action is to
1594 add the adjust value (which may be positive or negative) to it. The
1595 yield of the function is the new value. However, the value of the count
1596 is constrained to lie between 0 and 65535, inclusive. If the new value
1597 is outside these limits, it is forced to the appropriate limit value.
1598
1599 Except when it is zero, the reference count is not correctly preserved
1600 if a pattern is compiled on one host and then transferred to a host
1601 whose byte-order is different. (This seems a highly unlikely scenario.)
1602
1603
1604 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1605
1606 int pcre_exec(const pcre *code, const pcre_extra *extra,
1607 const char *subject, int length, int startoffset,
1608 int options, int *ovector, int ovecsize);
1609
1610 The function pcre_exec() is called to match a subject string against a
1611 compiled pattern, which is passed in the code argument. If the pattern
1612 has been studied, the result of the study should be passed in the extra
1613 argument. This function is the main matching facility of the library,
1614 and it operates in a Perl-like manner. For specialist use there is also
1615 an alternative matching function, which is described below in the sec-
1616 tion about the pcre_dfa_exec() function.
1617
1618 In most applications, the pattern will have been compiled (and option-
1619 ally studied) in the same process that calls pcre_exec(). However, it
1620 is possible to save compiled patterns and study data, and then use them
1621 later in different processes, possibly even on different hosts. For a
1622 discussion about this, see the pcreprecompile documentation.
1623
1624 Here is an example of a simple call to pcre_exec():
1625
1626 int rc;
1627 int ovector[30];
1628 rc = pcre_exec(
1629 re, /* result of pcre_compile() */
1630 NULL, /* we didn't study the pattern */
1631 "some string", /* the subject string */
1632 11, /* the length of the subject string */
1633 0, /* start at offset 0 in the subject */
1634 0, /* default options */
1635 ovector, /* vector of integers for substring information */
1636 30); /* number of elements (NOT size in bytes) */
1637
1638 Extra data for pcre_exec()
1639
1640 If the extra argument is not NULL, it must point to a pcre_extra data
1641 block. The pcre_study() function returns such a block (when it doesn't
1642 return NULL), but you can also create one for yourself, and pass addi-
1643 tional information in it. The pcre_extra block contains the following
1644 fields (not necessarily in this order):
1645
1646 unsigned long int flags;
1647 void *study_data;
1648 unsigned long int match_limit;
1649 unsigned long int match_limit_recursion;
1650 void *callout_data;
1651 const unsigned char *tables;
1652
1653 The flags field is a bitmap that specifies which of the other fields
1654 are set. The flag bits are:
1655
1656 PCRE_EXTRA_STUDY_DATA
1657 PCRE_EXTRA_MATCH_LIMIT
1658 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1659 PCRE_EXTRA_CALLOUT_DATA
1660 PCRE_EXTRA_TABLES
1661
1662 Other flag bits should be set to zero. The study_data field is set in
1663 the pcre_extra block that is returned by pcre_study(), together with
1664 the appropriate flag bit. You should not set this yourself, but you may
1665 add to the block by setting the other fields and their corresponding
1666 flag bits.
1667
1668 The match_limit field provides a means of preventing PCRE from using up
1669 a vast amount of resources when running patterns that are not going to
1670 match, but which have a very large number of possibilities in their
1671 search trees. The classic example is the use of nested unlimited
1672 repeats.
1673
1674 Internally, PCRE uses a function called match() which it calls repeat-
1675 edly (sometimes recursively). The limit set by match_limit is imposed
1676 on the number of times this function is called during a match, which
1677 has the effect of limiting the amount of backtracking that can take
1678 place. For patterns that are not anchored, the count restarts from zero
1679 for each position in the subject string.
1680
1681 The default value for the limit can be set when PCRE is built; the
1682 default default is 10 million, which handles all but the most extreme
1683 cases. You can override the default by suppling pcre_exec() with a
1684 pcre_extra block in which match_limit is set, and
1685 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1686 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1687
1688 The match_limit_recursion field is similar to match_limit, but instead
1689 of limiting the total number of times that match() is called, it limits
1690 the depth of recursion. The recursion depth is a smaller number than
1691 the total number of calls, because not all calls to match() are recur-
1692 sive. This limit is of use only if it is set smaller than match_limit.
1693
1694 Limiting the recursion depth limits the amount of stack that can be
1695 used, or, when PCRE has been compiled to use memory on the heap instead
1696 of the stack, the amount of heap memory that can be used.
1697
1698 The default value for match_limit_recursion can be set when PCRE is
1699 built; the default default is the same value as the default for
1700 match_limit. You can override the default by suppling pcre_exec() with
1701 a pcre_extra block in which match_limit_recursion is set, and
1702 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1703 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1704
1705 The pcre_callout field is used in conjunction with the "callout" fea-
1706 ture, which is described in the pcrecallout documentation.
1707
1708 The tables field is used to pass a character tables pointer to
1709 pcre_exec(); this overrides the value that is stored with the compiled
1710 pattern. A non-NULL value is stored with the compiled pattern only if
1711 custom tables were supplied to pcre_compile() via its tableptr argu-
1712 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1713 PCRE's internal tables to be used. This facility is helpful when re-
1714 using patterns that have been saved after compiling with an external
1715 set of tables, because the external tables might be at a different
1716 address when pcre_exec() is called. See the pcreprecompile documenta-
1717 tion for a discussion of saving compiled patterns for later use.
1718
1719 Option bits for pcre_exec()
1720
1721 The unused bits of the options argument for pcre_exec() must be zero.
1722 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1723 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1724 PCRE_PARTIAL.
1725
1726 PCRE_ANCHORED
1727
1728 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1729 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1730 turned out to be anchored by virtue of its contents, it cannot be made
1731 unachored at matching time.
1732
1733 PCRE_NEWLINE_CR
1734 PCRE_NEWLINE_LF
1735 PCRE_NEWLINE_CRLF
1736 PCRE_NEWLINE_ANYCRLF
1737 PCRE_NEWLINE_ANY
1738
1739 These options override the newline definition that was chosen or
1740 defaulted when the pattern was compiled. For details, see the descrip-
1741 tion of pcre_compile() above. During matching, the newline choice
1742 affects the behaviour of the dot, circumflex, and dollar metacharac-
1743 ters. It may also alter the way the match position is advanced after a
1744 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF,
1745 PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is set, and a match attempt
1746 fails when the current position is at a CRLF sequence, the match posi-
1747 tion is advanced by two characters instead of one, in other words, to
1748 after the CRLF.
1749
1750 PCRE_NOTBOL
1751
1752 This option specifies that first character of the subject string is not
1753 the beginning of a line, so the circumflex metacharacter should not
1754 match before it. Setting this without PCRE_MULTILINE (at compile time)
1755 causes circumflex never to match. This option affects only the behav-
1756 iour of the circumflex metacharacter. It does not affect \A.
1757
1758 PCRE_NOTEOL
1759
1760 This option specifies that the end of the subject string is not the end
1761 of a line, so the dollar metacharacter should not match it nor (except
1762 in multiline mode) a newline immediately before it. Setting this with-
1763 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1764 option affects only the behaviour of the dollar metacharacter. It does
1765 not affect \Z or \z.
1766
1767 PCRE_NOTEMPTY
1768
1769 An empty string is not considered to be a valid match if this option is
1770 set. If there are alternatives in the pattern, they are tried. If all
1771 the alternatives match the empty string, the entire match fails. For
1772 example, if the pattern
1773
1774 a?b?
1775
1776 is applied to a string not beginning with "a" or "b", it matches the
1777 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1778 match is not valid, so PCRE searches further into the string for occur-
1779 rences of "a" or "b".
1780
1781 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1782 cial case of a pattern match of the empty string within its split()
1783 function, and when using the /g modifier. It is possible to emulate
1784 Perl's behaviour after matching a null string by first trying the match
1785 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1786 if that fails by advancing the starting offset (see below) and trying
1787 an ordinary match again. There is some code that demonstrates how to do
1788 this in the pcredemo.c sample program.
1789
1790 PCRE_NO_UTF8_CHECK
1791
1792 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1793 UTF-8 string is automatically checked when pcre_exec() is subsequently
1794 called. The value of startoffset is also checked to ensure that it
1795 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1796 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1797 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1798 returned.
1799
1800 If you already know that your subject is valid, and you want to skip
1801 these checks for performance reasons, you can set the
1802 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1803 do this for the second and subsequent calls to pcre_exec() if you are
1804 making repeated calls to find all the matches in a single subject
1805 string. However, you should be sure that the value of startoffset
1806 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1807 set, the effect of passing an invalid UTF-8 string as a subject, or a
1808 value of startoffset that does not point to the start of a UTF-8 char-
1809 acter, is undefined. Your program may crash.
1810
1811 PCRE_PARTIAL
1812
1813 This option turns on the partial matching feature. If the subject
1814 string fails to match the pattern, but at some point during the match-
1815 ing process the end of the subject was reached (that is, the subject
1816 partially matches the pattern and the failure to match occurred only
1817 because there were not enough subject characters), pcre_exec() returns
1818 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1819 used, there are restrictions on what may appear in the pattern. These
1820 are discussed in the pcrepartial documentation.
1821
1822 The string to be matched by pcre_exec()
1823
1824 The subject string is passed to pcre_exec() as a pointer in subject, a
1825 length in length, and a starting byte offset in startoffset. In UTF-8
1826 mode, the byte offset must point to the start of a UTF-8 character.
1827 Unlike the pattern string, the subject may contain binary zero bytes.
1828 When the starting offset is zero, the search for a match starts at the
1829 beginning of the subject, and this is by far the most common case.
1830
1831 A non-zero starting offset is useful when searching for another match
1832 in the same subject by calling pcre_exec() again after a previous suc-
1833 cess. Setting startoffset differs from just passing over a shortened
1834 string and setting PCRE_NOTBOL in the case of a pattern that begins
1835 with any kind of lookbehind. For example, consider the pattern
1836
1837 \Biss\B
1838
1839 which finds occurrences of "iss" in the middle of words. (\B matches
1840 only if the current position in the subject is not a word boundary.)
1841 When applied to the string "Mississipi" the first call to pcre_exec()
1842 finds the first occurrence. If pcre_exec() is called again with just
1843 the remainder of the subject, namely "issipi", it does not match,
1844 because \B is always false at the start of the subject, which is deemed
1845 to be a word boundary. However, if pcre_exec() is passed the entire
1846 string again, but with startoffset set to 4, it finds the second occur-
1847 rence of "iss" because it is able to look behind the starting point to
1848 discover that it is preceded by a letter.
1849
1850 If a non-zero starting offset is passed when the pattern is anchored,
1851 one attempt to match at the given offset is made. This can only succeed
1852 if the pattern does not require the match to be at the start of the
1853 subject.
1854
1855 How pcre_exec() returns captured substrings
1856
1857 In general, a pattern matches a certain portion of the subject, and in
1858 addition, further substrings from the subject may be picked out by
1859 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1860 this is called "capturing" in what follows, and the phrase "capturing
1861 subpattern" is used for a fragment of a pattern that picks out a sub-
1862 string. PCRE supports several other kinds of parenthesized subpattern
1863 that do not cause substrings to be captured.
1864
1865 Captured substrings are returned to the caller via a vector of integer
1866 offsets whose address is passed in ovector. The number of elements in
1867 the vector is passed in ovecsize, which must be a non-negative number.
1868 Note: this argument is NOT the size of ovector in bytes.
1869
1870 The first two-thirds of the vector is used to pass back captured sub-
1871 strings, each substring using a pair of integers. The remaining third
1872 of the vector is used as workspace by pcre_exec() while matching cap-
1873 turing subpatterns, and is not available for passing back information.
1874 The length passed in ovecsize should always be a multiple of three. If
1875 it is not, it is rounded down.
1876
1877 When a match is successful, information about captured substrings is
1878 returned in pairs of integers, starting at the beginning of ovector,
1879 and continuing up to two-thirds of its length at the most. The first
1880 element of a pair is set to the offset of the first character in a sub-
1881 string, and the second is set to the offset of the first character
1882 after the end of a substring. The first pair, ovector[0] and ovec-
1883 tor[1], identify the portion of the subject string matched by the
1884 entire pattern. The next pair is used for the first capturing subpat-
1885 tern, and so on. The value returned by pcre_exec() is one more than the
1886 highest numbered pair that has been set. For example, if two substrings
1887 have been captured, the returned value is 3. If there are no capturing
1888 subpatterns, the return value from a successful match is 1, indicating
1889 that just the first pair of offsets has been set.
1890
1891 If a capturing subpattern is matched repeatedly, it is the last portion
1892 of the string that it matched that is returned.
1893
1894 If the vector is too small to hold all the captured substring offsets,
1895 it is used as far as possible (up to two-thirds of its length), and the
1896 function returns a value of zero. In particular, if the substring off-
1897 sets are not of interest, pcre_exec() may be called with ovector passed
1898 as NULL and ovecsize as zero. However, if the pattern contains back
1899 references and the ovector is not big enough to remember the related
1900 substrings, PCRE has to get additional memory for use during matching.
1901 Thus it is usually advisable to supply an ovector.
1902
1903 The pcre_info() function can be used to find out how many capturing
1904 subpatterns there are in a compiled pattern. The smallest size for
1905 ovector that will allow for n captured substrings, in addition to the
1906 offsets of the substring matched by the whole pattern, is (n+1)*3.
1907
1908 It is possible for capturing subpattern number n+1 to match some part
1909 of the subject when subpattern n has not been used at all. For example,
1910 if the string "abc" is matched against the pattern (a|(z))(bc) the
1911 return from the function is 4, and subpatterns 1 and 3 are matched, but
1912 2 is not. When this happens, both values in the offset pairs corre-
1913 sponding to unused subpatterns are set to -1.
1914
1915 Offset values that correspond to unused subpatterns at the end of the
1916 expression are also set to -1. For example, if the string "abc" is
1917 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1918 matched. The return from the function is 2, because the highest used
1919 capturing subpattern number is 1. However, you can refer to the offsets
1920 for the second and third capturing subpatterns if you wish (assuming
1921 the vector is large enough, of course).
1922
1923 Some convenience functions are provided for extracting the captured
1924 substrings as separate strings. These are described below.
1925
1926 Error return values from pcre_exec()
1927
1928 If pcre_exec() fails, it returns a negative number. The following are
1929 defined in the header file:
1930
1931 PCRE_ERROR_NOMATCH (-1)
1932
1933 The subject string did not match the pattern.
1934
1935 PCRE_ERROR_NULL (-2)
1936
1937 Either code or subject was passed as NULL, or ovector was NULL and
1938 ovecsize was not zero.
1939
1940 PCRE_ERROR_BADOPTION (-3)
1941
1942 An unrecognized bit was set in the options argument.
1943
1944 PCRE_ERROR_BADMAGIC (-4)
1945
1946 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1947 to catch the case when it is passed a junk pointer and to detect when a
1948 pattern that was compiled in an environment of one endianness is run in
1949 an environment with the other endianness. This is the error that PCRE
1950 gives when the magic number is not present.
1951
1952 PCRE_ERROR_UNKNOWN_OPCODE (-5)
1953
1954 While running the pattern match, an unknown item was encountered in the
1955 compiled pattern. This error could be caused by a bug in PCRE or by
1956 overwriting of the compiled pattern.
1957
1958 PCRE_ERROR_NOMEMORY (-6)
1959
1960 If a pattern contains back references, but the ovector that is passed
1961 to pcre_exec() is not big enough to remember the referenced substrings,
1962 PCRE gets a block of memory at the start of matching to use for this
1963 purpose. If the call via pcre_malloc() fails, this error is given. The
1964 memory is automatically freed at the end of matching.
1965
1966 PCRE_ERROR_NOSUBSTRING (-7)
1967
1968 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1969 and pcre_get_substring_list() functions (see below). It is never
1970 returned by pcre_exec().
1971
1972 PCRE_ERROR_MATCHLIMIT (-8)
1973
1974 The backtracking limit, as specified by the match_limit field in a
1975 pcre_extra structure (or defaulted) was reached. See the description
1976 above.
1977
1978 PCRE_ERROR_CALLOUT (-9)
1979
1980 This error is never generated by pcre_exec() itself. It is provided for
1981 use by callout functions that want to yield a distinctive error code.
1982 See the pcrecallout documentation for details.
1983
1984 PCRE_ERROR_BADUTF8 (-10)
1985
1986 A string that contains an invalid UTF-8 byte sequence was passed as a
1987 subject.
1988
1989 PCRE_ERROR_BADUTF8_OFFSET (-11)
1990
1991 The UTF-8 byte sequence that was passed as a subject was valid, but the
1992 value of startoffset did not point to the beginning of a UTF-8 charac-
1993 ter.
1994
1995 PCRE_ERROR_PARTIAL (-12)
1996
1997 The subject string did not match, but it did match partially. See the
1998 pcrepartial documentation for details of partial matching.
1999
2000 PCRE_ERROR_BADPARTIAL (-13)
2001
2002 The PCRE_PARTIAL option was used with a compiled pattern containing
2003 items that are not supported for partial matching. See the pcrepartial
2004 documentation for details of partial matching.
2005
2006 PCRE_ERROR_INTERNAL (-14)
2007
2008 An unexpected internal error has occurred. This error could be caused
2009 by a bug in PCRE or by overwriting of the compiled pattern.
2010
2011 PCRE_ERROR_BADCOUNT (-15)
2012
2013 This error is given if the value of the ovecsize argument is negative.
2014
2015 PCRE_ERROR_RECURSIONLIMIT (-21)
2016
2017 The internal recursion limit, as specified by the match_limit_recursion
2018 field in a pcre_extra structure (or defaulted) was reached. See the
2019 description above.
2020
2021 PCRE_ERROR_NULLWSLIMIT (-22)
2022
2023 When a group that can match an empty substring is repeated with an
2024 unbounded upper limit, the subject position at the start of the group
2025 must be remembered, so that a test for an empty string can be made when
2026 the end of the group is reached. Some workspace is required for this;
2027 if it runs out, this error is given.
2028
2029 PCRE_ERROR_BADNEWLINE (-23)
2030
2031 An invalid combination of PCRE_NEWLINE_xxx options was given.
2032
2033 Error numbers -16 to -20 are not used by pcre_exec().
2034
2035
2036 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2037
2038 int pcre_copy_substring(const char *subject, int *ovector,
2039 int stringcount, int stringnumber, char *buffer,
2040 int buffersize);
2041
2042 int pcre_get_substring(const char *subject, int *ovector,
2043 int stringcount, int stringnumber,
2044 const char **stringptr);
2045
2046 int pcre_get_substring_list(const char *subject,
2047 int *ovector, int stringcount, const char ***listptr);
2048
2049 Captured substrings can be accessed directly by using the offsets
2050 returned by pcre_exec() in ovector. For convenience, the functions
2051 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2052 string_list() are provided for extracting captured substrings as new,
2053 separate, zero-terminated strings. These functions identify substrings
2054 by number. The next section describes functions for extracting named
2055 substrings.
2056
2057 A substring that contains a binary zero is correctly extracted and has
2058 a further zero added on the end, but the result is not, of course, a C
2059 string. However, you can process such a string by referring to the
2060 length that is returned by pcre_copy_substring() and pcre_get_sub-
2061 string(). Unfortunately, the interface to pcre_get_substring_list() is
2062 not adequate for handling strings containing binary zeros, because the
2063 end of the final string is not independently indicated.
2064
2065 The first three arguments are the same for all three of these func-
2066 tions: subject is the subject string that has just been successfully
2067 matched, ovector is a pointer to the vector of integer offsets that was
2068 passed to pcre_exec(), and stringcount is the number of substrings that
2069 were captured by the match, including the substring that matched the
2070 entire regular expression. This is the value returned by pcre_exec() if
2071 it is greater than zero. If pcre_exec() returned zero, indicating that
2072 it ran out of space in ovector, the value passed as stringcount should
2073 be the number of elements in the vector divided by three.
2074
2075 The functions pcre_copy_substring() and pcre_get_substring() extract a
2076 single substring, whose number is given as stringnumber. A value of
2077 zero extracts the substring that matched the entire pattern, whereas
2078 higher values extract the captured substrings. For pcre_copy_sub-
2079 string(), the string is placed in buffer, whose length is given by
2080 buffersize, while for pcre_get_substring() a new block of memory is
2081 obtained via pcre_malloc, and its address is returned via stringptr.
2082 The yield of the function is the length of the string, not including
2083 the terminating zero, or one of these error codes:
2084
2085 PCRE_ERROR_NOMEMORY (-6)
2086
2087 The buffer was too small for pcre_copy_substring(), or the attempt to
2088 get memory failed for pcre_get_substring().
2089
2090 PCRE_ERROR_NOSUBSTRING (-7)
2091
2092 There is no substring whose number is stringnumber.
2093
2094 The pcre_get_substring_list() function extracts all available sub-
2095 strings and builds a list of pointers to them. All this is done in a
2096 single block of memory that is obtained via pcre_malloc. The address of
2097 the memory block is returned via listptr, which is also the start of
2098 the list of string pointers. The end of the list is marked by a NULL
2099 pointer. The yield of the function is zero if all went well, or the
2100 error code
2101
2102 PCRE_ERROR_NOMEMORY (-6)
2103
2104 if the attempt to get the memory block failed.
2105
2106 When any of these functions encounter a substring that is unset, which
2107 can happen when capturing subpattern number n+1 matches some part of
2108 the subject, but subpattern n has not been used at all, they return an
2109 empty string. This can be distinguished from a genuine zero-length sub-
2110 string by inspecting the appropriate offset in ovector, which is nega-
2111 tive for unset substrings.
2112
2113 The two convenience functions pcre_free_substring() and pcre_free_sub-
2114 string_list() can be used to free the memory returned by a previous
2115 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2116 tively. They do nothing more than call the function pointed to by
2117 pcre_free, which of course could be called directly from a C program.
2118 However, PCRE is used in some situations where it is linked via a spe-
2119 cial interface to another programming language that cannot use
2120 pcre_free directly; it is for these cases that the functions are pro-
2121 vided.
2122
2123
2124 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2125
2126 int pcre_get_stringnumber(const pcre *code,
2127 const char *name);
2128
2129 int pcre_copy_named_substring(const pcre *code,
2130 const char *subject, int *ovector,
2131 int stringcount, const char *stringname,
2132 char *buffer, int buffersize);
2133
2134 int pcre_get_named_substring(const pcre *code,
2135 const char *subject, int *ovector,
2136 int stringcount, const char *stringname,
2137 const char **stringptr);
2138
2139 To extract a substring by name, you first have to find associated num-
2140 ber. For example, for this pattern
2141
2142 (a+)b(?<xxx>\d+)...
2143
2144 the number of the subpattern called "xxx" is 2. If the name is known to
2145 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2146 name by calling pcre_get_stringnumber(). The first argument is the com-
2147 piled pattern, and the second is the name. The yield of the function is
2148 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2149 subpattern of that name.
2150
2151 Given the number, you can extract the substring directly, or use one of
2152 the functions described in the previous section. For convenience, there
2153 are also two functions that do the whole job.
2154
2155 Most of the arguments of pcre_copy_named_substring() and
2156 pcre_get_named_substring() are the same as those for the similarly
2157 named functions that extract by number. As these are described in the
2158 previous section, they are not re-described here. There are just two
2159 differences:
2160
2161 First, instead of a substring number, a substring name is given. Sec-
2162 ond, there is an extra argument, given at the start, which is a pointer
2163 to the compiled pattern. This is needed in order to gain access to the
2164 name-to-number translation table.
2165
2166 These functions call pcre_get_stringnumber(), and if it succeeds, they
2167 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2168 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2169 behaviour may not be what you want (see the next section).
2170
2171
2172 DUPLICATE SUBPATTERN NAMES
2173
2174 int pcre_get_stringtable_entries(const pcre *code,
2175 const char *name, char **first, char **last);
2176
2177 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2178 subpatterns are not required to be unique. Normally, patterns with
2179 duplicate names are such that in any one match, only one of the named
2180 subpatterns participates. An example is shown in the pcrepattern docu-
2181 mentation. When duplicates are present, pcre_copy_named_substring() and
2182 pcre_get_named_substring() return the first substring corresponding to
2183 the given name that is set. If none are set, an empty string is
2184 returned. The pcre_get_stringnumber() function returns one of the num-
2185 bers that are associated with the name, but it is not defined which it
2186 is.
2187
2188 If you want to get full details of all captured substrings for a given
2189 name, you must use the pcre_get_stringtable_entries() function. The
2190 first argument is the compiled pattern, and the second is the name. The
2191 third and fourth are pointers to variables which are updated by the
2192 function. After it has run, they point to the first and last entries in
2193 the name-to-number table for the given name. The function itself
2194 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2195 there are none. The format of the table is described above in the sec-
2196 tion entitled Information about a pattern. Given all the relevant
2197 entries for the name, you can extract each of their numbers, and hence
2198 the captured data, if any.
2199
2200
2201 FINDING ALL POSSIBLE MATCHES
2202
2203 The traditional matching function uses a similar algorithm to Perl,
2204 which stops when it finds the first match, starting at a given point in
2205 the subject. If you want to find all possible matches, or the longest
2206 possible match, consider using the alternative matching function (see
2207 below) instead. If you cannot use the alternative function, but still
2208 need to find all possible matches, you can kludge it up by making use
2209 of the callout facility, which is described in the pcrecallout documen-
2210 tation.
2211
2212 What you have to do is to insert a callout right at the end of the pat-
2213 tern. When your callout function is called, extract and save the cur-
2214 rent matched substring. Then return 1, which forces pcre_exec() to
2215 backtrack and try other alternatives. Ultimately, when it runs out of
2216 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2217
2218
2219 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2220
2221 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2222 const char *subject, int length, int startoffset,
2223 int options, int *ovector, int ovecsize,
2224 int *workspace, int wscount);
2225
2226 The function pcre_dfa_exec() is called to match a subject string
2227 against a compiled pattern, using a matching algorithm that scans the
2228 subject string just once, and does not backtrack. This has different
2229 characteristics to the normal algorithm, and is not compatible with
2230 Perl. Some of the features of PCRE patterns are not supported. Never-
2231 theless, there are times when this kind of matching can be useful. For
2232 a discussion of the two matching algorithms, see the pcrematching docu-
2233 mentation.
2234
2235 The arguments for the pcre_dfa_exec() function are the same as for
2236 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2237 ent way, and this is described below. The other common arguments are
2238 used in the same way as for pcre_exec(), so their description is not
2239 repeated here.
2240
2241 The two additional arguments provide workspace for the function. The
2242 workspace vector should contain at least 20 elements. It is used for
2243 keeping track of multiple paths through the pattern tree. More
2244 workspace will be needed for patterns and subjects where there are a
2245 lot of potential matches.
2246
2247 Here is an example of a simple call to pcre_dfa_exec():
2248
2249 int rc;
2250 int ovector[10];
2251 int wspace[20];
2252 rc = pcre_dfa_exec(
2253 re, /* result of pcre_compile() */
2254 NULL, /* we didn't study the pattern */
2255 "some string", /* the subject string */
2256 11, /* the length of the subject string */
2257 0, /* start at offset 0 in the subject */
2258 0, /* default options */
2259 ovector, /* vector of integers for substring information */
2260 10, /* number of elements (NOT size in bytes) */
2261 wspace, /* working space vector */
2262 20); /* number of elements (NOT size in bytes) */
2263
2264 Option bits for pcre_dfa_exec()
2265
2266 The unused bits of the options argument for pcre_dfa_exec() must be
2267 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2268 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2269 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2270 three of these are the same as for pcre_exec(), so their description is
2271 not repeated here.
2272
2273 PCRE_PARTIAL
2274
2275 This has the same general effect as it does for pcre_exec(), but the
2276 details are slightly different. When PCRE_PARTIAL is set for
2277 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2278 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2279 been no complete matches, but there is still at least one matching pos-
2280 sibility. The portion of the string that provided the partial match is
2281 set as the first matching string.
2282
2283 PCRE_DFA_SHORTEST
2284
2285 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2286 stop as soon as it has found one match. Because of the way the alterna-
2287 tive algorithm works, this is necessarily the shortest possible match
2288 at the first possible matching point in the subject string.
2289
2290 PCRE_DFA_RESTART
2291
2292 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2293 returns a partial match, it is possible to call it again, with addi-
2294 tional subject characters, and have it continue with the same match.
2295 The PCRE_DFA_RESTART option requests this action; when it is set, the
2296 workspace and wscount options must reference the same vector as before
2297 because data about the match so far is left in them after a partial
2298 match. There is more discussion of this facility in the pcrepartial
2299 documentation.
2300
2301 Successful returns from pcre_dfa_exec()
2302
2303 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2304 string in the subject. Note, however, that all the matches from one run
2305 of the function start at the same point in the subject. The shorter
2306 matches are all initial substrings of the longer matches. For example,
2307 if the pattern
2308
2309 <.*>
2310
2311 is matched against the string
2312
2313 This is <something> <something else> <something further> no more
2314
2315 the three matched strings are
2316
2317 <something>
2318 <something> <something else>
2319 <something> <something else> <something further>
2320
2321 On success, the yield of the function is a number greater than zero,
2322 which is the number of matched substrings. The substrings themselves
2323 are returned in ovector. Each string uses two elements; the first is
2324 the offset to the start, and the second is the offset to the end. In
2325 fact, all the strings have the same start offset. (Space could have
2326 been saved by giving this only once, but it was decided to retain some
2327 compatibility with the way pcre_exec() returns data, even though the
2328 meaning of the strings is different.)
2329
2330 The strings are returned in reverse order of length; that is, the long-
2331 est matching string is given first. If there were too many matches to
2332 fit into ovector, the yield of the function is zero, and the vector is
2333 filled with the longest matches.
2334
2335 Error returns from pcre_dfa_exec()
2336
2337 The pcre_dfa_exec() function returns a negative number when it fails.
2338 Many of the errors are the same as for pcre_exec(), and these are
2339 described above. There are in addition the following errors that are
2340 specific to pcre_dfa_exec():
2341
2342 PCRE_ERROR_DFA_UITEM (-16)
2343
2344 This return is given if pcre_dfa_exec() encounters an item in the pat-
2345 tern that it does not support, for instance, the use of \C or a back
2346 reference.
2347
2348 PCRE_ERROR_DFA_UCOND (-17)
2349
2350 This return is given if pcre_dfa_exec() encounters a condition item
2351 that uses a back reference for the condition, or a test for recursion
2352 in a specific group. These are not supported.
2353
2354 PCRE_ERROR_DFA_UMLIMIT (-18)
2355
2356 This return is given if pcre_dfa_exec() is called with an extra block
2357 that contains a setting of the match_limit field. This is not supported
2358 (it is meaningless).
2359
2360 PCRE_ERROR_DFA_WSSIZE (-19)
2361
2362 This return is given if pcre_dfa_exec() runs out of space in the
2363 workspace vector.
2364
2365 PCRE_ERROR_DFA_RECURSE (-20)
2366
2367 When a recursive subpattern is processed, the matching function calls
2368 itself recursively, using private vectors for ovector and workspace.
2369 This error is given if the output vector is not large enough. This
2370 should be extremely rare, as a vector of size 1000 is used.
2371
2372
2373 SEE ALSO
2374
2375 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2376 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2377
2378
2379 AUTHOR
2380
2381 Philip Hazel
2382 University Computing Service
2383 Cambridge CB2 3QH, England.
2384
2385
2386 REVISION
2387
2388 Last updated: 16 April 2007
2389 Copyright (c) 1997-2007 University of Cambridge.
2390 ------------------------------------------------------------------------------
2391
2392
2393 PCRECALLOUT(3) PCRECALLOUT(3)
2394
2395
2396 NAME
2397 PCRE - Perl-compatible regular expressions
2398
2399
2400 PCRE CALLOUTS
2401
2402 int (*pcre_callout)(pcre_callout_block *);
2403
2404 PCRE provides a feature called "callout", which is a means of temporar-
2405 ily passing control to the caller of PCRE in the middle of pattern
2406 matching. The caller of PCRE provides an external function by putting
2407 its entry point in the global variable pcre_callout. By default, this
2408 variable contains NULL, which disables all calling out.
2409
2410 Within a regular expression, (?C) indicates the points at which the
2411 external function is to be called. Different callout points can be
2412 identified by putting a number less than 256 after the letter C. The
2413 default value is zero. For example, this pattern has two callout
2414 points:
2415
2416 (?C1)eabc(?C2)def
2417
2418 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2419 called, PCRE automatically inserts callouts, all with number 255,
2420 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2421 used with the pattern
2422
2423 A(\d{2}|--)
2424
2425 it is processed as if it were
2426
2427 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2428
2429 Notice that there is a callout before and after each parenthesis and
2430 alternation bar. Automatic callouts can be used for tracking the
2431 progress of pattern matching. The pcretest command has an option that
2432 sets automatic callouts; when it is used, the output indicates how the
2433 pattern is matched. This is useful information when you are trying to
2434 optimize the performance of a particular pattern.
2435
2436
2437 MISSING CALLOUTS
2438
2439 You should be aware that, because of optimizations in the way PCRE
2440 matches patterns, callouts sometimes do not happen. For example, if the
2441 pattern is
2442
2443 ab(?C4)cd
2444
2445 PCRE knows that any matching string must contain the letter "d". If the
2446 subject string is "abyz", the lack of "d" means that matching doesn't
2447 ever start, and the callout is never reached. However, with "abyd",
2448 though the result is still no match, the callout is obeyed.
2449
2450
2451 THE CALLOUT INTERFACE
2452
2453 During matching, when PCRE reaches a callout point, the external func-
2454 tion defined by pcre_callout is called (if it is set). This applies to
2455 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2456 only argument to the callout function is a pointer to a pcre_callout
2457 block. This structure contains the following fields:
2458
2459 int version;
2460 int callout_number;
2461 int *offset_vector;
2462 const char *subject;
2463 int subject_length;
2464 int start_match;
2465 int current_position;
2466 int capture_top;
2467 int capture_last;
2468 void *callout_data;
2469 int pattern_position;
2470 int next_item_length;
2471
2472 The version field is an integer containing the version number of the
2473 block format. The initial version was 0; the current version is 1. The
2474 version number will change again in future if additional fields are
2475 added, but the intention is never to remove any of the existing fields.
2476
2477 The callout_number field contains the number of the callout, as com-
2478 piled into the pattern (that is, the number after ?C for manual call-
2479 outs, and 255 for automatically generated callouts).
2480
2481 The offset_vector field is a pointer to the vector of offsets that was
2482 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2483 pcre_exec() is used, the contents can be inspected in order to extract
2484 substrings that have been matched so far, in the same way as for
2485 extracting substrings after a match has completed. For pcre_dfa_exec()
2486 this field is not useful.
2487
2488 The subject and subject_length fields contain copies of the values that
2489 were passed to pcre_exec().
2490
2491 The start_match field contains the offset within the subject at which
2492 the current match attempt started. If the pattern is not anchored, the
2493 callout function may be called several times from the same point in the
2494 pattern for different starting points in the subject.
2495
2496 The current_position field contains the offset within the subject of
2497 the current match pointer.
2498
2499 When the pcre_exec() function is used, the capture_top field contains
2500 one more than the number of the highest numbered captured substring so
2501 far. If no substrings have been captured, the value of capture_top is
2502 one. This is always the case when pcre_dfa_exec() is used, because it
2503 does not support captured substrings.
2504
2505 The capture_last field contains the number of the most recently cap-
2506 tured substring. If no substrings have been captured, its value is -1.
2507 This is always the case when pcre_dfa_exec() is used.
2508
2509 The callout_data field contains a value that is passed to pcre_exec()
2510 or pcre_dfa_exec() specifically so that it can be passed back in call-
2511 outs. It is passed in the pcre_callout field of the pcre_extra data
2512 structure. If no such data was passed, the value of callout_data in a
2513 pcre_callout block is NULL. There is a description of the pcre_extra
2514 structure in the pcreapi documentation.
2515
2516 The pattern_position field is present from version 1 of the pcre_call-
2517 out structure. It contains the offset to the next item to be matched in
2518 the pattern string.
2519
2520 The next_item_length field is present from version 1 of the pcre_call-
2521 out structure. It contains the length of the next item to be matched in
2522 the pattern string. When the callout immediately precedes an alterna-
2523 tion bar, a closing parenthesis, or the end of the pattern, the length
2524 is zero. When the callout precedes an opening parenthesis, the length
2525 is that of the entire subpattern.
2526
2527 The pattern_position and next_item_length fields are intended to help
2528 in distinguishing between different automatic callouts, which all have
2529 the same callout number. However, they are set for all callouts.
2530
2531
2532 RETURN VALUES
2533
2534 The external callout function returns an integer to PCRE. If the value
2535 is zero, matching proceeds as normal. If the value is greater than
2536 zero, matching fails at the current point, but the testing of other
2537 matching possibilities goes ahead, just as if a lookahead assertion had
2538 failed. If the value is less than zero, the match is abandoned, and
2539 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2540
2541 Negative values should normally be chosen from the set of
2542 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2543 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2544 reserved for use by callout functions; it will never be used by PCRE
2545 itself.
2546
2547
2548 AUTHOR
2549
2550 Philip Hazel
2551 University Computing Service
2552 Cambridge CB2 3QH, England.
2553
2554
2555 REVISION
2556
2557 Last updated: 06 March 2007
2558 Copyright (c) 1997-2007 University of Cambridge.
2559 ------------------------------------------------------------------------------
2560
2561
2562 PCRECOMPAT(3) PCRECOMPAT(3)
2563
2564
2565 NAME
2566 PCRE - Perl-compatible regular expressions
2567
2568
2569 DIFFERENCES BETWEEN PCRE AND PERL
2570
2571 This document describes the differences in the ways that PCRE and Perl
2572 handle regular expressions. The differences described here are mainly
2573 with respect to Perl 5.8, though PCRE version 7.0 contains some fea-
2574 tures that are expected to be in the forthcoming Perl 5.10.
2575
2576 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2577 of what it does have are given in the section on UTF-8 support in the
2578 main pcre page.
2579
2580 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2581 permits them, but they do not mean what you might think. For example,
2582 (?!a){3} does not assert that the next three characters are not "a". It
2583 just asserts that the next character is not "a" three times.
2584
2585 3. Capturing subpatterns that occur inside negative lookahead asser-
2586 tions are counted, but their entries in the offsets vector are never
2587 set. Perl sets its numerical variables from any such patterns that are
2588 matched before the assertion fails to match something (thereby succeed-
2589 ing), but only if the negative lookahead assertion contains just one
2590 branch.
2591
2592 4. Though binary zero characters are supported in the subject string,
2593 they are not allowed in a pattern string because it is passed as a nor-
2594 mal C string, terminated by zero. The escape sequence \0 can be used in
2595 the pattern to represent a binary zero.
2596
2597 5. The following Perl escape sequences are not supported: \l, \u, \L,
2598 \U, and \N. In fact these are implemented by Perl's general string-han-
2599 dling and are not part of its pattern matching engine. If any of these
2600 are encountered by PCRE, an error is generated.
2601
2602 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2603 is built with Unicode character property support. The properties that
2604 can be tested with \p and \P are limited to the general category prop-
2605 erties such as Lu and Nd, script names such as Greek or Han, and the
2606 derived properties Any and L&.
2607
2608 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2609 ters in between are treated as literals. This is slightly different
2610 from Perl in that $ and @ are also handled as literals inside the
2611 quotes. In Perl, they cause variable interpolation (but of course PCRE
2612 does not have variables). Note the following examples:
2613
2614 Pattern PCRE matches Perl matches
2615
2616 \Qabc$xyz\E abc$xyz abc followed by the
2617 contents of $xyz
2618 \Qabc\$xyz\E abc\$xyz abc\$xyz
2619 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2620
2621 The \Q...\E sequence is recognized both inside and outside character
2622 classes.
2623
2624 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2625 constructions. However, there is support for recursive patterns. This
2626 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2627 "callout" feature allows an external function to be called during pat-
2628 tern matching. See the pcrecallout documentation for details.
2629
2630 9. Subpatterns that are called recursively or as "subroutines" are
2631 always treated as atomic groups in PCRE. This is like Python, but
2632 unlike Perl.
2633
2634 10. There are some differences that are concerned with the settings of
2635 captured strings when part of a pattern is repeated. For example,
2636 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2637 unset, but in PCRE it is set to "b".
2638
2639 11. PCRE provides some extensions to the Perl regular expression facil-
2640 ities. Perl 5.10 will include new features that are not in earlier
2641 versions, some of which (such as named parentheses) have been in PCRE
2642 for some time. This list is with respect to Perl 5.10:
2643
2644 (a) Although lookbehind assertions must match fixed length strings,
2645 each alternative branch of a lookbehind assertion can match a different
2646 length of string. Perl requires them all to have the same length.
2647
2648 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2649 meta-character matches only at the very end of the string.
2650
2651 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2652 cial meaning is faulted. Otherwise, like Perl, the backslash is
2653 ignored. (Perl can be made to issue a warning.)
2654
2655 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2656 fiers is inverted, that is, by default they are not greedy, but if fol-
2657 lowed by a question mark they are.
2658
2659 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2660 tried only at the first matching position in the subject string.
2661
2662 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2663 TURE options for pcre_exec() have no Perl equivalents.
2664
2665 (g) The callout facility is PCRE-specific.
2666
2667 (h) The partial matching facility is PCRE-specific.
2668
2669 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2670 even on different hosts that have the other endianness.
2671
2672 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2673 different way and is not Perl-compatible.
2674
2675
2676 AUTHOR
2677
2678 Philip Hazel
2679 University Computing Service
2680 Cambridge CB2 3QH, England.
2681
2682
2683 REVISION
2684
2685 Last updated: 06 March 2007
2686 Copyright (c) 1997-2007 University of Cambridge.
2687 ------------------------------------------------------------------------------
2688
2689
2690 PCREPATTERN(3) PCREPATTERN(3)
2691
2692
2693 NAME
2694 PCRE - Perl-compatible regular expressions
2695
2696
2697 PCRE REGULAR EXPRESSION DETAILS
2698
2699 The syntax and semantics of the regular expressions supported by PCRE
2700 are described below. Regular expressions are also described in the Perl
2701 documentation and in a number of books, some of which have copious
2702 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2703 by O'Reilly, covers regular expressions in great detail. This descrip-
2704 tion of PCRE's regular expressions is intended as reference material.
2705
2706 The original operation of PCRE was on strings of one-byte characters.
2707 However, there is now also support for UTF-8 character strings. To use
2708 this, you must build PCRE to include UTF-8 support, and then call
2709 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2710 matching is mentioned in several places below. There is also a summary
2711 of UTF-8 features in the section on UTF-8 support in the main pcre
2712 page.
2713
2714 The remainder of this document discusses the patterns that are sup-
2715 ported by PCRE when its main matching function, pcre_exec(), is used.
2716 From release 6.0, PCRE offers a second matching function,
2717 pcre_dfa_exec(), which matches using a different algorithm that is not
2718 Perl-compatible. The advantages and disadvantages of the alternative
2719 function, and how it differs from the normal function, are discussed in
2720 the pcrematching page.
2721
2722
2723 CHARACTERS AND METACHARACTERS
2724
2725 A regular expression is a pattern that is matched against a subject
2726 string from left to right. Most characters stand for themselves in a
2727 pattern, and match the corresponding characters in the subject. As a
2728 trivial example, the pattern
2729
2730 The quick brown fox
2731
2732 matches a portion of a subject string that is identical to itself. When
2733 caseless matching is specified (the PCRE_CASELESS option), letters are
2734 matched independently of case. In UTF-8 mode, PCRE always understands
2735 the concept of case for characters whose values are less than 128, so
2736 caseless matching is always possible. For characters with higher val-
2737 ues, the concept of case is supported if PCRE is compiled with Unicode
2738 property support, but not otherwise. If you want to use caseless
2739 matching for characters 128 and above, you must ensure that PCRE is
2740 compiled with Unicode property support as well as with UTF-8 support.
2741
2742 The power of regular expressions comes from the ability to include
2743 alternatives and repetitions in the pattern. These are encoded in the
2744 pattern by the use of metacharacters, which do not stand for themselves
2745 but instead are interpreted in some special way.
2746
2747 There are two different sets of metacharacters: those that are recog-
2748 nized anywhere in the pattern except within square brackets, and those
2749 that are recognized within square brackets. Outside square brackets,
2750 the metacharacters are as follows:
2751
2752 \ general escape character with several uses
2753 ^ assert start of string (or line, in multiline mode)
2754 $ assert end of string (or line, in multiline mode)
2755 . match any character except newline (by default)
2756 [ start character class definition
2757 | start of alternative branch
2758 ( start subpattern
2759 ) end subpattern
2760 ? extends the meaning of (
2761 also 0 or 1 quantifier
2762 also quantifier minimizer
2763 * 0 or more quantifier
2764 + 1 or more quantifier
2765 also "possessive quantifier"
2766 { start min/max quantifier
2767
2768 Part of a pattern that is in square brackets is called a "character
2769 class". In a character class the only metacharacters are:
2770
2771 \ general escape character
2772 ^ negate the class, but only if the first character
2773 - indicates character range
2774 [ POSIX character class (only if followed by POSIX
2775 syntax)
2776 ] terminates the character class
2777
2778 The following sections describe the use of each of the metacharacters.
2779
2780
2781 BACKSLASH
2782
2783 The backslash character has several uses. Firstly, if it is followed by
2784 a non-alphanumeric character, it takes away any special meaning that
2785 character may have. This use of backslash as an escape character
2786 applies both inside and outside character classes.
2787
2788 For example, if you want to match a * character, you write \* in the
2789 pattern. This escaping action applies whether or not the following
2790 character would otherwise be interpreted as a metacharacter, so it is
2791 always safe to precede a non-alphanumeric with backslash to specify
2792 that it stands for itself. In particular, if you want to match a back-
2793 slash, you write \\.
2794
2795 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2796 the pattern (other than in a character class) and characters between a
2797 # outside a character class and the next newline are ignored. An escap-
2798 ing backslash can be used to include a whitespace or # character as
2799 part of the pattern.
2800
2801 If you want to remove the special meaning from a sequence of charac-
2802 ters, you can do so by putting them between \Q and \E. This is differ-
2803 ent from Perl in that $ and @ are handled as literals in \Q...\E
2804 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2805 tion. Note the following examples:
2806
2807 Pattern PCRE matches Perl matches
2808
2809 \Qabc$xyz\E abc$xyz abc followed by the
2810 contents of $xyz
2811 \Qabc\$xyz\E abc\$xyz abc\$xyz
2812 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2813
2814 The \Q...\E sequence is recognized both inside and outside character
2815 classes.
2816
2817 Non-printing characters
2818
2819 A second use of backslash provides a way of encoding non-printing char-
2820 acters in patterns in a visible manner. There is no restriction on the
2821 appearance of non-printing characters, apart from the binary zero that
2822 terminates a pattern, but when a pattern is being prepared by text
2823 editing, it is usually easier to use one of the following escape
2824 sequences than the binary character it represents:
2825
2826 \a alarm, that is, the BEL character (hex 07)
2827 \cx "control-x", where x is any character
2828 \e escape (hex 1B)
2829 \f formfeed (hex 0C)
2830 \n newline (hex 0A)
2831 \r carriage return (hex 0D)
2832 \t tab (hex 09)
2833 \ddd character with octal code ddd, or backreference
2834 \xhh character with hex code hh
2835 \x{hhh..} character with hex code hhh..
2836
2837 The precise effect of \cx is as follows: if x is a lower case letter,
2838 it is converted to upper case. Then bit 6 of the character (hex 40) is
2839 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2840 becomes hex 7B.
2841
2842 After \x, from zero to two hexadecimal digits are read (letters can be
2843 in upper or lower case). Any number of hexadecimal digits may appear
2844 between \x{ and }, but the value of the character code must be less
2845 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2846 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2847 hexadecimal digits appear between \x{ and }, or if there is no termi-
2848 nating }, this form of escape is not recognized. Instead, the initial
2849 \x will be interpreted as a basic hexadecimal escape, with no following
2850 digits, giving a character whose value is zero.
2851
2852 Characters whose value is less than 256 can be defined by either of the
2853 two syntaxes for \x. There is no difference in the way they are han-
2854 dled. For example, \xdc is exactly the same as \x{dc}.
2855
2856 After \0 up to two further octal digits are read. If there are fewer
2857 than two digits, just those that are present are used. Thus the
2858 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2859 (code value 7). Make sure you supply two digits after the initial zero
2860 if the pattern character that follows is itself an octal digit.
2861
2862 The handling of a backslash followed by a digit other than 0 is compli-
2863 cated. Outside a character class, PCRE reads it and any following dig-
2864 its as a decimal number. If the number is less than 10, or if there
2865 have been at least that many previous capturing left parentheses in the
2866 expression, the entire sequence is taken as a back reference. A
2867 description of how this works is given later, following the discussion
2868 of parenthesized subpatterns.
2869
2870 Inside a character class, or if the decimal number is greater than 9
2871 and there have not been that many capturing subpatterns, PCRE re-reads
2872 up to three octal digits following the backslash, and uses them to gen-
2873 erate a data character. Any subsequent digits stand for themselves. In
2874 non-UTF-8 mode, the value of a character specified in octal must be
2875 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2876 example:
2877
2878 \040 is another way of writing a space
2879 \40 is the same, provided there are fewer than 40
2880 previous capturing subpatterns
2881 \7 is always a back reference
2882 \11 might be a back reference, or another way of
2883 writing a tab
2884 \011 is always a tab
2885 \0113 is a tab followed by the character "3"
2886 \113 might be a back reference, otherwise the
2887 character with octal code 113
2888 \377 might be a back reference, otherwise
2889 the byte consisting entirely of 1 bits
2890 \81 is either a back reference, or a binary zero
2891 followed by the two characters "8" and "1"
2892
2893 Note that octal values of 100 or greater must not be introduced by a
2894 leading zero, because no more than three octal digits are ever read.
2895
2896 All the sequences that define a single character value can be used both
2897 inside and outside character classes. In addition, inside a character
2898 class, the sequence \b is interpreted as the backspace character (hex
2899 08), and the sequences \R and \X are interpreted as the characters "R"
2900 and "X", respectively. Outside a character class, these sequences have
2901 different meanings (see below).
2902
2903 Absolute and relative back references
2904
2905 The sequence \g followed by a positive or negative number, optionally
2906 enclosed in braces, is an absolute or relative back reference. Back
2907 references are discussed later, following the discussion of parenthe-
2908 sized subpatterns.
2909
2910 Generic character types
2911
2912 Another use of backslash is for specifying generic character types. The
2913 following are always recognized:
2914
2915 \d any decimal digit
2916 \D any character that is not a decimal digit
2917 \s any whitespace character
2918 \S any character that is not a whitespace character
2919 \w any "word" character
2920 \W any "non-word" character
2921
2922 Each pair of escape sequences partitions the complete set of characters
2923 into two disjoint sets. Any given character matches one, and only one,
2924 of each pair.
2925
2926 These character type sequences can appear both inside and outside char-
2927 acter classes. They each match one character of the appropriate type.
2928 If the current matching point is at the end of the subject string, all
2929 of them fail, since there is no character to match.
2930
2931 For compatibility with Perl, \s does not match the VT character (code
2932 11). This makes it different from the the POSIX "space" class. The \s
2933 characters are HT (9), LF (10), FF (12), CR (13), and space (32). (If
2934 "use locale;" is included in a Perl script, \s may match the VT charac-
2935 ter. In PCRE, it never does.)
2936
2937 A "word" character is an underscore or any character less than 256 that
2938 is a letter or digit. The definition of letters and digits is con-
2939 trolled by PCRE's low-valued character tables, and may vary if locale-
2940 specific matching is taking place (see "Locale support" in the pcreapi
2941 page). For example, in a French locale such as "fr_FR" in Unix-like
2942 systems, or "french" in Windows, some character codes greater than 128
2943 are used for accented letters, and these are matched by \w.
2944
2945 In UTF-8 mode, characters with values greater than 128 never match \d,
2946 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2947 code character property support is available. The use of locales with
2948 Unicode is discouraged.
2949
2950 Newline sequences
2951
2952 Outside a character class, the escape sequence \R matches any Unicode
2953 newline sequence. This is an extension to Perl. In non-UTF-8 mode \R is
2954 equivalent to the following:
2955
2956 (?>\r\n|\n|\x0b|\f|\r|\x85)
2957
2958 This is an example of an "atomic group", details of which are given
2959 below. This particular group matches either the two-character sequence
2960 CR followed by LF, or one of the single characters LF (linefeed,
2961 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
2962 return, U+000D), or NEL (next line, U+0085). The two-character sequence
2963 is treated as a single unit that cannot be split.
2964
2965 In UTF-8 mode, two additional characters whose codepoints are greater
2966 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
2967 rator, U+2029). Unicode character property support is not needed for
2968 these characters to be recognized.
2969
2970 Inside a character class, \R matches the letter "R".
2971
2972 Unicode character properties
2973
2974 When PCRE is built with Unicode character property support, three addi-
2975 tional escape sequences to match character properties are available
2976 when UTF-8 mode is selected. They are:
2977
2978 \p{xx} a character with the xx property
2979 \P{xx} a character without the xx property
2980 \X an extended Unicode sequence
2981
2982 The property names represented by xx above are limited to the Unicode
2983 script names, the general category properties, and "Any", which matches
2984 any character (including newline). Other properties such as "InMusical-
2985 Symbols" are not currently supported by PCRE. Note that \P{Any} does
2986 not match any characters, so always causes a match failure.
2987
2988 Sets of Unicode characters are defined as belonging to certain scripts.
2989 A character from one of these sets can be matched using a script name.
2990 For example:
2991
2992 \p{Greek}
2993 \P{Han}
2994
2995 Those that are not part of an identified script are lumped together as
2996 "Common". The current list of scripts is:
2997
2998 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
2999 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3000 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3001 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3002 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3003 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3004 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3005 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3006 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3007
3008 Each character has exactly one general category property, specified by
3009 a two-letter abbreviation. For compatibility with Perl, negation can be
3010 specified by including a circumflex between the opening brace and the
3011 property name. For example, \p{^Lu} is the same as \P{Lu}.
3012
3013 If only one letter is specified with \p or \P, it includes all the gen-
3014 eral category properties that start with that letter. In this case, in
3015 the absence of negation, the curly brackets in the escape sequence are
3016 optional; these two examples have the same effect:
3017
3018 \p{L}
3019 \pL
3020
3021 The following general category property codes are supported:
3022
3023 C Other
3024 Cc Control
3025 Cf Format
3026 Cn Unassigned
3027 Co Private use
3028 Cs Surrogate
3029
3030 L Letter
3031 Ll Lower case letter
3032 Lm Modifier letter
3033 Lo Other letter
3034 Lt Title case letter
3035 Lu Upper case letter
3036
3037 M Mark
3038 Mc Spacing mark
3039 Me Enclosing mark
3040 Mn Non-spacing mark
3041
3042 N Number
3043 Nd Decimal number
3044 Nl Letter number
3045 No Other number
3046
3047 P Punctuation
3048 Pc Connector punctuation
3049 Pd Dash punctuation
3050 Pe Close punctuation
3051 Pf Final punctuation
3052 Pi Initial punctuation
3053 Po Other punctuation
3054 Ps Open punctuation
3055
3056 S Symbol
3057 Sc Currency symbol
3058 Sk Modifier symbol
3059 Sm Mathematical symbol
3060 So Other symbol
3061
3062 Z Separator
3063 Zl Line separator
3064 Zp Paragraph separator
3065 Zs Space separator
3066
3067 The special property L& is also supported: it matches a character that
3068 has the Lu, Ll, or Lt property, in other words, a letter that is not
3069 classified as a modifier or "other".
3070
3071 The long synonyms for these properties that Perl supports (such as
3072 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3073 any of these properties with "Is".
3074
3075 No character that is in the Unicode table has the Cn (unassigned) prop-
3076 erty. Instead, this property is assumed for any code point that is not
3077 in the Unicode table.
3078
3079 Specifying caseless matching does not affect these escape sequences.
3080 For example, \p{Lu} always matches only upper case letters.
3081
3082 The \X escape matches any number of Unicode characters that form an
3083 extended Unicode sequence. \X is equivalent to
3084
3085 (?>\PM\pM*)
3086
3087 That is, it matches a character without the "mark" property, followed
3088 by zero or more characters with the "mark" property, and treats the
3089 sequence as an atomic group (see below). Characters with the "mark"
3090 property are typically accents that affect the preceding character.
3091
3092 Matching characters by Unicode property is not fast, because PCRE has
3093 to search a structure that contains data for over fifteen thousand
3094 characters. That is why the traditional escape sequences such as \d and
3095 \w do not use Unicode properties in PCRE.
3096
3097 Simple assertions
3098
3099 The final use of backslash is for certain simple assertions. An asser-
3100 tion specifies a condition that has to be met at a particular point in
3101 a match, without consuming any characters from the subject string. The
3102 use of subpatterns for more complicated assertions is described below.
3103 The backslashed assertions are:
3104
3105 \b matches at a word boundary
3106 \B matches when not at a word boundary
3107 \A matches at the start of the subject
3108 \Z matches at the end of the subject
3109 also matches before a newline at the end of the subject
3110 \z matches only at the end of the subject
3111 \G matches at the first matching position in the subject
3112
3113 These assertions may not appear in character classes (but note that \b
3114 has a different meaning, namely the backspace character, inside a char-
3115 acter class).
3116
3117 A word boundary is a position in the subject string where the current
3118 character and the previous character do not both match \w or \W (i.e.
3119 one matches \w and the other matches \W), or the start or end of the
3120 string if the first or last character matches \w, respectively.
3121
3122 The \A, \Z, and \z assertions differ from the traditional circumflex
3123 and dollar (described in the next section) in that they only ever match
3124 at the very start and end of the subject string, whatever options are
3125 set. Thus, they are independent of multiline mode. These three asser-
3126 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3127 affect only the behaviour of the circumflex and dollar metacharacters.
3128 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3129 cating that matching is to start at a point other than the beginning of
3130 the subject, \A can never match. The difference between \Z and \z is
3131 that \Z matches before a newline at the end of the string as well as at
3132 the very end, whereas \z matches only at the end.
3133
3134 The \G assertion is true only when the current matching position is at
3135 the start point of the match, as specified by the startoffset argument
3136 of pcre_exec(). It differs from \A when the value of startoffset is
3137 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3138 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3139 mentation where \G can be useful.
3140
3141 Note, however, that PCRE's interpretation of \G, as the start of the
3142 current match, is subtly different from Perl's, which defines it as the
3143 end of the previous match. In Perl, these can be different when the
3144 previously matched string was empty. Because PCRE does just one match
3145 at a time, it cannot reproduce this behaviour.
3146
3147 If all the alternatives of a pattern begin with \G, the expression is
3148 anchored to the starting match position, and the "anchored" flag is set
3149 in the compiled regular expression.
3150
3151
3152 CIRCUMFLEX AND DOLLAR
3153
3154 Outside a character class, in the default matching mode, the circumflex
3155 character is an assertion that is true only if the current matching
3156 point is at the start of the subject string. If the startoffset argu-
3157 ment of pcre_exec() is non-zero, circumflex can never match if the
3158 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3159 has an entirely different meaning (see below).
3160
3161 Circumflex need not be the first character of the pattern if a number
3162 of alternatives are involved, but it should be the first thing in each
3163 alternative in which it appears if the pattern is ever to match that
3164 branch. If all possible alternatives start with a circumflex, that is,
3165 if the pattern is constrained to match only at the start of the sub-
3166 ject, it is said to be an "anchored" pattern. (There are also other
3167 constructs that can cause a pattern to be anchored.)
3168
3169 A dollar character is an assertion that is true only if the current
3170 matching point is at the end of the subject string, or immediately
3171 before a newline at the end of the string (by default). Dollar need not
3172 be the last character of the pattern if a number of alternatives are
3173 involved, but it should be the last item in any branch in which it
3174 appears. Dollar has no special meaning in a character class.
3175
3176 The meaning of dollar can be changed so that it matches only at the
3177 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3178 compile time. This does not affect the \Z assertion.
3179
3180 The meanings of the circumflex and dollar characters are changed if the
3181 PCRE_MULTILINE option is set. When this is the case, a circumflex
3182 matches immediately after internal newlines as well as at the start of
3183 the subject string. It does not match after a newline that ends the
3184 string. A dollar matches before any newlines in the string, as well as
3185 at the very end, when PCRE_MULTILINE is set. When newline is specified
3186 as the two-character sequence CRLF, isolated CR and LF characters do
3187 not indicate newlines.
3188
3189 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3190 (where \n represents a newline) in multiline mode, but not otherwise.
3191 Consequently, patterns that are anchored in single line mode because
3192 all branches start with ^ are not anchored in multiline mode, and a
3193 match for circumflex is possible when the startoffset argument of
3194 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3195 PCRE_MULTILINE is set.
3196
3197 Note that the sequences \A, \Z, and \z can be used to match the start
3198 and end of the subject in both modes, and if all branches of a pattern
3199 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3200 set.
3201
3202
3203 FULL STOP (PERIOD, DOT)
3204
3205 Outside a character class, a dot in the pattern matches any one charac-
3206 ter in the subject string except (by default) a character that signi-
3207 fies the end of a line. In UTF-8 mode, the matched character may be
3208 more than one byte long.
3209
3210 When a line ending is defined as a single character, dot never matches
3211 that character; when the two-character sequence CRLF is used, dot does
3212 not match CR if it is immediately followed by LF, but otherwise it
3213 matches all characters (including isolated CRs and LFs). When any Uni-
3214 code line endings are being recognized, dot does not match CR or LF or
3215 any of the other line ending characters.
3216
3217 The behaviour of dot with regard to newlines can be changed. If the
3218 PCRE_DOTALL option is set, a dot matches any one character, without
3219 exception. If the two-character sequence CRLF is present in the subject
3220 string, it takes two dots to match it.
3221
3222 The handling of dot is entirely independent of the handling of circum-
3223 flex and dollar, the only relationship being that they both involve
3224 newlines. Dot has no special meaning in a character class.
3225
3226
3227 MATCHING A SINGLE BYTE
3228
3229 Outside a character class, the escape sequence \C matches any one byte,
3230 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3231 line-ending characters. The feature is provided in Perl in order to
3232 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3233 acters into individual bytes, what remains in the string may be a mal-
3234 formed UTF-8 string. For this reason, the \C escape sequence is best
3235 avoided.
3236
3237 PCRE does not allow \C to appear in lookbehind assertions (described
3238 below), because in UTF-8 mode this would make it impossible to calcu-
3239 late the length of the lookbehind.
3240
3241
3242 SQUARE BRACKETS AND CHARACTER CLASSES
3243
3244 An opening square bracket introduces a character class, terminated by a
3245 closing square bracket. A closing square bracket on its own is not spe-
3246 cial. If a closing square bracket is required as a member of the class,
3247 it should be the first data character in the class (after an initial
3248 circumflex, if present) or escaped with a backslash.
3249
3250 A character class matches a single character in the subject. In UTF-8
3251 mode, the character may occupy more than one byte. A matched character
3252 must be in the set of characters defined by the class, unless the first
3253 character in the class definition is a circumflex, in which case the
3254 subject character must not be in the set defined by the class. If a
3255 circumflex is actually required as a member of the class, ensure it is
3256 not the first character, or escape it with a backslash.
3257
3258 For example, the character class [aeiou] matches any lower case vowel,
3259 while [^aeiou] matches any character that is not a lower case vowel.
3260 Note that a circumflex is just a convenient notation for specifying the
3261 characters that are in the class by enumerating those that are not. A
3262 class that starts with a circumflex is not an assertion: it still con-
3263 sumes a character from the subject string, and therefore it fails if
3264 the current pointer is at the end of the string.
3265
3266 In UTF-8 mode, characters with values greater than 255 can be included
3267 in a class as a literal string of bytes, or by using the \x{ escaping
3268 mechanism.
3269
3270 When caseless matching is set, any letters in a class represent both
3271 their upper case and lower case versions, so for example, a caseless
3272 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3273 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3274 understands the concept of case for characters whose values are less
3275 than 128, so caseless matching is always possible. For characters with
3276 higher values, the concept of case is supported if PCRE is compiled
3277 with Unicode property support, but not otherwise. If you want to use
3278 caseless matching for characters 128 and above, you must ensure that
3279 PCRE is compiled with Unicode property support as well as with UTF-8
3280 support.
3281
3282 Characters that might indicate line breaks are never treated in any
3283 special way when matching character classes, whatever line-ending
3284 sequence is in use, and whatever setting of the PCRE_DOTALL and
3285 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3286 of these characters.
3287
3288 The minus (hyphen) character can be used to specify a range of charac-
3289 ters in a character class. For example, [d-m] matches any letter
3290 between d and m, inclusive. If a minus character is required in a
3291 class, it must be escaped with a backslash or appear in a position
3292 where it cannot be interpreted as indicating a range, typically as the
3293 first or last character in the class.
3294
3295 It is not possible to have the literal character "]" as the end charac-
3296 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3297 two characters ("W" and "-") followed by a literal string "46]", so it
3298 would match "W46]" or "-46]". However, if the "]" is escaped with a
3299 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3300 preted as a class containing a range followed by two other characters.
3301 The octal or hexadecimal representation of "]" can also be used to end
3302 a range.
3303
3304 Ranges operate in the collating sequence of character values. They can
3305 also be used for characters specified numerically, for example
3306 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3307 are greater than 255, for example [\x{100}-\x{2ff}].
3308
3309 If a range that includes letters is used when caseless matching is set,
3310 it matches the letters in either case. For example, [W-c] is equivalent
3311 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3312 character tables for a French locale are in use, [\xc8-\xcb] matches
3313 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3314 concept of case for characters with values greater than 128 only when
3315 it is compiled with Unicode property support.
3316
3317 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3318 in a character class, and add the characters that they match to the
3319 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3320 flex can conveniently be used with the upper case character types to
3321 specify a more restricted set of characters than the matching lower
3322 case type. For example, the class [^\W_] matches any letter or digit,
3323 but not underscore.
3324
3325 The only metacharacters that are recognized in character classes are
3326 backslash, hyphen (only where it can be interpreted as specifying a
3327 range), circumflex (only at the start), opening square bracket (only
3328 when it can be interpreted as introducing a POSIX class name - see the
3329 next section), and the terminating closing square bracket. However,
3330 escaping other non-alphanumeric characters does no harm.
3331
3332
3333 POSIX CHARACTER CLASSES
3334
3335 Perl supports the POSIX notation for character classes. This uses names
3336 enclosed by [: and :] within the enclosing square brackets. PCRE also
3337 supports this notation. For example,
3338
3339 [01[:alpha:]%]
3340
3341 matches "0", "1", any alphabetic character, or "%". The supported class
3342 names are
3343
3344 alnum letters and digits
3345 alpha letters
3346 ascii character codes 0 - 127
3347 blank space or tab only
3348 cntrl control characters
3349 digit decimal digits (same as \d)
3350 graph printing characters, excluding space
3351 lower lower case letters
3352 print printing characters, including space
3353 punct printing characters, excluding letters and digits
3354 space white space (not quite the same as \s)
3355 upper upper case letters
3356 word "word" characters (same as \w)
3357 xdigit hexadecimal digits
3358
3359 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3360 and space (32). Notice that this list includes the VT character (code
3361 11). This makes "space" different to \s, which does not include VT (for
3362 Perl compatibility).
3363
3364 The name "word" is a Perl extension, and "blank" is a GNU extension
3365 from Perl 5.8. Another Perl extension is negation, which is indicated
3366 by a ^ character after the colon. For example,
3367
3368 [12[:^digit:]]
3369
3370 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3371 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3372 these are not supported, and an error is given if they are encountered.
3373
3374 In UTF-8 mode, characters with values greater than 128 do not match any
3375 of the POSIX character classes.
3376
3377
3378 VERTICAL BAR
3379
3380 Vertical bar characters are used to separate alternative patterns. For
3381 example, the pattern
3382
3383 gilbert|sullivan
3384
3385 matches either "gilbert" or "sullivan". Any number of alternatives may
3386 appear, and an empty alternative is permitted (matching the empty
3387 string). The matching process tries each alternative in turn, from left
3388 to right, and the first one that succeeds is used. If the alternatives
3389 are within a subpattern (defined below), "succeeds" means matching the
3390 rest of the main pattern as well as the alternative in the subpattern.
3391
3392
3393 INTERNAL OPTION SETTING
3394
3395 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3396 PCRE_EXTENDED options can be changed from within the pattern by a
3397 sequence of Perl option letters enclosed between "(?" and ")". The
3398 option letters are
3399
3400 i for PCRE_CASELESS
3401 m for PCRE_MULTILINE
3402 s for PCRE_DOTALL
3403 x for PCRE_EXTENDED
3404
3405 For example, (?im) sets caseless, multiline matching. It is also possi-
3406 ble to unset these options by preceding the letter with a hyphen, and a
3407 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3408 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3409 is also permitted. If a letter appears both before and after the
3410 hyphen, the option is unset.
3411
3412 When an option change occurs at top level (that is, not inside subpat-
3413 tern parentheses), the change applies to the remainder of the pattern
3414 that follows. If the change is placed right at the start of a pattern,
3415 PCRE extracts it into the global options (and it will therefore show up
3416 in data extracted by the pcre_fullinfo() function).
3417
3418 An option change within a subpattern (see below for a description of
3419 subpatterns) affects only that part of the current pattern that follows
3420 it, so
3421
3422 (a(?i)b)c
3423
3424 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3425 used). By this means, options can be made to have different settings
3426 in different parts of the pattern. Any changes made in one alternative
3427 do carry on into subsequent branches within the same subpattern. For
3428 example,
3429
3430 (a(?i)b|c)
3431
3432 matches "ab", "aB", "c", and "C", even though when matching "C" the
3433 first branch is abandoned before the option setting. This is because
3434 the effects of option settings happen at compile time. There would be
3435 some very weird behaviour otherwise.
3436
3437 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3438 can be changed in the same way as the Perl-compatible options by using
3439 the characters J, U and X respectively.
3440
3441
3442 SUBPATTERNS
3443
3444 Subpatterns are delimited by parentheses (round brackets), which can be
3445 nested. Turning part of a pattern into a subpattern does two things:
3446
3447 1. It localizes a set of alternatives. For example, the pattern
3448
3449 cat(aract|erpillar|)
3450
3451 matches one of the words "cat", "cataract", or "caterpillar". Without
3452 the parentheses, it would match "cataract", "erpillar" or an empty
3453 string.
3454
3455 2. It sets up the subpattern as a capturing subpattern. This means
3456 that, when the whole pattern matches, that portion of the subject
3457 string that matched the subpattern is passed back to the caller via the
3458 ovector argument of pcre_exec(). Opening parentheses are counted from
3459 left to right (starting from 1) to obtain numbers for the capturing
3460 subpatterns.
3461
3462 For example, if the string "the red king" is matched against the pat-
3463 tern
3464
3465 the ((red|white) (king|queen))
3466
3467 the captured substrings are "red king", "red", and "king", and are num-
3468 bered 1, 2, and 3, respectively.
3469
3470 The fact that plain parentheses fulfil two functions is not always
3471 helpful. There are often times when a grouping subpattern is required
3472 without a capturing requirement. If an opening parenthesis is followed
3473 by a question mark and a colon, the subpattern does not do any captur-
3474 ing, and is not counted when computing the number of any subsequent
3475 capturing subpatterns. For example, if the string "the white queen" is
3476 matched against the pattern
3477
3478 the ((?:red|white) (king|queen))
3479
3480 the captured substrings are "white queen" and "queen", and are numbered
3481 1 and 2. The maximum number of capturing subpatterns is 65535.
3482
3483 As a convenient shorthand, if any option settings are required at the
3484 start of a non-capturing subpattern, the option letters may appear
3485 between the "?" and the ":". Thus the two patterns
3486
3487 (?i:saturday|sunday)
3488 (?:(?i)saturday|sunday)
3489
3490 match exactly the same set of strings. Because alternative branches are
3491 tried from left to right, and options are not reset until the end of
3492 the subpattern is reached, an option setting in one branch does affect
3493 subsequent branches, so the above patterns match "SUNDAY" as well as
3494 "Saturday".
3495
3496
3497 NAMED SUBPATTERNS
3498
3499 Identifying capturing parentheses by number is simple, but it can be
3500 very hard to keep track of the numbers in complicated regular expres-
3501 sions. Furthermore, if an expression is modified, the numbers may
3502 change. To help with this difficulty, PCRE supports the naming of sub-
3503 patterns. This feature was not added to Perl until release 5.10. Python
3504 had the feature earlier, and PCRE introduced it at release 4.0, using
3505 the Python syntax. PCRE now supports both the Perl and the Python syn-
3506 tax.
3507
3508 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3509 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3510 to capturing parentheses from other parts of the pattern, such as back-
3511 references, recursion, and conditions, can be made by name as well as
3512 by number.
3513
3514 Names consist of up to 32 alphanumeric characters and underscores.
3515 Named capturing parentheses are still allocated numbers as well as
3516 names, exactly as if the names were not present. The PCRE API provides
3517 function calls for extracting the name-to-number translation table from
3518 a compiled pattern. There is also a convenience function for extracting
3519 a captured substring by name.
3520
3521 By default, a name must be unique within a pattern, but it is possible
3522 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3523 time. This can be useful for patterns where only one instance of the
3524 named parentheses can match. Suppose you want to match the name of a
3525 weekday, either as a 3-letter abbreviation or as the full name, and in
3526 both cases you want to extract the abbreviation. This pattern (ignoring
3527 the line breaks) does the job:
3528
3529 (?<DN>Mon|Fri|Sun)(?:day)?|
3530 (?<DN>Tue)(?:sday)?|
3531 (?<DN>Wed)(?:nesday)?|
3532 (?<DN>Thu)(?:rsday)?|
3533 (?<DN>Sat)(?:urday)?
3534
3535 There are five capturing substrings, but only one is ever set after a
3536 match. The convenience function for extracting the data by name
3537 returns the substring for the first (and in this example, the only)
3538 subpattern of that name that matched. This saves searching to find
3539 which numbered subpattern it was. If you make a reference to a non-
3540 unique named subpattern from elsewhere in the pattern, the one that
3541 corresponds to the lowest number is used. For further details of the
3542 interfaces for handling named subpatterns, see the pcreapi documenta-
3543 tion.
3544
3545
3546 REPETITION
3547
3548 Repetition is specified by quantifiers, which can follow any of the
3549 following items:
3550
3551 a literal data character
3552 the dot metacharacter
3553 the \C escape sequence
3554 the \X escape sequence (in UTF-8 mode with Unicode properties)
3555 the \R escape sequence
3556 an escape such as \d that matches a single character
3557 a character class
3558 a back reference (see next section)
3559 a parenthesized subpattern (unless it is an assertion)
3560
3561 The general repetition quantifier specifies a minimum and maximum num-
3562 ber of permitted matches, by giving the two numbers in curly brackets
3563 (braces), separated by a comma. The numbers must be less than 65536,
3564 and the first must be less than or equal to the second. For example:
3565
3566 z{2,4}
3567
3568 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3569 special character. If the second number is omitted, but the comma is
3570 present, there is no upper limit; if the second number and the comma
3571 are both omitted, the quantifier specifies an exact number of required
3572 matches. Thus
3573
3574 [aeiou]{3,}
3575
3576 matches at least 3 successive vowels, but may match many more, while
3577
3578 \d{8}
3579
3580 matches exactly 8 digits. An opening curly bracket that appears in a
3581 position where a quantifier is not allowed, or one that does not match
3582 the syntax of a quantifier, is taken as a literal character. For exam-
3583 ple, {,6} is not a quantifier, but a literal string of four characters.
3584
3585 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3586 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3587 acters, each of which is represented by a two-byte sequence. Similarly,
3588 when Unicode property support is available, \X{3} matches three Unicode
3589 extended sequences, each of which may be several bytes long (and they
3590 may be of different lengths).
3591
3592 The quantifier {0} is permitted, causing the expression to behave as if
3593 the previous item and the quantifier were not present.
3594
3595 For convenience, the three most common quantifiers have single-charac-
3596 ter abbreviations:
3597
3598 * is equivalent to {0,}
3599 + is equivalent to {1,}
3600 ? is equivalent to {0,1}
3601
3602 It is possible to construct infinite loops by following a subpattern
3603 that can match no characters with a quantifier that has no upper limit,
3604 for example:
3605
3606 (a?)*
3607
3608 Earlier versions of Perl and PCRE used to give an error at compile time
3609 for such patterns. However, because there are cases where this can be
3610 useful, such patterns are now accepted, but if any repetition of the
3611 subpattern does in fact match no characters, the loop is forcibly bro-
3612 ken.
3613
3614 By default, the quantifiers are "greedy", that is, they match as much
3615 as possible (up to the maximum number of permitted times), without
3616 causing the rest of the pattern to fail. The classic example of where
3617 this gives problems is in trying to match comments in C programs. These
3618 appear between /* and */ and within the comment, individual * and /
3619 characters may appear. An attempt to match C comments by applying the
3620 pattern
3621
3622 /\*.*\*/
3623
3624 to the string
3625
3626 /* first comment */ not comment /* second comment */
3627
3628 fails, because it matches the entire string owing to the greediness of
3629 the .* item.
3630
3631 However, if a quantifier is followed by a question mark, it ceases to
3632 be greedy, and instead matches the minimum number of times possible, so
3633 the pattern
3634
3635 /\*.*?\*/
3636
3637 does the right thing with the C comments. The meaning of the various
3638 quantifiers is not otherwise changed, just the preferred number of
3639 matches. Do not confuse this use of question mark with its use as a
3640 quantifier in its own right. Because it has two uses, it can sometimes
3641 appear doubled, as in
3642
3643 \d??\d
3644
3645 which matches one digit by preference, but can match two if that is the
3646 only way the rest of the pattern matches.
3647
3648 If the PCRE_UNGREEDY option is set (an option that is not available in
3649 Perl), the quantifiers are not greedy by default, but individual ones
3650 can be made greedy by following them with a question mark. In other
3651 words, it inverts the default behaviour.
3652
3653 When a parenthesized subpattern is quantified with a minimum repeat
3654 count that is greater than 1 or with a limited maximum, more memory is
3655 required for the compiled pattern, in proportion to the size of the
3656 minimum or maximum.
3657
3658 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3659 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3660 the pattern is implicitly anchored, because whatever follows will be
3661 tried against every character position in the subject string, so there
3662 is no point in retrying the overall match at any position after the
3663 first. PCRE normally treats such a pattern as though it were preceded
3664 by \A.
3665
3666 In cases where it is known that the subject string contains no new-
3667 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3668 mization, or alternatively using ^ to indicate anchoring explicitly.
3669
3670 However, there is one situation where the optimization cannot be used.
3671 When .* is inside capturing parentheses that are the subject of a
3672 backreference elsewhere in the pattern, a match at the start may fail
3673 where a later one succeeds. Consider, for example:
3674
3675 (.*)abc\1
3676
3677 If the subject is "xyz123abc123" the match point is the fourth charac-
3678 ter. For this reason, such a pattern is not implicitly anchored.
3679
3680 When a capturing subpattern is repeated, the value captured is the sub-
3681 string that matched the final iteration. For example, after
3682
3683 (tweedle[dume]{3}\s*)+
3684
3685 has matched "tweedledum tweedledee" the value of the captured substring
3686 is "tweedledee". However, if there are nested capturing subpatterns,
3687 the corresponding captured values may have been set in previous itera-
3688 tions. For example, after
3689
3690 /(a|(b))+/
3691
3692 matches "aba" the value of the second captured substring is "b".
3693
3694
3695 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3696
3697 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3698 repetition, failure of what follows normally causes the repeated item
3699 to be re-evaluated to see if a different number of repeats allows the
3700 rest of the pattern to match. Sometimes it is useful to prevent this,
3701 either to change the nature of the match, or to cause it fail earlier
3702 than it otherwise might, when the author of the pattern knows there is
3703 no point in carrying on.
3704
3705 Consider, for example, the pattern \d+foo when applied to the subject
3706 line
3707
3708 123456bar
3709
3710 After matching all 6 digits and then failing to match "foo", the normal
3711 action of the matcher is to try again with only 5 digits matching the
3712 \d+ item, and then with 4, and so on, before ultimately failing.
3713 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3714 the means for specifying that once a subpattern has matched, it is not
3715 to be re-evaluated in this way.
3716
3717 If we use atomic grouping for the previous example, the matcher gives
3718 up immediately on failing to match "foo" the first time. The notation
3719 is a kind of special parenthesis, starting with (?> as in this example:
3720
3721 (?>\d+)foo
3722
3723 This kind of parenthesis "locks up" the part of the pattern it con-
3724 tains once it has matched, and a failure further into the pattern is
3725 prevented from backtracking into it. Backtracking past it to previous
3726 items, however, works as normal.
3727
3728 An alternative description is that a subpattern of this type matches
3729 the string of characters that an identical standalone pattern would
3730 match, if anchored at the current point in the subject string.
3731
3732 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3733 such as the above example can be thought of as a maximizing repeat that
3734 must swallow everything it can. So, while both \d+ and \d+? are pre-
3735 pared to adjust the number of digits they match in order to make the
3736 rest of the pattern match, (?>\d+) can only match an entire sequence of
3737 digits.
3738
3739 Atomic groups in general can of course contain arbitrarily complicated
3740 subpatterns, and can be nested. However, when the subpattern for an
3741 atomic group is just a single repeated item, as in the example above, a
3742 simpler notation, called a "possessive quantifier" can be used. This
3743 consists of an additional + character following a quantifier. Using
3744 this notation, the previous example can be rewritten as
3745
3746 \d++foo
3747
3748 Possessive quantifiers are always greedy; the setting of the
3749 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3750 simpler forms of atomic group. However, there is no difference in the
3751 meaning of a possessive quantifier and the equivalent atomic group,
3752 though there may be a performance difference; possessive quantifiers
3753 should be slightly faster.
3754
3755 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3756 tax. Jeffrey Friedl originated the idea (and the name) in the first
3757 edition of his book. Mike McCloskey liked it, so implemented it when he
3758 built Sun's Java package, and PCRE copied it from there. It ultimately
3759 found its way into Perl at release 5.10.
3760
3761 PCRE has an optimization that automatically "possessifies" certain sim-
3762 ple pattern constructs. For example, the sequence A+B is treated as
3763 A++B because there is no point in backtracking into a sequence of A's
3764 when B must follow.
3765
3766 When a pattern contains an unlimited repeat inside a subpattern that
3767 can itself be repeated an unlimited number of times, the use of an
3768 atomic group is the only way to avoid some failing matches taking a
3769 very long time indeed. The pattern
3770
3771 (\D+|<\d+>)*[!?]
3772
3773 matches an unlimited number of substrings that either consist of non-
3774 digits, or digits enclosed in <>, followed by either ! or ?. When it
3775 matches, it runs quickly. However, if it is applied to
3776
3777 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3778
3779 it takes a long time before reporting failure. This is because the
3780 string can be divided between the internal \D+ repeat and the external
3781 * repeat in a large number of ways, and all have to be tried. (The
3782 example uses [!?] rather than a single character at the end, because
3783 both PCRE and Perl have an optimization that allows for fast failure
3784 when a single character is used. They remember the last single charac-
3785 ter that is required for a match, and fail early if it is not present
3786 in the string.) If the pattern is changed so that it uses an atomic
3787 group, like this:
3788
3789 ((?>\D+)|<\d+>)*[!?]
3790
3791 sequences of non-digits cannot be broken, and failure happens quickly.
3792
3793
3794 BACK REFERENCES
3795
3796 Outside a character class, a backslash followed by a digit greater than
3797 0 (and possibly further digits) is a back reference to a capturing sub-
3798 pattern earlier (that is, to its left) in the pattern, provided there
3799 have been that many previous capturing left parentheses.
3800
3801 However, if the decimal number following the backslash is less than 10,
3802 it is always taken as a back reference, and causes an error only if
3803 there are not that many capturing left parentheses in the entire pat-
3804 tern. In other words, the parentheses that are referenced need not be
3805 to the left of the reference for numbers less than 10. A "forward back
3806 reference" of this type can make sense when a repetition is involved
3807 and the subpattern to the right has participated in an earlier itera-
3808 tion.
3809
3810 It is not possible to have a numerical "forward back reference" to a
3811 subpattern whose number is 10 or more using this syntax because a
3812 sequence such as \50 is interpreted as a character defined in octal.
3813 See the subsection entitled "Non-printing characters" above for further
3814 details of the handling of digits following a backslash. There is no
3815 such problem when named parentheses are used. A back reference to any
3816 subpattern is possible using named parentheses (see below).
3817
3818 Another way of avoiding the ambiguity inherent in the use of digits
3819 following a backslash is to use the \g escape sequence, which is a fea-
3820 ture introduced in Perl 5.10. This escape must be followed by a posi-
3821 tive or a negative number, optionally enclosed in braces. These exam-
3822 ples are all identical:
3823
3824 (ring), \1
3825 (ring), \g1
3826 (ring), \g{1}
3827
3828 A positive number specifies an absolute reference without the ambiguity
3829 that is present in the older syntax. It is also useful when literal
3830 digits follow the reference. A negative number is a relative reference.
3831 Consider this example:
3832
3833 (abc(def)ghi)\g{-1}
3834
3835 The sequence \g{-1} is a reference to the most recently started captur-
3836 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
3837 \g{-2} would be equivalent to \1. The use of relative references can be
3838 helpful in long patterns, and also in patterns that are created by
3839 joining together fragments that contain references within themselves.
3840
3841 A back reference matches whatever actually matched the capturing sub-
3842 pattern in the current subject string, rather than anything matching
3843 the subpattern itself (see "Subpatterns as subroutines" below for a way
3844 of doing that). So the pattern
3845
3846 (sens|respons)e and \1ibility
3847
3848 matches "sense and sensibility" and "response and responsibility", but
3849 not "sense and responsibility". If caseful matching is in force at the
3850 time of the back reference, the case of letters is relevant. For exam-
3851 ple,
3852
3853 ((?i)rah)\s+\1
3854
3855 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3856 original capturing subpattern is matched caselessly.
3857
3858 Back references to named subpatterns use the Perl syntax \k<name> or
3859 \k'name' or the Python syntax (?P=name). We could rewrite the above
3860 example in either of the following ways:
3861
3862 (?<p1>(?i)rah)\s+\k<p1>
3863 (?P<p1>(?i)rah)\s+(?P=p1)
3864
3865 A subpattern that is referenced by name may appear in the pattern
3866 before or after the reference.
3867
3868 There may be more than one back reference to the same subpattern. If a
3869 subpattern has not actually been used in a particular match, any back
3870 references to it always fail. For example, the pattern
3871
3872 (a|(bc))\2
3873
3874 always fails if it starts to match "a" rather than "bc". Because there
3875 may be many capturing parentheses in a pattern, all digits following
3876 the backslash are taken as part of a potential back reference number.
3877 If the pattern continues with a digit character, some delimiter must be
3878 used to terminate the back reference. If the PCRE_EXTENDED option is
3879 set, this can be whitespace. Otherwise an empty comment (see "Com-
3880 ments" below) can be used.
3881
3882 A back reference that occurs inside the parentheses to which it refers
3883 fails when the subpattern is first used, so, for example, (a\1) never
3884 matches. However, such references can be useful inside repeated sub-
3885 patterns. For example, the pattern
3886
3887 (a|b\1)+
3888
3889 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3890 ation of the subpattern, the back reference matches the character
3891 string corresponding to the previous iteration. In order for this to
3892 work, the pattern must be such that the first iteration does not need
3893 to match the back reference. This can be done using alternation, as in
3894 the example above, or by a quantifier with a minimum of zero.
3895
3896
3897 ASSERTIONS
3898
3899 An assertion is a test on the characters following or preceding the
3900 current matching point that does not actually consume any characters.
3901 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3902 described above.
3903
3904 More complicated assertions are coded as subpatterns. There are two
3905 kinds: those that look ahead of the current position in the subject
3906 string, and those that look behind it. An assertion subpattern is
3907 matched in the normal way, except that it does not cause the current
3908 matching position to be changed.
3909
3910 Assertion subpatterns are not capturing subpatterns, and may not be
3911 repeated, because it makes no sense to assert the same thing several
3912 times. If any kind of assertion contains capturing subpatterns within
3913 it, these are counted for the purposes of numbering the capturing sub-
3914 patterns in the whole pattern. However, substring capturing is carried
3915 out only for positive assertions, because it does not make sense for
3916 negative assertions.
3917
3918 Lookahead assertions
3919
3920 Lookahead assertions start with (?= for positive assertions and (?! for
3921 negative assertions. For example,
3922
3923 \w+(?=;)
3924
3925 matches a word followed by a semicolon, but does not include the semi-
3926 colon in the match, and
3927
3928 foo(?!bar)
3929
3930 matches any occurrence of "foo" that is not followed by "bar". Note
3931 that the apparently similar pattern
3932
3933 (?!foo)bar
3934
3935 does not find an occurrence of "bar" that is preceded by something
3936 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3937 the assertion (?!foo) is always true when the next three characters are
3938 "bar". A lookbehind assertion is needed to achieve the other effect.
3939
3940 If you want to force a matching failure at some point in a pattern, the
3941 most convenient way to do it is with (?!) because an empty string
3942 always matches, so an assertion that requires there not to be an empty
3943 string must always fail.
3944
3945 Lookbehind assertions
3946
3947 Lookbehind assertions start with (?<= for positive assertions and (?<!
3948 for negative assertions. For example,
3949
3950 (?<!foo)bar
3951
3952 does find an occurrence of "bar" that is not preceded by "foo". The
3953 contents of a lookbehind assertion are restricted such that all the
3954 strings it matches must have a fixed length. However, if there are sev-
3955 eral top-level alternatives, they do not all have to have the same
3956 fixed length. Thus
3957
3958 (?<=bullock|donkey)
3959
3960 is permitted, but
3961
3962 (?<!dogs?|cats?)
3963
3964 causes an error at compile time. Branches that match different length
3965 strings are permitted only at the top level of a lookbehind assertion.
3966 This is an extension compared with Perl (at least for 5.8), which
3967 requires all branches to match the same length of string. An assertion
3968 such as
3969
3970 (?<=ab(c|de))
3971
3972 is not permitted, because its single top-level branch can match two
3973 different lengths, but it is acceptable if rewritten to use two top-
3974 level branches:
3975
3976 (?<=abc|abde)
3977
3978 The implementation of lookbehind assertions is, for each alternative,
3979 to temporarily move the current position back by the fixed length and
3980 then try to match. If there are insufficient characters before the cur-
3981 rent position, the assertion fails.
3982
3983 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3984 mode) to appear in lookbehind assertions, because it makes it impossi-
3985 ble to calculate the length of the lookbehind. The \X and \R escapes,
3986 which can match different numbers of bytes, are also not permitted.
3987
3988 Possessive quantifiers can be used in conjunction with lookbehind
3989 assertions to specify efficient matching at the end of the subject
3990 string. Consider a simple pattern such as
3991
3992 abcd$
3993
3994 when applied to a long string that does not match. Because matching
3995 proceeds from left to right, PCRE will look for each "a" in the subject
3996 and then see if what follows matches the rest of the pattern. If the
3997 pattern is specified as
3998
3999 ^.*abcd$
4000
4001 the initial .* matches the entire string at first, but when this fails
4002 (because there is no following "a"), it backtracks to match all but the
4003 last character, then all but the last two characters, and so on. Once
4004 again the search for "a" covers the entire string, from right to left,
4005 so we are no better off. However, if the pattern is written as
4006
4007 ^.*+(?<=abcd)
4008
4009 there can be no backtracking for the .*+ item; it can match only the
4010 entire string. The subsequent lookbehind assertion does a single test
4011 on the last four characters. If it fails, the match fails immediately.
4012 For long strings, this approach makes a significant difference to the
4013 processing time.
4014
4015 Using multiple assertions
4016
4017 Several assertions (of any sort) may occur in succession. For example,
4018
4019 (?<=\d{3})(?<!999)foo
4020
4021 matches "foo" preceded by three digits that are not "999". Notice that
4022 each of the assertions is applied independently at the same point in
4023 the subject string. First there is a check that the previous three
4024 characters are all digits, and then there is a check that the same
4025 three characters are not "999". This pattern does not match "foo" pre-
4026 ceded by six characters, the first of which are digits and the last
4027 three of which are not "999". For example, it doesn't match "123abc-
4028 foo". A pattern to do that is
4029
4030 (?<=\d{3}...)(?<!999)foo
4031
4032 This time the first assertion looks at the preceding six characters,
4033 checking that the first three are digits, and then the second assertion
4034 checks that the preceding three characters are not "999".
4035
4036 Assertions can be nested in any combination. For example,
4037
4038 (?<=(?<!foo)bar)baz
4039
4040 matches an occurrence of "baz" that is preceded by "bar" which in turn
4041 is not preceded by "foo", while
4042
4043 (?<=\d{3}(?!999)...)foo
4044
4045 is another pattern that matches "foo" preceded by three digits and any
4046 three characters that are not "999".
4047
4048
4049 CONDITIONAL SUBPATTERNS
4050
4051 It is possible to cause the matching process to obey a subpattern con-
4052 ditionally or to choose between two alternative subpatterns, depending
4053 on the result of an assertion, or whether a previous capturing subpat-
4054 tern matched or not. The two possible forms of conditional subpattern
4055 are
4056
4057 (?(condition)yes-pattern)
4058 (?(condition)yes-pattern|no-pattern)
4059
4060 If the condition is satisfied, the yes-pattern is used; otherwise the
4061 no-pattern (if present) is used. If there are more than two alterna-
4062 tives in the subpattern, a compile-time error occurs.
4063
4064 There are four kinds of condition: references to subpatterns, refer-
4065 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4066
4067 Checking for a used subpattern by number
4068
4069 If the text between the parentheses consists of a sequence of digits,
4070 the condition is true if the capturing subpattern of that number has
4071 previously matched.
4072
4073 Consider the following pattern, which contains non-significant white
4074 space to make it more readable (assume the PCRE_EXTENDED option) and to
4075 divide it into three parts for ease of discussion:
4076
4077 ( \( )? [^()]+ (?(1) \) )
4078
4079 The first part matches an optional opening parenthesis, and if that
4080 character is present, sets it as the first captured substring. The sec-
4081 ond part matches one or more characters that are not parentheses. The
4082 third part is a conditional subpattern that tests whether the first set
4083 of parentheses matched or not. If they did, that is, if subject started
4084 with an opening parenthesis, the condition is true, and so the yes-pat-
4085 tern is executed and a closing parenthesis is required. Otherwise,
4086 since no-pattern is not present, the subpattern matches nothing. In
4087 other words, this pattern matches a sequence of non-parentheses,
4088 optionally enclosed in parentheses.
4089
4090 Checking for a used subpattern by name
4091
4092 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4093 used subpattern by name. For compatibility with earlier versions of
4094 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4095 also recognized. However, there is a possible ambiguity with this syn-
4096 tax, because subpattern names may consist entirely of digits. PCRE
4097 looks first for a named subpattern; if it cannot find one and the name
4098 consists entirely of digits, PCRE looks for a subpattern of that num-
4099 ber, which must be greater than zero. Using subpattern names that con-
4100 sist entirely of digits is not recommended.
4101
4102 Rewriting the above example to use a named subpattern gives this:
4103
4104 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4105
4106
4107 Checking for pattern recursion
4108
4109 If the condition is the string (R), and there is no subpattern with the
4110 name R, the condition is true if a recursive call to the whole pattern
4111 or any subpattern has been made. If digits or a name preceded by amper-
4112 sand follow the letter R, for example:
4113
4114 (?(R3)...) or (?(R&name)...)
4115
4116 the condition is true if the most recent recursion is into the subpat-
4117 tern whose number or name is given. This condition does not check the
4118 entire recursion stack.
4119
4120 At "top level", all these recursion test conditions are false. Recur-
4121 sive patterns are described below.
4122
4123 Defining subpatterns for use by reference only
4124
4125 If the condition is the string (DEFINE), and there is no subpattern
4126 with the name DEFINE, the condition is always false. In this case,
4127 there may be only one alternative in the subpattern. It is always
4128 skipped if control reaches this point in the pattern; the idea of
4129 DEFINE is that it can be used to define "subroutines" that can be ref-
4130 erenced from elsewhere. (The use of "subroutines" is described below.)
4131 For example, a pattern to match an IPv4 address could be written like
4132 this (ignore whitespace and line breaks):
4133
4134 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4135 \b (?&byte) (\.(?&byte)){3} \b
4136
4137 The first part of the pattern is a DEFINE group inside which a another
4138 group named "byte" is defined. This matches an individual component of
4139 an IPv4 address (a number less than 256). When matching takes place,
4140 this part of the pattern is skipped because DEFINE acts like a false
4141 condition.
4142
4143 The rest of the pattern uses references to the named group to match the
4144 four dot-separated components of an IPv4 address, insisting on a word
4145 boundary at each end.
4146
4147 Assertion conditions
4148
4149 If the condition is not in any of the above formats, it must be an
4150 assertion. This may be a positive or negative lookahead or lookbehind
4151 assertion. Consider this pattern, again containing non-significant
4152 white space, and with the two alternatives on the second line:
4153
4154 (?(?=[^a-z]*[a-z])
4155 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4156
4157 The condition is a positive lookahead assertion that matches an
4158 optional sequence of non-letters followed by a letter. In other words,
4159 it tests for the presence of at least one letter in the subject. If a
4160 letter is found, the subject is matched against the first alternative;
4161 otherwise it is matched against the second. This pattern matches
4162 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4163 letters and dd are digits.
4164
4165
4166 COMMENTS
4167
4168 The sequence (?# marks the start of a comment that continues up to the
4169 next closing parenthesis. Nested parentheses are not permitted. The
4170 characters that make up a comment play no part in the pattern matching
4171 at all.
4172
4173 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4174 character class introduces a comment that continues to immediately
4175 after the next newline in the pattern.
4176
4177
4178 RECURSIVE PATTERNS
4179
4180 Consider the problem of matching a string in parentheses, allowing for
4181 unlimited nested parentheses. Without the use of recursion, the best
4182 that can be done is to use a pattern that matches up to some fixed
4183 depth of nesting. It is not possible to handle an arbitrary nesting
4184 depth.
4185
4186 For some time, Perl has provided a facility that allows regular expres-
4187 sions to recurse (amongst other things). It does this by interpolating
4188 Perl code in the expression at run time, and the code can refer to the
4189 expression itself. A Perl pattern using code interpolation to solve the
4190 parentheses problem can be created like this:
4191
4192 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4193
4194 The (?p{...}) item interpolates Perl code at run time, and in this case
4195 refers recursively to the pattern in which it appears.
4196
4197 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4198 it supports special syntax for recursion of the entire pattern, and
4199 also for individual subpattern recursion. After its introduction in
4200 PCRE and Python, this kind of recursion was introduced into Perl at
4201 release 5.10.
4202
4203 A special item that consists of (? followed by a number greater than
4204 zero and a closing parenthesis is a recursive call of the subpattern of
4205 the given number, provided that it occurs inside that subpattern. (If
4206 not, it is a "subroutine" call, which is described in the next sec-
4207 tion.) The special item (?R) or (?0) is a recursive call of the entire
4208 regular expression.
4209
4210 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4211 always treated as an atomic group. That is, once it has matched some of
4212 the subject string, it is never re-entered, even if it contains untried
4213 alternatives and there is a subsequent matching failure.
4214
4215 This PCRE pattern solves the nested parentheses problem (assume the
4216 PCRE_EXTENDED option is set so that white space is ignored):
4217
4218 \( ( (?>[^()]+) | (?R) )* \)
4219
4220 First it matches an opening parenthesis. Then it matches any number of
4221 substrings which can either be a sequence of non-parentheses, or a
4222 recursive match of the pattern itself (that is, a correctly parenthe-
4223 sized substring). Finally there is a closing parenthesis.
4224
4225 If this were part of a larger pattern, you would not want to recurse
4226 the entire pattern, so instead you could use this:
4227
4228 ( \( ( (?>[^()]+) | (?1) )* \) )
4229
4230 We have put the pattern into parentheses, and caused the recursion to
4231 refer to them instead of the whole pattern. In a larger pattern, keep-
4232 ing track of parenthesis numbers can be tricky. It may be more conve-
4233 nient to use named parentheses instead. The Perl syntax for this is
4234 (?&name); PCRE's earlier syntax (?P>name) is also supported. We could
4235 rewrite the above example as follows:
4236
4237 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4238
4239 If there is more than one subpattern with the same name, the earliest
4240 one is used. This particular example pattern contains nested unlimited
4241 repeats, and so the use of atomic grouping for matching strings of non-
4242 parentheses is important when applying the pattern to strings that do
4243 not match. For example, when this pattern is applied to
4244
4245 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4246
4247 it yields "no match" quickly. However, if atomic grouping is not used,
4248 the match runs for a very long time indeed because there are so many
4249 different ways the + and * repeats can carve up the subject, and all
4250 have to be tested before failure can be reported.
4251
4252 At the end of a match, the values set for any capturing subpatterns are
4253 those from the outermost level of the recursion at which the subpattern
4254 value is set. If you want to obtain intermediate values, a callout
4255 function can be used (see below and the pcrecallout documentation). If
4256 the pattern above is matched against
4257
4258 (ab(cd)ef)
4259
4260 the value for the capturing parentheses is "ef", which is the last
4261 value taken on at the top level. If additional parentheses are added,
4262 giving
4263
4264 \( ( ( (?>[^()]+) | (?R) )* ) \)
4265 ^ ^
4266 ^ ^
4267
4268 the string they capture is "ab(cd)ef", the contents of the top level
4269 parentheses. If there are more than 15 capturing parentheses in a pat-
4270 tern, PCRE has to obtain extra memory to store data during a recursion,
4271 which it does by using pcre_malloc, freeing it via pcre_free after-
4272 wards. If no memory can be obtained, the match fails with the
4273 PCRE_ERROR_NOMEMORY error.
4274
4275 Do not confuse the (?R) item with the condition (R), which tests for
4276 recursion. Consider this pattern, which matches text in angle brack-
4277 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4278 brackets (that is, when recursing), whereas any characters are permit-
4279 ted at the outer level.
4280
4281 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4282
4283 In this pattern, (?(R) is the start of a conditional subpattern, with
4284 two different alternatives for the recursive and non-recursive cases.
4285 The (?R) item is the actual recursive call.
4286
4287
4288 SUBPATTERNS AS SUBROUTINES
4289
4290 If the syntax for a recursive subpattern reference (either by number or
4291 by name) is used outside the parentheses to which it refers, it oper-
4292 ates like a subroutine in a programming language. The "called" subpat-
4293 tern may be defined before or after the reference. An earlier example
4294 pointed out that the pattern
4295
4296 (sens|respons)e and \1ibility
4297
4298 matches "sense and sensibility" and "response and responsibility", but
4299 not "sense and responsibility". If instead the pattern
4300
4301 (sens|respons)e and (?1)ibility
4302
4303 is used, it does match "sense and responsibility" as well as the other
4304 two strings. Another example is given in the discussion of DEFINE
4305 above.
4306
4307 Like recursive subpatterns, a "subroutine" call is always treated as an
4308 atomic group. That is, once it has matched some of the subject string,
4309 it is never re-entered, even if it contains untried alternatives and
4310 there is a subsequent matching failure.
4311
4312 When a subpattern is used as a subroutine, processing options such as
4313 case-independence are fixed when the subpattern is defined. They cannot
4314 be changed for different calls. For example, consider this pattern:
4315
4316 (abc)(?i:(?1))
4317
4318 It matches "abcabc". It does not match "abcABC" because the change of
4319 processing option does not affect the called subpattern.
4320
4321
4322 CALLOUTS
4323
4324 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4325 Perl code to be obeyed in the middle of matching a regular expression.
4326 This makes it possible, amongst other things, to extract different sub-
4327 strings that match the same pair of parentheses when there is a repeti-
4328 tion.
4329
4330 PCRE provides a similar feature, but of course it cannot obey arbitrary
4331 Perl code. The feature is called "callout". The caller of PCRE provides
4332 an external function by putting its entry point in the global variable
4333 pcre_callout. By default, this variable contains NULL, which disables
4334 all calling out.
4335
4336 Within a regular expression, (?C) indicates the points at which the
4337 external function is to be called. If you want to identify different
4338 callout points, you can put a number less than 256 after the letter C.
4339 The default value is zero. For example, this pattern has two callout
4340 points:
4341
4342 (?C1)abc(?C2)def
4343
4344 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4345 automatically installed before each item in the pattern. They are all
4346 numbered 255.
4347
4348 During matching, when PCRE reaches a callout point (and pcre_callout is
4349 set), the external function is called. It is provided with the number
4350 of the callout, the position in the pattern, and, optionally, one item
4351 of data originally supplied by the caller of pcre_exec(). The callout
4352 function may cause matching to proceed, to backtrack, or to fail alto-
4353 gether. A complete description of the interface to the callout function
4354 is given in the pcrecallout documentation.
4355
4356
4357 SEE ALSO
4358
4359 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4360
4361
4362 AUTHOR
4363
4364 Philip Hazel
4365 University Computing Service
4366 Cambridge CB2 3QH, England.
4367
4368
4369 REVISION
4370
4371 Last updated: 06 March 2007
4372 Copyright (c) 1997-2007 University of Cambridge.
4373 ------------------------------------------------------------------------------
4374
4375
4376 PCREPARTIAL(3) PCREPARTIAL(3)
4377
4378
4379 NAME
4380 PCRE - Perl-compatible regular expressions
4381
4382
4383 PARTIAL MATCHING IN PCRE
4384
4385 In normal use of PCRE, if the subject string that is passed to
4386 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4387 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4388 There are circumstances where it might be helpful to distinguish this
4389 case from other cases in which there is no match.
4390
4391 Consider, for example, an application where a human is required to type
4392 in data for a field with specific formatting requirements. An example
4393 might be a date in the form ddmmmyy, defined by this pattern:
4394
4395 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4396
4397 If the application sees the user's keystrokes one by one, and can check
4398 that what has been typed so far is potentially valid, it is able to
4399 raise an error as soon as a mistake is made, possibly beeping and not
4400 reflecting the character that has been typed. This immediate feedback
4401 is likely to be a better user interface than a check that is delayed
4402 until the entire string has been entered.
4403
4404 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4405 TIAL option, which can be set when calling pcre_exec() or
4406 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4407 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4408 during the matching process the last part of the subject string matched
4409 part of the pattern. Unfortunately, for non-anchored matching, it is
4410 not possible to obtain the position of the start of the partial match.
4411 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4412
4413 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4414 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4415 the subject is reached, there have been no complete matches, but there
4416 is still at least one matching possibility. The portion of the string
4417 that provided the partial match is set as the first matching string.
4418
4419 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4420 the last literal byte in a pattern, and abandons matching immediately
4421 if such a byte is not present in the subject string. This optimization
4422 cannot be used for a subject string that might match only partially.
4423
4424
4425 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4426
4427 Because of the way certain internal optimizations are implemented in
4428 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4429 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4430 used. For pcre_exec(), repeated single characters such as
4431
4432 a{2,4}
4433
4434 and repeated single metasequences such as
4435
4436 \d+
4437
4438 are not permitted if the maximum number of occurrences is greater than
4439 one. Optional items such as \d? (where the maximum is one) are permit-
4440 ted. Quantifiers with any values are permitted after parentheses, so
4441 the invalid examples above can be coded thus:
4442
4443 (a){2,4}
4444 (\d)+
4445
4446 These constructions run more slowly, but for the kinds of application
4447 that are envisaged for this facility, this is not felt to be a major
4448 restriction.
4449
4450 If PCRE_PARTIAL is set for a pattern that does not conform to the
4451 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4452 (-13).
4453
4454
4455 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4456
4457 If the escape sequence \P is present in a pcretest data line, the
4458 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4459 uses the date example quoted above:
4460
4461 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4462 data> 25jun04\P
4463 0: 25jun04
4464 1: jun
4465 data> 25dec3\P
4466 Partial match
4467 data> 3ju\P
4468 Partial match
4469 data> 3juj\P
4470 No match
4471 data> j\P
4472 No match
4473
4474 The first data string is matched completely, so pcretest shows the
4475 matched substrings. The remaining four strings do not match the com-
4476 plete pattern, but the first two are partial matches. The same test,
4477 using pcre_dfa_exec() matching (by means of the \D escape sequence),
4478 produces the following output:
4479
4480 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4481 data> 25jun04\P\D
4482 0: 25jun04
4483 data> 23dec3\P\D
4484 Partial match: 23dec3
4485 data> 3ju\P\D
4486 Partial match: 3ju
4487 data> 3juj\P\D
4488 No match
4489 data> j\P\D
4490 No match
4491
4492 Notice that in this case the portion of the string that was matched is
4493 made available.
4494
4495
4496 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4497
4498 When a partial match has been found using pcre_dfa_exec(), it is possi-
4499 ble to continue the match by providing additional subject data and
4500 calling pcre_dfa_exec() again with the same compiled regular expres-
4501 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
4502 the same working space as before, because this is where details of the
4503 previous partial match are stored. Here is an example using pcretest,
4504 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
4505 \D are as above):
4506
4507 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4508 data> 23ja\P\D
4509 Partial match: 23ja
4510 data> n05\R\D
4511 0: n05
4512
4513 The first call has "23ja" as the subject, and requests partial match-
4514 ing; the second call has "n05" as the subject for the continued
4515 (restarted) match. Notice that when the match is complete, only the
4516 last part is shown; PCRE does not retain the previously partially-
4517 matched string. It is up to the calling program to do that if it needs
4518 to.
4519
4520 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
4521 matching over multiple segments. This facility can be used to pass very
4522 long subject strings to pcre_dfa_exec(). However, some care is needed
4523 for certain types of pattern.
4524
4525 1. If the pattern contains tests for the beginning or end of a line,
4526 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4527 ate, when the subject string for any call does not contain the begin-
4528 ning or end of a line.
4529
4530 2. If the pattern contains backward assertions (including \b or \B),
4531 you need to arrange for some overlap in the subject strings to allow
4532 for this. For example, you could pass the subject in chunks that are
4533 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4534 set to 200 and the previous 200 bytes at the start of the buffer.
4535
4536 3. Matching a subject string that is split into multiple segments does
4537 not always produce exactly the same result as matching over one single
4538 long string. The difference arises when there are multiple matching
4539 possibilities, because a partial match result is given only when there
4540 are no completed matches in a call to pcre_dfa_exec(). This means that
4541 as soon as the shortest match has been found, continuation to a new
4542 subject segment is no longer possible. Consider this pcretest example:
4543
4544 re> /dog(sbody)?/
4545 data> do\P\D
4546 Partial match: do
4547 data> gsb\R\P\D
4548 0: g
4549 data> dogsbody\D
4550 0: dogsbody
4551 1: dog
4552
4553 The pattern matches the words "dog" or "dogsbody". When the subject is
4554 presented in several parts ("do" and "gsb" being the first two) the
4555 match stops when "dog" has been found, and it is not possible to con-
4556 tinue. On the other hand, if "dogsbody" is presented as a single
4557 string, both matches are found.
4558
4559 Because of this phenomenon, it does not usually make sense to end a
4560 pattern that is going to be matched in this way with a variable repeat.
4561
4562 4. Patterns that contain alternatives at the top level which do not all
4563 start with the same pattern item may not work as expected. For example,
4564 consider this pattern:
4565
4566 1234|3789
4567
4568 If the first part of the subject is "ABC123", a partial match of the
4569 first alternative is found at offset 3. There is no partial match for
4570 the second alternative, because such a match does not start at the same
4571 point in the subject string. Attempting to continue with the string
4572 "789" does not yield a match because only those alternatives that match
4573 at one point in the subject are remembered. The problem arises because
4574 the start of the second alternative matches within the first alterna-
4575 tive. There is no problem with anchored patterns or patterns such as:
4576
4577 1234|ABCD
4578
4579 where no string can be a partial match for both alternatives.
4580
4581
4582 AUTHOR
4583
4584 Philip Hazel
4585 University Computing Service
4586 Cambridge CB2 3QH, England.
4587
4588
4589 REVISION
4590
4591 Last updated: 06 March 2007
4592 Copyright (c) 1997-2007 University of Cambridge.
4593 ------------------------------------------------------------------------------
4594
4595
4596 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4597
4598
4599 NAME
4600 PCRE - Perl-compatible regular expressions
4601
4602
4603 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4604
4605 If you are running an application that uses a large number of regular
4606 expression patterns, it may be useful to store them in a precompiled
4607 form instead of having to compile them every time the application is
4608 run. If you are not using any private character tables (see the
4609 pcre_maketables() documentation), this is relatively straightforward.
4610 If you are using private tables, it is a little bit more complicated.
4611
4612 If you save compiled patterns to a file, you can copy them to a differ-
4613 ent host and run them there. This works even if the new host has the
4614 opposite endianness to the one on which the patterns were compiled.
4615 There may be a small performance penalty, but it should be insignifi-
4616 cant.
4617
4618
4619 SAVING A COMPILED PATTERN
4620 The value returned by pcre_compile() points to a single block of memory
4621 that holds the compiled pattern and associated data. You can find the
4622 length of this block in bytes by calling pcre_fullinfo() with an argu-
4623 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4624 manner. Here is sample code that compiles a pattern and writes it to a
4625 file. It assumes that the variable fd refers to a file that is open for
4626 output:
4627
4628 int erroroffset, rc, size;
4629 char *error;
4630 pcre *re;
4631
4632 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4633 if (re == NULL) { ... handle errors ... }
4634 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4635 if (rc < 0) { ... handle errors ... }
4636 rc = fwrite(re, 1, size, fd);
4637 if (rc != size) { ... handle errors ... }
4638
4639 In this example, the bytes that comprise the compiled pattern are
4640 copied exactly. Note that this is binary data that may contain any of
4641 the 256 possible byte values. On systems that make a distinction
4642 between binary and non-binary data, be sure that the file is opened for
4643 binary output.
4644
4645 If you want to write more than one pattern to a file, you will have to
4646 devise a way of separating them. For binary data, preceding each pat-
4647 tern with its length is probably the most straightforward approach.
4648 Another possibility is to write out the data in hexadecimal instead of
4649 binary, one pattern to a line.
4650
4651 Saving compiled patterns in a file is only one possible way of storing
4652 them for later use. They could equally well be saved in a database, or
4653 in the memory of some daemon process that passes them via sockets to
4654 the processes that want them.
4655
4656 If the pattern has been studied, it is also possible to save the study
4657 data in a similar way to the compiled pattern itself. When studying
4658 generates additional information, pcre_study() returns a pointer to a
4659 pcre_extra data block. Its format is defined in the section on matching
4660 a pattern in the pcreapi documentation. The study_data field points to
4661 the binary study data, and this is what you must save (not the
4662 pcre_extra block itself). The length of the study data can be obtained
4663 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4664 Remember to check that pcre_study() did return a non-NULL value before
4665 trying to save the study data.
4666
4667
4668 RE-USING A PRECOMPILED PATTERN
4669
4670 Re-using a precompiled pattern is straightforward. Having reloaded it
4671 into main memory, you pass its pointer to pcre_exec() or
4672 pcre_dfa_exec() in the usual way. This should work even on another
4673 host, and even if that host has the opposite endianness to the one
4674 where the pattern was compiled.
4675
4676 However, if you passed a pointer to custom character tables when the
4677 pattern was compiled (the tableptr argument of pcre_compile()), you
4678 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4679 because the value saved with the compiled pattern will obviously be
4680 nonsense. A field in a pcre_extra() block is used to pass this data, as
4681 described in the section on matching a pattern in the pcreapi documen-
4682 tation.
4683
4684 If you did not provide custom character tables when the pattern was
4685 compiled, the pointer in the compiled pattern is NULL, which causes
4686 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4687 take any special action at run time in this case.
4688
4689 If you saved study data with the compiled pattern, you need to create
4690 your own pcre_extra data block and set the study_data field to point to
4691 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4692 bit in the flags field to indicate that study data is present. Then
4693 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4694 usual way.
4695
4696
4697 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4698
4699 The layout of the control block that is at the start of the data that
4700 makes up a compiled pattern was changed for release 5.0. If you have
4701 any saved patterns that were compiled with previous releases (not a
4702 facility that was previously advertised), you will have to recompile
4703 them for release 5.0 and above.
4704
4705 If you have any saved patterns in UTF-8 mode that use \p or \P that
4706 were compiled with any release up to and including 6.4, you will have
4707 to recompile them for release 6.5 and above.
4708
4709 All saved patterns from earlier releases must be recompiled for release
4710 7.0 or higher, because there was an internal reorganization at that
4711 release.
4712
4713
4714 AUTHOR
4715
4716 Philip Hazel
4717 University Computing Service
4718 Cambridge CB2 3QH, England.
4719
4720
4721 REVISION
4722
4723 Last updated: 06 March 2007
4724 Copyright (c) 1997-2007 University of Cambridge.
4725 ------------------------------------------------------------------------------
4726
4727
4728 PCREPERFORM(3) PCREPERFORM(3)
4729
4730
4731 NAME
4732 PCRE - Perl-compatible regular expressions
4733
4734
4735 PCRE PERFORMANCE
4736
4737 Two aspects of performance are discussed below: memory usage and pro-
4738 cessing time. The way you express your pattern as a regular expression
4739 can affect both of them.
4740
4741
4742 MEMORY USAGE
4743
4744 Patterns are compiled by PCRE into a reasonably efficient byte code, so
4745 that most simple patterns do not use much memory. However, there is one
4746 case where memory usage can be unexpectedly large. When a parenthesized
4747 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
4748 ited maximum, the whole subpattern is repeated in the compiled code.
4749 For example, the pattern
4750
4751 (abc|def){2,4}
4752
4753 is compiled as if it were
4754
4755 (abc|def)(abc|def)((abc|def)(abc|def)?)?
4756
4757 (Technical aside: It is done this way so that backtrack points within
4758 each of the repetitions can be independently maintained.)
4759
4760 For regular expressions whose quantifiers use only small numbers, this
4761 is not usually a problem. However, if the numbers are large, and par-
4762 ticularly if such repetitions are nested, the memory usage can become
4763 an embarrassment. For example, the very simple pattern
4764
4765 ((ab){1,1000}c){1,3}
4766
4767 uses 51K bytes when compiled. When PCRE is compiled with its default
4768 internal pointer size of two bytes, the size limit on a compiled pat-
4769 tern is 64K, and this is reached with the above pattern if the outer
4770 repetition is increased from 3 to 4. PCRE can be compiled to use larger
4771 internal pointers and thus handle larger compiled patterns, but it is
4772 better to try to rewrite your pattern to use less memory if you can.
4773
4774 One way of reducing the memory usage for such patterns is to make use
4775 of PCRE's "subroutine" facility. Re-writing the above pattern as
4776
4777 ((ab)(?2){0,999}c)(?1){0,2}
4778
4779 reduces the memory requirements to 18K, and indeed it remains under 20K
4780 even with the outer repetition increased to 100. However, this pattern
4781 is not exactly equivalent, because the "subroutine" calls are treated
4782 as atomic groups into which there can be no backtracking if there is a
4783 subsequent matching failure. Therefore, PCRE cannot do this kind of
4784 rewriting automatically. Furthermore, there is a noticeable loss of
4785 speed when executing the modified pattern. Nevertheless, if the atomic
4786 grouping is not a problem and the loss of speed is acceptable, this
4787 kind of rewriting will allow you to process patterns that PCRE cannot
4788 otherwise handle.
4789
4790
4791 PROCESSING TIME
4792
4793 Certain items in regular expression patterns are processed more effi-
4794 ciently than others. It is more efficient to use a character class like
4795 [aeiou] than a set of single-character alternatives such as
4796 (a|e|i|o|u). In general, the simplest construction that provides the
4797 required behaviour is usually the most efficient. Jeffrey Friedl's book
4798 contains a lot of useful general discussion about optimizing regular
4799 expressions for efficient performance. This document contains a few
4800 observations about PCRE.
4801
4802 Using Unicode character properties (the \p, \P, and \X escapes) is
4803 slow, because PCRE has to scan a structure that contains data for over
4804 fifteen thousand characters whenever it needs a character's property.
4805 If you can find an alternative pattern that does not use character
4806 properties, it will probably be faster.
4807
4808 When a pattern begins with .* not in parentheses, or in parentheses
4809 that are not the subject of a backreference, and the PCRE_DOTALL option
4810 is set, the pattern is implicitly anchored by PCRE, since it can match
4811 only at the start of a subject string. However, if PCRE_DOTALL is not
4812 set, PCRE cannot make this optimization, because the . metacharacter
4813 does not then match a newline, and if the subject string contains new-
4814 lines, the pattern may match from the character immediately following
4815 one of them instead of from the very start. For example, the pattern
4816
4817 .*second
4818
4819 matches the subject "first\nand second" (where \n stands for a newline
4820 character), with the match starting at the seventh character. In order
4821 to do this, PCRE has to retry the match starting after every newline in
4822 the subject.
4823
4824 If you are using such a pattern with subject strings that do not con-
4825 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4826 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4827 ing. That saves PCRE from having to scan along the subject looking for
4828 a newline to restart at.
4829
4830 Beware of patterns that contain nested indefinite repeats. These can
4831 take a long time to run when applied to a string that does not match.
4832 Consider the pattern fragment
4833
4834 ^(a+)*
4835
4836 This can match "aaaa" in 16 different ways, and this number increases
4837 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4838 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
4839 repeats can match different numbers of times.) When the remainder of
4840 the pattern is such that the entire match is going to fail, PCRE has in
4841 principle to try every possible variation, and this can take an
4842 extremely long time, even for relatively short strings.
4843
4844 An optimization catches some of the more simple cases such as
4845
4846 (a+)*b
4847
4848 where a literal character follows. Before embarking on the standard
4849 matching procedure, PCRE checks that there is a "b" later in the sub-
4850 ject string, and if there is not, it fails the match immediately. How-
4851 ever, when there is no following literal this optimization cannot be
4852 used. You can see the difference by comparing the behaviour of
4853
4854 (a+)*\d
4855
4856 with the pattern above. The former gives a failure almost instantly
4857 when applied to a whole line of "a" characters, whereas the latter
4858 takes an appreciable time with strings longer than about 20 characters.
4859
4860 In many cases, the solution to this kind of performance issue is to use
4861 an atomic group or a possessive quantifier.
4862
4863
4864 AUTHOR
4865
4866 Philip Hazel
4867 University Computing Service
4868 Cambridge CB2 3QH, England.
4869
4870
4871 REVISION
4872
4873 Last updated: 06 March 2007
4874 Copyright (c) 1997-2007 University of Cambridge.
4875 ------------------------------------------------------------------------------
4876
4877
4878 PCREPOSIX(3) PCREPOSIX(3)
4879
4880
4881 NAME
4882 PCRE - Perl-compatible regular expressions.
4883
4884
4885 SYNOPSIS OF POSIX API
4886
4887 #include <pcreposix.h>
4888
4889 int regcomp(regex_t *preg, const char *pattern,
4890 int cflags);
4891
4892 int regexec(regex_t *preg, const char *string,
4893 size_t nmatch, regmatch_t pmatch[], int eflags);
4894
4895 size_t regerror(int errcode, const regex_t *preg,
4896 char *errbuf, size_t errbuf_size);
4897
4898 void regfree(regex_t *preg);
4899
4900
4901 DESCRIPTION
4902
4903 This set of functions provides a POSIX-style API to the PCRE regular
4904 expression package. See the pcreapi documentation for a description of
4905 PCRE's native API, which contains much additional functionality.
4906
4907 The functions described here are just wrapper functions that ultimately
4908 call the PCRE native API. Their prototypes are defined in the
4909 pcreposix.h header file, and on Unix systems the library itself is
4910 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4911 command for linking an application that uses them. Because the POSIX
4912 functions call the native ones, it is also necessary to add -lpcre.
4913
4914 I have implemented only those option bits that can be reasonably mapped
4915 to PCRE native options. In addition, the option REG_EXTENDED is defined
4916 with the value zero. This has no effect, but since programs that are
4917 written to the POSIX interface often use it, this makes it easier to
4918 slot in PCRE as a replacement library. Other POSIX options are not even
4919 defined.
4920
4921 When PCRE is called via these functions, it is only the API that is
4922 POSIX-like in style. The syntax and semantics of the regular expres-
4923 sions themselves are still those of Perl, subject to the setting of
4924 various PCRE options, as described below. "POSIX-like in style" means
4925 that the API approximates to the POSIX definition; it is not fully
4926 POSIX-compatible, and in multi-byte encoding domains it is probably
4927 even less compatible.
4928
4929 The header for these functions is supplied as pcreposix.h to avoid any
4930 potential clash with other POSIX libraries. It can, of course, be
4931 renamed or aliased as regex.h, which is the "correct" name. It provides
4932 two structure types, regex_t for compiled internal forms, and reg-
4933 match_t for returning captured substrings. It also defines some con-
4934 stants whose names start with "REG_"; these are used for setting
4935 options and identifying error codes.
4936
4937
4938 COMPILING A PATTERN
4939
4940 The function regcomp() is called to compile a pattern into an internal
4941 form. The pattern is a C string terminated by a binary zero, and is
4942 passed in the argument pattern. The preg argument is a pointer to a
4943 regex_t structure that is used as a base for storing information about
4944 the compiled regular expression.
4945
4946 The argument cflags is either zero, or contains one or more of the bits
4947 defined by the following macros:
4948
4949 REG_DOTALL
4950
4951 The PCRE_DOTALL option is set when the regular expression is passed for
4952 compilation to the native function. Note that REG_DOTALL is not part of
4953 the POSIX standard.
4954
4955 REG_ICASE
4956
4957 The PCRE_CASELESS option is set when the regular expression is passed
4958 for compilation to the native function.
4959
4960 REG_NEWLINE
4961
4962 The PCRE_MULTILINE option is set when the regular expression is passed
4963 for compilation to the native function. Note that this does not mimic
4964 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
4965 tion).
4966
4967 REG_NOSUB
4968
4969 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
4970 passed for compilation to the native function. In addition, when a pat-
4971 tern that is compiled with this flag is passed to regexec() for match-
4972 ing, the nmatch and pmatch arguments are ignored, and no captured
4973 strings are returned.
4974
4975 REG_UTF8
4976
4977 The PCRE_UTF8 option is set when the regular expression is passed for
4978 compilation to the native function. This causes the pattern itself and
4979 all data strings used for matching it to be treated as UTF-8 strings.
4980 Note that REG_UTF8 is not part of the POSIX standard.
4981
4982 In the absence of these flags, no options are passed to the native
4983 function. This means the the regex is compiled with PCRE default
4984 semantics. In particular, the way it handles newline characters in the
4985 subject string is the Perl way, not the POSIX way. Note that setting
4986 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4987 It does not affect the way newlines are matched by . (they aren't) or
4988 by a negative class such as [^a] (they are).
4989
4990 The yield of regcomp() is zero on success, and non-zero otherwise. The
4991 preg structure is filled in on success, and one member of the structure
4992 is public: re_nsub contains the number of capturing subpatterns in the
4993 regular expression. Various error codes are defined in the header file.
4994
4995
4996 MATCHING NEWLINE CHARACTERS
4997
4998 This area is not simple, because POSIX and Perl take different views of
4999 things. It is not possible to get PCRE to obey POSIX semantics, but
5000 then PCRE was never intended to be a POSIX engine. The following table
5001 lists the different possibilities for matching newline characters in
5002 PCRE:
5003
5004 Default Change with
5005
5006 . matches newline no PCRE_DOTALL
5007 newline matches [^a] yes not changeable
5008 $ matches \n at end yes PCRE_DOLLARENDONLY
5009 $ matches \n in middle no PCRE_MULTILINE
5010 ^ matches \n in middle no PCRE_MULTILINE
5011
5012 This is the equivalent table for POSIX:
5013
5014 Default Change with
5015
5016 . matches newline yes REG_NEWLINE
5017 newline matches [^a] yes REG_NEWLINE
5018 $ matches \n at end no REG_NEWLINE
5019 $ matches \n in middle no REG_NEWLINE
5020 ^ matches \n in middle no REG_NEWLINE
5021
5022 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5023 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5024 no way to stop newline from matching [^a].
5025
5026 The default POSIX newline handling can be obtained by setting
5027 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5028 behave exactly as for the REG_NEWLINE action.
5029
5030
5031 MATCHING A PATTERN
5032
5033 The function regexec() is called to match a compiled pattern preg
5034 against a given string, which is terminated by a zero byte, subject to
5035 the options in eflags. These can be:
5036
5037 REG_NOTBOL
5038
5039 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5040 function.
5041
5042 REG_NOTEOL
5043
5044 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5045 function.
5046
5047 If the pattern was compiled with the REG_NOSUB flag, no data about any
5048 matched strings is returned. The nmatch and pmatch arguments of
5049 regexec() are ignored.
5050
5051 Otherwise,the portion of the string that was matched, and also any cap-
5052 tured substrings, are returned via the pmatch argument, which points to
5053 an array of nmatch structures of type regmatch_t, containing the mem-
5054 bers rm_so and rm_eo. These contain the offset to the first character
5055 of each substring and the offset to the first character after the end
5056 of each substring, respectively. The 0th element of the vector relates
5057 to the entire portion of string that was matched; subsequent elements
5058 relate to the capturing subpatterns of the regular expression. Unused
5059 entries in the array have both structure members set to -1.
5060
5061 A successful match yields a zero return; various error codes are
5062 defined in the header file, of which REG_NOMATCH is the "expected"
5063 failure code.
5064
5065
5066 ERROR MESSAGES
5067
5068 The regerror() function maps a non-zero errorcode from either regcomp()
5069 or regexec() to a printable message. If preg is not NULL, the error
5070 should have arisen from the use of that structure. A message terminated
5071 by a binary zero is placed in errbuf. The length of the message,
5072 including the zero, is limited to errbuf_size. The yield of the func-
5073 tion is the size of buffer needed to hold the whole message.
5074
5075
5076 MEMORY USAGE
5077
5078 Compiling a regular expression causes memory to be allocated and asso-
5079 ciated with the preg structure. The function regfree() frees all such
5080 memory, after which preg may no longer be used as a compiled expres-
5081 sion.
5082
5083
5084 AUTHOR
5085
5086 Philip Hazel
5087 University Computing Service
5088 Cambridge CB2 3QH, England.
5089
5090
5091 REVISION
5092
5093 Last updated: 06 March 2007
5094 Copyright (c) 1997-2007 University of Cambridge.
5095 ------------------------------------------------------------------------------
5096
5097
5098 PCRECPP(3) PCRECPP(3)
5099
5100
5101 NAME
5102 PCRE - Perl-compatible regular expressions.
5103
5104
5105 SYNOPSIS OF C++ WRAPPER
5106
5107 #include <pcrecpp.h>
5108
5109
5110 DESCRIPTION
5111
5112 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5113 functionality was added by Giuseppe Maxia. This brief man page was con-
5114 structed from the notes in the pcrecpp.h file, which should be con-
5115 sulted for further details.
5116
5117
5118 MATCHING INTERFACE
5119
5120 The "FullMatch" operation checks that supplied text matches a supplied
5121 pattern exactly. If pointer arguments are supplied, it copies matched
5122 sub-strings that match sub-patterns into them.
5123
5124 Example: successful match
5125 pcrecpp::RE re("h.*o");
5126 re.FullMatch("hello");
5127
5128 Example: unsuccessful match (requires full match):
5129 pcrecpp::RE re("e");
5130 !re.FullMatch("hello");
5131
5132 Example: creating a temporary RE object:
5133 pcrecpp::RE("h.*o").FullMatch("hello");
5134
5135 You can pass in a "const char*" or a "string" for "text". The examples
5136 below tend to use a const char*. You can, as in the different examples
5137 above, store the RE object explicitly in a variable or use a temporary
5138 RE object. The examples below use one mode or the other arbitrarily.
5139 Either could correctly be used for any of these examples.
5140
5141 You must supply extra pointer arguments to extract matched subpieces.
5142
5143 Example: extracts "ruby" into "s" and 1234 into "i"
5144 int i;
5145 string s;
5146 pcrecpp::RE re("(\\w+):(\\d+)");
5147 re.FullMatch("ruby:1234", &s, &i);
5148
5149 Example: does not try to extract any extra sub-patterns
5150 re.FullMatch("ruby:1234", &s);
5151
5152 Example: does not try to extract into NULL
5153 re.FullMatch("ruby:1234", NULL, &i);
5154
5155 Example: integer overflow causes failure
5156 !re.FullMatch("ruby:1234567891234", NULL, &i);
5157
5158 Example: fails because there aren't enough sub-patterns:
5159 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5160
5161 Example: fails because string cannot be stored in integer
5162 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5163
5164 The provided pointer arguments can be pointers to any scalar numeric
5165 type, or one of:
5166
5167 string (matched piece is copied to string)
5168 StringPiece (StringPiece is mutated to point to matched piece)
5169 T (where "bool T::ParseFrom(const char*, int)" exists)
5170 NULL (the corresponding matched sub-pattern is not copied)
5171
5172 The function returns true iff all of the following conditions are sat-
5173 isfied:
5174
5175 a. "text" matches "pattern" exactly;
5176
5177 b. The number of matched sub-patterns is >= number of supplied
5178 pointers;
5179
5180 c. The "i"th argument has a suitable type for holding the
5181 string captured as the "i"th sub-pattern. If you pass in
5182 NULL for the "i"th argument, or pass fewer arguments than
5183 number of sub-patterns, "i"th captured sub-pattern is
5184 ignored.
5185
5186 CAVEAT: An optional sub-pattern that does not exist in the matched
5187 string is assigned the empty string. Therefore, the following will
5188 return false (because the empty string is not a valid number):
5189
5190 int number;
5191 pcrecpp::RE::FullMatch("abc", "[a-z]+(\d+)?", &number);
5192
5193 The matching interface supports at most 16 arguments per call. If you
5194 need more, consider using the more general interface
5195 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5196
5197
5198 QUOTING METACHARACTERS
5199
5200 You can use the "QuoteMeta" operation to insert backslashes before all
5201 potentially meaningful characters in a string. The returned string,
5202 used as a regular expression, will exactly match the original string.
5203
5204 Example:
5205 string quoted = RE::QuoteMeta(unquoted);
5206
5207 Note that it's legal to escape a character even if it has no special
5208 meaning in a regular expression -- so this function does that. (This
5209 also makes it identical to the perl function of the same name; see
5210 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5211 "1\.5\-2\.0\?".
5212
5213
5214 PARTIAL MATCHES
5215
5216 You can use the "PartialMatch" operation when you want the pattern to
5217 match any substring of the text.
5218
5219 Example: simple search for a string:
5220 pcrecpp::RE("ell").PartialMatch("hello");
5221
5222 Example: find first number in a string:
5223 int number;
5224 pcrecpp::RE re("(\\d+)");
5225 re.PartialMatch("x*100 + 20", &number);
5226 assert(number == 100);
5227
5228
5229 UTF-8 AND THE MATCHING INTERFACE
5230
5231 By default, pattern and text are plain text, one byte per character.
5232 The UTF8 flag, passed to the constructor, causes both pattern and
5233 string to be treated as UTF-8 text, still a byte stream but potentially
5234 multiple bytes per character. In practice, the text is likelier to be
5235 UTF-8 than the pattern, but the match returned may depend on the UTF8
5236 flag, so always use it when matching UTF8 text. For example, "." will
5237 match one byte normally but with UTF8 set may match up to three bytes
5238 of a multi-byte character.
5239
5240 Example:
5241 pcrecpp::RE_Options options;
5242 options.set_utf8();
5243 pcrecpp::RE re(utf8_pattern, options);
5244 re.FullMatch(utf8_string);
5245
5246 Example: using the convenience function UTF8():
5247 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5248 re.FullMatch(utf8_string);
5249
5250 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5251 --enable-utf8 flag.
5252
5253
5254 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5255
5256 PCRE defines some modifiers to change the behavior of the regular
5257 expression engine. The C++ wrapper defines an auxiliary class,
5258 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5259 rently, the following modifiers are supported:
5260
5261 modifier description Perl corresponding
5262
5263 PCRE_CASELESS case insensitive match /i
5264 PCRE_MULTILINE multiple lines match /m
5265 PCRE_DOTALL dot matches newlines /s
5266 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5267 PCRE_EXTRA strict escape parsing N/A
5268 PCRE_EXTENDED ignore whitespaces /x
5269 PCRE_UTF8 handles UTF8 chars built-in
5270 PCRE_UNGREEDY reverses * and *? N/A
5271 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5272
5273 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5274 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5275 ture, while (ab|cd) does.
5276
5277 For a full account on how each modifier works, please check the PCRE
5278 API reference page.
5279
5280 For each modifier, there are two member functions whose name is made
5281 out of the modifier in lowercase, without the "PCRE_" prefix. For
5282 instance, PCRE_CASELESS is handled by
5283
5284 bool caseless()
5285
5286 which returns true if the modifier is set, and
5287
5288 RE_Options & set_caseless(bool)
5289
5290 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5291 be accessed through the set_match_limit() and match_limit() member
5292 functions. Setting match_limit to a non-zero value will limit the exe-
5293 cution of pcre to keep it from doing bad things like blowing the stack
5294 or taking an eternity to return a result. A value of 5000 is good
5295 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5296 to zero disables match limiting. Alternatively, you can call
5297 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5298 limit how much PCRE recurses. match_limit() limits the number of
5299 matches PCRE does; match_limit_recursion() limits the depth of internal
5300 recursion, and therefore the amount of stack that is used.
5301
5302 Normally, to pass one or more modifiers to a RE class, you declare a
5303 RE_Options object, set the appropriate options, and pass this object to
5304 a RE constructor. Example:
5305
5306 RE_options opt;
5307 opt.set_caseless(true);
5308 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5309
5310 RE_options has two constructors. The default constructor takes no argu-
5311 ments and creates a set of flags that are off by default. The optional
5312 parameter option_flags is to facilitate transfer of legacy code from C
5313 programs. This lets you do
5314
5315 RE(pattern,
5316 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5317
5318 However, new code is better off doing
5319
5320 RE(pattern,
5321 RE_Options().set_caseless(true).set_multiline(true))
5322 .PartialMatch(str);
5323
5324 If you are going to pass one of the most used modifiers, there are some
5325 convenience functions that return a RE_Options class with the appropri-
5326 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5327 and EXTENDED().
5328
5329 If you need to set several options at once, and you don't want to go
5330 through the pains of declaring a RE_Options object and setting several
5331 options, there is a parallel method that give you such ability on the
5332 fly. You can concatenate several set_xxxxx() member functions, since
5333 each of them returns a reference to its class object. For example, to
5334 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5335 statement, you may write:
5336
5337 RE(" ^ xyz \\s+ .* blah$",
5338 RE_Options()
5339 .set_caseless(true)
5340 .set_extended(true)
5341 .set_multiline(true)).PartialMatch(sometext);
5342
5343
5344 SCANNING TEXT INCREMENTALLY
5345
5346 The "Consume" operation may be useful if you want to repeatedly match
5347 regular expressions at the front of a string and skip over them as they
5348 match. This requires use of the "StringPiece" type, which represents a
5349 sub-range of a real string. Like RE, StringPiece is defined in the
5350 pcrecpp namespace.
5351
5352 Example: read lines of the form "var = value" from a string.
5353 string contents = ...; // Fill string somehow
5354 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
5355
5356 string var;
5357 int value;
5358 pcrecpp::RE re("(\\w+) = (\\d+)\n");
5359 while (re.Consume(&input, &var, &value)) {
5360 ...;
5361 }
5362
5363 Each successful call to "Consume" will set "var/value", and also
5364 advance "input" so it points past the matched text.
5365
5366 The "FindAndConsume" operation is similar to "Consume" but does not
5367 anchor your match at the beginning of the string. For example, you
5368 could extract all words from a string by repeatedly calling
5369
5370 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
5371
5372
5373 PARSING HEX/OCTAL/C-RADIX NUMBERS
5374
5375 By default, if you pass a pointer to a numeric value, the corresponding
5376 text is interpreted as a base-10 number. You can instead wrap the
5377 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
5378 to interpret the text in another base. The CRadix operator interprets
5379 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
5380 base-10.
5381
5382 Example:
5383 int a, b, c, d;
5384 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
5385 re.FullMatch("100 40 0100 0x40",
5386 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
5387 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
5388
5389 will leave 64 in a, b, c, and d.
5390
5391
5392 REPLACING PARTS OF STRINGS
5393
5394 You can replace the first match of "pattern" in "str" with "rewrite".
5395 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
5396 insert text matching corresponding parenthesized group from the pat-
5397 tern. \0 in "rewrite" refers to the entire matching text. For example:
5398
5399 string s = "yabba dabba doo";
5400 pcrecpp::RE("b+").Replace("d", &s);
5401
5402 will leave "s" containing "yada dabba doo". The result is true if the
5403 pattern matches and a replacement occurs, false otherwise.
5404
5405 GlobalReplace is like Replace except that it replaces all occurrences
5406 of the pattern in the string with the rewrite. Replacements are not
5407 subject to re-matching. For example:
5408
5409 string s = "yabba dabba doo";
5410 pcrecpp::RE("b+").GlobalReplace("d", &s);
5411
5412 will leave "s" containing "yada dada doo". It returns the number of
5413 replacements made.
5414
5415 Extract is like Replace, except that if the pattern matches, "rewrite"
5416 is copied into "out" (an additional argument) with substitutions. The
5417 non-matching portions of "text" are ignored. Returns true iff a match
5418 occurred and the extraction happened successfully; if no match occurs,
5419 the string is left unaffected.
5420
5421
5422 AUTHOR
5423
5424 The C++ wrapper was contributed by Google Inc.
5425 Copyright (c) 2007 Google Inc.
5426
5427
5428 REVISION
5429
5430 Last updated: 06 March 2007
5431 ------------------------------------------------------------------------------
5432
5433
5434 PCRESAMPLE(3) PCRESAMPLE(3)
5435
5436
5437 NAME
5438 PCRE - Perl-compatible regular expressions
5439
5440
5441 PCRE SAMPLE PROGRAM
5442
5443 A simple, complete demonstration program, to get you started with using
5444 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
5445
5446 The program compiles the regular expression that is its first argument,
5447 and matches it against the subject string in its second argument. No
5448 PCRE options are set, and default character tables are used. If match-
5449 ing succeeds, the program outputs the portion of the subject that
5450 matched, together with the contents of any captured substrings.
5451
5452 If the -g option is given on the command line, the program then goes on
5453 to check for further matches of the same regular expression in the same
5454 subject string. The logic is a little bit tricky because of the possi-
5455 bility of matching an empty string. Comments in the code explain what
5456 is going on.
5457
5458 If PCRE is installed in the standard include and library directories
5459 for your system, you should be able to compile the demonstration pro-
5460 gram using this command:
5461
5462 gcc -o pcredemo pcredemo.c -lpcre
5463
5464 If PCRE is installed elsewhere, you may need to add additional options
5465 to the command line. For example, on a Unix-like system that has PCRE
5466 installed in /usr/local, you can compile the demonstration program
5467 using a command like this:
5468
5469 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5470 -L/usr/local/lib -lpcre
5471
5472 Once you have compiled the demonstration program, you can run simple
5473 tests like this:
5474
5475 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5476 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5477
5478 Note that there is a much more comprehensive test program, called
5479 pcretest, which supports many more facilities for testing regular
5480 expressions and the PCRE library. The pcredemo program is provided as a
5481 simple coding example.
5482
5483 On some operating systems (e.g. Solaris), when PCRE is not installed in
5484 the standard library directory, you may get an error like this when you
5485 try to run pcredemo:
5486
5487 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5488 directory
5489
5490 This is caused by the way shared library support works on those sys-
5491 tems. You need to add
5492
5493 -R/usr/local/lib
5494
5495 (for example) to the compile command to get round this problem.
5496
5497
5498 AUTHOR
5499
5500 Philip Hazel
5501 University Computing Service
5502 Cambridge CB2 3QH, England.
5503
5504
5505 REVISION
5506
5507 Last updated: 06 March 2007
5508 Copyright (c) 1997-2007 University of Cambridge.
5509 ------------------------------------------------------------------------------
5510 PCRESTACK(3) PCRESTACK(3)
5511
5512
5513 NAME
5514 PCRE - Perl-compatible regular expressions
5515
5516
5517 PCRE DISCUSSION OF STACK USAGE
5518
5519 When you call pcre_exec(), it makes use of an internal function called
5520 match(). This calls itself recursively at branch points in the pattern,
5521 in order to remember the state of the match so that it can back up and
5522 try a different alternative if the first one fails. As matching pro-
5523 ceeds deeper and deeper into the tree of possibilities, the recursion
5524 depth increases.
5525
5526 Not all calls of match() increase the recursion depth; for an item such
5527 as a* it may be called several times at the same level, after matching
5528 different numbers of a's. Furthermore, in a number of cases where the
5529 result of the recursive call would immediately be passed back as the
5530 result of the current call (a "tail recursion"), the function is just
5531 restarted instead.
5532
5533 The pcre_dfa_exec() function operates in an entirely different way, and
5534 hardly uses recursion at all. The limit on its complexity is the amount
5535 of workspace it is given. The comments that follow do NOT apply to
5536 pcre_dfa_exec(); they are relevant only for pcre_exec().
5537
5538 You can set limits on the number of times that match() is called, both
5539 in total and recursively. If the limit is exceeded, an error occurs.
5540 For details, see the section on extra data for pcre_exec() in the
5541 pcreapi documentation.
5542
5543 Each time that match() is actually called recursively, it uses memory
5544 from the process stack. For certain kinds of pattern and data, very
5545 large amounts of stack may be needed, despite the recognition of "tail
5546 recursion". You can often reduce the amount of recursion, and there-
5547 fore the amount of stack used, by modifying the pattern that is being
5548 matched. Consider, for example, this pattern:
5549
5550 ([^<]|<(?!inet))+
5551
5552 It matches from wherever it starts until it encounters "<inet" or the
5553 end of the data, and is the kind of pattern that might be used when
5554 processing an XML file. Each iteration of the outer parentheses matches
5555 either one character that is not "<" or a "<" that is not followed by
5556 "inet". However, each time a parenthesis is processed, a recursion
5557 occurs, so this formulation uses a stack frame for each matched charac-
5558 ter. For a long string, a lot of stack is required. Consider now this
5559 rewritten pattern, which matches exactly the same strings:
5560
5561 ([^<]++|<(?!inet))+
5562
5563 This uses very much less stack, because runs of characters that do not
5564 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5565 sion happens only when a "<" character that is not followed by "inet"
5566 is encountered (and we assume this is relatively rare). A possessive
5567 quantifier is used to stop any backtracking into the runs of non-"<"
5568 characters, but that is not related to stack usage.
5569
5570 This example shows that one way of avoiding stack problems when match-
5571 ing long subject strings is to write repeated parenthesized subpatterns
5572 to match more than one character whenever possible.
5573
5574 In environments where stack memory is constrained, you might want to
5575 compile PCRE to use heap memory instead of stack for remembering back-
5576 up points. This makes it run a lot more slowly, however. Details of how
5577 to do this are given in the pcrebuild documentation.
5578
5579 In Unix-like environments, there is not often a problem with the stack
5580 unless very long strings are involved, though the default limit on
5581 stack size varies from system to system. Values from 8Mb to 64Mb are
5582 common. You can find your default limit by running the command:
5583
5584 ulimit -s
5585
5586 Unfortunately, the effect of running out of stack is often SIGSEGV,
5587 though sometimes a more explicit error message is given. You can nor-
5588 mally increase the limit on stack size by code such as this:
5589
5590 struct rlimit rlim;
5591 getrlimit(RLIMIT_STACK, &rlim);
5592 rlim.rlim_cur = 100*1024*1024;
5593 setrlimit(RLIMIT_STACK, &rlim);
5594
5595 This reads the current limits (soft and hard) using getrlimit(), then
5596 attempts to increase the soft limit to 100Mb using setrlimit(). You
5597 must do this before calling pcre_exec().
5598
5599 PCRE has an internal counter that can be used to limit the depth of
5600 recursion, and thus cause pcre_exec() to give an error code before it
5601 runs out of stack. By default, the limit is very large, and unlikely
5602 ever to operate. It can be changed when PCRE is built, and it can also
5603 be set when pcre_exec() is called. For details of these interfaces, see
5604 the pcrebuild and pcreapi documentation.
5605
5606 As a very rough rule of thumb, you should reckon on about 500 bytes per
5607 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5608 should set the limit at 16000 recursions. A 64Mb stack, on the other
5609 hand, can support around 128000 recursions. The pcretest test program
5610 has a command line option (-S) that can be used to increase the size of
5611 its stack.
5612
5613
5614 AUTHOR
5615
5616 Philip Hazel
5617 University Computing Service
5618 Cambridge CB2 3QH, England.
5619
5620
5621 REVISION
5622
5623 Last updated: 12 March 2007
5624 Copyright (c) 1997-2007 University of Cambridge.
5625 ------------------------------------------------------------------------------
5626
5627

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12