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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 111 - (show annotations) (download)
Thu Mar 8 16:53:09 2007 UTC (7 years, 9 months ago) by ph10
File MIME type: text/plain
File size: 264910 byte(s)
Create the PrepareRelease script to process the documentation and create the 
.generic files for distribution, also to remove trailing spaces. Update a lot 
more of the build-time documentation. Arrange for PrepareRelease and its 
sub-scripts to be distributed.

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

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