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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 153 - (show annotations)
Wed Apr 18 09:12:14 2007 UTC (12 years, 5 months ago) by ph10
File MIME type: text/plain
File size: 266823 byte(s)
Added a man page for pcre-config.
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 pcre-config show PCRE installation configuration information
76 pcreapi details of PCRE's native C API
77 pcrebuild options for building PCRE
78 pcrecallout details of the callout feature
79 pcrecompat discussion of Perl compatibility
80 pcrecpp details of the C++ wrapper
81 pcregrep description of the pcregrep command
82 pcrematching discussion of the two matching algorithms
83 pcrepartial details of the partial matching facility
84 pcrepattern syntax and semantics of supported
85 regular expressions
86 pcreperform discussion of performance issues
87 pcreposix the POSIX-compatible C API
88 pcreprecompile details of saving and re-using precompiled patterns
89 pcresample discussion of the sample program
90 pcrestack discussion of stack usage
91 pcretest description of the pcretest testing command
92
93 In addition, in the "man" and HTML formats, there is a short page for
94 each C library function, listing its arguments and results.
95
96
97 LIMITATIONS
98
99 There are some size limitations in PCRE but it is hoped that they will
100 never in practice be relevant.
101
102 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
103 is compiled with the default internal linkage size of 2. If you want to
104 process regular expressions that are truly enormous, you can compile
105 PCRE with an internal linkage size of 3 or 4 (see the README file in
106 the source distribution and the pcrebuild documentation for details).
107 In these cases the limit is substantially larger. However, the speed
108 of execution is slower.
109
110 All values in repeating quantifiers must be less than 65536. The maxi-
111 mum compiled length of subpattern with an explicit repeat count is
112 30000 bytes. The maximum number of capturing subpatterns is 65535.
113
114 There is no limit to the number of parenthesized subpatterns, but there
115 can be no more than 65535 capturing subpatterns.
116
117 The maximum length of name for a named subpattern is 32 characters, and
118 the maximum number of named subpatterns is 10000.
119
120 The maximum length of a subject string is the largest positive number
121 that an integer variable can hold. However, when using the traditional
122 matching function, PCRE uses recursion to handle subpatterns and indef-
123 inite repetition. This means that the available stack space may limit
124 the size of a subject string that can be processed by certain patterns.
125 For a discussion of stack issues, see the pcrestack documentation.
126
127
128 UTF-8 AND UNICODE PROPERTY SUPPORT
129
130 From release 3.3, PCRE has had some support for character strings
131 encoded in the UTF-8 format. For release 4.0 this was greatly extended
132 to cover most common requirements, and in release 5.0 additional sup-
133 port for Unicode general category properties was added.
134
135 In order process UTF-8 strings, you must build PCRE to include UTF-8
136 support in the code, and, in addition, you must call pcre_compile()
137 with the PCRE_UTF8 option flag. When you do this, both the pattern and
138 any subject strings that are matched against it are treated as UTF-8
139 strings instead of just strings of bytes.
140
141 If you compile PCRE with UTF-8 support, but do not use it at run time,
142 the library will be a bit bigger, but the additional run time overhead
143 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
144 very big.
145
146 If PCRE is built with Unicode character property support (which implies
147 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
148 ported. The available properties that can be tested are limited to the
149 general category properties such as Lu for an upper case letter or Nd
150 for a decimal number, the Unicode script names such as Arabic or Han,
151 and the derived properties Any and L&. A full list is given in the
152 pcrepattern documentation. Only the short names for properties are sup-
153 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
154 ter}, is not supported. Furthermore, in Perl, many properties may
155 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
156 does not support this.
157
158 The following comments apply when PCRE is running in UTF-8 mode:
159
160 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
161 subjects are checked for validity on entry to the relevant functions.
162 If an invalid UTF-8 string is passed, an error return is given. In some
163 situations, you may already know that your strings are valid, and
164 therefore want to skip these checks in order to improve performance. If
165 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
166 PCRE assumes that the pattern or subject it is given (respectively)
167 contains only valid UTF-8 codes. In this case, it does not diagnose an
168 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
169 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
170 crash.
171
172 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
173 two-byte UTF-8 character if the value is greater than 127.
174
175 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
176 characters for values greater than \177.
177
178 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
179 vidual bytes, for example: \x{100}{3}.
180
181 5. The dot metacharacter matches one UTF-8 character instead of a sin-
182 gle byte.
183
184 6. The escape sequence \C can be used to match a single byte in UTF-8
185 mode, but its use can lead to some strange effects. This facility is
186 not available in the alternative matching function, pcre_dfa_exec().
187
188 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
189 test characters of any code value, but the characters that PCRE recog-
190 nizes as digits, spaces, or word characters remain the same set as
191 before, all with values less than 256. This remains true even when PCRE
192 includes Unicode property support, because to do otherwise would slow
193 down PCRE in many common cases. If you really want to test for a wider
194 sense of, say, "digit", you must use Unicode property tests such as
195 \p{Nd}.
196
197 8. Similarly, characters that match the POSIX named character classes
198 are all low-valued characters.
199
200 9. Case-insensitive matching applies only to characters whose values
201 are less than 128, unless PCRE is built with Unicode property support.
202 Even when Unicode property support is available, PCRE still uses its
203 own character tables when checking the case of low-valued characters,
204 so as not to degrade performance. The Unicode property information is
205 used only for characters with higher values. Even when Unicode property
206 support is available, PCRE supports case-insensitive matching only when
207 there is a one-to-one mapping between a letter's cases. There are a
208 small number of many-to-one mappings in Unicode; these are not sup-
209 ported by PCRE.
210
211
212 AUTHOR
213
214 Philip Hazel
215 University Computing Service
216 Cambridge CB2 3QH, England.
217
218 Putting an actual email address here seems to have been a spam magnet,
219 so I've taken it away. If you want to email me, use my two initials,
220 followed by the two digits 10, at the domain cam.ac.uk.
221
222
223 REVISION
224
225 Last updated: 18 April 2007
226 Copyright (c) 1997-2007 University of Cambridge.
227 ------------------------------------------------------------------------------
228
229
230 PCREBUILD(3) PCREBUILD(3)
231
232
233 NAME
234 PCRE - Perl-compatible regular expressions
235
236
237 PCRE BUILD-TIME OPTIONS
238
239 This document describes the optional features of PCRE that can be
240 selected when the library is compiled. They are all selected, or dese-
241 lected, by providing options to the configure script that is run before
242 the make command. The complete list of options for configure (which
243 includes the standard ones such as the selection of the installation
244 directory) can be obtained by running
245
246 ./configure --help
247
248 The following sections include descriptions of options whose names
249 begin with --enable or --disable. These settings specify changes to the
250 defaults for the configure command. Because of the way that configure
251 works, --enable and --disable always come in pairs, so the complemen-
252 tary option always exists as well, but as it specifies the default, it
253 is not described.
254
255
256 C++ SUPPORT
257
258 By default, the configure script will search for a C++ compiler and C++
259 header files. If it finds them, it automatically builds the C++ wrapper
260 library for PCRE. You can disable this by adding
261
262 --disable-cpp
263
264 to the configure command.
265
266
267 UTF-8 SUPPORT
268
269 To build PCRE with support for UTF-8 character strings, add
270
271 --enable-utf8
272
273 to the configure command. Of itself, this does not make PCRE treat
274 strings as UTF-8. As well as compiling PCRE with this option, you also
275 have have to set the PCRE_UTF8 option when you call the pcre_compile()
276 function.
277
278
279 UNICODE CHARACTER PROPERTY SUPPORT
280
281 UTF-8 support allows PCRE to process character values greater than 255
282 in the strings that it handles. On its own, however, it does not pro-
283 vide any facilities for accessing the properties of such characters. If
284 you want to be able to use the pattern escapes \P, \p, and \X, which
285 refer to Unicode character properties, you must add
286
287 --enable-unicode-properties
288
289 to the configure command. This implies UTF-8 support, even if you have
290 not explicitly requested it.
291
292 Including Unicode property support adds around 30K of tables to the
293 PCRE library. Only the general category properties such as Lu and Nd
294 are supported. Details are given in 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-anycrlf
317
318 which causes PCRE to recognize any of the three sequences CR, LF, or
319 CRLF as indicating a line ending. Finally, a fifth option, specified by
320
321 --enable-newline-is-any
322
323 causes PCRE to recognize any Unicode newline sequence.
324
325 Whatever line ending convention is selected when PCRE is built can be
326 overridden when the library functions are called. At build time it is
327 conventional to use the standard for your operating system.
328
329
330 BUILDING SHARED AND STATIC LIBRARIES
331
332 The PCRE building process uses libtool to build both shared and static
333 Unix libraries by default. You can suppress one of these by adding one
334 of
335
336 --disable-shared
337 --disable-static
338
339 to the configure command, as required.
340
341
342 POSIX MALLOC USAGE
343
344 When PCRE is called through the POSIX interface (see the pcreposix doc-
345 umentation), additional working storage is required for holding the
346 pointers to capturing substrings, because PCRE requires three integers
347 per substring, whereas the POSIX interface provides only two. If the
348 number of expected substrings is small, the wrapper function uses space
349 on the stack, because this is faster than using malloc() for each call.
350 The default threshold above which the stack is no longer used is 10; it
351 can be changed by adding a setting such as
352
353 --with-posix-malloc-threshold=20
354
355 to the configure command.
356
357
358 HANDLING VERY LARGE PATTERNS
359
360 Within a compiled pattern, offset values are used to point from one
361 part to another (for example, from an opening parenthesis to an alter-
362 nation metacharacter). By default, two-byte values are used for these
363 offsets, leading to a maximum size for a compiled pattern of around
364 64K. This is sufficient to handle all but the most gigantic patterns.
365 Nevertheless, some people do want to process enormous patterns, so it
366 is possible to compile PCRE to use three-byte or four-byte offsets by
367 adding a setting such as
368
369 --with-link-size=3
370
371 to the configure command. The value given must be 2, 3, or 4. Using
372 longer offsets slows down the operation of PCRE because it has to load
373 additional bytes when handling them.
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 CREATING CHARACTER TABLES AT BUILD TIME
433
434 PCRE uses fixed tables for processing characters whose code values are
435 less than 256. By default, PCRE is built with a set of tables that are
436 distributed in the file pcre_chartables.c.dist. These tables are for
437 ASCII codes only. If you add
438
439 --enable-rebuild-chartables
440
441 to the configure command, the distributed tables are no longer used.
442 Instead, a program called dftables is compiled and run. This outputs
443 the source for new set of tables, created in the default locale of your
444 C runtime system. (This method of replacing the tables does not work if
445 you are cross compiling, because dftables is run on the local host. If
446 you need to create alternative tables when cross compiling, you will
447 have to do so "by hand".)
448
449
450 USING EBCDIC CODE
451
452 PCRE assumes by default that it will run in an environment where the
453 character code is ASCII (or Unicode, which is a superset of ASCII).
454 PCRE can, however, be compiled to run in an EBCDIC environment by
455 adding
456
457 --enable-ebcdic
458
459 to the configure command. This setting implies --enable-rebuild-charta-
460 bles.
461
462
463 SEE ALSO
464
465 pcreapi(3), pcre_config(3).
466
467
468 AUTHOR
469
470 Philip Hazel
471 University Computing Service
472 Cambridge CB2 3QH, England.
473
474
475 REVISION
476
477 Last updated: 16 April 2007
478 Copyright (c) 1997-2007 University of Cambridge.
479 ------------------------------------------------------------------------------
480
481
482 PCREMATCHING(3) PCREMATCHING(3)
483
484
485 NAME
486 PCRE - Perl-compatible regular expressions
487
488
489 PCRE MATCHING ALGORITHMS
490
491 This document describes the two different algorithms that are available
492 in PCRE for matching a compiled regular expression against a given sub-
493 ject string. The "standard" algorithm is the one provided by the
494 pcre_exec() function. This works in the same was as Perl's matching
495 function, and provides a Perl-compatible matching operation.
496
497 An alternative algorithm is provided by the pcre_dfa_exec() function;
498 this operates in a different way, and is not Perl-compatible. It has
499 advantages and disadvantages compared with the standard algorithm, and
500 these are described below.
501
502 When there is only one possible way in which a given subject string can
503 match a pattern, the two algorithms give the same answer. A difference
504 arises, however, when there are multiple possibilities. For example, if
505 the pattern
506
507 ^<.*>
508
509 is matched against the string
510
511 <something> <something else> <something further>
512
513 there are three possible answers. The standard algorithm finds only one
514 of them, whereas the alternative algorithm finds all three.
515
516
517 REGULAR EXPRESSIONS AS TREES
518
519 The set of strings that are matched by a regular expression can be rep-
520 resented as a tree structure. An unlimited repetition in the pattern
521 makes the tree of infinite size, but it is still a tree. Matching the
522 pattern to a given subject string (from a given starting point) can be
523 thought of as a search of the tree. There are two ways to search a
524 tree: depth-first and breadth-first, and these correspond to the two
525 matching algorithms provided by PCRE.
526
527
528 THE STANDARD MATCHING ALGORITHM
529
530 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
531 sions", the standard algorithm is an "NFA algorithm". It conducts a
532 depth-first search of the pattern tree. That is, it proceeds along a
533 single path through the tree, checking that the subject matches what is
534 required. When there is a mismatch, the algorithm tries any alterna-
535 tives at the current point, and if they all fail, it backs up to the
536 previous branch point in the tree, and tries the next alternative
537 branch at that level. This often involves backing up (moving to the
538 left) in the subject string as well. The order in which repetition
539 branches are tried is controlled by the greedy or ungreedy nature of
540 the quantifier.
541
542 If a leaf node is reached, a matching string has been found, and at
543 that point the algorithm stops. Thus, if there is more than one possi-
544 ble match, this algorithm returns the first one that it finds. Whether
545 this is the shortest, the longest, or some intermediate length depends
546 on the way the greedy and ungreedy repetition quantifiers are specified
547 in the pattern.
548
549 Because it ends up with a single path through the tree, it is rela-
550 tively straightforward for this algorithm to keep track of the sub-
551 strings that are matched by portions of the pattern in parentheses.
552 This provides support for capturing parentheses and back references.
553
554
555 THE ALTERNATIVE MATCHING ALGORITHM
556
557 This algorithm conducts a breadth-first search of the tree. Starting
558 from the first matching point in the subject, it scans the subject
559 string from left to right, once, character by character, and as it does
560 this, it remembers all the paths through the tree that represent valid
561 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
562 though it is not implemented as a traditional finite state machine (it
563 keeps multiple states active simultaneously).
564
565 The scan continues until either the end of the subject is reached, or
566 there are no more unterminated paths. At this point, terminated paths
567 represent the different matching possibilities (if there are none, the
568 match has failed). Thus, if there is more than one possible match,
569 this algorithm finds all of them, and in particular, it finds the long-
570 est. In PCRE, there is an option to stop the algorithm after the first
571 match (which is necessarily the shortest) has been found.
572
573 Note that all the matches that are found start at the same point in the
574 subject. If the pattern
575
576 cat(er(pillar)?)
577
578 is matched against the string "the caterpillar catchment", the result
579 will be the three strings "cat", "cater", and "caterpillar" that start
580 at the fourth character of the subject. The algorithm does not automat-
581 ically move on to find matches that start at later positions.
582
583 There are a number of features of PCRE regular expressions that are not
584 supported by the alternative matching algorithm. They are as follows:
585
586 1. Because the algorithm finds all possible matches, the greedy or
587 ungreedy nature of repetition quantifiers is not relevant. Greedy and
588 ungreedy quantifiers are treated in exactly the same way. However, pos-
589 sessive quantifiers can make a difference when what follows could also
590 match what is quantified, for example in a pattern like this:
591
592 ^a++\w!
593
594 This pattern matches "aaab!" but not "aaa!", which would be matched by
595 a non-possessive quantifier. Similarly, if an atomic group is present,
596 it is matched as if it were a standalone pattern at the current point,
597 and the longest match is then "locked in" for the rest of the overall
598 pattern.
599
600 2. When dealing with multiple paths through the tree simultaneously, it
601 is not straightforward to keep track of captured substrings for the
602 different matching possibilities, and PCRE's implementation of this
603 algorithm does not attempt to do this. This means that no captured sub-
604 strings are available.
605
606 3. Because no substrings are captured, back references within the pat-
607 tern are not supported, and cause errors if encountered.
608
609 4. For the same reason, conditional expressions that use a backrefer-
610 ence as the condition or test for a specific group recursion are not
611 supported.
612
613 5. Callouts are supported, but the value of the capture_top field is
614 always 1, and the value of the capture_last field is always -1.
615
616 6. The \C escape sequence, which (in the standard algorithm) matches a
617 single byte, even in UTF-8 mode, is not supported because the alterna-
618 tive algorithm moves through the subject string one character at a
619 time, for all active paths through the tree.
620
621
622 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
623
624 Using the alternative matching algorithm provides the following advan-
625 tages:
626
627 1. All possible matches (at a single point in the subject) are automat-
628 ically found, and in particular, the longest match is found. To find
629 more than one match using the standard algorithm, you have to do kludgy
630 things with callouts.
631
632 2. There is much better support for partial matching. The restrictions
633 on the content of the pattern that apply when using the standard algo-
634 rithm for partial matching do not apply to the alternative algorithm.
635 For non-anchored patterns, the starting position of a partial match is
636 available.
637
638 3. Because the alternative algorithm scans the subject string just
639 once, and never needs to backtrack, it is possible to pass very long
640 subject strings to the matching function in several pieces, checking
641 for partial matching each time.
642
643
644 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
645
646 The alternative algorithm suffers from a number of disadvantages:
647
648 1. It is substantially slower than the standard algorithm. This is
649 partly because it has to search for all possible matches, but is also
650 because it is less susceptible to optimization.
651
652 2. Capturing parentheses and back references are not supported.
653
654 3. Although atomic groups are supported, their use does not provide the
655 performance advantage that it does for the standard algorithm.
656
657
658 AUTHOR
659
660 Philip Hazel
661 University Computing Service
662 Cambridge CB2 3QH, England.
663
664
665 REVISION
666
667 Last updated: 06 March 2007
668 Copyright (c) 1997-2007 University of Cambridge.
669 ------------------------------------------------------------------------------
670
671
672 PCREAPI(3) PCREAPI(3)
673
674
675 NAME
676 PCRE - Perl-compatible regular expressions
677
678
679 PCRE NATIVE API
680
681 #include <pcre.h>
682
683 pcre *pcre_compile(const char *pattern, int options,
684 const char **errptr, int *erroffset,
685 const unsigned char *tableptr);
686
687 pcre *pcre_compile2(const char *pattern, int options,
688 int *errorcodeptr,
689 const char **errptr, int *erroffset,
690 const unsigned char *tableptr);
691
692 pcre_extra *pcre_study(const pcre *code, int options,
693 const char **errptr);
694
695 int pcre_exec(const pcre *code, const pcre_extra *extra,
696 const char *subject, int length, int startoffset,
697 int options, int *ovector, int ovecsize);
698
699 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
700 const char *subject, int length, int startoffset,
701 int options, int *ovector, int ovecsize,
702 int *workspace, int wscount);
703
704 int pcre_copy_named_substring(const pcre *code,
705 const char *subject, int *ovector,
706 int stringcount, const char *stringname,
707 char *buffer, int buffersize);
708
709 int pcre_copy_substring(const char *subject, int *ovector,
710 int stringcount, int stringnumber, char *buffer,
711 int buffersize);
712
713 int pcre_get_named_substring(const pcre *code,
714 const char *subject, int *ovector,
715 int stringcount, const char *stringname,
716 const char **stringptr);
717
718 int pcre_get_stringnumber(const pcre *code,
719 const char *name);
720
721 int pcre_get_stringtable_entries(const pcre *code,
722 const char *name, char **first, char **last);
723
724 int pcre_get_substring(const char *subject, int *ovector,
725 int stringcount, int stringnumber,
726 const char **stringptr);
727
728 int pcre_get_substring_list(const char *subject,
729 int *ovector, int stringcount, const char ***listptr);
730
731 void pcre_free_substring(const char *stringptr);
732
733 void pcre_free_substring_list(const char **stringptr);
734
735 const unsigned char *pcre_maketables(void);
736
737 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
738 int what, void *where);
739
740 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
741
742 int pcre_refcount(pcre *code, int adjust);
743
744 int pcre_config(int what, void *where);
745
746 char *pcre_version(void);
747
748 void *(*pcre_malloc)(size_t);
749
750 void (*pcre_free)(void *);
751
752 void *(*pcre_stack_malloc)(size_t);
753
754 void (*pcre_stack_free)(void *);
755
756 int (*pcre_callout)(pcre_callout_block *);
757
758
759 PCRE API OVERVIEW
760
761 PCRE has its own native API, which is described in this document. There
762 are also some wrapper functions that correspond to the POSIX regular
763 expression API. These are described in the pcreposix documentation.
764 Both of these APIs define a set of C function calls. A C++ wrapper is
765 distributed with PCRE. It is documented in the pcrecpp page.
766
767 The native API C function prototypes are defined in the header file
768 pcre.h, and on Unix systems the library itself is called libpcre. It
769 can normally be accessed by adding -lpcre to the command for linking an
770 application that uses PCRE. The header file defines the macros
771 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
772 bers for the library. Applications can use these to include support
773 for different releases of PCRE.
774
775 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
776 pcre_exec() are used for compiling and matching regular expressions in
777 a Perl-compatible manner. A sample program that demonstrates the sim-
778 plest way of using them is provided in the file called pcredemo.c in
779 the source distribution. The pcresample documentation describes how to
780 run it.
781
782 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
783 ble, is also provided. This uses a different algorithm for the match-
784 ing. The alternative algorithm finds all possible matches (at a given
785 point in the subject), and scans the subject just once. However, this
786 algorithm does not return captured substrings. A description of the two
787 matching algorithms and their advantages and disadvantages is given in
788 the pcrematching documentation.
789
790 In addition to the main compiling and matching functions, there are
791 convenience functions for extracting captured substrings from a subject
792 string that is matched by pcre_exec(). They are:
793
794 pcre_copy_substring()
795 pcre_copy_named_substring()
796 pcre_get_substring()
797 pcre_get_named_substring()
798 pcre_get_substring_list()
799 pcre_get_stringnumber()
800 pcre_get_stringtable_entries()
801
802 pcre_free_substring() and pcre_free_substring_list() are also provided,
803 to free the memory used for extracted strings.
804
805 The function pcre_maketables() is used to build a set of character
806 tables in the current locale for passing to pcre_compile(),
807 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
808 provided for specialist use. Most commonly, no special tables are
809 passed, in which case internal tables that are generated when PCRE is
810 built are used.
811
812 The function pcre_fullinfo() is used to find out information about a
813 compiled pattern; pcre_info() is an obsolete version that returns only
814 some of the available information, but is retained for backwards com-
815 patibility. The function pcre_version() returns a pointer to a string
816 containing the version of PCRE and its date of release.
817
818 The function pcre_refcount() maintains a reference count in a data
819 block containing a compiled pattern. This is provided for the benefit
820 of object-oriented applications.
821
822 The global variables pcre_malloc and pcre_free initially contain the
823 entry points of the standard malloc() and free() functions, respec-
824 tively. PCRE calls the memory management functions via these variables,
825 so a calling program can replace them if it wishes to intercept the
826 calls. This should be done before calling any PCRE functions.
827
828 The global variables pcre_stack_malloc and pcre_stack_free are also
829 indirections to memory management functions. These special functions
830 are used only when PCRE is compiled to use the heap for remembering
831 data, instead of recursive function calls, when running the pcre_exec()
832 function. See the pcrebuild documentation for details of how to do
833 this. It is a non-standard way of building PCRE, for use in environ-
834 ments that have limited stacks. Because of the greater use of memory
835 management, it runs more slowly. Separate functions are provided so
836 that special-purpose external code can be used for this case. When
837 used, these functions are always called in a stack-like manner (last
838 obtained, first freed), and always for memory blocks of the same size.
839 There is a discussion about PCRE's stack usage in the pcrestack docu-
840 mentation.
841
842 The global variable pcre_callout initially contains NULL. It can be set
843 by the caller to a "callout" function, which PCRE will then call at
844 specified points during a matching operation. Details are given in the
845 pcrecallout documentation.
846
847
848 NEWLINES
849
850 PCRE supports five different conventions for indicating line breaks in
851 strings: a single CR (carriage return) character, a single LF (line-
852 feed) character, the two-character sequence CRLF, any of the three pre-
853 ceding, or any Unicode newline sequence. The Unicode newline sequences
854 are the three just mentioned, plus the single characters VT (vertical
855 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
856 separator, U+2028), and PS (paragraph separator, U+2029).
857
858 Each of the first three conventions is used by at least one operating
859 system as its standard newline sequence. When PCRE is built, a default
860 can be specified. The default default is LF, which is the Unix stan-
861 dard. When PCRE is run, the default can be overridden, either when a
862 pattern is compiled, or when it is matched.
863
864 In the PCRE documentation the word "newline" is used to mean "the char-
865 acter or pair of characters that indicate a line break". The choice of
866 newline convention affects the handling of the dot, circumflex, and
867 dollar metacharacters, the handling of #-comments in /x mode, and, when
868 CRLF is a recognized line ending sequence, the match position advance-
869 ment for a non-anchored pattern. The choice of newline convention does
870 not affect the interpretation of the \n or \r escape sequences.
871
872
873 MULTITHREADING
874
875 The PCRE functions can be used in multi-threading applications, with
876 the proviso that the memory management functions pointed to by
877 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
878 callout function pointed to by pcre_callout, are shared by all threads.
879
880 The compiled form of a regular expression is not altered during match-
881 ing, so the same compiled pattern can safely be used by several threads
882 at once.
883
884
885 SAVING PRECOMPILED PATTERNS FOR LATER USE
886
887 The compiled form of a regular expression can be saved and re-used at a
888 later time, possibly by a different program, and even on a host other
889 than the one on which it was compiled. Details are given in the
890 pcreprecompile documentation.
891
892
893 CHECKING BUILD-TIME OPTIONS
894
895 int pcre_config(int what, void *where);
896
897 The function pcre_config() makes it possible for a PCRE client to dis-
898 cover which optional features have been compiled into the PCRE library.
899 The pcrebuild documentation has more details about these optional fea-
900 tures.
901
902 The first argument for pcre_config() is an integer, specifying which
903 information is required; the second argument is a pointer to a variable
904 into which the information is placed. The following information is
905 available:
906
907 PCRE_CONFIG_UTF8
908
909 The output is an integer that is set to one if UTF-8 support is avail-
910 able; otherwise it is set to zero.
911
912 PCRE_CONFIG_UNICODE_PROPERTIES
913
914 The output is an integer that is set to one if support for Unicode
915 character properties is available; otherwise it is set to zero.
916
917 PCRE_CONFIG_NEWLINE
918
919 The output is an integer whose value specifies the default character
920 sequence that is recognized as meaning "newline". The four values that
921 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
922 and -1 for ANY. The default should normally be the standard sequence
923 for your operating system.
924
925 PCRE_CONFIG_LINK_SIZE
926
927 The output is an integer that contains the number of bytes used for
928 internal linkage in compiled regular expressions. The value is 2, 3, or
929 4. Larger values allow larger regular expressions to be compiled, at
930 the expense of slower matching. The default value of 2 is sufficient
931 for all but the most massive patterns, since it allows the compiled
932 pattern to be up to 64K in size.
933
934 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
935
936 The output is an integer that contains the threshold above which the
937 POSIX interface uses malloc() for output vectors. Further details are
938 given in the pcreposix documentation.
939
940 PCRE_CONFIG_MATCH_LIMIT
941
942 The output is an integer that gives the default limit for the number of
943 internal matching function calls in a pcre_exec() execution. Further
944 details are given with pcre_exec() below.
945
946 PCRE_CONFIG_MATCH_LIMIT_RECURSION
947
948 The output is an integer that gives the default limit for the depth of
949 recursion when calling the internal matching function in a pcre_exec()
950 execution. Further details are given with pcre_exec() below.
951
952 PCRE_CONFIG_STACKRECURSE
953
954 The output is an integer that is set to one if internal recursion when
955 running pcre_exec() is implemented by recursive function calls that use
956 the stack to remember their state. This is the usual way that PCRE is
957 compiled. The output is zero if PCRE was compiled to use blocks of data
958 on the heap instead of recursive function calls. In this case,
959 pcre_stack_malloc and pcre_stack_free are called to manage memory
960 blocks on the heap, thus avoiding the use of the stack.
961
962
963 COMPILING A PATTERN
964
965 pcre *pcre_compile(const char *pattern, int options,
966 const char **errptr, int *erroffset,
967 const unsigned char *tableptr);
968
969 pcre *pcre_compile2(const char *pattern, int options,
970 int *errorcodeptr,
971 const char **errptr, int *erroffset,
972 const unsigned char *tableptr);
973
974 Either of the functions pcre_compile() or pcre_compile2() can be called
975 to compile a pattern into an internal form. The only difference between
976 the two interfaces is that pcre_compile2() has an additional argument,
977 errorcodeptr, via which a numerical error code can be returned.
978
979 The pattern is a C string terminated by a binary zero, and is passed in
980 the pattern argument. A pointer to a single block of memory that is
981 obtained via pcre_malloc is returned. This contains the compiled code
982 and related data. The pcre type is defined for the returned block; this
983 is a typedef for a structure whose contents are not externally defined.
984 It is up to the caller to free the memory (via pcre_free) when it is no
985 longer required.
986
987 Although the compiled code of a PCRE regex is relocatable, that is, it
988 does not depend on memory location, the complete pcre data block is not
989 fully relocatable, because it may contain a copy of the tableptr argu-
990 ment, which is an address (see below).
991
992 The options argument contains various bit settings that affect the com-
993 pilation. It should be zero if no options are required. The available
994 options are described below. Some of them, in particular, those that
995 are compatible with Perl, can also be set and unset from within the
996 pattern (see the detailed description in the pcrepattern documenta-
997 tion). For these options, the contents of the options argument speci-
998 fies their initial settings at the start of compilation and execution.
999 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1000 of matching as well as at compile time.
1001
1002 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1003 if compilation of a pattern fails, pcre_compile() returns NULL, and
1004 sets the variable pointed to by errptr to point to a textual error mes-
1005 sage. This is a static string that is part of the library. You must not
1006 try to free it. The offset from the start of the pattern to the charac-
1007 ter where the error was discovered is placed in the variable pointed to
1008 by erroffset, which must not be NULL. If it is, an immediate error is
1009 given.
1010
1011 If pcre_compile2() is used instead of pcre_compile(), and the error-
1012 codeptr argument is not NULL, a non-zero error code number is returned
1013 via this argument in the event of an error. This is in addition to the
1014 textual error message. Error codes and messages are listed below.
1015
1016 If the final argument, tableptr, is NULL, PCRE uses a default set of
1017 character tables that are built when PCRE is compiled, using the
1018 default C locale. Otherwise, tableptr must be an address that is the
1019 result of a call to pcre_maketables(). This value is stored with the
1020 compiled pattern, and used again by pcre_exec(), unless another table
1021 pointer is passed to it. For more discussion, see the section on locale
1022 support below.
1023
1024 This code fragment shows a typical straightforward call to pcre_com-
1025 pile():
1026
1027 pcre *re;
1028 const char *error;
1029 int erroffset;
1030 re = pcre_compile(
1031 "^A.*Z", /* the pattern */
1032 0, /* default options */
1033 &error, /* for error message */
1034 &erroffset, /* for error offset */
1035 NULL); /* use default character tables */
1036
1037 The following names for option bits are defined in the pcre.h header
1038 file:
1039
1040 PCRE_ANCHORED
1041
1042 If this bit is set, the pattern is forced to be "anchored", that is, it
1043 is constrained to match only at the first matching point in the string
1044 that is being searched (the "subject string"). This effect can also be
1045 achieved by appropriate constructs in the pattern itself, which is the
1046 only way to do it in Perl.
1047
1048 PCRE_AUTO_CALLOUT
1049
1050 If this bit is set, pcre_compile() automatically inserts callout items,
1051 all with number 255, before each pattern item. For discussion of the
1052 callout facility, see the pcrecallout documentation.
1053
1054 PCRE_CASELESS
1055
1056 If this bit is set, letters in the pattern match both upper and lower
1057 case letters. It is equivalent to Perl's /i option, and it can be
1058 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1059 always understands the concept of case for characters whose values are
1060 less than 128, so caseless matching is always possible. For characters
1061 with higher values, the concept of case is supported if PCRE is com-
1062 piled with Unicode property support, but not otherwise. If you want to
1063 use caseless matching for characters 128 and above, you must ensure
1064 that PCRE is compiled with Unicode property support as well as with
1065 UTF-8 support.
1066
1067 PCRE_DOLLAR_ENDONLY
1068
1069 If this bit is set, a dollar metacharacter in the pattern matches only
1070 at the end of the subject string. Without this option, a dollar also
1071 matches immediately before a newline at the end of the string (but not
1072 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1073 if PCRE_MULTILINE is set. There is no equivalent to this option in
1074 Perl, and no way to set it within a pattern.
1075
1076 PCRE_DOTALL
1077
1078 If this bit is set, a dot metacharater in the pattern matches all char-
1079 acters, including those that indicate newline. Without it, a dot does
1080 not match when the current position is at a newline. This option is
1081 equivalent to Perl's /s option, and it can be changed within a pattern
1082 by a (?s) option setting. A negative class such as [^a] always matches
1083 newline characters, independent of the setting of this option.
1084
1085 PCRE_DUPNAMES
1086
1087 If this bit is set, names used to identify capturing subpatterns need
1088 not be unique. This can be helpful for certain types of pattern when it
1089 is known that only one instance of the named subpattern can ever be
1090 matched. There are more details of named subpatterns below; see also
1091 the pcrepattern documentation.
1092
1093 PCRE_EXTENDED
1094
1095 If this bit is set, whitespace data characters in the pattern are
1096 totally ignored except when escaped or inside a character class. White-
1097 space does not include the VT character (code 11). In addition, charac-
1098 ters between an unescaped # outside a character class and the next new-
1099 line, inclusive, are also ignored. This is equivalent to Perl's /x
1100 option, and it can be changed within a pattern by a (?x) option set-
1101 ting.
1102
1103 This option makes it possible to include comments inside complicated
1104 patterns. Note, however, that this applies only to data characters.
1105 Whitespace characters may never appear within special character
1106 sequences in a pattern, for example within the sequence (?( which
1107 introduces a conditional subpattern.
1108
1109 PCRE_EXTRA
1110
1111 This option was invented in order to turn on additional functionality
1112 of PCRE that is incompatible with Perl, but it is currently of very
1113 little use. When set, any backslash in a pattern that is followed by a
1114 letter that has no special meaning causes an error, thus reserving
1115 these combinations for future expansion. By default, as in Perl, a
1116 backslash followed by a letter with no special meaning is treated as a
1117 literal. (Perl can, however, be persuaded to give a warning for this.)
1118 There are at present no other features controlled by this option. It
1119 can also be set by a (?X) option setting within a pattern.
1120
1121 PCRE_FIRSTLINE
1122
1123 If this option is set, an unanchored pattern is required to match
1124 before or at the first newline in the subject string, though the
1125 matched text may continue over the newline.
1126
1127 PCRE_MULTILINE
1128
1129 By default, PCRE treats the subject string as consisting of a single
1130 line of characters (even if it actually contains newlines). The "start
1131 of line" metacharacter (^) matches only at the start of the string,
1132 while the "end of line" metacharacter ($) matches only at the end of
1133 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1134 is set). This is the same as Perl.
1135
1136 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1137 constructs match immediately following or immediately before internal
1138 newlines in the subject string, respectively, as well as at the very
1139 start and end. This is equivalent to Perl's /m option, and it can be
1140 changed within a pattern by a (?m) option setting. If there are no new-
1141 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1142 setting PCRE_MULTILINE has no effect.
1143
1144 PCRE_NEWLINE_CR
1145 PCRE_NEWLINE_LF
1146 PCRE_NEWLINE_CRLF
1147 PCRE_NEWLINE_ANYCRLF
1148 PCRE_NEWLINE_ANY
1149
1150 These options override the default newline definition that was chosen
1151 when PCRE was built. Setting the first or the second specifies that a
1152 newline is indicated by a single character (CR or LF, respectively).
1153 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1154 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1155 that any of the three preceding sequences should be recognized. Setting
1156 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1157 recognized. The Unicode newline sequences are the three just mentioned,
1158 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1159 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1160 (paragraph separator, U+2029). The last two are recognized only in
1161 UTF-8 mode.
1162
1163 The newline setting in the options word uses three bits that are
1164 treated as a number, giving eight possibilities. Currently only six are
1165 used (default plus the five values above). This means that if you set
1166 more than one newline option, the combination may or may not be sensi-
1167 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1168 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1169 cause an error.
1170
1171 The only time that a line break is specially recognized when compiling
1172 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1173 character class is encountered. This indicates a comment that lasts
1174 until after the next line break sequence. In other circumstances, line
1175 break sequences are treated as literal data, except that in
1176 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1177 and are therefore ignored.
1178
1179 The newline option that is set at compile time becomes the default that
1180 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1181
1182 PCRE_NO_AUTO_CAPTURE
1183
1184 If this option is set, it disables the use of numbered capturing paren-
1185 theses in the pattern. Any opening parenthesis that is not followed by
1186 ? behaves as if it were followed by ?: but named parentheses can still
1187 be used for capturing (and they acquire numbers in the usual way).
1188 There is no equivalent of this option in Perl.
1189
1190 PCRE_UNGREEDY
1191
1192 This option inverts the "greediness" of the quantifiers so that they
1193 are not greedy by default, but become greedy if followed by "?". It is
1194 not compatible with Perl. It can also be set by a (?U) option setting
1195 within the pattern.
1196
1197 PCRE_UTF8
1198
1199 This option causes PCRE to regard both the pattern and the subject as
1200 strings of UTF-8 characters instead of single-byte character strings.
1201 However, it is available only when PCRE is built to include UTF-8 sup-
1202 port. If not, the use of this option provokes an error. Details of how
1203 this option changes the behaviour of PCRE are given in the section on
1204 UTF-8 support in the main pcre page.
1205
1206 PCRE_NO_UTF8_CHECK
1207
1208 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1209 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1210 pcre_compile() returns an error. If you already know that your pattern
1211 is valid, and you want to skip this check for performance reasons, you
1212 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1213 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1214 your program to crash. Note that this option can also be passed to
1215 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1216 ing of subject strings.
1217
1218
1219 COMPILATION ERROR CODES
1220
1221 The following table lists the error codes than may be returned by
1222 pcre_compile2(), along with the error messages that may be returned by
1223 both compiling functions. As PCRE has developed, some error codes have
1224 fallen out of use. To avoid confusion, they have not been re-used.
1225
1226 0 no error
1227 1 \ at end of pattern
1228 2 \c at end of pattern
1229 3 unrecognized character follows \
1230 4 numbers out of order in {} quantifier
1231 5 number too big in {} quantifier
1232 6 missing terminating ] for character class
1233 7 invalid escape sequence in character class
1234 8 range out of order in character class
1235 9 nothing to repeat
1236 10 [this code is not in use]
1237 11 internal error: unexpected repeat
1238 12 unrecognized character after (?
1239 13 POSIX named classes are supported only within a class
1240 14 missing )
1241 15 reference to non-existent subpattern
1242 16 erroffset passed as NULL
1243 17 unknown option bit(s) set
1244 18 missing ) after comment
1245 19 [this code is not in use]
1246 20 regular expression too large
1247 21 failed to get memory
1248 22 unmatched parentheses
1249 23 internal error: code overflow
1250 24 unrecognized character after (?<
1251 25 lookbehind assertion is not fixed length
1252 26 malformed number or name after (?(
1253 27 conditional group contains more than two branches
1254 28 assertion expected after (?(
1255 29 (?R or (?digits must be followed by )
1256 30 unknown POSIX class name
1257 31 POSIX collating elements are not supported
1258 32 this version of PCRE is not compiled with PCRE_UTF8 support
1259 33 [this code is not in use]
1260 34 character value in \x{...} sequence is too large
1261 35 invalid condition (?(0)
1262 36 \C not allowed in lookbehind assertion
1263 37 PCRE does not support \L, \l, \N, \U, or \u
1264 38 number after (?C is > 255
1265 39 closing ) for (?C expected
1266 40 recursive call could loop indefinitely
1267 41 unrecognized character after (?P
1268 42 syntax error in subpattern name (missing terminator)
1269 43 two named subpatterns have the same name
1270 44 invalid UTF-8 string
1271 45 support for \P, \p, and \X has not been compiled
1272 46 malformed \P or \p sequence
1273 47 unknown property name after \P or \p
1274 48 subpattern name is too long (maximum 32 characters)
1275 49 too many named subpatterns (maximum 10,000)
1276 50 repeated subpattern is too long
1277 51 octal value is greater than \377 (not in UTF-8 mode)
1278 52 internal error: overran compiling workspace
1279 53 internal error: previously-checked referenced subpattern not
1280 found
1281 54 DEFINE group contains more than one branch
1282 55 repeating a DEFINE group is not allowed
1283 56 inconsistent NEWLINE options"
1284
1285
1286 STUDYING A PATTERN
1287
1288 pcre_extra *pcre_study(const pcre *code, int options
1289 const char **errptr);
1290
1291 If a compiled pattern is going to be used several times, it is worth
1292 spending more time analyzing it in order to speed up the time taken for
1293 matching. The function pcre_study() takes a pointer to a compiled pat-
1294 tern as its first argument. If studying the pattern produces additional
1295 information that will help speed up matching, pcre_study() returns a
1296 pointer to a pcre_extra block, in which the study_data field points to
1297 the results of the study.
1298
1299 The returned value from pcre_study() can be passed directly to
1300 pcre_exec(). However, a pcre_extra block also contains other fields
1301 that can be set by the caller before the block is passed; these are
1302 described below in the section on matching a pattern.
1303
1304 If studying the pattern does not produce any additional information
1305 pcre_study() returns NULL. In that circumstance, if the calling program
1306 wants to pass any of the other fields to pcre_exec(), it must set up
1307 its own pcre_extra block.
1308
1309 The second argument of pcre_study() contains option bits. At present,
1310 no options are defined, and this argument should always be zero.
1311
1312 The third argument for pcre_study() is a pointer for an error message.
1313 If studying succeeds (even if no data is returned), the variable it
1314 points to is set to NULL. Otherwise it is set to point to a textual
1315 error message. This is a static string that is part of the library. You
1316 must not try to free it. You should test the error pointer for NULL
1317 after calling pcre_study(), to be sure that it has run successfully.
1318
1319 This is a typical call to pcre_study():
1320
1321 pcre_extra *pe;
1322 pe = pcre_study(
1323 re, /* result of pcre_compile() */
1324 0, /* no options exist */
1325 &error); /* set to NULL or points to a message */
1326
1327 At present, studying a pattern is useful only for non-anchored patterns
1328 that do not have a single fixed starting character. A bitmap of possi-
1329 ble starting bytes is created.
1330
1331
1332 LOCALE SUPPORT
1333
1334 PCRE handles caseless matching, and determines whether characters are
1335 letters, digits, or whatever, by reference to a set of tables, indexed
1336 by character value. When running in UTF-8 mode, this applies only to
1337 characters with codes less than 128. Higher-valued codes never match
1338 escapes such as \w or \d, but can be tested with \p if PCRE is built
1339 with Unicode character property support. The use of locales with Uni-
1340 code is discouraged. If you are handling characters with codes greater
1341 than 128, you should either use UTF-8 and Unicode, or use locales, but
1342 not try to mix the two.
1343
1344 PCRE contains an internal set of tables that are used when the final
1345 argument of pcre_compile() is NULL. These are sufficient for many
1346 applications. Normally, the internal tables recognize only ASCII char-
1347 acters. However, when PCRE is built, it is possible to cause the inter-
1348 nal tables to be rebuilt in the default "C" locale of the local system,
1349 which may cause them to be different.
1350
1351 The internal tables can always be overridden by tables supplied by the
1352 application that calls PCRE. These may be created in a different locale
1353 from the default. As more and more applications change to using Uni-
1354 code, the need for this locale support is expected to die away.
1355
1356 External tables are built by calling the pcre_maketables() function,
1357 which has no arguments, in the relevant locale. The result can then be
1358 passed to pcre_compile() or pcre_exec() as often as necessary. For
1359 example, to build and use tables that are appropriate for the French
1360 locale (where accented characters with values greater than 128 are
1361 treated as letters), the following code could be used:
1362
1363 setlocale(LC_CTYPE, "fr_FR");
1364 tables = pcre_maketables();
1365 re = pcre_compile(..., tables);
1366
1367 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1368 if you are using Windows, the name for the French locale is "french".
1369
1370 When pcre_maketables() runs, the tables are built in memory that is
1371 obtained via pcre_malloc. It is the caller's responsibility to ensure
1372 that the memory containing the tables remains available for as long as
1373 it is needed.
1374
1375 The pointer that is passed to pcre_compile() is saved with the compiled
1376 pattern, and the same tables are used via this pointer by pcre_study()
1377 and normally also by pcre_exec(). Thus, by default, for any single pat-
1378 tern, compilation, studying and matching all happen in the same locale,
1379 but different patterns can be compiled in different locales.
1380
1381 It is possible to pass a table pointer or NULL (indicating the use of
1382 the internal tables) to pcre_exec(). Although not intended for this
1383 purpose, this facility could be used to match a pattern in a different
1384 locale from the one in which it was compiled. Passing table pointers at
1385 run time is discussed below in the section on matching a pattern.
1386
1387
1388 INFORMATION ABOUT A PATTERN
1389
1390 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1391 int what, void *where);
1392
1393 The pcre_fullinfo() function returns information about a compiled pat-
1394 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1395 less retained for backwards compability (and is documented below).
1396
1397 The first argument for pcre_fullinfo() is a pointer to the compiled
1398 pattern. The second argument is the result of pcre_study(), or NULL if
1399 the pattern was not studied. The third argument specifies which piece
1400 of information is required, and the fourth argument is a pointer to a
1401 variable to receive the data. The yield of the function is zero for
1402 success, or one of the following negative numbers:
1403
1404 PCRE_ERROR_NULL the argument code was NULL
1405 the argument where was NULL
1406 PCRE_ERROR_BADMAGIC the "magic number" was not found
1407 PCRE_ERROR_BADOPTION the value of what was invalid
1408
1409 The "magic number" is placed at the start of each compiled pattern as
1410 an simple check against passing an arbitrary memory pointer. Here is a
1411 typical call of pcre_fullinfo(), to obtain the length of the compiled
1412 pattern:
1413
1414 int rc;
1415 size_t length;
1416 rc = pcre_fullinfo(
1417 re, /* result of pcre_compile() */
1418 pe, /* result of pcre_study(), or NULL */
1419 PCRE_INFO_SIZE, /* what is required */
1420 &length); /* where to put the data */
1421
1422 The possible values for the third argument are defined in pcre.h, and
1423 are as follows:
1424
1425 PCRE_INFO_BACKREFMAX
1426
1427 Return the number of the highest back reference in the pattern. The
1428 fourth argument should point to an int variable. Zero is returned if
1429 there are no back references.
1430
1431 PCRE_INFO_CAPTURECOUNT
1432
1433 Return the number of capturing subpatterns in the pattern. The fourth
1434 argument should point to an int variable.
1435
1436 PCRE_INFO_DEFAULT_TABLES
1437
1438 Return a pointer to the internal default character tables within PCRE.
1439 The fourth argument should point to an unsigned char * variable. This
1440 information call is provided for internal use by the pcre_study() func-
1441 tion. External callers can cause PCRE to use its internal tables by
1442 passing a NULL table pointer.
1443
1444 PCRE_INFO_FIRSTBYTE
1445
1446 Return information about the first byte of any matched string, for a
1447 non-anchored pattern. The fourth argument should point to an int vari-
1448 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1449 is still recognized for backwards compatibility.)
1450
1451 If there is a fixed first byte, for example, from a pattern such as
1452 (cat|cow|coyote), its value is returned. Otherwise, if either
1453
1454 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1455 branch starts with "^", or
1456
1457 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1458 set (if it were set, the pattern would be anchored),
1459
1460 -1 is returned, indicating that the pattern matches only at the start
1461 of a subject string or after any newline within the string. Otherwise
1462 -2 is returned. For anchored patterns, -2 is returned.
1463
1464 PCRE_INFO_FIRSTTABLE
1465
1466 If the pattern was studied, and this resulted in the construction of a
1467 256-bit table indicating a fixed set of bytes for the first byte in any
1468 matching string, a pointer to the table is returned. Otherwise NULL is
1469 returned. The fourth argument should point to an unsigned char * vari-
1470 able.
1471
1472 PCRE_INFO_LASTLITERAL
1473
1474 Return the value of the rightmost literal byte that must exist in any
1475 matched string, other than at its start, if such a byte has been
1476 recorded. The fourth argument should point to an int variable. If there
1477 is no such byte, -1 is returned. For anchored patterns, a last literal
1478 byte is recorded only if it follows something of variable length. For
1479 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1480 /^a\dz\d/ the returned value is -1.
1481
1482 PCRE_INFO_NAMECOUNT
1483 PCRE_INFO_NAMEENTRYSIZE
1484 PCRE_INFO_NAMETABLE
1485
1486 PCRE supports the use of named as well as numbered capturing parenthe-
1487 ses. The names are just an additional way of identifying the parenthe-
1488 ses, which still acquire numbers. Several convenience functions such as
1489 pcre_get_named_substring() are provided for extracting captured sub-
1490 strings by name. It is also possible to extract the data directly, by
1491 first converting the name to a number in order to access the correct
1492 pointers in the output vector (described with pcre_exec() below). To do
1493 the conversion, you need to use the name-to-number map, which is
1494 described by these three values.
1495
1496 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1497 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1498 of each entry; both of these return an int value. The entry size
1499 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1500 a pointer to the first entry of the table (a pointer to char). The
1501 first two bytes of each entry are the number of the capturing parenthe-
1502 sis, most significant byte first. The rest of the entry is the corre-
1503 sponding name, zero terminated. The names are in alphabetical order.
1504 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1505 theses numbers. For example, consider the following pattern (assume
1506 PCRE_EXTENDED is set, so white space - including newlines - is
1507 ignored):
1508
1509 (?<date> (?<year>(\d\d)?\d\d) -
1510 (?<month>\d\d) - (?<day>\d\d) )
1511
1512 There are four named subpatterns, so the table has four entries, and
1513 each entry in the table is eight bytes long. The table is as follows,
1514 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1515 as ??:
1516
1517 00 01 d a t e 00 ??
1518 00 05 d a y 00 ?? ??
1519 00 04 m o n t h 00
1520 00 02 y e a r 00 ??
1521
1522 When writing code to extract data from named subpatterns using the
1523 name-to-number map, remember that the length of the entries is likely
1524 to be different for each compiled pattern.
1525
1526 PCRE_INFO_OPTIONS
1527
1528 Return a copy of the options with which the pattern was compiled. The
1529 fourth argument should point to an unsigned long int variable. These
1530 option bits are those specified in the call to pcre_compile(), modified
1531 by any top-level option settings within the pattern itself.
1532
1533 A pattern is automatically anchored by PCRE if all of its top-level
1534 alternatives begin with one of the following:
1535
1536 ^ unless PCRE_MULTILINE is set
1537 \A always
1538 \G always
1539 .* if PCRE_DOTALL is set and there are no back
1540 references to the subpattern in which .* appears
1541
1542 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1543 by pcre_fullinfo().
1544
1545 PCRE_INFO_SIZE
1546
1547 Return the size of the compiled pattern, that is, the value that was
1548 passed as the argument to pcre_malloc() when PCRE was getting memory in
1549 which to place the compiled data. The fourth argument should point to a
1550 size_t variable.
1551
1552 PCRE_INFO_STUDYSIZE
1553
1554 Return the size of the data block pointed to by the study_data field in
1555 a pcre_extra block. That is, it is the value that was passed to
1556 pcre_malloc() when PCRE was getting memory into which to place the data
1557 created by pcre_study(). The fourth argument should point to a size_t
1558 variable.
1559
1560
1561 OBSOLETE INFO FUNCTION
1562
1563 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1564
1565 The pcre_info() function is now obsolete because its interface is too
1566 restrictive to return all the available data about a compiled pattern.
1567 New programs should use pcre_fullinfo() instead. The yield of
1568 pcre_info() is the number of capturing subpatterns, or one of the fol-
1569 lowing negative numbers:
1570
1571 PCRE_ERROR_NULL the argument code was NULL
1572 PCRE_ERROR_BADMAGIC the "magic number" was not found
1573
1574 If the optptr argument is not NULL, a copy of the options with which
1575 the pattern was compiled is placed in the integer it points to (see
1576 PCRE_INFO_OPTIONS above).
1577
1578 If the pattern is not anchored and the firstcharptr argument is not
1579 NULL, it is used to pass back information about the first character of
1580 any matched string (see PCRE_INFO_FIRSTBYTE above).
1581
1582
1583 REFERENCE COUNTS
1584
1585 int pcre_refcount(pcre *code, int adjust);
1586
1587 The pcre_refcount() function is used to maintain a reference count in
1588 the data block that contains a compiled pattern. It is provided for the
1589 benefit of applications that operate in an object-oriented manner,
1590 where different parts of the application may be using the same compiled
1591 pattern, but you want to free the block when they are all done.
1592
1593 When a pattern is compiled, the reference count field is initialized to
1594 zero. It is changed only by calling this function, whose action is to
1595 add the adjust value (which may be positive or negative) to it. The
1596 yield of the function is the new value. However, the value of the count
1597 is constrained to lie between 0 and 65535, inclusive. If the new value
1598 is outside these limits, it is forced to the appropriate limit value.
1599
1600 Except when it is zero, the reference count is not correctly preserved
1601 if a pattern is compiled on one host and then transferred to a host
1602 whose byte-order is different. (This seems a highly unlikely scenario.)
1603
1604
1605 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1606
1607 int pcre_exec(const pcre *code, const pcre_extra *extra,
1608 const char *subject, int length, int startoffset,
1609 int options, int *ovector, int ovecsize);
1610
1611 The function pcre_exec() is called to match a subject string against a
1612 compiled pattern, which is passed in the code argument. If the pattern
1613 has been studied, the result of the study should be passed in the extra
1614 argument. This function is the main matching facility of the library,
1615 and it operates in a Perl-like manner. For specialist use there is also
1616 an alternative matching function, which is described below in the sec-
1617 tion about the pcre_dfa_exec() function.
1618
1619 In most applications, the pattern will have been compiled (and option-
1620 ally studied) in the same process that calls pcre_exec(). However, it
1621 is possible to save compiled patterns and study data, and then use them
1622 later in different processes, possibly even on different hosts. For a
1623 discussion about this, see the pcreprecompile documentation.
1624
1625 Here is an example of a simple call to pcre_exec():
1626
1627 int rc;
1628 int ovector[30];
1629 rc = pcre_exec(
1630 re, /* result of pcre_compile() */
1631 NULL, /* we didn't study the pattern */
1632 "some string", /* the subject string */
1633 11, /* the length of the subject string */
1634 0, /* start at offset 0 in the subject */
1635 0, /* default options */
1636 ovector, /* vector of integers for substring information */
1637 30); /* number of elements (NOT size in bytes) */
1638
1639 Extra data for pcre_exec()
1640
1641 If the extra argument is not NULL, it must point to a pcre_extra data
1642 block. The pcre_study() function returns such a block (when it doesn't
1643 return NULL), but you can also create one for yourself, and pass addi-
1644 tional information in it. The pcre_extra block contains the following
1645 fields (not necessarily in this order):
1646
1647 unsigned long int flags;
1648 void *study_data;
1649 unsigned long int match_limit;
1650 unsigned long int match_limit_recursion;
1651 void *callout_data;
1652 const unsigned char *tables;
1653
1654 The flags field is a bitmap that specifies which of the other fields
1655 are set. The flag bits are:
1656
1657 PCRE_EXTRA_STUDY_DATA
1658 PCRE_EXTRA_MATCH_LIMIT
1659 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1660 PCRE_EXTRA_CALLOUT_DATA
1661 PCRE_EXTRA_TABLES
1662
1663 Other flag bits should be set to zero. The study_data field is set in
1664 the pcre_extra block that is returned by pcre_study(), together with
1665 the appropriate flag bit. You should not set this yourself, but you may
1666 add to the block by setting the other fields and their corresponding
1667 flag bits.
1668
1669 The match_limit field provides a means of preventing PCRE from using up
1670 a vast amount of resources when running patterns that are not going to
1671 match, but which have a very large number of possibilities in their
1672 search trees. The classic example is the use of nested unlimited
1673 repeats.
1674
1675 Internally, PCRE uses a function called match() which it calls repeat-
1676 edly (sometimes recursively). The limit set by match_limit is imposed
1677 on the number of times this function is called during a match, which
1678 has the effect of limiting the amount of backtracking that can take
1679 place. For patterns that are not anchored, the count restarts from zero
1680 for each position in the subject string.
1681
1682 The default value for the limit can be set when PCRE is built; the
1683 default default is 10 million, which handles all but the most extreme
1684 cases. You can override the default by suppling pcre_exec() with a
1685 pcre_extra block in which match_limit is set, and
1686 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1687 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1688
1689 The match_limit_recursion field is similar to match_limit, but instead
1690 of limiting the total number of times that match() is called, it limits
1691 the depth of recursion. The recursion depth is a smaller number than
1692 the total number of calls, because not all calls to match() are recur-
1693 sive. This limit is of use only if it is set smaller than match_limit.
1694
1695 Limiting the recursion depth limits the amount of stack that can be
1696 used, or, when PCRE has been compiled to use memory on the heap instead
1697 of the stack, the amount of heap memory that can be used.
1698
1699 The default value for match_limit_recursion can be set when PCRE is
1700 built; the default default is the same value as the default for
1701 match_limit. You can override the default by suppling pcre_exec() with
1702 a pcre_extra block in which match_limit_recursion is set, and
1703 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1704 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1705
1706 The pcre_callout field is used in conjunction with the "callout" fea-
1707 ture, which is described in the pcrecallout documentation.
1708
1709 The tables field is used to pass a character tables pointer to
1710 pcre_exec(); this overrides the value that is stored with the compiled
1711 pattern. A non-NULL value is stored with the compiled pattern only if
1712 custom tables were supplied to pcre_compile() via its tableptr argu-
1713 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1714 PCRE's internal tables to be used. This facility is helpful when re-
1715 using patterns that have been saved after compiling with an external
1716 set of tables, because the external tables might be at a different
1717 address when pcre_exec() is called. See the pcreprecompile documenta-
1718 tion for a discussion of saving compiled patterns for later use.
1719
1720 Option bits for pcre_exec()
1721
1722 The unused bits of the options argument for pcre_exec() must be zero.
1723 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1724 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1725 PCRE_PARTIAL.
1726
1727 PCRE_ANCHORED
1728
1729 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1730 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1731 turned out to be anchored by virtue of its contents, it cannot be made
1732 unachored at matching time.
1733
1734 PCRE_NEWLINE_CR
1735 PCRE_NEWLINE_LF
1736 PCRE_NEWLINE_CRLF
1737 PCRE_NEWLINE_ANYCRLF
1738 PCRE_NEWLINE_ANY
1739
1740 These options override the newline definition that was chosen or
1741 defaulted when the pattern was compiled. For details, see the descrip-
1742 tion of pcre_compile() above. During matching, the newline choice
1743 affects the behaviour of the dot, circumflex, and dollar metacharac-
1744 ters. It may also alter the way the match position is advanced after a
1745 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF,
1746 PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is set, and a match attempt
1747 fails when the current position is at a CRLF sequence, the match posi-
1748 tion is advanced by two characters instead of one, in other words, to
1749 after the CRLF.
1750
1751 PCRE_NOTBOL
1752
1753 This option specifies that first character of the subject string is not
1754 the beginning of a line, so the circumflex metacharacter should not
1755 match before it. Setting this without PCRE_MULTILINE (at compile time)
1756 causes circumflex never to match. This option affects only the behav-
1757 iour of the circumflex metacharacter. It does not affect \A.
1758
1759 PCRE_NOTEOL
1760
1761 This option specifies that the end of the subject string is not the end
1762 of a line, so the dollar metacharacter should not match it nor (except
1763 in multiline mode) a newline immediately before it. Setting this with-
1764 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1765 option affects only the behaviour of the dollar metacharacter. It does
1766 not affect \Z or \z.
1767
1768 PCRE_NOTEMPTY
1769
1770 An empty string is not considered to be a valid match if this option is
1771 set. If there are alternatives in the pattern, they are tried. If all
1772 the alternatives match the empty string, the entire match fails. For
1773 example, if the pattern
1774
1775 a?b?
1776
1777 is applied to a string not beginning with "a" or "b", it matches the
1778 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1779 match is not valid, so PCRE searches further into the string for occur-
1780 rences of "a" or "b".
1781
1782 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1783 cial case of a pattern match of the empty string within its split()
1784 function, and when using the /g modifier. It is possible to emulate
1785 Perl's behaviour after matching a null string by first trying the match
1786 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1787 if that fails by advancing the starting offset (see below) and trying
1788 an ordinary match again. There is some code that demonstrates how to do
1789 this in the pcredemo.c sample program.
1790
1791 PCRE_NO_UTF8_CHECK
1792
1793 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1794 UTF-8 string is automatically checked when pcre_exec() is subsequently
1795 called. The value of startoffset is also checked to ensure that it
1796 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1797 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1798 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1799 returned.
1800
1801 If you already know that your subject is valid, and you want to skip
1802 these checks for performance reasons, you can set the
1803 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1804 do this for the second and subsequent calls to pcre_exec() if you are
1805 making repeated calls to find all the matches in a single subject
1806 string. However, you should be sure that the value of startoffset
1807 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1808 set, the effect of passing an invalid UTF-8 string as a subject, or a
1809 value of startoffset that does not point to the start of a UTF-8 char-
1810 acter, is undefined. Your program may crash.
1811
1812 PCRE_PARTIAL
1813
1814 This option turns on the partial matching feature. If the subject
1815 string fails to match the pattern, but at some point during the match-
1816 ing process the end of the subject was reached (that is, the subject
1817 partially matches the pattern and the failure to match occurred only
1818 because there were not enough subject characters), pcre_exec() returns
1819 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1820 used, there are restrictions on what may appear in the pattern. These
1821 are discussed in the pcrepartial documentation.
1822
1823 The string to be matched by pcre_exec()
1824
1825 The subject string is passed to pcre_exec() as a pointer in subject, a
1826 length in length, and a starting byte offset in startoffset. In UTF-8
1827 mode, the byte offset must point to the start of a UTF-8 character.
1828 Unlike the pattern string, the subject may contain binary zero bytes.
1829 When the starting offset is zero, the search for a match starts at the
1830 beginning of the subject, and this is by far the most common case.
1831
1832 A non-zero starting offset is useful when searching for another match
1833 in the same subject by calling pcre_exec() again after a previous suc-
1834 cess. Setting startoffset differs from just passing over a shortened
1835 string and setting PCRE_NOTBOL in the case of a pattern that begins
1836 with any kind of lookbehind. For example, consider the pattern
1837
1838 \Biss\B
1839
1840 which finds occurrences of "iss" in the middle of words. (\B matches
1841 only if the current position in the subject is not a word boundary.)
1842 When applied to the string "Mississipi" the first call to pcre_exec()
1843 finds the first occurrence. If pcre_exec() is called again with just
1844 the remainder of the subject, namely "issipi", it does not match,
1845 because \B is always false at the start of the subject, which is deemed
1846 to be a word boundary. However, if pcre_exec() is passed the entire
1847 string again, but with startoffset set to 4, it finds the second occur-
1848 rence of "iss" because it is able to look behind the starting point to
1849 discover that it is preceded by a letter.
1850
1851 If a non-zero starting offset is passed when the pattern is anchored,
1852 one attempt to match at the given offset is made. This can only succeed
1853 if the pattern does not require the match to be at the start of the
1854 subject.
1855
1856 How pcre_exec() returns captured substrings
1857
1858 In general, a pattern matches a certain portion of the subject, and in
1859 addition, further substrings from the subject may be picked out by
1860 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1861 this is called "capturing" in what follows, and the phrase "capturing
1862 subpattern" is used for a fragment of a pattern that picks out a sub-
1863 string. PCRE supports several other kinds of parenthesized subpattern
1864 that do not cause substrings to be captured.
1865
1866 Captured substrings are returned to the caller via a vector of integer
1867 offsets whose address is passed in ovector. The number of elements in
1868 the vector is passed in ovecsize, which must be a non-negative number.
1869 Note: this argument is NOT the size of ovector in bytes.
1870
1871 The first two-thirds of the vector is used to pass back captured sub-
1872 strings, each substring using a pair of integers. The remaining third
1873 of the vector is used as workspace by pcre_exec() while matching cap-
1874 turing subpatterns, and is not available for passing back information.
1875 The length passed in ovecsize should always be a multiple of three. If
1876 it is not, it is rounded down.
1877
1878 When a match is successful, information about captured substrings is
1879 returned in pairs of integers, starting at the beginning of ovector,
1880 and continuing up to two-thirds of its length at the most. The first
1881 element of a pair is set to the offset of the first character in a sub-
1882 string, and the second is set to the offset of the first character
1883 after the end of a substring. The first pair, ovector[0] and ovec-
1884 tor[1], identify the portion of the subject string matched by the
1885 entire pattern. The next pair is used for the first capturing subpat-
1886 tern, and so on. The value returned by pcre_exec() is one more than the
1887 highest numbered pair that has been set. For example, if two substrings
1888 have been captured, the returned value is 3. If there are no capturing
1889 subpatterns, the return value from a successful match is 1, indicating
1890 that just the first pair of offsets has been set.
1891
1892 If a capturing subpattern is matched repeatedly, it is the last portion
1893 of the string that it matched that is returned.
1894
1895 If the vector is too small to hold all the captured substring offsets,
1896 it is used as far as possible (up to two-thirds of its length), and the
1897 function returns a value of zero. In particular, if the substring off-
1898 sets are not of interest, pcre_exec() may be called with ovector passed
1899 as NULL and ovecsize as zero. However, if the pattern contains back
1900 references and the ovector is not big enough to remember the related
1901 substrings, PCRE has to get additional memory for use during matching.
1902 Thus it is usually advisable to supply an ovector.
1903
1904 The pcre_info() function can be used to find out how many capturing
1905 subpatterns there are in a compiled pattern. The smallest size for
1906 ovector that will allow for n captured substrings, in addition to the
1907 offsets of the substring matched by the whole pattern, is (n+1)*3.
1908
1909 It is possible for capturing subpattern number n+1 to match some part
1910 of the subject when subpattern n has not been used at all. For example,
1911 if the string "abc" is matched against the pattern (a|(z))(bc) the
1912 return from the function is 4, and subpatterns 1 and 3 are matched, but
1913 2 is not. When this happens, both values in the offset pairs corre-
1914 sponding to unused subpatterns are set to -1.
1915
1916 Offset values that correspond to unused subpatterns at the end of the
1917 expression are also set to -1. For example, if the string "abc" is
1918 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1919 matched. The return from the function is 2, because the highest used
1920 capturing subpattern number is 1. However, you can refer to the offsets
1921 for the second and third capturing subpatterns if you wish (assuming
1922 the vector is large enough, of course).
1923
1924 Some convenience functions are provided for extracting the captured
1925 substrings as separate strings. These are described below.
1926
1927 Error return values from pcre_exec()
1928
1929 If pcre_exec() fails, it returns a negative number. The following are
1930 defined in the header file:
1931
1932 PCRE_ERROR_NOMATCH (-1)
1933
1934 The subject string did not match the pattern.
1935
1936 PCRE_ERROR_NULL (-2)
1937
1938 Either code or subject was passed as NULL, or ovector was NULL and
1939 ovecsize was not zero.
1940
1941 PCRE_ERROR_BADOPTION (-3)
1942
1943 An unrecognized bit was set in the options argument.
1944
1945 PCRE_ERROR_BADMAGIC (-4)
1946
1947 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1948 to catch the case when it is passed a junk pointer and to detect when a
1949 pattern that was compiled in an environment of one endianness is run in
1950 an environment with the other endianness. This is the error that PCRE
1951 gives when the magic number is not present.
1952
1953 PCRE_ERROR_UNKNOWN_OPCODE (-5)
1954
1955 While running the pattern match, an unknown item was encountered in the
1956 compiled pattern. This error could be caused by a bug in PCRE or by
1957 overwriting of the compiled pattern.
1958
1959 PCRE_ERROR_NOMEMORY (-6)
1960
1961 If a pattern contains back references, but the ovector that is passed
1962 to pcre_exec() is not big enough to remember the referenced substrings,
1963 PCRE gets a block of memory at the start of matching to use for this
1964 purpose. If the call via pcre_malloc() fails, this error is given. The
1965 memory is automatically freed at the end of matching.
1966
1967 PCRE_ERROR_NOSUBSTRING (-7)
1968
1969 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1970 and pcre_get_substring_list() functions (see below). It is never
1971 returned by pcre_exec().
1972
1973 PCRE_ERROR_MATCHLIMIT (-8)
1974
1975 The backtracking limit, as specified by the match_limit field in a
1976 pcre_extra structure (or defaulted) was reached. See the description
1977 above.
1978
1979 PCRE_ERROR_CALLOUT (-9)
1980
1981 This error is never generated by pcre_exec() itself. It is provided for
1982 use by callout functions that want to yield a distinctive error code.
1983 See the pcrecallout documentation for details.
1984
1985 PCRE_ERROR_BADUTF8 (-10)
1986
1987 A string that contains an invalid UTF-8 byte sequence was passed as a
1988 subject.
1989
1990 PCRE_ERROR_BADUTF8_OFFSET (-11)
1991
1992 The UTF-8 byte sequence that was passed as a subject was valid, but the
1993 value of startoffset did not point to the beginning of a UTF-8 charac-
1994 ter.
1995
1996 PCRE_ERROR_PARTIAL (-12)
1997
1998 The subject string did not match, but it did match partially. See the
1999 pcrepartial documentation for details of partial matching.
2000
2001 PCRE_ERROR_BADPARTIAL (-13)
2002
2003 The PCRE_PARTIAL option was used with a compiled pattern containing
2004 items that are not supported for partial matching. See the pcrepartial
2005 documentation for details of partial matching.
2006
2007 PCRE_ERROR_INTERNAL (-14)
2008
2009 An unexpected internal error has occurred. This error could be caused
2010 by a bug in PCRE or by overwriting of the compiled pattern.
2011
2012 PCRE_ERROR_BADCOUNT (-15)
2013
2014 This error is given if the value of the ovecsize argument is negative.
2015
2016 PCRE_ERROR_RECURSIONLIMIT (-21)
2017
2018 The internal recursion limit, as specified by the match_limit_recursion
2019 field in a pcre_extra structure (or defaulted) was reached. See the
2020 description above.
2021
2022 PCRE_ERROR_NULLWSLIMIT (-22)
2023
2024 When a group that can match an empty substring is repeated with an
2025 unbounded upper limit, the subject position at the start of the group
2026 must be remembered, so that a test for an empty string can be made when
2027 the end of the group is reached. Some workspace is required for this;
2028 if it runs out, this error is given.
2029
2030 PCRE_ERROR_BADNEWLINE (-23)
2031
2032 An invalid combination of PCRE_NEWLINE_xxx options was given.
2033
2034 Error numbers -16 to -20 are not used by pcre_exec().
2035
2036
2037 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2038
2039 int pcre_copy_substring(const char *subject, int *ovector,
2040 int stringcount, int stringnumber, char *buffer,
2041 int buffersize);
2042
2043 int pcre_get_substring(const char *subject, int *ovector,
2044 int stringcount, int stringnumber,
2045 const char **stringptr);
2046
2047 int pcre_get_substring_list(const char *subject,
2048 int *ovector, int stringcount, const char ***listptr);
2049
2050 Captured substrings can be accessed directly by using the offsets
2051 returned by pcre_exec() in ovector. For convenience, the functions
2052 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2053 string_list() are provided for extracting captured substrings as new,
2054 separate, zero-terminated strings. These functions identify substrings
2055 by number. The next section describes functions for extracting named
2056 substrings.
2057
2058 A substring that contains a binary zero is correctly extracted and has
2059 a further zero added on the end, but the result is not, of course, a C
2060 string. However, you can process such a string by referring to the
2061 length that is returned by pcre_copy_substring() and pcre_get_sub-
2062 string(). Unfortunately, the interface to pcre_get_substring_list() is
2063 not adequate for handling strings containing binary zeros, because the
2064 end of the final string is not independently indicated.
2065
2066 The first three arguments are the same for all three of these func-
2067 tions: subject is the subject string that has just been successfully
2068 matched, ovector is a pointer to the vector of integer offsets that was
2069 passed to pcre_exec(), and stringcount is the number of substrings that
2070 were captured by the match, including the substring that matched the
2071 entire regular expression. This is the value returned by pcre_exec() if
2072 it is greater than zero. If pcre_exec() returned zero, indicating that
2073 it ran out of space in ovector, the value passed as stringcount should
2074 be the number of elements in the vector divided by three.
2075
2076 The functions pcre_copy_substring() and pcre_get_substring() extract a
2077 single substring, whose number is given as stringnumber. A value of
2078 zero extracts the substring that matched the entire pattern, whereas
2079 higher values extract the captured substrings. For pcre_copy_sub-
2080 string(), the string is placed in buffer, whose length is given by
2081 buffersize, while for pcre_get_substring() a new block of memory is
2082 obtained via pcre_malloc, and its address is returned via stringptr.
2083 The yield of the function is the length of the string, not including
2084 the terminating zero, or one of these error codes:
2085
2086 PCRE_ERROR_NOMEMORY (-6)
2087
2088 The buffer was too small for pcre_copy_substring(), or the attempt to
2089 get memory failed for pcre_get_substring().
2090
2091 PCRE_ERROR_NOSUBSTRING (-7)
2092
2093 There is no substring whose number is stringnumber.
2094
2095 The pcre_get_substring_list() function extracts all available sub-
2096 strings and builds a list of pointers to them. All this is done in a
2097 single block of memory that is obtained via pcre_malloc. The address of
2098 the memory block is returned via listptr, which is also the start of
2099 the list of string pointers. The end of the list is marked by a NULL
2100 pointer. The yield of the function is zero if all went well, or the
2101 error code
2102
2103 PCRE_ERROR_NOMEMORY (-6)
2104
2105 if the attempt to get the memory block failed.
2106
2107 When any of these functions encounter a substring that is unset, which
2108 can happen when capturing subpattern number n+1 matches some part of
2109 the subject, but subpattern n has not been used at all, they return an
2110 empty string. This can be distinguished from a genuine zero-length sub-
2111 string by inspecting the appropriate offset in ovector, which is nega-
2112 tive for unset substrings.
2113
2114 The two convenience functions pcre_free_substring() and pcre_free_sub-
2115 string_list() can be used to free the memory returned by a previous
2116 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2117 tively. They do nothing more than call the function pointed to by
2118 pcre_free, which of course could be called directly from a C program.
2119 However, PCRE is used in some situations where it is linked via a spe-
2120 cial interface to another programming language that cannot use
2121 pcre_free directly; it is for these cases that the functions are pro-
2122 vided.
2123
2124
2125 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2126
2127 int pcre_get_stringnumber(const pcre *code,
2128 const char *name);
2129
2130 int pcre_copy_named_substring(const pcre *code,
2131 const char *subject, int *ovector,
2132 int stringcount, const char *stringname,
2133 char *buffer, int buffersize);
2134
2135 int pcre_get_named_substring(const pcre *code,
2136 const char *subject, int *ovector,
2137 int stringcount, const char *stringname,
2138 const char **stringptr);
2139
2140 To extract a substring by name, you first have to find associated num-
2141 ber. For example, for this pattern
2142
2143 (a+)b(?<xxx>\d+)...
2144
2145 the number of the subpattern called "xxx" is 2. If the name is known to
2146 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2147 name by calling pcre_get_stringnumber(). The first argument is the com-
2148 piled pattern, and the second is the name. The yield of the function is
2149 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2150 subpattern of that name.
2151
2152 Given the number, you can extract the substring directly, or use one of
2153 the functions described in the previous section. For convenience, there
2154 are also two functions that do the whole job.
2155
2156 Most of the arguments of pcre_copy_named_substring() and
2157 pcre_get_named_substring() are the same as those for the similarly
2158 named functions that extract by number. As these are described in the
2159 previous section, they are not re-described here. There are just two
2160 differences:
2161
2162 First, instead of a substring number, a substring name is given. Sec-
2163 ond, there is an extra argument, given at the start, which is a pointer
2164 to the compiled pattern. This is needed in order to gain access to the
2165 name-to-number translation table.
2166
2167 These functions call pcre_get_stringnumber(), and if it succeeds, they
2168 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2169 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2170 behaviour may not be what you want (see the next section).
2171
2172
2173 DUPLICATE SUBPATTERN NAMES
2174
2175 int pcre_get_stringtable_entries(const pcre *code,
2176 const char *name, char **first, char **last);
2177
2178 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2179 subpatterns are not required to be unique. Normally, patterns with
2180 duplicate names are such that in any one match, only one of the named
2181 subpatterns participates. An example is shown in the pcrepattern docu-
2182 mentation. When duplicates are present, pcre_copy_named_substring() and
2183 pcre_get_named_substring() return the first substring corresponding to
2184 the given name that is set. If none are set, an empty string is
2185 returned. The pcre_get_stringnumber() function returns one of the num-
2186 bers that are associated with the name, but it is not defined which it
2187 is.
2188
2189 If you want to get full details of all captured substrings for a given
2190 name, you must use the pcre_get_stringtable_entries() function. The
2191 first argument is the compiled pattern, and the second is the name. The
2192 third and fourth are pointers to variables which are updated by the
2193 function. After it has run, they point to the first and last entries in
2194 the name-to-number table for the given name. The function itself
2195 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2196 there are none. The format of the table is described above in the sec-
2197 tion entitled Information about a pattern. Given all the relevant
2198 entries for the name, you can extract each of their numbers, and hence
2199 the captured data, if any.
2200
2201
2202 FINDING ALL POSSIBLE MATCHES
2203
2204 The traditional matching function uses a similar algorithm to Perl,
2205 which stops when it finds the first match, starting at a given point in
2206 the subject. If you want to find all possible matches, or the longest
2207 possible match, consider using the alternative matching function (see
2208 below) instead. If you cannot use the alternative function, but still
2209 need to find all possible matches, you can kludge it up by making use
2210 of the callout facility, which is described in the pcrecallout documen-
2211 tation.
2212
2213 What you have to do is to insert a callout right at the end of the pat-
2214 tern. When your callout function is called, extract and save the cur-
2215 rent matched substring. Then return 1, which forces pcre_exec() to
2216 backtrack and try other alternatives. Ultimately, when it runs out of
2217 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2218
2219
2220 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2221
2222 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2223 const char *subject, int length, int startoffset,
2224 int options, int *ovector, int ovecsize,
2225 int *workspace, int wscount);
2226
2227 The function pcre_dfa_exec() is called to match a subject string
2228 against a compiled pattern, using a matching algorithm that scans the
2229 subject string just once, and does not backtrack. This has different
2230 characteristics to the normal algorithm, and is not compatible with
2231 Perl. Some of the features of PCRE patterns are not supported. Never-
2232 theless, there are times when this kind of matching can be useful. For
2233 a discussion of the two matching algorithms, see the pcrematching docu-
2234 mentation.
2235
2236 The arguments for the pcre_dfa_exec() function are the same as for
2237 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2238 ent way, and this is described below. The other common arguments are
2239 used in the same way as for pcre_exec(), so their description is not
2240 repeated here.
2241
2242 The two additional arguments provide workspace for the function. The
2243 workspace vector should contain at least 20 elements. It is used for
2244 keeping track of multiple paths through the pattern tree. More
2245 workspace will be needed for patterns and subjects where there are a
2246 lot of potential matches.
2247
2248 Here is an example of a simple call to pcre_dfa_exec():
2249
2250 int rc;
2251 int ovector[10];
2252 int wspace[20];
2253 rc = pcre_dfa_exec(
2254 re, /* result of pcre_compile() */
2255 NULL, /* we didn't study the pattern */
2256 "some string", /* the subject string */
2257 11, /* the length of the subject string */
2258 0, /* start at offset 0 in the subject */
2259 0, /* default options */
2260 ovector, /* vector of integers for substring information */
2261 10, /* number of elements (NOT size in bytes) */
2262 wspace, /* working space vector */
2263 20); /* number of elements (NOT size in bytes) */
2264
2265 Option bits for pcre_dfa_exec()
2266
2267 The unused bits of the options argument for pcre_dfa_exec() must be
2268 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2269 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2270 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2271 three of these are the same as for pcre_exec(), so their description is
2272 not repeated here.
2273
2274 PCRE_PARTIAL
2275
2276 This has the same general effect as it does for pcre_exec(), but the
2277 details are slightly different. When PCRE_PARTIAL is set for
2278 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2279 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2280 been no complete matches, but there is still at least one matching pos-
2281 sibility. The portion of the string that provided the partial match is
2282 set as the first matching string.
2283
2284 PCRE_DFA_SHORTEST
2285
2286 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2287 stop as soon as it has found one match. Because of the way the alterna-
2288 tive algorithm works, this is necessarily the shortest possible match
2289 at the first possible matching point in the subject string.
2290
2291 PCRE_DFA_RESTART
2292
2293 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2294 returns a partial match, it is possible to call it again, with addi-
2295 tional subject characters, and have it continue with the same match.
2296 The PCRE_DFA_RESTART option requests this action; when it is set, the
2297 workspace and wscount options must reference the same vector as before
2298 because data about the match so far is left in them after a partial
2299 match. There is more discussion of this facility in the pcrepartial
2300 documentation.
2301
2302 Successful returns from pcre_dfa_exec()
2303
2304 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2305 string in the subject. Note, however, that all the matches from one run
2306 of the function start at the same point in the subject. The shorter
2307 matches are all initial substrings of the longer matches. For example,
2308 if the pattern
2309
2310 <.*>
2311
2312 is matched against the string
2313
2314 This is <something> <something else> <something further> no more
2315
2316 the three matched strings are
2317
2318 <something>
2319 <something> <something else>
2320 <something> <something else> <something further>
2321
2322 On success, the yield of the function is a number greater than zero,
2323 which is the number of matched substrings. The substrings themselves
2324 are returned in ovector. Each string uses two elements; the first is
2325 the offset to the start, and the second is the offset to the end. In
2326 fact, all the strings have the same start offset. (Space could have
2327 been saved by giving this only once, but it was decided to retain some
2328 compatibility with the way pcre_exec() returns data, even though the
2329 meaning of the strings is different.)
2330
2331 The strings are returned in reverse order of length; that is, the long-
2332 est matching string is given first. If there were too many matches to
2333 fit into ovector, the yield of the function is zero, and the vector is
2334 filled with the longest matches.
2335
2336 Error returns from pcre_dfa_exec()
2337
2338 The pcre_dfa_exec() function returns a negative number when it fails.
2339 Many of the errors are the same as for pcre_exec(), and these are
2340 described above. There are in addition the following errors that are
2341 specific to pcre_dfa_exec():
2342
2343 PCRE_ERROR_DFA_UITEM (-16)
2344
2345 This return is given if pcre_dfa_exec() encounters an item in the pat-
2346 tern that it does not support, for instance, the use of \C or a back
2347 reference.
2348
2349 PCRE_ERROR_DFA_UCOND (-17)
2350
2351 This return is given if pcre_dfa_exec() encounters a condition item
2352 that uses a back reference for the condition, or a test for recursion
2353 in a specific group. These are not supported.
2354
2355 PCRE_ERROR_DFA_UMLIMIT (-18)
2356
2357 This return is given if pcre_dfa_exec() is called with an extra block
2358 that contains a setting of the match_limit field. This is not supported
2359 (it is meaningless).
2360
2361 PCRE_ERROR_DFA_WSSIZE (-19)
2362
2363 This return is given if pcre_dfa_exec() runs out of space in the
2364 workspace vector.
2365
2366 PCRE_ERROR_DFA_RECURSE (-20)
2367
2368 When a recursive subpattern is processed, the matching function calls
2369 itself recursively, using private vectors for ovector and workspace.
2370 This error is given if the output vector is not large enough. This
2371 should be extremely rare, as a vector of size 1000 is used.
2372
2373
2374 SEE ALSO
2375
2376 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2377 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2378
2379
2380 AUTHOR
2381
2382 Philip Hazel
2383 University Computing Service
2384 Cambridge CB2 3QH, England.
2385
2386
2387 REVISION
2388
2389 Last updated: 16 April 2007
2390 Copyright (c) 1997-2007 University of Cambridge.
2391 ------------------------------------------------------------------------------
2392
2393
2394 PCRECALLOUT(3) PCRECALLOUT(3)
2395
2396
2397 NAME
2398 PCRE - Perl-compatible regular expressions
2399
2400
2401 PCRE CALLOUTS
2402
2403 int (*pcre_callout)(pcre_callout_block *);
2404
2405 PCRE provides a feature called "callout", which is a means of temporar-
2406 ily passing control to the caller of PCRE in the middle of pattern
2407 matching. The caller of PCRE provides an external function by putting
2408 its entry point in the global variable pcre_callout. By default, this
2409 variable contains NULL, which disables all calling out.
2410
2411 Within a regular expression, (?C) indicates the points at which the
2412 external function is to be called. Different callout points can be
2413 identified by putting a number less than 256 after the letter C. The
2414 default value is zero. For example, this pattern has two callout
2415 points:
2416
2417 (?C1)eabc(?C2)def
2418
2419 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2420 called, PCRE automatically inserts callouts, all with number 255,
2421 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2422 used with the pattern
2423
2424 A(\d{2}|--)
2425
2426 it is processed as if it were
2427
2428 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2429
2430 Notice that there is a callout before and after each parenthesis and
2431 alternation bar. Automatic callouts can be used for tracking the
2432 progress of pattern matching. The pcretest command has an option that
2433 sets automatic callouts; when it is used, the output indicates how the
2434 pattern is matched. This is useful information when you are trying to
2435 optimize the performance of a particular pattern.
2436
2437
2438 MISSING CALLOUTS
2439
2440 You should be aware that, because of optimizations in the way PCRE
2441 matches patterns, callouts sometimes do not happen. For example, if the
2442 pattern is
2443
2444 ab(?C4)cd
2445
2446 PCRE knows that any matching string must contain the letter "d". If the
2447 subject string is "abyz", the lack of "d" means that matching doesn't
2448 ever start, and the callout is never reached. However, with "abyd",
2449 though the result is still no match, the callout is obeyed.
2450
2451
2452 THE CALLOUT INTERFACE
2453
2454 During matching, when PCRE reaches a callout point, the external func-
2455 tion defined by pcre_callout is called (if it is set). This applies to
2456 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2457 only argument to the callout function is a pointer to a pcre_callout
2458 block. This structure contains the following fields:
2459
2460 int version;
2461 int callout_number;
2462 int *offset_vector;
2463 const char *subject;
2464 int subject_length;
2465 int start_match;
2466 int current_position;
2467 int capture_top;
2468 int capture_last;
2469 void *callout_data;
2470 int pattern_position;
2471 int next_item_length;
2472
2473 The version field is an integer containing the version number of the
2474 block format. The initial version was 0; the current version is 1. The
2475 version number will change again in future if additional fields are
2476 added, but the intention is never to remove any of the existing fields.
2477
2478 The callout_number field contains the number of the callout, as com-
2479 piled into the pattern (that is, the number after ?C for manual call-
2480 outs, and 255 for automatically generated callouts).
2481
2482 The offset_vector field is a pointer to the vector of offsets that was
2483 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2484 pcre_exec() is used, the contents can be inspected in order to extract
2485 substrings that have been matched so far, in the same way as for
2486 extracting substrings after a match has completed. For pcre_dfa_exec()
2487 this field is not useful.
2488
2489 The subject and subject_length fields contain copies of the values that
2490 were passed to pcre_exec().
2491
2492 The start_match field contains the offset within the subject at which
2493 the current match attempt started. If the pattern is not anchored, the
2494 callout function may be called several times from the same point in the
2495 pattern for different starting points in the subject.
2496
2497 The current_position field contains the offset within the subject of
2498 the current match pointer.
2499
2500 When the pcre_exec() function is used, the capture_top field contains
2501 one more than the number of the highest numbered captured substring so
2502 far. If no substrings have been captured, the value of capture_top is
2503 one. This is always the case when pcre_dfa_exec() is used, because it
2504 does not support captured substrings.
2505
2506 The capture_last field contains the number of the most recently cap-
2507 tured substring. If no substrings have been captured, its value is -1.
2508 This is always the case when pcre_dfa_exec() is used.
2509
2510 The callout_data field contains a value that is passed to pcre_exec()
2511 or pcre_dfa_exec() specifically so that it can be passed back in call-
2512 outs. It is passed in the pcre_callout field of the pcre_extra data
2513 structure. If no such data was passed, the value of callout_data in a
2514 pcre_callout block is NULL. There is a description of the pcre_extra
2515 structure in the pcreapi documentation.
2516
2517 The pattern_position field is present from version 1 of the pcre_call-
2518 out structure. It contains the offset to the next item to be matched in
2519 the pattern string.
2520
2521 The next_item_length field is present from version 1 of the pcre_call-
2522 out structure. It contains the length of the next item to be matched in
2523 the pattern string. When the callout immediately precedes an alterna-
2524 tion bar, a closing parenthesis, or the end of the pattern, the length
2525 is zero. When the callout precedes an opening parenthesis, the length
2526 is that of the entire subpattern.
2527
2528 The pattern_position and next_item_length fields are intended to help
2529 in distinguishing between different automatic callouts, which all have
2530 the same callout number. However, they are set for all callouts.
2531
2532
2533 RETURN VALUES
2534
2535 The external callout function returns an integer to PCRE. If the value
2536 is zero, matching proceeds as normal. If the value is greater than
2537 zero, matching fails at the current point, but the testing of other
2538 matching possibilities goes ahead, just as if a lookahead assertion had
2539 failed. If the value is less than zero, the match is abandoned, and
2540 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2541
2542 Negative values should normally be chosen from the set of
2543 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2544 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2545 reserved for use by callout functions; it will never be used by PCRE
2546 itself.
2547
2548
2549 AUTHOR
2550
2551 Philip Hazel
2552 University Computing Service
2553 Cambridge CB2 3QH, England.
2554
2555
2556 REVISION
2557
2558 Last updated: 06 March 2007
2559 Copyright (c) 1997-2007 University of Cambridge.
2560 ------------------------------------------------------------------------------
2561
2562
2563 PCRECOMPAT(3) PCRECOMPAT(3)
2564
2565
2566 NAME
2567 PCRE - Perl-compatible regular expressions
2568
2569
2570 DIFFERENCES BETWEEN PCRE AND PERL
2571
2572 This document describes the differences in the ways that PCRE and Perl
2573 handle regular expressions. The differences described here are mainly
2574 with respect to Perl 5.8, though PCRE version 7.0 contains some fea-
2575 tures that are expected to be in the forthcoming Perl 5.10.
2576
2577 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2578 of what it does have are given in the section on UTF-8 support in the
2579 main pcre page.
2580
2581 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2582 permits them, but they do not mean what you might think. For example,
2583 (?!a){3} does not assert that the next three characters are not "a". It
2584 just asserts that the next character is not "a" three times.
2585
2586 3. Capturing subpatterns that occur inside negative lookahead asser-
2587 tions are counted, but their entries in the offsets vector are never
2588 set. Perl sets its numerical variables from any such patterns that are
2589 matched before the assertion fails to match something (thereby succeed-
2590 ing), but only if the negative lookahead assertion contains just one
2591 branch.
2592
2593 4. Though binary zero characters are supported in the subject string,
2594 they are not allowed in a pattern string because it is passed as a nor-
2595 mal C string, terminated by zero. The escape sequence \0 can be used in
2596 the pattern to represent a binary zero.
2597
2598 5. The following Perl escape sequences are not supported: \l, \u, \L,
2599 \U, and \N. In fact these are implemented by Perl's general string-han-
2600 dling and are not part of its pattern matching engine. If any of these
2601 are encountered by PCRE, an error is generated.
2602
2603 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2604 is built with Unicode character property support. The properties that
2605 can be tested with \p and \P are limited to the general category prop-
2606 erties such as Lu and Nd, script names such as Greek or Han, and the
2607 derived properties Any and L&.
2608
2609 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2610 ters in between are treated as literals. This is slightly different
2611 from Perl in that $ and @ are also handled as literals inside the
2612 quotes. In Perl, they cause variable interpolation (but of course PCRE
2613 does not have variables). Note the following examples:
2614
2615 Pattern PCRE matches Perl matches
2616
2617 \Qabc$xyz\E abc$xyz abc followed by the
2618 contents of $xyz
2619 \Qabc\$xyz\E abc\$xyz abc\$xyz
2620 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2621
2622 The \Q...\E sequence is recognized both inside and outside character
2623 classes.
2624
2625 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2626 constructions. However, there is support for recursive patterns. This
2627 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2628 "callout" feature allows an external function to be called during pat-
2629 tern matching. See the pcrecallout documentation for details.
2630
2631 9. Subpatterns that are called recursively or as "subroutines" are
2632 always treated as atomic groups in PCRE. This is like Python, but
2633 unlike Perl.
2634
2635 10. There are some differences that are concerned with the settings of
2636 captured strings when part of a pattern is repeated. For example,
2637 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2638 unset, but in PCRE it is set to "b".
2639
2640 11. PCRE provides some extensions to the Perl regular expression facil-
2641 ities. Perl 5.10 will include new features that are not in earlier
2642 versions, some of which (such as named parentheses) have been in PCRE
2643 for some time. This list is with respect to Perl 5.10:
2644
2645 (a) Although lookbehind assertions must match fixed length strings,
2646 each alternative branch of a lookbehind assertion can match a different
2647 length of string. Perl requires them all to have the same length.
2648
2649 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2650 meta-character matches only at the very end of the string.
2651
2652 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2653 cial meaning is faulted. Otherwise, like Perl, the backslash is
2654 ignored. (Perl can be made to issue a warning.)
2655
2656 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2657 fiers is inverted, that is, by default they are not greedy, but if fol-
2658 lowed by a question mark they are.
2659
2660 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2661 tried only at the first matching position in the subject string.
2662
2663 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2664 TURE options for pcre_exec() have no Perl equivalents.
2665
2666 (g) The callout facility is PCRE-specific.
2667
2668 (h) The partial matching facility is PCRE-specific.
2669
2670 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2671 even on different hosts that have the other endianness.
2672
2673 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2674 different way and is not Perl-compatible.
2675
2676
2677 AUTHOR
2678
2679 Philip Hazel
2680 University Computing Service
2681 Cambridge CB2 3QH, England.
2682
2683
2684 REVISION
2685
2686 Last updated: 06 March 2007
2687 Copyright (c) 1997-2007 University of Cambridge.
2688 ------------------------------------------------------------------------------
2689
2690
2691 PCREPATTERN(3) PCREPATTERN(3)
2692
2693
2694 NAME
2695 PCRE - Perl-compatible regular expressions
2696
2697
2698 PCRE REGULAR EXPRESSION DETAILS
2699
2700 The syntax and semantics of the regular expressions supported by PCRE
2701 are described below. Regular expressions are also described in the Perl
2702 documentation and in a number of books, some of which have copious
2703 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2704 by O'Reilly, covers regular expressions in great detail. This descrip-
2705 tion of PCRE's regular expressions is intended as reference material.
2706
2707 The original operation of PCRE was on strings of one-byte characters.
2708 However, there is now also support for UTF-8 character strings. To use
2709 this, you must build PCRE to include UTF-8 support, and then call
2710 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2711 matching is mentioned in several places below. There is also a summary
2712 of UTF-8 features in the section on UTF-8 support in the main pcre
2713 page.
2714
2715 The remainder of this document discusses the patterns that are sup-
2716 ported by PCRE when its main matching function, pcre_exec(), is used.
2717 From release 6.0, PCRE offers a second matching function,
2718 pcre_dfa_exec(), which matches using a different algorithm that is not
2719 Perl-compatible. The advantages and disadvantages of the alternative
2720 function, and how it differs from the normal function, are discussed in
2721 the pcrematching page.
2722
2723
2724 CHARACTERS AND METACHARACTERS
2725
2726 A regular expression is a pattern that is matched against a subject
2727 string from left to right. Most characters stand for themselves in a
2728 pattern, and match the corresponding characters in the subject. As a
2729 trivial example, the pattern
2730
2731 The quick brown fox
2732
2733 matches a portion of a subject string that is identical to itself. When
2734 caseless matching is specified (the PCRE_CASELESS option), letters are
2735 matched independently of case. In UTF-8 mode, PCRE always understands
2736 the concept of case for characters whose values are less than 128, so
2737 caseless matching is always possible. For characters with higher val-
2738 ues, the concept of case is supported if PCRE is compiled with Unicode
2739 property support, but not otherwise. If you want to use caseless
2740 matching for characters 128 and above, you must ensure that PCRE is
2741 compiled with Unicode property support as well as with UTF-8 support.
2742
2743 The power of regular expressions comes from the ability to include
2744 alternatives and repetitions in the pattern. These are encoded in the
2745 pattern by the use of metacharacters, which do not stand for themselves
2746 but instead are interpreted in some special way.
2747
2748 There are two different sets of metacharacters: those that are recog-
2749 nized anywhere in the pattern except within square brackets, and those
2750 that are recognized within square brackets. Outside square brackets,
2751 the metacharacters are as follows:
2752
2753 \ general escape character with several uses
2754 ^ assert start of string (or line, in multiline mode)
2755 $ assert end of string (or line, in multiline mode)
2756 . match any character except newline (by default)
2757 [ start character class definition
2758 | start of alternative branch
2759 ( start subpattern
2760 ) end subpattern
2761 ? extends the meaning of (
2762 also 0 or 1 quantifier
2763 also quantifier minimizer
2764 * 0 or more quantifier
2765 + 1 or more quantifier
2766 also "possessive quantifier"
2767 { start min/max quantifier
2768
2769 Part of a pattern that is in square brackets is called a "character
2770 class". In a character class the only metacharacters are:
2771
2772 \ general escape character
2773 ^ negate the class, but only if the first character
2774 - indicates character range
2775 [ POSIX character class (only if followed by POSIX
2776 syntax)
2777 ] terminates the character class
2778
2779 The following sections describe the use of each of the metacharacters.
2780
2781
2782 BACKSLASH
2783
2784 The backslash character has several uses. Firstly, if it is followed by
2785 a non-alphanumeric character, it takes away any special meaning that
2786 character may have. This use of backslash as an escape character
2787 applies both inside and outside character classes.
2788
2789 For example, if you want to match a * character, you write \* in the
2790 pattern. This escaping action applies whether or not the following
2791 character would otherwise be interpreted as a metacharacter, so it is
2792 always safe to precede a non-alphanumeric with backslash to specify
2793 that it stands for itself. In particular, if you want to match a back-
2794 slash, you write \\.
2795
2796 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2797 the pattern (other than in a character class) and characters between a
2798 # outside a character class and the next newline are ignored. An escap-
2799 ing backslash can be used to include a whitespace or # character as
2800 part of the pattern.
2801
2802 If you want to remove the special meaning from a sequence of charac-
2803 ters, you can do so by putting them between \Q and \E. This is differ-
2804 ent from Perl in that $ and @ are handled as literals in \Q...\E
2805 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2806 tion. Note the following examples:
2807
2808 Pattern PCRE matches Perl matches
2809
2810 \Qabc$xyz\E abc$xyz abc followed by the
2811 contents of $xyz
2812 \Qabc\$xyz\E abc\$xyz abc\$xyz
2813 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2814
2815 The \Q...\E sequence is recognized both inside and outside character
2816 classes.
2817
2818 Non-printing characters
2819
2820 A second use of backslash provides a way of encoding non-printing char-
2821 acters in patterns in a visible manner. There is no restriction on the
2822 appearance of non-printing characters, apart from the binary zero that
2823 terminates a pattern, but when a pattern is being prepared by text
2824 editing, it is usually easier to use one of the following escape
2825 sequences than the binary character it represents:
2826
2827 \a alarm, that is, the BEL character (hex 07)
2828 \cx "control-x", where x is any character
2829 \e escape (hex 1B)
2830 \f formfeed (hex 0C)
2831 \n newline (hex 0A)
2832 \r carriage return (hex 0D)
2833 \t tab (hex 09)
2834 \ddd character with octal code ddd, or backreference
2835 \xhh character with hex code hh
2836 \x{hhh..} character with hex code hhh..
2837
2838 The precise effect of \cx is as follows: if x is a lower case letter,
2839 it is converted to upper case. Then bit 6 of the character (hex 40) is
2840 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2841 becomes hex 7B.
2842
2843 After \x, from zero to two hexadecimal digits are read (letters can be
2844 in upper or lower case). Any number of hexadecimal digits may appear
2845 between \x{ and }, but the value of the character code must be less
2846 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2847 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2848 hexadecimal digits appear between \x{ and }, or if there is no termi-
2849 nating }, this form of escape is not recognized. Instead, the initial
2850 \x will be interpreted as a basic hexadecimal escape, with no following
2851 digits, giving a character whose value is zero.
2852
2853 Characters whose value is less than 256 can be defined by either of the
2854 two syntaxes for \x. There is no difference in the way they are han-
2855 dled. For example, \xdc is exactly the same as \x{dc}.
2856
2857 After \0 up to two further octal digits are read. If there are fewer
2858 than two digits, just those that are present are used. Thus the
2859 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2860 (code value 7). Make sure you supply two digits after the initial zero
2861 if the pattern character that follows is itself an octal digit.
2862
2863 The handling of a backslash followed by a digit other than 0 is compli-
2864 cated. Outside a character class, PCRE reads it and any following dig-
2865 its as a decimal number. If the number is less than 10, or if there
2866 have been at least that many previous capturing left parentheses in the
2867 expression, the entire sequence is taken as a back reference. A
2868 description of how this works is given later, following the discussion
2869 of parenthesized subpatterns.
2870
2871 Inside a character class, or if the decimal number is greater than 9
2872 and there have not been that many capturing subpatterns, PCRE re-reads
2873 up to three octal digits following the backslash, and uses them to gen-
2874 erate a data character. Any subsequent digits stand for themselves. In
2875 non-UTF-8 mode, the value of a character specified in octal must be
2876 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2877 example:
2878
2879 \040 is another way of writing a space
2880 \40 is the same, provided there are fewer than 40
2881 previous capturing subpatterns
2882 \7 is always a back reference
2883 \11 might be a back reference, or another way of
2884 writing a tab
2885 \011 is always a tab
2886 \0113 is a tab followed by the character "3"
2887 \113 might be a back reference, otherwise the
2888 character with octal code 113
2889 \377 might be a back reference, otherwise
2890 the byte consisting entirely of 1 bits
2891 \81 is either a back reference, or a binary zero
2892 followed by the two characters "8" and "1"
2893
2894 Note that octal values of 100 or greater must not be introduced by a
2895 leading zero, because no more than three octal digits are ever read.
2896
2897 All the sequences that define a single character value can be used both
2898 inside and outside character classes. In addition, inside a character
2899 class, the sequence \b is interpreted as the backspace character (hex
2900 08), and the sequences \R and \X are interpreted as the characters "R"
2901 and "X", respectively. Outside a character class, these sequences have
2902 different meanings (see below).
2903
2904 Absolute and relative back references
2905
2906 The sequence \g followed by a positive or negative number, optionally
2907 enclosed in braces, is an absolute or relative back reference. Back
2908 references are discussed later, following the discussion of parenthe-
2909 sized subpatterns.
2910
2911 Generic character types
2912
2913 Another use of backslash is for specifying generic character types. The
2914 following are always recognized:
2915
2916 \d any decimal digit
2917 \D any character that is not a decimal digit
2918 \s any whitespace character
2919 \S any character that is not a whitespace character
2920 \w any "word" character
2921 \W any "non-word" character
2922
2923 Each pair of escape sequences partitions the complete set of characters
2924 into two disjoint sets. Any given character matches one, and only one,
2925 of each pair.
2926
2927 These character type sequences can appear both inside and outside char-
2928 acter classes. They each match one character of the appropriate type.
2929 If the current matching point is at the end of the subject string, all
2930 of them fail, since there is no character to match.
2931
2932 For compatibility with Perl, \s does not match the VT character (code
2933 11). This makes it different from the the POSIX "space" class. The \s
2934 characters are HT (9), LF (10), FF (12), CR (13), and space (32). (If
2935 "use locale;" is included in a Perl script, \s may match the VT charac-
2936 ter. In PCRE, it never does.)
2937
2938 A "word" character is an underscore or any character less than 256 that
2939 is a letter or digit. The definition of letters and digits is con-
2940 trolled by PCRE's low-valued character tables, and may vary if locale-
2941 specific matching is taking place (see "Locale support" in the pcreapi
2942 page). For example, in a French locale such as "fr_FR" in Unix-like
2943 systems, or "french" in Windows, some character codes greater than 128
2944 are used for accented letters, and these are matched by \w.
2945
2946 In UTF-8 mode, characters with values greater than 128 never match \d,
2947 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2948 code character property support is available. The use of locales with
2949 Unicode is discouraged.
2950
2951 Newline sequences
2952
2953 Outside a character class, the escape sequence \R matches any Unicode
2954 newline sequence. This is an extension to Perl. In non-UTF-8 mode \R is
2955 equivalent to the following:
2956
2957 (?>\r\n|\n|\x0b|\f|\r|\x85)
2958
2959 This is an example of an "atomic group", details of which are given
2960 below. This particular group matches either the two-character sequence
2961 CR followed by LF, or one of the single characters LF (linefeed,
2962 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
2963 return, U+000D), or NEL (next line, U+0085). The two-character sequence
2964 is treated as a single unit that cannot be split.
2965
2966 In UTF-8 mode, two additional characters whose codepoints are greater
2967 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
2968 rator, U+2029). Unicode character property support is not needed for
2969 these characters to be recognized.
2970
2971 Inside a character class, \R matches the letter "R".
2972
2973 Unicode character properties
2974
2975 When PCRE is built with Unicode character property support, three addi-
2976 tional escape sequences to match character properties are available
2977 when UTF-8 mode is selected. They are:
2978
2979 \p{xx} a character with the xx property
2980 \P{xx} a character without the xx property
2981 \X an extended Unicode sequence
2982
2983 The property names represented by xx above are limited to the Unicode
2984 script names, the general category properties, and "Any", which matches
2985 any character (including newline). Other properties such as "InMusical-
2986 Symbols" are not currently supported by PCRE. Note that \P{Any} does
2987 not match any characters, so always causes a match failure.
2988
2989 Sets of Unicode characters are defined as belonging to certain scripts.
2990 A character from one of these sets can be matched using a script name.
2991 For example:
2992
2993 \p{Greek}
2994 \P{Han}
2995
2996 Those that are not part of an identified script are lumped together as
2997 "Common". The current list of scripts is:
2998
2999 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3000 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3001 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3002 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3003 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3004 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3005 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3006 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3007 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3008
3009 Each character has exactly one general category property, specified by
3010 a two-letter abbreviation. For compatibility with Perl, negation can be
3011 specified by including a circumflex between the opening brace and the
3012 property name. For example, \p{^Lu} is the same as \P{Lu}.
3013
3014 If only one letter is specified with \p or \P, it includes all the gen-
3015 eral category properties that start with that letter. In this case, in
3016 the absence of negation, the curly brackets in the escape sequence are
3017 optional; these two examples have the same effect:
3018
3019 \p{L}
3020 \pL
3021
3022 The following general category property codes are supported:
3023
3024 C Other
3025 Cc Control
3026 Cf Format
3027 Cn Unassigned
3028 Co Private use
3029 Cs Surrogate
3030
3031 L Letter
3032 Ll Lower case letter
3033 Lm Modifier letter
3034 Lo Other letter
3035 Lt Title case letter
3036 Lu Upper case letter
3037
3038 M Mark
3039 Mc Spacing mark
3040 Me Enclosing mark
3041 Mn Non-spacing mark
3042
3043 N Number
3044 Nd Decimal number
3045 Nl Letter number
3046 No Other number
3047
3048 P Punctuation
3049 Pc Connector punctuation
3050 Pd Dash punctuation
3051 Pe Close punctuation
3052 Pf Final punctuation
3053 Pi Initial punctuation
3054 Po Other punctuation
3055 Ps Open punctuation
3056
3057 S Symbol
3058 Sc Currency symbol
3059 Sk Modifier symbol
3060 Sm Mathematical symbol
3061 So Other symbol
3062
3063 Z Separator
3064 Zl Line separator
3065 Zp Paragraph separator
3066 Zs Space separator
3067
3068 The special property L& is also supported: it matches a character that
3069 has the Lu, Ll, or Lt property, in other words, a letter that is not
3070 classified as a modifier or "other".
3071
3072 The long synonyms for these properties that Perl supports (such as
3073 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3074 any of these properties with "Is".
3075
3076 No character that is in the Unicode table has the Cn (unassigned) prop-
3077 erty. Instead, this property is assumed for any code point that is not
3078 in the Unicode table.
3079
3080 Specifying caseless matching does not affect these escape sequences.
3081 For example, \p{Lu} always matches only upper case letters.
3082
3083 The \X escape matches any number of Unicode characters that form an
3084 extended Unicode sequence. \X is equivalent to
3085
3086 (?>\PM\pM*)
3087
3088 That is, it matches a character without the "mark" property, followed
3089 by zero or more characters with the "mark" property, and treats the
3090 sequence as an atomic group (see below). Characters with the "mark"
3091 property are typically accents that affect the preceding character.
3092
3093 Matching characters by Unicode property is not fast, because PCRE has
3094 to search a structure that contains data for over fifteen thousand
3095 characters. That is why the traditional escape sequences such as \d and
3096 \w do not use Unicode properties in PCRE.
3097
3098 Simple assertions
3099
3100 The final use of backslash is for certain simple assertions. An asser-
3101 tion specifies a condition that has to be met at a particular point in
3102 a match, without consuming any characters from the subject string. The
3103 use of subpatterns for more complicated assertions is described below.
3104 The backslashed assertions are:
3105
3106 \b matches at a word boundary
3107 \B matches when not at a word boundary
3108 \A matches at the start of the subject
3109 \Z matches at the end of the subject
3110 also matches before a newline at the end of the subject
3111 \z matches only at the end of the subject
3112 \G matches at the first matching position in the subject
3113
3114 These assertions may not appear in character classes (but note that \b
3115 has a different meaning, namely the backspace character, inside a char-
3116 acter class).
3117
3118 A word boundary is a position in the subject string where the current
3119 character and the previous character do not both match \w or \W (i.e.
3120 one matches \w and the other matches \W), or the start or end of the
3121 string if the first or last character matches \w, respectively.
3122
3123 The \A, \Z, and \z assertions differ from the traditional circumflex
3124 and dollar (described in the next section) in that they only ever match
3125 at the very start and end of the subject string, whatever options are
3126 set. Thus, they are independent of multiline mode. These three asser-
3127 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3128 affect only the behaviour of the circumflex and dollar metacharacters.
3129 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3130 cating that matching is to start at a point other than the beginning of
3131 the subject, \A can never match. The difference between \Z and \z is
3132 that \Z matches before a newline at the end of the string as well as at
3133 the very end, whereas \z matches only at the end.
3134
3135 The \G assertion is true only when the current matching position is at
3136 the start point of the match, as specified by the startoffset argument
3137 of pcre_exec(). It differs from \A when the value of startoffset is
3138 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3139 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3140 mentation where \G can be useful.
3141
3142 Note, however, that PCRE's interpretation of \G, as the start of the
3143 current match, is subtly different from Perl's, which defines it as the
3144 end of the previous match. In Perl, these can be different when the
3145 previously matched string was empty. Because PCRE does just one match
3146 at a time, it cannot reproduce this behaviour.
3147
3148 If all the alternatives of a pattern begin with \G, the expression is
3149 anchored to the starting match position, and the "anchored" flag is set
3150 in the compiled regular expression.
3151
3152
3153 CIRCUMFLEX AND DOLLAR
3154
3155 Outside a character class, in the default matching mode, the circumflex
3156 character is an assertion that is true only if the current matching
3157 point is at the start of the subject string. If the startoffset argu-
3158 ment of pcre_exec() is non-zero, circumflex can never match if the
3159 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3160 has an entirely different meaning (see below).
3161
3162 Circumflex need not be the first character of the pattern if a number
3163 of alternatives are involved, but it should be the first thing in each
3164 alternative in which it appears if the pattern is ever to match that
3165 branch. If all possible alternatives start with a circumflex, that is,
3166 if the pattern is constrained to match only at the start of the sub-
3167 ject, it is said to be an "anchored" pattern. (There are also other
3168 constructs that can cause a pattern to be anchored.)
3169
3170 A dollar character is an assertion that is true only if the current
3171 matching point is at the end of the subject string, or immediately
3172 before a newline at the end of the string (by default). Dollar need not
3173 be the last character of the pattern if a number of alternatives are
3174 involved, but it should be the last item in any branch in which it
3175 appears. Dollar has no special meaning in a character class.
3176
3177 The meaning of dollar can be changed so that it matches only at the
3178 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3179 compile time. This does not affect the \Z assertion.
3180
3181 The meanings of the circumflex and dollar characters are changed if the
3182 PCRE_MULTILINE option is set. When this is the case, a circumflex
3183 matches immediately after internal newlines as well as at the start of
3184 the subject string. It does not match after a newline that ends the
3185 string. A dollar matches before any newlines in the string, as well as
3186 at the very end, when PCRE_MULTILINE is set. When newline is specified
3187 as the two-character sequence CRLF, isolated CR and LF characters do
3188 not indicate newlines.
3189
3190 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3191 (where \n represents a newline) in multiline mode, but not otherwise.
3192 Consequently, patterns that are anchored in single line mode because
3193 all branches start with ^ are not anchored in multiline mode, and a
3194 match for circumflex is possible when the startoffset argument of
3195 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3196 PCRE_MULTILINE is set.
3197
3198 Note that the sequences \A, \Z, and \z can be used to match the start
3199 and end of the subject in both modes, and if all branches of a pattern
3200 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3201 set.
3202
3203
3204 FULL STOP (PERIOD, DOT)
3205
3206 Outside a character class, a dot in the pattern matches any one charac-
3207 ter in the subject string except (by default) a character that signi-
3208 fies the end of a line. In UTF-8 mode, the matched character may be
3209 more than one byte long.
3210
3211 When a line ending is defined as a single character, dot never matches
3212 that character; when the two-character sequence CRLF is used, dot does
3213 not match CR if it is immediately followed by LF, but otherwise it
3214 matches all characters (including isolated CRs and LFs). When any Uni-
3215 code line endings are being recognized, dot does not match CR or LF or
3216 any of the other line ending characters.
3217
3218 The behaviour of dot with regard to newlines can be changed. If the
3219 PCRE_DOTALL option is set, a dot matches any one character, without
3220 exception. If the two-character sequence CRLF is present in the subject
3221 string, it takes two dots to match it.
3222
3223 The handling of dot is entirely independent of the handling of circum-
3224 flex and dollar, the only relationship being that they both involve
3225 newlines. Dot has no special meaning in a character class.
3226
3227
3228 MATCHING A SINGLE BYTE
3229
3230 Outside a character class, the escape sequence \C matches any one byte,
3231 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3232 line-ending characters. The feature is provided in Perl in order to
3233 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3234 acters into individual bytes, what remains in the string may be a mal-
3235 formed UTF-8 string. For this reason, the \C escape sequence is best
3236 avoided.
3237
3238 PCRE does not allow \C to appear in lookbehind assertions (described
3239 below), because in UTF-8 mode this would make it impossible to calcu-
3240 late the length of the lookbehind.
3241
3242
3243 SQUARE BRACKETS AND CHARACTER CLASSES
3244
3245 An opening square bracket introduces a character class, terminated by a
3246 closing square bracket. A closing square bracket on its own is not spe-
3247 cial. If a closing square bracket is required as a member of the class,
3248 it should be the first data character in the class (after an initial
3249 circumflex, if present) or escaped with a backslash.
3250
3251 A character class matches a single character in the subject. In UTF-8
3252 mode, the character may occupy more than one byte. A matched character
3253 must be in the set of characters defined by the class, unless the first
3254 character in the class definition is a circumflex, in which case the
3255 subject character must not be in the set defined by the class. If a
3256 circumflex is actually required as a member of the class, ensure it is
3257 not the first character, or escape it with a backslash.
3258
3259 For example, the character class [aeiou] matches any lower case vowel,
3260 while [^aeiou] matches any character that is not a lower case vowel.
3261 Note that a circumflex is just a convenient notation for specifying the
3262 characters that are in the class by enumerating those that are not. A
3263 class that starts with a circumflex is not an assertion: it still con-
3264 sumes a character from the subject string, and therefore it fails if
3265 the current pointer is at the end of the string.
3266
3267 In UTF-8 mode, characters with values greater than 255 can be included
3268 in a class as a literal string of bytes, or by using the \x{ escaping
3269 mechanism.
3270
3271 When caseless matching is set, any letters in a class represent both
3272 their upper case and lower case versions, so for example, a caseless
3273 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3274 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3275 understands the concept of case for characters whose values are less
3276 than 128, so caseless matching is always possible. For characters with
3277 higher values, the concept of case is supported if PCRE is compiled
3278 with Unicode property support, but not otherwise. If you want to use
3279 caseless matching for characters 128 and above, you must ensure that
3280 PCRE is compiled with Unicode property support as well as with UTF-8
3281 support.
3282
3283 Characters that might indicate line breaks are never treated in any
3284 special way when matching character classes, whatever line-ending
3285 sequence is in use, and whatever setting of the PCRE_DOTALL and
3286 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3287 of these characters.
3288
3289 The minus (hyphen) character can be used to specify a range of charac-
3290 ters in a character class. For example, [d-m] matches any letter
3291 between d and m, inclusive. If a minus character is required in a
3292 class, it must be escaped with a backslash or appear in a position
3293 where it cannot be interpreted as indicating a range, typically as the
3294 first or last character in the class.
3295
3296 It is not possible to have the literal character "]" as the end charac-
3297 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3298 two characters ("W" and "-") followed by a literal string "46]", so it
3299 would match "W46]" or "-46]". However, if the "]" is escaped with a
3300 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3301 preted as a class containing a range followed by two other characters.
3302 The octal or hexadecimal representation of "]" can also be used to end
3303 a range.
3304
3305 Ranges operate in the collating sequence of character values. They can
3306 also be used for characters specified numerically, for example
3307 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3308 are greater than 255, for example [\x{100}-\x{2ff}].
3309
3310 If a range that includes letters is used when caseless matching is set,
3311 it matches the letters in either case. For example, [W-c] is equivalent
3312 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3313 character tables for a French locale are in use, [\xc8-\xcb] matches
3314 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3315 concept of case for characters with values greater than 128 only when
3316 it is compiled with Unicode property support.
3317
3318 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3319 in a character class, and add the characters that they match to the
3320 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3321 flex can conveniently be used with the upper case character types to
3322 specify a more restricted set of characters than the matching lower
3323 case type. For example, the class [^\W_] matches any letter or digit,
3324 but not underscore.
3325
3326 The only metacharacters that are recognized in character classes are
3327 backslash, hyphen (only where it can be interpreted as specifying a
3328 range), circumflex (only at the start), opening square bracket (only
3329 when it can be interpreted as introducing a POSIX class name - see the
3330 next section), and the terminating closing square bracket. However,
3331 escaping other non-alphanumeric characters does no harm.
3332
3333
3334 POSIX CHARACTER CLASSES
3335
3336 Perl supports the POSIX notation for character classes. This uses names
3337 enclosed by [: and :] within the enclosing square brackets. PCRE also
3338 supports this notation. For example,
3339
3340 [01[:alpha:]%]
3341
3342 matches "0", "1", any alphabetic character, or "%". The supported class
3343 names are
3344
3345 alnum letters and digits
3346 alpha letters
3347 ascii character codes 0 - 127
3348 blank space or tab only
3349 cntrl control characters
3350 digit decimal digits (same as \d)
3351 graph printing characters, excluding space
3352 lower lower case letters
3353 print printing characters, including space
3354 punct printing characters, excluding letters and digits
3355 space white space (not quite the same as \s)
3356 upper upper case letters
3357 word "word" characters (same as \w)
3358 xdigit hexadecimal digits
3359
3360 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3361 and space (32). Notice that this list includes the VT character (code
3362 11). This makes "space" different to \s, which does not include VT (for
3363 Perl compatibility).
3364
3365 The name "word" is a Perl extension, and "blank" is a GNU extension
3366 from Perl 5.8. Another Perl extension is negation, which is indicated
3367 by a ^ character after the colon. For example,
3368
3369 [12[:^digit:]]
3370
3371 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3372 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3373 these are not supported, and an error is given if they are encountered.
3374
3375 In UTF-8 mode, characters with values greater than 128 do not match any
3376 of the POSIX character classes.
3377
3378
3379 VERTICAL BAR
3380
3381 Vertical bar characters are used to separate alternative patterns. For
3382 example, the pattern
3383
3384 gilbert|sullivan
3385
3386 matches either "gilbert" or "sullivan". Any number of alternatives may
3387 appear, and an empty alternative is permitted (matching the empty
3388 string). The matching process tries each alternative in turn, from left
3389 to right, and the first one that succeeds is used. If the alternatives
3390 are within a subpattern (defined below), "succeeds" means matching the
3391 rest of the main pattern as well as the alternative in the subpattern.
3392
3393
3394 INTERNAL OPTION SETTING
3395
3396 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3397 PCRE_EXTENDED options can be changed from within the pattern by a
3398 sequence of Perl option letters enclosed between "(?" and ")". The
3399 option letters are
3400
3401 i for PCRE_CASELESS
3402 m for PCRE_MULTILINE
3403 s for PCRE_DOTALL
3404 x for PCRE_EXTENDED
3405
3406 For example, (?im) sets caseless, multiline matching. It is also possi-
3407 ble to unset these options by preceding the letter with a hyphen, and a
3408 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3409 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3410 is also permitted. If a letter appears both before and after the
3411 hyphen, the option is unset.
3412
3413 When an option change occurs at top level (that is, not inside subpat-
3414 tern parentheses), the change applies to the remainder of the pattern
3415 that follows. If the change is placed right at the start of a pattern,
3416 PCRE extracts it into the global options (and it will therefore show up
3417 in data extracted by the pcre_fullinfo() function).
3418
3419 An option change within a subpattern (see below for a description of
3420 subpatterns) affects only that part of the current pattern that follows
3421 it, so
3422
3423 (a(?i)b)c
3424
3425 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3426 used). By this means, options can be made to have different settings
3427 in different parts of the pattern. Any changes made in one alternative
3428 do carry on into subsequent branches within the same subpattern. For
3429 example,
3430
3431 (a(?i)b|c)
3432
3433 matches "ab", "aB", "c", and "C", even though when matching "C" the
3434 first branch is abandoned before the option setting. This is because
3435 the effects of option settings happen at compile time. There would be
3436 some very weird behaviour otherwise.
3437
3438 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3439 can be changed in the same way as the Perl-compatible options by using
3440 the characters J, U and X respectively.
3441
3442
3443 SUBPATTERNS
3444
3445 Subpatterns are delimited by parentheses (round brackets), which can be
3446 nested. Turning part of a pattern into a subpattern does two things:
3447
3448 1. It localizes a set of alternatives. For example, the pattern
3449
3450 cat(aract|erpillar|)
3451
3452 matches one of the words "cat", "cataract", or "caterpillar". Without
3453 the parentheses, it would match "cataract", "erpillar" or an empty
3454 string.
3455
3456 2. It sets up the subpattern as a capturing subpattern. This means
3457 that, when the whole pattern matches, that portion of the subject
3458 string that matched the subpattern is passed back to the caller via the
3459 ovector argument of pcre_exec(). Opening parentheses are counted from
3460 left to right (starting from 1) to obtain numbers for the capturing
3461 subpatterns.
3462
3463 For example, if the string "the red king" is matched against the pat-
3464 tern
3465
3466 the ((red|white) (king|queen))
3467
3468 the captured substrings are "red king", "red", and "king", and are num-
3469 bered 1, 2, and 3, respectively.
3470
3471 The fact that plain parentheses fulfil two functions is not always
3472 helpful. There are often times when a grouping subpattern is required
3473 without a capturing requirement. If an opening parenthesis is followed
3474 by a question mark and a colon, the subpattern does not do any captur-
3475 ing, and is not counted when computing the number of any subsequent
3476 capturing subpatterns. For example, if the string "the white queen" is
3477 matched against the pattern
3478
3479 the ((?:red|white) (king|queen))
3480
3481 the captured substrings are "white queen" and "queen", and are numbered
3482 1 and 2. The maximum number of capturing subpatterns is 65535.
3483
3484 As a convenient shorthand, if any option settings are required at the
3485 start of a non-capturing subpattern, the option letters may appear
3486 between the "?" and the ":". Thus the two patterns
3487
3488 (?i:saturday|sunday)
3489 (?:(?i)saturday|sunday)
3490
3491 match exactly the same set of strings. Because alternative branches are
3492 tried from left to right, and options are not reset until the end of
3493 the subpattern is reached, an option setting in one branch does affect
3494 subsequent branches, so the above patterns match "SUNDAY" as well as
3495 "Saturday".
3496
3497
3498 NAMED SUBPATTERNS
3499
3500 Identifying capturing parentheses by number is simple, but it can be
3501 very hard to keep track of the numbers in complicated regular expres-
3502 sions. Furthermore, if an expression is modified, the numbers may
3503 change. To help with this difficulty, PCRE supports the naming of sub-
3504 patterns. This feature was not added to Perl until release 5.10. Python
3505 had the feature earlier, and PCRE introduced it at release 4.0, using
3506 the Python syntax. PCRE now supports both the Perl and the Python syn-
3507 tax.
3508
3509 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3510 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3511 to capturing parentheses from other parts of the pattern, such as back-
3512 references, recursion, and conditions, can be made by name as well as
3513 by number.
3514
3515 Names consist of up to 32 alphanumeric characters and underscores.
3516 Named capturing parentheses are still allocated numbers as well as
3517 names, exactly as if the names were not present. The PCRE API provides
3518 function calls for extracting the name-to-number translation table from
3519 a compiled pattern. There is also a convenience function for extracting
3520 a captured substring by name.
3521
3522 By default, a name must be unique within a pattern, but it is possible
3523 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3524 time. This can be useful for patterns where only one instance of the
3525 named parentheses can match. Suppose you want to match the name of a
3526 weekday, either as a 3-letter abbreviation or as the full name, and in
3527 both cases you want to extract the abbreviation. This pattern (ignoring
3528 the line breaks) does the job:
3529
3530 (?<DN>Mon|Fri|Sun)(?:day)?|
3531 (?<DN>Tue)(?:sday)?|
3532 (?<DN>Wed)(?:nesday)?|
3533 (?<DN>Thu)(?:rsday)?|
3534 (?<DN>Sat)(?:urday)?
3535
3536 There are five capturing substrings, but only one is ever set after a
3537 match. The convenience function for extracting the data by name
3538 returns the substring for the first (and in this example, the only)
3539 subpattern of that name that matched. This saves searching to find
3540 which numbered subpattern it was. If you make a reference to a non-
3541 unique named subpattern from elsewhere in the pattern, the one that
3542 corresponds to the lowest number is used. For further details of the
3543 interfaces for handling named subpatterns, see the pcreapi documenta-
3544 tion.
3545
3546
3547 REPETITION
3548
3549 Repetition is specified by quantifiers, which can follow any of the
3550 following items:
3551
3552 a literal data character
3553 the dot metacharacter
3554 the \C escape sequence
3555 the \X escape sequence (in UTF-8 mode with Unicode properties)
3556 the \R escape sequence
3557 an escape such as \d that matches a single character
3558 a character class
3559 a back reference (see next section)
3560 a parenthesized subpattern (unless it is an assertion)
3561
3562 The general repetition quantifier specifies a minimum and maximum num-
3563 ber of permitted matches, by giving the two numbers in curly brackets
3564 (braces), separated by a comma. The numbers must be less than 65536,
3565 and the first must be less than or equal to the second. For example:
3566
3567 z{2,4}
3568
3569 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3570 special character. If the second number is omitted, but the comma is
3571 present, there is no upper limit; if the second number and the comma
3572 are both omitted, the quantifier specifies an exact number of required
3573 matches. Thus
3574
3575 [aeiou]{3,}
3576
3577 matches at least 3 successive vowels, but may match many more, while
3578
3579 \d{8}
3580
3581 matches exactly 8 digits. An opening curly bracket that appears in a
3582 position where a quantifier is not allowed, or one that does not match
3583 the syntax of a quantifier, is taken as a literal character. For exam-
3584 ple, {,6} is not a quantifier, but a literal string of four characters.
3585
3586 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3587 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3588 acters, each of which is represented by a two-byte sequence. Similarly,
3589 when Unicode property support is available, \X{3} matches three Unicode
3590 extended sequences, each of which may be several bytes long (and they
3591 may be of different lengths).
3592
3593 The quantifier {0} is permitted, causing the expression to behave as if
3594 the previous item and the quantifier were not present.
3595
3596 For convenience, the three most common quantifiers have single-charac-
3597 ter abbreviations:
3598
3599 * is equivalent to {0,}
3600 + is equivalent to {1,}
3601 ? is equivalent to {0,1}
3602
3603 It is possible to construct infinite loops by following a subpattern
3604 that can match no characters with a quantifier that has no upper limit,
3605 for example:
3606
3607 (a?)*
3608
3609 Earlier versions of Perl and PCRE used to give an error at compile time
3610 for such patterns. However, because there are cases where this can be
3611 useful, such patterns are now accepted, but if any repetition of the
3612 subpattern does in fact match no characters, the loop is forcibly bro-
3613 ken.
3614
3615 By default, the quantifiers are "greedy", that is, they match as much
3616 as possible (up to the maximum number of permitted times), without
3617 causing the rest of the pattern to fail. The classic example of where
3618 this gives problems is in trying to match comments in C programs. These
3619 appear between /* and */ and within the comment, individual * and /
3620 characters may appear. An attempt to match C comments by applying the
3621 pattern
3622
3623 /\*.*\*/
3624
3625 to the string
3626
3627 /* first comment */ not comment /* second comment */
3628
3629 fails, because it matches the entire string owing to the greediness of
3630 the .* item.
3631
3632 However, if a quantifier is followed by a question mark, it ceases to
3633 be greedy, and instead matches the minimum number of times possible, so
3634 the pattern
3635
3636 /\*.*?\*/
3637
3638 does the right thing with the C comments. The meaning of the various
3639 quantifiers is not otherwise changed, just the preferred number of
3640 matches. Do not confuse this use of question mark with its use as a
3641 quantifier in its own right. Because it has two uses, it can sometimes
3642 appear doubled, as in
3643
3644 \d??\d
3645
3646 which matches one digit by preference, but can match two if that is the
3647 only way the rest of the pattern matches.
3648
3649 If the PCRE_UNGREEDY option is set (an option that is not available in
3650 Perl), the quantifiers are not greedy by default, but individual ones
3651 can be made greedy by following them with a question mark. In other
3652 words, it inverts the default behaviour.
3653
3654 When a parenthesized subpattern is quantified with a minimum repeat
3655 count that is greater than 1 or with a limited maximum, more memory is
3656 required for the compiled pattern, in proportion to the size of the
3657 minimum or maximum.
3658
3659 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3660 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3661 the pattern is implicitly anchored, because whatever follows will be
3662 tried against every character position in the subject string, so there
3663 is no point in retrying the overall match at any position after the
3664 first. PCRE normally treats such a pattern as though it were preceded
3665 by \A.
3666
3667 In cases where it is known that the subject string contains no new-
3668 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3669 mization, or alternatively using ^ to indicate anchoring explicitly.
3670
3671 However, there is one situation where the optimization cannot be used.
3672 When .* is inside capturing parentheses that are the subject of a
3673 backreference elsewhere in the pattern, a match at the start may fail
3674 where a later one succeeds. Consider, for example:
3675
3676 (.*)abc\1
3677
3678 If the subject is "xyz123abc123" the match point is the fourth charac-
3679 ter. For this reason, such a pattern is not implicitly anchored.
3680
3681 When a capturing subpattern is repeated, the value captured is the sub-
3682 string that matched the final iteration. For example, after
3683
3684 (tweedle[dume]{3}\s*)+
3685
3686 has matched "tweedledum tweedledee" the value of the captured substring
3687 is "tweedledee". However, if there are nested capturing subpatterns,
3688 the corresponding captured values may have been set in previous itera-
3689 tions. For example, after
3690
3691 /(a|(b))+/
3692
3693 matches "aba" the value of the second captured substring is "b".
3694
3695
3696 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3697
3698 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3699 repetition, failure of what follows normally causes the repeated item
3700 to be re-evaluated to see if a different number of repeats allows the
3701 rest of the pattern to match. Sometimes it is useful to prevent this,
3702 either to change the nature of the match, or to cause it fail earlier
3703 than it otherwise might, when the author of the pattern knows there is
3704 no point in carrying on.
3705
3706 Consider, for example, the pattern \d+foo when applied to the subject
3707 line
3708
3709 123456bar
3710
3711 After matching all 6 digits and then failing to match "foo", the normal
3712 action of the matcher is to try again with only 5 digits matching the
3713 \d+ item, and then with 4, and so on, before ultimately failing.
3714 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3715 the means for specifying that once a subpattern has matched, it is not
3716 to be re-evaluated in this way.
3717
3718 If we use atomic grouping for the previous example, the matcher gives
3719 up immediately on failing to match "foo" the first time. The notation
3720 is a kind of special parenthesis, starting with (?> as in this example:
3721
3722 (?>\d+)foo
3723
3724 This kind of parenthesis "locks up" the part of the pattern it con-
3725 tains once it has matched, and a failure further into the pattern is
3726 prevented from backtracking into it. Backtracking past it to previous
3727 items, however, works as normal.
3728
3729 An alternative description is that a subpattern of this type matches
3730 the string of characters that an identical standalone pattern would
3731 match, if anchored at the current point in the subject string.
3732
3733 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3734 such as the above example can be thought of as a maximizing repeat that
3735 must swallow everything it can. So, while both \d+ and \d+? are pre-
3736 pared to adjust the number of digits they match in order to make the
3737 rest of the pattern match, (?>\d+) can only match an entire sequence of
3738 digits.
3739
3740 Atomic groups in general can of course contain arbitrarily complicated
3741 subpatterns, and can be nested. However, when the subpattern for an
3742 atomic group is just a single repeated item, as in the example above, a
3743 simpler notation, called a "possessive quantifier" can be used. This
3744 consists of an additional + character following a quantifier. Using
3745 this notation, the previous example can be rewritten as
3746
3747 \d++foo
3748
3749 Possessive quantifiers are always greedy; the setting of the
3750 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3751 simpler forms of atomic group. However, there is no difference in the
3752 meaning of a possessive quantifier and the equivalent atomic group,
3753 though there may be a performance difference; possessive quantifiers
3754 should be slightly faster.
3755
3756 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3757 tax. Jeffrey Friedl originated the idea (and the name) in the first
3758 edition of his book. Mike McCloskey liked it, so implemented it when he
3759 built Sun's Java package, and PCRE copied it from there. It ultimately
3760 found its way into Perl at release 5.10.
3761
3762 PCRE has an optimization that automatically "possessifies" certain sim-
3763 ple pattern constructs. For example, the sequence A+B is treated as
3764 A++B because there is no point in backtracking into a sequence of A's
3765 when B must follow.
3766
3767 When a pattern contains an unlimited repeat inside a subpattern that
3768 can itself be repeated an unlimited number of times, the use of an
3769 atomic group is the only way to avoid some failing matches taking a
3770 very long time indeed. The pattern
3771
3772 (\D+|<\d+>)*[!?]
3773
3774 matches an unlimited number of substrings that either consist of non-
3775 digits, or digits enclosed in <>, followed by either ! or ?. When it
3776 matches, it runs quickly. However, if it is applied to
3777
3778 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3779
3780 it takes a long time before reporting failure. This is because the
3781 string can be divided between the internal \D+ repeat and the external
3782 * repeat in a large number of ways, and all have to be tried. (The
3783 example uses [!?] rather than a single character at the end, because
3784 both PCRE and Perl have an optimization that allows for fast failure
3785 when a single character is used. They remember the last single charac-
3786 ter that is required for a match, and fail early if it is not present
3787 in the string.) If the pattern is changed so that it uses an atomic
3788 group, like this:
3789
3790 ((?>\D+)|<\d+>)*[!?]
3791
3792 sequences of non-digits cannot be broken, and failure happens quickly.
3793
3794
3795 BACK REFERENCES
3796
3797 Outside a character class, a backslash followed by a digit greater than
3798 0 (and possibly further digits) is a back reference to a capturing sub-
3799 pattern earlier (that is, to its left) in the pattern, provided there
3800 have been that many previous capturing left parentheses.
3801
3802 However, if the decimal number following the backslash is less than 10,
3803 it is always taken as a back reference, and causes an error only if
3804 there are not that many capturing left parentheses in the entire pat-
3805 tern. In other words, the parentheses that are referenced need not be
3806 to the left of the reference for numbers less than 10. A "forward back
3807 reference" of this type can make sense when a repetition is involved
3808 and the subpattern to the right has participated in an earlier itera-
3809 tion.
3810
3811 It is not possible to have a numerical "forward back reference" to a
3812 subpattern whose number is 10 or more using this syntax because a
3813 sequence such as \50 is interpreted as a character defined in octal.
3814 See the subsection entitled "Non-printing characters" above for further
3815 details of the handling of digits following a backslash. There is no
3816 such problem when named parentheses are used. A back reference to any
3817 subpattern is possible using named parentheses (see below).
3818
3819 Another way of avoiding the ambiguity inherent in the use of digits
3820 following a backslash is to use the \g escape sequence, which is a fea-
3821 ture introduced in Perl 5.10. This escape must be followed by a posi-
3822 tive or a negative number, optionally enclosed in braces. These exam-
3823 ples are all identical:
3824
3825 (ring), \1
3826 (ring), \g1
3827 (ring), \g{1}
3828
3829 A positive number specifies an absolute reference without the ambiguity
3830 that is present in the older syntax. It is also useful when literal
3831 digits follow the reference. A negative number is a relative reference.
3832 Consider this example:
3833
3834 (abc(def)ghi)\g{-1}
3835
3836 The sequence \g{-1} is a reference to the most recently started captur-
3837 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
3838 \g{-2} would be equivalent to \1. The use of relative references can be
3839 helpful in long patterns, and also in patterns that are created by
3840 joining together fragments that contain references within themselves.
3841
3842 A back reference matches whatever actually matched the capturing sub-
3843 pattern in the current subject string, rather than anything matching
3844 the subpattern itself (see "Subpatterns as subroutines" below for a way
3845 of doing that). So the pattern
3846
3847 (sens|respons)e and \1ibility
3848
3849 matches "sense and sensibility" and "response and responsibility", but
3850 not "sense and responsibility". If caseful matching is in force at the
3851 time of the back reference, the case of letters is relevant. For exam-
3852 ple,
3853
3854 ((?i)rah)\s+\1
3855
3856 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3857 original capturing subpattern is matched caselessly.
3858
3859 Back references to named subpatterns use the Perl syntax \k<name> or
3860 \k'name' or the Python syntax (?P=name). We could rewrite the above
3861 example in either of the following ways:
3862
3863 (?<p1>(?i)rah)\s+\k<p1>
3864 (?P<p1>(?i)rah)\s+(?P=p1)
3865
3866 A subpattern that is referenced by name may appear in the pattern
3867 before or after the reference.
3868
3869 There may be more than one back reference to the same subpattern. If a
3870 subpattern has not actually been used in a particular match, any back
3871 references to it always fail. For example, the pattern
3872
3873 (a|(bc))\2
3874
3875 always fails if it starts to match "a" rather than "bc". Because there
3876 may be many capturing parentheses in a pattern, all digits following
3877 the backslash are taken as part of a potential back reference number.
3878 If the pattern continues with a digit character, some delimiter must be
3879 used to terminate the back reference. If the PCRE_EXTENDED option is
3880 set, this can be whitespace. Otherwise an empty comment (see "Com-
3881 ments" below) can be used.
3882
3883 A back reference that occurs inside the parentheses to which it refers
3884 fails when the subpattern is first used, so, for example, (a\1) never
3885 matches. However, such references can be useful inside repeated sub-
3886 patterns. For example, the pattern
3887
3888 (a|b\1)+
3889
3890 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3891 ation of the subpattern, the back reference matches the character
3892 string corresponding to the previous iteration. In order for this to
3893 work, the pattern must be such that the first iteration does not need
3894 to match the back reference. This can be done using alternation, as in
3895 the example above, or by a quantifier with a minimum of zero.
3896
3897
3898 ASSERTIONS
3899
3900 An assertion is a test on the characters following or preceding the
3901 current matching point that does not actually consume any characters.
3902 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3903 described above.
3904
3905 More complicated assertions are coded as subpatterns. There are two
3906 kinds: those that look ahead of the current position in the subject
3907 string, and those that look behind it. An assertion subpattern is
3908 matched in the normal way, except that it does not cause the current
3909 matching position to be changed.
3910
3911 Assertion subpatterns are not capturing subpatterns, and may not be
3912 repeated, because it makes no sense to assert the same thing several
3913 times. If any kind of assertion contains capturing subpatterns within
3914 it, these are counted for the purposes of numbering the capturing sub-
3915 patterns in the whole pattern. However, substring capturing is carried
3916 out only for positive assertions, because it does not make sense for
3917 negative assertions.
3918
3919 Lookahead assertions
3920
3921 Lookahead assertions start with (?= for positive assertions and (?! for
3922 negative assertions. For example,
3923
3924 \w+(?=;)
3925
3926 matches a word followed by a semicolon, but does not include the semi-
3927 colon in the match, and
3928
3929 foo(?!bar)
3930
3931 matches any occurrence of "foo" that is not followed by "bar". Note
3932 that the apparently similar pattern
3933
3934 (?!foo)bar
3935
3936 does not find an occurrence of "bar" that is preceded by something
3937 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3938 the assertion (?!foo) is always true when the next three characters are
3939 "bar". A lookbehind assertion is needed to achieve the other effect.
3940
3941 If you want to force a matching failure at some point in a pattern, the
3942 most convenient way to do it is with (?!) because an empty string
3943 always matches, so an assertion that requires there not to be an empty
3944 string must always fail.
3945
3946 Lookbehind assertions
3947
3948 Lookbehind assertions start with (?<= for positive assertions and (?<!
3949 for negative assertions. For example,
3950
3951 (?<!foo)bar
3952
3953 does find an occurrence of "bar" that is not preceded by "foo". The
3954 contents of a lookbehind assertion are restricted such that all the
3955 strings it matches must have a fixed length. However, if there are sev-
3956 eral top-level alternatives, they do not all have to have the same
3957 fixed length. Thus
3958
3959 (?<=bullock|donkey)
3960
3961 is permitted, but
3962
3963 (?<!dogs?|cats?)
3964
3965 causes an error at compile time. Branches that match different length
3966 strings are permitted only at the top level of a lookbehind assertion.
3967 This is an extension compared with Perl (at least for 5.8), which
3968 requires all branches to match the same length of string. An assertion
3969 such as
3970
3971 (?<=ab(c|de))
3972
3973 is not permitted, because its single top-level branch can match two
3974 different lengths, but it is acceptable if rewritten to use two top-
3975 level branches:
3976
3977 (?<=abc|abde)
3978
3979 The implementation of lookbehind assertions is, for each alternative,
3980 to temporarily move the current position back by the fixed length and
3981 then try to match. If there are insufficient characters before the cur-
3982 rent position, the assertion fails.
3983
3984 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3985 mode) to appear in lookbehind assertions, because it makes it impossi-
3986 ble to calculate the length of the lookbehind. The \X and \R escapes,
3987 which can match different numbers of bytes, are also not permitted.
3988
3989 Possessive quantifiers can be used in conjunction with lookbehind
3990 assertions to specify efficient matching at the end of the subject
3991 string. Consider a simple pattern such as
3992
3993 abcd$
3994
3995 when applied to a long string that does not match. Because matching
3996 proceeds from left to right, PCRE will look for each "a" in the subject
3997 and then see if what follows matches the rest of the pattern. If the
3998 pattern is specified as
3999
4000 ^.*abcd$
4001
4002 the initial .* matches the entire string at first, but when this fails
4003 (because there is no following "a"), it backtracks to match all but the
4004 last character, then all but the last two characters, and so on. Once
4005 again the search for "a" covers the entire string, from right to left,
4006 so we are no better off. However, if the pattern is written as
4007
4008 ^.*+(?<=abcd)
4009
4010 there can be no backtracking for the .*+ item; it can match only the
4011 entire string. The subsequent lookbehind assertion does a single test
4012 on the last four characters. If it fails, the match fails immediately.
4013 For long strings, this approach makes a significant difference to the
4014 processing time.
4015
4016 Using multiple assertions
4017
4018 Several assertions (of any sort) may occur in succession. For example,
4019
4020 (?<=\d{3})(?<!999)foo
4021
4022 matches "foo" preceded by three digits that are not "999". Notice that
4023 each of the assertions is applied independently at the same point in
4024 the subject string. First there is a check that the previous three
4025 characters are all digits, and then there is a check that the same
4026 three characters are not "999". This pattern does not match "foo" pre-
4027 ceded by six characters, the first of which are digits and the last
4028 three of which are not "999". For example, it doesn't match "123abc-
4029 foo". A pattern to do that is
4030
4031 (?<=\d{3}...)(?<!999)foo
4032
4033 This time the first assertion looks at the preceding six characters,
4034 checking that the first three are digits, and then the second assertion
4035 checks that the preceding three characters are not "999".
4036
4037 Assertions can be nested in any combination. For example,
4038
4039 (?<=(?<!foo)bar)baz
4040
4041 matches an occurrence of "baz" that is preceded by "bar" which in turn
4042 is not preceded by "foo", while
4043
4044 (?<=\d{3}(?!999)...)foo
4045
4046 is another pattern that matches "foo" preceded by three digits and any
4047 three characters that are not "999".
4048
4049
4050 CONDITIONAL SUBPATTERNS
4051
4052 It is possible to cause the matching process to obey a subpattern con-
4053 ditionally or to choose between two alternative subpatterns, depending
4054 on the result of an assertion, or whether a previous capturing subpat-
4055 tern matched or not. The two possible forms of conditional subpattern
4056 are
4057
4058 (?(condition)yes-pattern)
4059 (?(condition)yes-pattern|no-pattern)
4060
4061 If the condition is satisfied, the yes-pattern is used; otherwise the
4062 no-pattern (if present) is used. If there are more than two alterna-
4063 tives in the subpattern, a compile-time error occurs.
4064
4065 There are four kinds of condition: references to subpatterns, refer-
4066 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4067
4068 Checking for a used subpattern by number
4069
4070 If the text between the parentheses consists of a sequence of digits,
4071 the condition is true if the capturing subpattern of that number has
4072 previously matched.
4073
4074 Consider the following pattern, which contains non-significant white
4075 space to make it more readable (assume the PCRE_EXTENDED option) and to
4076 divide it into three parts for ease of discussion:
4077
4078 ( \( )? [^()]+ (?(1) \) )
4079
4080 The first part matches an optional opening parenthesis, and if that
4081 character is present, sets it as the first captured substring. The sec-
4082 ond part matches one or more characters that are not parentheses. The
4083 third part is a conditional subpattern that tests whether the first set
4084 of parentheses matched or not. If they did, that is, if subject started
4085 with an opening parenthesis, the condition is true, and so the yes-pat-
4086 tern is executed and a closing parenthesis is required. Otherwise,
4087 since no-pattern is not present, the subpattern matches nothing. In
4088 other words, this pattern matches a sequence of non-parentheses,
4089 optionally enclosed in parentheses.
4090
4091 Checking for a used subpattern by name
4092
4093 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4094 used subpattern by name. For compatibility with earlier versions of
4095 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4096 also recognized. However, there is a possible ambiguity with this syn-
4097 tax, because subpattern names may consist entirely of digits. PCRE
4098 looks first for a named subpattern; if it cannot find one and the name
4099 consists entirely of digits, PCRE looks for a subpattern of that num-
4100 ber, which must be greater than zero. Using subpattern names that con-
4101 sist entirely of digits is not recommended.
4102
4103 Rewriting the above example to use a named subpattern gives this:
4104
4105 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4106
4107
4108 Checking for pattern recursion
4109
4110 If the condition is the string (R), and there is no subpattern with the
4111 name R, the condition is true if a recursive call to the whole pattern
4112 or any subpattern has been made. If digits or a name preceded by amper-
4113 sand follow the letter R, for example:
4114
4115 (?(R3)...) or (?(R&name)...)
4116
4117 the condition is true if the most recent recursion is into the subpat-
4118 tern whose number or name is given. This condition does not check the
4119 entire recursion stack.
4120
4121 At "top level", all these recursion test conditions are false. Recur-
4122 sive patterns are described below.
4123
4124 Defining subpatterns for use by reference only
4125
4126 If the condition is the string (DEFINE), and there is no subpattern
4127 with the name DEFINE, the condition is always false. In this case,
4128 there may be only one alternative in the subpattern. It is always
4129 skipped if control reaches this point in the pattern; the idea of
4130 DEFINE is that it can be used to define "subroutines" that can be ref-
4131 erenced from elsewhere. (The use of "subroutines" is described below.)
4132 For example, a pattern to match an IPv4 address could be written like
4133 this (ignore whitespace and line breaks):
4134
4135 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4136 \b (?&byte) (\.(?&byte)){3} \b
4137
4138 The first part of the pattern is a DEFINE group inside which a another
4139 group named "byte" is defined. This matches an individual component of
4140 an IPv4 address (a number less than 256). When matching takes place,
4141 this part of the pattern is skipped because DEFINE acts like a false
4142 condition.
4143
4144 The rest of the pattern uses references to the named group to match the
4145 four dot-separated components of an IPv4 address, insisting on a word
4146 boundary at each end.
4147
4148 Assertion conditions
4149
4150 If the condition is not in any of the above formats, it must be an
4151 assertion. This may be a positive or negative lookahead or lookbehind
4152 assertion. Consider this pattern, again containing non-significant
4153 white space, and with the two alternatives on the second line:
4154
4155 (?(?=[^a-z]*[a-z])
4156 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4157
4158 The condition is a positive lookahead assertion that matches an
4159 optional sequence of non-letters followed by a letter. In other words,
4160 it tests for the presence of at least one letter in the subject. If a
4161 letter is found, the subject is matched against the first alternative;
4162 otherwise it is matched against the second. This pattern matches
4163 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4164 letters and dd are digits.
4165
4166
4167 COMMENTS
4168
4169 The sequence (?# marks the start of a comment that continues up to the
4170 next closing parenthesis. Nested parentheses are not permitted. The
4171 characters that make up a comment play no part in the pattern matching
4172 at all.
4173
4174 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4175 character class introduces a comment that continues to immediately
4176 after the next newline in the pattern.
4177
4178
4179 RECURSIVE PATTERNS
4180
4181 Consider the problem of matching a string in parentheses, allowing for
4182 unlimited nested parentheses. Without the use of recursion, the best
4183 that can be done is to use a pattern that matches up to some fixed
4184 depth of nesting. It is not possible to handle an arbitrary nesting
4185 depth.
4186
4187 For some time, Perl has provided a facility that allows regular expres-
4188 sions to recurse (amongst other things). It does this by interpolating
4189 Perl code in the expression at run time, and the code can refer to the
4190 expression itself. A Perl pattern using code interpolation to solve the
4191 parentheses problem can be created like this:
4192
4193 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4194
4195 The (?p{...}) item interpolates Perl code at run time, and in this case
4196 refers recursively to the pattern in which it appears.
4197
4198 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4199 it supports special syntax for recursion of the entire pattern, and
4200 also for individual subpattern recursion. After its introduction in
4201 PCRE and Python, this kind of recursion was introduced into Perl at
4202 release 5.10.
4203
4204 A special item that consists of (? followed by a number greater than
4205 zero and a closing parenthesis is a recursive call of the subpattern of
4206 the given number, provided that it occurs inside that subpattern. (If
4207 not, it is a "subroutine" call, which is described in the next sec-
4208 tion.) The special item (?R) or (?0) is a recursive call of the entire
4209 regular expression.
4210
4211 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4212 always treated as an atomic group. That is, once it has matched some of
4213 the subject string, it is never re-entered, even if it contains untried
4214 alternatives and there is a subsequent matching failure.
4215
4216 This PCRE pattern solves the nested parentheses problem (assume the
4217 PCRE_EXTENDED option is set so that white space is ignored):
4218
4219 \( ( (?>[^()]+) | (?R) )* \)
4220
4221 First it matches an opening parenthesis. Then it matches any number of
4222 substrings which can either be a sequence of non-parentheses, or a
4223 recursive match of the pattern itself (that is, a correctly parenthe-
4224 sized substring). Finally there is a closing parenthesis.
4225
4226 If this were part of a larger pattern, you would not want to recurse
4227 the entire pattern, so instead you could use this:
4228
4229 ( \( ( (?>[^()]+) | (?1) )* \) )
4230
4231 We have put the pattern into parentheses, and caused the recursion to
4232 refer to them instead of the whole pattern. In a larger pattern, keep-
4233 ing track of parenthesis numbers can be tricky. It may be more conve-
4234 nient to use named parentheses instead. The Perl syntax for this is
4235 (?&name); PCRE's earlier syntax (?P>name) is also supported. We could
4236 rewrite the above example as follows:
4237
4238 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4239
4240 If there is more than one subpattern with the same name, the earliest
4241 one is used. This particular example pattern contains nested unlimited
4242 repeats, and so the use of atomic grouping for matching strings of non-
4243 parentheses is important when applying the pattern to strings that do
4244 not match. For example, when this pattern is applied to
4245
4246 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4247
4248 it yields "no match" quickly. However, if atomic grouping is not used,
4249 the match runs for a very long time indeed because there are so many
4250 different ways the + and * repeats can carve up the subject, and all
4251 have to be tested before failure can be reported.
4252
4253 At the end of a match, the values set for any capturing subpatterns are
4254 those from the outermost level of the recursion at which the subpattern
4255 value is set. If you want to obtain intermediate values, a callout
4256 function can be used (see below and the pcrecallout documentation). If
4257 the pattern above is matched against
4258
4259 (ab(cd)ef)
4260
4261 the value for the capturing parentheses is "ef", which is the last
4262 value taken on at the top level. If additional parentheses are added,
4263 giving
4264
4265 \( ( ( (?>[^()]+) | (?R) )* ) \)
4266 ^ ^
4267 ^ ^
4268
4269 the string they capture is "ab(cd)ef", the contents of the top level
4270 parentheses. If there are more than 15 capturing parentheses in a pat-
4271 tern, PCRE has to obtain extra memory to store data during a recursion,
4272 which it does by using pcre_malloc, freeing it via pcre_free after-
4273 wards. If no memory can be obtained, the match fails with the
4274 PCRE_ERROR_NOMEMORY error.
4275
4276 Do not confuse the (?R) item with the condition (R), which tests for
4277 recursion. Consider this pattern, which matches text in angle brack-
4278 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4279 brackets (that is, when recursing), whereas any characters are permit-
4280 ted at the outer level.
4281
4282 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4283
4284 In this pattern, (?(R) is the start of a conditional subpattern, with
4285 two different alternatives for the recursive and non-recursive cases.
4286 The (?R) item is the actual recursive call.
4287
4288
4289 SUBPATTERNS AS SUBROUTINES
4290
4291 If the syntax for a recursive subpattern reference (either by number or
4292 by name) is used outside the parentheses to which it refers, it oper-
4293 ates like a subroutine in a programming language. The "called" subpat-
4294 tern may be defined before or after the reference. An earlier example
4295 pointed out that the pattern
4296
4297 (sens|respons)e and \1ibility
4298
4299 matches "sense and sensibility" and "response and responsibility", but
4300 not "sense and responsibility". If instead the pattern
4301
4302 (sens|respons)e and (?1)ibility
4303
4304 is used, it does match "sense and responsibility" as well as the other
4305 two strings. Another example is given in the discussion of DEFINE
4306 above.
4307
4308 Like recursive subpatterns, a "subroutine" call is always treated as an
4309 atomic group. That is, once it has matched some of the subject string,
4310 it is never re-entered, even if it contains untried alternatives and
4311 there is a subsequent matching failure.
4312
4313 When a subpattern is used as a subroutine, processing options such as
4314 case-independence are fixed when the subpattern is defined. They cannot
4315 be changed for different calls. For example, consider this pattern:
4316
4317 (abc)(?i:(?1))
4318
4319 It matches "abcabc". It does not match "abcABC" because the change of
4320 processing option does not affect the called subpattern.
4321
4322
4323 CALLOUTS
4324
4325 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4326 Perl code to be obeyed in the middle of matching a regular expression.
4327 This makes it possible, amongst other things, to extract different sub-
4328 strings that match the same pair of parentheses when there is a repeti-
4329 tion.
4330
4331 PCRE provides a similar feature, but of course it cannot obey arbitrary
4332 Perl code. The feature is called "callout". The caller of PCRE provides
4333 an external function by putting its entry point in the global variable
4334 pcre_callout. By default, this variable contains NULL, which disables
4335 all calling out.
4336
4337 Within a regular expression, (?C) indicates the points at which the
4338 external function is to be called. If you want to identify different
4339 callout points, you can put a number less than 256 after the letter C.
4340 The default value is zero. For example, this pattern has two callout
4341 points:
4342
4343 (?C1)abc(?C2)def
4344
4345 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4346 automatically installed before each item in the pattern. They are all
4347 numbered 255.
4348
4349 During matching, when PCRE reaches a callout point (and pcre_callout is
4350 set), the external function is called. It is provided with the number
4351 of the callout, the position in the pattern, and, optionally, one item
4352 of data originally supplied by the caller of pcre_exec(). The callout
4353 function may cause matching to proceed, to backtrack, or to fail alto-
4354 gether. A complete description of the interface to the callout function
4355 is given in the pcrecallout documentation.
4356
4357
4358 SEE ALSO
4359
4360 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4361
4362
4363 AUTHOR
4364
4365 Philip Hazel
4366 University Computing Service
4367 Cambridge CB2 3QH, England.
4368
4369
4370 REVISION
4371
4372 Last updated: 06 March 2007
4373 Copyright (c) 1997-2007 University of Cambridge.
4374 ------------------------------------------------------------------------------
4375
4376
4377 PCREPARTIAL(3) PCREPARTIAL(3)
4378
4379
4380 NAME
4381 PCRE - Perl-compatible regular expressions
4382
4383
4384 PARTIAL MATCHING IN PCRE
4385
4386 In normal use of PCRE, if the subject string that is passed to
4387 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4388 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4389 There are circumstances where it might be helpful to distinguish this
4390 case from other cases in which there is no match.
4391
4392 Consider, for example, an application where a human is required to type
4393 in data for a field with specific formatting requirements. An example
4394 might be a date in the form ddmmmyy, defined by this pattern:
4395
4396 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4397
4398 If the application sees the user's keystrokes one by one, and can check
4399 that what has been typed so far is potentially valid, it is able to
4400 raise an error as soon as a mistake is made, possibly beeping and not
4401 reflecting the character that has been typed. This immediate feedback
4402 is likely to be a better user interface than a check that is delayed
4403 until the entire string has been entered.
4404
4405 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4406 TIAL option, which can be set when calling pcre_exec() or
4407 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4408 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4409 during the matching process the last part of the subject string matched
4410 part of the pattern. Unfortunately, for non-anchored matching, it is
4411 not possible to obtain the position of the start of the partial match.
4412 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4413
4414 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4415 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4416 the subject is reached, there have been no complete matches, but there
4417 is still at least one matching possibility. The portion of the string
4418 that provided the partial match is set as the first matching string.
4419
4420 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4421 the last literal byte in a pattern, and abandons matching immediately
4422 if such a byte is not present in the subject string. This optimization
4423 cannot be used for a subject string that might match only partially.
4424
4425
4426 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4427
4428 Because of the way certain internal optimizations are implemented in
4429 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4430 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4431 used. For pcre_exec(), repeated single characters such as
4432
4433 a{2,4}
4434
4435 and repeated single metasequences such as
4436
4437 \d+
4438
4439 are not permitted if the maximum number of occurrences is greater than
4440 one. Optional items such as \d? (where the maximum is one) are permit-
4441 ted. Quantifiers with any values are permitted after parentheses, so
4442 the invalid examples above can be coded thus:
4443
4444 (a){2,4}
4445 (\d)+
4446
4447 These constructions run more slowly, but for the kinds of application
4448 that are envisaged for this facility, this is not felt to be a major
4449 restriction.
4450
4451 If PCRE_PARTIAL is set for a pattern that does not conform to the
4452 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4453 (-13).
4454
4455
4456 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4457
4458 If the escape sequence \P is present in a pcretest data line, the
4459 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4460 uses the date example quoted above:
4461
4462 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4463 data> 25jun04\P
4464 0: 25jun04
4465 1: jun
4466 data> 25dec3\P
4467 Partial match
4468 data> 3ju\P
4469 Partial match
4470 data> 3juj\P
4471 No match
4472 data> j\P
4473 No match
4474
4475 The first data string is matched completely, so pcretest shows the
4476 matched substrings. The remaining four strings do not match the com-
4477 plete pattern, but the first two are partial matches. The same test,
4478 using pcre_dfa_exec() matching (by means of the \D escape sequence),
4479 produces the following output:
4480
4481 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4482 data> 25jun04\P\D
4483 0: 25jun04
4484 data> 23dec3\P\D
4485 Partial match: 23dec3
4486 data> 3ju\P\D
4487 Partial match: 3ju
4488 data> 3juj\P\D
4489 No match
4490 data> j\P\D
4491 No match
4492
4493 Notice that in this case the portion of the string that was matched is
4494 made available.
4495
4496
4497 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4498
4499 When a partial match has been found using pcre_dfa_exec(), it is possi-
4500 ble to continue the match by providing additional subject data and
4501 calling pcre_dfa_exec() again with the same compiled regular expres-
4502 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
4503 the same working space as before, because this is where details of the
4504 previous partial match are stored. Here is an example using pcretest,
4505 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
4506 \D are as above):
4507
4508 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4509 data> 23ja\P\D
4510 Partial match: 23ja
4511 data> n05\R\D
4512 0: n05
4513
4514 The first call has "23ja" as the subject, and requests partial match-
4515 ing; the second call has "n05" as the subject for the continued
4516 (restarted) match. Notice that when the match is complete, only the
4517 last part is shown; PCRE does not retain the previously partially-
4518 matched string. It is up to the calling program to do that if it needs
4519 to.
4520
4521 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
4522 matching over multiple segments. This facility can be used to pass very
4523 long subject strings to pcre_dfa_exec(). However, some care is needed
4524 for certain types of pattern.
4525
4526 1. If the pattern contains tests for the beginning or end of a line,
4527 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4528 ate, when the subject string for any call does not contain the begin-
4529 ning or end of a line.
4530
4531 2. If the pattern contains backward assertions (including \b or \B),
4532 you need to arrange for some overlap in the subject strings to allow
4533 for this. For example, you could pass the subject in chunks that are
4534 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4535 set to 200 and the previous 200 bytes at the start of the buffer.
4536
4537 3. Matching a subject string that is split into multiple segments does
4538 not always produce exactly the same result as matching over one single
4539 long string. The difference arises when there are multiple matching
4540 possibilities, because a partial match result is given only when there
4541 are no completed matches in a call to pcre_dfa_exec(). This means that
4542 as soon as the shortest match has been found, continuation to a new
4543 subject segment is no longer possible. Consider this pcretest example:
4544
4545 re> /dog(sbody)?/
4546 data> do\P\D
4547 Partial match: do
4548 data> gsb\R\P\D
4549 0: g
4550 data> dogsbody\D
4551 0: dogsbody
4552 1: dog
4553
4554 The pattern matches the words "dog" or "dogsbody". When the subject is
4555 presented in several parts ("do" and "gsb" being the first two) the
4556 match stops when "dog" has been found, and it is not possible to con-
4557 tinue. On the other hand, if "dogsbody" is presented as a single
4558 string, both matches are found.
4559
4560 Because of this phenomenon, it does not usually make sense to end a
4561 pattern that is going to be matched in this way with a variable repeat.
4562
4563 4. Patterns that contain alternatives at the top level which do not all
4564 start with the same pattern item may not work as expected. For example,
4565 consider this pattern:
4566
4567 1234|3789
4568
4569 If the first part of the subject is "ABC123", a partial match of the
4570 first alternative is found at offset 3. There is no partial match for
4571 the second alternative, because such a match does not start at the same
4572 point in the subject string. Attempting to continue with the string
4573 "789" does not yield a match because only those alternatives that match
4574 at one point in the subject are remembered. The problem arises because
4575 the start of the second alternative matches within the first alterna-
4576 tive. There is no problem with anchored patterns or patterns such as:
4577
4578 1234|ABCD
4579
4580 where no string can be a partial match for both alternatives.
4581
4582
4583 AUTHOR
4584
4585 Philip Hazel
4586 University Computing Service
4587 Cambridge CB2 3QH, England.
4588
4589
4590 REVISION
4591
4592 Last updated: 06 March 2007
4593 Copyright (c) 1997-2007 University of Cambridge.
4594 ------------------------------------------------------------------------------
4595
4596
4597 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4598
4599
4600 NAME
4601 PCRE - Perl-compatible regular expressions
4602
4603
4604 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4605
4606 If you are running an application that uses a large number of regular
4607 expression patterns, it may be useful to store them in a precompiled
4608 form instead of having to compile them every time the application is
4609 run. If you are not using any private character tables (see the
4610 pcre_maketables() documentation), this is relatively straightforward.
4611 If you are using private tables, it is a little bit more complicated.
4612
4613 If you save compiled patterns to a file, you can copy them to a differ-
4614 ent host and run them there. This works even if the new host has the
4615 opposite endianness to the one on which the patterns were compiled.
4616 There may be a small performance penalty, but it should be insignifi-
4617 cant.
4618
4619
4620 SAVING A COMPILED PATTERN
4621 The value returned by pcre_compile() points to a single block of memory
4622 that holds the compiled pattern and associated data. You can find the
4623 length of this block in bytes by calling pcre_fullinfo() with an argu-
4624 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4625 manner. Here is sample code that compiles a pattern and writes it to a
4626 file. It assumes that the variable fd refers to a file that is open for
4627 output:
4628
4629 int erroroffset, rc, size;
4630 char *error;
4631 pcre *re;
4632
4633 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4634 if (re == NULL) { ... handle errors ... }
4635 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4636 if (rc < 0) { ... handle errors ... }
4637 rc = fwrite(re, 1, size, fd);
4638 if (rc != size) { ... handle errors ... }
4639
4640 In this example, the bytes that comprise the compiled pattern are
4641 copied exactly. Note that this is binary data that may contain any of
4642 the 256 possible byte values. On systems that make a distinction
4643 between binary and non-binary data, be sure that the file is opened for
4644 binary output.
4645
4646 If you want to write more than one pattern to a file, you will have to
4647 devise a way of separating them. For binary data, preceding each pat-
4648 tern with its length is probably the most straightforward approach.
4649 Another possibility is to write out the data in hexadecimal instead of
4650 binary, one pattern to a line.
4651
4652 Saving compiled patterns in a file is only one possible way of storing
4653 them for later use. They could equally well be saved in a database, or
4654 in the memory of some daemon process that passes them via sockets to
4655 the processes that want them.
4656
4657 If the pattern has been studied, it is also possible to save the study
4658 data in a similar way to the compiled pattern itself. When studying
4659 generates additional information, pcre_study() returns a pointer to a
4660 pcre_extra data block. Its format is defined in the section on matching
4661 a pattern in the pcreapi documentation. The study_data field points to
4662 the binary study data, and this is what you must save (not the
4663 pcre_extra block itself). The length of the study data can be obtained
4664 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4665 Remember to check that pcre_study() did return a non-NULL value before
4666 trying to save the study data.
4667
4668
4669 RE-USING A PRECOMPILED PATTERN
4670
4671 Re-using a precompiled pattern is straightforward. Having reloaded it
4672 into main memory, you pass its pointer to pcre_exec() or
4673 pcre_dfa_exec() in the usual way. This should work even on another
4674 host, and even if that host has the opposite endianness to the one
4675 where the pattern was compiled.
4676
4677 However, if you passed a pointer to custom character tables when the
4678 pattern was compiled (the tableptr argument of pcre_compile()), you
4679 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4680 because the value saved with the compiled pattern will obviously be
4681 nonsense. A field in a pcre_extra() block is used to pass this data, as
4682 described in the section on matching a pattern in the pcreapi documen-
4683 tation.
4684
4685 If you did not provide custom character tables when the pattern was
4686 compiled, the pointer in the compiled pattern is NULL, which causes
4687 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4688 take any special action at run time in this case.
4689
4690 If you saved study data with the compiled pattern, you need to create
4691 your own pcre_extra data block and set the study_data field to point to
4692 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4693 bit in the flags field to indicate that study data is present. Then
4694 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4695 usual way.
4696
4697
4698 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4699
4700 The layout of the control block that is at the start of the data that
4701 makes up a compiled pattern was changed for release 5.0. If you have
4702 any saved patterns that were compiled with previous releases (not a
4703 facility that was previously advertised), you will have to recompile
4704 them for release 5.0 and above.
4705
4706 If you have any saved patterns in UTF-8 mode that use \p or \P that
4707 were compiled with any release up to and including 6.4, you will have
4708 to recompile them for release 6.5 and above.
4709
4710 All saved patterns from earlier releases must be recompiled for release
4711 7.0 or higher, because there was an internal reorganization at that
4712 release.
4713
4714
4715 AUTHOR
4716
4717 Philip Hazel
4718 University Computing Service
4719 Cambridge CB2 3QH, England.
4720
4721
4722 REVISION
4723
4724 Last updated: 06 March 2007
4725 Copyright (c) 1997-2007 University of Cambridge.
4726 ------------------------------------------------------------------------------
4727
4728
4729 PCREPERFORM(3) PCREPERFORM(3)
4730
4731
4732 NAME
4733 PCRE - Perl-compatible regular expressions
4734
4735
4736 PCRE PERFORMANCE
4737
4738 Two aspects of performance are discussed below: memory usage and pro-
4739 cessing time. The way you express your pattern as a regular expression
4740 can affect both of them.
4741
4742
4743 MEMORY USAGE
4744
4745 Patterns are compiled by PCRE into a reasonably efficient byte code, so
4746 that most simple patterns do not use much memory. However, there is one
4747 case where memory usage can be unexpectedly large. When a parenthesized
4748 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
4749 ited maximum, the whole subpattern is repeated in the compiled code.
4750 For example, the pattern
4751
4752 (abc|def){2,4}
4753
4754 is compiled as if it were
4755
4756 (abc|def)(abc|def)((abc|def)(abc|def)?)?
4757
4758 (Technical aside: It is done this way so that backtrack points within
4759 each of the repetitions can be independently maintained.)
4760
4761 For regular expressions whose quantifiers use only small numbers, this
4762 is not usually a problem. However, if the numbers are large, and par-
4763 ticularly if such repetitions are nested, the memory usage can become
4764 an embarrassment. For example, the very simple pattern
4765
4766 ((ab){1,1000}c){1,3}
4767
4768 uses 51K bytes when compiled. When PCRE is compiled with its default
4769 internal pointer size of two bytes, the size limit on a compiled pat-
4770 tern is 64K, and this is reached with the above pattern if the outer
4771 repetition is increased from 3 to 4. PCRE can be compiled to use larger
4772 internal pointers and thus handle larger compiled patterns, but it is
4773 better to try to rewrite your pattern to use less memory if you can.
4774
4775 One way of reducing the memory usage for such patterns is to make use
4776 of PCRE's "subroutine" facility. Re-writing the above pattern as
4777
4778 ((ab)(?2){0,999}c)(?1){0,2}
4779
4780 reduces the memory requirements to 18K, and indeed it remains under 20K
4781 even with the outer repetition increased to 100. However, this pattern
4782 is not exactly equivalent, because the "subroutine" calls are treated
4783 as atomic groups into which there can be no backtracking if there is a
4784 subsequent matching failure. Therefore, PCRE cannot do this kind of
4785 rewriting automatically. Furthermore, there is a noticeable loss of
4786 speed when executing the modified pattern. Nevertheless, if the atomic
4787 grouping is not a problem and the loss of speed is acceptable, this
4788 kind of rewriting will allow you to process patterns that PCRE cannot
4789 otherwise handle.
4790
4791
4792 PROCESSING TIME
4793
4794 Certain items in regular expression patterns are processed more effi-
4795 ciently than others. It is more efficient to use a character class like
4796 [aeiou] than a set of single-character alternatives such as
4797 (a|e|i|o|u). In general, the simplest construction that provides the
4798 required behaviour is usually the most efficient. Jeffrey Friedl's book
4799 contains a lot of useful general discussion about optimizing regular
4800 expressions for efficient performance. This document contains a few
4801 observations about PCRE.
4802
4803 Using Unicode character properties (the \p, \P, and \X escapes) is
4804 slow, because PCRE has to scan a structure that contains data for over
4805 fifteen thousand characters whenever it needs a character's property.
4806 If you can find an alternative pattern that does not use character
4807 properties, it will probably be faster.
4808
4809 When a pattern begins with .* not in parentheses, or in parentheses
4810 that are not the subject of a backreference, and the PCRE_DOTALL option
4811 is set, the pattern is implicitly anchored by PCRE, since it can match
4812 only at the start of a subject string. However, if PCRE_DOTALL is not
4813 set, PCRE cannot make this optimization, because the . metacharacter
4814 does not then match a newline, and if the subject string contains new-
4815 lines, the pattern may match from the character immediately following
4816 one of them instead of from the very start. For example, the pattern
4817
4818 .*second
4819
4820 matches the subject "first\nand second" (where \n stands for a newline
4821 character), with the match starting at the seventh character. In order
4822 to do this, PCRE has to retry the match starting after every newline in
4823 the subject.
4824
4825 If you are using such a pattern with subject strings that do not con-
4826 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4827 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4828 ing. That saves PCRE from having to scan along the subject looking for
4829 a newline to restart at.
4830
4831 Beware of patterns that contain nested indefinite repeats. These can
4832 take a long time to run when applied to a string that does not match.
4833 Consider the pattern fragment
4834
4835 ^(a+)*
4836
4837 This can match "aaaa" in 16 different ways, and this number increases
4838 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4839 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
4840 repeats can match different numbers of times.) When the remainder of
4841 the pattern is such that the entire match is going to fail, PCRE has in
4842 principle to try every possible variation, and this can take an
4843 extremely long time, even for relatively short strings.
4844
4845 An optimization catches some of the more simple cases such as
4846
4847 (a+)*b
4848
4849 where a literal character follows. Before embarking on the standard
4850 matching procedure, PCRE checks that there is a "b" later in the sub-
4851 ject string, and if there is not, it fails the match immediately. How-
4852 ever, when there is no following literal this optimization cannot be
4853 used. You can see the difference by comparing the behaviour of
4854
4855 (a+)*\d
4856
4857 with the pattern above. The former gives a failure almost instantly
4858 when applied to a whole line of "a" characters, whereas the latter
4859 takes an appreciable time with strings longer than about 20 characters.
4860
4861 In many cases, the solution to this kind of performance issue is to use
4862 an atomic group or a possessive quantifier.
4863
4864
4865 AUTHOR
4866
4867 Philip Hazel
4868 University Computing Service
4869 Cambridge CB2 3QH, England.
4870
4871
4872 REVISION
4873
4874 Last updated: 06 March 2007
4875 Copyright (c) 1997-2007 University of Cambridge.
4876 ------------------------------------------------------------------------------
4877
4878
4879 PCREPOSIX(3) PCREPOSIX(3)
4880
4881
4882 NAME
4883 PCRE - Perl-compatible regular expressions.
4884
4885
4886 SYNOPSIS OF POSIX API
4887
4888 #include <pcreposix.h>
4889
4890 int regcomp(regex_t *preg, const char *pattern,
4891 int cflags);
4892
4893 int regexec(regex_t *preg, const char *string,
4894 size_t nmatch, regmatch_t pmatch[], int eflags);
4895
4896 size_t regerror(int errcode, const regex_t *preg,
4897 char *errbuf, size_t errbuf_size);
4898
4899 void regfree(regex_t *preg);
4900
4901
4902 DESCRIPTION
4903
4904 This set of functions provides a POSIX-style API to the PCRE regular
4905 expression package. See the pcreapi documentation for a description of
4906 PCRE's native API, which contains much additional functionality.
4907
4908 The functions described here are just wrapper functions that ultimately
4909 call the PCRE native API. Their prototypes are defined in the
4910 pcreposix.h header file, and on Unix systems the library itself is
4911 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4912 command for linking an application that uses them. Because the POSIX
4913 functions call the native ones, it is also necessary to add -lpcre.
4914
4915 I have implemented only those option bits that can be reasonably mapped
4916 to PCRE native options. In addition, the option REG_EXTENDED is defined
4917 with the value zero. This has no effect, but since programs that are
4918 written to the POSIX interface often use it, this makes it easier to
4919 slot in PCRE as a replacement library. Other POSIX options are not even
4920 defined.
4921
4922 When PCRE is called via these functions, it is only the API that is
4923 POSIX-like in style. The syntax and semantics of the regular expres-
4924 sions themselves are still those of Perl, subject to the setting of
4925 various PCRE options, as described below. "POSIX-like in style" means
4926 that the API approximates to the POSIX definition; it is not fully
4927 POSIX-compatible, and in multi-byte encoding domains it is probably
4928 even less compatible.
4929
4930 The header for these functions is supplied as pcreposix.h to avoid any
4931 potential clash with other POSIX libraries. It can, of course, be
4932 renamed or aliased as regex.h, which is the "correct" name. It provides
4933 two structure types, regex_t for compiled internal forms, and reg-
4934 match_t for returning captured substrings. It also defines some con-
4935 stants whose names start with "REG_"; these are used for setting
4936 options and identifying error codes.
4937
4938
4939 COMPILING A PATTERN
4940
4941 The function regcomp() is called to compile a pattern into an internal
4942 form. The pattern is a C string terminated by a binary zero, and is
4943 passed in the argument pattern. The preg argument is a pointer to a
4944 regex_t structure that is used as a base for storing information about
4945 the compiled regular expression.
4946
4947 The argument cflags is either zero, or contains one or more of the bits
4948 defined by the following macros:
4949
4950 REG_DOTALL
4951
4952 The PCRE_DOTALL option is set when the regular expression is passed for
4953 compilation to the native function. Note that REG_DOTALL is not part of
4954 the POSIX standard.
4955
4956 REG_ICASE
4957
4958 The PCRE_CASELESS option is set when the regular expression is passed
4959 for compilation to the native function.
4960
4961 REG_NEWLINE
4962
4963 The PCRE_MULTILINE option is set when the regular expression is passed
4964 for compilation to the native function. Note that this does not mimic
4965 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
4966 tion).
4967
4968 REG_NOSUB
4969
4970 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
4971 passed for compilation to the native function. In addition, when a pat-
4972 tern that is compiled with this flag is passed to regexec() for match-
4973 ing, the nmatch and pmatch arguments are ignored, and no captured
4974 strings are returned.
4975
4976 REG_UTF8
4977
4978 The PCRE_UTF8 option is set when the regular expression is passed for
4979 compilation to the native function. This causes the pattern itself and
4980 all data strings used for matching it to be treated as UTF-8 strings.
4981 Note that REG_UTF8 is not part of the POSIX standard.
4982
4983 In the absence of these flags, no options are passed to the native
4984 function. This means the the regex is compiled with PCRE default
4985 semantics. In particular, the way it handles newline characters in the
4986 subject string is the Perl way, not the POSIX way. Note that setting
4987 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4988 It does not affect the way newlines are matched by . (they aren't) or
4989 by a negative class such as [^a] (they are).
4990
4991 The yield of regcomp() is zero on success, and non-zero otherwise. The
4992 preg structure is filled in on success, and one member of the structure
4993 is public: re_nsub contains the number of capturing subpatterns in the
4994 regular expression. Various error codes are defined in the header file.
4995
4996
4997 MATCHING NEWLINE CHARACTERS
4998
4999 This area is not simple, because POSIX and Perl take different views of
5000 things. It is not possible to get PCRE to obey POSIX semantics, but
5001 then PCRE was never intended to be a POSIX engine. The following table
5002 lists the different possibilities for matching newline characters in
5003 PCRE:
5004
5005 Default Change with
5006
5007 . matches newline no PCRE_DOTALL
5008 newline matches [^a] yes not changeable
5009 $ matches \n at end yes PCRE_DOLLARENDONLY
5010 $ matches \n in middle no PCRE_MULTILINE
5011 ^ matches \n in middle no PCRE_MULTILINE
5012
5013 This is the equivalent table for POSIX:
5014
5015 Default Change with
5016
5017 . matches newline yes REG_NEWLINE
5018 newline matches [^a] yes REG_NEWLINE
5019 $ matches \n at end no REG_NEWLINE
5020 $ matches \n in middle no REG_NEWLINE
5021 ^ matches \n in middle no REG_NEWLINE
5022
5023 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5024 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5025 no way to stop newline from matching [^a].
5026
5027 The default POSIX newline handling can be obtained by setting
5028 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5029 behave exactly as for the REG_NEWLINE action.
5030
5031
5032 MATCHING A PATTERN
5033
5034 The function regexec() is called to match a compiled pattern preg
5035 against a given string, which is terminated by a zero byte, subject to
5036 the options in eflags. These can be:
5037
5038 REG_NOTBOL
5039
5040 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5041 function.
5042
5043 REG_NOTEOL
5044
5045 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5046 function.
5047
5048 If the pattern was compiled with the REG_NOSUB flag, no data about any
5049 matched strings is returned. The nmatch and pmatch arguments of
5050 regexec() are ignored.
5051
5052 Otherwise,the portion of the string that was matched, and also any cap-
5053 tured substrings, are returned via the pmatch argument, which points to
5054 an array of nmatch structures of type regmatch_t, containing the mem-
5055 bers rm_so and rm_eo. These contain the offset to the first character
5056 of each substring and the offset to the first character after the end
5057 of each substring, respectively. The 0th element of the vector relates
5058 to the entire portion of string that was matched; subsequent elements
5059 relate to the capturing subpatterns of the regular expression. Unused
5060 entries in the array have both structure members set to -1.
5061
5062 A successful match yields a zero return; various error codes are
5063 defined in the header file, of which REG_NOMATCH is the "expected"
5064 failure code.
5065
5066
5067 ERROR MESSAGES
5068
5069 The regerror() function maps a non-zero errorcode from either regcomp()
5070 or regexec() to a printable message. If preg is not NULL, the error
5071 should have arisen from the use of that structure. A message terminated
5072 by a binary zero is placed in errbuf. The length of the message,
5073 including the zero, is limited to errbuf_size. The yield of the func-
5074 tion is the size of buffer needed to hold the whole message.
5075
5076
5077 MEMORY USAGE
5078
5079 Compiling a regular expression causes memory to be allocated and asso-
5080 ciated with the preg structure. The function regfree() frees all such
5081 memory, after which preg may no longer be used as a compiled expres-
5082 sion.
5083
5084
5085 AUTHOR
5086
5087 Philip Hazel
5088 University Computing Service
5089 Cambridge CB2 3QH, England.
5090
5091
5092 REVISION
5093
5094 Last updated: 06 March 2007
5095 Copyright (c) 1997-2007 University of Cambridge.
5096 ------------------------------------------------------------------------------
5097
5098
5099 PCRECPP(3) PCRECPP(3)
5100
5101
5102 NAME
5103 PCRE - Perl-compatible regular expressions.
5104
5105
5106 SYNOPSIS OF C++ WRAPPER
5107
5108 #include <pcrecpp.h>
5109
5110
5111 DESCRIPTION
5112
5113 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5114 functionality was added by Giuseppe Maxia. This brief man page was con-
5115 structed from the notes in the pcrecpp.h file, which should be con-
5116 sulted for further details.
5117
5118
5119 MATCHING INTERFACE
5120
5121 The "FullMatch" operation checks that supplied text matches a supplied
5122 pattern exactly. If pointer arguments are supplied, it copies matched
5123 sub-strings that match sub-patterns into them.
5124
5125 Example: successful match
5126 pcrecpp::RE re("h.*o");
5127 re.FullMatch("hello");
5128
5129 Example: unsuccessful match (requires full match):
5130 pcrecpp::RE re("e");
5131 !re.FullMatch("hello");
5132
5133 Example: creating a temporary RE object:
5134 pcrecpp::RE("h.*o").FullMatch("hello");
5135
5136 You can pass in a "const char*" or a "string" for "text". The examples
5137 below tend to use a const char*. You can, as in the different examples
5138 above, store the RE object explicitly in a variable or use a temporary
5139 RE object. The examples below use one mode or the other arbitrarily.
5140 Either could correctly be used for any of these examples.
5141
5142 You must supply extra pointer arguments to extract matched subpieces.
5143
5144 Example: extracts "ruby" into "s" and 1234 into "i"
5145 int i;
5146 string s;
5147 pcrecpp::RE re("(\\w+):(\\d+)");
5148 re.FullMatch("ruby:1234", &s, &i);
5149
5150 Example: does not try to extract any extra sub-patterns
5151 re.FullMatch("ruby:1234", &s);
5152
5153 Example: does not try to extract into NULL
5154 re.FullMatch("ruby:1234", NULL, &i);
5155
5156 Example: integer overflow causes failure
5157 !re.FullMatch("ruby:1234567891234", NULL, &i);
5158
5159 Example: fails because there aren't enough sub-patterns:
5160 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5161
5162 Example: fails because string cannot be stored in integer
5163 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5164
5165 The provided pointer arguments can be pointers to any scalar numeric
5166 type, or one of:
5167
5168 string (matched piece is copied to string)
5169 StringPiece (StringPiece is mutated to point to matched piece)
5170 T (where "bool T::ParseFrom(const char*, int)" exists)
5171 NULL (the corresponding matched sub-pattern is not copied)
5172
5173 The function returns true iff all of the following conditions are sat-
5174 isfied:
5175
5176 a. "text" matches "pattern" exactly;
5177
5178 b. The number of matched sub-patterns is >= number of supplied
5179 pointers;
5180
5181 c. The "i"th argument has a suitable type for holding the
5182 string captured as the "i"th sub-pattern. If you pass in
5183 NULL for the "i"th argument, or pass fewer arguments than
5184 number of sub-patterns, "i"th captured sub-pattern is
5185 ignored.
5186
5187 CAVEAT: An optional sub-pattern that does not exist in the matched
5188 string is assigned the empty string. Therefore, the following will
5189 return false (because the empty string is not a valid number):
5190
5191 int number;
5192 pcrecpp::RE::FullMatch("abc", "[a-z]+(\d+)?", &number);
5193
5194 The matching interface supports at most 16 arguments per call. If you
5195 need more, consider using the more general interface
5196 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5197
5198
5199 QUOTING METACHARACTERS
5200
5201 You can use the "QuoteMeta" operation to insert backslashes before all
5202 potentially meaningful characters in a string. The returned string,
5203 used as a regular expression, will exactly match the original string.
5204
5205 Example:
5206 string quoted = RE::QuoteMeta(unquoted);
5207
5208 Note that it's legal to escape a character even if it has no special
5209 meaning in a regular expression -- so this function does that. (This
5210 also makes it identical to the perl function of the same name; see
5211 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5212 "1\.5\-2\.0\?".
5213
5214
5215 PARTIAL MATCHES
5216
5217 You can use the "PartialMatch" operation when you want the pattern to
5218 match any substring of the text.
5219
5220 Example: simple search for a string:
5221 pcrecpp::RE("ell").PartialMatch("hello");
5222
5223 Example: find first number in a string:
5224 int number;
5225 pcrecpp::RE re("(\\d+)");
5226 re.PartialMatch("x*100 + 20", &number);
5227 assert(number == 100);
5228
5229
5230 UTF-8 AND THE MATCHING INTERFACE
5231
5232 By default, pattern and text are plain text, one byte per character.
5233 The UTF8 flag, passed to the constructor, causes both pattern and
5234 string to be treated as UTF-8 text, still a byte stream but potentially
5235 multiple bytes per character. In practice, the text is likelier to be
5236 UTF-8 than the pattern, but the match returned may depend on the UTF8
5237 flag, so always use it when matching UTF8 text. For example, "." will
5238 match one byte normally but with UTF8 set may match up to three bytes
5239 of a multi-byte character.
5240
5241 Example:
5242 pcrecpp::RE_Options options;
5243 options.set_utf8();
5244 pcrecpp::RE re(utf8_pattern, options);
5245 re.FullMatch(utf8_string);
5246
5247 Example: using the convenience function UTF8():
5248 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5249 re.FullMatch(utf8_string);
5250
5251 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5252 --enable-utf8 flag.
5253
5254
5255 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5256
5257 PCRE defines some modifiers to change the behavior of the regular
5258 expression engine. The C++ wrapper defines an auxiliary class,
5259 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5260 rently, the following modifiers are supported:
5261
5262 modifier description Perl corresponding
5263
5264 PCRE_CASELESS case insensitive match /i
5265 PCRE_MULTILINE multiple lines match /m
5266 PCRE_DOTALL dot matches newlines /s
5267 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5268 PCRE_EXTRA strict escape parsing N/A
5269 PCRE_EXTENDED ignore whitespaces /x
5270 PCRE_UTF8 handles UTF8 chars built-in
5271 PCRE_UNGREEDY reverses * and *? N/A
5272 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5273
5274 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5275 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5276 ture, while (ab|cd) does.
5277
5278 For a full account on how each modifier works, please check the PCRE
5279 API reference page.
5280
5281 For each modifier, there are two member functions whose name is made
5282 out of the modifier in lowercase, without the "PCRE_" prefix. For
5283 instance, PCRE_CASELESS is handled by
5284
5285 bool caseless()
5286
5287 which returns true if the modifier is set, and
5288
5289 RE_Options & set_caseless(bool)
5290
5291 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5292 be accessed through the set_match_limit() and match_limit() member
5293 functions. Setting match_limit to a non-zero value will limit the exe-
5294 cution of pcre to keep it from doing bad things like blowing the stack
5295 or taking an eternity to return a result. A value of 5000 is good
5296 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5297 to zero disables match limiting. Alternatively, you can call
5298 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5299 limit how much PCRE recurses. match_limit() limits the number of
5300 matches PCRE does; match_limit_recursion() limits the depth of internal
5301 recursion, and therefore the amount of stack that is used.
5302
5303 Normally, to pass one or more modifiers to a RE class, you declare a
5304 RE_Options object, set the appropriate options, and pass this object to
5305 a RE constructor. Example:
5306
5307 RE_options opt;
5308 opt.set_caseless(true);
5309 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5310
5311 RE_options has two constructors. The default constructor takes no argu-
5312 ments and creates a set of flags that are off by default. The optional
5313 parameter option_flags is to facilitate transfer of legacy code from C
5314 programs. This lets you do
5315
5316 RE(pattern,
5317 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5318
5319 However, new code is better off doing
5320
5321 RE(pattern,
5322 RE_Options().set_caseless(true).set_multiline(true))
5323 .PartialMatch(str);
5324
5325 If you are going to pass one of the most used modifiers, there are some
5326 convenience functions that return a RE_Options class with the appropri-
5327 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5328 and EXTENDED().
5329
5330 If you need to set several options at once, and you don't want to go
5331 through the pains of declaring a RE_Options object and setting several
5332 options, there is a parallel method that give you such ability on the
5333 fly. You can concatenate several set_xxxxx() member functions, since
5334 each of them returns a reference to its class object. For example, to
5335 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5336 statement, you may write:
5337
5338 RE(" ^ xyz \\s+ .* blah$",
5339 RE_Options()
5340 .set_caseless(true)
5341 .set_extended(true)
5342 .set_multiline(true)).PartialMatch(sometext);
5343
5344
5345 SCANNING TEXT INCREMENTALLY
5346
5347 The "Consume" operation may be useful if you want to repeatedly match
5348 regular expressions at the front of a string and skip over them as they
5349 match. This requires use of the "StringPiece" type, which represents a
5350 sub-range of a real string. Like RE, StringPiece is defined in the
5351 pcrecpp namespace.
5352
5353 Example: read lines of the form "var = value" from a string.
5354 string contents = ...; // Fill string somehow
5355 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
5356
5357 string var;
5358 int value;
5359 pcrecpp::RE re("(\\w+) = (\\d+)\n");
5360 while (re.Consume(&input, &var, &value)) {
5361 ...;
5362 }
5363
5364 Each successful call to "Consume" will set "var/value", and also
5365 advance "input" so it points past the matched text.
5366
5367 The "FindAndConsume" operation is similar to "Consume" but does not
5368 anchor your match at the beginning of the string. For example, you
5369 could extract all words from a string by repeatedly calling
5370
5371 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
5372
5373
5374 PARSING HEX/OCTAL/C-RADIX NUMBERS
5375
5376 By default, if you pass a pointer to a numeric value, the corresponding
5377 text is interpreted as a base-10 number. You can instead wrap the
5378 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
5379 to interpret the text in another base. The CRadix operator interprets
5380 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
5381 base-10.
5382
5383 Example:
5384 int a, b, c, d;
5385 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
5386 re.FullMatch("100 40 0100 0x40",
5387 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
5388 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
5389
5390 will leave 64 in a, b, c, and d.
5391
5392
5393 REPLACING PARTS OF STRINGS
5394
5395 You can replace the first match of "pattern" in "str" with "rewrite".
5396 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
5397 insert text matching corresponding parenthesized group from the pat-
5398 tern. \0 in "rewrite" refers to the entire matching text. For example:
5399
5400 string s = "yabba dabba doo";
5401 pcrecpp::RE("b+").Replace("d", &s);
5402
5403 will leave "s" containing "yada dabba doo". The result is true if the
5404 pattern matches and a replacement occurs, false otherwise.
5405
5406 GlobalReplace is like Replace except that it replaces all occurrences
5407 of the pattern in the string with the rewrite. Replacements are not
5408 subject to re-matching. For example:
5409
5410 string s = "yabba dabba doo";
5411 pcrecpp::RE("b+").GlobalReplace("d", &s);
5412
5413 will leave "s" containing "yada dada doo". It returns the number of
5414 replacements made.
5415
5416 Extract is like Replace, except that if the pattern matches, "rewrite"
5417 is copied into "out" (an additional argument) with substitutions. The
5418 non-matching portions of "text" are ignored. Returns true iff a match
5419 occurred and the extraction happened successfully; if no match occurs,
5420 the string is left unaffected.
5421
5422
5423 AUTHOR
5424
5425 The C++ wrapper was contributed by Google Inc.
5426 Copyright (c) 2007 Google Inc.
5427
5428
5429 REVISION
5430
5431 Last updated: 06 March 2007
5432 ------------------------------------------------------------------------------
5433
5434
5435 PCRESAMPLE(3) PCRESAMPLE(3)
5436
5437
5438 NAME
5439 PCRE - Perl-compatible regular expressions
5440
5441
5442 PCRE SAMPLE PROGRAM
5443
5444 A simple, complete demonstration program, to get you started with using
5445 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
5446
5447 The program compiles the regular expression that is its first argument,
5448 and matches it against the subject string in its second argument. No
5449 PCRE options are set, and default character tables are used. If match-
5450 ing succeeds, the program outputs the portion of the subject that
5451 matched, together with the contents of any captured substrings.
5452
5453 If the -g option is given on the command line, the program then goes on
5454 to check for further matches of the same regular expression in the same
5455 subject string. The logic is a little bit tricky because of the possi-
5456 bility of matching an empty string. Comments in the code explain what
5457 is going on.
5458
5459 If PCRE is installed in the standard include and library directories
5460 for your system, you should be able to compile the demonstration pro-
5461 gram using this command:
5462
5463 gcc -o pcredemo pcredemo.c -lpcre
5464
5465 If PCRE is installed elsewhere, you may need to add additional options
5466 to the command line. For example, on a Unix-like system that has PCRE
5467 installed in /usr/local, you can compile the demonstration program
5468 using a command like this:
5469
5470 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5471 -L/usr/local/lib -lpcre
5472
5473 Once you have compiled the demonstration program, you can run simple
5474 tests like this:
5475
5476 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5477 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5478
5479 Note that there is a much more comprehensive test program, called
5480 pcretest, which supports many more facilities for testing regular
5481 expressions and the PCRE library. The pcredemo program is provided as a
5482 simple coding example.
5483
5484 On some operating systems (e.g. Solaris), when PCRE is not installed in
5485 the standard library directory, you may get an error like this when you
5486 try to run pcredemo:
5487
5488 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5489 directory
5490
5491 This is caused by the way shared library support works on those sys-
5492 tems. You need to add
5493
5494 -R/usr/local/lib
5495
5496 (for example) to the compile command to get round this problem.
5497
5498
5499 AUTHOR
5500
5501 Philip Hazel
5502 University Computing Service
5503 Cambridge CB2 3QH, England.
5504
5505
5506 REVISION
5507
5508 Last updated: 06 March 2007
5509 Copyright (c) 1997-2007 University of Cambridge.
5510 ------------------------------------------------------------------------------
5511 PCRESTACK(3) PCRESTACK(3)
5512
5513
5514 NAME
5515 PCRE - Perl-compatible regular expressions
5516
5517
5518 PCRE DISCUSSION OF STACK USAGE
5519
5520 When you call pcre_exec(), it makes use of an internal function called
5521 match(). This calls itself recursively at branch points in the pattern,
5522 in order to remember the state of the match so that it can back up and
5523 try a different alternative if the first one fails. As matching pro-
5524 ceeds deeper and deeper into the tree of possibilities, the recursion
5525 depth increases.
5526
5527 Not all calls of match() increase the recursion depth; for an item such
5528 as a* it may be called several times at the same level, after matching
5529 different numbers of a's. Furthermore, in a number of cases where the
5530 result of the recursive call would immediately be passed back as the
5531 result of the current call (a "tail recursion"), the function is just
5532 restarted instead.
5533
5534 The pcre_dfa_exec() function operates in an entirely different way, and
5535 hardly uses recursion at all. The limit on its complexity is the amount
5536 of workspace it is given. The comments that follow do NOT apply to
5537 pcre_dfa_exec(); they are relevant only for pcre_exec().
5538
5539 You can set limits on the number of times that match() is called, both
5540 in total and recursively. If the limit is exceeded, an error occurs.
5541 For details, see the section on extra data for pcre_exec() in the
5542 pcreapi documentation.
5543
5544 Each time that match() is actually called recursively, it uses memory
5545 from the process stack. For certain kinds of pattern and data, very
5546 large amounts of stack may be needed, despite the recognition of "tail
5547 recursion". You can often reduce the amount of recursion, and there-
5548 fore the amount of stack used, by modifying the pattern that is being
5549 matched. Consider, for example, this pattern:
5550
5551 ([^<]|<(?!inet))+
5552
5553 It matches from wherever it starts until it encounters "<inet" or the
5554 end of the data, and is the kind of pattern that might be used when
5555 processing an XML file. Each iteration of the outer parentheses matches
5556 either one character that is not "<" or a "<" that is not followed by
5557 "inet". However, each time a parenthesis is processed, a recursion
5558 occurs, so this formulation uses a stack frame for each matched charac-
5559 ter. For a long string, a lot of stack is required. Consider now this
5560 rewritten pattern, which matches exactly the same strings:
5561
5562 ([^<]++|<(?!inet))+
5563
5564 This uses very much less stack, because runs of characters that do not
5565 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5566 sion happens only when a "<" character that is not followed by "inet"
5567 is encountered (and we assume this is relatively rare). A possessive
5568 quantifier is used to stop any backtracking into the runs of non-"<"
5569 characters, but that is not related to stack usage.
5570
5571 This example shows that one way of avoiding stack problems when match-
5572 ing long subject strings is to write repeated parenthesized subpatterns
5573 to match more than one character whenever possible.
5574
5575 In environments where stack memory is constrained, you might want to
5576 compile PCRE to use heap memory instead of stack for remembering back-
5577 up points. This makes it run a lot more slowly, however. Details of how
5578 to do this are given in the pcrebuild documentation.
5579
5580 In Unix-like environments, there is not often a problem with the stack
5581 unless very long strings are involved, though the default limit on
5582 stack size varies from system to system. Values from 8Mb to 64Mb are
5583 common. You can find your default limit by running the command:
5584
5585 ulimit -s
5586
5587 Unfortunately, the effect of running out of stack is often SIGSEGV,
5588 though sometimes a more explicit error message is given. You can nor-
5589 mally increase the limit on stack size by code such as this:
5590
5591 struct rlimit rlim;
5592 getrlimit(RLIMIT_STACK, &rlim);
5593 rlim.rlim_cur = 100*1024*1024;
5594 setrlimit(RLIMIT_STACK, &rlim);
5595
5596 This reads the current limits (soft and hard) using getrlimit(), then
5597 attempts to increase the soft limit to 100Mb using setrlimit(). You
5598 must do this before calling pcre_exec().
5599
5600 PCRE has an internal counter that can be used to limit the depth of
5601 recursion, and thus cause pcre_exec() to give an error code before it
5602 runs out of stack. By default, the limit is very large, and unlikely
5603 ever to operate. It can be changed when PCRE is built, and it can also
5604 be set when pcre_exec() is called. For details of these interfaces, see
5605 the pcrebuild and pcreapi documentation.
5606
5607 As a very rough rule of thumb, you should reckon on about 500 bytes per
5608 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5609 should set the limit at 16000 recursions. A 64Mb stack, on the other
5610 hand, can support around 128000 recursions. The pcretest test program
5611 has a command line option (-S) that can be used to increase the size of
5612 its stack.
5613
5614
5615 AUTHOR
5616
5617 Philip Hazel
5618 University Computing Service
5619 Cambridge CB2 3QH, England.
5620
5621
5622 REVISION
5623
5624 Last updated: 12 March 2007
5625 Copyright (c) 1997-2007 University of Cambridge.
5626 ------------------------------------------------------------------------------
5627
5628

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5