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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 128 - (show annotations)
Tue Mar 20 11:46:50 2007 UTC (12 years, 7 months ago) by ph10
File MIME type: text/plain
File size: 265674 byte(s)
Make the use of dftables optional, and not the default. Further tidies to 
documentation.
1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. (Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages.
49
50 Some features of PCRE can be included, excluded, or changed when the
51 library is built. The pcre_config() function makes it possible for a
52 client to discover which features are available. The features them-
53 selves are described in the pcrebuild page. Documentation about build-
54 ing PCRE for various operating systems can be found in the README file
55 in the source distribution.
56
57 The library contains a number of undocumented internal functions and
58 data tables that are used by more than one of the exported external
59 functions, but which are not intended for use by external callers.
60 Their names all begin with "_pcre_", which hopefully will not provoke
61 any name clashes. In some environments, it is possible to control which
62 external symbols are exported when a shared library is built, and in
63 these cases the undocumented symbols are not exported.
64
65
66 USER DOCUMENTATION
67
68 The user documentation for PCRE comprises a number of different sec-
69 tions. In the "man" format, each of these is a separate "man page". In
70 the HTML format, each is a separate page, linked from the index page.
71 In the plain text format, all the sections are concatenated, for ease
72 of searching. The sections are as follows:
73
74 pcre this document
75 pcreapi details of PCRE's native C API
76 pcrebuild options for building PCRE
77 pcrecallout details of the callout feature
78 pcrecompat discussion of Perl compatibility
79 pcrecpp details of the C++ wrapper
80 pcregrep description of the pcregrep command
81 pcrematching discussion of the two matching algorithms
82 pcrepartial details of the partial matching facility
83 pcrepattern syntax and semantics of supported
84 regular expressions
85 pcreperform discussion of performance issues
86 pcreposix the POSIX-compatible C API
87 pcreprecompile details of saving and re-using precompiled patterns
88 pcresample discussion of the sample program
89 pcrestack discussion of stack usage
90 pcretest description of the pcretest testing command
91
92 In addition, in the "man" and HTML formats, there is a short page for
93 each C library function, listing its arguments and results.
94
95
96 LIMITATIONS
97
98 There are some size limitations in PCRE but it is hoped that they will
99 never in practice be relevant.
100
101 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
102 is compiled with the default internal linkage size of 2. If you want to
103 process regular expressions that are truly enormous, you can compile
104 PCRE with an internal linkage size of 3 or 4 (see the README file in
105 the source distribution and the pcrebuild documentation for details).
106 In these cases the limit is substantially larger. However, the speed
107 of execution is slower.
108
109 All values in repeating quantifiers must be less than 65536. The maxi-
110 mum compiled length of subpattern with an explicit repeat count is
111 30000 bytes. The maximum number of capturing subpatterns is 65535.
112
113 There is no limit to the number of parenthesized subpatterns, but there
114 can be no more than 65535 capturing subpatterns.
115
116 The maximum length of name for a named subpattern is 32 characters, and
117 the maximum number of named subpatterns is 10000.
118
119 The maximum length of a subject string is the largest positive number
120 that an integer variable can hold. However, when using the traditional
121 matching function, PCRE uses recursion to handle subpatterns and indef-
122 inite repetition. This means that the available stack space may limit
123 the size of a subject string that can be processed by certain patterns.
124 For a discussion of stack issues, see the pcrestack documentation.
125
126
127 UTF-8 AND UNICODE PROPERTY SUPPORT
128
129 From release 3.3, PCRE has had some support for character strings
130 encoded in the UTF-8 format. For release 4.0 this was greatly extended
131 to cover most common requirements, and in release 5.0 additional sup-
132 port for Unicode general category properties was added.
133
134 In order process UTF-8 strings, you must build PCRE to include UTF-8
135 support in the code, and, in addition, you must call pcre_compile()
136 with the PCRE_UTF8 option flag. When you do this, both the pattern and
137 any subject strings that are matched against it are treated as UTF-8
138 strings instead of just strings of bytes.
139
140 If you compile PCRE with UTF-8 support, but do not use it at run time,
141 the library will be a bit bigger, but the additional run time overhead
142 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
143 very big.
144
145 If PCRE is built with Unicode character property support (which implies
146 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
147 ported. The available properties that can be tested are limited to the
148 general category properties such as Lu for an upper case letter or Nd
149 for a decimal number, the Unicode script names such as Arabic or Han,
150 and the derived properties Any and L&. A full list is given in the
151 pcrepattern documentation. Only the short names for properties are sup-
152 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
153 ter}, is not supported. Furthermore, in Perl, many properties may
154 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
155 does not support this.
156
157 The following comments apply when PCRE is running in UTF-8 mode:
158
159 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
160 subjects are checked for validity on entry to the relevant functions.
161 If an invalid UTF-8 string is passed, an error return is given. In some
162 situations, you may already know that your strings are valid, and
163 therefore want to skip these checks in order to improve performance. If
164 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
165 PCRE assumes that the pattern or subject it is given (respectively)
166 contains only valid UTF-8 codes. In this case, it does not diagnose an
167 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
168 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
169 crash.
170
171 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
172 two-byte UTF-8 character if the value is greater than 127.
173
174 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
175 characters for values greater than \177.
176
177 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
178 vidual bytes, for example: \x{100}{3}.
179
180 5. The dot metacharacter matches one UTF-8 character instead of a sin-
181 gle byte.
182
183 6. The escape sequence \C can be used to match a single byte in UTF-8
184 mode, but its use can lead to some strange effects. This facility is
185 not available in the alternative matching function, pcre_dfa_exec().
186
187 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
188 test characters of any code value, but the characters that PCRE recog-
189 nizes as digits, spaces, or word characters remain the same set as
190 before, all with values less than 256. This remains true even when PCRE
191 includes Unicode property support, because to do otherwise would slow
192 down PCRE in many common cases. If you really want to test for a wider
193 sense of, say, "digit", you must use Unicode property tests such as
194 \p{Nd}.
195
196 8. Similarly, characters that match the POSIX named character classes
197 are all low-valued characters.
198
199 9. Case-insensitive matching applies only to characters whose values
200 are less than 128, unless PCRE is built with Unicode property support.
201 Even when Unicode property support is available, PCRE still uses its
202 own character tables when checking the case of low-valued characters,
203 so as not to degrade performance. The Unicode property information is
204 used only for characters with higher values. Even when Unicode property
205 support is available, PCRE supports case-insensitive matching only when
206 there is a one-to-one mapping between a letter's cases. There are a
207 small number of many-to-one mappings in Unicode; these are not sup-
208 ported by PCRE.
209
210
211 AUTHOR
212
213 Philip Hazel
214 University Computing Service
215 Cambridge CB2 3QH, England.
216
217 Putting an actual email address here seems to have been a spam magnet,
218 so I've taken it away. If you want to email me, use my initial and sur-
219 name, separated by a dot, at the domain ucs.cam.ac.uk.
220
221
222 REVISION
223
224 Last updated: 06 March 2007
225 Copyright (c) 1997-2007 University of Cambridge.
226 ------------------------------------------------------------------------------
227
228
229 PCREBUILD(3) PCREBUILD(3)
230
231
232 NAME
233 PCRE - Perl-compatible regular expressions
234
235
236 PCRE BUILD-TIME OPTIONS
237
238 This document describes the optional features of PCRE that can be
239 selected when the library is compiled. They are all selected, or dese-
240 lected, by providing options to the configure script that is run before
241 the make command. The complete list of options for configure (which
242 includes the standard ones such as the selection of the installation
243 directory) can be obtained by running
244
245 ./configure --help
246
247 The following sections include descriptions of options whose names
248 begin with --enable or --disable. These settings specify changes to the
249 defaults for the configure command. Because of the way that configure
250 works, --enable and --disable always come in pairs, so the complemen-
251 tary option always exists as well, but as it specifies the default, it
252 is not described.
253
254
255 C++ SUPPORT
256
257 By default, the configure script will search for a C++ compiler and C++
258 header files. If it finds them, it automatically builds the C++ wrapper
259 library for PCRE. You can disable this by adding
260
261 --disable-cpp
262
263 to the configure command.
264
265
266 UTF-8 SUPPORT
267
268 To build PCRE with support for UTF-8 character strings, add
269
270 --enable-utf8
271
272 to the configure command. Of itself, this does not make PCRE treat
273 strings as UTF-8. As well as compiling PCRE with this option, you also
274 have have to set the PCRE_UTF8 option when you call the pcre_compile()
275 function.
276
277
278 UNICODE CHARACTER PROPERTY SUPPORT
279
280 UTF-8 support allows PCRE to process character values greater than 255
281 in the strings that it handles. On its own, however, it does not pro-
282 vide any facilities for accessing the properties of such characters. If
283 you want to be able to use the pattern escapes \P, \p, and \X, which
284 refer to Unicode character properties, you must add
285
286 --enable-unicode-properties
287
288 to the configure command. This implies UTF-8 support, even if you have
289 not explicitly requested it.
290
291 Including Unicode property support adds around 30K of tables to the
292 PCRE library. Only the general category properties such as Lu and Nd
293 are supported. Details are given in the pcrepattern documentation.
294
295
296 CODE VALUE OF NEWLINE
297
298 By default, PCRE interprets character 10 (linefeed, LF) as indicating
299 the end of a line. This is the normal newline character on Unix-like
300 systems. You can compile PCRE to use character 13 (carriage return, CR)
301 instead, by adding
302
303 --enable-newline-is-cr
304
305 to the configure command. There is also a --enable-newline-is-lf
306 option, which explicitly specifies linefeed as the newline character.
307
308 Alternatively, you can specify that line endings are to be indicated by
309 the two character sequence CRLF. If you want this, add
310
311 --enable-newline-is-crlf
312
313 to the configure command. There is a fourth option, specified by
314
315 --enable-newline-is-any
316
317 which causes PCRE to recognize any Unicode newline sequence.
318
319 Whatever line ending convention is selected when PCRE is built can be
320 overridden when the library functions are called. At build time it is
321 conventional to use the standard for your operating system.
322
323
324 BUILDING SHARED AND STATIC LIBRARIES
325
326 The PCRE building process uses libtool to build both shared and static
327 Unix libraries by default. You can suppress one of these by adding one
328 of
329
330 --disable-shared
331 --disable-static
332
333 to the configure command, as required.
334
335
336 POSIX MALLOC USAGE
337
338 When PCRE is called through the POSIX interface (see the pcreposix doc-
339 umentation), additional working storage is required for holding the
340 pointers to capturing substrings, because PCRE requires three integers
341 per substring, whereas the POSIX interface provides only two. If the
342 number of expected substrings is small, the wrapper function uses space
343 on the stack, because this is faster than using malloc() for each call.
344 The default threshold above which the stack is no longer used is 10; it
345 can be changed by adding a setting such as
346
347 --with-posix-malloc-threshold=20
348
349 to the configure command.
350
351
352 HANDLING VERY LARGE PATTERNS
353
354 Within a compiled pattern, offset values are used to point from one
355 part to another (for example, from an opening parenthesis to an alter-
356 nation metacharacter). By default, two-byte values are used for these
357 offsets, leading to a maximum size for a compiled pattern of around
358 64K. This is sufficient to handle all but the most gigantic patterns.
359 Nevertheless, some people do want to process enormous patterns, so it
360 is possible to compile PCRE to use three-byte or four-byte offsets by
361 adding a setting such as
362
363 --with-link-size=3
364
365 to the configure command. The value given must be 2, 3, or 4. Using
366 longer offsets slows down the operation of PCRE because it has to load
367 additional bytes when handling them.
368
369
370 AVOIDING EXCESSIVE STACK USAGE
371
372 When matching with the pcre_exec() function, PCRE implements backtrack-
373 ing by making recursive calls to an internal function called match().
374 In environments where the size of the stack is limited, this can se-
375 verely limit PCRE's operation. (The Unix environment does not usually
376 suffer from this problem, but it may sometimes be necessary to increase
377 the maximum stack size. There is a discussion in the pcrestack docu-
378 mentation.) An alternative approach to recursion that uses memory from
379 the heap to remember data, instead of using recursive function calls,
380 has been implemented to work round the problem of limited stack size.
381 If you want to build a version of PCRE that works this way, add
382
383 --disable-stack-for-recursion
384
385 to the configure command. With this configuration, PCRE will use the
386 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
387 ment functions. Separate functions are provided because the usage is
388 very predictable: the block sizes requested are always the same, and
389 the blocks are always freed in reverse order. A calling program might
390 be able to implement optimized functions that perform better than the
391 standard malloc() and free() functions. PCRE runs noticeably more
392 slowly when built in this way. This option affects only the pcre_exec()
393 function; it is not relevant for the the pcre_dfa_exec() function.
394
395
396 LIMITING PCRE RESOURCE USAGE
397
398 Internally, PCRE has a function called match(), which it calls repeat-
399 edly (sometimes recursively) when matching a pattern with the
400 pcre_exec() function. By controlling the maximum number of times this
401 function may be called during a single matching operation, a limit can
402 be placed on the resources used by a single call to pcre_exec(). The
403 limit can be changed at run time, as described in the pcreapi documen-
404 tation. The default is 10 million, but this can be changed by adding a
405 setting such as
406
407 --with-match-limit=500000
408
409 to the configure command. This setting has no effect on the
410 pcre_dfa_exec() matching function.
411
412 In some environments it is desirable to limit the depth of recursive
413 calls of match() more strictly than the total number of calls, in order
414 to restrict the maximum amount of stack (or heap, if --disable-stack-
415 for-recursion is specified) that is used. A second limit controls this;
416 it defaults to the value that is set for --with-match-limit, which
417 imposes no additional constraints. However, you can set a lower limit
418 by adding, for example,
419
420 --with-match-limit-recursion=10000
421
422 to the configure command. This value can also be overridden at run
423 time.
424
425
426 CREATING CHARACTER TABLES AT BUILD TIME
427
428 PCRE uses fixed tables for processing characters whose code values are
429 less than 256. By default, PCRE is built with a set of tables that are
430 distributed in the file pcre_chartables.c.dist. These tables are for
431 ASCII codes only. If you add
432
433 --enable-rebuild-chartables
434
435 to the configure command, the distributed tables are no longer used.
436 Instead, a program called dftables is compiled and run. This outputs
437 the source for new set of tables, created in the default locale of your
438 C runtime system. (This method of replacing the tables does not work if
439 you are cross compiling, because dftables is run on the local host. If
440 you need to create alternative tables when cross compiling, you will
441 have to do so "by hand".)
442
443
444 USING EBCDIC CODE
445
446 PCRE assumes by default that it will run in an environment where the
447 character code is ASCII (or Unicode, which is a superset of ASCII).
448 PCRE can, however, be compiled to run in an EBCDIC environment by
449 adding
450
451 --enable-ebcdic
452
453 to the configure command. This setting implies --enable-rebuild-charta-
454 bles.
455
456
457 SEE ALSO
458
459 pcreapi(3), pcre_config(3).
460
461
462 AUTHOR
463
464 Philip Hazel
465 University Computing Service
466 Cambridge CB2 3QH, England.
467
468
469 REVISION
470
471 Last updated: 20 March 2007
472 Copyright (c) 1997-2007 University of Cambridge.
473 ------------------------------------------------------------------------------
474
475
476 PCREMATCHING(3) PCREMATCHING(3)
477
478
479 NAME
480 PCRE - Perl-compatible regular expressions
481
482
483 PCRE MATCHING ALGORITHMS
484
485 This document describes the two different algorithms that are available
486 in PCRE for matching a compiled regular expression against a given sub-
487 ject string. The "standard" algorithm is the one provided by the
488 pcre_exec() function. This works in the same was as Perl's matching
489 function, and provides a Perl-compatible matching operation.
490
491 An alternative algorithm is provided by the pcre_dfa_exec() function;
492 this operates in a different way, and is not Perl-compatible. It has
493 advantages and disadvantages compared with the standard algorithm, and
494 these are described below.
495
496 When there is only one possible way in which a given subject string can
497 match a pattern, the two algorithms give the same answer. A difference
498 arises, however, when there are multiple possibilities. For example, if
499 the pattern
500
501 ^<.*>
502
503 is matched against the string
504
505 <something> <something else> <something further>
506
507 there are three possible answers. The standard algorithm finds only one
508 of them, whereas the alternative algorithm finds all three.
509
510
511 REGULAR EXPRESSIONS AS TREES
512
513 The set of strings that are matched by a regular expression can be rep-
514 resented as a tree structure. An unlimited repetition in the pattern
515 makes the tree of infinite size, but it is still a tree. Matching the
516 pattern to a given subject string (from a given starting point) can be
517 thought of as a search of the tree. There are two ways to search a
518 tree: depth-first and breadth-first, and these correspond to the two
519 matching algorithms provided by PCRE.
520
521
522 THE STANDARD MATCHING ALGORITHM
523
524 In the terminology of Jeffrey Friedl's book Mastering Regular Expres-
525 sions, the standard algorithm is an "NFA algorithm". It conducts a
526 depth-first search of the pattern tree. That is, it proceeds along a
527 single path through the tree, checking that the subject matches what is
528 required. When there is a mismatch, the algorithm tries any alterna-
529 tives at the current point, and if they all fail, it backs up to the
530 previous branch point in the tree, and tries the next alternative
531 branch at that level. This often involves backing up (moving to the
532 left) in the subject string as well. The order in which repetition
533 branches are tried is controlled by the greedy or ungreedy nature of
534 the quantifier.
535
536 If a leaf node is reached, a matching string has been found, and at
537 that point the algorithm stops. Thus, if there is more than one possi-
538 ble match, this algorithm returns the first one that it finds. Whether
539 this is the shortest, the longest, or some intermediate length depends
540 on the way the greedy and ungreedy repetition quantifiers are specified
541 in the pattern.
542
543 Because it ends up with a single path through the tree, it is rela-
544 tively straightforward for this algorithm to keep track of the sub-
545 strings that are matched by portions of the pattern in parentheses.
546 This provides support for capturing parentheses and back references.
547
548
549 THE ALTERNATIVE MATCHING ALGORITHM
550
551 This algorithm conducts a breadth-first search of the tree. Starting
552 from the first matching point in the subject, it scans the subject
553 string from left to right, once, character by character, and as it does
554 this, it remembers all the paths through the tree that represent valid
555 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
556 though it is not implemented as a traditional finite state machine (it
557 keeps multiple states active simultaneously).
558
559 The scan continues until either the end of the subject is reached, or
560 there are no more unterminated paths. At this point, terminated paths
561 represent the different matching possibilities (if there are none, the
562 match has failed). Thus, if there is more than one possible match,
563 this algorithm finds all of them, and in particular, it finds the long-
564 est. In PCRE, there is an option to stop the algorithm after the first
565 match (which is necessarily the shortest) has been found.
566
567 Note that all the matches that are found start at the same point in the
568 subject. If the pattern
569
570 cat(er(pillar)?)
571
572 is matched against the string "the caterpillar catchment", the result
573 will be the three strings "cat", "cater", and "caterpillar" that start
574 at the fourth character of the subject. The algorithm does not automat-
575 ically move on to find matches that start at later positions.
576
577 There are a number of features of PCRE regular expressions that are not
578 supported by the alternative matching algorithm. They are as follows:
579
580 1. Because the algorithm finds all possible matches, the greedy or
581 ungreedy nature of repetition quantifiers is not relevant. Greedy and
582 ungreedy quantifiers are treated in exactly the same way. However, pos-
583 sessive quantifiers can make a difference when what follows could also
584 match what is quantified, for example in a pattern like this:
585
586 ^a++\w!
587
588 This pattern matches "aaab!" but not "aaa!", which would be matched by
589 a non-possessive quantifier. Similarly, if an atomic group is present,
590 it is matched as if it were a standalone pattern at the current point,
591 and the longest match is then "locked in" for the rest of the overall
592 pattern.
593
594 2. When dealing with multiple paths through the tree simultaneously, it
595 is not straightforward to keep track of captured substrings for the
596 different matching possibilities, and PCRE's implementation of this
597 algorithm does not attempt to do this. This means that no captured sub-
598 strings are available.
599
600 3. Because no substrings are captured, back references within the pat-
601 tern are not supported, and cause errors if encountered.
602
603 4. For the same reason, conditional expressions that use a backrefer-
604 ence as the condition or test for a specific group recursion are not
605 supported.
606
607 5. Callouts are supported, but the value of the capture_top field is
608 always 1, and the value of the capture_last field is always -1.
609
610 6. The \C escape sequence, which (in the standard algorithm) matches a
611 single byte, even in UTF-8 mode, is not supported because the alterna-
612 tive algorithm moves through the subject string one character at a
613 time, for all active paths through the tree.
614
615
616 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
617
618 Using the alternative matching algorithm provides the following advan-
619 tages:
620
621 1. All possible matches (at a single point in the subject) are automat-
622 ically found, and in particular, the longest match is found. To find
623 more than one match using the standard algorithm, you have to do kludgy
624 things with callouts.
625
626 2. There is much better support for partial matching. The restrictions
627 on the content of the pattern that apply when using the standard algo-
628 rithm for partial matching do not apply to the alternative algorithm.
629 For non-anchored patterns, the starting position of a partial match is
630 available.
631
632 3. Because the alternative algorithm scans the subject string just
633 once, and never needs to backtrack, it is possible to pass very long
634 subject strings to the matching function in several pieces, checking
635 for partial matching each time.
636
637
638 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
639
640 The alternative algorithm suffers from a number of disadvantages:
641
642 1. It is substantially slower than the standard algorithm. This is
643 partly because it has to search for all possible matches, but is also
644 because it is less susceptible to optimization.
645
646 2. Capturing parentheses and back references are not supported.
647
648 3. Although atomic groups are supported, their use does not provide the
649 performance advantage that it does for the standard algorithm.
650
651
652 AUTHOR
653
654 Philip Hazel
655 University Computing Service
656 Cambridge CB2 3QH, England.
657
658
659 REVISION
660
661 Last updated: 06 March 2007
662 Copyright (c) 1997-2007 University of Cambridge.
663 ------------------------------------------------------------------------------
664
665
666 PCREAPI(3) PCREAPI(3)
667
668
669 NAME
670 PCRE - Perl-compatible regular expressions
671
672
673 PCRE NATIVE API
674
675 #include <pcre.h>
676
677 pcre *pcre_compile(const char *pattern, int options,
678 const char **errptr, int *erroffset,
679 const unsigned char *tableptr);
680
681 pcre *pcre_compile2(const char *pattern, int options,
682 int *errorcodeptr,
683 const char **errptr, int *erroffset,
684 const unsigned char *tableptr);
685
686 pcre_extra *pcre_study(const pcre *code, int options,
687 const char **errptr);
688
689 int pcre_exec(const pcre *code, const pcre_extra *extra,
690 const char *subject, int length, int startoffset,
691 int options, int *ovector, int ovecsize);
692
693 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
694 const char *subject, int length, int startoffset,
695 int options, int *ovector, int ovecsize,
696 int *workspace, int wscount);
697
698 int pcre_copy_named_substring(const pcre *code,
699 const char *subject, int *ovector,
700 int stringcount, const char *stringname,
701 char *buffer, int buffersize);
702
703 int pcre_copy_substring(const char *subject, int *ovector,
704 int stringcount, int stringnumber, char *buffer,
705 int buffersize);
706
707 int pcre_get_named_substring(const pcre *code,
708 const char *subject, int *ovector,
709 int stringcount, const char *stringname,
710 const char **stringptr);
711
712 int pcre_get_stringnumber(const pcre *code,
713 const char *name);
714
715 int pcre_get_stringtable_entries(const pcre *code,
716 const char *name, char **first, char **last);
717
718 int pcre_get_substring(const char *subject, int *ovector,
719 int stringcount, int stringnumber,
720 const char **stringptr);
721
722 int pcre_get_substring_list(const char *subject,
723 int *ovector, int stringcount, const char ***listptr);
724
725 void pcre_free_substring(const char *stringptr);
726
727 void pcre_free_substring_list(const char **stringptr);
728
729 const unsigned char *pcre_maketables(void);
730
731 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
732 int what, void *where);
733
734 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
735
736 int pcre_refcount(pcre *code, int adjust);
737
738 int pcre_config(int what, void *where);
739
740 char *pcre_version(void);
741
742 void *(*pcre_malloc)(size_t);
743
744 void (*pcre_free)(void *);
745
746 void *(*pcre_stack_malloc)(size_t);
747
748 void (*pcre_stack_free)(void *);
749
750 int (*pcre_callout)(pcre_callout_block *);
751
752
753 PCRE API OVERVIEW
754
755 PCRE has its own native API, which is described in this document. There
756 are also some wrapper functions that correspond to the POSIX regular
757 expression API. These are described in the pcreposix documentation.
758 Both of these APIs define a set of C function calls. A C++ wrapper is
759 distributed with PCRE. It is documented in the pcrecpp page.
760
761 The native API C function prototypes are defined in the header file
762 pcre.h, and on Unix systems the library itself is called libpcre. It
763 can normally be accessed by adding -lpcre to the command for linking an
764 application that uses PCRE. The header file defines the macros
765 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
766 bers for the library. Applications can use these to include support
767 for different releases of PCRE.
768
769 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
770 pcre_exec() are used for compiling and matching regular expressions in
771 a Perl-compatible manner. A sample program that demonstrates the sim-
772 plest way of using them is provided in the file called pcredemo.c in
773 the source distribution. The pcresample documentation describes how to
774 run it.
775
776 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
777 ble, is also provided. This uses a different algorithm for the match-
778 ing. The alternative algorithm finds all possible matches (at a given
779 point in the subject), and scans the subject just once. However, this
780 algorithm does not return captured substrings. A description of the two
781 matching algorithms and their advantages and disadvantages is given in
782 the pcrematching documentation.
783
784 In addition to the main compiling and matching functions, there are
785 convenience functions for extracting captured substrings from a subject
786 string that is matched by pcre_exec(). They are:
787
788 pcre_copy_substring()
789 pcre_copy_named_substring()
790 pcre_get_substring()
791 pcre_get_named_substring()
792 pcre_get_substring_list()
793 pcre_get_stringnumber()
794 pcre_get_stringtable_entries()
795
796 pcre_free_substring() and pcre_free_substring_list() are also provided,
797 to free the memory used for extracted strings.
798
799 The function pcre_maketables() is used to build a set of character
800 tables in the current locale for passing to pcre_compile(),
801 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
802 provided for specialist use. Most commonly, no special tables are
803 passed, in which case internal tables that are generated when PCRE is
804 built are used.
805
806 The function pcre_fullinfo() is used to find out information about a
807 compiled pattern; pcre_info() is an obsolete version that returns only
808 some of the available information, but is retained for backwards com-
809 patibility. The function pcre_version() returns a pointer to a string
810 containing the version of PCRE and its date of release.
811
812 The function pcre_refcount() maintains a reference count in a data
813 block containing a compiled pattern. This is provided for the benefit
814 of object-oriented applications.
815
816 The global variables pcre_malloc and pcre_free initially contain the
817 entry points of the standard malloc() and free() functions, respec-
818 tively. PCRE calls the memory management functions via these variables,
819 so a calling program can replace them if it wishes to intercept the
820 calls. This should be done before calling any PCRE functions.
821
822 The global variables pcre_stack_malloc and pcre_stack_free are also
823 indirections to memory management functions. These special functions
824 are used only when PCRE is compiled to use the heap for remembering
825 data, instead of recursive function calls, when running the pcre_exec()
826 function. See the pcrebuild documentation for details of how to do
827 this. It is a non-standard way of building PCRE, for use in environ-
828 ments that have limited stacks. Because of the greater use of memory
829 management, it runs more slowly. Separate functions are provided so
830 that special-purpose external code can be used for this case. When
831 used, these functions are always called in a stack-like manner (last
832 obtained, first freed), and always for memory blocks of the same size.
833 There is a discussion about PCRE's stack usage in the pcrestack docu-
834 mentation.
835
836 The global variable pcre_callout initially contains NULL. It can be set
837 by the caller to a "callout" function, which PCRE will then call at
838 specified points during a matching operation. Details are given in the
839 pcrecallout documentation.
840
841
842 NEWLINES
843
844 PCRE supports four different conventions for indicating line breaks in
845 strings: a single CR (carriage return) character, a single LF (line-
846 feed) character, the two-character sequence CRLF, or any Unicode new-
847 line sequence. The Unicode newline sequences are the three just men-
848 tioned, plus the single characters VT (vertical tab, U+000B), FF (form-
849 feed, U+000C), NEL (next line, U+0085), LS (line separator, U+2028),
850 and PS (paragraph separator, U+2029).
851
852 Each of the first three conventions is used by at least one operating
853 system as its standard newline sequence. When PCRE is built, a default
854 can be specified. The default default is LF, which is the Unix stan-
855 dard. When PCRE is run, the default can be overridden, either when a
856 pattern is compiled, or when it is matched.
857
858 In the PCRE documentation the word "newline" is used to mean "the char-
859 acter or pair of characters that indicate a line break". The choice of
860 newline convention affects the handling of the dot, circumflex, and
861 dollar metacharacters, the handling of #-comments in /x mode, and, when
862 CRLF is a recognized line ending sequence, the match position advance-
863 ment for a non-anchored pattern. The choice of newline convention does
864 not affect the interpretation of the \n or \r escape sequences.
865
866
867 MULTITHREADING
868
869 The PCRE functions can be used in multi-threading applications, with
870 the proviso that the memory management functions pointed to by
871 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
872 callout function pointed to by pcre_callout, are shared by all threads.
873
874 The compiled form of a regular expression is not altered during match-
875 ing, so the same compiled pattern can safely be used by several threads
876 at once.
877
878
879 SAVING PRECOMPILED PATTERNS FOR LATER USE
880
881 The compiled form of a regular expression can be saved and re-used at a
882 later time, possibly by a different program, and even on a host other
883 than the one on which it was compiled. Details are given in the
884 pcreprecompile documentation.
885
886
887 CHECKING BUILD-TIME OPTIONS
888
889 int pcre_config(int what, void *where);
890
891 The function pcre_config() makes it possible for a PCRE client to dis-
892 cover which optional features have been compiled into the PCRE library.
893 The pcrebuild documentation has more details about these optional fea-
894 tures.
895
896 The first argument for pcre_config() is an integer, specifying which
897 information is required; the second argument is a pointer to a variable
898 into which the information is placed. The following information is
899 available:
900
901 PCRE_CONFIG_UTF8
902
903 The output is an integer that is set to one if UTF-8 support is avail-
904 able; otherwise it is set to zero.
905
906 PCRE_CONFIG_UNICODE_PROPERTIES
907
908 The output is an integer that is set to one if support for Unicode
909 character properties is available; otherwise it is set to zero.
910
911 PCRE_CONFIG_NEWLINE
912
913 The output is an integer whose value specifies the default character
914 sequence that is recognized as meaning "newline". The four values that
915 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, and -1 for ANY.
916 The default should normally be the standard sequence for your operating
917 system.
918
919 PCRE_CONFIG_LINK_SIZE
920
921 The output is an integer that contains the number of bytes used for
922 internal linkage in compiled regular expressions. The value is 2, 3, or
923 4. Larger values allow larger regular expressions to be compiled, at
924 the expense of slower matching. The default value of 2 is sufficient
925 for all but the most massive patterns, since it allows the compiled
926 pattern to be up to 64K in size.
927
928 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
929
930 The output is an integer that contains the threshold above which the
931 POSIX interface uses malloc() for output vectors. Further details are
932 given in the pcreposix documentation.
933
934 PCRE_CONFIG_MATCH_LIMIT
935
936 The output is an integer that gives the default limit for the number of
937 internal matching function calls in a pcre_exec() execution. Further
938 details are given with pcre_exec() below.
939
940 PCRE_CONFIG_MATCH_LIMIT_RECURSION
941
942 The output is an integer that gives the default limit for the depth of
943 recursion when calling the internal matching function in a pcre_exec()
944 execution. Further details are given with pcre_exec() below.
945
946 PCRE_CONFIG_STACKRECURSE
947
948 The output is an integer that is set to one if internal recursion when
949 running pcre_exec() is implemented by recursive function calls that use
950 the stack to remember their state. This is the usual way that PCRE is
951 compiled. The output is zero if PCRE was compiled to use blocks of data
952 on the heap instead of recursive function calls. In this case,
953 pcre_stack_malloc and pcre_stack_free are called to manage memory
954 blocks on the heap, thus avoiding the use of the stack.
955
956
957 COMPILING A PATTERN
958
959 pcre *pcre_compile(const char *pattern, int options,
960 const char **errptr, int *erroffset,
961 const unsigned char *tableptr);
962
963 pcre *pcre_compile2(const char *pattern, int options,
964 int *errorcodeptr,
965 const char **errptr, int *erroffset,
966 const unsigned char *tableptr);
967
968 Either of the functions pcre_compile() or pcre_compile2() can be called
969 to compile a pattern into an internal form. The only difference between
970 the two interfaces is that pcre_compile2() has an additional argument,
971 errorcodeptr, via which a numerical error code can be returned.
972
973 The pattern is a C string terminated by a binary zero, and is passed in
974 the pattern argument. A pointer to a single block of memory that is
975 obtained via pcre_malloc is returned. This contains the compiled code
976 and related data. The pcre type is defined for the returned block; this
977 is a typedef for a structure whose contents are not externally defined.
978 It is up to the caller to free the memory (via pcre_free) when it is no
979 longer required.
980
981 Although the compiled code of a PCRE regex is relocatable, that is, it
982 does not depend on memory location, the complete pcre data block is not
983 fully relocatable, because it may contain a copy of the tableptr argu-
984 ment, which is an address (see below).
985
986 The options argument contains various bit settings that affect the com-
987 pilation. It should be zero if no options are required. The available
988 options are described below. Some of them, in particular, those that
989 are compatible with Perl, can also be set and unset from within the
990 pattern (see the detailed description in the pcrepattern documenta-
991 tion). For these options, the contents of the options argument speci-
992 fies their initial settings at the start of compilation and execution.
993 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
994 of matching as well as at compile time.
995
996 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
997 if compilation of a pattern fails, pcre_compile() returns NULL, and
998 sets the variable pointed to by errptr to point to a textual error mes-
999 sage. This is a static string that is part of the library. You must not
1000 try to free it. The offset from the start of the pattern to the charac-
1001 ter where the error was discovered is placed in the variable pointed to
1002 by erroffset, which must not be NULL. If it is, an immediate error is
1003 given.
1004
1005 If pcre_compile2() is used instead of pcre_compile(), and the error-
1006 codeptr argument is not NULL, a non-zero error code number is returned
1007 via this argument in the event of an error. This is in addition to the
1008 textual error message. Error codes and messages are listed below.
1009
1010 If the final argument, tableptr, is NULL, PCRE uses a default set of
1011 character tables that are built when PCRE is compiled, using the
1012 default C locale. Otherwise, tableptr must be an address that is the
1013 result of a call to pcre_maketables(). This value is stored with the
1014 compiled pattern, and used again by pcre_exec(), unless another table
1015 pointer is passed to it. For more discussion, see the section on locale
1016 support below.
1017
1018 This code fragment shows a typical straightforward call to pcre_com-
1019 pile():
1020
1021 pcre *re;
1022 const char *error;
1023 int erroffset;
1024 re = pcre_compile(
1025 "^A.*Z", /* the pattern */
1026 0, /* default options */
1027 &error, /* for error message */
1028 &erroffset, /* for error offset */
1029 NULL); /* use default character tables */
1030
1031 The following names for option bits are defined in the pcre.h header
1032 file:
1033
1034 PCRE_ANCHORED
1035
1036 If this bit is set, the pattern is forced to be "anchored", that is, it
1037 is constrained to match only at the first matching point in the string
1038 that is being searched (the "subject string"). This effect can also be
1039 achieved by appropriate constructs in the pattern itself, which is the
1040 only way to do it in Perl.
1041
1042 PCRE_AUTO_CALLOUT
1043
1044 If this bit is set, pcre_compile() automatically inserts callout items,
1045 all with number 255, before each pattern item. For discussion of the
1046 callout facility, see the pcrecallout documentation.
1047
1048 PCRE_CASELESS
1049
1050 If this bit is set, letters in the pattern match both upper and lower
1051 case letters. It is equivalent to Perl's /i option, and it can be
1052 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1053 always understands the concept of case for characters whose values are
1054 less than 128, so caseless matching is always possible. For characters
1055 with higher values, the concept of case is supported if PCRE is com-
1056 piled with Unicode property support, but not otherwise. If you want to
1057 use caseless matching for characters 128 and above, you must ensure
1058 that PCRE is compiled with Unicode property support as well as with
1059 UTF-8 support.
1060
1061 PCRE_DOLLAR_ENDONLY
1062
1063 If this bit is set, a dollar metacharacter in the pattern matches only
1064 at the end of the subject string. Without this option, a dollar also
1065 matches immediately before a newline at the end of the string (but not
1066 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1067 if PCRE_MULTILINE is set. There is no equivalent to this option in
1068 Perl, and no way to set it within a pattern.
1069
1070 PCRE_DOTALL
1071
1072 If this bit is set, a dot metacharater in the pattern matches all char-
1073 acters, including those that indicate newline. Without it, a dot does
1074 not match when the current position is at a newline. This option is
1075 equivalent to Perl's /s option, and it can be changed within a pattern
1076 by a (?s) option setting. A negative class such as [^a] always matches
1077 newline characters, independent of the setting of this option.
1078
1079 PCRE_DUPNAMES
1080
1081 If this bit is set, names used to identify capturing subpatterns need
1082 not be unique. This can be helpful for certain types of pattern when it
1083 is known that only one instance of the named subpattern can ever be
1084 matched. There are more details of named subpatterns below; see also
1085 the pcrepattern documentation.
1086
1087 PCRE_EXTENDED
1088
1089 If this bit is set, whitespace data characters in the pattern are
1090 totally ignored except when escaped or inside a character class. White-
1091 space does not include the VT character (code 11). In addition, charac-
1092 ters between an unescaped # outside a character class and the next new-
1093 line, inclusive, are also ignored. This is equivalent to Perl's /x
1094 option, and it can be changed within a pattern by a (?x) option set-
1095 ting.
1096
1097 This option makes it possible to include comments inside complicated
1098 patterns. Note, however, that this applies only to data characters.
1099 Whitespace characters may never appear within special character
1100 sequences in a pattern, for example within the sequence (?( which
1101 introduces a conditional subpattern.
1102
1103 PCRE_EXTRA
1104
1105 This option was invented in order to turn on additional functionality
1106 of PCRE that is incompatible with Perl, but it is currently of very
1107 little use. When set, any backslash in a pattern that is followed by a
1108 letter that has no special meaning causes an error, thus reserving
1109 these combinations for future expansion. By default, as in Perl, a
1110 backslash followed by a letter with no special meaning is treated as a
1111 literal. (Perl can, however, be persuaded to give a warning for this.)
1112 There are at present no other features controlled by this option. It
1113 can also be set by a (?X) option setting within a pattern.
1114
1115 PCRE_FIRSTLINE
1116
1117 If this option is set, an unanchored pattern is required to match
1118 before or at the first newline in the subject string, though the
1119 matched text may continue over the newline.
1120
1121 PCRE_MULTILINE
1122
1123 By default, PCRE treats the subject string as consisting of a single
1124 line of characters (even if it actually contains newlines). The "start
1125 of line" metacharacter (^) matches only at the start of the string,
1126 while the "end of line" metacharacter ($) matches only at the end of
1127 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1128 is set). This is the same as Perl.
1129
1130 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1131 constructs match immediately following or immediately before internal
1132 newlines in the subject string, respectively, as well as at the very
1133 start and end. This is equivalent to Perl's /m option, and it can be
1134 changed within a pattern by a (?m) option setting. If there are no new-
1135 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1136 setting PCRE_MULTILINE has no effect.
1137
1138 PCRE_NEWLINE_CR
1139 PCRE_NEWLINE_LF
1140 PCRE_NEWLINE_CRLF
1141 PCRE_NEWLINE_ANY
1142
1143 These options override the default newline definition that was chosen
1144 when PCRE was built. Setting the first or the second specifies that a
1145 newline is indicated by a single character (CR or LF, respectively).
1146 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1147 two-character CRLF sequence. Setting PCRE_NEWLINE_ANY specifies that
1148 any Unicode newline sequence should be recognized. The Unicode newline
1149 sequences are the three just mentioned, plus the single characters VT
1150 (vertical tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085),
1151 LS (line separator, U+2028), and PS (paragraph separator, U+2029). The
1152 last two are recognized only in UTF-8 mode.
1153
1154 The newline setting in the options word uses three bits that are
1155 treated as a number, giving eight possibilities. Currently only five
1156 are used (default plus the four values above). This means that if you
1157 set more than one newline option, the combination may or may not be
1158 sensible. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equiva-
1159 lent to PCRE_NEWLINE_CRLF, but other combinations yield unused numbers
1160 and cause an error.
1161
1162 The only time that a line break is specially recognized when compiling
1163 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1164 character class is encountered. This indicates a comment that lasts
1165 until after the next line break sequence. In other circumstances, line
1166 break sequences are treated as literal data, except that in
1167 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1168 and are therefore ignored.
1169
1170 The newline option that is set at compile time becomes the default that
1171 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1172
1173 PCRE_NO_AUTO_CAPTURE
1174
1175 If this option is set, it disables the use of numbered capturing paren-
1176 theses in the pattern. Any opening parenthesis that is not followed by
1177 ? behaves as if it were followed by ?: but named parentheses can still
1178 be used for capturing (and they acquire numbers in the usual way).
1179 There is no equivalent of this option in Perl.
1180
1181 PCRE_UNGREEDY
1182
1183 This option inverts the "greediness" of the quantifiers so that they
1184 are not greedy by default, but become greedy if followed by "?". It is
1185 not compatible with Perl. It can also be set by a (?U) option setting
1186 within the pattern.
1187
1188 PCRE_UTF8
1189
1190 This option causes PCRE to regard both the pattern and the subject as
1191 strings of UTF-8 characters instead of single-byte character strings.
1192 However, it is available only when PCRE is built to include UTF-8 sup-
1193 port. If not, the use of this option provokes an error. Details of how
1194 this option changes the behaviour of PCRE are given in the section on
1195 UTF-8 support in the main pcre page.
1196
1197 PCRE_NO_UTF8_CHECK
1198
1199 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1200 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1201 pcre_compile() returns an error. If you already know that your pattern
1202 is valid, and you want to skip this check for performance reasons, you
1203 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1204 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1205 your program to crash. Note that this option can also be passed to
1206 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1207 ing of subject strings.
1208
1209
1210 COMPILATION ERROR CODES
1211
1212 The following table lists the error codes than may be returned by
1213 pcre_compile2(), along with the error messages that may be returned by
1214 both compiling functions. As PCRE has developed, some error codes have
1215 fallen out of use. To avoid confusion, they have not been re-used.
1216
1217 0 no error
1218 1 \ at end of pattern
1219 2 \c at end of pattern
1220 3 unrecognized character follows \
1221 4 numbers out of order in {} quantifier
1222 5 number too big in {} quantifier
1223 6 missing terminating ] for character class
1224 7 invalid escape sequence in character class
1225 8 range out of order in character class
1226 9 nothing to repeat
1227 10 [this code is not in use]
1228 11 internal error: unexpected repeat
1229 12 unrecognized character after (?
1230 13 POSIX named classes are supported only within a class
1231 14 missing )
1232 15 reference to non-existent subpattern
1233 16 erroffset passed as NULL
1234 17 unknown option bit(s) set
1235 18 missing ) after comment
1236 19 [this code is not in use]
1237 20 regular expression too large
1238 21 failed to get memory
1239 22 unmatched parentheses
1240 23 internal error: code overflow
1241 24 unrecognized character after (?<
1242 25 lookbehind assertion is not fixed length
1243 26 malformed number or name after (?(
1244 27 conditional group contains more than two branches
1245 28 assertion expected after (?(
1246 29 (?R or (?digits must be followed by )
1247 30 unknown POSIX class name
1248 31 POSIX collating elements are not supported
1249 32 this version of PCRE is not compiled with PCRE_UTF8 support
1250 33 [this code is not in use]
1251 34 character value in \x{...} sequence is too large
1252 35 invalid condition (?(0)
1253 36 \C not allowed in lookbehind assertion
1254 37 PCRE does not support \L, \l, \N, \U, or \u
1255 38 number after (?C is > 255
1256 39 closing ) for (?C expected
1257 40 recursive call could loop indefinitely
1258 41 unrecognized character after (?P
1259 42 syntax error in subpattern name (missing terminator)
1260 43 two named subpatterns have the same name
1261 44 invalid UTF-8 string
1262 45 support for \P, \p, and \X has not been compiled
1263 46 malformed \P or \p sequence
1264 47 unknown property name after \P or \p
1265 48 subpattern name is too long (maximum 32 characters)
1266 49 too many named subpatterns (maximum 10,000)
1267 50 repeated subpattern is too long
1268 51 octal value is greater than \377 (not in UTF-8 mode)
1269 52 internal error: overran compiling workspace
1270 53 internal error: previously-checked referenced subpattern not
1271 found
1272 54 DEFINE group contains more than one branch
1273 55 repeating a DEFINE group is not allowed
1274 56 inconsistent NEWLINE options"
1275
1276
1277 STUDYING A PATTERN
1278
1279 pcre_extra *pcre_study(const pcre *code, int options
1280 const char **errptr);
1281
1282 If a compiled pattern is going to be used several times, it is worth
1283 spending more time analyzing it in order to speed up the time taken for
1284 matching. The function pcre_study() takes a pointer to a compiled pat-
1285 tern as its first argument. If studying the pattern produces additional
1286 information that will help speed up matching, pcre_study() returns a
1287 pointer to a pcre_extra block, in which the study_data field points to
1288 the results of the study.
1289
1290 The returned value from pcre_study() can be passed directly to
1291 pcre_exec(). However, a pcre_extra block also contains other fields
1292 that can be set by the caller before the block is passed; these are
1293 described below in the section on matching a pattern.
1294
1295 If studying the pattern does not produce any additional information
1296 pcre_study() returns NULL. In that circumstance, if the calling program
1297 wants to pass any of the other fields to pcre_exec(), it must set up
1298 its own pcre_extra block.
1299
1300 The second argument of pcre_study() contains option bits. At present,
1301 no options are defined, and this argument should always be zero.
1302
1303 The third argument for pcre_study() is a pointer for an error message.
1304 If studying succeeds (even if no data is returned), the variable it
1305 points to is set to NULL. Otherwise it is set to point to a textual
1306 error message. This is a static string that is part of the library. You
1307 must not try to free it. You should test the error pointer for NULL
1308 after calling pcre_study(), to be sure that it has run successfully.
1309
1310 This is a typical call to pcre_study():
1311
1312 pcre_extra *pe;
1313 pe = pcre_study(
1314 re, /* result of pcre_compile() */
1315 0, /* no options exist */
1316 &error); /* set to NULL or points to a message */
1317
1318 At present, studying a pattern is useful only for non-anchored patterns
1319 that do not have a single fixed starting character. A bitmap of possi-
1320 ble starting bytes is created.
1321
1322
1323 LOCALE SUPPORT
1324
1325 PCRE handles caseless matching, and determines whether characters are
1326 letters digits, or whatever, by reference to a set of tables, indexed
1327 by character value. When running in UTF-8 mode, this applies only to
1328 characters with codes less than 128. Higher-valued codes never match
1329 escapes such as \w or \d, but can be tested with \p if PCRE is built
1330 with Unicode character property support. The use of locales with Uni-
1331 code is discouraged.
1332
1333 An internal set of tables is created in the default C locale when PCRE
1334 is built. This is used when the final argument of pcre_compile() is
1335 NULL, and is sufficient for many applications. An alternative set of
1336 tables can, however, be supplied. These may be created in a different
1337 locale from the default. As more and more applications change to using
1338 Unicode, the need for this locale support is expected to die away.
1339
1340 External tables are built by calling the pcre_maketables() function,
1341 which has no arguments, in the relevant locale. The result can then be
1342 passed to pcre_compile() or pcre_exec() as often as necessary. For
1343 example, to build and use tables that are appropriate for the French
1344 locale (where accented characters with values greater than 128 are
1345 treated as letters), the following code could be used:
1346
1347 setlocale(LC_CTYPE, "fr_FR");
1348 tables = pcre_maketables();
1349 re = pcre_compile(..., tables);
1350
1351 When pcre_maketables() runs, the tables are built in memory that is
1352 obtained via pcre_malloc. It is the caller's responsibility to ensure
1353 that the memory containing the tables remains available for as long as
1354 it is needed.
1355
1356 The pointer that is passed to pcre_compile() is saved with the compiled
1357 pattern, and the same tables are used via this pointer by pcre_study()
1358 and normally also by pcre_exec(). Thus, by default, for any single pat-
1359 tern, compilation, studying and matching all happen in the same locale,
1360 but different patterns can be compiled in different locales.
1361
1362 It is possible to pass a table pointer or NULL (indicating the use of
1363 the internal tables) to pcre_exec(). Although not intended for this
1364 purpose, this facility could be used to match a pattern in a different
1365 locale from the one in which it was compiled. Passing table pointers at
1366 run time is discussed below in the section on matching a pattern.
1367
1368
1369 INFORMATION ABOUT A PATTERN
1370
1371 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1372 int what, void *where);
1373
1374 The pcre_fullinfo() function returns information about a compiled pat-
1375 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1376 less retained for backwards compability (and is documented below).
1377
1378 The first argument for pcre_fullinfo() is a pointer to the compiled
1379 pattern. The second argument is the result of pcre_study(), or NULL if
1380 the pattern was not studied. The third argument specifies which piece
1381 of information is required, and the fourth argument is a pointer to a
1382 variable to receive the data. The yield of the function is zero for
1383 success, or one of the following negative numbers:
1384
1385 PCRE_ERROR_NULL the argument code was NULL
1386 the argument where was NULL
1387 PCRE_ERROR_BADMAGIC the "magic number" was not found
1388 PCRE_ERROR_BADOPTION the value of what was invalid
1389
1390 The "magic number" is placed at the start of each compiled pattern as
1391 an simple check against passing an arbitrary memory pointer. Here is a
1392 typical call of pcre_fullinfo(), to obtain the length of the compiled
1393 pattern:
1394
1395 int rc;
1396 size_t length;
1397 rc = pcre_fullinfo(
1398 re, /* result of pcre_compile() */
1399 pe, /* result of pcre_study(), or NULL */
1400 PCRE_INFO_SIZE, /* what is required */
1401 &length); /* where to put the data */
1402
1403 The possible values for the third argument are defined in pcre.h, and
1404 are as follows:
1405
1406 PCRE_INFO_BACKREFMAX
1407
1408 Return the number of the highest back reference in the pattern. The
1409 fourth argument should point to an int variable. Zero is returned if
1410 there are no back references.
1411
1412 PCRE_INFO_CAPTURECOUNT
1413
1414 Return the number of capturing subpatterns in the pattern. The fourth
1415 argument should point to an int variable.
1416
1417 PCRE_INFO_DEFAULT_TABLES
1418
1419 Return a pointer to the internal default character tables within PCRE.
1420 The fourth argument should point to an unsigned char * variable. This
1421 information call is provided for internal use by the pcre_study() func-
1422 tion. External callers can cause PCRE to use its internal tables by
1423 passing a NULL table pointer.
1424
1425 PCRE_INFO_FIRSTBYTE
1426
1427 Return information about the first byte of any matched string, for a
1428 non-anchored pattern. The fourth argument should point to an int vari-
1429 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1430 is still recognized for backwards compatibility.)
1431
1432 If there is a fixed first byte, for example, from a pattern such as
1433 (cat|cow|coyote), its value is returned. Otherwise, if either
1434
1435 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1436 branch starts with "^", or
1437
1438 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1439 set (if it were set, the pattern would be anchored),
1440
1441 -1 is returned, indicating that the pattern matches only at the start
1442 of a subject string or after any newline within the string. Otherwise
1443 -2 is returned. For anchored patterns, -2 is returned.
1444
1445 PCRE_INFO_FIRSTTABLE
1446
1447 If the pattern was studied, and this resulted in the construction of a
1448 256-bit table indicating a fixed set of bytes for the first byte in any
1449 matching string, a pointer to the table is returned. Otherwise NULL is
1450 returned. The fourth argument should point to an unsigned char * vari-
1451 able.
1452
1453 PCRE_INFO_LASTLITERAL
1454
1455 Return the value of the rightmost literal byte that must exist in any
1456 matched string, other than at its start, if such a byte has been
1457 recorded. The fourth argument should point to an int variable. If there
1458 is no such byte, -1 is returned. For anchored patterns, a last literal
1459 byte is recorded only if it follows something of variable length. For
1460 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1461 /^a\dz\d/ the returned value is -1.
1462
1463 PCRE_INFO_NAMECOUNT
1464 PCRE_INFO_NAMEENTRYSIZE
1465 PCRE_INFO_NAMETABLE
1466
1467 PCRE supports the use of named as well as numbered capturing parenthe-
1468 ses. The names are just an additional way of identifying the parenthe-
1469 ses, which still acquire numbers. Several convenience functions such as
1470 pcre_get_named_substring() are provided for extracting captured sub-
1471 strings by name. It is also possible to extract the data directly, by
1472 first converting the name to a number in order to access the correct
1473 pointers in the output vector (described with pcre_exec() below). To do
1474 the conversion, you need to use the name-to-number map, which is
1475 described by these three values.
1476
1477 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1478 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1479 of each entry; both of these return an int value. The entry size
1480 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1481 a pointer to the first entry of the table (a pointer to char). The
1482 first two bytes of each entry are the number of the capturing parenthe-
1483 sis, most significant byte first. The rest of the entry is the corre-
1484 sponding name, zero terminated. The names are in alphabetical order.
1485 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1486 theses numbers. For example, consider the following pattern (assume
1487 PCRE_EXTENDED is set, so white space - including newlines - is
1488 ignored):
1489
1490 (?<date> (?<year>(\d\d)?\d\d) -
1491 (?<month>\d\d) - (?<day>\d\d) )
1492
1493 There are four named subpatterns, so the table has four entries, and
1494 each entry in the table is eight bytes long. The table is as follows,
1495 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1496 as ??:
1497
1498 00 01 d a t e 00 ??
1499 00 05 d a y 00 ?? ??
1500 00 04 m o n t h 00
1501 00 02 y e a r 00 ??
1502
1503 When writing code to extract data from named subpatterns using the
1504 name-to-number map, remember that the length of the entries is likely
1505 to be different for each compiled pattern.
1506
1507 PCRE_INFO_OPTIONS
1508
1509 Return a copy of the options with which the pattern was compiled. The
1510 fourth argument should point to an unsigned long int variable. These
1511 option bits are those specified in the call to pcre_compile(), modified
1512 by any top-level option settings within the pattern itself.
1513
1514 A pattern is automatically anchored by PCRE if all of its top-level
1515 alternatives begin with one of the following:
1516
1517 ^ unless PCRE_MULTILINE is set
1518 \A always
1519 \G always
1520 .* if PCRE_DOTALL is set and there are no back
1521 references to the subpattern in which .* appears
1522
1523 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1524 by pcre_fullinfo().
1525
1526 PCRE_INFO_SIZE
1527
1528 Return the size of the compiled pattern, that is, the value that was
1529 passed as the argument to pcre_malloc() when PCRE was getting memory in
1530 which to place the compiled data. The fourth argument should point to a
1531 size_t variable.
1532
1533 PCRE_INFO_STUDYSIZE
1534
1535 Return the size of the data block pointed to by the study_data field in
1536 a pcre_extra block. That is, it is the value that was passed to
1537 pcre_malloc() when PCRE was getting memory into which to place the data
1538 created by pcre_study(). The fourth argument should point to a size_t
1539 variable.
1540
1541
1542 OBSOLETE INFO FUNCTION
1543
1544 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1545
1546 The pcre_info() function is now obsolete because its interface is too
1547 restrictive to return all the available data about a compiled pattern.
1548 New programs should use pcre_fullinfo() instead. The yield of
1549 pcre_info() is the number of capturing subpatterns, or one of the fol-
1550 lowing negative numbers:
1551
1552 PCRE_ERROR_NULL the argument code was NULL
1553 PCRE_ERROR_BADMAGIC the "magic number" was not found
1554
1555 If the optptr argument is not NULL, a copy of the options with which
1556 the pattern was compiled is placed in the integer it points to (see
1557 PCRE_INFO_OPTIONS above).
1558
1559 If the pattern is not anchored and the firstcharptr argument is not
1560 NULL, it is used to pass back information about the first character of
1561 any matched string (see PCRE_INFO_FIRSTBYTE above).
1562
1563
1564 REFERENCE COUNTS
1565
1566 int pcre_refcount(pcre *code, int adjust);
1567
1568 The pcre_refcount() function is used to maintain a reference count in
1569 the data block that contains a compiled pattern. It is provided for the
1570 benefit of applications that operate in an object-oriented manner,
1571 where different parts of the application may be using the same compiled
1572 pattern, but you want to free the block when they are all done.
1573
1574 When a pattern is compiled, the reference count field is initialized to
1575 zero. It is changed only by calling this function, whose action is to
1576 add the adjust value (which may be positive or negative) to it. The
1577 yield of the function is the new value. However, the value of the count
1578 is constrained to lie between 0 and 65535, inclusive. If the new value
1579 is outside these limits, it is forced to the appropriate limit value.
1580
1581 Except when it is zero, the reference count is not correctly preserved
1582 if a pattern is compiled on one host and then transferred to a host
1583 whose byte-order is different. (This seems a highly unlikely scenario.)
1584
1585
1586 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1587
1588 int pcre_exec(const pcre *code, const pcre_extra *extra,
1589 const char *subject, int length, int startoffset,
1590 int options, int *ovector, int ovecsize);
1591
1592 The function pcre_exec() is called to match a subject string against a
1593 compiled pattern, which is passed in the code argument. If the pattern
1594 has been studied, the result of the study should be passed in the extra
1595 argument. This function is the main matching facility of the library,
1596 and it operates in a Perl-like manner. For specialist use there is also
1597 an alternative matching function, which is described below in the sec-
1598 tion about the pcre_dfa_exec() function.
1599
1600 In most applications, the pattern will have been compiled (and option-
1601 ally studied) in the same process that calls pcre_exec(). However, it
1602 is possible to save compiled patterns and study data, and then use them
1603 later in different processes, possibly even on different hosts. For a
1604 discussion about this, see the pcreprecompile documentation.
1605
1606 Here is an example of a simple call to pcre_exec():
1607
1608 int rc;
1609 int ovector[30];
1610 rc = pcre_exec(
1611 re, /* result of pcre_compile() */
1612 NULL, /* we didn't study the pattern */
1613 "some string", /* the subject string */
1614 11, /* the length of the subject string */
1615 0, /* start at offset 0 in the subject */
1616 0, /* default options */
1617 ovector, /* vector of integers for substring information */
1618 30); /* number of elements (NOT size in bytes) */
1619
1620 Extra data for pcre_exec()
1621
1622 If the extra argument is not NULL, it must point to a pcre_extra data
1623 block. The pcre_study() function returns such a block (when it doesn't
1624 return NULL), but you can also create one for yourself, and pass addi-
1625 tional information in it. The pcre_extra block contains the following
1626 fields (not necessarily in this order):
1627
1628 unsigned long int flags;
1629 void *study_data;
1630 unsigned long int match_limit;
1631 unsigned long int match_limit_recursion;
1632 void *callout_data;
1633 const unsigned char *tables;
1634
1635 The flags field is a bitmap that specifies which of the other fields
1636 are set. The flag bits are:
1637
1638 PCRE_EXTRA_STUDY_DATA
1639 PCRE_EXTRA_MATCH_LIMIT
1640 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1641 PCRE_EXTRA_CALLOUT_DATA
1642 PCRE_EXTRA_TABLES
1643
1644 Other flag bits should be set to zero. The study_data field is set in
1645 the pcre_extra block that is returned by pcre_study(), together with
1646 the appropriate flag bit. You should not set this yourself, but you may
1647 add to the block by setting the other fields and their corresponding
1648 flag bits.
1649
1650 The match_limit field provides a means of preventing PCRE from using up
1651 a vast amount of resources when running patterns that are not going to
1652 match, but which have a very large number of possibilities in their
1653 search trees. The classic example is the use of nested unlimited
1654 repeats.
1655
1656 Internally, PCRE uses a function called match() which it calls repeat-
1657 edly (sometimes recursively). The limit set by match_limit is imposed
1658 on the number of times this function is called during a match, which
1659 has the effect of limiting the amount of backtracking that can take
1660 place. For patterns that are not anchored, the count restarts from zero
1661 for each position in the subject string.
1662
1663 The default value for the limit can be set when PCRE is built; the
1664 default default is 10 million, which handles all but the most extreme
1665 cases. You can override the default by suppling pcre_exec() with a
1666 pcre_extra block in which match_limit is set, and
1667 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1668 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1669
1670 The match_limit_recursion field is similar to match_limit, but instead
1671 of limiting the total number of times that match() is called, it limits
1672 the depth of recursion. The recursion depth is a smaller number than
1673 the total number of calls, because not all calls to match() are recur-
1674 sive. This limit is of use only if it is set smaller than match_limit.
1675
1676 Limiting the recursion depth limits the amount of stack that can be
1677 used, or, when PCRE has been compiled to use memory on the heap instead
1678 of the stack, the amount of heap memory that can be used.
1679
1680 The default value for match_limit_recursion can be set when PCRE is
1681 built; the default default is the same value as the default for
1682 match_limit. You can override the default by suppling pcre_exec() with
1683 a pcre_extra block in which match_limit_recursion is set, and
1684 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1685 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1686
1687 The pcre_callout field is used in conjunction with the "callout" fea-
1688 ture, which is described in the pcrecallout documentation.
1689
1690 The tables field is used to pass a character tables pointer to
1691 pcre_exec(); this overrides the value that is stored with the compiled
1692 pattern. A non-NULL value is stored with the compiled pattern only if
1693 custom tables were supplied to pcre_compile() via its tableptr argu-
1694 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1695 PCRE's internal tables to be used. This facility is helpful when re-
1696 using patterns that have been saved after compiling with an external
1697 set of tables, because the external tables might be at a different
1698 address when pcre_exec() is called. See the pcreprecompile documenta-
1699 tion for a discussion of saving compiled patterns for later use.
1700
1701 Option bits for pcre_exec()
1702
1703 The unused bits of the options argument for pcre_exec() must be zero.
1704 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1705 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1706 PCRE_PARTIAL.
1707
1708 PCRE_ANCHORED
1709
1710 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1711 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1712 turned out to be anchored by virtue of its contents, it cannot be made
1713 unachored at matching time.
1714
1715 PCRE_NEWLINE_CR
1716 PCRE_NEWLINE_LF
1717 PCRE_NEWLINE_CRLF
1718 PCRE_NEWLINE_ANY
1719
1720 These options override the newline definition that was chosen or
1721 defaulted when the pattern was compiled. For details, see the descrip-
1722 tion of pcre_compile() above. During matching, the newline choice
1723 affects the behaviour of the dot, circumflex, and dollar metacharac-
1724 ters. It may also alter the way the match position is advanced after a
1725 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF or
1726 PCRE_NEWLINE_ANY is set, and a match attempt fails when the current
1727 position is at a CRLF sequence, the match position is advanced by two
1728 characters instead of one, in other words, to after the CRLF.
1729
1730 PCRE_NOTBOL
1731
1732 This option specifies that first character of the subject string is not
1733 the beginning of a line, so the circumflex metacharacter should not
1734 match before it. Setting this without PCRE_MULTILINE (at compile time)
1735 causes circumflex never to match. This option affects only the behav-
1736 iour of the circumflex metacharacter. It does not affect \A.
1737
1738 PCRE_NOTEOL
1739
1740 This option specifies that the end of the subject string is not the end
1741 of a line, so the dollar metacharacter should not match it nor (except
1742 in multiline mode) a newline immediately before it. Setting this with-
1743 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1744 option affects only the behaviour of the dollar metacharacter. It does
1745 not affect \Z or \z.
1746
1747 PCRE_NOTEMPTY
1748
1749 An empty string is not considered to be a valid match if this option is
1750 set. If there are alternatives in the pattern, they are tried. If all
1751 the alternatives match the empty string, the entire match fails. For
1752 example, if the pattern
1753
1754 a?b?
1755
1756 is applied to a string not beginning with "a" or "b", it matches the
1757 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1758 match is not valid, so PCRE searches further into the string for occur-
1759 rences of "a" or "b".
1760
1761 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1762 cial case of a pattern match of the empty string within its split()
1763 function, and when using the /g modifier. It is possible to emulate
1764 Perl's behaviour after matching a null string by first trying the match
1765 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1766 if that fails by advancing the starting offset (see below) and trying
1767 an ordinary match again. There is some code that demonstrates how to do
1768 this in the pcredemo.c sample program.
1769
1770 PCRE_NO_UTF8_CHECK
1771
1772 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1773 UTF-8 string is automatically checked when pcre_exec() is subsequently
1774 called. The value of startoffset is also checked to ensure that it
1775 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1776 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1777 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1778 returned.
1779
1780 If you already know that your subject is valid, and you want to skip
1781 these checks for performance reasons, you can set the
1782 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1783 do this for the second and subsequent calls to pcre_exec() if you are
1784 making repeated calls to find all the matches in a single subject
1785 string. However, you should be sure that the value of startoffset
1786 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1787 set, the effect of passing an invalid UTF-8 string as a subject, or a
1788 value of startoffset that does not point to the start of a UTF-8 char-
1789 acter, is undefined. Your program may crash.
1790
1791 PCRE_PARTIAL
1792
1793 This option turns on the partial matching feature. If the subject
1794 string fails to match the pattern, but at some point during the match-
1795 ing process the end of the subject was reached (that is, the subject
1796 partially matches the pattern and the failure to match occurred only
1797 because there were not enough subject characters), pcre_exec() returns
1798 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1799 used, there are restrictions on what may appear in the pattern. These
1800 are discussed in the pcrepartial documentation.
1801
1802 The string to be matched by pcre_exec()
1803
1804 The subject string is passed to pcre_exec() as a pointer in subject, a
1805 length in length, and a starting byte offset in startoffset. In UTF-8
1806 mode, the byte offset must point to the start of a UTF-8 character.
1807 Unlike the pattern string, the subject may contain binary zero bytes.
1808 When the starting offset is zero, the search for a match starts at the
1809 beginning of the subject, and this is by far the most common case.
1810
1811 A non-zero starting offset is useful when searching for another match
1812 in the same subject by calling pcre_exec() again after a previous suc-
1813 cess. Setting startoffset differs from just passing over a shortened
1814 string and setting PCRE_NOTBOL in the case of a pattern that begins
1815 with any kind of lookbehind. For example, consider the pattern
1816
1817 \Biss\B
1818
1819 which finds occurrences of "iss" in the middle of words. (\B matches
1820 only if the current position in the subject is not a word boundary.)
1821 When applied to the string "Mississipi" the first call to pcre_exec()
1822 finds the first occurrence. If pcre_exec() is called again with just
1823 the remainder of the subject, namely "issipi", it does not match,
1824 because \B is always false at the start of the subject, which is deemed
1825 to be a word boundary. However, if pcre_exec() is passed the entire
1826 string again, but with startoffset set to 4, it finds the second occur-
1827 rence of "iss" because it is able to look behind the starting point to
1828 discover that it is preceded by a letter.
1829
1830 If a non-zero starting offset is passed when the pattern is anchored,
1831 one attempt to match at the given offset is made. This can only succeed
1832 if the pattern does not require the match to be at the start of the
1833 subject.
1834
1835 How pcre_exec() returns captured substrings
1836
1837 In general, a pattern matches a certain portion of the subject, and in
1838 addition, further substrings from the subject may be picked out by
1839 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1840 this is called "capturing" in what follows, and the phrase "capturing
1841 subpattern" is used for a fragment of a pattern that picks out a sub-
1842 string. PCRE supports several other kinds of parenthesized subpattern
1843 that do not cause substrings to be captured.
1844
1845 Captured substrings are returned to the caller via a vector of integer
1846 offsets whose address is passed in ovector. The number of elements in
1847 the vector is passed in ovecsize, which must be a non-negative number.
1848 Note: this argument is NOT the size of ovector in bytes.
1849
1850 The first two-thirds of the vector is used to pass back captured sub-
1851 strings, each substring using a pair of integers. The remaining third
1852 of the vector is used as workspace by pcre_exec() while matching cap-
1853 turing subpatterns, and is not available for passing back information.
1854 The length passed in ovecsize should always be a multiple of three. If
1855 it is not, it is rounded down.
1856
1857 When a match is successful, information about captured substrings is
1858 returned in pairs of integers, starting at the beginning of ovector,
1859 and continuing up to two-thirds of its length at the most. The first
1860 element of a pair is set to the offset of the first character in a sub-
1861 string, and the second is set to the offset of the first character
1862 after the end of a substring. The first pair, ovector[0] and ovec-
1863 tor[1], identify the portion of the subject string matched by the
1864 entire pattern. The next pair is used for the first capturing subpat-
1865 tern, and so on. The value returned by pcre_exec() is one more than the
1866 highest numbered pair that has been set. For example, if two substrings
1867 have been captured, the returned value is 3. If there are no capturing
1868 subpatterns, the return value from a successful match is 1, indicating
1869 that just the first pair of offsets has been set.
1870
1871 If a capturing subpattern is matched repeatedly, it is the last portion
1872 of the string that it matched that is returned.
1873
1874 If the vector is too small to hold all the captured substring offsets,
1875 it is used as far as possible (up to two-thirds of its length), and the
1876 function returns a value of zero. In particular, if the substring off-
1877 sets are not of interest, pcre_exec() may be called with ovector passed
1878 as NULL and ovecsize as zero. However, if the pattern contains back
1879 references and the ovector is not big enough to remember the related
1880 substrings, PCRE has to get additional memory for use during matching.
1881 Thus it is usually advisable to supply an ovector.
1882
1883 The pcre_info() function can be used to find out how many capturing
1884 subpatterns there are in a compiled pattern. The smallest size for
1885 ovector that will allow for n captured substrings, in addition to the
1886 offsets of the substring matched by the whole pattern, is (n+1)*3.
1887
1888 It is possible for capturing subpattern number n+1 to match some part
1889 of the subject when subpattern n has not been used at all. For example,
1890 if the string "abc" is matched against the pattern (a|(z))(bc) the
1891 return from the function is 4, and subpatterns 1 and 3 are matched, but
1892 2 is not. When this happens, both values in the offset pairs corre-
1893 sponding to unused subpatterns are set to -1.
1894
1895 Offset values that correspond to unused subpatterns at the end of the
1896 expression are also set to -1. For example, if the string "abc" is
1897 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1898 matched. The return from the function is 2, because the highest used
1899 capturing subpattern number is 1. However, you can refer to the offsets
1900 for the second and third capturing subpatterns if you wish (assuming
1901 the vector is large enough, of course).
1902
1903 Some convenience functions are provided for extracting the captured
1904 substrings as separate strings. These are described below.
1905
1906 Error return values from pcre_exec()
1907
1908 If pcre_exec() fails, it returns a negative number. The following are
1909 defined in the header file:
1910
1911 PCRE_ERROR_NOMATCH (-1)
1912
1913 The subject string did not match the pattern.
1914
1915 PCRE_ERROR_NULL (-2)
1916
1917 Either code or subject was passed as NULL, or ovector was NULL and
1918 ovecsize was not zero.
1919
1920 PCRE_ERROR_BADOPTION (-3)
1921
1922 An unrecognized bit was set in the options argument.
1923
1924 PCRE_ERROR_BADMAGIC (-4)
1925
1926 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1927 to catch the case when it is passed a junk pointer and to detect when a
1928 pattern that was compiled in an environment of one endianness is run in
1929 an environment with the other endianness. This is the error that PCRE
1930 gives when the magic number is not present.
1931
1932 PCRE_ERROR_UNKNOWN_OPCODE (-5)
1933
1934 While running the pattern match, an unknown item was encountered in the
1935 compiled pattern. This error could be caused by a bug in PCRE or by
1936 overwriting of the compiled pattern.
1937
1938 PCRE_ERROR_NOMEMORY (-6)
1939
1940 If a pattern contains back references, but the ovector that is passed
1941 to pcre_exec() is not big enough to remember the referenced substrings,
1942 PCRE gets a block of memory at the start of matching to use for this
1943 purpose. If the call via pcre_malloc() fails, this error is given. The
1944 memory is automatically freed at the end of matching.
1945
1946 PCRE_ERROR_NOSUBSTRING (-7)
1947
1948 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1949 and pcre_get_substring_list() functions (see below). It is never
1950 returned by pcre_exec().
1951
1952 PCRE_ERROR_MATCHLIMIT (-8)
1953
1954 The backtracking limit, as specified by the match_limit field in a
1955 pcre_extra structure (or defaulted) was reached. See the description
1956 above.
1957
1958 PCRE_ERROR_CALLOUT (-9)
1959
1960 This error is never generated by pcre_exec() itself. It is provided for
1961 use by callout functions that want to yield a distinctive error code.
1962 See the pcrecallout documentation for details.
1963
1964 PCRE_ERROR_BADUTF8 (-10)
1965
1966 A string that contains an invalid UTF-8 byte sequence was passed as a
1967 subject.
1968
1969 PCRE_ERROR_BADUTF8_OFFSET (-11)
1970
1971 The UTF-8 byte sequence that was passed as a subject was valid, but the
1972 value of startoffset did not point to the beginning of a UTF-8 charac-
1973 ter.
1974
1975 PCRE_ERROR_PARTIAL (-12)
1976
1977 The subject string did not match, but it did match partially. See the
1978 pcrepartial documentation for details of partial matching.
1979
1980 PCRE_ERROR_BADPARTIAL (-13)
1981
1982 The PCRE_PARTIAL option was used with a compiled pattern containing
1983 items that are not supported for partial matching. See the pcrepartial
1984 documentation for details of partial matching.
1985
1986 PCRE_ERROR_INTERNAL (-14)
1987
1988 An unexpected internal error has occurred. This error could be caused
1989 by a bug in PCRE or by overwriting of the compiled pattern.
1990
1991 PCRE_ERROR_BADCOUNT (-15)
1992
1993 This error is given if the value of the ovecsize argument is negative.
1994
1995 PCRE_ERROR_RECURSIONLIMIT (-21)
1996
1997 The internal recursion limit, as specified by the match_limit_recursion
1998 field in a pcre_extra structure (or defaulted) was reached. See the
1999 description above.
2000
2001 PCRE_ERROR_NULLWSLIMIT (-22)
2002
2003 When a group that can match an empty substring is repeated with an
2004 unbounded upper limit, the subject position at the start of the group
2005 must be remembered, so that a test for an empty string can be made when
2006 the end of the group is reached. Some workspace is required for this;
2007 if it runs out, this error is given.
2008
2009 PCRE_ERROR_BADNEWLINE (-23)
2010
2011 An invalid combination of PCRE_NEWLINE_xxx options was given.
2012
2013 Error numbers -16 to -20 are not used by pcre_exec().
2014
2015
2016 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2017
2018 int pcre_copy_substring(const char *subject, int *ovector,
2019 int stringcount, int stringnumber, char *buffer,
2020 int buffersize);
2021
2022 int pcre_get_substring(const char *subject, int *ovector,
2023 int stringcount, int stringnumber,
2024 const char **stringptr);
2025
2026 int pcre_get_substring_list(const char *subject,
2027 int *ovector, int stringcount, const char ***listptr);
2028
2029 Captured substrings can be accessed directly by using the offsets
2030 returned by pcre_exec() in ovector. For convenience, the functions
2031 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2032 string_list() are provided for extracting captured substrings as new,
2033 separate, zero-terminated strings. These functions identify substrings
2034 by number. The next section describes functions for extracting named
2035 substrings.
2036
2037 A substring that contains a binary zero is correctly extracted and has
2038 a further zero added on the end, but the result is not, of course, a C
2039 string. However, you can process such a string by referring to the
2040 length that is returned by pcre_copy_substring() and pcre_get_sub-
2041 string(). Unfortunately, the interface to pcre_get_substring_list() is
2042 not adequate for handling strings containing binary zeros, because the
2043 end of the final string is not independently indicated.
2044
2045 The first three arguments are the same for all three of these func-
2046 tions: subject is the subject string that has just been successfully
2047 matched, ovector is a pointer to the vector of integer offsets that was
2048 passed to pcre_exec(), and stringcount is the number of substrings that
2049 were captured by the match, including the substring that matched the
2050 entire regular expression. This is the value returned by pcre_exec() if
2051 it is greater than zero. If pcre_exec() returned zero, indicating that
2052 it ran out of space in ovector, the value passed as stringcount should
2053 be the number of elements in the vector divided by three.
2054
2055 The functions pcre_copy_substring() and pcre_get_substring() extract a
2056 single substring, whose number is given as stringnumber. A value of
2057 zero extracts the substring that matched the entire pattern, whereas
2058 higher values extract the captured substrings. For pcre_copy_sub-
2059 string(), the string is placed in buffer, whose length is given by
2060 buffersize, while for pcre_get_substring() a new block of memory is
2061 obtained via pcre_malloc, and its address is returned via stringptr.
2062 The yield of the function is the length of the string, not including
2063 the terminating zero, or one of these error codes:
2064
2065 PCRE_ERROR_NOMEMORY (-6)
2066
2067 The buffer was too small for pcre_copy_substring(), or the attempt to
2068 get memory failed for pcre_get_substring().
2069
2070 PCRE_ERROR_NOSUBSTRING (-7)
2071
2072 There is no substring whose number is stringnumber.
2073
2074 The pcre_get_substring_list() function extracts all available sub-
2075 strings and builds a list of pointers to them. All this is done in a
2076 single block of memory that is obtained via pcre_malloc. The address of
2077 the memory block is returned via listptr, which is also the start of
2078 the list of string pointers. The end of the list is marked by a NULL
2079 pointer. The yield of the function is zero if all went well, or the
2080 error code
2081
2082 PCRE_ERROR_NOMEMORY (-6)
2083
2084 if the attempt to get the memory block failed.
2085
2086 When any of these functions encounter a substring that is unset, which
2087 can happen when capturing subpattern number n+1 matches some part of
2088 the subject, but subpattern n has not been used at all, they return an
2089 empty string. This can be distinguished from a genuine zero-length sub-
2090 string by inspecting the appropriate offset in ovector, which is nega-
2091 tive for unset substrings.
2092
2093 The two convenience functions pcre_free_substring() and pcre_free_sub-
2094 string_list() can be used to free the memory returned by a previous
2095 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2096 tively. They do nothing more than call the function pointed to by
2097 pcre_free, which of course could be called directly from a C program.
2098 However, PCRE is used in some situations where it is linked via a spe-
2099 cial interface to another programming language that cannot use
2100 pcre_free directly; it is for these cases that the functions are pro-
2101 vided.
2102
2103
2104 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2105
2106 int pcre_get_stringnumber(const pcre *code,
2107 const char *name);
2108
2109 int pcre_copy_named_substring(const pcre *code,
2110 const char *subject, int *ovector,
2111 int stringcount, const char *stringname,
2112 char *buffer, int buffersize);
2113
2114 int pcre_get_named_substring(const pcre *code,
2115 const char *subject, int *ovector,
2116 int stringcount, const char *stringname,
2117 const char **stringptr);
2118
2119 To extract a substring by name, you first have to find associated num-
2120 ber. For example, for this pattern
2121
2122 (a+)b(?<xxx>\d+)...
2123
2124 the number of the subpattern called "xxx" is 2. If the name is known to
2125 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2126 name by calling pcre_get_stringnumber(). The first argument is the com-
2127 piled pattern, and the second is the name. The yield of the function is
2128 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2129 subpattern of that name.
2130
2131 Given the number, you can extract the substring directly, or use one of
2132 the functions described in the previous section. For convenience, there
2133 are also two functions that do the whole job.
2134
2135 Most of the arguments of pcre_copy_named_substring() and
2136 pcre_get_named_substring() are the same as those for the similarly
2137 named functions that extract by number. As these are described in the
2138 previous section, they are not re-described here. There are just two
2139 differences:
2140
2141 First, instead of a substring number, a substring name is given. Sec-
2142 ond, there is an extra argument, given at the start, which is a pointer
2143 to the compiled pattern. This is needed in order to gain access to the
2144 name-to-number translation table.
2145
2146 These functions call pcre_get_stringnumber(), and if it succeeds, they
2147 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2148 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2149 behaviour may not be what you want (see the next section).
2150
2151
2152 DUPLICATE SUBPATTERN NAMES
2153
2154 int pcre_get_stringtable_entries(const pcre *code,
2155 const char *name, char **first, char **last);
2156
2157 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2158 subpatterns are not required to be unique. Normally, patterns with
2159 duplicate names are such that in any one match, only one of the named
2160 subpatterns participates. An example is shown in the pcrepattern docu-
2161 mentation. When duplicates are present, pcre_copy_named_substring() and
2162 pcre_get_named_substring() return the first substring corresponding to
2163 the given name that is set. If none are set, an empty string is
2164 returned. The pcre_get_stringnumber() function returns one of the num-
2165 bers that are associated with the name, but it is not defined which it
2166 is.
2167
2168 If you want to get full details of all captured substrings for a given
2169 name, you must use the pcre_get_stringtable_entries() function. The
2170 first argument is the compiled pattern, and the second is the name. The
2171 third and fourth are pointers to variables which are updated by the
2172 function. After it has run, they point to the first and last entries in
2173 the name-to-number table for the given name. The function itself
2174 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2175 there are none. The format of the table is described above in the sec-
2176 tion entitled Information about a pattern. Given all the relevant
2177 entries for the name, you can extract each of their numbers, and hence
2178 the captured data, if any.
2179
2180
2181 FINDING ALL POSSIBLE MATCHES
2182
2183 The traditional matching function uses a similar algorithm to Perl,
2184 which stops when it finds the first match, starting at a given point in
2185 the subject. If you want to find all possible matches, or the longest
2186 possible match, consider using the alternative matching function (see
2187 below) instead. If you cannot use the alternative function, but still
2188 need to find all possible matches, you can kludge it up by making use
2189 of the callout facility, which is described in the pcrecallout documen-
2190 tation.
2191
2192 What you have to do is to insert a callout right at the end of the pat-
2193 tern. When your callout function is called, extract and save the cur-
2194 rent matched substring. Then return 1, which forces pcre_exec() to
2195 backtrack and try other alternatives. Ultimately, when it runs out of
2196 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2197
2198
2199 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2200
2201 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2202 const char *subject, int length, int startoffset,
2203 int options, int *ovector, int ovecsize,
2204 int *workspace, int wscount);
2205
2206 The function pcre_dfa_exec() is called to match a subject string
2207 against a compiled pattern, using a matching algorithm that scans the
2208 subject string just once, and does not backtrack. This has different
2209 characteristics to the normal algorithm, and is not compatible with
2210 Perl. Some of the features of PCRE patterns are not supported. Never-
2211 theless, there are times when this kind of matching can be useful. For
2212 a discussion of the two matching algorithms, see the pcrematching docu-
2213 mentation.
2214
2215 The arguments for the pcre_dfa_exec() function are the same as for
2216 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2217 ent way, and this is described below. The other common arguments are
2218 used in the same way as for pcre_exec(), so their description is not
2219 repeated here.
2220
2221 The two additional arguments provide workspace for the function. The
2222 workspace vector should contain at least 20 elements. It is used for
2223 keeping track of multiple paths through the pattern tree. More
2224 workspace will be needed for patterns and subjects where there are a
2225 lot of potential matches.
2226
2227 Here is an example of a simple call to pcre_dfa_exec():
2228
2229 int rc;
2230 int ovector[10];
2231 int wspace[20];
2232 rc = pcre_dfa_exec(
2233 re, /* result of pcre_compile() */
2234 NULL, /* we didn't study the pattern */
2235 "some string", /* the subject string */
2236 11, /* the length of the subject string */
2237 0, /* start at offset 0 in the subject */
2238 0, /* default options */
2239 ovector, /* vector of integers for substring information */
2240 10, /* number of elements (NOT size in bytes) */
2241 wspace, /* working space vector */
2242 20); /* number of elements (NOT size in bytes) */
2243
2244 Option bits for pcre_dfa_exec()
2245
2246 The unused bits of the options argument for pcre_dfa_exec() must be
2247 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2248 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2249 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2250 three of these are the same as for pcre_exec(), so their description is
2251 not repeated here.
2252
2253 PCRE_PARTIAL
2254
2255 This has the same general effect as it does for pcre_exec(), but the
2256 details are slightly different. When PCRE_PARTIAL is set for
2257 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2258 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2259 been no complete matches, but there is still at least one matching pos-
2260 sibility. The portion of the string that provided the partial match is
2261 set as the first matching string.
2262
2263 PCRE_DFA_SHORTEST
2264
2265 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2266 stop as soon as it has found one match. Because of the way the alterna-
2267 tive algorithm works, this is necessarily the shortest possible match
2268 at the first possible matching point in the subject string.
2269
2270 PCRE_DFA_RESTART
2271
2272 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2273 returns a partial match, it is possible to call it again, with addi-
2274 tional subject characters, and have it continue with the same match.
2275 The PCRE_DFA_RESTART option requests this action; when it is set, the
2276 workspace and wscount options must reference the same vector as before
2277 because data about the match so far is left in them after a partial
2278 match. There is more discussion of this facility in the pcrepartial
2279 documentation.
2280
2281 Successful returns from pcre_dfa_exec()
2282
2283 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2284 string in the subject. Note, however, that all the matches from one run
2285 of the function start at the same point in the subject. The shorter
2286 matches are all initial substrings of the longer matches. For example,
2287 if the pattern
2288
2289 <.*>
2290
2291 is matched against the string
2292
2293 This is <something> <something else> <something further> no more
2294
2295 the three matched strings are
2296
2297 <something>
2298 <something> <something else>
2299 <something> <something else> <something further>
2300
2301 On success, the yield of the function is a number greater than zero,
2302 which is the number of matched substrings. The substrings themselves
2303 are returned in ovector. Each string uses two elements; the first is
2304 the offset to the start, and the second is the offset to the end. In
2305 fact, all the strings have the same start offset. (Space could have
2306 been saved by giving this only once, but it was decided to retain some
2307 compatibility with the way pcre_exec() returns data, even though the
2308 meaning of the strings is different.)
2309
2310 The strings are returned in reverse order of length; that is, the long-
2311 est matching string is given first. If there were too many matches to
2312 fit into ovector, the yield of the function is zero, and the vector is
2313 filled with the longest matches.
2314
2315 Error returns from pcre_dfa_exec()
2316
2317 The pcre_dfa_exec() function returns a negative number when it fails.
2318 Many of the errors are the same as for pcre_exec(), and these are
2319 described above. There are in addition the following errors that are
2320 specific to pcre_dfa_exec():
2321
2322 PCRE_ERROR_DFA_UITEM (-16)
2323
2324 This return is given if pcre_dfa_exec() encounters an item in the pat-
2325 tern that it does not support, for instance, the use of \C or a back
2326 reference.
2327
2328 PCRE_ERROR_DFA_UCOND (-17)
2329
2330 This return is given if pcre_dfa_exec() encounters a condition item
2331 that uses a back reference for the condition, or a test for recursion
2332 in a specific group. These are not supported.
2333
2334 PCRE_ERROR_DFA_UMLIMIT (-18)
2335
2336 This return is given if pcre_dfa_exec() is called with an extra block
2337 that contains a setting of the match_limit field. This is not supported
2338 (it is meaningless).
2339
2340 PCRE_ERROR_DFA_WSSIZE (-19)
2341
2342 This return is given if pcre_dfa_exec() runs out of space in the
2343 workspace vector.
2344
2345 PCRE_ERROR_DFA_RECURSE (-20)
2346
2347 When a recursive subpattern is processed, the matching function calls
2348 itself recursively, using private vectors for ovector and workspace.
2349 This error is given if the output vector is not large enough. This
2350 should be extremely rare, as a vector of size 1000 is used.
2351
2352
2353 SEE ALSO
2354
2355 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2356 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2357
2358
2359 AUTHOR
2360
2361 Philip Hazel
2362 University Computing Service
2363 Cambridge CB2 3QH, England.
2364
2365
2366 REVISION
2367
2368 Last updated: 06 March 2007
2369 Copyright (c) 1997-2007 University of Cambridge.
2370 ------------------------------------------------------------------------------
2371
2372
2373 PCRECALLOUT(3) PCRECALLOUT(3)
2374
2375
2376 NAME
2377 PCRE - Perl-compatible regular expressions
2378
2379
2380 PCRE CALLOUTS
2381
2382 int (*pcre_callout)(pcre_callout_block *);
2383
2384 PCRE provides a feature called "callout", which is a means of temporar-
2385 ily passing control to the caller of PCRE in the middle of pattern
2386 matching. The caller of PCRE provides an external function by putting
2387 its entry point in the global variable pcre_callout. By default, this
2388 variable contains NULL, which disables all calling out.
2389
2390 Within a regular expression, (?C) indicates the points at which the
2391 external function is to be called. Different callout points can be
2392 identified by putting a number less than 256 after the letter C. The
2393 default value is zero. For example, this pattern has two callout
2394 points:
2395
2396 (?C1)eabc(?C2)def
2397
2398 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2399 called, PCRE automatically inserts callouts, all with number 255,
2400 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2401 used with the pattern
2402
2403 A(\d{2}|--)
2404
2405 it is processed as if it were
2406
2407 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2408
2409 Notice that there is a callout before and after each parenthesis and
2410 alternation bar. Automatic callouts can be used for tracking the
2411 progress of pattern matching. The pcretest command has an option that
2412 sets automatic callouts; when it is used, the output indicates how the
2413 pattern is matched. This is useful information when you are trying to
2414 optimize the performance of a particular pattern.
2415
2416
2417 MISSING CALLOUTS
2418
2419 You should be aware that, because of optimizations in the way PCRE
2420 matches patterns, callouts sometimes do not happen. For example, if the
2421 pattern is
2422
2423 ab(?C4)cd
2424
2425 PCRE knows that any matching string must contain the letter "d". If the
2426 subject string is "abyz", the lack of "d" means that matching doesn't
2427 ever start, and the callout is never reached. However, with "abyd",
2428 though the result is still no match, the callout is obeyed.
2429
2430
2431 THE CALLOUT INTERFACE
2432
2433 During matching, when PCRE reaches a callout point, the external func-
2434 tion defined by pcre_callout is called (if it is set). This applies to
2435 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2436 only argument to the callout function is a pointer to a pcre_callout
2437 block. This structure contains the following fields:
2438
2439 int version;
2440 int callout_number;
2441 int *offset_vector;
2442 const char *subject;
2443 int subject_length;
2444 int start_match;
2445 int current_position;
2446 int capture_top;
2447 int capture_last;
2448 void *callout_data;
2449 int pattern_position;
2450 int next_item_length;
2451
2452 The version field is an integer containing the version number of the
2453 block format. The initial version was 0; the current version is 1. The
2454 version number will change again in future if additional fields are
2455 added, but the intention is never to remove any of the existing fields.
2456
2457 The callout_number field contains the number of the callout, as com-
2458 piled into the pattern (that is, the number after ?C for manual call-
2459 outs, and 255 for automatically generated callouts).
2460
2461 The offset_vector field is a pointer to the vector of offsets that was
2462 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2463 pcre_exec() is used, the contents can be inspected in order to extract
2464 substrings that have been matched so far, in the same way as for
2465 extracting substrings after a match has completed. For pcre_dfa_exec()
2466 this field is not useful.
2467
2468 The subject and subject_length fields contain copies of the values that
2469 were passed to pcre_exec().
2470
2471 The start_match field contains the offset within the subject at which
2472 the current match attempt started. If the pattern is not anchored, the
2473 callout function may be called several times from the same point in the
2474 pattern for different starting points in the subject.
2475
2476 The current_position field contains the offset within the subject of
2477 the current match pointer.
2478
2479 When the pcre_exec() function is used, the capture_top field contains
2480 one more than the number of the highest numbered captured substring so
2481 far. If no substrings have been captured, the value of capture_top is
2482 one. This is always the case when pcre_dfa_exec() is used, because it
2483 does not support captured substrings.
2484
2485 The capture_last field contains the number of the most recently cap-
2486 tured substring. If no substrings have been captured, its value is -1.
2487 This is always the case when pcre_dfa_exec() is used.
2488
2489 The callout_data field contains a value that is passed to pcre_exec()
2490 or pcre_dfa_exec() specifically so that it can be passed back in call-
2491 outs. It is passed in the pcre_callout field of the pcre_extra data
2492 structure. If no such data was passed, the value of callout_data in a
2493 pcre_callout block is NULL. There is a description of the pcre_extra
2494 structure in the pcreapi documentation.
2495
2496 The pattern_position field is present from version 1 of the pcre_call-
2497 out structure. It contains the offset to the next item to be matched in
2498 the pattern string.
2499
2500 The next_item_length field is present from version 1 of the pcre_call-
2501 out structure. It contains the length of the next item to be matched in
2502 the pattern string. When the callout immediately precedes an alterna-
2503 tion bar, a closing parenthesis, or the end of the pattern, the length
2504 is zero. When the callout precedes an opening parenthesis, the length
2505 is that of the entire subpattern.
2506
2507 The pattern_position and next_item_length fields are intended to help
2508 in distinguishing between different automatic callouts, which all have
2509 the same callout number. However, they are set for all callouts.
2510
2511
2512 RETURN VALUES
2513
2514 The external callout function returns an integer to PCRE. If the value
2515 is zero, matching proceeds as normal. If the value is greater than
2516 zero, matching fails at the current point, but the testing of other
2517 matching possibilities goes ahead, just as if a lookahead assertion had
2518 failed. If the value is less than zero, the match is abandoned, and
2519 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2520
2521 Negative values should normally be chosen from the set of
2522 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2523 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2524 reserved for use by callout functions; it will never be used by PCRE
2525 itself.
2526
2527
2528 AUTHOR
2529
2530 Philip Hazel
2531 University Computing Service
2532 Cambridge CB2 3QH, England.
2533
2534
2535 REVISION
2536
2537 Last updated: 06 March 2007
2538 Copyright (c) 1997-2007 University of Cambridge.
2539 ------------------------------------------------------------------------------
2540
2541
2542 PCRECOMPAT(3) PCRECOMPAT(3)
2543
2544
2545 NAME
2546 PCRE - Perl-compatible regular expressions
2547
2548
2549 DIFFERENCES BETWEEN PCRE AND PERL
2550
2551 This document describes the differences in the ways that PCRE and Perl
2552 handle regular expressions. The differences described here are mainly
2553 with respect to Perl 5.8, though PCRE version 7.0 contains some fea-
2554 tures that are expected to be in the forthcoming Perl 5.10.
2555
2556 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2557 of what it does have are given in the section on UTF-8 support in the
2558 main pcre page.
2559
2560 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2561 permits them, but they do not mean what you might think. For example,
2562 (?!a){3} does not assert that the next three characters are not "a". It
2563 just asserts that the next character is not "a" three times.
2564
2565 3. Capturing subpatterns that occur inside negative lookahead asser-
2566 tions are counted, but their entries in the offsets vector are never
2567 set. Perl sets its numerical variables from any such patterns that are
2568 matched before the assertion fails to match something (thereby succeed-
2569 ing), but only if the negative lookahead assertion contains just one
2570 branch.
2571
2572 4. Though binary zero characters are supported in the subject string,
2573 they are not allowed in a pattern string because it is passed as a nor-
2574 mal C string, terminated by zero. The escape sequence \0 can be used in
2575 the pattern to represent a binary zero.
2576
2577 5. The following Perl escape sequences are not supported: \l, \u, \L,
2578 \U, and \N. In fact these are implemented by Perl's general string-han-
2579 dling and are not part of its pattern matching engine. If any of these
2580 are encountered by PCRE, an error is generated.
2581
2582 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2583 is built with Unicode character property support. The properties that
2584 can be tested with \p and \P are limited to the general category prop-
2585 erties such as Lu and Nd, script names such as Greek or Han, and the
2586 derived properties Any and L&.
2587
2588 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2589 ters in between are treated as literals. This is slightly different
2590 from Perl in that $ and @ are also handled as literals inside the
2591 quotes. In Perl, they cause variable interpolation (but of course PCRE
2592 does not have variables). Note the following examples:
2593
2594 Pattern PCRE matches Perl matches
2595
2596 \Qabc$xyz\E abc$xyz abc followed by the
2597 contents of $xyz
2598 \Qabc\$xyz\E abc\$xyz abc\$xyz
2599 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2600
2601 The \Q...\E sequence is recognized both inside and outside character
2602 classes.
2603
2604 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2605 constructions. However, there is support for recursive patterns. This
2606 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2607 "callout" feature allows an external function to be called during pat-
2608 tern matching. See the pcrecallout documentation for details.
2609
2610 9. Subpatterns that are called recursively or as "subroutines" are
2611 always treated as atomic groups in PCRE. This is like Python, but
2612 unlike Perl.
2613
2614 10. There are some differences that are concerned with the settings of
2615 captured strings when part of a pattern is repeated. For example,
2616 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2617 unset, but in PCRE it is set to "b".
2618
2619 11. PCRE provides some extensions to the Perl regular expression facil-
2620 ities. Perl 5.10 will include new features that are not in earlier
2621 versions, some of which (such as named parentheses) have been in PCRE
2622 for some time. This list is with respect to Perl 5.10:
2623
2624 (a) Although lookbehind assertions must match fixed length strings,
2625 each alternative branch of a lookbehind assertion can match a different
2626 length of string. Perl requires them all to have the same length.
2627
2628 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2629 meta-character matches only at the very end of the string.
2630
2631 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2632 cial meaning is faulted. Otherwise, like Perl, the backslash is
2633 ignored. (Perl can be made to issue a warning.)
2634
2635 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2636 fiers is inverted, that is, by default they are not greedy, but if fol-
2637 lowed by a question mark they are.
2638
2639 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2640 tried only at the first matching position in the subject string.
2641
2642 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2643 TURE options for pcre_exec() have no Perl equivalents.
2644
2645 (g) The callout facility is PCRE-specific.
2646
2647 (h) The partial matching facility is PCRE-specific.
2648
2649 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2650 even on different hosts that have the other endianness.
2651
2652 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2653 different way and is not Perl-compatible.
2654
2655
2656 AUTHOR
2657
2658 Philip Hazel
2659 University Computing Service
2660 Cambridge CB2 3QH, England.
2661
2662
2663 REVISION
2664
2665 Last updated: 06 March 2007
2666 Copyright (c) 1997-2007 University of Cambridge.
2667 ------------------------------------------------------------------------------
2668
2669
2670 PCREPATTERN(3) PCREPATTERN(3)
2671
2672
2673 NAME
2674 PCRE - Perl-compatible regular expressions
2675
2676
2677 PCRE REGULAR EXPRESSION DETAILS
2678
2679 The syntax and semantics of the regular expressions supported by PCRE
2680 are described below. Regular expressions are also described in the Perl
2681 documentation and in a number of books, some of which have copious
2682 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2683 by O'Reilly, covers regular expressions in great detail. This descrip-
2684 tion of PCRE's regular expressions is intended as reference material.
2685
2686 The original operation of PCRE was on strings of one-byte characters.
2687 However, there is now also support for UTF-8 character strings. To use
2688 this, you must build PCRE to include UTF-8 support, and then call
2689 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2690 matching is mentioned in several places below. There is also a summary
2691 of UTF-8 features in the section on UTF-8 support in the main pcre
2692 page.
2693
2694 The remainder of this document discusses the patterns that are sup-
2695 ported by PCRE when its main matching function, pcre_exec(), is used.
2696 From release 6.0, PCRE offers a second matching function,
2697 pcre_dfa_exec(), which matches using a different algorithm that is not
2698 Perl-compatible. The advantages and disadvantages of the alternative
2699 function, and how it differs from the normal function, are discussed in
2700 the pcrematching page.
2701
2702
2703 CHARACTERS AND METACHARACTERS
2704
2705 A regular expression is a pattern that is matched against a subject
2706 string from left to right. Most characters stand for themselves in a
2707 pattern, and match the corresponding characters in the subject. As a
2708 trivial example, the pattern
2709
2710 The quick brown fox
2711
2712 matches a portion of a subject string that is identical to itself. When
2713 caseless matching is specified (the PCRE_CASELESS option), letters are
2714 matched independently of case. In UTF-8 mode, PCRE always understands
2715 the concept of case for characters whose values are less than 128, so
2716 caseless matching is always possible. For characters with higher val-
2717 ues, the concept of case is supported if PCRE is compiled with Unicode
2718 property support, but not otherwise. If you want to use caseless
2719 matching for characters 128 and above, you must ensure that PCRE is
2720 compiled with Unicode property support as well as with UTF-8 support.
2721
2722 The power of regular expressions comes from the ability to include
2723 alternatives and repetitions in the pattern. These are encoded in the
2724 pattern by the use of metacharacters, which do not stand for themselves
2725 but instead are interpreted in some special way.
2726
2727 There are two different sets of metacharacters: those that are recog-
2728 nized anywhere in the pattern except within square brackets, and those
2729 that are recognized within square brackets. Outside square brackets,
2730 the metacharacters are as follows:
2731
2732 \ general escape character with several uses
2733 ^ assert start of string (or line, in multiline mode)
2734 $ assert end of string (or line, in multiline mode)
2735 . match any character except newline (by default)
2736 [ start character class definition
2737 | start of alternative branch
2738 ( start subpattern
2739 ) end subpattern
2740 ? extends the meaning of (
2741 also 0 or 1 quantifier
2742 also quantifier minimizer
2743 * 0 or more quantifier
2744 + 1 or more quantifier
2745 also "possessive quantifier"
2746 { start min/max quantifier
2747
2748 Part of a pattern that is in square brackets is called a "character
2749 class". In a character class the only metacharacters are:
2750
2751 \ general escape character
2752 ^ negate the class, but only if the first character
2753 - indicates character range
2754 [ POSIX character class (only if followed by POSIX
2755 syntax)
2756 ] terminates the character class
2757
2758 The following sections describe the use of each of the metacharacters.
2759
2760
2761 BACKSLASH
2762
2763 The backslash character has several uses. Firstly, if it is followed by
2764 a non-alphanumeric character, it takes away any special meaning that
2765 character may have. This use of backslash as an escape character
2766 applies both inside and outside character classes.
2767
2768 For example, if you want to match a * character, you write \* in the
2769 pattern. This escaping action applies whether or not the following
2770 character would otherwise be interpreted as a metacharacter, so it is
2771 always safe to precede a non-alphanumeric with backslash to specify
2772 that it stands for itself. In particular, if you want to match a back-
2773 slash, you write \\.
2774
2775 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2776 the pattern (other than in a character class) and characters between a
2777 # outside a character class and the next newline are ignored. An escap-
2778 ing backslash can be used to include a whitespace or # character as
2779 part of the pattern.
2780
2781 If you want to remove the special meaning from a sequence of charac-
2782 ters, you can do so by putting them between \Q and \E. This is differ-
2783 ent from Perl in that $ and @ are handled as literals in \Q...\E
2784 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2785 tion. Note the following examples:
2786
2787 Pattern PCRE matches Perl matches
2788
2789 \Qabc$xyz\E abc$xyz abc followed by the
2790 contents of $xyz
2791 \Qabc\$xyz\E abc\$xyz abc\$xyz
2792 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2793
2794 The \Q...\E sequence is recognized both inside and outside character
2795 classes.
2796
2797 Non-printing characters
2798
2799 A second use of backslash provides a way of encoding non-printing char-
2800 acters in patterns in a visible manner. There is no restriction on the
2801 appearance of non-printing characters, apart from the binary zero that
2802 terminates a pattern, but when a pattern is being prepared by text
2803 editing, it is usually easier to use one of the following escape
2804 sequences than the binary character it represents:
2805
2806 \a alarm, that is, the BEL character (hex 07)
2807 \cx "control-x", where x is any character
2808 \e escape (hex 1B)
2809 \f formfeed (hex 0C)
2810 \n newline (hex 0A)
2811 \r carriage return (hex 0D)
2812 \t tab (hex 09)
2813 \ddd character with octal code ddd, or backreference
2814 \xhh character with hex code hh
2815 \x{hhh..} character with hex code hhh..
2816
2817 The precise effect of \cx is as follows: if x is a lower case letter,
2818 it is converted to upper case. Then bit 6 of the character (hex 40) is
2819 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2820 becomes hex 7B.
2821
2822 After \x, from zero to two hexadecimal digits are read (letters can be
2823 in upper or lower case). Any number of hexadecimal digits may appear
2824 between \x{ and }, but the value of the character code must be less
2825 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2826 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2827 hexadecimal digits appear between \x{ and }, or if there is no termi-
2828 nating }, this form of escape is not recognized. Instead, the initial
2829 \x will be interpreted as a basic hexadecimal escape, with no following
2830 digits, giving a character whose value is zero.
2831
2832 Characters whose value is less than 256 can be defined by either of the
2833 two syntaxes for \x. There is no difference in the way they are han-
2834 dled. For example, \xdc is exactly the same as \x{dc}.
2835
2836 After \0 up to two further octal digits are read. If there are fewer
2837 than two digits, just those that are present are used. Thus the
2838 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2839 (code value 7). Make sure you supply two digits after the initial zero
2840 if the pattern character that follows is itself an octal digit.
2841
2842 The handling of a backslash followed by a digit other than 0 is compli-
2843 cated. Outside a character class, PCRE reads it and any following dig-
2844 its as a decimal number. If the number is less than 10, or if there
2845 have been at least that many previous capturing left parentheses in the
2846 expression, the entire sequence is taken as a back reference. A
2847 description of how this works is given later, following the discussion
2848 of parenthesized subpatterns.
2849
2850 Inside a character class, or if the decimal number is greater than 9
2851 and there have not been that many capturing subpatterns, PCRE re-reads
2852 up to three octal digits following the backslash, and uses them to gen-
2853 erate a data character. Any subsequent digits stand for themselves. In
2854 non-UTF-8 mode, the value of a character specified in octal must be
2855 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2856 example:
2857
2858 \040 is another way of writing a space
2859 \40 is the same, provided there are fewer than 40
2860 previous capturing subpatterns
2861 \7 is always a back reference
2862 \11 might be a back reference, or another way of
2863 writing a tab
2864 \011 is always a tab
2865 \0113 is a tab followed by the character "3"
2866 \113 might be a back reference, otherwise the
2867 character with octal code 113
2868 \377 might be a back reference, otherwise
2869 the byte consisting entirely of 1 bits
2870 \81 is either a back reference, or a binary zero
2871 followed by the two characters "8" and "1"
2872
2873 Note that octal values of 100 or greater must not be introduced by a
2874 leading zero, because no more than three octal digits are ever read.
2875
2876 All the sequences that define a single character value can be used both
2877 inside and outside character classes. In addition, inside a character
2878 class, the sequence \b is interpreted as the backspace character (hex
2879 08), and the sequences \R and \X are interpreted as the characters "R"
2880 and "X", respectively. Outside a character class, these sequences have
2881 different meanings (see below).
2882
2883 Absolute and relative back references
2884
2885 The sequence \g followed by a positive or negative number, optionally
2886 enclosed in braces, is an absolute or relative back reference. Back
2887 references are discussed later, following the discussion of parenthe-
2888 sized subpatterns.
2889
2890 Generic character types
2891
2892 Another use of backslash is for specifying generic character types. The
2893 following are always recognized:
2894
2895 \d any decimal digit
2896 \D any character that is not a decimal digit
2897 \s any whitespace character
2898 \S any character that is not a whitespace character
2899 \w any "word" character
2900 \W any "non-word" character
2901
2902 Each pair of escape sequences partitions the complete set of characters
2903 into two disjoint sets. Any given character matches one, and only one,
2904 of each pair.
2905
2906 These character type sequences can appear both inside and outside char-
2907 acter classes. They each match one character of the appropriate type.
2908 If the current matching point is at the end of the subject string, all
2909 of them fail, since there is no character to match.
2910
2911 For compatibility with Perl, \s does not match the VT character (code
2912 11). This makes it different from the the POSIX "space" class. The \s
2913 characters are HT (9), LF (10), FF (12), CR (13), and space (32). (If
2914 "use locale;" is included in a Perl script, \s may match the VT charac-
2915 ter. In PCRE, it never does.)
2916
2917 A "word" character is an underscore or any character less than 256 that
2918 is a letter or digit. The definition of letters and digits is con-
2919 trolled by PCRE's low-valued character tables, and may vary if locale-
2920 specific matching is taking place (see "Locale support" in the pcreapi
2921 page). For example, in the "fr_FR" (French) locale, some character
2922 codes greater than 128 are used for accented letters, and these are
2923 matched by \w.
2924
2925 In UTF-8 mode, characters with values greater than 128 never match \d,
2926 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2927 code character property support is available. The use of locales with
2928 Unicode is discouraged.
2929
2930 Newline sequences
2931
2932 Outside a character class, the escape sequence \R matches any Unicode
2933 newline sequence. This is an extension to Perl. In non-UTF-8 mode \R is
2934 equivalent to the following:
2935
2936 (?>\r\n|\n|\x0b|\f|\r|\x85)
2937
2938 This is an example of an "atomic group", details of which are given
2939 below. This particular group matches either the two-character sequence
2940 CR followed by LF, or one of the single characters LF (linefeed,
2941 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
2942 return, U+000D), or NEL (next line, U+0085). The two-character sequence
2943 is treated as a single unit that cannot be split.
2944
2945 In UTF-8 mode, two additional characters whose codepoints are greater
2946 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
2947 rator, U+2029). Unicode character property support is not needed for
2948 these characters to be recognized.
2949
2950 Inside a character class, \R matches the letter "R".
2951
2952 Unicode character properties
2953
2954 When PCRE is built with Unicode character property support, three addi-
2955 tional escape sequences to match character properties are available
2956 when UTF-8 mode is selected. They are:
2957
2958 \p{xx} a character with the xx property
2959 \P{xx} a character without the xx property
2960 \X an extended Unicode sequence
2961
2962 The property names represented by xx above are limited to the Unicode
2963 script names, the general category properties, and "Any", which matches
2964 any character (including newline). Other properties such as "InMusical-
2965 Symbols" are not currently supported by PCRE. Note that \P{Any} does
2966 not match any characters, so always causes a match failure.
2967
2968 Sets of Unicode characters are defined as belonging to certain scripts.
2969 A character from one of these sets can be matched using a script name.
2970 For example:
2971
2972 \p{Greek}
2973 \P{Han}
2974
2975 Those that are not part of an identified script are lumped together as
2976 "Common". The current list of scripts is:
2977
2978 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
2979 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
2980 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
2981 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
2982 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
2983 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
2984 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
2985 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
2986 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
2987
2988 Each character has exactly one general category property, specified by
2989 a two-letter abbreviation. For compatibility with Perl, negation can be
2990 specified by including a circumflex between the opening brace and the
2991 property name. For example, \p{^Lu} is the same as \P{Lu}.
2992
2993 If only one letter is specified with \p or \P, it includes all the gen-
2994 eral category properties that start with that letter. In this case, in
2995 the absence of negation, the curly brackets in the escape sequence are
2996 optional; these two examples have the same effect:
2997
2998 \p{L}
2999 \pL
3000
3001 The following general category property codes are supported:
3002
3003 C Other
3004 Cc Control
3005 Cf Format
3006 Cn Unassigned
3007 Co Private use
3008 Cs Surrogate
3009
3010 L Letter
3011 Ll Lower case letter
3012 Lm Modifier letter
3013 Lo Other letter
3014 Lt Title case letter
3015 Lu Upper case letter
3016
3017 M Mark
3018 Mc Spacing mark
3019 Me Enclosing mark
3020 Mn Non-spacing mark
3021
3022 N Number
3023 Nd Decimal number
3024 Nl Letter number
3025 No Other number
3026
3027 P Punctuation
3028 Pc Connector punctuation
3029 Pd Dash punctuation
3030 Pe Close punctuation
3031 Pf Final punctuation
3032 Pi Initial punctuation
3033 Po Other punctuation
3034 Ps Open punctuation
3035
3036 S Symbol
3037 Sc Currency symbol
3038 Sk Modifier symbol
3039 Sm Mathematical symbol
3040 So Other symbol
3041
3042 Z Separator
3043 Zl Line separator
3044 Zp Paragraph separator
3045 Zs Space separator
3046
3047 The special property L& is also supported: it matches a character that
3048 has the Lu, Ll, or Lt property, in other words, a letter that is not
3049 classified as a modifier or "other".
3050
3051 The long synonyms for these properties that Perl supports (such as
3052 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3053 any of these properties with "Is".
3054
3055 No character that is in the Unicode table has the Cn (unassigned) prop-
3056 erty. Instead, this property is assumed for any code point that is not
3057 in the Unicode table.
3058
3059 Specifying caseless matching does not affect these escape sequences.
3060 For example, \p{Lu} always matches only upper case letters.
3061
3062 The \X escape matches any number of Unicode characters that form an
3063 extended Unicode sequence. \X is equivalent to
3064
3065 (?>\PM\pM*)
3066
3067 That is, it matches a character without the "mark" property, followed
3068 by zero or more characters with the "mark" property, and treats the
3069 sequence as an atomic group (see below). Characters with the "mark"
3070 property are typically accents that affect the preceding character.
3071
3072 Matching characters by Unicode property is not fast, because PCRE has
3073 to search a structure that contains data for over fifteen thousand
3074 characters. That is why the traditional escape sequences such as \d and
3075 \w do not use Unicode properties in PCRE.
3076
3077 Simple assertions
3078
3079 The final use of backslash is for certain simple assertions. An asser-
3080 tion specifies a condition that has to be met at a particular point in
3081 a match, without consuming any characters from the subject string. The
3082 use of subpatterns for more complicated assertions is described below.
3083 The backslashed assertions are:
3084
3085 \b matches at a word boundary
3086 \B matches when not at a word boundary
3087 \A matches at the start of the subject
3088 \Z matches at the end of the subject
3089 also matches before a newline at the end of the subject
3090 \z matches only at the end of the subject
3091 \G matches at the first matching position in the subject
3092
3093 These assertions may not appear in character classes (but note that \b
3094 has a different meaning, namely the backspace character, inside a char-
3095 acter class).
3096
3097 A word boundary is a position in the subject string where the current
3098 character and the previous character do not both match \w or \W (i.e.
3099 one matches \w and the other matches \W), or the start or end of the
3100 string if the first or last character matches \w, respectively.
3101
3102 The \A, \Z, and \z assertions differ from the traditional circumflex
3103 and dollar (described in the next section) in that they only ever match
3104 at the very start and end of the subject string, whatever options are
3105 set. Thus, they are independent of multiline mode. These three asser-
3106 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3107 affect only the behaviour of the circumflex and dollar metacharacters.
3108 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3109 cating that matching is to start at a point other than the beginning of
3110 the subject, \A can never match. The difference between \Z and \z is
3111 that \Z matches before a newline at the end of the string as well as at
3112 the very end, whereas \z matches only at the end.
3113
3114 The \G assertion is true only when the current matching position is at
3115 the start point of the match, as specified by the startoffset argument
3116 of pcre_exec(). It differs from \A when the value of startoffset is
3117 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3118 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3119 mentation where \G can be useful.
3120
3121 Note, however, that PCRE's interpretation of \G, as the start of the
3122 current match, is subtly different from Perl's, which defines it as the
3123 end of the previous match. In Perl, these can be different when the
3124 previously matched string was empty. Because PCRE does just one match
3125 at a time, it cannot reproduce this behaviour.
3126
3127 If all the alternatives of a pattern begin with \G, the expression is
3128 anchored to the starting match position, and the "anchored" flag is set
3129 in the compiled regular expression.
3130
3131
3132 CIRCUMFLEX AND DOLLAR
3133
3134 Outside a character class, in the default matching mode, the circumflex
3135 character is an assertion that is true only if the current matching
3136 point is at the start of the subject string. If the startoffset argu-
3137 ment of pcre_exec() is non-zero, circumflex can never match if the
3138 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3139 has an entirely different meaning (see below).
3140
3141 Circumflex need not be the first character of the pattern if a number
3142 of alternatives are involved, but it should be the first thing in each
3143 alternative in which it appears if the pattern is ever to match that
3144 branch. If all possible alternatives start with a circumflex, that is,
3145 if the pattern is constrained to match only at the start of the sub-
3146 ject, it is said to be an "anchored" pattern. (There are also other
3147 constructs that can cause a pattern to be anchored.)
3148
3149 A dollar character is an assertion that is true only if the current
3150 matching point is at the end of the subject string, or immediately
3151 before a newline at the end of the string (by default). Dollar need not
3152 be the last character of the pattern if a number of alternatives are
3153 involved, but it should be the last item in any branch in which it
3154 appears. Dollar has no special meaning in a character class.
3155
3156 The meaning of dollar can be changed so that it matches only at the
3157 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3158 compile time. This does not affect the \Z assertion.
3159
3160 The meanings of the circumflex and dollar characters are changed if the
3161 PCRE_MULTILINE option is set. When this is the case, a circumflex
3162 matches immediately after internal newlines as well as at the start of
3163 the subject string. It does not match after a newline that ends the
3164 string. A dollar matches before any newlines in the string, as well as
3165 at the very end, when PCRE_MULTILINE is set. When newline is specified
3166 as the two-character sequence CRLF, isolated CR and LF characters do
3167 not indicate newlines.
3168
3169 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3170 (where \n represents a newline) in multiline mode, but not otherwise.
3171 Consequently, patterns that are anchored in single line mode because
3172 all branches start with ^ are not anchored in multiline mode, and a
3173 match for circumflex is possible when the startoffset argument of
3174 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3175 PCRE_MULTILINE is set.
3176
3177 Note that the sequences \A, \Z, and \z can be used to match the start
3178 and end of the subject in both modes, and if all branches of a pattern
3179 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3180 set.
3181
3182
3183 FULL STOP (PERIOD, DOT)
3184
3185 Outside a character class, a dot in the pattern matches any one charac-
3186 ter in the subject string except (by default) a character that signi-
3187 fies the end of a line. In UTF-8 mode, the matched character may be
3188 more than one byte long.
3189
3190 When a line ending is defined as a single character, dot never matches
3191 that character; when the two-character sequence CRLF is used, dot does
3192 not match CR if it is immediately followed by LF, but otherwise it
3193 matches all characters (including isolated CRs and LFs). When any Uni-
3194 code line endings are being recognized, dot does not match CR or LF or
3195 any of the other line ending characters.
3196
3197 The behaviour of dot with regard to newlines can be changed. If the
3198 PCRE_DOTALL option is set, a dot matches any one character, without
3199 exception. If the two-character sequence CRLF is present in the subject
3200 string, it takes two dots to match it.
3201
3202 The handling of dot is entirely independent of the handling of circum-
3203 flex and dollar, the only relationship being that they both involve
3204 newlines. Dot has no special meaning in a character class.
3205
3206
3207 MATCHING A SINGLE BYTE
3208
3209 Outside a character class, the escape sequence \C matches any one byte,
3210 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3211 line-ending characters. The feature is provided in Perl in order to
3212 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3213 acters into individual bytes, what remains in the string may be a mal-
3214 formed UTF-8 string. For this reason, the \C escape sequence is best
3215 avoided.
3216
3217 PCRE does not allow \C to appear in lookbehind assertions (described
3218 below), because in UTF-8 mode this would make it impossible to calcu-
3219 late the length of the lookbehind.
3220
3221
3222 SQUARE BRACKETS AND CHARACTER CLASSES
3223
3224 An opening square bracket introduces a character class, terminated by a
3225 closing square bracket. A closing square bracket on its own is not spe-
3226 cial. If a closing square bracket is required as a member of the class,
3227 it should be the first data character in the class (after an initial
3228 circumflex, if present) or escaped with a backslash.
3229
3230 A character class matches a single character in the subject. In UTF-8
3231 mode, the character may occupy more than one byte. A matched character
3232 must be in the set of characters defined by the class, unless the first
3233 character in the class definition is a circumflex, in which case the
3234 subject character must not be in the set defined by the class. If a
3235 circumflex is actually required as a member of the class, ensure it is
3236 not the first character, or escape it with a backslash.
3237
3238 For example, the character class [aeiou] matches any lower case vowel,
3239 while [^aeiou] matches any character that is not a lower case vowel.
3240 Note that a circumflex is just a convenient notation for specifying the
3241 characters that are in the class by enumerating those that are not. A
3242 class that starts with a circumflex is not an assertion: it still con-
3243 sumes a character from the subject string, and therefore it fails if
3244 the current pointer is at the end of the string.
3245
3246 In UTF-8 mode, characters with values greater than 255 can be included
3247 in a class as a literal string of bytes, or by using the \x{ escaping
3248 mechanism.
3249
3250 When caseless matching is set, any letters in a class represent both
3251 their upper case and lower case versions, so for example, a caseless
3252 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3253 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3254 understands the concept of case for characters whose values are less
3255 than 128, so caseless matching is always possible. For characters with
3256 higher values, the concept of case is supported if PCRE is compiled
3257 with Unicode property support, but not otherwise. If you want to use
3258 caseless matching for characters 128 and above, you must ensure that
3259 PCRE is compiled with Unicode property support as well as with UTF-8
3260 support.
3261
3262 Characters that might indicate line breaks are never treated in any
3263 special way when matching character classes, whatever line-ending
3264 sequence is in use, and whatever setting of the PCRE_DOTALL and
3265 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3266 of these characters.
3267
3268 The minus (hyphen) character can be used to specify a range of charac-
3269 ters in a character class. For example, [d-m] matches any letter
3270 between d and m, inclusive. If a minus character is required in a
3271 class, it must be escaped with a backslash or appear in a position
3272 where it cannot be interpreted as indicating a range, typically as the
3273 first or last character in the class.
3274
3275 It is not possible to have the literal character "]" as the end charac-
3276 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3277 two characters ("W" and "-") followed by a literal string "46]", so it
3278 would match "W46]" or "-46]". However, if the "]" is escaped with a
3279 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3280 preted as a class containing a range followed by two other characters.
3281 The octal or hexadecimal representation of "]" can also be used to end
3282 a range.
3283
3284 Ranges operate in the collating sequence of character values. They can
3285 also be used for characters specified numerically, for example
3286 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3287 are greater than 255, for example [\x{100}-\x{2ff}].
3288
3289 If a range that includes letters is used when caseless matching is set,
3290 it matches the letters in either case. For example, [W-c] is equivalent
3291 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3292 character tables for the "fr_FR" locale are in use, [\xc8-\xcb] matches
3293 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3294 concept of case for characters with values greater than 128 only when
3295 it is compiled with Unicode property support.
3296
3297 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3298 in a character class, and add the characters that they match to the
3299 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3300 flex can conveniently be used with the upper case character types to
3301 specify a more restricted set of characters than the matching lower
3302 case type. For example, the class [^\W_] matches any letter or digit,
3303 but not underscore.
3304
3305 The only metacharacters that are recognized in character classes are
3306 backslash, hyphen (only where it can be interpreted as specifying a
3307 range), circumflex (only at the start), opening square bracket (only
3308 when it can be interpreted as introducing a POSIX class name - see the
3309 next section), and the terminating closing square bracket. However,
3310 escaping other non-alphanumeric characters does no harm.
3311
3312
3313 POSIX CHARACTER CLASSES
3314
3315 Perl supports the POSIX notation for character classes. This uses names
3316 enclosed by [: and :] within the enclosing square brackets. PCRE also
3317 supports this notation. For example,
3318
3319 [01[:alpha:]%]
3320
3321 matches "0", "1", any alphabetic character, or "%". The supported class
3322 names are
3323
3324 alnum letters and digits
3325 alpha letters
3326 ascii character codes 0 - 127
3327 blank space or tab only
3328 cntrl control characters
3329 digit decimal digits (same as \d)
3330 graph printing characters, excluding space
3331 lower lower case letters
3332 print printing characters, including space
3333 punct printing characters, excluding letters and digits
3334 space white space (not quite the same as \s)
3335 upper upper case letters
3336 word "word" characters (same as \w)
3337 xdigit hexadecimal digits
3338
3339 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3340 and space (32). Notice that this list includes the VT character (code
3341 11). This makes "space" different to \s, which does not include VT (for
3342 Perl compatibility).
3343
3344 The name "word" is a Perl extension, and "blank" is a GNU extension
3345 from Perl 5.8. Another Perl extension is negation, which is indicated
3346 by a ^ character after the colon. For example,
3347
3348 [12[:^digit:]]
3349
3350 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3351 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3352 these are not supported, and an error is given if they are encountered.
3353
3354 In UTF-8 mode, characters with values greater than 128 do not match any
3355 of the POSIX character classes.
3356
3357
3358 VERTICAL BAR
3359
3360 Vertical bar characters are used to separate alternative patterns. For
3361 example, the pattern
3362
3363 gilbert|sullivan
3364
3365 matches either "gilbert" or "sullivan". Any number of alternatives may
3366 appear, and an empty alternative is permitted (matching the empty
3367 string). The matching process tries each alternative in turn, from left
3368 to right, and the first one that succeeds is used. If the alternatives
3369 are within a subpattern (defined below), "succeeds" means matching the
3370 rest of the main pattern as well as the alternative in the subpattern.
3371
3372
3373 INTERNAL OPTION SETTING
3374
3375 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3376 PCRE_EXTENDED options can be changed from within the pattern by a
3377 sequence of Perl option letters enclosed between "(?" and ")". The
3378 option letters are
3379
3380 i for PCRE_CASELESS
3381 m for PCRE_MULTILINE
3382 s for PCRE_DOTALL
3383 x for PCRE_EXTENDED
3384
3385 For example, (?im) sets caseless, multiline matching. It is also possi-
3386 ble to unset these options by preceding the letter with a hyphen, and a
3387 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3388 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3389 is also permitted. If a letter appears both before and after the
3390 hyphen, the option is unset.
3391
3392 When an option change occurs at top level (that is, not inside subpat-
3393 tern parentheses), the change applies to the remainder of the pattern
3394 that follows. If the change is placed right at the start of a pattern,
3395 PCRE extracts it into the global options (and it will therefore show up
3396 in data extracted by the pcre_fullinfo() function).
3397
3398 An option change within a subpattern (see below for a description of
3399 subpatterns) affects only that part of the current pattern that follows
3400 it, so
3401
3402 (a(?i)b)c
3403
3404 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3405 used). By this means, options can be made to have different settings
3406 in different parts of the pattern. Any changes made in one alternative
3407 do carry on into subsequent branches within the same subpattern. For
3408 example,
3409
3410 (a(?i)b|c)
3411
3412 matches "ab", "aB", "c", and "C", even though when matching "C" the
3413 first branch is abandoned before the option setting. This is because
3414 the effects of option settings happen at compile time. There would be
3415 some very weird behaviour otherwise.
3416
3417 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3418 can be changed in the same way as the Perl-compatible options by using
3419 the characters J, U and X respectively.
3420
3421
3422 SUBPATTERNS
3423
3424 Subpatterns are delimited by parentheses (round brackets), which can be
3425 nested. Turning part of a pattern into a subpattern does two things:
3426
3427 1. It localizes a set of alternatives. For example, the pattern
3428
3429 cat(aract|erpillar|)
3430
3431 matches one of the words "cat", "cataract", or "caterpillar". Without
3432 the parentheses, it would match "cataract", "erpillar" or an empty
3433 string.
3434
3435 2. It sets up the subpattern as a capturing subpattern. This means
3436 that, when the whole pattern matches, that portion of the subject
3437 string that matched the subpattern is passed back to the caller via the
3438 ovector argument of pcre_exec(). Opening parentheses are counted from
3439 left to right (starting from 1) to obtain numbers for the capturing
3440 subpatterns.
3441
3442 For example, if the string "the red king" is matched against the pat-
3443 tern
3444
3445 the ((red|white) (king|queen))
3446
3447 the captured substrings are "red king", "red", and "king", and are num-
3448 bered 1, 2, and 3, respectively.
3449
3450 The fact that plain parentheses fulfil two functions is not always
3451 helpful. There are often times when a grouping subpattern is required
3452 without a capturing requirement. If an opening parenthesis is followed
3453 by a question mark and a colon, the subpattern does not do any captur-
3454 ing, and is not counted when computing the number of any subsequent
3455 capturing subpatterns. For example, if the string "the white queen" is
3456 matched against the pattern
3457
3458 the ((?:red|white) (king|queen))
3459
3460 the captured substrings are "white queen" and "queen", and are numbered
3461 1 and 2. The maximum number of capturing subpatterns is 65535.
3462
3463 As a convenient shorthand, if any option settings are required at the
3464 start of a non-capturing subpattern, the option letters may appear
3465 between the "?" and the ":". Thus the two patterns
3466
3467 (?i:saturday|sunday)
3468 (?:(?i)saturday|sunday)
3469
3470 match exactly the same set of strings. Because alternative branches are
3471 tried from left to right, and options are not reset until the end of
3472 the subpattern is reached, an option setting in one branch does affect
3473 subsequent branches, so the above patterns match "SUNDAY" as well as
3474 "Saturday".
3475
3476
3477 NAMED SUBPATTERNS
3478
3479 Identifying capturing parentheses by number is simple, but it can be
3480 very hard to keep track of the numbers in complicated regular expres-
3481 sions. Furthermore, if an expression is modified, the numbers may
3482 change. To help with this difficulty, PCRE supports the naming of sub-
3483 patterns. This feature was not added to Perl until release 5.10. Python
3484 had the feature earlier, and PCRE introduced it at release 4.0, using
3485 the Python syntax. PCRE now supports both the Perl and the Python syn-
3486 tax.
3487
3488 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3489 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3490 to capturing parentheses from other parts of the pattern, such as back-
3491 references, recursion, and conditions, can be made by name as well as
3492 by number.
3493
3494 Names consist of up to 32 alphanumeric characters and underscores.
3495 Named capturing parentheses are still allocated numbers as well as
3496 names, exactly as if the names were not present. The PCRE API provides
3497 function calls for extracting the name-to-number translation table from
3498 a compiled pattern. There is also a convenience function for extracting
3499 a captured substring by name.
3500
3501 By default, a name must be unique within a pattern, but it is possible
3502 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3503 time. This can be useful for patterns where only one instance of the
3504 named parentheses can match. Suppose you want to match the name of a
3505 weekday, either as a 3-letter abbreviation or as the full name, and in
3506 both cases you want to extract the abbreviation. This pattern (ignoring
3507 the line breaks) does the job:
3508
3509 (?<DN>Mon|Fri|Sun)(?:day)?|
3510 (?<DN>Tue)(?:sday)?|
3511 (?<DN>Wed)(?:nesday)?|
3512 (?<DN>Thu)(?:rsday)?|
3513 (?<DN>Sat)(?:urday)?
3514
3515 There are five capturing substrings, but only one is ever set after a
3516 match. The convenience function for extracting the data by name
3517 returns the substring for the first (and in this example, the only)
3518 subpattern of that name that matched. This saves searching to find
3519 which numbered subpattern it was. If you make a reference to a non-
3520 unique named subpattern from elsewhere in the pattern, the one that
3521 corresponds to the lowest number is used. For further details of the
3522 interfaces for handling named subpatterns, see the pcreapi documenta-
3523 tion.
3524
3525
3526 REPETITION
3527
3528 Repetition is specified by quantifiers, which can follow any of the
3529 following items:
3530
3531 a literal data character
3532 the dot metacharacter
3533 the \C escape sequence
3534 the \X escape sequence (in UTF-8 mode with Unicode properties)
3535 the \R escape sequence
3536 an escape such as \d that matches a single character
3537 a character class
3538 a back reference (see next section)
3539 a parenthesized subpattern (unless it is an assertion)
3540
3541 The general repetition quantifier specifies a minimum and maximum num-
3542 ber of permitted matches, by giving the two numbers in curly brackets
3543 (braces), separated by a comma. The numbers must be less than 65536,
3544 and the first must be less than or equal to the second. For example:
3545
3546 z{2,4}
3547
3548 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3549 special character. If the second number is omitted, but the comma is
3550 present, there is no upper limit; if the second number and the comma
3551 are both omitted, the quantifier specifies an exact number of required
3552 matches. Thus
3553
3554 [aeiou]{3,}
3555
3556 matches at least 3 successive vowels, but may match many more, while
3557
3558 \d{8}
3559
3560 matches exactly 8 digits. An opening curly bracket that appears in a
3561 position where a quantifier is not allowed, or one that does not match
3562 the syntax of a quantifier, is taken as a literal character. For exam-
3563 ple, {,6} is not a quantifier, but a literal string of four characters.
3564
3565 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3566 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3567 acters, each of which is represented by a two-byte sequence. Similarly,
3568 when Unicode property support is available, \X{3} matches three Unicode
3569 extended sequences, each of which may be several bytes long (and they
3570 may be of different lengths).
3571
3572 The quantifier {0} is permitted, causing the expression to behave as if
3573 the previous item and the quantifier were not present.
3574
3575 For convenience, the three most common quantifiers have single-charac-
3576 ter abbreviations:
3577
3578 * is equivalent to {0,}
3579 + is equivalent to {1,}
3580 ? is equivalent to {0,1}
3581
3582 It is possible to construct infinite loops by following a subpattern
3583 that can match no characters with a quantifier that has no upper limit,
3584 for example:
3585
3586 (a?)*
3587
3588 Earlier versions of Perl and PCRE used to give an error at compile time
3589 for such patterns. However, because there are cases where this can be
3590 useful, such patterns are now accepted, but if any repetition of the
3591 subpattern does in fact match no characters, the loop is forcibly bro-
3592 ken.
3593
3594 By default, the quantifiers are "greedy", that is, they match as much
3595 as possible (up to the maximum number of permitted times), without
3596 causing the rest of the pattern to fail. The classic example of where
3597 this gives problems is in trying to match comments in C programs. These
3598 appear between /* and */ and within the comment, individual * and /
3599 characters may appear. An attempt to match C comments by applying the
3600 pattern
3601
3602 /\*.*\*/
3603
3604 to the string
3605
3606 /* first comment */ not comment /* second comment */
3607
3608 fails, because it matches the entire string owing to the greediness of
3609 the .* item.
3610
3611 However, if a quantifier is followed by a question mark, it ceases to
3612 be greedy, and instead matches the minimum number of times possible, so
3613 the pattern
3614
3615 /\*.*?\*/
3616
3617 does the right thing with the C comments. The meaning of the various
3618 quantifiers is not otherwise changed, just the preferred number of
3619 matches. Do not confuse this use of question mark with its use as a
3620 quantifier in its own right. Because it has two uses, it can sometimes
3621 appear doubled, as in
3622
3623 \d??\d
3624
3625 which matches one digit by preference, but can match two if that is the
3626 only way the rest of the pattern matches.
3627
3628 If the PCRE_UNGREEDY option is set (an option that is not available in
3629 Perl), the quantifiers are not greedy by default, but individual ones
3630 can be made greedy by following them with a question mark. In other
3631 words, it inverts the default behaviour.
3632
3633 When a parenthesized subpattern is quantified with a minimum repeat
3634 count that is greater than 1 or with a limited maximum, more memory is
3635 required for the compiled pattern, in proportion to the size of the
3636 minimum or maximum.
3637
3638 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3639 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3640 the pattern is implicitly anchored, because whatever follows will be
3641 tried against every character position in the subject string, so there
3642 is no point in retrying the overall match at any position after the
3643 first. PCRE normally treats such a pattern as though it were preceded
3644 by \A.
3645
3646 In cases where it is known that the subject string contains no new-
3647 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3648 mization, or alternatively using ^ to indicate anchoring explicitly.
3649
3650 However, there is one situation where the optimization cannot be used.
3651 When .* is inside capturing parentheses that are the subject of a
3652 backreference elsewhere in the pattern, a match at the start may fail
3653 where a later one succeeds. Consider, for example:
3654
3655 (.*)abc\1
3656
3657 If the subject is "xyz123abc123" the match point is the fourth charac-
3658 ter. For this reason, such a pattern is not implicitly anchored.
3659
3660 When a capturing subpattern is repeated, the value captured is the sub-
3661 string that matched the final iteration. For example, after
3662
3663 (tweedle[dume]{3}\s*)+
3664
3665 has matched "tweedledum tweedledee" the value of the captured substring
3666 is "tweedledee". However, if there are nested capturing subpatterns,
3667 the corresponding captured values may have been set in previous itera-
3668 tions. For example, after
3669
3670 /(a|(b))+/
3671
3672 matches "aba" the value of the second captured substring is "b".
3673
3674
3675 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3676
3677 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3678 repetition, failure of what follows normally causes the repeated item
3679 to be re-evaluated to see if a different number of repeats allows the
3680 rest of the pattern to match. Sometimes it is useful to prevent this,
3681 either to change the nature of the match, or to cause it fail earlier
3682 than it otherwise might, when the author of the pattern knows there is
3683 no point in carrying on.
3684
3685 Consider, for example, the pattern \d+foo when applied to the subject
3686 line
3687
3688 123456bar
3689
3690 After matching all 6 digits and then failing to match "foo", the normal
3691 action of the matcher is to try again with only 5 digits matching the
3692 \d+ item, and then with 4, and so on, before ultimately failing.
3693 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3694 the means for specifying that once a subpattern has matched, it is not
3695 to be re-evaluated in this way.
3696
3697 If we use atomic grouping for the previous example, the matcher gives
3698 up immediately on failing to match "foo" the first time. The notation
3699 is a kind of special parenthesis, starting with (?> as in this example:
3700
3701 (?>\d+)foo
3702
3703 This kind of parenthesis "locks up" the part of the pattern it con-
3704 tains once it has matched, and a failure further into the pattern is
3705 prevented from backtracking into it. Backtracking past it to previous
3706 items, however, works as normal.
3707
3708 An alternative description is that a subpattern of this type matches
3709 the string of characters that an identical standalone pattern would
3710 match, if anchored at the current point in the subject string.
3711
3712 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3713 such as the above example can be thought of as a maximizing repeat that
3714 must swallow everything it can. So, while both \d+ and \d+? are pre-
3715 pared to adjust the number of digits they match in order to make the
3716 rest of the pattern match, (?>\d+) can only match an entire sequence of
3717 digits.
3718
3719 Atomic groups in general can of course contain arbitrarily complicated
3720 subpatterns, and can be nested. However, when the subpattern for an
3721 atomic group is just a single repeated item, as in the example above, a
3722 simpler notation, called a "possessive quantifier" can be used. This
3723 consists of an additional + character following a quantifier. Using
3724 this notation, the previous example can be rewritten as
3725
3726 \d++foo
3727
3728 Possessive quantifiers are always greedy; the setting of the
3729 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3730 simpler forms of atomic group. However, there is no difference in the
3731 meaning of a possessive quantifier and the equivalent atomic group,
3732 though there may be a performance difference; possessive quantifiers
3733 should be slightly faster.
3734
3735 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3736 tax. Jeffrey Friedl originated the idea (and the name) in the first
3737 edition of his book. Mike McCloskey liked it, so implemented it when he
3738 built Sun's Java package, and PCRE copied it from there. It ultimately
3739 found its way into Perl at release 5.10.
3740
3741 PCRE has an optimization that automatically "possessifies" certain sim-
3742 ple pattern constructs. For example, the sequence A+B is treated as
3743 A++B because there is no point in backtracking into a sequence of A's
3744 when B must follow.
3745
3746 When a pattern contains an unlimited repeat inside a subpattern that
3747 can itself be repeated an unlimited number of times, the use of an
3748 atomic group is the only way to avoid some failing matches taking a
3749 very long time indeed. The pattern
3750
3751 (\D+|<\d+>)*[!?]
3752
3753 matches an unlimited number of substrings that either consist of non-
3754 digits, or digits enclosed in <>, followed by either ! or ?. When it
3755 matches, it runs quickly. However, if it is applied to
3756
3757 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3758
3759 it takes a long time before reporting failure. This is because the
3760 string can be divided between the internal \D+ repeat and the external
3761 * repeat in a large number of ways, and all have to be tried. (The
3762 example uses [!?] rather than a single character at the end, because
3763 both PCRE and Perl have an optimization that allows for fast failure
3764 when a single character is used. They remember the last single charac-
3765 ter that is required for a match, and fail early if it is not present
3766 in the string.) If the pattern is changed so that it uses an atomic
3767 group, like this:
3768
3769 ((?>\D+)|<\d+>)*[!?]
3770
3771 sequences of non-digits cannot be broken, and failure happens quickly.
3772
3773
3774 BACK REFERENCES
3775
3776 Outside a character class, a backslash followed by a digit greater than
3777 0 (and possibly further digits) is a back reference to a capturing sub-
3778 pattern earlier (that is, to its left) in the pattern, provided there
3779 have been that many previous capturing left parentheses.
3780
3781 However, if the decimal number following the backslash is less than 10,
3782 it is always taken as a back reference, and causes an error only if
3783 there are not that many capturing left parentheses in the entire pat-
3784 tern. In other words, the parentheses that are referenced need not be
3785 to the left of the reference for numbers less than 10. A "forward back
3786 reference" of this type can make sense when a repetition is involved
3787 and the subpattern to the right has participated in an earlier itera-
3788 tion.
3789
3790 It is not possible to have a numerical "forward back reference" to a
3791 subpattern whose number is 10 or more using this syntax because a
3792 sequence such as \50 is interpreted as a character defined in octal.
3793 See the subsection entitled "Non-printing characters" above for further
3794 details of the handling of digits following a backslash. There is no
3795 such problem when named parentheses are used. A back reference to any
3796 subpattern is possible using named parentheses (see below).
3797
3798 Another way of avoiding the ambiguity inherent in the use of digits
3799 following a backslash is to use the \g escape sequence, which is a fea-
3800 ture introduced in Perl 5.10. This escape must be followed by a posi-
3801 tive or a negative number, optionally enclosed in braces. These exam-
3802 ples are all identical:
3803
3804 (ring), \1
3805 (ring), \g1
3806 (ring), \g{1}
3807
3808 A positive number specifies an absolute reference without the ambiguity
3809 that is present in the older syntax. It is also useful when literal
3810 digits follow the reference. A negative number is a relative reference.
3811 Consider this example:
3812
3813 (abc(def)ghi)\g{-1}
3814
3815 The sequence \g{-1} is a reference to the most recently started captur-
3816 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
3817 \g{-2} would be equivalent to \1. The use of relative references can be
3818 helpful in long patterns, and also in patterns that are created by
3819 joining together fragments that contain references within themselves.
3820
3821 A back reference matches whatever actually matched the capturing sub-
3822 pattern in the current subject string, rather than anything matching
3823 the subpattern itself (see "Subpatterns as subroutines" below for a way
3824 of doing that). So the pattern
3825
3826 (sens|respons)e and \1ibility
3827
3828 matches "sense and sensibility" and "response and responsibility", but
3829 not "sense and responsibility". If caseful matching is in force at the
3830 time of the back reference, the case of letters is relevant. For exam-
3831 ple,
3832
3833 ((?i)rah)\s+\1
3834
3835 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3836 original capturing subpattern is matched caselessly.
3837
3838 Back references to named subpatterns use the Perl syntax \k<name> or
3839 \k'name' or the Python syntax (?P=name). We could rewrite the above
3840 example in either of the following ways:
3841
3842 (?<p1>(?i)rah)\s+\k<p1>
3843 (?P<p1>(?i)rah)\s+(?P=p1)
3844
3845 A subpattern that is referenced by name may appear in the pattern
3846 before or after the reference.
3847
3848 There may be more than one back reference to the same subpattern. If a
3849 subpattern has not actually been used in a particular match, any back
3850 references to it always fail. For example, the pattern
3851
3852 (a|(bc))\2
3853
3854 always fails if it starts to match "a" rather than "bc". Because there
3855 may be many capturing parentheses in a pattern, all digits following
3856 the backslash are taken as part of a potential back reference number.
3857 If the pattern continues with a digit character, some delimiter must be
3858 used to terminate the back reference. If the PCRE_EXTENDED option is
3859 set, this can be whitespace. Otherwise an empty comment (see "Com-
3860 ments" below) can be used.
3861
3862 A back reference that occurs inside the parentheses to which it refers
3863 fails when the subpattern is first used, so, for example, (a\1) never
3864 matches. However, such references can be useful inside repeated sub-
3865 patterns. For example, the pattern
3866
3867 (a|b\1)+
3868
3869 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3870 ation of the subpattern, the back reference matches the character
3871 string corresponding to the previous iteration. In order for this to
3872 work, the pattern must be such that the first iteration does not need
3873 to match the back reference. This can be done using alternation, as in
3874 the example above, or by a quantifier with a minimum of zero.
3875
3876
3877 ASSERTIONS
3878
3879 An assertion is a test on the characters following or preceding the
3880 current matching point that does not actually consume any characters.
3881 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3882 described above.
3883
3884 More complicated assertions are coded as subpatterns. There are two
3885 kinds: those that look ahead of the current position in the subject
3886 string, and those that look behind it. An assertion subpattern is
3887 matched in the normal way, except that it does not cause the current
3888 matching position to be changed.
3889
3890 Assertion subpatterns are not capturing subpatterns, and may not be
3891 repeated, because it makes no sense to assert the same thing several
3892 times. If any kind of assertion contains capturing subpatterns within
3893 it, these are counted for the purposes of numbering the capturing sub-
3894 patterns in the whole pattern. However, substring capturing is carried
3895 out only for positive assertions, because it does not make sense for
3896 negative assertions.
3897
3898 Lookahead assertions
3899
3900 Lookahead assertions start with (?= for positive assertions and (?! for
3901 negative assertions. For example,
3902
3903 \w+(?=;)
3904
3905 matches a word followed by a semicolon, but does not include the semi-
3906 colon in the match, and
3907
3908 foo(?!bar)
3909
3910 matches any occurrence of "foo" that is not followed by "bar". Note
3911 that the apparently similar pattern
3912
3913 (?!foo)bar
3914
3915 does not find an occurrence of "bar" that is preceded by something
3916 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3917 the assertion (?!foo) is always true when the next three characters are
3918 "bar". A lookbehind assertion is needed to achieve the other effect.
3919
3920 If you want to force a matching failure at some point in a pattern, the
3921 most convenient way to do it is with (?!) because an empty string
3922 always matches, so an assertion that requires there not to be an empty
3923 string must always fail.
3924
3925 Lookbehind assertions
3926
3927 Lookbehind assertions start with (?<= for positive assertions and (?<!
3928 for negative assertions. For example,
3929
3930 (?<!foo)bar
3931
3932 does find an occurrence of "bar" that is not preceded by "foo". The
3933 contents of a lookbehind assertion are restricted such that all the
3934 strings it matches must have a fixed length. However, if there are sev-
3935 eral top-level alternatives, they do not all have to have the same
3936 fixed length. Thus
3937
3938 (?<=bullock|donkey)
3939
3940 is permitted, but
3941
3942 (?<!dogs?|cats?)
3943
3944 causes an error at compile time. Branches that match different length
3945 strings are permitted only at the top level of a lookbehind assertion.
3946 This is an extension compared with Perl (at least for 5.8), which
3947 requires all branches to match the same length of string. An assertion
3948 such as
3949
3950 (?<=ab(c|de))
3951
3952 is not permitted, because its single top-level branch can match two
3953 different lengths, but it is acceptable if rewritten to use two top-
3954 level branches:
3955
3956 (?<=abc|abde)
3957
3958 The implementation of lookbehind assertions is, for each alternative,
3959 to temporarily move the current position back by the fixed length and
3960 then try to match. If there are insufficient characters before the cur-
3961 rent position, the assertion fails.
3962
3963 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3964 mode) to appear in lookbehind assertions, because it makes it impossi-
3965 ble to calculate the length of the lookbehind. The \X and \R escapes,
3966 which can match different numbers of bytes, are also not permitted.
3967
3968 Possessive quantifiers can be used in conjunction with lookbehind
3969 assertions to specify efficient matching at the end of the subject
3970 string. Consider a simple pattern such as
3971
3972 abcd$
3973
3974 when applied to a long string that does not match. Because matching
3975 proceeds from left to right, PCRE will look for each "a" in the subject
3976 and then see if what follows matches the rest of the pattern. If the
3977 pattern is specified as
3978
3979 ^.*abcd$
3980
3981 the initial .* matches the entire string at first, but when this fails
3982 (because there is no following "a"), it backtracks to match all but the
3983 last character, then all but the last two characters, and so on. Once
3984 again the search for "a" covers the entire string, from right to left,
3985 so we are no better off. However, if the pattern is written as
3986
3987 ^.*+(?<=abcd)
3988
3989 there can be no backtracking for the .*+ item; it can match only the
3990 entire string. The subsequent lookbehind assertion does a single test
3991 on the last four characters. If it fails, the match fails immediately.
3992 For long strings, this approach makes a significant difference to the
3993 processing time.
3994
3995 Using multiple assertions
3996
3997 Several assertions (of any sort) may occur in succession. For example,
3998
3999 (?<=\d{3})(?<!999)foo
4000
4001 matches "foo" preceded by three digits that are not "999". Notice that
4002 each of the assertions is applied independently at the same point in
4003 the subject string. First there is a check that the previous three
4004 characters are all digits, and then there is a check that the same
4005 three characters are not "999". This pattern does not match "foo" pre-
4006 ceded by six characters, the first of which are digits and the last
4007 three of which are not "999". For example, it doesn't match "123abc-
4008 foo". A pattern to do that is
4009
4010 (?<=\d{3}...)(?<!999)foo
4011
4012 This time the first assertion looks at the preceding six characters,
4013 checking that the first three are digits, and then the second assertion
4014 checks that the preceding three characters are not "999".
4015
4016 Assertions can be nested in any combination. For example,
4017
4018 (?<=(?<!foo)bar)baz
4019
4020 matches an occurrence of "baz" that is preceded by "bar" which in turn
4021 is not preceded by "foo", while
4022
4023 (?<=\d{3}(?!999)...)foo
4024
4025 is another pattern that matches "foo" preceded by three digits and any
4026 three characters that are not "999".
4027
4028
4029 CONDITIONAL SUBPATTERNS
4030
4031 It is possible to cause the matching process to obey a subpattern con-
4032 ditionally or to choose between two alternative subpatterns, depending
4033 on the result of an assertion, or whether a previous capturing subpat-
4034 tern matched or not. The two possible forms of conditional subpattern
4035 are
4036
4037 (?(condition)yes-pattern)
4038 (?(condition)yes-pattern|no-pattern)
4039
4040 If the condition is satisfied, the yes-pattern is used; otherwise the
4041 no-pattern (if present) is used. If there are more than two alterna-
4042 tives in the subpattern, a compile-time error occurs.
4043
4044 There are four kinds of condition: references to subpatterns, refer-
4045 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4046
4047 Checking for a used subpattern by number
4048
4049 If the text between the parentheses consists of a sequence of digits,
4050 the condition is true if the capturing subpattern of that number has
4051 previously matched.
4052
4053 Consider the following pattern, which contains non-significant white
4054 space to make it more readable (assume the PCRE_EXTENDED option) and to
4055 divide it into three parts for ease of discussion:
4056
4057 ( \( )? [^()]+ (?(1) \) )
4058
4059 The first part matches an optional opening parenthesis, and if that
4060 character is present, sets it as the first captured substring. The sec-
4061 ond part matches one or more characters that are not parentheses. The
4062 third part is a conditional subpattern that tests whether the first set
4063 of parentheses matched or not. If they did, that is, if subject started
4064 with an opening parenthesis, the condition is true, and so the yes-pat-
4065 tern is executed and a closing parenthesis is required. Otherwise,
4066 since no-pattern is not present, the subpattern matches nothing. In
4067 other words, this pattern matches a sequence of non-parentheses,
4068 optionally enclosed in parentheses.
4069
4070 Checking for a used subpattern by name
4071
4072 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4073 used subpattern by name. For compatibility with earlier versions of
4074 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4075 also recognized. However, there is a possible ambiguity with this syn-
4076 tax, because subpattern names may consist entirely of digits. PCRE
4077 looks first for a named subpattern; if it cannot find one and the name
4078 consists entirely of digits, PCRE looks for a subpattern of that num-
4079 ber, which must be greater than zero. Using subpattern names that con-
4080 sist entirely of digits is not recommended.
4081
4082 Rewriting the above example to use a named subpattern gives this:
4083
4084 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4085
4086
4087 Checking for pattern recursion
4088
4089 If the condition is the string (R), and there is no subpattern with the
4090 name R, the condition is true if a recursive call to the whole pattern
4091 or any subpattern has been made. If digits or a name preceded by amper-
4092 sand follow the letter R, for example:
4093
4094 (?(R3)...) or (?(R&name)...)
4095
4096 the condition is true if the most recent recursion is into the subpat-
4097 tern whose number or name is given. This condition does not check the
4098 entire recursion stack.
4099
4100 At "top level", all these recursion test conditions are false. Recur-
4101 sive patterns are described below.
4102
4103 Defining subpatterns for use by reference only
4104
4105 If the condition is the string (DEFINE), and there is no subpattern
4106 with the name DEFINE, the condition is always false. In this case,
4107 there may be only one alternative in the subpattern. It is always
4108 skipped if control reaches this point in the pattern; the idea of
4109 DEFINE is that it can be used to define "subroutines" that can be ref-
4110 erenced from elsewhere. (The use of "subroutines" is described below.)
4111 For example, a pattern to match an IPv4 address could be written like
4112 this (ignore whitespace and line breaks):
4113
4114 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4115 \b (?&byte) (\.(?&byte)){3} \b
4116
4117 The first part of the pattern is a DEFINE group inside which a another
4118 group named "byte" is defined. This matches an individual component of
4119 an IPv4 address (a number less than 256). When matching takes place,
4120 this part of the pattern is skipped because DEFINE acts like a false
4121 condition.
4122
4123 The rest of the pattern uses references to the named group to match the
4124 four dot-separated components of an IPv4 address, insisting on a word
4125 boundary at each end.
4126
4127 Assertion conditions
4128
4129 If the condition is not in any of the above formats, it must be an
4130 assertion. This may be a positive or negative lookahead or lookbehind
4131 assertion. Consider this pattern, again containing non-significant
4132 white space, and with the two alternatives on the second line:
4133
4134 (?(?=[^a-z]*[a-z])
4135 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4136
4137 The condition is a positive lookahead assertion that matches an
4138 optional sequence of non-letters followed by a letter. In other words,
4139 it tests for the presence of at least one letter in the subject. If a
4140 letter is found, the subject is matched against the first alternative;
4141 otherwise it is matched against the second. This pattern matches
4142 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4143 letters and dd are digits.
4144
4145
4146 COMMENTS
4147
4148 The sequence (?# marks the start of a comment that continues up to the
4149 next closing parenthesis. Nested parentheses are not permitted. The
4150 characters that make up a comment play no part in the pattern matching
4151 at all.
4152
4153 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4154 character class introduces a comment that continues to immediately
4155 after the next newline in the pattern.
4156
4157
4158 RECURSIVE PATTERNS
4159
4160 Consider the problem of matching a string in parentheses, allowing for
4161 unlimited nested parentheses. Without the use of recursion, the best
4162 that can be done is to use a pattern that matches up to some fixed
4163 depth of nesting. It is not possible to handle an arbitrary nesting
4164 depth.
4165
4166 For some time, Perl has provided a facility that allows regular expres-
4167 sions to recurse (amongst other things). It does this by interpolating
4168 Perl code in the expression at run time, and the code can refer to the
4169 expression itself. A Perl pattern using code interpolation to solve the
4170 parentheses problem can be created like this:
4171
4172 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4173
4174 The (?p{...}) item interpolates Perl code at run time, and in this case
4175 refers recursively to the pattern in which it appears.
4176
4177 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4178 it supports special syntax for recursion of the entire pattern, and
4179 also for individual subpattern recursion. After its introduction in
4180 PCRE and Python, this kind of recursion was introduced into Perl at
4181 release 5.10.
4182
4183 A special item that consists of (? followed by a number greater than
4184 zero and a closing parenthesis is a recursive call of the subpattern of
4185 the given number, provided that it occurs inside that subpattern. (If
4186 not, it is a "subroutine" call, which is described in the next sec-
4187 tion.) The special item (?R) or (?0) is a recursive call of the entire
4188 regular expression.
4189
4190 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4191 always treated as an atomic group. That is, once it has matched some of
4192 the subject string, it is never re-entered, even if it contains untried
4193 alternatives and there is a subsequent matching failure.
4194
4195 This PCRE pattern solves the nested parentheses problem (assume the
4196 PCRE_EXTENDED option is set so that white space is ignored):
4197
4198 \( ( (?>[^()]+) | (?R) )* \)
4199
4200 First it matches an opening parenthesis. Then it matches any number of
4201 substrings which can either be a sequence of non-parentheses, or a
4202 recursive match of the pattern itself (that is, a correctly parenthe-
4203 sized substring). Finally there is a closing parenthesis.
4204
4205 If this were part of a larger pattern, you would not want to recurse
4206 the entire pattern, so instead you could use this:
4207
4208 ( \( ( (?>[^()]+) | (?1) )* \) )
4209
4210 We have put the pattern into parentheses, and caused the recursion to
4211 refer to them instead of the whole pattern. In a larger pattern, keep-
4212 ing track of parenthesis numbers can be tricky. It may be more conve-
4213 nient to use named parentheses instead. The Perl syntax for this is
4214 (?&name); PCRE's earlier syntax (?P>name) is also supported. We could
4215 rewrite the above example as follows:
4216
4217 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4218
4219 If there is more than one subpattern with the same name, the earliest
4220 one is used. This particular example pattern contains nested unlimited
4221 repeats, and so the use of atomic grouping for matching strings of non-
4222 parentheses is important when applying the pattern to strings that do
4223 not match. For example, when this pattern is applied to
4224
4225 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4226
4227 it yields "no match" quickly. However, if atomic grouping is not used,
4228 the match runs for a very long time indeed because there are so many
4229 different ways the + and * repeats can carve up the subject, and all
4230 have to be tested before failure can be reported.
4231
4232 At the end of a match, the values set for any capturing subpatterns are
4233 those from the outermost level of the recursion at which the subpattern
4234 value is set. If you want to obtain intermediate values, a callout
4235 function can be used (see below and the pcrecallout documentation). If
4236 the pattern above is matched against
4237
4238 (ab(cd)ef)
4239
4240 the value for the capturing parentheses is "ef", which is the last
4241 value taken on at the top level. If additional parentheses are added,
4242 giving
4243
4244 \( ( ( (?>[^()]+) | (?R) )* ) \)
4245 ^ ^
4246 ^ ^
4247
4248 the string they capture is "ab(cd)ef", the contents of the top level
4249 parentheses. If there are more than 15 capturing parentheses in a pat-
4250 tern, PCRE has to obtain extra memory to store data during a recursion,
4251 which it does by using pcre_malloc, freeing it via pcre_free after-
4252 wards. If no memory can be obtained, the match fails with the
4253 PCRE_ERROR_NOMEMORY error.
4254
4255 Do not confuse the (?R) item with the condition (R), which tests for
4256 recursion. Consider this pattern, which matches text in angle brack-
4257 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4258 brackets (that is, when recursing), whereas any characters are permit-
4259 ted at the outer level.
4260
4261 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4262
4263 In this pattern, (?(R) is the start of a conditional subpattern, with
4264 two different alternatives for the recursive and non-recursive cases.
4265 The (?R) item is the actual recursive call.
4266
4267
4268 SUBPATTERNS AS SUBROUTINES
4269
4270 If the syntax for a recursive subpattern reference (either by number or
4271 by name) is used outside the parentheses to which it refers, it oper-
4272 ates like a subroutine in a programming language. The "called" subpat-
4273 tern may be defined before or after the reference. An earlier example
4274 pointed out that the pattern
4275
4276 (sens|respons)e and \1ibility
4277
4278 matches "sense and sensibility" and "response and responsibility", but
4279 not "sense and responsibility". If instead the pattern
4280
4281 (sens|respons)e and (?1)ibility
4282
4283 is used, it does match "sense and responsibility" as well as the other
4284 two strings. Another example is given in the discussion of DEFINE
4285 above.
4286
4287 Like recursive subpatterns, a "subroutine" call is always treated as an
4288 atomic group. That is, once it has matched some of the subject string,
4289 it is never re-entered, even if it contains untried alternatives and
4290 there is a subsequent matching failure.
4291
4292 When a subpattern is used as a subroutine, processing options such as
4293 case-independence are fixed when the subpattern is defined. They cannot
4294 be changed for different calls. For example, consider this pattern:
4295
4296 (abc)(?i:(?1))
4297
4298 It matches "abcabc". It does not match "abcABC" because the change of
4299 processing option does not affect the called subpattern.
4300
4301
4302 CALLOUTS
4303
4304 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4305 Perl code to be obeyed in the middle of matching a regular expression.
4306 This makes it possible, amongst other things, to extract different sub-
4307 strings that match the same pair of parentheses when there is a repeti-
4308 tion.
4309
4310 PCRE provides a similar feature, but of course it cannot obey arbitrary
4311 Perl code. The feature is called "callout". The caller of PCRE provides
4312 an external function by putting its entry point in the global variable
4313 pcre_callout. By default, this variable contains NULL, which disables
4314 all calling out.
4315
4316 Within a regular expression, (?C) indicates the points at which the
4317 external function is to be called. If you want to identify different
4318 callout points, you can put a number less than 256 after the letter C.
4319 The default value is zero. For example, this pattern has two callout
4320 points:
4321
4322 (?C1)abc(?C2)def
4323
4324 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4325 automatically installed before each item in the pattern. They are all
4326 numbered 255.
4327
4328 During matching, when PCRE reaches a callout point (and pcre_callout is
4329 set), the external function is called. It is provided with the number
4330 of the callout, the position in the pattern, and, optionally, one item
4331 of data originally supplied by the caller of pcre_exec(). The callout
4332 function may cause matching to proceed, to backtrack, or to fail alto-
4333 gether. A complete description of the interface to the callout function
4334 is given in the pcrecallout documentation.
4335
4336
4337 SEE ALSO
4338
4339 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4340
4341
4342 AUTHOR
4343
4344 Philip Hazel
4345 University Computing Service
4346 Cambridge CB2 3QH, England.
4347
4348
4349 REVISION
4350
4351 Last updated: 06 March 2007
4352 Copyright (c) 1997-2007 University of Cambridge.
4353 ------------------------------------------------------------------------------
4354
4355
4356 PCREPARTIAL(3) PCREPARTIAL(3)
4357
4358
4359 NAME
4360 PCRE - Perl-compatible regular expressions
4361
4362
4363 PARTIAL MATCHING IN PCRE
4364
4365 In normal use of PCRE, if the subject string that is passed to
4366 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4367 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4368 There are circumstances where it might be helpful to distinguish this
4369 case from other cases in which there is no match.
4370
4371 Consider, for example, an application where a human is required to type
4372 in data for a field with specific formatting requirements. An example
4373 might be a date in the form ddmmmyy, defined by this pattern:
4374
4375 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4376
4377 If the application sees the user's keystrokes one by one, and can check
4378 that what has been typed so far is potentially valid, it is able to
4379 raise an error as soon as a mistake is made, possibly beeping and not
4380 reflecting the character that has been typed. This immediate feedback
4381 is likely to be a better user interface than a check that is delayed
4382 until the entire string has been entered.
4383
4384 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4385 TIAL option, which can be set when calling pcre_exec() or
4386 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4387 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4388 during the matching process the last part of the subject string matched
4389 part of the pattern. Unfortunately, for non-anchored matching, it is
4390 not possible to obtain the position of the start of the partial match.
4391 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4392
4393 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4394 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4395 the subject is reached, there have been no complete matches, but there
4396 is still at least one matching possibility. The portion of the string
4397 that provided the partial match is set as the first matching string.
4398
4399 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4400 the last literal byte in a pattern, and abandons matching immediately
4401 if such a byte is not present in the subject string. This optimization
4402 cannot be used for a subject string that might match only partially.
4403
4404
4405 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4406
4407 Because of the way certain internal optimizations are implemented in
4408 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4409 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4410 used. For pcre_exec(), repeated single characters such as
4411
4412 a{2,4}
4413
4414 and repeated single metasequences such as
4415
4416 \d+
4417
4418 are not permitted if the maximum number of occurrences is greater than
4419 one. Optional items such as \d? (where the maximum is one) are permit-
4420 ted. Quantifiers with any values are permitted after parentheses, so
4421 the invalid examples above can be coded thus:
4422
4423 (a){2,4}
4424 (\d)+
4425
4426 These constructions run more slowly, but for the kinds of application
4427 that are envisaged for this facility, this is not felt to be a major
4428 restriction.
4429
4430 If PCRE_PARTIAL is set for a pattern that does not conform to the
4431 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4432 (-13).
4433
4434
4435 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4436
4437 If the escape sequence \P is present in a pcretest data line, the
4438 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4439 uses the date example quoted above:
4440
4441 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4442 data> 25jun04\P
4443 0: 25jun04
4444 1: jun
4445 data> 25dec3\P
4446 Partial match
4447 data> 3ju\P
4448 Partial match
4449 data> 3juj\P
4450 No match
4451 data> j\P
4452 No match
4453
4454 The first data string is matched completely, so pcretest shows the
4455 matched substrings. The remaining four strings do not match the com-
4456 plete pattern, but the first two are partial matches. The same test,
4457 using pcre_dfa_exec() matching (by means of the \D escape sequence),
4458 produces the following output:
4459
4460 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4461 data> 25jun04\P\D
4462 0: 25jun04
4463 data> 23dec3\P\D
4464 Partial match: 23dec3
4465 data> 3ju\P\D
4466 Partial match: 3ju
4467 data> 3juj\P\D
4468 No match
4469 data> j\P\D
4470 No match
4471
4472 Notice that in this case the portion of the string that was matched is
4473 made available.
4474
4475
4476 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4477
4478 When a partial match has been found using pcre_dfa_exec(), it is possi-
4479 ble to continue the match by providing additional subject data and
4480 calling pcre_dfa_exec() again with the same compiled regular expres-
4481 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
4482 the same working space as before, because this is where details of the
4483 previous partial match are stored. Here is an example using pcretest,
4484 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
4485 \D are as above):
4486
4487 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4488 data> 23ja\P\D
4489 Partial match: 23ja
4490 data> n05\R\D
4491 0: n05
4492
4493 The first call has "23ja" as the subject, and requests partial match-
4494 ing; the second call has "n05" as the subject for the continued
4495 (restarted) match. Notice that when the match is complete, only the
4496 last part is shown; PCRE does not retain the previously partially-
4497 matched string. It is up to the calling program to do that if it needs
4498 to.
4499
4500 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
4501 matching over multiple segments. This facility can be used to pass very
4502 long subject strings to pcre_dfa_exec(). However, some care is needed
4503 for certain types of pattern.
4504
4505 1. If the pattern contains tests for the beginning or end of a line,
4506 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4507 ate, when the subject string for any call does not contain the begin-
4508 ning or end of a line.
4509
4510 2. If the pattern contains backward assertions (including \b or \B),
4511 you need to arrange for some overlap in the subject strings to allow
4512 for this. For example, you could pass the subject in chunks that are
4513 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4514 set to 200 and the previous 200 bytes at the start of the buffer.
4515
4516 3. Matching a subject string that is split into multiple segments does
4517 not always produce exactly the same result as matching over one single
4518 long string. The difference arises when there are multiple matching
4519 possibilities, because a partial match result is given only when there
4520 are no completed matches in a call to fBpcre_dfa_exec(). This means
4521 that as soon as the shortest match has been found, continuation to a
4522 new subject segment is no longer possible. Consider this pcretest
4523 example:
4524
4525 re> /dog(sbody)?/
4526 data> do\P\D
4527 Partial match: do
4528 data> gsb\R\P\D
4529 0: g
4530 data> dogsbody\D
4531 0: dogsbody
4532 1: dog
4533
4534 The pattern matches the words "dog" or "dogsbody". When the subject is
4535 presented in several parts ("do" and "gsb" being the first two) the
4536 match stops when "dog" has been found, and it is not possible to con-
4537 tinue. On the other hand, if "dogsbody" is presented as a single
4538 string, both matches are found.
4539
4540 Because of this phenomenon, it does not usually make sense to end a
4541 pattern that is going to be matched in this way with a variable repeat.
4542
4543 4. Patterns that contain alternatives at the top level which do not all
4544 start with the same pattern item may not work as expected. For example,
4545 consider this pattern:
4546
4547 1234|3789
4548
4549 If the first part of the subject is "ABC123", a partial match of the
4550 first alternative is found at offset 3. There is no partial match for
4551 the second alternative, because such a match does not start at the same
4552 point in the subject string. Attempting to continue with the string
4553 "789" does not yield a match because only those alternatives that match
4554 at one point in the subject are remembered. The problem arises because
4555 the start of the second alternative matches within the first alterna-
4556 tive. There is no problem with anchored patterns or patterns such as:
4557
4558 1234|ABCD
4559
4560 where no string can be a partial match for both alternatives.
4561
4562
4563 AUTHOR
4564
4565 Philip Hazel
4566 University Computing Service
4567 Cambridge CB2 3QH, England.
4568
4569
4570 REVISION
4571
4572 Last updated: 06 March 2007
4573 Copyright (c) 1997-2007 University of Cambridge.
4574 ------------------------------------------------------------------------------
4575
4576
4577 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4578
4579
4580 NAME
4581 PCRE - Perl-compatible regular expressions
4582
4583
4584 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4585
4586 If you are running an application that uses a large number of regular
4587 expression patterns, it may be useful to store them in a precompiled
4588 form instead of having to compile them every time the application is
4589 run. If you are not using any private character tables (see the
4590 pcre_maketables() documentation), this is relatively straightforward.
4591 If you are using private tables, it is a little bit more complicated.
4592
4593 If you save compiled patterns to a file, you can copy them to a differ-
4594 ent host and run them there. This works even if the new host has the
4595 opposite endianness to the one on which the patterns were compiled.
4596 There may be a small performance penalty, but it should be insignifi-
4597 cant.
4598
4599
4600 SAVING A COMPILED PATTERN
4601 The value returned by pcre_compile() points to a single block of memory
4602 that holds the compiled pattern and associated data. You can find the
4603 length of this block in bytes by calling pcre_fullinfo() with an argu-
4604 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4605 manner. Here is sample code that compiles a pattern and writes it to a
4606 file. It assumes that the variable fd refers to a file that is open for
4607 output:
4608
4609 int erroroffset, rc, size;
4610 char *error;
4611 pcre *re;
4612
4613 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4614 if (re == NULL) { ... handle errors ... }
4615 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4616 if (rc < 0) { ... handle errors ... }
4617 rc = fwrite(re, 1, size, fd);
4618 if (rc != size) { ... handle errors ... }
4619
4620 In this example, the bytes that comprise the compiled pattern are
4621 copied exactly. Note that this is binary data that may contain any of
4622 the 256 possible byte values. On systems that make a distinction
4623 between binary and non-binary data, be sure that the file is opened for
4624 binary output.
4625
4626 If you want to write more than one pattern to a file, you will have to
4627 devise a way of separating them. For binary data, preceding each pat-
4628 tern with its length is probably the most straightforward approach.
4629 Another possibility is to write out the data in hexadecimal instead of
4630 binary, one pattern to a line.
4631
4632 Saving compiled patterns in a file is only one possible way of storing
4633 them for later use. They could equally well be saved in a database, or
4634 in the memory of some daemon process that passes them via sockets to
4635 the processes that want them.
4636
4637 If the pattern has been studied, it is also possible to save the study
4638 data in a similar way to the compiled pattern itself. When studying
4639 generates additional information, pcre_study() returns a pointer to a
4640 pcre_extra data block. Its format is defined in the section on matching
4641 a pattern in the pcreapi documentation. The study_data field points to
4642 the binary study data, and this is what you must save (not the
4643 pcre_extra block itself). The length of the study data can be obtained
4644 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4645 Remember to check that pcre_study() did return a non-NULL value before
4646 trying to save the study data.
4647
4648
4649 RE-USING A PRECOMPILED PATTERN
4650
4651 Re-using a precompiled pattern is straightforward. Having reloaded it
4652 into main memory, you pass its pointer to pcre_exec() or
4653 pcre_dfa_exec() in the usual way. This should work even on another
4654 host, and even if that host has the opposite endianness to the one
4655 where the pattern was compiled.
4656
4657 However, if you passed a pointer to custom character tables when the
4658 pattern was compiled (the tableptr argument of pcre_compile()), you
4659 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4660 because the value saved with the compiled pattern will obviously be
4661 nonsense. A field in a pcre_extra() block is used to pass this data, as
4662 described in the section on matching a pattern in the pcreapi documen-
4663 tation.
4664
4665 If you did not provide custom character tables when the pattern was
4666 compiled, the pointer in the compiled pattern is NULL, which causes
4667 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4668 take any special action at run time in this case.
4669
4670 If you saved study data with the compiled pattern, you need to create
4671 your own pcre_extra data block and set the study_data field to point to
4672 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4673 bit in the flags field to indicate that study data is present. Then
4674 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4675 usual way.
4676
4677
4678 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4679
4680 The layout of the control block that is at the start of the data that
4681 makes up a compiled pattern was changed for release 5.0. If you have
4682 any saved patterns that were compiled with previous releases (not a
4683 facility that was previously advertised), you will have to recompile
4684 them for release 5.0 and above.
4685
4686 If you have any saved patterns in UTF-8 mode that use \p or \P that
4687 were compiled with any release up to and including 6.4, you will have
4688 to recompile them for release 6.5 and above.
4689
4690 All saved patterns from earlier releases must be recompiled for release
4691 7.0 or higher, because there was an internal reorganization at that
4692 release.
4693
4694
4695 AUTHOR
4696
4697 Philip Hazel
4698 University Computing Service
4699 Cambridge CB2 3QH, England.
4700
4701
4702 REVISION
4703
4704 Last updated: 06 March 2007
4705 Copyright (c) 1997-2007 University of Cambridge.
4706 ------------------------------------------------------------------------------
4707
4708
4709 PCREPERFORM(3) PCREPERFORM(3)
4710
4711
4712 NAME
4713 PCRE - Perl-compatible regular expressions
4714
4715
4716 PCRE PERFORMANCE
4717
4718 Two aspects of performance are discussed below: memory usage and pro-
4719 cessing time. The way you express your pattern as a regular expression
4720 can affect both of them.
4721
4722
4723 MEMORY USAGE
4724
4725 Patterns are compiled by PCRE into a reasonably efficient byte code, so
4726 that most simple patterns do not use much memory. However, there is one
4727 case where memory usage can be unexpectedly large. When a parenthesized
4728 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
4729 ited maximum, the whole subpattern is repeated in the compiled code.
4730 For example, the pattern
4731
4732 (abc|def){2,4}
4733
4734 is compiled as if it were
4735
4736 (abc|def)(abc|def)((abc|def)(abc|def)?)?
4737
4738 (Technical aside: It is done this way so that backtrack points within
4739 each of the repetitions can be independently maintained.)
4740
4741 For regular expressions whose quantifiers use only small numbers, this
4742 is not usually a problem. However, if the numbers are large, and par-
4743 ticularly if such repetitions are nested, the memory usage can become
4744 an embarrassment. For example, the very simple pattern
4745
4746 ((ab){1,1000}c){1,3}
4747
4748 uses 51K bytes when compiled. When PCRE is compiled with its default
4749 internal pointer size of two bytes, the size limit on a compiled pat-
4750 tern is 64K, and this is reached with the above pattern if the outer
4751 repetition is increased from 3 to 4. PCRE can be compiled to use larger
4752 internal pointers and thus handle larger compiled patterns, but it is
4753 better to try to rewrite your pattern to use less memory if you can.
4754
4755 One way of reducing the memory usage for such patterns is to make use
4756 of PCRE's "subroutine" facility. Re-writing the above pattern as
4757
4758 ((ab)(?2){0,999}c)(?1){0,2}
4759
4760 reduces the memory requirements to 18K, and indeed it remains under 20K
4761 even with the outer repetition increased to 100. However, this pattern
4762 is not exactly equivalent, because the "subroutine" calls are treated
4763 as atomic groups into which there can be no backtracking if there is a
4764 subsequent matching failure. Therefore, PCRE cannot do this kind of
4765 rewriting automatically. Furthermore, there is a noticeable loss of
4766 speed when executing the modified pattern. Nevertheless, if the atomic
4767 grouping is not a problem and the loss of speed is acceptable, this
4768 kind of rewriting will allow you to process patterns that PCRE cannot
4769 otherwise handle.
4770
4771
4772 PROCESSING TIME
4773
4774 Certain items in regular expression patterns are processed more effi-
4775 ciently than others. It is more efficient to use a character class like
4776 [aeiou] than a set of single-character alternatives such as
4777 (a|e|i|o|u). In general, the simplest construction that provides the
4778 required behaviour is usually the most efficient. Jeffrey Friedl's book
4779 contains a lot of useful general discussion about optimizing regular
4780 expressions for efficient performance. This document contains a few
4781 observations about PCRE.
4782
4783 Using Unicode character properties (the \p, \P, and \X escapes) is
4784 slow, because PCRE has to scan a structure that contains data for over
4785 fifteen thousand characters whenever it needs a character's property.
4786 If you can find an alternative pattern that does not use character
4787 properties, it will probably be faster.
4788
4789 When a pattern begins with .* not in parentheses, or in parentheses
4790 that are not the subject of a backreference, and the PCRE_DOTALL option
4791 is set, the pattern is implicitly anchored by PCRE, since it can match
4792 only at the start of a subject string. However, if PCRE_DOTALL is not
4793 set, PCRE cannot make this optimization, because the . metacharacter
4794 does not then match a newline, and if the subject string contains new-
4795 lines, the pattern may match from the character immediately following
4796 one of them instead of from the very start. For example, the pattern
4797
4798 .*second
4799
4800 matches the subject "first\nand second" (where \n stands for a newline
4801 character), with the match starting at the seventh character. In order
4802 to do this, PCRE has to retry the match starting after every newline in
4803 the subject.
4804
4805 If you are using such a pattern with subject strings that do not con-
4806 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4807 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4808 ing. That saves PCRE from having to scan along the subject looking for
4809 a newline to restart at.
4810
4811 Beware of patterns that contain nested indefinite repeats. These can
4812 take a long time to run when applied to a string that does not match.
4813 Consider the pattern fragment
4814
4815 ^(a+)*
4816
4817 This can match "aaaa" in 16 different ways, and this number increases
4818 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4819 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
4820 repeats can match different numbers of times.) When the remainder of
4821 the pattern is such that the entire match is going to fail, PCRE has in
4822 principle to try every possible variation, and this can take an
4823 extremely long time, even for relatively short strings.
4824
4825 An optimization catches some of the more simple cases such as
4826
4827 (a+)*b
4828
4829 where a literal character follows. Before embarking on the standard
4830 matching procedure, PCRE checks that there is a "b" later in the sub-
4831 ject string, and if there is not, it fails the match immediately. How-
4832 ever, when there is no following literal this optimization cannot be
4833 used. You can see the difference by comparing the behaviour of
4834
4835 (a+)*\d
4836
4837 with the pattern above. The former gives a failure almost instantly
4838 when applied to a whole line of "a" characters, whereas the latter
4839 takes an appreciable time with strings longer than about 20 characters.
4840
4841 In many cases, the solution to this kind of performance issue is to use
4842 an atomic group or a possessive quantifier.
4843
4844
4845 AUTHOR
4846
4847 Philip Hazel
4848 University Computing Service
4849 Cambridge CB2 3QH, England.
4850
4851
4852 REVISION
4853
4854 Last updated: 06 March 2007
4855 Copyright (c) 1997-2007 University of Cambridge.
4856 ------------------------------------------------------------------------------
4857
4858
4859 PCREPOSIX(3) PCREPOSIX(3)
4860
4861
4862 NAME
4863 PCRE - Perl-compatible regular expressions.
4864
4865
4866 SYNOPSIS OF POSIX API
4867
4868 #include <pcreposix.h>
4869
4870 int regcomp(regex_t *preg, const char *pattern,
4871 int cflags);
4872
4873 int regexec(regex_t *preg, const char *string,
4874 size_t nmatch, regmatch_t pmatch[], int eflags);
4875
4876 size_t regerror(int errcode, const regex_t *preg,
4877 char *errbuf, size_t errbuf_size);
4878
4879 void regfree(regex_t *preg);
4880
4881
4882 DESCRIPTION
4883
4884 This set of functions provides a POSIX-style API to the PCRE regular
4885 expression package. See the pcreapi documentation for a description of
4886 PCRE's native API, which contains much additional functionality.
4887
4888 The functions described here are just wrapper functions that ultimately
4889 call the PCRE native API. Their prototypes are defined in the
4890 pcreposix.h header file, and on Unix systems the library itself is
4891 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4892 command for linking an application that uses them. Because the POSIX
4893 functions call the native ones, it is also necessary to add -lpcre.
4894
4895 I have implemented only those option bits that can be reasonably mapped
4896 to PCRE native options. In addition, the option REG_EXTENDED is defined
4897 with the value zero. This has no effect, but since programs that are
4898 written to the POSIX interface often use it, this makes it easier to
4899 slot in PCRE as a replacement library. Other POSIX options are not even
4900 defined.
4901
4902 When PCRE is called via these functions, it is only the API that is
4903 POSIX-like in style. The syntax and semantics of the regular expres-
4904 sions themselves are still those of Perl, subject to the setting of
4905 various PCRE options, as described below. "POSIX-like in style" means
4906 that the API approximates to the POSIX definition; it is not fully
4907 POSIX-compatible, and in multi-byte encoding domains it is probably
4908 even less compatible.
4909
4910 The header for these functions is supplied as pcreposix.h to avoid any
4911 potential clash with other POSIX libraries. It can, of course, be
4912 renamed or aliased as regex.h, which is the "correct" name. It provides
4913 two structure types, regex_t for compiled internal forms, and reg-
4914 match_t for returning captured substrings. It also defines some con-
4915 stants whose names start with "REG_"; these are used for setting
4916 options and identifying error codes.
4917
4918
4919 COMPILING A PATTERN
4920
4921 The function regcomp() is called to compile a pattern into an internal
4922 form. The pattern is a C string terminated by a binary zero, and is
4923 passed in the argument pattern. The preg argument is a pointer to a
4924 regex_t structure that is used as a base for storing information about
4925 the compiled regular expression.
4926
4927 The argument cflags is either zero, or contains one or more of the bits
4928 defined by the following macros:
4929
4930 REG_DOTALL
4931
4932 The PCRE_DOTALL option is set when the regular expression is passed for
4933 compilation to the native function. Note that REG_DOTALL is not part of
4934 the POSIX standard.
4935
4936 REG_ICASE
4937
4938 The PCRE_CASELESS option is set when the regular expression is passed
4939 for compilation to the native function.
4940
4941 REG_NEWLINE
4942
4943 The PCRE_MULTILINE option is set when the regular expression is passed
4944 for compilation to the native function. Note that this does not mimic
4945 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
4946 tion).
4947
4948 REG_NOSUB
4949
4950 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
4951 passed for compilation to the native function. In addition, when a pat-
4952 tern that is compiled with this flag is passed to regexec() for match-
4953 ing, the nmatch and pmatch arguments are ignored, and no captured
4954 strings are returned.
4955
4956 REG_UTF8
4957
4958 The PCRE_UTF8 option is set when the regular expression is passed for
4959 compilation to the native function. This causes the pattern itself and
4960 all data strings used for matching it to be treated as UTF-8 strings.
4961 Note that REG_UTF8 is not part of the POSIX standard.
4962
4963 In the absence of these flags, no options are passed to the native
4964 function. This means the the regex is compiled with PCRE default
4965 semantics. In particular, the way it handles newline characters in the
4966 subject string is the Perl way, not the POSIX way. Note that setting
4967 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4968 It does not affect the way newlines are matched by . (they aren't) or
4969 by a negative class such as [^a] (they are).
4970
4971 The yield of regcomp() is zero on success, and non-zero otherwise. The
4972 preg structure is filled in on success, and one member of the structure
4973 is public: re_nsub contains the number of capturing subpatterns in the
4974 regular expression. Various error codes are defined in the header file.
4975
4976
4977 MATCHING NEWLINE CHARACTERS
4978
4979 This area is not simple, because POSIX and Perl take different views of
4980 things. It is not possible to get PCRE to obey POSIX semantics, but
4981 then PCRE was never intended to be a POSIX engine. The following table
4982 lists the different possibilities for matching newline characters in
4983 PCRE:
4984
4985 Default Change with
4986
4987 . matches newline no PCRE_DOTALL
4988 newline matches [^a] yes not changeable
4989 $ matches \n at end yes PCRE_DOLLARENDONLY
4990 $ matches \n in middle no PCRE_MULTILINE
4991 ^ matches \n in middle no PCRE_MULTILINE
4992
4993 This is the equivalent table for POSIX:
4994
4995 Default Change with
4996
4997 . matches newline yes REG_NEWLINE
4998 newline matches [^a] yes REG_NEWLINE
4999 $ matches \n at end no REG_NEWLINE
5000 $ matches \n in middle no REG_NEWLINE
5001 ^ matches \n in middle no REG_NEWLINE
5002
5003 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5004 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5005 no way to stop newline from matching [^a].
5006
5007 The default POSIX newline handling can be obtained by setting
5008 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5009 behave exactly as for the REG_NEWLINE action.
5010
5011
5012 MATCHING A PATTERN
5013
5014 The function regexec() is called to match a compiled pattern preg
5015 against a given string, which is terminated by a zero byte, subject to
5016 the options in eflags. These can be:
5017
5018 REG_NOTBOL
5019
5020 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5021 function.
5022
5023 REG_NOTEOL
5024
5025 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5026 function.
5027
5028 If the pattern was compiled with the REG_NOSUB flag, no data about any
5029 matched strings is returned. The nmatch and pmatch arguments of
5030 regexec() are ignored.
5031
5032 Otherwise,the portion of the string that was matched, and also any cap-
5033 tured substrings, are returned via the pmatch argument, which points to
5034 an array of nmatch structures of type regmatch_t, containing the mem-
5035 bers rm_so and rm_eo. These contain the offset to the first character
5036 of each substring and the offset to the first character after the end
5037 of each substring, respectively. The 0th element of the vector relates
5038 to the entire portion of string that was matched; subsequent elements
5039 relate to the capturing subpatterns of the regular expression. Unused
5040 entries in the array have both structure members set to -1.
5041
5042 A successful match yields a zero return; various error codes are
5043 defined in the header file, of which REG_NOMATCH is the "expected"
5044 failure code.
5045
5046
5047 ERROR MESSAGES
5048
5049 The regerror() function maps a non-zero errorcode from either regcomp()
5050 or regexec() to a printable message. If preg is not NULL, the error
5051 should have arisen from the use of that structure. A message terminated
5052 by a binary zero is placed in errbuf. The length of the message,
5053 including the zero, is limited to errbuf_size. The yield of the func-
5054 tion is the size of buffer needed to hold the whole message.
5055
5056
5057 MEMORY USAGE
5058
5059 Compiling a regular expression causes memory to be allocated and asso-
5060 ciated with the preg structure. The function regfree() frees all such
5061 memory, after which preg may no longer be used as a compiled expres-
5062 sion.
5063
5064
5065 AUTHOR
5066
5067 Philip Hazel
5068 University Computing Service
5069 Cambridge CB2 3QH, England.
5070
5071
5072 REVISION
5073
5074 Last updated: 06 March 2007
5075 Copyright (c) 1997-2007 University of Cambridge.
5076 ------------------------------------------------------------------------------
5077
5078
5079 PCRECPP(3) PCRECPP(3)
5080
5081
5082 NAME
5083 PCRE - Perl-compatible regular expressions.
5084
5085
5086 SYNOPSIS OF C++ WRAPPER
5087
5088 #include <pcrecpp.h>
5089
5090
5091 DESCRIPTION
5092
5093 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5094 functionality was added by Giuseppe Maxia. This brief man page was con-
5095 structed from the notes in the pcrecpp.h file, which should be con-
5096 sulted for further details.
5097
5098
5099 MATCHING INTERFACE
5100
5101 The "FullMatch" operation checks that supplied text matches a supplied
5102 pattern exactly. If pointer arguments are supplied, it copies matched
5103 sub-strings that match sub-patterns into them.
5104
5105 Example: successful match
5106 pcrecpp::RE re("h.*o");
5107 re.FullMatch("hello");
5108
5109 Example: unsuccessful match (requires full match):
5110 pcrecpp::RE re("e");
5111 !re.FullMatch("hello");
5112
5113 Example: creating a temporary RE object:
5114 pcrecpp::RE("h.*o").FullMatch("hello");
5115
5116 You can pass in a "const char*" or a "string" for "text". The examples
5117 below tend to use a const char*. You can, as in the different examples
5118 above, store the RE object explicitly in a variable or use a temporary
5119 RE object. The examples below use one mode or the other arbitrarily.
5120 Either could correctly be used for any of these examples.
5121
5122 You must supply extra pointer arguments to extract matched subpieces.
5123
5124 Example: extracts "ruby" into "s" and 1234 into "i"
5125 int i;
5126 string s;
5127 pcrecpp::RE re("(\\w+):(\\d+)");
5128 re.FullMatch("ruby:1234", &s, &i);
5129
5130 Example: does not try to extract any extra sub-patterns
5131 re.FullMatch("ruby:1234", &s);
5132
5133 Example: does not try to extract into NULL
5134 re.FullMatch("ruby:1234", NULL, &i);
5135
5136 Example: integer overflow causes failure
5137 !re.FullMatch("ruby:1234567891234", NULL, &i);
5138
5139 Example: fails because there aren't enough sub-patterns:
5140 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5141
5142 Example: fails because string cannot be stored in integer
5143 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5144
5145 The provided pointer arguments can be pointers to any scalar numeric
5146 type, or one of:
5147
5148 string (matched piece is copied to string)
5149 StringPiece (StringPiece is mutated to point to matched piece)
5150 T (where "bool T::ParseFrom(const char*, int)" exists)
5151 NULL (the corresponding matched sub-pattern is not copied)
5152
5153 The function returns true iff all of the following conditions are sat-
5154 isfied:
5155
5156 a. "text" matches "pattern" exactly;
5157
5158 b. The number of matched sub-patterns is >= number of supplied
5159 pointers;
5160
5161 c. The "i"th argument has a suitable type for holding the
5162 string captured as the "i"th sub-pattern. If you pass in
5163 NULL for the "i"th argument, or pass fewer arguments than
5164 number of sub-patterns, "i"th captured sub-pattern is
5165 ignored.
5166
5167 CAVEAT: An optional sub-pattern that does not exist in the matched
5168 string is assigned the empty string. Therefore, the following will
5169 return false (because the empty string is not a valid number):
5170
5171 int number;
5172 pcrecpp::RE::FullMatch("abc", "[a-z]+(\d+)?", &number);
5173
5174 The matching interface supports at most 16 arguments per call. If you
5175 need more, consider using the more general interface
5176 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5177
5178
5179 QUOTING METACHARACTERS
5180
5181 You can use the "QuoteMeta" operation to insert backslashes before all
5182 potentially meaningful characters in a string. The returned string,
5183 used as a regular expression, will exactly match the original string.
5184
5185 Example:
5186 string quoted = RE::QuoteMeta(unquoted);
5187
5188 Note that it's legal to escape a character even if it has no special
5189 meaning in a regular expression -- so this function does that. (This
5190 also makes it identical to the perl function of the same name; see
5191 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5192 "1\.5\-2\.0\?".
5193
5194
5195 PARTIAL MATCHES
5196
5197 You can use the "PartialMatch" operation when you want the pattern to
5198 match any substring of the text.
5199
5200 Example: simple search for a string:
5201 pcrecpp::RE("ell").PartialMatch("hello");
5202
5203 Example: find first number in a string:
5204 int number;
5205 pcrecpp::RE re("(\\d+)");
5206 re.PartialMatch("x*100 + 20", &number);
5207 assert(number == 100);
5208
5209
5210 UTF-8 AND THE MATCHING INTERFACE
5211
5212 By default, pattern and text are plain text, one byte per character.
5213 The UTF8 flag, passed to the constructor, causes both pattern and
5214 string to be treated as UTF-8 text, still a byte stream but potentially
5215 multiple bytes per character. In practice, the text is likelier to be
5216 UTF-8 than the pattern, but the match returned may depend on the UTF8
5217 flag, so always use it when matching UTF8 text. For example, "." will
5218 match one byte normally but with UTF8 set may match up to three bytes
5219 of a multi-byte character.
5220
5221 Example:
5222 pcrecpp::RE_Options options;
5223 options.set_utf8();
5224 pcrecpp::RE re(utf8_pattern, options);
5225 re.FullMatch(utf8_string);
5226
5227 Example: using the convenience function UTF8():
5228 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5229 re.FullMatch(utf8_string);
5230
5231 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5232 --enable-utf8 flag.
5233
5234
5235 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5236
5237 PCRE defines some modifiers to change the behavior of the regular
5238 expression engine. The C++ wrapper defines an auxiliary class,
5239 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5240 rently, the following modifiers are supported:
5241
5242 modifier description Perl corresponding
5243
5244 PCRE_CASELESS case insensitive match /i
5245 PCRE_MULTILINE multiple lines match /m
5246 PCRE_DOTALL dot matches newlines /s
5247 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5248 PCRE_EXTRA strict escape parsing N/A
5249 PCRE_EXTENDED ignore whitespaces /x
5250 PCRE_UTF8 handles UTF8 chars built-in
5251 PCRE_UNGREEDY reverses * and *? N/A
5252 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5253
5254 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5255 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5256 ture, while (ab|cd) does.
5257
5258 For a full account on how each modifier works, please check the PCRE
5259 API reference page.
5260
5261 For each modifier, there are two member functions whose name is made
5262 out of the modifier in lowercase, without the "PCRE_" prefix. For
5263 instance, PCRE_CASELESS is handled by
5264
5265 bool caseless()
5266
5267 which returns true if the modifier is set, and
5268
5269 RE_Options & set_caseless(bool)
5270
5271 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5272 be accessed through the set_match_limit() and match_limit() member
5273 functions. Setting match_limit to a non-zero value will limit the exe-
5274 cution of pcre to keep it from doing bad things like blowing the stack
5275 or taking an eternity to return a result. A value of 5000 is good
5276 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5277 to zero disables match limiting. Alternatively, you can call
5278 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5279 limit how much PCRE recurses. match_limit() limits the number of
5280 matches PCRE does; match_limit_recursion() limits the depth of internal
5281 recursion, and therefore the amount of stack that is used.
5282
5283 Normally, to pass one or more modifiers to a RE class, you declare a
5284 RE_Options object, set the appropriate options, and pass this object to
5285 a RE constructor. Example:
5286
5287 RE_options opt;
5288 opt.set_caseless(true);
5289 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5290
5291 RE_options has two constructors. The default constructor takes no argu-
5292 ments and creates a set of flags that are off by default. The optional
5293 parameter option_flags is to facilitate transfer of legacy code from C
5294 programs. This lets you do
5295
5296 RE(pattern,
5297 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5298
5299 However, new code is better off doing
5300
5301 RE(pattern,
5302 RE_Options().set_caseless(true).set_multiline(true))
5303 .PartialMatch(str);
5304
5305 If you are going to pass one of the most used modifiers, there are some
5306 convenience functions that return a RE_Options class with the appropri-
5307 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5308 and EXTENDED().
5309
5310 If you need to set several options at once, and you don't want to go
5311 through the pains of declaring a RE_Options object and setting several
5312 options, there is a parallel method that give you such ability on the
5313 fly. You can concatenate several set_xxxxx() member functions, since
5314 each of them returns a reference to its class object. For example, to
5315 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5316 statement, you may write:
5317
5318 RE(" ^ xyz \\s+ .* blah$",
5319 RE_Options()
5320 .set_caseless(true)
5321 .set_extended(true)
5322 .set_multiline(true)).PartialMatch(sometext);
5323
5324
5325 SCANNING TEXT INCREMENTALLY
5326
5327 The "Consume" operation may be useful if you want to repeatedly match
5328 regular expressions at the front of a string and skip over them as they
5329 match. This requires use of the "StringPiece" type, which represents a
5330 sub-range of a real string. Like RE, StringPiece is defined in the
5331 pcrecpp namespace.
5332
5333 Example: read lines of the form "var = value" from a string.
5334 string contents = ...; // Fill string somehow
5335 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
5336
5337 string var;
5338 int value;
5339 pcrecpp::RE re("(\\w+) = (\\d+)\n");
5340 while (re.Consume(&input, &var, &value)) {
5341 ...;
5342 }
5343
5344 Each successful call to "Consume" will set "var/value", and also
5345 advance "input" so it points past the matched text.
5346
5347 The "FindAndConsume" operation is similar to "Consume" but does not
5348 anchor your match at the beginning of the string. For example, you
5349 could extract all words from a string by repeatedly calling
5350
5351 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
5352
5353
5354 PARSING HEX/OCTAL/C-RADIX NUMBERS
5355
5356 By default, if you pass a pointer to a numeric value, the corresponding
5357 text is interpreted as a base-10 number. You can instead wrap the
5358 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
5359 to interpret the text in another base. The CRadix operator interprets
5360 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
5361 base-10.
5362
5363 Example:
5364 int a, b, c, d;
5365 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
5366 re.FullMatch("100 40 0100 0x40",
5367 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
5368 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
5369
5370 will leave 64 in a, b, c, and d.
5371
5372
5373 REPLACING PARTS OF STRINGS
5374
5375 You can replace the first match of "pattern" in "str" with "rewrite".
5376 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
5377 insert text matching corresponding parenthesized group from the pat-
5378 tern. \0 in "rewrite" refers to the entire matching text. For example:
5379
5380 string s = "yabba dabba doo";
5381 pcrecpp::RE("b+").Replace("d", &s);
5382
5383 will leave "s" containing "yada dabba doo". The result is true if the
5384 pattern matches and a replacement occurs, false otherwise.
5385
5386 GlobalReplace is like Replace except that it replaces all occurrences
5387 of the pattern in the string with the rewrite. Replacements are not
5388 subject to re-matching. For example:
5389
5390 string s = "yabba dabba doo";
5391 pcrecpp::RE("b+").GlobalReplace("d", &s);
5392
5393 will leave "s" containing "yada dada doo". It returns the number of
5394 replacements made.
5395
5396 Extract is like Replace, except that if the pattern matches, "rewrite"
5397 is copied into "out" (an additional argument) with substitutions. The
5398 non-matching portions of "text" are ignored. Returns true iff a match
5399 occurred and the extraction happened successfully; if no match occurs,
5400 the string is left unaffected.
5401
5402
5403 AUTHOR
5404
5405 The C++ wrapper was contributed by Google Inc.
5406 Copyright (c) 2007 Google Inc.
5407
5408
5409 REVISION
5410
5411 Last updated: 06 March 2007
5412 ------------------------------------------------------------------------------
5413
5414
5415 PCRESAMPLE(3) PCRESAMPLE(3)
5416
5417
5418 NAME
5419 PCRE - Perl-compatible regular expressions
5420
5421
5422 PCRE SAMPLE PROGRAM
5423
5424 A simple, complete demonstration program, to get you started with using
5425 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
5426
5427 The program compiles the regular expression that is its first argument,
5428 and matches it against the subject string in its second argument. No
5429 PCRE options are set, and default character tables are used. If match-
5430 ing succeeds, the program outputs the portion of the subject that
5431 matched, together with the contents of any captured substrings.
5432
5433 If the -g option is given on the command line, the program then goes on
5434 to check for further matches of the same regular expression in the same
5435 subject string. The logic is a little bit tricky because of the possi-
5436 bility of matching an empty string. Comments in the code explain what
5437 is going on.
5438
5439 If PCRE is installed in the standard include and library directories
5440 for your system, you should be able to compile the demonstration pro-
5441 gram using this command:
5442
5443 gcc -o pcredemo pcredemo.c -lpcre
5444
5445 If PCRE is installed elsewhere, you may need to add additional options
5446 to the command line. For example, on a Unix-like system that has PCRE
5447 installed in /usr/local, you can compile the demonstration program
5448 using a command like this:
5449
5450 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5451 -L/usr/local/lib -lpcre
5452
5453 Once you have compiled the demonstration program, you can run simple
5454 tests like this:
5455
5456 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5457 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5458
5459 Note that there is a much more comprehensive test program, called
5460 pcretest, which supports many more facilities for testing regular
5461 expressions and the PCRE library. The pcredemo program is provided as a
5462 simple coding example.
5463
5464 On some operating systems (e.g. Solaris), when PCRE is not installed in
5465 the standard library directory, you may get an error like this when you
5466 try to run pcredemo:
5467
5468 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5469 directory
5470
5471 This is caused by the way shared library support works on those sys-
5472 tems. You need to add
5473
5474 -R/usr/local/lib
5475
5476 (for example) to the compile command to get round this problem.
5477
5478
5479 AUTHOR
5480
5481 Philip Hazel
5482 University Computing Service
5483 Cambridge CB2 3QH, England.
5484
5485
5486 REVISION
5487
5488 Last updated: 06 March 2007
5489 Copyright (c) 1997-2007 University of Cambridge.
5490 ------------------------------------------------------------------------------
5491 PCRESTACK(3) PCRESTACK(3)
5492
5493
5494 NAME
5495 PCRE - Perl-compatible regular expressions
5496
5497
5498 PCRE DISCUSSION OF STACK USAGE
5499
5500 When you call pcre_exec(), it makes use of an internal function called
5501 match(). This calls itself recursively at branch points in the pattern,
5502 in order to remember the state of the match so that it can back up and
5503 try a different alternative if the first one fails. As matching pro-
5504 ceeds deeper and deeper into the tree of possibilities, the recursion
5505 depth increases.
5506
5507 Not all calls of match() increase the recursion depth; for an item such
5508 as a* it may be called several times at the same level, after matching
5509 different numbers of a's. Furthermore, in a number of cases where the
5510 result of the recursive call would immediately be passed back as the
5511 result of the current call (a "tail recursion"), the function is just
5512 restarted instead.
5513
5514 The pcre_dfa_exec() function operates in an entirely different way, and
5515 hardly uses recursion at all. The limit on its complexity is the amount
5516 of workspace it is given. The comments that follow do NOT apply to
5517 pcre_dfa_exec(); they are relevant only for pcre_exec().
5518
5519 You can set limits on the number of times that match() is called, both
5520 in total and recursively. If the limit is exceeded, an error occurs.
5521 For details, see the section on extra data for pcre_exec() in the
5522 pcreapi documentation.
5523
5524 Each time that match() is actually called recursively, it uses memory
5525 from the process stack. For certain kinds of pattern and data, very
5526 large amounts of stack may be needed, despite the recognition of "tail
5527 recursion". You can often reduce the amount of recursion, and there-
5528 fore the amount of stack used, by modifying the pattern that is being
5529 matched. Consider, for example, this pattern:
5530
5531 ([^<]|<(?!inet))+
5532
5533 It matches from wherever it starts until it encounters "<inet" or the
5534 end of the data, and is the kind of pattern that might be used when
5535 processing an XML file. Each iteration of the outer parentheses matches
5536 either one character that is not "<" or a "<" that is not followed by
5537 "inet". However, each time a parenthesis is processed, a recursion
5538 occurs, so this formulation uses a stack frame for each matched charac-
5539 ter. For a long string, a lot of stack is required. Consider now this
5540 rewritten pattern, which matches exactly the same strings:
5541
5542 ([^<]++|<(?!inet))+
5543
5544 This uses very much less stack, because runs of characters that do not
5545 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5546 sion happens only when a "<" character that is not followed by "inet"
5547 is encountered (and we assume this is relatively rare). A possessive
5548 quantifier is used to stop any backtracking into the runs of non-"<"
5549 characters, but that is not related to stack usage.
5550
5551 This example shows that one way of avoiding stack problems when match-
5552 ing long subject strings is to write repeated parenthesized subpatterns
5553 to match more than one character whenever possible.
5554
5555 In environments where stack memory is constrained, you might want to
5556 compile PCRE to use heap memory instead of stack for remembering back-
5557 up points. This makes it run a lot more slowly, however. Details of how
5558 to do this are given in the pcrebuild documentation.
5559
5560 In Unix-like environments, there is not often a problem with the stack
5561 unless very long strings are involved, though the default limit on
5562 stack size varies from system to system. Values from 8Mb to 64Mb are
5563 common. You can find your default limit by running the command:
5564
5565 ulimit -s
5566
5567 Unfortunately, the effect of running out of stack is often SIGSEGV,
5568 though sometimes a more explicit error message is given. You can nor-
5569 mally increase the limit on stack size by code such as this:
5570
5571 struct rlimit rlim;
5572 getrlimit(RLIMIT_STACK, &rlim);
5573 rlim.rlim_cur = 100*1024*1024;
5574 setrlimit(RLIMIT_STACK, &rlim);
5575
5576 This reads the current limits (soft and hard) using getrlimit(), then
5577 attempts to increase the soft limit to 100Mb using setrlimit(). You
5578 must do this before calling pcre_exec().
5579
5580 PCRE has an internal counter that can be used to limit the depth of
5581 recursion, and thus cause pcre_exec() to give an error code before it
5582 runs out of stack. By default, the limit is very large, and unlikely
5583 ever to operate. It can be changed when PCRE is built, and it can also
5584 be set when pcre_exec() is called. For details of these interfaces, see
5585 the pcrebuild and pcreapi documentation.
5586
5587 As a very rough rule of thumb, you should reckon on about 500 bytes per
5588 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5589 should set the limit at 16000 recursions. A 64Mb stack, on the other
5590 hand, can support around 128000 recursions. The pcretest test program
5591 has a command line option (-S) that can be used to increase the size of
5592 its stack.
5593
5594
5595 AUTHOR
5596
5597 Philip Hazel
5598 University Computing Service
5599 Cambridge CB2 3QH, England.
5600
5601
5602 REVISION
5603
5604 Last updated: 12 March 2007
5605 Copyright (c) 1997-2007 University of Cambridge.
5606 ------------------------------------------------------------------------------
5607
5608

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5