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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 406 - (show annotations)
Mon Mar 23 12:05:43 2009 UTC (10 years, 6 months ago) by ph10
File MIME type: text/plain
File size: 312284 byte(s)
Trailing space tidies
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. There is also some support for certain .NET and Oniguruma syn-
24 tax items, and there is an option for requesting some minor changes
25 that give better JavaScript compatibility.
26
27 The current implementation of PCRE (release 7.x) corresponds approxi-
28 mately with Perl 5.10, including support for UTF-8 encoded strings and
29 Unicode general category properties. However, UTF-8 and Unicode support
30 has to be explicitly enabled; it is not the default. The Unicode tables
31 correspond to Unicode release 5.0.0.
32
33 In addition to the Perl-compatible matching function, PCRE contains an
34 alternative matching function that matches the same compiled patterns
35 in a different way. In certain circumstances, the alternative function
36 has some advantages. For a discussion of the two matching algorithms,
37 see the pcrematching page.
38
39 PCRE is written in C and released as a C library. A number of people
40 have written wrappers and interfaces of various kinds. In particular,
41 Google Inc. have provided a comprehensive C++ wrapper. This is now
42 included as part of the PCRE distribution. The pcrecpp page has details
43 of this interface. Other people's contributions can be found in the
44 Contrib directory at the primary FTP site, which is:
45
46 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
47
48 Details of exactly which Perl regular expression features are and are
49 not supported by PCRE are given in separate documents. See the pcrepat-
50 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
51 page.
52
53 Some features of PCRE can be included, excluded, or changed when the
54 library is built. The pcre_config() function makes it possible for a
55 client to discover which features are available. The features them-
56 selves are described in the pcrebuild page. Documentation about build-
57 ing PCRE for various operating systems can be found in the README file
58 in the source distribution.
59
60 The library contains a number of undocumented internal functions and
61 data tables that are used by more than one of the exported external
62 functions, but which are not intended for use by external callers.
63 Their names all begin with "_pcre_", which hopefully will not provoke
64 any name clashes. In some environments, it is possible to control which
65 external symbols are exported when a shared library is built, and in
66 these cases the undocumented symbols are not exported.
67
68
69 USER DOCUMENTATION
70
71 The user documentation for PCRE comprises a number of different sec-
72 tions. In the "man" format, each of these is a separate "man page". In
73 the HTML format, each is a separate page, linked from the index page.
74 In the plain text format, all the sections are concatenated, for ease
75 of searching. The sections are as follows:
76
77 pcre this document
78 pcre-config show PCRE installation configuration information
79 pcreapi details of PCRE's native C API
80 pcrebuild options for building PCRE
81 pcrecallout details of the callout feature
82 pcrecompat discussion of Perl compatibility
83 pcrecpp details of the C++ wrapper
84 pcregrep description of the pcregrep command
85 pcrematching discussion of the two matching algorithms
86 pcrepartial details of the partial matching facility
87 pcrepattern syntax and semantics of supported
88 regular expressions
89 pcresyntax quick syntax reference
90 pcreperform discussion of performance issues
91 pcreposix the POSIX-compatible C API
92 pcreprecompile details of saving and re-using precompiled patterns
93 pcresample discussion of the sample program
94 pcrestack discussion of stack usage
95 pcretest description of the pcretest testing command
96
97 In addition, in the "man" and HTML formats, there is a short page for
98 each C library function, listing its arguments and results.
99
100
101 LIMITATIONS
102
103 There are some size limitations in PCRE but it is hoped that they will
104 never in practice be relevant.
105
106 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
107 is compiled with the default internal linkage size of 2. If you want to
108 process regular expressions that are truly enormous, you can compile
109 PCRE with an internal linkage size of 3 or 4 (see the README file in
110 the source distribution and the pcrebuild documentation for details).
111 In these cases the limit is substantially larger. However, the speed
112 of execution is slower.
113
114 All values in repeating quantifiers must be less than 65536.
115
116 There is no limit to the number of parenthesized subpatterns, but there
117 can be no more than 65535 capturing subpatterns.
118
119 The maximum length of name for a named subpattern is 32 characters, and
120 the maximum number of named subpatterns is 10000.
121
122 The maximum length of a subject string is the largest positive number
123 that an integer variable can hold. However, when using the traditional
124 matching function, PCRE uses recursion to handle subpatterns and indef-
125 inite repetition. This means that the available stack space may limit
126 the size of a subject string that can be processed by certain patterns.
127 For a discussion of stack issues, see the pcrestack documentation.
128
129
130 UTF-8 AND UNICODE PROPERTY SUPPORT
131
132 From release 3.3, PCRE has had some support for character strings
133 encoded in the UTF-8 format. For release 4.0 this was greatly extended
134 to cover most common requirements, and in release 5.0 additional sup-
135 port for Unicode general category properties was added.
136
137 In order process UTF-8 strings, you must build PCRE to include UTF-8
138 support in the code, and, in addition, you must call pcre_compile()
139 with the PCRE_UTF8 option flag. When you do this, both the pattern and
140 any subject strings that are matched against it are treated as UTF-8
141 strings instead of just strings of bytes.
142
143 If you compile PCRE with UTF-8 support, but do not use it at run time,
144 the library will be a bit bigger, but the additional run time overhead
145 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
146 very big.
147
148 If PCRE is built with Unicode character property support (which implies
149 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
150 ported. The available properties that can be tested are limited to the
151 general category properties such as Lu for an upper case letter or Nd
152 for a decimal number, the Unicode script names such as Arabic or Han,
153 and the derived properties Any and L&. A full list is given in the
154 pcrepattern documentation. Only the short names for properties are sup-
155 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
156 ter}, is not supported. Furthermore, in Perl, many properties may
157 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
158 does not support this.
159
160 Validity of UTF-8 strings
161
162 When you set the PCRE_UTF8 flag, the strings passed as patterns and
163 subjects are (by default) checked for validity on entry to the relevant
164 functions. From release 7.3 of PCRE, the check is according the rules
165 of RFC 3629, which are themselves derived from the Unicode specifica-
166 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
167 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
168 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
169 to U+DFFF.
170
171 The excluded code points are the "Low Surrogate Area" of Unicode, of
172 which the Unicode Standard says this: "The Low Surrogate Area does not
173 contain any character assignments, consequently no character code
174 charts or namelists are provided for this area. Surrogates are reserved
175 for use with UTF-16 and then must be used in pairs." The code points
176 that are encoded by UTF-16 pairs are available as independent code
177 points in the UTF-8 encoding. (In other words, the whole surrogate
178 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
179
180 If an invalid UTF-8 string is passed to PCRE, an error return
181 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
182 that your strings are valid, and therefore want to skip these checks in
183 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
184 compile time or at run time, PCRE assumes that the pattern or subject
185 it is given (respectively) contains only valid UTF-8 codes. In this
186 case, it does not diagnose an invalid UTF-8 string.
187
188 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
189 what happens depends on why the string is invalid. If the string con-
190 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
191 string of characters in the range 0 to 0x7FFFFFFF. In other words,
192 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
193 strings according to the more liberal rules of RFC 2279. However, if
194 the string does not even conform to RFC 2279, the result is undefined.
195 Your program may crash.
196
197 If you want to process strings of values in the full range 0 to
198 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
199 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
200 this situation, you will have to apply your own validity check.
201
202 General comments about UTF-8 mode
203
204 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
205 two-byte UTF-8 character if the value is greater than 127.
206
207 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
208 characters for values greater than \177.
209
210 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
211 vidual bytes, for example: \x{100}{3}.
212
213 4. The dot metacharacter matches one UTF-8 character instead of a sin-
214 gle byte.
215
216 5. The escape sequence \C can be used to match a single byte in UTF-8
217 mode, but its use can lead to some strange effects. This facility is
218 not available in the alternative matching function, pcre_dfa_exec().
219
220 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
221 test characters of any code value, but the characters that PCRE recog-
222 nizes as digits, spaces, or word characters remain the same set as
223 before, all with values less than 256. This remains true even when PCRE
224 includes Unicode property support, because to do otherwise would slow
225 down PCRE in many common cases. If you really want to test for a wider
226 sense of, say, "digit", you must use Unicode property tests such as
227 \p{Nd}. Note that this also applies to \b, because it is defined in
228 terms of \w and \W.
229
230 7. Similarly, characters that match the POSIX named character classes
231 are all low-valued characters.
232
233 8. However, the Perl 5.10 horizontal and vertical whitespace matching
234 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
235 acters.
236
237 9. Case-insensitive matching applies only to characters whose values
238 are less than 128, unless PCRE is built with Unicode property support.
239 Even when Unicode property support is available, PCRE still uses its
240 own character tables when checking the case of low-valued characters,
241 so as not to degrade performance. The Unicode property information is
242 used only for characters with higher values. Even when Unicode property
243 support is available, PCRE supports case-insensitive matching only when
244 there is a one-to-one mapping between a letter's cases. There are a
245 small number of many-to-one mappings in Unicode; these are not sup-
246 ported by PCRE.
247
248
249 AUTHOR
250
251 Philip Hazel
252 University Computing Service
253 Cambridge CB2 3QH, England.
254
255 Putting an actual email address here seems to have been a spam magnet,
256 so I've taken it away. If you want to email me, use my two initials,
257 followed by the two digits 10, at the domain cam.ac.uk.
258
259
260 REVISION
261
262 Last updated: 18 March 2009
263 Copyright (c) 1997-2009 University of Cambridge.
264 ------------------------------------------------------------------------------
265
266
267 PCREBUILD(3) PCREBUILD(3)
268
269
270 NAME
271 PCRE - Perl-compatible regular expressions
272
273
274 PCRE BUILD-TIME OPTIONS
275
276 This document describes the optional features of PCRE that can be
277 selected when the library is compiled. It assumes use of the configure
278 script, where the optional features are selected or deselected by pro-
279 viding options to configure before running the make command. However,
280 the same options can be selected in both Unix-like and non-Unix-like
281 environments using the GUI facility of CMakeSetup if you are using
282 CMake instead of configure to build PCRE.
283
284 The complete list of options for configure (which includes the standard
285 ones such as the selection of the installation directory) can be
286 obtained by running
287
288 ./configure --help
289
290 The following sections include descriptions of options whose names
291 begin with --enable or --disable. These settings specify changes to the
292 defaults for the configure command. Because of the way that configure
293 works, --enable and --disable always come in pairs, so the complemen-
294 tary option always exists as well, but as it specifies the default, it
295 is not described.
296
297
298 C++ SUPPORT
299
300 By default, the configure script will search for a C++ compiler and C++
301 header files. If it finds them, it automatically builds the C++ wrapper
302 library for PCRE. You can disable this by adding
303
304 --disable-cpp
305
306 to the configure command.
307
308
309 UTF-8 SUPPORT
310
311 To build PCRE with support for UTF-8 Unicode character strings, add
312
313 --enable-utf8
314
315 to the configure command. Of itself, this does not make PCRE treat
316 strings as UTF-8. As well as compiling PCRE with this option, you also
317 have have to set the PCRE_UTF8 option when you call the pcre_compile()
318 function.
319
320 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
321 expects its input to be either ASCII or UTF-8 (depending on the runtime
322 option). It is not possible to support both EBCDIC and UTF-8 codes in
323 the same version of the library. Consequently, --enable-utf8 and
324 --enable-ebcdic are mutually exclusive.
325
326
327 UNICODE CHARACTER PROPERTY SUPPORT
328
329 UTF-8 support allows PCRE to process character values greater than 255
330 in the strings that it handles. On its own, however, it does not pro-
331 vide any facilities for accessing the properties of such characters. If
332 you want to be able to use the pattern escapes \P, \p, and \X, which
333 refer to Unicode character properties, you must add
334
335 --enable-unicode-properties
336
337 to the configure command. This implies UTF-8 support, even if you have
338 not explicitly requested it.
339
340 Including Unicode property support adds around 30K of tables to the
341 PCRE library. Only the general category properties such as Lu and Nd
342 are supported. Details are given in the pcrepattern documentation.
343
344
345 CODE VALUE OF NEWLINE
346
347 By default, PCRE interprets the linefeed (LF) character as indicating
348 the end of a line. This is the normal newline character on Unix-like
349 systems. You can compile PCRE to use carriage return (CR) instead, by
350 adding
351
352 --enable-newline-is-cr
353
354 to the configure command. There is also a --enable-newline-is-lf
355 option, which explicitly specifies linefeed as the newline character.
356
357 Alternatively, you can specify that line endings are to be indicated by
358 the two character sequence CRLF. If you want this, add
359
360 --enable-newline-is-crlf
361
362 to the configure command. There is a fourth option, specified by
363
364 --enable-newline-is-anycrlf
365
366 which causes PCRE to recognize any of the three sequences CR, LF, or
367 CRLF as indicating a line ending. Finally, a fifth option, specified by
368
369 --enable-newline-is-any
370
371 causes PCRE to recognize any Unicode newline sequence.
372
373 Whatever line ending convention is selected when PCRE is built can be
374 overridden when the library functions are called. At build time it is
375 conventional to use the standard for your operating system.
376
377
378 WHAT \R MATCHES
379
380 By default, the sequence \R in a pattern matches any Unicode newline
381 sequence, whatever has been selected as the line ending sequence. If
382 you specify
383
384 --enable-bsr-anycrlf
385
386 the default is changed so that \R matches only CR, LF, or CRLF. What-
387 ever is selected when PCRE is built can be overridden when the library
388 functions are called.
389
390
391 BUILDING SHARED AND STATIC LIBRARIES
392
393 The PCRE building process uses libtool to build both shared and static
394 Unix libraries by default. You can suppress one of these by adding one
395 of
396
397 --disable-shared
398 --disable-static
399
400 to the configure command, as required.
401
402
403 POSIX MALLOC USAGE
404
405 When PCRE is called through the POSIX interface (see the pcreposix doc-
406 umentation), additional working storage is required for holding the
407 pointers to capturing substrings, because PCRE requires three integers
408 per substring, whereas the POSIX interface provides only two. If the
409 number of expected substrings is small, the wrapper function uses space
410 on the stack, because this is faster than using malloc() for each call.
411 The default threshold above which the stack is no longer used is 10; it
412 can be changed by adding a setting such as
413
414 --with-posix-malloc-threshold=20
415
416 to the configure command.
417
418
419 HANDLING VERY LARGE PATTERNS
420
421 Within a compiled pattern, offset values are used to point from one
422 part to another (for example, from an opening parenthesis to an alter-
423 nation metacharacter). By default, two-byte values are used for these
424 offsets, leading to a maximum size for a compiled pattern of around
425 64K. This is sufficient to handle all but the most gigantic patterns.
426 Nevertheless, some people do want to process enormous patterns, so it
427 is possible to compile PCRE to use three-byte or four-byte offsets by
428 adding a setting such as
429
430 --with-link-size=3
431
432 to the configure command. The value given must be 2, 3, or 4. Using
433 longer offsets slows down the operation of PCRE because it has to load
434 additional bytes when handling them.
435
436
437 AVOIDING EXCESSIVE STACK USAGE
438
439 When matching with the pcre_exec() function, PCRE implements backtrack-
440 ing by making recursive calls to an internal function called match().
441 In environments where the size of the stack is limited, this can se-
442 verely limit PCRE's operation. (The Unix environment does not usually
443 suffer from this problem, but it may sometimes be necessary to increase
444 the maximum stack size. There is a discussion in the pcrestack docu-
445 mentation.) An alternative approach to recursion that uses memory from
446 the heap to remember data, instead of using recursive function calls,
447 has been implemented to work round the problem of limited stack size.
448 If you want to build a version of PCRE that works this way, add
449
450 --disable-stack-for-recursion
451
452 to the configure command. With this configuration, PCRE will use the
453 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
454 ment functions. By default these point to malloc() and free(), but you
455 can replace the pointers so that your own functions are used.
456
457 Separate functions are provided rather than using pcre_malloc and
458 pcre_free because the usage is very predictable: the block sizes
459 requested are always the same, and the blocks are always freed in
460 reverse order. A calling program might be able to implement optimized
461 functions that perform better than malloc() and free(). PCRE runs
462 noticeably more slowly when built in this way. This option affects only
463 the pcre_exec() function; it is not relevant for the the
464 pcre_dfa_exec() function.
465
466
467 LIMITING PCRE RESOURCE USAGE
468
469 Internally, PCRE has a function called match(), which it calls repeat-
470 edly (sometimes recursively) when matching a pattern with the
471 pcre_exec() function. By controlling the maximum number of times this
472 function may be called during a single matching operation, a limit can
473 be placed on the resources used by a single call to pcre_exec(). The
474 limit can be changed at run time, as described in the pcreapi documen-
475 tation. The default is 10 million, but this can be changed by adding a
476 setting such as
477
478 --with-match-limit=500000
479
480 to the configure command. This setting has no effect on the
481 pcre_dfa_exec() matching function.
482
483 In some environments it is desirable to limit the depth of recursive
484 calls of match() more strictly than the total number of calls, in order
485 to restrict the maximum amount of stack (or heap, if --disable-stack-
486 for-recursion is specified) that is used. A second limit controls this;
487 it defaults to the value that is set for --with-match-limit, which
488 imposes no additional constraints. However, you can set a lower limit
489 by adding, for example,
490
491 --with-match-limit-recursion=10000
492
493 to the configure command. This value can also be overridden at run
494 time.
495
496
497 CREATING CHARACTER TABLES AT BUILD TIME
498
499 PCRE uses fixed tables for processing characters whose code values are
500 less than 256. By default, PCRE is built with a set of tables that are
501 distributed in the file pcre_chartables.c.dist. These tables are for
502 ASCII codes only. If you add
503
504 --enable-rebuild-chartables
505
506 to the configure command, the distributed tables are no longer used.
507 Instead, a program called dftables is compiled and run. This outputs
508 the source for new set of tables, created in the default locale of your
509 C runtime system. (This method of replacing the tables does not work if
510 you are cross compiling, because dftables is run on the local host. If
511 you need to create alternative tables when cross compiling, you will
512 have to do so "by hand".)
513
514
515 USING EBCDIC CODE
516
517 PCRE assumes by default that it will run in an environment where the
518 character code is ASCII (or Unicode, which is a superset of ASCII).
519 This is the case for most computer operating systems. PCRE can, how-
520 ever, be compiled to run in an EBCDIC environment by adding
521
522 --enable-ebcdic
523
524 to the configure command. This setting implies --enable-rebuild-charta-
525 bles. You should only use it if you know that you are in an EBCDIC
526 environment (for example, an IBM mainframe operating system). The
527 --enable-ebcdic option is incompatible with --enable-utf8.
528
529
530 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
531
532 By default, pcregrep reads all files as plain text. You can build it so
533 that it recognizes files whose names end in .gz or .bz2, and reads them
534 with libz or libbz2, respectively, by adding one or both of
535
536 --enable-pcregrep-libz
537 --enable-pcregrep-libbz2
538
539 to the configure command. These options naturally require that the rel-
540 evant libraries are installed on your system. Configuration will fail
541 if they are not.
542
543
544 PCRETEST OPTION FOR LIBREADLINE SUPPORT
545
546 If you add
547
548 --enable-pcretest-libreadline
549
550 to the configure command, pcretest is linked with the libreadline
551 library, and when its input is from a terminal, it reads it using the
552 readline() function. This provides line-editing and history facilities.
553 Note that libreadline is GPL-licenced, so if you distribute a binary of
554 pcretest linked in this way, there may be licensing issues.
555
556 Setting this option causes the -lreadline option to be added to the
557 pcretest build. In many operating environments with a sytem-installed
558 libreadline this is sufficient. However, in some environments (e.g. if
559 an unmodified distribution version of readline is in use), some extra
560 configuration may be necessary. The INSTALL file for libreadline says
561 this:
562
563 "Readline uses the termcap functions, but does not link with the
564 termcap or curses library itself, allowing applications which link
565 with readline the to choose an appropriate library."
566
567 If your environment has not been set up so that an appropriate library
568 is automatically included, you may need to add something like
569
570 LIBS="-ncurses"
571
572 immediately before the configure command.
573
574
575 SEE ALSO
576
577 pcreapi(3), pcre_config(3).
578
579
580 AUTHOR
581
582 Philip Hazel
583 University Computing Service
584 Cambridge CB2 3QH, England.
585
586
587 REVISION
588
589 Last updated: 17 March 2009
590 Copyright (c) 1997-2009 University of Cambridge.
591 ------------------------------------------------------------------------------
592
593
594 PCREMATCHING(3) PCREMATCHING(3)
595
596
597 NAME
598 PCRE - Perl-compatible regular expressions
599
600
601 PCRE MATCHING ALGORITHMS
602
603 This document describes the two different algorithms that are available
604 in PCRE for matching a compiled regular expression against a given sub-
605 ject string. The "standard" algorithm is the one provided by the
606 pcre_exec() function. This works in the same was as Perl's matching
607 function, and provides a Perl-compatible matching operation.
608
609 An alternative algorithm is provided by the pcre_dfa_exec() function;
610 this operates in a different way, and is not Perl-compatible. It has
611 advantages and disadvantages compared with the standard algorithm, and
612 these are described below.
613
614 When there is only one possible way in which a given subject string can
615 match a pattern, the two algorithms give the same answer. A difference
616 arises, however, when there are multiple possibilities. For example, if
617 the pattern
618
619 ^<.*>
620
621 is matched against the string
622
623 <something> <something else> <something further>
624
625 there are three possible answers. The standard algorithm finds only one
626 of them, whereas the alternative algorithm finds all three.
627
628
629 REGULAR EXPRESSIONS AS TREES
630
631 The set of strings that are matched by a regular expression can be rep-
632 resented as a tree structure. An unlimited repetition in the pattern
633 makes the tree of infinite size, but it is still a tree. Matching the
634 pattern to a given subject string (from a given starting point) can be
635 thought of as a search of the tree. There are two ways to search a
636 tree: depth-first and breadth-first, and these correspond to the two
637 matching algorithms provided by PCRE.
638
639
640 THE STANDARD MATCHING ALGORITHM
641
642 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
643 sions", the standard algorithm is an "NFA algorithm". It conducts a
644 depth-first search of the pattern tree. That is, it proceeds along a
645 single path through the tree, checking that the subject matches what is
646 required. When there is a mismatch, the algorithm tries any alterna-
647 tives at the current point, and if they all fail, it backs up to the
648 previous branch point in the tree, and tries the next alternative
649 branch at that level. This often involves backing up (moving to the
650 left) in the subject string as well. The order in which repetition
651 branches are tried is controlled by the greedy or ungreedy nature of
652 the quantifier.
653
654 If a leaf node is reached, a matching string has been found, and at
655 that point the algorithm stops. Thus, if there is more than one possi-
656 ble match, this algorithm returns the first one that it finds. Whether
657 this is the shortest, the longest, or some intermediate length depends
658 on the way the greedy and ungreedy repetition quantifiers are specified
659 in the pattern.
660
661 Because it ends up with a single path through the tree, it is rela-
662 tively straightforward for this algorithm to keep track of the sub-
663 strings that are matched by portions of the pattern in parentheses.
664 This provides support for capturing parentheses and back references.
665
666
667 THE ALTERNATIVE MATCHING ALGORITHM
668
669 This algorithm conducts a breadth-first search of the tree. Starting
670 from the first matching point in the subject, it scans the subject
671 string from left to right, once, character by character, and as it does
672 this, it remembers all the paths through the tree that represent valid
673 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
674 though it is not implemented as a traditional finite state machine (it
675 keeps multiple states active simultaneously).
676
677 The scan continues until either the end of the subject is reached, or
678 there are no more unterminated paths. At this point, terminated paths
679 represent the different matching possibilities (if there are none, the
680 match has failed). Thus, if there is more than one possible match,
681 this algorithm finds all of them, and in particular, it finds the long-
682 est. In PCRE, there is an option to stop the algorithm after the first
683 match (which is necessarily the shortest) has been found.
684
685 Note that all the matches that are found start at the same point in the
686 subject. If the pattern
687
688 cat(er(pillar)?)
689
690 is matched against the string "the caterpillar catchment", the result
691 will be the three strings "cat", "cater", and "caterpillar" that start
692 at the fourth character of the subject. The algorithm does not automat-
693 ically move on to find matches that start at later positions.
694
695 There are a number of features of PCRE regular expressions that are not
696 supported by the alternative matching algorithm. They are as follows:
697
698 1. Because the algorithm finds all possible matches, the greedy or
699 ungreedy nature of repetition quantifiers is not relevant. Greedy and
700 ungreedy quantifiers are treated in exactly the same way. However, pos-
701 sessive quantifiers can make a difference when what follows could also
702 match what is quantified, for example in a pattern like this:
703
704 ^a++\w!
705
706 This pattern matches "aaab!" but not "aaa!", which would be matched by
707 a non-possessive quantifier. Similarly, if an atomic group is present,
708 it is matched as if it were a standalone pattern at the current point,
709 and the longest match is then "locked in" for the rest of the overall
710 pattern.
711
712 2. When dealing with multiple paths through the tree simultaneously, it
713 is not straightforward to keep track of captured substrings for the
714 different matching possibilities, and PCRE's implementation of this
715 algorithm does not attempt to do this. This means that no captured sub-
716 strings are available.
717
718 3. Because no substrings are captured, back references within the pat-
719 tern are not supported, and cause errors if encountered.
720
721 4. For the same reason, conditional expressions that use a backrefer-
722 ence as the condition or test for a specific group recursion are not
723 supported.
724
725 5. Because many paths through the tree may be active, the \K escape
726 sequence, which resets the start of the match when encountered (but may
727 be on some paths and not on others), is not supported. It causes an
728 error if encountered.
729
730 6. Callouts are supported, but the value of the capture_top field is
731 always 1, and the value of the capture_last field is always -1.
732
733 7. The \C escape sequence, which (in the standard algorithm) matches a
734 single byte, even in UTF-8 mode, is not supported because the alterna-
735 tive algorithm moves through the subject string one character at a
736 time, for all active paths through the tree.
737
738 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
739 are not supported. (*FAIL) is supported, and behaves like a failing
740 negative assertion.
741
742
743 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
744
745 Using the alternative matching algorithm provides the following advan-
746 tages:
747
748 1. All possible matches (at a single point in the subject) are automat-
749 ically found, and in particular, the longest match is found. To find
750 more than one match using the standard algorithm, you have to do kludgy
751 things with callouts.
752
753 2. There is much better support for partial matching. The restrictions
754 on the content of the pattern that apply when using the standard algo-
755 rithm for partial matching do not apply to the alternative algorithm.
756 For non-anchored patterns, the starting position of a partial match is
757 available.
758
759 3. Because the alternative algorithm scans the subject string just
760 once, and never needs to backtrack, it is possible to pass very long
761 subject strings to the matching function in several pieces, checking
762 for partial matching each time.
763
764
765 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
766
767 The alternative algorithm suffers from a number of disadvantages:
768
769 1. It is substantially slower than the standard algorithm. This is
770 partly because it has to search for all possible matches, but is also
771 because it is less susceptible to optimization.
772
773 2. Capturing parentheses and back references are not supported.
774
775 3. Although atomic groups are supported, their use does not provide the
776 performance advantage that it does for the standard algorithm.
777
778
779 AUTHOR
780
781 Philip Hazel
782 University Computing Service
783 Cambridge CB2 3QH, England.
784
785
786 REVISION
787
788 Last updated: 19 April 2008
789 Copyright (c) 1997-2008 University of Cambridge.
790 ------------------------------------------------------------------------------
791
792
793 PCREAPI(3) PCREAPI(3)
794
795
796 NAME
797 PCRE - Perl-compatible regular expressions
798
799
800 PCRE NATIVE API
801
802 #include <pcre.h>
803
804 pcre *pcre_compile(const char *pattern, int options,
805 const char **errptr, int *erroffset,
806 const unsigned char *tableptr);
807
808 pcre *pcre_compile2(const char *pattern, int options,
809 int *errorcodeptr,
810 const char **errptr, int *erroffset,
811 const unsigned char *tableptr);
812
813 pcre_extra *pcre_study(const pcre *code, int options,
814 const char **errptr);
815
816 int pcre_exec(const pcre *code, const pcre_extra *extra,
817 const char *subject, int length, int startoffset,
818 int options, int *ovector, int ovecsize);
819
820 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
821 const char *subject, int length, int startoffset,
822 int options, int *ovector, int ovecsize,
823 int *workspace, int wscount);
824
825 int pcre_copy_named_substring(const pcre *code,
826 const char *subject, int *ovector,
827 int stringcount, const char *stringname,
828 char *buffer, int buffersize);
829
830 int pcre_copy_substring(const char *subject, int *ovector,
831 int stringcount, int stringnumber, char *buffer,
832 int buffersize);
833
834 int pcre_get_named_substring(const pcre *code,
835 const char *subject, int *ovector,
836 int stringcount, const char *stringname,
837 const char **stringptr);
838
839 int pcre_get_stringnumber(const pcre *code,
840 const char *name);
841
842 int pcre_get_stringtable_entries(const pcre *code,
843 const char *name, char **first, char **last);
844
845 int pcre_get_substring(const char *subject, int *ovector,
846 int stringcount, int stringnumber,
847 const char **stringptr);
848
849 int pcre_get_substring_list(const char *subject,
850 int *ovector, int stringcount, const char ***listptr);
851
852 void pcre_free_substring(const char *stringptr);
853
854 void pcre_free_substring_list(const char **stringptr);
855
856 const unsigned char *pcre_maketables(void);
857
858 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
859 int what, void *where);
860
861 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
862
863 int pcre_refcount(pcre *code, int adjust);
864
865 int pcre_config(int what, void *where);
866
867 char *pcre_version(void);
868
869 void *(*pcre_malloc)(size_t);
870
871 void (*pcre_free)(void *);
872
873 void *(*pcre_stack_malloc)(size_t);
874
875 void (*pcre_stack_free)(void *);
876
877 int (*pcre_callout)(pcre_callout_block *);
878
879
880 PCRE API OVERVIEW
881
882 PCRE has its own native API, which is described in this document. There
883 are also some wrapper functions that correspond to the POSIX regular
884 expression API. These are described in the pcreposix documentation.
885 Both of these APIs define a set of C function calls. A C++ wrapper is
886 distributed with PCRE. It is documented in the pcrecpp page.
887
888 The native API C function prototypes are defined in the header file
889 pcre.h, and on Unix systems the library itself is called libpcre. It
890 can normally be accessed by adding -lpcre to the command for linking an
891 application that uses PCRE. The header file defines the macros
892 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
893 bers for the library. Applications can use these to include support
894 for different releases of PCRE.
895
896 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
897 pcre_exec() are used for compiling and matching regular expressions in
898 a Perl-compatible manner. A sample program that demonstrates the sim-
899 plest way of using them is provided in the file called pcredemo.c in
900 the source distribution. The pcresample documentation describes how to
901 compile and run it.
902
903 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
904 ble, is also provided. This uses a different algorithm for the match-
905 ing. The alternative algorithm finds all possible matches (at a given
906 point in the subject), and scans the subject just once. However, this
907 algorithm does not return captured substrings. A description of the two
908 matching algorithms and their advantages and disadvantages is given in
909 the pcrematching documentation.
910
911 In addition to the main compiling and matching functions, there are
912 convenience functions for extracting captured substrings from a subject
913 string that is matched by pcre_exec(). They are:
914
915 pcre_copy_substring()
916 pcre_copy_named_substring()
917 pcre_get_substring()
918 pcre_get_named_substring()
919 pcre_get_substring_list()
920 pcre_get_stringnumber()
921 pcre_get_stringtable_entries()
922
923 pcre_free_substring() and pcre_free_substring_list() are also provided,
924 to free the memory used for extracted strings.
925
926 The function pcre_maketables() is used to build a set of character
927 tables in the current locale for passing to pcre_compile(),
928 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
929 provided for specialist use. Most commonly, no special tables are
930 passed, in which case internal tables that are generated when PCRE is
931 built are used.
932
933 The function pcre_fullinfo() is used to find out information about a
934 compiled pattern; pcre_info() is an obsolete version that returns only
935 some of the available information, but is retained for backwards com-
936 patibility. The function pcre_version() returns a pointer to a string
937 containing the version of PCRE and its date of release.
938
939 The function pcre_refcount() maintains a reference count in a data
940 block containing a compiled pattern. This is provided for the benefit
941 of object-oriented applications.
942
943 The global variables pcre_malloc and pcre_free initially contain the
944 entry points of the standard malloc() and free() functions, respec-
945 tively. PCRE calls the memory management functions via these variables,
946 so a calling program can replace them if it wishes to intercept the
947 calls. This should be done before calling any PCRE functions.
948
949 The global variables pcre_stack_malloc and pcre_stack_free are also
950 indirections to memory management functions. These special functions
951 are used only when PCRE is compiled to use the heap for remembering
952 data, instead of recursive function calls, when running the pcre_exec()
953 function. See the pcrebuild documentation for details of how to do
954 this. It is a non-standard way of building PCRE, for use in environ-
955 ments that have limited stacks. Because of the greater use of memory
956 management, it runs more slowly. Separate functions are provided so
957 that special-purpose external code can be used for this case. When
958 used, these functions are always called in a stack-like manner (last
959 obtained, first freed), and always for memory blocks of the same size.
960 There is a discussion about PCRE's stack usage in the pcrestack docu-
961 mentation.
962
963 The global variable pcre_callout initially contains NULL. It can be set
964 by the caller to a "callout" function, which PCRE will then call at
965 specified points during a matching operation. Details are given in the
966 pcrecallout documentation.
967
968
969 NEWLINES
970
971 PCRE supports five different conventions for indicating line breaks in
972 strings: a single CR (carriage return) character, a single LF (line-
973 feed) character, the two-character sequence CRLF, any of the three pre-
974 ceding, or any Unicode newline sequence. The Unicode newline sequences
975 are the three just mentioned, plus the single characters VT (vertical
976 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
977 separator, U+2028), and PS (paragraph separator, U+2029).
978
979 Each of the first three conventions is used by at least one operating
980 system as its standard newline sequence. When PCRE is built, a default
981 can be specified. The default default is LF, which is the Unix stan-
982 dard. When PCRE is run, the default can be overridden, either when a
983 pattern is compiled, or when it is matched.
984
985 At compile time, the newline convention can be specified by the options
986 argument of pcre_compile(), or it can be specified by special text at
987 the start of the pattern itself; this overrides any other settings. See
988 the pcrepattern page for details of the special character sequences.
989
990 In the PCRE documentation the word "newline" is used to mean "the char-
991 acter or pair of characters that indicate a line break". The choice of
992 newline convention affects the handling of the dot, circumflex, and
993 dollar metacharacters, the handling of #-comments in /x mode, and, when
994 CRLF is a recognized line ending sequence, the match position advance-
995 ment for a non-anchored pattern. There is more detail about this in the
996 section on pcre_exec() options below.
997
998 The choice of newline convention does not affect the interpretation of
999 the \n or \r escape sequences, nor does it affect what \R matches,
1000 which is controlled in a similar way, but by separate options.
1001
1002
1003 MULTITHREADING
1004
1005 The PCRE functions can be used in multi-threading applications, with
1006 the proviso that the memory management functions pointed to by
1007 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1008 callout function pointed to by pcre_callout, are shared by all threads.
1009
1010 The compiled form of a regular expression is not altered during match-
1011 ing, so the same compiled pattern can safely be used by several threads
1012 at once.
1013
1014
1015 SAVING PRECOMPILED PATTERNS FOR LATER USE
1016
1017 The compiled form of a regular expression can be saved and re-used at a
1018 later time, possibly by a different program, and even on a host other
1019 than the one on which it was compiled. Details are given in the
1020 pcreprecompile documentation. However, compiling a regular expression
1021 with one version of PCRE for use with a different version is not guar-
1022 anteed to work and may cause crashes.
1023
1024
1025 CHECKING BUILD-TIME OPTIONS
1026
1027 int pcre_config(int what, void *where);
1028
1029 The function pcre_config() makes it possible for a PCRE client to dis-
1030 cover which optional features have been compiled into the PCRE library.
1031 The pcrebuild documentation has more details about these optional fea-
1032 tures.
1033
1034 The first argument for pcre_config() is an integer, specifying which
1035 information is required; the second argument is a pointer to a variable
1036 into which the information is placed. The following information is
1037 available:
1038
1039 PCRE_CONFIG_UTF8
1040
1041 The output is an integer that is set to one if UTF-8 support is avail-
1042 able; otherwise it is set to zero.
1043
1044 PCRE_CONFIG_UNICODE_PROPERTIES
1045
1046 The output is an integer that is set to one if support for Unicode
1047 character properties is available; otherwise it is set to zero.
1048
1049 PCRE_CONFIG_NEWLINE
1050
1051 The output is an integer whose value specifies the default character
1052 sequence that is recognized as meaning "newline". The four values that
1053 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1054 and -1 for ANY. Though they are derived from ASCII, the same values
1055 are returned in EBCDIC environments. The default should normally corre-
1056 spond to the standard sequence for your operating system.
1057
1058 PCRE_CONFIG_BSR
1059
1060 The output is an integer whose value indicates what character sequences
1061 the \R escape sequence matches by default. A value of 0 means that \R
1062 matches any Unicode line ending sequence; a value of 1 means that \R
1063 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1064 tern is compiled or matched.
1065
1066 PCRE_CONFIG_LINK_SIZE
1067
1068 The output is an integer that contains the number of bytes used for
1069 internal linkage in compiled regular expressions. The value is 2, 3, or
1070 4. Larger values allow larger regular expressions to be compiled, at
1071 the expense of slower matching. The default value of 2 is sufficient
1072 for all but the most massive patterns, since it allows the compiled
1073 pattern to be up to 64K in size.
1074
1075 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1076
1077 The output is an integer that contains the threshold above which the
1078 POSIX interface uses malloc() for output vectors. Further details are
1079 given in the pcreposix documentation.
1080
1081 PCRE_CONFIG_MATCH_LIMIT
1082
1083 The output is a long integer that gives the default limit for the num-
1084 ber of internal matching function calls in a pcre_exec() execution.
1085 Further details are given with pcre_exec() below.
1086
1087 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1088
1089 The output is a long integer that gives the default limit for the depth
1090 of recursion when calling the internal matching function in a
1091 pcre_exec() execution. Further details are given with pcre_exec()
1092 below.
1093
1094 PCRE_CONFIG_STACKRECURSE
1095
1096 The output is an integer that is set to one if internal recursion when
1097 running pcre_exec() is implemented by recursive function calls that use
1098 the stack to remember their state. This is the usual way that PCRE is
1099 compiled. The output is zero if PCRE was compiled to use blocks of data
1100 on the heap instead of recursive function calls. In this case,
1101 pcre_stack_malloc and pcre_stack_free are called to manage memory
1102 blocks on the heap, thus avoiding the use of the stack.
1103
1104
1105 COMPILING A PATTERN
1106
1107 pcre *pcre_compile(const char *pattern, int options,
1108 const char **errptr, int *erroffset,
1109 const unsigned char *tableptr);
1110
1111 pcre *pcre_compile2(const char *pattern, int options,
1112 int *errorcodeptr,
1113 const char **errptr, int *erroffset,
1114 const unsigned char *tableptr);
1115
1116 Either of the functions pcre_compile() or pcre_compile2() can be called
1117 to compile a pattern into an internal form. The only difference between
1118 the two interfaces is that pcre_compile2() has an additional argument,
1119 errorcodeptr, via which a numerical error code can be returned.
1120
1121 The pattern is a C string terminated by a binary zero, and is passed in
1122 the pattern argument. A pointer to a single block of memory that is
1123 obtained via pcre_malloc is returned. This contains the compiled code
1124 and related data. The pcre type is defined for the returned block; this
1125 is a typedef for a structure whose contents are not externally defined.
1126 It is up to the caller to free the memory (via pcre_free) when it is no
1127 longer required.
1128
1129 Although the compiled code of a PCRE regex is relocatable, that is, it
1130 does not depend on memory location, the complete pcre data block is not
1131 fully relocatable, because it may contain a copy of the tableptr argu-
1132 ment, which is an address (see below).
1133
1134 The options argument contains various bit settings that affect the com-
1135 pilation. It should be zero if no options are required. The available
1136 options are described below. Some of them, in particular, those that
1137 are compatible with Perl, can also be set and unset from within the
1138 pattern (see the detailed description in the pcrepattern documenta-
1139 tion). For these options, the contents of the options argument speci-
1140 fies their initial settings at the start of compilation and execution.
1141 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1142 of matching as well as at compile time.
1143
1144 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1145 if compilation of a pattern fails, pcre_compile() returns NULL, and
1146 sets the variable pointed to by errptr to point to a textual error mes-
1147 sage. This is a static string that is part of the library. You must not
1148 try to free it. The offset from the start of the pattern to the charac-
1149 ter where the error was discovered is placed in the variable pointed to
1150 by erroffset, which must not be NULL. If it is, an immediate error is
1151 given.
1152
1153 If pcre_compile2() is used instead of pcre_compile(), and the error-
1154 codeptr argument is not NULL, a non-zero error code number is returned
1155 via this argument in the event of an error. This is in addition to the
1156 textual error message. Error codes and messages are listed below.
1157
1158 If the final argument, tableptr, is NULL, PCRE uses a default set of
1159 character tables that are built when PCRE is compiled, using the
1160 default C locale. Otherwise, tableptr must be an address that is the
1161 result of a call to pcre_maketables(). This value is stored with the
1162 compiled pattern, and used again by pcre_exec(), unless another table
1163 pointer is passed to it. For more discussion, see the section on locale
1164 support below.
1165
1166 This code fragment shows a typical straightforward call to pcre_com-
1167 pile():
1168
1169 pcre *re;
1170 const char *error;
1171 int erroffset;
1172 re = pcre_compile(
1173 "^A.*Z", /* the pattern */
1174 0, /* default options */
1175 &error, /* for error message */
1176 &erroffset, /* for error offset */
1177 NULL); /* use default character tables */
1178
1179 The following names for option bits are defined in the pcre.h header
1180 file:
1181
1182 PCRE_ANCHORED
1183
1184 If this bit is set, the pattern is forced to be "anchored", that is, it
1185 is constrained to match only at the first matching point in the string
1186 that is being searched (the "subject string"). This effect can also be
1187 achieved by appropriate constructs in the pattern itself, which is the
1188 only way to do it in Perl.
1189
1190 PCRE_AUTO_CALLOUT
1191
1192 If this bit is set, pcre_compile() automatically inserts callout items,
1193 all with number 255, before each pattern item. For discussion of the
1194 callout facility, see the pcrecallout documentation.
1195
1196 PCRE_BSR_ANYCRLF
1197 PCRE_BSR_UNICODE
1198
1199 These options (which are mutually exclusive) control what the \R escape
1200 sequence matches. The choice is either to match only CR, LF, or CRLF,
1201 or to match any Unicode newline sequence. The default is specified when
1202 PCRE is built. It can be overridden from within the pattern, or by set-
1203 ting an option when a compiled pattern is matched.
1204
1205 PCRE_CASELESS
1206
1207 If this bit is set, letters in the pattern match both upper and lower
1208 case letters. It is equivalent to Perl's /i option, and it can be
1209 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1210 always understands the concept of case for characters whose values are
1211 less than 128, so caseless matching is always possible. For characters
1212 with higher values, the concept of case is supported if PCRE is com-
1213 piled with Unicode property support, but not otherwise. If you want to
1214 use caseless matching for characters 128 and above, you must ensure
1215 that PCRE is compiled with Unicode property support as well as with
1216 UTF-8 support.
1217
1218 PCRE_DOLLAR_ENDONLY
1219
1220 If this bit is set, a dollar metacharacter in the pattern matches only
1221 at the end of the subject string. Without this option, a dollar also
1222 matches immediately before a newline at the end of the string (but not
1223 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1224 if PCRE_MULTILINE is set. There is no equivalent to this option in
1225 Perl, and no way to set it within a pattern.
1226
1227 PCRE_DOTALL
1228
1229 If this bit is set, a dot metacharater in the pattern matches all char-
1230 acters, including those that indicate newline. Without it, a dot does
1231 not match when the current position is at a newline. This option is
1232 equivalent to Perl's /s option, and it can be changed within a pattern
1233 by a (?s) option setting. A negative class such as [^a] always matches
1234 newline characters, independent of the setting of this option.
1235
1236 PCRE_DUPNAMES
1237
1238 If this bit is set, names used to identify capturing subpatterns need
1239 not be unique. This can be helpful for certain types of pattern when it
1240 is known that only one instance of the named subpattern can ever be
1241 matched. There are more details of named subpatterns below; see also
1242 the pcrepattern documentation.
1243
1244 PCRE_EXTENDED
1245
1246 If this bit is set, whitespace data characters in the pattern are
1247 totally ignored except when escaped or inside a character class. White-
1248 space does not include the VT character (code 11). In addition, charac-
1249 ters between an unescaped # outside a character class and the next new-
1250 line, inclusive, are also ignored. This is equivalent to Perl's /x
1251 option, and it can be changed within a pattern by a (?x) option set-
1252 ting.
1253
1254 This option makes it possible to include comments inside complicated
1255 patterns. Note, however, that this applies only to data characters.
1256 Whitespace characters may never appear within special character
1257 sequences in a pattern, for example within the sequence (?( which
1258 introduces a conditional subpattern.
1259
1260 PCRE_EXTRA
1261
1262 This option was invented in order to turn on additional functionality
1263 of PCRE that is incompatible with Perl, but it is currently of very
1264 little use. When set, any backslash in a pattern that is followed by a
1265 letter that has no special meaning causes an error, thus reserving
1266 these combinations for future expansion. By default, as in Perl, a
1267 backslash followed by a letter with no special meaning is treated as a
1268 literal. (Perl can, however, be persuaded to give a warning for this.)
1269 There are at present no other features controlled by this option. It
1270 can also be set by a (?X) option setting within a pattern.
1271
1272 PCRE_FIRSTLINE
1273
1274 If this option is set, an unanchored pattern is required to match
1275 before or at the first newline in the subject string, though the
1276 matched text may continue over the newline.
1277
1278 PCRE_JAVASCRIPT_COMPAT
1279
1280 If this option is set, PCRE's behaviour is changed in some ways so that
1281 it is compatible with JavaScript rather than Perl. The changes are as
1282 follows:
1283
1284 (1) A lone closing square bracket in a pattern causes a compile-time
1285 error, because this is illegal in JavaScript (by default it is treated
1286 as a data character). Thus, the pattern AB]CD becomes illegal when this
1287 option is set.
1288
1289 (2) At run time, a back reference to an unset subpattern group matches
1290 an empty string (by default this causes the current matching alterna-
1291 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1292 set (assuming it can find an "a" in the subject), whereas it fails by
1293 default, for Perl compatibility.
1294
1295 PCRE_MULTILINE
1296
1297 By default, PCRE treats the subject string as consisting of a single
1298 line of characters (even if it actually contains newlines). The "start
1299 of line" metacharacter (^) matches only at the start of the string,
1300 while the "end of line" metacharacter ($) matches only at the end of
1301 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1302 is set). This is the same as Perl.
1303
1304 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1305 constructs match immediately following or immediately before internal
1306 newlines in the subject string, respectively, as well as at the very
1307 start and end. This is equivalent to Perl's /m option, and it can be
1308 changed within a pattern by a (?m) option setting. If there are no new-
1309 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1310 setting PCRE_MULTILINE has no effect.
1311
1312 PCRE_NEWLINE_CR
1313 PCRE_NEWLINE_LF
1314 PCRE_NEWLINE_CRLF
1315 PCRE_NEWLINE_ANYCRLF
1316 PCRE_NEWLINE_ANY
1317
1318 These options override the default newline definition that was chosen
1319 when PCRE was built. Setting the first or the second specifies that a
1320 newline is indicated by a single character (CR or LF, respectively).
1321 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1322 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1323 that any of the three preceding sequences should be recognized. Setting
1324 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1325 recognized. The Unicode newline sequences are the three just mentioned,
1326 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1327 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1328 (paragraph separator, U+2029). The last two are recognized only in
1329 UTF-8 mode.
1330
1331 The newline setting in the options word uses three bits that are
1332 treated as a number, giving eight possibilities. Currently only six are
1333 used (default plus the five values above). This means that if you set
1334 more than one newline option, the combination may or may not be sensi-
1335 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1336 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1337 cause an error.
1338
1339 The only time that a line break is specially recognized when compiling
1340 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1341 character class is encountered. This indicates a comment that lasts
1342 until after the next line break sequence. In other circumstances, line
1343 break sequences are treated as literal data, except that in
1344 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1345 and are therefore ignored.
1346
1347 The newline option that is set at compile time becomes the default that
1348 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1349
1350 PCRE_NO_AUTO_CAPTURE
1351
1352 If this option is set, it disables the use of numbered capturing paren-
1353 theses in the pattern. Any opening parenthesis that is not followed by
1354 ? behaves as if it were followed by ?: but named parentheses can still
1355 be used for capturing (and they acquire numbers in the usual way).
1356 There is no equivalent of this option in Perl.
1357
1358 PCRE_UNGREEDY
1359
1360 This option inverts the "greediness" of the quantifiers so that they
1361 are not greedy by default, but become greedy if followed by "?". It is
1362 not compatible with Perl. It can also be set by a (?U) option setting
1363 within the pattern.
1364
1365 PCRE_UTF8
1366
1367 This option causes PCRE to regard both the pattern and the subject as
1368 strings of UTF-8 characters instead of single-byte character strings.
1369 However, it is available only when PCRE is built to include UTF-8 sup-
1370 port. If not, the use of this option provokes an error. Details of how
1371 this option changes the behaviour of PCRE are given in the section on
1372 UTF-8 support in the main pcre page.
1373
1374 PCRE_NO_UTF8_CHECK
1375
1376 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1377 automatically checked. There is a discussion about the validity of
1378 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1379 bytes is found, pcre_compile() returns an error. If you already know
1380 that your pattern is valid, and you want to skip this check for perfor-
1381 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1382 set, the effect of passing an invalid UTF-8 string as a pattern is
1383 undefined. It may cause your program to crash. Note that this option
1384 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1385 UTF-8 validity checking of subject strings.
1386
1387
1388 COMPILATION ERROR CODES
1389
1390 The following table lists the error codes than may be returned by
1391 pcre_compile2(), along with the error messages that may be returned by
1392 both compiling functions. As PCRE has developed, some error codes have
1393 fallen out of use. To avoid confusion, they have not been re-used.
1394
1395 0 no error
1396 1 \ at end of pattern
1397 2 \c at end of pattern
1398 3 unrecognized character follows \
1399 4 numbers out of order in {} quantifier
1400 5 number too big in {} quantifier
1401 6 missing terminating ] for character class
1402 7 invalid escape sequence in character class
1403 8 range out of order in character class
1404 9 nothing to repeat
1405 10 [this code is not in use]
1406 11 internal error: unexpected repeat
1407 12 unrecognized character after (? or (?-
1408 13 POSIX named classes are supported only within a class
1409 14 missing )
1410 15 reference to non-existent subpattern
1411 16 erroffset passed as NULL
1412 17 unknown option bit(s) set
1413 18 missing ) after comment
1414 19 [this code is not in use]
1415 20 regular expression is too large
1416 21 failed to get memory
1417 22 unmatched parentheses
1418 23 internal error: code overflow
1419 24 unrecognized character after (?<
1420 25 lookbehind assertion is not fixed length
1421 26 malformed number or name after (?(
1422 27 conditional group contains more than two branches
1423 28 assertion expected after (?(
1424 29 (?R or (?[+-]digits must be followed by )
1425 30 unknown POSIX class name
1426 31 POSIX collating elements are not supported
1427 32 this version of PCRE is not compiled with PCRE_UTF8 support
1428 33 [this code is not in use]
1429 34 character value in \x{...} sequence is too large
1430 35 invalid condition (?(0)
1431 36 \C not allowed in lookbehind assertion
1432 37 PCRE does not support \L, \l, \N, \U, or \u
1433 38 number after (?C is > 255
1434 39 closing ) for (?C expected
1435 40 recursive call could loop indefinitely
1436 41 unrecognized character after (?P
1437 42 syntax error in subpattern name (missing terminator)
1438 43 two named subpatterns have the same name
1439 44 invalid UTF-8 string
1440 45 support for \P, \p, and \X has not been compiled
1441 46 malformed \P or \p sequence
1442 47 unknown property name after \P or \p
1443 48 subpattern name is too long (maximum 32 characters)
1444 49 too many named subpatterns (maximum 10000)
1445 50 [this code is not in use]
1446 51 octal value is greater than \377 (not in UTF-8 mode)
1447 52 internal error: overran compiling workspace
1448 53 internal error: previously-checked referenced subpattern not
1449 found
1450 54 DEFINE group contains more than one branch
1451 55 repeating a DEFINE group is not allowed
1452 56 inconsistent NEWLINE options
1453 57 \g is not followed by a braced, angle-bracketed, or quoted
1454 name/number or by a plain number
1455 58 a numbered reference must not be zero
1456 59 (*VERB) with an argument is not supported
1457 60 (*VERB) not recognized
1458 61 number is too big
1459 62 subpattern name expected
1460 63 digit expected after (?+
1461 64 ] is an invalid data character in JavaScript compatibility mode
1462
1463 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1464 values may be used if the limits were changed when PCRE was built.
1465
1466
1467 STUDYING A PATTERN
1468
1469 pcre_extra *pcre_study(const pcre *code, int options
1470 const char **errptr);
1471
1472 If a compiled pattern is going to be used several times, it is worth
1473 spending more time analyzing it in order to speed up the time taken for
1474 matching. The function pcre_study() takes a pointer to a compiled pat-
1475 tern as its first argument. If studying the pattern produces additional
1476 information that will help speed up matching, pcre_study() returns a
1477 pointer to a pcre_extra block, in which the study_data field points to
1478 the results of the study.
1479
1480 The returned value from pcre_study() can be passed directly to
1481 pcre_exec(). However, a pcre_extra block also contains other fields
1482 that can be set by the caller before the block is passed; these are
1483 described below in the section on matching a pattern.
1484
1485 If studying the pattern does not produce any additional information
1486 pcre_study() returns NULL. In that circumstance, if the calling program
1487 wants to pass any of the other fields to pcre_exec(), it must set up
1488 its own pcre_extra block.
1489
1490 The second argument of pcre_study() contains option bits. At present,
1491 no options are defined, and this argument should always be zero.
1492
1493 The third argument for pcre_study() is a pointer for an error message.
1494 If studying succeeds (even if no data is returned), the variable it
1495 points to is set to NULL. Otherwise it is set to point to a textual
1496 error message. This is a static string that is part of the library. You
1497 must not try to free it. You should test the error pointer for NULL
1498 after calling pcre_study(), to be sure that it has run successfully.
1499
1500 This is a typical call to pcre_study():
1501
1502 pcre_extra *pe;
1503 pe = pcre_study(
1504 re, /* result of pcre_compile() */
1505 0, /* no options exist */
1506 &error); /* set to NULL or points to a message */
1507
1508 At present, studying a pattern is useful only for non-anchored patterns
1509 that do not have a single fixed starting character. A bitmap of possi-
1510 ble starting bytes is created.
1511
1512
1513 LOCALE SUPPORT
1514
1515 PCRE handles caseless matching, and determines whether characters are
1516 letters, digits, or whatever, by reference to a set of tables, indexed
1517 by character value. When running in UTF-8 mode, this applies only to
1518 characters with codes less than 128. Higher-valued codes never match
1519 escapes such as \w or \d, but can be tested with \p if PCRE is built
1520 with Unicode character property support. The use of locales with Uni-
1521 code is discouraged. If you are handling characters with codes greater
1522 than 128, you should either use UTF-8 and Unicode, or use locales, but
1523 not try to mix the two.
1524
1525 PCRE contains an internal set of tables that are used when the final
1526 argument of pcre_compile() is NULL. These are sufficient for many
1527 applications. Normally, the internal tables recognize only ASCII char-
1528 acters. However, when PCRE is built, it is possible to cause the inter-
1529 nal tables to be rebuilt in the default "C" locale of the local system,
1530 which may cause them to be different.
1531
1532 The internal tables can always be overridden by tables supplied by the
1533 application that calls PCRE. These may be created in a different locale
1534 from the default. As more and more applications change to using Uni-
1535 code, the need for this locale support is expected to die away.
1536
1537 External tables are built by calling the pcre_maketables() function,
1538 which has no arguments, in the relevant locale. The result can then be
1539 passed to pcre_compile() or pcre_exec() as often as necessary. For
1540 example, to build and use tables that are appropriate for the French
1541 locale (where accented characters with values greater than 128 are
1542 treated as letters), the following code could be used:
1543
1544 setlocale(LC_CTYPE, "fr_FR");
1545 tables = pcre_maketables();
1546 re = pcre_compile(..., tables);
1547
1548 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1549 if you are using Windows, the name for the French locale is "french".
1550
1551 When pcre_maketables() runs, the tables are built in memory that is
1552 obtained via pcre_malloc. It is the caller's responsibility to ensure
1553 that the memory containing the tables remains available for as long as
1554 it is needed.
1555
1556 The pointer that is passed to pcre_compile() is saved with the compiled
1557 pattern, and the same tables are used via this pointer by pcre_study()
1558 and normally also by pcre_exec(). Thus, by default, for any single pat-
1559 tern, compilation, studying and matching all happen in the same locale,
1560 but different patterns can be compiled in different locales.
1561
1562 It is possible to pass a table pointer or NULL (indicating the use of
1563 the internal tables) to pcre_exec(). Although not intended for this
1564 purpose, this facility could be used to match a pattern in a different
1565 locale from the one in which it was compiled. Passing table pointers at
1566 run time is discussed below in the section on matching a pattern.
1567
1568
1569 INFORMATION ABOUT A PATTERN
1570
1571 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1572 int what, void *where);
1573
1574 The pcre_fullinfo() function returns information about a compiled pat-
1575 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1576 less retained for backwards compability (and is documented below).
1577
1578 The first argument for pcre_fullinfo() is a pointer to the compiled
1579 pattern. The second argument is the result of pcre_study(), or NULL if
1580 the pattern was not studied. The third argument specifies which piece
1581 of information is required, and the fourth argument is a pointer to a
1582 variable to receive the data. The yield of the function is zero for
1583 success, or one of the following negative numbers:
1584
1585 PCRE_ERROR_NULL the argument code was NULL
1586 the argument where was NULL
1587 PCRE_ERROR_BADMAGIC the "magic number" was not found
1588 PCRE_ERROR_BADOPTION the value of what was invalid
1589
1590 The "magic number" is placed at the start of each compiled pattern as
1591 an simple check against passing an arbitrary memory pointer. Here is a
1592 typical call of pcre_fullinfo(), to obtain the length of the compiled
1593 pattern:
1594
1595 int rc;
1596 size_t length;
1597 rc = pcre_fullinfo(
1598 re, /* result of pcre_compile() */
1599 pe, /* result of pcre_study(), or NULL */
1600 PCRE_INFO_SIZE, /* what is required */
1601 &length); /* where to put the data */
1602
1603 The possible values for the third argument are defined in pcre.h, and
1604 are as follows:
1605
1606 PCRE_INFO_BACKREFMAX
1607
1608 Return the number of the highest back reference in the pattern. The
1609 fourth argument should point to an int variable. Zero is returned if
1610 there are no back references.
1611
1612 PCRE_INFO_CAPTURECOUNT
1613
1614 Return the number of capturing subpatterns in the pattern. The fourth
1615 argument should point to an int variable.
1616
1617 PCRE_INFO_DEFAULT_TABLES
1618
1619 Return a pointer to the internal default character tables within PCRE.
1620 The fourth argument should point to an unsigned char * variable. This
1621 information call is provided for internal use by the pcre_study() func-
1622 tion. External callers can cause PCRE to use its internal tables by
1623 passing a NULL table pointer.
1624
1625 PCRE_INFO_FIRSTBYTE
1626
1627 Return information about the first byte of any matched string, for a
1628 non-anchored pattern. The fourth argument should point to an int vari-
1629 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1630 is still recognized for backwards compatibility.)
1631
1632 If there is a fixed first byte, for example, from a pattern such as
1633 (cat|cow|coyote), its value is returned. Otherwise, if either
1634
1635 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1636 branch starts with "^", or
1637
1638 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1639 set (if it were set, the pattern would be anchored),
1640
1641 -1 is returned, indicating that the pattern matches only at the start
1642 of a subject string or after any newline within the string. Otherwise
1643 -2 is returned. For anchored patterns, -2 is returned.
1644
1645 PCRE_INFO_FIRSTTABLE
1646
1647 If the pattern was studied, and this resulted in the construction of a
1648 256-bit table indicating a fixed set of bytes for the first byte in any
1649 matching string, a pointer to the table is returned. Otherwise NULL is
1650 returned. The fourth argument should point to an unsigned char * vari-
1651 able.
1652
1653 PCRE_INFO_HASCRORLF
1654
1655 Return 1 if the pattern contains any explicit matches for CR or LF
1656 characters, otherwise 0. The fourth argument should point to an int
1657 variable. An explicit match is either a literal CR or LF character, or
1658 \r or \n.
1659
1660 PCRE_INFO_JCHANGED
1661
1662 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1663 otherwise 0. The fourth argument should point to an int variable. (?J)
1664 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1665
1666 PCRE_INFO_LASTLITERAL
1667
1668 Return the value of the rightmost literal byte that must exist in any
1669 matched string, other than at its start, if such a byte has been
1670 recorded. The fourth argument should point to an int variable. If there
1671 is no such byte, -1 is returned. For anchored patterns, a last literal
1672 byte is recorded only if it follows something of variable length. For
1673 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1674 /^a\dz\d/ the returned value is -1.
1675
1676 PCRE_INFO_NAMECOUNT
1677 PCRE_INFO_NAMEENTRYSIZE
1678 PCRE_INFO_NAMETABLE
1679
1680 PCRE supports the use of named as well as numbered capturing parenthe-
1681 ses. The names are just an additional way of identifying the parenthe-
1682 ses, which still acquire numbers. Several convenience functions such as
1683 pcre_get_named_substring() are provided for extracting captured sub-
1684 strings by name. It is also possible to extract the data directly, by
1685 first converting the name to a number in order to access the correct
1686 pointers in the output vector (described with pcre_exec() below). To do
1687 the conversion, you need to use the name-to-number map, which is
1688 described by these three values.
1689
1690 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1691 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1692 of each entry; both of these return an int value. The entry size
1693 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1694 a pointer to the first entry of the table (a pointer to char). The
1695 first two bytes of each entry are the number of the capturing parenthe-
1696 sis, most significant byte first. The rest of the entry is the corre-
1697 sponding name, zero terminated. The names are in alphabetical order.
1698 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1699 theses numbers. For example, consider the following pattern (assume
1700 PCRE_EXTENDED is set, so white space - including newlines - is
1701 ignored):
1702
1703 (?<date> (?<year>(\d\d)?\d\d) -
1704 (?<month>\d\d) - (?<day>\d\d) )
1705
1706 There are four named subpatterns, so the table has four entries, and
1707 each entry in the table is eight bytes long. The table is as follows,
1708 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1709 as ??:
1710
1711 00 01 d a t e 00 ??
1712 00 05 d a y 00 ?? ??
1713 00 04 m o n t h 00
1714 00 02 y e a r 00 ??
1715
1716 When writing code to extract data from named subpatterns using the
1717 name-to-number map, remember that the length of the entries is likely
1718 to be different for each compiled pattern.
1719
1720 PCRE_INFO_OKPARTIAL
1721
1722 Return 1 if the pattern can be used for partial matching, otherwise 0.
1723 The fourth argument should point to an int variable. The pcrepartial
1724 documentation lists the restrictions that apply to patterns when par-
1725 tial matching is used.
1726
1727 PCRE_INFO_OPTIONS
1728
1729 Return a copy of the options with which the pattern was compiled. The
1730 fourth argument should point to an unsigned long int variable. These
1731 option bits are those specified in the call to pcre_compile(), modified
1732 by any top-level option settings at the start of the pattern itself. In
1733 other words, they are the options that will be in force when matching
1734 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1735 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1736 and PCRE_EXTENDED.
1737
1738 A pattern is automatically anchored by PCRE if all of its top-level
1739 alternatives begin with one of the following:
1740
1741 ^ unless PCRE_MULTILINE is set
1742 \A always
1743 \G always
1744 .* if PCRE_DOTALL is set and there are no back
1745 references to the subpattern in which .* appears
1746
1747 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1748 by pcre_fullinfo().
1749
1750 PCRE_INFO_SIZE
1751
1752 Return the size of the compiled pattern, that is, the value that was
1753 passed as the argument to pcre_malloc() when PCRE was getting memory in
1754 which to place the compiled data. The fourth argument should point to a
1755 size_t variable.
1756
1757 PCRE_INFO_STUDYSIZE
1758
1759 Return the size of the data block pointed to by the study_data field in
1760 a pcre_extra block. That is, it is the value that was passed to
1761 pcre_malloc() when PCRE was getting memory into which to place the data
1762 created by pcre_study(). The fourth argument should point to a size_t
1763 variable.
1764
1765
1766 OBSOLETE INFO FUNCTION
1767
1768 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1769
1770 The pcre_info() function is now obsolete because its interface is too
1771 restrictive to return all the available data about a compiled pattern.
1772 New programs should use pcre_fullinfo() instead. The yield of
1773 pcre_info() is the number of capturing subpatterns, or one of the fol-
1774 lowing negative numbers:
1775
1776 PCRE_ERROR_NULL the argument code was NULL
1777 PCRE_ERROR_BADMAGIC the "magic number" was not found
1778
1779 If the optptr argument is not NULL, a copy of the options with which
1780 the pattern was compiled is placed in the integer it points to (see
1781 PCRE_INFO_OPTIONS above).
1782
1783 If the pattern is not anchored and the firstcharptr argument is not
1784 NULL, it is used to pass back information about the first character of
1785 any matched string (see PCRE_INFO_FIRSTBYTE above).
1786
1787
1788 REFERENCE COUNTS
1789
1790 int pcre_refcount(pcre *code, int adjust);
1791
1792 The pcre_refcount() function is used to maintain a reference count in
1793 the data block that contains a compiled pattern. It is provided for the
1794 benefit of applications that operate in an object-oriented manner,
1795 where different parts of the application may be using the same compiled
1796 pattern, but you want to free the block when they are all done.
1797
1798 When a pattern is compiled, the reference count field is initialized to
1799 zero. It is changed only by calling this function, whose action is to
1800 add the adjust value (which may be positive or negative) to it. The
1801 yield of the function is the new value. However, the value of the count
1802 is constrained to lie between 0 and 65535, inclusive. If the new value
1803 is outside these limits, it is forced to the appropriate limit value.
1804
1805 Except when it is zero, the reference count is not correctly preserved
1806 if a pattern is compiled on one host and then transferred to a host
1807 whose byte-order is different. (This seems a highly unlikely scenario.)
1808
1809
1810 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1811
1812 int pcre_exec(const pcre *code, const pcre_extra *extra,
1813 const char *subject, int length, int startoffset,
1814 int options, int *ovector, int ovecsize);
1815
1816 The function pcre_exec() is called to match a subject string against a
1817 compiled pattern, which is passed in the code argument. If the pattern
1818 has been studied, the result of the study should be passed in the extra
1819 argument. This function is the main matching facility of the library,
1820 and it operates in a Perl-like manner. For specialist use there is also
1821 an alternative matching function, which is described below in the sec-
1822 tion about the pcre_dfa_exec() function.
1823
1824 In most applications, the pattern will have been compiled (and option-
1825 ally studied) in the same process that calls pcre_exec(). However, it
1826 is possible to save compiled patterns and study data, and then use them
1827 later in different processes, possibly even on different hosts. For a
1828 discussion about this, see the pcreprecompile documentation.
1829
1830 Here is an example of a simple call to pcre_exec():
1831
1832 int rc;
1833 int ovector[30];
1834 rc = pcre_exec(
1835 re, /* result of pcre_compile() */
1836 NULL, /* we didn't study the pattern */
1837 "some string", /* the subject string */
1838 11, /* the length of the subject string */
1839 0, /* start at offset 0 in the subject */
1840 0, /* default options */
1841 ovector, /* vector of integers for substring information */
1842 30); /* number of elements (NOT size in bytes) */
1843
1844 Extra data for pcre_exec()
1845
1846 If the extra argument is not NULL, it must point to a pcre_extra data
1847 block. The pcre_study() function returns such a block (when it doesn't
1848 return NULL), but you can also create one for yourself, and pass addi-
1849 tional information in it. The pcre_extra block contains the following
1850 fields (not necessarily in this order):
1851
1852 unsigned long int flags;
1853 void *study_data;
1854 unsigned long int match_limit;
1855 unsigned long int match_limit_recursion;
1856 void *callout_data;
1857 const unsigned char *tables;
1858
1859 The flags field is a bitmap that specifies which of the other fields
1860 are set. The flag bits are:
1861
1862 PCRE_EXTRA_STUDY_DATA
1863 PCRE_EXTRA_MATCH_LIMIT
1864 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1865 PCRE_EXTRA_CALLOUT_DATA
1866 PCRE_EXTRA_TABLES
1867
1868 Other flag bits should be set to zero. The study_data field is set in
1869 the pcre_extra block that is returned by pcre_study(), together with
1870 the appropriate flag bit. You should not set this yourself, but you may
1871 add to the block by setting the other fields and their corresponding
1872 flag bits.
1873
1874 The match_limit field provides a means of preventing PCRE from using up
1875 a vast amount of resources when running patterns that are not going to
1876 match, but which have a very large number of possibilities in their
1877 search trees. The classic example is the use of nested unlimited
1878 repeats.
1879
1880 Internally, PCRE uses a function called match() which it calls repeat-
1881 edly (sometimes recursively). The limit set by match_limit is imposed
1882 on the number of times this function is called during a match, which
1883 has the effect of limiting the amount of backtracking that can take
1884 place. For patterns that are not anchored, the count restarts from zero
1885 for each position in the subject string.
1886
1887 The default value for the limit can be set when PCRE is built; the
1888 default default is 10 million, which handles all but the most extreme
1889 cases. You can override the default by suppling pcre_exec() with a
1890 pcre_extra block in which match_limit is set, and
1891 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1892 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1893
1894 The match_limit_recursion field is similar to match_limit, but instead
1895 of limiting the total number of times that match() is called, it limits
1896 the depth of recursion. The recursion depth is a smaller number than
1897 the total number of calls, because not all calls to match() are recur-
1898 sive. This limit is of use only if it is set smaller than match_limit.
1899
1900 Limiting the recursion depth limits the amount of stack that can be
1901 used, or, when PCRE has been compiled to use memory on the heap instead
1902 of the stack, the amount of heap memory that can be used.
1903
1904 The default value for match_limit_recursion can be set when PCRE is
1905 built; the default default is the same value as the default for
1906 match_limit. You can override the default by suppling pcre_exec() with
1907 a pcre_extra block in which match_limit_recursion is set, and
1908 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1909 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1910
1911 The pcre_callout field is used in conjunction with the "callout" fea-
1912 ture, which is described in the pcrecallout documentation.
1913
1914 The tables field is used to pass a character tables pointer to
1915 pcre_exec(); this overrides the value that is stored with the compiled
1916 pattern. A non-NULL value is stored with the compiled pattern only if
1917 custom tables were supplied to pcre_compile() via its tableptr argu-
1918 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1919 PCRE's internal tables to be used. This facility is helpful when re-
1920 using patterns that have been saved after compiling with an external
1921 set of tables, because the external tables might be at a different
1922 address when pcre_exec() is called. See the pcreprecompile documenta-
1923 tion for a discussion of saving compiled patterns for later use.
1924
1925 Option bits for pcre_exec()
1926
1927 The unused bits of the options argument for pcre_exec() must be zero.
1928 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1929 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_START_OPTIMIZE,
1930 PCRE_NO_UTF8_CHECK and PCRE_PARTIAL.
1931
1932 PCRE_ANCHORED
1933
1934 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1935 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1936 turned out to be anchored by virtue of its contents, it cannot be made
1937 unachored at matching time.
1938
1939 PCRE_BSR_ANYCRLF
1940 PCRE_BSR_UNICODE
1941
1942 These options (which are mutually exclusive) control what the \R escape
1943 sequence matches. The choice is either to match only CR, LF, or CRLF,
1944 or to match any Unicode newline sequence. These options override the
1945 choice that was made or defaulted when the pattern was compiled.
1946
1947 PCRE_NEWLINE_CR
1948 PCRE_NEWLINE_LF
1949 PCRE_NEWLINE_CRLF
1950 PCRE_NEWLINE_ANYCRLF
1951 PCRE_NEWLINE_ANY
1952
1953 These options override the newline definition that was chosen or
1954 defaulted when the pattern was compiled. For details, see the descrip-
1955 tion of pcre_compile() above. During matching, the newline choice
1956 affects the behaviour of the dot, circumflex, and dollar metacharac-
1957 ters. It may also alter the way the match position is advanced after a
1958 match failure for an unanchored pattern.
1959
1960 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1961 set, and a match attempt for an unanchored pattern fails when the cur-
1962 rent position is at a CRLF sequence, and the pattern contains no
1963 explicit matches for CR or LF characters, the match position is
1964 advanced by two characters instead of one, in other words, to after the
1965 CRLF.
1966
1967 The above rule is a compromise that makes the most common cases work as
1968 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1969 option is not set), it does not match the string "\r\nA" because, after
1970 failing at the start, it skips both the CR and the LF before retrying.
1971 However, the pattern [\r\n]A does match that string, because it con-
1972 tains an explicit CR or LF reference, and so advances only by one char-
1973 acter after the first failure.
1974
1975 An explicit match for CR of LF is either a literal appearance of one of
1976 those characters, or one of the \r or \n escape sequences. Implicit
1977 matches such as [^X] do not count, nor does \s (which includes CR and
1978 LF in the characters that it matches).
1979
1980 Notwithstanding the above, anomalous effects may still occur when CRLF
1981 is a valid newline sequence and explicit \r or \n escapes appear in the
1982 pattern.
1983
1984 PCRE_NOTBOL
1985
1986 This option specifies that first character of the subject string is not
1987 the beginning of a line, so the circumflex metacharacter should not
1988 match before it. Setting this without PCRE_MULTILINE (at compile time)
1989 causes circumflex never to match. This option affects only the behav-
1990 iour of the circumflex metacharacter. It does not affect \A.
1991
1992 PCRE_NOTEOL
1993
1994 This option specifies that the end of the subject string is not the end
1995 of a line, so the dollar metacharacter should not match it nor (except
1996 in multiline mode) a newline immediately before it. Setting this with-
1997 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1998 option affects only the behaviour of the dollar metacharacter. It does
1999 not affect \Z or \z.
2000
2001 PCRE_NOTEMPTY
2002
2003 An empty string is not considered to be a valid match if this option is
2004 set. If there are alternatives in the pattern, they are tried. If all
2005 the alternatives match the empty string, the entire match fails. For
2006 example, if the pattern
2007
2008 a?b?
2009
2010 is applied to a string not beginning with "a" or "b", it matches the
2011 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2012 match is not valid, so PCRE searches further into the string for occur-
2013 rences of "a" or "b".
2014
2015 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
2016 cial case of a pattern match of the empty string within its split()
2017 function, and when using the /g modifier. It is possible to emulate
2018 Perl's behaviour after matching a null string by first trying the match
2019 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
2020 if that fails by advancing the starting offset (see below) and trying
2021 an ordinary match again. There is some code that demonstrates how to do
2022 this in the pcredemo.c sample program.
2023
2024 PCRE_NO_START_OPTIMIZE
2025
2026 There are a number of optimizations that pcre_exec() uses at the start
2027 of a match, in order to speed up the process. For example, if it is
2028 known that a match must start with a specific character, it searches
2029 the subject for that character, and fails immediately if it cannot find
2030 it, without actually running the main matching function. When callouts
2031 are in use, these optimizations can cause them to be skipped. This
2032 option disables the "start-up" optimizations, causing performance to
2033 suffer, but ensuring that the callouts do occur.
2034
2035 PCRE_NO_UTF8_CHECK
2036
2037 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2038 UTF-8 string is automatically checked when pcre_exec() is subsequently
2039 called. The value of startoffset is also checked to ensure that it
2040 points to the start of a UTF-8 character. There is a discussion about
2041 the validity of UTF-8 strings in the section on UTF-8 support in the
2042 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2043 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
2044 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
2045
2046 If you already know that your subject is valid, and you want to skip
2047 these checks for performance reasons, you can set the
2048 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2049 do this for the second and subsequent calls to pcre_exec() if you are
2050 making repeated calls to find all the matches in a single subject
2051 string. However, you should be sure that the value of startoffset
2052 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
2053 set, the effect of passing an invalid UTF-8 string as a subject, or a
2054 value of startoffset that does not point to the start of a UTF-8 char-
2055 acter, is undefined. Your program may crash.
2056
2057 PCRE_PARTIAL
2058
2059 This option turns on the partial matching feature. If the subject
2060 string fails to match the pattern, but at some point during the match-
2061 ing process the end of the subject was reached (that is, the subject
2062 partially matches the pattern and the failure to match occurred only
2063 because there were not enough subject characters), pcre_exec() returns
2064 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
2065 used, there are restrictions on what may appear in the pattern. These
2066 are discussed in the pcrepartial documentation.
2067
2068 The string to be matched by pcre_exec()
2069
2070 The subject string is passed to pcre_exec() as a pointer in subject, a
2071 length (in bytes) in length, and a starting byte offset in startoffset.
2072 In UTF-8 mode, the byte offset must point to the start of a UTF-8 char-
2073 acter. Unlike the pattern string, the subject may contain binary zero
2074 bytes. When the starting offset is zero, the search for a match starts
2075 at the beginning of the subject, and this is by far the most common
2076 case.
2077
2078 A non-zero starting offset is useful when searching for another match
2079 in the same subject by calling pcre_exec() again after a previous suc-
2080 cess. Setting startoffset differs from just passing over a shortened
2081 string and setting PCRE_NOTBOL in the case of a pattern that begins
2082 with any kind of lookbehind. For example, consider the pattern
2083
2084 \Biss\B
2085
2086 which finds occurrences of "iss" in the middle of words. (\B matches
2087 only if the current position in the subject is not a word boundary.)
2088 When applied to the string "Mississipi" the first call to pcre_exec()
2089 finds the first occurrence. If pcre_exec() is called again with just
2090 the remainder of the subject, namely "issipi", it does not match,
2091 because \B is always false at the start of the subject, which is deemed
2092 to be a word boundary. However, if pcre_exec() is passed the entire
2093 string again, but with startoffset set to 4, it finds the second occur-
2094 rence of "iss" because it is able to look behind the starting point to
2095 discover that it is preceded by a letter.
2096
2097 If a non-zero starting offset is passed when the pattern is anchored,
2098 one attempt to match at the given offset is made. This can only succeed
2099 if the pattern does not require the match to be at the start of the
2100 subject.
2101
2102 How pcre_exec() returns captured substrings
2103
2104 In general, a pattern matches a certain portion of the subject, and in
2105 addition, further substrings from the subject may be picked out by
2106 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2107 this is called "capturing" in what follows, and the phrase "capturing
2108 subpattern" is used for a fragment of a pattern that picks out a sub-
2109 string. PCRE supports several other kinds of parenthesized subpattern
2110 that do not cause substrings to be captured.
2111
2112 Captured substrings are returned to the caller via a vector of integers
2113 whose address is passed in ovector. The number of elements in the vec-
2114 tor is passed in ovecsize, which must be a non-negative number. Note:
2115 this argument is NOT the size of ovector in bytes.
2116
2117 The first two-thirds of the vector is used to pass back captured sub-
2118 strings, each substring using a pair of integers. The remaining third
2119 of the vector is used as workspace by pcre_exec() while matching cap-
2120 turing subpatterns, and is not available for passing back information.
2121 The number passed in ovecsize should always be a multiple of three. If
2122 it is not, it is rounded down.
2123
2124 When a match is successful, information about captured substrings is
2125 returned in pairs of integers, starting at the beginning of ovector,
2126 and continuing up to two-thirds of its length at the most. The first
2127 element of each pair is set to the byte offset of the first character
2128 in a substring, and the second is set to the byte offset of the first
2129 character after the end of a substring. Note: these values are always
2130 byte offsets, even in UTF-8 mode. They are not character counts.
2131
2132 The first pair of integers, ovector[0] and ovector[1], identify the
2133 portion of the subject string matched by the entire pattern. The next
2134 pair is used for the first capturing subpattern, and so on. The value
2135 returned by pcre_exec() is one more than the highest numbered pair that
2136 has been set. For example, if two substrings have been captured, the
2137 returned value is 3. If there are no capturing subpatterns, the return
2138 value from a successful match is 1, indicating that just the first pair
2139 of offsets has been set.
2140
2141 If a capturing subpattern is matched repeatedly, it is the last portion
2142 of the string that it matched that is returned.
2143
2144 If the vector is too small to hold all the captured substring offsets,
2145 it is used as far as possible (up to two-thirds of its length), and the
2146 function returns a value of zero. If the substring offsets are not of
2147 interest, pcre_exec() may be called with ovector passed as NULL and
2148 ovecsize as zero. However, if the pattern contains back references and
2149 the ovector is not big enough to remember the related substrings, PCRE
2150 has to get additional memory for use during matching. Thus it is usu-
2151 ally advisable to supply an ovector.
2152
2153 The pcre_info() function can be used to find out how many capturing
2154 subpatterns there are in a compiled pattern. The smallest size for
2155 ovector that will allow for n captured substrings, in addition to the
2156 offsets of the substring matched by the whole pattern, is (n+1)*3.
2157
2158 It is possible for capturing subpattern number n+1 to match some part
2159 of the subject when subpattern n has not been used at all. For example,
2160 if the string "abc" is matched against the pattern (a|(z))(bc) the
2161 return from the function is 4, and subpatterns 1 and 3 are matched, but
2162 2 is not. When this happens, both values in the offset pairs corre-
2163 sponding to unused subpatterns are set to -1.
2164
2165 Offset values that correspond to unused subpatterns at the end of the
2166 expression are also set to -1. For example, if the string "abc" is
2167 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2168 matched. The return from the function is 2, because the highest used
2169 capturing subpattern number is 1. However, you can refer to the offsets
2170 for the second and third capturing subpatterns if you wish (assuming
2171 the vector is large enough, of course).
2172
2173 Some convenience functions are provided for extracting the captured
2174 substrings as separate strings. These are described below.
2175
2176 Error return values from pcre_exec()
2177
2178 If pcre_exec() fails, it returns a negative number. The following are
2179 defined in the header file:
2180
2181 PCRE_ERROR_NOMATCH (-1)
2182
2183 The subject string did not match the pattern.
2184
2185 PCRE_ERROR_NULL (-2)
2186
2187 Either code or subject was passed as NULL, or ovector was NULL and
2188 ovecsize was not zero.
2189
2190 PCRE_ERROR_BADOPTION (-3)
2191
2192 An unrecognized bit was set in the options argument.
2193
2194 PCRE_ERROR_BADMAGIC (-4)
2195
2196 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2197 to catch the case when it is passed a junk pointer and to detect when a
2198 pattern that was compiled in an environment of one endianness is run in
2199 an environment with the other endianness. This is the error that PCRE
2200 gives when the magic number is not present.
2201
2202 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2203
2204 While running the pattern match, an unknown item was encountered in the
2205 compiled pattern. This error could be caused by a bug in PCRE or by
2206 overwriting of the compiled pattern.
2207
2208 PCRE_ERROR_NOMEMORY (-6)
2209
2210 If a pattern contains back references, but the ovector that is passed
2211 to pcre_exec() is not big enough to remember the referenced substrings,
2212 PCRE gets a block of memory at the start of matching to use for this
2213 purpose. If the call via pcre_malloc() fails, this error is given. The
2214 memory is automatically freed at the end of matching.
2215
2216 PCRE_ERROR_NOSUBSTRING (-7)
2217
2218 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2219 and pcre_get_substring_list() functions (see below). It is never
2220 returned by pcre_exec().
2221
2222 PCRE_ERROR_MATCHLIMIT (-8)
2223
2224 The backtracking limit, as specified by the match_limit field in a
2225 pcre_extra structure (or defaulted) was reached. See the description
2226 above.
2227
2228 PCRE_ERROR_CALLOUT (-9)
2229
2230 This error is never generated by pcre_exec() itself. It is provided for
2231 use by callout functions that want to yield a distinctive error code.
2232 See the pcrecallout documentation for details.
2233
2234 PCRE_ERROR_BADUTF8 (-10)
2235
2236 A string that contains an invalid UTF-8 byte sequence was passed as a
2237 subject.
2238
2239 PCRE_ERROR_BADUTF8_OFFSET (-11)
2240
2241 The UTF-8 byte sequence that was passed as a subject was valid, but the
2242 value of startoffset did not point to the beginning of a UTF-8 charac-
2243 ter.
2244
2245 PCRE_ERROR_PARTIAL (-12)
2246
2247 The subject string did not match, but it did match partially. See the
2248 pcrepartial documentation for details of partial matching.
2249
2250 PCRE_ERROR_BADPARTIAL (-13)
2251
2252 The PCRE_PARTIAL option was used with a compiled pattern containing
2253 items that are not supported for partial matching. See the pcrepartial
2254 documentation for details of partial matching.
2255
2256 PCRE_ERROR_INTERNAL (-14)
2257
2258 An unexpected internal error has occurred. This error could be caused
2259 by a bug in PCRE or by overwriting of the compiled pattern.
2260
2261 PCRE_ERROR_BADCOUNT (-15)
2262
2263 This error is given if the value of the ovecsize argument is negative.
2264
2265 PCRE_ERROR_RECURSIONLIMIT (-21)
2266
2267 The internal recursion limit, as specified by the match_limit_recursion
2268 field in a pcre_extra structure (or defaulted) was reached. See the
2269 description above.
2270
2271 PCRE_ERROR_BADNEWLINE (-23)
2272
2273 An invalid combination of PCRE_NEWLINE_xxx options was given.
2274
2275 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2276
2277
2278 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2279
2280 int pcre_copy_substring(const char *subject, int *ovector,
2281 int stringcount, int stringnumber, char *buffer,
2282 int buffersize);
2283
2284 int pcre_get_substring(const char *subject, int *ovector,
2285 int stringcount, int stringnumber,
2286 const char **stringptr);
2287
2288 int pcre_get_substring_list(const char *subject,
2289 int *ovector, int stringcount, const char ***listptr);
2290
2291 Captured substrings can be accessed directly by using the offsets
2292 returned by pcre_exec() in ovector. For convenience, the functions
2293 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2294 string_list() are provided for extracting captured substrings as new,
2295 separate, zero-terminated strings. These functions identify substrings
2296 by number. The next section describes functions for extracting named
2297 substrings.
2298
2299 A substring that contains a binary zero is correctly extracted and has
2300 a further zero added on the end, but the result is not, of course, a C
2301 string. However, you can process such a string by referring to the
2302 length that is returned by pcre_copy_substring() and pcre_get_sub-
2303 string(). Unfortunately, the interface to pcre_get_substring_list() is
2304 not adequate for handling strings containing binary zeros, because the
2305 end of the final string is not independently indicated.
2306
2307 The first three arguments are the same for all three of these func-
2308 tions: subject is the subject string that has just been successfully
2309 matched, ovector is a pointer to the vector of integer offsets that was
2310 passed to pcre_exec(), and stringcount is the number of substrings that
2311 were captured by the match, including the substring that matched the
2312 entire regular expression. This is the value returned by pcre_exec() if
2313 it is greater than zero. If pcre_exec() returned zero, indicating that
2314 it ran out of space in ovector, the value passed as stringcount should
2315 be the number of elements in the vector divided by three.
2316
2317 The functions pcre_copy_substring() and pcre_get_substring() extract a
2318 single substring, whose number is given as stringnumber. A value of
2319 zero extracts the substring that matched the entire pattern, whereas
2320 higher values extract the captured substrings. For pcre_copy_sub-
2321 string(), the string is placed in buffer, whose length is given by
2322 buffersize, while for pcre_get_substring() a new block of memory is
2323 obtained via pcre_malloc, and its address is returned via stringptr.
2324 The yield of the function is the length of the string, not including
2325 the terminating zero, or one of these error codes:
2326
2327 PCRE_ERROR_NOMEMORY (-6)
2328
2329 The buffer was too small for pcre_copy_substring(), or the attempt to
2330 get memory failed for pcre_get_substring().
2331
2332 PCRE_ERROR_NOSUBSTRING (-7)
2333
2334 There is no substring whose number is stringnumber.
2335
2336 The pcre_get_substring_list() function extracts all available sub-
2337 strings and builds a list of pointers to them. All this is done in a
2338 single block of memory that is obtained via pcre_malloc. The address of
2339 the memory block is returned via listptr, which is also the start of
2340 the list of string pointers. The end of the list is marked by a NULL
2341 pointer. The yield of the function is zero if all went well, or the
2342 error code
2343
2344 PCRE_ERROR_NOMEMORY (-6)
2345
2346 if the attempt to get the memory block failed.
2347
2348 When any of these functions encounter a substring that is unset, which
2349 can happen when capturing subpattern number n+1 matches some part of
2350 the subject, but subpattern n has not been used at all, they return an
2351 empty string. This can be distinguished from a genuine zero-length sub-
2352 string by inspecting the appropriate offset in ovector, which is nega-
2353 tive for unset substrings.
2354
2355 The two convenience functions pcre_free_substring() and pcre_free_sub-
2356 string_list() can be used to free the memory returned by a previous
2357 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2358 tively. They do nothing more than call the function pointed to by
2359 pcre_free, which of course could be called directly from a C program.
2360 However, PCRE is used in some situations where it is linked via a spe-
2361 cial interface to another programming language that cannot use
2362 pcre_free directly; it is for these cases that the functions are pro-
2363 vided.
2364
2365
2366 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2367
2368 int pcre_get_stringnumber(const pcre *code,
2369 const char *name);
2370
2371 int pcre_copy_named_substring(const pcre *code,
2372 const char *subject, int *ovector,
2373 int stringcount, const char *stringname,
2374 char *buffer, int buffersize);
2375
2376 int pcre_get_named_substring(const pcre *code,
2377 const char *subject, int *ovector,
2378 int stringcount, const char *stringname,
2379 const char **stringptr);
2380
2381 To extract a substring by name, you first have to find associated num-
2382 ber. For example, for this pattern
2383
2384 (a+)b(?<xxx>\d+)...
2385
2386 the number of the subpattern called "xxx" is 2. If the name is known to
2387 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2388 name by calling pcre_get_stringnumber(). The first argument is the com-
2389 piled pattern, and the second is the name. The yield of the function is
2390 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2391 subpattern of that name.
2392
2393 Given the number, you can extract the substring directly, or use one of
2394 the functions described in the previous section. For convenience, there
2395 are also two functions that do the whole job.
2396
2397 Most of the arguments of pcre_copy_named_substring() and
2398 pcre_get_named_substring() are the same as those for the similarly
2399 named functions that extract by number. As these are described in the
2400 previous section, they are not re-described here. There are just two
2401 differences:
2402
2403 First, instead of a substring number, a substring name is given. Sec-
2404 ond, there is an extra argument, given at the start, which is a pointer
2405 to the compiled pattern. This is needed in order to gain access to the
2406 name-to-number translation table.
2407
2408 These functions call pcre_get_stringnumber(), and if it succeeds, they
2409 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2410 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2411 behaviour may not be what you want (see the next section).
2412
2413 Warning: If the pattern uses the "(?|" feature to set up multiple sub-
2414 patterns with the same number, you cannot use names to distinguish
2415 them, because names are not included in the compiled code. The matching
2416 process uses only numbers.
2417
2418
2419 DUPLICATE SUBPATTERN NAMES
2420
2421 int pcre_get_stringtable_entries(const pcre *code,
2422 const char *name, char **first, char **last);
2423
2424 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2425 subpatterns are not required to be unique. Normally, patterns with
2426 duplicate names are such that in any one match, only one of the named
2427 subpatterns participates. An example is shown in the pcrepattern docu-
2428 mentation.
2429
2430 When duplicates are present, pcre_copy_named_substring() and
2431 pcre_get_named_substring() return the first substring corresponding to
2432 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2433 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2434 function returns one of the numbers that are associated with the name,
2435 but it is not defined which it is.
2436
2437 If you want to get full details of all captured substrings for a given
2438 name, you must use the pcre_get_stringtable_entries() function. The
2439 first argument is the compiled pattern, and the second is the name. The
2440 third and fourth are pointers to variables which are updated by the
2441 function. After it has run, they point to the first and last entries in
2442 the name-to-number table for the given name. The function itself
2443 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2444 there are none. The format of the table is described above in the sec-
2445 tion entitled Information about a pattern. Given all the relevant
2446 entries for the name, you can extract each of their numbers, and hence
2447 the captured data, if any.
2448
2449
2450 FINDING ALL POSSIBLE MATCHES
2451
2452 The traditional matching function uses a similar algorithm to Perl,
2453 which stops when it finds the first match, starting at a given point in
2454 the subject. If you want to find all possible matches, or the longest
2455 possible match, consider using the alternative matching function (see
2456 below) instead. If you cannot use the alternative function, but still
2457 need to find all possible matches, you can kludge it up by making use
2458 of the callout facility, which is described in the pcrecallout documen-
2459 tation.
2460
2461 What you have to do is to insert a callout right at the end of the pat-
2462 tern. When your callout function is called, extract and save the cur-
2463 rent matched substring. Then return 1, which forces pcre_exec() to
2464 backtrack and try other alternatives. Ultimately, when it runs out of
2465 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2466
2467
2468 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2469
2470 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2471 const char *subject, int length, int startoffset,
2472 int options, int *ovector, int ovecsize,
2473 int *workspace, int wscount);
2474
2475 The function pcre_dfa_exec() is called to match a subject string
2476 against a compiled pattern, using a matching algorithm that scans the
2477 subject string just once, and does not backtrack. This has different
2478 characteristics to the normal algorithm, and is not compatible with
2479 Perl. Some of the features of PCRE patterns are not supported. Never-
2480 theless, there are times when this kind of matching can be useful. For
2481 a discussion of the two matching algorithms, see the pcrematching docu-
2482 mentation.
2483
2484 The arguments for the pcre_dfa_exec() function are the same as for
2485 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2486 ent way, and this is described below. The other common arguments are
2487 used in the same way as for pcre_exec(), so their description is not
2488 repeated here.
2489
2490 The two additional arguments provide workspace for the function. The
2491 workspace vector should contain at least 20 elements. It is used for
2492 keeping track of multiple paths through the pattern tree. More
2493 workspace will be needed for patterns and subjects where there are a
2494 lot of potential matches.
2495
2496 Here is an example of a simple call to pcre_dfa_exec():
2497
2498 int rc;
2499 int ovector[10];
2500 int wspace[20];
2501 rc = pcre_dfa_exec(
2502 re, /* result of pcre_compile() */
2503 NULL, /* we didn't study the pattern */
2504 "some string", /* the subject string */
2505 11, /* the length of the subject string */
2506 0, /* start at offset 0 in the subject */
2507 0, /* default options */
2508 ovector, /* vector of integers for substring information */
2509 10, /* number of elements (NOT size in bytes) */
2510 wspace, /* working space vector */
2511 20); /* number of elements (NOT size in bytes) */
2512
2513 Option bits for pcre_dfa_exec()
2514
2515 The unused bits of the options argument for pcre_dfa_exec() must be
2516 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2517 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2518 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2519 three of these are the same as for pcre_exec(), so their description is
2520 not repeated here.
2521
2522 PCRE_PARTIAL
2523
2524 This has the same general effect as it does for pcre_exec(), but the
2525 details are slightly different. When PCRE_PARTIAL is set for
2526 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2527 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2528 been no complete matches, but there is still at least one matching pos-
2529 sibility. The portion of the string that provided the partial match is
2530 set as the first matching string.
2531
2532 PCRE_DFA_SHORTEST
2533
2534 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2535 stop as soon as it has found one match. Because of the way the alterna-
2536 tive algorithm works, this is necessarily the shortest possible match
2537 at the first possible matching point in the subject string.
2538
2539 PCRE_DFA_RESTART
2540
2541 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2542 returns a partial match, it is possible to call it again, with addi-
2543 tional subject characters, and have it continue with the same match.
2544 The PCRE_DFA_RESTART option requests this action; when it is set, the
2545 workspace and wscount options must reference the same vector as before
2546 because data about the match so far is left in them after a partial
2547 match. There is more discussion of this facility in the pcrepartial
2548 documentation.
2549
2550 Successful returns from pcre_dfa_exec()
2551
2552 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2553 string in the subject. Note, however, that all the matches from one run
2554 of the function start at the same point in the subject. The shorter
2555 matches are all initial substrings of the longer matches. For example,
2556 if the pattern
2557
2558 <.*>
2559
2560 is matched against the string
2561
2562 This is <something> <something else> <something further> no more
2563
2564 the three matched strings are
2565
2566 <something>
2567 <something> <something else>
2568 <something> <something else> <something further>
2569
2570 On success, the yield of the function is a number greater than zero,
2571 which is the number of matched substrings. The substrings themselves
2572 are returned in ovector. Each string uses two elements; the first is
2573 the offset to the start, and the second is the offset to the end. In
2574 fact, all the strings have the same start offset. (Space could have
2575 been saved by giving this only once, but it was decided to retain some
2576 compatibility with the way pcre_exec() returns data, even though the
2577 meaning of the strings is different.)
2578
2579 The strings are returned in reverse order of length; that is, the long-
2580 est matching string is given first. If there were too many matches to
2581 fit into ovector, the yield of the function is zero, and the vector is
2582 filled with the longest matches.
2583
2584 Error returns from pcre_dfa_exec()
2585
2586 The pcre_dfa_exec() function returns a negative number when it fails.
2587 Many of the errors are the same as for pcre_exec(), and these are
2588 described above. There are in addition the following errors that are
2589 specific to pcre_dfa_exec():
2590
2591 PCRE_ERROR_DFA_UITEM (-16)
2592
2593 This return is given if pcre_dfa_exec() encounters an item in the pat-
2594 tern that it does not support, for instance, the use of \C or a back
2595 reference.
2596
2597 PCRE_ERROR_DFA_UCOND (-17)
2598
2599 This return is given if pcre_dfa_exec() encounters a condition item
2600 that uses a back reference for the condition, or a test for recursion
2601 in a specific group. These are not supported.
2602
2603 PCRE_ERROR_DFA_UMLIMIT (-18)
2604
2605 This return is given if pcre_dfa_exec() is called with an extra block
2606 that contains a setting of the match_limit field. This is not supported
2607 (it is meaningless).
2608
2609 PCRE_ERROR_DFA_WSSIZE (-19)
2610
2611 This return is given if pcre_dfa_exec() runs out of space in the
2612 workspace vector.
2613
2614 PCRE_ERROR_DFA_RECURSE (-20)
2615
2616 When a recursive subpattern is processed, the matching function calls
2617 itself recursively, using private vectors for ovector and workspace.
2618 This error is given if the output vector is not large enough. This
2619 should be extremely rare, as a vector of size 1000 is used.
2620
2621
2622 SEE ALSO
2623
2624 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2625 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2626
2627
2628 AUTHOR
2629
2630 Philip Hazel
2631 University Computing Service
2632 Cambridge CB2 3QH, England.
2633
2634
2635 REVISION
2636
2637 Last updated: 17 March 2009
2638 Copyright (c) 1997-2009 University of Cambridge.
2639 ------------------------------------------------------------------------------
2640
2641
2642 PCRECALLOUT(3) PCRECALLOUT(3)
2643
2644
2645 NAME
2646 PCRE - Perl-compatible regular expressions
2647
2648
2649 PCRE CALLOUTS
2650
2651 int (*pcre_callout)(pcre_callout_block *);
2652
2653 PCRE provides a feature called "callout", which is a means of temporar-
2654 ily passing control to the caller of PCRE in the middle of pattern
2655 matching. The caller of PCRE provides an external function by putting
2656 its entry point in the global variable pcre_callout. By default, this
2657 variable contains NULL, which disables all calling out.
2658
2659 Within a regular expression, (?C) indicates the points at which the
2660 external function is to be called. Different callout points can be
2661 identified by putting a number less than 256 after the letter C. The
2662 default value is zero. For example, this pattern has two callout
2663 points:
2664
2665 (?C1)abc(?C2)def
2666
2667 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2668 called, PCRE automatically inserts callouts, all with number 255,
2669 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2670 used with the pattern
2671
2672 A(\d{2}|--)
2673
2674 it is processed as if it were
2675
2676 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2677
2678 Notice that there is a callout before and after each parenthesis and
2679 alternation bar. Automatic callouts can be used for tracking the
2680 progress of pattern matching. The pcretest command has an option that
2681 sets automatic callouts; when it is used, the output indicates how the
2682 pattern is matched. This is useful information when you are trying to
2683 optimize the performance of a particular pattern.
2684
2685
2686 MISSING CALLOUTS
2687
2688 You should be aware that, because of optimizations in the way PCRE
2689 matches patterns by default, callouts sometimes do not happen. For
2690 example, if the pattern is
2691
2692 ab(?C4)cd
2693
2694 PCRE knows that any matching string must contain the letter "d". If the
2695 subject string is "abyz", the lack of "d" means that matching doesn't
2696 ever start, and the callout is never reached. However, with "abyd",
2697 though the result is still no match, the callout is obeyed.
2698
2699 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2700 MIZE option to pcre_exec() or pcre_dfa_exec(). This slows down the
2701 matching process, but does ensure that callouts such as the example
2702 above are obeyed.
2703
2704
2705 THE CALLOUT INTERFACE
2706
2707 During matching, when PCRE reaches a callout point, the external func-
2708 tion defined by pcre_callout is called (if it is set). This applies to
2709 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2710 only argument to the callout function is a pointer to a pcre_callout
2711 block. This structure contains the following fields:
2712
2713 int version;
2714 int callout_number;
2715 int *offset_vector;
2716 const char *subject;
2717 int subject_length;
2718 int start_match;
2719 int current_position;
2720 int capture_top;
2721 int capture_last;
2722 void *callout_data;
2723 int pattern_position;
2724 int next_item_length;
2725
2726 The version field is an integer containing the version number of the
2727 block format. The initial version was 0; the current version is 1. The
2728 version number will change again in future if additional fields are
2729 added, but the intention is never to remove any of the existing fields.
2730
2731 The callout_number field contains the number of the callout, as com-
2732 piled into the pattern (that is, the number after ?C for manual call-
2733 outs, and 255 for automatically generated callouts).
2734
2735 The offset_vector field is a pointer to the vector of offsets that was
2736 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2737 pcre_exec() is used, the contents can be inspected in order to extract
2738 substrings that have been matched so far, in the same way as for
2739 extracting substrings after a match has completed. For pcre_dfa_exec()
2740 this field is not useful.
2741
2742 The subject and subject_length fields contain copies of the values that
2743 were passed to pcre_exec().
2744
2745 The start_match field normally contains the offset within the subject
2746 at which the current match attempt started. However, if the escape
2747 sequence \K has been encountered, this value is changed to reflect the
2748 modified starting point. If the pattern is not anchored, the callout
2749 function may be called several times from the same point in the pattern
2750 for different starting points in the subject.
2751
2752 The current_position field contains the offset within the subject of
2753 the current match pointer.
2754
2755 When the pcre_exec() function is used, the capture_top field contains
2756 one more than the number of the highest numbered captured substring so
2757 far. If no substrings have been captured, the value of capture_top is
2758 one. This is always the case when pcre_dfa_exec() is used, because it
2759 does not support captured substrings.
2760
2761 The capture_last field contains the number of the most recently cap-
2762 tured substring. If no substrings have been captured, its value is -1.
2763 This is always the case when pcre_dfa_exec() is used.
2764
2765 The callout_data field contains a value that is passed to pcre_exec()
2766 or pcre_dfa_exec() specifically so that it can be passed back in call-
2767 outs. It is passed in the pcre_callout field of the pcre_extra data
2768 structure. If no such data was passed, the value of callout_data in a
2769 pcre_callout block is NULL. There is a description of the pcre_extra
2770 structure in the pcreapi documentation.
2771
2772 The pattern_position field is present from version 1 of the pcre_call-
2773 out structure. It contains the offset to the next item to be matched in
2774 the pattern string.
2775
2776 The next_item_length field is present from version 1 of the pcre_call-
2777 out structure. It contains the length of the next item to be matched in
2778 the pattern string. When the callout immediately precedes an alterna-
2779 tion bar, a closing parenthesis, or the end of the pattern, the length
2780 is zero. When the callout precedes an opening parenthesis, the length
2781 is that of the entire subpattern.
2782
2783 The pattern_position and next_item_length fields are intended to help
2784 in distinguishing between different automatic callouts, which all have
2785 the same callout number. However, they are set for all callouts.
2786
2787
2788 RETURN VALUES
2789
2790 The external callout function returns an integer to PCRE. If the value
2791 is zero, matching proceeds as normal. If the value is greater than
2792 zero, matching fails at the current point, but the testing of other
2793 matching possibilities goes ahead, just as if a lookahead assertion had
2794 failed. If the value is less than zero, the match is abandoned, and
2795 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2796
2797 Negative values should normally be chosen from the set of
2798 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2799 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2800 reserved for use by callout functions; it will never be used by PCRE
2801 itself.
2802
2803
2804 AUTHOR
2805
2806 Philip Hazel
2807 University Computing Service
2808 Cambridge CB2 3QH, England.
2809
2810
2811 REVISION
2812
2813 Last updated: 15 March 2009
2814 Copyright (c) 1997-2009 University of Cambridge.
2815 ------------------------------------------------------------------------------
2816
2817
2818 PCRECOMPAT(3) PCRECOMPAT(3)
2819
2820
2821 NAME
2822 PCRE - Perl-compatible regular expressions
2823
2824
2825 DIFFERENCES BETWEEN PCRE AND PERL
2826
2827 This document describes the differences in the ways that PCRE and Perl
2828 handle regular expressions. The differences described here are mainly
2829 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2830 some features that are expected to be in the forthcoming Perl 5.10.
2831
2832 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2833 of what it does have are given in the section on UTF-8 support in the
2834 main pcre page.
2835
2836 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2837 permits them, but they do not mean what you might think. For example,
2838 (?!a){3} does not assert that the next three characters are not "a". It
2839 just asserts that the next character is not "a" three times.
2840
2841 3. Capturing subpatterns that occur inside negative lookahead asser-
2842 tions are counted, but their entries in the offsets vector are never
2843 set. Perl sets its numerical variables from any such patterns that are
2844 matched before the assertion fails to match something (thereby succeed-
2845 ing), but only if the negative lookahead assertion contains just one
2846 branch.
2847
2848 4. Though binary zero characters are supported in the subject string,
2849 they are not allowed in a pattern string because it is passed as a nor-
2850 mal C string, terminated by zero. The escape sequence \0 can be used in
2851 the pattern to represent a binary zero.
2852
2853 5. The following Perl escape sequences are not supported: \l, \u, \L,
2854 \U, and \N. In fact these are implemented by Perl's general string-han-
2855 dling and are not part of its pattern matching engine. If any of these
2856 are encountered by PCRE, an error is generated.
2857
2858 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2859 is built with Unicode character property support. The properties that
2860 can be tested with \p and \P are limited to the general category prop-
2861 erties such as Lu and Nd, script names such as Greek or Han, and the
2862 derived properties Any and L&.
2863
2864 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2865 ters in between are treated as literals. This is slightly different
2866 from Perl in that $ and @ are also handled as literals inside the
2867 quotes. In Perl, they cause variable interpolation (but of course PCRE
2868 does not have variables). Note the following examples:
2869
2870 Pattern PCRE matches Perl matches
2871
2872 \Qabc$xyz\E abc$xyz abc followed by the
2873 contents of $xyz
2874 \Qabc\$xyz\E abc\$xyz abc\$xyz
2875 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2876
2877 The \Q...\E sequence is recognized both inside and outside character
2878 classes.
2879
2880 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2881 constructions. However, there is support for recursive patterns. This
2882 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2883 "callout" feature allows an external function to be called during pat-
2884 tern matching. See the pcrecallout documentation for details.
2885
2886 9. Subpatterns that are called recursively or as "subroutines" are
2887 always treated as atomic groups in PCRE. This is like Python, but
2888 unlike Perl.
2889
2890 10. There are some differences that are concerned with the settings of
2891 captured strings when part of a pattern is repeated. For example,
2892 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2893 unset, but in PCRE it is set to "b".
2894
2895 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2896 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2897 the forms without an argument. PCRE does not support (*MARK). If
2898 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2899 ture group; this is different to Perl.
2900
2901 12. PCRE provides some extensions to the Perl regular expression facil-
2902 ities. Perl 5.10 will include new features that are not in earlier
2903 versions, some of which (such as named parentheses) have been in PCRE
2904 for some time. This list is with respect to Perl 5.10:
2905
2906 (a) Although lookbehind assertions must match fixed length strings,
2907 each alternative branch of a lookbehind assertion can match a different
2908 length of string. Perl requires them all to have the same length.
2909
2910 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2911 meta-character matches only at the very end of the string.
2912
2913 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2914 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2915 ignored. (Perl can be made to issue a warning.)
2916
2917 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2918 fiers is inverted, that is, by default they are not greedy, but if fol-
2919 lowed by a question mark they are.
2920
2921 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2922 tried only at the first matching position in the subject string.
2923
2924 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2925 TURE options for pcre_exec() have no Perl equivalents.
2926
2927 (g) The \R escape sequence can be restricted to match only CR, LF, or
2928 CRLF by the PCRE_BSR_ANYCRLF option.
2929
2930 (h) The callout facility is PCRE-specific.
2931
2932 (i) The partial matching facility is PCRE-specific.
2933
2934 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2935 even on different hosts that have the other endianness.
2936
2937 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2938 different way and is not Perl-compatible.
2939
2940 (l) PCRE recognizes some special sequences such as (*CR) at the start
2941 of a pattern that set overall options that cannot be changed within the
2942 pattern.
2943
2944
2945 AUTHOR
2946
2947 Philip Hazel
2948 University Computing Service
2949 Cambridge CB2 3QH, England.
2950
2951
2952 REVISION
2953
2954 Last updated: 11 September 2007
2955 Copyright (c) 1997-2007 University of Cambridge.
2956 ------------------------------------------------------------------------------
2957
2958
2959 PCREPATTERN(3) PCREPATTERN(3)
2960
2961
2962 NAME
2963 PCRE - Perl-compatible regular expressions
2964
2965
2966 PCRE REGULAR EXPRESSION DETAILS
2967
2968 The syntax and semantics of the regular expressions that are supported
2969 by PCRE are described in detail below. There is a quick-reference syn-
2970 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
2971 semantics as closely as it can. PCRE also supports some alternative
2972 regular expression syntax (which does not conflict with the Perl syn-
2973 tax) in order to provide some compatibility with regular expressions in
2974 Python, .NET, and Oniguruma.
2975
2976 Perl's regular expressions are described in its own documentation, and
2977 regular expressions in general are covered in a number of books, some
2978 of which have copious examples. Jeffrey Friedl's "Mastering Regular
2979 Expressions", published by O'Reilly, covers regular expressions in
2980 great detail. This description of PCRE's regular expressions is
2981 intended as reference material.
2982
2983 The original operation of PCRE was on strings of one-byte characters.
2984 However, there is now also support for UTF-8 character strings. To use
2985 this, you must build PCRE to include UTF-8 support, and then call
2986 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2987 matching is mentioned in several places below. There is also a summary
2988 of UTF-8 features in the section on UTF-8 support in the main pcre
2989 page.
2990
2991 The remainder of this document discusses the patterns that are sup-
2992 ported by PCRE when its main matching function, pcre_exec(), is used.
2993 From release 6.0, PCRE offers a second matching function,
2994 pcre_dfa_exec(), which matches using a different algorithm that is not
2995 Perl-compatible. Some of the features discussed below are not available
2996 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2997 alternative function, and how it differs from the normal function, are
2998 discussed in the pcrematching page.
2999
3000
3001 NEWLINE CONVENTIONS
3002
3003 PCRE supports five different conventions for indicating line breaks in
3004 strings: a single CR (carriage return) character, a single LF (line-
3005 feed) character, the two-character sequence CRLF, any of the three pre-
3006 ceding, or any Unicode newline sequence. The pcreapi page has further
3007 discussion about newlines, and shows how to set the newline convention
3008 in the options arguments for the compiling and matching functions.
3009
3010 It is also possible to specify a newline convention by starting a pat-
3011 tern string with one of the following five sequences:
3012
3013 (*CR) carriage return
3014 (*LF) linefeed
3015 (*CRLF) carriage return, followed by linefeed
3016 (*ANYCRLF) any of the three above
3017 (*ANY) all Unicode newline sequences
3018
3019 These override the default and the options given to pcre_compile(). For
3020 example, on a Unix system where LF is the default newline sequence, the
3021 pattern
3022
3023 (*CR)a.b
3024
3025 changes the convention to CR. That pattern matches "a\nb" because LF is
3026 no longer a newline. Note that these special settings, which are not
3027 Perl-compatible, are recognized only at the very start of a pattern,
3028 and that they must be in upper case. If more than one of them is
3029 present, the last one is used.
3030
3031 The newline convention does not affect what the \R escape sequence
3032 matches. By default, this is any Unicode newline sequence, for Perl
3033 compatibility. However, this can be changed; see the description of \R
3034 in the section entitled "Newline sequences" below. A change of \R set-
3035 ting can be combined with a change of newline convention.
3036
3037
3038 CHARACTERS AND METACHARACTERS
3039
3040 A regular expression is a pattern that is matched against a subject
3041 string from left to right. Most characters stand for themselves in a
3042 pattern, and match the corresponding characters in the subject. As a
3043 trivial example, the pattern
3044
3045 The quick brown fox
3046
3047 matches a portion of a subject string that is identical to itself. When
3048 caseless matching is specified (the PCRE_CASELESS option), letters are
3049 matched independently of case. In UTF-8 mode, PCRE always understands
3050 the concept of case for characters whose values are less than 128, so
3051 caseless matching is always possible. For characters with higher val-
3052 ues, the concept of case is supported if PCRE is compiled with Unicode
3053 property support, but not otherwise. If you want to use caseless
3054 matching for characters 128 and above, you must ensure that PCRE is
3055 compiled with Unicode property support as well as with UTF-8 support.
3056
3057 The power of regular expressions comes from the ability to include
3058 alternatives and repetitions in the pattern. These are encoded in the
3059 pattern by the use of metacharacters, which do not stand for themselves
3060 but instead are interpreted in some special way.
3061
3062 There are two different sets of metacharacters: those that are recog-
3063 nized anywhere in the pattern except within square brackets, and those
3064 that are recognized within square brackets. Outside square brackets,
3065 the metacharacters are as follows:
3066
3067 \ general escape character with several uses
3068 ^ assert start of string (or line, in multiline mode)
3069 $ assert end of string (or line, in multiline mode)
3070 . match any character except newline (by default)
3071 [ start character class definition
3072 | start of alternative branch
3073 ( start subpattern
3074 ) end subpattern
3075 ? extends the meaning of (
3076 also 0 or 1 quantifier
3077 also quantifier minimizer
3078 * 0 or more quantifier
3079 + 1 or more quantifier
3080 also "possessive quantifier"
3081 { start min/max quantifier
3082
3083 Part of a pattern that is in square brackets is called a "character
3084 class". In a character class the only metacharacters are:
3085
3086 \ general escape character
3087 ^ negate the class, but only if the first character
3088 - indicates character range
3089 [ POSIX character class (only if followed by POSIX
3090 syntax)
3091 ] terminates the character class
3092
3093 The following sections describe the use of each of the metacharacters.
3094
3095
3096 BACKSLASH
3097
3098 The backslash character has several uses. Firstly, if it is followed by
3099 a non-alphanumeric character, it takes away any special meaning that
3100 character may have. This use of backslash as an escape character
3101 applies both inside and outside character classes.
3102
3103 For example, if you want to match a * character, you write \* in the
3104 pattern. This escaping action applies whether or not the following
3105 character would otherwise be interpreted as a metacharacter, so it is
3106 always safe to precede a non-alphanumeric with backslash to specify
3107 that it stands for itself. In particular, if you want to match a back-
3108 slash, you write \\.
3109
3110 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3111 the pattern (other than in a character class) and characters between a
3112 # outside a character class and the next newline are ignored. An escap-
3113 ing backslash can be used to include a whitespace or # character as
3114 part of the pattern.
3115
3116 If you want to remove the special meaning from a sequence of charac-
3117 ters, you can do so by putting them between \Q and \E. This is differ-
3118 ent from Perl in that $ and @ are handled as literals in \Q...\E
3119 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3120 tion. Note the following examples:
3121
3122 Pattern PCRE matches Perl matches
3123
3124 \Qabc$xyz\E abc$xyz abc followed by the
3125 contents of $xyz
3126 \Qabc\$xyz\E abc\$xyz abc\$xyz
3127 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3128
3129 The \Q...\E sequence is recognized both inside and outside character
3130 classes.
3131
3132 Non-printing characters
3133
3134 A second use of backslash provides a way of encoding non-printing char-
3135 acters in patterns in a visible manner. There is no restriction on the
3136 appearance of non-printing characters, apart from the binary zero that
3137 terminates a pattern, but when a pattern is being prepared by text
3138 editing, it is usually easier to use one of the following escape
3139 sequences than the binary character it represents:
3140
3141 \a alarm, that is, the BEL character (hex 07)
3142 \cx "control-x", where x is any character
3143 \e escape (hex 1B)
3144 \f formfeed (hex 0C)
3145 \n linefeed (hex 0A)
3146 \r carriage return (hex 0D)
3147 \t tab (hex 09)
3148 \ddd character with octal code ddd, or backreference
3149 \xhh character with hex code hh
3150 \x{hhh..} character with hex code hhh..
3151
3152 The precise effect of \cx is as follows: if x is a lower case letter,
3153 it is converted to upper case. Then bit 6 of the character (hex 40) is
3154 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3155 becomes hex 7B.
3156
3157 After \x, from zero to two hexadecimal digits are read (letters can be
3158 in upper or lower case). Any number of hexadecimal digits may appear
3159 between \x{ and }, but the value of the character code must be less
3160 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3161 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3162 than the largest Unicode code point, which is 10FFFF.
3163
3164 If characters other than hexadecimal digits appear between \x{ and },
3165 or if there is no terminating }, this form of escape is not recognized.
3166 Instead, the initial \x will be interpreted as a basic hexadecimal
3167 escape, with no following digits, giving a character whose value is
3168 zero.
3169
3170 Characters whose value is less than 256 can be defined by either of the
3171 two syntaxes for \x. There is no difference in the way they are han-
3172 dled. For example, \xdc is exactly the same as \x{dc}.
3173
3174 After \0 up to two further octal digits are read. If there are fewer
3175 than two digits, just those that are present are used. Thus the
3176 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3177 (code value 7). Make sure you supply two digits after the initial zero
3178 if the pattern character that follows is itself an octal digit.
3179
3180 The handling of a backslash followed by a digit other than 0 is compli-
3181 cated. Outside a character class, PCRE reads it and any following dig-
3182 its as a decimal number. If the number is less than 10, or if there
3183 have been at least that many previous capturing left parentheses in the
3184 expression, the entire sequence is taken as a back reference. A
3185 description of how this works is given later, following the discussion
3186 of parenthesized subpatterns.
3187
3188 Inside a character class, or if the decimal number is greater than 9
3189 and there have not been that many capturing subpatterns, PCRE re-reads
3190 up to three octal digits following the backslash, and uses them to gen-
3191 erate a data character. Any subsequent digits stand for themselves. In
3192 non-UTF-8 mode, the value of a character specified in octal must be
3193 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3194 example:
3195
3196 \040 is another way of writing a space
3197 \40 is the same, provided there are fewer than 40
3198 previous capturing subpatterns
3199 \7 is always a back reference
3200 \11 might be a back reference, or another way of
3201 writing a tab
3202 \011 is always a tab
3203 \0113 is a tab followed by the character "3"
3204 \113 might be a back reference, otherwise the
3205 character with octal code 113
3206 \377 might be a back reference, otherwise
3207 the byte consisting entirely of 1 bits
3208 \81 is either a back reference, or a binary zero
3209 followed by the two characters "8" and "1"
3210
3211 Note that octal values of 100 or greater must not be introduced by a
3212 leading zero, because no more than three octal digits are ever read.
3213
3214 All the sequences that define a single character value can be used both
3215 inside and outside character classes. In addition, inside a character
3216 class, the sequence \b is interpreted as the backspace character (hex
3217 08), and the sequences \R and \X are interpreted as the characters "R"
3218 and "X", respectively. Outside a character class, these sequences have
3219 different meanings (see below).
3220
3221 Absolute and relative back references
3222
3223 The sequence \g followed by an unsigned or a negative number, option-
3224 ally enclosed in braces, is an absolute or relative back reference. A
3225 named back reference can be coded as \g{name}. Back references are dis-
3226 cussed later, following the discussion of parenthesized subpatterns.
3227
3228 Absolute and relative subroutine calls
3229
3230 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3231 name or a number enclosed either in angle brackets or single quotes, is
3232 an alternative syntax for referencing a subpattern as a "subroutine".
3233 Details are discussed later. Note that \g{...} (Perl syntax) and
3234 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3235 reference; the latter is a subroutine call.
3236
3237 Generic character types
3238
3239 Another use of backslash is for specifying generic character types. The
3240 following are always recognized:
3241
3242 \d any decimal digit
3243 \D any character that is not a decimal digit
3244 \h any horizontal whitespace character
3245 \H any character that is not a horizontal whitespace character
3246 \s any whitespace character
3247 \S any character that is not a whitespace character
3248 \v any vertical whitespace character
3249 \V any character that is not a vertical whitespace character
3250 \w any "word" character
3251 \W any "non-word" character
3252
3253 Each pair of escape sequences partitions the complete set of characters
3254 into two disjoint sets. Any given character matches one, and only one,
3255 of each pair.
3256
3257 These character type sequences can appear both inside and outside char-
3258 acter classes. They each match one character of the appropriate type.
3259 If the current matching point is at the end of the subject string, all
3260 of them fail, since there is no character to match.
3261
3262 For compatibility with Perl, \s does not match the VT character (code
3263 11). This makes it different from the the POSIX "space" class. The \s
3264 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3265 "use locale;" is included in a Perl script, \s may match the VT charac-
3266 ter. In PCRE, it never does.
3267
3268 In UTF-8 mode, characters with values greater than 128 never match \d,
3269 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3270 code character property support is available. These sequences retain
3271 their original meanings from before UTF-8 support was available, mainly
3272 for efficiency reasons. Note that this also affects \b, because it is
3273 defined in terms of \w and \W.
3274
3275 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3276 the other sequences, these do match certain high-valued codepoints in
3277 UTF-8 mode. The horizontal space characters are:
3278
3279 U+0009 Horizontal tab
3280 U+0020 Space
3281 U+00A0 Non-break space
3282 U+1680 Ogham space mark
3283 U+180E Mongolian vowel separator
3284 U+2000 En quad
3285 U+2001 Em quad
3286 U+2002 En space
3287 U+2003 Em space
3288 U+2004 Three-per-em space
3289 U+2005 Four-per-em space
3290 U+2006 Six-per-em space
3291 U+2007 Figure space
3292 U+2008 Punctuation space
3293 U+2009 Thin space
3294 U+200A Hair space
3295 U+202F Narrow no-break space
3296 U+205F Medium mathematical space
3297 U+3000 Ideographic space
3298
3299 The vertical space characters are:
3300
3301 U+000A Linefeed
3302 U+000B Vertical tab
3303 U+000C Formfeed
3304 U+000D Carriage return
3305 U+0085 Next line
3306 U+2028 Line separator
3307 U+2029 Paragraph separator
3308
3309 A "word" character is an underscore or any character less than 256 that
3310 is a letter or digit. The definition of letters and digits is con-
3311 trolled by PCRE's low-valued character tables, and may vary if locale-
3312 specific matching is taking place (see "Locale support" in the pcreapi
3313 page). For example, in a French locale such as "fr_FR" in Unix-like
3314 systems, or "french" in Windows, some character codes greater than 128
3315 are used for accented letters, and these are matched by \w. The use of
3316 locales with Unicode is discouraged.
3317
3318 Newline sequences
3319
3320 Outside a character class, by default, the escape sequence \R matches
3321 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3322 mode \R is equivalent to the following:
3323
3324 (?>\r\n|\n|\x0b|\f|\r|\x85)
3325
3326 This is an example of an "atomic group", details of which are given
3327 below. This particular group matches either the two-character sequence
3328 CR followed by LF, or one of the single characters LF (linefeed,
3329 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3330 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3331 is treated as a single unit that cannot be split.
3332
3333 In UTF-8 mode, two additional characters whose codepoints are greater
3334 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3335 rator, U+2029). Unicode character property support is not needed for
3336 these characters to be recognized.
3337
3338 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3339 the complete set of Unicode line endings) by setting the option
3340 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3341 (BSR is an abbrevation for "backslash R".) This can be made the default
3342 when PCRE is built; if this is the case, the other behaviour can be
3343 requested via the PCRE_BSR_UNICODE option. It is also possible to
3344 specify these settings by starting a pattern string with one of the
3345 following sequences:
3346
3347 (*BSR_ANYCRLF) CR, LF, or CRLF only
3348 (*BSR_UNICODE) any Unicode newline sequence
3349
3350 These override the default and the options given to pcre_compile(), but
3351 they can be overridden by options given to pcre_exec(). Note that these
3352 special settings, which are not Perl-compatible, are recognized only at
3353 the very start of a pattern, and that they must be in upper case. If
3354 more than one of them is present, the last one is used. They can be
3355 combined with a change of newline convention, for example, a pattern
3356 can start with:
3357
3358 (*ANY)(*BSR_ANYCRLF)
3359
3360 Inside a character class, \R matches the letter "R".
3361
3362 Unicode character properties
3363
3364 When PCRE is built with Unicode character property support, three addi-
3365 tional escape sequences that match characters with specific properties
3366 are available. When not in UTF-8 mode, these sequences are of course
3367 limited to testing characters whose codepoints are less than 256, but
3368 they do work in this mode. The extra escape sequences are:
3369
3370 \p{xx} a character with the xx property
3371 \P{xx} a character without the xx property
3372 \X an extended Unicode sequence
3373
3374 The property names represented by xx above are limited to the Unicode
3375 script names, the general category properties, and "Any", which matches
3376 any character (including newline). Other properties such as "InMusical-
3377 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3378 not match any characters, so always causes a match failure.
3379
3380 Sets of Unicode characters are defined as belonging to certain scripts.
3381 A character from one of these sets can be matched using a script name.
3382 For example:
3383
3384 \p{Greek}
3385 \P{Han}
3386
3387 Those that are not part of an identified script are lumped together as
3388 "Common". The current list of scripts is:
3389
3390 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3391 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3392 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3393 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3394 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3395 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3396 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3397 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3398 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3399
3400 Each character has exactly one general category property, specified by
3401 a two-letter abbreviation. For compatibility with Perl, negation can be
3402 specified by including a circumflex between the opening brace and the
3403 property name. For example, \p{^Lu} is the same as \P{Lu}.
3404
3405 If only one letter is specified with \p or \P, it includes all the gen-
3406 eral category properties that start with that letter. In this case, in
3407 the absence of negation, the curly brackets in the escape sequence are
3408 optional; these two examples have the same effect:
3409
3410 \p{L}
3411 \pL
3412
3413 The following general category property codes are supported:
3414
3415 C Other
3416 Cc Control
3417 Cf Format
3418 Cn Unassigned
3419 Co Private use
3420 Cs Surrogate
3421
3422 L Letter
3423 Ll Lower case letter
3424 Lm Modifier letter
3425 Lo Other letter
3426 Lt Title case letter
3427 Lu Upper case letter
3428
3429 M Mark
3430 Mc Spacing mark
3431 Me Enclosing mark
3432 Mn Non-spacing mark
3433
3434 N Number
3435 Nd Decimal number
3436 Nl Letter number
3437 No Other number
3438
3439 P Punctuation
3440 Pc Connector punctuation
3441 Pd Dash punctuation
3442 Pe Close punctuation
3443 Pf Final punctuation
3444 Pi Initial punctuation
3445 Po Other punctuation
3446 Ps Open punctuation
3447
3448 S Symbol
3449 Sc Currency symbol
3450 Sk Modifier symbol
3451 Sm Mathematical symbol
3452 So Other symbol
3453
3454 Z Separator
3455 Zl Line separator
3456 Zp Paragraph separator
3457 Zs Space separator
3458
3459 The special property L& is also supported: it matches a character that
3460 has the Lu, Ll, or Lt property, in other words, a letter that is not
3461 classified as a modifier or "other".
3462
3463 The Cs (Surrogate) property applies only to characters in the range
3464 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3465 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3466 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3467 the pcreapi page).
3468
3469 The long synonyms for these properties that Perl supports (such as
3470 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3471 any of these properties with "Is".
3472
3473 No character that is in the Unicode table has the Cn (unassigned) prop-
3474 erty. Instead, this property is assumed for any code point that is not
3475 in the Unicode table.
3476
3477 Specifying caseless matching does not affect these escape sequences.
3478 For example, \p{Lu} always matches only upper case letters.
3479
3480 The \X escape matches any number of Unicode characters that form an
3481 extended Unicode sequence. \X is equivalent to
3482
3483 (?>\PM\pM*)
3484
3485 That is, it matches a character without the "mark" property, followed
3486 by zero or more characters with the "mark" property, and treats the
3487 sequence as an atomic group (see below). Characters with the "mark"
3488 property are typically accents that affect the preceding character.
3489 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3490 matches any one character.
3491
3492 Matching characters by Unicode property is not fast, because PCRE has
3493 to search a structure that contains data for over fifteen thousand
3494 characters. That is why the traditional escape sequences such as \d and
3495 \w do not use Unicode properties in PCRE.
3496
3497 Resetting the match start
3498
3499 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3500 ously matched characters not to be included in the final matched
3501 sequence. For example, the pattern:
3502
3503 foo\Kbar
3504
3505 matches "foobar", but reports that it has matched "bar". This feature
3506 is similar to a lookbehind assertion (described below). However, in
3507 this case, the part of the subject before the real match does not have
3508 to be of fixed length, as lookbehind assertions do. The use of \K does
3509 not interfere with the setting of captured substrings. For example,
3510 when the pattern
3511
3512 (foo)\Kbar
3513
3514 matches "foobar", the first substring is still set to "foo".
3515
3516 Simple assertions
3517
3518 The final use of backslash is for certain simple assertions. An asser-
3519 tion specifies a condition that has to be met at a particular point in
3520 a match, without consuming any characters from the subject string. The
3521 use of subpatterns for more complicated assertions is described below.
3522 The backslashed assertions are:
3523
3524 \b matches at a word boundary
3525 \B matches when not at a word boundary
3526 \A matches at the start of the subject
3527 \Z matches at the end of the subject
3528 also matches before a newline at the end of the subject
3529 \z matches only at the end of the subject
3530 \G matches at the first matching position in the subject
3531
3532 These assertions may not appear in character classes (but note that \b
3533 has a different meaning, namely the backspace character, inside a char-
3534 acter class).
3535
3536 A word boundary is a position in the subject string where the current
3537 character and the previous character do not both match \w or \W (i.e.
3538 one matches \w and the other matches \W), or the start or end of the
3539 string if the first or last character matches \w, respectively.
3540
3541 The \A, \Z, and \z assertions differ from the traditional circumflex
3542 and dollar (described in the next section) in that they only ever match
3543 at the very start and end of the subject string, whatever options are
3544 set. Thus, they are independent of multiline mode. These three asser-
3545 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3546 affect only the behaviour of the circumflex and dollar metacharacters.
3547 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3548 cating that matching is to start at a point other than the beginning of
3549 the subject, \A can never match. The difference between \Z and \z is
3550 that \Z matches before a newline at the end of the string as well as at
3551 the very end, whereas \z matches only at the end.
3552
3553 The \G assertion is true only when the current matching position is at
3554 the start point of the match, as specified by the startoffset argument
3555 of pcre_exec(). It differs from \A when the value of startoffset is
3556 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3557 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3558 mentation where \G can be useful.
3559
3560 Note, however, that PCRE's interpretation of \G, as the start of the
3561 current match, is subtly different from Perl's, which defines it as the
3562 end of the previous match. In Perl, these can be different when the
3563 previously matched string was empty. Because PCRE does just one match
3564 at a time, it cannot reproduce this behaviour.
3565
3566 If all the alternatives of a pattern begin with \G, the expression is
3567 anchored to the starting match position, and the "anchored" flag is set
3568 in the compiled regular expression.
3569
3570
3571 CIRCUMFLEX AND DOLLAR
3572
3573 Outside a character class, in the default matching mode, the circumflex
3574 character is an assertion that is true only if the current matching
3575 point is at the start of the subject string. If the startoffset argu-
3576 ment of pcre_exec() is non-zero, circumflex can never match if the
3577 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3578 has an entirely different meaning (see below).
3579
3580 Circumflex need not be the first character of the pattern if a number
3581 of alternatives are involved, but it should be the first thing in each
3582 alternative in which it appears if the pattern is ever to match that
3583 branch. If all possible alternatives start with a circumflex, that is,
3584 if the pattern is constrained to match only at the start of the sub-
3585 ject, it is said to be an "anchored" pattern. (There are also other
3586 constructs that can cause a pattern to be anchored.)
3587
3588 A dollar character is an assertion that is true only if the current
3589 matching point is at the end of the subject string, or immediately
3590 before a newline at the end of the string (by default). Dollar need not
3591 be the last character of the pattern if a number of alternatives are
3592 involved, but it should be the last item in any branch in which it
3593 appears. Dollar has no special meaning in a character class.
3594
3595 The meaning of dollar can be changed so that it matches only at the
3596 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3597 compile time. This does not affect the \Z assertion.
3598
3599 The meanings of the circumflex and dollar characters are changed if the
3600 PCRE_MULTILINE option is set. When this is the case, a circumflex
3601 matches immediately after internal newlines as well as at the start of
3602 the subject string. It does not match after a newline that ends the
3603 string. A dollar matches before any newlines in the string, as well as
3604 at the very end, when PCRE_MULTILINE is set. When newline is specified
3605 as the two-character sequence CRLF, isolated CR and LF characters do
3606 not indicate newlines.
3607
3608 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3609 (where \n represents a newline) in multiline mode, but not otherwise.
3610 Consequently, patterns that are anchored in single line mode because
3611 all branches start with ^ are not anchored in multiline mode, and a
3612 match for circumflex is possible when the startoffset argument of
3613 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3614 PCRE_MULTILINE is set.
3615
3616 Note that the sequences \A, \Z, and \z can be used to match the start
3617 and end of the subject in both modes, and if all branches of a pattern
3618 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3619 set.
3620
3621
3622 FULL STOP (PERIOD, DOT)
3623
3624 Outside a character class, a dot in the pattern matches any one charac-
3625 ter in the subject string except (by default) a character that signi-
3626 fies the end of a line. In UTF-8 mode, the matched character may be
3627 more than one byte long.
3628
3629 When a line ending is defined as a single character, dot never matches
3630 that character; when the two-character sequence CRLF is used, dot does
3631 not match CR if it is immediately followed by LF, but otherwise it
3632 matches all characters (including isolated CRs and LFs). When any Uni-
3633 code line endings are being recognized, dot does not match CR or LF or
3634 any of the other line ending characters.
3635
3636 The behaviour of dot with regard to newlines can be changed. If the
3637 PCRE_DOTALL option is set, a dot matches any one character, without
3638 exception. If the two-character sequence CRLF is present in the subject
3639 string, it takes two dots to match it.
3640
3641 The handling of dot is entirely independent of the handling of circum-
3642 flex and dollar, the only relationship being that they both involve
3643 newlines. Dot has no special meaning in a character class.
3644
3645
3646 MATCHING A SINGLE BYTE
3647
3648 Outside a character class, the escape sequence \C matches any one byte,
3649 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3650 line-ending characters. The feature is provided in Perl in order to
3651 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3652 acters into individual bytes, what remains in the string may be a mal-
3653 formed UTF-8 string. For this reason, the \C escape sequence is best
3654 avoided.
3655
3656 PCRE does not allow \C to appear in lookbehind assertions (described
3657 below), because in UTF-8 mode this would make it impossible to calcu-
3658 late the length of the lookbehind.
3659
3660
3661 SQUARE BRACKETS AND CHARACTER CLASSES
3662
3663 An opening square bracket introduces a character class, terminated by a
3664 closing square bracket. A closing square bracket on its own is not spe-
3665 cial. If a closing square bracket is required as a member of the class,
3666 it should be the first data character in the class (after an initial
3667 circumflex, if present) or escaped with a backslash.
3668
3669 A character class matches a single character in the subject. In UTF-8
3670 mode, the character may occupy more than one byte. A matched character
3671 must be in the set of characters defined by the class, unless the first
3672 character in the class definition is a circumflex, in which case the
3673 subject character must not be in the set defined by the class. If a
3674 circumflex is actually required as a member of the class, ensure it is
3675 not the first character, or escape it with a backslash.
3676
3677 For example, the character class [aeiou] matches any lower case vowel,
3678 while [^aeiou] matches any character that is not a lower case vowel.
3679 Note that a circumflex is just a convenient notation for specifying the
3680 characters that are in the class by enumerating those that are not. A
3681 class that starts with a circumflex is not an assertion: it still con-
3682 sumes a character from the subject string, and therefore it fails if
3683 the current pointer is at the end of the string.
3684
3685 In UTF-8 mode, characters with values greater than 255 can be included
3686 in a class as a literal string of bytes, or by using the \x{ escaping
3687 mechanism.
3688
3689 When caseless matching is set, any letters in a class represent both
3690 their upper case and lower case versions, so for example, a caseless
3691 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3692 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3693 understands the concept of case for characters whose values are less
3694 than 128, so caseless matching is always possible. For characters with
3695 higher values, the concept of case is supported if PCRE is compiled
3696 with Unicode property support, but not otherwise. If you want to use
3697 caseless matching for characters 128 and above, you must ensure that
3698 PCRE is compiled with Unicode property support as well as with UTF-8
3699 support.
3700
3701 Characters that might indicate line breaks are never treated in any
3702 special way when matching character classes, whatever line-ending
3703 sequence is in use, and whatever setting of the PCRE_DOTALL and
3704 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3705 of these characters.
3706
3707 The minus (hyphen) character can be used to specify a range of charac-
3708 ters in a character class. For example, [d-m] matches any letter
3709 between d and m, inclusive. If a minus character is required in a
3710 class, it must be escaped with a backslash or appear in a position
3711 where it cannot be interpreted as indicating a range, typically as the
3712 first or last character in the class.
3713
3714 It is not possible to have the literal character "]" as the end charac-
3715 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3716 two characters ("W" and "-") followed by a literal string "46]", so it
3717 would match "W46]" or "-46]". However, if the "]" is escaped with a
3718 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3719 preted as a class containing a range followed by two other characters.
3720 The octal or hexadecimal representation of "]" can also be used to end
3721 a range.
3722
3723 Ranges operate in the collating sequence of character values. They can
3724 also be used for characters specified numerically, for example
3725 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3726 are greater than 255, for example [\x{100}-\x{2ff}].
3727
3728 If a range that includes letters is used when caseless matching is set,
3729 it matches the letters in either case. For example, [W-c] is equivalent
3730 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3731 character tables for a French locale are in use, [\xc8-\xcb] matches
3732 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3733 concept of case for characters with values greater than 128 only when
3734 it is compiled with Unicode property support.
3735
3736 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3737 in a character class, and add the characters that they match to the
3738 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3739 flex can conveniently be used with the upper case character types to
3740 specify a more restricted set of characters than the matching lower
3741 case type. For example, the class [^\W_] matches any letter or digit,
3742 but not underscore.
3743
3744 The only metacharacters that are recognized in character classes are
3745 backslash, hyphen (only where it can be interpreted as specifying a
3746 range), circumflex (only at the start), opening square bracket (only
3747 when it can be interpreted as introducing a POSIX class name - see the
3748 next section), and the terminating closing square bracket. However,
3749 escaping other non-alphanumeric characters does no harm.
3750
3751
3752 POSIX CHARACTER CLASSES
3753
3754 Perl supports the POSIX notation for character classes. This uses names
3755 enclosed by [: and :] within the enclosing square brackets. PCRE also
3756 supports this notation. For example,
3757
3758 [01[:alpha:]%]
3759
3760 matches "0", "1", any alphabetic character, or "%". The supported class
3761 names are
3762
3763 alnum letters and digits
3764 alpha letters
3765 ascii character codes 0 - 127
3766 blank space or tab only
3767 cntrl control characters
3768 digit decimal digits (same as \d)
3769 graph printing characters, excluding space
3770 lower lower case letters
3771 print printing characters, including space
3772 punct printing characters, excluding letters and digits
3773 space white space (not quite the same as \s)
3774 upper upper case letters
3775 word "word" characters (same as \w)
3776 xdigit hexadecimal digits
3777
3778 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3779 and space (32). Notice that this list includes the VT character (code
3780 11). This makes "space" different to \s, which does not include VT (for
3781 Perl compatibility).
3782
3783 The name "word" is a Perl extension, and "blank" is a GNU extension
3784 from Perl 5.8. Another Perl extension is negation, which is indicated
3785 by a ^ character after the colon. For example,
3786
3787 [12[:^digit:]]
3788
3789 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3790 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3791 these are not supported, and an error is given if they are encountered.
3792
3793 In UTF-8 mode, characters with values greater than 128 do not match any
3794 of the POSIX character classes.
3795
3796
3797 VERTICAL BAR
3798
3799 Vertical bar characters are used to separate alternative patterns. For
3800 example, the pattern
3801
3802 gilbert|sullivan
3803
3804 matches either "gilbert" or "sullivan". Any number of alternatives may
3805 appear, and an empty alternative is permitted (matching the empty
3806 string). The matching process tries each alternative in turn, from left
3807 to right, and the first one that succeeds is used. If the alternatives
3808 are within a subpattern (defined below), "succeeds" means matching the
3809 rest of the main pattern as well as the alternative in the subpattern.
3810
3811
3812 INTERNAL OPTION SETTING
3813
3814 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3815 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3816 within the pattern by a sequence of Perl option letters enclosed
3817 between "(?" and ")". The option letters are
3818
3819 i for PCRE_CASELESS
3820 m for PCRE_MULTILINE
3821 s for PCRE_DOTALL
3822 x for PCRE_EXTENDED
3823
3824 For example, (?im) sets caseless, multiline matching. It is also possi-
3825 ble to unset these options by preceding the letter with a hyphen, and a
3826 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3827 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3828 is also permitted. If a letter appears both before and after the
3829 hyphen, the option is unset.
3830
3831 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3832 can be changed in the same way as the Perl-compatible options by using
3833 the characters J, U and X respectively.
3834
3835 When an option change occurs at top level (that is, not inside subpat-
3836 tern parentheses), the change applies to the remainder of the pattern
3837 that follows. If the change is placed right at the start of a pattern,
3838 PCRE extracts it into the global options (and it will therefore show up
3839 in data extracted by the pcre_fullinfo() function).
3840
3841 An option change within a subpattern (see below for a description of
3842 subpatterns) affects only that part of the current pattern that follows
3843 it, so
3844
3845 (a(?i)b)c
3846
3847 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3848 used). By this means, options can be made to have different settings
3849 in different parts of the pattern. Any changes made in one alternative
3850 do carry on into subsequent branches within the same subpattern. For
3851 example,
3852
3853 (a(?i)b|c)
3854
3855 matches "ab", "aB", "c", and "C", even though when matching "C" the
3856 first branch is abandoned before the option setting. This is because
3857 the effects of option settings happen at compile time. There would be
3858 some very weird behaviour otherwise.
3859
3860 Note: There are other PCRE-specific options that can be set by the
3861 application when the compile or match functions are called. In some
3862 cases the pattern can contain special leading sequences to override
3863 what the application has set or what has been defaulted. Details are
3864 given in the section entitled "Newline sequences" above.
3865
3866
3867 SUBPATTERNS
3868
3869 Subpatterns are delimited by parentheses (round brackets), which can be
3870 nested. Turning part of a pattern into a subpattern does two things:
3871
3872 1. It localizes a set of alternatives. For example, the pattern
3873
3874 cat(aract|erpillar|)
3875
3876 matches one of the words "cat", "cataract", or "caterpillar". Without
3877 the parentheses, it would match "cataract", "erpillar" or an empty
3878 string.
3879
3880 2. It sets up the subpattern as a capturing subpattern. This means
3881 that, when the whole pattern matches, that portion of the subject
3882 string that matched the subpattern is passed back to the caller via the
3883 ovector argument of pcre_exec(). Opening parentheses are counted from
3884 left to right (starting from 1) to obtain numbers for the capturing
3885 subpatterns.
3886
3887 For example, if the string "the red king" is matched against the pat-
3888 tern
3889
3890 the ((red|white) (king|queen))
3891
3892 the captured substrings are "red king", "red", and "king", and are num-
3893 bered 1, 2, and 3, respectively.
3894
3895 The fact that plain parentheses fulfil two functions is not always
3896 helpful. There are often times when a grouping subpattern is required
3897 without a capturing requirement. If an opening parenthesis is followed
3898 by a question mark and a colon, the subpattern does not do any captur-
3899 ing, and is not counted when computing the number of any subsequent
3900 capturing subpatterns. For example, if the string "the white queen" is
3901 matched against the pattern
3902
3903 the ((?:red|white) (king|queen))
3904
3905 the captured substrings are "white queen" and "queen", and are numbered
3906 1 and 2. The maximum number of capturing subpatterns is 65535.
3907
3908 As a convenient shorthand, if any option settings are required at the
3909 start of a non-capturing subpattern, the option letters may appear
3910 between the "?" and the ":". Thus the two patterns
3911
3912 (?i:saturday|sunday)
3913 (?:(?i)saturday|sunday)
3914
3915 match exactly the same set of strings. Because alternative branches are
3916 tried from left to right, and options are not reset until the end of
3917 the subpattern is reached, an option setting in one branch does affect
3918 subsequent branches, so the above patterns match "SUNDAY" as well as
3919 "Saturday".
3920
3921
3922 DUPLICATE SUBPATTERN NUMBERS
3923
3924 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3925 uses the same numbers for its capturing parentheses. Such a subpattern
3926 starts with (?| and is itself a non-capturing subpattern. For example,
3927 consider this pattern:
3928
3929 (?|(Sat)ur|(Sun))day
3930
3931 Because the two alternatives are inside a (?| group, both sets of cap-
3932 turing parentheses are numbered one. Thus, when the pattern matches,
3933 you can look at captured substring number one, whichever alternative
3934 matched. This construct is useful when you want to capture part, but
3935 not all, of one of a number of alternatives. Inside a (?| group, paren-
3936 theses are numbered as usual, but the number is reset at the start of
3937 each branch. The numbers of any capturing buffers that follow the sub-
3938 pattern start after the highest number used in any branch. The follow-
3939 ing example is taken from the Perl documentation. The numbers under-
3940 neath show in which buffer the captured content will be stored.
3941
3942 # before ---------------branch-reset----------- after
3943 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3944 # 1 2 2 3 2 3 4
3945
3946 A backreference or a recursive call to a numbered subpattern always
3947 refers to the first one in the pattern with the given number.
3948
3949 An alternative approach to using this "branch reset" feature is to use
3950 duplicate named subpatterns, as described in the next section.
3951
3952
3953 NAMED SUBPATTERNS
3954
3955 Identifying capturing parentheses by number is simple, but it can be
3956 very hard to keep track of the numbers in complicated regular expres-
3957 sions. Furthermore, if an expression is modified, the numbers may
3958 change. To help with this difficulty, PCRE supports the naming of sub-
3959 patterns. This feature was not added to Perl until release 5.10. Python
3960 had the feature earlier, and PCRE introduced it at release 4.0, using
3961 the Python syntax. PCRE now supports both the Perl and the Python syn-
3962 tax.
3963
3964 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3965 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3966 to capturing parentheses from other parts of the pattern, such as back-
3967 references, recursion, and conditions, can be made by name as well as
3968 by number.
3969
3970 Names consist of up to 32 alphanumeric characters and underscores.
3971 Named capturing parentheses are still allocated numbers as well as
3972 names, exactly as if the names were not present. The PCRE API provides
3973 function calls for extracting the name-to-number translation table from
3974 a compiled pattern. There is also a convenience function for extracting
3975 a captured substring by name.
3976
3977 By default, a name must be unique within a pattern, but it is possible
3978 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3979 time. This can be useful for patterns where only one instance of the
3980 named parentheses can match. Suppose you want to match the name of a
3981 weekday, either as a 3-letter abbreviation or as the full name, and in
3982 both cases you want to extract the abbreviation. This pattern (ignoring
3983 the line breaks) does the job:
3984
3985 (?<DN>Mon|Fri|Sun)(?:day)?|
3986 (?<DN>Tue)(?:sday)?|
3987 (?<DN>Wed)(?:nesday)?|
3988 (?<DN>Thu)(?:rsday)?|
3989 (?<DN>Sat)(?:urday)?
3990
3991 There are five capturing substrings, but only one is ever set after a
3992 match. (An alternative way of solving this problem is to use a "branch
3993 reset" subpattern, as described in the previous section.)
3994
3995 The convenience function for extracting the data by name returns the
3996 substring for the first (and in this example, the only) subpattern of
3997 that name that matched. This saves searching to find which numbered
3998 subpattern it was. If you make a reference to a non-unique named sub-
3999 pattern from elsewhere in the pattern, the one that corresponds to the
4000 lowest number is used. For further details of the interfaces for han-
4001 dling named subpatterns, see the pcreapi documentation.
4002
4003 Warning: You cannot use different names to distinguish between two sub-
4004 patterns with the same number (see the previous section) because PCRE
4005 uses only the numbers when matching.
4006
4007
4008 REPETITION
4009
4010 Repetition is specified by quantifiers, which can follow any of the
4011 following items:
4012
4013 a literal data character
4014 the dot metacharacter
4015 the \C escape sequence
4016 the \X escape sequence (in UTF-8 mode with Unicode properties)
4017 the \R escape sequence
4018 an escape such as \d that matches a single character
4019 a character class
4020 a back reference (see next section)
4021 a parenthesized subpattern (unless it is an assertion)
4022
4023 The general repetition quantifier specifies a minimum and maximum num-
4024 ber of permitted matches, by giving the two numbers in curly brackets
4025 (braces), separated by a comma. The numbers must be less than 65536,
4026 and the first must be less than or equal to the second. For example:
4027
4028 z{2,4}
4029
4030 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4031 special character. If the second number is omitted, but the comma is
4032 present, there is no upper limit; if the second number and the comma
4033 are both omitted, the quantifier specifies an exact number of required
4034 matches. Thus
4035
4036 [aeiou]{3,}
4037
4038 matches at least 3 successive vowels, but may match many more, while
4039
4040 \d{8}
4041
4042 matches exactly 8 digits. An opening curly bracket that appears in a
4043 position where a quantifier is not allowed, or one that does not match
4044 the syntax of a quantifier, is taken as a literal character. For exam-
4045 ple, {,6} is not a quantifier, but a literal string of four characters.
4046
4047 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4048 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4049 acters, each of which is represented by a two-byte sequence. Similarly,
4050 when Unicode property support is available, \X{3} matches three Unicode
4051 extended sequences, each of which may be several bytes long (and they
4052 may be of different lengths).
4053
4054 The quantifier {0} is permitted, causing the expression to behave as if
4055 the previous item and the quantifier were not present. This may be use-
4056 ful for subpatterns that are referenced as subroutines from elsewhere
4057 in the pattern. Items other than subpatterns that have a {0} quantifier
4058 are omitted from the compiled pattern.
4059
4060 For convenience, the three most common quantifiers have single-charac-
4061 ter abbreviations:
4062
4063 * is equivalent to {0,}
4064 + is equivalent to {1,}
4065 ? is equivalent to {0,1}
4066
4067 It is possible to construct infinite loops by following a subpattern
4068 that can match no characters with a quantifier that has no upper limit,
4069 for example:
4070
4071 (a?)*
4072
4073 Earlier versions of Perl and PCRE used to give an error at compile time
4074 for such patterns. However, because there are cases where this can be
4075 useful, such patterns are now accepted, but if any repetition of the
4076 subpattern does in fact match no characters, the loop is forcibly bro-
4077 ken.
4078
4079 By default, the quantifiers are "greedy", that is, they match as much
4080 as possible (up to the maximum number of permitted times), without
4081 causing the rest of the pattern to fail. The classic example of where
4082 this gives problems is in trying to match comments in C programs. These
4083 appear between /* and */ and within the comment, individual * and /
4084 characters may appear. An attempt to match C comments by applying the
4085 pattern
4086
4087 /\*.*\*/
4088
4089 to the string
4090
4091 /* first comment */ not comment /* second comment */
4092
4093 fails, because it matches the entire string owing to the greediness of
4094 the .* item.
4095
4096 However, if a quantifier is followed by a question mark, it ceases to
4097 be greedy, and instead matches the minimum number of times possible, so
4098 the pattern
4099
4100 /\*.*?\*/
4101
4102 does the right thing with the C comments. The meaning of the various
4103 quantifiers is not otherwise changed, just the preferred number of
4104 matches. Do not confuse this use of question mark with its use as a
4105 quantifier in its own right. Because it has two uses, it can sometimes
4106 appear doubled, as in
4107
4108 \d??\d
4109
4110 which matches one digit by preference, but can match two if that is the
4111 only way the rest of the pattern matches.
4112
4113 If the PCRE_UNGREEDY option is set (an option that is not available in
4114 Perl), the quantifiers are not greedy by default, but individual ones
4115 can be made greedy by following them with a question mark. In other
4116 words, it inverts the default behaviour.
4117
4118 When a parenthesized subpattern is quantified with a minimum repeat
4119 count that is greater than 1 or with a limited maximum, more memory is
4120 required for the compiled pattern, in proportion to the size of the
4121 minimum or maximum.
4122
4123 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4124 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4125 the pattern is implicitly anchored, because whatever follows will be
4126 tried against every character position in the subject string, so there
4127 is no point in retrying the overall match at any position after the
4128 first. PCRE normally treats such a pattern as though it were preceded
4129 by \A.
4130
4131 In cases where it is known that the subject string contains no new-
4132 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4133 mization, or alternatively using ^ to indicate anchoring explicitly.
4134
4135 However, there is one situation where the optimization cannot be used.
4136 When .* is inside capturing parentheses that are the subject of a
4137 backreference elsewhere in the pattern, a match at the start may fail
4138 where a later one succeeds. Consider, for example:
4139
4140 (.*)abc\1
4141
4142 If the subject is "xyz123abc123" the match point is the fourth charac-
4143 ter. For this reason, such a pattern is not implicitly anchored.
4144
4145 When a capturing subpattern is repeated, the value captured is the sub-
4146 string that matched the final iteration. For example, after
4147
4148 (tweedle[dume]{3}\s*)+
4149
4150 has matched "tweedledum tweedledee" the value of the captured substring
4151 is "tweedledee". However, if there are nested capturing subpatterns,
4152 the corresponding captured values may have been set in previous itera-
4153 tions. For example, after
4154
4155 /(a|(b))+/
4156
4157 matches "aba" the value of the second captured substring is "b".
4158
4159
4160 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4161
4162 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4163 repetition, failure of what follows normally causes the repeated item
4164 to be re-evaluated to see if a different number of repeats allows the
4165 rest of the pattern to match. Sometimes it is useful to prevent this,
4166 either to change the nature of the match, or to cause it fail earlier
4167 than it otherwise might, when the author of the pattern knows there is
4168 no point in carrying on.
4169
4170 Consider, for example, the pattern \d+foo when applied to the subject
4171 line
4172
4173 123456bar
4174
4175 After matching all 6 digits and then failing to match "foo", the normal
4176 action of the matcher is to try again with only 5 digits matching the
4177 \d+ item, and then with 4, and so on, before ultimately failing.
4178 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4179 the means for specifying that once a subpattern has matched, it is not
4180 to be re-evaluated in this way.
4181
4182 If we use atomic grouping for the previous example, the matcher gives
4183 up immediately on failing to match "foo" the first time. The notation
4184 is a kind of special parenthesis, starting with (?> as in this example:
4185
4186 (?>\d+)foo
4187
4188 This kind of parenthesis "locks up" the part of the pattern it con-
4189 tains once it has matched, and a failure further into the pattern is
4190 prevented from backtracking into it. Backtracking past it to previous
4191 items, however, works as normal.
4192
4193 An alternative description is that a subpattern of this type matches
4194 the string of characters that an identical standalone pattern would
4195 match, if anchored at the current point in the subject string.
4196
4197 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4198 such as the above example can be thought of as a maximizing repeat that
4199 must swallow everything it can. So, while both \d+ and \d+? are pre-
4200 pared to adjust the number of digits they match in order to make the
4201 rest of the pattern match, (?>\d+) can only match an entire sequence of
4202 digits.
4203
4204 Atomic groups in general can of course contain arbitrarily complicated
4205 subpatterns, and can be nested. However, when the subpattern for an
4206 atomic group is just a single repeated item, as in the example above, a
4207 simpler notation, called a "possessive quantifier" can be used. This
4208 consists of an additional + character following a quantifier. Using
4209 this notation, the previous example can be rewritten as
4210
4211 \d++foo
4212
4213 Note that a possessive quantifier can be used with an entire group, for
4214 example:
4215
4216 (abc|xyz){2,3}+
4217
4218 Possessive quantifiers are always greedy; the setting of the
4219 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4220 simpler forms of atomic group. However, there is no difference in the
4221 meaning of a possessive quantifier and the equivalent atomic group,
4222 though there may be a performance difference; possessive quantifiers
4223 should be slightly faster.
4224
4225 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4226 tax. Jeffrey Friedl originated the idea (and the name) in the first
4227 edition of his book. Mike McCloskey liked it, so implemented it when he
4228 built Sun's Java package, and PCRE copied it from there. It ultimately
4229 found its way into Perl at release 5.10.
4230
4231 PCRE has an optimization that automatically "possessifies" certain sim-
4232 ple pattern constructs. For example, the sequence A+B is treated as
4233 A++B because there is no point in backtracking into a sequence of A's
4234 when B must follow.
4235
4236 When a pattern contains an unlimited repeat inside a subpattern that
4237 can itself be repeated an unlimited number of times, the use of an
4238 atomic group is the only way to avoid some failing matches taking a
4239 very long time indeed. The pattern
4240
4241 (\D+|<\d+>)*[!?]
4242
4243 matches an unlimited number of substrings that either consist of non-
4244 digits, or digits enclosed in <>, followed by either ! or ?. When it
4245 matches, it runs quickly. However, if it is applied to
4246
4247 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4248
4249 it takes a long time before reporting failure. This is because the
4250 string can be divided between the internal \D+ repeat and the external
4251 * repeat in a large number of ways, and all have to be tried. (The
4252 example uses [!?] rather than a single character at the end, because
4253 both PCRE and Perl have an optimization that allows for fast failure
4254 when a single character is used. They remember the last single charac-
4255 ter that is required for a match, and fail early if it is not present
4256 in the string.) If the pattern is changed so that it uses an atomic
4257 group, like this:
4258
4259 ((?>\D+)|<\d+>)*[!?]
4260
4261 sequences of non-digits cannot be broken, and failure happens quickly.
4262
4263
4264 BACK REFERENCES
4265
4266 Outside a character class, a backslash followed by a digit greater than
4267 0 (and possibly further digits) is a back reference to a capturing sub-
4268 pattern earlier (that is, to its left) in the pattern, provided there
4269 have been that many previous capturing left parentheses.
4270
4271 However, if the decimal number following the backslash is less than 10,
4272 it is always taken as a back reference, and causes an error only if
4273 there are not that many capturing left parentheses in the entire pat-
4274 tern. In other words, the parentheses that are referenced need not be
4275 to the left of the reference for numbers less than 10. A "forward back
4276 reference" of this type can make sense when a repetition is involved
4277 and the subpattern to the right has participated in an earlier itera-
4278 tion.
4279
4280 It is not possible to have a numerical "forward back reference" to a
4281 subpattern whose number is 10 or more using this syntax because a
4282 sequence such as \50 is interpreted as a character defined in octal.
4283 See the subsection entitled "Non-printing characters" above for further
4284 details of the handling of digits following a backslash. There is no
4285 such problem when named parentheses are used. A back reference to any
4286 subpattern is possible using named parentheses (see below).
4287
4288 Another way of avoiding the ambiguity inherent in the use of digits
4289 following a backslash is to use the \g escape sequence, which is a fea-
4290 ture introduced in Perl 5.10. This escape must be followed by an
4291 unsigned number or a negative number, optionally enclosed in braces.
4292 These examples are all identical:
4293
4294 (ring), \1
4295 (ring), \g1
4296 (ring), \g{1}
4297
4298 An unsigned number specifies an absolute reference without the ambigu-
4299 ity that is present in the older syntax. It is also useful when literal
4300 digits follow the reference. A negative number is a relative reference.
4301 Consider this example:
4302
4303 (abc(def)ghi)\g{-1}
4304
4305 The sequence \g{-1} is a reference to the most recently started captur-
4306 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4307 \g{-2} would be equivalent to \1. The use of relative references can be
4308 helpful in long patterns, and also in patterns that are created by
4309 joining together fragments that contain references within themselves.
4310
4311 A back reference matches whatever actually matched the capturing sub-
4312 pattern in the current subject string, rather than anything matching
4313 the subpattern itself (see "Subpatterns as subroutines" below for a way
4314 of doing that). So the pattern
4315
4316 (sens|respons)e and \1ibility
4317
4318 matches "sense and sensibility" and "response and responsibility", but
4319 not "sense and responsibility". If caseful matching is in force at the
4320 time of the back reference, the case of letters is relevant. For exam-
4321 ple,
4322
4323 ((?i)rah)\s+\1
4324
4325 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4326 original capturing subpattern is matched caselessly.
4327
4328 There are several different ways of writing back references to named
4329 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4330 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4331 unified back reference syntax, in which \g can be used for both numeric
4332 and named references, is also supported. We could rewrite the above
4333 example in any of the following ways:
4334
4335 (?<p1>(?i)rah)\s+\k<p1>
4336 (?'p1'(?i)rah)\s+\k{p1}
4337 (?P<p1>(?i)rah)\s+(?P=p1)
4338 (?<p1>(?i)rah)\s+\g{p1}
4339
4340 A subpattern that is referenced by name may appear in the pattern
4341 before or after the reference.
4342
4343 There may be more than one back reference to the same subpattern. If a
4344 subpattern has not actually been used in a particular match, any back
4345 references to it always fail. For example, the pattern
4346
4347 (a|(bc))\2
4348
4349 always fails if it starts to match "a" rather than "bc". Because there
4350 may be many capturing parentheses in a pattern, all digits following
4351 the backslash are taken as part of a potential back reference number.
4352 If the pattern continues with a digit character, some delimiter must be
4353 used to terminate the back reference. If the PCRE_EXTENDED option is
4354 set, this can be whitespace. Otherwise an empty comment (see "Com-
4355 ments" below) can be used.
4356
4357 A back reference that occurs inside the parentheses to which it refers
4358 fails when the subpattern is first used, so, for example, (a\1) never
4359 matches. However, such references can be useful inside repeated sub-
4360 patterns. For example, the pattern
4361
4362 (a|b\1)+
4363
4364 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4365 ation of the subpattern, the back reference matches the character
4366 string corresponding to the previous iteration. In order for this to
4367 work, the pattern must be such that the first iteration does not need
4368 to match the back reference. This can be done using alternation, as in
4369 the example above, or by a quantifier with a minimum of zero.
4370
4371
4372 ASSERTIONS
4373
4374 An assertion is a test on the characters following or preceding the
4375 current matching point that does not actually consume any characters.
4376 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4377 described above.
4378
4379 More complicated assertions are coded as subpatterns. There are two
4380 kinds: those that look ahead of the current position in the subject
4381 string, and those that look behind it. An assertion subpattern is
4382 matched in the normal way, except that it does not cause the current
4383 matching position to be changed.
4384
4385 Assertion subpatterns are not capturing subpatterns, and may not be
4386 repeated, because it makes no sense to assert the same thing several
4387 times. If any kind of assertion contains capturing subpatterns within
4388 it, these are counted for the purposes of numbering the capturing sub-
4389 patterns in the whole pattern. However, substring capturing is carried
4390 out only for positive assertions, because it does not make sense for
4391 negative assertions.
4392
4393 Lookahead assertions
4394
4395 Lookahead assertions start with (?= for positive assertions and (?! for
4396 negative assertions. For example,
4397
4398 \w+(?=;)
4399
4400 matches a word followed by a semicolon, but does not include the semi-
4401 colon in the match, and
4402
4403 foo(?!bar)
4404
4405 matches any occurrence of "foo" that is not followed by "bar". Note
4406 that the apparently similar pattern
4407
4408 (?!foo)bar
4409
4410 does not find an occurrence of "bar" that is preceded by something
4411 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4412 the assertion (?!foo) is always true when the next three characters are
4413 "bar". A lookbehind assertion is needed to achieve the other effect.
4414
4415 If you want to force a matching failure at some point in a pattern, the
4416 most convenient way to do it is with (?!) because an empty string
4417 always matches, so an assertion that requires there not to be an empty
4418 string must always fail.
4419
4420 Lookbehind assertions
4421
4422 Lookbehind assertions start with (?<= for positive assertions and (?<!
4423 for negative assertions. For example,
4424
4425 (?<!foo)bar
4426
4427 does find an occurrence of "bar" that is not preceded by "foo". The
4428 contents of a lookbehind assertion are restricted such that all the
4429 strings it matches must have a fixed length. However, if there are sev-
4430 eral top-level alternatives, they do not all have to have the same
4431 fixed length. Thus
4432
4433 (?<=bullock|donkey)
4434
4435 is permitted, but
4436
4437 (?<!dogs?|cats?)
4438
4439 causes an error at compile time. Branches that match different length
4440 strings are permitted only at the top level of a lookbehind assertion.
4441 This is an extension compared with Perl (at least for 5.8), which
4442 requires all branches to match the same length of string. An assertion
4443 such as
4444
4445 (?<=ab(c|de))
4446
4447 is not permitted, because its single top-level branch can match two
4448 different lengths, but it is acceptable if rewritten to use two top-
4449 level branches:
4450
4451 (?<=abc|abde)
4452
4453 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4454 instead of a lookbehind assertion; this is not restricted to a fixed-
4455 length.
4456
4457 The implementation of lookbehind assertions is, for each alternative,
4458 to temporarily move the current position back by the fixed length and
4459 then try to match. If there are insufficient characters before the cur-
4460 rent position, the assertion fails.
4461
4462 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4463 mode) to appear in lookbehind assertions, because it makes it impossi-
4464 ble to calculate the length of the lookbehind. The \X and \R escapes,
4465 which can match different numbers of bytes, are also not permitted.
4466
4467 Possessive quantifiers can be used in conjunction with lookbehind
4468 assertions to specify efficient matching at the end of the subject
4469 string. Consider a simple pattern such as
4470
4471 abcd$
4472
4473 when applied to a long string that does not match. Because matching
4474 proceeds from left to right, PCRE will look for each "a" in the subject
4475 and then see if what follows matches the rest of the pattern. If the
4476 pattern is specified as
4477
4478 ^.*abcd$
4479
4480 the initial .* matches the entire string at first, but when this fails
4481 (because there is no following "a"), it backtracks to match all but the
4482 last character, then all but the last two characters, and so on. Once
4483 again the search for "a" covers the entire string, from right to left,
4484 so we are no better off. However, if the pattern is written as
4485
4486 ^.*+(?<=abcd)
4487
4488 there can be no backtracking for the .*+ item; it can match only the
4489 entire string. The subsequent lookbehind assertion does a single test
4490 on the last four characters. If it fails, the match fails immediately.
4491 For long strings, this approach makes a significant difference to the
4492 processing time.
4493
4494 Using multiple assertions
4495
4496 Several assertions (of any sort) may occur in succession. For example,
4497
4498 (?<=\d{3})(?<!999)foo
4499
4500 matches "foo" preceded by three digits that are not "999". Notice that
4501 each of the assertions is applied independently at the same point in
4502 the subject string. First there is a check that the previous three
4503 characters are all digits, and then there is a check that the same
4504 three characters are not "999". This pattern does not match "foo" pre-
4505 ceded by six characters, the first of which are digits and the last
4506 three of which are not "999". For example, it doesn't match "123abc-
4507 foo". A pattern to do that is
4508
4509 (?<=\d{3}...)(?<!999)foo
4510
4511 This time the first assertion looks at the preceding six characters,
4512 checking that the first three are digits, and then the second assertion
4513 checks that the preceding three characters are not "999".
4514
4515 Assertions can be nested in any combination. For example,
4516
4517 (?<=(?<!foo)bar)baz
4518
4519 matches an occurrence of "baz" that is preceded by "bar" which in turn
4520 is not preceded by "foo", while
4521
4522 (?<=\d{3}(?!999)...)foo
4523
4524 is another pattern that matches "foo" preceded by three digits and any
4525 three characters that are not "999".
4526
4527
4528 CONDITIONAL SUBPATTERNS
4529
4530 It is possible to cause the matching process to obey a subpattern con-
4531 ditionally or to choose between two alternative subpatterns, depending
4532 on the result of an assertion, or whether a previous capturing subpat-
4533 tern matched or not. The two possible forms of conditional subpattern
4534 are
4535
4536 (?(condition)yes-pattern)
4537 (?(condition)yes-pattern|no-pattern)
4538
4539 If the condition is satisfied, the yes-pattern is used; otherwise the
4540 no-pattern (if present) is used. If there are more than two alterna-
4541 tives in the subpattern, a compile-time error occurs.
4542
4543 There are four kinds of condition: references to subpatterns, refer-
4544 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4545
4546 Checking for a used subpattern by number
4547
4548 If the text between the parentheses consists of a sequence of digits,
4549 the condition is true if the capturing subpattern of that number has
4550 previously matched. An alternative notation is to precede the digits
4551 with a plus or minus sign. In this case, the subpattern number is rela-
4552 tive rather than absolute. The most recently opened parentheses can be
4553 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4554 looping constructs it can also make sense to refer to subsequent groups
4555 with constructs such as (?(+2).
4556
4557 Consider the following pattern, which contains non-significant white
4558 space to make it more readable (assume the PCRE_EXTENDED option) and to
4559 divide it into three parts for ease of discussion:
4560
4561 ( \( )? [^()]+ (?(1) \) )
4562
4563 The first part matches an optional opening parenthesis, and if that
4564 character is present, sets it as the first captured substring. The sec-
4565 ond part matches one or more characters that are not parentheses. The
4566 third part is a conditional subpattern that tests whether the first set
4567 of parentheses matched or not. If they did, that is, if subject started
4568 with an opening parenthesis, the condition is true, and so the yes-pat-
4569 tern is executed and a closing parenthesis is required. Otherwise,
4570 since no-pattern is not present, the subpattern matches nothing. In
4571 other words, this pattern matches a sequence of non-parentheses,
4572 optionally enclosed in parentheses.
4573
4574 If you were embedding this pattern in a larger one, you could use a
4575 relative reference:
4576
4577 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4578
4579 This makes the fragment independent of the parentheses in the larger
4580 pattern.
4581
4582 Checking for a used subpattern by name
4583
4584 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4585 used subpattern by name. For compatibility with earlier versions of
4586 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4587 also recognized. However, there is a possible ambiguity with this syn-
4588 tax, because subpattern names may consist entirely of digits. PCRE
4589 looks first for a named subpattern; if it cannot find one and the name
4590 consists entirely of digits, PCRE looks for a subpattern of that num-
4591 ber, which must be greater than zero. Using subpattern names that con-
4592 sist entirely of digits is not recommended.
4593
4594 Rewriting the above example to use a named subpattern gives this:
4595
4596 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4597
4598
4599 Checking for pattern recursion
4600
4601 If the condition is the string (R), and there is no subpattern with the
4602 name R, the condition is true if a recursive call to the whole pattern
4603 or any subpattern has been made. If digits or a name preceded by amper-
4604 sand follow the letter R, for example:
4605
4606 (?(R3)...) or (?(R&name)...)
4607
4608 the condition is true if the most recent recursion is into the subpat-
4609 tern whose number or name is given. This condition does not check the
4610 entire recursion stack.
4611
4612 At "top level", all these recursion test conditions are false. Recur-
4613 sive patterns are described below.
4614
4615 Defining subpatterns for use by reference only
4616
4617 If the condition is the string (DEFINE), and there is no subpattern
4618 with the name DEFINE, the condition is always false. In this case,
4619 there may be only one alternative in the subpattern. It is always
4620 skipped if control reaches this point in the pattern; the idea of
4621 DEFINE is that it can be used to define "subroutines" that can be ref-
4622 erenced from elsewhere. (The use of "subroutines" is described below.)
4623 For example, a pattern to match an IPv4 address could be written like
4624 this (ignore whitespace and line breaks):
4625
4626 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4627 \b (?&byte) (\.(?&byte)){3} \b
4628
4629 The first part of the pattern is a DEFINE group inside which a another
4630 group named "byte" is defined. This matches an individual component of
4631 an IPv4 address (a number less than 256). When matching takes place,
4632 this part of the pattern is skipped because DEFINE acts like a false
4633 condition.
4634
4635 The rest of the pattern uses references to the named group to match the
4636 four dot-separated components of an IPv4 address, insisting on a word
4637 boundary at each end.
4638
4639 Assertion conditions
4640
4641 If the condition is not in any of the above formats, it must be an
4642 assertion. This may be a positive or negative lookahead or lookbehind
4643 assertion. Consider this pattern, again containing non-significant
4644 white space, and with the two alternatives on the second line:
4645
4646 (?(?=[^a-z]*[a-z])
4647 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4648
4649 The condition is a positive lookahead assertion that matches an
4650 optional sequence of non-letters followed by a letter. In other words,
4651 it tests for the presence of at least one letter in the subject. If a
4652 letter is found, the subject is matched against the first alternative;
4653 otherwise it is matched against the second. This pattern matches
4654 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4655 letters and dd are digits.
4656
4657
4658 COMMENTS
4659
4660 The sequence (?# marks the start of a comment that continues up to the
4661 next closing parenthesis. Nested parentheses are not permitted. The
4662 characters that make up a comment play no part in the pattern matching
4663 at all.
4664
4665 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4666 character class introduces a comment that continues to immediately
4667 after the next newline in the pattern.
4668
4669
4670 RECURSIVE PATTERNS
4671
4672 Consider the problem of matching a string in parentheses, allowing for
4673 unlimited nested parentheses. Without the use of recursion, the best
4674 that can be done is to use a pattern that matches up to some fixed
4675 depth of nesting. It is not possible to handle an arbitrary nesting
4676 depth.
4677
4678 For some time, Perl has provided a facility that allows regular expres-
4679 sions to recurse (amongst other things). It does this by interpolating
4680 Perl code in the expression at run time, and the code can refer to the
4681 expression itself. A Perl pattern using code interpolation to solve the
4682 parentheses problem can be created like this:
4683
4684 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4685
4686 The (?p{...}) item interpolates Perl code at run time, and in this case
4687 refers recursively to the pattern in which it appears.
4688
4689 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4690 it supports special syntax for recursion of the entire pattern, and
4691 also for individual subpattern recursion. After its introduction in
4692 PCRE and Python, this kind of recursion was introduced into Perl at
4693 release 5.10.
4694
4695 A special item that consists of (? followed by a number greater than
4696 zero and a closing parenthesis is a recursive call of the subpattern of
4697 the given number, provided that it occurs inside that subpattern. (If
4698 not, it is a "subroutine" call, which is described in the next sec-
4699 tion.) The special item (?R) or (?0) is a recursive call of the entire
4700 regular expression.
4701
4702 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4703 always treated as an atomic group. That is, once it has matched some of
4704 the subject string, it is never re-entered, even if it contains untried
4705 alternatives and there is a subsequent matching failure.
4706
4707 This PCRE pattern solves the nested parentheses problem (assume the
4708 PCRE_EXTENDED option is set so that white space is ignored):
4709
4710 \( ( (?>[^()]+) | (?R) )* \)
4711
4712 First it matches an opening parenthesis. Then it matches any number of
4713 substrings which can either be a sequence of non-parentheses, or a
4714 recursive match of the pattern itself (that is, a correctly parenthe-
4715 sized substring). Finally there is a closing parenthesis.
4716
4717 If this were part of a larger pattern, you would not want to recurse
4718 the entire pattern, so instead you could use this:
4719
4720 ( \( ( (?>[^()]+) | (?1) )* \) )
4721
4722 We have put the pattern into parentheses, and caused the recursion to
4723 refer to them instead of the whole pattern.
4724
4725 In a larger pattern, keeping track of parenthesis numbers can be
4726 tricky. This is made easier by the use of relative references. (A Perl
4727 5.10 feature.) Instead of (?1) in the pattern above you can write
4728 (?-2) to refer to the second most recently opened parentheses preceding
4729 the recursion. In other words, a negative number counts capturing
4730 parentheses leftwards from the point at which it is encountered.
4731
4732 It is also possible to refer to subsequently opened parentheses, by
4733 writing references such as (?+2). However, these cannot be recursive
4734 because the reference is not inside the parentheses that are refer-
4735 enced. They are always "subroutine" calls, as described in the next
4736 section.
4737
4738 An alternative approach is to use named parentheses instead. The Perl
4739 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4740 supported. We could rewrite the above example as follows:
4741
4742 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4743
4744 If there is more than one subpattern with the same name, the earliest
4745 one is used.
4746
4747 This particular example pattern that we have been looking at contains
4748 nested unlimited repeats, and so the use of atomic grouping for match-
4749 ing strings of non-parentheses is important when applying the pattern
4750 to strings that do not match. For example, when this pattern is applied
4751 to
4752
4753 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4754
4755 it yields "no match" quickly. However, if atomic grouping is not used,
4756 the match runs for a very long time indeed because there are so many
4757 different ways the + and * repeats can carve up the subject, and all
4758 have to be tested before failure can be reported.
4759
4760 At the end of a match, the values set for any capturing subpatterns are
4761 those from the outermost level of the recursion at which the subpattern
4762 value is set. If you want to obtain intermediate values, a callout
4763 function can be used (see below and the pcrecallout documentation). If
4764 the pattern above is matched against
4765
4766 (ab(cd)ef)
4767
4768 the value for the capturing parentheses is "ef", which is the last
4769 value taken on at the top level. If additional parentheses are added,
4770 giving
4771
4772 \( ( ( (?>[^()]+) | (?R) )* ) \)
4773 ^ ^
4774 ^ ^
4775
4776 the string they capture is "ab(cd)ef", the contents of the top level
4777 parentheses. If there are more than 15 capturing parentheses in a pat-
4778 tern, PCRE has to obtain extra memory to store data during a recursion,
4779 which it does by using pcre_malloc, freeing it via pcre_free after-
4780 wards. If no memory can be obtained, the match fails with the
4781 PCRE_ERROR_NOMEMORY error.
4782
4783 Do not confuse the (?R) item with the condition (R), which tests for
4784 recursion. Consider this pattern, which matches text in angle brack-
4785 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4786 brackets (that is, when recursing), whereas any characters are permit-
4787 ted at the outer level.
4788
4789 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4790
4791 In this pattern, (?(R) is the start of a conditional subpattern, with
4792 two different alternatives for the recursive and non-recursive cases.
4793 The (?R) item is the actual recursive call.
4794
4795
4796 SUBPATTERNS AS SUBROUTINES
4797
4798 If the syntax for a recursive subpattern reference (either by number or
4799 by name) is used outside the parentheses to which it refers, it oper-
4800 ates like a subroutine in a programming language. The "called" subpat-
4801 tern may be defined before or after the reference. A numbered reference
4802 can be absolute or relative, as in these examples:
4803
4804 (...(absolute)...)...(?2)...
4805 (...(relative)...)...(?-1)...
4806 (...(?+1)...(relative)...
4807
4808 An earlier example pointed out that the pattern
4809
4810 (sens|respons)e and \1ibility
4811
4812 matches "sense and sensibility" and "response and responsibility", but
4813 not "sense and responsibility". If instead the pattern
4814
4815 (sens|respons)e and (?1)ibility
4816
4817 is used, it does match "sense and responsibility" as well as the other
4818 two strings. Another example is given in the discussion of DEFINE
4819 above.
4820
4821 Like recursive subpatterns, a "subroutine" call is always treated as an
4822 atomic group. That is, once it has matched some of the subject string,
4823 it is never re-entered, even if it contains untried alternatives and
4824 there is a subsequent matching failure.
4825
4826 When a subpattern is used as a subroutine, processing options such as
4827 case-independence are fixed when the subpattern is defined. They cannot
4828 be changed for different calls. For example, consider this pattern:
4829
4830 (abc)(?i:(?-1))
4831
4832 It matches "abcabc". It does not match "abcABC" because the change of
4833 processing option does not affect the called subpattern.
4834
4835
4836 ONIGURUMA SUBROUTINE SYNTAX
4837
4838 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4839 name or a number enclosed either in angle brackets or single quotes, is
4840 an alternative syntax for referencing a subpattern as a subroutine,
4841 possibly recursively. Here are two of the examples used above, rewrit-
4842 ten using this syntax:
4843
4844 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
4845 (sens|respons)e and \g'1'ibility
4846
4847 PCRE supports an extension to Oniguruma: if a number is preceded by a
4848 plus or a minus sign it is taken as a relative reference. For example:
4849
4850 (abc)(?i:\g<-1>)
4851
4852 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
4853 synonymous. The former is a back reference; the latter is a subroutine
4854 call.
4855
4856
4857 CALLOUTS
4858
4859 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4860 Perl code to be obeyed in the middle of matching a regular expression.
4861 This makes it possible, amongst other things, to extract different sub-
4862 strings that match the same pair of parentheses when there is a repeti-
4863 tion.
4864
4865 PCRE provides a similar feature, but of course it cannot obey arbitrary
4866 Perl code. The feature is called "callout". The caller of PCRE provides
4867 an external function by putting its entry point in the global variable
4868 pcre_callout. By default, this variable contains NULL, which disables
4869 all calling out.
4870
4871 Within a regular expression, (?C) indicates the points at which the
4872 external function is to be called. If you want to identify different
4873 callout points, you can put a number less than 256 after the letter C.
4874 The default value is zero. For example, this pattern has two callout
4875 points:
4876
4877 (?C1)abc(?C2)def
4878
4879 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4880 automatically installed before each item in the pattern. They are all
4881 numbered 255.
4882
4883 During matching, when PCRE reaches a callout point (and pcre_callout is
4884 set), the external function is called. It is provided with the number
4885 of the callout, the position in the pattern, and, optionally, one item
4886 of data originally supplied by the caller of pcre_exec(). The callout
4887 function may cause matching to proceed, to backtrack, or to fail alto-
4888 gether. A complete description of the interface to the callout function
4889 is given in the pcrecallout documentation.
4890
4891
4892 BACKTRACKING CONTROL
4893
4894 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4895 which are described in the Perl documentation as "experimental and sub-
4896 ject to change or removal in a future version of Perl". It goes on to
4897 say: "Their usage in production code should be noted to avoid problems
4898 during upgrades." The same remarks apply to the PCRE features described
4899 in this section.
4900
4901 Since these verbs are specifically related to backtracking, most of
4902 them can be used only when the pattern is to be matched using
4903 pcre_exec(), which uses a backtracking algorithm. With the exception of
4904 (*FAIL), which behaves like a failing negative assertion, they cause an
4905 error if encountered by pcre_dfa_exec().
4906
4907 The new verbs make use of what was previously invalid syntax: an open-
4908 ing parenthesis followed by an asterisk. In Perl, they are generally of
4909 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4910 its general form is just (*VERB). Any number of these verbs may occur
4911 in a pattern. There are two kinds:
4912
4913 Verbs that act immediately
4914
4915 The following verbs act as soon as they are encountered:
4916
4917 (*ACCEPT)
4918
4919 This verb causes the match to end successfully, skipping the remainder
4920 of the pattern. When inside a recursion, only the innermost pattern is
4921 ended immediately. PCRE differs from Perl in what happens if the
4922 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4923 captured: in PCRE no data is captured. For example:
4924
4925 A(A|B(*ACCEPT)|C)D
4926
4927 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4928 is captured.
4929
4930 (*FAIL) or (*F)
4931
4932 This verb causes the match to fail, forcing backtracking to occur. It
4933 is equivalent to (?!) but easier to read. The Perl documentation notes
4934 that it is probably useful only when combined with (?{}) or (??{}).
4935 Those are, of course, Perl features that are not present in PCRE. The
4936 nearest equivalent is the callout feature, as for example in this pat-
4937 tern:
4938
4939 a+(?C)(*FAIL)
4940
4941 A match with the string "aaaa" always fails, but the callout is taken
4942 before each backtrack happens (in this example, 10 times).
4943
4944 Verbs that act after backtracking
4945
4946 The following verbs do nothing when they are encountered. Matching con-
4947 tinues with what follows, but if there is no subsequent match, a fail-
4948 ure is forced. The verbs differ in exactly what kind of failure
4949 occurs.
4950
4951 (*COMMIT)
4952
4953 This verb causes the whole match to fail outright if the rest of the
4954 pattern does not match. Even if the pattern is unanchored, no further
4955 attempts to find a match by advancing the start point take place. Once
4956 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4957 at the current starting point, or not at all. For example:
4958
4959 a+(*COMMIT)b
4960
4961 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4962 of dynamic anchor, or "I've started, so I must finish."
4963
4964 (*PRUNE)
4965
4966 This verb causes the match to fail at the current position if the rest
4967 of the pattern does not match. If the pattern is unanchored, the normal
4968 "bumpalong" advance to the next starting character then happens. Back-
4969 tracking can occur as usual to the left of (*PRUNE), or when matching
4970 to the right of (*PRUNE), but if there is no match to the right, back-
4971 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4972 is just an alternative to an atomic group or possessive quantifier, but
4973 there are some uses of (*PRUNE) that cannot be expressed in any other
4974 way.
4975
4976 (*SKIP)
4977
4978 This verb is like (*PRUNE), except that if the pattern is unanchored,
4979 the "bumpalong" advance is not to the next character, but to the posi-
4980 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4981 that whatever text was matched leading up to it cannot be part of a
4982 successful match. Consider:
4983
4984 a+(*SKIP)b
4985
4986 If the subject is "aaaac...", after the first match attempt fails
4987 (starting at the first character in the string), the starting point
4988 skips on to start the next attempt at "c". Note that a possessive quan-
4989 tifer does not have the same effect in this example; although it would
4990 suppress backtracking during the first match attempt, the second
4991 attempt would start at the second character instead of skipping on to
4992 "c".
4993
4994 (*THEN)
4995
4996 This verb causes a skip to the next alternation if the rest of the pat-
4997 tern does not match. That is, it cancels pending backtracking, but only
4998 within the current alternation. Its name comes from the observation
4999 that it can be used for a pattern-based if-then-else block:
5000
5001 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5002
5003 If the COND1 pattern matches, FOO is tried (and possibly further items
5004 after the end of the group if FOO succeeds); on failure the matcher
5005 skips to the second alternative and tries COND2, without backtracking
5006 into COND1. If (*THEN) is used outside of any alternation, it acts
5007 exactly like (*PRUNE).
5008
5009
5010 SEE ALSO
5011
5012 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5013
5014
5015 AUTHOR
5016
5017 Philip Hazel
5018 University Computing Service
5019 Cambridge CB2 3QH, England.
5020
5021
5022 REVISION
5023
5024 Last updated: 18 March 2009
5025 Copyright (c) 1997-2009 University of Cambridge.
5026 ------------------------------------------------------------------------------
5027
5028
5029 PCRESYNTAX(3) PCRESYNTAX(3)
5030
5031
5032 NAME
5033 PCRE - Perl-compatible regular expressions
5034
5035
5036 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5037
5038 The full syntax and semantics of the regular expressions that are sup-
5039 ported by PCRE are described in the pcrepattern documentation. This
5040 document contains just a quick-reference summary of the syntax.
5041
5042
5043 QUOTING
5044
5045 \x where x is non-alphanumeric is a literal x
5046 \Q...\E treat enclosed characters as literal
5047
5048
5049 CHARACTERS
5050
5051 \a alarm, that is, the BEL character (hex 07)
5052 \cx "control-x", where x is any character
5053 \e escape (hex 1B)
5054 \f formfeed (hex 0C)
5055 \n newline (hex 0A)
5056 \r carriage return (hex 0D)
5057 \t tab (hex 09)
5058 \ddd character with octal code ddd, or backreference
5059 \xhh character with hex code hh
5060 \x{hhh..} character with hex code hhh..
5061
5062
5063 CHARACTER TYPES
5064
5065 . any character except newline;
5066 in dotall mode, any character whatsoever
5067 \C one byte, even in UTF-8 mode (best avoided)
5068 \d a decimal digit
5069 \D a character that is not a decimal digit
5070 \h a horizontal whitespace character
5071 \H a character that is not a horizontal whitespace character
5072 \p{xx} a character with the xx property
5073 \P{xx} a character without the xx property
5074 \R a newline sequence
5075 \s a whitespace character
5076 \S a character that is not a whitespace character
5077 \v a vertical whitespace character
5078 \V a character that is not a vertical whitespace character
5079 \w a "word" character
5080 \W a "non-word" character
5081 \X an extended Unicode sequence
5082
5083 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
5084
5085
5086 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
5087
5088 C Other
5089 Cc Control
5090 Cf Format
5091 Cn Unassigned
5092 Co Private use
5093 Cs Surrogate
5094
5095 L Letter
5096 Ll Lower case letter
5097 Lm Modifier letter
5098 Lo Other letter
5099 Lt Title case letter
5100 Lu Upper case letter
5101 L& Ll, Lu, or Lt
5102
5103 M Mark
5104 Mc Spacing mark
5105 Me Enclosing mark
5106 Mn Non-spacing mark
5107
5108 N Number
5109 Nd Decimal number
5110 Nl Letter number
5111 No Other number
5112
5113 P Punctuation
5114 Pc Connector punctuation
5115 Pd Dash punctuation
5116 Pe Close punctuation
5117 Pf Final punctuation
5118 Pi Initial punctuation
5119 Po Other punctuation
5120 Ps Open punctuation
5121
5122 S Symbol
5123 Sc Currency symbol
5124 Sk Modifier symbol
5125 Sm Mathematical symbol
5126 So Other symbol
5127
5128 Z Separator
5129 Zl Line separator
5130 Zp Paragraph separator
5131 Zs Space separator
5132
5133
5134 SCRIPT NAMES FOR \p AND \P
5135
5136 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
5137 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
5138 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
5139 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
5140 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
5141 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
5142 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
5143 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
5144 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
5145
5146
5147 CHARACTER CLASSES
5148
5149 [...] positive character class
5150 [^...] negative character class
5151 [x-y] range (can be used for hex characters)
5152 [[:xxx:]] positive POSIX named set
5153 [[:^xxx:]] negative POSIX named set
5154
5155 alnum alphanumeric
5156 alpha alphabetic
5157 ascii 0-127
5158 blank space or tab
5159 cntrl control character
5160 digit decimal digit
5161 graph printing, excluding space
5162 lower lower case letter
5163 print printing, including space
5164 punct printing, excluding alphanumeric
5165 space whitespace
5166 upper upper case letter
5167 word same as \w
5168 xdigit hexadecimal digit
5169
5170 In PCRE, POSIX character set names recognize only ASCII characters. You
5171 can use \Q...\E inside a character class.
5172
5173
5174 QUANTIFIERS
5175
5176 ? 0 or 1, greedy
5177 ?+ 0 or 1, possessive
5178 ?? 0 or 1, lazy
5179 * 0 or more, greedy
5180 *+ 0 or more, possessive
5181 *? 0 or more, lazy
5182 + 1 or more, greedy
5183 ++ 1 or more, possessive
5184 +? 1 or more, lazy
5185 {n} exactly n
5186 {n,m} at least n, no more than m, greedy
5187 {n,m}+ at least n, no more than m, possessive
5188 {n,m}? at least n, no more than m, lazy
5189 {n,} n or more, greedy
5190 {n,}+ n or more, possessive
5191 {n,}? n or more, lazy
5192
5193
5194 ANCHORS AND SIMPLE ASSERTIONS
5195
5196 \b word boundary
5197 \B not a word boundary
5198 ^ start of subject
5199 also after internal newline in multiline mode
5200 \A start of subject
5201 $ end of subject
5202 also before newline at end of subject
5203 also before internal newline in multiline mode
5204 \Z end of subject
5205 also before newline at end of subject
5206 \z end of subject
5207 \G first matching position in subject
5208
5209
5210 MATCH POINT RESET
5211
5212 \K reset start of match
5213
5214
5215 ALTERNATION
5216
5217 expr|expr|expr...
5218
5219
5220 CAPTURING
5221
5222 (...) capturing group
5223 (?<name>...) named capturing group (Perl)
5224 (?'name'...) named capturing group (Perl)
5225 (?P<name>...) named capturing group (Python)
5226 (?:...) non-capturing group
5227 (?|...) non-capturing group; reset group numbers for
5228 capturing groups in each alternative
5229
5230
5231 ATOMIC GROUPS
5232
5233 (?>...) atomic, non-capturing group
5234
5235
5236 COMMENT
5237
5238 (?#....) comment (not nestable)
5239
5240
5241 OPTION SETTING
5242
5243 (?i) caseless
5244 (?J) allow duplicate names
5245 (?m) multiline
5246 (?s) single line (dotall)
5247 (?U) default ungreedy (lazy)
5248 (?x) extended (ignore white space)
5249 (?-...) unset option(s)
5250
5251
5252 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5253
5254 (?=...) positive look ahead
5255 (?!...) negative look ahead
5256 (?<=...) positive look behind
5257 (?<!...) negative look behind
5258
5259 Each top-level branch of a look behind must be of a fixed length.
5260
5261
5262 BACKREFERENCES
5263
5264 \n reference by number (can be ambiguous)
5265 \gn reference by number
5266 \g{n} reference by number
5267 \g{-n} relative reference by number
5268 \k<name> reference by name (Perl)
5269 \k'name' reference by name (Perl)
5270 \g{name} reference by name (Perl)
5271 \k{name} reference by name (.NET)
5272 (?P=name) reference by name (Python)
5273
5274
5275 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5276
5277 (?R) recurse whole pattern
5278 (?n) call subpattern by absolute number
5279 (?+n) call subpattern by relative number
5280 (?-n) call subpattern by relative number
5281 (?&name) call subpattern by name (Perl)
5282 (?P>name) call subpattern by name (Python)
5283 \g<name> call subpattern by name (Oniguruma)
5284 \g'name' call subpattern by name (Oniguruma)
5285 \g<n> call subpattern by absolute number (Oniguruma)
5286 \g'n' call subpattern by absolute number (Oniguruma)
5287 \g<+n> call subpattern by relative number (PCRE extension)
5288 \g'+n' call subpattern by relative number (PCRE extension)
5289 \g<-n> call subpattern by relative number (PCRE extension)
5290 \g'-n' call subpattern by relative number (PCRE extension)
5291
5292
5293 CONDITIONAL PATTERNS
5294
5295 (?(condition)yes-pattern)
5296 (?(condition)yes-pattern|no-pattern)
5297
5298 (?(n)... absolute reference condition
5299 (?(+n)... relative reference condition
5300 (?(-n)... relative reference condition
5301 (?(<name>)... named reference condition (Perl)
5302 (?('name')... named reference condition (Perl)
5303 (?(name)... named reference condition (PCRE)
5304 (?(R)... overall recursion condition
5305 (?(Rn)... specific group recursion condition
5306 (?(R&name)... specific recursion condition
5307 (?(DEFINE)... define subpattern for reference
5308 (?(assert)... assertion condition
5309
5310
5311 BACKTRACKING CONTROL
5312
5313 The following act immediately they are reached:
5314
5315 (*ACCEPT) force successful match
5316 (*FAIL) force backtrack; synonym (*F)
5317
5318 The following act only when a subsequent match failure causes a back-
5319 track to reach them. They all force a match failure, but they differ in
5320 what happens afterwards. Those that advance the start-of-match point do
5321 so only if the pattern is not anchored.
5322
5323 (*COMMIT) overall failure, no advance of starting point
5324 (*PRUNE) advance to next starting character
5325 (*SKIP) advance start to current matching position
5326 (*THEN) local failure, backtrack to next alternation
5327
5328
5329 NEWLINE CONVENTIONS
5330
5331 These are recognized only at the very start of the pattern or after a
5332 (*BSR_...) option.
5333
5334 (*CR)
5335 (*LF)
5336 (*CRLF)
5337 (*ANYCRLF)
5338 (*ANY)
5339
5340
5341 WHAT \R MATCHES
5342
5343 These are recognized only at the very start of the pattern or after a
5344 (*...) option that sets the newline convention.
5345
5346 (*BSR_ANYCRLF)
5347 (*BSR_UNICODE)
5348
5349
5350 CALLOUTS
5351
5352 (?C) callout
5353 (?Cn) callout with data n
5354
5355
5356 SEE ALSO
5357
5358 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5359
5360
5361 AUTHOR
5362
5363 Philip Hazel
5364 University Computing Service
5365 Cambridge CB2 3QH, England.
5366
5367
5368 REVISION
5369
5370 Last updated: 09 April 2008
5371 Copyright (c) 1997-2008 University of Cambridge.
5372 ------------------------------------------------------------------------------
5373
5374
5375 PCREPARTIAL(3) PCREPARTIAL(3)
5376
5377
5378 NAME
5379 PCRE - Perl-compatible regular expressions
5380
5381
5382 PARTIAL MATCHING IN PCRE
5383
5384 In normal use of PCRE, if the subject string that is passed to
5385 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5386 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5387 There are circumstances where it might be helpful to distinguish this
5388 case from other cases in which there is no match.
5389
5390 Consider, for example, an application where a human is required to type
5391 in data for a field with specific formatting requirements. An example
5392 might be a date in the form ddmmmyy, defined by this pattern:
5393
5394 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5395
5396 If the application sees the user's keystrokes one by one, and can check
5397 that what has been typed so far is potentially valid, it is able to
5398 raise an error as soon as a mistake is made, possibly beeping and not
5399 reflecting the character that has been typed. This immediate feedback
5400 is likely to be a better user interface than a check that is delayed
5401 until the entire string has been entered.
5402
5403 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5404 TIAL option, which can be set when calling pcre_exec() or
5405 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5406 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5407 during the matching process the last part of the subject string matched
5408 part of the pattern. Unfortunately, for non-anchored matching, it is
5409 not possible to obtain the position of the start of the partial match.
5410 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5411
5412 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5413 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5414 the subject is reached, there have been no complete matches, but there
5415 is still at least one matching possibility. The portion of the string
5416 that provided the partial match is set as the first matching string.
5417
5418 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5419 the last literal byte in a pattern, and abandons matching immediately
5420 if such a byte is not present in the subject string. This optimization
5421 cannot be used for a subject string that might match only partially.
5422
5423
5424 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5425
5426 Because of the way certain internal optimizations are implemented in
5427 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5428 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5429 used. For pcre_exec(), repeated single characters such as
5430
5431 a{2,4}
5432
5433 and repeated single metasequences such as
5434
5435 \d+
5436
5437 are not permitted if the maximum number of occurrences is greater than
5438 one. Optional items such as \d? (where the maximum is one) are permit-
5439 ted. Quantifiers with any values are permitted after parentheses, so
5440 the invalid examples above can be coded thus:
5441
5442 (a){2,4}
5443 (\d)+
5444
5445 These constructions run more slowly, but for the kinds of application
5446 that are envisaged for this facility, this is not felt to be a major
5447 restriction.
5448
5449 If PCRE_PARTIAL is set for a pattern that does not conform to the
5450 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5451 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5452 find out if a compiled pattern can be used for partial matching.
5453
5454
5455 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5456
5457 If the escape sequence \P is present in a pcretest data line, the
5458 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5459 uses the date example quoted above:
5460
5461 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5462 data> 25jun04\P
5463 0: 25jun04
5464 1: jun
5465 data> 25dec3\P
5466 Partial match
5467 data> 3ju\P
5468 Partial match
5469 data> 3juj\P
5470 No match
5471 data> j\P
5472 No match
5473
5474 The first data string is matched completely, so pcretest shows the
5475 matched substrings. The remaining four strings do not match the com-
5476 plete pattern, but the first two are partial matches. The same test,
5477 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5478 produces the following output:
5479
5480 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5481 data> 25jun04\P\D
5482 0: 25jun04
5483 data> 23dec3\P\D
5484 Partial match: 23dec3
5485 data> 3ju\P\D
5486 Partial match: 3ju
5487 data> 3juj\P\D
5488 No match
5489 data> j\P\D
5490 No match
5491
5492 Notice that in this case the portion of the string that was matched is
5493 made available.
5494
5495
5496 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5497
5498 When a partial match has been found using pcre_dfa_exec(), it is possi-
5499 ble to continue the match by providing additional subject data and
5500 calling pcre_dfa_exec() again with the same compiled regular expres-
5501 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5502 the same working space as before, because this is where details of the
5503 previous partial match are stored. Here is an example using pcretest,
5504 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5505 \D are as above):
5506
5507 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5508 data> 23ja\P\D
5509 Partial match: 23ja
5510 data> n05\R\D
5511 0: n05
5512
5513 The first call has "23ja" as the subject, and requests partial match-
5514 ing; the second call has "n05" as the subject for the continued
5515 (restarted) match. Notice that when the match is complete, only the
5516 last part is shown; PCRE does not retain the previously partially-
5517 matched string. It is up to the calling program to do that if it needs
5518 to.
5519
5520 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5521 matching over multiple segments. This facility can be used to pass very
5522 long subject strings to pcre_dfa_exec(). However, some care is needed
5523 for certain types of pattern.
5524
5525 1. If the pattern contains tests for the beginning or end of a line,
5526 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5527 ate, when the subject string for any call does not contain the begin-
5528 ning or end of a line.
5529
5530 2. If the pattern contains backward assertions (including \b or \B),
5531 you need to arrange for some overlap in the subject strings to allow
5532 for this. For example, you could pass the subject in chunks that are
5533 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5534 set to 200 and the previous 200 bytes at the start of the buffer.
5535
5536 3. Matching a subject string that is split into multiple segments does
5537 not always produce exactly the same result as matching over one single
5538 long string. The difference arises when there are multiple matching
5539 possibilities, because a partial match result is given only when there
5540 are no completed matches in a call to pcre_dfa_exec(). This means that
5541 as soon as the shortest match has been found, continuation to a new
5542 subject segment is no longer possible. Consider this pcretest example:
5543
5544 re> /dog(sbody)?/
5545 data> do\P\D
5546 Partial match: do
5547 data> gsb\R\P\D
5548 0: g
5549 data> dogsbody\D
5550 0: dogsbody
5551 1: dog
5552
5553 The pattern matches the words "dog" or "dogsbody". When the subject is
5554 presented in several parts ("do" and "gsb" being the first two) the
5555 match stops when "dog" has been found, and it is not possible to con-
5556 tinue. On the other hand, if "dogsbody" is presented as a single
5557 string, both matches are found.
5558
5559 Because of this phenomenon, it does not usually make sense to end a
5560 pattern that is going to be matched in this way with a variable repeat.
5561
5562 4. Patterns that contain alternatives at the top level which do not all
5563 start with the same pattern item may not work as expected. For example,
5564 consider this pattern:
5565
5566 1234|3789
5567
5568 If the first part of the subject is "ABC123", a partial match of the
5569 first alternative is found at offset 3. There is no partial match for
5570 the second alternative, because such a match does not start at the same
5571 point in the subject string. Attempting to continue with the string
5572 "789" does not yield a match because only those alternatives that match
5573 at one point in the subject are remembered. The problem arises because
5574 the start of the second alternative matches within the first alterna-
5575 tive. There is no problem with anchored patterns or patterns such as:
5576
5577 1234|ABCD
5578
5579 where no string can be a partial match for both alternatives.
5580
5581
5582 AUTHOR
5583
5584 Philip Hazel
5585 University Computing Service
5586 Cambridge CB2 3QH, England.
5587
5588
5589 REVISION
5590
5591 Last updated: 04 June 2007
5592 Copyright (c) 1997-2007 University of Cambridge.
5593 ------------------------------------------------------------------------------
5594
5595
5596 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5597
5598
5599 NAME
5600 PCRE - Perl-compatible regular expressions
5601
5602
5603 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5604
5605 If you are running an application that uses a large number of regular
5606 expression patterns, it may be useful to store them in a precompiled
5607 form instead of having to compile them every time the application is
5608 run. If you are not using any private character tables (see the
5609 pcre_maketables() documentation), this is relatively straightforward.
5610 If you are using private tables, it is a little bit more complicated.
5611
5612 If you save compiled patterns to a file, you can copy them to a differ-
5613 ent host and run them there. This works even if the new host has the
5614 opposite endianness to the one on which the patterns were compiled.
5615 There may be a small performance penalty, but it should be insignifi-
5616 cant. However, compiling regular expressions with one version of PCRE
5617 for use with a different version is not guaranteed to work and may
5618 cause crashes.
5619
5620
5621 SAVING A COMPILED PATTERN
5622 The value returned by pcre_compile() points to a single block of memory
5623 that holds the compiled pattern and associated data. You can find the
5624 length of this block in bytes by calling pcre_fullinfo() with an argu-
5625 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5626 manner. Here is sample code that compiles a pattern and writes it to a
5627 file. It assumes that the variable fd refers to a file that is open for
5628 output:
5629
5630 int erroroffset, rc, size;
5631 char *error;
5632 pcre *re;
5633
5634 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5635 if (re == NULL) { ... handle errors ... }
5636 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5637 if (rc < 0) { ... handle errors ... }
5638 rc = fwrite(re, 1, size, fd);
5639 if (rc != size) { ... handle errors ... }
5640
5641 In this example, the bytes that comprise the compiled pattern are
5642 copied exactly. Note that this is binary data that may contain any of
5643 the 256 possible byte values. On systems that make a distinction
5644 between binary and non-binary data, be sure that the file is opened for
5645 binary output.
5646
5647 If you want to write more than one pattern to a file, you will have to
5648 devise a way of separating them. For binary data, preceding each pat-
5649 tern with its length is probably the most straightforward approach.
5650 Another possibility is to write out the data in hexadecimal instead of
5651 binary, one pattern to a line.
5652
5653 Saving compiled patterns in a file is only one possible way of storing
5654 them for later use. They could equally well be saved in a database, or
5655 in the memory of some daemon process that passes them via sockets to
5656 the processes that want them.
5657
5658 If the pattern has been studied, it is also possible to save the study
5659 data in a similar way to the compiled pattern itself. When studying
5660 generates additional information, pcre_study() returns a pointer to a
5661 pcre_extra data block. Its format is defined in the section on matching
5662 a pattern in the pcreapi documentation. The study_data field points to
5663 the binary study data, and this is what you must save (not the
5664 pcre_extra block itself). The length of the study data can be obtained
5665 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5666 Remember to check that pcre_study() did return a non-NULL value before
5667 trying to save the study data.
5668
5669
5670 RE-USING A PRECOMPILED PATTERN
5671
5672 Re-using a precompiled pattern is straightforward. Having reloaded it
5673 into main memory, you pass its pointer to pcre_exec() or
5674 pcre_dfa_exec() in the usual way. This should work even on another
5675 host, and even if that host has the opposite endianness to the one
5676 where the pattern was compiled.
5677
5678 However, if you passed a pointer to custom character tables when the
5679 pattern was compiled (the tableptr argument of pcre_compile()), you
5680 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5681 because the value saved with the compiled pattern will obviously be
5682 nonsense. A field in a pcre_extra() block is used to pass this data, as
5683 described in the section on matching a pattern in the pcreapi documen-
5684 tation.
5685
5686 If you did not provide custom character tables when the pattern was
5687 compiled, the pointer in the compiled pattern is NULL, which causes
5688 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5689 take any special action at run time in this case.
5690
5691 If you saved study data with the compiled pattern, you need to create
5692 your own pcre_extra data block and set the study_data field to point to
5693 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5694 bit in the flags field to indicate that study data is present. Then
5695 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5696 usual way.
5697
5698
5699 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5700
5701 In general, it is safest to recompile all saved patterns when you
5702 update to a new PCRE release, though not all updates actually require
5703 this. Recompiling is definitely needed for release 7.2.
5704
5705
5706 AUTHOR
5707
5708 Philip Hazel
5709 University Computing Service
5710 Cambridge CB2 3QH, England.
5711
5712
5713 REVISION
5714
5715 Last updated: 13 June 2007
5716 Copyright (c) 1997-2007 University of Cambridge.
5717 ------------------------------------------------------------------------------
5718
5719
5720 PCREPERFORM(3) PCREPERFORM(3)
5721
5722
5723 NAME
5724 PCRE - Perl-compatible regular expressions
5725
5726
5727 PCRE PERFORMANCE
5728
5729 Two aspects of performance are discussed below: memory usage and pro-
5730 cessing time. The way you express your pattern as a regular expression
5731 can affect both of them.
5732
5733
5734 MEMORY USAGE
5735
5736 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5737 that most simple patterns do not use much memory. However, there is one
5738 case where memory usage can be unexpectedly large. When a parenthesized
5739 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5740 ited maximum, the whole subpattern is repeated in the compiled code.
5741 For example, the pattern
5742
5743 (abc|def){2,4}
5744
5745 is compiled as if it were
5746
5747 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5748
5749 (Technical aside: It is done this way so that backtrack points within
5750 each of the repetitions can be independently maintained.)
5751
5752 For regular expressions whose quantifiers use only small numbers, this
5753 is not usually a problem. However, if the numbers are large, and par-
5754 ticularly if such repetitions are nested, the memory usage can become
5755 an embarrassment. For example, the very simple pattern
5756
5757 ((ab){1,1000}c){1,3}
5758
5759 uses 51K bytes when compiled. When PCRE is compiled with its default
5760 internal pointer size of two bytes, the size limit on a compiled pat-
5761 tern is 64K, and this is reached with the above pattern if the outer
5762 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5763 internal pointers and thus handle larger compiled patterns, but it is
5764 better to try to rewrite your pattern to use less memory if you can.
5765
5766 One way of reducing the memory usage for such patterns is to make use
5767 of PCRE's "subroutine" facility. Re-writing the above pattern as
5768
5769 ((ab)(?2){0,999}c)(?1){0,2}
5770
5771 reduces the memory requirements to 18K, and indeed it remains under 20K
5772 even with the outer repetition increased to 100. However, this pattern
5773 is not exactly equivalent, because the "subroutine" calls are treated
5774 as atomic groups into which there can be no backtracking if there is a
5775 subsequent matching failure. Therefore, PCRE cannot do this kind of
5776 rewriting automatically. Furthermore, there is a noticeable loss of
5777 speed when executing the modified pattern. Nevertheless, if the atomic
5778 grouping is not a problem and the loss of speed is acceptable, this
5779 kind of rewriting will allow you to process patterns that PCRE cannot
5780 otherwise handle.
5781
5782
5783 PROCESSING TIME
5784
5785 Certain items in regular expression patterns are processed more effi-
5786 ciently than others. It is more efficient to use a character class like
5787 [aeiou] than a set of single-character alternatives such as
5788 (a|e|i|o|u). In general, the simplest construction that provides the
5789 required behaviour is usually the most efficient. Jeffrey Friedl's book
5790 contains a lot of useful general discussion about optimizing regular
5791 expressions for efficient performance. This document contains a few
5792 observations about PCRE.
5793
5794 Using Unicode character properties (the \p, \P, and \X escapes) is
5795 slow, because PCRE has to scan a structure that contains data for over
5796 fifteen thousand characters whenever it needs a character's property.
5797 If you can find an alternative pattern that does not use character
5798 properties, it will probably be faster.
5799
5800 When a pattern begins with .* not in parentheses, or in parentheses
5801 that are not the subject of a backreference, and the PCRE_DOTALL option
5802 is set, the pattern is implicitly anchored by PCRE, since it can match
5803 only at the start of a subject string. However, if PCRE_DOTALL is not
5804 set, PCRE cannot make this optimization, because the . metacharacter
5805 does not then match a newline, and if the subject string contains new-
5806 lines, the pattern may match from the character immediately following
5807 one of them instead of from the very start. For example, the pattern
5808
5809 .*second
5810
5811 matches the subject "first\nand second" (where \n stands for a newline
5812 character), with the match starting at the seventh character. In order
5813 to do this, PCRE has to retry the match starting after every newline in
5814 the subject.
5815
5816 If you are using such a pattern with subject strings that do not con-
5817 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5818 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5819 ing. That saves PCRE from having to scan along the subject looking for
5820 a newline to restart at.
5821
5822 Beware of patterns that contain nested indefinite repeats. These can
5823 take a long time to run when applied to a string that does not match.
5824 Consider the pattern fragment
5825
5826 ^(a+)*
5827
5828 This can match "aaaa" in 16 different ways, and this number increases
5829 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5830 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5831 repeats can match different numbers of times.) When the remainder of
5832 the pattern is such that the entire match is going to fail, PCRE has in
5833 principle to try every possible variation, and this can take an
5834 extremely long time, even for relatively short strings.
5835
5836 An optimization catches some of the more simple cases such as
5837
5838 (a+)*b
5839
5840 where a literal character follows. Before embarking on the standard
5841 matching procedure, PCRE checks that there is a "b" later in the sub-
5842 ject string, and if there is not, it fails the match immediately. How-
5843 ever, when there is no following literal this optimization cannot be
5844 used. You can see the difference by comparing the behaviour of
5845
5846 (a+)*\d
5847
5848 with the pattern above. The former gives a failure almost instantly
5849 when applied to a whole line of "a" characters, whereas the latter
5850 takes an appreciable time with strings longer than about 20 characters.
5851
5852 In many cases, the solution to this kind of performance issue is to use
5853 an atomic group or a possessive quantifier.
5854
5855
5856 AUTHOR
5857
5858 Philip Hazel
5859 University Computing Service
5860 Cambridge CB2 3QH, England.
5861
5862
5863 REVISION
5864
5865 Last updated: 06 March 2007
5866 Copyright (c) 1997-2007 University of Cambridge.
5867 ------------------------------------------------------------------------------
5868
5869
5870 PCREPOSIX(3) PCREPOSIX(3)
5871
5872
5873 NAME
5874 PCRE - Perl-compatible regular expressions.
5875
5876
5877 SYNOPSIS OF POSIX API
5878
5879 #include <pcreposix.h>
5880
5881 int regcomp(regex_t *preg, const char *pattern,
5882 int cflags);
5883
5884 int regexec(regex_t *preg, const char *string,
5885 size_t nmatch, regmatch_t pmatch[], int eflags);
5886
5887 size_t regerror(int errcode, const regex_t *preg,
5888 char *errbuf, size_t errbuf_size);
5889
5890 void regfree(regex_t *preg);
5891
5892
5893 DESCRIPTION
5894
5895 This set of functions provides a POSIX-style API to the PCRE regular
5896 expression package. See the pcreapi documentation for a description of
5897 PCRE's native API, which contains much additional functionality.
5898
5899 The functions described here are just wrapper functions that ultimately
5900 call the PCRE native API. Their prototypes are defined in the
5901 pcreposix.h header file, and on Unix systems the library itself is
5902 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5903 command for linking an application that uses them. Because the POSIX
5904 functions call the native ones, it is also necessary to add -lpcre.
5905
5906 I have implemented only those POSIX option bits that can be reasonably
5907 mapped to PCRE native options. In addition, the option REG_EXTENDED is
5908 defined with the value zero. This has no effect, but since programs
5909 that are written to the POSIX interface often use it, this makes it
5910 easier to slot in PCRE as a replacement library. Other POSIX options
5911 are not even defined.
5912
5913 When PCRE is called via these functions, it is only the API that is
5914 POSIX-like in style. The syntax and semantics of the regular expres-
5915 sions themselves are still those of Perl, subject to the setting of
5916 various PCRE options, as described below. "POSIX-like in style" means
5917 that the API approximates to the POSIX definition; it is not fully
5918 POSIX-compatible, and in multi-byte encoding domains it is probably
5919 even less compatible.
5920
5921 The header for these functions is supplied as pcreposix.h to avoid any
5922 potential clash with other POSIX libraries. It can, of course, be
5923 renamed or aliased as regex.h, which is the "correct" name. It provides
5924 two structure types, regex_t for compiled internal forms, and reg-
5925 match_t for returning captured substrings. It also defines some con-
5926 stants whose names start with "REG_"; these are used for setting
5927 options and identifying error codes.
5928
5929
5930 COMPILING A PATTERN
5931
5932 The function regcomp() is called to compile a pattern into an internal
5933 form. The pattern is a C string terminated by a binary zero, and is
5934 passed in the argument pattern. The preg argument is a pointer to a
5935 regex_t structure that is used as a base for storing information about
5936 the compiled regular expression.
5937
5938 The argument cflags is either zero, or contains one or more of the bits
5939 defined by the following macros:
5940
5941 REG_DOTALL
5942
5943 The PCRE_DOTALL option is set when the regular expression is passed for
5944 compilation to the native function. Note that REG_DOTALL is not part of
5945 the POSIX standard.
5946
5947 REG_ICASE
5948
5949 The PCRE_CASELESS option is set when the regular expression is passed
5950 for compilation to the native function.
5951
5952 REG_NEWLINE
5953
5954 The PCRE_MULTILINE option is set when the regular expression is passed
5955 for compilation to the native function. Note that this does not mimic
5956 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5957 tion).
5958
5959 REG_NOSUB
5960
5961 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5962 passed for compilation to the native function. In addition, when a pat-
5963 tern that is compiled with this flag is passed to regexec() for match-
5964 ing, the nmatch and pmatch arguments are ignored, and no captured
5965 strings are returned.
5966
5967 REG_UTF8
5968
5969 The PCRE_UTF8 option is set when the regular expression is passed for
5970 compilation to the native function. This causes the pattern itself and
5971 all data strings used for matching it to be treated as UTF-8 strings.
5972 Note that REG_UTF8 is not part of the POSIX standard.
5973
5974 In the absence of these flags, no options are passed to the native
5975 function. This means the the regex is compiled with PCRE default
5976 semantics. In particular, the way it handles newline characters in the
5977 subject string is the Perl way, not the POSIX way. Note that setting
5978 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5979 It does not affect the way newlines are matched by . (they aren't) or
5980 by a negative class such as [^a] (they are).
5981
5982 The yield of regcomp() is zero on success, and non-zero otherwise. The
5983 preg structure is filled in on success, and one member of the structure
5984 is public: re_nsub contains the number of capturing subpatterns in the
5985 regular expression. Various error codes are defined in the header file.
5986
5987
5988 MATCHING NEWLINE CHARACTERS
5989
5990 This area is not simple, because POSIX and Perl take different views of
5991 things. It is not possible to get PCRE to obey POSIX semantics, but
5992 then PCRE was never intended to be a POSIX engine. The following table
5993 lists the different possibilities for matching newline characters in
5994 PCRE:
5995
5996 Default Change with
5997
5998 . matches newline no PCRE_DOTALL
5999 newline matches [^a] yes not changeable
6000 $ matches \n at end yes PCRE_DOLLARENDONLY
6001 $ matches \n in middle no PCRE_MULTILINE
6002 ^ matches \n in middle no PCRE_MULTILINE
6003
6004 This is the equivalent table for POSIX:
6005
6006 Default Change with
6007
6008 . matches newline yes REG_NEWLINE
6009 newline matches [^a] yes REG_NEWLINE
6010 $ matches \n at end no REG_NEWLINE
6011 $ matches \n in middle no REG_NEWLINE
6012 ^ matches \n in middle no REG_NEWLINE
6013
6014 PCRE's behaviour is the same as Perl's, except that there is no equiva-
6015 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
6016 no way to stop newline from matching [^a].
6017
6018 The default POSIX newline handling can be obtained by setting
6019 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
6020 behave exactly as for the REG_NEWLINE action.
6021
6022
6023 MATCHING A PATTERN
6024
6025 The function regexec() is called to match a compiled pattern preg
6026 against a given string, which is by default terminated by a zero byte
6027 (but see REG_STARTEND below), subject to the options in eflags. These
6028 can be:
6029
6030 REG_NOTBOL
6031
6032 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
6033 function.
6034
6035 REG_NOTEMPTY
6036
6037 The PCRE_NOTEMPTY option is set when calling the underlying PCRE match-
6038 ing function. Note that REG_NOTEMPTY is not part of the POSIX standard.
6039 However, setting this option can give more POSIX-like behaviour in some
6040 situations.
6041
6042 REG_NOTEOL
6043
6044 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
6045 function.
6046
6047 REG_STARTEND
6048
6049 The string is considered to start at string + pmatch[0].rm_so and to
6050 have a terminating NUL located at string + pmatch[0].rm_eo (there need
6051 not actually be a NUL at that location), regardless of the value of
6052 nmatch. This is a BSD extension, compatible with but not specified by
6053 IEEE Standard 1003.2 (POSIX.2), and should be used with caution in
6054 software intended to be portable to other systems. Note that a non-zero
6055 rm_so does not imply REG_NOTBOL; REG_STARTEND affects only the location
6056 of the string, not how it is matched.
6057
6058 If the pattern was compiled with the REG_NOSUB flag, no data about any
6059 matched strings is returned. The nmatch and pmatch arguments of
6060 regexec() are ignored.
6061
6062 Otherwise,the portion of the string that was matched, and also any cap-
6063 tured substrings, are returned via the pmatch argument, which points to
6064 an array of nmatch structures of type regmatch_t, containing the mem-
6065 bers rm_so and rm_eo. These contain the offset to the first character
6066 of each substring and the offset to the first character after the end
6067 of each substring, respectively. The 0th element of the vector relates
6068 to the entire portion of string that was matched; subsequent elements
6069 relate to the capturing subpatterns of the regular expression. Unused
6070 entries in the array have both structure members set to -1.
6071
6072 A successful match yields a zero return; various error codes are
6073 defined in the header file, of which REG_NOMATCH is the "expected"
6074 failure code.
6075
6076
6077 ERROR MESSAGES
6078
6079 The regerror() function maps a non-zero errorcode from either regcomp()
6080 or regexec() to a printable message. If preg is not NULL, the error
6081 should have arisen from the use of that structure. A message terminated
6082 by a binary zero is placed in errbuf. The length of the message,
6083 including the zero, is limited to errbuf_size. The yield of the func-
6084 tion is the size of buffer needed to hold the whole message.
6085
6086
6087 MEMORY USAGE
6088
6089 Compiling a regular expression causes memory to be allocated and asso-
6090 ciated with the preg structure. The function regfree() frees all such
6091 memory, after which preg may no longer be used as a compiled expres-
6092 sion.
6093
6094
6095 AUTHOR
6096
6097 Philip Hazel
6098 University Computing Service
6099 Cambridge CB2 3QH, England.