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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 416 - (show annotations) (download)
Sat Apr 11 14:34:02 2009 UTC (5 years, 5 months ago) by ph10
File MIME type: text/plain
File size: 313515 byte(s)
File tidies for 7.9 release.

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 5.1.
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, or the pattern must start with the
140 sequence (*UTF8). When either of these is the case, both the pattern
141 and any subject strings that are matched against it are treated as
142 UTF-8 strings instead of just strings of bytes.
143
144 If you compile PCRE with UTF-8 support, but do not use it at run time,
145 the library will be a bit bigger, but the additional run time overhead
146 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
147 very big.
148
149 If PCRE is built with Unicode character property support (which implies
150 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
151 ported. The available properties that can be tested are limited to the
152 general category properties such as Lu for an upper case letter or Nd
153 for a decimal number, the Unicode script names such as Arabic or Han,
154 and the derived properties Any and L&. A full list is given in the
155 pcrepattern documentation. Only the short names for properties are sup-
156 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
157 ter}, is not supported. Furthermore, in Perl, many properties may
158 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
159 does not support this.
160
161 Validity of UTF-8 strings
162
163 When you set the PCRE_UTF8 flag, the strings passed as patterns and
164 subjects are (by default) checked for validity on entry to the relevant
165 functions. From release 7.3 of PCRE, the check is according the rules
166 of RFC 3629, which are themselves derived from the Unicode specifica-
167 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
168 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
169 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
170 to U+DFFF.
171
172 The excluded code points are the "Low Surrogate Area" of Unicode, of
173 which the Unicode Standard says this: "The Low Surrogate Area does not
174 contain any character assignments, consequently no character code
175 charts or namelists are provided for this area. Surrogates are reserved
176 for use with UTF-16 and then must be used in pairs." The code points
177 that are encoded by UTF-16 pairs are available as independent code
178 points in the UTF-8 encoding. (In other words, the whole surrogate
179 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
180
181 If an invalid UTF-8 string is passed to PCRE, an error return
182 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
183 that your strings are valid, and therefore want to skip these checks in
184 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
185 compile time or at run time, PCRE assumes that the pattern or subject
186 it is given (respectively) contains only valid UTF-8 codes. In this
187 case, it does not diagnose an invalid UTF-8 string.
188
189 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
190 what happens depends on why the string is invalid. If the string con-
191 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
192 string of characters in the range 0 to 0x7FFFFFFF. In other words,
193 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
194 strings according to the more liberal rules of RFC 2279. However, if
195 the string does not even conform to RFC 2279, the result is undefined.
196 Your program may crash.
197
198 If you want to process strings of values in the full range 0 to
199 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
200 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
201 this situation, you will have to apply your own validity check.
202
203 General comments about UTF-8 mode
204
205 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
206 two-byte UTF-8 character if the value is greater than 127.
207
208 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
209 characters for values greater than \177.
210
211 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
212 vidual bytes, for example: \x{100}{3}.
213
214 4. The dot metacharacter matches one UTF-8 character instead of a sin-
215 gle byte.
216
217 5. The escape sequence \C can be used to match a single byte in UTF-8
218 mode, but its use can lead to some strange effects. This facility is
219 not available in the alternative matching function, pcre_dfa_exec().
220
221 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
222 test characters of any code value, but the characters that PCRE recog-
223 nizes as digits, spaces, or word characters remain the same set as
224 before, all with values less than 256. This remains true even when PCRE
225 includes Unicode property support, because to do otherwise would slow
226 down PCRE in many common cases. If you really want to test for a wider
227 sense of, say, "digit", you must use Unicode property tests such as
228 \p{Nd}. Note that this also applies to \b, because it is defined in
229 terms of \w and \W.
230
231 7. Similarly, characters that match the POSIX named character classes
232 are all low-valued characters.
233
234 8. However, the Perl 5.10 horizontal and vertical whitespace matching
235 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
236 acters.
237
238 9. Case-insensitive matching applies only to characters whose values
239 are less than 128, unless PCRE is built with Unicode property support.
240 Even when Unicode property support is available, PCRE still uses its
241 own character tables when checking the case of low-valued characters,
242 so as not to degrade performance. The Unicode property information is
243 used only for characters with higher values. Even when Unicode property
244 support is available, PCRE supports case-insensitive matching only when
245 there is a one-to-one mapping between a letter's cases. There are a
246 small number of many-to-one mappings in Unicode; these are not sup-
247 ported by PCRE.
248
249
250 AUTHOR
251
252 Philip Hazel
253 University Computing Service
254 Cambridge CB2 3QH, England.
255
256 Putting an actual email address here seems to have been a spam magnet,
257 so I've taken it away. If you want to email me, use my two initials,
258 followed by the two digits 10, at the domain cam.ac.uk.
259
260
261 REVISION
262
263 Last updated: 11 April 2009
264 Copyright (c) 1997-2009 University of Cambridge.
265 ------------------------------------------------------------------------------
266
267
268 PCREBUILD(3) PCREBUILD(3)
269
270
271 NAME
272 PCRE - Perl-compatible regular expressions
273
274
275 PCRE BUILD-TIME OPTIONS
276
277 This document describes the optional features of PCRE that can be
278 selected when the library is compiled. It assumes use of the configure
279 script, where the optional features are selected or deselected by pro-
280 viding options to configure before running the make command. However,
281 the same options can be selected in both Unix-like and non-Unix-like
282 environments using the GUI facility of CMakeSetup if you are using
283 CMake instead of configure to build PCRE.
284
285 The complete list of options for configure (which includes the standard
286 ones such as the selection of the installation directory) can be
287 obtained by running
288
289 ./configure --help
290
291 The following sections include descriptions of options whose names
292 begin with --enable or --disable. These settings specify changes to the
293 defaults for the configure command. Because of the way that configure
294 works, --enable and --disable always come in pairs, so the complemen-
295 tary option always exists as well, but as it specifies the default, it
296 is not described.
297
298
299 C++ SUPPORT
300
301 By default, the configure script will search for a C++ compiler and C++
302 header files. If it finds them, it automatically builds the C++ wrapper
303 library for PCRE. You can disable this by adding
304
305 --disable-cpp
306
307 to the configure command.
308
309
310 UTF-8 SUPPORT
311
312 To build PCRE with support for UTF-8 Unicode character strings, add
313
314 --enable-utf8
315
316 to the configure command. Of itself, this does not make PCRE treat
317 strings as UTF-8. As well as compiling PCRE with this option, you also
318 have have to set the PCRE_UTF8 option when you call the pcre_compile()
319 function.
320
321 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
322 expects its input to be either ASCII or UTF-8 (depending on the runtime
323 option). It is not possible to support both EBCDIC and UTF-8 codes in
324 the same version of the library. Consequently, --enable-utf8 and
325 --enable-ebcdic are mutually exclusive.
326
327
328 UNICODE CHARACTER PROPERTY SUPPORT
329
330 UTF-8 support allows PCRE to process character values greater than 255
331 in the strings that it handles. On its own, however, it does not pro-
332 vide any facilities for accessing the properties of such characters. If
333 you want to be able to use the pattern escapes \P, \p, and \X, which
334 refer to Unicode character properties, you must add
335
336 --enable-unicode-properties
337
338 to the configure command. This implies UTF-8 support, even if you have
339 not explicitly requested it.
340
341 Including Unicode property support adds around 30K of tables to the
342 PCRE library. Only the general category properties such as Lu and Nd
343 are supported. Details are given in the pcrepattern documentation.
344
345
346 CODE VALUE OF NEWLINE
347
348 By default, PCRE interprets the linefeed (LF) character as indicating
349 the end of a line. This is the normal newline character on Unix-like
350 systems. You can compile PCRE to use carriage return (CR) instead, by
351 adding
352
353 --enable-newline-is-cr
354
355 to the configure command. There is also a --enable-newline-is-lf
356 option, which explicitly specifies linefeed as the newline character.
357
358 Alternatively, you can specify that line endings are to be indicated by
359 the two character sequence CRLF. If you want this, add
360
361 --enable-newline-is-crlf
362
363 to the configure command. There is a fourth option, specified by
364
365 --enable-newline-is-anycrlf
366
367 which causes PCRE to recognize any of the three sequences CR, LF, or
368 CRLF as indicating a line ending. Finally, a fifth option, specified by
369
370 --enable-newline-is-any
371
372 causes PCRE to recognize any Unicode newline sequence.
373
374 Whatever line ending convention is selected when PCRE is built can be
375 overridden when the library functions are called. At build time it is
376 conventional to use the standard for your operating system.
377
378
379 WHAT \R MATCHES
380
381 By default, the sequence \R in a pattern matches any Unicode newline
382 sequence, whatever has been selected as the line ending sequence. If
383 you specify
384
385 --enable-bsr-anycrlf
386
387 the default is changed so that \R matches only CR, LF, or CRLF. What-
388 ever is selected when PCRE is built can be overridden when the library
389 functions are called.
390
391
392 BUILDING SHARED AND STATIC LIBRARIES
393
394 The PCRE building process uses libtool to build both shared and static
395 Unix libraries by default. You can suppress one of these by adding one
396 of
397
398 --disable-shared
399 --disable-static
400
401 to the configure command, as required.
402
403
404 POSIX MALLOC USAGE
405
406 When PCRE is called through the POSIX interface (see the pcreposix doc-
407 umentation), additional working storage is required for holding the
408 pointers to capturing substrings, because PCRE requires three integers
409 per substring, whereas the POSIX interface provides only two. If the
410 number of expected substrings is small, the wrapper function uses space
411 on the stack, because this is faster than using malloc() for each call.
412 The default threshold above which the stack is no longer used is 10; it
413 can be changed by adding a setting such as
414
415 --with-posix-malloc-threshold=20
416
417 to the configure command.
418
419
420 HANDLING VERY LARGE PATTERNS
421
422 Within a compiled pattern, offset values are used to point from one
423 part to another (for example, from an opening parenthesis to an alter-
424 nation metacharacter). By default, two-byte values are used for these
425 offsets, leading to a maximum size for a compiled pattern of around
426 64K. This is sufficient to handle all but the most gigantic patterns.
427 Nevertheless, some people do want to process enormous patterns, so it
428 is possible to compile PCRE to use three-byte or four-byte offsets by
429 adding a setting such as
430
431 --with-link-size=3
432
433 to the configure command. The value given must be 2, 3, or 4. Using
434 longer offsets slows down the operation of PCRE because it has to load
435 additional bytes when handling them.
436
437
438 AVOIDING EXCESSIVE STACK USAGE
439
440 When matching with the pcre_exec() function, PCRE implements backtrack-
441 ing by making recursive calls to an internal function called match().
442 In environments where the size of the stack is limited, this can se-
443 verely limit PCRE's operation. (The Unix environment does not usually
444 suffer from this problem, but it may sometimes be necessary to increase
445 the maximum stack size. There is a discussion in the pcrestack docu-
446 mentation.) An alternative approach to recursion that uses memory from
447 the heap to remember data, instead of using recursive function calls,
448 has been implemented to work round the problem of limited stack size.
449 If you want to build a version of PCRE that works this way, add
450
451 --disable-stack-for-recursion
452
453 to the configure command. With this configuration, PCRE will use the
454 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
455 ment functions. By default these point to malloc() and free(), but you
456 can replace the pointers so that your own functions are used.
457
458 Separate functions are provided rather than using pcre_malloc and
459 pcre_free because the usage is very predictable: the block sizes
460 requested are always the same, and the blocks are always freed in
461 reverse order. A calling program might be able to implement optimized
462 functions that perform better than malloc() and free(). PCRE runs
463 noticeably more slowly when built in this way. This option affects only
464 the pcre_exec() function; it is not relevant for the the
465 pcre_dfa_exec() function.
466
467
468 LIMITING PCRE RESOURCE USAGE
469
470 Internally, PCRE has a function called match(), which it calls repeat-
471 edly (sometimes recursively) when matching a pattern with the
472 pcre_exec() function. By controlling the maximum number of times this
473 function may be called during a single matching operation, a limit can
474 be placed on the resources used by a single call to pcre_exec(). The
475 limit can be changed at run time, as described in the pcreapi documen-
476 tation. The default is 10 million, but this can be changed by adding a
477 setting such as
478
479 --with-match-limit=500000
480
481 to the configure command. This setting has no effect on the
482 pcre_dfa_exec() matching function.
483
484 In some environments it is desirable to limit the depth of recursive
485 calls of match() more strictly than the total number of calls, in order
486 to restrict the maximum amount of stack (or heap, if --disable-stack-
487 for-recursion is specified) that is used. A second limit controls this;
488 it defaults to the value that is set for --with-match-limit, which
489 imposes no additional constraints. However, you can set a lower limit
490 by adding, for example,
491
492 --with-match-limit-recursion=10000
493
494 to the configure command. This value can also be overridden at run
495 time.
496
497
498 CREATING CHARACTER TABLES AT BUILD TIME
499
500 PCRE uses fixed tables for processing characters whose code values are
501 less than 256. By default, PCRE is built with a set of tables that are
502 distributed in the file pcre_chartables.c.dist. These tables are for
503 ASCII codes only. If you add
504
505 --enable-rebuild-chartables
506
507 to the configure command, the distributed tables are no longer used.
508 Instead, a program called dftables is compiled and run. This outputs
509 the source for new set of tables, created in the default locale of your
510 C runtime system. (This method of replacing the tables does not work if
511 you are cross compiling, because dftables is run on the local host. If
512 you need to create alternative tables when cross compiling, you will
513 have to do so "by hand".)
514
515
516 USING EBCDIC CODE
517
518 PCRE assumes by default that it will run in an environment where the
519 character code is ASCII (or Unicode, which is a superset of ASCII).
520 This is the case for most computer operating systems. PCRE can, how-
521 ever, be compiled to run in an EBCDIC environment by adding
522
523 --enable-ebcdic
524
525 to the configure command. This setting implies --enable-rebuild-charta-
526 bles. You should only use it if you know that you are in an EBCDIC
527 environment (for example, an IBM mainframe operating system). The
528 --enable-ebcdic option is incompatible with --enable-utf8.
529
530
531 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
532
533 By default, pcregrep reads all files as plain text. You can build it so
534 that it recognizes files whose names end in .gz or .bz2, and reads them
535 with libz or libbz2, respectively, by adding one or both of
536
537 --enable-pcregrep-libz
538 --enable-pcregrep-libbz2
539
540 to the configure command. These options naturally require that the rel-
541 evant libraries are installed on your system. Configuration will fail
542 if they are not.
543
544
545 PCRETEST OPTION FOR LIBREADLINE SUPPORT
546
547 If you add
548
549 --enable-pcretest-libreadline
550
551 to the configure command, pcretest is linked with the libreadline
552 library, and when its input is from a terminal, it reads it using the
553 readline() function. This provides line-editing and history facilities.
554 Note that libreadline is GPL-licenced, so if you distribute a binary of
555 pcretest linked in this way, there may be licensing issues.
556
557 Setting this option causes the -lreadline option to be added to the
558 pcretest build. In many operating environments with a sytem-installed
559 libreadline this is sufficient. However, in some environments (e.g. if
560 an unmodified distribution version of readline is in use), some extra
561 configuration may be necessary. The INSTALL file for libreadline says
562 this:
563
564 "Readline uses the termcap functions, but does not link with the
565 termcap or curses library itself, allowing applications which link
566 with readline the to choose an appropriate library."
567
568 If your environment has not been set up so that an appropriate library
569 is automatically included, you may need to add something like
570
571 LIBS="-ncurses"
572
573 immediately before the configure command.
574
575
576 SEE ALSO
577
578 pcreapi(3), pcre_config(3).
579
580
581 AUTHOR
582
583 Philip Hazel
584 University Computing Service
585 Cambridge CB2 3QH, England.
586
587
588 REVISION
589
590 Last updated: 17 March 2009
591 Copyright (c) 1997-2009 University of Cambridge.
592 ------------------------------------------------------------------------------
593
594
595 PCREMATCHING(3) PCREMATCHING(3)
596
597
598 NAME
599 PCRE - Perl-compatible regular expressions
600
601
602 PCRE MATCHING ALGORITHMS
603
604 This document describes the two different algorithms that are available
605 in PCRE for matching a compiled regular expression against a given sub-
606 ject string. The "standard" algorithm is the one provided by the
607 pcre_exec() function. This works in the same was as Perl's matching
608 function, and provides a Perl-compatible matching operation.
609
610 An alternative algorithm is provided by the pcre_dfa_exec() function;
611 this operates in a different way, and is not Perl-compatible. It has
612 advantages and disadvantages compared with the standard algorithm, and
613 these are described below.
614
615 When there is only one possible way in which a given subject string can
616 match a pattern, the two algorithms give the same answer. A difference
617 arises, however, when there are multiple possibilities. For example, if
618 the pattern
619
620 ^<.*>
621
622 is matched against the string
623
624 <something> <something else> <something further>
625
626 there are three possible answers. The standard algorithm finds only one
627 of them, whereas the alternative algorithm finds all three.
628
629
630 REGULAR EXPRESSIONS AS TREES
631
632 The set of strings that are matched by a regular expression can be rep-
633 resented as a tree structure. An unlimited repetition in the pattern
634 makes the tree of infinite size, but it is still a tree. Matching the
635 pattern to a given subject string (from a given starting point) can be
636 thought of as a search of the tree. There are two ways to search a
637 tree: depth-first and breadth-first, and these correspond to the two
638 matching algorithms provided by PCRE.
639
640
641 THE STANDARD MATCHING ALGORITHM
642
643 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
644 sions", the standard algorithm is an "NFA algorithm". It conducts a
645 depth-first search of the pattern tree. That is, it proceeds along a
646 single path through the tree, checking that the subject matches what is
647 required. When there is a mismatch, the algorithm tries any alterna-
648 tives at the current point, and if they all fail, it backs up to the
649 previous branch point in the tree, and tries the next alternative
650 branch at that level. This often involves backing up (moving to the
651 left) in the subject string as well. The order in which repetition
652 branches are tried is controlled by the greedy or ungreedy nature of
653 the quantifier.
654
655 If a leaf node is reached, a matching string has been found, and at
656 that point the algorithm stops. Thus, if there is more than one possi-
657 ble match, this algorithm returns the first one that it finds. Whether
658 this is the shortest, the longest, or some intermediate length depends
659 on the way the greedy and ungreedy repetition quantifiers are specified
660 in the pattern.
661
662 Because it ends up with a single path through the tree, it is rela-
663 tively straightforward for this algorithm to keep track of the sub-
664 strings that are matched by portions of the pattern in parentheses.
665 This provides support for capturing parentheses and back references.
666
667
668 THE ALTERNATIVE MATCHING ALGORITHM
669
670 This algorithm conducts a breadth-first search of the tree. Starting
671 from the first matching point in the subject, it scans the subject
672 string from left to right, once, character by character, and as it does
673 this, it remembers all the paths through the tree that represent valid
674 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
675 though it is not implemented as a traditional finite state machine (it
676 keeps multiple states active simultaneously).
677
678 The scan continues until either the end of the subject is reached, or
679 there are no more unterminated paths. At this point, terminated paths
680 represent the different matching possibilities (if there are none, the
681 match has failed). Thus, if there is more than one possible match,
682 this algorithm finds all of them, and in particular, it finds the long-
683 est. In PCRE, there is an option to stop the algorithm after the first
684 match (which is necessarily the shortest) has been found.
685
686 Note that all the matches that are found start at the same point in the
687 subject. If the pattern
688
689 cat(er(pillar)?)
690
691 is matched against the string "the caterpillar catchment", the result
692 will be the three strings "cat", "cater", and "caterpillar" that start
693 at the fourth character of the subject. The algorithm does not automat-
694 ically move on to find matches that start at later positions.
695
696 There are a number of features of PCRE regular expressions that are not
697 supported by the alternative matching algorithm. They are as follows:
698
699 1. Because the algorithm finds all possible matches, the greedy or
700 ungreedy nature of repetition quantifiers is not relevant. Greedy and
701 ungreedy quantifiers are treated in exactly the same way. However, pos-
702 sessive quantifiers can make a difference when what follows could also
703 match what is quantified, for example in a pattern like this:
704
705 ^a++\w!
706
707 This pattern matches "aaab!" but not "aaa!", which would be matched by
708 a non-possessive quantifier. Similarly, if an atomic group is present,
709 it is matched as if it were a standalone pattern at the current point,
710 and the longest match is then "locked in" for the rest of the overall
711 pattern.
712
713 2. When dealing with multiple paths through the tree simultaneously, it
714 is not straightforward to keep track of captured substrings for the
715 different matching possibilities, and PCRE's implementation of this
716 algorithm does not attempt to do this. This means that no captured sub-
717 strings are available.
718
719 3. Because no substrings are captured, back references within the pat-
720 tern are not supported, and cause errors if encountered.
721
722 4. For the same reason, conditional expressions that use a backrefer-
723 ence as the condition or test for a specific group recursion are not
724 supported.
725
726 5. Because many paths through the tree may be active, the \K escape
727 sequence, which resets the start of the match when encountered (but may
728 be on some paths and not on others), is not supported. It causes an
729 error if encountered.
730
731 6. Callouts are supported, but the value of the capture_top field is
732 always 1, and the value of the capture_last field is always -1.
733
734 7. The \C escape sequence, which (in the standard algorithm) matches a
735 single byte, even in UTF-8 mode, is not supported because the alterna-
736 tive algorithm moves through the subject string one character at a
737 time, for all active paths through the tree.
738
739 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
740 are not supported. (*FAIL) is supported, and behaves like a failing
741 negative assertion.
742
743
744 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
745
746 Using the alternative matching algorithm provides the following advan-
747 tages:
748
749 1. All possible matches (at a single point in the subject) are automat-
750 ically found, and in particular, the longest match is found. To find
751 more than one match using the standard algorithm, you have to do kludgy
752 things with callouts.
753
754 2. There is much better support for partial matching. The restrictions
755 on the content of the pattern that apply when using the standard algo-
756 rithm for partial matching do not apply to the alternative algorithm.
757 For non-anchored patterns, the starting position of a partial match is
758 available.
759
760 3. Because the alternative algorithm scans the subject string just
761 once, and never needs to backtrack, it is possible to pass very long
762 subject strings to the matching function in several pieces, checking
763 for partial matching each time.
764
765
766 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
767
768 The alternative algorithm suffers from a number of disadvantages:
769
770 1. It is substantially slower than the standard algorithm. This is
771 partly because it has to search for all possible matches, but is also
772 because it is less susceptible to optimization.
773
774 2. Capturing parentheses and back references are not supported.
775
776 3. Although atomic groups are supported, their use does not provide the
777 performance advantage that it does for the standard algorithm.
778
779
780 AUTHOR
781
782 Philip Hazel
783 University Computing Service
784 Cambridge CB2 3QH, England.
785
786
787 REVISION
788
789 Last updated: 19 April 2008
790 Copyright (c) 1997-2008 University of Cambridge.
791 ------------------------------------------------------------------------------
792
793
794 PCREAPI(3) PCREAPI(3)
795
796
797 NAME
798 PCRE - Perl-compatible regular expressions
799
800
801 PCRE NATIVE API
802
803 #include <pcre.h>
804
805 pcre *pcre_compile(const char *pattern, int options,
806 const char **errptr, int *erroffset,
807 const unsigned char *tableptr);
808
809 pcre *pcre_compile2(const char *pattern, int options,
810 int *errorcodeptr,
811 const char **errptr, int *erroffset,
812 const unsigned char *tableptr);
813
814 pcre_extra *pcre_study(const pcre *code, int options,
815 const char **errptr);
816
817 int pcre_exec(const pcre *code, const pcre_extra *extra,
818 const char *subject, int length, int startoffset,
819 int options, int *ovector, int ovecsize);
820
821 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
822 const char *subject, int length, int startoffset,
823 int options, int *ovector, int ovecsize,
824 int *workspace, int wscount);
825
826 int pcre_copy_named_substring(const pcre *code,
827 const char *subject, int *ovector,
828 int stringcount, const char *stringname,
829 char *buffer, int buffersize);
830
831 int pcre_copy_substring(const char *subject, int *ovector,
832 int stringcount, int stringnumber, char *buffer,
833 int buffersize);
834
835 int pcre_get_named_substring(const pcre *code,
836 const char *subject, int *ovector,
837 int stringcount, const char *stringname,
838 const char **stringptr);
839
840 int pcre_get_stringnumber(const pcre *code,
841 const char *name);
842
843 int pcre_get_stringtable_entries(const pcre *code,
844 const char *name, char **first, char **last);
845
846 int pcre_get_substring(const char *subject, int *ovector,
847 int stringcount, int stringnumber,
848 const char **stringptr);
849
850 int pcre_get_substring_list(const char *subject,
851 int *ovector, int stringcount, const char ***listptr);
852
853 void pcre_free_substring(const char *stringptr);
854
855 void pcre_free_substring_list(const char **stringptr);
856
857 const unsigned char *pcre_maketables(void);
858
859 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
860 int what, void *where);
861
862 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
863
864 int pcre_refcount(pcre *code, int adjust);
865
866 int pcre_config(int what, void *where);
867
868 char *pcre_version(void);
869
870 void *(*pcre_malloc)(size_t);
871
872 void (*pcre_free)(void *);
873
874 void *(*pcre_stack_malloc)(size_t);
875
876 void (*pcre_stack_free)(void *);
877
878 int (*pcre_callout)(pcre_callout_block *);
879
880
881 PCRE API OVERVIEW
882
883 PCRE has its own native API, which is described in this document. There
884 are also some wrapper functions that correspond to the POSIX regular
885 expression API. These are described in the pcreposix documentation.
886 Both of these APIs define a set of C function calls. A C++ wrapper is
887 distributed with PCRE. It is documented in the pcrecpp page.
888
889 The native API C function prototypes are defined in the header file
890 pcre.h, and on Unix systems the library itself is called libpcre. It
891 can normally be accessed by adding -lpcre to the command for linking an
892 application that uses PCRE. The header file defines the macros
893 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
894 bers for the library. Applications can use these to include support
895 for different releases of PCRE.
896
897 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
898 pcre_exec() are used for compiling and matching regular expressions in
899 a Perl-compatible manner. A sample program that demonstrates the sim-
900 plest way of using them is provided in the file called pcredemo.c in
901 the source distribution. The pcresample documentation describes how to
902 compile and run it.
903
904 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
905 ble, is also provided. This uses a different algorithm for the match-
906 ing. The alternative algorithm finds all possible matches (at a given
907 point in the subject), and scans the subject just once. However, this
908 algorithm does not return captured substrings. A description of the two
909 matching algorithms and their advantages and disadvantages is given in
910 the pcrematching documentation.
911
912 In addition to the main compiling and matching functions, there are
913 convenience functions for extracting captured substrings from a subject
914 string that is matched by pcre_exec(). They are:
915
916 pcre_copy_substring()
917 pcre_copy_named_substring()
918 pcre_get_substring()
919 pcre_get_named_substring()
920 pcre_get_substring_list()
921 pcre_get_stringnumber()
922 pcre_get_stringtable_entries()
923
924 pcre_free_substring() and pcre_free_substring_list() are also provided,
925 to free the memory used for extracted strings.
926
927 The function pcre_maketables() is used to build a set of character
928 tables in the current locale for passing to pcre_compile(),
929 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
930 provided for specialist use. Most commonly, no special tables are
931 passed, in which case internal tables that are generated when PCRE is
932 built are used.
933
934 The function pcre_fullinfo() is used to find out information about a
935 compiled pattern; pcre_info() is an obsolete version that returns only
936 some of the available information, but is retained for backwards com-
937 patibility. The function pcre_version() returns a pointer to a string
938 containing the version of PCRE and its date of release.
939
940 The function pcre_refcount() maintains a reference count in a data
941 block containing a compiled pattern. This is provided for the benefit
942 of object-oriented applications.
943
944 The global variables pcre_malloc and pcre_free initially contain the
945 entry points of the standard malloc() and free() functions, respec-
946 tively. PCRE calls the memory management functions via these variables,
947 so a calling program can replace them if it wishes to intercept the
948 calls. This should be done before calling any PCRE functions.
949
950 The global variables pcre_stack_malloc and pcre_stack_free are also
951 indirections to memory management functions. These special functions
952 are used only when PCRE is compiled to use the heap for remembering
953 data, instead of recursive function calls, when running the pcre_exec()
954 function. See the pcrebuild documentation for details of how to do
955 this. It is a non-standard way of building PCRE, for use in environ-
956 ments that have limited stacks. Because of the greater use of memory
957 management, it runs more slowly. Separate functions are provided so
958 that special-purpose external code can be used for this case. When
959 used, these functions are always called in a stack-like manner (last
960 obtained, first freed), and always for memory blocks of the same size.
961 There is a discussion about PCRE's stack usage in the pcrestack docu-
962 mentation.
963
964 The global variable pcre_callout initially contains NULL. It can be set
965 by the caller to a "callout" function, which PCRE will then call at
966 specified points during a matching operation. Details are given in the
967 pcrecallout documentation.
968
969
970 NEWLINES
971
972 PCRE supports five different conventions for indicating line breaks in
973 strings: a single CR (carriage return) character, a single LF (line-
974 feed) character, the two-character sequence CRLF, any of the three pre-
975 ceding, or any Unicode newline sequence. The Unicode newline sequences
976 are the three just mentioned, plus the single characters VT (vertical
977 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
978 separator, U+2028), and PS (paragraph separator, U+2029).
979
980 Each of the first three conventions is used by at least one operating
981 system as its standard newline sequence. When PCRE is built, a default
982 can be specified. The default default is LF, which is the Unix stan-
983 dard. When PCRE is run, the default can be overridden, either when a
984 pattern is compiled, or when it is matched.
985
986 At compile time, the newline convention can be specified by the options
987 argument of pcre_compile(), or it can be specified by special text at
988 the start of the pattern itself; this overrides any other settings. See
989 the pcrepattern page for details of the special character sequences.
990
991 In the PCRE documentation the word "newline" is used to mean "the char-
992 acter or pair of characters that indicate a line break". The choice of
993 newline convention affects the handling of the dot, circumflex, and
994 dollar metacharacters, the handling of #-comments in /x mode, and, when
995 CRLF is a recognized line ending sequence, the match position advance-
996 ment for a non-anchored pattern. There is more detail about this in the
997 section on pcre_exec() options below.
998
999 The choice of newline convention does not affect the interpretation of
1000 the \n or \r escape sequences, nor does it affect what \R matches,
1001 which is controlled in a similar way, but by separate options.
1002
1003
1004 MULTITHREADING
1005
1006 The PCRE functions can be used in multi-threading applications, with
1007 the proviso that the memory management functions pointed to by
1008 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1009 callout function pointed to by pcre_callout, are shared by all threads.
1010
1011 The compiled form of a regular expression is not altered during match-
1012 ing, so the same compiled pattern can safely be used by several threads
1013 at once.
1014
1015
1016 SAVING PRECOMPILED PATTERNS FOR LATER USE
1017
1018 The compiled form of a regular expression can be saved and re-used at a
1019 later time, possibly by a different program, and even on a host other
1020 than the one on which it was compiled. Details are given in the
1021 pcreprecompile documentation. However, compiling a regular expression
1022 with one version of PCRE for use with a different version is not guar-
1023 anteed to work and may cause crashes.
1024
1025
1026 CHECKING BUILD-TIME OPTIONS
1027
1028 int pcre_config(int what, void *where);
1029
1030 The function pcre_config() makes it possible for a PCRE client to dis-
1031 cover which optional features have been compiled into the PCRE library.
1032 The pcrebuild documentation has more details about these optional fea-
1033 tures.
1034
1035 The first argument for pcre_config() is an integer, specifying which
1036 information is required; the second argument is a pointer to a variable
1037 into which the information is placed. The following information is
1038 available:
1039
1040 PCRE_CONFIG_UTF8
1041
1042 The output is an integer that is set to one if UTF-8 support is avail-
1043 able; otherwise it is set to zero.
1044
1045 PCRE_CONFIG_UNICODE_PROPERTIES
1046
1047 The output is an integer that is set to one if support for Unicode
1048 character properties is available; otherwise it is set to zero.
1049
1050 PCRE_CONFIG_NEWLINE
1051
1052 The output is an integer whose value specifies the default character
1053 sequence that is recognized as meaning "newline". The four values that
1054 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1055 and -1 for ANY. Though they are derived from ASCII, the same values
1056 are returned in EBCDIC environments. The default should normally corre-
1057 spond to the standard sequence for your operating system.
1058
1059 PCRE_CONFIG_BSR
1060
1061 The output is an integer whose value indicates what character sequences
1062 the \R escape sequence matches by default. A value of 0 means that \R
1063 matches any Unicode line ending sequence; a value of 1 means that \R
1064 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1065 tern is compiled or matched.
1066
1067 PCRE_CONFIG_LINK_SIZE
1068
1069 The output is an integer that contains the number of bytes used for
1070 internal linkage in compiled regular expressions. The value is 2, 3, or
1071 4. Larger values allow larger regular expressions to be compiled, at
1072 the expense of slower matching. The default value of 2 is sufficient
1073 for all but the most massive patterns, since it allows the compiled
1074 pattern to be up to 64K in size.
1075
1076 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1077
1078 The output is an integer that contains the threshold above which the
1079 POSIX interface uses malloc() for output vectors. Further details are
1080 given in the pcreposix documentation.
1081
1082 PCRE_CONFIG_MATCH_LIMIT
1083
1084 The output is a long integer that gives the default limit for the num-
1085 ber of internal matching function calls in a pcre_exec() execution.
1086 Further details are given with pcre_exec() below.
1087
1088 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1089
1090 The output is a long integer that gives the default limit for the depth
1091 of recursion when calling the internal matching function in a
1092 pcre_exec() execution. Further details are given with pcre_exec()
1093 below.
1094
1095 PCRE_CONFIG_STACKRECURSE
1096
1097 The output is an integer that is set to one if internal recursion when
1098 running pcre_exec() is implemented by recursive function calls that use
1099 the stack to remember their state. This is the usual way that PCRE is
1100 compiled. The output is zero if PCRE was compiled to use blocks of data
1101 on the heap instead of recursive function calls. In this case,
1102 pcre_stack_malloc and pcre_stack_free are called to manage memory
1103 blocks on the heap, thus avoiding the use of the stack.
1104
1105
1106 COMPILING A PATTERN
1107
1108 pcre *pcre_compile(const char *pattern, int options,
1109 const char **errptr, int *erroffset,
1110 const unsigned char *tableptr);
1111
1112 pcre *pcre_compile2(const char *pattern, int options,
1113 int *errorcodeptr,
1114 const char **errptr, int *erroffset,
1115 const unsigned char *tableptr);
1116
1117 Either of the functions pcre_compile() or pcre_compile2() can be called
1118 to compile a pattern into an internal form. The only difference between
1119 the two interfaces is that pcre_compile2() has an additional argument,
1120 errorcodeptr, via which a numerical error code can be returned.
1121
1122 The pattern is a C string terminated by a binary zero, and is passed in
1123 the pattern argument. A pointer to a single block of memory that is
1124 obtained via pcre_malloc is returned. This contains the compiled code
1125 and related data. The pcre type is defined for the returned block; this
1126 is a typedef for a structure whose contents are not externally defined.
1127 It is up to the caller to free the memory (via pcre_free) when it is no
1128 longer required.
1129
1130 Although the compiled code of a PCRE regex is relocatable, that is, it
1131 does not depend on memory location, the complete pcre data block is not
1132 fully relocatable, because it may contain a copy of the tableptr argu-
1133 ment, which is an address (see below).
1134
1135 The options argument contains various bit settings that affect the com-
1136 pilation. It should be zero if no options are required. The available
1137 options are described below. Some of them (in particular, those that
1138 are compatible with Perl, but also some others) can also be set and
1139 unset from within the pattern (see the detailed description in the
1140 pcrepattern documentation). For those options that can be different in
1141 different parts of the pattern, the contents of the options argument
1142 specifies their initial settings at the start of compilation and execu-
1143 tion. The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the
1144 time of matching as well as at compile time.
1145
1146 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1147 if compilation of a pattern fails, pcre_compile() returns NULL, and
1148 sets the variable pointed to by errptr to point to a textual error mes-
1149 sage. This is a static string that is part of the library. You must not
1150 try to free it. The offset from the start of the pattern to the charac-
1151 ter where the error was discovered is placed in the variable pointed to
1152 by erroffset, which must not be NULL. If it is, an immediate error is
1153 given.
1154
1155 If pcre_compile2() is used instead of pcre_compile(), and the error-
1156 codeptr argument is not NULL, a non-zero error code number is returned
1157 via this argument in the event of an error. This is in addition to the
1158 textual error message. Error codes and messages are listed below.
1159
1160 If the final argument, tableptr, is NULL, PCRE uses a default set of
1161 character tables that are built when PCRE is compiled, using the
1162 default C locale. Otherwise, tableptr must be an address that is the
1163 result of a call to pcre_maketables(). This value is stored with the
1164 compiled pattern, and used again by pcre_exec(), unless another table
1165 pointer is passed to it. For more discussion, see the section on locale
1166 support below.
1167
1168 This code fragment shows a typical straightforward call to pcre_com-
1169 pile():
1170
1171 pcre *re;
1172 const char *error;
1173 int erroffset;
1174 re = pcre_compile(
1175 "^A.*Z", /* the pattern */
1176 0, /* default options */
1177 &error, /* for error message */
1178 &erroffset, /* for error offset */
1179 NULL); /* use default character tables */
1180
1181 The following names for option bits are defined in the pcre.h header
1182 file:
1183
1184 PCRE_ANCHORED
1185
1186 If this bit is set, the pattern is forced to be "anchored", that is, it
1187 is constrained to match only at the first matching point in the string
1188 that is being searched (the "subject string"). This effect can also be
1189 achieved by appropriate constructs in the pattern itself, which is the
1190 only way to do it in Perl.
1191
1192 PCRE_AUTO_CALLOUT
1193
1194 If this bit is set, pcre_compile() automatically inserts callout items,
1195 all with number 255, before each pattern item. For discussion of the
1196 callout facility, see the pcrecallout documentation.
1197
1198 PCRE_BSR_ANYCRLF
1199 PCRE_BSR_UNICODE
1200
1201 These options (which are mutually exclusive) control what the \R escape
1202 sequence matches. The choice is either to match only CR, LF, or CRLF,
1203 or to match any Unicode newline sequence. The default is specified when
1204 PCRE is built. It can be overridden from within the pattern, or by set-
1205 ting an option when a compiled pattern is matched.
1206
1207 PCRE_CASELESS
1208
1209 If this bit is set, letters in the pattern match both upper and lower
1210 case letters. It is equivalent to Perl's /i option, and it can be
1211 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1212 always understands the concept of case for characters whose values are
1213 less than 128, so caseless matching is always possible. For characters
1214 with higher values, the concept of case is supported if PCRE is com-
1215 piled with Unicode property support, but not otherwise. If you want to
1216 use caseless matching for characters 128 and above, you must ensure
1217 that PCRE is compiled with Unicode property support as well as with
1218 UTF-8 support.
1219
1220 PCRE_DOLLAR_ENDONLY
1221
1222 If this bit is set, a dollar metacharacter in the pattern matches only
1223 at the end of the subject string. Without this option, a dollar also
1224 matches immediately before a newline at the end of the string (but not
1225 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1226 if PCRE_MULTILINE is set. There is no equivalent to this option in
1227 Perl, and no way to set it within a pattern.
1228
1229 PCRE_DOTALL
1230
1231 If this bit is set, a dot metacharater in the pattern matches all char-
1232 acters, including those that indicate newline. Without it, a dot does
1233 not match when the current position is at a newline. This option is
1234 equivalent to Perl's /s option, and it can be changed within a pattern
1235 by a (?s) option setting. A negative class such as [^a] always matches
1236 newline characters, independent of the setting of this option.
1237
1238 PCRE_DUPNAMES
1239
1240 If this bit is set, names used to identify capturing subpatterns need
1241 not be unique. This can be helpful for certain types of pattern when it
1242 is known that only one instance of the named subpattern can ever be
1243 matched. There are more details of named subpatterns below; see also
1244 the pcrepattern documentation.
1245
1246 PCRE_EXTENDED
1247
1248 If this bit is set, whitespace data characters in the pattern are
1249 totally ignored except when escaped or inside a character class. White-
1250 space does not include the VT character (code 11). In addition, charac-
1251 ters between an unescaped # outside a character class and the next new-
1252 line, inclusive, are also ignored. This is equivalent to Perl's /x
1253 option, and it can be changed within a pattern by a (?x) option set-
1254 ting.
1255
1256 This option makes it possible to include comments inside complicated
1257 patterns. Note, however, that this applies only to data characters.
1258 Whitespace characters may never appear within special character
1259 sequences in a pattern, for example within the sequence (?( which
1260 introduces a conditional subpattern.
1261
1262 PCRE_EXTRA
1263
1264 This option was invented in order to turn on additional functionality
1265 of PCRE that is incompatible with Perl, but it is currently of very
1266 little use. When set, any backslash in a pattern that is followed by a
1267 letter that has no special meaning causes an error, thus reserving
1268 these combinations for future expansion. By default, as in Perl, a
1269 backslash followed by a letter with no special meaning is treated as a
1270 literal. (Perl can, however, be persuaded to give a warning for this.)
1271 There are at present no other features controlled by this option. It
1272 can also be set by a (?X) option setting within a pattern.
1273
1274 PCRE_FIRSTLINE
1275
1276 If this option is set, an unanchored pattern is required to match
1277 before or at the first newline in the subject string, though the
1278 matched text may continue over the newline.
1279
1280 PCRE_JAVASCRIPT_COMPAT
1281
1282 If this option is set, PCRE's behaviour is changed in some ways so that
1283 it is compatible with JavaScript rather than Perl. The changes are as
1284 follows:
1285
1286 (1) A lone closing square bracket in a pattern causes a compile-time
1287 error, because this is illegal in JavaScript (by default it is treated
1288 as a data character). Thus, the pattern AB]CD becomes illegal when this
1289 option is set.
1290
1291 (2) At run time, a back reference to an unset subpattern group matches
1292 an empty string (by default this causes the current matching alterna-
1293 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1294 set (assuming it can find an "a" in the subject), whereas it fails by
1295 default, for Perl compatibility.
1296
1297 PCRE_MULTILINE
1298
1299 By default, PCRE treats the subject string as consisting of a single
1300 line of characters (even if it actually contains newlines). The "start
1301 of line" metacharacter (^) matches only at the start of the string,
1302 while the "end of line" metacharacter ($) matches only at the end of
1303 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1304 is set). This is the same as Perl.
1305
1306 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1307 constructs match immediately following or immediately before internal
1308 newlines in the subject string, respectively, as well as at the very
1309 start and end. This is equivalent to Perl's /m option, and it can be
1310 changed within a pattern by a (?m) option setting. If there are no new-
1311 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1312 setting PCRE_MULTILINE has no effect.
1313
1314 PCRE_NEWLINE_CR
1315 PCRE_NEWLINE_LF
1316 PCRE_NEWLINE_CRLF
1317 PCRE_NEWLINE_ANYCRLF
1318 PCRE_NEWLINE_ANY
1319
1320 These options override the default newline definition that was chosen
1321 when PCRE was built. Setting the first or the second specifies that a
1322 newline is indicated by a single character (CR or LF, respectively).
1323 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1324 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1325 that any of the three preceding sequences should be recognized. Setting
1326 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1327 recognized. The Unicode newline sequences are the three just mentioned,
1328 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1329 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1330 (paragraph separator, U+2029). The last two are recognized only in
1331 UTF-8 mode.
1332
1333 The newline setting in the options word uses three bits that are
1334 treated as a number, giving eight possibilities. Currently only six are
1335 used (default plus the five values above). This means that if you set
1336 more than one newline option, the combination may or may not be sensi-
1337 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1338 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1339 cause an error.
1340
1341 The only time that a line break is specially recognized when compiling
1342 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1343 character class is encountered. This indicates a comment that lasts
1344 until after the next line break sequence. In other circumstances, line
1345 break sequences are treated as literal data, except that in
1346 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1347 and are therefore ignored.
1348
1349 The newline option that is set at compile time becomes the default that
1350 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1351
1352 PCRE_NO_AUTO_CAPTURE
1353
1354 If this option is set, it disables the use of numbered capturing paren-
1355 theses in the pattern. Any opening parenthesis that is not followed by
1356 ? behaves as if it were followed by ?: but named parentheses can still
1357 be used for capturing (and they acquire numbers in the usual way).
1358 There is no equivalent of this option in Perl.
1359
1360 PCRE_UNGREEDY
1361
1362 This option inverts the "greediness" of the quantifiers so that they
1363 are not greedy by default, but become greedy if followed by "?". It is
1364 not compatible with Perl. It can also be set by a (?U) option setting
1365 within the pattern.
1366
1367 PCRE_UTF8
1368
1369 This option causes PCRE to regard both the pattern and the subject as
1370 strings of UTF-8 characters instead of single-byte character strings.
1371 However, it is available only when PCRE is built to include UTF-8 sup-
1372 port. If not, the use of this option provokes an error. Details of how
1373 this option changes the behaviour of PCRE are given in the section on
1374 UTF-8 support in the main pcre page.
1375
1376 PCRE_NO_UTF8_CHECK
1377
1378 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1379 automatically checked. There is a discussion about the validity of
1380 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1381 bytes is found, pcre_compile() returns an error. If you already know
1382 that your pattern is valid, and you want to skip this check for perfor-
1383 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1384 set, the effect of passing an invalid UTF-8 string as a pattern is
1385 undefined. It may cause your program to crash. Note that this option
1386 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1387 UTF-8 validity checking of subject strings.
1388
1389
1390 COMPILATION ERROR CODES
1391
1392 The following table lists the error codes than may be returned by
1393 pcre_compile2(), along with the error messages that may be returned by
1394 both compiling functions. As PCRE has developed, some error codes have
1395 fallen out of use. To avoid confusion, they have not been re-used.
1396
1397 0 no error
1398 1 \ at end of pattern
1399 2 \c at end of pattern
1400 3 unrecognized character follows \
1401 4 numbers out of order in {} quantifier
1402 5 number too big in {} quantifier
1403 6 missing terminating ] for character class
1404 7 invalid escape sequence in character class
1405 8 range out of order in character class
1406 9 nothing to repeat
1407 10 [this code is not in use]
1408 11 internal error: unexpected repeat
1409 12 unrecognized character after (? or (?-
1410 13 POSIX named classes are supported only within a class
1411 14 missing )
1412 15 reference to non-existent subpattern
1413 16 erroffset passed as NULL
1414 17 unknown option bit(s) set
1415 18 missing ) after comment
1416 19 [this code is not in use]
1417 20 regular expression is too large
1418 21 failed to get memory
1419 22 unmatched parentheses
1420 23 internal error: code overflow
1421 24 unrecognized character after (?<
1422 25 lookbehind assertion is not fixed length
1423 26 malformed number or name after (?(
1424 27 conditional group contains more than two branches
1425 28 assertion expected after (?(
1426 29 (?R or (?[+-]digits must be followed by )
1427 30 unknown POSIX class name
1428 31 POSIX collating elements are not supported
1429 32 this version of PCRE is not compiled with PCRE_UTF8 support
1430 33 [this code is not in use]
1431 34 character value in \x{...} sequence is too large
1432 35 invalid condition (?(0)
1433 36 \C not allowed in lookbehind assertion
1434 37 PCRE does not support \L, \l, \N, \U, or \u
1435 38 number after (?C is > 255
1436 39 closing ) for (?C expected
1437 40 recursive call could loop indefinitely
1438 41 unrecognized character after (?P
1439 42 syntax error in subpattern name (missing terminator)
1440 43 two named subpatterns have the same name
1441 44 invalid UTF-8 string
1442 45 support for \P, \p, and \X has not been compiled
1443 46 malformed \P or \p sequence
1444 47 unknown property name after \P or \p
1445 48 subpattern name is too long (maximum 32 characters)
1446 49 too many named subpatterns (maximum 10000)
1447 50 [this code is not in use]
1448 51 octal value is greater than \377 (not in UTF-8 mode)
1449 52 internal error: overran compiling workspace
1450 53 internal error: previously-checked referenced subpattern not
1451 found
1452 54 DEFINE group contains more than one branch
1453 55 repeating a DEFINE group is not allowed
1454 56 inconsistent NEWLINE options
1455 57 \g is not followed by a braced, angle-bracketed, or quoted
1456 name/number or by a plain number
1457 58 a numbered reference must not be zero
1458 59 (*VERB) with an argument is not supported
1459 60 (*VERB) not recognized
1460 61 number is too big
1461 62 subpattern name expected
1462 63 digit expected after (?+
1463 64 ] is an invalid data character in JavaScript compatibility mode
1464
1465 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1466 values may be used if the limits were changed when PCRE was built.
1467
1468
1469 STUDYING A PATTERN
1470
1471 pcre_extra *pcre_study(const pcre *code, int options
1472 const char **errptr);
1473
1474 If a compiled pattern is going to be used several times, it is worth
1475 spending more time analyzing it in order to speed up the time taken for
1476 matching. The function pcre_study() takes a pointer to a compiled pat-
1477 tern as its first argument. If studying the pattern produces additional
1478 information that will help speed up matching, pcre_study() returns a
1479 pointer to a pcre_extra block, in which the study_data field points to
1480 the results of the study.
1481
1482 The returned value from pcre_study() can be passed directly to
1483 pcre_exec(). However, a pcre_extra block also contains other fields
1484 that can be set by the caller before the block is passed; these are
1485 described below in the section on matching a pattern.
1486
1487 If studying the pattern does not produce any additional information
1488 pcre_study() returns NULL. In that circumstance, if the calling program
1489 wants to pass any of the other fields to pcre_exec(), it must set up
1490 its own pcre_extra block.
1491
1492 The second argument of pcre_study() contains option bits. At present,
1493 no options are defined, and this argument should always be zero.
1494
1495 The third argument for pcre_study() is a pointer for an error message.
1496 If studying succeeds (even if no data is returned), the variable it
1497 points to is set to NULL. Otherwise it is set to point to a textual
1498 error message. This is a static string that is part of the library. You
1499 must not try to free it. You should test the error pointer for NULL
1500 after calling pcre_study(), to be sure that it has run successfully.
1501
1502 This is a typical call to pcre_study():
1503
1504 pcre_extra *pe;
1505 pe = pcre_study(
1506 re, /* result of pcre_compile() */
1507 0, /* no options exist */
1508 &error); /* set to NULL or points to a message */
1509
1510 At present, studying a pattern is useful only for non-anchored patterns
1511 that do not have a single fixed starting character. A bitmap of possi-
1512 ble starting bytes is created.
1513
1514
1515 LOCALE SUPPORT
1516
1517 PCRE handles caseless matching, and determines whether characters are
1518 letters, digits, or whatever, by reference to a set of tables, indexed
1519 by character value. When running in UTF-8 mode, this applies only to
1520 characters with codes less than 128. Higher-valued codes never match
1521 escapes such as \w or \d, but can be tested with \p if PCRE is built
1522 with Unicode character property support. The use of locales with Uni-
1523 code is discouraged. If you are handling characters with codes greater
1524 than 128, you should either use UTF-8 and Unicode, or use locales, but
1525 not try to mix the two.
1526
1527 PCRE contains an internal set of tables that are used when the final
1528 argument of pcre_compile() is NULL. These are sufficient for many
1529 applications. Normally, the internal tables recognize only ASCII char-
1530 acters. However, when PCRE is built, it is possible to cause the inter-
1531 nal tables to be rebuilt in the default "C" locale of the local system,
1532 which may cause them to be different.
1533
1534 The internal tables can always be overridden by tables supplied by the
1535 application that calls PCRE. These may be created in a different locale
1536 from the default. As more and more applications change to using Uni-
1537 code, the need for this locale support is expected to die away.
1538
1539 External tables are built by calling the pcre_maketables() function,
1540 which has no arguments, in the relevant locale. The result can then be
1541 passed to pcre_compile() or pcre_exec() as often as necessary. For
1542 example, to build and use tables that are appropriate for the French
1543 locale (where accented characters with values greater than 128 are
1544 treated as letters), the following code could be used:
1545
1546 setlocale(LC_CTYPE, "fr_FR");
1547 tables = pcre_maketables();
1548 re = pcre_compile(..., tables);
1549
1550 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1551 if you are using Windows, the name for the French locale is "french".
1552
1553 When pcre_maketables() runs, the tables are built in memory that is
1554 obtained via pcre_malloc. It is the caller's responsibility to ensure
1555 that the memory containing the tables remains available for as long as
1556 it is needed.
1557
1558 The pointer that is passed to pcre_compile() is saved with the compiled
1559 pattern, and the same tables are used via this pointer by pcre_study()
1560 and normally also by pcre_exec(). Thus, by default, for any single pat-
1561 tern, compilation, studying and matching all happen in the same locale,
1562 but different patterns can be compiled in different locales.
1563
1564 It is possible to pass a table pointer or NULL (indicating the use of
1565 the internal tables) to pcre_exec(). Although not intended for this
1566 purpose, this facility could be used to match a pattern in a different
1567 locale from the one in which it was compiled. Passing table pointers at
1568 run time is discussed below in the section on matching a pattern.
1569
1570
1571 INFORMATION ABOUT A PATTERN
1572
1573 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1574 int what, void *where);
1575
1576 The pcre_fullinfo() function returns information about a compiled pat-
1577 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1578 less retained for backwards compability (and is documented below).
1579
1580 The first argument for pcre_fullinfo() is a pointer to the compiled
1581 pattern. The second argument is the result of pcre_study(), or NULL if
1582 the pattern was not studied. The third argument specifies which piece
1583 of information is required, and the fourth argument is a pointer to a
1584 variable to receive the data. The yield of the function is zero for
1585 success, or one of the following negative numbers:
1586
1587 PCRE_ERROR_NULL the argument code was NULL
1588 the argument where was NULL
1589 PCRE_ERROR_BADMAGIC the "magic number" was not found
1590 PCRE_ERROR_BADOPTION the value of what was invalid
1591
1592 The "magic number" is placed at the start of each compiled pattern as
1593 an simple check against passing an arbitrary memory pointer. Here is a
1594 typical call of pcre_fullinfo(), to obtain the length of the compiled
1595 pattern:
1596
1597 int rc;
1598 size_t length;
1599 rc = pcre_fullinfo(
1600 re, /* result of pcre_compile() */
1601 pe, /* result of pcre_study(), or NULL */
1602 PCRE_INFO_SIZE, /* what is required */
1603 &length); /* where to put the data */
1604
1605 The possible values for the third argument are defined in pcre.h, and
1606 are as follows:
1607
1608 PCRE_INFO_BACKREFMAX
1609
1610 Return the number of the highest back reference in the pattern. The
1611 fourth argument should point to an int variable. Zero is returned if
1612 there are no back references.
1613
1614 PCRE_INFO_CAPTURECOUNT
1615
1616 Return the number of capturing subpatterns in the pattern. The fourth
1617 argument should point to an int variable.
1618
1619 PCRE_INFO_DEFAULT_TABLES
1620
1621 Return a pointer to the internal default character tables within PCRE.
1622 The fourth argument should point to an unsigned char * variable. This
1623 information call is provided for internal use by the pcre_study() func-
1624 tion. External callers can cause PCRE to use its internal tables by
1625 passing a NULL table pointer.
1626
1627 PCRE_INFO_FIRSTBYTE
1628
1629 Return information about the first byte of any matched string, for a
1630 non-anchored pattern. The fourth argument should point to an int vari-
1631 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1632 is still recognized for backwards compatibility.)
1633
1634 If there is a fixed first byte, for example, from a pattern such as
1635 (cat|cow|coyote), its value is returned. Otherwise, if either
1636
1637 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1638 branch starts with "^", or
1639
1640 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1641 set (if it were set, the pattern would be anchored),
1642
1643 -1 is returned, indicating that the pattern matches only at the start
1644 of a subject string or after any newline within the string. Otherwise
1645 -2 is returned. For anchored patterns, -2 is returned.
1646
1647 PCRE_INFO_FIRSTTABLE
1648
1649 If the pattern was studied, and this resulted in the construction of a
1650 256-bit table indicating a fixed set of bytes for the first byte in any
1651 matching string, a pointer to the table is returned. Otherwise NULL is
1652 returned. The fourth argument should point to an unsigned char * vari-
1653 able.
1654
1655 PCRE_INFO_HASCRORLF
1656
1657 Return 1 if the pattern contains any explicit matches for CR or LF
1658 characters, otherwise 0. The fourth argument should point to an int
1659 variable. An explicit match is either a literal CR or LF character, or
1660 \r or \n.
1661
1662 PCRE_INFO_JCHANGED
1663
1664 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1665 otherwise 0. The fourth argument should point to an int variable. (?J)
1666 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1667
1668 PCRE_INFO_LASTLITERAL
1669
1670 Return the value of the rightmost literal byte that must exist in any
1671 matched string, other than at its start, if such a byte has been
1672 recorded. The fourth argument should point to an int variable. If there
1673 is no such byte, -1 is returned. For anchored patterns, a last literal
1674 byte is recorded only if it follows something of variable length. For
1675 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1676 /^a\dz\d/ the returned value is -1.
1677
1678 PCRE_INFO_NAMECOUNT
1679 PCRE_INFO_NAMEENTRYSIZE
1680 PCRE_INFO_NAMETABLE
1681
1682 PCRE supports the use of named as well as numbered capturing parenthe-
1683 ses. The names are just an additional way of identifying the parenthe-
1684 ses, which still acquire numbers. Several convenience functions such as
1685 pcre_get_named_substring() are provided for extracting captured sub-
1686 strings by name. It is also possible to extract the data directly, by
1687 first converting the name to a number in order to access the correct
1688 pointers in the output vector (described with pcre_exec() below). To do
1689 the conversion, you need to use the name-to-number map, which is
1690 described by these three values.
1691
1692 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1693 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1694 of each entry; both of these return an int value. The entry size
1695 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1696 a pointer to the first entry of the table (a pointer to char). The
1697 first two bytes of each entry are the number of the capturing parenthe-
1698 sis, most significant byte first. The rest of the entry is the corre-
1699 sponding name, zero terminated. The names are in alphabetical order.
1700 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1701 theses numbers. For example, consider the following pattern (assume
1702 PCRE_EXTENDED is set, so white space - including newlines - is
1703 ignored):
1704
1705 (?<date> (?<year>(\d\d)?\d\d) -
1706 (?<month>\d\d) - (?<day>\d\d) )
1707
1708 There are four named subpatterns, so the table has four entries, and
1709 each entry in the table is eight bytes long. The table is as follows,
1710 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1711 as ??:
1712
1713 00 01 d a t e 00 ??
1714 00 05 d a y 00 ?? ??
1715 00 04 m o n t h 00
1716 00 02 y e a r 00 ??
1717
1718 When writing code to extract data from named subpatterns using the
1719 name-to-number map, remember that the length of the entries is likely
1720 to be different for each compiled pattern.
1721
1722 PCRE_INFO_OKPARTIAL
1723
1724 Return 1 if the pattern can be used for partial matching, otherwise 0.
1725 The fourth argument should point to an int variable. The pcrepartial
1726 documentation lists the restrictions that apply to patterns when par-
1727 tial matching is used.
1728
1729 PCRE_INFO_OPTIONS
1730
1731 Return a copy of the options with which the pattern was compiled. The
1732 fourth argument should point to an unsigned long int variable. These
1733 option bits are those specified in the call to pcre_compile(), modified
1734 by any top-level option settings at the start of the pattern itself. In
1735 other words, they are the options that will be in force when matching
1736 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1737 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1738 and PCRE_EXTENDED.
1739
1740 A pattern is automatically anchored by PCRE if all of its top-level
1741 alternatives begin with one of the following:
1742
1743 ^ unless PCRE_MULTILINE is set
1744 \A always
1745 \G always
1746 .* if PCRE_DOTALL is set and there are no back
1747 references to the subpattern in which .* appears
1748
1749 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1750 by pcre_fullinfo().
1751
1752 PCRE_INFO_SIZE
1753
1754 Return the size of the compiled pattern, that is, the value that was
1755 passed as the argument to pcre_malloc() when PCRE was getting memory in
1756 which to place the compiled data. The fourth argument should point to a
1757 size_t variable.
1758
1759 PCRE_INFO_STUDYSIZE
1760
1761 Return the size of the data block pointed to by the study_data field in
1762 a pcre_extra block. That is, it is the value that was passed to
1763 pcre_malloc() when PCRE was getting memory into which to place the data
1764 created by pcre_study(). The fourth argument should point to a size_t
1765 variable.
1766
1767
1768 OBSOLETE INFO FUNCTION
1769
1770 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1771
1772 The pcre_info() function is now obsolete because its interface is too
1773 restrictive to return all the available data about a compiled pattern.
1774 New programs should use pcre_fullinfo() instead. The yield of
1775 pcre_info() is the number of capturing subpatterns, or one of the fol-
1776 lowing negative numbers:
1777
1778 PCRE_ERROR_NULL the argument code was NULL
1779 PCRE_ERROR_BADMAGIC the "magic number" was not found
1780
1781 If the optptr argument is not NULL, a copy of the options with which
1782 the pattern was compiled is placed in the integer it points to (see
1783 PCRE_INFO_OPTIONS above).
1784
1785 If the pattern is not anchored and the firstcharptr argument is not
1786 NULL, it is used to pass back information about the first character of
1787 any matched string (see PCRE_INFO_FIRSTBYTE above).
1788
1789
1790 REFERENCE COUNTS
1791
1792 int pcre_refcount(pcre *code, int adjust);
1793
1794 The pcre_refcount() function is used to maintain a reference count in
1795 the data block that contains a compiled pattern. It is provided for the
1796 benefit of applications that operate in an object-oriented manner,
1797 where different parts of the application may be using the same compiled
1798 pattern, but you want to free the block when they are all done.
1799
1800 When a pattern is compiled, the reference count field is initialized to
1801 zero. It is changed only by calling this function, whose action is to
1802 add the adjust value (which may be positive or negative) to it. The
1803 yield of the function is the new value. However, the value of the count
1804 is constrained to lie between 0 and 65535, inclusive. If the new value
1805 is outside these limits, it is forced to the appropriate limit value.
1806
1807 Except when it is zero, the reference count is not correctly preserved
1808 if a pattern is compiled on one host and then transferred to a host
1809 whose byte-order is different. (This seems a highly unlikely scenario.)
1810
1811
1812 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1813
1814 int pcre_exec(const pcre *code, const pcre_extra *extra,
1815 const char *subject, int length, int startoffset,
1816 int options, int *ovector, int ovecsize);
1817
1818 The function pcre_exec() is called to match a subject string against a
1819 compiled pattern, which is passed in the code argument. If the pattern
1820 has been studied, the result of the study should be passed in the extra
1821 argument. This function is the main matching facility of the library,
1822 and it operates in a Perl-like manner. For specialist use there is also
1823 an alternative matching function, which is described below in the sec-
1824 tion about the pcre_dfa_exec() function.
1825
1826 In most applications, the pattern will have been compiled (and option-
1827 ally studied) in the same process that calls pcre_exec(). However, it
1828 is possible to save compiled patterns and study data, and then use them
1829 later in different processes, possibly even on different hosts. For a
1830 discussion about this, see the pcreprecompile documentation.
1831
1832 Here is an example of a simple call to pcre_exec():
1833
1834 int rc;
1835 int ovector[30];
1836 rc = pcre_exec(
1837 re, /* result of pcre_compile() */
1838 NULL, /* we didn't study the pattern */
1839 "some string", /* the subject string */
1840 11, /* the length of the subject string */
1841 0, /* start at offset 0 in the subject */
1842 0, /* default options */
1843 ovector, /* vector of integers for substring information */
1844 30); /* number of elements (NOT size in bytes) */
1845
1846 Extra data for pcre_exec()
1847
1848 If the extra argument is not NULL, it must point to a pcre_extra data
1849 block. The pcre_study() function returns such a block (when it doesn't
1850 return NULL), but you can also create one for yourself, and pass addi-
1851 tional information in it. The pcre_extra block contains the following
1852 fields (not necessarily in this order):
1853
1854 unsigned long int flags;
1855 void *study_data;
1856 unsigned long int match_limit;
1857 unsigned long int match_limit_recursion;
1858 void *callout_data;
1859 const unsigned char *tables;
1860
1861 The flags field is a bitmap that specifies which of the other fields
1862 are set. The flag bits are:
1863
1864 PCRE_EXTRA_STUDY_DATA
1865 PCRE_EXTRA_MATCH_LIMIT
1866 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1867 PCRE_EXTRA_CALLOUT_DATA
1868 PCRE_EXTRA_TABLES
1869
1870 Other flag bits should be set to zero. The study_data field is set in
1871 the pcre_extra block that is returned by pcre_study(), together with
1872 the appropriate flag bit. You should not set this yourself, but you may
1873 add to the block by setting the other fields and their corresponding
1874 flag bits.
1875
1876 The match_limit field provides a means of preventing PCRE from using up
1877 a vast amount of resources when running patterns that are not going to
1878 match, but which have a very large number of possibilities in their
1879 search trees. The classic example is the use of nested unlimited
1880 repeats.
1881
1882 Internally, PCRE uses a function called match() which it calls repeat-
1883 edly (sometimes recursively). The limit set by match_limit is imposed
1884 on the number of times this function is called during a match, which
1885 has the effect of limiting the amount of backtracking that can take
1886 place. For patterns that are not anchored, the count restarts from zero
1887 for each position in the subject string.
1888
1889 The default value for the limit can be set when PCRE is built; the
1890 default default is 10 million, which handles all but the most extreme
1891 cases. You can override the default by suppling pcre_exec() with a
1892 pcre_extra block in which match_limit is set, and
1893 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1894 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1895
1896 The match_limit_recursion field is similar to match_limit, but instead
1897 of limiting the total number of times that match() is called, it limits
1898 the depth of recursion. The recursion depth is a smaller number than
1899 the total number of calls, because not all calls to match() are recur-
1900 sive. This limit is of use only if it is set smaller than match_limit.
1901
1902 Limiting the recursion depth limits the amount of stack that can be
1903 used, or, when PCRE has been compiled to use memory on the heap instead
1904 of the stack, the amount of heap memory that can be used.
1905
1906 The default value for match_limit_recursion can be set when PCRE is
1907 built; the default default is the same value as the default for
1908 match_limit. You can override the default by suppling pcre_exec() with
1909 a pcre_extra block in which match_limit_recursion is set, and
1910 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1911 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1912
1913 The pcre_callout field is used in conjunction with the "callout" fea-
1914 ture, which is described in the pcrecallout documentation.
1915
1916 The tables field is used to pass a character tables pointer to
1917 pcre_exec(); this overrides the value that is stored with the compiled
1918 pattern. A non-NULL value is stored with the compiled pattern only if
1919 custom tables were supplied to pcre_compile() via its tableptr argu-
1920 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1921 PCRE's internal tables to be used. This facility is helpful when re-
1922 using patterns that have been saved after compiling with an external
1923 set of tables, because the external tables might be at a different
1924 address when pcre_exec() is called. See the pcreprecompile documenta-
1925 tion for a discussion of saving compiled patterns for later use.
1926
1927 Option bits for pcre_exec()
1928
1929 The unused bits of the options argument for pcre_exec() must be zero.
1930 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1931 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_START_OPTIMIZE,
1932 PCRE_NO_UTF8_CHECK and PCRE_PARTIAL.
1933
1934 PCRE_ANCHORED
1935
1936 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1937 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1938 turned out to be anchored by virtue of its contents, it cannot be made
1939 unachored at matching time.
1940
1941 PCRE_BSR_ANYCRLF
1942 PCRE_BSR_UNICODE
1943
1944 These options (which are mutually exclusive) control what the \R escape
1945 sequence matches. The choice is either to match only CR, LF, or CRLF,
1946 or to match any Unicode newline sequence. These options override the
1947 choice that was made or defaulted when the pattern was compiled.
1948
1949 PCRE_NEWLINE_CR
1950 PCRE_NEWLINE_LF
1951 PCRE_NEWLINE_CRLF
1952 PCRE_NEWLINE_ANYCRLF
1953 PCRE_NEWLINE_ANY
1954
1955 These options override the newline definition that was chosen or
1956 defaulted when the pattern was compiled. For details, see the descrip-
1957 tion of pcre_compile() above. During matching, the newline choice
1958 affects the behaviour of the dot, circumflex, and dollar metacharac-
1959 ters. It may also alter the way the match position is advanced after a
1960 match failure for an unanchored pattern.
1961
1962 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1963 set, and a match attempt for an unanchored pattern fails when the cur-
1964 rent position is at a CRLF sequence, and the pattern contains no
1965 explicit matches for CR or LF characters, the match position is
1966 advanced by two characters instead of one, in other words, to after the
1967 CRLF.
1968
1969 The above rule is a compromise that makes the most common cases work as
1970 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1971 option is not set), it does not match the string "\r\nA" because, after
1972 failing at the start, it skips both the CR and the LF before retrying.
1973 However, the pattern [\r\n]A does match that string, because it con-
1974 tains an explicit CR or LF reference, and so advances only by one char-
1975 acter after the first failure.
1976
1977 An explicit match for CR of LF is either a literal appearance of one of
1978 those characters, or one of the \r or \n escape sequences. Implicit
1979 matches such as [^X] do not count, nor does \s (which includes CR and
1980 LF in the characters that it matches).
1981
1982 Notwithstanding the above, anomalous effects may still occur when CRLF
1983 is a valid newline sequence and explicit \r or \n escapes appear in the
1984 pattern.
1985
1986 PCRE_NOTBOL
1987
1988 This option specifies that first character of the subject string is not
1989 the beginning of a line, so the circumflex metacharacter should not
1990 match before it. Setting this without PCRE_MULTILINE (at compile time)
1991 causes circumflex never to match. This option affects only the behav-
1992 iour of the circumflex metacharacter. It does not affect \A.
1993
1994 PCRE_NOTEOL
1995
1996 This option specifies that the end of the subject string is not the end
1997 of a line, so the dollar metacharacter should not match it nor (except
1998 in multiline mode) a newline immediately before it. Setting this with-
1999 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2000 option affects only the behaviour of the dollar metacharacter. It does
2001 not affect \Z or \z.
2002
2003 PCRE_NOTEMPTY
2004
2005 An empty string is not considered to be a valid match if this option is
2006 set. If there are alternatives in the pattern, they are tried. If all
2007 the alternatives match the empty string, the entire match fails. For
2008 example, if the pattern
2009
2010 a?b?
2011
2012 is applied to a string not beginning with "a" or "b", it matches the
2013 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2014 match is not valid, so PCRE searches further into the string for occur-
2015 rences of "a" or "b".
2016
2017 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
2018 cial case of a pattern match of the empty string within its split()
2019 function, and when using the /g modifier. It is possible to emulate
2020 Perl's behaviour after matching a null string by first trying the match
2021 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
2022 if that fails by advancing the starting offset (see below) and trying
2023 an ordinary match again. There is some code that demonstrates how to do
2024 this in the pcredemo.c sample program.
2025
2026 PCRE_NO_START_OPTIMIZE
2027
2028 There are a number of optimizations that pcre_exec() uses at the start
2029 of a match, in order to speed up the process. For example, if it is
2030 known that a match must start with a specific character, it searches
2031 the subject for that character, and fails immediately if it cannot find
2032 it, without actually running the main matching function. When callouts
2033 are in use, these optimizations can cause them to be skipped. This
2034 option disables the "start-up" optimizations, causing performance to
2035 suffer, but ensuring that the callouts do occur.
2036
2037 PCRE_NO_UTF8_CHECK
2038
2039 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2040 UTF-8 string is automatically checked when pcre_exec() is subsequently
2041 called. The value of startoffset is also checked to ensure that it
2042 points to the start of a UTF-8 character. There is a discussion about
2043 the validity of UTF-8 strings in the section on UTF-8 support in the
2044 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2045 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
2046 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
2047
2048 If you already know that your subject is valid, and you want to skip
2049 these checks for performance reasons, you can set the
2050 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2051 do this for the second and subsequent calls to pcre_exec() if you are
2052 making repeated calls to find all the matches in a single subject
2053 string. However, you should be sure that the value of startoffset
2054 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
2055 set, the effect of passing an invalid UTF-8 string as a subject, or a
2056 value of startoffset that does not point to the start of a UTF-8 char-
2057 acter, is undefined. Your program may crash.
2058
2059 PCRE_PARTIAL
2060
2061 This option turns on the partial matching feature. If the subject
2062 string fails to match the pattern, but at some point during the match-
2063 ing process the end of the subject was reached (that is, the subject
2064 partially matches the pattern and the failure to match occurred only
2065 because there were not enough subject characters), pcre_exec() returns
2066 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
2067 used, there are restrictions on what may appear in the pattern. These
2068 are discussed in the pcrepartial documentation.
2069
2070 The string to be matched by pcre_exec()
2071
2072 The subject string is passed to pcre_exec() as a pointer in subject, a
2073 length (in bytes) in length, and a starting byte offset in startoffset.
2074 In UTF-8 mode, the byte offset must point to the start of a UTF-8 char-
2075 acter. Unlike the pattern string, the subject may contain binary zero
2076 bytes. When the starting offset is zero, the search for a match starts
2077 at the beginning of the subject, and this is by far the most common
2078 case.
2079
2080 A non-zero starting offset is useful when searching for another match
2081 in the same subject by calling pcre_exec() again after a previous suc-
2082 cess. Setting startoffset differs from just passing over a shortened
2083 string and setting PCRE_NOTBOL in the case of a pattern that begins
2084 with any kind of lookbehind. For example, consider the pattern
2085
2086 \Biss\B
2087
2088 which finds occurrences of "iss" in the middle of words. (\B matches
2089 only if the current position in the subject is not a word boundary.)
2090 When applied to the string "Mississipi" the first call to pcre_exec()
2091 finds the first occurrence. If pcre_exec() is called again with just
2092 the remainder of the subject, namely "issipi", it does not match,
2093 because \B is always false at the start of the subject, which is deemed
2094 to be a word boundary. However, if pcre_exec() is passed the entire
2095 string again, but with startoffset set to 4, it finds the second occur-
2096 rence of "iss" because it is able to look behind the starting point to
2097 discover that it is preceded by a letter.
2098
2099 If a non-zero starting offset is passed when the pattern is anchored,
2100 one attempt to match at the given offset is made. This can only succeed
2101 if the pattern does not require the match to be at the start of the
2102 subject.
2103
2104 How pcre_exec() returns captured substrings
2105
2106 In general, a pattern matches a certain portion of the subject, and in
2107 addition, further substrings from the subject may be picked out by
2108 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2109 this is called "capturing" in what follows, and the phrase "capturing
2110 subpattern" is used for a fragment of a pattern that picks out a sub-
2111 string. PCRE supports several other kinds of parenthesized subpattern
2112 that do not cause substrings to be captured.
2113
2114 Captured substrings are returned to the caller via a vector of integers
2115 whose address is passed in ovector. The number of elements in the vec-
2116 tor is passed in ovecsize, which must be a non-negative number. Note:
2117 this argument is NOT the size of ovector in bytes.
2118
2119 The first two-thirds of the vector is used to pass back captured sub-
2120 strings, each substring using a pair of integers. The remaining third
2121 of the vector is used as workspace by pcre_exec() while matching cap-
2122 turing subpatterns, and is not available for passing back information.
2123 The number passed in ovecsize should always be a multiple of three. If
2124 it is not, it is rounded down.
2125
2126 When a match is successful, information about captured substrings is
2127 returned in pairs of integers, starting at the beginning of ovector,
2128 and continuing up to two-thirds of its length at the most. The first
2129 element of each pair is set to the byte offset of the first character
2130 in a substring, and the second is set to the byte offset of the first
2131 character after the end of a substring. Note: these values are always
2132 byte offsets, even in UTF-8 mode. They are not character counts.
2133
2134 The first pair of integers, ovector[0] and ovector[1], identify the
2135 portion of the subject string matched by the entire pattern. The next
2136 pair is used for the first capturing subpattern, and so on. The value
2137 returned by pcre_exec() is one more than the highest numbered pair that
2138 has been set. For example, if two substrings have been captured, the
2139 returned value is 3. If there are no capturing subpatterns, the return
2140 value from a successful match is 1, indicating that just the first pair
2141 of offsets has been set.
2142
2143 If a capturing subpattern is matched repeatedly, it is the last portion
2144 of the string that it matched that is returned.
2145
2146 If the vector is too small to hold all the captured substring offsets,
2147 it is used as far as possible (up to two-thirds of its length), and the
2148 function returns a value of zero. If the substring offsets are not of
2149 interest, pcre_exec() may be called with ovector passed as NULL and
2150 ovecsize as zero. However, if the pattern contains back references and
2151 the ovector is not big enough to remember the related substrings, PCRE
2152 has to get additional memory for use during matching. Thus it is usu-
2153 ally advisable to supply an ovector.
2154
2155 The pcre_info() function can be used to find out how many capturing
2156 subpatterns there are in a compiled pattern. The smallest size for
2157 ovector that will allow for n captured substrings, in addition to the
2158 offsets of the substring matched by the whole pattern, is (n+1)*3.
2159
2160 It is possible for capturing subpattern number n+1 to match some part
2161 of the subject when subpattern n has not been used at all. For example,
2162 if the string "abc" is matched against the pattern (a|(z))(bc) the
2163 return from the function is 4, and subpatterns 1 and 3 are matched, but
2164 2 is not. When this happens, both values in the offset pairs corre-
2165 sponding to unused subpatterns are set to -1.
2166
2167 Offset values that correspond to unused subpatterns at the end of the
2168 expression are also set to -1. For example, if the string "abc" is
2169 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2170 matched. The return from the function is 2, because the highest used
2171 capturing subpattern number is 1. However, you can refer to the offsets
2172 for the second and third capturing subpatterns if you wish (assuming
2173 the vector is large enough, of course).
2174
2175 Some convenience functions are provided for extracting the captured
2176 substrings as separate strings. These are described below.
2177
2178 Error return values from pcre_exec()
2179
2180 If pcre_exec() fails, it returns a negative number. The following are
2181 defined in the header file:
2182
2183 PCRE_ERROR_NOMATCH (-1)
2184
2185 The subject string did not match the pattern.
2186
2187 PCRE_ERROR_NULL (-2)
2188
2189 Either code or subject was passed as NULL, or ovector was NULL and
2190 ovecsize was not zero.
2191
2192 PCRE_ERROR_BADOPTION (-3)
2193
2194 An unrecognized bit was set in the options argument.
2195
2196 PCRE_ERROR_BADMAGIC (-4)
2197
2198 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2199 to catch the case when it is passed a junk pointer and to detect when a
2200 pattern that was compiled in an environment of one endianness is run in
2201 an environment with the other endianness. This is the error that PCRE
2202 gives when the magic number is not present.
2203
2204 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2205
2206 While running the pattern match, an unknown item was encountered in the
2207 compiled pattern. This error could be caused by a bug in PCRE or by
2208 overwriting of the compiled pattern.
2209
2210 PCRE_ERROR_NOMEMORY (-6)
2211
2212 If a pattern contains back references, but the ovector that is passed
2213 to pcre_exec() is not big enough to remember the referenced substrings,
2214 PCRE gets a block of memory at the start of matching to use for this
2215 purpose. If the call via pcre_malloc() fails, this error is given. The
2216 memory is automatically freed at the end of matching.
2217
2218 PCRE_ERROR_NOSUBSTRING (-7)
2219
2220 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2221 and pcre_get_substring_list() functions (see below). It is never
2222 returned by pcre_exec().
2223
2224 PCRE_ERROR_MATCHLIMIT (-8)
2225
2226 The backtracking limit, as specified by the match_limit field in a
2227 pcre_extra structure (or defaulted) was reached. See the description
2228 above.
2229
2230 PCRE_ERROR_CALLOUT (-9)
2231
2232 This error is never generated by pcre_exec() itself. It is provided for
2233 use by callout functions that want to yield a distinctive error code.
2234 See the pcrecallout documentation for details.
2235
2236 PCRE_ERROR_BADUTF8 (-10)
2237
2238 A string that contains an invalid UTF-8 byte sequence was passed as a
2239 subject.
2240
2241 PCRE_ERROR_BADUTF8_OFFSET (-11)
2242
2243 The UTF-8 byte sequence that was passed as a subject was valid, but the
2244 value of startoffset did not point to the beginning of a UTF-8 charac-
2245 ter.
2246
2247 PCRE_ERROR_PARTIAL (-12)
2248
2249 The subject string did not match, but it did match partially. See the
2250 pcrepartial documentation for details of partial matching.
2251
2252 PCRE_ERROR_BADPARTIAL (-13)
2253
2254 The PCRE_PARTIAL option was used with a compiled pattern containing
2255 items that are not supported for partial matching. See the pcrepartial
2256 documentation for details of partial matching.
2257
2258 PCRE_ERROR_INTERNAL (-14)
2259
2260 An unexpected internal error has occurred. This error could be caused
2261 by a bug in PCRE or by overwriting of the compiled pattern.
2262
2263 PCRE_ERROR_BADCOUNT (-15)
2264
2265 This error is given if the value of the ovecsize argument is negative.
2266
2267 PCRE_ERROR_RECURSIONLIMIT (-21)
2268
2269 The internal recursion limit, as specified by the match_limit_recursion
2270 field in a pcre_extra structure (or defaulted) was reached. See the
2271 description above.
2272
2273 PCRE_ERROR_BADNEWLINE (-23)
2274
2275 An invalid combination of PCRE_NEWLINE_xxx options was given.
2276
2277 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2278
2279
2280 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2281
2282 int pcre_copy_substring(const char *subject, int *ovector,
2283 int stringcount, int stringnumber, char *buffer,
2284 int buffersize);
2285
2286 int pcre_get_substring(const char *subject, int *ovector,
2287 int stringcount, int stringnumber,
2288 const char **stringptr);
2289
2290 int pcre_get_substring_list(const char *subject,
2291 int *ovector, int stringcount, const char ***listptr);
2292
2293 Captured substrings can be accessed directly by using the offsets
2294 returned by pcre_exec() in ovector. For convenience, the functions
2295 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2296 string_list() are provided for extracting captured substrings as new,
2297 separate, zero-terminated strings. These functions identify substrings
2298 by number. The next section describes functions for extracting named
2299 substrings.
2300
2301 A substring that contains a binary zero is correctly extracted and has
2302 a further zero added on the end, but the result is not, of course, a C
2303 string. However, you can process such a string by referring to the
2304 length that is returned by pcre_copy_substring() and pcre_get_sub-
2305 string(). Unfortunately, the interface to pcre_get_substring_list() is
2306 not adequate for handling strings containing binary zeros, because the
2307 end of the final string is not independently indicated.
2308
2309 The first three arguments are the same for all three of these func-
2310 tions: subject is the subject string that has just been successfully
2311 matched, ovector is a pointer to the vector of integer offsets that was
2312 passed to pcre_exec(), and stringcount is the number of substrings that
2313 were captured by the match, including the substring that matched the
2314 entire regular expression. This is the value returned by pcre_exec() if
2315 it is greater than zero. If pcre_exec() returned zero, indicating that
2316 it ran out of space in ovector, the value passed as stringcount should
2317 be the number of elements in the vector divided by three.
2318
2319 The functions pcre_copy_substring() and pcre_get_substring() extract a
2320 single substring, whose number is given as stringnumber. A value of
2321 zero extracts the substring that matched the entire pattern, whereas
2322 higher values extract the captured substrings. For pcre_copy_sub-
2323 string(), the string is placed in buffer, whose length is given by
2324 buffersize, while for pcre_get_substring() a new block of memory is
2325 obtained via pcre_malloc, and its address is returned via stringptr.
2326 The yield of the function is the length of the string, not including
2327 the terminating zero, or one of these error codes:
2328
2329 PCRE_ERROR_NOMEMORY (-6)
2330
2331 The buffer was too small for pcre_copy_substring(), or the attempt to
2332 get memory failed for pcre_get_substring().
2333
2334 PCRE_ERROR_NOSUBSTRING (-7)
2335
2336 There is no substring whose number is stringnumber.
2337
2338 The pcre_get_substring_list() function extracts all available sub-
2339 strings and builds a list of pointers to them. All this is done in a
2340 single block of memory that is obtained via pcre_malloc. The address of
2341 the memory block is returned via listptr, which is also the start of
2342 the list of string pointers. The end of the list is marked by a NULL
2343 pointer. The yield of the function is zero if all went well, or the
2344 error code
2345
2346 PCRE_ERROR_NOMEMORY (-6)
2347
2348 if the attempt to get the memory block failed.
2349
2350 When any of these functions encounter a substring that is unset, which
2351 can happen when capturing subpattern number n+1 matches some part of
2352 the subject, but subpattern n has not been used at all, they return an
2353 empty string. This can be distinguished from a genuine zero-length sub-
2354 string by inspecting the appropriate offset in ovector, which is nega-
2355 tive for unset substrings.
2356
2357 The two convenience functions pcre_free_substring() and pcre_free_sub-
2358 string_list() can be used to free the memory returned by a previous
2359 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2360 tively. They do nothing more than call the function pointed to by
2361 pcre_free, which of course could be called directly from a C program.
2362 However, PCRE is used in some situations where it is linked via a spe-
2363 cial interface to another programming language that cannot use
2364 pcre_free directly; it is for these cases that the functions are pro-
2365 vided.
2366
2367
2368 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2369
2370 int pcre_get_stringnumber(const pcre *code,
2371 const char *name);
2372
2373 int pcre_copy_named_substring(const pcre *code,
2374 const char *subject, int *ovector,
2375 int stringcount, const char *stringname,
2376 char *buffer, int buffersize);
2377
2378 int pcre_get_named_substring(const pcre *code,
2379 const char *subject, int *ovector,
2380 int stringcount, const char *stringname,
2381 const char **stringptr);
2382
2383 To extract a substring by name, you first have to find associated num-
2384 ber. For example, for this pattern
2385
2386 (a+)b(?<xxx>\d+)...
2387
2388 the number of the subpattern called "xxx" is 2. If the name is known to
2389 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2390 name by calling pcre_get_stringnumber(). The first argument is the com-
2391 piled pattern, and the second is the name. The yield of the function is
2392 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2393 subpattern of that name.
2394
2395 Given the number, you can extract the substring directly, or use one of
2396 the functions described in the previous section. For convenience, there
2397 are also two functions that do the whole job.
2398
2399 Most of the arguments of pcre_copy_named_substring() and
2400 pcre_get_named_substring() are the same as those for the similarly
2401 named functions that extract by number. As these are described in the
2402 previous section, they are not re-described here. There are just two
2403 differences:
2404
2405 First, instead of a substring number, a substring name is given. Sec-
2406 ond, there is an extra argument, given at the start, which is a pointer
2407 to the compiled pattern. This is needed in order to gain access to the
2408 name-to-number translation table.
2409
2410 These functions call pcre_get_stringnumber(), and if it succeeds, they
2411 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2412 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2413 behaviour may not be what you want (see the next section).
2414
2415 Warning: If the pattern uses the "(?|" feature to set up multiple sub-
2416 patterns with the same number, you cannot use names to distinguish
2417 them, because names are not included in the compiled code. The matching
2418 process uses only numbers.
2419
2420
2421 DUPLICATE SUBPATTERN NAMES
2422
2423 int pcre_get_stringtable_entries(const pcre *code,
2424 const char *name, char **first, char **last);
2425
2426 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2427 subpatterns are not required to be unique. Normally, patterns with
2428 duplicate names are such that in any one match, only one of the named
2429 subpatterns participates. An example is shown in the pcrepattern docu-
2430 mentation.
2431
2432 When duplicates are present, pcre_copy_named_substring() and
2433 pcre_get_named_substring() return the first substring corresponding to
2434 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2435 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2436 function returns one of the numbers that are associated with the name,
2437 but it is not defined which it is.
2438
2439 If you want to get full details of all captured substrings for a given
2440 name, you must use the pcre_get_stringtable_entries() function. The
2441 first argument is the compiled pattern, and the second is the name. The
2442 third and fourth are pointers to variables which are updated by the
2443 function. After it has run, they point to the first and last entries in
2444 the name-to-number table for the given name. The function itself
2445 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2446 there are none. The format of the table is described above in the sec-
2447 tion entitled Information about a pattern. Given all the relevant
2448 entries for the name, you can extract each of their numbers, and hence
2449 the captured data, if any.
2450
2451
2452 FINDING ALL POSSIBLE MATCHES
2453
2454 The traditional matching function uses a similar algorithm to Perl,
2455 which stops when it finds the first match, starting at a given point in
2456 the subject. If you want to find all possible matches, or the longest
2457 possible match, consider using the alternative matching function (see
2458 below) instead. If you cannot use the alternative function, but still
2459 need to find all possible matches, you can kludge it up by making use
2460 of the callout facility, which is described in the pcrecallout documen-
2461 tation.
2462
2463 What you have to do is to insert a callout right at the end of the pat-
2464 tern. When your callout function is called, extract and save the cur-
2465 rent matched substring. Then return 1, which forces pcre_exec() to
2466 backtrack and try other alternatives. Ultimately, when it runs out of
2467 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2468
2469
2470 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2471
2472 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2473 const char *subject, int length, int startoffset,
2474 int options, int *ovector, int ovecsize,
2475 int *workspace, int wscount);
2476
2477 The function pcre_dfa_exec() is called to match a subject string
2478 against a compiled pattern, using a matching algorithm that scans the
2479 subject string just once, and does not backtrack. This has different
2480 characteristics to the normal algorithm, and is not compatible with
2481 Perl. Some of the features of PCRE patterns are not supported. Never-
2482 theless, there are times when this kind of matching can be useful. For
2483 a discussion of the two matching algorithms, see the pcrematching docu-
2484 mentation.
2485
2486 The arguments for the pcre_dfa_exec() function are the same as for
2487 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2488 ent way, and this is described below. The other common arguments are
2489 used in the same way as for pcre_exec(), so their description is not
2490 repeated here.
2491
2492 The two additional arguments provide workspace for the function. The
2493 workspace vector should contain at least 20 elements. It is used for
2494 keeping track of multiple paths through the pattern tree. More
2495 workspace will be needed for patterns and subjects where there are a
2496 lot of potential matches.
2497
2498 Here is an example of a simple call to pcre_dfa_exec():
2499
2500 int rc;
2501 int ovector[10];
2502 int wspace[20];
2503 rc = pcre_dfa_exec(
2504 re, /* result of pcre_compile() */
2505 NULL, /* we didn't study the pattern */
2506 "some string", /* the subject string */
2507 11, /* the length of the subject string */
2508 0, /* start at offset 0 in the subject */
2509 0, /* default options */
2510 ovector, /* vector of integers for substring information */
2511 10, /* number of elements (NOT size in bytes) */
2512 wspace, /* working space vector */
2513 20); /* number of elements (NOT size in bytes) */
2514
2515 Option bits for pcre_dfa_exec()
2516
2517 The unused bits of the options argument for pcre_dfa_exec() must be
2518 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2519 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2520 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2521 three of these are the same as for pcre_exec(), so their description is
2522 not repeated here.
2523
2524 PCRE_PARTIAL
2525
2526 This has the same general effect as it does for pcre_exec(), but the
2527 details are slightly different. When PCRE_PARTIAL is set for
2528 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2529 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2530 been no complete matches, but there is still at least one matching pos-
2531 sibility. The portion of the string that provided the partial match is
2532 set as the first matching string.
2533
2534 PCRE_DFA_SHORTEST
2535
2536 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2537 stop as soon as it has found one match. Because of the way the alterna-
2538 tive algorithm works, this is necessarily the shortest possible match
2539 at the first possible matching point in the subject string.
2540
2541 PCRE_DFA_RESTART
2542
2543 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2544 returns a partial match, it is possible to call it again, with addi-
2545 tional subject characters, and have it continue with the same match.
2546 The PCRE_DFA_RESTART option requests this action; when it is set, the
2547 workspace and wscount options must reference the same vector as before
2548 because data about the match so far is left in them after a partial
2549 match. There is more discussion of this facility in the pcrepartial
2550 documentation.
2551
2552 Successful returns from pcre_dfa_exec()
2553
2554 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2555 string in the subject. Note, however, that all the matches from one run
2556 of the function start at the same point in the subject. The shorter
2557 matches are all initial substrings of the longer matches. For example,
2558 if the pattern
2559
2560 <.*>
2561
2562 is matched against the string
2563
2564 This is <something> <something else> <something further> no more
2565
2566 the three matched strings are
2567
2568 <something>
2569 <something> <something else>
2570 <something> <something else> <something further>
2571
2572 On success, the yield of the function is a number greater than zero,
2573 which is the number of matched substrings. The substrings themselves
2574 are returned in ovector. Each string uses two elements; the first is
2575 the offset to the start, and the second is the offset to the end. In
2576 fact, all the strings have the same start offset. (Space could have
2577 been saved by giving this only once, but it was decided to retain some
2578 compatibility with the way pcre_exec() returns data, even though the
2579 meaning of the strings is different.)
2580
2581 The strings are returned in reverse order of length; that is, the long-
2582 est matching string is given first. If there were too many matches to
2583 fit into ovector, the yield of the function is zero, and the vector is
2584 filled with the longest matches.
2585
2586 Error returns from pcre_dfa_exec()
2587
2588 The pcre_dfa_exec() function returns a negative number when it fails.
2589 Many of the errors are the same as for pcre_exec(), and these are
2590 described above. There are in addition the following errors that are
2591 specific to pcre_dfa_exec():
2592
2593 PCRE_ERROR_DFA_UITEM (-16)
2594
2595 This return is given if pcre_dfa_exec() encounters an item in the pat-
2596 tern that it does not support, for instance, the use of \C or a back
2597 reference.
2598
2599 PCRE_ERROR_DFA_UCOND (-17)
2600
2601 This return is given if pcre_dfa_exec() encounters a condition item
2602 that uses a back reference for the condition, or a test for recursion
2603 in a specific group. These are not supported.
2604
2605 PCRE_ERROR_DFA_UMLIMIT (-18)
2606
2607 This return is given if pcre_dfa_exec() is called with an extra block
2608 that contains a setting of the match_limit field. This is not supported
2609 (it is meaningless).
2610
2611 PCRE_ERROR_DFA_WSSIZE (-19)
2612
2613 This return is given if pcre_dfa_exec() runs out of space in the
2614 workspace vector.
2615
2616 PCRE_ERROR_DFA_RECURSE (-20)
2617
2618 When a recursive subpattern is processed, the matching function calls
2619 itself recursively, using private vectors for ovector and workspace.
2620 This error is given if the output vector is not large enough. This
2621 should be extremely rare, as a vector of size 1000 is used.
2622
2623
2624 SEE ALSO
2625
2626 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2627 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2628
2629
2630 AUTHOR
2631
2632 Philip Hazel
2633 University Computing Service
2634 Cambridge CB2 3QH, England.
2635
2636
2637 REVISION
2638
2639 Last updated: 11 April 2009
2640 Copyright (c) 1997-2009 University of Cambridge.
2641 ------------------------------------------------------------------------------
2642
2643
2644 PCRECALLOUT(3) PCRECALLOUT(3)
2645
2646
2647 NAME
2648 PCRE - Perl-compatible regular expressions
2649
2650
2651 PCRE CALLOUTS
2652
2653 int (*pcre_callout)(pcre_callout_block *);
2654
2655 PCRE provides a feature called "callout", which is a means of temporar-
2656 ily passing control to the caller of PCRE in the middle of pattern
2657 matching. The caller of PCRE provides an external function by putting
2658 its entry point in the global variable pcre_callout. By default, this
2659 variable contains NULL, which disables all calling out.
2660
2661 Within a regular expression, (?C) indicates the points at which the
2662 external function is to be called. Different callout points can be
2663 identified by putting a number less than 256 after the letter C. The
2664 default value is zero. For example, this pattern has two callout
2665 points:
2666
2667 (?C1)abc(?C2)def
2668
2669 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2670 called, PCRE automatically inserts callouts, all with number 255,
2671 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2672 used with the pattern
2673
2674 A(\d{2}|--)
2675
2676 it is processed as if it were
2677
2678 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2679
2680 Notice that there is a callout before and after each parenthesis and
2681 alternation bar. Automatic callouts can be used for tracking the
2682 progress of pattern matching. The pcretest command has an option that
2683 sets automatic callouts; when it is used, the output indicates how the
2684 pattern is matched. This is useful information when you are trying to
2685 optimize the performance of a particular pattern.
2686
2687
2688 MISSING CALLOUTS
2689
2690 You should be aware that, because of optimizations in the way PCRE
2691 matches patterns by default, callouts sometimes do not happen. For
2692 example, if the pattern is
2693
2694 ab(?C4)cd
2695
2696 PCRE knows that any matching string must contain the letter "d". If the
2697 subject string is "abyz", the lack of "d" means that matching doesn't
2698 ever start, and the callout is never reached. However, with "abyd",
2699 though the result is still no match, the callout is obeyed.
2700
2701 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2702 MIZE option to pcre_exec() or pcre_dfa_exec(). This slows down the
2703 matching process, but does ensure that callouts such as the example
2704 above are obeyed.
2705
2706
2707 THE CALLOUT INTERFACE
2708
2709 During matching, when PCRE reaches a callout point, the external func-
2710 tion defined by pcre_callout is called (if it is set). This applies to
2711 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2712 only argument to the callout function is a pointer to a pcre_callout
2713 block. This structure contains the following fields:
2714
2715 int version;
2716 int callout_number;
2717 int *offset_vector;
2718 const char *subject;
2719 int subject_length;
2720 int start_match;
2721 int current_position;
2722 int capture_top;
2723 int capture_last;
2724 void *callout_data;
2725 int pattern_position;
2726 int next_item_length;
2727
2728 The version field is an integer containing the version number of the
2729 block format. The initial version was 0; the current version is 1. The
2730 version number will change again in future if additional fields are
2731 added, but the intention is never to remove any of the existing fields.
2732
2733 The callout_number field contains the number of the callout, as com-
2734 piled into the pattern (that is, the number after ?C for manual call-
2735 outs, and 255 for automatically generated callouts).
2736
2737 The offset_vector field is a pointer to the vector of offsets that was
2738 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2739 pcre_exec() is used, the contents can be inspected in order to extract
2740 substrings that have been matched so far, in the same way as for
2741 extracting substrings after a match has completed. For pcre_dfa_exec()
2742 this field is not useful.
2743
2744 The subject and subject_length fields contain copies of the values that
2745 were passed to pcre_exec().
2746
2747 The start_match field normally contains the offset within the subject
2748 at which the current match attempt started. However, if the escape
2749 sequence \K has been encountered, this value is changed to reflect the
2750 modified starting point. If the pattern is not anchored, the callout
2751 function may be called several times from the same point in the pattern
2752 for different starting points in the subject.
2753
2754 The current_position field contains the offset within the subject of
2755 the current match pointer.
2756
2757 When the pcre_exec() function is used, the capture_top field contains
2758 one more than the number of the highest numbered captured substring so
2759 far. If no substrings have been captured, the value of capture_top is
2760 one. This is always the case when pcre_dfa_exec() is used, because it
2761 does not support captured substrings.
2762
2763 The capture_last field contains the number of the most recently cap-
2764 tured substring. If no substrings have been captured, its value is -1.
2765 This is always the case when pcre_dfa_exec() is used.
2766
2767 The callout_data field contains a value that is passed to pcre_exec()
2768 or pcre_dfa_exec() specifically so that it can be passed back in call-
2769 outs. It is passed in the pcre_callout field of the pcre_extra data
2770 structure. If no such data was passed, the value of callout_data in a
2771 pcre_callout block is NULL. There is a description of the pcre_extra
2772 structure in the pcreapi documentation.
2773
2774 The pattern_position field is present from version 1 of the pcre_call-
2775 out structure. It contains the offset to the next item to be matched in
2776 the pattern string.
2777
2778 The next_item_length field is present from version 1 of the pcre_call-
2779 out structure. It contains the length of the next item to be matched in
2780 the pattern string. When the callout immediately precedes an alterna-
2781 tion bar, a closing parenthesis, or the end of the pattern, the length
2782 is zero. When the callout precedes an opening parenthesis, the length
2783 is that of the entire subpattern.
2784
2785 The pattern_position and next_item_length fields are intended to help
2786 in distinguishing between different automatic callouts, which all have
2787 the same callout number. However, they are set for all callouts.
2788
2789
2790 RETURN VALUES
2791
2792 The external callout function returns an integer to PCRE. If the value
2793 is zero, matching proceeds as normal. If the value is greater than
2794 zero, matching fails at the current point, but the testing of other
2795 matching possibilities goes ahead, just as if a lookahead assertion had
2796 failed. If the value is less than zero, the match is abandoned, and
2797 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2798
2799 Negative values should normally be chosen from the set of
2800 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2801 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2802 reserved for use by callout functions; it will never be used by PCRE
2803 itself.
2804
2805
2806 AUTHOR
2807
2808 Philip Hazel
2809 University Computing Service
2810 Cambridge CB2 3QH, England.
2811
2812
2813 REVISION
2814
2815 Last updated: 15 March 2009
2816 Copyright (c) 1997-2009 University of Cambridge.
2817 ------------------------------------------------------------------------------
2818
2819
2820 PCRECOMPAT(3) PCRECOMPAT(3)
2821
2822
2823 NAME
2824 PCRE - Perl-compatible regular expressions
2825
2826
2827 DIFFERENCES BETWEEN PCRE AND PERL
2828
2829 This document describes the differences in the ways that PCRE and Perl
2830 handle regular expressions. The differences described here are mainly
2831 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2832 some features that are expected to be in the forthcoming Perl 5.10.
2833
2834 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2835 of what it does have are given in the section on UTF-8 support in the
2836 main pcre page.
2837
2838 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2839 permits them, but they do not mean what you might think. For example,
2840 (?!a){3} does not assert that the next three characters are not "a". It
2841 just asserts that the next character is not "a" three times.
2842
2843 3. Capturing subpatterns that occur inside negative lookahead asser-
2844 tions are counted, but their entries in the offsets vector are never
2845 set. Perl sets its numerical variables from any such patterns that are
2846 matched before the assertion fails to match something (thereby succeed-
2847 ing), but only if the negative lookahead assertion contains just one
2848 branch.
2849
2850 4. Though binary zero characters are supported in the subject string,
2851 they are not allowed in a pattern string because it is passed as a nor-
2852 mal C string, terminated by zero. The escape sequence \0 can be used in
2853 the pattern to represent a binary zero.
2854
2855 5. The following Perl escape sequences are not supported: \l, \u, \L,
2856 \U, and \N. In fact these are implemented by Perl's general string-han-
2857 dling and are not part of its pattern matching engine. If any of these
2858 are encountered by PCRE, an error is generated.
2859
2860 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2861 is built with Unicode character property support. The properties that
2862 can be tested with \p and \P are limited to the general category prop-
2863 erties such as Lu and Nd, script names such as Greek or Han, and the
2864 derived properties Any and L&.
2865
2866 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2867 ters in between are treated as literals. This is slightly different
2868 from Perl in that $ and @ are also handled as literals inside the
2869 quotes. In Perl, they cause variable interpolation (but of course PCRE
2870 does not have variables). Note the following examples:
2871
2872 Pattern PCRE matches Perl matches
2873
2874 \Qabc$xyz\E abc$xyz abc followed by the
2875 contents of $xyz
2876 \Qabc\$xyz\E abc\$xyz abc\$xyz
2877 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2878
2879 The \Q...\E sequence is recognized both inside and outside character
2880 classes.
2881
2882 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2883 constructions. However, there is support for recursive patterns. This
2884 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2885 "callout" feature allows an external function to be called during pat-
2886 tern matching. See the pcrecallout documentation for details.
2887
2888 9. Subpatterns that are called recursively or as "subroutines" are
2889 always treated as atomic groups in PCRE. This is like Python, but
2890 unlike Perl.
2891
2892 10. There are some differences that are concerned with the settings of
2893 captured strings when part of a pattern is repeated. For example,
2894 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2895 unset, but in PCRE it is set to "b".
2896
2897 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2898 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2899 the forms without an argument. PCRE does not support (*MARK). If
2900 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2901 ture group; this is different to Perl.
2902
2903 12. PCRE provides some extensions to the Perl regular expression facil-
2904 ities. Perl 5.10 will include new features that are not in earlier
2905 versions, some of which (such as named parentheses) have been in PCRE
2906 for some time. This list is with respect to Perl 5.10:
2907
2908 (a) Although lookbehind assertions must match fixed length strings,
2909 each alternative branch of a lookbehind assertion can match a different
2910 length of string. Perl requires them all to have the same length.
2911
2912 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2913 meta-character matches only at the very end of the string.
2914
2915 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2916 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2917 ignored. (Perl can be made to issue a warning.)
2918
2919 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2920 fiers is inverted, that is, by default they are not greedy, but if fol-
2921 lowed by a question mark they are.
2922
2923 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2924 tried only at the first matching position in the subject string.
2925
2926 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2927 TURE options for pcre_exec() have no Perl equivalents.
2928
2929 (g) The \R escape sequence can be restricted to match only CR, LF, or
2930 CRLF by the PCRE_BSR_ANYCRLF option.
2931
2932 (h) The callout facility is PCRE-specific.
2933
2934 (i) The partial matching facility is PCRE-specific.
2935
2936 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2937 even on different hosts that have the other endianness.
2938
2939 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2940 different way and is not Perl-compatible.
2941
2942 (l) PCRE recognizes some special sequences such as (*CR) at the start
2943 of a pattern that set overall options that cannot be changed within the
2944 pattern.
2945
2946
2947 AUTHOR
2948
2949 Philip Hazel
2950 University Computing Service
2951 Cambridge CB2 3QH, England.
2952
2953
2954 REVISION
2955
2956 Last updated: 11 September 2007
2957 Copyright (c) 1997-2007 University of Cambridge.
2958 ------------------------------------------------------------------------------
2959
2960
2961 PCREPATTERN(3) PCREPATTERN(3)
2962
2963
2964 NAME
2965 PCRE - Perl-compatible regular expressions
2966
2967
2968 PCRE REGULAR EXPRESSION DETAILS
2969
2970 The syntax and semantics of the regular expressions that are supported
2971 by PCRE are described in detail below. There is a quick-reference syn-
2972 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
2973 semantics as closely as it can. PCRE also supports some alternative
2974 regular expression syntax (which does not conflict with the Perl syn-
2975 tax) in order to provide some compatibility with regular expressions in
2976 Python, .NET, and Oniguruma.
2977
2978 Perl's regular expressions are described in its own documentation, and
2979 regular expressions in general are covered in a number of books, some
2980 of which have copious examples. Jeffrey Friedl's "Mastering Regular
2981 Expressions", published by O'Reilly, covers regular expressions in
2982 great detail. This description of PCRE's regular expressions is
2983 intended as reference material.
2984
2985 The original operation of PCRE was on strings of one-byte characters.
2986 However, there is now also support for UTF-8 character strings. To use
2987 this, you must build PCRE to include UTF-8 support, and then call
2988 pcre_compile() with the PCRE_UTF8 option. There is also a special
2989 sequence that can be given at the start of a pattern:
2990
2991 (*UTF8)
2992
2993 Starting a pattern with this sequence is equivalent to setting the
2994 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
2995 UTF-8 mode affects pattern matching is mentioned in several places
2996 below. There is also a summary of UTF-8 features in the section on
2997 UTF-8 support in the main pcre page.
2998
2999 The remainder of this document discusses the patterns that are sup-
3000 ported by PCRE when its main matching function, pcre_exec(), is used.
3001 From release 6.0, PCRE offers a second matching function,
3002 pcre_dfa_exec(), which matches using a different algorithm that is not
3003 Perl-compatible. Some of the features discussed below are not available
3004 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3005 alternative function, and how it differs from the normal function, are
3006 discussed in the pcrematching page.
3007
3008
3009 NEWLINE CONVENTIONS
3010
3011 PCRE supports five different conventions for indicating line breaks in
3012 strings: a single CR (carriage return) character, a single LF (line-
3013 feed) character, the two-character sequence CRLF, any of the three pre-
3014 ceding, or any Unicode newline sequence. The pcreapi page has further
3015 discussion about newlines, and shows how to set the newline convention
3016 in the options arguments for the compiling and matching functions.
3017
3018 It is also possible to specify a newline convention by starting a pat-
3019 tern string with one of the following five sequences:
3020
3021 (*CR) carriage return
3022 (*LF) linefeed
3023 (*CRLF) carriage return, followed by linefeed
3024 (*ANYCRLF) any of the three above
3025 (*ANY) all Unicode newline sequences
3026
3027 These override the default and the options given to pcre_compile(). For
3028 example, on a Unix system where LF is the default newline sequence, the
3029 pattern
3030
3031 (*CR)a.b
3032
3033 changes the convention to CR. That pattern matches "a\nb" because LF is
3034 no longer a newline. Note that these special settings, which are not
3035 Perl-compatible, are recognized only at the very start of a pattern,
3036 and that they must be in upper case. If more than one of them is
3037 present, the last one is used.
3038
3039 The newline convention does not affect what the \R escape sequence
3040 matches. By default, this is any Unicode newline sequence, for Perl
3041 compatibility. However, this can be changed; see the description of \R
3042 in the section entitled "Newline sequences" below. A change of \R set-
3043 ting can be combined with a change of newline convention.
3044
3045
3046 CHARACTERS AND METACHARACTERS
3047
3048 A regular expression is a pattern that is matched against a subject
3049 string from left to right. Most characters stand for themselves in a
3050 pattern, and match the corresponding characters in the subject. As a
3051 trivial example, the pattern
3052
3053 The quick brown fox
3054
3055 matches a portion of a subject string that is identical to itself. When
3056 caseless matching is specified (the PCRE_CASELESS option), letters are
3057 matched independently of case. In UTF-8 mode, PCRE always understands
3058 the concept of case for characters whose values are less than 128, so
3059 caseless matching is always possible. For characters with higher val-
3060 ues, the concept of case is supported if PCRE is compiled with Unicode
3061 property support, but not otherwise. If you want to use caseless
3062 matching for characters 128 and above, you must ensure that PCRE is
3063 compiled with Unicode property support as well as with UTF-8 support.
3064
3065 The power of regular expressions comes from the ability to include
3066 alternatives and repetitions in the pattern. These are encoded in the
3067 pattern by the use of metacharacters, which do not stand for themselves
3068 but instead are interpreted in some special way.
3069
3070 There are two different sets of metacharacters: those that are recog-
3071 nized anywhere in the pattern except within square brackets, and those
3072 that are recognized within square brackets. Outside square brackets,
3073 the metacharacters are as follows:
3074
3075 \ general escape character with several uses
3076 ^ assert start of string (or line, in multiline mode)
3077 $ assert end of string (or line, in multiline mode)
3078 . match any character except newline (by default)
3079 [ start character class definition
3080 | start of alternative branch
3081 ( start subpattern
3082 ) end subpattern
3083 ? extends the meaning of (
3084 also 0 or 1 quantifier
3085 also quantifier minimizer
3086 * 0 or more quantifier
3087 + 1 or more quantifier
3088 also "possessive quantifier"
3089 { start min/max quantifier
3090
3091 Part of a pattern that is in square brackets is called a "character
3092 class". In a character class the only metacharacters are:
3093
3094 \ general escape character
3095 ^ negate the class, but only if the first character
3096 - indicates character range
3097 [ POSIX character class (only if followed by POSIX
3098 syntax)
3099 ] terminates the character class
3100
3101 The following sections describe the use of each of the metacharacters.
3102
3103
3104 BACKSLASH
3105
3106 The backslash character has several uses. Firstly, if it is followed by
3107 a non-alphanumeric character, it takes away any special meaning that
3108 character may have. This use of backslash as an escape character
3109 applies both inside and outside character classes.
3110
3111 For example, if you want to match a * character, you write \* in the
3112 pattern. This escaping action applies whether or not the following
3113 character would otherwise be interpreted as a metacharacter, so it is
3114 always safe to precede a non-alphanumeric with backslash to specify
3115 that it stands for itself. In particular, if you want to match a back-
3116 slash, you write \\.
3117
3118 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3119 the pattern (other than in a character class) and characters between a
3120 # outside a character class and the next newline are ignored. An escap-
3121 ing backslash can be used to include a whitespace or # character as
3122 part of the pattern.
3123
3124 If you want to remove the special meaning from a sequence of charac-
3125 ters, you can do so by putting them between \Q and \E. This is differ-
3126 ent from Perl in that $ and @ are handled as literals in \Q...\E
3127 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3128 tion. Note the following examples:
3129
3130 Pattern PCRE matches Perl matches
3131
3132 \Qabc$xyz\E abc$xyz abc followed by the
3133 contents of $xyz
3134 \Qabc\$xyz\E abc\$xyz abc\$xyz
3135 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3136
3137 The \Q...\E sequence is recognized both inside and outside character
3138 classes.
3139
3140 Non-printing characters
3141
3142 A second use of backslash provides a way of encoding non-printing char-
3143 acters in patterns in a visible manner. There is no restriction on the
3144 appearance of non-printing characters, apart from the binary zero that
3145 terminates a pattern, but when a pattern is being prepared by text
3146 editing, it is usually easier to use one of the following escape
3147 sequences than the binary character it represents:
3148
3149 \a alarm, that is, the BEL character (hex 07)
3150 \cx "control-x", where x is any character
3151 \e escape (hex 1B)
3152 \f formfeed (hex 0C)
3153 \n linefeed (hex 0A)
3154 \r carriage return (hex 0D)
3155 \t tab (hex 09)
3156 \ddd character with octal code ddd, or backreference
3157 \xhh character with hex code hh
3158 \x{hhh..} character with hex code hhh..
3159
3160 The precise effect of \cx is as follows: if x is a lower case letter,
3161 it is converted to upper case. Then bit 6 of the character (hex 40) is
3162 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3163 becomes hex 7B.
3164
3165 After \x, from zero to two hexadecimal digits are read (letters can be
3166 in upper or lower case). Any number of hexadecimal digits may appear
3167 between \x{ and }, but the value of the character code must be less
3168 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3169 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3170 than the largest Unicode code point, which is 10FFFF.
3171
3172 If characters other than hexadecimal digits appear between \x{ and },
3173 or if there is no terminating }, this form of escape is not recognized.
3174 Instead, the initial \x will be interpreted as a basic hexadecimal
3175 escape, with no following digits, giving a character whose value is
3176 zero.
3177
3178 Characters whose value is less than 256 can be defined by either of the
3179 two syntaxes for \x. There is no difference in the way they are han-
3180 dled. For example, \xdc is exactly the same as \x{dc}.
3181
3182 After \0 up to two further octal digits are read. If there are fewer
3183 than two digits, just those that are present are used. Thus the
3184 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3185 (code value 7). Make sure you supply two digits after the initial zero
3186 if the pattern character that follows is itself an octal digit.
3187
3188 The handling of a backslash followed by a digit other than 0 is compli-
3189 cated. Outside a character class, PCRE reads it and any following dig-
3190 its as a decimal number. If the number is less than 10, or if there
3191 have been at least that many previous capturing left parentheses in the
3192 expression, the entire sequence is taken as a back reference. A
3193 description of how this works is given later, following the discussion
3194 of parenthesized subpatterns.
3195
3196 Inside a character class, or if the decimal number is greater than 9
3197 and there have not been that many capturing subpatterns, PCRE re-reads
3198 up to three octal digits following the backslash, and uses them to gen-
3199 erate a data character. Any subsequent digits stand for themselves. In
3200 non-UTF-8 mode, the value of a character specified in octal must be
3201 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3202 example:
3203
3204 \040 is another way of writing a space
3205 \40 is the same, provided there are fewer than 40
3206 previous capturing subpatterns
3207 \7 is always a back reference
3208 \11 might be a back reference, or another way of
3209 writing a tab
3210 \011 is always a tab
3211 \0113 is a tab followed by the character "3"
3212 \113 might be a back reference, otherwise the
3213 character with octal code 113
3214 \377 might be a back reference, otherwise
3215 the byte consisting entirely of 1 bits
3216 \81 is either a back reference, or a binary zero
3217 followed by the two characters "8" and "1"
3218
3219 Note that octal values of 100 or greater must not be introduced by a
3220 leading zero, because no more than three octal digits are ever read.
3221
3222 All the sequences that define a single character value can be used both
3223 inside and outside character classes. In addition, inside a character
3224 class, the sequence \b is interpreted as the backspace character (hex
3225 08), and the sequences \R and \X are interpreted as the characters "R"
3226 and "X", respectively. Outside a character class, these sequences have
3227 different meanings (see below).
3228
3229 Absolute and relative back references
3230
3231 The sequence \g followed by an unsigned or a negative number, option-
3232 ally enclosed in braces, is an absolute or relative back reference. A
3233 named back reference can be coded as \g{name}. Back references are dis-
3234 cussed later, following the discussion of parenthesized subpatterns.
3235
3236 Absolute and relative subroutine calls
3237
3238 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3239 name or a number enclosed either in angle brackets or single quotes, is
3240 an alternative syntax for referencing a subpattern as a "subroutine".
3241 Details are discussed later. Note that \g{...} (Perl syntax) and
3242 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3243 reference; the latter is a subroutine call.
3244
3245 Generic character types
3246
3247 Another use of backslash is for specifying generic character types. The
3248 following are always recognized:
3249
3250 \d any decimal digit
3251 \D any character that is not a decimal digit
3252 \h any horizontal whitespace character
3253 \H any character that is not a horizontal whitespace character
3254 \s any whitespace character
3255 \S any character that is not a whitespace character
3256 \v any vertical whitespace character
3257 \V any character that is not a vertical whitespace character
3258 \w any "word" character
3259 \W any "non-word" character
3260
3261 Each pair of escape sequences partitions the complete set of characters
3262 into two disjoint sets. Any given character matches one, and only one,
3263 of each pair.
3264
3265 These character type sequences can appear both inside and outside char-
3266 acter classes. They each match one character of the appropriate type.
3267 If the current matching point is at the end of the subject string, all
3268 of them fail, since there is no character to match.
3269
3270 For compatibility with Perl, \s does not match the VT character (code
3271 11). This makes it different from the the POSIX "space" class. The \s
3272 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3273 "use locale;" is included in a Perl script, \s may match the VT charac-
3274 ter. In PCRE, it never does.
3275
3276 In UTF-8 mode, characters with values greater than 128 never match \d,
3277 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3278 code character property support is available. These sequences retain
3279 their original meanings from before UTF-8 support was available, mainly
3280 for efficiency reasons. Note that this also affects \b, because it is
3281 defined in terms of \w and \W.
3282
3283 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3284 the other sequences, these do match certain high-valued codepoints in
3285 UTF-8 mode. The horizontal space characters are:
3286
3287 U+0009 Horizontal tab
3288 U+0020 Space
3289 U+00A0 Non-break space
3290 U+1680 Ogham space mark
3291 U+180E Mongolian vowel separator
3292 U+2000 En quad
3293 U+2001 Em quad
3294 U+2002 En space
3295 U+2003 Em space
3296 U+2004 Three-per-em space
3297 U+2005 Four-per-em space
3298 U+2006 Six-per-em space
3299 U+2007 Figure space
3300 U+2008 Punctuation space
3301 U+2009 Thin space
3302 U+200A Hair space
3303 U+202F Narrow no-break space
3304 U+205F Medium mathematical space
3305 U+3000 Ideographic space
3306
3307 The vertical space characters are:
3308
3309 U+000A Linefeed
3310 U+000B Vertical tab
3311 U+000C Formfeed
3312 U+000D Carriage return
3313 U+0085 Next line
3314 U+2028 Line separator
3315 U+2029 Paragraph separator
3316
3317 A "word" character is an underscore or any character less than 256 that
3318 is a letter or digit. The definition of letters and digits is con-
3319 trolled by PCRE's low-valued character tables, and may vary if locale-
3320 specific matching is taking place (see "Locale support" in the pcreapi
3321 page). For example, in a French locale such as "fr_FR" in Unix-like
3322 systems, or "french" in Windows, some character codes greater than 128
3323 are used for accented letters, and these are matched by \w. The use of
3324 locales with Unicode is discouraged.
3325
3326 Newline sequences
3327
3328 Outside a character class, by default, the escape sequence \R matches
3329 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3330 mode \R is equivalent to the following:
3331
3332 (?>\r\n|\n|\x0b|\f|\r|\x85)
3333
3334 This is an example of an "atomic group", details of which are given
3335 below. This particular group matches either the two-character sequence
3336 CR followed by LF, or one of the single characters LF (linefeed,
3337 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3338 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3339 is treated as a single unit that cannot be split.
3340
3341 In UTF-8 mode, two additional characters whose codepoints are greater
3342 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3343 rator, U+2029). Unicode character property support is not needed for
3344 these characters to be recognized.
3345
3346 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3347 the complete set of Unicode line endings) by setting the option
3348 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3349 (BSR is an abbrevation for "backslash R".) This can be made the default
3350 when PCRE is built; if this is the case, the other behaviour can be
3351 requested via the PCRE_BSR_UNICODE option. It is also possible to
3352 specify these settings by starting a pattern string with one of the
3353 following sequences:
3354
3355 (*BSR_ANYCRLF) CR, LF, or CRLF only
3356 (*BSR_UNICODE) any Unicode newline sequence
3357
3358 These override the default and the options given to pcre_compile(), but
3359 they can be overridden by options given to pcre_exec(). Note that these
3360 special settings, which are not Perl-compatible, are recognized only at
3361 the very start of a pattern, and that they must be in upper case. If
3362 more than one of them is present, the last one is used. They can be
3363 combined with a change of newline convention, for example, a pattern
3364 can start with:
3365
3366 (*ANY)(*BSR_ANYCRLF)
3367
3368 Inside a character class, \R matches the letter "R".
3369
3370 Unicode character properties
3371
3372 When PCRE is built with Unicode character property support, three addi-
3373 tional escape sequences that match characters with specific properties
3374 are available. When not in UTF-8 mode, these sequences are of course
3375 limited to testing characters whose codepoints are less than 256, but
3376 they do work in this mode. The extra escape sequences are:
3377
3378 \p{xx} a character with the xx property
3379 \P{xx} a character without the xx property
3380 \X an extended Unicode sequence
3381
3382 The property names represented by xx above are limited to the Unicode
3383 script names, the general category properties, and "Any", which matches
3384 any character (including newline). Other properties such as "InMusical-
3385 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3386 not match any characters, so always causes a match failure.
3387
3388 Sets of Unicode characters are defined as belonging to certain scripts.
3389 A character from one of these sets can be matched using a script name.
3390 For example:
3391
3392 \p{Greek}
3393 \P{Han}
3394
3395 Those that are not part of an identified script are lumped together as
3396 "Common". The current list of scripts is:
3397
3398 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3399 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3400 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3401 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3402 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3403 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3404 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3405 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3406 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3407
3408 Each character has exactly one general category property, specified by
3409 a two-letter abbreviation. For compatibility with Perl, negation can be
3410 specified by including a circumflex between the opening brace and the
3411 property name. For example, \p{^Lu} is the same as \P{Lu}.
3412
3413 If only one letter is specified with \p or \P, it includes all the gen-
3414 eral category properties that start with that letter. In this case, in
3415 the absence of negation, the curly brackets in the escape sequence are
3416 optional; these two examples have the same effect:
3417
3418 \p{L}
3419 \pL
3420
3421 The following general category property codes are supported:
3422
3423 C Other
3424 Cc Control
3425 Cf Format
3426 Cn Unassigned
3427 Co Private use
3428 Cs Surrogate
3429
3430 L Letter
3431 Ll Lower case letter
3432 Lm Modifier letter
3433 Lo Other letter
3434 Lt Title case letter
3435 Lu Upper case letter
3436
3437 M Mark
3438 Mc Spacing mark
3439 Me Enclosing mark
3440 Mn Non-spacing mark
3441
3442 N Number
3443 Nd Decimal number
3444 Nl Letter number
3445 No Other number
3446
3447 P Punctuation
3448 Pc Connector punctuation
3449 Pd Dash punctuation
3450 Pe Close punctuation
3451 Pf Final punctuation
3452 Pi Initial punctuation
3453 Po Other punctuation
3454 Ps Open punctuation
3455
3456 S Symbol
3457 Sc Currency symbol
3458 Sk Modifier symbol
3459 Sm Mathematical symbol
3460 So Other symbol
3461
3462 Z Separator
3463 Zl Line separator
3464 Zp Paragraph separator
3465 Zs Space separator
3466
3467 The special property L& is also supported: it matches a character that
3468 has the Lu, Ll, or Lt property, in other words, a letter that is not
3469 classified as a modifier or "other".
3470
3471 The Cs (Surrogate) property applies only to characters in the range
3472 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3473 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3474 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3475 the pcreapi page).
3476
3477 The long synonyms for these properties that Perl supports (such as
3478 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3479 any of these properties with "Is".
3480
3481 No character that is in the Unicode table has the Cn (unassigned) prop-
3482 erty. Instead, this property is assumed for any code point that is not
3483 in the Unicode table.
3484
3485 Specifying caseless matching does not affect these escape sequences.
3486 For example, \p{Lu} always matches only upper case letters.
3487
3488 The \X escape matches any number of Unicode characters that form an
3489 extended Unicode sequence. \X is equivalent to
3490
3491 (?>\PM\pM*)
3492
3493 That is, it matches a character without the "mark" property, followed
3494 by zero or more characters with the "mark" property, and treats the
3495 sequence as an atomic group (see below). Characters with the "mark"
3496 property are typically accents that affect the preceding character.
3497 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3498 matches any one character.
3499
3500 Matching characters by Unicode property is not fast, because PCRE has
3501 to search a structure that contains data for over fifteen thousand
3502 characters. That is why the traditional escape sequences such as \d and
3503 \w do not use Unicode properties in PCRE.
3504
3505 Resetting the match start
3506
3507 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3508 ously matched characters not to be included in the final matched
3509 sequence. For example, the pattern:
3510
3511 foo\Kbar
3512
3513 matches "foobar", but reports that it has matched "bar". This feature
3514 is similar to a lookbehind assertion (described below). However, in
3515 this case, the part of the subject before the real match does not have
3516 to be of fixed length, as lookbehind assertions do. The use of \K does
3517 not interfere with the setting of captured substrings. For example,
3518 when the pattern
3519
3520 (foo)\Kbar
3521
3522 matches "foobar", the first substring is still set to "foo".
3523
3524 Simple assertions
3525
3526 The final use of backslash is for certain simple assertions. An asser-
3527 tion specifies a condition that has to be met at a particular point in
3528 a match, without consuming any characters from the subject string. The
3529 use of subpatterns for more complicated assertions is described below.
3530 The backslashed assertions are:
3531
3532 \b matches at a word boundary
3533 \B matches when not at a word boundary
3534 \A matches at the start of the subject
3535 \Z matches at the end of the subject
3536 also matches before a newline at the end of the subject
3537 \z matches only at the end of the subject
3538 \G matches at the first matching position in the subject
3539
3540 These assertions may not appear in character classes (but note that \b
3541 has a different meaning, namely the backspace character, inside a char-
3542 acter class).
3543
3544 A word boundary is a position in the subject string where the current
3545 character and the previous character do not both match \w or \W (i.e.
3546 one matches \w and the other matches \W), or the start or end of the
3547 string if the first or last character matches \w, respectively.
3548
3549 The \A, \Z, and \z assertions differ from the traditional circumflex
3550 and dollar (described in the next section) in that they only ever match
3551 at the very start and end of the subject string, whatever options are
3552 set. Thus, they are independent of multiline mode. These three asser-
3553 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3554 affect only the behaviour of the circumflex and dollar metacharacters.
3555 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3556 cating that matching is to start at a point other than the beginning of
3557 the subject, \A can never match. The difference between \Z and \z is
3558 that \Z matches before a newline at the end of the string as well as at
3559 the very end, whereas \z matches only at the end.
3560
3561 The \G assertion is true only when the current matching position is at
3562 the start point of the match, as specified by the startoffset argument
3563 of pcre_exec(). It differs from \A when the value of startoffset is
3564 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3565 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3566 mentation where \G can be useful.
3567
3568 Note, however, that PCRE's interpretation of \G, as the start of the
3569 current match, is subtly different from Perl's, which defines it as the
3570 end of the previous match. In Perl, these can be different when the
3571 previously matched string was empty. Because PCRE does just one match
3572 at a time, it cannot reproduce this behaviour.
3573
3574 If all the alternatives of a pattern begin with \G, the expression is
3575 anchored to the starting match position, and the "anchored" flag is set
3576 in the compiled regular expression.
3577
3578
3579 CIRCUMFLEX AND DOLLAR
3580
3581 Outside a character class, in the default matching mode, the circumflex
3582 character is an assertion that is true only if the current matching
3583 point is at the start of the subject string. If the startoffset argu-
3584 ment of pcre_exec() is non-zero, circumflex can never match if the
3585 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3586 has an entirely different meaning (see below).
3587
3588 Circumflex need not be the first character of the pattern if a number
3589 of alternatives are involved, but it should be the first thing in each
3590 alternative in which it appears if the pattern is ever to match that
3591 branch. If all possible alternatives start with a circumflex, that is,
3592 if the pattern is constrained to match only at the start of the sub-
3593 ject, it is said to be an "anchored" pattern. (There are also other
3594 constructs that can cause a pattern to be anchored.)
3595
3596 A dollar character is an assertion that is true only if the current
3597 matching point is at the end of the subject string, or immediately
3598 before a newline at the end of the string (by default). Dollar need not
3599 be the last character of the pattern if a number of alternatives are
3600 involved, but it should be the last item in any branch in which it
3601 appears. Dollar has no special meaning in a character class.
3602
3603 The meaning of dollar can be changed so that it matches only at the
3604 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3605 compile time. This does not affect the \Z assertion.
3606
3607 The meanings of the circumflex and dollar characters are changed if the
3608 PCRE_MULTILINE option is set. When this is the case, a circumflex
3609 matches immediately after internal newlines as well as at the start of
3610 the subject string. It does not match after a newline that ends the
3611 string. A dollar matches before any newlines in the string, as well as
3612 at the very end, when PCRE_MULTILINE is set. When newline is specified
3613 as the two-character sequence CRLF, isolated CR and LF characters do
3614 not indicate newlines.
3615
3616 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3617 (where \n represents a newline) in multiline mode, but not otherwise.
3618 Consequently, patterns that are anchored in single line mode because
3619 all branches start with ^ are not anchored in multiline mode, and a
3620 match for circumflex is possible when the startoffset argument of
3621 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3622 PCRE_MULTILINE is set.
3623
3624 Note that the sequences \A, \Z, and \z can be used to match the start
3625 and end of the subject in both modes, and if all branches of a pattern
3626 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3627 set.
3628
3629
3630 FULL STOP (PERIOD, DOT)
3631
3632 Outside a character class, a dot in the pattern matches any one charac-
3633 ter in the subject string except (by default) a character that signi-
3634 fies the end of a line. In UTF-8 mode, the matched character may be
3635 more than one byte long.
3636
3637 When a line ending is defined as a single character, dot never matches
3638 that character; when the two-character sequence CRLF is used, dot does
3639 not match CR if it is immediately followed by LF, but otherwise it
3640 matches all characters (including isolated CRs and LFs). When any Uni-
3641 code line endings are being recognized, dot does not match CR or LF or
3642 any of the other line ending characters.
3643
3644 The behaviour of dot with regard to newlines can be changed. If the
3645 PCRE_DOTALL option is set, a dot matches any one character, without
3646 exception. If the two-character sequence CRLF is present in the subject
3647 string, it takes two dots to match it.
3648
3649 The handling of dot is entirely independent of the handling of circum-
3650 flex and dollar, the only relationship being that they both involve
3651 newlines. Dot has no special meaning in a character class.
3652
3653
3654 MATCHING A SINGLE BYTE
3655
3656 Outside a character class, the escape sequence \C matches any one byte,
3657 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3658 line-ending characters. The feature is provided in Perl in order to
3659 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3660 acters into individual bytes, what remains in the string may be a mal-
3661 formed UTF-8 string. For this reason, the \C escape sequence is best
3662 avoided.
3663
3664 PCRE does not allow \C to appear in lookbehind assertions (described
3665 below), because in UTF-8 mode this would make it impossible to calcu-
3666 late the length of the lookbehind.
3667
3668
3669 SQUARE BRACKETS AND CHARACTER CLASSES
3670
3671 An opening square bracket introduces a character class, terminated by a
3672 closing square bracket. A closing square bracket on its own is not spe-
3673 cial. If a closing square bracket is required as a member of the class,
3674 it should be the first data character in the class (after an initial
3675 circumflex, if present) or escaped with a backslash.
3676
3677 A character class matches a single character in the subject. In UTF-8
3678 mode, the character may occupy more than one byte. A matched character
3679 must be in the set of characters defined by the class, unless the first
3680 character in the class definition is a circumflex, in which case the
3681 subject character must not be in the set defined by the class. If a
3682 circumflex is actually required as a member of the class, ensure it is
3683 not the first character, or escape it with a backslash.
3684
3685 For example, the character class [aeiou] matches any lower case vowel,
3686 while [^aeiou] matches any character that is not a lower case vowel.
3687 Note that a circumflex is just a convenient notation for specifying the
3688 characters that are in the class by enumerating those that are not. A
3689 class that starts with a circumflex is not an assertion: it still con-
3690 sumes a character from the subject string, and therefore it fails if
3691 the current pointer is at the end of the string.
3692
3693 In UTF-8 mode, characters with values greater than 255 can be included
3694 in a class as a literal string of bytes, or by using the \x{ escaping
3695 mechanism.
3696
3697 When caseless matching is set, any letters in a class represent both
3698 their upper case and lower case versions, so for example, a caseless
3699 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3700 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3701 understands the concept of case for characters whose values are less
3702 than 128, so caseless matching is always possible. For characters with
3703 higher values, the concept of case is supported if PCRE is compiled
3704 with Unicode property support, but not otherwise. If you want to use
3705 caseless matching for characters 128 and above, you must ensure that
3706 PCRE is compiled with Unicode property support as well as with UTF-8
3707 support.
3708
3709 Characters that might indicate line breaks are never treated in any
3710 special way when matching character classes, whatever line-ending
3711 sequence is in use, and whatever setting of the PCRE_DOTALL and
3712 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3713 of these characters.
3714
3715 The minus (hyphen) character can be used to specify a range of charac-
3716 ters in a character class. For example, [d-m] matches any letter
3717 between d and m, inclusive. If a minus character is required in a
3718 class, it must be escaped with a backslash or appear in a position
3719 where it cannot be interpreted as indicating a range, typically as the
3720 first or last character in the class.
3721
3722 It is not possible to have the literal character "]" as the end charac-
3723 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3724 two characters ("W" and "-") followed by a literal string "46]", so it
3725 would match "W46]" or "-46]". However, if the "]" is escaped with a
3726 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3727 preted as a class containing a range followed by two other characters.
3728 The octal or hexadecimal representation of "]" can also be used to end
3729 a range.
3730
3731 Ranges operate in the collating sequence of character values. They can
3732 also be used for characters specified numerically, for example
3733 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3734 are greater than 255, for example [\x{100}-\x{2ff}].
3735
3736 If a range that includes letters is used when caseless matching is set,
3737 it matches the letters in either case. For example, [W-c] is equivalent
3738 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3739 character tables for a French locale are in use, [\xc8-\xcb] matches
3740 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3741 concept of case for characters with values greater than 128 only when
3742 it is compiled with Unicode property support.
3743
3744 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3745 in a character class, and add the characters that they match to the
3746 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3747 flex can conveniently be used with the upper case character types to
3748 specify a more restricted set of characters than the matching lower
3749 case type. For example, the class [^\W_] matches any letter or digit,
3750 but not underscore.
3751
3752 The only metacharacters that are recognized in character classes are
3753 backslash, hyphen (only where it can be interpreted as specifying a
3754 range), circumflex (only at the start), opening square bracket (only
3755 when it can be interpreted as introducing a POSIX class name - see the
3756 next section), and the terminating closing square bracket. However,
3757 escaping other non-alphanumeric characters does no harm.
3758
3759
3760 POSIX CHARACTER CLASSES
3761
3762 Perl supports the POSIX notation for character classes. This uses names
3763 enclosed by [: and :] within the enclosing square brackets. PCRE also
3764 supports this notation. For example,
3765
3766 [01[:alpha:]%]
3767
3768 matches "0", "1", any alphabetic character, or "%". The supported class
3769 names are
3770
3771 alnum letters and digits
3772 alpha letters
3773 ascii character codes 0 - 127
3774 blank space or tab only
3775 cntrl control characters
3776 digit decimal digits (same as \d)
3777 graph printing characters, excluding space
3778 lower lower case letters
3779 print printing characters, including space
3780 punct printing characters, excluding letters and digits
3781 space white space (not quite the same as \s)
3782 upper upper case letters
3783 word "word" characters (same as \w)
3784 xdigit hexadecimal digits
3785
3786 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3787 and space (32). Notice that this list includes the VT character (code
3788 11). This makes "space" different to \s, which does not include VT (for
3789 Perl compatibility).
3790
3791 The name "word" is a Perl extension, and "blank" is a GNU extension
3792 from Perl 5.8. Another Perl extension is negation, which is indicated
3793 by a ^ character after the colon. For example,
3794
3795 [12[:^digit:]]
3796
3797 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3798 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3799 these are not supported, and an error is given if they are encountered.
3800
3801 In UTF-8 mode, characters with values greater than 128 do not match any
3802 of the POSIX character classes.
3803
3804
3805 VERTICAL BAR
3806
3807 Vertical bar characters are used to separate alternative patterns. For
3808 example, the pattern
3809
3810 gilbert|sullivan
3811
3812 matches either "gilbert" or "sullivan". Any number of alternatives may
3813 appear, and an empty alternative is permitted (matching the empty
3814 string). The matching process tries each alternative in turn, from left
3815 to right, and the first one that succeeds is used. If the alternatives
3816 are within a subpattern (defined below), "succeeds" means matching the
3817 rest of the main pattern as well as the alternative in the subpattern.
3818
3819
3820 INTERNAL OPTION SETTING
3821
3822 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3823 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3824 within the pattern by a sequence of Perl option letters enclosed
3825 between "(?" and ")". The option letters are
3826
3827 i for PCRE_CASELESS
3828 m for PCRE_MULTILINE
3829 s for PCRE_DOTALL
3830 x for PCRE_EXTENDED
3831
3832 For example, (?im) sets caseless, multiline matching. It is also possi-
3833 ble to unset these options by preceding the letter with a hyphen, and a
3834 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3835 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3836 is also permitted. If a letter appears both before and after the
3837 hyphen, the option is unset.
3838
3839 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3840 can be changed in the same way as the Perl-compatible options by using
3841 the characters J, U and X respectively.
3842
3843 When one of these option changes occurs at top level (that is, not
3844 inside subpattern parentheses), the change applies to the remainder of
3845 the pattern that follows. If the change is placed right at the start of
3846 a pattern, PCRE extracts it into the global options (and it will there-
3847 fore show up in data extracted by the pcre_fullinfo() function).
3848
3849 An option change within a subpattern (see below for a description of
3850 subpatterns) affects only that part of the current pattern that follows
3851 it, so
3852
3853 (a(?i)b)c
3854
3855 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3856 used). By this means, options can be made to have different settings
3857 in different parts of the pattern. Any changes made in one alternative
3858 do carry on into subsequent branches within the same subpattern. For
3859 example,
3860
3861 (a(?i)b|c)
3862
3863 matches "ab", "aB", "c", and "C", even though when matching "C" the
3864 first branch is abandoned before the option setting. This is because
3865 the effects of option settings happen at compile time. There would be
3866 some very weird behaviour otherwise.
3867
3868 Note: There are other PCRE-specific options that can be set by the
3869 application when the compile or match functions are called. In some
3870 cases the pattern can contain special leading sequences such as (*CRLF)
3871 to override what the application has set or what has been defaulted.
3872 Details are given in the section entitled "Newline sequences" above.
3873 There is also the (*UTF8) leading sequence that can be used to set
3874 UTF-8 mode; this is equivalent to setting the PCRE_UTF8 option.
3875
3876
3877 SUBPATTERNS
3878
3879 Subpatterns are delimited by parentheses (round brackets), which can be
3880 nested. Turning part of a pattern into a subpattern does two things:
3881
3882 1. It localizes a set of alternatives. For example, the pattern
3883
3884 cat(aract|erpillar|)
3885
3886 matches one of the words "cat", "cataract", or "caterpillar". Without
3887 the parentheses, it would match "cataract", "erpillar" or an empty
3888 string.
3889
3890 2. It sets up the subpattern as a capturing subpattern. This means
3891 that, when the whole pattern matches, that portion of the subject
3892 string that matched the subpattern is passed back to the caller via the
3893 ovector argument of pcre_exec(). Opening parentheses are counted from
3894 left to right (starting from 1) to obtain numbers for the capturing
3895 subpatterns.
3896
3897 For example, if the string "the red king" is matched against the pat-
3898 tern
3899
3900 the ((red|white) (king|queen))
3901
3902 the captured substrings are "red king", "red", and "king", and are num-
3903 bered 1, 2, and 3, respectively.
3904
3905 The fact that plain parentheses fulfil two functions is not always
3906 helpful. There are often times when a grouping subpattern is required
3907 without a capturing requirement. If an opening parenthesis is followed
3908 by a question mark and a colon, the subpattern does not do any captur-
3909 ing, and is not counted when computing the number of any subsequent
3910 capturing subpatterns. For example, if the string "the white queen" is
3911 matched against the pattern
3912
3913 the ((?:red|white) (king|queen))
3914
3915 the captured substrings are "white queen" and "queen", and are numbered
3916 1 and 2. The maximum number of capturing subpatterns is 65535.
3917
3918 As a convenient shorthand, if any option settings are required at the
3919 start of a non-capturing subpattern, the option letters may appear
3920 between the "?" and the ":". Thus the two patterns
3921
3922 (?i:saturday|sunday)
3923 (?:(?i)saturday|sunday)
3924
3925 match exactly the same set of strings. Because alternative branches are
3926 tried from left to right, and options are not reset until the end of
3927 the subpattern is reached, an option setting in one branch does affect
3928 subsequent branches, so the above patterns match "SUNDAY" as well as
3929 "Saturday".
3930
3931
3932 DUPLICATE SUBPATTERN NUMBERS
3933
3934 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3935 uses the same numbers for its capturing parentheses. Such a subpattern
3936 starts with (?| and is itself a non-capturing subpattern. For example,
3937 consider this pattern:
3938
3939 (?|(Sat)ur|(Sun))day
3940
3941 Because the two alternatives are inside a (?| group, both sets of cap-
3942 turing parentheses are numbered one. Thus, when the pattern matches,
3943 you can look at captured substring number one, whichever alternative
3944 matched. This construct is useful when you want to capture part, but
3945 not all, of one of a number of alternatives. Inside a (?| group, paren-
3946 theses are numbered as usual, but the number is reset at the start of
3947 each branch. The numbers of any capturing buffers that follow the sub-
3948 pattern start after the highest number used in any branch. The follow-
3949 ing example is taken from the Perl documentation. The numbers under-
3950 neath show in which buffer the captured content will be stored.
3951
3952 # before ---------------branch-reset----------- after
3953 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3954 # 1 2 2 3 2 3 4
3955
3956 A backreference or a recursive call to a numbered subpattern always
3957 refers to the first one in the pattern with the given number.
3958
3959 An alternative approach to using this "branch reset" feature is to use
3960 duplicate named subpatterns, as described in the next section.
3961
3962
3963 NAMED SUBPATTERNS
3964
3965 Identifying capturing parentheses by number is simple, but it can be
3966 very hard to keep track of the numbers in complicated regular expres-
3967 sions. Furthermore, if an expression is modified, the numbers may
3968 change. To help with this difficulty, PCRE supports the naming of sub-
3969 patterns. This feature was not added to Perl until release 5.10. Python
3970 had the feature earlier, and PCRE introduced it at release 4.0, using
3971 the Python syntax. PCRE now supports both the Perl and the Python syn-
3972 tax.
3973
3974 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3975 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3976 to capturing parentheses from other parts of the pattern, such as back-
3977 references, recursion, and conditions, can be made by name as well as
3978 by number.
3979
3980 Names consist of up to 32 alphanumeric characters and underscores.
3981 Named capturing parentheses are still allocated numbers as well as
3982 names, exactly as if the names were not present. The PCRE API provides
3983 function calls for extracting the name-to-number translation table from
3984 a compiled pattern. There is also a convenience function for extracting
3985 a captured substring by name.
3986
3987 By default, a name must be unique within a pattern, but it is possible
3988 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3989 time. This can be useful for patterns where only one instance of the
3990 named parentheses can match. Suppose you want to match the name of a
3991 weekday, either as a 3-letter abbreviation or as the full name, and in
3992 both cases you want to extract the abbreviation. This pattern (ignoring
3993 the line breaks) does the job:
3994
3995 (?<DN>Mon|Fri|Sun)(?:day)?|
3996 (?<DN>Tue)(?:sday)?|
3997 (?<DN>Wed)(?:nesday)?|
3998 (?<DN>Thu)(?:rsday)?|
3999 (?<DN>Sat)(?:urday)?
4000
4001 There are five capturing substrings, but only one is ever set after a
4002 match. (An alternative way of solving this problem is to use a "branch
4003 reset" subpattern, as described in the previous section.)
4004
4005 The convenience function for extracting the data by name returns the
4006 substring for the first (and in this example, the only) subpattern of
4007 that name that matched. This saves searching to find which numbered
4008 subpattern it was. If you make a reference to a non-unique named sub-
4009 pattern from elsewhere in the pattern, the one that corresponds to the
4010 lowest number is used. For further details of the interfaces for han-
4011 dling named subpatterns, see the pcreapi documentation.
4012
4013 Warning: You cannot use different names to distinguish between two sub-
4014 patterns with the same number (see the previous section) because PCRE
4015 uses only the numbers when matching.
4016
4017
4018 REPETITION
4019
4020 Repetition is specified by quantifiers, which can follow any of the
4021 following items:
4022
4023 a literal data character
4024 the dot metacharacter
4025 the \C escape sequence
4026 the \X escape sequence (in UTF-8 mode with Unicode properties)
4027 the \R escape sequence
4028 an escape such as \d that matches a single character
4029 a character class
4030 a back reference (see next section)
4031 a parenthesized subpattern (unless it is an assertion)
4032
4033 The general repetition quantifier specifies a minimum and maximum num-
4034 ber of permitted matches, by giving the two numbers in curly brackets
4035 (braces), separated by a comma. The numbers must be less than 65536,
4036 and the first must be less than or equal to the second. For example:
4037
4038 z{2,4}
4039
4040 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4041 special character. If the second number is omitted, but the comma is
4042 present, there is no upper limit; if the second number and the comma
4043 are both omitted, the quantifier specifies an exact number of required
4044 matches. Thus
4045
4046 [aeiou]{3,}
4047
4048 matches at least 3 successive vowels, but may match many more, while
4049
4050 \d{8}
4051
4052 matches exactly 8 digits. An opening curly bracket that appears in a
4053 position where a quantifier is not allowed, or one that does not match
4054 the syntax of a quantifier, is taken as a literal character. For exam-
4055 ple, {,6} is not a quantifier, but a literal string of four characters.
4056
4057 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4058 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4059 acters, each of which is represented by a two-byte sequence. Similarly,
4060 when Unicode property support is available, \X{3} matches three Unicode
4061 extended sequences, each of which may be several bytes long (and they
4062 may be of different lengths).
4063
4064 The quantifier {0} is permitted, causing the expression to behave as if
4065 the previous item and the quantifier were not present. This may be use-
4066 ful for subpatterns that are referenced as subroutines from elsewhere
4067 in the pattern. Items other than subpatterns that have a {0} quantifier
4068 are omitted from the compiled pattern.
4069
4070 For convenience, the three most common quantifiers have single-charac-
4071 ter abbreviations:
4072
4073 * is equivalent to {0,}
4074 + is equivalent to {1,}
4075 ? is equivalent to {0,1}
4076
4077 It is possible to construct infinite loops by following a subpattern
4078 that can match no characters with a quantifier that has no upper limit,
4079 for example:
4080
4081 (a?)*
4082
4083 Earlier versions of Perl and PCRE used to give an error at compile time
4084 for such patterns. However, because there are cases where this can be
4085 useful, such patterns are now accepted, but if any repetition of the
4086 subpattern does in fact match no characters, the loop is forcibly bro-
4087 ken.
4088
4089 By default, the quantifiers are "greedy", that is, they match as much
4090 as possible (up to the maximum number of permitted times), without
4091 causing the rest of the pattern to fail. The classic example of where
4092 this gives problems is in trying to match comments in C programs. These
4093 appear between /* and */ and within the comment, individual * and /
4094 characters may appear. An attempt to match C comments by applying the
4095 pattern
4096
4097 /\*.*\*/
4098
4099 to the string
4100
4101 /* first comment */ not comment /* second comment */
4102
4103 fails, because it matches the entire string owing to the greediness of
4104 the .* item.
4105
4106 However, if a quantifier is followed by a question mark, it ceases to
4107 be greedy, and instead matches the minimum number of times possible, so
4108 the pattern
4109
4110 /\*.*?\*/
4111
4112 does the right thing with the C comments. The meaning of the various
4113 quantifiers is not otherwise changed, just the preferred number of
4114 matches. Do not confuse this use of question mark with its use as a
4115 quantifier in its own right. Because it has two uses, it can sometimes
4116 appear doubled, as in
4117
4118 \d??\d
4119
4120 which matches one digit by preference, but can match two if that is the
4121 only way the rest of the pattern matches.
4122
4123 If the PCRE_UNGREEDY option is set (an option that is not available in
4124 Perl), the quantifiers are not greedy by default, but individual ones
4125 can be made greedy by following them with a question mark. In other
4126 words, it inverts the default behaviour.
4127
4128 When a parenthesized subpattern is quantified with a minimum repeat
4129 count that is greater than 1 or with a limited maximum, more memory is
4130 required for the compiled pattern, in proportion to the size of the
4131 minimum or maximum.
4132
4133 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4134 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4135 the pattern is implicitly anchored, because whatever follows will be
4136 tried against every character position in the subject string, so there
4137 is no point in retrying the overall match at any position after the
4138 first. PCRE normally treats such a pattern as though it were preceded
4139 by \A.
4140
4141 In cases where it is known that the subject string contains no new-
4142 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4143 mization, or alternatively using ^ to indicate anchoring explicitly.
4144
4145 However, there is one situation where the optimization cannot be used.
4146 When .* is inside capturing parentheses that are the subject of a
4147 backreference elsewhere in the pattern, a match at the start may fail
4148 where a later one succeeds. Consider, for example:
4149
4150 (.*)abc\1
4151
4152 If the subject is "xyz123abc123" the match point is the fourth charac-
4153 ter. For this reason, such a pattern is not implicitly anchored.
4154
4155 When a capturing subpattern is repeated, the value captured is the sub-
4156 string that matched the final iteration. For example, after
4157
4158 (tweedle[dume]{3}\s*)+
4159
4160 has matched "tweedledum tweedledee" the value of the captured substring
4161 is "tweedledee". However, if there are nested capturing subpatterns,
4162 the corresponding captured values may have been set in previous itera-
4163 tions. For example, after
4164
4165 /(a|(b))+/
4166
4167 matches "aba" the value of the second captured substring is "b".
4168
4169
4170 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4171
4172 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4173 repetition, failure of what follows normally causes the repeated item
4174 to be re-evaluated to see if a different number of repeats allows the
4175 rest of the pattern to match. Sometimes it is useful to prevent this,
4176 either to change the nature of the match, or to cause it fail earlier
4177 than it otherwise might, when the author of the pattern knows there is
4178 no point in carrying on.
4179
4180 Consider, for example, the pattern \d+foo when applied to the subject
4181 line
4182
4183 123456bar
4184
4185 After matching all 6 digits and then failing to match "foo", the normal
4186 action of the matcher is to try again with only 5 digits matching the
4187 \d+ item, and then with 4, and so on, before ultimately failing.
4188 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4189 the means for specifying that once a subpattern has matched, it is not
4190 to be re-evaluated in this way.
4191
4192 If we use atomic grouping for the previous example, the matcher gives
4193 up immediately on failing to match "foo" the first time. The notation
4194 is a kind of special parenthesis, starting with (?> as in this example:
4195
4196 (?>\d+)foo
4197
4198 This kind of parenthesis "locks up" the part of the pattern it con-
4199 tains once it has matched, and a failure further into the pattern is
4200 prevented from backtracking into it. Backtracking past it to previous
4201 items, however, works as normal.
4202
4203 An alternative description is that a subpattern of this type matches
4204 the string of characters that an identical standalone pattern would
4205 match, if anchored at the current point in the subject string.
4206
4207 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4208 such as the above example can be thought of as a maximizing repeat that
4209 must swallow everything it can. So, while both \d+ and \d+? are pre-
4210 pared to adjust the number of digits they match in order to make the
4211 rest of the pattern match, (?>\d+) can only match an entire sequence of
4212 digits.
4213
4214 Atomic groups in general can of course contain arbitrarily complicated
4215 subpatterns, and can be nested. However, when the subpattern for an
4216 atomic group is just a single repeated item, as in the example above, a
4217 simpler notation, called a "possessive quantifier" can be used. This
4218 consists of an additional + character following a quantifier. Using
4219 this notation, the previous example can be rewritten as
4220
4221 \d++foo
4222
4223 Note that a possessive quantifier can be used with an entire group, for
4224 example:
4225
4226 (abc|xyz){2,3}+
4227
4228 Possessive quantifiers are always greedy; the setting of the
4229 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4230 simpler forms of atomic group. However, there is no difference in the
4231 meaning of a possessive quantifier and the equivalent atomic group,
4232 though there may be a performance difference; possessive quantifiers
4233 should be slightly faster.
4234
4235 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4236 tax. Jeffrey Friedl originated the idea (and the name) in the first
4237 edition of his book. Mike McCloskey liked it, so implemented it when he
4238 built Sun's Java package, and PCRE copied it from there. It ultimately
4239 found its way into Perl at release 5.10.
4240
4241 PCRE has an optimization that automatically "possessifies" certain sim-
4242 ple pattern constructs. For example, the sequence A+B is treated as
4243 A++B because there is no point in backtracking into a sequence of A's
4244 when B must follow.
4245
4246 When a pattern contains an unlimited repeat inside a subpattern that
4247 can itself be repeated an unlimited number of times, the use of an
4248 atomic group is the only way to avoid some failing matches taking a
4249 very long time indeed. The pattern
4250
4251 (\D+|<\d+>)*[!?]
4252
4253 matches an unlimited number of substrings that either consist of non-
4254 digits, or digits enclosed in <>, followed by either ! or ?. When it
4255 matches, it runs quickly. However, if it is applied to
4256
4257 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4258
4259 it takes a long time before reporting failure. This is because the
4260 string can be divided between the internal \D+ repeat and the external
4261 * repeat in a large number of ways, and all have to be tried. (The
4262 example uses [!?] rather than a single character at the end, because
4263 both PCRE and Perl have an optimization that allows for fast failure
4264 when a single character is used. They remember the last single charac-
4265 ter that is required for a match, and fail early if it is not present
4266 in the string.) If the pattern is changed so that it uses an atomic
4267 group, like this:
4268
4269 ((?>\D+)|<\d+>)*[!?]
4270
4271 sequences of non-digits cannot be broken, and failure happens quickly.
4272
4273
4274 BACK REFERENCES
4275
4276 Outside a character class, a backslash followed by a digit greater than
4277 0 (and possibly further digits) is a back reference to a capturing sub-
4278 pattern earlier (that is, to its left) in the pattern, provided there
4279 have been that many previous capturing left parentheses.
4280
4281 However, if the decimal number following the backslash is less than 10,
4282 it is always taken as a back reference, and causes an error only if
4283 there are not that many capturing left parentheses in the entire pat-
4284 tern. In other words, the parentheses that are referenced need not be
4285 to the left of the reference for numbers less than 10. A "forward back
4286 reference" of this type can make sense when a repetition is involved
4287 and the subpattern to the right has participated in an earlier itera-
4288 tion.
4289
4290 It is not possible to have a numerical "forward back reference" to a
4291 subpattern whose number is 10 or more using this syntax because a
4292 sequence such as \50 is interpreted as a character defined in octal.
4293 See the subsection entitled "Non-printing characters" above for further
4294 details of the handling of digits following a backslash. There is no
4295 such problem when named parentheses are used. A back reference to any
4296 subpattern is possible using named parentheses (see below).
4297
4298 Another way of avoiding the ambiguity inherent in the use of digits
4299 following a backslash is to use the \g escape sequence, which is a fea-
4300 ture introduced in Perl 5.10. This escape must be followed by an
4301 unsigned number or a negative number, optionally enclosed in braces.
4302 These examples are all identical:
4303
4304 (ring), \1
4305 (ring), \g1
4306 (ring), \g{1}
4307
4308 An unsigned number specifies an absolute reference without the ambigu-
4309 ity that is present in the older syntax. It is also useful when literal
4310 digits follow the reference. A negative number is a relative reference.
4311 Consider this example:
4312
4313 (abc(def)ghi)\g{-1}
4314
4315 The sequence \g{-1} is a reference to the most recently started captur-
4316 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4317 \g{-2} would be equivalent to \1. The use of relative references can be
4318 helpful in long patterns, and also in patterns that are created by
4319 joining together fragments that contain references within themselves.
4320
4321 A back reference matches whatever actually matched the capturing sub-
4322 pattern in the current subject string, rather than anything matching
4323 the subpattern itself (see "Subpatterns as subroutines" below for a way
4324 of doing that). So the pattern
4325
4326 (sens|respons)e and \1ibility
4327
4328 matches "sense and sensibility" and "response and responsibility", but
4329 not "sense and responsibility". If caseful matching is in force at the
4330 time of the back reference, the case of letters is relevant. For exam-
4331 ple,
4332
4333 ((?i)rah)\s+\1
4334
4335 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4336 original capturing subpattern is matched caselessly.
4337
4338 There are several different ways of writing back references to named
4339 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4340 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4341 unified back reference syntax, in which \g can be used for both numeric
4342 and named references, is also supported. We could rewrite the above
4343 example in any of the following ways:
4344
4345 (?<p1>(?i)rah)\s+\k<p1>
4346 (?'p1'(?i)rah)\s+\k{p1}
4347 (?P<p1>(?i)rah)\s+(?P=p1)
4348 (?<p1>(?i)rah)\s+\g{p1}
4349
4350 A subpattern that is referenced by name may appear in the pattern
4351 before or after the reference.
4352
4353 There may be more than one back reference to the same subpattern. If a
4354 subpattern has not actually been used in a particular match, any back
4355 references to it always fail. For example, the pattern
4356
4357 (a|(bc))\2
4358
4359 always fails if it starts to match "a" rather than "bc". Because there
4360 may be many capturing parentheses in a pattern, all digits following
4361 the backslash are taken as part of a potential back reference number.
4362 If the pattern continues with a digit character, some delimiter must be
4363 used to terminate the back reference. If the PCRE_EXTENDED option is
4364 set, this can be whitespace. Otherwise an empty comment (see "Com-
4365 ments" below) can be used.
4366
4367 A back reference that occurs inside the parentheses to which it refers
4368 fails when the subpattern is first used, so, for example, (a\1) never
4369 matches. However, such references can be useful inside repeated sub-
4370 patterns. For example, the pattern
4371
4372 (a|b\1)+
4373
4374 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4375 ation of the subpattern, the back reference matches the character
4376 string corresponding to the previous iteration. In order for this to
4377 work, the pattern must be such that the first iteration does not need
4378 to match the back reference. This can be done using alternation, as in
4379 the example above, or by a quantifier with a minimum of zero.
4380
4381
4382 ASSERTIONS
4383
4384 An assertion is a test on the characters following or preceding the
4385 current matching point that does not actually consume any characters.
4386 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4387 described above.
4388
4389 More complicated assertions are coded as subpatterns. There are two
4390 kinds: those that look ahead of the current position in the subject
4391 string, and those that look behind it. An assertion subpattern is
4392 matched in the normal way, except that it does not cause the current
4393 matching position to be changed.
4394
4395 Assertion subpatterns are not capturing subpatterns, and may not be
4396 repeated, because it makes no sense to assert the same thing several
4397 times. If any kind of assertion contains capturing subpatterns within
4398 it, these are counted for the purposes of numbering the capturing sub-
4399 patterns in the whole pattern. However, substring capturing is carried
4400 out only for positive assertions, because it does not make sense for
4401 negative assertions.
4402
4403 Lookahead assertions
4404
4405 Lookahead assertions start with (?= for positive assertions and (?! for
4406 negative assertions. For example,
4407
4408 \w+(?=;)
4409
4410 matches a word followed by a semicolon, but does not include the semi-
4411 colon in the match, and
4412
4413 foo(?!bar)
4414
4415 matches any occurrence of "foo" that is not followed by "bar". Note
4416 that the apparently similar pattern
4417
4418 (?!foo)bar
4419
4420 does not find an occurrence of "bar" that is preceded by something
4421 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4422 the assertion (?!foo) is always true when the next three characters are
4423 "bar". A lookbehind assertion is needed to achieve the other effect.
4424
4425 If you want to force a matching failure at some point in a pattern, the
4426 most convenient way to do it is with (?!) because an empty string
4427 always matches, so an assertion that requires there not to be an empty
4428 string must always fail.
4429
4430 Lookbehind assertions
4431
4432 Lookbehind assertions start with (?<= for positive assertions and (?<!
4433 for negative assertions. For example,
4434
4435 (?<!foo)bar
4436
4437 does find an occurrence of "bar" that is not preceded by "foo". The
4438 contents of a lookbehind assertion are restricted such that all the
4439 strings it matches must have a fixed length. However, if there are sev-
4440 eral top-level alternatives, they do not all have to have the same
4441 fixed length. Thus
4442
4443 (?<=bullock|donkey)
4444
4445 is permitted, but
4446
4447 (?<!dogs?|cats?)
4448
4449 causes an error at compile time. Branches that match different length
4450 strings are permitted only at the top level of a lookbehind assertion.
4451 This is an extension compared with Perl (at least for 5.8), which
4452 requires all branches to match the same length of string. An assertion
4453 such as
4454
4455 (?<=ab(c|de))
4456
4457 is not permitted, because its single top-level branch can match two
4458 different lengths, but it is acceptable if rewritten to use two top-
4459 level branches:
4460
4461 (?<=abc|abde)
4462
4463 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4464 instead of a lookbehind assertion; this is not restricted to a fixed-
4465 length.
4466
4467 The implementation of lookbehind assertions is, for each alternative,
4468 to temporarily move the current position back by the fixed length and
4469 then try to match. If there are insufficient characters before the cur-
4470 rent position, the assertion fails.
4471
4472 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4473 mode) to appear in lookbehind assertions, because it makes it impossi-
4474 ble to calculate the length of the lookbehind. The \X and \R escapes,
4475 which can match different numbers of bytes, are also not permitted.
4476
4477 Possessive quantifiers can be used in conjunction with lookbehind
4478 assertions to specify efficient matching at the end of the subject
4479 string. Consider a simple pattern such as
4480
4481 abcd$
4482
4483 when applied to a long string that does not match. Because matching
4484 proceeds from left to right, PCRE will look for each "a" in the subject
4485 and then see if what follows matches the rest of the pattern. If the
4486 pattern is specified as
4487
4488 ^.*abcd$
4489
4490 the initial .* matches the entire string at first, but when this fails
4491 (because there is no following "a"), it backtracks to match all but the
4492 last character, then all but the last two characters, and so on. Once
4493 again the search for "a" covers the entire string, from right to left,
4494 so we are no better off. However, if the pattern is written as
4495
4496 ^.*+(?<=abcd)
4497
4498 there can be no backtracking for the .*+ item; it can match only the
4499 entire string. The subsequent lookbehind assertion does a single test
4500 on the last four characters. If it fails, the match fails immediately.
4501 For long strings, this approach makes a significant difference to the
4502 processing time.
4503
4504 Using multiple assertions
4505
4506 Several assertions (of any sort) may occur in succession. For example,
4507
4508 (?<=\d{3})(?<!999)foo
4509
4510 matches "foo" preceded by three digits that are not "999". Notice that
4511 each of the assertions is applied independently at the same point in
4512 the subject string. First there is a check that the previous three
4513 characters are all digits, and then there is a check that the same
4514 three characters are not "999". This pattern does not match "foo" pre-
4515 ceded by six characters, the first of which are digits and the last
4516 three of which are not "999". For example, it doesn't match "123abc-
4517 foo". A pattern to do that is
4518
4519 (?<=\d{3}...)(?<!999)foo
4520
4521 This time the first assertion looks at the preceding six characters,
4522 checking that the first three are digits, and then the second assertion
4523 checks that the preceding three characters are not "999".
4524
4525 Assertions can be nested in any combination. For example,
4526
4527 (?<=(?<!foo)bar)baz
4528
4529 matches an occurrence of "baz" that is preceded by "bar" which in turn
4530 is not preceded by "foo", while
4531
4532 (?<=\d{3}(?!999)...)foo
4533
4534 is another pattern that matches "foo" preceded by three digits and any
4535 three characters that are not "999".
4536
4537
4538 CONDITIONAL SUBPATTERNS
4539
4540 It is possible to cause the matching process to obey a subpattern con-
4541 ditionally or to choose between two alternative subpatterns, depending
4542 on the result of an assertion, or whether a previous capturing subpat-
4543 tern matched or not. The two possible forms of conditional subpattern
4544 are
4545
4546 (?(condition)yes-pattern)
4547 (?(condition)yes-pattern|no-pattern)
4548
4549 If the condition is satisfied, the yes-pattern is used; otherwise the
4550 no-pattern (if present) is used. If there are more than two alterna-
4551 tives in the subpattern, a compile-time error occurs.
4552
4553 There are four kinds of condition: references to subpatterns, refer-
4554 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4555
4556 Checking for a used subpattern by number
4557
4558 If the text between the parentheses consists of a sequence of digits,
4559 the condition is true if the capturing subpattern of that number has
4560 previously matched. An alternative notation is to precede the digits
4561 with a plus or minus sign. In this case, the subpattern number is rela-
4562 tive rather than absolute. The most recently opened parentheses can be
4563 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4564 looping constructs it can also make sense to refer to subsequent groups
4565 with constructs such as (?(+2).
4566
4567 Consider the following pattern, which contains non-significant white
4568 space to make it more readable (assume the PCRE_EXTENDED option) and to
4569 divide it into three parts for ease of discussion:
4570
4571 ( \( )? [^()]+ (?(1) \) )
4572
4573 The first part matches an optional opening parenthesis, and if that
4574 character is present, sets it as the first captured substring. The sec-
4575 ond part matches one or more characters that are not parentheses. The
4576 third part is a conditional subpattern that tests whether the first set
4577 of parentheses matched or not. If they did, that is, if subject started
4578 with an opening parenthesis, the condition is true, and so the yes-pat-
4579 tern is executed and a closing parenthesis is required. Otherwise,
4580 since no-pattern is not present, the subpattern matches nothing. In
4581 other words, this pattern matches a sequence of non-parentheses,
4582 optionally enclosed in parentheses.
4583
4584 If you were embedding this pattern in a larger one, you could use a
4585 relative reference:
4586
4587 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4588
4589 This makes the fragment independent of the parentheses in the larger
4590 pattern.
4591
4592 Checking for a used subpattern by name
4593
4594 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4595 used subpattern by name. For compatibility with earlier versions of
4596 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4597 also recognized. However, there is a possible ambiguity with this syn-
4598 tax, because subpattern names may consist entirely of digits. PCRE
4599 looks first for a named subpattern; if it cannot find one and the name
4600 consists entirely of digits, PCRE looks for a subpattern of that num-
4601 ber, which must be greater than zero. Using subpattern names that con-
4602 sist entirely of digits is not recommended.
4603
4604 Rewriting the above example to use a named subpattern gives this:
4605
4606 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4607
4608
4609 Checking for pattern recursion
4610
4611 If the condition is the string (R), and there is no subpattern with the
4612 name R, the condition is true if a recursive call to the whole pattern
4613 or any subpattern has been made. If digits or a name preceded by amper-
4614 sand follow the letter R, for example:
4615
4616 (?(R3)...) or (?(R&name)...)
4617
4618 the condition is true if the most recent recursion is into the subpat-
4619 tern whose number or name is given. This condition does not check the
4620 entire recursion stack.
4621
4622 At "top level", all these recursion test conditions are false. Recur-
4623 sive patterns are described below.
4624
4625 Defining subpatterns for use by reference only
4626
4627 If the condition is the string (DEFINE), and there is no subpattern
4628 with the name DEFINE, the condition is always false. In this case,
4629 there may be only one alternative in the subpattern. It is always
4630 skipped if control reaches this point in the pattern; the idea of
4631 DEFINE is that it can be used to define "subroutines" that can be ref-
4632 erenced from elsewhere. (The use of "subroutines" is described below.)
4633 For example, a pattern to match an IPv4 address could be written like
4634 this (ignore whitespace and line breaks):
4635
4636 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4637 \b (?&byte) (\.(?&byte)){3} \b
4638
4639 The first part of the pattern is a DEFINE group inside which a another
4640 group named "byte" is defined. This matches an individual component of
4641 an IPv4 address (a number less than 256). When matching takes place,
4642 this part of the pattern is skipped because DEFINE acts like a false
4643 condition.
4644
4645 The rest of the pattern uses references to the named group to match the
4646 four dot-separated components of an IPv4 address, insisting on a word
4647 boundary at each end.
4648
4649 Assertion conditions
4650
4651 If the condition is not in any of the above formats, it must be an
4652 assertion. This may be a positive or negative lookahead or lookbehind
4653 assertion. Consider this pattern, again containing non-significant
4654 white space, and with the two alternatives on the second line:
4655
4656 (?(?=[^a-z]*[a-z])
4657 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4658
4659 The condition is a positive lookahead assertion that matches an
4660 optional sequence of non-letters followed by a letter. In other words,
4661 it tests for the presence of at least one letter in the subject. If a
4662 letter is found, the subject is matched against the first alternative;
4663 otherwise it is matched against the second. This pattern matches
4664 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4665 letters and dd are digits.
4666
4667
4668 COMMENTS
4669
4670 The sequence (?# marks the start of a comment that continues up to the
4671 next closing parenthesis. Nested parentheses are not permitted. The
4672 characters that make up a comment play no part in the pattern matching
4673 at all.
4674
4675 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4676 character class introduces a comment that continues to immediately
4677 after the next newline in the pattern.
4678
4679
4680 RECURSIVE PATTERNS
4681
4682 Consider the problem of matching a string in parentheses, allowing for
4683 unlimited nested parentheses. Without the use of recursion, the best
4684 that can be done is to use a pattern that matches up to some fixed
4685 depth of nesting. It is not possible to handle an arbitrary nesting
4686 depth.
4687
4688 For some time, Perl has provided a facility that allows regular expres-
4689 sions to recurse (amongst other things). It does this by interpolating
4690 Perl code in the expression at run time, and the code can refer to the
4691 expression itself. A Perl pattern using code interpolation to solve the
4692 parentheses problem can be created like this:
4693
4694 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4695
4696 The (?p{...}) item interpolates Perl code at run time, and in this case
4697 refers recursively to the pattern in which it appears.
4698
4699 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4700 it supports special syntax for recursion of the entire pattern, and
4701 also for individual subpattern recursion. After its introduction in
4702 PCRE and Python, this kind of recursion was introduced into Perl at
4703 release 5.10.
4704
4705 A special item that consists of (? followed by a number greater than
4706 zero and a closing parenthesis is a recursive call of the subpattern of
4707 the given number, provided that it occurs inside that subpattern. (If
4708 not, it is a "subroutine" call, which is described in the next sec-
4709 tion.) The special item (?R) or (?0) is a recursive call of the entire
4710 regular expression.
4711
4712 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4713 always treated as an atomic group. That is, once it has matched some of
4714 the subject string, it is never re-entered, even if it contains untried
4715 alternatives and there is a subsequent matching failure.
4716
4717 This PCRE pattern solves the nested parentheses problem (assume the
4718 PCRE_EXTENDED option is set so that white space is ignored):
4719
4720 \( ( (?>[^()]+) | (?R) )* \)
4721
4722 First it matches an opening parenthesis. Then it matches any number of
4723 substrings which can either be a sequence of non-parentheses, or a
4724 recursive match of the pattern itself (that is, a correctly parenthe-
4725 sized substring). Finally there is a closing parenthesis.
4726
4727 If this were part of a larger pattern, you would not want to recurse
4728 the entire pattern, so instead you could use this:
4729
4730 ( \( ( (?>[^()]+) | (?1) )* \) )
4731
4732 We have put the pattern into parentheses, and caused the recursion to
4733 refer to them instead of the whole pattern.
4734
4735 In a larger pattern, keeping track of parenthesis numbers can be
4736 tricky. This is made easier by the use of relative references. (A Perl
4737 5.10 feature.) Instead of (?1) in the pattern above you can write
4738 (?-2) to refer to the second most recently opened parentheses preceding
4739 the recursion. In other words, a negative number counts capturing
4740 parentheses leftwards from the point at which it is encountered.
4741
4742 It is also possible to refer to subsequently opened parentheses, by
4743 writing references such as (?+2). However, these cannot be recursive
4744 because the reference is not inside the parentheses that are refer-
4745 enced. They are always "subroutine" calls, as described in the next
4746 section.
4747
4748 An alternative approach is to use named parentheses instead. The Perl
4749 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4750 supported. We could rewrite the above example as follows:
4751
4752 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4753
4754 If there is more than one subpattern with the same name, the earliest
4755 one is used.
4756
4757 This particular example pattern that we have been looking at contains
4758 nested unlimited repeats, and so the use of atomic grouping for match-
4759 ing strings of non-parentheses is important when applying the pattern
4760 to strings that do not match. For example, when this pattern is applied
4761 to
4762
4763 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4764
4765 it yields "no match" quickly. However, if atomic grouping is not used,
4766 the match runs for a very long time indeed because there are so many
4767 different ways the + and * repeats can carve up the subject, and all
4768 have to be tested before failure can be reported.
4769
4770 At the end of a match, the values set for any capturing subpatterns are
4771 those from the outermost level of the recursion at which the subpattern
4772 value is set. If you want to obtain intermediate values, a callout
4773 function can be used (see below and the pcrecallout documentation). If
4774 the pattern above is matched against
4775
4776 (ab(cd)ef)
4777
4778 the value for the capturing parentheses is "ef", which is the last
4779 value taken on at the top level. If additional parentheses are added,
4780 giving
4781
4782 \( ( ( (?>[^()]+) | (?R) )* ) \)
4783 ^ ^
4784 ^ ^
4785
4786 the string they capture is "ab(cd)ef", the contents of the top level
4787 parentheses. If there are more than 15 capturing parentheses in a pat-
4788 tern, PCRE has to obtain extra memory to store data during a recursion,
4789 which it does by using pcre_malloc, freeing it via pcre_free after-
4790 wards. If no memory can be obtained, the match fails with the
4791 PCRE_ERROR_NOMEMORY error.
4792
4793 Do not confuse the (?R) item with the condition (R), which tests for
4794 recursion. Consider this pattern, which matches text in angle brack-
4795 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4796 brackets (that is, when recursing), whereas any characters are permit-
4797 ted at the outer level.
4798
4799 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4800
4801 In this pattern, (?(R) is the start of a conditional subpattern, with
4802 two different alternatives for the recursive and non-recursive cases.
4803 The (?R) item is the actual recursive call.
4804
4805
4806 SUBPATTERNS AS SUBROUTINES
4807
4808 If the syntax for a recursive subpattern reference (either by number or
4809 by name) is used outside the parentheses to which it refers, it oper-
4810 ates like a subroutine in a programming language. The "called" subpat-
4811 tern may be defined before or after the reference. A numbered reference
4812 can be absolute or relative, as in these examples:
4813
4814 (...(absolute)...)...(?2)...
4815 (...(relative)...)...(?-1)...
4816 (...(?+1)...(relative)...
4817
4818 An earlier example pointed out that the pattern
4819
4820 (sens|respons)e and \1ibility
4821
4822 matches "sense and sensibility" and "response and responsibility", but
4823 not "sense and responsibility". If instead the pattern
4824
4825 (sens|respons)e and (?1)ibility
4826
4827 is used, it does match "sense and responsibility" as well as the other
4828 two strings. Another example is given in the discussion of DEFINE
4829 above.
4830
4831 Like recursive subpatterns, a "subroutine" call is always treated as an
4832 atomic group. That is, once it has matched some of the subject string,
4833 it is never re-entered, even if it contains untried alternatives and
4834 there is a subsequent matching failure.
4835
4836 When a subpattern is used as a subroutine, processing options such as
4837 case-independence are fixed when the subpattern is defined. They cannot
4838 be changed for different calls. For example, consider this pattern:
4839
4840 (abc)(?i:(?-1))
4841
4842 It matches "abcabc". It does not match "abcABC" because the change of
4843 processing option does not affect the called subpattern.
4844
4845
4846 ONIGURUMA SUBROUTINE SYNTAX
4847
4848 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4849 name or a number enclosed either in angle brackets or single quotes, is
4850 an alternative syntax for referencing a subpattern as a subroutine,
4851 possibly recursively. Here are two of the examples used above, rewrit-
4852 ten using this syntax:
4853
4854 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
4855 (sens|respons)e and \g'1'ibility
4856
4857 PCRE supports an extension to Oniguruma: if a number is preceded by a
4858 plus or a minus sign it is taken as a relative reference. For example:
4859
4860 (abc)(?i:\g<-1>)
4861
4862 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
4863 synonymous. The former is a back reference; the latter is a subroutine
4864 call.
4865
4866
4867 CALLOUTS
4868
4869 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4870 Perl code to be obeyed in the middle of matching a regular expression.
4871 This makes it possible, amongst other things, to extract different sub-
4872 strings that match the same pair of parentheses when there is a repeti-
4873 tion.
4874
4875 PCRE provides a similar feature, but of course it cannot obey arbitrary
4876 Perl code. The feature is called "callout". The caller of PCRE provides
4877 an external function by putting its entry point in the global variable
4878 pcre_callout. By default, this variable contains NULL, which disables
4879 all calling out.
4880
4881 Within a regular expression, (?C) indicates the points at which the
4882 external function is to be called. If you want to identify different
4883 callout points, you can put a number less than 256 after the letter C.
4884 The default value is zero. For example, this pattern has two callout
4885 points:
4886
4887 (?C1)abc(?C2)def
4888
4889 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4890 automatically installed before each item in the pattern. They are all
4891 numbered 255.
4892
4893 During matching, when PCRE reaches a callout point (and pcre_callout is
4894 set), the external function is called. It is provided with the number
4895 of the callout, the position in the pattern, and, optionally, one item
4896 of data originally supplied by the caller of pcre_exec(). The callout
4897 function may cause matching to proceed, to backtrack, or to fail alto-
4898 gether. A complete description of the interface to the callout function
4899 is given in the pcrecallout documentation.
4900
4901
4902 BACKTRACKING CONTROL
4903
4904 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4905 which are described in the Perl documentation as "experimental and sub-
4906 ject to change or removal in a future version of Perl". It goes on to
4907 say: "Their usage in production code should be noted to avoid problems
4908 during upgrades." The same remarks apply to the PCRE features described
4909 in this section.
4910
4911 Since these verbs are specifically related to backtracking, most of
4912 them can be used only when the pattern is to be matched using
4913 pcre_exec(), which uses a backtracking algorithm. With the exception of
4914 (*FAIL), which behaves like a failing negative assertion, they cause an
4915 error if encountered by pcre_dfa_exec().
4916
4917 The new verbs make use of what was previously invalid syntax: an open-
4918 ing parenthesis followed by an asterisk. In Perl, they are generally of
4919 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4920 its general form is just (*VERB). Any number of these verbs may occur
4921 in a pattern. There are two kinds:
4922
4923 Verbs that act immediately
4924
4925 The following verbs act as soon as they are encountered:
4926
4927 (*ACCEPT)
4928
4929 This verb causes the match to end successfully, skipping the remainder
4930 of the pattern. When inside a recursion, only the innermost pattern is
4931 ended immediately. PCRE differs from Perl in what happens if the
4932 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4933 captured: in PCRE no data is captured. For example:
4934
4935 A(A|B(*ACCEPT)|C)D
4936
4937 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4938 is captured.
4939
4940 (*FAIL) or (*F)
4941
4942 This verb causes the match to fail, forcing backtracking to occur. It
4943 is equivalent to (?!) but easier to read. The Perl documentation notes
4944 that it is probably useful only when combined with (?{}) or (??{}).
4945 Those are, of course, Perl features that are not present in PCRE. The
4946 nearest equivalent is the callout feature, as for example in this pat-
4947 tern:
4948
4949 a+(?C)(*FAIL)
4950
4951 A match with the string "aaaa" always fails, but the callout is taken
4952 before each backtrack happens (in this example, 10 times).
4953
4954 Verbs that act after backtracking
4955
4956 The following verbs do nothing when they are encountered. Matching con-
4957 tinues with what follows, but if there is no subsequent match, a fail-
4958 ure is forced. The verbs differ in exactly what kind of failure
4959 occurs.
4960
4961 (*COMMIT)
4962
4963 This verb causes the whole match to fail outright if the rest of the
4964 pattern does not match. Even if the pattern is unanchored, no further
4965 attempts to find a match by advancing the start point take place. Once
4966 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4967 at the current starting point, or not at all. For example:
4968
4969 a+(*COMMIT)b
4970
4971 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4972 of dynamic anchor, or "I've started, so I must finish."
4973
4974 (*PRUNE)
4975
4976 This verb causes the match to fail at the current position if the rest
4977 of the pattern does not match. If the pattern is unanchored, the normal
4978 "bumpalong" advance to the next starting character then happens. Back-
4979 tracking can occur as usual to the left of (*PRUNE), or when matching
4980 to the right of (*PRUNE), but if there is no match to the right, back-
4981 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4982 is just an alternative to an atomic group or possessive quantifier, but
4983 there are some uses of (*PRUNE) that cannot be expressed in any other
4984 way.
4985
4986 (*SKIP)
4987
4988 This verb is like (*PRUNE), except that if the pattern is unanchored,
4989 the "bumpalong" advance is not to the next character, but to the posi-
4990 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4991 that whatever text was matched leading up to it cannot be part of a
4992 successful match. Consider:
4993
4994 a+(*SKIP)b
4995
4996 If the subject is "aaaac...", after the first match attempt fails
4997 (starting at the first character in the string), the starting point
4998 skips on to start the next attempt at "c". Note that a possessive quan-
4999 tifer does not have the same effect in this example; although it would
5000 suppress backtracking during the first match attempt, the second
5001 attempt would start at the second character instead of skipping on to
5002 "c".
5003
5004 (*THEN)
5005
5006 This verb causes a skip to the next alternation if the rest of the pat-
5007 tern does not match. That is, it cancels pending backtracking, but only
5008 within the current alternation. Its name comes from the observation
5009 that it can be used for a pattern-based if-then-else block:
5010
5011 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5012
5013 If the COND1 pattern matches, FOO is tried (and possibly further items
5014 after the end of the group if FOO succeeds); on failure the matcher
5015 skips to the second alternative and tries COND2, without backtracking
5016 into COND1. If (*THEN) is used outside of any alternation, it acts
5017 exactly like (*PRUNE).
5018
5019
5020 SEE ALSO
5021
5022 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5023
5024
5025 AUTHOR
5026
5027 Philip Hazel
5028 University Computing Service
5029 Cambridge CB2 3QH, England.
5030
5031
5032 REVISION
5033
5034 Last updated: 11 April 2009
5035 Copyright (c) 1997-2009 University of Cambridge.
5036 ------------------------------------------------------------------------------
5037
5038
5039 PCRESYNTAX(3) PCRESYNTAX(3)
5040
5041
5042 NAME
5043 PCRE - Perl-compatible regular expressions
5044
5045
5046 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5047
5048 The full syntax and semantics of the regular expressions that are sup-
5049 ported by PCRE are described in the pcrepattern documentation. This
5050 document contains just a quick-reference summary of the syntax.
5051
5052
5053 QUOTING
5054
5055 \x where x is non-alphanumeric is a literal x
5056 \Q...\E treat enclosed characters as literal
5057
5058
5059 CHARACTERS
5060
5061 \a alarm, that is, the BEL character (hex 07)
5062 \cx "control-x", where x is any character
5063 \e escape (hex 1B)
5064 \f formfeed (hex 0C)
5065 \n newline (hex 0A)
5066 \r carriage return (hex 0D)
5067 \t tab (hex 09)
5068 \ddd character with octal code ddd, or backreference
5069 \xhh character with hex code hh
5070 \x{hhh..} character with hex code hhh..
5071
5072
5073 CHARACTER TYPES
5074
5075 . any character except newline;
5076 in dotall mode, any character whatsoever
5077 \C one byte, even in UTF-8 mode (best avoided)
5078 \d a decimal digit
5079 \D a character that is not a decimal digit
5080 \h a horizontal whitespace character
5081 \H a character that is not a horizontal whitespace character
5082 \p{xx} a character with the xx property
5083 \P{xx} a character without the xx property
5084 \R a newline sequence
5085 \s a whitespace character
5086 \S a character that is not a whitespace character
5087 \v a vertical whitespace character
5088 \V a character that is not a vertical whitespace character
5089 \w a "word" character
5090 \W a "non-word" character
5091 \X an extended Unicode sequence
5092
5093 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
5094
5095
5096 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
5097
5098 C Other
5099 Cc Control
5100 Cf Format
5101 Cn Unassigned
5102 Co Private use
5103 Cs Surrogate
5104
5105 L Letter
5106 Ll Lower case letter
5107 Lm Modifier letter
5108 Lo Other letter
5109 Lt Title case letter
5110 Lu Upper case letter
5111 L& Ll, Lu, or Lt
5112
5113 M Mark
5114 Mc Spacing mark
5115 Me Enclosing mark
5116 Mn Non-spacing mark
5117
5118 N Number
5119 Nd Decimal number
5120 Nl Letter number
5121 No Other number
5122
5123 P Punctuation
5124 Pc Connector punctuation
5125 Pd Dash punctuation
5126 Pe Close punctuation
5127 Pf Final punctuation
5128 Pi Initial punctuation
5129 Po Other punctuation
5130 Ps Open punctuation
5131
5132 S Symbol
5133 Sc Currency symbol
5134 Sk Modifier symbol
5135 Sm Mathematical symbol
5136 So Other symbol
5137
5138 Z Separator
5139 Zl Line separator
5140 Zp Paragraph separator
5141 Zs Space separator
5142
5143
5144 SCRIPT NAMES FOR \p AND \P
5145
5146 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
5147 Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common, Coptic, Cu-
5148 neiform, Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian,
5149 Glagolitic, Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo,
5150 Hebrew, Hiragana, Inherited, Kannada, Katakana, Kayah_Li, Kharoshthi,
5151 Khmer, Lao, Latin, Lepcha, Limbu, Linear_B, Lycian, Lydian, Malayalam,
5152 Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic, Old_Persian,
5153 Ol_Chiki, Oriya, Osmanya, Phags_Pa, Phoenician, Rejang, Runic, Saurash-
5154 tra, Shavian, Sinhala, Sudanese, Syloti_Nagri, Syriac, Tagalog, Tag-
5155 banwa, Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5156 Ugaritic, Vai, Yi.
5157
5158
5159 CHARACTER CLASSES
5160
5161 [...] positive character class
5162 [^...] negative character class
5163 [x-y] range (can be used for hex characters)
5164 [[:xxx:]] positive POSIX named set
5165 [[:^xxx:]] negative POSIX named set
5166
5167 alnum alphanumeric
5168 alpha alphabetic
5169 ascii 0-127
5170 blank space or tab
5171 cntrl control character
5172 digit decimal digit
5173 graph printing, excluding space
5174 lower lower case letter
5175 print printing, including space
5176 punct printing, excluding alphanumeric
5177 space whitespace
5178 upper upper case letter
5179 word same as \w
5180 xdigit hexadecimal digit
5181
5182 In PCRE, POSIX character set names recognize only ASCII characters. You
5183 can use \Q...\E inside a character class.
5184
5185
5186 QUANTIFIERS
5187
5188 ? 0 or 1, greedy
5189 ?+ 0 or 1, possessive
5190 ?? 0 or 1, lazy
5191 * 0 or more, greedy
5192 *+ 0 or more, possessive
5193 *? 0 or more, lazy
5194 + 1 or more, greedy
5195 ++ 1 or more, possessive
5196 +? 1 or more, lazy
5197 {n} exactly n
5198 {n,m} at least n, no more than m, greedy
5199 {n,m}+ at least n, no more than m, possessive
5200 {n,m}? at least n, no more than m, lazy
5201 {n,} n or more, greedy
5202 {n,}+ n or more, possessive
5203 {n,}? n or more, lazy
5204
5205
5206 ANCHORS AND SIMPLE ASSERTIONS
5207
5208 \b word boundary (only ASCII letters recognized)
5209 \B not a word boundary
5210 ^ start of subject
5211 also after internal newline in multiline mode
5212 \A start of subject
5213 $ end of subject
5214 also before newline at end of subject
5215 also before internal newline in multiline mode
5216 \Z end of subject
5217 also before newline at end of subject
5218 \z end of subject
5219 \G first matching position in subject
5220
5221
5222 MATCH POINT RESET
5223
5224 \K reset start of match
5225
5226
5227 ALTERNATION
5228
5229 expr|expr|expr...
5230
5231
5232 CAPTURING
5233
5234 (...) capturing group
5235 (?<name>...) named capturing group (Perl)
5236 (?'name'...) named capturing group (Perl)
5237 (?P<name>...) named capturing group (Python)
5238 (?:...) non-capturing group
5239 (?|...) non-capturing group; reset group numbers for
5240 capturing groups in each alternative
5241
5242
5243 ATOMIC GROUPS
5244
5245 (?>...) atomic, non-capturing group
5246
5247
5248 COMMENT
5249
5250 (?#....) comment (not nestable)
5251
5252
5253 OPTION SETTING
5254
5255 (?i) caseless
5256 (?J) allow duplicate names
5257 (?m) multiline
5258 (?s) single line (dotall)
5259 (?U) default ungreedy (lazy)
5260 (?x) extended (ignore white space)
5261 (?-...) unset option(s)
5262
5263 The following is recognized only at the start of a pattern or after one
5264 of the newline-setting options with similar syntax:
5265
5266 (*UTF8) set UTF-8 mode
5267
5268
5269 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5270
5271 (?=...) positive look ahead
5272 (?!...) negative look ahead
5273 (?<=...) positive look behind
5274 (?<!...) negative look behind
5275
5276 Each top-level branch of a look behind must be of a fixed length.
5277
5278
5279 BACKREFERENCES
5280
5281 \n reference by number (can be ambiguous)
5282 \gn reference by number
5283 \g{n} reference by number
5284 \g{-n} relative reference by number
5285 \k<name> reference by name (Perl)
5286 \k'name' reference by name (Perl)
5287 \g{name} reference by name (Perl)
5288 \k{name} reference by name (.NET)
5289 (?P=name) reference by name (Python)
5290
5291
5292 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5293
5294 (?R) recurse whole pattern
5295 (?n) call subpattern by absolute number
5296 (?+n) call subpattern by relative number
5297 (?-n) call subpattern by relative number
5298 (?&name) call subpattern by name (Perl)
5299 (?P>name) call subpattern by name (Python)
5300 \g<name> call subpattern by name (Oniguruma)
5301 \g'name' call subpattern by name (Oniguruma)
5302 \g<n> call subpattern by absolute number (Oniguruma)
5303 \g'n' call subpattern by absolute number (Oniguruma)
5304 \g<+n> call subpattern by relative number (PCRE extension)
5305 \g'+n' call subpattern by relative number (PCRE extension)
5306 \g<-n> call subpattern by relative number (PCRE extension)
5307 \g'-n' call subpattern by relative number (PCRE extension)
5308
5309
5310 CONDITIONAL PATTERNS
5311
5312 (?(condition)yes-pattern)
5313 (?(condition)yes-pattern|no-pattern)
5314
5315 (?(n)... absolute reference condition
5316 (?(+n)... relative reference condition
5317 (?(-n)... relative reference condition
5318 (?(<name>)... named reference condition (Perl)
5319 (?('name')... named reference condition (Perl)
5320 (?(name)... named reference condition (PCRE)
5321 (?(R)... overall recursion condition
5322 (?(Rn)... specific group recursion condition
5323 (?(R&name)... specific recursion condition
5324 (?(DEFINE)... define subpattern for reference
5325 (?(assert)... assertion condition
5326
5327
5328 BACKTRACKING CONTROL
5329
5330 The following act immediately they are reached:
5331
5332 (*ACCEPT) force successful match
5333 (*FAIL) force backtrack; synonym (*F)
5334
5335 The following act only when a subsequent match failure causes a back-
5336 track to reach them. They all force a match failure, but they differ in
5337 what happens afterwards. Those that advance the start-of-match point do
5338 so only if the pattern is not anchored.
5339
5340 (*COMMIT) overall failure, no advance of starting point
5341 (*PRUNE) advance to next starting character
5342 (*SKIP) advance start to current matching position
5343 (*THEN) local failure, backtrack to next alternation
5344
5345
5346 NEWLINE CONVENTIONS
5347
5348 These are recognized only at the very start of the pattern or after a
5349 (*BSR_...) or (*UTF8) option.
5350
5351 (*CR) carriage return only
5352 (*LF) linefeed only
5353 (*CRLF) carriage return followed by linefeed
5354 (*ANYCRLF) all three of the above
5355 (*ANY) any Unicode newline sequence
5356
5357
5358 WHAT \R MATCHES
5359
5360 These are recognized only at the very start of the pattern or after a
5361 (*...) option that sets the newline convention or UTF-8 mode.
5362
5363 (*BSR_ANYCRLF) CR, LF, or CRLF
5364 (*BSR_UNICODE) any Unicode newline sequence
5365
5366
5367 CALLOUTS
5368
5369 (?C) callout
5370 (?Cn) callout with data n
5371
5372
5373 SEE ALSO
5374
5375 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5376
5377
5378 AUTHOR
5379
5380 Philip Hazel
5381 University Computing Service
5382 Cambridge CB2 3QH, England.
5383
5384
5385 REVISION
5386
5387 Last updated: 11 April 2009
5388 Copyright (c) 1997-2009 University of Cambridge.
5389 ------------------------------------------------------------------------------
5390
5391
5392 PCREPARTIAL(3) PCREPARTIAL(3)
5393
5394
5395 NAME
5396 PCRE - Perl-compatible regular expressions
5397
5398
5399 PARTIAL MATCHING IN PCRE
5400
5401 In normal use of PCRE, if the subject string that is passed to
5402 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5403 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5404 There are circumstances where it might be helpful to distinguish this
5405 case from other cases in which there is no match.
5406
5407 Consider, for example, an application where a human is required to type
5408 in data for a field with specific formatting requirements. An example
5409 might be a date in the form ddmmmyy, defined by this pattern:
5410
5411 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5412
5413 If the application sees the user's keystrokes one by one, and can check
5414 that what has been typed so far is potentially valid, it is able to
5415 raise an error as soon as a mistake is made, possibly beeping and not
5416 reflecting the character that has been typed. This immediate feedback
5417 is likely to be a better user interface than a check that is delayed
5418 until the entire string has been entered.
5419
5420 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5421 TIAL option, which can be set when calling pcre_exec() or
5422 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5423 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5424 during the matching process the last part of the subject string matched
5425 part of the pattern. Unfortunately, for non-anchored matching, it is
5426 not possible to obtain the position of the start of the partial match.
5427 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5428
5429 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5430 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5431 the subject is reached, there have been no complete matches, but there
5432 is still at least one matching possibility. The portion of the string
5433 that provided the partial match is set as the first matching string.
5434
5435 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5436 the last literal byte in a pattern, and abandons matching immediately
5437 if such a byte is not present in the subject string. This optimization
5438 cannot be used for a subject string that might match only partially.
5439
5440
5441 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5442
5443 Because of the way certain internal optimizations are implemented in
5444 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5445 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5446 used. For pcre_exec(), repeated single characters such as
5447
5448 a{2,4}
5449
5450 and repeated single metasequences such as
5451
5452 \d+
5453
5454 are not permitted if the maximum number of occurrences is greater than
5455 one. Optional items such as \d? (where the maximum is one) are permit-
5456 ted. Quantifiers with any values are permitted after parentheses, so
5457 the invalid examples above can be coded thus:
5458
5459 (a){2,4}
5460 (\d)+
5461
5462 These constructions run more slowly, but for the kinds of application
5463 that are envisaged for this facility, this is not felt to be a major
5464 restriction.
5465
5466 If PCRE_PARTIAL is set for a pattern that does not conform to the
5467 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5468 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5469 find out if a compiled pattern can be used for partial matching.
5470
5471
5472 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5473
5474 If the escape sequence \P is present in a pcretest data line, the
5475 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5476 uses the date example quoted above:
5477
5478 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5479 data> 25jun04\P
5480 0: 25jun04
5481 1: jun
5482 data> 25dec3\P
5483 Partial match
5484 data> 3ju\P
5485 Partial match
5486 data> 3juj\P
5487 No match
5488 data> j\P
5489 No match
5490
5491 The first data string is matched completely, so pcretest shows the
5492 matched substrings. The remaining four strings do not match the com-
5493 plete pattern, but the first two are partial matches. The same test,
5494 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5495 produces the following output:
5496
5497 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5498 data> 25jun04\P\D
5499 0: 25jun04
5500 data> 23dec3\P\D
5501 Partial match: 23dec3
5502 data> 3ju\P\D
5503 Partial match: 3ju
5504 data> 3juj\P\D
5505 No match
5506 data> j\P\D
5507 No match
5508
5509 Notice that in this case the portion of the string that was matched is
5510 made available.
5511
5512
5513 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5514
5515 When a partial match has been found using pcre_dfa_exec(), it is possi-
5516 ble to continue the match by providing additional subject data and
5517 calling pcre_dfa_exec() again with the same compiled regular expres-
5518 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5519 the same working space as before, because this is where details of the
5520 previous partial match are stored. Here is an example using pcretest,
5521 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5522 \D are as above):
5523
5524 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5525 data> 23ja\P\D
5526 Partial match: 23ja
5527 data> n05\R\D
5528 0: n05
5529
5530 The first call has "23ja" as the subject, and requests partial match-
5531 ing; the second call has "n05" as the subject for the continued
5532 (restarted) match. Notice that when the match is complete, only the
5533 last part is shown; PCRE does not retain the previously partially-
5534 matched string. It is up to the calling program to do that if it needs
5535 to.
5536
5537 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5538 matching over multiple segments. This facility can be used to pass very
5539 long subject strings to pcre_dfa_exec(). However, some care is needed
5540 for certain types of pattern.
5541
5542 1. If the pattern contains tests for the beginning or end of a line,
5543 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5544 ate, when the subject string for any call does not contain the begin-
5545 ning or end of a line.
5546
5547 2. If the pattern contains backward assertions (including \b or \B),
5548 you need to arrange for some overlap in the subject strings to allow
5549 for this. For example, you could pass the subject in chunks that are
5550 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5551 set to 200 and the previous 200 bytes at the start of the buffer.
5552
5553 3. Matching a subject string that is split into multiple segments does
5554 not always produce exactly the same result as matching over one single
5555 long string. The difference arises when there are multiple matching
5556 possibilities, because a partial match result is given only when there
5557 are no completed matches in a call to pcre_dfa_exec(). This means that
5558 as soon as the shortest match has been found, continuation to a new
5559 subject segment is no longer possible. Consider this pcretest example:
5560
5561 re> /dog(sbody)?/
5562 data> do\P\D
5563 Partial match: do
5564 data> gsb\R\P\D
5565 0: g
5566 data> dogsbody\D
5567 0: dogsbody
5568 1: dog
5569
5570 The pattern matches the words "dog" or "dogsbody". When the subject is
5571 presented in several parts ("do" and "gsb" being the first two) the
5572 match stops when "dog" has been found, and it is not possible to con-
5573 tinue. On the other hand, if "dogsbody" is presented as a single
5574 string, both matches are found.
5575
5576 Because of this phenomenon, it does not usually make sense to end a
5577 pattern that is going to be matched in this way with a variable repeat.
5578
5579 4. Patterns that contain alternatives at the top level which do not all
5580 start with the same pattern item may not work as expected. For example,
5581 consider this pattern:
5582
5583 1234|3789
5584
5585 If the first part of the subject is "ABC123", a partial match of the
5586 first alternative is found at offset 3. There is no partial match for
5587 the second alternative, because such a match does not start at the same
5588 point in the subject string. Attempting to continue with the string
5589 "789" does not yield a match because only those alternatives that match
5590 at one point in the subject are remembered. The problem arises because
5591 the start of the second alternative matches within the first alterna-
5592 tive. There is no problem with anchored patterns or patterns such as:
5593
5594 1234|ABCD
5595
5596 where no string can be a partial match for both alternatives.
5597
5598
5599 AUTHOR
5600
5601 Philip Hazel
5602 University Computing Service
5603 Cambridge CB2 3QH, England.
5604
5605
5606 REVISION
5607
5608 Last updated: 04 June 2007
5609 Copyright (c) 1997-2007 University of Cambridge.
5610 ------------------------------------------------------------------------------
5611
5612
5613 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5614
5615
5616 NAME
5617 PCRE - Perl-compatible regular expressions
5618
5619
5620 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5621
5622 If you are running an application that uses a large number of regular
5623 expression patterns, it may be useful to store them in a precompiled
5624 form instead of having to compile them every time the application is
5625 run. If you are not using any private character tables (see the
5626 pcre_maketables() documentation), this is relatively straightforward.
5627 If you are using private tables, it is a little bit more complicated.
5628
5629 If you save compiled patterns to a file, you can copy them to a differ-
5630 ent host and run them there. This works even if the new host has the
5631 opposite endianness to the one on which the patterns were compiled.
5632 There may be a small performance penalty, but it should be insignifi-
5633 cant. However, compiling regular expressions with one version of PCRE
5634 for use with a different version is not guaranteed to work and may
5635 cause crashes.
5636
5637
5638 SAVING A COMPILED PATTERN
5639 The value returned by pcre_compile() points to a single block of memory
5640 that holds the compiled pattern and associated data. You can find the
5641 length of this block in bytes by calling pcre_fullinfo() with an argu-
5642 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5643 manner. Here is sample code that compiles a pattern and writes it to a
5644 file. It assumes that the variable fd refers to a file that is open for
5645 output:
5646
5647 int erroroffset, rc, size;
5648 char *error;
5649 pcre *re;
5650
5651 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5652 if (re == NULL) { ... handle errors ... }
5653 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5654 if (rc < 0) { ... handle errors ... }
5655 rc = fwrite(re, 1, size, fd);
5656 if (rc != size) { ... handle errors ... }
5657
5658 In this example, the bytes that comprise the compiled pattern are
5659 copied exactly. Note that this is binary data that may contain any of
5660 the 256 possible byte values. On systems that make a distinction
5661 between binary and non-binary data, be sure that the file is opened for
5662 binary output.
5663
5664 If you want to write more than one pattern to a file, you will have to
5665 devise a way of separating them. For binary data, preceding each pat-
5666 tern with its length is probably the most straightforward approach.
5667 Another possibility is to write out the data in hexadecimal instead of
5668 binary, one pattern to a line.
5669
5670 Saving compiled patterns in a file is only one possible way of storing
5671 them for later use. They could equally well be saved in a database, or
5672 in the memory of some daemon process that passes them via sockets to
5673 the processes that want them.
5674
5675 If the pattern has been studied, it is also possible to save the study
5676 data in a similar way to the compiled pattern itself. When studying
5677 generates additional information, pcre_study() returns a pointer to a
5678 pcre_extra data block. Its format is defined in the section on matching
5679 a pattern in the pcreapi documentation. The study_data field points to
5680 the binary study data, and this is what you must save (not the
5681 pcre_extra block itself). The length of the study data can be obtained
5682 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5683 Remember to check that pcre_study() did return a non-NULL value before
5684 trying to save the study data.
5685
5686
5687 RE-USING A PRECOMPILED PATTERN
5688
5689 Re-using a precompiled pattern is straightforward. Having reloaded it
5690 into main memory, you pass its pointer to pcre_exec() or
5691 pcre_dfa_exec() in the usual way. This should work even on another
5692 host, and even if that host has the opposite endianness to the one
5693 where the pattern was compiled.
5694
5695 However, if you passed a pointer to custom character tables when the
5696 pattern was compiled (the tableptr argument of pcre_compile()), you
5697 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5698 because the value saved with the compiled pattern will obviously be
5699 nonsense. A field in a pcre_extra() block is used to pass this data, as
5700 described in the section on matching a pattern in the pcreapi documen-
5701 tation.
5702
5703 If you did not provide custom character tables when the pattern was
5704 compiled, the pointer in the compiled pattern is NULL, which causes
5705 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5706 take any special action at run time in this case.
5707
5708 If you saved study data with the compiled pattern, you need to create
5709 your own pcre_extra data block and set the study_data field to point to
5710 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5711 bit in the flags field to indicate that study data is present. Then
5712 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5713 usual way.
5714
5715
5716 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5717
5718 In general, it is safest to recompile all saved patterns when you
5719 update to a new PCRE release, though not all updates actually require
5720 this. Recompiling is definitely needed for release 7.2.
5721
5722
5723 AUTHOR
5724
5725 Philip Hazel
5726 University Computing Service
5727 Cambridge CB2 3QH, England.
5728
5729
5730 REVISION
5731
5732 Last updated: 13 June 2007
5733 Copyright (c) 1997-2007 University of Cambridge.
5734 ------------------------------------------------------------------------------
5735
5736
5737 PCREPERFORM(3) PCREPERFORM(3)
5738
5739
5740 NAME
5741 PCRE - Perl-compatible regular expressions
5742
5743
5744 PCRE PERFORMANCE
5745
5746 Two aspects of performance are discussed below: memory usage and pro-
5747 cessing time. The way you express your pattern as a regular expression
5748 can affect both of them.
5749
5750
5751 MEMORY USAGE
5752
5753 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5754 that most simple patterns do not use much memory. However, there is one
5755 case where memory usage can be unexpectedly large. When a parenthesized
5756 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5757 ited maximum, the whole subpattern is repeated in the compiled code.
5758 For example, the pattern
5759
5760 (abc|def){2,4}
5761
5762 is compiled as if it were
5763
5764 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5765
5766 (Technical aside: It is done this way so that backtrack points within
5767 each of the repetitions can be independently maintained.)
5768
5769 For regular expressions whose quantifiers use only small numbers, this
5770 is not usually a problem. However, if the numbers are large, and par-
5771 ticularly if such repetitions are nested, the memory usage can become
5772 an embarrassment. For example, the very simple pattern
5773
5774 ((ab){1,1000}c){1,3}
5775
5776 uses 51K bytes when compiled. When PCRE is compiled with its default
5777 internal pointer size of two bytes, the size limit on a compiled pat-
5778 tern is 64K, and this is reached with the above pattern if the outer
5779 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5780 internal pointers and thus handle larger compiled patterns, but it is
5781 better to try to rewrite your pattern to use less memory if you can.
5782
5783 One way of reducing the memory usage for such patterns is to make use
5784 of PCRE's "subroutine" facility. Re-writing the above pattern as
5785
5786 ((ab)(?2){0,999}c)(?1){0,2}
5787
5788 reduces the memory requirements to 18K, and indeed it remains under 20K
5789 even with the outer repetition increased to 100. However, this pattern
5790 is not exactly equivalent, because the "subroutine" calls are treated
5791 as atomic groups into which there can be no backtracking if there is a
5792 subsequent matching failure. Therefore, PCRE cannot do this kind of
5793 rewriting automatically. Furthermore, there is a noticeable loss of
5794 speed when executing the modified pattern. Nevertheless, if the atomic
5795 grouping is not a problem and the loss of speed is acceptable, this
5796 kind of rewriting will allow you to process patterns that PCRE cannot
5797 otherwise handle.
5798
5799
5800 PROCESSING TIME
5801
5802 Certain items in regular expression patterns are processed more effi-
5803 ciently than others. It is more efficient to use a character class like
5804 [aeiou] than a set of single-character alternatives such as
5805 (a|e|i|o|u). In general, the simplest construction that provides the
5806 required behaviour is usually the most efficient. Jeffrey Friedl's book
5807 contains a lot of useful general discussion about optimizing regular
5808 expressions for efficient performance. This document contains a few
5809 observations about PCRE.
5810
5811 Using Unicode character properties (the \p, \P, and \X escapes) is
5812 slow, because PCRE has to scan a structure that contains data for over
5813 fifteen thousand characters whenever it needs a character's property.
5814 If you can find an alternative pattern that does not use character
5815 properties, it will probably be faster.
5816
5817 When a pattern begins with .* not in parentheses, or in parentheses
5818 that are not the subject of a backreference, and the PCRE_DOTALL option
5819 is set, the pattern is implicitly anchored by PCRE, since it can match
5820 only at the start of a subject string. However, if PCRE_DOTALL is not
5821 set, PCRE cannot make this optimization, because the . metacharacter
5822 does not then match a newline, and if the subject string contains new-
5823 lines, the pattern may match from the character immediately following
5824 one of them instead of from the very start. For example, the pattern
5825
5826 .*second
5827
5828 matches the subject "first\nand second" (where \n stands for a newline
5829 character), with the match starting at the seventh character. In order
5830 to do this, PCRE has to retry the match starting after every newline in
5831 the subject.
5832
5833 If you are using such a pattern with subject strings that do not con-
5834 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5835 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5836 ing. That saves PCRE from having to scan along the subject looking for
5837 a newline to restart at.
5838
5839 Beware of patterns that contain nested indefinite repeats. These can
5840 take a long time to run when applied to a string that does not match.
5841 Consider the pattern fragment
5842
5843 ^(a+)*
5844
5845 This can match "aaaa" in 16 different ways, and this number increases
5846 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5847 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5848 repeats can match different numbers of times.) When the remainder of
5849 the pattern is such that the entire match is going to fail, PCRE has in
5850 principle to try every possible variation, and this can take an
5851 extremely long time, even for relatively short strings.
5852
5853 An optimization catches some of the more simple cases such as
5854
5855 (a+)*b
5856
5857 where a literal character follows. Before embarking on the standard
5858 matching procedure, PCRE checks that there is a "b" later in the sub-
5859 ject string, and if there is not, it fails the match immediately. How-
5860 ever, when there is no following literal this optimization cannot be
5861 used. You can see the difference by comparing the behaviour of
5862
5863 (a+)*\d
5864
5865 with the pattern above. The former gives a failure almost instantly
5866 when applied to a whole line of "a" characters, whereas the latter
5867 takes an appreciable time with strings longer than about 20 characters.
5868
5869 In many cases, the solution to this kind of performance issue is to use
5870 an atomic group or a possessive quantifier.
5871
5872
5873 AUTHOR
5874
5875 Philip Hazel
5876 University Computing Service
5877 Cambridge CB2 3QH, England.
5878
5879
5880 REVISION
5881
5882 Last updated: 06 March 2007
5883 Copyright (c) 1997-2007 University of Cambridge.
5884 ------------------------------------------------------------------------------
5885
5886
5887 PCREPOSIX(3) PCREPOSIX(3)
5888
5889
5890 NAME
5891 PCRE - Perl-compatible regular expressions.
5892
5893
5894 SYNOPSIS OF POSIX API
5895
5896 #include <pcreposix.h>
5897
5898 int regcomp(regex_t *preg, const char *pattern,
5899 int cflags);
5900
5901 int regexec(regex_t *preg, const char *string,
5902 size_t nmatch, regmatch_t pmatch[], int eflags);
5903
5904 size_t regerror(int errcode, const regex_t *preg,
5905 char *errbuf, size_t errbuf_size);
5906
5907 void regfree(regex_t *preg);
5908
5909
5910 DESCRIPTION
5911
5912 This set of functions provides a POSIX-style API to the PCRE regular
5913 expression package. See the pcreapi documentation for a description of
5914 PCRE's native API, which contains much additional functionality.
5915
5916 The functions described here are just wrapper functions that ultimately
5917 call the PCRE native API. Their prototypes are defined in the
5918 pcreposix.h header file, and on Unix systems the library itself is
5919 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5920 command for linking an application that uses them. Because the POSIX
5921 functions call the native ones, it is also necessary to add -lpcre.
5922
5923 I have implemented only those POSIX option bits that can be reasonably
5924 mapped to PCRE native options. In addition, the option REG_EXTENDED is
5925 defined with the value zero. This has no effect, but since programs
5926 that are written to the POSIX interface often use it, this makes it
5927 easier to slot in PCRE as a replacement library. Other POSIX options
5928 are not even defined.
5929
5930 When PCRE is called via these functions, it is only the API that is
5931 POSIX-like in style. The syntax and semantics of the regular expres-
5932 sions themselves are still those of Perl, subject to the setting of
5933 various PCRE options, as described below. "POSIX-like in style" means
5934 that the API approximates to the POSIX definition; it is not fully
5935 POSIX-compatible, and in multi-byte encoding domains it is probably
5936 even less compatible.
5937
5938 The header for these functions is supplied as pcreposix.h to avoid any
5939 potential clash with other POSIX libraries. It can, of course, be
5940 renamed or aliased as regex.h, which is the "correct" name. It provides
5941 two structure types, regex_t for compiled internal forms, and reg-
5942 match_t for returning captured substrings. It also defines some con-
5943 stants whose names start with "REG_"; these are used for setting
5944 options and identifying error codes.
5945
5946
5947 COMPILING A PATTERN
5948
5949 The function regcomp() is called to compile a pattern into an internal
5950 form. The pattern is a C string terminated by a binary zero, and is
5951 passed in the argument pattern. The preg argument is a pointer to a
5952 regex_t structure that is used as a base for storing information about
5953 the compiled regular expression.
5954
5955 The argument cflags is either zero, or contains one or more of the bits
5956 defined by the following macros:
5957
5958 REG_DOTALL
5959
5960 The PCRE_DOTALL option is set when the regular expression is passed for
5961 compilation to the native function. Note that REG_DOTALL is not part of
5962 the POSIX standard.
5963
5964 REG_ICASE
5965
5966 The PCRE_CASELESS option is set when the regular expression is passed
5967 for compilation to the native function.
5968
5969 REG_NEWLINE
5970
5971 The PCRE_MULTILINE option is set when the regular expression is passed
5972 for compilation to the native function. Note that this does not mimic
5973 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5974 tion).
5975
5976 REG_NOSUB
5977
5978 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5979 passed for compilation to the native function. In addition, when a pat-
5980 tern that is compiled with this flag is passed to regexec() for match-
5981 ing, the nmatch and pmatch arguments are ignored, and no captured
5982 strings are returned.
5983
5984 REG_UTF8
5985
5986 The PCRE_UTF8 option is set when the regular expression is passed for
5987 compilation to the native function. This causes the pattern itself and
5988 all data strings used for matching it to be treated as UTF-8 strings.
5989 Note that REG_UTF8 is not part of the POSIX standard.
5990
5991 In the absence of these flags, no options are passed to the native
5992 function. This means the the regex is compiled with PCRE default
5993 semantics. In particular, the way it handles newline characters in the
5994 subject string is the Perl way, not the POSIX way. Note that setting
5995 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5996 It does not affect the way newlines are matched by . (they aren't) or
5997 by a negative class such as [^a] (they are).
5998
5999 The yield of regcomp() is zero on success, and non-zero otherwise. The
6000 preg structure is filled in on success, and one member of the structure
6001 is public: re_nsub contains the number of capturing subpatterns in the
6002 regular expression. Various error codes are defined in the header file.
6003
6004
6005 MATCHING NEWLINE CHARACTERS
6006
6007 This area is not simple, because POSIX and Perl take different views of
6008 things. It is not possible to get PCRE to obey POSIX semantics, but
6009 then PCRE was never intended to be a POSIX engine. The following table
6010 lists the different possibilities for matching newline characters in
6011 PCRE:
6012
6013 Default Change with
6014
6015 . matches newline no PCRE_DOTALL
6016 newline matches [^a] yes not changeable
6017 $ matches \n at end yes PCRE_DOLLARENDONLY
6018 $ matches \n in middle no PCRE_MULTILINE
6019 ^ matches \n in middle no PCRE_MULTILINE
6020
6021 This is the equivalent table for POSIX:
6022
6023 Default Change with
6024
6025 . matches newline yes REG_NEWLINE
6026 newline matches [^a] yes REG_NEWLINE
6027 $ matches \n at end no REG_NEWLINE
6028 $ matches \n in middle no REG_NEWLINE
6029 ^ matches \n in middle no REG_NEWLINE
6030
6031 PCRE's behaviour is the same as Perl's, except that there is no equiva-
6032 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
6033 no way to stop newline from matching [^a].
6034
6035 The default POSIX newline handling can be obtained by setting
6036 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
6037 behave exactly as for the REG_NEWLINE action.
6038
6039
6040 MATCHING A PATTERN
6041
6042 The function regexec() is called to match a compiled pattern preg
6043 against a given string, which is by default terminated by a zero byte
6044 (but see REG_STARTEND below), subject to the options in eflags. These
6045 can be:
6046
6047 REG_NOTBOL
6048
6049 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
6050 function.
6051
6052 REG_NOTEMPTY
6053
6054 The PCRE_NOTEMPTY option is set when calling the underlying PCRE match-
6055 ing function. Note that REG_NOTEMPTY is not part of the POSIX standard.
6056 However, setting this option can give more POSIX-like behaviour in some
6057 situations.
6058
6059 REG_NOTEOL
6060
6061 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
6062 function.
6063
6064 REG_STARTEND
6065
6066 The string is considered to start at string + pmatch[0].rm_so and to
6067 have a terminating NUL located at string + pmatch[0].rm_eo (there need
6068 not actually be a NUL at that location), regardless of the value of
6069 nmatch. This is a BSD extension, compatible with but not specified by
6070 IEEE Standard 1003.2 (POSIX.2), and should be used with caution in
6071 software intended to be portable to other systems. Note that a non-zero
6072 rm_so does not imply REG_NOTBOL; REG_STARTEND affects only the location
6073 of the string, not how it is matched.
6074
6075 If the pattern was compiled with the REG_NOSUB flag, no data about any
6076 matched strings is returned. The nmatch and pmatch arguments of
6077 regexec() are ignored.
6078
6079 Otherwise,the portion of the string that was matched, and also any cap-
6080 tured substrings, are returned via the pmatch argument, which points to
6081 an array of nmatch structures of type regmatch_t, containing the mem-
6082 bers rm_so and rm_eo. These contain the offset to the first character
6083 of each substring and the offset to the first character after the end
6084 of each substring, respectively. The 0th element of the vector relates
6085 to the entire portion of string that was matched; subsequent elements
6086 relate to the capturing subpatterns of the regular expression. Unused
6087 entries in the array have both structure members set to -1.
6088
6089 A successful match yields a zero return; various error codes are