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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 589 - (show annotations)
Sat Jan 15 11:31:39 2011 UTC (4 years, 6 months ago) by ph10
File MIME type: text/plain
File size: 368093 byte(s)
Error occurred while calculating annotation data.
Tidies for Release 8.12
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. Neither has
6 the pcredemo program. There are separate text files for the pcregrep and
7 pcretest commands.
8 -----------------------------------------------------------------------------
9
10
11 PCRE(3) PCRE(3)
12
13
14 NAME
15 PCRE - Perl-compatible regular expressions
16
17
18 INTRODUCTION
19
20 The PCRE library is a set of functions that implement regular expres-
21 sion pattern matching using the same syntax and semantics as Perl, with
22 just a few differences. Some features that appeared in Python and PCRE
23 before they appeared in Perl are also available using the Python syn-
24 tax, there is some support for one or two .NET and Oniguruma syntax
25 items, and there is an option for requesting some minor changes that
26 give better JavaScript compatibility.
27
28 The current implementation of PCRE corresponds approximately with Perl
29 5.12, including support for UTF-8 encoded strings and Unicode general
30 category properties. However, UTF-8 and Unicode support has to be
31 explicitly enabled; it is not the default. The Unicode tables corre-
32 spond to Unicode release 5.2.0.
33
34 In addition to the Perl-compatible matching function, PCRE contains an
35 alternative function that matches the same compiled patterns in a dif-
36 ferent way. In certain circumstances, the alternative function has some
37 advantages. For a discussion of the two matching algorithms, see the
38 pcrematching page.
39
40 PCRE is written in C and released as a C library. A number of people
41 have written wrappers and interfaces of various kinds. In particular,
42 Google Inc. have provided a comprehensive C++ wrapper. This is now
43 included as part of the PCRE distribution. The pcrecpp page has details
44 of this interface. Other people's contributions can be found in the
45 Contrib directory at the primary FTP site, which is:
46
47 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
48
49 Details of exactly which Perl regular expression features are and are
50 not supported by PCRE are given in separate documents. See the pcrepat-
51 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
52 page.
53
54 Some features of PCRE can be included, excluded, or changed when the
55 library is built. The pcre_config() function makes it possible for a
56 client to discover which features are available. The features them-
57 selves are described in the pcrebuild page. Documentation about build-
58 ing PCRE for various operating systems can be found in the README and
59 NON-UNIX-USE files in the source distribution.
60
61 The library contains a number of undocumented internal functions and
62 data tables that are used by more than one of the exported external
63 functions, but which are not intended for use by external callers.
64 Their names all begin with "_pcre_", which hopefully will not provoke
65 any name clashes. In some environments, it is possible to control which
66 external symbols are exported when a shared library is built, and in
67 these cases the undocumented symbols are not exported.
68
69
70 USER DOCUMENTATION
71
72 The user documentation for PCRE comprises a number of different sec-
73 tions. In the "man" format, each of these is a separate "man page". In
74 the HTML format, each is a separate page, linked from the index page.
75 In the plain text format, all the sections, except the pcredemo sec-
76 tion, are concatenated, for ease of searching. The sections are as fol-
77 lows:
78
79 pcre this document
80 pcre-config show PCRE installation configuration information
81 pcreapi details of PCRE's native C API
82 pcrebuild options for building PCRE
83 pcrecallout details of the callout feature
84 pcrecompat discussion of Perl compatibility
85 pcrecpp details of the C++ wrapper
86 pcredemo a demonstration C program that uses PCRE
87 pcregrep description of the pcregrep command
88 pcrematching discussion of the two matching algorithms
89 pcrepartial details of the partial matching facility
90 pcrepattern syntax and semantics of supported
91 regular expressions
92 pcreperform discussion of performance issues
93 pcreposix the POSIX-compatible C API
94 pcreprecompile details of saving and re-using precompiled patterns
95 pcresample discussion of the pcredemo program
96 pcrestack discussion of stack usage
97 pcresyntax quick syntax reference
98 pcretest description of the pcretest testing command
99
100 In addition, in the "man" and HTML formats, there is a short page for
101 each C library function, listing its arguments and results.
102
103
104 LIMITATIONS
105
106 There are some size limitations in PCRE but it is hoped that they will
107 never in practice be relevant.
108
109 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
110 is compiled with the default internal linkage size of 2. If you want to
111 process regular expressions that are truly enormous, you can compile
112 PCRE with an internal linkage size of 3 or 4 (see the README file in
113 the source distribution and the pcrebuild documentation for details).
114 In these cases the limit is substantially larger. However, the speed
115 of execution is slower.
116
117 All values in repeating quantifiers must be less than 65536.
118
119 There is no limit to the number of parenthesized subpatterns, but there
120 can be no more than 65535 capturing subpatterns.
121
122 The maximum length of name for a named subpattern is 32 characters, and
123 the maximum number of named subpatterns is 10000.
124
125 The maximum length of a subject string is the largest positive number
126 that an integer variable can hold. However, when using the traditional
127 matching function, PCRE uses recursion to handle subpatterns and indef-
128 inite repetition. This means that the available stack space may limit
129 the size of a subject string that can be processed by certain patterns.
130 For a discussion of stack issues, see the pcrestack documentation.
131
132
133 UTF-8 AND UNICODE PROPERTY SUPPORT
134
135 From release 3.3, PCRE has had some support for character strings
136 encoded in the UTF-8 format. For release 4.0 this was greatly extended
137 to cover most common requirements, and in release 5.0 additional sup-
138 port for Unicode general category properties was added.
139
140 In order process UTF-8 strings, you must build PCRE to include UTF-8
141 support in the code, and, in addition, you must call pcre_compile()
142 with the PCRE_UTF8 option flag, or the pattern must start with the
143 sequence (*UTF8). When either of these is the case, both the pattern
144 and any subject strings that are matched against it are treated as
145 UTF-8 strings instead of strings of 1-byte characters.
146
147 If you compile PCRE with UTF-8 support, but do not use it at run time,
148 the library will be a bit bigger, but the additional run time overhead
149 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
150 very big.
151
152 If PCRE is built with Unicode character property support (which implies
153 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
154 ported. The available properties that can be tested are limited to the
155 general category properties such as Lu for an upper case letter or Nd
156 for a decimal number, the Unicode script names such as Arabic or Han,
157 and the derived properties Any and L&. A full list is given in the
158 pcrepattern documentation. Only the short names for properties are sup-
159 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
160 ter}, is not supported. Furthermore, in Perl, many properties may
161 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
162 does not support this.
163
164 Validity of UTF-8 strings
165
166 When you set the PCRE_UTF8 flag, the strings passed as patterns and
167 subjects are (by default) checked for validity on entry to the relevant
168 functions. From release 7.3 of PCRE, the check is according the rules
169 of RFC 3629, which are themselves derived from the Unicode specifica-
170 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
171 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
172 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
173 to U+DFFF.
174
175 The excluded code points are the "Low Surrogate Area" of Unicode, of
176 which the Unicode Standard says this: "The Low Surrogate Area does not
177 contain any character assignments, consequently no character code
178 charts or namelists are provided for this area. Surrogates are reserved
179 for use with UTF-16 and then must be used in pairs." The code points
180 that are encoded by UTF-16 pairs are available as independent code
181 points in the UTF-8 encoding. (In other words, the whole surrogate
182 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
183
184 If an invalid UTF-8 string is passed to PCRE, an error return
185 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
186 that your strings are valid, and therefore want to skip these checks in
187 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
188 compile time or at run time, PCRE assumes that the pattern or subject
189 it is given (respectively) contains only valid UTF-8 codes. In this
190 case, it does not diagnose an invalid UTF-8 string.
191
192 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
193 what happens depends on why the string is invalid. If the string con-
194 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
195 string of characters in the range 0 to 0x7FFFFFFF. In other words,
196 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
197 strings according to the more liberal rules of RFC 2279. However, if
198 the string does not even conform to RFC 2279, the result is undefined.
199 Your program may crash.
200
201 If you want to process strings of values in the full range 0 to
202 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
203 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
204 this situation, you will have to apply your own validity check.
205
206 General comments about UTF-8 mode
207
208 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
209 two-byte UTF-8 character if the value is greater than 127.
210
211 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
212 characters for values greater than \177.
213
214 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
215 vidual bytes, for example: \x{100}{3}.
216
217 4. The dot metacharacter matches one UTF-8 character instead of a sin-
218 gle byte.
219
220 5. The escape sequence \C can be used to match a single byte in UTF-8
221 mode, but its use can lead to some strange effects. This facility is
222 not available in the alternative matching function, pcre_dfa_exec().
223
224 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
225 test characters of any code value, but, by default, the characters that
226 PCRE recognizes as digits, spaces, or word characters remain the same
227 set as before, all with values less than 256. This remains true even
228 when PCRE is built to include Unicode property support, because to do
229 otherwise would slow down PCRE in many common cases. Note in particular
230 that this applies to \b and \B, because they are defined in terms of \w
231 and \W. If you really want to test for a wider sense of, say, "digit",
232 you can use explicit Unicode property tests such as \p{Nd}. Alterna-
233 tively, if you set the PCRE_UCP option, the way that the character
234 escapes work is changed so that Unicode properties are used to deter-
235 mine which characters match. There are more details in the section on
236 generic character types in the pcrepattern documentation.
237
238 7. Similarly, characters that match the POSIX named character classes
239 are all low-valued characters, unless the PCRE_UCP option is set.
240
241 8. However, the horizontal and vertical whitespace matching escapes
242 (\h, \H, \v, and \V) do match all the appropriate Unicode characters,
243 whether or not PCRE_UCP is set.
244
245 9. Case-insensitive matching applies only to characters whose values
246 are less than 128, unless PCRE is built with Unicode property support.
247 Even when Unicode property support is available, PCRE still uses its
248 own character tables when checking the case of low-valued characters,
249 so as not to degrade performance. The Unicode property information is
250 used only for characters with higher values. Furthermore, PCRE supports
251 case-insensitive matching only when there is a one-to-one mapping
252 between a letter's cases. There are a small number of many-to-one map-
253 pings in Unicode; these are not supported by PCRE.
254
255
256 AUTHOR
257
258 Philip Hazel
259 University Computing Service
260 Cambridge CB2 3QH, England.
261
262 Putting an actual email address here seems to have been a spam magnet,
263 so I've taken it away. If you want to email me, use my two initials,
264 followed by the two digits 10, at the domain cam.ac.uk.
265
266
267 REVISION
268
269 Last updated: 13 November 2010
270 Copyright (c) 1997-2010 University of Cambridge.
271 ------------------------------------------------------------------------------
272
273
274 PCREBUILD(3) PCREBUILD(3)
275
276
277 NAME
278 PCRE - Perl-compatible regular expressions
279
280
281 PCRE BUILD-TIME OPTIONS
282
283 This document describes the optional features of PCRE that can be
284 selected when the library is compiled. It assumes use of the configure
285 script, where the optional features are selected or deselected by pro-
286 viding options to configure before running the make command. However,
287 the same options can be selected in both Unix-like and non-Unix-like
288 environments using the GUI facility of cmake-gui if you are using CMake
289 instead of configure to build PCRE.
290
291 There is a lot more information about building PCRE in non-Unix-like
292 environments in the file called NON_UNIX_USE, which is part of the PCRE
293 distribution. You should consult this file as well as the README file
294 if you are building in a non-Unix-like environment.
295
296 The complete list of options for configure (which includes the standard
297 ones such as the selection of the installation directory) can be
298 obtained by running
299
300 ./configure --help
301
302 The following sections include descriptions of options whose names
303 begin with --enable or --disable. These settings specify changes to the
304 defaults for the configure command. Because of the way that configure
305 works, --enable and --disable always come in pairs, so the complemen-
306 tary option always exists as well, but as it specifies the default, it
307 is not described.
308
309
310 C++ SUPPORT
311
312 By default, the configure script will search for a C++ compiler and C++
313 header files. If it finds them, it automatically builds the C++ wrapper
314 library for PCRE. You can disable this by adding
315
316 --disable-cpp
317
318 to the configure command.
319
320
321 UTF-8 SUPPORT
322
323 To build PCRE with support for UTF-8 Unicode character strings, add
324
325 --enable-utf8
326
327 to the configure command. Of itself, this does not make PCRE treat
328 strings as UTF-8. As well as compiling PCRE with this option, you also
329 have have to set the PCRE_UTF8 option when you call the pcre_compile()
330 or pcre_compile2() functions.
331
332 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
333 expects its input to be either ASCII or UTF-8 (depending on the runtime
334 option). It is not possible to support both EBCDIC and UTF-8 codes in
335 the same version of the library. Consequently, --enable-utf8 and
336 --enable-ebcdic are mutually exclusive.
337
338
339 UNICODE CHARACTER PROPERTY SUPPORT
340
341 UTF-8 support allows PCRE to process character values greater than 255
342 in the strings that it handles. On its own, however, it does not pro-
343 vide any facilities for accessing the properties of such characters. If
344 you want to be able to use the pattern escapes \P, \p, and \X, which
345 refer to Unicode character properties, you must add
346
347 --enable-unicode-properties
348
349 to the configure command. This implies UTF-8 support, even if you have
350 not explicitly requested it.
351
352 Including Unicode property support adds around 30K of tables to the
353 PCRE library. Only the general category properties such as Lu and Nd
354 are supported. Details are given in the pcrepattern documentation.
355
356
357 CODE VALUE OF NEWLINE
358
359 By default, PCRE interprets the linefeed (LF) character as indicating
360 the end of a line. This is the normal newline character on Unix-like
361 systems. You can compile PCRE to use carriage return (CR) instead, by
362 adding
363
364 --enable-newline-is-cr
365
366 to the configure command. There is also a --enable-newline-is-lf
367 option, which explicitly specifies linefeed as the newline character.
368
369 Alternatively, you can specify that line endings are to be indicated by
370 the two character sequence CRLF. If you want this, add
371
372 --enable-newline-is-crlf
373
374 to the configure command. There is a fourth option, specified by
375
376 --enable-newline-is-anycrlf
377
378 which causes PCRE to recognize any of the three sequences CR, LF, or
379 CRLF as indicating a line ending. Finally, a fifth option, specified by
380
381 --enable-newline-is-any
382
383 causes PCRE to recognize any Unicode newline sequence.
384
385 Whatever line ending convention is selected when PCRE is built can be
386 overridden when the library functions are called. At build time it is
387 conventional to use the standard for your operating system.
388
389
390 WHAT \R MATCHES
391
392 By default, the sequence \R in a pattern matches any Unicode newline
393 sequence, whatever has been selected as the line ending sequence. If
394 you specify
395
396 --enable-bsr-anycrlf
397
398 the default is changed so that \R matches only CR, LF, or CRLF. What-
399 ever is selected when PCRE is built can be overridden when the library
400 functions are called.
401
402
403 BUILDING SHARED AND STATIC LIBRARIES
404
405 The PCRE building process uses libtool to build both shared and static
406 Unix libraries by default. You can suppress one of these by adding one
407 of
408
409 --disable-shared
410 --disable-static
411
412 to the configure command, as required.
413
414
415 POSIX MALLOC USAGE
416
417 When PCRE is called through the POSIX interface (see the pcreposix doc-
418 umentation), additional working storage is required for holding the
419 pointers to capturing substrings, because PCRE requires three integers
420 per substring, whereas the POSIX interface provides only two. If the
421 number of expected substrings is small, the wrapper function uses space
422 on the stack, because this is faster than using malloc() for each call.
423 The default threshold above which the stack is no longer used is 10; it
424 can be changed by adding a setting such as
425
426 --with-posix-malloc-threshold=20
427
428 to the configure command.
429
430
431 HANDLING VERY LARGE PATTERNS
432
433 Within a compiled pattern, offset values are used to point from one
434 part to another (for example, from an opening parenthesis to an alter-
435 nation metacharacter). By default, two-byte values are used for these
436 offsets, leading to a maximum size for a compiled pattern of around
437 64K. This is sufficient to handle all but the most gigantic patterns.
438 Nevertheless, some people do want to process truyl enormous patterns,
439 so it is possible to compile PCRE to use three-byte or four-byte off-
440 sets by adding a setting such as
441
442 --with-link-size=3
443
444 to the configure command. The value given must be 2, 3, or 4. Using
445 longer offsets slows down the operation of PCRE because it has to load
446 additional bytes when handling them.
447
448
449 AVOIDING EXCESSIVE STACK USAGE
450
451 When matching with the pcre_exec() function, PCRE implements backtrack-
452 ing by making recursive calls to an internal function called match().
453 In environments where the size of the stack is limited, this can se-
454 verely limit PCRE's operation. (The Unix environment does not usually
455 suffer from this problem, but it may sometimes be necessary to increase
456 the maximum stack size. There is a discussion in the pcrestack docu-
457 mentation.) An alternative approach to recursion that uses memory from
458 the heap to remember data, instead of using recursive function calls,
459 has been implemented to work round the problem of limited stack size.
460 If you want to build a version of PCRE that works this way, add
461
462 --disable-stack-for-recursion
463
464 to the configure command. With this configuration, PCRE will use the
465 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
466 ment functions. By default these point to malloc() and free(), but you
467 can replace the pointers so that your own functions are used instead.
468
469 Separate functions are provided rather than using pcre_malloc and
470 pcre_free because the usage is very predictable: the block sizes
471 requested are always the same, and the blocks are always freed in
472 reverse order. A calling program might be able to implement optimized
473 functions that perform better than malloc() and free(). PCRE runs
474 noticeably more slowly when built in this way. This option affects only
475 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
476
477
478 LIMITING PCRE RESOURCE USAGE
479
480 Internally, PCRE has a function called match(), which it calls repeat-
481 edly (sometimes recursively) when matching a pattern with the
482 pcre_exec() function. By controlling the maximum number of times this
483 function may be called during a single matching operation, a limit can
484 be placed on the resources used by a single call to pcre_exec(). The
485 limit can be changed at run time, as described in the pcreapi documen-
486 tation. The default is 10 million, but this can be changed by adding a
487 setting such as
488
489 --with-match-limit=500000
490
491 to the configure command. This setting has no effect on the
492 pcre_dfa_exec() matching function.
493
494 In some environments it is desirable to limit the depth of recursive
495 calls of match() more strictly than the total number of calls, in order
496 to restrict the maximum amount of stack (or heap, if --disable-stack-
497 for-recursion is specified) that is used. A second limit controls this;
498 it defaults to the value that is set for --with-match-limit, which
499 imposes no additional constraints. However, you can set a lower limit
500 by adding, for example,
501
502 --with-match-limit-recursion=10000
503
504 to the configure command. This value can also be overridden at run
505 time.
506
507
508 CREATING CHARACTER TABLES AT BUILD TIME
509
510 PCRE uses fixed tables for processing characters whose code values are
511 less than 256. By default, PCRE is built with a set of tables that are
512 distributed in the file pcre_chartables.c.dist. These tables are for
513 ASCII codes only. If you add
514
515 --enable-rebuild-chartables
516
517 to the configure command, the distributed tables are no longer used.
518 Instead, a program called dftables is compiled and run. This outputs
519 the source for new set of tables, created in the default locale of your
520 C runtime system. (This method of replacing the tables does not work if
521 you are cross compiling, because dftables is run on the local host. If
522 you need to create alternative tables when cross compiling, you will
523 have to do so "by hand".)
524
525
526 USING EBCDIC CODE
527
528 PCRE assumes by default that it will run in an environment where the
529 character code is ASCII (or Unicode, which is a superset of ASCII).
530 This is the case for most computer operating systems. PCRE can, how-
531 ever, be compiled to run in an EBCDIC environment by adding
532
533 --enable-ebcdic
534
535 to the configure command. This setting implies --enable-rebuild-charta-
536 bles. You should only use it if you know that you are in an EBCDIC
537 environment (for example, an IBM mainframe operating system). The
538 --enable-ebcdic option is incompatible with --enable-utf8.
539
540
541 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
542
543 By default, pcregrep reads all files as plain text. You can build it so
544 that it recognizes files whose names end in .gz or .bz2, and reads them
545 with libz or libbz2, respectively, by adding one or both of
546
547 --enable-pcregrep-libz
548 --enable-pcregrep-libbz2
549
550 to the configure command. These options naturally require that the rel-
551 evant libraries are installed on your system. Configuration will fail
552 if they are not.
553
554
555 PCRETEST OPTION FOR LIBREADLINE SUPPORT
556
557 If you add
558
559 --enable-pcretest-libreadline
560
561 to the configure command, pcretest is linked with the libreadline
562 library, and when its input is from a terminal, it reads it using the
563 readline() function. This provides line-editing and history facilities.
564 Note that libreadline is GPL-licensed, so if you distribute a binary of
565 pcretest linked in this way, there may be licensing issues.
566
567 Setting this option causes the -lreadline option to be added to the
568 pcretest build. In many operating environments with a sytem-installed
569 libreadline this is sufficient. However, in some environments (e.g. if
570 an unmodified distribution version of readline is in use), some extra
571 configuration may be necessary. The INSTALL file for libreadline says
572 this:
573
574 "Readline uses the termcap functions, but does not link with the
575 termcap or curses library itself, allowing applications which link
576 with readline the to choose an appropriate library."
577
578 If your environment has not been set up so that an appropriate library
579 is automatically included, you may need to add something like
580
581 LIBS="-ncurses"
582
583 immediately before the configure command.
584
585
586 SEE ALSO
587
588 pcreapi(3), pcre_config(3).
589
590
591 AUTHOR
592
593 Philip Hazel
594 University Computing Service
595 Cambridge CB2 3QH, England.
596
597
598 REVISION
599
600 Last updated: 29 September 2009
601 Copyright (c) 1997-2009 University of Cambridge.
602 ------------------------------------------------------------------------------
603
604
605 PCREMATCHING(3) PCREMATCHING(3)
606
607
608 NAME
609 PCRE - Perl-compatible regular expressions
610
611
612 PCRE MATCHING ALGORITHMS
613
614 This document describes the two different algorithms that are available
615 in PCRE for matching a compiled regular expression against a given sub-
616 ject string. The "standard" algorithm is the one provided by the
617 pcre_exec() function. This works in the same was as Perl's matching
618 function, and provides a Perl-compatible matching operation.
619
620 An alternative algorithm is provided by the pcre_dfa_exec() function;
621 this operates in a different way, and is not Perl-compatible. It has
622 advantages and disadvantages compared with the standard algorithm, and
623 these are described below.
624
625 When there is only one possible way in which a given subject string can
626 match a pattern, the two algorithms give the same answer. A difference
627 arises, however, when there are multiple possibilities. For example, if
628 the pattern
629
630 ^<.*>
631
632 is matched against the string
633
634 <something> <something else> <something further>
635
636 there are three possible answers. The standard algorithm finds only one
637 of them, whereas the alternative algorithm finds all three.
638
639
640 REGULAR EXPRESSIONS AS TREES
641
642 The set of strings that are matched by a regular expression can be rep-
643 resented as a tree structure. An unlimited repetition in the pattern
644 makes the tree of infinite size, but it is still a tree. Matching the
645 pattern to a given subject string (from a given starting point) can be
646 thought of as a search of the tree. There are two ways to search a
647 tree: depth-first and breadth-first, and these correspond to the two
648 matching algorithms provided by PCRE.
649
650
651 THE STANDARD MATCHING ALGORITHM
652
653 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
654 sions", the standard algorithm is an "NFA algorithm". It conducts a
655 depth-first search of the pattern tree. That is, it proceeds along a
656 single path through the tree, checking that the subject matches what is
657 required. When there is a mismatch, the algorithm tries any alterna-
658 tives at the current point, and if they all fail, it backs up to the
659 previous branch point in the tree, and tries the next alternative
660 branch at that level. This often involves backing up (moving to the
661 left) in the subject string as well. The order in which repetition
662 branches are tried is controlled by the greedy or ungreedy nature of
663 the quantifier.
664
665 If a leaf node is reached, a matching string has been found, and at
666 that point the algorithm stops. Thus, if there is more than one possi-
667 ble match, this algorithm returns the first one that it finds. Whether
668 this is the shortest, the longest, or some intermediate length depends
669 on the way the greedy and ungreedy repetition quantifiers are specified
670 in the pattern.
671
672 Because it ends up with a single path through the tree, it is rela-
673 tively straightforward for this algorithm to keep track of the sub-
674 strings that are matched by portions of the pattern in parentheses.
675 This provides support for capturing parentheses and back references.
676
677
678 THE ALTERNATIVE MATCHING ALGORITHM
679
680 This algorithm conducts a breadth-first search of the tree. Starting
681 from the first matching point in the subject, it scans the subject
682 string from left to right, once, character by character, and as it does
683 this, it remembers all the paths through the tree that represent valid
684 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
685 though it is not implemented as a traditional finite state machine (it
686 keeps multiple states active simultaneously).
687
688 Although the general principle of this matching algorithm is that it
689 scans the subject string only once, without backtracking, there is one
690 exception: when a lookaround assertion is encountered, the characters
691 following or preceding the current point have to be independently
692 inspected.
693
694 The scan continues until either the end of the subject is reached, or
695 there are no more unterminated paths. At this point, terminated paths
696 represent the different matching possibilities (if there are none, the
697 match has failed). Thus, if there is more than one possible match,
698 this algorithm finds all of them, and in particular, it finds the long-
699 est. The matches are returned in decreasing order of length. There is
700 an option to stop the algorithm after the first match (which is neces-
701 sarily the shortest) is found.
702
703 Note that all the matches that are found start at the same point in the
704 subject. If the pattern
705
706 cat(er(pillar)?)?
707
708 is matched against the string "the caterpillar catchment", the result
709 will be the three strings "caterpillar", "cater", and "cat" that start
710 at the fifth character of the subject. The algorithm does not automati-
711 cally move on to find matches that start at later positions.
712
713 There are a number of features of PCRE regular expressions that are not
714 supported by the alternative matching algorithm. They are as follows:
715
716 1. Because the algorithm finds all possible matches, the greedy or
717 ungreedy nature of repetition quantifiers is not relevant. Greedy and
718 ungreedy quantifiers are treated in exactly the same way. However, pos-
719 sessive quantifiers can make a difference when what follows could also
720 match what is quantified, for example in a pattern like this:
721
722 ^a++\w!
723
724 This pattern matches "aaab!" but not "aaa!", which would be matched by
725 a non-possessive quantifier. Similarly, if an atomic group is present,
726 it is matched as if it were a standalone pattern at the current point,
727 and the longest match is then "locked in" for the rest of the overall
728 pattern.
729
730 2. When dealing with multiple paths through the tree simultaneously, it
731 is not straightforward to keep track of captured substrings for the
732 different matching possibilities, and PCRE's implementation of this
733 algorithm does not attempt to do this. This means that no captured sub-
734 strings are available.
735
736 3. Because no substrings are captured, back references within the pat-
737 tern are not supported, and cause errors if encountered.
738
739 4. For the same reason, conditional expressions that use a backrefer-
740 ence as the condition or test for a specific group recursion are not
741 supported.
742
743 5. Because many paths through the tree may be active, the \K escape
744 sequence, which resets the start of the match when encountered (but may
745 be on some paths and not on others), is not supported. It causes an
746 error if encountered.
747
748 6. Callouts are supported, but the value of the capture_top field is
749 always 1, and the value of the capture_last field is always -1.
750
751 7. The \C escape sequence, which (in the standard algorithm) matches a
752 single byte, even in UTF-8 mode, is not supported because the alterna-
753 tive algorithm moves through the subject string one character at a
754 time, for all active paths through the tree.
755
756 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
757 are not supported. (*FAIL) is supported, and behaves like a failing
758 negative assertion.
759
760
761 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
762
763 Using the alternative matching algorithm provides the following advan-
764 tages:
765
766 1. All possible matches (at a single point in the subject) are automat-
767 ically found, and in particular, the longest match is found. To find
768 more than one match using the standard algorithm, you have to do kludgy
769 things with callouts.
770
771 2. Because the alternative algorithm scans the subject string just
772 once, and never needs to backtrack, it is possible to pass very long
773 subject strings to the matching function in several pieces, checking
774 for partial matching each time. Although it is possible to do multi-
775 segment matching using the standard algorithm (pcre_exec()), by retain-
776 ing partially matched substrings, it is more complicated. The pcrepar-
777 tial documentation gives details of partial matching and discusses
778 multi-segment matching.
779
780
781 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
782
783 The alternative algorithm suffers from a number of disadvantages:
784
785 1. It is substantially slower than the standard algorithm. This is
786 partly because it has to search for all possible matches, but is also
787 because it is less susceptible to optimization.
788
789 2. Capturing parentheses and back references are not supported.
790
791 3. Although atomic groups are supported, their use does not provide the
792 performance advantage that it does for the standard algorithm.
793
794
795 AUTHOR
796
797 Philip Hazel
798 University Computing Service
799 Cambridge CB2 3QH, England.
800
801
802 REVISION
803
804 Last updated: 17 November 2010
805 Copyright (c) 1997-2010 University of Cambridge.
806 ------------------------------------------------------------------------------
807
808
809 PCREAPI(3) PCREAPI(3)
810
811
812 NAME
813 PCRE - Perl-compatible regular expressions
814
815
816 PCRE NATIVE API
817
818 #include <pcre.h>
819
820 pcre *pcre_compile(const char *pattern, int options,
821 const char **errptr, int *erroffset,
822 const unsigned char *tableptr);
823
824 pcre *pcre_compile2(const char *pattern, int options,
825 int *errorcodeptr,
826 const char **errptr, int *erroffset,
827 const unsigned char *tableptr);
828
829 pcre_extra *pcre_study(const pcre *code, int options,
830 const char **errptr);
831
832 int pcre_exec(const pcre *code, const pcre_extra *extra,
833 const char *subject, int length, int startoffset,
834 int options, int *ovector, int ovecsize);
835
836 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
837 const char *subject, int length, int startoffset,
838 int options, int *ovector, int ovecsize,
839 int *workspace, int wscount);
840
841 int pcre_copy_named_substring(const pcre *code,
842 const char *subject, int *ovector,
843 int stringcount, const char *stringname,
844 char *buffer, int buffersize);
845
846 int pcre_copy_substring(const char *subject, int *ovector,
847 int stringcount, int stringnumber, char *buffer,
848 int buffersize);
849
850 int pcre_get_named_substring(const pcre *code,
851 const char *subject, int *ovector,
852 int stringcount, const char *stringname,
853 const char **stringptr);
854
855 int pcre_get_stringnumber(const pcre *code,
856 const char *name);
857
858 int pcre_get_stringtable_entries(const pcre *code,
859 const char *name, char **first, char **last);
860
861 int pcre_get_substring(const char *subject, int *ovector,
862 int stringcount, int stringnumber,
863 const char **stringptr);
864
865 int pcre_get_substring_list(const char *subject,
866 int *ovector, int stringcount, const char ***listptr);
867
868 void pcre_free_substring(const char *stringptr);
869
870 void pcre_free_substring_list(const char **stringptr);
871
872 const unsigned char *pcre_maketables(void);
873
874 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
875 int what, void *where);
876
877 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
878
879 int pcre_refcount(pcre *code, int adjust);
880
881 int pcre_config(int what, void *where);
882
883 char *pcre_version(void);
884
885 void *(*pcre_malloc)(size_t);
886
887 void (*pcre_free)(void *);
888
889 void *(*pcre_stack_malloc)(size_t);
890
891 void (*pcre_stack_free)(void *);
892
893 int (*pcre_callout)(pcre_callout_block *);
894
895
896 PCRE API OVERVIEW
897
898 PCRE has its own native API, which is described in this document. There
899 are also some wrapper functions that correspond to the POSIX regular
900 expression API. These are described in the pcreposix documentation.
901 Both of these APIs define a set of C function calls. A C++ wrapper is
902 distributed with PCRE. It is documented in the pcrecpp page.
903
904 The native API C function prototypes are defined in the header file
905 pcre.h, and on Unix systems the library itself is called libpcre. It
906 can normally be accessed by adding -lpcre to the command for linking an
907 application that uses PCRE. The header file defines the macros
908 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
909 bers for the library. Applications can use these to include support
910 for different releases of PCRE.
911
912 In a Windows environment, if you want to statically link an application
913 program against a non-dll pcre.a file, you must define PCRE_STATIC
914 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
915 loc() and pcre_free() exported functions will be declared
916 __declspec(dllimport), with unwanted results.
917
918 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
919 pcre_exec() are used for compiling and matching regular expressions in
920 a Perl-compatible manner. A sample program that demonstrates the sim-
921 plest way of using them is provided in the file called pcredemo.c in
922 the PCRE source distribution. A listing of this program is given in the
923 pcredemo documentation, and the pcresample documentation describes how
924 to compile and run it.
925
926 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
927 ble, is also provided. This uses a different algorithm for the match-
928 ing. The alternative algorithm finds all possible matches (at a given
929 point in the subject), and scans the subject just once (unless there
930 are lookbehind assertions). However, this algorithm does not return
931 captured substrings. A description of the two matching algorithms and
932 their advantages and disadvantages is given in the pcrematching docu-
933 mentation.
934
935 In addition to the main compiling and matching functions, there are
936 convenience functions for extracting captured substrings from a subject
937 string that is matched by pcre_exec(). They are:
938
939 pcre_copy_substring()
940 pcre_copy_named_substring()
941 pcre_get_substring()
942 pcre_get_named_substring()
943 pcre_get_substring_list()
944 pcre_get_stringnumber()
945 pcre_get_stringtable_entries()
946
947 pcre_free_substring() and pcre_free_substring_list() are also provided,
948 to free the memory used for extracted strings.
949
950 The function pcre_maketables() is used to build a set of character
951 tables in the current locale for passing to pcre_compile(),
952 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
953 provided for specialist use. Most commonly, no special tables are
954 passed, in which case internal tables that are generated when PCRE is
955 built are used.
956
957 The function pcre_fullinfo() is used to find out information about a
958 compiled pattern; pcre_info() is an obsolete version that returns only
959 some of the available information, but is retained for backwards com-
960 patibility. The function pcre_version() returns a pointer to a string
961 containing the version of PCRE and its date of release.
962
963 The function pcre_refcount() maintains a reference count in a data
964 block containing a compiled pattern. This is provided for the benefit
965 of object-oriented applications.
966
967 The global variables pcre_malloc and pcre_free initially contain the
968 entry points of the standard malloc() and free() functions, respec-
969 tively. PCRE calls the memory management functions via these variables,
970 so a calling program can replace them if it wishes to intercept the
971 calls. This should be done before calling any PCRE functions.
972
973 The global variables pcre_stack_malloc and pcre_stack_free are also
974 indirections to memory management functions. These special functions
975 are used only when PCRE is compiled to use the heap for remembering
976 data, instead of recursive function calls, when running the pcre_exec()
977 function. See the pcrebuild documentation for details of how to do
978 this. It is a non-standard way of building PCRE, for use in environ-
979 ments that have limited stacks. Because of the greater use of memory
980 management, it runs more slowly. Separate functions are provided so
981 that special-purpose external code can be used for this case. When
982 used, these functions are always called in a stack-like manner (last
983 obtained, first freed), and always for memory blocks of the same size.
984 There is a discussion about PCRE's stack usage in the pcrestack docu-
985 mentation.
986
987 The global variable pcre_callout initially contains NULL. It can be set
988 by the caller to a "callout" function, which PCRE will then call at
989 specified points during a matching operation. Details are given in the
990 pcrecallout documentation.
991
992
993 NEWLINES
994
995 PCRE supports five different conventions for indicating line breaks in
996 strings: a single CR (carriage return) character, a single LF (line-
997 feed) character, the two-character sequence CRLF, any of the three pre-
998 ceding, or any Unicode newline sequence. The Unicode newline sequences
999 are the three just mentioned, plus the single characters VT (vertical
1000 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
1001 separator, U+2028), and PS (paragraph separator, U+2029).
1002
1003 Each of the first three conventions is used by at least one operating
1004 system as its standard newline sequence. When PCRE is built, a default
1005 can be specified. The default default is LF, which is the Unix stan-
1006 dard. When PCRE is run, the default can be overridden, either when a
1007 pattern is compiled, or when it is matched.
1008
1009 At compile time, the newline convention can be specified by the options
1010 argument of pcre_compile(), or it can be specified by special text at
1011 the start of the pattern itself; this overrides any other settings. See
1012 the pcrepattern page for details of the special character sequences.
1013
1014 In the PCRE documentation the word "newline" is used to mean "the char-
1015 acter or pair of characters that indicate a line break". The choice of
1016 newline convention affects the handling of the dot, circumflex, and
1017 dollar metacharacters, the handling of #-comments in /x mode, and, when
1018 CRLF is a recognized line ending sequence, the match position advance-
1019 ment for a non-anchored pattern. There is more detail about this in the
1020 section on pcre_exec() options below.
1021
1022 The choice of newline convention does not affect the interpretation of
1023 the \n or \r escape sequences, nor does it affect what \R matches,
1024 which is controlled in a similar way, but by separate options.
1025
1026
1027 MULTITHREADING
1028
1029 The PCRE functions can be used in multi-threading applications, with
1030 the proviso that the memory management functions pointed to by
1031 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1032 callout function pointed to by pcre_callout, are shared by all threads.
1033
1034 The compiled form of a regular expression is not altered during match-
1035 ing, so the same compiled pattern can safely be used by several threads
1036 at once.
1037
1038
1039 SAVING PRECOMPILED PATTERNS FOR LATER USE
1040
1041 The compiled form of a regular expression can be saved and re-used at a
1042 later time, possibly by a different program, and even on a host other
1043 than the one on which it was compiled. Details are given in the
1044 pcreprecompile documentation. However, compiling a regular expression
1045 with one version of PCRE for use with a different version is not guar-
1046 anteed to work and may cause crashes.
1047
1048
1049 CHECKING BUILD-TIME OPTIONS
1050
1051 int pcre_config(int what, void *where);
1052
1053 The function pcre_config() makes it possible for a PCRE client to dis-
1054 cover which optional features have been compiled into the PCRE library.
1055 The pcrebuild documentation has more details about these optional fea-
1056 tures.
1057
1058 The first argument for pcre_config() is an integer, specifying which
1059 information is required; the second argument is a pointer to a variable
1060 into which the information is placed. The following information is
1061 available:
1062
1063 PCRE_CONFIG_UTF8
1064
1065 The output is an integer that is set to one if UTF-8 support is avail-
1066 able; otherwise it is set to zero.
1067
1068 PCRE_CONFIG_UNICODE_PROPERTIES
1069
1070 The output is an integer that is set to one if support for Unicode
1071 character properties is available; otherwise it is set to zero.
1072
1073 PCRE_CONFIG_NEWLINE
1074
1075 The output is an integer whose value specifies the default character
1076 sequence that is recognized as meaning "newline". The four values that
1077 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1078 and -1 for ANY. Though they are derived from ASCII, the same values
1079 are returned in EBCDIC environments. The default should normally corre-
1080 spond to the standard sequence for your operating system.
1081
1082 PCRE_CONFIG_BSR
1083
1084 The output is an integer whose value indicates what character sequences
1085 the \R escape sequence matches by default. A value of 0 means that \R
1086 matches any Unicode line ending sequence; a value of 1 means that \R
1087 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1088 tern is compiled or matched.
1089
1090 PCRE_CONFIG_LINK_SIZE
1091
1092 The output is an integer that contains the number of bytes used for
1093 internal linkage in compiled regular expressions. The value is 2, 3, or
1094 4. Larger values allow larger regular expressions to be compiled, at
1095 the expense of slower matching. The default value of 2 is sufficient
1096 for all but the most massive patterns, since it allows the compiled
1097 pattern to be up to 64K in size.
1098
1099 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1100
1101 The output is an integer that contains the threshold above which the
1102 POSIX interface uses malloc() for output vectors. Further details are
1103 given in the pcreposix documentation.
1104
1105 PCRE_CONFIG_MATCH_LIMIT
1106
1107 The output is a long integer that gives the default limit for the num-
1108 ber of internal matching function calls in a pcre_exec() execution.
1109 Further details are given with pcre_exec() below.
1110
1111 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1112
1113 The output is a long integer that gives the default limit for the depth
1114 of recursion when calling the internal matching function in a
1115 pcre_exec() execution. Further details are given with pcre_exec()
1116 below.
1117
1118 PCRE_CONFIG_STACKRECURSE
1119
1120 The output is an integer that is set to one if internal recursion when
1121 running pcre_exec() is implemented by recursive function calls that use
1122 the stack to remember their state. This is the usual way that PCRE is
1123 compiled. The output is zero if PCRE was compiled to use blocks of data
1124 on the heap instead of recursive function calls. In this case,
1125 pcre_stack_malloc and pcre_stack_free are called to manage memory
1126 blocks on the heap, thus avoiding the use of the stack.
1127
1128
1129 COMPILING A PATTERN
1130
1131 pcre *pcre_compile(const char *pattern, int options,
1132 const char **errptr, int *erroffset,
1133 const unsigned char *tableptr);
1134
1135 pcre *pcre_compile2(const char *pattern, int options,
1136 int *errorcodeptr,
1137 const char **errptr, int *erroffset,
1138 const unsigned char *tableptr);
1139
1140 Either of the functions pcre_compile() or pcre_compile2() can be called
1141 to compile a pattern into an internal form. The only difference between
1142 the two interfaces is that pcre_compile2() has an additional argument,
1143 errorcodeptr, via which a numerical error code can be returned. To
1144 avoid too much repetition, we refer just to pcre_compile() below, but
1145 the information applies equally to pcre_compile2().
1146
1147 The pattern is a C string terminated by a binary zero, and is passed in
1148 the pattern argument. A pointer to a single block of memory that is
1149 obtained via pcre_malloc is returned. This contains the compiled code
1150 and related data. The pcre type is defined for the returned block; this
1151 is a typedef for a structure whose contents are not externally defined.
1152 It is up to the caller to free the memory (via pcre_free) when it is no
1153 longer required.
1154
1155 Although the compiled code of a PCRE regex is relocatable, that is, it
1156 does not depend on memory location, the complete pcre data block is not
1157 fully relocatable, because it may contain a copy of the tableptr argu-
1158 ment, which is an address (see below).
1159
1160 The options argument contains various bit settings that affect the com-
1161 pilation. It should be zero if no options are required. The available
1162 options are described below. Some of them (in particular, those that
1163 are compatible with Perl, but some others as well) can also be set and
1164 unset from within the pattern (see the detailed description in the
1165 pcrepattern documentation). For those options that can be different in
1166 different parts of the pattern, the contents of the options argument
1167 specifies their settings at the start of compilation and execution. The
1168 PCRE_ANCHORED, PCRE_BSR_xxx, PCRE_NEWLINE_xxx, PCRE_NO_UTF8_CHECK, and
1169 PCRE_NO_START_OPT options can be set at the time of matching as well as
1170 at compile time.
1171
1172 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1173 if compilation of a pattern fails, pcre_compile() returns NULL, and
1174 sets the variable pointed to by errptr to point to a textual error mes-
1175 sage. This is a static string that is part of the library. You must not
1176 try to free it. The offset from the start of the pattern to the byte
1177 that was being processed when the error was discovered is placed in the
1178 variable pointed to by erroffset, which must not be NULL. If it is, an
1179 immediate error is given. Some errors are not detected until checks are
1180 carried out when the whole pattern has been scanned; in this case the
1181 offset is set to the end of the pattern.
1182
1183 Note that the offset is in bytes, not characters, even in UTF-8 mode.
1184 It may point into the middle of a UTF-8 character (for example, when
1185 PCRE_ERROR_BADUTF8 is returned for an invalid UTF-8 string).
1186
1187 If pcre_compile2() is used instead of pcre_compile(), and the error-
1188 codeptr argument is not NULL, a non-zero error code number is returned
1189 via this argument in the event of an error. This is in addition to the
1190 textual error message. Error codes and messages are listed below.
1191
1192 If the final argument, tableptr, is NULL, PCRE uses a default set of
1193 character tables that are built when PCRE is compiled, using the
1194 default C locale. Otherwise, tableptr must be an address that is the
1195 result of a call to pcre_maketables(). This value is stored with the
1196 compiled pattern, and used again by pcre_exec(), unless another table
1197 pointer is passed to it. For more discussion, see the section on locale
1198 support below.
1199
1200 This code fragment shows a typical straightforward call to pcre_com-
1201 pile():
1202
1203 pcre *re;
1204 const char *error;
1205 int erroffset;
1206 re = pcre_compile(
1207 "^A.*Z", /* the pattern */
1208 0, /* default options */
1209 &error, /* for error message */
1210 &erroffset, /* for error offset */
1211 NULL); /* use default character tables */
1212
1213 The following names for option bits are defined in the pcre.h header
1214 file:
1215
1216 PCRE_ANCHORED
1217
1218 If this bit is set, the pattern is forced to be "anchored", that is, it
1219 is constrained to match only at the first matching point in the string
1220 that is being searched (the "subject string"). This effect can also be
1221 achieved by appropriate constructs in the pattern itself, which is the
1222 only way to do it in Perl.
1223
1224 PCRE_AUTO_CALLOUT
1225
1226 If this bit is set, pcre_compile() automatically inserts callout items,
1227 all with number 255, before each pattern item. For discussion of the
1228 callout facility, see the pcrecallout documentation.
1229
1230 PCRE_BSR_ANYCRLF
1231 PCRE_BSR_UNICODE
1232
1233 These options (which are mutually exclusive) control what the \R escape
1234 sequence matches. The choice is either to match only CR, LF, or CRLF,
1235 or to match any Unicode newline sequence. The default is specified when
1236 PCRE is built. It can be overridden from within the pattern, or by set-
1237 ting an option when a compiled pattern is matched.
1238
1239 PCRE_CASELESS
1240
1241 If this bit is set, letters in the pattern match both upper and lower
1242 case letters. It is equivalent to Perl's /i option, and it can be
1243 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1244 always understands the concept of case for characters whose values are
1245 less than 128, so caseless matching is always possible. For characters
1246 with higher values, the concept of case is supported if PCRE is com-
1247 piled with Unicode property support, but not otherwise. If you want to
1248 use caseless matching for characters 128 and above, you must ensure
1249 that PCRE is compiled with Unicode property support as well as with
1250 UTF-8 support.
1251
1252 PCRE_DOLLAR_ENDONLY
1253
1254 If this bit is set, a dollar metacharacter in the pattern matches only
1255 at the end of the subject string. Without this option, a dollar also
1256 matches immediately before a newline at the end of the string (but not
1257 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1258 if PCRE_MULTILINE is set. There is no equivalent to this option in
1259 Perl, and no way to set it within a pattern.
1260
1261 PCRE_DOTALL
1262
1263 If this bit is set, a dot metacharacter in the pattern matches a char-
1264 acter of any value, including one that indicates a newline. However, it
1265 only ever matches one character, even if newlines are coded as CRLF.
1266 Without this option, a dot does not match when the current position is
1267 at a newline. This option is equivalent to Perl's /s option, and it can
1268 be changed within a pattern by a (?s) option setting. A negative class
1269 such as [^a] always matches newline characters, independent of the set-
1270 ting of this option.
1271
1272 PCRE_DUPNAMES
1273
1274 If this bit is set, names used to identify capturing subpatterns need
1275 not be unique. This can be helpful for certain types of pattern when it
1276 is known that only one instance of the named subpattern can ever be
1277 matched. There are more details of named subpatterns below; see also
1278 the pcrepattern documentation.
1279
1280 PCRE_EXTENDED
1281
1282 If this bit is set, whitespace data characters in the pattern are
1283 totally ignored except when escaped or inside a character class. White-
1284 space does not include the VT character (code 11). In addition, charac-
1285 ters between an unescaped # outside a character class and the next new-
1286 line, inclusive, are also ignored. This is equivalent to Perl's /x
1287 option, and it can be changed within a pattern by a (?x) option set-
1288 ting.
1289
1290 Which characters are interpreted as newlines is controlled by the
1291 options passed to pcre_compile() or by a special sequence at the start
1292 of the pattern, as described in the section entitled "Newline conven-
1293 tions" in the pcrepattern documentation. Note that the end of this type
1294 of comment is a literal newline sequence in the pattern; escape
1295 sequences that happen to represent a newline do not count.
1296
1297 This option makes it possible to include comments inside complicated
1298 patterns. Note, however, that this applies only to data characters.
1299 Whitespace characters may never appear within special character
1300 sequences in a pattern, for example within the sequence (?( that intro-
1301 duces a conditional subpattern.
1302
1303 PCRE_EXTRA
1304
1305 This option was invented in order to turn on additional functionality
1306 of PCRE that is incompatible with Perl, but it is currently of very
1307 little use. When set, any backslash in a pattern that is followed by a
1308 letter that has no special meaning causes an error, thus reserving
1309 these combinations for future expansion. By default, as in Perl, a
1310 backslash followed by a letter with no special meaning is treated as a
1311 literal. (Perl can, however, be persuaded to give an error for this, by
1312 running it with the -w option.) There are at present no other features
1313 controlled by this option. It can also be set by a (?X) option setting
1314 within a pattern.
1315
1316 PCRE_FIRSTLINE
1317
1318 If this option is set, an unanchored pattern is required to match
1319 before or at the first newline in the subject string, though the
1320 matched text may continue over the newline.
1321
1322 PCRE_JAVASCRIPT_COMPAT
1323
1324 If this option is set, PCRE's behaviour is changed in some ways so that
1325 it is compatible with JavaScript rather than Perl. The changes are as
1326 follows:
1327
1328 (1) A lone closing square bracket in a pattern causes a compile-time
1329 error, because this is illegal in JavaScript (by default it is treated
1330 as a data character). Thus, the pattern AB]CD becomes illegal when this
1331 option is set.
1332
1333 (2) At run time, a back reference to an unset subpattern group matches
1334 an empty string (by default this causes the current matching alterna-
1335 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1336 set (assuming it can find an "a" in the subject), whereas it fails by
1337 default, for Perl compatibility.
1338
1339 PCRE_MULTILINE
1340
1341 By default, PCRE treats the subject string as consisting of a single
1342 line of characters (even if it actually contains newlines). The "start
1343 of line" metacharacter (^) matches only at the start of the string,
1344 while the "end of line" metacharacter ($) matches only at the end of
1345 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1346 is set). This is the same as Perl.
1347
1348 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1349 constructs match immediately following or immediately before internal
1350 newlines in the subject string, respectively, as well as at the very
1351 start and end. This is equivalent to Perl's /m option, and it can be
1352 changed within a pattern by a (?m) option setting. If there are no new-
1353 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1354 setting PCRE_MULTILINE has no effect.
1355
1356 PCRE_NEWLINE_CR
1357 PCRE_NEWLINE_LF
1358 PCRE_NEWLINE_CRLF
1359 PCRE_NEWLINE_ANYCRLF
1360 PCRE_NEWLINE_ANY
1361
1362 These options override the default newline definition that was chosen
1363 when PCRE was built. Setting the first or the second specifies that a
1364 newline is indicated by a single character (CR or LF, respectively).
1365 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1366 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1367 that any of the three preceding sequences should be recognized. Setting
1368 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1369 recognized. The Unicode newline sequences are the three just mentioned,
1370 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1371 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1372 (paragraph separator, U+2029). The last two are recognized only in
1373 UTF-8 mode.
1374
1375 The newline setting in the options word uses three bits that are
1376 treated as a number, giving eight possibilities. Currently only six are
1377 used (default plus the five values above). This means that if you set
1378 more than one newline option, the combination may or may not be sensi-
1379 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1380 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1381 cause an error.
1382
1383 The only time that a line break in a pattern is specially recognized
1384 when compiling is when PCRE_EXTENDED is set. CR and LF are whitespace
1385 characters, and so are ignored in this mode. Also, an unescaped # out-
1386 side a character class indicates a comment that lasts until after the
1387 next line break sequence. In other circumstances, line break sequences
1388 in patterns are treated as literal data.
1389
1390 The newline option that is set at compile time becomes the default that
1391 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1392
1393 PCRE_NO_AUTO_CAPTURE
1394
1395 If this option is set, it disables the use of numbered capturing paren-
1396 theses in the pattern. Any opening parenthesis that is not followed by
1397 ? behaves as if it were followed by ?: but named parentheses can still
1398 be used for capturing (and they acquire numbers in the usual way).
1399 There is no equivalent of this option in Perl.
1400
1401 NO_START_OPTIMIZE
1402
1403 This is an option that acts at matching time; that is, it is really an
1404 option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
1405 time, it is remembered with the compiled pattern and assumed at match-
1406 ing time. For details see the discussion of PCRE_NO_START_OPTIMIZE
1407 below.
1408
1409 PCRE_UCP
1410
1411 This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
1412 \w, and some of the POSIX character classes. By default, only ASCII
1413 characters are recognized, but if PCRE_UCP is set, Unicode properties
1414 are used instead to classify characters. More details are given in the
1415 section on generic character types in the pcrepattern page. If you set
1416 PCRE_UCP, matching one of the items it affects takes much longer. The
1417 option is available only if PCRE has been compiled with Unicode prop-
1418 erty support.
1419
1420 PCRE_UNGREEDY
1421
1422 This option inverts the "greediness" of the quantifiers so that they
1423 are not greedy by default, but become greedy if followed by "?". It is
1424 not compatible with Perl. It can also be set by a (?U) option setting
1425 within the pattern.
1426
1427 PCRE_UTF8
1428
1429 This option causes PCRE to regard both the pattern and the subject as
1430 strings of UTF-8 characters instead of single-byte character strings.
1431 However, it is available only when PCRE is built to include UTF-8 sup-
1432 port. If not, the use of this option provokes an error. Details of how
1433 this option changes the behaviour of PCRE are given in the section on
1434 UTF-8 support in the main pcre page.
1435
1436 PCRE_NO_UTF8_CHECK
1437
1438 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1439 automatically checked. There is a discussion about the validity of
1440 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1441 bytes is found, pcre_compile() returns an error. If you already know
1442 that your pattern is valid, and you want to skip this check for perfor-
1443 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1444 set, the effect of passing an invalid UTF-8 string as a pattern is
1445 undefined. It may cause your program to crash. Note that this option
1446 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1447 UTF-8 validity checking of subject strings.
1448
1449
1450 COMPILATION ERROR CODES
1451
1452 The following table lists the error codes than may be returned by
1453 pcre_compile2(), along with the error messages that may be returned by
1454 both compiling functions. As PCRE has developed, some error codes have
1455 fallen out of use. To avoid confusion, they have not been re-used.
1456
1457 0 no error
1458 1 \ at end of pattern
1459 2 \c at end of pattern
1460 3 unrecognized character follows \
1461 4 numbers out of order in {} quantifier
1462 5 number too big in {} quantifier
1463 6 missing terminating ] for character class
1464 7 invalid escape sequence in character class
1465 8 range out of order in character class
1466 9 nothing to repeat
1467 10 [this code is not in use]
1468 11 internal error: unexpected repeat
1469 12 unrecognized character after (? or (?-
1470 13 POSIX named classes are supported only within a class
1471 14 missing )
1472 15 reference to non-existent subpattern
1473 16 erroffset passed as NULL
1474 17 unknown option bit(s) set
1475 18 missing ) after comment
1476 19 [this code is not in use]
1477 20 regular expression is too large
1478 21 failed to get memory
1479 22 unmatched parentheses
1480 23 internal error: code overflow
1481 24 unrecognized character after (?<
1482 25 lookbehind assertion is not fixed length
1483 26 malformed number or name after (?(
1484 27 conditional group contains more than two branches
1485 28 assertion expected after (?(
1486 29 (?R or (?[+-]digits must be followed by )
1487 30 unknown POSIX class name
1488 31 POSIX collating elements are not supported
1489 32 this version of PCRE is not compiled with PCRE_UTF8 support
1490 33 [this code is not in use]
1491 34 character value in \x{...} sequence is too large
1492 35 invalid condition (?(0)
1493 36 \C not allowed in lookbehind assertion
1494 37 PCRE does not support \L, \l, \N, \U, or \u
1495 38 number after (?C is > 255
1496 39 closing ) for (?C expected
1497 40 recursive call could loop indefinitely
1498 41 unrecognized character after (?P
1499 42 syntax error in subpattern name (missing terminator)
1500 43 two named subpatterns have the same name
1501 44 invalid UTF-8 string
1502 45 support for \P, \p, and \X has not been compiled
1503 46 malformed \P or \p sequence
1504 47 unknown property name after \P or \p
1505 48 subpattern name is too long (maximum 32 characters)
1506 49 too many named subpatterns (maximum 10000)
1507 50 [this code is not in use]
1508 51 octal value is greater than \377 (not in UTF-8 mode)
1509 52 internal error: overran compiling workspace
1510 53 internal error: previously-checked referenced subpattern
1511 not found
1512 54 DEFINE group contains more than one branch
1513 55 repeating a DEFINE group is not allowed
1514 56 inconsistent NEWLINE options
1515 57 \g is not followed by a braced, angle-bracketed, or quoted
1516 name/number or by a plain number
1517 58 a numbered reference must not be zero
1518 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1519 60 (*VERB) not recognized
1520 61 number is too big
1521 62 subpattern name expected
1522 63 digit expected after (?+
1523 64 ] is an invalid data character in JavaScript compatibility mode
1524 65 different names for subpatterns of the same number are
1525 not allowed
1526 66 (*MARK) must have an argument
1527 67 this version of PCRE is not compiled with PCRE_UCP support
1528
1529 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1530 values may be used if the limits were changed when PCRE was built.
1531
1532
1533 STUDYING A PATTERN
1534
1535 pcre_extra *pcre_study(const pcre *code, int options
1536 const char **errptr);
1537
1538 If a compiled pattern is going to be used several times, it is worth
1539 spending more time analyzing it in order to speed up the time taken for
1540 matching. The function pcre_study() takes a pointer to a compiled pat-
1541 tern as its first argument. If studying the pattern produces additional
1542 information that will help speed up matching, pcre_study() returns a
1543 pointer to a pcre_extra block, in which the study_data field points to
1544 the results of the study.
1545
1546 The returned value from pcre_study() can be passed directly to
1547 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1548 tains other fields that can be set by the caller before the block is
1549 passed; these are described below in the section on matching a pattern.
1550
1551 If studying the pattern does not produce any useful information,
1552 pcre_study() returns NULL. In that circumstance, if the calling program
1553 wants to pass any of the other fields to pcre_exec() or
1554 pcre_dfa_exec(), it must set up its own pcre_extra block.
1555
1556 The second argument of pcre_study() contains option bits. At present,
1557 no options are defined, and this argument should always be zero.
1558
1559 The third argument for pcre_study() is a pointer for an error message.
1560 If studying succeeds (even if no data is returned), the variable it
1561 points to is set to NULL. Otherwise it is set to point to a textual
1562 error message. This is a static string that is part of the library. You
1563 must not try to free it. You should test the error pointer for NULL
1564 after calling pcre_study(), to be sure that it has run successfully.
1565
1566 This is a typical call to pcre_study():
1567
1568 pcre_extra *pe;
1569 pe = pcre_study(
1570 re, /* result of pcre_compile() */
1571 0, /* no options exist */
1572 &error); /* set to NULL or points to a message */
1573
1574 Studying a pattern does two things: first, a lower bound for the length
1575 of subject string that is needed to match the pattern is computed. This
1576 does not mean that there are any strings of that length that match, but
1577 it does guarantee that no shorter strings match. The value is used by
1578 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1579 match strings that are shorter than the lower bound. You can find out
1580 the value in a calling program via the pcre_fullinfo() function.
1581
1582 Studying a pattern is also useful for non-anchored patterns that do not
1583 have a single fixed starting character. A bitmap of possible starting
1584 bytes is created. This speeds up finding a position in the subject at
1585 which to start matching.
1586
1587 The two optimizations just described can be disabled by setting the
1588 PCRE_NO_START_OPTIMIZE option when calling pcre_exec() or
1589 pcre_dfa_exec(). You might want to do this if your pattern contains
1590 callouts or (*MARK), and you want to make use of these facilities in
1591 cases where matching fails. See the discussion of PCRE_NO_START_OPTI-
1592 MIZE below.
1593
1594
1595 LOCALE SUPPORT
1596
1597 PCRE handles caseless matching, and determines whether characters are
1598 letters, digits, or whatever, by reference to a set of tables, indexed
1599 by character value. When running in UTF-8 mode, this applies only to
1600 characters with codes less than 128. By default, higher-valued codes
1601 never match escapes such as \w or \d, but they can be tested with \p if
1602 PCRE is built with Unicode character property support. Alternatively,
1603 the PCRE_UCP option can be set at compile time; this causes \w and
1604 friends to use Unicode property support instead of built-in tables. The
1605 use of locales with Unicode is discouraged. If you are handling charac-
1606 ters with codes greater than 128, you should either use UTF-8 and Uni-
1607 code, or use locales, but not try to mix the two.
1608
1609 PCRE contains an internal set of tables that are used when the final
1610 argument of pcre_compile() is NULL. These are sufficient for many
1611 applications. Normally, the internal tables recognize only ASCII char-
1612 acters. However, when PCRE is built, it is possible to cause the inter-
1613 nal tables to be rebuilt in the default "C" locale of the local system,
1614 which may cause them to be different.
1615
1616 The internal tables can always be overridden by tables supplied by the
1617 application that calls PCRE. These may be created in a different locale
1618 from the default. As more and more applications change to using Uni-
1619 code, the need for this locale support is expected to die away.
1620
1621 External tables are built by calling the pcre_maketables() function,
1622 which has no arguments, in the relevant locale. The result can then be
1623 passed to pcre_compile() or pcre_exec() as often as necessary. For
1624 example, to build and use tables that are appropriate for the French
1625 locale (where accented characters with values greater than 128 are
1626 treated as letters), the following code could be used:
1627
1628 setlocale(LC_CTYPE, "fr_FR");
1629 tables = pcre_maketables();
1630 re = pcre_compile(..., tables);
1631
1632 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1633 if you are using Windows, the name for the French locale is "french".
1634
1635 When pcre_maketables() runs, the tables are built in memory that is
1636 obtained via pcre_malloc. It is the caller's responsibility to ensure
1637 that the memory containing the tables remains available for as long as
1638 it is needed.
1639
1640 The pointer that is passed to pcre_compile() is saved with the compiled
1641 pattern, and the same tables are used via this pointer by pcre_study()
1642 and normally also by pcre_exec(). Thus, by default, for any single pat-
1643 tern, compilation, studying and matching all happen in the same locale,
1644 but different patterns can be compiled in different locales.
1645
1646 It is possible to pass a table pointer or NULL (indicating the use of
1647 the internal tables) to pcre_exec(). Although not intended for this
1648 purpose, this facility could be used to match a pattern in a different
1649 locale from the one in which it was compiled. Passing table pointers at
1650 run time is discussed below in the section on matching a pattern.
1651
1652
1653 INFORMATION ABOUT A PATTERN
1654
1655 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1656 int what, void *where);
1657
1658 The pcre_fullinfo() function returns information about a compiled pat-
1659 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1660 less retained for backwards compability (and is documented below).
1661
1662 The first argument for pcre_fullinfo() is a pointer to the compiled
1663 pattern. The second argument is the result of pcre_study(), or NULL if
1664 the pattern was not studied. The third argument specifies which piece
1665 of information is required, and the fourth argument is a pointer to a
1666 variable to receive the data. The yield of the function is zero for
1667 success, or one of the following negative numbers:
1668
1669 PCRE_ERROR_NULL the argument code was NULL
1670 the argument where was NULL
1671 PCRE_ERROR_BADMAGIC the "magic number" was not found
1672 PCRE_ERROR_BADOPTION the value of what was invalid
1673
1674 The "magic number" is placed at the start of each compiled pattern as
1675 an simple check against passing an arbitrary memory pointer. Here is a
1676 typical call of pcre_fullinfo(), to obtain the length of the compiled
1677 pattern:
1678
1679 int rc;
1680 size_t length;
1681 rc = pcre_fullinfo(
1682 re, /* result of pcre_compile() */
1683 pe, /* result of pcre_study(), or NULL */
1684 PCRE_INFO_SIZE, /* what is required */
1685 &length); /* where to put the data */
1686
1687 The possible values for the third argument are defined in pcre.h, and
1688 are as follows:
1689
1690 PCRE_INFO_BACKREFMAX
1691
1692 Return the number of the highest back reference in the pattern. The
1693 fourth argument should point to an int variable. Zero is returned if
1694 there are no back references.
1695
1696 PCRE_INFO_CAPTURECOUNT
1697
1698 Return the number of capturing subpatterns in the pattern. The fourth
1699 argument should point to an int variable.
1700
1701 PCRE_INFO_DEFAULT_TABLES
1702
1703 Return a pointer to the internal default character tables within PCRE.
1704 The fourth argument should point to an unsigned char * variable. This
1705 information call is provided for internal use by the pcre_study() func-
1706 tion. External callers can cause PCRE to use its internal tables by
1707 passing a NULL table pointer.
1708
1709 PCRE_INFO_FIRSTBYTE
1710
1711 Return information about the first byte of any matched string, for a
1712 non-anchored pattern. The fourth argument should point to an int vari-
1713 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1714 is still recognized for backwards compatibility.)
1715
1716 If there is a fixed first byte, for example, from a pattern such as
1717 (cat|cow|coyote), its value is returned. Otherwise, if either
1718
1719 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1720 branch starts with "^", or
1721
1722 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1723 set (if it were set, the pattern would be anchored),
1724
1725 -1 is returned, indicating that the pattern matches only at the start
1726 of a subject string or after any newline within the string. Otherwise
1727 -2 is returned. For anchored patterns, -2 is returned.
1728
1729 PCRE_INFO_FIRSTTABLE
1730
1731 If the pattern was studied, and this resulted in the construction of a
1732 256-bit table indicating a fixed set of bytes for the first byte in any
1733 matching string, a pointer to the table is returned. Otherwise NULL is
1734 returned. The fourth argument should point to an unsigned char * vari-
1735 able.
1736
1737 PCRE_INFO_HASCRORLF
1738
1739 Return 1 if the pattern contains any explicit matches for CR or LF
1740 characters, otherwise 0. The fourth argument should point to an int
1741 variable. An explicit match is either a literal CR or LF character, or
1742 \r or \n.
1743
1744 PCRE_INFO_JCHANGED
1745
1746 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1747 otherwise 0. The fourth argument should point to an int variable. (?J)
1748 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1749
1750 PCRE_INFO_LASTLITERAL
1751
1752 Return the value of the rightmost literal byte that must exist in any
1753 matched string, other than at its start, if such a byte has been
1754 recorded. The fourth argument should point to an int variable. If there
1755 is no such byte, -1 is returned. For anchored patterns, a last literal
1756 byte is recorded only if it follows something of variable length. For
1757 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1758 /^a\dz\d/ the returned value is -1.
1759
1760 PCRE_INFO_MINLENGTH
1761
1762 If the pattern was studied and a minimum length for matching subject
1763 strings was computed, its value is returned. Otherwise the returned
1764 value is -1. The value is a number of characters, not bytes (this may
1765 be relevant in UTF-8 mode). The fourth argument should point to an int
1766 variable. A non-negative value is a lower bound to the length of any
1767 matching string. There may not be any strings of that length that do
1768 actually match, but every string that does match is at least that long.
1769
1770 PCRE_INFO_NAMECOUNT
1771 PCRE_INFO_NAMEENTRYSIZE
1772 PCRE_INFO_NAMETABLE
1773
1774 PCRE supports the use of named as well as numbered capturing parenthe-
1775 ses. The names are just an additional way of identifying the parenthe-
1776 ses, which still acquire numbers. Several convenience functions such as
1777 pcre_get_named_substring() are provided for extracting captured sub-
1778 strings by name. It is also possible to extract the data directly, by
1779 first converting the name to a number in order to access the correct
1780 pointers in the output vector (described with pcre_exec() below). To do
1781 the conversion, you need to use the name-to-number map, which is
1782 described by these three values.
1783
1784 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1785 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1786 of each entry; both of these return an int value. The entry size
1787 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1788 a pointer to the first entry of the table (a pointer to char). The
1789 first two bytes of each entry are the number of the capturing parenthe-
1790 sis, most significant byte first. The rest of the entry is the corre-
1791 sponding name, zero terminated.
1792
1793 The names are in alphabetical order. Duplicate names may appear if (?|
1794 is used to create multiple groups with the same number, as described in
1795 the section on duplicate subpattern numbers in the pcrepattern page.
1796 Duplicate names for subpatterns with different numbers are permitted
1797 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
1798 appear in the table in the order in which they were found in the pat-
1799 tern. In the absence of (?| this is the order of increasing number;
1800 when (?| is used this is not necessarily the case because later subpat-
1801 terns may have lower numbers.
1802
1803 As a simple example of the name/number table, consider the following
1804 pattern (assume PCRE_EXTENDED is set, so white space - including new-
1805 lines - is ignored):
1806
1807 (?<date> (?<year>(\d\d)?\d\d) -
1808 (?<month>\d\d) - (?<day>\d\d) )
1809
1810 There are four named subpatterns, so the table has four entries, and
1811 each entry in the table is eight bytes long. The table is as follows,
1812 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1813 as ??:
1814
1815 00 01 d a t e 00 ??
1816 00 05 d a y 00 ?? ??
1817 00 04 m o n t h 00
1818 00 02 y e a r 00 ??
1819
1820 When writing code to extract data from named subpatterns using the
1821 name-to-number map, remember that the length of the entries is likely
1822 to be different for each compiled pattern.
1823
1824 PCRE_INFO_OKPARTIAL
1825
1826 Return 1 if the pattern can be used for partial matching with
1827 pcre_exec(), otherwise 0. The fourth argument should point to an int
1828 variable. From release 8.00, this always returns 1, because the
1829 restrictions that previously applied to partial matching have been
1830 lifted. The pcrepartial documentation gives details of partial match-
1831 ing.
1832
1833 PCRE_INFO_OPTIONS
1834
1835 Return a copy of the options with which the pattern was compiled. The
1836 fourth argument should point to an unsigned long int variable. These
1837 option bits are those specified in the call to pcre_compile(), modified
1838 by any top-level option settings at the start of the pattern itself. In
1839 other words, they are the options that will be in force when matching
1840 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1841 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1842 and PCRE_EXTENDED.
1843
1844 A pattern is automatically anchored by PCRE if all of its top-level
1845 alternatives begin with one of the following:
1846
1847 ^ unless PCRE_MULTILINE is set
1848 \A always
1849 \G always
1850 .* if PCRE_DOTALL is set and there are no back
1851 references to the subpattern in which .* appears
1852
1853 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1854 by pcre_fullinfo().
1855
1856 PCRE_INFO_SIZE
1857
1858 Return the size of the compiled pattern, that is, the value that was
1859 passed as the argument to pcre_malloc() when PCRE was getting memory in
1860 which to place the compiled data. The fourth argument should point to a
1861 size_t variable.
1862
1863 PCRE_INFO_STUDYSIZE
1864
1865 Return the size of the data block pointed to by the study_data field in
1866 a pcre_extra block. That is, it is the value that was passed to
1867 pcre_malloc() when PCRE was getting memory into which to place the data
1868 created by pcre_study(). If pcre_extra is NULL, or there is no study
1869 data, zero is returned. The fourth argument should point to a size_t
1870 variable.
1871
1872
1873 OBSOLETE INFO FUNCTION
1874
1875 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1876
1877 The pcre_info() function is now obsolete because its interface is too
1878 restrictive to return all the available data about a compiled pattern.
1879 New programs should use pcre_fullinfo() instead. The yield of
1880 pcre_info() is the number of capturing subpatterns, or one of the fol-
1881 lowing negative numbers:
1882
1883 PCRE_ERROR_NULL the argument code was NULL
1884 PCRE_ERROR_BADMAGIC the "magic number" was not found
1885
1886 If the optptr argument is not NULL, a copy of the options with which
1887 the pattern was compiled is placed in the integer it points to (see
1888 PCRE_INFO_OPTIONS above).
1889
1890 If the pattern is not anchored and the firstcharptr argument is not
1891 NULL, it is used to pass back information about the first character of
1892 any matched string (see PCRE_INFO_FIRSTBYTE above).
1893
1894
1895 REFERENCE COUNTS
1896
1897 int pcre_refcount(pcre *code, int adjust);
1898
1899 The pcre_refcount() function is used to maintain a reference count in
1900 the data block that contains a compiled pattern. It is provided for the
1901 benefit of applications that operate in an object-oriented manner,
1902 where different parts of the application may be using the same compiled
1903 pattern, but you want to free the block when they are all done.
1904
1905 When a pattern is compiled, the reference count field is initialized to
1906 zero. It is changed only by calling this function, whose action is to
1907 add the adjust value (which may be positive or negative) to it. The
1908 yield of the function is the new value. However, the value of the count
1909 is constrained to lie between 0 and 65535, inclusive. If the new value
1910 is outside these limits, it is forced to the appropriate limit value.
1911
1912 Except when it is zero, the reference count is not correctly preserved
1913 if a pattern is compiled on one host and then transferred to a host
1914 whose byte-order is different. (This seems a highly unlikely scenario.)
1915
1916
1917 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1918
1919 int pcre_exec(const pcre *code, const pcre_extra *extra,
1920 const char *subject, int length, int startoffset,
1921 int options, int *ovector, int ovecsize);
1922
1923 The function pcre_exec() is called to match a subject string against a
1924 compiled pattern, which is passed in the code argument. If the pattern
1925 was studied, the result of the study should be passed in the extra
1926 argument. This function is the main matching facility of the library,
1927 and it operates in a Perl-like manner. For specialist use there is also
1928 an alternative matching function, which is described below in the sec-
1929 tion about the pcre_dfa_exec() function.
1930
1931 In most applications, the pattern will have been compiled (and option-
1932 ally studied) in the same process that calls pcre_exec(). However, it
1933 is possible to save compiled patterns and study data, and then use them
1934 later in different processes, possibly even on different hosts. For a
1935 discussion about this, see the pcreprecompile documentation.
1936
1937 Here is an example of a simple call to pcre_exec():
1938
1939 int rc;
1940 int ovector[30];
1941 rc = pcre_exec(
1942 re, /* result of pcre_compile() */
1943 NULL, /* we didn't study the pattern */
1944 "some string", /* the subject string */
1945 11, /* the length of the subject string */
1946 0, /* start at offset 0 in the subject */
1947 0, /* default options */
1948 ovector, /* vector of integers for substring information */
1949 30); /* number of elements (NOT size in bytes) */
1950
1951 Extra data for pcre_exec()
1952
1953 If the extra argument is not NULL, it must point to a pcre_extra data
1954 block. The pcre_study() function returns such a block (when it doesn't
1955 return NULL), but you can also create one for yourself, and pass addi-
1956 tional information in it. The pcre_extra block contains the following
1957 fields (not necessarily in this order):
1958
1959 unsigned long int flags;
1960 void *study_data;
1961 unsigned long int match_limit;
1962 unsigned long int match_limit_recursion;
1963 void *callout_data;
1964 const unsigned char *tables;
1965 unsigned char **mark;
1966
1967 The flags field is a bitmap that specifies which of the other fields
1968 are set. The flag bits are:
1969
1970 PCRE_EXTRA_STUDY_DATA
1971 PCRE_EXTRA_MATCH_LIMIT
1972 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1973 PCRE_EXTRA_CALLOUT_DATA
1974 PCRE_EXTRA_TABLES
1975 PCRE_EXTRA_MARK
1976
1977 Other flag bits should be set to zero. The study_data field is set in
1978 the pcre_extra block that is returned by pcre_study(), together with
1979 the appropriate flag bit. You should not set this yourself, but you may
1980 add to the block by setting the other fields and their corresponding
1981 flag bits.
1982
1983 The match_limit field provides a means of preventing PCRE from using up
1984 a vast amount of resources when running patterns that are not going to
1985 match, but which have a very large number of possibilities in their
1986 search trees. The classic example is a pattern that uses nested unlim-
1987 ited repeats.
1988
1989 Internally, PCRE uses a function called match() which it calls repeat-
1990 edly (sometimes recursively). The limit set by match_limit is imposed
1991 on the number of times this function is called during a match, which
1992 has the effect of limiting the amount of backtracking that can take
1993 place. For patterns that are not anchored, the count restarts from zero
1994 for each position in the subject string.
1995
1996 The default value for the limit can be set when PCRE is built; the
1997 default default is 10 million, which handles all but the most extreme
1998 cases. You can override the default by suppling pcre_exec() with a
1999 pcre_extra block in which match_limit is set, and
2000 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
2001 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2002
2003 The match_limit_recursion field is similar to match_limit, but instead
2004 of limiting the total number of times that match() is called, it limits
2005 the depth of recursion. The recursion depth is a smaller number than
2006 the total number of calls, because not all calls to match() are recur-
2007 sive. This limit is of use only if it is set smaller than match_limit.
2008
2009 Limiting the recursion depth limits the amount of stack that can be
2010 used, or, when PCRE has been compiled to use memory on the heap instead
2011 of the stack, the amount of heap memory that can be used.
2012
2013 The default value for match_limit_recursion can be set when PCRE is
2014 built; the default default is the same value as the default for
2015 match_limit. You can override the default by suppling pcre_exec() with
2016 a pcre_extra block in which match_limit_recursion is set, and
2017 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
2018 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
2019
2020 The callout_data field is used in conjunction with the "callout" fea-
2021 ture, and is described in the pcrecallout documentation.
2022
2023 The tables field is used to pass a character tables pointer to
2024 pcre_exec(); this overrides the value that is stored with the compiled
2025 pattern. A non-NULL value is stored with the compiled pattern only if
2026 custom tables were supplied to pcre_compile() via its tableptr argu-
2027 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
2028 PCRE's internal tables to be used. This facility is helpful when re-
2029 using patterns that have been saved after compiling with an external
2030 set of tables, because the external tables might be at a different
2031 address when pcre_exec() is called. See the pcreprecompile documenta-
2032 tion for a discussion of saving compiled patterns for later use.
2033
2034 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
2035 set to point to a char * variable. If the pattern contains any back-
2036 tracking control verbs such as (*MARK:NAME), and the execution ends up
2037 with a name to pass back, a pointer to the name string (zero termi-
2038 nated) is placed in the variable pointed to by the mark field. The
2039 names are within the compiled pattern; if you wish to retain such a
2040 name you must copy it before freeing the memory of a compiled pattern.
2041 If there is no name to pass back, the variable pointed to by the mark
2042 field set to NULL. For details of the backtracking control verbs, see
2043 the section entitled "Backtracking control" in the pcrepattern documen-
2044 tation.
2045
2046 Option bits for pcre_exec()
2047
2048 The unused bits of the options argument for pcre_exec() must be zero.
2049 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
2050 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
2051 PCRE_NO_START_OPTIMIZE, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL_SOFT, and
2052 PCRE_PARTIAL_HARD.
2053
2054 PCRE_ANCHORED
2055
2056 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
2057 matching position. If a pattern was compiled with PCRE_ANCHORED, or
2058 turned out to be anchored by virtue of its contents, it cannot be made
2059 unachored at matching time.
2060
2061 PCRE_BSR_ANYCRLF
2062 PCRE_BSR_UNICODE
2063
2064 These options (which are mutually exclusive) control what the \R escape
2065 sequence matches. The choice is either to match only CR, LF, or CRLF,
2066 or to match any Unicode newline sequence. These options override the
2067 choice that was made or defaulted when the pattern was compiled.
2068
2069 PCRE_NEWLINE_CR
2070 PCRE_NEWLINE_LF
2071 PCRE_NEWLINE_CRLF
2072 PCRE_NEWLINE_ANYCRLF
2073 PCRE_NEWLINE_ANY
2074
2075 These options override the newline definition that was chosen or
2076 defaulted when the pattern was compiled. For details, see the descrip-
2077 tion of pcre_compile() above. During matching, the newline choice
2078 affects the behaviour of the dot, circumflex, and dollar metacharac-
2079 ters. It may also alter the way the match position is advanced after a
2080 match failure for an unanchored pattern.
2081
2082 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
2083 set, and a match attempt for an unanchored pattern fails when the cur-
2084 rent position is at a CRLF sequence, and the pattern contains no
2085 explicit matches for CR or LF characters, the match position is
2086 advanced by two characters instead of one, in other words, to after the
2087 CRLF.
2088
2089 The above rule is a compromise that makes the most common cases work as
2090 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
2091 option is not set), it does not match the string "\r\nA" because, after
2092 failing at the start, it skips both the CR and the LF before retrying.
2093 However, the pattern [\r\n]A does match that string, because it con-
2094 tains an explicit CR or LF reference, and so advances only by one char-
2095 acter after the first failure.
2096
2097 An explicit match for CR of LF is either a literal appearance of one of
2098 those characters, or one of the \r or \n escape sequences. Implicit
2099 matches such as [^X] do not count, nor does \s (which includes CR and
2100 LF in the characters that it matches).
2101
2102 Notwithstanding the above, anomalous effects may still occur when CRLF
2103 is a valid newline sequence and explicit \r or \n escapes appear in the
2104 pattern.
2105
2106 PCRE_NOTBOL
2107
2108 This option specifies that first character of the subject string is not
2109 the beginning of a line, so the circumflex metacharacter should not
2110 match before it. Setting this without PCRE_MULTILINE (at compile time)
2111 causes circumflex never to match. This option affects only the behav-
2112 iour of the circumflex metacharacter. It does not affect \A.
2113
2114 PCRE_NOTEOL
2115
2116 This option specifies that the end of the subject string is not the end
2117 of a line, so the dollar metacharacter should not match it nor (except
2118 in multiline mode) a newline immediately before it. Setting this with-
2119 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2120 option affects only the behaviour of the dollar metacharacter. It does
2121 not affect \Z or \z.
2122
2123 PCRE_NOTEMPTY
2124
2125 An empty string is not considered to be a valid match if this option is
2126 set. If there are alternatives in the pattern, they are tried. If all
2127 the alternatives match the empty string, the entire match fails. For
2128 example, if the pattern
2129
2130 a?b?
2131
2132 is applied to a string not beginning with "a" or "b", it matches an
2133 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2134 match is not valid, so PCRE searches further into the string for occur-
2135 rences of "a" or "b".
2136
2137 PCRE_NOTEMPTY_ATSTART
2138
2139 This is like PCRE_NOTEMPTY, except that an empty string match that is
2140 not at the start of the subject is permitted. If the pattern is
2141 anchored, such a match can occur only if the pattern contains \K.
2142
2143 Perl has no direct equivalent of PCRE_NOTEMPTY or
2144 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2145 match of the empty string within its split() function, and when using
2146 the /g modifier. It is possible to emulate Perl's behaviour after
2147 matching a null string by first trying the match again at the same off-
2148 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2149 fails, by advancing the starting offset (see below) and trying an ordi-
2150 nary match again. There is some code that demonstrates how to do this
2151 in the pcredemo sample program. In the most general case, you have to
2152 check to see if the newline convention recognizes CRLF as a newline,
2153 and if so, and the current character is CR followed by LF, advance the
2154 starting offset by two characters instead of one.
2155
2156 PCRE_NO_START_OPTIMIZE
2157
2158 There are a number of optimizations that pcre_exec() uses at the start
2159 of a match, in order to speed up the process. For example, if it is
2160 known that an unanchored match must start with a specific character, it
2161 searches the subject for that character, and fails immediately if it
2162 cannot find it, without actually running the main matching function.
2163 This means that a special item such as (*COMMIT) at the start of a pat-
2164 tern is not considered until after a suitable starting point for the
2165 match has been found. When callouts or (*MARK) items are in use, these
2166 "start-up" optimizations can cause them to be skipped if the pattern is
2167 never actually used. The start-up optimizations are in effect a pre-
2168 scan of the subject that takes place before the pattern is run.
2169
2170 The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
2171 possibly causing performance to suffer, but ensuring that in cases
2172 where the result is "no match", the callouts do occur, and that items
2173 such as (*COMMIT) and (*MARK) are considered at every possible starting
2174 position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
2175 compile time, it cannot be unset at matching time.
2176
2177 Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
2178 operation. Consider the pattern
2179
2180 (*COMMIT)ABC
2181
2182 When this is compiled, PCRE records the fact that a match must start
2183 with the character "A". Suppose the subject string is "DEFABC". The
2184 start-up optimization scans along the subject, finds "A" and runs the
2185 first match attempt from there. The (*COMMIT) item means that the pat-
2186 tern must match the current starting position, which in this case, it
2187 does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
2188 set, the initial scan along the subject string does not happen. The
2189 first match attempt is run starting from "D" and when this fails,
2190 (*COMMIT) prevents any further matches being tried, so the overall
2191 result is "no match". If the pattern is studied, more start-up opti-
2192 mizations may be used. For example, a minimum length for the subject
2193 may be recorded. Consider the pattern
2194
2195 (*MARK:A)(X|Y)
2196
2197 The minimum length for a match is one character. If the subject is
2198 "ABC", there will be attempts to match "ABC", "BC", "C", and then
2199 finally an empty string. If the pattern is studied, the final attempt
2200 does not take place, because PCRE knows that the subject is too short,
2201 and so the (*MARK) is never encountered. In this case, studying the
2202 pattern does not affect the overall match result, which is still "no
2203 match", but it does affect the auxiliary information that is returned.
2204
2205 PCRE_NO_UTF8_CHECK
2206
2207 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2208 UTF-8 string is automatically checked when pcre_exec() is subsequently
2209 called. The value of startoffset is also checked to ensure that it
2210 points to the start of a UTF-8 character. There is a discussion about
2211 the validity of UTF-8 strings in the section on UTF-8 support in the
2212 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2213 pcre_exec() returns the error PCRE_ERROR_BADUTF8 or, if PCRE_PAR-
2214 TIAL_HARD is set and the problem is a truncated UTF-8 character at the
2215 end of the subject, PCRE_ERROR_SHORTUTF8. If startoffset contains a
2216 value that does not point to the start of a UTF-8 character (or to the
2217 end of the subject), PCRE_ERROR_BADUTF8_OFFSET is returned.
2218
2219 If you already know that your subject is valid, and you want to skip
2220 these checks for performance reasons, you can set the
2221 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2222 do this for the second and subsequent calls to pcre_exec() if you are
2223 making repeated calls to find all the matches in a single subject
2224 string. However, you should be sure that the value of startoffset
2225 points to the start of a UTF-8 character (or the end of the subject).
2226 When PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid UTF-8
2227 string as a subject or an invalid value of startoffset is undefined.
2228 Your program may crash.
2229
2230 PCRE_PARTIAL_HARD
2231 PCRE_PARTIAL_SOFT
2232
2233 These options turn on the partial matching feature. For backwards com-
2234 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2235 match occurs if the end of the subject string is reached successfully,
2236 but there are not enough subject characters to complete the match. If
2237 this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
2238 matching continues by testing any remaining alternatives. Only if no
2239 complete match can be found is PCRE_ERROR_PARTIAL returned instead of
2240 PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
2241 caller is prepared to handle a partial match, but only if no complete
2242 match can be found.
2243
2244 If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
2245 case, if a partial match is found, pcre_exec() immediately returns
2246 PCRE_ERROR_PARTIAL, without considering any other alternatives. In
2247 other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
2248 ered to be more important that an alternative complete match.
2249
2250 In both cases, the portion of the string that was inspected when the
2251 partial match was found is set as the first matching string. There is a
2252 more detailed discussion of partial and multi-segment matching, with
2253 examples, in the pcrepartial documentation.
2254
2255 The string to be matched by pcre_exec()
2256
2257 The subject string is passed to pcre_exec() as a pointer in subject, a
2258 length (in bytes) in length, and a starting byte offset in startoffset.
2259 If this is negative or greater than the length of the subject,
2260 pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
2261 zero, the search for a match starts at the beginning of the subject,
2262 and this is by far the most common case. In UTF-8 mode, the byte offset
2263 must point to the start of a UTF-8 character (or the end of the sub-
2264 ject). Unlike the pattern string, the subject may contain binary zero
2265 bytes.
2266
2267 A non-zero starting offset is useful when searching for another match
2268 in the same subject by calling pcre_exec() again after a previous suc-
2269 cess. Setting startoffset differs from just passing over a shortened
2270 string and setting PCRE_NOTBOL in the case of a pattern that begins
2271 with any kind of lookbehind. For example, consider the pattern
2272
2273 \Biss\B
2274
2275 which finds occurrences of "iss" in the middle of words. (\B matches
2276 only if the current position in the subject is not a word boundary.)
2277 When applied to the string "Mississipi" the first call to pcre_exec()
2278 finds the first occurrence. If pcre_exec() is called again with just
2279 the remainder of the subject, namely "issipi", it does not match,
2280 because \B is always false at the start of the subject, which is deemed
2281 to be a word boundary. However, if pcre_exec() is passed the entire
2282 string again, but with startoffset set to 4, it finds the second occur-
2283 rence of "iss" because it is able to look behind the starting point to
2284 discover that it is preceded by a letter.
2285
2286 Finding all the matches in a subject is tricky when the pattern can
2287 match an empty string. It is possible to emulate Perl's /g behaviour by
2288 first trying the match again at the same offset, with the
2289 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
2290 fails, advancing the starting offset and trying an ordinary match
2291 again. There is some code that demonstrates how to do this in the pcre-
2292 demo sample program. In the most general case, you have to check to see
2293 if the newline convention recognizes CRLF as a newline, and if so, and
2294 the current character is CR followed by LF, advance the starting offset
2295 by two characters instead of one.
2296
2297 If a non-zero starting offset is passed when the pattern is anchored,
2298 one attempt to match at the given offset is made. This can only succeed
2299 if the pattern does not require the match to be at the start of the
2300 subject.
2301
2302 How pcre_exec() returns captured substrings
2303
2304 In general, a pattern matches a certain portion of the subject, and in
2305 addition, further substrings from the subject may be picked out by
2306 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2307 this is called "capturing" in what follows, and the phrase "capturing
2308 subpattern" is used for a fragment of a pattern that picks out a sub-
2309 string. PCRE supports several other kinds of parenthesized subpattern
2310 that do not cause substrings to be captured.
2311
2312 Captured substrings are returned to the caller via a vector of integers
2313 whose address is passed in ovector. The number of elements in the vec-
2314 tor is passed in ovecsize, which must be a non-negative number. Note:
2315 this argument is NOT the size of ovector in bytes.
2316
2317 The first two-thirds of the vector is used to pass back captured sub-
2318 strings, each substring using a pair of integers. The remaining third
2319 of the vector is used as workspace by pcre_exec() while matching cap-
2320 turing subpatterns, and is not available for passing back information.
2321 The number passed in ovecsize should always be a multiple of three. If
2322 it is not, it is rounded down.
2323
2324 When a match is successful, information about captured substrings is
2325 returned in pairs of integers, starting at the beginning of ovector,
2326 and continuing up to two-thirds of its length at the most. The first
2327 element of each pair is set to the byte offset of the first character
2328 in a substring, and the second is set to the byte offset of the first
2329 character after the end of a substring. Note: these values are always
2330 byte offsets, even in UTF-8 mode. They are not character counts.
2331
2332 The first pair of integers, ovector[0] and ovector[1], identify the
2333 portion of the subject string matched by the entire pattern. The next
2334 pair is used for the first capturing subpattern, and so on. The value
2335 returned by pcre_exec() is one more than the highest numbered pair that
2336 has been set. For example, if two substrings have been captured, the
2337 returned value is 3. If there are no capturing subpatterns, the return
2338 value from a successful match is 1, indicating that just the first pair
2339 of offsets has been set.
2340
2341 If a capturing subpattern is matched repeatedly, it is the last portion
2342 of the string that it matched that is returned.
2343
2344 If the vector is too small to hold all the captured substring offsets,
2345 it is used as far as possible (up to two-thirds of its length), and the
2346 function returns a value of zero. If the substring offsets are not of
2347 interest, pcre_exec() may be called with ovector passed as NULL and
2348 ovecsize as zero. However, if the pattern contains back references and
2349 the ovector is not big enough to remember the related substrings, PCRE
2350 has to get additional memory for use during matching. Thus it is usu-
2351 ally advisable to supply an ovector.
2352
2353 The pcre_fullinfo() function can be used to find out how many capturing
2354 subpatterns there are in a compiled pattern. The smallest size for
2355 ovector that will allow for n captured substrings, in addition to the
2356 offsets of the substring matched by the whole pattern, is (n+1)*3.
2357
2358 It is possible for capturing subpattern number n+1 to match some part
2359 of the subject when subpattern n has not been used at all. For example,
2360 if the string "abc" is matched against the pattern (a|(z))(bc) the
2361 return from the function is 4, and subpatterns 1 and 3 are matched, but
2362 2 is not. When this happens, both values in the offset pairs corre-
2363 sponding to unused subpatterns are set to -1.
2364
2365 Offset values that correspond to unused subpatterns at the end of the
2366 expression are also set to -1. For example, if the string "abc" is
2367 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2368 matched. The return from the function is 2, because the highest used
2369 capturing subpattern number is 1, and the offsets for for the second
2370 and third capturing subpatterns (assuming the vector is large enough,
2371 of course) are set to -1.
2372
2373 Note: Elements of ovector that do not correspond to capturing parenthe-
2374 ses in the pattern are never changed. That is, if a pattern contains n
2375 capturing parentheses, no more than ovector[0] to ovector[2n+1] are set
2376 by pcre_exec(). The other elements retain whatever values they previ-
2377 ously had.
2378
2379 Some convenience functions are provided for extracting the captured
2380 substrings as separate strings. These are described below.
2381
2382 Error return values from pcre_exec()
2383
2384 If pcre_exec() fails, it returns a negative number. The following are
2385 defined in the header file:
2386
2387 PCRE_ERROR_NOMATCH (-1)
2388
2389 The subject string did not match the pattern.
2390
2391 PCRE_ERROR_NULL (-2)
2392
2393 Either code or subject was passed as NULL, or ovector was NULL and
2394 ovecsize was not zero.
2395
2396 PCRE_ERROR_BADOPTION (-3)
2397
2398 An unrecognized bit was set in the options argument.
2399
2400 PCRE_ERROR_BADMAGIC (-4)
2401
2402 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2403 to catch the case when it is passed a junk pointer and to detect when a
2404 pattern that was compiled in an environment of one endianness is run in
2405 an environment with the other endianness. This is the error that PCRE
2406 gives when the magic number is not present.
2407
2408 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2409
2410 While running the pattern match, an unknown item was encountered in the
2411 compiled pattern. This error could be caused by a bug in PCRE or by
2412 overwriting of the compiled pattern.
2413
2414 PCRE_ERROR_NOMEMORY (-6)
2415
2416 If a pattern contains back references, but the ovector that is passed
2417 to pcre_exec() is not big enough to remember the referenced substrings,
2418 PCRE gets a block of memory at the start of matching to use for this
2419 purpose. If the call via pcre_malloc() fails, this error is given. The
2420 memory is automatically freed at the end of matching.
2421
2422 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2423 This can happen only when PCRE has been compiled with --disable-stack-
2424 for-recursion.
2425
2426 PCRE_ERROR_NOSUBSTRING (-7)
2427
2428 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2429 and pcre_get_substring_list() functions (see below). It is never
2430 returned by pcre_exec().
2431
2432 PCRE_ERROR_MATCHLIMIT (-8)
2433
2434 The backtracking limit, as specified by the match_limit field in a
2435 pcre_extra structure (or defaulted) was reached. See the description
2436 above.
2437
2438 PCRE_ERROR_CALLOUT (-9)
2439
2440 This error is never generated by pcre_exec() itself. It is provided for
2441 use by callout functions that want to yield a distinctive error code.
2442 See the pcrecallout documentation for details.
2443
2444 PCRE_ERROR_BADUTF8 (-10)
2445
2446 A string that contains an invalid UTF-8 byte sequence was passed as a
2447 subject. However, if PCRE_PARTIAL_HARD is set and the problem is a
2448 truncated UTF-8 character at the end of the subject, PCRE_ERROR_SHORT-
2449 UTF8 is used instead.
2450
2451 PCRE_ERROR_BADUTF8_OFFSET (-11)
2452
2453 The UTF-8 byte sequence that was passed as a subject was valid, but the
2454 value of startoffset did not point to the beginning of a UTF-8 charac-
2455 ter or the end of the subject.
2456
2457 PCRE_ERROR_PARTIAL (-12)
2458
2459 The subject string did not match, but it did match partially. See the
2460 pcrepartial documentation for details of partial matching.
2461
2462 PCRE_ERROR_BADPARTIAL (-13)
2463
2464 This code is no longer in use. It was formerly returned when the
2465 PCRE_PARTIAL option was used with a compiled pattern containing items
2466 that were not supported for partial matching. From release 8.00
2467 onwards, there are no restrictions on partial matching.
2468
2469 PCRE_ERROR_INTERNAL (-14)
2470
2471 An unexpected internal error has occurred. This error could be caused
2472 by a bug in PCRE or by overwriting of the compiled pattern.
2473
2474 PCRE_ERROR_BADCOUNT (-15)
2475
2476 This error is given if the value of the ovecsize argument is negative.
2477
2478 PCRE_ERROR_RECURSIONLIMIT (-21)
2479
2480 The internal recursion limit, as specified by the match_limit_recursion
2481 field in a pcre_extra structure (or defaulted) was reached. See the
2482 description above.
2483
2484 PCRE_ERROR_BADNEWLINE (-23)
2485
2486 An invalid combination of PCRE_NEWLINE_xxx options was given.
2487
2488 PCRE_ERROR_BADOFFSET (-24)
2489
2490 The value of startoffset was negative or greater than the length of the
2491 subject, that is, the value in length.
2492
2493 PCRE_ERROR_SHORTUTF8 (-25)
2494
2495 The subject string ended with an incomplete (truncated) UTF-8 charac-
2496 ter, and the PCRE_PARTIAL_HARD option was set. Without this option,
2497 PCRE_ERROR_BADUTF8 is returned in this situation.
2498
2499 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2500
2501
2502 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2503
2504 int pcre_copy_substring(const char *subject, int *ovector,
2505 int stringcount, int stringnumber, char *buffer,
2506 int buffersize);
2507
2508 int pcre_get_substring(const char *subject, int *ovector,
2509 int stringcount, int stringnumber,
2510 const char **stringptr);
2511
2512 int pcre_get_substring_list(const char *subject,
2513 int *ovector, int stringcount, const char ***listptr);
2514
2515 Captured substrings can be accessed directly by using the offsets
2516 returned by pcre_exec() in ovector. For convenience, the functions
2517 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2518 string_list() are provided for extracting captured substrings as new,
2519 separate, zero-terminated strings. These functions identify substrings
2520 by number. The next section describes functions for extracting named
2521 substrings.
2522
2523 A substring that contains a binary zero is correctly extracted and has
2524 a further zero added on the end, but the result is not, of course, a C
2525 string. However, you can process such a string by referring to the
2526 length that is returned by pcre_copy_substring() and pcre_get_sub-
2527 string(). Unfortunately, the interface to pcre_get_substring_list() is
2528 not adequate for handling strings containing binary zeros, because the
2529 end of the final string is not independently indicated.
2530
2531 The first three arguments are the same for all three of these func-
2532 tions: subject is the subject string that has just been successfully
2533 matched, ovector is a pointer to the vector of integer offsets that was
2534 passed to pcre_exec(), and stringcount is the number of substrings that
2535 were captured by the match, including the substring that matched the
2536 entire regular expression. This is the value returned by pcre_exec() if
2537 it is greater than zero. If pcre_exec() returned zero, indicating that
2538 it ran out of space in ovector, the value passed as stringcount should
2539 be the number of elements in the vector divided by three.
2540
2541 The functions pcre_copy_substring() and pcre_get_substring() extract a
2542 single substring, whose number is given as stringnumber. A value of
2543 zero extracts the substring that matched the entire pattern, whereas
2544 higher values extract the captured substrings. For pcre_copy_sub-
2545 string(), the string is placed in buffer, whose length is given by
2546 buffersize, while for pcre_get_substring() a new block of memory is
2547 obtained via pcre_malloc, and its address is returned via stringptr.
2548 The yield of the function is the length of the string, not including
2549 the terminating zero, or one of these error codes:
2550
2551 PCRE_ERROR_NOMEMORY (-6)
2552
2553 The buffer was too small for pcre_copy_substring(), or the attempt to
2554 get memory failed for pcre_get_substring().
2555
2556 PCRE_ERROR_NOSUBSTRING (-7)
2557
2558 There is no substring whose number is stringnumber.
2559
2560 The pcre_get_substring_list() function extracts all available sub-
2561 strings and builds a list of pointers to them. All this is done in a
2562 single block of memory that is obtained via pcre_malloc. The address of
2563 the memory block is returned via listptr, which is also the start of
2564 the list of string pointers. The end of the list is marked by a NULL
2565 pointer. The yield of the function is zero if all went well, or the
2566 error code
2567
2568 PCRE_ERROR_NOMEMORY (-6)
2569
2570 if the attempt to get the memory block failed.
2571
2572 When any of these functions encounter a substring that is unset, which
2573 can happen when capturing subpattern number n+1 matches some part of
2574 the subject, but subpattern n has not been used at all, they return an
2575 empty string. This can be distinguished from a genuine zero-length sub-
2576 string by inspecting the appropriate offset in ovector, which is nega-
2577 tive for unset substrings.
2578
2579 The two convenience functions pcre_free_substring() and pcre_free_sub-
2580 string_list() can be used to free the memory returned by a previous
2581 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2582 tively. They do nothing more than call the function pointed to by
2583 pcre_free, which of course could be called directly from a C program.
2584 However, PCRE is used in some situations where it is linked via a spe-
2585 cial interface to another programming language that cannot use
2586 pcre_free directly; it is for these cases that the functions are pro-
2587 vided.
2588
2589
2590 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2591
2592 int pcre_get_stringnumber(const pcre *code,
2593 const char *name);
2594
2595 int pcre_copy_named_substring(const pcre *code,
2596 const char *subject, int *ovector,
2597 int stringcount, const char *stringname,
2598 char *buffer, int buffersize);
2599
2600 int pcre_get_named_substring(const pcre *code,
2601 const char *subject, int *ovector,
2602 int stringcount, const char *stringname,
2603 const char **stringptr);
2604
2605 To extract a substring by name, you first have to find associated num-
2606 ber. For example, for this pattern
2607
2608 (a+)b(?<xxx>\d+)...
2609
2610 the number of the subpattern called "xxx" is 2. If the name is known to
2611 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2612 name by calling pcre_get_stringnumber(). The first argument is the com-
2613 piled pattern, and the second is the name. The yield of the function is
2614 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2615 subpattern of that name.
2616
2617 Given the number, you can extract the substring directly, or use one of
2618 the functions described in the previous section. For convenience, there
2619 are also two functions that do the whole job.
2620
2621 Most of the arguments of pcre_copy_named_substring() and
2622 pcre_get_named_substring() are the same as those for the similarly
2623 named functions that extract by number. As these are described in the
2624 previous section, they are not re-described here. There are just two
2625 differences:
2626
2627 First, instead of a substring number, a substring name is given. Sec-
2628 ond, there is an extra argument, given at the start, which is a pointer
2629 to the compiled pattern. This is needed in order to gain access to the
2630 name-to-number translation table.
2631
2632 These functions call pcre_get_stringnumber(), and if it succeeds, they
2633 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2634 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2635 behaviour may not be what you want (see the next section).
2636
2637 Warning: If the pattern uses the (?| feature to set up multiple subpat-
2638 terns with the same number, as described in the section on duplicate
2639 subpattern numbers in the pcrepattern page, you cannot use names to
2640 distinguish the different subpatterns, because names are not included
2641 in the compiled code. The matching process uses only numbers. For this
2642 reason, the use of different names for subpatterns of the same number
2643 causes an error at compile time.
2644
2645
2646 DUPLICATE SUBPATTERN NAMES
2647
2648 int pcre_get_stringtable_entries(const pcre *code,
2649 const char *name, char **first, char **last);
2650
2651 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2652 subpatterns are not required to be unique. (Duplicate names are always
2653 allowed for subpatterns with the same number, created by using the (?|
2654 feature. Indeed, if such subpatterns are named, they are required to
2655 use the same names.)
2656
2657 Normally, patterns with duplicate names are such that in any one match,
2658 only one of the named subpatterns participates. An example is shown in
2659 the pcrepattern documentation.
2660
2661 When duplicates are present, pcre_copy_named_substring() and
2662 pcre_get_named_substring() return the first substring corresponding to
2663 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2664 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2665 function returns one of the numbers that are associated with the name,
2666 but it is not defined which it is.
2667
2668 If you want to get full details of all captured substrings for a given
2669 name, you must use the pcre_get_stringtable_entries() function. The
2670 first argument is the compiled pattern, and the second is the name. The
2671 third and fourth are pointers to variables which are updated by the
2672 function. After it has run, they point to the first and last entries in
2673 the name-to-number table for the given name. The function itself
2674 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2675 there are none. The format of the table is described above in the sec-
2676 tion entitled Information about a pattern. Given all the relevant
2677 entries for the name, you can extract each of their numbers, and hence
2678 the captured data, if any.
2679
2680
2681 FINDING ALL POSSIBLE MATCHES
2682
2683 The traditional matching function uses a similar algorithm to Perl,
2684 which stops when it finds the first match, starting at a given point in
2685 the subject. If you want to find all possible matches, or the longest
2686 possible match, consider using the alternative matching function (see
2687 below) instead. If you cannot use the alternative function, but still
2688 need to find all possible matches, you can kludge it up by making use
2689 of the callout facility, which is described in the pcrecallout documen-
2690 tation.
2691
2692 What you have to do is to insert a callout right at the end of the pat-
2693 tern. When your callout function is called, extract and save the cur-
2694 rent matched substring. Then return 1, which forces pcre_exec() to
2695 backtrack and try other alternatives. Ultimately, when it runs out of
2696 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2697
2698
2699 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2700
2701 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2702 const char *subject, int length, int startoffset,
2703 int options, int *ovector, int ovecsize,
2704 int *workspace, int wscount);
2705
2706 The function pcre_dfa_exec() is called to match a subject string
2707 against a compiled pattern, using a matching algorithm that scans the
2708 subject string just once, and does not backtrack. This has different
2709 characteristics to the normal algorithm, and is not compatible with
2710 Perl. Some of the features of PCRE patterns are not supported. Never-
2711 theless, there are times when this kind of matching can be useful. For
2712 a discussion of the two matching algorithms, and a list of features
2713 that pcre_dfa_exec() does not support, see the pcrematching documenta-
2714 tion.
2715
2716 The arguments for the pcre_dfa_exec() function are the same as for
2717 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2718 ent way, and this is described below. The other common arguments are
2719 used in the same way as for pcre_exec(), so their description is not
2720 repeated here.
2721
2722 The two additional arguments provide workspace for the function. The
2723 workspace vector should contain at least 20 elements. It is used for
2724 keeping track of multiple paths through the pattern tree. More
2725 workspace will be needed for patterns and subjects where there are a
2726 lot of potential matches.
2727
2728 Here is an example of a simple call to pcre_dfa_exec():
2729
2730 int rc;
2731 int ovector[10];
2732 int wspace[20];
2733 rc = pcre_dfa_exec(
2734 re, /* result of pcre_compile() */
2735 NULL, /* we didn't study the pattern */
2736 "some string", /* the subject string */
2737 11, /* the length of the subject string */
2738 0, /* start at offset 0 in the subject */
2739 0, /* default options */
2740 ovector, /* vector of integers for substring information */
2741 10, /* number of elements (NOT size in bytes) */
2742 wspace, /* working space vector */
2743 20); /* number of elements (NOT size in bytes) */
2744
2745 Option bits for pcre_dfa_exec()
2746
2747 The unused bits of the options argument for pcre_dfa_exec() must be
2748 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2749 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
2750 PCRE_NOTEMPTY_ATSTART, PCRE_NO_UTF8_CHECK, PCRE_BSR_ANYCRLF,
2751 PCRE_BSR_UNICODE, PCRE_NO_START_OPTIMIZE, PCRE_PARTIAL_HARD, PCRE_PAR-
2752 TIAL_SOFT, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2753 four of these are exactly the same as for pcre_exec(), so their
2754 description is not repeated here.
2755
2756 PCRE_PARTIAL_HARD
2757 PCRE_PARTIAL_SOFT
2758
2759 These have the same general effect as they do for pcre_exec(), but the
2760 details are slightly different. When PCRE_PARTIAL_HARD is set for
2761 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
2762 ject is reached and there is still at least one matching possibility
2763 that requires additional characters. This happens even if some complete
2764 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
2765 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
2766 of the subject is reached, there have been no complete matches, but
2767 there is still at least one matching possibility. The portion of the
2768 string that was inspected when the longest partial match was found is
2769 set as the first matching string in both cases. There is a more
2770 detailed discussion of partial and multi-segment matching, with exam-
2771 ples, in the pcrepartial documentation.
2772
2773 PCRE_DFA_SHORTEST
2774
2775 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2776 stop as soon as it has found one match. Because of the way the alterna-
2777 tive algorithm works, this is necessarily the shortest possible match
2778 at the first possible matching point in the subject string.
2779
2780 PCRE_DFA_RESTART
2781
2782 When pcre_dfa_exec() returns a partial match, it is possible to call it
2783 again, with additional subject characters, and have it continue with
2784 the same match. The PCRE_DFA_RESTART option requests this action; when
2785 it is set, the workspace and wscount options must reference the same
2786 vector as before because data about the match so far is left in them
2787 after a partial match. There is more discussion of this facility in the
2788 pcrepartial documentation.
2789
2790 Successful returns from pcre_dfa_exec()
2791
2792 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2793 string in the subject. Note, however, that all the matches from one run
2794 of the function start at the same point in the subject. The shorter
2795 matches are all initial substrings of the longer matches. For example,
2796 if the pattern
2797
2798 <.*>
2799
2800 is matched against the string
2801
2802 This is <something> <something else> <something further> no more
2803
2804 the three matched strings are
2805
2806 <something>
2807 <something> <something else>
2808 <something> <something else> <something further>
2809
2810 On success, the yield of the function is a number greater than zero,
2811 which is the number of matched substrings. The substrings themselves
2812 are returned in ovector. Each string uses two elements; the first is
2813 the offset to the start, and the second is the offset to the end. In
2814 fact, all the strings have the same start offset. (Space could have
2815 been saved by giving this only once, but it was decided to retain some
2816 compatibility with the way pcre_exec() returns data, even though the
2817 meaning of the strings is different.)
2818
2819 The strings are returned in reverse order of length; that is, the long-
2820 est matching string is given first. If there were too many matches to
2821 fit into ovector, the yield of the function is zero, and the vector is
2822 filled with the longest matches.
2823
2824 Error returns from pcre_dfa_exec()
2825
2826 The pcre_dfa_exec() function returns a negative number when it fails.
2827 Many of the errors are the same as for pcre_exec(), and these are
2828 described above. There are in addition the following errors that are
2829 specific to pcre_dfa_exec():
2830
2831 PCRE_ERROR_DFA_UITEM (-16)
2832
2833 This return is given if pcre_dfa_exec() encounters an item in the pat-
2834 tern that it does not support, for instance, the use of \C or a back
2835 reference.
2836
2837 PCRE_ERROR_DFA_UCOND (-17)
2838
2839 This return is given if pcre_dfa_exec() encounters a condition item
2840 that uses a back reference for the condition, or a test for recursion
2841 in a specific group. These are not supported.
2842
2843 PCRE_ERROR_DFA_UMLIMIT (-18)
2844
2845 This return is given if pcre_dfa_exec() is called with an extra block
2846 that contains a setting of the match_limit field. This is not supported
2847 (it is meaningless).
2848
2849 PCRE_ERROR_DFA_WSSIZE (-19)
2850
2851 This return is given if pcre_dfa_exec() runs out of space in the
2852 workspace vector.
2853
2854 PCRE_ERROR_DFA_RECURSE (-20)
2855
2856 When a recursive subpattern is processed, the matching function calls
2857 itself recursively, using private vectors for ovector and workspace.
2858 This error is given if the output vector is not large enough. This
2859 should be extremely rare, as a vector of size 1000 is used.
2860
2861
2862 SEE ALSO
2863
2864 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2865 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2866
2867
2868 AUTHOR
2869
2870 Philip Hazel
2871 University Computing Service
2872 Cambridge CB2 3QH, England.
2873
2874
2875 REVISION
2876
2877 Last updated: 21 November 2010
2878 Copyright (c) 1997-2010 University of Cambridge.
2879 ------------------------------------------------------------------------------
2880
2881
2882 PCRECALLOUT(3) PCRECALLOUT(3)
2883
2884
2885 NAME
2886 PCRE - Perl-compatible regular expressions
2887
2888
2889 PCRE CALLOUTS
2890
2891 int (*pcre_callout)(pcre_callout_block *);
2892
2893 PCRE provides a feature called "callout", which is a means of temporar-
2894 ily passing control to the caller of PCRE in the middle of pattern
2895 matching. The caller of PCRE provides an external function by putting
2896 its entry point in the global variable pcre_callout. By default, this
2897 variable contains NULL, which disables all calling out.
2898
2899 Within a regular expression, (?C) indicates the points at which the
2900 external function is to be called. Different callout points can be
2901 identified by putting a number less than 256 after the letter C. The
2902 default value is zero. For example, this pattern has two callout
2903 points:
2904
2905 (?C1)abc(?C2)def
2906
2907 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() or
2908 pcre_compile2() is called, PCRE automatically inserts callouts, all
2909 with number 255, before each item in the pattern. For example, if
2910 PCRE_AUTO_CALLOUT is used with the pattern
2911
2912 A(\d{2}|--)
2913
2914 it is processed as if it were
2915
2916 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2917
2918 Notice that there is a callout before and after each parenthesis and
2919 alternation bar. Automatic callouts can be used for tracking the
2920 progress of pattern matching. The pcretest command has an option that
2921 sets automatic callouts; when it is used, the output indicates how the
2922 pattern is matched. This is useful information when you are trying to
2923 optimize the performance of a particular pattern.
2924
2925
2926 MISSING CALLOUTS
2927
2928 You should be aware that, because of optimizations in the way PCRE
2929 matches patterns by default, callouts sometimes do not happen. For
2930 example, if the pattern is
2931
2932 ab(?C4)cd
2933
2934 PCRE knows that any matching string must contain the letter "d". If the
2935 subject string is "abyz", the lack of "d" means that matching doesn't
2936 ever start, and the callout is never reached. However, with "abyd",
2937 though the result is still no match, the callout is obeyed.
2938
2939 If the pattern is studied, PCRE knows the minimum length of a matching
2940 string, and will immediately give a "no match" return without actually
2941 running a match if the subject is not long enough, or, for unanchored
2942 patterns, if it has been scanned far enough.
2943
2944 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2945 MIZE option to pcre_compile(), pcre_exec(), or pcre_dfa_exec(), or by
2946 starting the pattern with (*NO_START_OPT). This slows down the matching
2947 process, but does ensure that callouts such as the example above are
2948 obeyed.
2949
2950
2951 THE CALLOUT INTERFACE
2952
2953 During matching, when PCRE reaches a callout point, the external func-
2954 tion defined by pcre_callout is called (if it is set). This applies to
2955 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2956 only argument to the callout function is a pointer to a pcre_callout
2957 block. This structure contains the following fields:
2958
2959 int version;
2960 int callout_number;
2961 int *offset_vector;
2962 const char *subject;
2963 int subject_length;
2964 int start_match;
2965 int current_position;
2966 int capture_top;
2967 int capture_last;
2968 void *callout_data;
2969 int pattern_position;
2970 int next_item_length;
2971
2972 The version field is an integer containing the version number of the
2973 block format. The initial version was 0; the current version is 1. The
2974 version number will change again in future if additional fields are
2975 added, but the intention is never to remove any of the existing fields.
2976
2977 The callout_number field contains the number of the callout, as com-
2978 piled into the pattern (that is, the number after ?C for manual call-
2979 outs, and 255 for automatically generated callouts).
2980
2981 The offset_vector field is a pointer to the vector of offsets that was
2982 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2983 pcre_exec() is used, the contents can be inspected in order to extract
2984 substrings that have been matched so far, in the same way as for
2985 extracting substrings after a match has completed. For pcre_dfa_exec()
2986 this field is not useful.
2987
2988 The subject and subject_length fields contain copies of the values that
2989 were passed to pcre_exec().
2990
2991 The start_match field normally contains the offset within the subject
2992 at which the current match attempt started. However, if the escape
2993 sequence \K has been encountered, this value is changed to reflect the
2994 modified starting point. If the pattern is not anchored, the callout
2995 function may be called several times from the same point in the pattern
2996 for different starting points in the subject.
2997
2998 The current_position field contains the offset within the subject of
2999 the current match pointer.
3000
3001 When the pcre_exec() function is used, the capture_top field contains
3002 one more than the number of the highest numbered captured substring so
3003 far. If no substrings have been captured, the value of capture_top is
3004 one. This is always the case when pcre_dfa_exec() is used, because it
3005 does not support captured substrings.
3006
3007 The capture_last field contains the number of the most recently cap-
3008 tured substring. If no substrings have been captured, its value is -1.
3009 This is always the case when pcre_dfa_exec() is used.
3010
3011 The callout_data field contains a value that is passed to pcre_exec()
3012 or pcre_dfa_exec() specifically so that it can be passed back in call-
3013 outs. It is passed in the pcre_callout field of the pcre_extra data
3014 structure. If no such data was passed, the value of callout_data in a
3015 pcre_callout block is NULL. There is a description of the pcre_extra
3016 structure in the pcreapi documentation.
3017
3018 The pattern_position field is present from version 1 of the pcre_call-
3019 out structure. It contains the offset to the next item to be matched in
3020 the pattern string.
3021
3022 The next_item_length field is present from version 1 of the pcre_call-
3023 out structure. It contains the length of the next item to be matched in
3024 the pattern string. When the callout immediately precedes an alterna-
3025 tion bar, a closing parenthesis, or the end of the pattern, the length
3026 is zero. When the callout precedes an opening parenthesis, the length
3027 is that of the entire subpattern.
3028
3029 The pattern_position and next_item_length fields are intended to help
3030 in distinguishing between different automatic callouts, which all have
3031 the same callout number. However, they are set for all callouts.
3032
3033
3034 RETURN VALUES
3035
3036 The external callout function returns an integer to PCRE. If the value
3037 is zero, matching proceeds as normal. If the value is greater than
3038 zero, matching fails at the current point, but the testing of other
3039 matching possibilities goes ahead, just as if a lookahead assertion had
3040 failed. If the value is less than zero, the match is abandoned, and
3041 pcre_exec() or pcre_dfa_exec() returns the negative value.
3042
3043 Negative values should normally be chosen from the set of
3044 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
3045 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
3046 reserved for use by callout functions; it will never be used by PCRE
3047 itself.
3048
3049
3050 AUTHOR
3051
3052 Philip Hazel
3053 University Computing Service
3054 Cambridge CB2 3QH, England.
3055
3056
3057 REVISION
3058
3059 Last updated: 21 November 2010
3060 Copyright (c) 1997-2010 University of Cambridge.
3061 ------------------------------------------------------------------------------
3062
3063
3064 PCRECOMPAT(3) PCRECOMPAT(3)
3065
3066
3067 NAME
3068 PCRE - Perl-compatible regular expressions
3069
3070
3071 DIFFERENCES BETWEEN PCRE AND PERL
3072
3073 This document describes the differences in the ways that PCRE and Perl
3074 handle regular expressions. The differences described here are with
3075 respect to Perl versions 5.10 and above.
3076
3077 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
3078 of what it does have are given in the section on UTF-8 support in the
3079 main pcre page.
3080
3081 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
3082 permits them, but they do not mean what you might think. For example,
3083 (?!a){3} does not assert that the next three characters are not "a". It
3084 just asserts that the next character is not "a" three times.
3085
3086 3. Capturing subpatterns that occur inside negative lookahead asser-
3087 tions are counted, but their entries in the offsets vector are never
3088 set. Perl sets its numerical variables from any such patterns that are
3089 matched before the assertion fails to match something (thereby succeed-
3090 ing), but only if the negative lookahead assertion contains just one
3091 branch.
3092
3093 4. Though binary zero characters are supported in the subject string,
3094 they are not allowed in a pattern string because it is passed as a nor-
3095 mal C string, terminated by zero. The escape sequence \0 can be used in
3096 the pattern to represent a binary zero.
3097
3098 5. The following Perl escape sequences are not supported: \l, \u, \L,
3099 \U, and \N. In fact these are implemented by Perl's general string-han-
3100 dling and are not part of its pattern matching engine. If any of these
3101 are encountered by PCRE, an error is generated.
3102
3103 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
3104 is built with Unicode character property support. The properties that
3105 can be tested with \p and \P are limited to the general category prop-
3106 erties such as Lu and Nd, script names such as Greek or Han, and the
3107 derived properties Any and L&. PCRE does support the Cs (surrogate)
3108 property, which Perl does not; the Perl documentation says "Because
3109 Perl hides the need for the user to understand the internal representa-
3110 tion of Unicode characters, there is no need to implement the somewhat
3111 messy concept of surrogates."
3112
3113 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
3114 ters in between are treated as literals. This is slightly different
3115 from Perl in that $ and @ are also handled as literals inside the
3116 quotes. In Perl, they cause variable interpolation (but of course PCRE
3117 does not have variables). Note the following examples:
3118
3119 Pattern PCRE matches Perl matches
3120
3121 \Qabc$xyz\E abc$xyz abc followed by the
3122 contents of $xyz
3123 \Qabc\$xyz\E abc\$xyz abc\$xyz
3124 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3125
3126 The \Q...\E sequence is recognized both inside and outside character
3127 classes.
3128
3129 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3130 constructions. However, there is support for recursive patterns. This
3131 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3132 "callout" feature allows an external function to be called during pat-
3133 tern matching. See the pcrecallout documentation for details.
3134
3135 9. Subpatterns that are called recursively or as "subroutines" are
3136 always treated as atomic groups in PCRE. This is like Python, but
3137 unlike Perl. There is a discussion of an example that explains this in
3138 more detail in the section on recursion differences from Perl in the
3139 pcrepattern page.
3140
3141 10. There are some differences that are concerned with the settings of
3142 captured strings when part of a pattern is repeated. For example,
3143 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3144 unset, but in PCRE it is set to "b".
3145
3146 11. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3147 pattern names is not as general as Perl's. This is a consequence of the
3148 fact the PCRE works internally just with numbers, using an external ta-
3149 ble to translate between numbers and names. In particular, a pattern
3150 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3151 the same number but different names, is not supported, and causes an
3152 error at compile time. If it were allowed, it would not be possible to
3153 distinguish which parentheses matched, because both names map to cap-
3154 turing subpattern number 1. To avoid this confusing situation, an error
3155 is given at compile time.
3156
3157 12. Perl recognizes comments in some places that PCRE doesn't, for
3158 example, between the ( and ? at the start of a subpattern.
3159
3160 13. PCRE provides some extensions to the Perl regular expression facil-
3161 ities. Perl 5.10 includes new features that are not in earlier ver-
3162 sions of Perl, some of which (such as named parentheses) have been in
3163 PCRE for some time. This list is with respect to Perl 5.10:
3164
3165 (a) Although lookbehind assertions in PCRE must match fixed length
3166 strings, each alternative branch of a lookbehind assertion can match a
3167 different length of string. Perl requires them all to have the same
3168 length.
3169
3170 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3171 meta-character matches only at the very end of the string.
3172
3173 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3174 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3175 ignored. (Perl can be made to issue a warning.)
3176
3177 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3178 fiers is inverted, that is, by default they are not greedy, but if fol-
3179 lowed by a question mark they are.
3180
3181 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3182 tried only at the first matching position in the subject string.
3183
3184 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
3185 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3186 lents.
3187
3188 (g) The \R escape sequence can be restricted to match only CR, LF, or
3189 CRLF by the PCRE_BSR_ANYCRLF option.
3190
3191 (h) The callout facility is PCRE-specific.
3192
3193 (i) The partial matching facility is PCRE-specific.
3194
3195 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3196 even on different hosts that have the other endianness.
3197
3198 (k) The alternative matching function (pcre_dfa_exec()) matches in a
3199 different way and is not Perl-compatible.
3200
3201 (l) PCRE recognizes some special sequences such as (*CR) at the start
3202 of a pattern that set overall options that cannot be changed within the
3203 pattern.
3204
3205
3206 AUTHOR
3207
3208 Philip Hazel
3209 University Computing Service
3210 Cambridge CB2 3QH, England.
3211
3212
3213 REVISION
3214
3215 Last updated: 31 October 2010
3216 Copyright (c) 1997-2010 University of Cambridge.
3217 ------------------------------------------------------------------------------
3218
3219
3220 PCREPATTERN(3) PCREPATTERN(3)
3221
3222
3223 NAME
3224 PCRE - Perl-compatible regular expressions
3225
3226
3227 PCRE REGULAR EXPRESSION DETAILS
3228
3229 The syntax and semantics of the regular expressions that are supported
3230 by PCRE are described in detail below. There is a quick-reference syn-
3231 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3232 semantics as closely as it can. PCRE also supports some alternative
3233 regular expression syntax (which does not conflict with the Perl syn-
3234 tax) in order to provide some compatibility with regular expressions in
3235 Python, .NET, and Oniguruma.
3236
3237 Perl's regular expressions are described in its own documentation, and
3238 regular expressions in general are covered in a number of books, some
3239 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3240 Expressions", published by O'Reilly, covers regular expressions in
3241 great detail. This description of PCRE's regular expressions is
3242 intended as reference material.
3243
3244 The original operation of PCRE was on strings of one-byte characters.
3245 However, there is now also support for UTF-8 character strings. To use
3246 this, PCRE must be built to include UTF-8 support, and you must call
3247 pcre_compile() or pcre_compile2() with the PCRE_UTF8 option. There is
3248 also a special sequence that can be given at the start of a pattern:
3249
3250 (*UTF8)
3251
3252 Starting a pattern with this sequence is equivalent to setting the
3253 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
3254 UTF-8 mode affects pattern matching is mentioned in several places
3255 below. There is also a summary of UTF-8 features in the section on
3256 UTF-8 support in the main pcre page.
3257
3258 Another special sequence that may appear at the start of a pattern or
3259 in combination with (*UTF8) is:
3260
3261 (*UCP)
3262
3263 This has the same effect as setting the PCRE_UCP option: it causes
3264 sequences such as \d and \w to use Unicode properties to determine
3265 character types, instead of recognizing only characters with codes less
3266 than 128 via a lookup table.
3267
3268 If a pattern starts with (*NO_START_OPT), it has the same effect as
3269 setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
3270 time. There are also some more of these special sequences that are con-
3271 cerned with the handling of newlines; they are described below.
3272
3273 The remainder of this document discusses the patterns that are sup-
3274 ported by PCRE when its main matching function, pcre_exec(), is used.
3275 From release 6.0, PCRE offers a second matching function,
3276 pcre_dfa_exec(), which matches using a different algorithm that is not
3277 Perl-compatible. Some of the features discussed below are not available
3278 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3279 alternative function, and how it differs from the normal function, are
3280 discussed in the pcrematching page.
3281
3282
3283 NEWLINE CONVENTIONS
3284
3285 PCRE supports five different conventions for indicating line breaks in
3286 strings: a single CR (carriage return) character, a single LF (line-
3287 feed) character, the two-character sequence CRLF, any of the three pre-
3288 ceding, or any Unicode newline sequence. The pcreapi page has further
3289 discussion about newlines, and shows how to set the newline convention
3290 in the options arguments for the compiling and matching functions.
3291
3292 It is also possible to specify a newline convention by starting a pat-
3293 tern string with one of the following five sequences:
3294
3295 (*CR) carriage return
3296 (*LF) linefeed
3297 (*CRLF) carriage return, followed by linefeed
3298 (*ANYCRLF) any of the three above
3299 (*ANY) all Unicode newline sequences
3300
3301 These override the default and the options given to pcre_compile() or
3302 pcre_compile2(). For example, on a Unix system where LF is the default
3303 newline sequence, the pattern
3304
3305 (*CR)a.b
3306
3307 changes the convention to CR. That pattern matches "a\nb" because LF is
3308 no longer a newline. Note that these special settings, which are not
3309 Perl-compatible, are recognized only at the very start of a pattern,
3310 and that they must be in upper case. If more than one of them is
3311 present, the last one is used.
3312
3313 The newline convention affects the interpretation of the dot metachar-
3314 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3315 ever, it does not affect what the \R escape sequence matches. By
3316 default, this is any Unicode newline sequence, for Perl compatibility.
3317 However, this can be changed; see the description of \R in the section
3318 entitled "Newline sequences" below. A change of \R setting can be com-
3319 bined with a change of newline convention.
3320
3321
3322 CHARACTERS AND METACHARACTERS
3323
3324 A regular expression is a pattern that is matched against a subject
3325 string from left to right. Most characters stand for themselves in a
3326 pattern, and match the corresponding characters in the subject. As a
3327 trivial example, the pattern
3328
3329 The quick brown fox
3330
3331 matches a portion of a subject string that is identical to itself. When
3332 caseless matching is specified (the PCRE_CASELESS option), letters are
3333 matched independently of case. In UTF-8 mode, PCRE always understands
3334 the concept of case for characters whose values are less than 128, so
3335 caseless matching is always possible. For characters with higher val-
3336 ues, the concept of case is supported if PCRE is compiled with Unicode
3337 property support, but not otherwise. If you want to use caseless
3338 matching for characters 128 and above, you must ensure that PCRE is
3339 compiled with Unicode property support as well as with UTF-8 support.
3340
3341 The power of regular expressions comes from the ability to include
3342 alternatives and repetitions in the pattern. These are encoded in the
3343 pattern by the use of metacharacters, which do not stand for themselves
3344 but instead are interpreted in some special way.
3345
3346 There are two different sets of metacharacters: those that are recog-
3347 nized anywhere in the pattern except within square brackets, and those
3348 that are recognized within square brackets. Outside square brackets,
3349 the metacharacters are as follows:
3350
3351 \ general escape character with several uses
3352 ^ assert start of string (or line, in multiline mode)
3353 $ assert end of string (or line, in multiline mode)
3354 . match any character except newline (by default)
3355 [ start character class definition
3356 | start of alternative branch
3357 ( start subpattern
3358 ) end subpattern
3359 ? extends the meaning of (
3360 also 0 or 1 quantifier
3361 also quantifier minimizer
3362 * 0 or more quantifier
3363 + 1 or more quantifier
3364 also "possessive quantifier"
3365 { start min/max quantifier
3366
3367 Part of a pattern that is in square brackets is called a "character
3368 class". In a character class the only metacharacters are:
3369
3370 \ general escape character
3371 ^ negate the class, but only if the first character
3372 - indicates character range
3373 [ POSIX character class (only if followed by POSIX
3374 syntax)
3375 ] terminates the character class
3376
3377 The following sections describe the use of each of the metacharacters.
3378
3379
3380 BACKSLASH
3381
3382 The backslash character has several uses. Firstly, if it is followed by
3383 a character that is not a number or a letter, it takes away any special
3384 meaning that character may have. This use of backslash as an escape
3385 character applies both inside and outside character classes.
3386
3387 For example, if you want to match a * character, you write \* in the
3388 pattern. This escaping action applies whether or not the following
3389 character would otherwise be interpreted as a metacharacter, so it is
3390 always safe to precede a non-alphanumeric with backslash to specify
3391 that it stands for itself. In particular, if you want to match a back-
3392 slash, you write \\.
3393
3394 In UTF-8 mode, only ASCII numbers and letters have any special meaning
3395 after a backslash. All other characters (in particular, those whose
3396 codepoints are greater than 127) are treated as literals.
3397
3398 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3399 the pattern (other than in a character class) and characters between a
3400 # outside a character class and the next newline are ignored. An escap-
3401 ing backslash can be used to include a whitespace or # character as
3402 part of the pattern.
3403
3404 If you want to remove the special meaning from a sequence of charac-
3405 ters, you can do so by putting them between \Q and \E. This is differ-
3406 ent from Perl in that $ and @ are handled as literals in \Q...\E
3407 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3408 tion. Note the following examples:
3409
3410 Pattern PCRE matches Perl matches
3411
3412 \Qabc$xyz\E abc$xyz abc followed by the
3413 contents of $xyz
3414 \Qabc\$xyz\E abc\$xyz abc\$xyz
3415 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3416
3417 The \Q...\E sequence is recognized both inside and outside character
3418 classes. An isolated \E that is not preceded by \Q is ignored.
3419
3420 Non-printing characters
3421
3422 A second use of backslash provides a way of encoding non-printing char-
3423 acters in patterns in a visible manner. There is no restriction on the
3424 appearance of non-printing characters, apart from the binary zero that
3425 terminates a pattern, but when a pattern is being prepared by text
3426 editing, it is often easier to use one of the following escape
3427 sequences than the binary character it represents:
3428
3429 \a alarm, that is, the BEL character (hex 07)
3430 \cx "control-x", where x is any ASCII character
3431 \e escape (hex 1B)
3432 \f formfeed (hex 0C)
3433 \n linefeed (hex 0A)
3434 \r carriage return (hex 0D)
3435 \t tab (hex 09)
3436 \ddd character with octal code ddd, or back reference
3437 \xhh character with hex code hh
3438 \x{hhh..} character with hex code hhh..
3439
3440 The precise effect of \cx is as follows: if x is a lower case letter,
3441 it is converted to upper case. Then bit 6 of the character (hex 40) is
3442 inverted. Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({
3443 is 7B), while \c; becomes hex 7B (; is 3B). If the byte following \c
3444 has a value greater than 127, a compile-time error occurs. This locks
3445 out non-ASCII characters in both byte mode and UTF-8 mode. (When PCRE
3446 is compiled in EBCDIC mode, all byte values are valid. A lower case
3447 letter is converted to upper case, and then the 0xc0 bits are flipped.)
3448
3449 After \x, from zero to two hexadecimal digits are read (letters can be
3450 in upper or lower case). Any number of hexadecimal digits may appear
3451 between \x{ and }, but the value of the character code must be less
3452 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3453 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3454 than the largest Unicode code point, which is 10FFFF.
3455
3456 If characters other than hexadecimal digits appear between \x{ and },
3457 or if there is no terminating }, this form of escape is not recognized.
3458 Instead, the initial \x will be interpreted as a basic hexadecimal
3459 escape, with no following digits, giving a character whose value is
3460 zero.
3461
3462 Characters whose value is less than 256 can be defined by either of the
3463 two syntaxes for \x. There is no difference in the way they are han-
3464 dled. For example, \xdc is exactly the same as \x{dc}.
3465
3466 After \0 up to two further octal digits are read. If there are fewer
3467 than two digits, just those that are present are used. Thus the
3468 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3469 (code value 7). Make sure you supply two digits after the initial zero
3470 if the pattern character that follows is itself an octal digit.
3471
3472 The handling of a backslash followed by a digit other than 0 is compli-
3473 cated. Outside a character class, PCRE reads it and any following dig-
3474 its as a decimal number. If the number is less than 10, or if there
3475 have been at least that many previous capturing left parentheses in the
3476 expression, the entire sequence is taken as a back reference. A
3477 description of how this works is given later, following the discussion
3478 of parenthesized subpatterns.
3479
3480 Inside a character class, or if the decimal number is greater than 9
3481 and there have not been that many capturing subpatterns, PCRE re-reads
3482 up to three octal digits following the backslash, and uses them to gen-
3483 erate a data character. Any subsequent digits stand for themselves. In
3484 non-UTF-8 mode, the value of a character specified in octal must be
3485 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3486 example:
3487
3488 \040 is another way of writing a space
3489 \40 is the same, provided there are fewer than 40
3490 previous capturing subpatterns
3491 \7 is always a back reference
3492 \11 might be a back reference, or another way of
3493 writing a tab
3494 \011 is always a tab
3495 \0113 is a tab followed by the character "3"
3496 \113 might be a back reference, otherwise the
3497 character with octal code 113
3498 \377 might be a back reference, otherwise
3499 the byte consisting entirely of 1 bits
3500 \81 is either a back reference, or a binary zero
3501 followed by the two characters "8" and "1"
3502
3503 Note that octal values of 100 or greater must not be introduced by a
3504 leading zero, because no more than three octal digits are ever read.
3505
3506 All the sequences that define a single character value can be used both
3507 inside and outside character classes. In addition, inside a character
3508 class, the sequence \b is interpreted as the backspace character (hex
3509 08). The sequences \B, \N, \R, and \X are not special inside a charac-
3510 ter class. Like any other unrecognized escape sequences, they are
3511 treated as the literal characters "B", "N", "R", and "X" by default,
3512 but cause an error if the PCRE_EXTRA option is set. Outside a character
3513 class, these sequences have different meanings.
3514
3515 Absolute and relative back references
3516
3517 The sequence \g followed by an unsigned or a negative number, option-
3518 ally enclosed in braces, is an absolute or relative back reference. A
3519 named back reference can be coded as \g{name}. Back references are dis-
3520 cussed later, following the discussion of parenthesized subpatterns.
3521
3522 Absolute and relative subroutine calls
3523
3524 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3525 name or a number enclosed either in angle brackets or single quotes, is
3526 an alternative syntax for referencing a subpattern as a "subroutine".
3527 Details are discussed later. Note that \g{...} (Perl syntax) and
3528 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3529 reference; the latter is a subroutine call.
3530
3531 Generic character types
3532
3533 Another use of backslash is for specifying generic character types:
3534
3535 \d any decimal digit
3536 \D any character that is not a decimal digit
3537 \h any horizontal whitespace character
3538 \H any character that is not a horizontal whitespace character
3539 \s any whitespace character
3540 \S any character that is not a whitespace character
3541 \v any vertical whitespace character
3542 \V any character that is not a vertical whitespace character
3543 \w any "word" character
3544 \W any "non-word" character
3545
3546 There is also the single sequence \N, which matches a non-newline char-
3547 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
3548 not set.
3549
3550 Each pair of lower and upper case escape sequences partitions the com-
3551 plete set of characters into two disjoint sets. Any given character
3552 matches one, and only one, of each pair. The sequences can appear both
3553 inside and outside character classes. They each match one character of
3554 the appropriate type. If the current matching point is at the end of
3555 the subject string, all of them fail, because there is no character to
3556 match.
3557
3558 For compatibility with Perl, \s does not match the VT character (code
3559 11). This makes it different from the the POSIX "space" class. The \s
3560 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3561 "use locale;" is included in a Perl script, \s may match the VT charac-
3562 ter. In PCRE, it never does.
3563
3564 A "word" character is an underscore or any character that is a letter
3565 or digit. By default, the definition of letters and digits is con-
3566 trolled by PCRE's low-valued character tables, and may vary if locale-
3567 specific matching is taking place (see "Locale support" in the pcreapi
3568 page). For example, in a French locale such as "fr_FR" in Unix-like
3569 systems, or "french" in Windows, some character codes greater than 128
3570 are used for accented letters, and these are then matched by \w. The
3571 use of locales with Unicode is discouraged.
3572
3573 By default, in UTF-8 mode, characters with values greater than 128
3574 never match \d, \s, or \w, and always match \D, \S, and \W. These
3575 sequences retain their original meanings from before UTF-8 support was
3576 available, mainly for efficiency reasons. However, if PCRE is compiled
3577 with Unicode property support, and the PCRE_UCP option is set, the be-
3578 haviour is changed so that Unicode properties are used to determine
3579 character types, as follows:
3580
3581 \d any character that \p{Nd} matches (decimal digit)
3582 \s any character that \p{Z} matches, plus HT, LF, FF, CR
3583 \w any character that \p{L} or \p{N} matches, plus underscore
3584
3585 The upper case escapes match the inverse sets of characters. Note that
3586 \d matches only decimal digits, whereas \w matches any Unicode digit,
3587 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
3588 affects \b, and \B because they are defined in terms of \w and \W.
3589 Matching these sequences is noticeably slower when PCRE_UCP is set.
3590
3591 The sequences \h, \H, \v, and \V are features that were added to Perl
3592 at release 5.10. In contrast to the other sequences, which match only
3593 ASCII characters by default, these always match certain high-valued
3594 codepoints in UTF-8 mode, whether or not PCRE_UCP is set. The horizon-
3595 tal space characters are:
3596
3597 U+0009 Horizontal tab
3598 U+0020 Space
3599 U+00A0 Non-break space
3600 U+1680 Ogham space mark
3601 U+180E Mongolian vowel separator
3602 U+2000 En quad
3603 U+2001 Em quad
3604 U+2002 En space
3605 U+2003 Em space
3606 U+2004 Three-per-em space
3607 U+2005 Four-per-em space
3608 U+2006 Six-per-em space
3609 U+2007 Figure space
3610 U+2008 Punctuation space
3611 U+2009 Thin space
3612 U+200A Hair space
3613 U+202F Narrow no-break space
3614 U+205F Medium mathematical space
3615 U+3000 Ideographic space
3616
3617 The vertical space characters are:
3618
3619 U+000A Linefeed
3620 U+000B Vertical tab
3621 U+000C Formfeed
3622 U+000D Carriage return
3623 U+0085 Next line
3624 U+2028 Line separator
3625 U+2029 Paragraph separator
3626
3627 Newline sequences
3628
3629 Outside a character class, by default, the escape sequence \R matches
3630 any Unicode newline sequence. In non-UTF-8 mode \R is equivalent to the
3631 following:
3632
3633 (?>\r\n|\n|\x0b|\f|\r|\x85)
3634
3635 This is an example of an "atomic group", details of which are given
3636 below. This particular group matches either the two-character sequence
3637 CR followed by LF, or one of the single characters LF (linefeed,
3638 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3639 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3640 is treated as a single unit that cannot be split.
3641
3642 In UTF-8 mode, two additional characters whose codepoints are greater
3643 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3644 rator, U+2029). Unicode character property support is not needed for
3645 these characters to be recognized.
3646
3647 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3648 the complete set of Unicode line endings) by setting the option
3649 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3650 (BSR is an abbrevation for "backslash R".) This can be made the default
3651 when PCRE is built; if this is the case, the other behaviour can be
3652 requested via the PCRE_BSR_UNICODE option. It is also possible to
3653 specify these settings by starting a pattern string with one of the
3654 following sequences:
3655
3656 (*BSR_ANYCRLF) CR, LF, or CRLF only
3657 (*BSR_UNICODE) any Unicode newline sequence
3658
3659 These override the default and the options given to pcre_compile() or
3660 pcre_compile2(), but they can be overridden by options given to
3661 pcre_exec() or pcre_dfa_exec(). Note that these special settings, which
3662 are not Perl-compatible, are recognized only at the very start of a
3663 pattern, and that they must be in upper case. If more than one of them
3664 is present, the last one is used. They can be combined with a change of
3665 newline convention; for example, a pattern can start with:
3666
3667 (*ANY)(*BSR_ANYCRLF)
3668
3669 They can also be combined with the (*UTF8) or (*UCP) special sequences.
3670 Inside a character class, \R is treated as an unrecognized escape
3671 sequence, and so matches the letter "R" by default, but causes an error
3672 if PCRE_EXTRA is set.
3673
3674 Unicode character properties
3675
3676 When PCRE is built with Unicode character property support, three addi-
3677 tional escape sequences that match characters with specific properties
3678 are available. When not in UTF-8 mode, these sequences are of course
3679 limited to testing characters whose codepoints are less than 256, but
3680 they do work in this mode. The extra escape sequences are:
3681
3682 \p{xx} a character with the xx property
3683 \P{xx} a character without the xx property
3684 \X an extended Unicode sequence
3685
3686 The property names represented by xx above are limited to the Unicode
3687 script names, the general category properties, "Any", which matches any
3688 character (including newline), and some special PCRE properties
3689 (described in the next section). Other Perl properties such as "InMu-
3690 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
3691 does not match any characters, so always causes a match failure.
3692
3693 Sets of Unicode characters are defined as belonging to certain scripts.
3694 A character from one of these sets can be matched using a script name.
3695 For example:
3696
3697 \p{Greek}
3698 \P{Han}
3699
3700 Those that are not part of an identified script are lumped together as
3701 "Common". The current list of scripts is:
3702
3703 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
3704 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
3705 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
3706 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
3707 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
3708 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
3709 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
3710 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
3711 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
3712 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
3713 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
3714 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
3715 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
3716 Ugaritic, Vai, Yi.
3717
3718 Each character has exactly one Unicode general category property, spec-
3719 ified by a two-letter abbreviation. For compatibility with Perl, nega-
3720 tion can be specified by including a circumflex between the opening
3721 brace and the property name. For example, \p{^Lu} is the same as
3722 \P{Lu}.
3723
3724 If only one letter is specified with \p or \P, it includes all the gen-
3725 eral category properties that start with that letter. In this case, in
3726 the absence of negation, the curly brackets in the escape sequence are
3727 optional; these two examples have the same effect:
3728
3729 \p{L}
3730 \pL
3731
3732 The following general category property codes are supported:
3733
3734 C Other
3735 Cc Control
3736 Cf Format
3737 Cn Unassigned
3738 Co Private use
3739 Cs Surrogate
3740
3741 L Letter
3742 Ll Lower case letter
3743 Lm Modifier letter
3744 Lo Other letter
3745 Lt Title case letter
3746 Lu Upper case letter
3747
3748 M Mark
3749 Mc Spacing mark
3750 Me Enclosing mark
3751 Mn Non-spacing mark
3752
3753 N Number
3754 Nd Decimal number
3755 Nl Letter number
3756 No Other number
3757
3758 P Punctuation
3759 Pc Connector punctuation
3760 Pd Dash punctuation
3761 Pe Close punctuation
3762 Pf Final punctuation
3763 Pi Initial punctuation
3764 Po Other punctuation
3765 Ps Open punctuation
3766
3767 S Symbol
3768 Sc Currency symbol
3769 Sk Modifier symbol
3770 Sm Mathematical symbol
3771 So Other symbol
3772
3773 Z Separator
3774 Zl Line separator
3775 Zp Paragraph separator
3776 Zs Space separator
3777
3778 The special property L& is also supported: it matches a character that
3779 has the Lu, Ll, or Lt property, in other words, a letter that is not
3780 classified as a modifier or "other".
3781
3782 The Cs (Surrogate) property applies only to characters in the range
3783 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3784 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3785 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3786 the pcreapi page). Perl does not support the Cs property.
3787
3788 The long synonyms for property names that Perl supports (such as
3789 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3790 any of these properties with "Is".
3791
3792 No character that is in the Unicode table has the Cn (unassigned) prop-
3793 erty. Instead, this property is assumed for any code point that is not
3794 in the Unicode table.
3795
3796 Specifying caseless matching does not affect these escape sequences.
3797 For example, \p{Lu} always matches only upper case letters.
3798
3799 The \X escape matches any number of Unicode characters that form an
3800 extended Unicode sequence. \X is equivalent to
3801
3802 (?>\PM\pM*)
3803
3804 That is, it matches a character without the "mark" property, followed
3805 by zero or more characters with the "mark" property, and treats the
3806 sequence as an atomic group (see below). Characters with the "mark"
3807 property are typically accents that affect the preceding character.
3808 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3809 matches any one character.
3810
3811 Matching characters by Unicode property is not fast, because PCRE has
3812 to search a structure that contains data for over fifteen thousand
3813 characters. That is why the traditional escape sequences such as \d and
3814 \w do not use Unicode properties in PCRE by default, though you can
3815 make them do so by setting the PCRE_UCP option for pcre_compile() or by
3816 starting the pattern with (*UCP).
3817
3818 PCRE's additional properties
3819
3820 As well as the standard Unicode properties described in the previous
3821 section, PCRE supports four more that make it possible to convert tra-
3822 ditional escape sequences such as \w and \s and POSIX character classes
3823 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
3824 erties internally when PCRE_UCP is set. They are:
3825
3826 Xan Any alphanumeric character
3827 Xps Any POSIX space character
3828 Xsp Any Perl space character
3829 Xwd Any Perl "word" character
3830
3831 Xan matches characters that have either the L (letter) or the N (num-
3832 ber) property. Xps matches the characters tab, linefeed, vertical tab,
3833 formfeed, or carriage return, and any other character that has the Z
3834 (separator) property. Xsp is the same as Xps, except that vertical tab
3835 is excluded. Xwd matches the same characters as Xan, plus underscore.
3836
3837 Resetting the match start
3838
3839 The escape sequence \K causes any previously matched characters not to
3840 be included in the final matched sequence. For example, the pattern:
3841
3842 foo\Kbar
3843
3844 matches "foobar", but reports that it has matched "bar". This feature
3845 is similar to a lookbehind assertion (described below). However, in
3846 this case, the part of the subject before the real match does not have
3847 to be of fixed length, as lookbehind assertions do. The use of \K does
3848 not interfere with the setting of captured substrings. For example,
3849 when the pattern
3850
3851 (foo)\Kbar
3852
3853 matches "foobar", the first substring is still set to "foo".
3854
3855 Perl documents that the use of \K within assertions is "not well
3856 defined". In PCRE, \K is acted upon when it occurs inside positive
3857 assertions, but is ignored in negative assertions.
3858
3859 Simple assertions
3860
3861 The final use of backslash is for certain simple assertions. An asser-
3862 tion specifies a condition that has to be met at a particular point in
3863 a match, without consuming any characters from the subject string. The
3864 use of subpatterns for more complicated assertions is described below.
3865 The backslashed assertions are:
3866
3867 \b matches at a word boundary
3868 \B matches when not at a word boundary
3869 \A matches at the start of the subject
3870 \Z matches at the end of the subject
3871 also matches before a newline at the end of the subject
3872 \z matches only at the end of the subject
3873 \G matches at the first matching position in the subject
3874
3875 Inside a character class, \b has a different meaning; it matches the
3876 backspace character. If any other of these assertions appears in a
3877 character class, by default it matches the corresponding literal char-
3878 acter (for example, \B matches the letter B). However, if the
3879 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
3880 ated instead.
3881
3882 A word boundary is a position in the subject string where the current
3883 character and the previous character do not both match \w or \W (i.e.
3884 one matches \w and the other matches \W), or the start or end of the
3885 string if the first or last character matches \w, respectively. In
3886 UTF-8 mode, the meanings of \w and \W can be changed by setting the
3887 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
3888 PCRE nor Perl has a separate "start of word" or "end of word" metase-
3889 quence. However, whatever follows \b normally determines which it is.
3890 For example, the fragment \ba matches "a" at the start of a word.
3891
3892 The \A, \Z, and \z assertions differ from the traditional circumflex
3893 and dollar (described in the next section) in that they only ever match
3894 at the very start and end of the subject string, whatever options are
3895 set. Thus, they are independent of multiline mode. These three asser-
3896 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3897 affect only the behaviour of the circumflex and dollar metacharacters.
3898 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3899 cating that matching is to start at a point other than the beginning of
3900 the subject, \A can never match. The difference between \Z and \z is
3901 that \Z matches before a newline at the end of the string as well as at
3902 the very end, whereas \z matches only at the end.
3903
3904 The \G assertion is true only when the current matching position is at
3905 the start point of the match, as specified by the startoffset argument
3906 of pcre_exec(). It differs from \A when the value of startoffset is
3907 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3908 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3909 mentation where \G can be useful.
3910
3911 Note, however, that PCRE's interpretation of \G, as the start of the
3912 current match, is subtly different from Perl's, which defines it as the
3913 end of the previous match. In Perl, these can be different when the
3914 previously matched string was empty. Because PCRE does just one match
3915 at a time, it cannot reproduce this behaviour.
3916
3917 If all the alternatives of a pattern begin with \G, the expression is
3918 anchored to the starting match position, and the "anchored" flag is set
3919 in the compiled regular expression.
3920
3921
3922 CIRCUMFLEX AND DOLLAR
3923
3924 Outside a character class, in the default matching mode, the circumflex
3925 character is an assertion that is true only if the current matching
3926 point is at the start of the subject string. If the startoffset argu-
3927 ment of pcre_exec() is non-zero, circumflex can never match if the
3928 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3929 has an entirely different meaning (see below).
3930
3931 Circumflex need not be the first character of the pattern if a number
3932 of alternatives are involved, but it should be the first thing in each
3933 alternative in which it appears if the pattern is ever to match that
3934 branch. If all possible alternatives start with a circumflex, that is,
3935 if the pattern is constrained to match only at the start of the sub-
3936 ject, it is said to be an "anchored" pattern. (There are also other
3937 constructs that can cause a pattern to be anchored.)
3938
3939 A dollar character is an assertion that is true only if the current
3940 matching point is at the end of the subject string, or immediately
3941 before a newline at the end of the string (by default). Dollar need not
3942 be the last character of the pattern if a number of alternatives are
3943 involved, but it should be the last item in any branch in which it
3944 appears. Dollar has no special meaning in a character class.
3945
3946 The meaning of dollar can be changed so that it matches only at the
3947 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3948 compile time. This does not affect the \Z assertion.
3949
3950 The meanings of the circumflex and dollar characters are changed if the
3951 PCRE_MULTILINE option is set. When this is the case, a circumflex
3952 matches immediately after internal newlines as well as at the start of
3953 the subject string. It does not match after a newline that ends the
3954 string. A dollar matches before any newlines in the string, as well as
3955 at the very end, when PCRE_MULTILINE is set. When newline is specified
3956 as the two-character sequence CRLF, isolated CR and LF characters do
3957 not indicate newlines.
3958
3959 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3960 (where \n represents a newline) in multiline mode, but not otherwise.
3961 Consequently, patterns that are anchored in single line mode because
3962 all branches start with ^ are not anchored in multiline mode, and a
3963 match for circumflex is possible when the startoffset argument of
3964 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3965 PCRE_MULTILINE is set.
3966
3967 Note that the sequences \A, \Z, and \z can be used to match the start
3968 and end of the subject in both modes, and if all branches of a pattern
3969 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3970 set.
3971
3972
3973 FULL STOP (PERIOD, DOT) AND \N
3974
3975 Outside a character class, a dot in the pattern matches any one charac-
3976 ter in the subject string except (by default) a character that signi-
3977 fies the end of a line. In UTF-8 mode, the matched character may be
3978 more than one byte long.
3979
3980 When a line ending is defined as a single character, dot never matches
3981 that character; when the two-character sequence CRLF is used, dot does
3982 not match CR if it is immediately followed by LF, but otherwise it
3983 matches all characters (including isolated CRs and LFs). When any Uni-
3984 code line endings are being recognized, dot does not match CR or LF or
3985 any of the other line ending characters.
3986
3987 The behaviour of dot with regard to newlines can be changed. If the
3988 PCRE_DOTALL option is set, a dot matches any one character, without
3989 exception. If the two-character sequence CRLF is present in the subject
3990 string, it takes two dots to match it.
3991
3992 The handling of dot is entirely independent of the handling of circum-
3993 flex and dollar, the only relationship being that they both involve
3994 newlines. Dot has no special meaning in a character class.
3995
3996 The escape sequence \N behaves like a dot, except that it is not
3997 affected by the PCRE_DOTALL option. In other words, it matches any
3998 character except one that signifies the end of a line.
3999
4000
4001 MATCHING A SINGLE BYTE
4002
4003 Outside a character class, the escape sequence \C matches any one byte,
4004 both in and out of UTF-8 mode. Unlike a dot, it always matches any
4005 line-ending characters. The feature is provided in Perl in order to
4006 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
4007 acters into individual bytes, the rest of the string may start with a
4008 malformed UTF-8 character. For this reason, the \C escape sequence is
4009 best avoided.
4010
4011 PCRE does not allow \C to appear in lookbehind assertions (described
4012 below), because in UTF-8 mode this would make it impossible to calcu-
4013 late the length of the lookbehind.
4014
4015
4016 SQUARE BRACKETS AND CHARACTER CLASSES
4017
4018 An opening square bracket introduces a character class, terminated by a
4019 closing square bracket. A closing square bracket on its own is not spe-
4020 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
4021 a lone closing square bracket causes a compile-time error. If a closing
4022 square bracket is required as a member of the class, it should be the
4023 first data character in the class (after an initial circumflex, if
4024 present) or escaped with a backslash.
4025
4026 A character class matches a single character in the subject. In UTF-8
4027 mode, the character may be more than one byte long. A matched character
4028 must be in the set of characters defined by the class, unless the first
4029 character in the class definition is a circumflex, in which case the
4030 subject character must not be in the set defined by the class. If a
4031 circumflex is actually required as a member of the class, ensure it is
4032 not the first character, or escape it with a backslash.
4033
4034 For example, the character class [aeiou] matches any lower case vowel,
4035 while [^aeiou] matches any character that is not a lower case vowel.
4036 Note that a circumflex is just a convenient notation for specifying the
4037 characters that are in the class by enumerating those that are not. A
4038 class that starts with a circumflex is not an assertion; it still con-
4039 sumes a character from the subject string, and therefore it fails if
4040 the current pointer is at the end of the string.
4041
4042 In UTF-8 mode, characters with values greater than 255 can be included
4043 in a class as a literal string of bytes, or by using the \x{ escaping
4044 mechanism.
4045
4046 When caseless matching is set, any letters in a class represent both
4047 their upper case and lower case versions, so for example, a caseless
4048 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
4049 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
4050 understands the concept of case for characters whose values are less
4051 than 128, so caseless matching is always possible. For characters with
4052 higher values, the concept of case is supported if PCRE is compiled
4053 with Unicode property support, but not otherwise. If you want to use
4054 caseless matching in UTF8-mode for characters 128 and above, you must
4055 ensure that PCRE is compiled with Unicode property support as well as
4056 with UTF-8 support.
4057
4058 Characters that might indicate line breaks are never treated in any
4059 special way when matching character classes, whatever line-ending
4060 sequence is in use, and whatever setting of the PCRE_DOTALL and
4061 PCRE_MULTILINE options is used. A class such as [^a] always matches one
4062 of these characters.
4063
4064 The minus (hyphen) character can be used to specify a range of charac-
4065 ters in a character class. For example, [d-m] matches any letter
4066 between d and m, inclusive. If a minus character is required in a
4067 class, it must be escaped with a backslash or appear in a position
4068 where it cannot be interpreted as indicating a range, typically as the
4069 first or last character in the class.
4070
4071 It is not possible to have the literal character "]" as the end charac-
4072 ter of a range. A pattern such as [W-]46] is interpreted as a class of
4073 two characters ("W" and "-") followed by a literal string "46]", so it
4074 would match "W46]" or "-46]". However, if the "]" is escaped with a
4075 backslash it is interpreted as the end of range, so [W-\]46] is inter-
4076 preted as a class containing a range followed by two other characters.
4077 The octal or hexadecimal representation of "]" can also be used to end
4078 a range.
4079
4080 Ranges operate in the collating sequence of character values. They can
4081 also be used for characters specified numerically, for example
4082 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
4083 are greater than 255, for example [\x{100}-\x{2ff}].
4084
4085 If a range that includes letters is used when caseless matching is set,
4086 it matches the letters in either case. For example, [W-c] is equivalent
4087 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
4088 character tables for a French locale are in use, [\xc8-\xcb] matches
4089 accented E characters in both cases. In UTF-8 mode, PCRE supports the
4090 concept of case for characters with values greater than 128 only when
4091 it is compiled with Unicode property support.
4092
4093 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
4094 \w, and \W may appear in a character class, and add the characters that
4095 they match to the class. For example, [\dABCDEF] matches any hexadeci-
4096 mal digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of
4097 \d, \s, \w and their upper case partners, just as it does when they
4098 appear outside a character class, as described in the section entitled
4099 "Generic character types" above. The escape sequence \b has a different
4100 meaning inside a character class; it matches the backspace character.
4101 The sequences \B, \N, \R, and \X are not special inside a character
4102 class. Like any other unrecognized escape sequences, they are treated
4103 as the literal characters "B", "N", "R", and "X" by default, but cause
4104 an error if the PCRE_EXTRA option is set.
4105
4106 A circumflex can conveniently be used with the upper case character
4107 types to specify a more restricted set of characters than the matching
4108 lower case type. For example, the class [^\W_] matches any letter or
4109 digit, but not underscore, whereas [\w] includes underscore. A positive
4110 character class should be read as "something OR something OR ..." and a
4111 negative class as "NOT something AND NOT something AND NOT ...".
4112
4113 The only metacharacters that are recognized in character classes are
4114 backslash, hyphen (only where it can be interpreted as specifying a
4115 range), circumflex (only at the start), opening square bracket (only
4116 when it can be interpreted as introducing a POSIX class name - see the
4117 next section), and the terminating closing square bracket. However,
4118 escaping other non-alphanumeric characters does no harm.
4119
4120
4121 POSIX CHARACTER CLASSES
4122
4123 Perl supports the POSIX notation for character classes. This uses names
4124 enclosed by [: and :] within the enclosing square brackets. PCRE also
4125 supports this notation. For example,
4126
4127 [01[:alpha:]%]
4128
4129 matches "0", "1", any alphabetic character, or "%". The supported class
4130 names are:
4131
4132 alnum letters and digits
4133 alpha letters
4134 ascii character codes 0 - 127
4135 blank space or tab only
4136 cntrl control characters
4137 digit decimal digits (same as \d)
4138 graph printing characters, excluding space
4139 lower lower case letters
4140 print printing characters, including space
4141 punct printing characters, excluding letters and digits and space
4142 space white space (not quite the same as \s)
4143 upper upper case letters
4144 word "word" characters (same as \w)
4145 xdigit hexadecimal digits
4146
4147 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
4148 and space (32). Notice that this list includes the VT character (code
4149 11). This makes "space" different to \s, which does not include VT (for
4150 Perl compatibility).
4151
4152 The name "word" is a Perl extension, and "blank" is a GNU extension
4153 from Perl 5.8. Another Perl extension is negation, which is indicated
4154 by a ^ character after the colon. For example,
4155
4156 [12[:^digit:]]
4157
4158 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4159 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4160 these are not supported, and an error is given if they are encountered.
4161
4162 By default, in UTF-8 mode, characters with values greater than 128 do
4163 not match any of the POSIX character classes. However, if the PCRE_UCP
4164 option is passed to pcre_compile(), some of the classes are changed so
4165 that Unicode character properties are used. This is achieved by replac-
4166 ing the POSIX classes by other sequences, as follows:
4167
4168 [:alnum:] becomes \p{Xan}
4169 [:alpha:] becomes \p{L}
4170 [:blank:] becomes \h
4171 [:digit:] becomes \p{Nd}
4172 [:lower:] becomes \p{Ll}
4173 [:space:] becomes \p{Xps}
4174 [:upper:] becomes \p{Lu}
4175 [:word:] becomes \p{Xwd}
4176
4177 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4178 POSIX classes are unchanged, and match only characters with code points
4179 less than 128.
4180
4181
4182 VERTICAL BAR
4183
4184 Vertical bar characters are used to separate alternative patterns. For
4185 example, the pattern
4186
4187 gilbert|sullivan
4188
4189 matches either "gilbert" or "sullivan". Any number of alternatives may
4190 appear, and an empty alternative is permitted (matching the empty
4191 string). The matching process tries each alternative in turn, from left
4192 to right, and the first one that succeeds is used. If the alternatives
4193 are within a subpattern (defined below), "succeeds" means matching the
4194 rest of the main pattern as well as the alternative in the subpattern.
4195
4196
4197 INTERNAL OPTION SETTING
4198
4199 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4200 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4201 within the pattern by a sequence of Perl option letters enclosed
4202 between "(?" and ")". The option letters are
4203
4204 i for PCRE_CASELESS
4205 m for PCRE_MULTILINE
4206 s for PCRE_DOTALL
4207 x for PCRE_EXTENDED
4208
4209 For example, (?im) sets caseless, multiline matching. It is also possi-
4210 ble to unset these options by preceding the letter with a hyphen, and a
4211 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4212 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
4213 is also permitted. If a letter appears both before and after the
4214 hyphen, the option is unset.
4215
4216 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
4217 can be changed in the same way as the Perl-compatible options by using
4218 the characters J, U and X respectively.
4219
4220 When one of these option changes occurs at top level (that is, not
4221 inside subpattern parentheses), the change applies to the remainder of
4222 the pattern that follows. If the change is placed right at the start of
4223 a pattern, PCRE extracts it into the global options (and it will there-
4224 fore show up in data extracted by the pcre_fullinfo() function).
4225
4226 An option change within a subpattern (see below for a description of
4227 subpatterns) affects only that part of the subpattern that follows it,
4228 so
4229
4230 (a(?i)b)c
4231
4232 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4233 used). By this means, options can be made to have different settings
4234 in different parts of the pattern. Any changes made in one alternative
4235 do carry on into subsequent branches within the same subpattern. For
4236 example,
4237
4238 (a(?i)b|c)
4239
4240 matches "ab", "aB", "c", and "C", even though when matching "C" the
4241 first branch is abandoned before the option setting. This is because
4242 the effects of option settings happen at compile time. There would be
4243 some very weird behaviour otherwise.
4244
4245 Note: There are other PCRE-specific options that can be set by the
4246 application when the compile or match functions are called. In some
4247 cases the pattern can contain special leading sequences such as (*CRLF)
4248 to override what the application has set or what has been defaulted.
4249 Details are given in the section entitled "Newline sequences" above.
4250 There are also the (*UTF8) and (*UCP) leading sequences that can be
4251 used to set UTF-8 and Unicode property modes; they are equivalent to
4252 setting the PCRE_UTF8 and the PCRE_UCP options, respectively.
4253
4254
4255 SUBPATTERNS
4256
4257 Subpatterns are delimited by parentheses (round brackets), which can be
4258 nested. Turning part of a pattern into a subpattern does two things:
4259
4260 1. It localizes a set of alternatives. For example, the pattern
4261
4262 cat(aract|erpillar|)
4263
4264 matches "cataract", "caterpillar", or "cat". Without the parentheses,
4265 it would match "cataract", "erpillar" or an empty string.
4266
4267 2. It sets up the subpattern as a capturing subpattern. This means
4268 that, when the whole pattern matches, that portion of the subject
4269 string that matched the subpattern is passed back to the caller via the
4270 ovector argument of pcre_exec(). Opening parentheses are counted from
4271 left to right (starting from 1) to obtain numbers for the capturing
4272 subpatterns. For example, if the string "the red king" is matched
4273 against the pattern
4274
4275 the ((red|white) (king|queen))
4276
4277 the captured substrings are "red king", "red", and "king", and are num-
4278 bered 1, 2, and 3, respectively.
4279
4280 The fact that plain parentheses fulfil two functions is not always
4281 helpful. There are often times when a grouping subpattern is required
4282 without a capturing requirement. If an opening parenthesis is followed
4283 by a question mark and a colon, the subpattern does not do any captur-
4284 ing, and is not counted when computing the number of any subsequent
4285 capturing subpatterns. For example, if the string "the white queen" is
4286 matched against the pattern
4287
4288 the ((?:red|white) (king|queen))
4289
4290 the captured substrings are "white queen" and "queen", and are numbered
4291 1 and 2. The maximum number of capturing subpatterns is 65535.
4292
4293 As a convenient shorthand, if any option settings are required at the
4294 start of a non-capturing subpattern, the option letters may appear
4295 between the "?" and the ":". Thus the two patterns
4296
4297 (?i:saturday|sunday)
4298 (?:(?i)saturday|sunday)
4299
4300 match exactly the same set of strings. Because alternative branches are
4301 tried from left to right, and options are not reset until the end of
4302 the subpattern is reached, an option setting in one branch does affect
4303 subsequent branches, so the above patterns match "SUNDAY" as well as
4304 "Saturday".
4305
4306
4307 DUPLICATE SUBPATTERN NUMBERS
4308
4309 Perl 5.10 introduced a feature whereby each alternative in a subpattern
4310 uses the same numbers for its capturing parentheses. Such a subpattern
4311 starts with (?| and is itself a non-capturing subpattern. For example,
4312 consider this pattern:
4313
4314 (?|(Sat)ur|(Sun))day
4315
4316 Because the two alternatives are inside a (?| group, both sets of cap-
4317 turing parentheses are numbered one. Thus, when the pattern matches,
4318 you can look at captured substring number one, whichever alternative
4319 matched. This construct is useful when you want to capture part, but
4320 not all, of one of a number of alternatives. Inside a (?| group, paren-
4321 theses are numbered as usual, but the number is reset at the start of
4322 each branch. The numbers of any capturing parentheses that follow the
4323 subpattern start after the highest number used in any branch. The fol-
4324 lowing example is taken from the Perl documentation. The numbers under-
4325 neath show in which buffer the captured content will be stored.
4326
4327 # before ---------------branch-reset----------- after
4328 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
4329 # 1 2 2 3 2 3 4
4330
4331 A back reference to a numbered subpattern uses the most recent value
4332 that is set for that number by any subpattern. The following pattern
4333 matches "abcabc" or "defdef":
4334
4335 /(?|(abc)|(def))\1/
4336
4337 In contrast, a recursive or "subroutine" call to a numbered subpattern
4338 always refers to the first one in the pattern with the given number.
4339 The following pattern matches "abcabc" or "defabc":
4340
4341 /(?|(abc)|(def))(?1)/
4342
4343 If a condition test for a subpattern's having matched refers to a non-
4344 unique number, the test is true if any of the subpatterns of that num-
4345 ber have matched.
4346
4347 An alternative approach to using this "branch reset" feature is to use
4348 duplicate named subpatterns, as described in the next section.
4349
4350
4351 NAMED SUBPATTERNS
4352
4353 Identifying capturing parentheses by number is simple, but it can be
4354 very hard to keep track of the numbers in complicated regular expres-
4355 sions. Furthermore, if an expression is modified, the numbers may
4356 change. To help with this difficulty, PCRE supports the naming of sub-
4357 patterns. This feature was not added to Perl until release 5.10. Python
4358 had the feature earlier, and PCRE introduced it at release 4.0, using
4359 the Python syntax. PCRE now supports both the Perl and the Python syn-
4360 tax. Perl allows identically numbered subpatterns to have different
4361 names, but PCRE does not.
4362
4363 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
4364 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
4365 to capturing parentheses from other parts of the pattern, such as back
4366 references, recursion, and conditions, can be made by name as well as
4367 by number.
4368
4369 Names consist of up to 32 alphanumeric characters and underscores.
4370 Named capturing parentheses are still allocated numbers as well as
4371 names, exactly as if the names were not present. The PCRE API provides
4372 function calls for extracting the name-to-number translation table from
4373 a compiled pattern. There is also a convenience function for extracting
4374 a captured substring by name.
4375
4376 By default, a name must be unique within a pattern, but it is possible
4377 to relax this constraint by setting the PCRE_DUPNAMES option at compile
4378 time. (Duplicate names are also always permitted for subpatterns with
4379 the same number, set up as described in the previous section.) Dupli-
4380 cate names can be useful for patterns where only one instance of the
4381 named parentheses can match. Suppose you want to match the name of a
4382 weekday, either as a 3-letter abbreviation or as the full name, and in
4383 both cases you want to extract the abbreviation. This pattern (ignoring
4384 the line breaks) does the job:
4385
4386 (?<DN>Mon|Fri|Sun)(?:day)?|
4387 (?<DN>Tue)(?:sday)?|
4388 (?<DN>Wed)(?:nesday)?|
4389 (?<DN>Thu)(?:rsday)?|
4390 (?<DN>Sat)(?:urday)?
4391
4392 There are five capturing substrings, but only one is ever set after a
4393 match. (An alternative way of solving this problem is to use a "branch
4394 reset" subpattern, as described in the previous section.)
4395
4396 The convenience function for extracting the data by name returns the
4397 substring for the first (and in this example, the only) subpattern of
4398 that name that matched. This saves searching to find which numbered
4399 subpattern it was.
4400
4401 If you make a back reference to a non-unique named subpattern from
4402 elsewhere in the pattern, the one that corresponds to the first occur-
4403 rence of the name is used. In the absence of duplicate numbers (see the
4404 previous section) this is the one with the lowest number. If you use a
4405 named reference in a condition test (see the section about conditions
4406 below), either to check whether a subpattern has matched, or to check
4407 for recursion, all subpatterns with the same name are tested. If the
4408 condition is true for any one of them, the overall condition is true.
4409 This is the same behaviour as testing by number. For further details of
4410 the interfaces for handling named subpatterns, see the pcreapi documen-
4411 tation.
4412
4413 Warning: You cannot use different names to distinguish between two sub-
4414 patterns with the same number because PCRE uses only the numbers when
4415 matching. For this reason, an error is given at compile time if differ-
4416 ent names are given to subpatterns with the same number. However, you
4417 can give the same name to subpatterns with the same number, even when
4418 PCRE_DUPNAMES is not set.
4419
4420
4421 REPETITION
4422
4423 Repetition is specified by quantifiers, which can follow any of the
4424 following items:
4425
4426 a literal data character
4427 the dot metacharacter
4428 the \C escape sequence
4429 the \X escape sequence (in UTF-8 mode with Unicode properties)
4430 the \R escape sequence
4431 an escape such as \d or \pL that matches a single character
4432 a character class
4433 a back reference (see next section)
4434 a parenthesized subpattern (unless it is an assertion)
4435 a recursive or "subroutine" call to a subpattern
4436
4437 The general repetition quantifier specifies a minimum and maximum num-
4438 ber of permitted matches, by giving the two numbers in curly brackets
4439 (braces), separated by a comma. The numbers must be less than 65536,
4440 and the first must be less than or equal to the second. For example:
4441
4442 z{2,4}
4443
4444 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4445 special character. If the second number is omitted, but the comma is
4446 present, there is no upper limit; if the second number and the comma
4447 are both omitted, the quantifier specifies an exact number of required
4448 matches. Thus
4449
4450 [aeiou]{3,}
4451
4452 matches at least 3 successive vowels, but may match many more, while
4453
4454 \d{8}
4455
4456 matches exactly 8 digits. An opening curly bracket that appears in a
4457 position where a quantifier is not allowed, or one that does not match
4458 the syntax of a quantifier, is taken as a literal character. For exam-
4459 ple, {,6} is not a quantifier, but a literal string of four characters.
4460
4461 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4462 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4463 acters, each of which is represented by a two-byte sequence. Similarly,
4464 when Unicode property support is available, \X{3} matches three Unicode
4465 extended sequences, each of which may be several bytes long (and they
4466 may be of different lengths).
4467
4468 The quantifier {0} is permitted, causing the expression to behave as if
4469 the previous item and the quantifier were not present. This may be use-
4470 ful for subpatterns that are referenced as subroutines from elsewhere
4471 in the pattern (but see also the section entitled "Defining subpatterns
4472 for use by reference only" below). Items other than subpatterns that
4473 have a {0} quantifier are omitted from the compiled pattern.
4474
4475 For convenience, the three most common quantifiers have single-charac-
4476 ter abbreviations:
4477
4478 * is equivalent to {0,}
4479 + is equivalent to {1,}
4480 ? is equivalent to {0,1}
4481
4482 It is possible to construct infinite loops by following a subpattern
4483 that can match no characters with a quantifier that has no upper limit,
4484 for example:
4485
4486 (a?)*
4487
4488 Earlier versions of Perl and PCRE used to give an error at compile time
4489 for such patterns. However, because there are cases where this can be
4490 useful, such patterns are now accepted, but if any repetition of the
4491 subpattern does in fact match no characters, the loop is forcibly bro-
4492 ken.
4493
4494 By default, the quantifiers are "greedy", that is, they match as much
4495 as possible (up to the maximum number of permitted times), without
4496 causing the rest of the pattern to fail. The classic example of where
4497 this gives problems is in trying to match comments in C programs. These
4498 appear between /* and */ and within the comment, individual * and /
4499 characters may appear. An attempt to match C comments by applying the
4500 pattern
4501
4502 /\*.*\*/
4503
4504 to the string
4505
4506 /* first comment */ not comment /* second comment */
4507
4508 fails, because it matches the entire string owing to the greediness of
4509 the .* item.
4510
4511 However, if a quantifier is followed by a question mark, it ceases to
4512 be greedy, and instead matches the minimum number of times possible, so
4513 the pattern
4514
4515 /\*.*?\*/
4516
4517 does the right thing with the C comments. The meaning of the various
4518 quantifiers is not otherwise changed, just the preferred number of
4519 matches. Do not confuse this use of question mark with its use as a
4520 quantifier in its own right. Because it has two uses, it can sometimes
4521 appear doubled, as in
4522
4523 \d??\d
4524
4525 which matches one digit by preference, but can match two if that is the
4526 only way the rest of the pattern matches.
4527
4528 If the PCRE_UNGREEDY option is set (an option that is not available in
4529 Perl), the quantifiers are not greedy by default, but individual ones
4530 can be made greedy by following them with a question mark. In other
4531 words, it inverts the default behaviour.
4532
4533 When a parenthesized subpattern is quantified with a minimum repeat
4534 count that is greater than 1 or with a limited maximum, more memory is
4535 required for the compiled pattern, in proportion to the size of the
4536 minimum or maximum.
4537
4538 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4539 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4540 the pattern is implicitly anchored, because whatever follows will be
4541 tried against every character position in the subject string, so there
4542 is no point in retrying the overall match at any position after the
4543 first. PCRE normally treats such a pattern as though it were preceded
4544 by \A.
4545
4546 In cases where it is known that the subject string contains no new-
4547 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4548 mization, or alternatively using ^ to indicate anchoring explicitly.
4549
4550 However, there is one situation where the optimization cannot be used.
4551 When .* is inside capturing parentheses that are the subject of a back
4552 reference elsewhere in the pattern, a match at the start may fail where
4553 a later one succeeds. Consider, for example:
4554
4555 (.*)abc\1
4556
4557 If the subject is "xyz123abc123" the match point is the fourth charac-
4558 ter. For this reason, such a pattern is not implicitly anchored.
4559
4560 When a capturing subpattern is repeated, the value captured is the sub-
4561 string that matched the final iteration. For example, after
4562
4563 (tweedle[dume]{3}\s*)+
4564
4565 has matched "tweedledum tweedledee" the value of the captured substring
4566 is "tweedledee". However, if there are nested capturing subpatterns,
4567 the corresponding captured values may have been set in previous itera-
4568 tions. For example, after
4569
4570 /(a|(b))+/
4571
4572 matches "aba" the value of the second captured substring is "b".
4573
4574
4575 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4576
4577 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4578 repetition, failure of what follows normally causes the repeated item
4579 to be re-evaluated to see if a different number of repeats allows the
4580 rest of the pattern to match. Sometimes it is useful to prevent this,
4581 either to change the nature of the match, or to cause it fail earlier
4582 than it otherwise might, when the author of the pattern knows there is
4583 no point in carrying on.
4584
4585 Consider, for example, the pattern \d+foo when applied to the subject
4586 line
4587
4588 123456bar
4589
4590 After matching all 6 digits and then failing to match "foo", the normal
4591 action of the matcher is to try again with only 5 digits matching the
4592 \d+ item, and then with 4, and so on, before ultimately failing.
4593 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4594 the means for specifying that once a subpattern has matched, it is not
4595 to be re-evaluated in this way.
4596
4597 If we use atomic grouping for the previous example, the matcher gives
4598 up immediately on failing to match "foo" the first time. The notation
4599 is a kind of special parenthesis, starting with (?> as in this example:
4600
4601 (?>\d+)foo
4602
4603 This kind of parenthesis "locks up" the part of the pattern it con-
4604 tains once it has matched, and a failure further into the pattern is
4605 prevented from backtracking into it. Backtracking past it to previous
4606 items, however, works as normal.
4607
4608 An alternative description is that a subpattern of this type matches
4609 the string of characters that an identical standalone pattern would
4610 match, if anchored at the current point in the subject string.
4611
4612 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4613 such as the above example can be thought of as a maximizing repeat that
4614 must swallow everything it can. So, while both \d+ and \d+? are pre-
4615 pared to adjust the number of digits they match in order to make the
4616 rest of the pattern match, (?>\d+) can only match an entire sequence of
4617 digits.
4618
4619 Atomic groups in general can of course contain arbitrarily complicated
4620 subpatterns, and can be nested. However, when the subpattern for an
4621 atomic group is just a single repeated item, as in the example above, a
4622 simpler notation, called a "possessive quantifier" can be used. This
4623 consists of an additional + character following a quantifier. Using
4624 this notation, the previous example can be rewritten as
4625
4626 \d++foo
4627
4628 Note that a possessive quantifier can be used with an entire group, for
4629 example:
4630
4631 (abc|xyz){2,3}+
4632
4633 Possessive quantifiers are always greedy; the setting of the
4634 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4635 simpler forms of atomic group. However, there is no difference in the
4636 meaning of a possessive quantifier and the equivalent atomic group,
4637 though there may be a performance difference; possessive quantifiers
4638 should be slightly faster.
4639
4640 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4641 tax. Jeffrey Friedl originated the idea (and the name) in the first
4642 edition of his book. Mike McCloskey liked it, so implemented it when he
4643 built Sun's Java package, and PCRE copied it from there. It ultimately
4644 found its way into Perl at release 5.10.
4645
4646 PCRE has an optimization that automatically "possessifies" certain sim-
4647 ple pattern constructs. For example, the sequence A+B is treated as
4648 A++B because there is no point in backtracking into a sequence of A's
4649 when B must follow.
4650
4651 When a pattern contains an unlimited repeat inside a subpattern that
4652 can itself be repeated an unlimited number of times, the use of an
4653 atomic group is the only way to avoid some failing matches taking a
4654 very long time indeed. The pattern
4655
4656 (\D+|<\d+>)*[!?]
4657
4658 matches an unlimited number of substrings that either consist of non-
4659 digits, or digits enclosed in <>, followed by either ! or ?. When it
4660 matches, it runs quickly. However, if it is applied to
4661
4662 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4663
4664 it takes a long time before reporting failure. This is because the
4665 string can be divided between the internal \D+ repeat and the external
4666 * repeat in a large number of ways, and all have to be tried. (The
4667 example uses [!?] rather than a single character at the end, because
4668 both PCRE and Perl have an optimization that allows for fast failure
4669 when a single character is used. They remember the last single charac-
4670 ter that is required for a match, and fail early if it is not present
4671 in the string.) If the pattern is changed so that it uses an atomic
4672 group, like this:
4673
4674 ((?>\D+)|<\d+>)*[!?]
4675
4676 sequences of non-digits cannot be broken, and failure happens quickly.
4677
4678
4679 BACK REFERENCES
4680
4681 Outside a character class, a backslash followed by a digit greater than
4682 0 (and possibly further digits) is a back reference to a capturing sub-
4683 pattern earlier (that is, to its left) in the pattern, provided there
4684 have been that many previous capturing left parentheses.
4685
4686 However, if the decimal number following the backslash is less than 10,
4687 it is always taken as a back reference, and causes an error only if
4688 there are not that many capturing left parentheses in the entire pat-
4689 tern. In other words, the parentheses that are referenced need not be
4690 to the left of the reference for numbers less than 10. A "forward back
4691 reference" of this type can make sense when a repetition is involved
4692 and the subpattern to the right has participated in an earlier itera-
4693 tion.
4694
4695 It is not possible to have a numerical "forward back reference" to a
4696 subpattern whose number is 10 or more using this syntax because a
4697 sequence such as \50 is interpreted as a character defined in octal.
4698 See the subsection entitled "Non-printing characters" above for further
4699 details of the handling of digits following a backslash. There is no
4700 such problem when named parentheses are used. A back reference to any
4701 subpattern is possible using named parentheses (see below).
4702
4703 Another way of avoiding the ambiguity inherent in the use of digits
4704 following a backslash is to use the \g escape sequence. This escape
4705 must be followed by an unsigned number or a negative number, optionally
4706 enclosed in braces. These examples are all identical:
4707
4708 (ring), \1
4709 (ring), \g1
4710 (ring), \g{1}
4711
4712 An unsigned number specifies an absolute reference without the ambigu-
4713 ity that is present in the older syntax. It is also useful when literal
4714 digits follow the reference. A negative number is a relative reference.
4715 Consider this example:
4716
4717 (abc(def)ghi)\g{-1}
4718
4719 The sequence \g{-1} is a reference to the most recently started captur-
4720 ing subpattern before \g, that is, is it equivalent to \2 in this exam-
4721 ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
4722 references can be helpful in long patterns, and also in patterns that
4723 are created by joining together fragments that contain references
4724 within themselves.
4725
4726 A back reference matches whatever actually matched the capturing sub-
4727 pattern in the current subject string, rather than anything matching
4728 the subpattern itself (see "Subpatterns as subroutines" below for a way
4729 of doing that). So the pattern
4730
4731 (sens|respons)e and \1ibility
4732
4733 matches "sense and sensibility" and "response and responsibility", but
4734 not "sense and responsibility". If caseful matching is in force at the
4735 time of the back reference, the case of letters is relevant. For exam-
4736 ple,
4737
4738 ((?i)rah)\s+\1
4739
4740 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4741 original capturing subpattern is matched caselessly.
4742
4743 There are several different ways of writing back references to named
4744 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4745 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4746 unified back reference syntax, in which \g can be used for both numeric
4747 and named references, is also supported. We could rewrite the above
4748 example in any of the following ways:
4749
4750 (?<p1>(?i)rah)\s+\k<p1>
4751 (?'p1'(?i)rah)\s+\k{p1}
4752 (?P<p1>(?i)rah)\s+(?P=p1)
4753 (?<p1>(?i)rah)\s+\g{p1}
4754
4755 A subpattern that is referenced by name may appear in the pattern
4756 before or after the reference.
4757
4758 There may be more than one back reference to the same subpattern. If a
4759 subpattern has not actually been used in a particular match, any back
4760 references to it always fail by default. For example, the pattern
4761
4762 (a|(bc))\2
4763
4764 always fails if it starts to match "a" rather than "bc". However, if
4765 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
4766 ence to an unset value matches an empty string.
4767
4768 Because there may be many capturing parentheses in a pattern, all dig-
4769 its following a backslash are taken as part of a potential back refer-
4770 ence number. If the pattern continues with a digit character, some
4771 delimiter must be used to terminate the back reference. If the
4772 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
4773 syntax or an empty comment (see "Comments" below) can be used.
4774
4775 Recursive back references
4776
4777 A back reference that occurs inside the parentheses to which it refers
4778 fails when the subpattern is first used, so, for example, (a\1) never
4779 matches. However, such references can be useful inside repeated sub-
4780 patterns. For example, the pattern
4781
4782 (a|b\1)+
4783
4784 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4785 ation of the subpattern, the back reference matches the character
4786 string corresponding to the previous iteration. In order for this to
4787 work, the pattern must be such that the first iteration does not need
4788 to match the back reference. This can be done using alternation, as in
4789 the example above, or by a quantifier with a minimum of zero.
4790
4791 Back references of this type cause the group that they reference to be
4792 treated as an atomic group. Once the whole group has been matched, a
4793 subsequent matching failure cannot cause backtracking into the middle
4794 of the group.
4795
4796
4797 ASSERTIONS
4798
4799 An assertion is a test on the characters following or preceding the
4800 current matching point that does not actually consume any characters.
4801 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4802 described above.
4803
4804 More complicated assertions are coded as subpatterns. There are two
4805 kinds: those that look ahead of the current position in the subject
4806 string, and those that look behind it. An assertion subpattern is
4807 matched in the normal way, except that it does not cause the current
4808 matching position to be changed.
4809
4810 Assertion subpatterns are not capturing subpatterns, and may not be
4811 repeated, because it makes no sense to assert the same thing several
4812 times. If any kind of assertion contains capturing subpatterns within
4813 it, these are counted for the purposes of numbering the capturing sub-
4814 patterns in the whole pattern. However, substring capturing is carried
4815 out only for positive assertions, because it does not make sense for
4816 negative assertions.
4817
4818 Lookahead assertions
4819
4820 Lookahead assertions start with (?= for positive assertions and (?! for
4821 negative assertions. For example,
4822
4823 \w+(?=;)
4824
4825 matches a word followed by a semicolon, but does not include the semi-
4826 colon in the match, and
4827
4828 foo(?!bar)
4829
4830 matches any occurrence of "foo" that is not followed by "bar". Note
4831 that the apparently similar pattern
4832
4833 (?!foo)bar
4834
4835 does not find an occurrence of "bar" that is preceded by something
4836 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4837 the assertion (?!foo) is always true when the next three characters are
4838 "bar". A lookbehind assertion is needed to achieve the other effect.
4839
4840 If you want to force a matching failure at some point in a pattern, the
4841 most convenient way to do it is with (?!) because an empty string
4842 always matches, so an assertion that requires there not to be an empty
4843 string must always fail. The backtracking control verb (*FAIL) or (*F)
4844 is a synonym for (?!).
4845
4846 Lookbehind assertions
4847
4848 Lookbehind assertions start with (?<= for positive assertions and (?<!
4849 for negative assertions. For example,
4850
4851 (?<!foo)bar
4852
4853 does find an occurrence of "bar" that is not preceded by "foo". The
4854 contents of a lookbehind assertion are restricted such that all the
4855 strings it matches must have a fixed length. However, if there are sev-
4856 eral top-level alternatives, they do not all have to have the same
4857 fixed length. Thus
4858
4859 (?<=bullock|donkey)
4860
4861 is permitted, but
4862
4863 (?<!dogs?|cats?)
4864
4865 causes an error at compile time. Branches that match different length
4866 strings are permitted only at the top level of a lookbehind assertion.
4867 This is an extension compared with Perl, which requires all branches to
4868 match the same length of string. An assertion such as
4869
4870 (?<=ab(c|de))
4871
4872 is not permitted, because its single top-level branch can match two
4873 different lengths, but it is acceptable to PCRE if rewritten to use two
4874 top-level branches:
4875
4876 (?<=abc|abde)
4877
4878 In some cases, the escape sequence \K (see above) can be used instead
4879 of a lookbehind assertion to get round the fixed-length restriction.
4880
4881 The implementation of lookbehind assertions is, for each alternative,
4882 to temporarily move the current position back by the fixed length and
4883 then try to match. If there are insufficient characters before the cur-
4884 rent position, the assertion fails.
4885
4886 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4887 mode) to appear in lookbehind assertions, because it makes it impossi-
4888 ble to calculate the length of the lookbehind. The \X and \R escapes,
4889 which can match different numbers of bytes, are also not permitted.
4890
4891 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
4892 lookbehinds, as long as the subpattern matches a fixed-length string.
4893 Recursion, however, is not supported.
4894
4895 Possessive quantifiers can be used in conjunction with lookbehind
4896 assertions to specify efficient matching of fixed-length strings at the
4897 end of subject strings. Consider a simple pattern such as
4898
4899 abcd$
4900
4901 when applied to a long string that does not match. Because matching
4902 proceeds from left to right, PCRE will look for each "a" in the subject
4903 and then see if what follows matches the rest of the pattern. If the
4904 pattern is specified as
4905
4906 ^.*abcd$
4907
4908 the initial .* matches the entire string at first, but when this fails
4909 (because there is no following "a"), it backtracks to match all but the
4910 last character, then all but the last two characters, and so on. Once
4911 again the search for "a" covers the entire string, from right to left,
4912 so we are no better off. However, if the pattern is written as
4913
4914 ^.*+(?<=abcd)
4915
4916 there can be no backtracking for the .*+ item; it can match only the
4917 entire string. The subsequent lookbehind assertion does a single test
4918 on the last four characters. If it fails, the match fails immediately.
4919 For long strings, this approach makes a significant difference to the
4920 processing time.
4921
4922 Using multiple assertions
4923
4924 Several assertions (of any sort) may occur in succession. For example,
4925
4926 (?<=\d{3})(?<!999)foo
4927
4928 matches "foo" preceded by three digits that are not "999". Notice that
4929 each of the assertions is applied independently at the same point in
4930 the subject string. First there is a check that the previous three
4931 characters are all digits, and then there is a check that the same
4932 three characters are not "999". This pattern does not match "foo" pre-
4933 ceded by six characters, the first of which are digits and the last
4934 three of which are not "999". For example, it doesn't match "123abc-
4935 foo". A pattern to do that is
4936
4937 (?<=\d{3}...)(?<!999)foo
4938
4939 This time the first assertion looks at the preceding six characters,
4940 checking that the first three are digits, and then the second assertion
4941 checks that the preceding three characters are not "999".
4942
4943 Assertions can be nested in any combination. For example,
4944
4945 (?<=(?<!foo)bar)baz
4946
4947 matches an occurrence of "baz" that is preceded by "bar" which in turn
4948 is not preceded by "foo", while
4949
4950 (?<=\d{3}(?!999)...)foo
4951
4952 is another pattern that matches "foo" preceded by three digits and any
4953 three characters that are not "999".
4954
4955
4956 CONDITIONAL SUBPATTERNS
4957
4958 It is possible to cause the matching process to obey a subpattern con-
4959 ditionally or to choose between two alternative subpatterns, depending
4960 on the result of an assertion, or whether a specific capturing subpat-
4961 tern has already been matched. The two possible forms of conditional
4962 subpattern are:
4963
4964 (?(condition)yes-pattern)
4965 (?(condition)yes-pattern|no-pattern)
4966
4967 If the condition is satisfied, the yes-pattern is used; otherwise the
4968 no-pattern (if present) is used. If there are more than two alterna-
4969 tives in the subpattern, a compile-time error occurs. Each of the two
4970 alternatives may itself contain nested subpatterns of any form, includ-
4971 ing conditional subpatterns; the restriction to two alternatives
4972 applies only at the level of the condition. This pattern fragment is an
4973 example where the alternatives are complex:
4974
4975 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
4976
4977
4978 There are four kinds of condition: references to subpatterns, refer-
4979 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4980
4981 Checking for a used subpattern by number
4982
4983 If the text between the parentheses consists of a sequence of digits,
4984 the condition is true if a capturing subpattern of that number has pre-
4985 viously matched. If there is more than one capturing subpattern with
4986 the same number (see the earlier section about duplicate subpattern
4987 numbers), the condition is true if any of them have matched. An alter-
4988 native notation is to precede the digits with a plus or minus sign. In
4989 this case, the subpattern number is relative rather than absolute. The
4990 most recently opened parentheses can be referenced by (?(-1), the next
4991 most recent by (?(-2), and so on. Inside loops it can also make sense
4992 to refer to subsequent groups. The next parentheses to be opened can be
4993 referenced as (?(+1), and so on. (The value zero in any of these forms
4994 is not used; it provokes a compile-time error.)
4995
4996 Consider the following pattern, which contains non-significant white
4997 space to make it more readable (assume the PCRE_EXTENDED option) and to
4998 divide it into three parts for ease of discussion:
4999
5000 ( \( )? [^()]+ (?(1) \) )
5001
5002 The first part matches an optional opening parenthesis, and if that
5003 character is present, sets it as the first captured substring. The sec-
5004 ond part matches one or more characters that are not parentheses. The
5005 third part is a conditional subpattern that tests whether or not the
5006 first set of parentheses matched. If they did, that is, if subject
5007 started with an opening parenthesis, the condition is true, and so the
5008 yes-pattern is executed and a closing parenthesis is required. Other-
5009 wise, since no-pattern is not present, the subpattern matches nothing.
5010 In other words, this pattern matches a sequence of non-parentheses,
5011 optionally enclosed in parentheses.
5012
5013 If you were embedding this pattern in a larger one, you could use a
5014 relative reference:
5015
5016 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
5017
5018 This makes the fragment independent of the parentheses in the larger
5019 pattern.
5020
5021 Checking for a used subpattern by name
5022
5023 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
5024 used subpattern by name. For compatibility with earlier versions of
5025 PCRE, which had this facility before Perl, the syntax (?(name)...) is
5026 also recognized. However, there is a possible ambiguity with this syn-
5027 tax, because subpattern names may consist entirely of digits. PCRE
5028 looks first for a named subpattern; if it cannot find one and the name
5029 consists entirely of digits, PCRE looks for a subpattern of that num-
5030 ber, which must be greater than zero. Using subpattern names that con-
5031 sist entirely of digits is not recommended.
5032
5033 Rewriting the above example to use a named subpattern gives this:
5034
5035 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
5036
5037 If the name used in a condition of this kind is a duplicate, the test
5038 is applied to all subpatterns of the same name, and is true if any one
5039 of them has matched.
5040
5041 Checking for pattern recursion
5042
5043 If the condition is the string (R), and there is no subpattern with the
5044 name R, the condition is true if a recursive call to the whole pattern
5045 or any subpattern has been made. If digits or a name preceded by amper-
5046 sand follow the letter R, for example:
5047
5048 (?(R3)...) or (?(R&name)...)
5049
5050 the condition is true if the most recent recursion is into a subpattern
5051 whose number or name is given. This condition does not check the entire
5052 recursion stack. If the name used in a condition of this kind is a
5053 duplicate, the test is applied to all subpatterns of the same name, and
5054 is true if any one of them is the most recent recursion.
5055
5056 At "top level", all these recursion test conditions are false. The
5057 syntax for recursive patterns is described below.
5058
5059 Defining subpatterns for use by reference only
5060
5061 If the condition is the string (DEFINE), and there is no subpattern
5062 with the name DEFINE, the condition is always false. In this case,
5063 there may be only one alternative in the subpattern. It is always
5064 skipped if control reaches this point in the pattern; the idea of
5065 DEFINE is that it can be used to define "subroutines" that can be ref-
5066 erenced from elsewhere. (The use of "subroutines" is described below.)
5067 For example, a pattern to match an IPv4 address such as
5068 "192.168.23.245" could be written like this (ignore whitespace and line
5069 breaks):
5070
5071 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
5072 \b (?&byte) (\.(?&byte)){3} \b
5073
5074 The first part of the pattern is a DEFINE group inside which a another
5075 group named "byte" is defined. This matches an individual component of
5076 an IPv4 address (a number less than 256). When matching takes place,
5077 this part of the pattern is skipped because DEFINE acts like a false
5078 condition. The rest of the pattern uses references to the named group
5079 to match the four dot-separated components of an IPv4 address, insist-
5080 ing on a word boundary at each end.
5081
5082 Assertion conditions
5083
5084 If the condition is not in any of the above formats, it must be an
5085 assertion. This may be a positive or negative lookahead or lookbehind
5086 assertion. Consider this pattern, again containing non-significant
5087 white space, and with the two alternatives on the second line:
5088
5089 (?(?=[^a-z]*[a-z])
5090 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
5091
5092 The condition is a positive lookahead assertion that matches an
5093 optional sequence of non-letters followed by a letter. In other words,
5094 it tests for the presence of at least one letter in the subject. If a
5095 letter is found, the subject is matched against the first alternative;
5096 otherwise it is matched against the second. This pattern matches
5097 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
5098 letters and dd are digits.
5099
5100
5101 COMMENTS
5102
5103 There are two ways of including comments in patterns that are processed
5104 by PCRE. In both cases, the start of the comment must not be in a char-
5105 acter class, nor in the middle of any other sequence of related charac-
5106 ters such as (?: or a subpattern name or number. The characters that
5107 make up a comment play no part in the pattern matching.
5108
5109 The sequence (?# marks the start of a comment that continues up to the
5110 next closing parenthesis. Nested parentheses are not permitted. If the
5111 PCRE_EXTENDED option is set, an unescaped # character also introduces a
5112 comment, which in this case continues to immediately after the next
5113 newline character or character sequence in the pattern. Which charac-
5114 ters are interpreted as newlines is controlled by the options passed to
5115 pcre_compile() or by a special sequence at the start of the pattern, as
5116 described in the section entitled "Newline conventions" above. Note
5117 that the end of this type of comment is a literal newline sequence in
5118 the pattern; escape sequences that happen to represent a newline do not
5119 count. For example, consider this pattern when PCRE_EXTENDED is set,
5120 and the default newline convention is in force:
5121
5122 abc #comment \n still comment
5123
5124 On encountering the # character, pcre_compile() skips along, looking
5125 for a newline in the pattern. The sequence \n is still literal at this
5126 stage, so it does not terminate the comment. Only an actual character
5127 with the code value 0x0a (the default newline) does so.
5128
5129
5130 RECURSIVE PATTERNS
5131
5132 Consider the problem of matching a string in parentheses, allowing for
5133 unlimited nested parentheses. Without the use of recursion, the best
5134 that can be done is to use a pattern that matches up to some fixed
5135 depth of nesting. It is not possible to handle an arbitrary nesting
5136 depth.
5137
5138 For some time, Perl has provided a facility that allows regular expres-
5139 sions to recurse (amongst other things). It does this by interpolating
5140 Perl code in the expression at run time, and the code can refer to the
5141 expression itself. A Perl pattern using code interpolation to solve the
5142 parentheses problem can be created like this:
5143
5144 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
5145
5146 The (?p{...}) item interpolates Perl code at run time, and in this case
5147 refers recursively to the pattern in which it appears.
5148
5149 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
5150 it supports special syntax for recursion of the entire pattern, and
5151 also for individual subpattern recursion. After its introduction in
5152 PCRE and Python, this kind of recursion was subsequently introduced
5153 into Perl at release 5.10.
5154
5155 A special item that consists of (? followed by a number greater than
5156 zero and a closing parenthesis is a recursive call of the subpattern of
5157 the given number, provided that it occurs inside that subpattern. (If
5158 not, it is a "subroutine" call, which is described in the next sec-
5159 tion.) The special item (?R) or (?0) is a recursive call of the entire
5160 regular expression.
5161
5162 This PCRE pattern solves the nested parentheses problem (assume the
5163 PCRE_EXTENDED option is set so that white space is ignored):
5164
5165 \( ( [^()]++ | (?R) )* \)
5166
5167 First it matches an opening parenthesis. Then it matches any number of
5168 substrings which can either be a sequence of non-parentheses, or a
5169 recursive match of the pattern itself (that is, a correctly parenthe-
5170 sized substring). Finally there is a closing parenthesis. Note the use
5171 of a possessive quantifier to avoid backtracking into sequences of non-
5172 parentheses.
5173
5174 If this were part of a larger pattern, you would not want to recurse
5175 the entire pattern, so instead you could use this:
5176
5177 ( \( ( [^()]++ | (?1) )* \) )
5178
5179 We have put the pattern into parentheses, and caused the recursion to
5180 refer to them instead of the whole pattern.
5181
5182 In a larger pattern, keeping track of parenthesis numbers can be
5183 tricky. This is made easier by the use of relative references. Instead
5184 of (?1) in the pattern above you can write (?-2) to refer to the second
5185 most recently opened parentheses preceding the recursion. In other
5186 words, a negative number counts capturing parentheses leftwards from
5187 the point at which it is encountered.
5188
5189 It is also possible to refer to subsequently opened parentheses, by
5190 writing references such as (?+2). However, these cannot be recursive
5191 because the reference is not inside the parentheses that are refer-
5192 enced. They are always "subroutine" calls, as described in the next
5193 section.
5194
5195 An alternative approach is to use named parentheses instead. The Perl
5196 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5197 supported. We could rewrite the above example as follows:
5198
5199 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5200
5201 If there is more than one subpattern with the same name, the earliest
5202 one is used.
5203
5204 This particular example pattern that we have been looking at contains
5205 nested unlimited repeats, and so the use of a possessive quantifier for
5206 matching strings of non-parentheses is important when applying the pat-
5207 tern to strings that do not match. For example, when this pattern is
5208 applied to
5209
5210 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5211
5212 it yields "no match" quickly. However, if a possessive quantifier is
5213 not used, the match runs for a very long time indeed because there are
5214 so many different ways the + and * repeats can carve up the subject,
5215 and all have to be tested before failure can be reported.
5216
5217 At the end of a match, the values of capturing parentheses are those
5218 from the outermost level. If you want to obtain intermediate values, a
5219 callout function can be used (see below and the pcrecallout documenta-
5220 tion). If the pattern above is matched against
5221
5222 (ab(cd)ef)
5223
5224 the value for the inner capturing parentheses (numbered 2) is "ef",
5225 which is the last value taken on at the top level. If a capturing sub-
5226 pattern is not matched at the top level, its final value is unset, even
5227 if it is (temporarily) set at a deeper level.
5228
5229 If there are more than 15 capturing parentheses in a pattern, PCRE has
5230 to obtain extra memory to store data during a recursion, which it does
5231 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5232 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5233
5234 Do not confuse the (?R) item with the condition (R), which tests for
5235 recursion. Consider this pattern, which matches text in angle brack-
5236 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5237 brackets (that is, when recursing), whereas any characters are permit-
5238 ted at the outer level.
5239
5240 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5241
5242 In this pattern, (?(R) is the start of a conditional subpattern, with
5243 two different alternatives for the recursive and non-recursive cases.
5244 The (?R) item is the actual recursive call.
5245
5246 Recursion difference from Perl
5247
5248 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5249 always treated as an atomic group. That is, once it has matched some of
5250 the subject string, it is never re-entered, even if it contains untried
5251 alternatives and there is a subsequent matching failure. This can be
5252 illustrated by the following pattern, which purports to match a palin-
5253 dromic string that contains an odd number of characters (for example,
5254 "a", "aba", "abcba", "abcdcba"):
5255
5256 ^(.|(.)(?1)\2)$
5257
5258 The idea is that it either matches a single character, or two identical
5259 characters surrounding a sub-palindrome. In Perl, this pattern works;
5260 in PCRE it does not if the pattern is longer than three characters.
5261 Consider the subject string "abcba":
5262
5263 At the top level, the first character is matched, but as it is not at
5264 the end of the string, the first alternative fails; the second alterna-
5265 tive is taken and the recursion kicks in. The recursive call to subpat-
5266 tern 1 successfully matches the next character ("b"). (Note that the
5267 beginning and end of line tests are not part of the recursion).
5268
5269 Back at the top level, the next character ("c") is compared with what
5270 subpattern 2 matched, which was "a". This fails. Because the recursion
5271 is treated as an atomic group, there are now no backtracking points,
5272 and so the entire match fails. (Perl is able, at this point, to re-
5273 enter the recursion and try the second alternative.) However, if the
5274 pattern is written with the alternatives in the other order, things are
5275 different:
5276
5277 ^((.)(?1)\2|.)$
5278
5279 This time, the recursing alternative is tried first, and continues to
5280 recurse until it runs out of characters, at which point the recursion
5281 fails. But this time we do have another alternative to try at the
5282 higher level. That is the big difference: in the previous case the
5283 remaining alternative is at a deeper recursion level, which PCRE cannot
5284 use.
5285
5286 To change the pattern so that it matches all palindromic strings, not
5287 just those with an odd number of characters, it is tempting to change
5288 the pattern to this:
5289
5290 ^((.)(?1)\2|.?)$
5291
5292 Again, this works in Perl, but not in PCRE, and for the same reason.
5293 When a deeper recursion has matched a single character, it cannot be
5294 entered again in order to match an empty string. The solution is to
5295 separate the two cases, and write out the odd and even cases as alter-
5296 natives at the higher level:
5297
5298 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
5299
5300 If you want to match typical palindromic phrases, the pattern has to
5301 ignore all non-word characters, which can be done like this:
5302
5303 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
5304
5305 If run with the PCRE_CASELESS option, this pattern matches phrases such
5306 as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
5307 Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
5308 ing into sequences of non-word characters. Without this, PCRE takes a
5309 great deal longer (ten times or more) to match typical phrases, and
5310 Perl takes so long that you think it has gone into a loop.
5311
5312 WARNING: The palindrome-matching patterns above work only if the sub-
5313 ject string does not start with a palindrome that is shorter than the
5314 entire string. For example, although "abcba" is correctly matched, if
5315 the subject is "ababa", PCRE finds the palindrome "aba" at the start,
5316 then fails at top level because the end of the string does not follow.
5317 Once again, it cannot jump back into the recursion to try other alter-
5318 natives, so the entire match fails.
5319
5320
5321 SUBPATTERNS AS SUBROUTINES
5322
5323 If the syntax for a recursive subpattern reference (either by number or
5324 by name) is used outside the parentheses to which it refers, it oper-
5325 ates like a subroutine in a programming language. The "called" subpat-
5326 tern may be defined before or after the reference. A numbered reference
5327 can be absolute or relative, as in these examples:
5328
5329 (...(absolute)...)...(?2)...
5330 (...(relative)...)...(?-1)...
5331 (...(?+1)...(relative)...
5332
5333 An earlier example pointed out that the pattern
5334
5335 (sens|respons)e and \1ibility
5336
5337 matches "sense and sensibility" and "response and responsibility", but
5338 not "sense and responsibility". If instead the pattern
5339
5340 (sens|respons)e and (?1)ibility
5341
5342 is used, it does match "sense and responsibility" as well as the other
5343 two strings. Another example is given in the discussion of DEFINE
5344 above.
5345
5346 Like recursive subpatterns, a subroutine call is always treated as an
5347 atomic group. That is, once it has matched some of the subject string,
5348 it is never re-entered, even if it contains untried alternatives and
5349 there is a subsequent matching failure. Any capturing parentheses that
5350 are set during the subroutine call revert to their previous values
5351 afterwards.
5352
5353 When a subpattern is used as a subroutine, processing options such as
5354 case-independence are fixed when the subpattern is defined. They cannot
5355 be changed for different calls. For example, consider this pattern:
5356
5357 (abc)(?i:(?-1))
5358
5359 It matches "abcabc". It does not match "abcABC" because the change of
5360 processing option does not affect the called subpattern.
5361
5362
5363 ONIGURUMA SUBROUTINE SYNTAX
5364
5365 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
5366 name or a number enclosed either in angle brackets or single quotes, is
5367 an alternative syntax for referencing a subpattern as a subroutine,
5368 possibly recursively. Here are two of the examples used above, rewrit-
5369 ten using this syntax:
5370
5371 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
5372 (sens|respons)e and \g'1'ibility
5373
5374 PCRE supports an extension to Oniguruma: if a number is preceded by a
5375 plus or a minus sign it is taken as a relative reference. For example:
5376
5377 (abc)(?i:\g<-1>)
5378
5379 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
5380 synonymous. The former is a back reference; the latter is a subroutine
5381 call.
5382
5383
5384 CALLOUTS
5385
5386 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
5387 Perl code to be obeyed in the middle of matching a regular expression.
5388 This makes it possible, amongst other things, to extract different sub-
5389 strings that match the same pair of parentheses when there is a repeti-
5390 tion.
5391
5392 PCRE provides a similar feature, but of course it cannot obey arbitrary
5393 Perl code. The feature is called "callout". The caller of PCRE provides
5394 an external function by putting its entry point in the global variable
5395 pcre_callout. By default, this variable contains NULL, which disables
5396 all calling out.
5397
5398 Within a regular expression, (?C) indicates the points at which the
5399 external function is to be called. If you want to identify different
5400 callout points, you can put a number less than 256 after the letter C.
5401 The default value is zero. For example, this pattern has two callout
5402 points:
5403
5404 (?C1)abc(?C2)def
5405
5406 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
5407 automatically installed before each item in the pattern. They are all
5408 numbered 255.
5409
5410 During matching, when PCRE reaches a callout point (and pcre_callout is
5411 set), the external function is called. It is provided with the number
5412 of the callout, the position in the pattern, and, optionally, one item
5413 of data originally supplied by the caller of pcre_exec(). The callout
5414 function may cause matching to proceed, to backtrack, or to fail alto-
5415 gether. A complete description of the interface to the callout function
5416 is given in the pcrecallout documentation.
5417
5418
5419 BACKTRACKING CONTROL
5420
5421 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
5422 which are described in the Perl documentation as "experimental and sub-
5423 ject to change or removal in a future version of Perl". It goes on to
5424 say: "Their usage in production code should be noted to avoid problems
5425 during upgrades." The same remarks apply to the PCRE features described
5426 in this section.
5427
5428 Since these verbs are specifically related to backtracking, most of
5429 them can be used only when the pattern is to be matched using
5430 pcre_exec(), which uses a backtracking algorithm. With the exception of
5431 (*FAIL), which behaves like a failing negative assertion, they cause an
5432 error if encountered by pcre_dfa_exec().
5433
5434 If any of these verbs are used in an assertion or subroutine subpattern
5435 (including recursive subpatterns), their effect is confined to that
5436 subpattern; it does not extend to the surrounding pattern. Note that
5437 such subpatterns are processed as anchored at the point where they are
5438 tested.
5439
5440 The new verbs make use of what was previously invalid syntax: an open-
5441 ing parenthesis followed by an asterisk. They are generally of the form
5442 (*VERB) or (*VERB:NAME). Some may take either form, with differing be-
5443 haviour, depending on whether or not an argument is present. An name is
5444 a sequence of letters, digits, and underscores. If the name is empty,
5445 that is, if the closing parenthesis immediately follows the colon, the
5446 effect is as if the colon were not there. Any number of these verbs may
5447 occur in a pattern.
5448
5449 PCRE contains some optimizations that are used to speed up matching by
5450 running some checks at the start of each match attempt. For example, it
5451 may know the minimum length of matching subject, or that a particular
5452 character must be present. When one of these optimizations suppresses
5453 the running of a match, any included backtracking verbs will not, of
5454 course, be processed. You can suppress the start-of-match optimizations
5455 by setting the PCRE_NO_START_OPTIMIZE option when calling pcre_com-
5456 pile() or pcre_exec(), or by starting the pattern with (*NO_START_OPT).
5457
5458 Verbs that act immediately
5459
5460 The following verbs act as soon as they are encountered. They may not
5461 be followed by a name.
5462
5463 (*ACCEPT)
5464
5465 This verb causes the match to end successfully, skipping the remainder
5466 of the pattern. When inside a recursion, only the innermost pattern is
5467 ended immediately. If (*ACCEPT) is inside capturing parentheses, the
5468 data so far is captured. (This feature was added to PCRE at release
5469 8.00.) For example:
5470
5471 A((?:A|B(*ACCEPT)|C)D)
5472
5473 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
5474 tured by the outer parentheses.
5475
5476 (*FAIL) or (*F)
5477
5478 This verb causes the match to fail, forcing backtracking to occur. It
5479 is equivalent to (?!) but easier to read. The Perl documentation notes
5480 that it is probably useful only when combined with (?{}) or (??{}).
5481 Those are, of course, Perl features that are not present in PCRE. The
5482 nearest equivalent is the callout feature, as for example in this pat-
5483 tern:
5484
5485 a+(?C)(*FAIL)
5486
5487 A match with the string "aaaa" always fails, but the callout is taken
5488 before each backtrack happens (in this example, 10 times).
5489
5490 Recording which path was taken
5491
5492 There is one verb whose main purpose is to track how a match was
5493 arrived at, though it also has a secondary use in conjunction with
5494 advancing the match starting point (see (*SKIP) below).
5495
5496 (*MARK:NAME) or (*:NAME)
5497
5498 A name is always required with this verb. There may be as many
5499 instances of (*MARK) as you like in a pattern, and their names do not
5500 have to be unique.
5501
5502 When a match succeeds, the name of the last-encountered (*MARK) is
5503 passed back to the caller via the pcre_extra data structure, as
5504 described in the section on pcre_extra in the pcreapi documentation. No
5505 data is returned for a partial match. Here is an example of pcretest
5506 output, where the /K modifier requests the retrieval and outputting of
5507 (*MARK) data:
5508
5509 /X(*MARK:A)Y|X(*MARK:B)Z/K
5510 XY
5511 0: XY
5512 MK: A
5513 XZ
5514 0: XZ
5515 MK: B
5516
5517 The (*MARK) name is tagged with "MK:" in this output, and in this exam-
5518 ple it indicates which of the two alternatives matched. This is a more
5519 efficient way of obtaining this information than putting each alterna-
5520 tive in its own capturing parentheses.
5521
5522 A name may also be returned after a failed match if the final path
5523 through the pattern involves (*MARK). However, unless (*MARK) used in
5524 conjunction with (*COMMIT), this is unlikely to happen for an unan-
5525 chored pattern because, as the starting point for matching is advanced,
5526 the final check is often with an empty string, causing a failure before
5527 (*MARK) is reached. For example:
5528
5529 /X(*MARK:A)Y|X(*MARK:B)Z/K
5530 XP
5531 No match
5532
5533 There are three potential starting points for this match (starting with
5534 X, starting with P, and with an empty string). If the pattern is
5535 anchored, the result is different:
5536
5537 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
5538 XP
5539 No match, mark = B
5540
5541 PCRE's start-of-match optimizations can also interfere with this. For
5542 example, if, as a result of a call to pcre_study(), it knows the mini-
5543 mum subject length for a match, a shorter subject will not be scanned
5544 at all.
5545
5546 Note that similar anomalies (though different in detail) exist in Perl,
5547 no doubt for the same reasons. The use of (*MARK) data after a failed
5548 match of an unanchored pattern is not recommended, unless (*COMMIT) is
5549 involved.
5550
5551 Verbs that act after backtracking
5552
5553 The following verbs do nothing when they are encountered. Matching con-
5554 tinues with what follows, but if there is no subsequent match, causing
5555 a backtrack to the verb, a failure is forced. That is, backtracking
5556 cannot pass to the left of the verb. However, when one of these verbs
5557 appears inside an atomic group, its effect is confined to that group,
5558 because once the group has been matched, there is never any backtrack-
5559 ing into it. In this situation, backtracking can "jump back" to the
5560 left of the entire atomic group. (Remember also, as stated above, that
5561 this localization also applies in subroutine calls and assertions.)
5562
5563 These verbs differ in exactly what kind of failure occurs when back-
5564 tracking reaches them.
5565
5566 (*COMMIT)
5567
5568 This verb, which may not be followed by a name, causes the whole match
5569 to fail outright if the rest of the pattern does not match. Even if the
5570 pattern is unanchored, no further attempts to find a match by advancing
5571 the starting point take place. Once (*COMMIT) has been passed,
5572 pcre_exec() is committed to finding a match at the current starting
5573 point, or not at all. For example:
5574
5575 a+(*COMMIT)b
5576
5577 This matches "xxaab" but not "aacaab". It can be thought of as a kind
5578 of dynamic anchor, or "I've started, so I must finish." The name of the
5579 most recently passed (*MARK) in the path is passed back when (*COMMIT)
5580 forces a match failure.
5581
5582 Note that (*COMMIT) at the start of a pattern is not the same as an
5583 anchor, unless PCRE's start-of-match optimizations are turned off, as
5584 shown in this pcretest example:
5585
5586 /(*COMMIT)abc/
5587 xyzabc
5588 0: abc
5589 xyzabc\Y
5590 No match
5591
5592 PCRE knows that any match must start with "a", so the optimization
5593 skips along the subject to "a" before running the first match attempt,
5594 which succeeds. When the optimization is disabled by the \Y escape in
5595 the second subject, the match starts at "x" and so the (*COMMIT) causes
5596 it to fail without trying any other starting points.
5597
5598 (*PRUNE) or (*PRUNE:NAME)
5599
5600 This verb causes the match to fail at the current starting position in
5601 the subject if the rest of the pattern does not match. If the pattern
5602 is unanchored, the normal "bumpalong" advance to the next starting
5603 character then happens. Backtracking can occur as usual to the left of
5604 (*PRUNE), before it is reached, or when matching to the right of
5605 (*PRUNE), but if there is no match to the right, backtracking cannot
5606 cross (*PRUNE). In simple cases, the use of (*PRUNE) is just an alter-
5607 native to an atomic group or possessive quantifier, but there are some
5608 uses of (*PRUNE) that cannot be expressed in any other way. The behav-
5609 iour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
5610 match fails completely; the name is passed back if this is the final
5611 attempt. (*PRUNE:NAME) does not pass back a name if the match suc-
5612 ceeds. In an anchored pattern (*PRUNE) has the same effect as (*COM-
5613 MIT).
5614
5615 (*SKIP)
5616
5617 This verb, when given without a name, is like (*PRUNE), except that if
5618 the pattern is unanchored, the "bumpalong" advance is not to the next
5619 character, but to the position in the subject where (*SKIP) was encoun-
5620 tered. (*SKIP) signifies that whatever text was matched leading up to
5621 it cannot be part of a successful match. Consider:
5622
5623 a+(*SKIP)b
5624
5625 If the subject is "aaaac...", after the first match attempt fails
5626 (starting at the first character in the string), the starting point
5627 skips on to start the next attempt at "c". Note that a possessive quan-
5628 tifer does not have the same effect as this example; although it would
5629 suppress backtracking during the first match attempt, the second
5630 attempt would start at the second character instead of skipping on to
5631 "c".
5632
5633 (*SKIP:NAME)
5634
5635 When (*SKIP) has an associated name, its behaviour is modified. If the
5636 following pattern fails to match, the previous path through the pattern
5637 is searched for the most recent (*MARK) that has the same name. If one
5638 is found, the "bumpalong" advance is to the subject position that cor-
5639 responds to that (*MARK) instead of to where (*SKIP) was encountered.
5640 If no (*MARK) with a matching name is found, normal "bumpalong" of one
5641 character happens (the (*SKIP) is ignored).
5642
5643 (*THEN) or (*THEN:NAME)
5644
5645 This verb causes a skip to the next alternation in the innermost
5646 enclosing group if the rest of the pattern does not match. That is, it
5647 cancels pending backtracking, but only within the current alternation.
5648 Its name comes from the observation that it can be used for a pattern-
5649 based if-then-else block:
5650
5651 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5652
5653 If the COND1 pattern matches, FOO is tried (and possibly further items
5654 after the end of the group if FOO succeeds); on failure the matcher
5655 skips to the second alternative and tries COND2, without backtracking
5656 into COND1. The behaviour of (*THEN:NAME) is exactly the same as
5657 (*MARK:NAME)(*THEN) if the overall match fails. If (*THEN) is not
5658 directly inside an alternation, it acts like (*PRUNE).
5659
5660 The above verbs provide four different "strengths" of control when sub-
5661 sequent matching fails. (*THEN) is the weakest, carrying on the match
5662 at the next alternation. (*PRUNE) comes next, failing the match at the
5663 current starting position, but allowing an advance to the next charac-
5664 ter (for an unanchored pattern). (*SKIP) is similar, except that the
5665 advance may be more than one character. (*COMMIT) is the strongest,
5666 causing the entire match to fail.
5667
5668 If more than one is present in a pattern, the "stongest" one wins. For
5669 example, consider this pattern, where A, B, etc. are complex pattern
5670 fragments:
5671
5672 (A(*COMMIT)B(*THEN)C|D)
5673
5674 Once A has matched, PCRE is committed to this match, at the current
5675 starting position. If subsequently B matches, but C does not, the nor-
5676 mal (*THEN) action of trying the next alternation (that is, D) does not
5677 happen because (*COMMIT) overrides.
5678
5679
5680 SEE ALSO
5681
5682 pcreapi(3), pcrecallout(3), pcrematching(3), pcresyntax(3), pcre(3).
5683
5684
5685 AUTHOR
5686
5687 Philip Hazel
5688 University Computing Service
5689 Cambridge CB2 3QH, England.
5690
5691
5692 REVISION
5693
5694 Last updated: 21 November 2010
5695 Copyright (c) 1997-2010 University of Cambridge.
5696 ------------------------------------------------------------------------------
5697
5698
5699 PCRESYNTAX(3) PCRESYNTAX(3)
5700
5701
5702 NAME
5703 PCRE - Perl-compatible regular expressions
5704
5705
5706 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5707
5708 The full syntax and semantics of the regular expressions that are sup-
5709 ported by PCRE are described in the pcrepattern documentation. This
5710 document contains just a quick-reference summary of the syntax.
5711
5712
5713 QUOTING
5714
5715 \x where x is non-alphanumeric is a literal x
5716 \Q...\E treat enclosed characters as literal
5717
5718
5719 CHARACTERS
5720
5721 \a alarm, that is, the BEL character (hex 07)
5722 \cx "control-x", where x is any ASCII character
5723 \e escape (hex 1B)
5724 \f formfeed (hex 0C)
5725 \n newline (hex 0A)
5726 \r carriage return (hex 0D)
5727 \t tab (hex 09)
5728 \ddd character with octal code ddd, or backreference
5729 \xhh character with hex code hh
5730 \x{hhh..} character with hex code hhh..
5731
5732
5733 CHARACTER TYPES
5734
5735 . any character except newline;
5736 in dotall mode, any character whatsoever
5737 \C one byte, even in UTF-8 mode (best avoided)
5738 \d a decimal digit
5739 \D a character that is not a decimal digit
5740 \h a horizontal whitespace character
5741 \H a character that is not a horizontal whitespace character
5742 \N a character that is not a newline
5743 \p{xx} a character with the xx property
5744 \P{xx} a character without the xx property
5745 \R a newline sequence
5746 \s a whitespace character
5747 \S a character that is not a whitespace character
5748 \v a vertical whitespace character
5749 \V a character that is not a vertical whitespace character
5750 \w a "word" character
5751 \W a "non-word" character
5752 \X an extended Unicode sequence
5753
5754 In PCRE, by default, \d, \D, \s, \S, \w, and \W recognize only ASCII
5755 characters, even in UTF-8 mode. However, this can be changed by setting
5756 the PCRE_UCP option.
5757
5758
5759 GENERAL CATEGORY PROPERTIES FOR \p and \P
5760
5761 C Other
5762 Cc Control
5763 Cf Format
5764 Cn Unassigned
5765 Co Private use
5766 Cs Surrogate
5767
5768 L Letter
5769 Ll Lower case letter
5770 Lm Modifier letter
5771 Lo Other letter
5772 Lt Title case letter
5773 Lu Upper case letter
5774 L& Ll, Lu, or Lt
5775
5776 M Mark
5777 Mc Spacing mark
5778 Me Enclosing mark
5779 Mn Non-spacing mark
5780
5781 N Number
5782 Nd Decimal number
5783 Nl Letter number
5784 No Other number
5785
5786 P Punctuation
5787 Pc Connector punctuation
5788 Pd Dash punctuation
5789 Pe Close punctuation
5790 Pf Final punctuation
5791 Pi Initial punctuation
5792 Po Other punctuation
5793 Ps Open punctuation
5794
5795 S Symbol
5796 Sc Currency symbol
5797 Sk Modifier symbol
5798 Sm Mathematical symbol
5799 So Other symbol
5800
5801 Z Separator
5802 Zl Line separator
5803 Zp Paragraph separator
5804 Zs Space separator
5805
5806
5807 PCRE SPECIAL CATEGORY PROPERTIES FOR \p and \P
5808
5809 Xan Alphanumeric: union of properties L and N
5810 Xps POSIX space: property Z or tab, NL, VT, FF, CR
5811 Xsp Perl space: property Z or tab, NL, FF, CR
5812 Xwd Perl word: property Xan or underscore
5813
5814
5815 SCRIPT NAMES FOR \p AND \P
5816
5817 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
5818 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
5819 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
5820 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
5821 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
5822 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
5823 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
5824 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
5825 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
5826 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
5827 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
5828 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
5829 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5830 Ugaritic, Vai, Yi.
5831
5832
5833 CHARACTER CLASSES
5834
5835 [...] positive character class
5836 [^...] negative character class
5837 [x-y] range (can be used for hex characters)
5838 [[:xxx:]] positive POSIX named set
5839 [[:^xxx:]] negative POSIX named set
5840
5841 alnum alphanumeric
5842 alpha alphabetic
5843 ascii 0-127
5844 blank space or tab
5845 cntrl control character
5846 digit decimal digit
5847 graph printing, excluding space
5848 lower lower case letter
5849 print printing, including space
5850 punct printing, excluding alphanumeric
5851 space whitespace
5852 upper upper case letter
5853 word same as \w
5854 xdigit hexadecimal digit
5855
5856 In PCRE, POSIX character set names recognize only ASCII characters by
5857 default, but some of them use Unicode properties if PCRE_UCP is set.
5858 You can use \Q...\E inside a character class.
5859
5860
5861 QUANTIFIERS
5862
5863 ? 0 or 1, greedy
5864 ?+ 0 or 1, possessive
5865 ?? 0 or 1, lazy
5866 * 0 or more, greedy
5867 *+ 0 or more, possessive
5868 *? 0 or more, lazy
5869 + 1 or more, greedy
5870 ++ 1 or more, possessive
5871 +? 1 or more, lazy
5872 {n} exactly n
5873 {n,m} at least n, no more than m, greedy
5874 {n,m}+ at least n, no more than m, possessive
5875 {n,m}? at least n, no more than m, lazy
5876 {n,} n or more, greedy
5877 {n,}+ n or more, possessive
5878 {n,}? n or more, lazy
5879
5880
5881 ANCHORS AND SIMPLE ASSERTIONS
5882
5883 \b word boundary
5884 \B not a word boundary
5885 ^ start of subject
5886 also after internal newline in multiline mode
5887 \A start of subject
5888 $ end of subject
5889 also before newline at end of subject
5890 also before internal newline in multiline mode
5891 \Z end of subject
5892 also before newline at end of subject
5893 \z end of subject
5894 \G first matching position in subject
5895
5896
5897 MATCH POINT RESET
5898
5899 \K reset start of match
5900
5901
5902 ALTERNATION
5903
5904 expr|expr|expr...
5905
5906
5907 CAPTURING
5908
5909 (...) capturing group
5910 (?<name>...) named capturing group (Perl)
5911 (?'name'...) named capturing group (Perl)
5912 (?P<name>...) named capturing group (Python)
5913 (?:...) non-capturing group
5914 (?|...) non-capturing group; reset group numbers for
5915 capturing groups in each alternative
5916
5917
5918 ATOMIC GROUPS
5919
5920 (?>...) atomic, non-capturing group
5921
5922
5923 COMMENT
5924
5925 (?#....) comment (not nestable)
5926
5927
5928 OPTION SETTING
5929
5930 (?i) caseless
5931 (?J) allow duplicate names
5932 (?m) multiline
5933 (?s) single line (dotall)
5934 (?U) default ungreedy (lazy)
5935 (?x) extended (ignore white space)
5936 (?-...) unset option(s)
5937
5938 The following are recognized only at the start of a pattern or after
5939 one of the newline-setting options with similar syntax:
5940
5941 (*NO_START_OPT) no start-match optimization (PCRE_NO_START_OPTIMIZE)
5942 (*UTF8) set UTF-8 mode (PCRE_UTF8)
5943 (*UCP) set PCRE_UCP (use Unicode properties for \d etc)
5944
5945
5946 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5947
5948 (?=...) positive look ahead
5949 (?!...) negative look ahead
5950 (?<=...) positive look behind
5951 (?<!...) negative look behind
5952
5953 Each top-level branch of a look behind must be of a fixed length.
5954
5955
5956 BACKREFERENCES
5957
5958 \n reference by number (can be ambiguous)
5959 \gn reference by number
5960 \g{n} reference by number
5961 \g{-n} relative reference by number
5962 \k<name> reference by name (Perl)
5963 \k'name' reference by name (Perl)
5964 \g{name} reference by name (Perl)
5965 \k{name} reference by name (.NET)
5966 (?P=name) reference by name (Python)
5967
5968
5969 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5970
5971 (?R) recurse whole pattern
5972 (?n) call subpattern by absolute number
5973 (?+n) call subpattern by relative number
5974 (?-n) call subpattern by relative number
5975 (?&name) call subpattern by name (Perl)
5976 (?P>name) call subpattern by name (Python)
5977 \g<name> call subpattern by name (Oniguruma)
5978 \g'name' call subpattern by name (Oniguruma)
5979 \g<n> call subpattern by absolute number (Oniguruma)
5980 \g'n' call subpattern by absolute number (Oniguruma)
5981 \g<+n> call subpattern by relative number (PCRE extension)
5982 \g'+n' call subpattern by relative number (PCRE extension)
5983 \g<-n> call subpattern by relative number (PCRE extension)
5984 \g'-n' call subpattern by relative number (PCRE extension)
5985
5986
5987 CONDITIONAL PATTERNS
5988
5989 (?(condition)yes-pattern)
5990 (?(condition)yes-pattern|no-pattern)
5991
5992 (?(n)... absolute reference condition
5993 (?(+n)... relative reference condition
5994 (?(-n)... relative reference condition
5995 (?(<name>)... named reference condition (Perl)
5996 (?('name')... named reference condition (Perl)
5997 (?(name)... named reference condition (PCRE)
5998 (?(R)... overall recursion condition
5999 (?(Rn)... specific group recursion condition
6000 (?(R&name)...