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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 392 - (show annotations)
Tue Mar 17 21:30:30 2009 UTC (10 years, 6 months ago) by ph10
File MIME type: text/plain
File size: 312107 byte(s)
Error occurred while calculating annotation data.
Update after detrailing for a test release.
1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax. There is also some support for certain .NET and Oniguruma syn-
24 tax items, and there is an option for requesting some minor changes
25 that give better JavaScript compatibility.
26
27 The current implementation of PCRE (release 7.x) corresponds approxi-
28 mately with Perl 5.10, including support for UTF-8 encoded strings and
29 Unicode general category properties. However, UTF-8 and Unicode support
30 has to be explicitly enabled; it is not the default. The Unicode tables
31 correspond to Unicode release 5.0.0.
32
33 In addition to the Perl-compatible matching function, PCRE contains an
34 alternative matching function that matches the same compiled patterns
35 in a different way. In certain circumstances, the alternative function
36 has some advantages. For a discussion of the two matching algorithms,
37 see the pcrematching page.
38
39 PCRE is written in C and released as a C library. A number of people
40 have written wrappers and interfaces of various kinds. In particular,
41 Google Inc. have provided a comprehensive C++ wrapper. This is now
42 included as part of the PCRE distribution. The pcrecpp page has details
43 of this interface. Other people's contributions can be found in the
44 Contrib directory at the primary FTP site, which is:
45
46 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
47
48 Details of exactly which Perl regular expression features are and are
49 not supported by PCRE are given in separate documents. See the pcrepat-
50 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
51 page.
52
53 Some features of PCRE can be included, excluded, or changed when the
54 library is built. The pcre_config() function makes it possible for a
55 client to discover which features are available. The features them-
56 selves are described in the pcrebuild page. Documentation about build-
57 ing PCRE for various operating systems can be found in the README file
58 in the source distribution.
59
60 The library contains a number of undocumented internal functions and
61 data tables that are used by more than one of the exported external
62 functions, but which are not intended for use by external callers.
63 Their names all begin with "_pcre_", which hopefully will not provoke
64 any name clashes. In some environments, it is possible to control which
65 external symbols are exported when a shared library is built, and in
66 these cases the undocumented symbols are not exported.
67
68
69 USER DOCUMENTATION
70
71 The user documentation for PCRE comprises a number of different sec-
72 tions. In the "man" format, each of these is a separate "man page". In
73 the HTML format, each is a separate page, linked from the index page.
74 In the plain text format, all the sections are concatenated, for ease
75 of searching. The sections are as follows:
76
77 pcre this document
78 pcre-config show PCRE installation configuration information
79 pcreapi details of PCRE's native C API
80 pcrebuild options for building PCRE
81 pcrecallout details of the callout feature
82 pcrecompat discussion of Perl compatibility
83 pcrecpp details of the C++ wrapper
84 pcregrep description of the pcregrep command
85 pcrematching discussion of the two matching algorithms
86 pcrepartial details of the partial matching facility
87 pcrepattern syntax and semantics of supported
88 regular expressions
89 pcresyntax quick syntax reference
90 pcreperform discussion of performance issues
91 pcreposix the POSIX-compatible C API
92 pcreprecompile details of saving and re-using precompiled patterns
93 pcresample discussion of the sample program
94 pcrestack discussion of stack usage
95 pcretest description of the pcretest testing command
96
97 In addition, in the "man" and HTML formats, there is a short page for
98 each C library function, listing its arguments and results.
99
100
101 LIMITATIONS
102
103 There are some size limitations in PCRE but it is hoped that they will
104 never in practice be relevant.
105
106 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
107 is compiled with the default internal linkage size of 2. If you want to
108 process regular expressions that are truly enormous, you can compile
109 PCRE with an internal linkage size of 3 or 4 (see the README file in
110 the source distribution and the pcrebuild documentation for details).
111 In these cases the limit is substantially larger. However, the speed
112 of execution is slower.
113
114 All values in repeating quantifiers must be less than 65536.
115
116 There is no limit to the number of parenthesized subpatterns, but there
117 can be no more than 65535 capturing subpatterns.
118
119 The maximum length of name for a named subpattern is 32 characters, and
120 the maximum number of named subpatterns is 10000.
121
122 The maximum length of a subject string is the largest positive number
123 that an integer variable can hold. However, when using the traditional
124 matching function, PCRE uses recursion to handle subpatterns and indef-
125 inite repetition. This means that the available stack space may limit
126 the size of a subject string that can be processed by certain patterns.
127 For a discussion of stack issues, see the pcrestack documentation.
128
129
130 UTF-8 AND UNICODE PROPERTY SUPPORT
131
132 From release 3.3, PCRE has had some support for character strings
133 encoded in the UTF-8 format. For release 4.0 this was greatly extended
134 to cover most common requirements, and in release 5.0 additional sup-
135 port for Unicode general category properties was added.
136
137 In order process UTF-8 strings, you must build PCRE to include UTF-8
138 support in the code, and, in addition, you must call pcre_compile()
139 with the PCRE_UTF8 option flag. When you do this, both the pattern and
140 any subject strings that are matched against it are treated as UTF-8
141 strings instead of just strings of bytes.
142
143 If you compile PCRE with UTF-8 support, but do not use it at run time,
144 the library will be a bit bigger, but the additional run time overhead
145 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
146 very big.
147
148 If PCRE is built with Unicode character property support (which implies
149 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
150 ported. The available properties that can be tested are limited to the
151 general category properties such as Lu for an upper case letter or Nd
152 for a decimal number, the Unicode script names such as Arabic or Han,
153 and the derived properties Any and L&. A full list is given in the
154 pcrepattern documentation. Only the short names for properties are sup-
155 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
156 ter}, is not supported. Furthermore, in Perl, many properties may
157 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
158 does not support this.
159
160 Validity of UTF-8 strings
161
162 When you set the PCRE_UTF8 flag, the strings passed as patterns and
163 subjects are (by default) checked for validity on entry to the relevant
164 functions. From release 7.3 of PCRE, the check is according the rules
165 of RFC 3629, which are themselves derived from the Unicode specifica-
166 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
167 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
168 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
169 to U+DFFF.
170
171 The excluded code points are the "Low Surrogate Area" of Unicode, of
172 which the Unicode Standard says this: "The Low Surrogate Area does not
173 contain any character assignments, consequently no character code
174 charts or namelists are provided for this area. Surrogates are reserved
175 for use with UTF-16 and then must be used in pairs." The code points
176 that are encoded by UTF-16 pairs are available as independent code
177 points in the UTF-8 encoding. (In other words, the whole surrogate
178 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
179
180 If an invalid UTF-8 string is passed to PCRE, an error return
181 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
182 that your strings are valid, and therefore want to skip these checks in
183 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
184 compile time or at run time, PCRE assumes that the pattern or subject
185 it is given (respectively) contains only valid UTF-8 codes. In this
186 case, it does not diagnose an invalid UTF-8 string.
187
188 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
189 what happens depends on why the string is invalid. If the string con-
190 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
191 string of characters in the range 0 to 0x7FFFFFFF. In other words,
192 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
193 strings according to the more liberal rules of RFC 2279. However, if
194 the string does not even conform to RFC 2279, the result is undefined.
195 Your program may crash.
196
197 If you want to process strings of values in the full range 0 to
198 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
199 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
200 this situation, you will have to apply your own validity check.
201
202 General comments about UTF-8 mode
203
204 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
205 two-byte UTF-8 character if the value is greater than 127.
206
207 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
208 characters for values greater than \177.
209
210 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
211 vidual bytes, for example: \x{100}{3}.
212
213 4. The dot metacharacter matches one UTF-8 character instead of a sin-
214 gle byte.
215
216 5. The escape sequence \C can be used to match a single byte in UTF-8
217 mode, but its use can lead to some strange effects. This facility is
218 not available in the alternative matching function, pcre_dfa_exec().
219
220 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
221 test characters of any code value, but the characters that PCRE recog-
222 nizes as digits, spaces, or word characters remain the same set as
223 before, all with values less than 256. This remains true even when PCRE
224 includes Unicode property support, because to do otherwise would slow
225 down PCRE in many common cases. If you really want to test for a wider
226 sense of, say, "digit", you must use Unicode property tests such as
227 \p{Nd}.
228
229 7. Similarly, characters that match the POSIX named character classes
230 are all low-valued characters.
231
232 8. However, the Perl 5.10 horizontal and vertical whitespace matching
233 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
234 acters.
235
236 9. Case-insensitive matching applies only to characters whose values
237 are less than 128, unless PCRE is built with Unicode property support.
238 Even when Unicode property support is available, PCRE still uses its
239 own character tables when checking the case of low-valued characters,
240 so as not to degrade performance. The Unicode property information is
241 used only for characters with higher values. Even when Unicode property
242 support is available, PCRE supports case-insensitive matching only when
243 there is a one-to-one mapping between a letter's cases. There are a
244 small number of many-to-one mappings in Unicode; these are not sup-
245 ported by PCRE.
246
247
248 AUTHOR
249
250 Philip Hazel
251 University Computing Service
252 Cambridge CB2 3QH, England.
253
254 Putting an actual email address here seems to have been a spam magnet,
255 so I've taken it away. If you want to email me, use my two initials,
256 followed by the two digits 10, at the domain cam.ac.uk.
257
258
259 REVISION
260
261 Last updated: 12 April 2008
262 Copyright (c) 1997-2008 University of Cambridge.
263 ------------------------------------------------------------------------------
264
265
266 PCREBUILD(3) PCREBUILD(3)
267
268
269 NAME
270 PCRE - Perl-compatible regular expressions
271
272
273 PCRE BUILD-TIME OPTIONS
274
275 This document describes the optional features of PCRE that can be
276 selected when the library is compiled. It assumes use of the configure
277 script, where the optional features are selected or deselected by pro-
278 viding options to configure before running the make command. However,
279 the same options can be selected in both Unix-like and non-Unix-like
280 environments using the GUI facility of CMakeSetup if you are using
281 CMake instead of configure to build PCRE.
282
283 The complete list of options for configure (which includes the standard
284 ones such as the selection of the installation directory) can be
285 obtained by running
286
287 ./configure --help
288
289 The following sections include descriptions of options whose names
290 begin with --enable or --disable. These settings specify changes to the
291 defaults for the configure command. Because of the way that configure
292 works, --enable and --disable always come in pairs, so the complemen-
293 tary option always exists as well, but as it specifies the default, it
294 is not described.
295
296
297 C++ SUPPORT
298
299 By default, the configure script will search for a C++ compiler and C++
300 header files. If it finds them, it automatically builds the C++ wrapper
301 library for PCRE. You can disable this by adding
302
303 --disable-cpp
304
305 to the configure command.
306
307
308 UTF-8 SUPPORT
309
310 To build PCRE with support for UTF-8 Unicode character strings, add
311
312 --enable-utf8
313
314 to the configure command. Of itself, this does not make PCRE treat
315 strings as UTF-8. As well as compiling PCRE with this option, you also
316 have have to set the PCRE_UTF8 option when you call the pcre_compile()
317 function.
318
319 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
320 expects its input to be either ASCII or UTF-8 (depending on the runtime
321 option). It is not possible to support both EBCDIC and UTF-8 codes in
322 the same version of the library. Consequently, --enable-utf8 and
323 --enable-ebcdic are mutually exclusive.
324
325
326 UNICODE CHARACTER PROPERTY SUPPORT
327
328 UTF-8 support allows PCRE to process character values greater than 255
329 in the strings that it handles. On its own, however, it does not pro-
330 vide any facilities for accessing the properties of such characters. If
331 you want to be able to use the pattern escapes \P, \p, and \X, which
332 refer to Unicode character properties, you must add
333
334 --enable-unicode-properties
335
336 to the configure command. This implies UTF-8 support, even if you have
337 not explicitly requested it.
338
339 Including Unicode property support adds around 30K of tables to the
340 PCRE library. Only the general category properties such as Lu and Nd
341 are supported. Details are given in the pcrepattern documentation.
342
343
344 CODE VALUE OF NEWLINE
345
346 By default, PCRE interprets the linefeed (LF) character as indicating
347 the end of a line. This is the normal newline character on Unix-like
348 systems. You can compile PCRE to use carriage return (CR) instead, by
349 adding
350
351 --enable-newline-is-cr
352
353 to the configure command. There is also a --enable-newline-is-lf
354 option, which explicitly specifies linefeed as the newline character.
355
356 Alternatively, you can specify that line endings are to be indicated by
357 the two character sequence CRLF. If you want this, add
358
359 --enable-newline-is-crlf
360
361 to the configure command. There is a fourth option, specified by
362
363 --enable-newline-is-anycrlf
364
365 which causes PCRE to recognize any of the three sequences CR, LF, or
366 CRLF as indicating a line ending. Finally, a fifth option, specified by
367
368 --enable-newline-is-any
369
370 causes PCRE to recognize any Unicode newline sequence.
371
372 Whatever line ending convention is selected when PCRE is built can be
373 overridden when the library functions are called. At build time it is
374 conventional to use the standard for your operating system.
375
376
377 WHAT \R MATCHES
378
379 By default, the sequence \R in a pattern matches any Unicode newline
380 sequence, whatever has been selected as the line ending sequence. If
381 you specify
382
383 --enable-bsr-anycrlf
384
385 the default is changed so that \R matches only CR, LF, or CRLF. What-
386 ever is selected when PCRE is built can be overridden when the library
387 functions are called.
388
389
390 BUILDING SHARED AND STATIC LIBRARIES
391
392 The PCRE building process uses libtool to build both shared and static
393 Unix libraries by default. You can suppress one of these by adding one
394 of
395
396 --disable-shared
397 --disable-static
398
399 to the configure command, as required.
400
401
402 POSIX MALLOC USAGE
403
404 When PCRE is called through the POSIX interface (see the pcreposix doc-
405 umentation), additional working storage is required for holding the
406 pointers to capturing substrings, because PCRE requires three integers
407 per substring, whereas the POSIX interface provides only two. If the
408 number of expected substrings is small, the wrapper function uses space
409 on the stack, because this is faster than using malloc() for each call.
410 The default threshold above which the stack is no longer used is 10; it
411 can be changed by adding a setting such as
412
413 --with-posix-malloc-threshold=20
414
415 to the configure command.
416
417
418 HANDLING VERY LARGE PATTERNS
419
420 Within a compiled pattern, offset values are used to point from one
421 part to another (for example, from an opening parenthesis to an alter-
422 nation metacharacter). By default, two-byte values are used for these
423 offsets, leading to a maximum size for a compiled pattern of around
424 64K. This is sufficient to handle all but the most gigantic patterns.
425 Nevertheless, some people do want to process enormous patterns, so it
426 is possible to compile PCRE to use three-byte or four-byte offsets by
427 adding a setting such as
428
429 --with-link-size=3
430
431 to the configure command. The value given must be 2, 3, or 4. Using
432 longer offsets slows down the operation of PCRE because it has to load
433 additional bytes when handling them.
434
435
436 AVOIDING EXCESSIVE STACK USAGE
437
438 When matching with the pcre_exec() function, PCRE implements backtrack-
439 ing by making recursive calls to an internal function called match().
440 In environments where the size of the stack is limited, this can se-
441 verely limit PCRE's operation. (The Unix environment does not usually
442 suffer from this problem, but it may sometimes be necessary to increase
443 the maximum stack size. There is a discussion in the pcrestack docu-
444 mentation.) An alternative approach to recursion that uses memory from
445 the heap to remember data, instead of using recursive function calls,
446 has been implemented to work round the problem of limited stack size.
447 If you want to build a version of PCRE that works this way, add
448
449 --disable-stack-for-recursion
450
451 to the configure command. With this configuration, PCRE will use the
452 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
453 ment functions. By default these point to malloc() and free(), but you
454 can replace the pointers so that your own functions are used.
455
456 Separate functions are provided rather than using pcre_malloc and
457 pcre_free because the usage is very predictable: the block sizes
458 requested are always the same, and the blocks are always freed in
459 reverse order. A calling program might be able to implement optimized
460 functions that perform better than malloc() and free(). PCRE runs
461 noticeably more slowly when built in this way. This option affects only
462 the pcre_exec() function; it is not relevant for the the
463 pcre_dfa_exec() function.
464
465
466 LIMITING PCRE RESOURCE USAGE
467
468 Internally, PCRE has a function called match(), which it calls repeat-
469 edly (sometimes recursively) when matching a pattern with the
470 pcre_exec() function. By controlling the maximum number of times this
471 function may be called during a single matching operation, a limit can
472 be placed on the resources used by a single call to pcre_exec(). The
473 limit can be changed at run time, as described in the pcreapi documen-
474 tation. The default is 10 million, but this can be changed by adding a
475 setting such as
476
477 --with-match-limit=500000
478
479 to the configure command. This setting has no effect on the
480 pcre_dfa_exec() matching function.
481
482 In some environments it is desirable to limit the depth of recursive
483 calls of match() more strictly than the total number of calls, in order
484 to restrict the maximum amount of stack (or heap, if --disable-stack-
485 for-recursion is specified) that is used. A second limit controls this;
486 it defaults to the value that is set for --with-match-limit, which
487 imposes no additional constraints. However, you can set a lower limit
488 by adding, for example,
489
490 --with-match-limit-recursion=10000
491
492 to the configure command. This value can also be overridden at run
493 time.
494
495
496 CREATING CHARACTER TABLES AT BUILD TIME
497
498 PCRE uses fixed tables for processing characters whose code values are
499 less than 256. By default, PCRE is built with a set of tables that are
500 distributed in the file pcre_chartables.c.dist. These tables are for
501 ASCII codes only. If you add
502
503 --enable-rebuild-chartables
504
505 to the configure command, the distributed tables are no longer used.
506 Instead, a program called dftables is compiled and run. This outputs
507 the source for new set of tables, created in the default locale of your
508 C runtime system. (This method of replacing the tables does not work if
509 you are cross compiling, because dftables is run on the local host. If
510 you need to create alternative tables when cross compiling, you will
511 have to do so "by hand".)
512
513
514 USING EBCDIC CODE
515
516 PCRE assumes by default that it will run in an environment where the
517 character code is ASCII (or Unicode, which is a superset of ASCII).
518 This is the case for most computer operating systems. PCRE can, how-
519 ever, be compiled to run in an EBCDIC environment by adding
520
521 --enable-ebcdic
522
523 to the configure command. This setting implies --enable-rebuild-charta-
524 bles. You should only use it if you know that you are in an EBCDIC
525 environment (for example, an IBM mainframe operating system). The
526 --enable-ebcdic option is incompatible with --enable-utf8.
527
528
529 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
530
531 By default, pcregrep reads all files as plain text. You can build it so
532 that it recognizes files whose names end in .gz or .bz2, and reads them
533 with libz or libbz2, respectively, by adding one or both of
534
535 --enable-pcregrep-libz
536 --enable-pcregrep-libbz2
537
538 to the configure command. These options naturally require that the rel-
539 evant libraries are installed on your system. Configuration will fail
540 if they are not.
541
542
543 PCRETEST OPTION FOR LIBREADLINE SUPPORT
544
545 If you add
546
547 --enable-pcretest-libreadline
548
549 to the configure command, pcretest is linked with the libreadline
550 library, and when its input is from a terminal, it reads it using the
551 readline() function. This provides line-editing and history facilities.
552 Note that libreadline is GPL-licenced, so if you distribute a binary of
553 pcretest linked in this way, there may be licensing issues.
554
555 Setting this option causes the -lreadline option to be added to the
556 pcretest build. In many operating environments with a sytem-installed
557 libreadline this is sufficient. However, in some environments (e.g. if
558 an unmodified distribution version of readline is in use), some extra
559 configuration may be necessary. The INSTALL file for libreadline says
560 this:
561
562 "Readline uses the termcap functions, but does not link with the
563 termcap or curses library itself, allowing applications which link
564 with readline the to choose an appropriate library."
565
566 If your environment has not been set up so that an appropriate library
567 is automatically included, you may need to add something like
568
569 LIBS="-ncurses"
570
571 immediately before the configure command.
572
573
574 SEE ALSO
575
576 pcreapi(3), pcre_config(3).
577
578
579 AUTHOR
580
581 Philip Hazel
582 University Computing Service
583 Cambridge CB2 3QH, England.
584
585
586 REVISION
587
588 Last updated: 17 March 2009
589 Copyright (c) 1997-2009 University of Cambridge.
590 ------------------------------------------------------------------------------
591
592
593 PCREMATCHING(3) PCREMATCHING(3)
594
595
596 NAME
597 PCRE - Perl-compatible regular expressions
598
599
600 PCRE MATCHING ALGORITHMS
601
602 This document describes the two different algorithms that are available
603 in PCRE for matching a compiled regular expression against a given sub-
604 ject string. The "standard" algorithm is the one provided by the
605 pcre_exec() function. This works in the same was as Perl's matching
606 function, and provides a Perl-compatible matching operation.
607
608 An alternative algorithm is provided by the pcre_dfa_exec() function;
609 this operates in a different way, and is not Perl-compatible. It has
610 advantages and disadvantages compared with the standard algorithm, and
611 these are described below.
612
613 When there is only one possible way in which a given subject string can
614 match a pattern, the two algorithms give the same answer. A difference
615 arises, however, when there are multiple possibilities. For example, if
616 the pattern
617
618 ^<.*>
619
620 is matched against the string
621
622 <something> <something else> <something further>
623
624 there are three possible answers. The standard algorithm finds only one
625 of them, whereas the alternative algorithm finds all three.
626
627
628 REGULAR EXPRESSIONS AS TREES
629
630 The set of strings that are matched by a regular expression can be rep-
631 resented as a tree structure. An unlimited repetition in the pattern
632 makes the tree of infinite size, but it is still a tree. Matching the
633 pattern to a given subject string (from a given starting point) can be
634 thought of as a search of the tree. There are two ways to search a
635 tree: depth-first and breadth-first, and these correspond to the two
636 matching algorithms provided by PCRE.
637
638
639 THE STANDARD MATCHING ALGORITHM
640
641 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
642 sions", the standard algorithm is an "NFA algorithm". It conducts a
643 depth-first search of the pattern tree. That is, it proceeds along a
644 single path through the tree, checking that the subject matches what is
645 required. When there is a mismatch, the algorithm tries any alterna-
646 tives at the current point, and if they all fail, it backs up to the
647 previous branch point in the tree, and tries the next alternative
648 branch at that level. This often involves backing up (moving to the
649 left) in the subject string as well. The order in which repetition
650 branches are tried is controlled by the greedy or ungreedy nature of
651 the quantifier.
652
653 If a leaf node is reached, a matching string has been found, and at
654 that point the algorithm stops. Thus, if there is more than one possi-
655 ble match, this algorithm returns the first one that it finds. Whether
656 this is the shortest, the longest, or some intermediate length depends
657 on the way the greedy and ungreedy repetition quantifiers are specified
658 in the pattern.
659
660 Because it ends up with a single path through the tree, it is rela-
661 tively straightforward for this algorithm to keep track of the sub-
662 strings that are matched by portions of the pattern in parentheses.
663 This provides support for capturing parentheses and back references.
664
665
666 THE ALTERNATIVE MATCHING ALGORITHM
667
668 This algorithm conducts a breadth-first search of the tree. Starting
669 from the first matching point in the subject, it scans the subject
670 string from left to right, once, character by character, and as it does
671 this, it remembers all the paths through the tree that represent valid
672 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
673 though it is not implemented as a traditional finite state machine (it
674 keeps multiple states active simultaneously).
675
676 The scan continues until either the end of the subject is reached, or
677 there are no more unterminated paths. At this point, terminated paths
678 represent the different matching possibilities (if there are none, the
679 match has failed). Thus, if there is more than one possible match,
680 this algorithm finds all of them, and in particular, it finds the long-
681 est. In PCRE, there is an option to stop the algorithm after the first
682 match (which is necessarily the shortest) has been found.
683
684 Note that all the matches that are found start at the same point in the
685 subject. If the pattern
686
687 cat(er(pillar)?)
688
689 is matched against the string "the caterpillar catchment", the result
690 will be the three strings "cat", "cater", and "caterpillar" that start
691 at the fourth character of the subject. The algorithm does not automat-
692 ically move on to find matches that start at later positions.
693
694 There are a number of features of PCRE regular expressions that are not
695 supported by the alternative matching algorithm. They are as follows:
696
697 1. Because the algorithm finds all possible matches, the greedy or
698 ungreedy nature of repetition quantifiers is not relevant. Greedy and
699 ungreedy quantifiers are treated in exactly the same way. However, pos-
700 sessive quantifiers can make a difference when what follows could also
701 match what is quantified, for example in a pattern like this:
702
703 ^a++\w!
704
705 This pattern matches "aaab!" but not "aaa!", which would be matched by
706 a non-possessive quantifier. Similarly, if an atomic group is present,
707 it is matched as if it were a standalone pattern at the current point,
708 and the longest match is then "locked in" for the rest of the overall
709 pattern.
710
711 2. When dealing with multiple paths through the tree simultaneously, it
712 is not straightforward to keep track of captured substrings for the
713 different matching possibilities, and PCRE's implementation of this
714 algorithm does not attempt to do this. This means that no captured sub-
715 strings are available.
716
717 3. Because no substrings are captured, back references within the pat-
718 tern are not supported, and cause errors if encountered.
719
720 4. For the same reason, conditional expressions that use a backrefer-
721 ence as the condition or test for a specific group recursion are not
722 supported.
723
724 5. Because many paths through the tree may be active, the \K escape
725 sequence, which resets the start of the match when encountered (but may
726 be on some paths and not on others), is not supported. It causes an
727 error if encountered.
728
729 6. Callouts are supported, but the value of the capture_top field is
730 always 1, and the value of the capture_last field is always -1.
731
732 7. The \C escape sequence, which (in the standard algorithm) matches a
733 single byte, even in UTF-8 mode, is not supported because the alterna-
734 tive algorithm moves through the subject string one character at a
735 time, for all active paths through the tree.
736
737 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
738 are not supported. (*FAIL) is supported, and behaves like a failing
739 negative assertion.
740
741
742 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
743
744 Using the alternative matching algorithm provides the following advan-
745 tages:
746
747 1. All possible matches (at a single point in the subject) are automat-
748 ically found, and in particular, the longest match is found. To find
749 more than one match using the standard algorithm, you have to do kludgy
750 things with callouts.
751
752 2. There is much better support for partial matching. The restrictions
753 on the content of the pattern that apply when using the standard algo-
754 rithm for partial matching do not apply to the alternative algorithm.
755 For non-anchored patterns, the starting position of a partial match is
756 available.
757
758 3. Because the alternative algorithm scans the subject string just
759 once, and never needs to backtrack, it is possible to pass very long
760 subject strings to the matching function in several pieces, checking
761 for partial matching each time.
762
763
764 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
765
766 The alternative algorithm suffers from a number of disadvantages:
767
768 1. It is substantially slower than the standard algorithm. This is
769 partly because it has to search for all possible matches, but is also
770 because it is less susceptible to optimization.
771
772 2. Capturing parentheses and back references are not supported.
773
774 3. Although atomic groups are supported, their use does not provide the
775 performance advantage that it does for the standard algorithm.
776
777
778 AUTHOR
779
780 Philip Hazel
781 University Computing Service
782 Cambridge CB2 3QH, England.
783
784
785 REVISION
786
787 Last updated: 19 April 2008
788 Copyright (c) 1997-2008 University of Cambridge.
789 ------------------------------------------------------------------------------
790
791
792 PCREAPI(3) PCREAPI(3)
793
794
795 NAME
796 PCRE - Perl-compatible regular expressions
797
798
799 PCRE NATIVE API
800
801 #include <pcre.h>
802
803 pcre *pcre_compile(const char *pattern, int options,
804 const char **errptr, int *erroffset,
805 const unsigned char *tableptr);
806
807 pcre *pcre_compile2(const char *pattern, int options,
808 int *errorcodeptr,
809 const char **errptr, int *erroffset,
810 const unsigned char *tableptr);
811
812 pcre_extra *pcre_study(const pcre *code, int options,
813 const char **errptr);
814
815 int pcre_exec(const pcre *code, const pcre_extra *extra,
816 const char *subject, int length, int startoffset,
817 int options, int *ovector, int ovecsize);
818
819 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
820 const char *subject, int length, int startoffset,
821 int options, int *ovector, int ovecsize,
822 int *workspace, int wscount);
823
824 int pcre_copy_named_substring(const pcre *code,
825 const char *subject, int *ovector,
826 int stringcount, const char *stringname,
827 char *buffer, int buffersize);
828
829 int pcre_copy_substring(const char *subject, int *ovector,
830 int stringcount, int stringnumber, char *buffer,
831 int buffersize);
832
833 int pcre_get_named_substring(const pcre *code,
834 const char *subject, int *ovector,
835 int stringcount, const char *stringname,
836 const char **stringptr);
837
838 int pcre_get_stringnumber(const pcre *code,
839 const char *name);
840
841 int pcre_get_stringtable_entries(const pcre *code,
842 const char *name, char **first, char **last);
843
844 int pcre_get_substring(const char *subject, int *ovector,
845 int stringcount, int stringnumber,
846 const char **stringptr);
847
848 int pcre_get_substring_list(const char *subject,
849 int *ovector, int stringcount, const char ***listptr);
850
851 void pcre_free_substring(const char *stringptr);
852
853 void pcre_free_substring_list(const char **stringptr);
854
855 const unsigned char *pcre_maketables(void);
856
857 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
858 int what, void *where);
859
860 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
861
862 int pcre_refcount(pcre *code, int adjust);
863
864 int pcre_config(int what, void *where);
865
866 char *pcre_version(void);
867
868 void *(*pcre_malloc)(size_t);
869
870 void (*pcre_free)(void *);
871
872 void *(*pcre_stack_malloc)(size_t);
873
874 void (*pcre_stack_free)(void *);
875
876 int (*pcre_callout)(pcre_callout_block *);
877
878
879 PCRE API OVERVIEW
880
881 PCRE has its own native API, which is described in this document. There
882 are also some wrapper functions that correspond to the POSIX regular
883 expression API. These are described in the pcreposix documentation.
884 Both of these APIs define a set of C function calls. A C++ wrapper is
885 distributed with PCRE. It is documented in the pcrecpp page.
886
887 The native API C function prototypes are defined in the header file
888 pcre.h, and on Unix systems the library itself is called libpcre. It
889 can normally be accessed by adding -lpcre to the command for linking an
890 application that uses PCRE. The header file defines the macros
891 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
892 bers for the library. Applications can use these to include support
893 for different releases of PCRE.
894
895 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
896 pcre_exec() are used for compiling and matching regular expressions in
897 a Perl-compatible manner. A sample program that demonstrates the sim-
898 plest way of using them is provided in the file called pcredemo.c in
899 the source distribution. The pcresample documentation describes how to
900 compile and run it.
901
902 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
903 ble, is also provided. This uses a different algorithm for the match-
904 ing. The alternative algorithm finds all possible matches (at a given
905 point in the subject), and scans the subject just once. However, this
906 algorithm does not return captured substrings. A description of the two
907 matching algorithms and their advantages and disadvantages is given in
908 the pcrematching documentation.
909
910 In addition to the main compiling and matching functions, there are
911 convenience functions for extracting captured substrings from a subject
912 string that is matched by pcre_exec(). They are:
913
914 pcre_copy_substring()
915 pcre_copy_named_substring()
916 pcre_get_substring()
917 pcre_get_named_substring()
918 pcre_get_substring_list()
919 pcre_get_stringnumber()
920 pcre_get_stringtable_entries()
921
922 pcre_free_substring() and pcre_free_substring_list() are also provided,
923 to free the memory used for extracted strings.
924
925 The function pcre_maketables() is used to build a set of character
926 tables in the current locale for passing to pcre_compile(),
927 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
928 provided for specialist use. Most commonly, no special tables are
929 passed, in which case internal tables that are generated when PCRE is
930 built are used.
931
932 The function pcre_fullinfo() is used to find out information about a
933 compiled pattern; pcre_info() is an obsolete version that returns only
934 some of the available information, but is retained for backwards com-
935 patibility. The function pcre_version() returns a pointer to a string
936 containing the version of PCRE and its date of release.
937
938 The function pcre_refcount() maintains a reference count in a data
939 block containing a compiled pattern. This is provided for the benefit
940 of object-oriented applications.
941
942 The global variables pcre_malloc and pcre_free initially contain the
943 entry points of the standard malloc() and free() functions, respec-
944 tively. PCRE calls the memory management functions via these variables,
945 so a calling program can replace them if it wishes to intercept the
946 calls. This should be done before calling any PCRE functions.
947
948 The global variables pcre_stack_malloc and pcre_stack_free are also
949 indirections to memory management functions. These special functions
950 are used only when PCRE is compiled to use the heap for remembering
951 data, instead of recursive function calls, when running the pcre_exec()
952 function. See the pcrebuild documentation for details of how to do
953 this. It is a non-standard way of building PCRE, for use in environ-
954 ments that have limited stacks. Because of the greater use of memory
955 management, it runs more slowly. Separate functions are provided so
956 that special-purpose external code can be used for this case. When
957 used, these functions are always called in a stack-like manner (last
958 obtained, first freed), and always for memory blocks of the same size.
959 There is a discussion about PCRE's stack usage in the pcrestack docu-
960 mentation.
961
962 The global variable pcre_callout initially contains NULL. It can be set
963 by the caller to a "callout" function, which PCRE will then call at
964 specified points during a matching operation. Details are given in the
965 pcrecallout documentation.
966
967
968 NEWLINES
969
970 PCRE supports five different conventions for indicating line breaks in
971 strings: a single CR (carriage return) character, a single LF (line-
972 feed) character, the two-character sequence CRLF, any of the three pre-
973 ceding, or any Unicode newline sequence. The Unicode newline sequences
974 are the three just mentioned, plus the single characters VT (vertical
975 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
976 separator, U+2028), and PS (paragraph separator, U+2029).
977
978 Each of the first three conventions is used by at least one operating
979 system as its standard newline sequence. When PCRE is built, a default
980 can be specified. The default default is LF, which is the Unix stan-
981 dard. When PCRE is run, the default can be overridden, either when a
982 pattern is compiled, or when it is matched.
983
984 At compile time, the newline convention can be specified by the options
985 argument of pcre_compile(), or it can be specified by special text at
986 the start of the pattern itself; this overrides any other settings. See
987 the pcrepattern page for details of the special character sequences.
988
989 In the PCRE documentation the word "newline" is used to mean "the char-
990 acter or pair of characters that indicate a line break". The choice of
991 newline convention affects the handling of the dot, circumflex, and
992 dollar metacharacters, the handling of #-comments in /x mode, and, when
993 CRLF is a recognized line ending sequence, the match position advance-
994 ment for a non-anchored pattern. There is more detail about this in the
995 section on pcre_exec() options below.
996
997 The choice of newline convention does not affect the interpretation of
998 the \n or \r escape sequences, nor does it affect what \R matches,
999 which is controlled in a similar way, but by separate options.
1000
1001
1002 MULTITHREADING
1003
1004 The PCRE functions can be used in multi-threading applications, with
1005 the proviso that the memory management functions pointed to by
1006 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1007 callout function pointed to by pcre_callout, are shared by all threads.
1008
1009 The compiled form of a regular expression is not altered during match-
1010 ing, so the same compiled pattern can safely be used by several threads
1011 at once.
1012
1013
1014 SAVING PRECOMPILED PATTERNS FOR LATER USE
1015
1016 The compiled form of a regular expression can be saved and re-used at a
1017 later time, possibly by a different program, and even on a host other
1018 than the one on which it was compiled. Details are given in the
1019 pcreprecompile documentation. However, compiling a regular expression
1020 with one version of PCRE for use with a different version is not guar-
1021 anteed to work and may cause crashes.
1022
1023
1024 CHECKING BUILD-TIME OPTIONS
1025
1026 int pcre_config(int what, void *where);
1027
1028 The function pcre_config() makes it possible for a PCRE client to dis-
1029 cover which optional features have been compiled into the PCRE library.
1030 The pcrebuild documentation has more details about these optional fea-
1031 tures.
1032
1033 The first argument for pcre_config() is an integer, specifying which
1034 information is required; the second argument is a pointer to a variable
1035 into which the information is placed. The following information is
1036 available:
1037
1038 PCRE_CONFIG_UTF8
1039
1040 The output is an integer that is set to one if UTF-8 support is avail-
1041 able; otherwise it is set to zero.
1042
1043 PCRE_CONFIG_UNICODE_PROPERTIES
1044
1045 The output is an integer that is set to one if support for Unicode
1046 character properties is available; otherwise it is set to zero.
1047
1048 PCRE_CONFIG_NEWLINE
1049
1050 The output is an integer whose value specifies the default character
1051 sequence that is recognized as meaning "newline". The four values that
1052 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1053 and -1 for ANY. Though they are derived from ASCII, the same values
1054 are returned in EBCDIC environments. The default should normally corre-
1055 spond to the standard sequence for your operating system.
1056
1057 PCRE_CONFIG_BSR
1058
1059 The output is an integer whose value indicates what character sequences
1060 the \R escape sequence matches by default. A value of 0 means that \R
1061 matches any Unicode line ending sequence; a value of 1 means that \R
1062 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1063 tern is compiled or matched.
1064
1065 PCRE_CONFIG_LINK_SIZE
1066
1067 The output is an integer that contains the number of bytes used for
1068 internal linkage in compiled regular expressions. The value is 2, 3, or
1069 4. Larger values allow larger regular expressions to be compiled, at
1070 the expense of slower matching. The default value of 2 is sufficient
1071 for all but the most massive patterns, since it allows the compiled
1072 pattern to be up to 64K in size.
1073
1074 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1075
1076 The output is an integer that contains the threshold above which the
1077 POSIX interface uses malloc() for output vectors. Further details are
1078 given in the pcreposix documentation.
1079
1080 PCRE_CONFIG_MATCH_LIMIT
1081
1082 The output is a long integer that gives the default limit for the num-
1083 ber of internal matching function calls in a pcre_exec() execution.
1084 Further details are given with pcre_exec() below.
1085
1086 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1087
1088 The output is a long integer that gives the default limit for the depth
1089 of recursion when calling the internal matching function in a
1090 pcre_exec() execution. Further details are given with pcre_exec()
1091 below.
1092
1093 PCRE_CONFIG_STACKRECURSE
1094
1095 The output is an integer that is set to one if internal recursion when
1096 running pcre_exec() is implemented by recursive function calls that use
1097 the stack to remember their state. This is the usual way that PCRE is
1098 compiled. The output is zero if PCRE was compiled to use blocks of data
1099 on the heap instead of recursive function calls. In this case,
1100 pcre_stack_malloc and pcre_stack_free are called to manage memory
1101 blocks on the heap, thus avoiding the use of the stack.
1102
1103
1104 COMPILING A PATTERN
1105
1106 pcre *pcre_compile(const char *pattern, int options,
1107 const char **errptr, int *erroffset,
1108 const unsigned char *tableptr);
1109
1110 pcre *pcre_compile2(const char *pattern, int options,
1111 int *errorcodeptr,
1112 const char **errptr, int *erroffset,
1113 const unsigned char *tableptr);
1114
1115 Either of the functions pcre_compile() or pcre_compile2() can be called
1116 to compile a pattern into an internal form. The only difference between
1117 the two interfaces is that pcre_compile2() has an additional argument,
1118 errorcodeptr, via which a numerical error code can be returned.
1119
1120 The pattern is a C string terminated by a binary zero, and is passed in
1121 the pattern argument. A pointer to a single block of memory that is
1122 obtained via pcre_malloc is returned. This contains the compiled code
1123 and related data. The pcre type is defined for the returned block; this
1124 is a typedef for a structure whose contents are not externally defined.
1125 It is up to the caller to free the memory (via pcre_free) when it is no
1126 longer required.
1127
1128 Although the compiled code of a PCRE regex is relocatable, that is, it
1129 does not depend on memory location, the complete pcre data block is not
1130 fully relocatable, because it may contain a copy of the tableptr argu-
1131 ment, which is an address (see below).
1132
1133 The options argument contains various bit settings that affect the com-
1134 pilation. It should be zero if no options are required. The available
1135 options are described below. Some of them, in particular, those that
1136 are compatible with Perl, can also be set and unset from within the
1137 pattern (see the detailed description in the pcrepattern documenta-
1138 tion). For these options, the contents of the options argument speci-
1139 fies their initial settings at the start of compilation and execution.
1140 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1141 of matching as well as at compile time.
1142
1143 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1144 if compilation of a pattern fails, pcre_compile() returns NULL, and
1145 sets the variable pointed to by errptr to point to a textual error mes-
1146 sage. This is a static string that is part of the library. You must not
1147 try to free it. The offset from the start of the pattern to the charac-
1148 ter where the error was discovered is placed in the variable pointed to
1149 by erroffset, which must not be NULL. If it is, an immediate error is
1150 given.
1151
1152 If pcre_compile2() is used instead of pcre_compile(), and the error-
1153 codeptr argument is not NULL, a non-zero error code number is returned
1154 via this argument in the event of an error. This is in addition to the
1155 textual error message. Error codes and messages are listed below.
1156
1157 If the final argument, tableptr, is NULL, PCRE uses a default set of
1158 character tables that are built when PCRE is compiled, using the
1159 default C locale. Otherwise, tableptr must be an address that is the
1160 result of a call to pcre_maketables(). This value is stored with the
1161 compiled pattern, and used again by pcre_exec(), unless another table
1162 pointer is passed to it. For more discussion, see the section on locale
1163 support below.
1164
1165 This code fragment shows a typical straightforward call to pcre_com-
1166 pile():
1167
1168 pcre *re;
1169 const char *error;
1170 int erroffset;
1171 re = pcre_compile(
1172 "^A.*Z", /* the pattern */
1173 0, /* default options */
1174 &error, /* for error message */
1175 &erroffset, /* for error offset */
1176 NULL); /* use default character tables */
1177
1178 The following names for option bits are defined in the pcre.h header
1179 file:
1180
1181 PCRE_ANCHORED
1182
1183 If this bit is set, the pattern is forced to be "anchored", that is, it
1184 is constrained to match only at the first matching point in the string
1185 that is being searched (the "subject string"). This effect can also be
1186 achieved by appropriate constructs in the pattern itself, which is the
1187 only way to do it in Perl.
1188
1189 PCRE_AUTO_CALLOUT
1190
1191 If this bit is set, pcre_compile() automatically inserts callout items,
1192 all with number 255, before each pattern item. For discussion of the
1193 callout facility, see the pcrecallout documentation.
1194
1195 PCRE_BSR_ANYCRLF
1196 PCRE_BSR_UNICODE
1197
1198 These options (which are mutually exclusive) control what the \R escape
1199 sequence matches. The choice is either to match only CR, LF, or CRLF,
1200 or to match any Unicode newline sequence. The default is specified when
1201 PCRE is built. It can be overridden from within the pattern, or by set-
1202 ting an option when a compiled pattern is matched.
1203
1204 PCRE_CASELESS
1205
1206 If this bit is set, letters in the pattern match both upper and lower
1207 case letters. It is equivalent to Perl's /i option, and it can be
1208 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1209 always understands the concept of case for characters whose values are
1210 less than 128, so caseless matching is always possible. For characters
1211 with higher values, the concept of case is supported if PCRE is com-
1212 piled with Unicode property support, but not otherwise. If you want to
1213 use caseless matching for characters 128 and above, you must ensure
1214 that PCRE is compiled with Unicode property support as well as with
1215 UTF-8 support.
1216
1217 PCRE_DOLLAR_ENDONLY
1218
1219 If this bit is set, a dollar metacharacter in the pattern matches only
1220 at the end of the subject string. Without this option, a dollar also
1221 matches immediately before a newline at the end of the string (but not
1222 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1223 if PCRE_MULTILINE is set. There is no equivalent to this option in
1224 Perl, and no way to set it within a pattern.
1225
1226 PCRE_DOTALL
1227
1228 If this bit is set, a dot metacharater in the pattern matches all char-
1229 acters, including those that indicate newline. Without it, a dot does
1230 not match when the current position is at a newline. This option is
1231 equivalent to Perl's /s option, and it can be changed within a pattern
1232 by a (?s) option setting. A negative class such as [^a] always matches
1233 newline characters, independent of the setting of this option.
1234
1235 PCRE_DUPNAMES
1236
1237 If this bit is set, names used to identify capturing subpatterns need
1238 not be unique. This can be helpful for certain types of pattern when it
1239 is known that only one instance of the named subpattern can ever be
1240 matched. There are more details of named subpatterns below; see also
1241 the pcrepattern documentation.
1242
1243 PCRE_EXTENDED
1244
1245 If this bit is set, whitespace data characters in the pattern are
1246 totally ignored except when escaped or inside a character class. White-
1247 space does not include the VT character (code 11). In addition, charac-
1248 ters between an unescaped # outside a character class and the next new-
1249 line, inclusive, are also ignored. This is equivalent to Perl's /x
1250 option, and it can be changed within a pattern by a (?x) option set-
1251 ting.
1252
1253 This option makes it possible to include comments inside complicated
1254 patterns. Note, however, that this applies only to data characters.
1255 Whitespace characters may never appear within special character
1256 sequences in a pattern, for example within the sequence (?( which
1257 introduces a conditional subpattern.
1258
1259 PCRE_EXTRA
1260
1261 This option was invented in order to turn on additional functionality
1262 of PCRE that is incompatible with Perl, but it is currently of very
1263 little use. When set, any backslash in a pattern that is followed by a
1264 letter that has no special meaning causes an error, thus reserving
1265 these combinations for future expansion. By default, as in Perl, a
1266 backslash followed by a letter with no special meaning is treated as a
1267 literal. (Perl can, however, be persuaded to give a warning for this.)
1268 There are at present no other features controlled by this option. It
1269 can also be set by a (?X) option setting within a pattern.
1270
1271 PCRE_FIRSTLINE
1272
1273 If this option is set, an unanchored pattern is required to match
1274 before or at the first newline in the subject string, though the
1275 matched text may continue over the newline.
1276
1277 PCRE_JAVASCRIPT_COMPAT
1278
1279 If this option is set, PCRE's behaviour is changed in some ways so that
1280 it is compatible with JavaScript rather than Perl. The changes are as
1281 follows:
1282
1283 (1) A lone closing square bracket in a pattern causes a compile-time
1284 error, because this is illegal in JavaScript (by default it is treated
1285 as a data character). Thus, the pattern AB]CD becomes illegal when this
1286 option is set.
1287
1288 (2) At run time, a back reference to an unset subpattern group matches
1289 an empty string (by default this causes the current matching alterna-
1290 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1291 set (assuming it can find an "a" in the subject), whereas it fails by
1292 default, for Perl compatibility.
1293
1294 PCRE_MULTILINE
1295
1296 By default, PCRE treats the subject string as consisting of a single
1297 line of characters (even if it actually contains newlines). The "start
1298 of line" metacharacter (^) matches only at the start of the string,
1299 while the "end of line" metacharacter ($) matches only at the end of
1300 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1301 is set). This is the same as Perl.
1302
1303 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1304 constructs match immediately following or immediately before internal
1305 newlines in the subject string, respectively, as well as at the very
1306 start and end. This is equivalent to Perl's /m option, and it can be
1307 changed within a pattern by a (?m) option setting. If there are no new-
1308 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1309 setting PCRE_MULTILINE has no effect.
1310
1311 PCRE_NEWLINE_CR
1312 PCRE_NEWLINE_LF
1313 PCRE_NEWLINE_CRLF
1314 PCRE_NEWLINE_ANYCRLF
1315 PCRE_NEWLINE_ANY
1316
1317 These options override the default newline definition that was chosen
1318 when PCRE was built. Setting the first or the second specifies that a
1319 newline is indicated by a single character (CR or LF, respectively).
1320 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1321 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1322 that any of the three preceding sequences should be recognized. Setting
1323 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1324 recognized. The Unicode newline sequences are the three just mentioned,
1325 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1326 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1327 (paragraph separator, U+2029). The last two are recognized only in
1328 UTF-8 mode.
1329
1330 The newline setting in the options word uses three bits that are
1331 treated as a number, giving eight possibilities. Currently only six are
1332 used (default plus the five values above). This means that if you set
1333 more than one newline option, the combination may or may not be sensi-
1334 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1335 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1336 cause an error.
1337
1338 The only time that a line break is specially recognized when compiling
1339 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1340 character class is encountered. This indicates a comment that lasts
1341 until after the next line break sequence. In other circumstances, line
1342 break sequences are treated as literal data, except that in
1343 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1344 and are therefore ignored.
1345
1346 The newline option that is set at compile time becomes the default that
1347 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1348
1349 PCRE_NO_AUTO_CAPTURE
1350
1351 If this option is set, it disables the use of numbered capturing paren-
1352 theses in the pattern. Any opening parenthesis that is not followed by
1353 ? behaves as if it were followed by ?: but named parentheses can still
1354 be used for capturing (and they acquire numbers in the usual way).
1355 There is no equivalent of this option in Perl.
1356
1357 PCRE_UNGREEDY
1358
1359 This option inverts the "greediness" of the quantifiers so that they
1360 are not greedy by default, but become greedy if followed by "?". It is
1361 not compatible with Perl. It can also be set by a (?U) option setting
1362 within the pattern.
1363
1364 PCRE_UTF8
1365
1366 This option causes PCRE to regard both the pattern and the subject as
1367 strings of UTF-8 characters instead of single-byte character strings.
1368 However, it is available only when PCRE is built to include UTF-8 sup-
1369 port. If not, the use of this option provokes an error. Details of how
1370 this option changes the behaviour of PCRE are given in the section on
1371 UTF-8 support in the main pcre page.
1372
1373 PCRE_NO_UTF8_CHECK
1374
1375 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1376 automatically checked. There is a discussion about the validity of
1377 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1378 bytes is found, pcre_compile() returns an error. If you already know
1379 that your pattern is valid, and you want to skip this check for perfor-
1380 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1381 set, the effect of passing an invalid UTF-8 string as a pattern is
1382 undefined. It may cause your program to crash. Note that this option
1383 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1384 UTF-8 validity checking of subject strings.
1385
1386
1387 COMPILATION ERROR CODES
1388
1389 The following table lists the error codes than may be returned by
1390 pcre_compile2(), along with the error messages that may be returned by
1391 both compiling functions. As PCRE has developed, some error codes have
1392 fallen out of use. To avoid confusion, they have not been re-used.
1393
1394 0 no error
1395 1 \ at end of pattern
1396 2 \c at end of pattern
1397 3 unrecognized character follows \
1398 4 numbers out of order in {} quantifier
1399 5 number too big in {} quantifier
1400 6 missing terminating ] for character class
1401 7 invalid escape sequence in character class
1402 8 range out of order in character class
1403 9 nothing to repeat
1404 10 [this code is not in use]
1405 11 internal error: unexpected repeat
1406 12 unrecognized character after (? or (?-
1407 13 POSIX named classes are supported only within a class
1408 14 missing )
1409 15 reference to non-existent subpattern
1410 16 erroffset passed as NULL
1411 17 unknown option bit(s) set
1412 18 missing ) after comment
1413 19 [this code is not in use]
1414 20 regular expression is too large
1415 21 failed to get memory
1416 22 unmatched parentheses
1417 23 internal error: code overflow
1418 24 unrecognized character after (?<
1419 25 lookbehind assertion is not fixed length
1420 26 malformed number or name after (?(
1421 27 conditional group contains more than two branches
1422 28 assertion expected after (?(
1423 29 (?R or (?[+-]digits must be followed by )
1424 30 unknown POSIX class name
1425 31 POSIX collating elements are not supported
1426 32 this version of PCRE is not compiled with PCRE_UTF8 support
1427 33 [this code is not in use]
1428 34 character value in \x{...} sequence is too large
1429 35 invalid condition (?(0)
1430 36 \C not allowed in lookbehind assertion
1431 37 PCRE does not support \L, \l, \N, \U, or \u
1432 38 number after (?C is > 255
1433 39 closing ) for (?C expected
1434 40 recursive call could loop indefinitely
1435 41 unrecognized character after (?P
1436 42 syntax error in subpattern name (missing terminator)
1437 43 two named subpatterns have the same name
1438 44 invalid UTF-8 string
1439 45 support for \P, \p, and \X has not been compiled
1440 46 malformed \P or \p sequence
1441 47 unknown property name after \P or \p
1442 48 subpattern name is too long (maximum 32 characters)
1443 49 too many named subpatterns (maximum 10000)
1444 50 [this code is not in use]
1445 51 octal value is greater than \377 (not in UTF-8 mode)
1446 52 internal error: overran compiling workspace
1447 53 internal error: previously-checked referenced subpattern not
1448 found
1449 54 DEFINE group contains more than one branch
1450 55 repeating a DEFINE group is not allowed
1451 56 inconsistent NEWLINE options
1452 57 \g is not followed by a braced, angle-bracketed, or quoted
1453 name/number or by a plain number
1454 58 a numbered reference must not be zero
1455 59 (*VERB) with an argument is not supported
1456 60 (*VERB) not recognized
1457 61 number is too big
1458 62 subpattern name expected
1459 63 digit expected after (?+
1460 64 ] is an invalid data character in JavaScript compatibility mode
1461
1462 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1463 values may be used if the limits were changed when PCRE was built.
1464
1465
1466 STUDYING A PATTERN
1467
1468 pcre_extra *pcre_study(const pcre *code, int options
1469 const char **errptr);
1470
1471 If a compiled pattern is going to be used several times, it is worth
1472 spending more time analyzing it in order to speed up the time taken for
1473 matching. The function pcre_study() takes a pointer to a compiled pat-
1474 tern as its first argument. If studying the pattern produces additional
1475 information that will help speed up matching, pcre_study() returns a
1476 pointer to a pcre_extra block, in which the study_data field points to
1477 the results of the study.
1478
1479 The returned value from pcre_study() can be passed directly to
1480 pcre_exec(). However, a pcre_extra block also contains other fields
1481 that can be set by the caller before the block is passed; these are
1482 described below in the section on matching a pattern.
1483
1484 If studying the pattern does not produce any additional information
1485 pcre_study() returns NULL. In that circumstance, if the calling program
1486 wants to pass any of the other fields to pcre_exec(), it must set up
1487 its own pcre_extra block.
1488
1489 The second argument of pcre_study() contains option bits. At present,
1490 no options are defined, and this argument should always be zero.
1491
1492 The third argument for pcre_study() is a pointer for an error message.
1493 If studying succeeds (even if no data is returned), the variable it
1494 points to is set to NULL. Otherwise it is set to point to a textual
1495 error message. This is a static string that is part of the library. You
1496 must not try to free it. You should test the error pointer for NULL
1497 after calling pcre_study(), to be sure that it has run successfully.
1498
1499 This is a typical call to pcre_study():
1500
1501 pcre_extra *pe;
1502 pe = pcre_study(
1503 re, /* result of pcre_compile() */
1504 0, /* no options exist */
1505 &error); /* set to NULL or points to a message */
1506
1507 At present, studying a pattern is useful only for non-anchored patterns
1508 that do not have a single fixed starting character. A bitmap of possi-
1509 ble starting bytes is created.
1510
1511
1512 LOCALE SUPPORT
1513
1514 PCRE handles caseless matching, and determines whether characters are
1515 letters, digits, or whatever, by reference to a set of tables, indexed
1516 by character value. When running in UTF-8 mode, this applies only to
1517 characters with codes less than 128. Higher-valued codes never match
1518 escapes such as \w or \d, but can be tested with \p if PCRE is built
1519 with Unicode character property support. The use of locales with Uni-
1520 code is discouraged. If you are handling characters with codes greater
1521 than 128, you should either use UTF-8 and Unicode, or use locales, but
1522 not try to mix the two.
1523
1524 PCRE contains an internal set of tables that are used when the final
1525 argument of pcre_compile() is NULL. These are sufficient for many
1526 applications. Normally, the internal tables recognize only ASCII char-
1527 acters. However, when PCRE is built, it is possible to cause the inter-
1528 nal tables to be rebuilt in the default "C" locale of the local system,
1529 which may cause them to be different.
1530
1531 The internal tables can always be overridden by tables supplied by the
1532 application that calls PCRE. These may be created in a different locale
1533 from the default. As more and more applications change to using Uni-
1534 code, the need for this locale support is expected to die away.
1535
1536 External tables are built by calling the pcre_maketables() function,
1537 which has no arguments, in the relevant locale. The result can then be
1538 passed to pcre_compile() or pcre_exec() as often as necessary. For
1539 example, to build and use tables that are appropriate for the French
1540 locale (where accented characters with values greater than 128 are
1541 treated as letters), the following code could be used:
1542
1543 setlocale(LC_CTYPE, "fr_FR");
1544 tables = pcre_maketables();
1545 re = pcre_compile(..., tables);
1546
1547 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1548 if you are using Windows, the name for the French locale is "french".
1549
1550 When pcre_maketables() runs, the tables are built in memory that is
1551 obtained via pcre_malloc. It is the caller's responsibility to ensure
1552 that the memory containing the tables remains available for as long as
1553 it is needed.
1554
1555 The pointer that is passed to pcre_compile() is saved with the compiled
1556 pattern, and the same tables are used via this pointer by pcre_study()
1557 and normally also by pcre_exec(). Thus, by default, for any single pat-
1558 tern, compilation, studying and matching all happen in the same locale,
1559 but different patterns can be compiled in different locales.
1560
1561 It is possible to pass a table pointer or NULL (indicating the use of
1562 the internal tables) to pcre_exec(). Although not intended for this
1563 purpose, this facility could be used to match a pattern in a different
1564 locale from the one in which it was compiled. Passing table pointers at
1565 run time is discussed below in the section on matching a pattern.
1566
1567
1568 INFORMATION ABOUT A PATTERN
1569
1570 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1571 int what, void *where);
1572
1573 The pcre_fullinfo() function returns information about a compiled pat-
1574 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1575 less retained for backwards compability (and is documented below).
1576
1577 The first argument for pcre_fullinfo() is a pointer to the compiled
1578 pattern. The second argument is the result of pcre_study(), or NULL if
1579 the pattern was not studied. The third argument specifies which piece
1580 of information is required, and the fourth argument is a pointer to a
1581 variable to receive the data. The yield of the function is zero for
1582 success, or one of the following negative numbers:
1583
1584 PCRE_ERROR_NULL the argument code was NULL
1585 the argument where was NULL
1586 PCRE_ERROR_BADMAGIC the "magic number" was not found
1587 PCRE_ERROR_BADOPTION the value of what was invalid
1588
1589 The "magic number" is placed at the start of each compiled pattern as
1590 an simple check against passing an arbitrary memory pointer. Here is a
1591 typical call of pcre_fullinfo(), to obtain the length of the compiled
1592 pattern:
1593
1594 int rc;
1595 size_t length;
1596 rc = pcre_fullinfo(
1597 re, /* result of pcre_compile() */
1598 pe, /* result of pcre_study(), or NULL */
1599 PCRE_INFO_SIZE, /* what is required */
1600 &length); /* where to put the data */
1601
1602 The possible values for the third argument are defined in pcre.h, and
1603 are as follows:
1604
1605 PCRE_INFO_BACKREFMAX
1606
1607 Return the number of the highest back reference in the pattern. The
1608 fourth argument should point to an int variable. Zero is returned if
1609 there are no back references.
1610
1611 PCRE_INFO_CAPTURECOUNT
1612
1613 Return the number of capturing subpatterns in the pattern. The fourth
1614 argument should point to an int variable.
1615
1616 PCRE_INFO_DEFAULT_TABLES
1617
1618 Return a pointer to the internal default character tables within PCRE.
1619 The fourth argument should point to an unsigned char * variable. This
1620 information call is provided for internal use by the pcre_study() func-
1621 tion. External callers can cause PCRE to use its internal tables by
1622 passing a NULL table pointer.
1623
1624 PCRE_INFO_FIRSTBYTE
1625
1626 Return information about the first byte of any matched string, for a
1627 non-anchored pattern. The fourth argument should point to an int vari-
1628 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1629 is still recognized for backwards compatibility.)
1630
1631 If there is a fixed first byte, for example, from a pattern such as
1632 (cat|cow|coyote), its value is returned. Otherwise, if either
1633
1634 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1635 branch starts with "^", or
1636
1637 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1638 set (if it were set, the pattern would be anchored),
1639
1640 -1 is returned, indicating that the pattern matches only at the start
1641 of a subject string or after any newline within the string. Otherwise
1642 -2 is returned. For anchored patterns, -2 is returned.
1643
1644 PCRE_INFO_FIRSTTABLE
1645
1646 If the pattern was studied, and this resulted in the construction of a
1647 256-bit table indicating a fixed set of bytes for the first byte in any
1648 matching string, a pointer to the table is returned. Otherwise NULL is
1649 returned. The fourth argument should point to an unsigned char * vari-
1650 able.
1651
1652 PCRE_INFO_HASCRORLF
1653
1654 Return 1 if the pattern contains any explicit matches for CR or LF
1655 characters, otherwise 0. The fourth argument should point to an int
1656 variable. An explicit match is either a literal CR or LF character, or
1657 \r or \n.
1658
1659 PCRE_INFO_JCHANGED
1660
1661 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1662 otherwise 0. The fourth argument should point to an int variable. (?J)
1663 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1664
1665 PCRE_INFO_LASTLITERAL
1666
1667 Return the value of the rightmost literal byte that must exist in any
1668 matched string, other than at its start, if such a byte has been
1669 recorded. The fourth argument should point to an int variable. If there
1670 is no such byte, -1 is returned. For anchored patterns, a last literal
1671 byte is recorded only if it follows something of variable length. For
1672 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1673 /^a\dz\d/ the returned value is -1.
1674
1675 PCRE_INFO_NAMECOUNT
1676 PCRE_INFO_NAMEENTRYSIZE
1677 PCRE_INFO_NAMETABLE
1678
1679 PCRE supports the use of named as well as numbered capturing parenthe-
1680 ses. The names are just an additional way of identifying the parenthe-
1681 ses, which still acquire numbers. Several convenience functions such as
1682 pcre_get_named_substring() are provided for extracting captured sub-
1683 strings by name. It is also possible to extract the data directly, by
1684 first converting the name to a number in order to access the correct
1685 pointers in the output vector (described with pcre_exec() below). To do
1686 the conversion, you need to use the name-to-number map, which is
1687 described by these three values.
1688
1689 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1690 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1691 of each entry; both of these return an int value. The entry size
1692 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1693 a pointer to the first entry of the table (a pointer to char). The
1694 first two bytes of each entry are the number of the capturing parenthe-
1695 sis, most significant byte first. The rest of the entry is the corre-
1696 sponding name, zero terminated. The names are in alphabetical order.
1697 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1698 theses numbers. For example, consider the following pattern (assume
1699 PCRE_EXTENDED is set, so white space - including newlines - is
1700 ignored):
1701
1702 (?<date> (?<year>(\d\d)?\d\d) -
1703 (?<month>\d\d) - (?<day>\d\d) )
1704
1705 There are four named subpatterns, so the table has four entries, and
1706 each entry in the table is eight bytes long. The table is as follows,
1707 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1708 as ??:
1709
1710 00 01 d a t e 00 ??
1711 00 05 d a y 00 ?? ??
1712 00 04 m o n t h 00
1713 00 02 y e a r 00 ??
1714
1715 When writing code to extract data from named subpatterns using the
1716 name-to-number map, remember that the length of the entries is likely
1717 to be different for each compiled pattern.
1718
1719 PCRE_INFO_OKPARTIAL
1720
1721 Return 1 if the pattern can be used for partial matching, otherwise 0.
1722 The fourth argument should point to an int variable. The pcrepartial
1723 documentation lists the restrictions that apply to patterns when par-
1724 tial matching is used.
1725
1726 PCRE_INFO_OPTIONS
1727
1728 Return a copy of the options with which the pattern was compiled. The
1729 fourth argument should point to an unsigned long int variable. These
1730 option bits are those specified in the call to pcre_compile(), modified
1731 by any top-level option settings at the start of the pattern itself. In
1732 other words, they are the options that will be in force when matching
1733 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1734 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1735 and PCRE_EXTENDED.
1736
1737 A pattern is automatically anchored by PCRE if all of its top-level
1738 alternatives begin with one of the following:
1739
1740 ^ unless PCRE_MULTILINE is set
1741 \A always
1742 \G always
1743 .* if PCRE_DOTALL is set and there are no back
1744 references to the subpattern in which .* appears
1745
1746 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1747 by pcre_fullinfo().
1748
1749 PCRE_INFO_SIZE
1750
1751 Return the size of the compiled pattern, that is, the value that was
1752 passed as the argument to pcre_malloc() when PCRE was getting memory in
1753 which to place the compiled data. The fourth argument should point to a
1754 size_t variable.
1755
1756 PCRE_INFO_STUDYSIZE
1757
1758 Return the size of the data block pointed to by the study_data field in
1759 a pcre_extra block. That is, it is the value that was passed to
1760 pcre_malloc() when PCRE was getting memory into which to place the data
1761 created by pcre_study(). The fourth argument should point to a size_t
1762 variable.
1763
1764
1765 OBSOLETE INFO FUNCTION
1766
1767 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1768
1769 The pcre_info() function is now obsolete because its interface is too
1770 restrictive to return all the available data about a compiled pattern.
1771 New programs should use pcre_fullinfo() instead. The yield of
1772 pcre_info() is the number of capturing subpatterns, or one of the fol-
1773 lowing negative numbers:
1774
1775 PCRE_ERROR_NULL the argument code was NULL
1776 PCRE_ERROR_BADMAGIC the "magic number" was not found
1777
1778 If the optptr argument is not NULL, a copy of the options with which
1779 the pattern was compiled is placed in the integer it points to (see
1780 PCRE_INFO_OPTIONS above).
1781
1782 If the pattern is not anchored and the firstcharptr argument is not
1783 NULL, it is used to pass back information about the first character of
1784 any matched string (see PCRE_INFO_FIRSTBYTE above).
1785
1786
1787 REFERENCE COUNTS
1788
1789 int pcre_refcount(pcre *code, int adjust);
1790
1791 The pcre_refcount() function is used to maintain a reference count in
1792 the data block that contains a compiled pattern. It is provided for the
1793 benefit of applications that operate in an object-oriented manner,
1794 where different parts of the application may be using the same compiled
1795 pattern, but you want to free the block when they are all done.
1796
1797 When a pattern is compiled, the reference count field is initialized to
1798 zero. It is changed only by calling this function, whose action is to
1799 add the adjust value (which may be positive or negative) to it. The
1800 yield of the function is the new value. However, the value of the count
1801 is constrained to lie between 0 and 65535, inclusive. If the new value
1802 is outside these limits, it is forced to the appropriate limit value.
1803
1804 Except when it is zero, the reference count is not correctly preserved
1805 if a pattern is compiled on one host and then transferred to a host
1806 whose byte-order is different. (This seems a highly unlikely scenario.)
1807
1808
1809 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1810
1811 int pcre_exec(const pcre *code, const pcre_extra *extra,
1812 const char *subject, int length, int startoffset,
1813 int options, int *ovector, int ovecsize);
1814
1815 The function pcre_exec() is called to match a subject string against a
1816 compiled pattern, which is passed in the code argument. If the pattern
1817 has been studied, the result of the study should be passed in the extra
1818 argument. This function is the main matching facility of the library,
1819 and it operates in a Perl-like manner. For specialist use there is also
1820 an alternative matching function, which is described below in the sec-
1821 tion about the pcre_dfa_exec() function.
1822
1823 In most applications, the pattern will have been compiled (and option-
1824 ally studied) in the same process that calls pcre_exec(). However, it
1825 is possible to save compiled patterns and study data, and then use them
1826 later in different processes, possibly even on different hosts. For a
1827 discussion about this, see the pcreprecompile documentation.
1828
1829 Here is an example of a simple call to pcre_exec():
1830
1831 int rc;
1832 int ovector[30];
1833 rc = pcre_exec(
1834 re, /* result of pcre_compile() */
1835 NULL, /* we didn't study the pattern */
1836 "some string", /* the subject string */
1837 11, /* the length of the subject string */
1838 0, /* start at offset 0 in the subject */
1839 0, /* default options */
1840 ovector, /* vector of integers for substring information */
1841 30); /* number of elements (NOT size in bytes) */
1842
1843 Extra data for pcre_exec()
1844
1845 If the extra argument is not NULL, it must point to a pcre_extra data
1846 block. The pcre_study() function returns such a block (when it doesn't
1847 return NULL), but you can also create one for yourself, and pass addi-
1848 tional information in it. The pcre_extra block contains the following
1849 fields (not necessarily in this order):
1850
1851 unsigned long int flags;
1852 void *study_data;
1853 unsigned long int match_limit;
1854 unsigned long int match_limit_recursion;
1855 void *callout_data;
1856 const unsigned char *tables;
1857
1858 The flags field is a bitmap that specifies which of the other fields
1859 are set. The flag bits are:
1860
1861 PCRE_EXTRA_STUDY_DATA
1862 PCRE_EXTRA_MATCH_LIMIT
1863 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1864 PCRE_EXTRA_CALLOUT_DATA
1865 PCRE_EXTRA_TABLES
1866
1867 Other flag bits should be set to zero. The study_data field is set in
1868 the pcre_extra block that is returned by pcre_study(), together with
1869 the appropriate flag bit. You should not set this yourself, but you may
1870 add to the block by setting the other fields and their corresponding
1871 flag bits.
1872
1873 The match_limit field provides a means of preventing PCRE from using up
1874 a vast amount of resources when running patterns that are not going to
1875 match, but which have a very large number of possibilities in their
1876 search trees. The classic example is the use of nested unlimited
1877 repeats.
1878
1879 Internally, PCRE uses a function called match() which it calls repeat-
1880 edly (sometimes recursively). The limit set by match_limit is imposed
1881 on the number of times this function is called during a match, which
1882 has the effect of limiting the amount of backtracking that can take
1883 place. For patterns that are not anchored, the count restarts from zero
1884 for each position in the subject string.
1885
1886 The default value for the limit can be set when PCRE is built; the
1887 default default is 10 million, which handles all but the most extreme
1888 cases. You can override the default by suppling pcre_exec() with a
1889 pcre_extra block in which match_limit is set, and
1890 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1891 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1892
1893 The match_limit_recursion field is similar to match_limit, but instead
1894 of limiting the total number of times that match() is called, it limits
1895 the depth of recursion. The recursion depth is a smaller number than
1896 the total number of calls, because not all calls to match() are recur-
1897 sive. This limit is of use only if it is set smaller than match_limit.
1898
1899 Limiting the recursion depth limits the amount of stack that can be
1900 used, or, when PCRE has been compiled to use memory on the heap instead
1901 of the stack, the amount of heap memory that can be used.
1902
1903 The default value for match_limit_recursion can be set when PCRE is
1904 built; the default default is the same value as the default for
1905 match_limit. You can override the default by suppling pcre_exec() with
1906 a pcre_extra block in which match_limit_recursion is set, and
1907 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1908 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1909
1910 The pcre_callout field is used in conjunction with the "callout" fea-
1911 ture, which is described in the pcrecallout documentation.
1912
1913 The tables field is used to pass a character tables pointer to
1914 pcre_exec(); this overrides the value that is stored with the compiled
1915 pattern. A non-NULL value is stored with the compiled pattern only if
1916 custom tables were supplied to pcre_compile() via its tableptr argu-
1917 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1918 PCRE's internal tables to be used. This facility is helpful when re-
1919 using patterns that have been saved after compiling with an external
1920 set of tables, because the external tables might be at a different
1921 address when pcre_exec() is called. See the pcreprecompile documenta-
1922 tion for a discussion of saving compiled patterns for later use.
1923
1924 Option bits for pcre_exec()
1925
1926 The unused bits of the options argument for pcre_exec() must be zero.
1927 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1928 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_START_OPTIMIZE,
1929 PCRE_NO_UTF8_CHECK and PCRE_PARTIAL.
1930
1931 PCRE_ANCHORED
1932
1933 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1934 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1935 turned out to be anchored by virtue of its contents, it cannot be made
1936 unachored at matching time.
1937
1938 PCRE_BSR_ANYCRLF
1939 PCRE_BSR_UNICODE
1940
1941 These options (which are mutually exclusive) control what the \R escape
1942 sequence matches. The choice is either to match only CR, LF, or CRLF,
1943 or to match any Unicode newline sequence. These options override the
1944 choice that was made or defaulted when the pattern was compiled.
1945
1946 PCRE_NEWLINE_CR
1947 PCRE_NEWLINE_LF
1948 PCRE_NEWLINE_CRLF
1949 PCRE_NEWLINE_ANYCRLF
1950 PCRE_NEWLINE_ANY
1951
1952 These options override the newline definition that was chosen or
1953 defaulted when the pattern was compiled. For details, see the descrip-
1954 tion of pcre_compile() above. During matching, the newline choice
1955 affects the behaviour of the dot, circumflex, and dollar metacharac-
1956 ters. It may also alter the way the match position is advanced after a
1957 match failure for an unanchored pattern.
1958
1959 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1960 set, and a match attempt for an unanchored pattern fails when the cur-
1961 rent position is at a CRLF sequence, and the pattern contains no
1962 explicit matches for CR or LF characters, the match position is
1963 advanced by two characters instead of one, in other words, to after the
1964 CRLF.
1965
1966 The above rule is a compromise that makes the most common cases work as
1967 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1968 option is not set), it does not match the string "\r\nA" because, after
1969 failing at the start, it skips both the CR and the LF before retrying.
1970 However, the pattern [\r\n]A does match that string, because it con-
1971 tains an explicit CR or LF reference, and so advances only by one char-
1972 acter after the first failure.
1973
1974 An explicit match for CR of LF is either a literal appearance of one of
1975 those characters, or one of the \r or \n escape sequences. Implicit
1976 matches such as [^X] do not count, nor does \s (which includes CR and
1977 LF in the characters that it matches).
1978
1979 Notwithstanding the above, anomalous effects may still occur when CRLF
1980 is a valid newline sequence and explicit \r or \n escapes appear in the
1981 pattern.
1982
1983 PCRE_NOTBOL
1984
1985 This option specifies that first character of the subject string is not
1986 the beginning of a line, so the circumflex metacharacter should not
1987 match before it. Setting this without PCRE_MULTILINE (at compile time)
1988 causes circumflex never to match. This option affects only the behav-
1989 iour of the circumflex metacharacter. It does not affect \A.
1990
1991 PCRE_NOTEOL
1992
1993 This option specifies that the end of the subject string is not the end
1994 of a line, so the dollar metacharacter should not match it nor (except
1995 in multiline mode) a newline immediately before it. Setting this with-
1996 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1997 option affects only the behaviour of the dollar metacharacter. It does
1998 not affect \Z or \z.
1999
2000 PCRE_NOTEMPTY
2001
2002 An empty string is not considered to be a valid match if this option is
2003 set. If there are alternatives in the pattern, they are tried. If all
2004 the alternatives match the empty string, the entire match fails. For
2005 example, if the pattern
2006
2007 a?b?
2008
2009 is applied to a string not beginning with "a" or "b", it matches the
2010 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2011 match is not valid, so PCRE searches further into the string for occur-
2012 rences of "a" or "b".
2013
2014 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
2015 cial case of a pattern match of the empty string within its split()
2016 function, and when using the /g modifier. It is possible to emulate
2017 Perl's behaviour after matching a null string by first trying the match
2018 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
2019 if that fails by advancing the starting offset (see below) and trying
2020 an ordinary match again. There is some code that demonstrates how to do
2021 this in the pcredemo.c sample program.
2022
2023 PCRE_NO_START_OPTIMIZE
2024
2025 There are a number of optimizations that pcre_exec() uses at the start
2026 of a match, in order to speed up the process. For example, if it is
2027 known that a match must start with a specific character, it searches
2028 the subject for that character, and fails immediately if it cannot find
2029 it, without actually running the main matching function. When callouts
2030 are in use, these optimizations can cause them to be skipped. This
2031 option disables the "start-up" optimizations, causing performance to
2032 suffer, but ensuring that the callouts do occur.
2033
2034 PCRE_NO_UTF8_CHECK
2035
2036 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2037 UTF-8 string is automatically checked when pcre_exec() is subsequently
2038 called. The value of startoffset is also checked to ensure that it
2039 points to the start of a UTF-8 character. There is a discussion about
2040 the validity of UTF-8 strings in the section on UTF-8 support in the
2041 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2042 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
2043 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
2044
2045 If you already know that your subject is valid, and you want to skip
2046 these checks for performance reasons, you can set the
2047 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2048 do this for the second and subsequent calls to pcre_exec() if you are
2049 making repeated calls to find all the matches in a single subject
2050 string. However, you should be sure that the value of startoffset
2051 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
2052 set, the effect of passing an invalid UTF-8 string as a subject, or a
2053 value of startoffset that does not point to the start of a UTF-8 char-
2054 acter, is undefined. Your program may crash.
2055
2056 PCRE_PARTIAL
2057
2058 This option turns on the partial matching feature. If the subject
2059 string fails to match the pattern, but at some point during the match-
2060 ing process the end of the subject was reached (that is, the subject
2061 partially matches the pattern and the failure to match occurred only
2062 because there were not enough subject characters), pcre_exec() returns
2063 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
2064 used, there are restrictions on what may appear in the pattern. These
2065 are discussed in the pcrepartial documentation.
2066
2067 The string to be matched by pcre_exec()
2068
2069 The subject string is passed to pcre_exec() as a pointer in subject, a
2070 length (in bytes) in length, and a starting byte offset in startoffset.
2071 In UTF-8 mode, the byte offset must point to the start of a UTF-8 char-
2072 acter. Unlike the pattern string, the subject may contain binary zero
2073 bytes. When the starting offset is zero, the search for a match starts
2074 at the beginning of the subject, and this is by far the most common
2075 case.
2076
2077 A non-zero starting offset is useful when searching for another match
2078 in the same subject by calling pcre_exec() again after a previous suc-
2079 cess. Setting startoffset differs from just passing over a shortened
2080 string and setting PCRE_NOTBOL in the case of a pattern that begins
2081 with any kind of lookbehind. For example, consider the pattern
2082
2083 \Biss\B
2084
2085 which finds occurrences of "iss" in the middle of words. (\B matches
2086 only if the current position in the subject is not a word boundary.)
2087 When applied to the string "Mississipi" the first call to pcre_exec()
2088 finds the first occurrence. If pcre_exec() is called again with just
2089 the remainder of the subject, namely "issipi", it does not match,
2090 because \B is always false at the start of the subject, which is deemed
2091 to be a word boundary. However, if pcre_exec() is passed the entire
2092 string again, but with startoffset set to 4, it finds the second occur-
2093 rence of "iss" because it is able to look behind the starting point to
2094 discover that it is preceded by a letter.
2095
2096 If a non-zero starting offset is passed when the pattern is anchored,
2097 one attempt to match at the given offset is made. This can only succeed
2098 if the pattern does not require the match to be at the start of the
2099 subject.
2100
2101 How pcre_exec() returns captured substrings
2102
2103 In general, a pattern matches a certain portion of the subject, and in
2104 addition, further substrings from the subject may be picked out by
2105 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2106 this is called "capturing" in what follows, and the phrase "capturing
2107 subpattern" is used for a fragment of a pattern that picks out a sub-
2108 string. PCRE supports several other kinds of parenthesized subpattern
2109 that do not cause substrings to be captured.
2110
2111 Captured substrings are returned to the caller via a vector of integers
2112 whose address is passed in ovector. The number of elements in the vec-
2113 tor is passed in ovecsize, which must be a non-negative number. Note:
2114 this argument is NOT the size of ovector in bytes.
2115
2116 The first two-thirds of the vector is used to pass back captured sub-
2117 strings, each substring using a pair of integers. The remaining third
2118 of the vector is used as workspace by pcre_exec() while matching cap-
2119 turing subpatterns, and is not available for passing back information.
2120 The number passed in ovecsize should always be a multiple of three. If
2121 it is not, it is rounded down.
2122
2123 When a match is successful, information about captured substrings is
2124 returned in pairs of integers, starting at the beginning of ovector,
2125 and continuing up to two-thirds of its length at the most. The first
2126 element of each pair is set to the byte offset of the first character
2127 in a substring, and the second is set to the byte offset of the first
2128 character after the end of a substring. Note: these values are always
2129 byte offsets, even in UTF-8 mode. They are not character counts.
2130
2131 The first pair of integers, ovector[0] and ovector[1], identify the
2132 portion of the subject string matched by the entire pattern. The next
2133 pair is used for the first capturing subpattern, and so on. The value
2134 returned by pcre_exec() is one more than the highest numbered pair that
2135 has been set. For example, if two substrings have been captured, the
2136 returned value is 3. If there are no capturing subpatterns, the return
2137 value from a successful match is 1, indicating that just the first pair
2138 of offsets has been set.
2139
2140 If a capturing subpattern is matched repeatedly, it is the last portion
2141 of the string that it matched that is returned.
2142
2143 If the vector is too small to hold all the captured substring offsets,
2144 it is used as far as possible (up to two-thirds of its length), and the
2145 function returns a value of zero. If the substring offsets are not of
2146 interest, pcre_exec() may be called with ovector passed as NULL and
2147 ovecsize as zero. However, if the pattern contains back references and
2148 the ovector is not big enough to remember the related substrings, PCRE
2149 has to get additional memory for use during matching. Thus it is usu-
2150 ally advisable to supply an ovector.
2151
2152 The pcre_info() function can be used to find out how many capturing
2153 subpatterns there are in a compiled pattern. The smallest size for
2154 ovector that will allow for n captured substrings, in addition to the
2155 offsets of the substring matched by the whole pattern, is (n+1)*3.
2156
2157 It is possible for capturing subpattern number n+1 to match some part
2158 of the subject when subpattern n has not been used at all. For example,
2159 if the string "abc" is matched against the pattern (a|(z))(bc) the
2160 return from the function is 4, and subpatterns 1 and 3 are matched, but
2161 2 is not. When this happens, both values in the offset pairs corre-
2162 sponding to unused subpatterns are set to -1.
2163
2164 Offset values that correspond to unused subpatterns at the end of the
2165 expression are also set to -1. For example, if the string "abc" is
2166 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2167 matched. The return from the function is 2, because the highest used
2168 capturing subpattern number is 1. However, you can refer to the offsets
2169 for the second and third capturing subpatterns if you wish (assuming
2170 the vector is large enough, of course).
2171
2172 Some convenience functions are provided for extracting the captured
2173 substrings as separate strings. These are described below.
2174
2175 Error return values from pcre_exec()
2176
2177 If pcre_exec() fails, it returns a negative number. The following are
2178 defined in the header file:
2179
2180 PCRE_ERROR_NOMATCH (-1)
2181
2182 The subject string did not match the pattern.
2183
2184 PCRE_ERROR_NULL (-2)
2185
2186 Either code or subject was passed as NULL, or ovector was NULL and
2187 ovecsize was not zero.
2188
2189 PCRE_ERROR_BADOPTION (-3)
2190
2191 An unrecognized bit was set in the options argument.
2192
2193 PCRE_ERROR_BADMAGIC (-4)
2194
2195 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2196 to catch the case when it is passed a junk pointer and to detect when a
2197 pattern that was compiled in an environment of one endianness is run in
2198 an environment with the other endianness. This is the error that PCRE
2199 gives when the magic number is not present.
2200
2201 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2202
2203 While running the pattern match, an unknown item was encountered in the
2204 compiled pattern. This error could be caused by a bug in PCRE or by
2205 overwriting of the compiled pattern.
2206
2207 PCRE_ERROR_NOMEMORY (-6)
2208
2209 If a pattern contains back references, but the ovector that is passed
2210 to pcre_exec() is not big enough to remember the referenced substrings,
2211 PCRE gets a block of memory at the start of matching to use for this
2212 purpose. If the call via pcre_malloc() fails, this error is given. The
2213 memory is automatically freed at the end of matching.
2214
2215 PCRE_ERROR_NOSUBSTRING (-7)
2216
2217 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2218 and pcre_get_substring_list() functions (see below). It is never
2219 returned by pcre_exec().
2220
2221 PCRE_ERROR_MATCHLIMIT (-8)
2222
2223 The backtracking limit, as specified by the match_limit field in a
2224 pcre_extra structure (or defaulted) was reached. See the description
2225 above.
2226
2227 PCRE_ERROR_CALLOUT (-9)
2228
2229 This error is never generated by pcre_exec() itself. It is provided for
2230 use by callout functions that want to yield a distinctive error code.
2231 See the pcrecallout documentation for details.
2232
2233 PCRE_ERROR_BADUTF8 (-10)
2234
2235 A string that contains an invalid UTF-8 byte sequence was passed as a
2236 subject.
2237
2238 PCRE_ERROR_BADUTF8_OFFSET (-11)
2239
2240 The UTF-8 byte sequence that was passed as a subject was valid, but the
2241 value of startoffset did not point to the beginning of a UTF-8 charac-
2242 ter.
2243
2244 PCRE_ERROR_PARTIAL (-12)
2245
2246 The subject string did not match, but it did match partially. See the
2247 pcrepartial documentation for details of partial matching.
2248
2249 PCRE_ERROR_BADPARTIAL (-13)
2250
2251 The PCRE_PARTIAL option was used with a compiled pattern containing
2252 items that are not supported for partial matching. See the pcrepartial
2253 documentation for details of partial matching.
2254
2255 PCRE_ERROR_INTERNAL (-14)
2256
2257 An unexpected internal error has occurred. This error could be caused
2258 by a bug in PCRE or by overwriting of the compiled pattern.
2259
2260 PCRE_ERROR_BADCOUNT (-15)
2261
2262 This error is given if the value of the ovecsize argument is negative.
2263
2264 PCRE_ERROR_RECURSIONLIMIT (-21)
2265
2266 The internal recursion limit, as specified by the match_limit_recursion
2267 field in a pcre_extra structure (or defaulted) was reached. See the
2268 description above.
2269
2270 PCRE_ERROR_BADNEWLINE (-23)
2271
2272 An invalid combination of PCRE_NEWLINE_xxx options was given.
2273
2274 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2275
2276
2277 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2278
2279 int pcre_copy_substring(const char *subject, int *ovector,
2280 int stringcount, int stringnumber, char *buffer,
2281 int buffersize);
2282
2283 int pcre_get_substring(const char *subject, int *ovector,
2284 int stringcount, int stringnumber,
2285 const char **stringptr);
2286
2287 int pcre_get_substring_list(const char *subject,
2288 int *ovector, int stringcount, const char ***listptr);
2289
2290 Captured substrings can be accessed directly by using the offsets
2291 returned by pcre_exec() in ovector. For convenience, the functions
2292 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2293 string_list() are provided for extracting captured substrings as new,
2294 separate, zero-terminated strings. These functions identify substrings
2295 by number. The next section describes functions for extracting named
2296 substrings.
2297
2298 A substring that contains a binary zero is correctly extracted and has
2299 a further zero added on the end, but the result is not, of course, a C
2300 string. However, you can process such a string by referring to the
2301 length that is returned by pcre_copy_substring() and pcre_get_sub-
2302 string(). Unfortunately, the interface to pcre_get_substring_list() is
2303 not adequate for handling strings containing binary zeros, because the
2304 end of the final string is not independently indicated.
2305
2306 The first three arguments are the same for all three of these func-
2307 tions: subject is the subject string that has just been successfully
2308 matched, ovector is a pointer to the vector of integer offsets that was
2309 passed to pcre_exec(), and stringcount is the number of substrings that
2310 were captured by the match, including the substring that matched the
2311 entire regular expression. This is the value returned by pcre_exec() if
2312 it is greater than zero. If pcre_exec() returned zero, indicating that
2313 it ran out of space in ovector, the value passed as stringcount should
2314 be the number of elements in the vector divided by three.
2315
2316 The functions pcre_copy_substring() and pcre_get_substring() extract a
2317 single substring, whose number is given as stringnumber. A value of
2318 zero extracts the substring that matched the entire pattern, whereas
2319 higher values extract the captured substrings. For pcre_copy_sub-
2320 string(), the string is placed in buffer, whose length is given by
2321 buffersize, while for pcre_get_substring() a new block of memory is
2322 obtained via pcre_malloc, and its address is returned via stringptr.
2323 The yield of the function is the length of the string, not including
2324 the terminating zero, or one of these error codes:
2325
2326 PCRE_ERROR_NOMEMORY (-6)
2327
2328 The buffer was too small for pcre_copy_substring(), or the attempt to
2329 get memory failed for pcre_get_substring().
2330
2331 PCRE_ERROR_NOSUBSTRING (-7)
2332
2333 There is no substring whose number is stringnumber.
2334
2335 The pcre_get_substring_list() function extracts all available sub-
2336 strings and builds a list of pointers to them. All this is done in a
2337 single block of memory that is obtained via pcre_malloc. The address of
2338 the memory block is returned via listptr, which is also the start of
2339 the list of string pointers. The end of the list is marked by a NULL
2340 pointer. The yield of the function is zero if all went well, or the
2341 error code
2342
2343 PCRE_ERROR_NOMEMORY (-6)
2344
2345 if the attempt to get the memory block failed.
2346
2347 When any of these functions encounter a substring that is unset, which
2348 can happen when capturing subpattern number n+1 matches some part of
2349 the subject, but subpattern n has not been used at all, they return an
2350 empty string. This can be distinguished from a genuine zero-length sub-
2351 string by inspecting the appropriate offset in ovector, which is nega-
2352 tive for unset substrings.
2353
2354 The two convenience functions pcre_free_substring() and pcre_free_sub-
2355 string_list() can be used to free the memory returned by a previous
2356 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2357 tively. They do nothing more than call the function pointed to by
2358 pcre_free, which of course could be called directly from a C program.
2359 However, PCRE is used in some situations where it is linked via a spe-
2360 cial interface to another programming language that cannot use
2361 pcre_free directly; it is for these cases that the functions are pro-
2362 vided.
2363
2364
2365 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2366
2367 int pcre_get_stringnumber(const pcre *code,
2368 const char *name);
2369
2370 int pcre_copy_named_substring(const pcre *code,
2371 const char *subject, int *ovector,
2372 int stringcount, const char *stringname,
2373 char *buffer, int buffersize);
2374
2375 int pcre_get_named_substring(const pcre *code,
2376 const char *subject, int *ovector,
2377 int stringcount, const char *stringname,
2378 const char **stringptr);
2379
2380 To extract a substring by name, you first have to find associated num-
2381 ber. For example, for this pattern
2382
2383 (a+)b(?<xxx>\d+)...
2384
2385 the number of the subpattern called "xxx" is 2. If the name is known to
2386 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2387 name by calling pcre_get_stringnumber(). The first argument is the com-
2388 piled pattern, and the second is the name. The yield of the function is
2389 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2390 subpattern of that name.
2391
2392 Given the number, you can extract the substring directly, or use one of
2393 the functions described in the previous section. For convenience, there
2394 are also two functions that do the whole job.
2395
2396 Most of the arguments of pcre_copy_named_substring() and
2397 pcre_get_named_substring() are the same as those for the similarly
2398 named functions that extract by number. As these are described in the
2399 previous section, they are not re-described here. There are just two
2400 differences:
2401
2402 First, instead of a substring number, a substring name is given. Sec-
2403 ond, there is an extra argument, given at the start, which is a pointer
2404 to the compiled pattern. This is needed in order to gain access to the
2405 name-to-number translation table.
2406
2407 These functions call pcre_get_stringnumber(), and if it succeeds, they
2408 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2409 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2410 behaviour may not be what you want (see the next section).
2411
2412 Warning: If the pattern uses the "(?|" feature to set up multiple sub-
2413 patterns with the same number, you cannot use names to distinguish
2414 them, because names are not included in the compiled code. The matching
2415 process uses only numbers.
2416
2417
2418 DUPLICATE SUBPATTERN NAMES
2419
2420 int pcre_get_stringtable_entries(const pcre *code,
2421 const char *name, char **first, char **last);
2422
2423 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2424 subpatterns are not required to be unique. Normally, patterns with
2425 duplicate names are such that in any one match, only one of the named
2426 subpatterns participates. An example is shown in the pcrepattern docu-
2427 mentation.
2428
2429 When duplicates are present, pcre_copy_named_substring() and
2430 pcre_get_named_substring() return the first substring corresponding to
2431 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2432 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2433 function returns one of the numbers that are associated with the name,
2434 but it is not defined which it is.
2435
2436 If you want to get full details of all captured substrings for a given
2437 name, you must use the pcre_get_stringtable_entries() function. The
2438 first argument is the compiled pattern, and the second is the name. The
2439 third and fourth are pointers to variables which are updated by the
2440 function. After it has run, they point to the first and last entries in
2441 the name-to-number table for the given name. The function itself
2442 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2443 there are none. The format of the table is described above in the sec-
2444 tion entitled Information about a pattern. Given all the relevant
2445 entries for the name, you can extract each of their numbers, and hence
2446 the captured data, if any.
2447
2448
2449 FINDING ALL POSSIBLE MATCHES
2450
2451 The traditional matching function uses a similar algorithm to Perl,
2452 which stops when it finds the first match, starting at a given point in
2453 the subject. If you want to find all possible matches, or the longest
2454 possible match, consider using the alternative matching function (see
2455 below) instead. If you cannot use the alternative function, but still
2456 need to find all possible matches, you can kludge it up by making use
2457 of the callout facility, which is described in the pcrecallout documen-
2458 tation.
2459
2460 What you have to do is to insert a callout right at the end of the pat-
2461 tern. When your callout function is called, extract and save the cur-
2462 rent matched substring. Then return 1, which forces pcre_exec() to
2463 backtrack and try other alternatives. Ultimately, when it runs out of
2464 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2465
2466
2467 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2468
2469 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2470 const char *subject, int length, int startoffset,
2471 int options, int *ovector, int ovecsize,
2472 int *workspace, int wscount);
2473
2474 The function pcre_dfa_exec() is called to match a subject string
2475 against a compiled pattern, using a matching algorithm that scans the
2476 subject string just once, and does not backtrack. This has different
2477 characteristics to the normal algorithm, and is not compatible with
2478 Perl. Some of the features of PCRE patterns are not supported. Never-
2479 theless, there are times when this kind of matching can be useful. For
2480 a discussion of the two matching algorithms, see the pcrematching docu-
2481 mentation.
2482
2483 The arguments for the pcre_dfa_exec() function are the same as for
2484 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2485 ent way, and this is described below. The other common arguments are
2486 used in the same way as for pcre_exec(), so their description is not
2487 repeated here.
2488
2489 The two additional arguments provide workspace for the function. The
2490 workspace vector should contain at least 20 elements. It is used for
2491 keeping track of multiple paths through the pattern tree. More
2492 workspace will be needed for patterns and subjects where there are a
2493 lot of potential matches.
2494
2495 Here is an example of a simple call to pcre_dfa_exec():
2496
2497 int rc;
2498 int ovector[10];
2499 int wspace[20];
2500 rc = pcre_dfa_exec(
2501 re, /* result of pcre_compile() */
2502 NULL, /* we didn't study the pattern */
2503 "some string", /* the subject string */
2504 11, /* the length of the subject string */
2505 0, /* start at offset 0 in the subject */
2506 0, /* default options */
2507 ovector, /* vector of integers for substring information */
2508 10, /* number of elements (NOT size in bytes) */
2509 wspace, /* working space vector */
2510 20); /* number of elements (NOT size in bytes) */
2511
2512 Option bits for pcre_dfa_exec()
2513
2514 The unused bits of the options argument for pcre_dfa_exec() must be
2515 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2516 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2517 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2518 three of these are the same as for pcre_exec(), so their description is
2519 not repeated here.
2520
2521 PCRE_PARTIAL
2522
2523 This has the same general effect as it does for pcre_exec(), but the
2524 details are slightly different. When PCRE_PARTIAL is set for
2525 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2526 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2527 been no complete matches, but there is still at least one matching pos-
2528 sibility. The portion of the string that provided the partial match is
2529 set as the first matching string.
2530
2531 PCRE_DFA_SHORTEST
2532
2533 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2534 stop as soon as it has found one match. Because of the way the alterna-
2535 tive algorithm works, this is necessarily the shortest possible match
2536 at the first possible matching point in the subject string.
2537
2538 PCRE_DFA_RESTART
2539
2540 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2541 returns a partial match, it is possible to call it again, with addi-
2542 tional subject characters, and have it continue with the same match.
2543 The PCRE_DFA_RESTART option requests this action; when it is set, the
2544 workspace and wscount options must reference the same vector as before
2545 because data about the match so far is left in them after a partial
2546 match. There is more discussion of this facility in the pcrepartial
2547 documentation.
2548
2549 Successful returns from pcre_dfa_exec()
2550
2551 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2552 string in the subject. Note, however, that all the matches from one run
2553 of the function start at the same point in the subject. The shorter
2554 matches are all initial substrings of the longer matches. For example,
2555 if the pattern
2556
2557 <.*>
2558
2559 is matched against the string
2560
2561 This is <something> <something else> <something further> no more
2562
2563 the three matched strings are
2564
2565 <something>
2566 <something> <something else>
2567 <something> <something else> <something further>
2568
2569 On success, the yield of the function is a number greater than zero,
2570 which is the number of matched substrings. The substrings themselves
2571 are returned in ovector. Each string uses two elements; the first is
2572 the offset to the start, and the second is the offset to the end. In
2573 fact, all the strings have the same start offset. (Space could have
2574 been saved by giving this only once, but it was decided to retain some
2575 compatibility with the way pcre_exec() returns data, even though the
2576 meaning of the strings is different.)
2577
2578 The strings are returned in reverse order of length; that is, the long-
2579 est matching string is given first. If there were too many matches to
2580 fit into ovector, the yield of the function is zero, and the vector is
2581 filled with the longest matches.
2582
2583 Error returns from pcre_dfa_exec()
2584
2585 The pcre_dfa_exec() function returns a negative number when it fails.
2586 Many of the errors are the same as for pcre_exec(), and these are
2587 described above. There are in addition the following errors that are
2588 specific to pcre_dfa_exec():
2589
2590 PCRE_ERROR_DFA_UITEM (-16)
2591
2592 This return is given if pcre_dfa_exec() encounters an item in the pat-
2593 tern that it does not support, for instance, the use of \C or a back
2594 reference.
2595
2596 PCRE_ERROR_DFA_UCOND (-17)
2597
2598 This return is given if pcre_dfa_exec() encounters a condition item
2599 that uses a back reference for the condition, or a test for recursion
2600 in a specific group. These are not supported.
2601
2602 PCRE_ERROR_DFA_UMLIMIT (-18)
2603
2604 This return is given if pcre_dfa_exec() is called with an extra block
2605 that contains a setting of the match_limit field. This is not supported
2606 (it is meaningless).
2607
2608 PCRE_ERROR_DFA_WSSIZE (-19)
2609
2610 This return is given if pcre_dfa_exec() runs out of space in the
2611 workspace vector.
2612
2613 PCRE_ERROR_DFA_RECURSE (-20)
2614
2615 When a recursive subpattern is processed, the matching function calls
2616 itself recursively, using private vectors for ovector and workspace.
2617 This error is given if the output vector is not large enough. This
2618 should be extremely rare, as a vector of size 1000 is used.
2619
2620
2621 SEE ALSO
2622
2623 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2624 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2625
2626
2627 AUTHOR
2628
2629 Philip Hazel
2630 University Computing Service
2631 Cambridge CB2 3QH, England.
2632
2633
2634 REVISION
2635
2636 Last updated: 17 March 2009
2637 Copyright (c) 1997-2009 University of Cambridge.
2638 ------------------------------------------------------------------------------
2639
2640
2641 PCRECALLOUT(3) PCRECALLOUT(3)
2642
2643
2644 NAME
2645 PCRE - Perl-compatible regular expressions
2646
2647
2648 PCRE CALLOUTS
2649
2650 int (*pcre_callout)(pcre_callout_block *);
2651
2652 PCRE provides a feature called "callout", which is a means of temporar-
2653 ily passing control to the caller of PCRE in the middle of pattern
2654 matching. The caller of PCRE provides an external function by putting
2655 its entry point in the global variable pcre_callout. By default, this
2656 variable contains NULL, which disables all calling out.
2657
2658 Within a regular expression, (?C) indicates the points at which the
2659 external function is to be called. Different callout points can be
2660 identified by putting a number less than 256 after the letter C. The
2661 default value is zero. For example, this pattern has two callout
2662 points:
2663
2664 (?C1)abc(?C2)def
2665
2666 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2667 called, PCRE automatically inserts callouts, all with number 255,
2668 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2669 used with the pattern
2670
2671 A(\d{2}|--)
2672
2673 it is processed as if it were
2674
2675 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2676
2677 Notice that there is a callout before and after each parenthesis and
2678 alternation bar. Automatic callouts can be used for tracking the
2679 progress of pattern matching. The pcretest command has an option that
2680 sets automatic callouts; when it is used, the output indicates how the
2681 pattern is matched. This is useful information when you are trying to
2682 optimize the performance of a particular pattern.
2683
2684
2685 MISSING CALLOUTS
2686
2687 You should be aware that, because of optimizations in the way PCRE
2688 matches patterns by default, callouts sometimes do not happen. For
2689 example, if the pattern is
2690
2691 ab(?C4)cd
2692
2693 PCRE knows that any matching string must contain the letter "d". If the
2694 subject string is "abyz", the lack of "d" means that matching doesn't
2695 ever start, and the callout is never reached. However, with "abyd",
2696 though the result is still no match, the callout is obeyed.
2697
2698 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2699 MIZE option to pcre_exec() or pcre_dfa_exec(). This slows down the
2700 matching process, but does ensure that callouts such as the example
2701 above are obeyed.
2702
2703
2704 THE CALLOUT INTERFACE
2705
2706 During matching, when PCRE reaches a callout point, the external func-
2707 tion defined by pcre_callout is called (if it is set). This applies to
2708 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2709 only argument to the callout function is a pointer to a pcre_callout
2710 block. This structure contains the following fields:
2711
2712 int version;
2713 int callout_number;
2714 int *offset_vector;
2715 const char *subject;
2716 int subject_length;
2717 int start_match;
2718 int current_position;
2719 int capture_top;
2720 int capture_last;
2721 void *callout_data;
2722 int pattern_position;
2723 int next_item_length;
2724
2725 The version field is an integer containing the version number of the
2726 block format. The initial version was 0; the current version is 1. The
2727 version number will change again in future if additional fields are
2728 added, but the intention is never to remove any of the existing fields.
2729
2730 The callout_number field contains the number of the callout, as com-
2731 piled into the pattern (that is, the number after ?C for manual call-
2732 outs, and 255 for automatically generated callouts).
2733
2734 The offset_vector field is a pointer to the vector of offsets that was
2735 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2736 pcre_exec() is used, the contents can be inspected in order to extract
2737 substrings that have been matched so far, in the same way as for
2738 extracting substrings after a match has completed. For pcre_dfa_exec()
2739 this field is not useful.
2740
2741 The subject and subject_length fields contain copies of the values that
2742 were passed to pcre_exec().
2743
2744 The start_match field normally contains the offset within the subject
2745 at which the current match attempt started. However, if the escape
2746 sequence \K has been encountered, this value is changed to reflect the
2747 modified starting point. If the pattern is not anchored, the callout
2748 function may be called several times from the same point in the pattern
2749 for different starting points in the subject.
2750
2751 The current_position field contains the offset within the subject of
2752 the current match pointer.
2753
2754 When the pcre_exec() function is used, the capture_top field contains
2755 one more than the number of the highest numbered captured substring so
2756 far. If no substrings have been captured, the value of capture_top is
2757 one. This is always the case when pcre_dfa_exec() is used, because it
2758 does not support captured substrings.
2759
2760 The capture_last field contains the number of the most recently cap-
2761 tured substring. If no substrings have been captured, its value is -1.
2762 This is always the case when pcre_dfa_exec() is used.
2763
2764 The callout_data field contains a value that is passed to pcre_exec()
2765 or pcre_dfa_exec() specifically so that it can be passed back in call-
2766 outs. It is passed in the pcre_callout field of the pcre_extra data
2767 structure. If no such data was passed, the value of callout_data in a
2768 pcre_callout block is NULL. There is a description of the pcre_extra
2769 structure in the pcreapi documentation.
2770
2771 The pattern_position field is present from version 1 of the pcre_call-
2772 out structure. It contains the offset to the next item to be matched in
2773 the pattern string.
2774
2775 The next_item_length field is present from version 1 of the pcre_call-
2776 out structure. It contains the length of the next item to be matched in
2777 the pattern string. When the callout immediately precedes an alterna-
2778 tion bar, a closing parenthesis, or the end of the pattern, the length
2779 is zero. When the callout precedes an opening parenthesis, the length
2780 is that of the entire subpattern.
2781
2782 The pattern_position and next_item_length fields are intended to help
2783 in distinguishing between different automatic callouts, which all have
2784 the same callout number. However, they are set for all callouts.
2785
2786
2787 RETURN VALUES
2788
2789 The external callout function returns an integer to PCRE. If the value
2790 is zero, matching proceeds as normal. If the value is greater than
2791 zero, matching fails at the current point, but the testing of other
2792 matching possibilities goes ahead, just as if a lookahead assertion had
2793 failed. If the value is less than zero, the match is abandoned, and
2794 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2795
2796 Negative values should normally be chosen from the set of
2797 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2798 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2799 reserved for use by callout functions; it will never be used by PCRE
2800 itself.
2801
2802
2803 AUTHOR
2804
2805 Philip Hazel
2806 University Computing Service
2807 Cambridge CB2 3QH, England.
2808
2809
2810 REVISION
2811
2812 Last updated: 15 March 2009
2813 Copyright (c) 1997-2009 University of Cambridge.
2814 ------------------------------------------------------------------------------
2815
2816
2817 PCRECOMPAT(3) PCRECOMPAT(3)
2818
2819
2820 NAME
2821 PCRE - Perl-compatible regular expressions
2822
2823
2824 DIFFERENCES BETWEEN PCRE AND PERL
2825
2826 This document describes the differences in the ways that PCRE and Perl
2827 handle regular expressions. The differences described here are mainly
2828 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2829 some features that are expected to be in the forthcoming Perl 5.10.
2830
2831 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2832 of what it does have are given in the section on UTF-8 support in the
2833 main pcre page.
2834
2835 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2836 permits them, but they do not mean what you might think. For example,
2837 (?!a){3} does not assert that the next three characters are not "a". It
2838 just asserts that the next character is not "a" three times.
2839
2840 3. Capturing subpatterns that occur inside negative lookahead asser-
2841 tions are counted, but their entries in the offsets vector are never
2842 set. Perl sets its numerical variables from any such patterns that are
2843 matched before the assertion fails to match something (thereby succeed-
2844 ing), but only if the negative lookahead assertion contains just one
2845 branch.
2846
2847 4. Though binary zero characters are supported in the subject string,
2848 they are not allowed in a pattern string because it is passed as a nor-
2849 mal C string, terminated by zero. The escape sequence \0 can be used in
2850 the pattern to represent a binary zero.
2851
2852 5. The following Perl escape sequences are not supported: \l, \u, \L,
2853 \U, and \N. In fact these are implemented by Perl's general string-han-
2854 dling and are not part of its pattern matching engine. If any of these
2855 are encountered by PCRE, an error is generated.
2856
2857 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2858 is built with Unicode character property support. The properties that
2859 can be tested with \p and \P are limited to the general category prop-
2860 erties such as Lu and Nd, script names such as Greek or Han, and the
2861 derived properties Any and L&.
2862
2863 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2864 ters in between are treated as literals. This is slightly different
2865 from Perl in that $ and @ are also handled as literals inside the
2866 quotes. In Perl, they cause variable interpolation (but of course PCRE
2867 does not have variables). Note the following examples:
2868
2869 Pattern PCRE matches Perl matches
2870
2871 \Qabc$xyz\E abc$xyz abc followed by the
2872 contents of $xyz
2873 \Qabc\$xyz\E abc\$xyz abc\$xyz
2874 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2875
2876 The \Q...\E sequence is recognized both inside and outside character
2877 classes.
2878
2879 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2880 constructions. However, there is support for recursive patterns. This
2881 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2882 "callout" feature allows an external function to be called during pat-
2883 tern matching. See the pcrecallout documentation for details.
2884
2885 9. Subpatterns that are called recursively or as "subroutines" are
2886 always treated as atomic groups in PCRE. This is like Python, but
2887 unlike Perl.
2888
2889 10. There are some differences that are concerned with the settings of
2890 captured strings when part of a pattern is repeated. For example,
2891 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2892 unset, but in PCRE it is set to "b".
2893
2894 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2895 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2896 the forms without an argument. PCRE does not support (*MARK). If
2897 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2898 ture group; this is different to Perl.
2899
2900 12. PCRE provides some extensions to the Perl regular expression facil-
2901 ities. Perl 5.10 will include new features that are not in earlier
2902 versions, some of which (such as named parentheses) have been in PCRE
2903 for some time. This list is with respect to Perl 5.10:
2904
2905 (a) Although lookbehind assertions must match fixed length strings,
2906 each alternative branch of a lookbehind assertion can match a different
2907 length of string. Perl requires them all to have the same length.
2908
2909 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2910 meta-character matches only at the very end of the string.
2911
2912 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2913 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2914 ignored. (Perl can be made to issue a warning.)
2915
2916 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2917 fiers is inverted, that is, by default they are not greedy, but if fol-
2918 lowed by a question mark they are.
2919
2920 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2921 tried only at the first matching position in the subject string.
2922
2923 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2924 TURE options for pcre_exec() have no Perl equivalents.
2925
2926 (g) The \R escape sequence can be restricted to match only CR, LF, or
2927 CRLF by the PCRE_BSR_ANYCRLF option.
2928
2929 (h) The callout facility is PCRE-specific.
2930
2931 (i) The partial matching facility is PCRE-specific.
2932
2933 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2934 even on different hosts that have the other endianness.
2935
2936 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2937 different way and is not Perl-compatible.
2938
2939 (l) PCRE recognizes some special sequences such as (*CR) at the start
2940 of a pattern that set overall options that cannot be changed within the
2941 pattern.
2942
2943
2944 AUTHOR
2945
2946 Philip Hazel
2947 University Computing Service
2948 Cambridge CB2 3QH, England.
2949
2950
2951 REVISION
2952
2953 Last updated: 11 September 2007
2954 Copyright (c) 1997-2007 University of Cambridge.
2955 ------------------------------------------------------------------------------
2956
2957
2958 PCREPATTERN(3) PCREPATTERN(3)
2959
2960
2961 NAME
2962 PCRE - Perl-compatible regular expressions
2963
2964
2965 PCRE REGULAR EXPRESSION DETAILS
2966
2967 The syntax and semantics of the regular expressions that are supported
2968 by PCRE are described in detail below. There is a quick-reference syn-
2969 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
2970 semantics as closely as it can. PCRE also supports some alternative
2971 regular expression syntax (which does not conflict with the Perl syn-
2972 tax) in order to provide some compatibility with regular expressions in
2973 Python, .NET, and Oniguruma.
2974
2975 Perl's regular expressions are described in its own documentation, and
2976 regular expressions in general are covered in a number of books, some
2977 of which have copious examples. Jeffrey Friedl's "Mastering Regular
2978 Expressions", published by O'Reilly, covers regular expressions in
2979 great detail. This description of PCRE's regular expressions is
2980 intended as reference material.
2981
2982 The original operation of PCRE was on strings of one-byte characters.
2983 However, there is now also support for UTF-8 character strings. To use
2984 this, you must build PCRE to include UTF-8 support, and then call
2985 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2986 matching is mentioned in several places below. There is also a summary
2987 of UTF-8 features in the section on UTF-8 support in the main pcre
2988 page.
2989
2990 The remainder of this document discusses the patterns that are sup-
2991 ported by PCRE when its main matching function, pcre_exec(), is used.
2992 From release 6.0, PCRE offers a second matching function,
2993 pcre_dfa_exec(), which matches using a different algorithm that is not
2994 Perl-compatible. Some of the features discussed below are not available
2995 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2996 alternative function, and how it differs from the normal function, are
2997 discussed in the pcrematching page.
2998
2999
3000 NEWLINE CONVENTIONS
3001
3002 PCRE supports five different conventions for indicating line breaks in
3003 strings: a single CR (carriage return) character, a single LF (line-
3004 feed) character, the two-character sequence CRLF, any of the three pre-
3005 ceding, or any Unicode newline sequence. The pcreapi page has further
3006 discussion about newlines, and shows how to set the newline convention
3007 in the options arguments for the compiling and matching functions.
3008
3009 It is also possible to specify a newline convention by starting a pat-
3010 tern string with one of the following five sequences:
3011
3012 (*CR) carriage return
3013 (*LF) linefeed
3014 (*CRLF) carriage return, followed by linefeed
3015 (*ANYCRLF) any of the three above
3016 (*ANY) all Unicode newline sequences
3017
3018 These override the default and the options given to pcre_compile(). For
3019 example, on a Unix system where LF is the default newline sequence, the
3020 pattern
3021
3022 (*CR)a.b
3023
3024 changes the convention to CR. That pattern matches "a\nb" because LF is
3025 no longer a newline. Note that these special settings, which are not
3026 Perl-compatible, are recognized only at the very start of a pattern,
3027 and that they must be in upper case. If more than one of them is
3028 present, the last one is used.
3029
3030 The newline convention does not affect what the \R escape sequence
3031 matches. By default, this is any Unicode newline sequence, for Perl
3032 compatibility. However, this can be changed; see the description of \R
3033 in the section entitled "Newline sequences" below. A change of \R set-
3034 ting can be combined with a change of newline convention.
3035
3036
3037 CHARACTERS AND METACHARACTERS
3038
3039 A regular expression is a pattern that is matched against a subject
3040 string from left to right. Most characters stand for themselves in a
3041 pattern, and match the corresponding characters in the subject. As a
3042 trivial example, the pattern
3043
3044 The quick brown fox
3045
3046 matches a portion of a subject string that is identical to itself. When
3047 caseless matching is specified (the PCRE_CASELESS option), letters are
3048 matched independently of case. In UTF-8 mode, PCRE always understands
3049 the concept of case for characters whose values are less than 128, so
3050 caseless matching is always possible. For characters with higher val-
3051 ues, the concept of case is supported if PCRE is compiled with Unicode
3052 property support, but not otherwise. If you want to use caseless
3053 matching for characters 128 and above, you must ensure that PCRE is
3054 compiled with Unicode property support as well as with UTF-8 support.
3055
3056 The power of regular expressions comes from the ability to include
3057 alternatives and repetitions in the pattern. These are encoded in the
3058 pattern by the use of metacharacters, which do not stand for themselves
3059 but instead are interpreted in some special way.
3060
3061 There are two different sets of metacharacters: those that are recog-
3062 nized anywhere in the pattern except within square brackets, and those
3063 that are recognized within square brackets. Outside square brackets,
3064 the metacharacters are as follows:
3065
3066 \ general escape character with several uses
3067 ^ assert start of string (or line, in multiline mode)
3068 $ assert end of string (or line, in multiline mode)
3069 . match any character except newline (by default)
3070 [ start character class definition
3071 | start of alternative branch
3072 ( start subpattern
3073 ) end subpattern
3074 ? extends the meaning of (
3075 also 0 or 1 quantifier
3076 also quantifier minimizer
3077 * 0 or more quantifier
3078 + 1 or more quantifier
3079 also "possessive quantifier"
3080 { start min/max quantifier
3081
3082 Part of a pattern that is in square brackets is called a "character
3083 class". In a character class the only metacharacters are:
3084
3085 \ general escape character
3086 ^ negate the class, but only if the first character
3087 - indicates character range
3088 [ POSIX character class (only if followed by POSIX
3089 syntax)
3090 ] terminates the character class
3091
3092 The following sections describe the use of each of the metacharacters.
3093
3094
3095 BACKSLASH
3096
3097 The backslash character has several uses. Firstly, if it is followed by
3098 a non-alphanumeric character, it takes away any special meaning that
3099 character may have. This use of backslash as an escape character
3100 applies both inside and outside character classes.
3101
3102 For example, if you want to match a * character, you write \* in the
3103 pattern. This escaping action applies whether or not the following
3104 character would otherwise be interpreted as a metacharacter, so it is
3105 always safe to precede a non-alphanumeric with backslash to specify
3106 that it stands for itself. In particular, if you want to match a back-
3107 slash, you write \\.
3108
3109 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3110 the pattern (other than in a character class) and characters between a
3111 # outside a character class and the next newline are ignored. An escap-
3112 ing backslash can be used to include a whitespace or # character as
3113 part of the pattern.
3114
3115 If you want to remove the special meaning from a sequence of charac-
3116 ters, you can do so by putting them between \Q and \E. This is differ-
3117 ent from Perl in that $ and @ are handled as literals in \Q...\E
3118 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3119 tion. Note the following examples:
3120
3121 Pattern PCRE matches Perl matches
3122
3123 \Qabc$xyz\E abc$xyz abc followed by the
3124 contents of $xyz
3125 \Qabc\$xyz\E abc\$xyz abc\$xyz
3126 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3127
3128 The \Q...\E sequence is recognized both inside and outside character
3129 classes.
3130
3131 Non-printing characters
3132
3133 A second use of backslash provides a way of encoding non-printing char-
3134 acters in patterns in a visible manner. There is no restriction on the
3135 appearance of non-printing characters, apart from the binary zero that
3136 terminates a pattern, but when a pattern is being prepared by text
3137 editing, it is usually easier to use one of the following escape
3138 sequences than the binary character it represents:
3139
3140 \a alarm, that is, the BEL character (hex 07)
3141 \cx "control-x", where x is any character
3142 \e escape (hex 1B)
3143 \f formfeed (hex 0C)
3144 \n linefeed (hex 0A)
3145 \r carriage return (hex 0D)
3146 \t tab (hex 09)
3147 \ddd character with octal code ddd, or backreference
3148 \xhh character with hex code hh
3149 \x{hhh..} character with hex code hhh..
3150
3151 The precise effect of \cx is as follows: if x is a lower case letter,
3152 it is converted to upper case. Then bit 6 of the character (hex 40) is
3153 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3154 becomes hex 7B.
3155
3156 After \x, from zero to two hexadecimal digits are read (letters can be
3157 in upper or lower case). Any number of hexadecimal digits may appear
3158 between \x{ and }, but the value of the character code must be less
3159 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3160 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3161 than the largest Unicode code point, which is 10FFFF.
3162
3163 If characters other than hexadecimal digits appear between \x{ and },
3164 or if there is no terminating }, this form of escape is not recognized.
3165 Instead, the initial \x will be interpreted as a basic hexadecimal
3166 escape, with no following digits, giving a character whose value is
3167 zero.
3168
3169 Characters whose value is less than 256 can be defined by either of the
3170 two syntaxes for \x. There is no difference in the way they are han-
3171 dled. For example, \xdc is exactly the same as \x{dc}.
3172
3173 After \0 up to two further octal digits are read. If there are fewer
3174 than two digits, just those that are present are used. Thus the
3175 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3176 (code value 7). Make sure you supply two digits after the initial zero
3177 if the pattern character that follows is itself an octal digit.
3178
3179 The handling of a backslash followed by a digit other than 0 is compli-
3180 cated. Outside a character class, PCRE reads it and any following dig-
3181 its as a decimal number. If the number is less than 10, or if there
3182 have been at least that many previous capturing left parentheses in the
3183 expression, the entire sequence is taken as a back reference. A
3184 description of how this works is given later, following the discussion
3185 of parenthesized subpatterns.
3186
3187 Inside a character class, or if the decimal number is greater than 9
3188 and there have not been that many capturing subpatterns, PCRE re-reads
3189 up to three octal digits following the backslash, and uses them to gen-
3190 erate a data character. Any subsequent digits stand for themselves. In
3191 non-UTF-8 mode, the value of a character specified in octal must be
3192 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3193 example:
3194
3195 \040 is another way of writing a space
3196 \40 is the same, provided there are fewer than 40
3197 previous capturing subpatterns
3198 \7 is always a back reference
3199 \11 might be a back reference, or another way of
3200 writing a tab
3201 \011 is always a tab
3202 \0113 is a tab followed by the character "3"
3203 \113 might be a back reference, otherwise the
3204 character with octal code 113
3205 \377 might be a back reference, otherwise
3206 the byte consisting entirely of 1 bits
3207 \81 is either a back reference, or a binary zero
3208 followed by the two characters "8" and "1"
3209
3210 Note that octal values of 100 or greater must not be introduced by a
3211 leading zero, because no more than three octal digits are ever read.
3212
3213 All the sequences that define a single character value can be used both
3214 inside and outside character classes. In addition, inside a character
3215 class, the sequence \b is interpreted as the backspace character (hex
3216 08), and the sequences \R and \X are interpreted as the characters "R"
3217 and "X", respectively. Outside a character class, these sequences have
3218 different meanings (see below).
3219
3220 Absolute and relative back references
3221
3222 The sequence \g followed by an unsigned or a negative number, option-
3223 ally enclosed in braces, is an absolute or relative back reference. A
3224 named back reference can be coded as \g{name}. Back references are dis-
3225 cussed later, following the discussion of parenthesized subpatterns.
3226
3227 Absolute and relative subroutine calls
3228
3229 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3230 name or a number enclosed either in angle brackets or single quotes, is
3231 an alternative syntax for referencing a subpattern as a "subroutine".
3232 Details are discussed later. Note that \g{...} (Perl syntax) and
3233 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3234 reference; the latter is a subroutine call.
3235
3236 Generic character types
3237
3238 Another use of backslash is for specifying generic character types. The
3239 following are always recognized:
3240
3241 \d any decimal digit
3242 \D any character that is not a decimal digit
3243 \h any horizontal whitespace character
3244 \H any character that is not a horizontal whitespace character
3245 \s any whitespace character
3246 \S any character that is not a whitespace character
3247 \v any vertical whitespace character
3248 \V any character that is not a vertical whitespace character
3249 \w any "word" character
3250 \W any "non-word" character
3251
3252 Each pair of escape sequences partitions the complete set of characters
3253 into two disjoint sets. Any given character matches one, and only one,
3254 of each pair.
3255
3256 These character type sequences can appear both inside and outside char-
3257 acter classes. They each match one character of the appropriate type.
3258 If the current matching point is at the end of the subject string, all
3259 of them fail, since there is no character to match.
3260
3261 For compatibility with Perl, \s does not match the VT character (code
3262 11). This makes it different from the the POSIX "space" class. The \s
3263 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3264 "use locale;" is included in a Perl script, \s may match the VT charac-
3265 ter. In PCRE, it never does.
3266
3267 In UTF-8 mode, characters with values greater than 128 never match \d,
3268 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3269 code character property support is available. These sequences retain
3270 their original meanings from before UTF-8 support was available, mainly
3271 for efficiency reasons.
3272
3273 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3274 the other sequences, these do match certain high-valued codepoints in
3275 UTF-8 mode. The horizontal space characters are:
3276
3277 U+0009 Horizontal tab
3278 U+0020 Space
3279 U+00A0 Non-break space
3280 U+1680 Ogham space mark
3281 U+180E Mongolian vowel separator
3282 U+2000 En quad
3283 U+2001 Em quad
3284 U+2002 En space
3285 U+2003 Em space
3286 U+2004 Three-per-em space
3287 U+2005 Four-per-em space
3288 U+2006 Six-per-em space
3289 U+2007 Figure space
3290 U+2008 Punctuation space
3291 U+2009 Thin space
3292 U+200A Hair space
3293 U+202F Narrow no-break space
3294 U+205F Medium mathematical space
3295 U+3000 Ideographic space
3296
3297 The vertical space characters are:
3298
3299 U+000A Linefeed
3300 U+000B Vertical tab
3301 U+000C Formfeed
3302 U+000D Carriage return
3303 U+0085 Next line
3304 U+2028 Line separator
3305 U+2029 Paragraph separator
3306
3307 A "word" character is an underscore or any character less than 256 that
3308 is a letter or digit. The definition of letters and digits is con-
3309 trolled by PCRE's low-valued character tables, and may vary if locale-
3310 specific matching is taking place (see "Locale support" in the pcreapi
3311 page). For example, in a French locale such as "fr_FR" in Unix-like
3312 systems, or "french" in Windows, some character codes greater than 128
3313 are used for accented letters, and these are matched by \w. The use of
3314 locales with Unicode is discouraged.
3315
3316 Newline sequences
3317
3318 Outside a character class, by default, the escape sequence \R matches
3319 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3320 mode \R is equivalent to the following:
3321
3322 (?>\r\n|\n|\x0b|\f|\r|\x85)
3323
3324 This is an example of an "atomic group", details of which are given
3325 below. This particular group matches either the two-character sequence
3326 CR followed by LF, or one of the single characters LF (linefeed,
3327 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3328 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3329 is treated as a single unit that cannot be split.
3330
3331 In UTF-8 mode, two additional characters whose codepoints are greater
3332 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3333 rator, U+2029). Unicode character property support is not needed for
3334 these characters to be recognized.
3335
3336 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3337 the complete set of Unicode line endings) by setting the option
3338 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3339 (BSR is an abbrevation for "backslash R".) This can be made the default
3340 when PCRE is built; if this is the case, the other behaviour can be
3341 requested via the PCRE_BSR_UNICODE option. It is also possible to
3342 specify these settings by starting a pattern string with one of the
3343 following sequences:
3344
3345 (*BSR_ANYCRLF) CR, LF, or CRLF only
3346 (*BSR_UNICODE) any Unicode newline sequence
3347
3348 These override the default and the options given to pcre_compile(), but
3349 they can be overridden by options given to pcre_exec(). Note that these
3350 special settings, which are not Perl-compatible, are recognized only at
3351 the very start of a pattern, and that they must be in upper case. If
3352 more than one of them is present, the last one is used. They can be
3353 combined with a change of newline convention, for example, a pattern
3354 can start with:
3355
3356 (*ANY)(*BSR_ANYCRLF)
3357
3358 Inside a character class, \R matches the letter "R".
3359
3360 Unicode character properties
3361
3362 When PCRE is built with Unicode character property support, three addi-
3363 tional escape sequences that match characters with specific properties
3364 are available. When not in UTF-8 mode, these sequences are of course
3365 limited to testing characters whose codepoints are less than 256, but
3366 they do work in this mode. The extra escape sequences are:
3367
3368 \p{xx} a character with the xx property
3369 \P{xx} a character without the xx property
3370 \X an extended Unicode sequence
3371
3372 The property names represented by xx above are limited to the Unicode
3373 script names, the general category properties, and "Any", which matches
3374 any character (including newline). Other properties such as "InMusical-
3375 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3376 not match any characters, so always causes a match failure.
3377
3378 Sets of Unicode characters are defined as belonging to certain scripts.
3379 A character from one of these sets can be matched using a script name.
3380 For example:
3381
3382 \p{Greek}
3383 \P{Han}
3384
3385 Those that are not part of an identified script are lumped together as
3386 "Common". The current list of scripts is:
3387
3388 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3389 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3390 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3391 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3392 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3393 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3394 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3395 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3396 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3397
3398 Each character has exactly one general category property, specified by
3399 a two-letter abbreviation. For compatibility with Perl, negation can be
3400 specified by including a circumflex between the opening brace and the
3401 property name. For example, \p{^Lu} is the same as \P{Lu}.
3402
3403 If only one letter is specified with \p or \P, it includes all the gen-
3404 eral category properties that start with that letter. In this case, in
3405 the absence of negation, the curly brackets in the escape sequence are
3406 optional; these two examples have the same effect:
3407
3408 \p{L}
3409 \pL
3410
3411 The following general category property codes are supported:
3412
3413 C Other
3414 Cc Control
3415 Cf Format
3416 Cn Unassigned
3417 Co Private use
3418 Cs Surrogate
3419
3420 L Letter
3421 Ll Lower case letter
3422 Lm Modifier letter
3423 Lo Other letter
3424 Lt Title case letter
3425 Lu Upper case letter
3426
3427 M Mark
3428 Mc Spacing mark
3429 Me Enclosing mark
3430 Mn Non-spacing mark
3431
3432 N Number
3433 Nd Decimal number
3434 Nl Letter number
3435 No Other number
3436
3437 P Punctuation
3438 Pc Connector punctuation
3439 Pd Dash punctuation
3440 Pe Close punctuation
3441 Pf Final punctuation
3442 Pi Initial punctuation
3443 Po Other punctuation
3444 Ps Open punctuation
3445
3446 S Symbol
3447 Sc Currency symbol
3448 Sk Modifier symbol
3449 Sm Mathematical symbol
3450 So Other symbol
3451
3452 Z Separator
3453 Zl Line separator
3454 Zp Paragraph separator
3455 Zs Space separator
3456
3457 The special property L& is also supported: it matches a character that
3458 has the Lu, Ll, or Lt property, in other words, a letter that is not
3459 classified as a modifier or "other".
3460
3461 The Cs (Surrogate) property applies only to characters in the range
3462 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3463 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3464 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3465 the pcreapi page).
3466
3467 The long synonyms for these properties that Perl supports (such as
3468 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3469 any of these properties with "Is".
3470
3471 No character that is in the Unicode table has the Cn (unassigned) prop-
3472 erty. Instead, this property is assumed for any code point that is not
3473 in the Unicode table.
3474
3475 Specifying caseless matching does not affect these escape sequences.
3476 For example, \p{Lu} always matches only upper case letters.
3477
3478 The \X escape matches any number of Unicode characters that form an
3479 extended Unicode sequence. \X is equivalent to
3480
3481 (?>\PM\pM*)
3482
3483 That is, it matches a character without the "mark" property, followed
3484 by zero or more characters with the "mark" property, and treats the
3485 sequence as an atomic group (see below). Characters with the "mark"
3486 property are typically accents that affect the preceding character.
3487 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3488 matches any one character.
3489
3490 Matching characters by Unicode property is not fast, because PCRE has
3491 to search a structure that contains data for over fifteen thousand
3492 characters. That is why the traditional escape sequences such as \d and
3493 \w do not use Unicode properties in PCRE.
3494
3495 Resetting the match start
3496
3497 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3498 ously matched characters not to be included in the final matched
3499 sequence. For example, the pattern:
3500
3501 foo\Kbar
3502
3503 matches "foobar", but reports that it has matched "bar". This feature
3504 is similar to a lookbehind assertion (described below). However, in
3505 this case, the part of the subject before the real match does not have
3506 to be of fixed length, as lookbehind assertions do. The use of \K does
3507 not interfere with the setting of captured substrings. For example,
3508 when the pattern
3509
3510 (foo)\Kbar
3511
3512 matches "foobar", the first substring is still set to "foo".
3513
3514 Simple assertions
3515
3516 The final use of backslash is for certain simple assertions. An asser-
3517 tion specifies a condition that has to be met at a particular point in
3518 a match, without consuming any characters from the subject string. The
3519 use of subpatterns for more complicated assertions is described below.
3520 The backslashed assertions are:
3521
3522 \b matches at a word boundary
3523 \B matches when not at a word boundary
3524 \A matches at the start of the subject
3525 \Z matches at the end of the subject
3526 also matches before a newline at the end of the subject
3527 \z matches only at the end of the subject
3528 \G matches at the first matching position in the subject
3529
3530 These assertions may not appear in character classes (but note that \b
3531 has a different meaning, namely the backspace character, inside a char-
3532 acter class).
3533
3534 A word boundary is a position in the subject string where the current
3535 character and the previous character do not both match \w or \W (i.e.
3536 one matches \w and the other matches \W), or the start or end of the
3537 string if the first or last character matches \w, respectively.
3538
3539 The \A, \Z, and \z assertions differ from the traditional circumflex
3540 and dollar (described in the next section) in that they only ever match
3541 at the very start and end of the subject string, whatever options are
3542 set. Thus, they are independent of multiline mode. These three asser-
3543 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3544 affect only the behaviour of the circumflex and dollar metacharacters.
3545 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3546 cating that matching is to start at a point other than the beginning of
3547 the subject, \A can never match. The difference between \Z and \z is
3548 that \Z matches before a newline at the end of the string as well as at
3549 the very end, whereas \z matches only at the end.
3550
3551 The \G assertion is true only when the current matching position is at
3552 the start point of the match, as specified by the startoffset argument
3553 of pcre_exec(). It differs from \A when the value of startoffset is
3554 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3555 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3556 mentation where \G can be useful.
3557
3558 Note, however, that PCRE's interpretation of \G, as the start of the
3559 current match, is subtly different from Perl's, which defines it as the
3560 end of the previous match. In Perl, these can be different when the
3561 previously matched string was empty. Because PCRE does just one match
3562 at a time, it cannot reproduce this behaviour.
3563
3564 If all the alternatives of a pattern begin with \G, the expression is
3565 anchored to the starting match position, and the "anchored" flag is set
3566 in the compiled regular expression.
3567
3568
3569 CIRCUMFLEX AND DOLLAR
3570
3571 Outside a character class, in the default matching mode, the circumflex
3572 character is an assertion that is true only if the current matching
3573 point is at the start of the subject string. If the startoffset argu-
3574 ment of pcre_exec() is non-zero, circumflex can never match if the
3575 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3576 has an entirely different meaning (see below).
3577
3578 Circumflex need not be the first character of the pattern if a number
3579 of alternatives are involved, but it should be the first thing in each
3580 alternative in which it appears if the pattern is ever to match that
3581 branch. If all possible alternatives start with a circumflex, that is,
3582 if the pattern is constrained to match only at the start of the sub-
3583 ject, it is said to be an "anchored" pattern. (There are also other
3584 constructs that can cause a pattern to be anchored.)
3585
3586 A dollar character is an assertion that is true only if the current
3587 matching point is at the end of the subject string, or immediately
3588 before a newline at the end of the string (by default). Dollar need not
3589 be the last character of the pattern if a number of alternatives are
3590 involved, but it should be the last item in any branch in which it
3591 appears. Dollar has no special meaning in a character class.
3592
3593 The meaning of dollar can be changed so that it matches only at the
3594 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3595 compile time. This does not affect the \Z assertion.
3596
3597 The meanings of the circumflex and dollar characters are changed if the
3598 PCRE_MULTILINE option is set. When this is the case, a circumflex
3599 matches immediately after internal newlines as well as at the start of
3600 the subject string. It does not match after a newline that ends the
3601 string. A dollar matches before any newlines in the string, as well as
3602 at the very end, when PCRE_MULTILINE is set. When newline is specified
3603 as the two-character sequence CRLF, isolated CR and LF characters do
3604 not indicate newlines.
3605
3606 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3607 (where \n represents a newline) in multiline mode, but not otherwise.
3608 Consequently, patterns that are anchored in single line mode because
3609 all branches start with ^ are not anchored in multiline mode, and a
3610 match for circumflex is possible when the startoffset argument of
3611 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3612 PCRE_MULTILINE is set.
3613
3614 Note that the sequences \A, \Z, and \z can be used to match the start
3615 and end of the subject in both modes, and if all branches of a pattern
3616 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3617 set.
3618
3619
3620 FULL STOP (PERIOD, DOT)
3621
3622 Outside a character class, a dot in the pattern matches any one charac-
3623 ter in the subject string except (by default) a character that signi-
3624 fies the end of a line. In UTF-8 mode, the matched character may be
3625 more than one byte long.
3626
3627 When a line ending is defined as a single character, dot never matches
3628 that character; when the two-character sequence CRLF is used, dot does
3629 not match CR if it is immediately followed by LF, but otherwise it
3630 matches all characters (including isolated CRs and LFs). When any Uni-
3631 code line endings are being recognized, dot does not match CR or LF or
3632 any of the other line ending characters.
3633
3634 The behaviour of dot with regard to newlines can be changed. If the
3635 PCRE_DOTALL option is set, a dot matches any one character, without
3636 exception. If the two-character sequence CRLF is present in the subject
3637 string, it takes two dots to match it.
3638
3639 The handling of dot is entirely independent of the handling of circum-
3640 flex and dollar, the only relationship being that they both involve
3641 newlines. Dot has no special meaning in a character class.
3642
3643
3644 MATCHING A SINGLE BYTE
3645
3646 Outside a character class, the escape sequence \C matches any one byte,
3647 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3648 line-ending characters. The feature is provided in Perl in order to
3649 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3650 acters into individual bytes, what remains in the string may be a mal-
3651 formed UTF-8 string. For this reason, the \C escape sequence is best
3652 avoided.
3653
3654 PCRE does not allow \C to appear in lookbehind assertions (described
3655 below), because in UTF-8 mode this would make it impossible to calcu-
3656 late the length of the lookbehind.
3657
3658
3659 SQUARE BRACKETS AND CHARACTER CLASSES
3660
3661 An opening square bracket introduces a character class, terminated by a
3662 closing square bracket. A closing square bracket on its own is not spe-
3663 cial. If a closing square bracket is required as a member of the class,
3664 it should be the first data character in the class (after an initial
3665 circumflex, if present) or escaped with a backslash.
3666
3667 A character class matches a single character in the subject. In UTF-8
3668 mode, the character may occupy more than one byte. A matched character
3669 must be in the set of characters defined by the class, unless the first
3670 character in the class definition is a circumflex, in which case the
3671 subject character must not be in the set defined by the class. If a
3672 circumflex is actually required as a member of the class, ensure it is
3673 not the first character, or escape it with a backslash.
3674
3675 For example, the character class [aeiou] matches any lower case vowel,
3676 while [^aeiou] matches any character that is not a lower case vowel.
3677 Note that a circumflex is just a convenient notation for specifying the
3678 characters that are in the class by enumerating those that are not. A
3679 class that starts with a circumflex is not an assertion: it still con-
3680 sumes a character from the subject string, and therefore it fails if
3681 the current pointer is at the end of the string.
3682
3683 In UTF-8 mode, characters with values greater than 255 can be included
3684 in a class as a literal string of bytes, or by using the \x{ escaping
3685 mechanism.
3686
3687 When caseless matching is set, any letters in a class represent both
3688 their upper case and lower case versions, so for example, a caseless
3689 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3690 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3691 understands the concept of case for characters whose values are less
3692 than 128, so caseless matching is always possible. For characters with
3693 higher values, the concept of case is supported if PCRE is compiled
3694 with Unicode property support, but not otherwise. If you want to use
3695 caseless matching for characters 128 and above, you must ensure that
3696 PCRE is compiled with Unicode property support as well as with UTF-8
3697 support.
3698
3699 Characters that might indicate line breaks are never treated in any
3700 special way when matching character classes, whatever line-ending
3701 sequence is in use, and whatever setting of the PCRE_DOTALL and
3702 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3703 of these characters.
3704
3705 The minus (hyphen) character can be used to specify a range of charac-
3706 ters in a character class. For example, [d-m] matches any letter
3707 between d and m, inclusive. If a minus character is required in a
3708 class, it must be escaped with a backslash or appear in a position
3709 where it cannot be interpreted as indicating a range, typically as the
3710 first or last character in the class.
3711
3712 It is not possible to have the literal character "]" as the end charac-
3713 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3714 two characters ("W" and "-") followed by a literal string "46]", so it
3715 would match "W46]" or "-46]". However, if the "]" is escaped with a
3716 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3717 preted as a class containing a range followed by two other characters.
3718 The octal or hexadecimal representation of "]" can also be used to end
3719 a range.
3720
3721 Ranges operate in the collating sequence of character values. They can
3722 also be used for characters specified numerically, for example
3723 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3724 are greater than 255, for example [\x{100}-\x{2ff}].
3725
3726 If a range that includes letters is used when caseless matching is set,
3727 it matches the letters in either case. For example, [W-c] is equivalent
3728 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3729 character tables for a French locale are in use, [\xc8-\xcb] matches
3730 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3731 concept of case for characters with values greater than 128 only when
3732 it is compiled with Unicode property support.
3733
3734 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3735 in a character class, and add the characters that they match to the
3736 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3737 flex can conveniently be used with the upper case character types to
3738 specify a more restricted set of characters than the matching lower
3739 case type. For example, the class [^\W_] matches any letter or digit,
3740 but not underscore.
3741
3742 The only metacharacters that are recognized in character classes are
3743 backslash, hyphen (only where it can be interpreted as specifying a
3744 range), circumflex (only at the start), opening square bracket (only
3745 when it can be interpreted as introducing a POSIX class name - see the
3746 next section), and the terminating closing square bracket. However,
3747 escaping other non-alphanumeric characters does no harm.
3748
3749
3750 POSIX CHARACTER CLASSES
3751
3752 Perl supports the POSIX notation for character classes. This uses names
3753 enclosed by [: and :] within the enclosing square brackets. PCRE also
3754 supports this notation. For example,
3755
3756 [01[:alpha:]%]
3757
3758 matches "0", "1", any alphabetic character, or "%". The supported class
3759 names are
3760
3761 alnum letters and digits
3762 alpha letters
3763 ascii character codes 0 - 127
3764 blank space or tab only
3765 cntrl control characters
3766 digit decimal digits (same as \d)
3767 graph printing characters, excluding space
3768 lower lower case letters
3769 print printing characters, including space
3770 punct printing characters, excluding letters and digits
3771 space white space (not quite the same as \s)
3772 upper upper case letters
3773 word "word" characters (same as \w)
3774 xdigit hexadecimal digits
3775
3776 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3777 and space (32). Notice that this list includes the VT character (code
3778 11). This makes "space" different to \s, which does not include VT (for
3779 Perl compatibility).
3780
3781 The name "word" is a Perl extension, and "blank" is a GNU extension
3782 from Perl 5.8. Another Perl extension is negation, which is indicated
3783 by a ^ character after the colon. For example,
3784
3785 [12[:^digit:]]
3786
3787 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3788 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3789 these are not supported, and an error is given if they are encountered.
3790
3791 In UTF-8 mode, characters with values greater than 128 do not match any
3792 of the POSIX character classes.
3793
3794
3795 VERTICAL BAR
3796
3797 Vertical bar characters are used to separate alternative patterns. For
3798 example, the pattern
3799
3800 gilbert|sullivan
3801
3802 matches either "gilbert" or "sullivan". Any number of alternatives may
3803 appear, and an empty alternative is permitted (matching the empty
3804 string). The matching process tries each alternative in turn, from left
3805 to right, and the first one that succeeds is used. If the alternatives
3806 are within a subpattern (defined below), "succeeds" means matching the
3807 rest of the main pattern as well as the alternative in the subpattern.
3808
3809
3810 INTERNAL OPTION SETTING
3811
3812 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3813 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3814 within the pattern by a sequence of Perl option letters enclosed
3815 between "(?" and ")". The option letters are
3816
3817 i for PCRE_CASELESS
3818 m for PCRE_MULTILINE
3819 s for PCRE_DOTALL
3820 x for PCRE_EXTENDED
3821
3822 For example, (?im) sets caseless, multiline matching. It is also possi-
3823 ble to unset these options by preceding the letter with a hyphen, and a
3824 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3825 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3826 is also permitted. If a letter appears both before and after the
3827 hyphen, the option is unset.
3828
3829 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3830 can be changed in the same way as the Perl-compatible options by using
3831 the characters J, U and X respectively.
3832
3833 When an option change occurs at top level (that is, not inside subpat-
3834 tern parentheses), the change applies to the remainder of the pattern
3835 that follows. If the change is placed right at the start of a pattern,
3836 PCRE extracts it into the global options (and it will therefore show up
3837 in data extracted by the pcre_fullinfo() function).
3838
3839 An option change within a subpattern (see below for a description of
3840 subpatterns) affects only that part of the current pattern that follows
3841 it, so
3842
3843 (a(?i)b)c
3844
3845 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3846 used). By this means, options can be made to have different settings
3847 in different parts of the pattern. Any changes made in one alternative
3848 do carry on into subsequent branches within the same subpattern. For
3849 example,
3850
3851 (a(?i)b|c)
3852
3853 matches "ab", "aB", "c", and "C", even though when matching "C" the
3854 first branch is abandoned before the option setting. This is because
3855 the effects of option settings happen at compile time. There would be
3856 some very weird behaviour otherwise.
3857
3858 Note: There are other PCRE-specific options that can be set by the
3859 application when the compile or match functions are called. In some
3860 cases the pattern can contain special leading sequences to override
3861 what the application has set or what has been defaulted. Details are
3862 given in the section entitled "Newline sequences" above.
3863
3864
3865 SUBPATTERNS
3866
3867 Subpatterns are delimited by parentheses (round brackets), which can be
3868 nested. Turning part of a pattern into a subpattern does two things:
3869
3870 1. It localizes a set of alternatives. For example, the pattern
3871
3872 cat(aract|erpillar|)
3873
3874 matches one of the words "cat", "cataract", or "caterpillar". Without
3875 the parentheses, it would match "cataract", "erpillar" or an empty
3876 string.
3877
3878 2. It sets up the subpattern as a capturing subpattern. This means
3879 that, when the whole pattern matches, that portion of the subject
3880 string that matched the subpattern is passed back to the caller via the
3881 ovector argument of pcre_exec(). Opening parentheses are counted from
3882 left to right (starting from 1) to obtain numbers for the capturing
3883 subpatterns.
3884
3885 For example, if the string "the red king" is matched against the pat-
3886 tern
3887
3888 the ((red|white) (king|queen))
3889
3890 the captured substrings are "red king", "red", and "king", and are num-
3891 bered 1, 2, and 3, respectively.
3892
3893 The fact that plain parentheses fulfil two functions is not always
3894 helpful. There are often times when a grouping subpattern is required
3895 without a capturing requirement. If an opening parenthesis is followed
3896 by a question mark and a colon, the subpattern does not do any captur-
3897 ing, and is not counted when computing the number of any subsequent
3898 capturing subpatterns. For example, if the string "the white queen" is
3899 matched against the pattern
3900
3901 the ((?:red|white) (king|queen))
3902
3903 the captured substrings are "white queen" and "queen", and are numbered
3904 1 and 2. The maximum number of capturing subpatterns is 65535.
3905
3906 As a convenient shorthand, if any option settings are required at the
3907 start of a non-capturing subpattern, the option letters may appear
3908 between the "?" and the ":". Thus the two patterns
3909
3910 (?i:saturday|sunday)
3911 (?:(?i)saturday|sunday)
3912
3913 match exactly the same set of strings. Because alternative branches are
3914 tried from left to right, and options are not reset until the end of
3915 the subpattern is reached, an option setting in one branch does affect
3916 subsequent branches, so the above patterns match "SUNDAY" as well as
3917 "Saturday".
3918
3919
3920 DUPLICATE SUBPATTERN NUMBERS
3921
3922 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3923 uses the same numbers for its capturing parentheses. Such a subpattern
3924 starts with (?| and is itself a non-capturing subpattern. For example,
3925 consider this pattern:
3926
3927 (?|(Sat)ur|(Sun))day
3928
3929 Because the two alternatives are inside a (?| group, both sets of cap-
3930 turing parentheses are numbered one. Thus, when the pattern matches,
3931 you can look at captured substring number one, whichever alternative
3932 matched. This construct is useful when you want to capture part, but
3933 not all, of one of a number of alternatives. Inside a (?| group, paren-
3934 theses are numbered as usual, but the number is reset at the start of
3935 each branch. The numbers of any capturing buffers that follow the sub-
3936 pattern start after the highest number used in any branch. The follow-
3937 ing example is taken from the Perl documentation. The numbers under-
3938 neath show in which buffer the captured content will be stored.
3939
3940 # before ---------------branch-reset----------- after
3941 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3942 # 1 2 2 3 2 3 4
3943
3944 A backreference or a recursive call to a numbered subpattern always
3945 refers to the first one in the pattern with the given number.
3946
3947 An alternative approach to using this "branch reset" feature is to use
3948 duplicate named subpatterns, as described in the next section.
3949
3950
3951 NAMED SUBPATTERNS
3952
3953 Identifying capturing parentheses by number is simple, but it can be
3954 very hard to keep track of the numbers in complicated regular expres-
3955 sions. Furthermore, if an expression is modified, the numbers may
3956 change. To help with this difficulty, PCRE supports the naming of sub-
3957 patterns. This feature was not added to Perl until release 5.10. Python
3958 had the feature earlier, and PCRE introduced it at release 4.0, using
3959 the Python syntax. PCRE now supports both the Perl and the Python syn-
3960 tax.
3961
3962 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3963 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3964 to capturing parentheses from other parts of the pattern, such as back-
3965 references, recursion, and conditions, can be made by name as well as
3966 by number.
3967
3968 Names consist of up to 32 alphanumeric characters and underscores.
3969 Named capturing parentheses are still allocated numbers as well as
3970 names, exactly as if the names were not present. The PCRE API provides
3971 function calls for extracting the name-to-number translation table from
3972 a compiled pattern. There is also a convenience function for extracting
3973 a captured substring by name.
3974
3975 By default, a name must be unique within a pattern, but it is possible
3976 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3977 time. This can be useful for patterns where only one instance of the
3978 named parentheses can match. Suppose you want to match the name of a
3979 weekday, either as a 3-letter abbreviation or as the full name, and in
3980 both cases you want to extract the abbreviation. This pattern (ignoring
3981 the line breaks) does the job:
3982
3983 (?<DN>Mon|Fri|Sun)(?:day)?|
3984 (?<DN>Tue)(?:sday)?|
3985 (?<DN>Wed)(?:nesday)?|
3986 (?<DN>Thu)(?:rsday)?|
3987 (?<DN>Sat)(?:urday)?
3988
3989 There are five capturing substrings, but only one is ever set after a
3990 match. (An alternative way of solving this problem is to use a "branch
3991 reset" subpattern, as described in the previous section.)
3992
3993 The convenience function for extracting the data by name returns the
3994 substring for the first (and in this example, the only) subpattern of
3995 that name that matched. This saves searching to find which numbered
3996 subpattern it was. If you make a reference to a non-unique named sub-
3997 pattern from elsewhere in the pattern, the one that corresponds to the
3998 lowest number is used. For further details of the interfaces for han-
3999 dling named subpatterns, see the pcreapi documentation.
4000
4001 Warning: You cannot use different names to distinguish between two sub-
4002 patterns with the same number (see the previous section) because PCRE
4003 uses only the numbers when matching.
4004
4005
4006 REPETITION
4007
4008 Repetition is specified by quantifiers, which can follow any of the
4009 following items:
4010
4011 a literal data character
4012 the dot metacharacter
4013 the \C escape sequence
4014 the \X escape sequence (in UTF-8 mode with Unicode properties)
4015 the \R escape sequence
4016 an escape such as \d that matches a single character
4017 a character class
4018 a back reference (see next section)
4019 a parenthesized subpattern (unless it is an assertion)
4020
4021 The general repetition quantifier specifies a minimum and maximum num-
4022 ber of permitted matches, by giving the two numbers in curly brackets
4023 (braces), separated by a comma. The numbers must be less than 65536,
4024 and the first must be less than or equal to the second. For example:
4025
4026 z{2,4}
4027
4028 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4029 special character. If the second number is omitted, but the comma is
4030 present, there is no upper limit; if the second number and the comma
4031 are both omitted, the quantifier specifies an exact number of required
4032 matches. Thus
4033
4034 [aeiou]{3,}
4035
4036 matches at least 3 successive vowels, but may match many more, while
4037
4038 \d{8}
4039
4040 matches exactly 8 digits. An opening curly bracket that appears in a
4041 position where a quantifier is not allowed, or one that does not match
4042 the syntax of a quantifier, is taken as a literal character. For exam-
4043 ple, {,6} is not a quantifier, but a literal string of four characters.
4044
4045 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4046 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4047 acters, each of which is represented by a two-byte sequence. Similarly,
4048 when Unicode property support is available, \X{3} matches three Unicode
4049 extended sequences, each of which may be several bytes long (and they
4050 may be of different lengths).
4051
4052 The quantifier {0} is permitted, causing the expression to behave as if
4053 the previous item and the quantifier were not present. This may be use-
4054 ful for subpatterns that are referenced as subroutines from elsewhere
4055 in the pattern. Items other than subpatterns that have a {0} quantifier
4056 are omitted from the compiled pattern.
4057
4058 For convenience, the three most common quantifiers have single-charac-
4059 ter abbreviations:
4060
4061 * is equivalent to {0,}
4062 + is equivalent to {1,}
4063 ? is equivalent to {0,1}
4064
4065 It is possible to construct infinite loops by following a subpattern
4066 that can match no characters with a quantifier that has no upper limit,
4067 for example:
4068
4069 (a?)*
4070
4071 Earlier versions of Perl and PCRE used to give an error at compile time
4072 for such patterns. However, because there are cases where this can be
4073 useful, such patterns are now accepted, but if any repetition of the
4074 subpattern does in fact match no characters, the loop is forcibly bro-
4075 ken.
4076
4077 By default, the quantifiers are "greedy", that is, they match as much
4078 as possible (up to the maximum number of permitted times), without
4079 causing the rest of the pattern to fail. The classic example of where
4080 this gives problems is in trying to match comments in C programs. These
4081 appear between /* and */ and within the comment, individual * and /
4082 characters may appear. An attempt to match C comments by applying the
4083 pattern
4084
4085 /\*.*\*/
4086
4087 to the string
4088
4089 /* first comment */ not comment /* second comment */
4090
4091 fails, because it matches the entire string owing to the greediness of
4092 the .* item.
4093
4094 However, if a quantifier is followed by a question mark, it ceases to
4095 be greedy, and instead matches the minimum number of times possible, so
4096 the pattern
4097
4098 /\*.*?\*/
4099
4100 does the right thing with the C comments. The meaning of the various
4101 quantifiers is not otherwise changed, just the preferred number of
4102 matches. Do not confuse this use of question mark with its use as a
4103 quantifier in its own right. Because it has two uses, it can sometimes
4104 appear doubled, as in
4105
4106 \d??\d
4107
4108 which matches one digit by preference, but can match two if that is the
4109 only way the rest of the pattern matches.
4110
4111 If the PCRE_UNGREEDY option is set (an option that is not available in
4112 Perl), the quantifiers are not greedy by default, but individual ones
4113 can be made greedy by following them with a question mark. In other
4114 words, it inverts the default behaviour.
4115
4116 When a parenthesized subpattern is quantified with a minimum repeat
4117 count that is greater than 1 or with a limited maximum, more memory is
4118 required for the compiled pattern, in proportion to the size of the
4119 minimum or maximum.
4120
4121 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4122 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4123 the pattern is implicitly anchored, because whatever follows will be
4124 tried against every character position in the subject string, so there
4125 is no point in retrying the overall match at any position after the
4126 first. PCRE normally treats such a pattern as though it were preceded
4127 by \A.
4128
4129 In cases where it is known that the subject string contains no new-
4130 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4131 mization, or alternatively using ^ to indicate anchoring explicitly.
4132
4133 However, there is one situation where the optimization cannot be used.
4134 When .* is inside capturing parentheses that are the subject of a
4135 backreference elsewhere in the pattern, a match at the start may fail
4136 where a later one succeeds. Consider, for example:
4137
4138 (.*)abc\1
4139
4140 If the subject is "xyz123abc123" the match point is the fourth charac-
4141 ter. For this reason, such a pattern is not implicitly anchored.
4142
4143 When a capturing subpattern is repeated, the value captured is the sub-
4144 string that matched the final iteration. For example, after
4145
4146 (tweedle[dume]{3}\s*)+
4147
4148 has matched "tweedledum tweedledee" the value of the captured substring
4149 is "tweedledee". However, if there are nested capturing subpatterns,
4150 the corresponding captured values may have been set in previous itera-
4151 tions. For example, after
4152
4153 /(a|(b))+/
4154
4155 matches "aba" the value of the second captured substring is "b".
4156
4157
4158 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4159
4160 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4161 repetition, failure of what follows normally causes the repeated item
4162 to be re-evaluated to see if a different number of repeats allows the
4163 rest of the pattern to match. Sometimes it is useful to prevent this,
4164 either to change the nature of the match, or to cause it fail earlier
4165 than it otherwise might, when the author of the pattern knows there is
4166 no point in carrying on.
4167
4168 Consider, for example, the pattern \d+foo when applied to the subject
4169 line
4170
4171 123456bar
4172
4173 After matching all 6 digits and then failing to match "foo", the normal
4174 action of the matcher is to try again with only 5 digits matching the
4175 \d+ item, and then with 4, and so on, before ultimately failing.
4176 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4177 the means for specifying that once a subpattern has matched, it is not
4178 to be re-evaluated in this way.
4179
4180 If we use atomic grouping for the previous example, the matcher gives
4181 up immediately on failing to match "foo" the first time. The notation
4182 is a kind of special parenthesis, starting with (?> as in this example:
4183
4184 (?>\d+)foo
4185
4186 This kind of parenthesis "locks up" the part of the pattern it con-
4187 tains once it has matched, and a failure further into the pattern is
4188 prevented from backtracking into it. Backtracking past it to previous
4189 items, however, works as normal.
4190
4191 An alternative description is that a subpattern of this type matches
4192 the string of characters that an identical standalone pattern would
4193 match, if anchored at the current point in the subject string.
4194
4195 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4196 such as the above example can be thought of as a maximizing repeat that
4197 must swallow everything it can. So, while both \d+ and \d+? are pre-
4198 pared to adjust the number of digits they match in order to make the
4199 rest of the pattern match, (?>\d+) can only match an entire sequence of
4200 digits.
4201
4202 Atomic groups in general can of course contain arbitrarily complicated
4203 subpatterns, and can be nested. However, when the subpattern for an
4204 atomic group is just a single repeated item, as in the example above, a
4205 simpler notation, called a "possessive quantifier" can be used. This
4206 consists of an additional + character following a quantifier. Using
4207 this notation, the previous example can be rewritten as
4208
4209 \d++foo
4210
4211 Note that a possessive quantifier can be used with an entire group, for
4212 example:
4213
4214 (abc|xyz){2,3}+
4215
4216 Possessive quantifiers are always greedy; the setting of the
4217 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4218 simpler forms of atomic group. However, there is no difference in the
4219 meaning of a possessive quantifier and the equivalent atomic group,
4220 though there may be a performance difference; possessive quantifiers
4221 should be slightly faster.
4222
4223 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4224 tax. Jeffrey Friedl originated the idea (and the name) in the first
4225 edition of his book. Mike McCloskey liked it, so implemented it when he
4226 built Sun's Java package, and PCRE copied it from there. It ultimately
4227 found its way into Perl at release 5.10.
4228
4229 PCRE has an optimization that automatically "possessifies" certain sim-
4230 ple pattern constructs. For example, the sequence A+B is treated as
4231 A++B because there is no point in backtracking into a sequence of A's
4232 when B must follow.
4233
4234 When a pattern contains an unlimited repeat inside a subpattern that
4235 can itself be repeated an unlimited number of times, the use of an
4236 atomic group is the only way to avoid some failing matches taking a
4237 very long time indeed. The pattern
4238
4239 (\D+|<\d+>)*[!?]
4240
4241 matches an unlimited number of substrings that either consist of non-
4242 digits, or digits enclosed in <>, followed by either ! or ?. When it
4243 matches, it runs quickly. However, if it is applied to
4244
4245 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4246
4247 it takes a long time before reporting failure. This is because the
4248 string can be divided between the internal \D+ repeat and the external
4249 * repeat in a large number of ways, and all have to be tried. (The
4250 example uses [!?] rather than a single character at the end, because
4251 both PCRE and Perl have an optimization that allows for fast failure
4252 when a single character is used. They remember the last single charac-
4253 ter that is required for a match, and fail early if it is not present
4254 in the string.) If the pattern is changed so that it uses an atomic
4255 group, like this:
4256
4257 ((?>\D+)|<\d+>)*[!?]
4258
4259 sequences of non-digits cannot be broken, and failure happens quickly.
4260
4261
4262 BACK REFERENCES
4263
4264 Outside a character class, a backslash followed by a digit greater than
4265 0 (and possibly further digits) is a back reference to a capturing sub-
4266 pattern earlier (that is, to its left) in the pattern, provided there
4267 have been that many previous capturing left parentheses.
4268
4269 However, if the decimal number following the backslash is less than 10,
4270 it is always taken as a back reference, and causes an error only if
4271 there are not that many capturing left parentheses in the entire pat-
4272 tern. In other words, the parentheses that are referenced need not be
4273 to the left of the reference for numbers less than 10. A "forward back
4274 reference" of this type can make sense when a repetition is involved
4275 and the subpattern to the right has participated in an earlier itera-
4276 tion.
4277
4278 It is not possible to have a numerical "forward back reference" to a
4279 subpattern whose number is 10 or more using this syntax because a
4280 sequence such as \50 is interpreted as a character defined in octal.
4281 See the subsection entitled "Non-printing characters" above for further
4282 details of the handling of digits following a backslash. There is no
4283 such problem when named parentheses are used. A back reference to any
4284 subpattern is possible using named parentheses (see below).
4285
4286 Another way of avoiding the ambiguity inherent in the use of digits
4287 following a backslash is to use the \g escape sequence, which is a fea-
4288 ture introduced in Perl 5.10. This escape must be followed by an
4289 unsigned number or a negative number, optionally enclosed in braces.
4290 These examples are all identical:
4291
4292 (ring), \1
4293 (ring), \g1
4294 (ring), \g{1}
4295
4296 An unsigned number specifies an absolute reference without the ambigu-
4297 ity that is present in the older syntax. It is also useful when literal
4298 digits follow the reference. A negative number is a relative reference.
4299 Consider this example:
4300
4301 (abc(def)ghi)\g{-1}
4302
4303 The sequence \g{-1} is a reference to the most recently started captur-
4304 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4305 \g{-2} would be equivalent to \1. The use of relative references can be
4306 helpful in long patterns, and also in patterns that are created by
4307 joining together fragments that contain references within themselves.
4308
4309 A back reference matches whatever actually matched the capturing sub-
4310 pattern in the current subject string, rather than anything matching
4311 the subpattern itself (see "Subpatterns as subroutines" below for a way
4312 of doing that). So the pattern
4313
4314 (sens|respons)e and \1ibility
4315
4316 matches "sense and sensibility" and "response and responsibility", but
4317 not "sense and responsibility". If caseful matching is in force at the
4318 time of the back reference, the case of letters is relevant. For exam-
4319 ple,
4320
4321 ((?i)rah)\s+\1
4322
4323 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4324 original capturing subpattern is matched caselessly.
4325
4326 There are several different ways of writing back references to named
4327 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4328 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4329 unified back reference syntax, in which \g can be used for both numeric
4330 and named references, is also supported. We could rewrite the above
4331 example in any of the following ways:
4332
4333 (?<p1>(?i)rah)\s+\k<p1>
4334 (?'p1'(?i)rah)\s+\k{p1}
4335 (?P<p1>(?i)rah)\s+(?P=p1)
4336 (?<p1>(?i)rah)\s+\g{p1}
4337
4338 A subpattern that is referenced by name may appear in the pattern
4339 before or after the reference.
4340
4341 There may be more than one back reference to the same subpattern. If a
4342 subpattern has not actually been used in a particular match, any back
4343 references to it always fail. For example, the pattern
4344
4345 (a|(bc))\2
4346
4347 always fails if it starts to match "a" rather than "bc". Because there
4348 may be many capturing parentheses in a pattern, all digits following
4349 the backslash are taken as part of a potential back reference number.
4350 If the pattern continues with a digit character, some delimiter must be
4351 used to terminate the back reference. If the PCRE_EXTENDED option is
4352 set, this can be whitespace. Otherwise an empty comment (see "Com-
4353 ments" below) can be used.
4354
4355 A back reference that occurs inside the parentheses to which it refers
4356 fails when the subpattern is first used, so, for example, (a\1) never
4357 matches. However, such references can be useful inside repeated sub-
4358 patterns. For example, the pattern
4359
4360 (a|b\1)+
4361
4362 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4363 ation of the subpattern, the back reference matches the character
4364 string corresponding to the previous iteration. In order for this to
4365 work, the pattern must be such that the first iteration does not need
4366 to match the back reference. This can be done using alternation, as in
4367 the example above, or by a quantifier with a minimum of zero.
4368
4369
4370 ASSERTIONS
4371
4372 An assertion is a test on the characters following or preceding the
4373 current matching point that does not actually consume any characters.
4374 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4375 described above.
4376
4377 More complicated assertions are coded as subpatterns. There are two
4378 kinds: those that look ahead of the current position in the subject
4379 string, and those that look behind it. An assertion subpattern is
4380 matched in the normal way, except that it does not cause the current
4381 matching position to be changed.
4382
4383 Assertion subpatterns are not capturing subpatterns, and may not be
4384 repeated, because it makes no sense to assert the same thing several
4385 times. If any kind of assertion contains capturing subpatterns within
4386 it, these are counted for the purposes of numbering the capturing sub-
4387 patterns in the whole pattern. However, substring capturing is carried
4388 out only for positive assertions, because it does not make sense for
4389 negative assertions.
4390
4391 Lookahead assertions
4392
4393 Lookahead assertions start with (?= for positive assertions and (?! for
4394 negative assertions. For example,
4395
4396 \w+(?=;)
4397
4398 matches a word followed by a semicolon, but does not include the semi-
4399 colon in the match, and
4400
4401 foo(?!bar)
4402
4403 matches any occurrence of "foo" that is not followed by "bar". Note
4404 that the apparently similar pattern
4405
4406 (?!foo)bar
4407
4408 does not find an occurrence of "bar" that is preceded by something
4409 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4410 the assertion (?!foo) is always true when the next three characters are
4411 "bar". A lookbehind assertion is needed to achieve the other effect.
4412
4413 If you want to force a matching failure at some point in a pattern, the
4414 most convenient way to do it is with (?!) because an empty string
4415 always matches, so an assertion that requires there not to be an empty
4416 string must always fail.
4417
4418 Lookbehind assertions
4419
4420 Lookbehind assertions start with (?<= for positive assertions and (?<!
4421 for negative assertions. For example,
4422
4423 (?<!foo)bar
4424
4425 does find an occurrence of "bar" that is not preceded by "foo". The
4426 contents of a lookbehind assertion are restricted such that all the
4427 strings it matches must have a fixed length. However, if there are sev-
4428 eral top-level alternatives, they do not all have to have the same
4429 fixed length. Thus
4430
4431 (?<=bullock|donkey)
4432
4433 is permitted, but
4434
4435 (?<!dogs?|cats?)
4436
4437 causes an error at compile time. Branches that match different length
4438 strings are permitted only at the top level of a lookbehind assertion.
4439 This is an extension compared with Perl (at least for 5.8), which
4440 requires all branches to match the same length of string. An assertion
4441 such as
4442
4443 (?<=ab(c|de))
4444
4445 is not permitted, because its single top-level branch can match two
4446 different lengths, but it is acceptable if rewritten to use two top-
4447 level branches:
4448
4449 (?<=abc|abde)
4450
4451 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4452 instead of a lookbehind assertion; this is not restricted to a fixed-
4453 length.
4454
4455 The implementation of lookbehind assertions is, for each alternative,
4456 to temporarily move the current position back by the fixed length and
4457 then try to match. If there are insufficient characters before the cur-
4458 rent position, the assertion fails.
4459
4460 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4461 mode) to appear in lookbehind assertions, because it makes it impossi-
4462 ble to calculate the length of the lookbehind. The \X and \R escapes,
4463 which can match different numbers of bytes, are also not permitted.
4464
4465 Possessive quantifiers can be used in conjunction with lookbehind
4466 assertions to specify efficient matching at the end of the subject
4467 string. Consider a simple pattern such as
4468
4469 abcd$
4470
4471 when applied to a long string that does not match. Because matching
4472 proceeds from left to right, PCRE will look for each "a" in the subject
4473 and then see if what follows matches the rest of the pattern. If the
4474 pattern is specified as
4475
4476 ^.*abcd$
4477
4478 the initial .* matches the entire string at first, but when this fails
4479 (because there is no following "a"), it backtracks to match all but the
4480 last character, then all but the last two characters, and so on. Once
4481 again the search for "a" covers the entire string, from right to left,
4482 so we are no better off. However, if the pattern is written as
4483
4484 ^.*+(?<=abcd)
4485
4486 there can be no backtracking for the .*+ item; it can match only the
4487 entire string. The subsequent lookbehind assertion does a single test
4488 on the last four characters. If it fails, the match fails immediately.
4489 For long strings, this approach makes a significant difference to the
4490 processing time.
4491
4492 Using multiple assertions
4493
4494 Several assertions (of any sort) may occur in succession. For example,
4495
4496 (?<=\d{3})(?<!999)foo
4497
4498 matches "foo" preceded by three digits that are not "999". Notice that
4499 each of the assertions is applied independently at the same point in
4500 the subject string. First there is a check that the previous three
4501 characters are all digits, and then there is a check that the same
4502 three characters are not "999". This pattern does not match "foo" pre-
4503 ceded by six characters, the first of which are digits and the last
4504 three of which are not "999". For example, it doesn't match "123abc-
4505 foo". A pattern to do that is
4506
4507 (?<=\d{3}...)(?<!999)foo
4508
4509 This time the first assertion looks at the preceding six characters,
4510 checking that the first three are digits, and then the second assertion
4511 checks that the preceding three characters are not "999".
4512
4513 Assertions can be nested in any combination. For example,
4514
4515 (?<=(?<!foo)bar)baz
4516
4517 matches an occurrence of "baz" that is preceded by "bar" which in turn
4518 is not preceded by "foo", while
4519
4520 (?<=\d{3}(?!999)...)foo
4521
4522 is another pattern that matches "foo" preceded by three digits and any
4523 three characters that are not "999".
4524
4525
4526 CONDITIONAL SUBPATTERNS
4527
4528 It is possible to cause the matching process to obey a subpattern con-
4529 ditionally or to choose between two alternative subpatterns, depending
4530 on the result of an assertion, or whether a previous capturing subpat-
4531 tern matched or not. The two possible forms of conditional subpattern
4532 are
4533
4534 (?(condition)yes-pattern)
4535 (?(condition)yes-pattern|no-pattern)
4536
4537 If the condition is satisfied, the yes-pattern is used; otherwise the
4538 no-pattern (if present) is used. If there are more than two alterna-
4539 tives in the subpattern, a compile-time error occurs.
4540
4541 There are four kinds of condition: references to subpatterns, refer-
4542 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4543
4544 Checking for a used subpattern by number
4545
4546 If the text between the parentheses consists of a sequence of digits,
4547 the condition is true if the capturing subpattern of that number has
4548 previously matched. An alternative notation is to precede the digits
4549 with a plus or minus sign. In this case, the subpattern number is rela-
4550 tive rather than absolute. The most recently opened parentheses can be
4551 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4552 looping constructs it can also make sense to refer to subsequent groups
4553 with constructs such as (?(+2).
4554
4555 Consider the following pattern, which contains non-significant white
4556 space to make it more readable (assume the PCRE_EXTENDED option) and to
4557 divide it into three parts for ease of discussion:
4558
4559 ( \( )? [^()]+ (?(1) \) )
4560
4561 The first part matches an optional opening parenthesis, and if that
4562 character is present, sets it as the first captured substring. The sec-
4563 ond part matches one or more characters that are not parentheses. The
4564 third part is a conditional subpattern that tests whether the first set
4565 of parentheses matched or not. If they did, that is, if subject started
4566 with an opening parenthesis, the condition is true, and so the yes-pat-
4567 tern is executed and a closing parenthesis is required. Otherwise,
4568 since no-pattern is not present, the subpattern matches nothing. In
4569 other words, this pattern matches a sequence of non-parentheses,
4570 optionally enclosed in parentheses.
4571
4572 If you were embedding this pattern in a larger one, you could use a
4573 relative reference:
4574
4575 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4576
4577 This makes the fragment independent of the parentheses in the larger
4578 pattern.
4579
4580 Checking for a used subpattern by name
4581
4582 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4583 used subpattern by name. For compatibility with earlier versions of
4584 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4585 also recognized. However, there is a possible ambiguity with this syn-
4586 tax, because subpattern names may consist entirely of digits. PCRE
4587 looks first for a named subpattern; if it cannot find one and the name
4588 consists entirely of digits, PCRE looks for a subpattern of that num-
4589 ber, which must be greater than zero. Using subpattern names that con-
4590 sist entirely of digits is not recommended.
4591
4592 Rewriting the above example to use a named subpattern gives this:
4593
4594 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4595
4596
4597 Checking for pattern recursion
4598
4599 If the condition is the string (R), and there is no subpattern with the
4600 name R, the condition is true if a recursive call to the whole pattern
4601 or any subpattern has been made. If digits or a name preceded by amper-
4602 sand follow the letter R, for example:
4603
4604 (?(R3)...) or (?(R&name)...)
4605
4606 the condition is true if the most recent recursion is into the subpat-
4607 tern whose number or name is given. This condition does not check the
4608 entire recursion stack.
4609
4610 At "top level", all these recursion test conditions are false. Recur-
4611 sive patterns are described below.
4612
4613 Defining subpatterns for use by reference only
4614
4615 If the condition is the string (DEFINE), and there is no subpattern
4616 with the name DEFINE, the condition is always false. In this case,
4617 there may be only one alternative in the subpattern. It is always
4618 skipped if control reaches this point in the pattern; the idea of
4619 DEFINE is that it can be used to define "subroutines" that can be ref-
4620 erenced from elsewhere. (The use of "subroutines" is described below.)
4621 For example, a pattern to match an IPv4 address could be written like
4622 this (ignore whitespace and line breaks):
4623
4624 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4625 \b (?&byte) (\.(?&byte)){3} \b
4626
4627 The first part of the pattern is a DEFINE group inside which a another
4628 group named "byte" is defined. This matches an individual component of
4629 an IPv4 address (a number less than 256). When matching takes place,
4630 this part of the pattern is skipped because DEFINE acts like a false
4631 condition.
4632
4633 The rest of the pattern uses references to the named group to match the
4634 four dot-separated components of an IPv4 address, insisting on a word
4635 boundary at each end.
4636
4637 Assertion conditions
4638
4639 If the condition is not in any of the above formats, it must be an
4640 assertion. This may be a positive or negative lookahead or lookbehind
4641 assertion. Consider this pattern, again containing non-significant
4642 white space, and with the two alternatives on the second line:
4643
4644 (?(?=[^a-z]*[a-z])
4645 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4646
4647 The condition is a positive lookahead assertion that matches an
4648 optional sequence of non-letters followed by a letter. In other words,
4649 it tests for the presence of at least one letter in the subject. If a
4650 letter is found, the subject is matched against the first alternative;
4651 otherwise it is matched against the second. This pattern matches
4652 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4653 letters and dd are digits.
4654
4655
4656 COMMENTS
4657
4658 The sequence (?# marks the start of a comment that continues up to the
4659 next closing parenthesis. Nested parentheses are not permitted. The
4660 characters that make up a comment play no part in the pattern matching
4661 at all.
4662
4663 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4664 character class introduces a comment that continues to immediately
4665 after the next newline in the pattern.
4666
4667
4668 RECURSIVE PATTERNS
4669
4670 Consider the problem of matching a string in parentheses, allowing for
4671 unlimited nested parentheses. Without the use of recursion, the best
4672 that can be done is to use a pattern that matches up to some fixed
4673 depth of nesting. It is not possible to handle an arbitrary nesting
4674 depth.
4675
4676 For some time, Perl has provided a facility that allows regular expres-
4677 sions to recurse (amongst other things). It does this by interpolating
4678 Perl code in the expression at run time, and the code can refer to the
4679 expression itself. A Perl pattern using code interpolation to solve the
4680 parentheses problem can be created like this:
4681
4682 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4683
4684 The (?p{...}) item interpolates Perl code at run time, and in this case
4685 refers recursively to the pattern in which it appears.
4686
4687 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4688 it supports special syntax for recursion of the entire pattern, and
4689 also for individual subpattern recursion. After its introduction in
4690 PCRE and Python, this kind of recursion was introduced into Perl at
4691 release 5.10.
4692
4693 A special item that consists of (? followed by a number greater than
4694 zero and a closing parenthesis is a recursive call of the subpattern of
4695 the given number, provided that it occurs inside that subpattern. (If
4696 not, it is a "subroutine" call, which is described in the next sec-
4697 tion.) The special item (?R) or (?0) is a recursive call of the entire
4698 regular expression.
4699
4700 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4701 always treated as an atomic group. That is, once it has matched some of
4702 the subject string, it is never re-entered, even if it contains untried
4703 alternatives and there is a subsequent matching failure.
4704
4705 This PCRE pattern solves the nested parentheses problem (assume the
4706 PCRE_EXTENDED option is set so that white space is ignored):
4707
4708 \( ( (?>[^()]+) | (?R) )* \)
4709
4710 First it matches an opening parenthesis. Then it matches any number of
4711 substrings which can either be a sequence of non-parentheses, or a
4712 recursive match of the pattern itself (that is, a correctly parenthe-
4713 sized substring). Finally there is a closing parenthesis.
4714
4715 If this were part of a larger pattern, you would not want to recurse
4716 the entire pattern, so instead you could use this:
4717
4718 ( \( ( (?>[^()]+) | (?1) )* \) )
4719
4720 We have put the pattern into parentheses, and caused the recursion to
4721 refer to them instead of the whole pattern.
4722
4723 In a larger pattern, keeping track of parenthesis numbers can be
4724 tricky. This is made easier by the use of relative references. (A Perl
4725 5.10 feature.) Instead of (?1) in the pattern above you can write
4726 (?-2) to refer to the second most recently opened parentheses preceding
4727 the recursion. In other words, a negative number counts capturing
4728 parentheses leftwards from the point at which it is encountered.
4729
4730 It is also possible to refer to subsequently opened parentheses, by
4731 writing references such as (?+2). However, these cannot be recursive
4732 because the reference is not inside the parentheses that are refer-
4733 enced. They are always "subroutine" calls, as described in the next
4734 section.
4735
4736 An alternative approach is to use named parentheses instead. The Perl
4737 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4738 supported. We could rewrite the above example as follows:
4739
4740 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4741
4742 If there is more than one subpattern with the same name, the earliest
4743 one is used.
4744
4745 This particular example pattern that we have been looking at contains
4746 nested unlimited repeats, and so the use of atomic grouping for match-
4747 ing strings of non-parentheses is important when applying the pattern
4748 to strings that do not match. For example, when this pattern is applied
4749 to
4750
4751 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4752
4753 it yields "no match" quickly. However, if atomic grouping is not used,
4754 the match runs for a very long time indeed because there are so many
4755 different ways the + and * repeats can carve up the subject, and all
4756 have to be tested before failure can be reported.
4757
4758 At the end of a match, the values set for any capturing subpatterns are
4759 those from the outermost level of the recursion at which the subpattern
4760 value is set. If you want to obtain intermediate values, a callout
4761 function can be used (see below and the pcrecallout documentation). If
4762 the pattern above is matched against
4763
4764 (ab(cd)ef)
4765
4766 the value for the capturing parentheses is "ef", which is the last
4767 value taken on at the top level. If additional parentheses are added,
4768 giving
4769
4770 \( ( ( (?>[^()]+) | (?R) )* ) \)
4771 ^ ^
4772 ^ ^
4773
4774 the string they capture is "ab(cd)ef", the contents of the top level
4775 parentheses. If there are more than 15 capturing parentheses in a pat-
4776 tern, PCRE has to obtain extra memory to store data during a recursion,
4777 which it does by using pcre_malloc, freeing it via pcre_free after-
4778 wards. If no memory can be obtained, the match fails with the
4779 PCRE_ERROR_NOMEMORY error.
4780
4781 Do not confuse the (?R) item with the condition (R), which tests for
4782 recursion. Consider this pattern, which matches text in angle brack-
4783 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4784 brackets (that is, when recursing), whereas any characters are permit-
4785 ted at the outer level.
4786
4787 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4788
4789 In this pattern, (?(R) is the start of a conditional subpattern, with
4790 two different alternatives for the recursive and non-recursive cases.
4791 The (?R) item is the actual recursive call.
4792
4793
4794 SUBPATTERNS AS SUBROUTINES
4795
4796 If the syntax for a recursive subpattern reference (either by number or
4797 by name) is used outside the parentheses to which it refers, it oper-
4798 ates like a subroutine in a programming language. The "called" subpat-
4799 tern may be defined before or after the reference. A numbered reference
4800 can be absolute or relative, as in these examples:
4801
4802 (...(absolute)...)...(?2)...
4803 (...(relative)...)...(?-1)...
4804 (...(?+1)...(relative)...
4805
4806 An earlier example pointed out that the pattern
4807
4808 (sens|respons)e and \1ibility
4809
4810 matches "sense and sensibility" and "response and responsibility", but
4811 not "sense and responsibility". If instead the pattern
4812
4813 (sens|respons)e and (?1)ibility
4814
4815 is used, it does match "sense and responsibility" as well as the other
4816 two strings. Another example is given in the discussion of DEFINE
4817 above.
4818
4819 Like recursive subpatterns, a "subroutine" call is always treated as an
4820 atomic group. That is, once it has matched some of the subject string,
4821 it is never re-entered, even if it contains untried alternatives and
4822 there is a subsequent matching failure.
4823
4824 When a subpattern is used as a subroutine, processing options such as
4825 case-independence are fixed when the subpattern is defined. They cannot
4826 be changed for different calls. For example, consider this pattern:
4827
4828 (abc)(?i:(?-1))
4829
4830 It matches "abcabc". It does not match "abcABC" because the change of
4831 processing option does not affect the called subpattern.
4832
4833
4834 ONIGURUMA SUBROUTINE SYNTAX
4835
4836 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4837 name or a number enclosed either in angle brackets or single quotes, is
4838 an alternative syntax for referencing a subpattern as a subroutine,
4839 possibly recursively. Here are two of the examples used above, rewrit-
4840 ten using this syntax:
4841
4842 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
4843 (sens|respons)e and \g'1'ibility
4844
4845 PCRE supports an extension to Oniguruma: if a number is preceded by a
4846 plus or a minus sign it is taken as a relative reference. For example:
4847
4848 (abc)(?i:\g<-1>)
4849
4850 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
4851 synonymous. The former is a back reference; the latter is a subroutine
4852 call.
4853
4854
4855 CALLOUTS
4856
4857 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4858 Perl code to be obeyed in the middle of matching a regular expression.
4859 This makes it possible, amongst other things, to extract different sub-
4860 strings that match the same pair of parentheses when there is a repeti-
4861 tion.
4862
4863 PCRE provides a similar feature, but of course it cannot obey arbitrary
4864 Perl code. The feature is called "callout". The caller of PCRE provides
4865 an external function by putting its entry point in the global variable
4866 pcre_callout. By default, this variable contains NULL, which disables
4867 all calling out.
4868
4869 Within a regular expression, (?C) indicates the points at which the
4870 external function is to be called. If you want to identify different
4871 callout points, you can put a number less than 256 after the letter C.
4872 The default value is zero. For example, this pattern has two callout
4873 points:
4874
4875 (?C1)abc(?C2)def
4876
4877 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4878 automatically installed before each item in the pattern. They are all
4879 numbered 255.
4880
4881 During matching, when PCRE reaches a callout point (and pcre_callout is
4882 set), the external function is called. It is provided with the number
4883 of the callout, the position in the pattern, and, optionally, one item
4884 of data originally supplied by the caller of pcre_exec(). The callout
4885 function may cause matching to proceed, to backtrack, or to fail alto-
4886 gether. A complete description of the interface to the callout function
4887 is given in the pcrecallout documentation.
4888
4889
4890 BACKTRACKING CONTROL
4891
4892 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4893 which are described in the Perl documentation as "experimental and sub-
4894 ject to change or removal in a future version of Perl". It goes on to
4895 say: "Their usage in production code should be noted to avoid problems
4896 during upgrades." The same remarks apply to the PCRE features described
4897 in this section.
4898
4899 Since these verbs are specifically related to backtracking, most of
4900 them can be used only when the pattern is to be matched using
4901 pcre_exec(), which uses a backtracking algorithm. With the exception of
4902 (*FAIL), which behaves like a failing negative assertion, they cause an
4903 error if encountered by pcre_dfa_exec().
4904
4905 The new verbs make use of what was previously invalid syntax: an open-
4906 ing parenthesis followed by an asterisk. In Perl, they are generally of
4907 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4908 its general form is just (*VERB). Any number of these verbs may occur
4909 in a pattern. There are two kinds:
4910
4911 Verbs that act immediately
4912
4913 The following verbs act as soon as they are encountered:
4914
4915 (*ACCEPT)
4916
4917 This verb causes the match to end successfully, skipping the remainder
4918 of the pattern. When inside a recursion, only the innermost pattern is
4919 ended immediately. PCRE differs from Perl in what happens if the
4920 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4921 captured: in PCRE no data is captured. For example:
4922
4923 A(A|B(*ACCEPT)|C)D
4924
4925 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4926 is captured.
4927
4928 (*FAIL) or (*F)
4929
4930 This verb causes the match to fail, forcing backtracking to occur. It
4931 is equivalent to (?!) but easier to read. The Perl documentation notes
4932 that it is probably useful only when combined with (?{}) or (??{}).
4933 Those are, of course, Perl features that are not present in PCRE. The
4934 nearest equivalent is the callout feature, as for example in this pat-
4935 tern:
4936
4937 a+(?C)(*FAIL)
4938
4939 A match with the string "aaaa" always fails, but the callout is taken
4940 before each backtrack happens (in this example, 10 times).
4941
4942 Verbs that act after backtracking
4943
4944 The following verbs do nothing when they are encountered. Matching con-
4945 tinues with what follows, but if there is no subsequent match, a fail-
4946 ure is forced. The verbs differ in exactly what kind of failure
4947 occurs.
4948
4949 (*COMMIT)
4950
4951 This verb causes the whole match to fail outright if the rest of the
4952 pattern does not match. Even if the pattern is unanchored, no further
4953 attempts to find a match by advancing the start point take place. Once
4954 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4955 at the current starting point, or not at all. For example:
4956
4957 a+(*COMMIT)b
4958
4959 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4960 of dynamic anchor, or "I've started, so I must finish."
4961
4962 (*PRUNE)
4963
4964 This verb causes the match to fail at the current position if the rest
4965 of the pattern does not match. If the pattern is unanchored, the normal
4966 "bumpalong" advance to the next starting character then happens. Back-
4967 tracking can occur as usual to the left of (*PRUNE), or when matching
4968 to the right of (*PRUNE), but if there is no match to the right, back-
4969 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4970 is just an alternative to an atomic group or possessive quantifier, but
4971 there are some uses of (*PRUNE) that cannot be expressed in any other
4972 way.
4973
4974 (*SKIP)
4975
4976 This verb is like (*PRUNE), except that if the pattern is unanchored,
4977 the "bumpalong" advance is not to the next character, but to the posi-
4978 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4979 that whatever text was matched leading up to it cannot be part of a
4980 successful match. Consider:
4981
4982 a+(*SKIP)b
4983
4984 If the subject is "aaaac...", after the first match attempt fails
4985 (starting at the first character in the string), the starting point
4986 skips on to start the next attempt at "c". Note that a possessive quan-
4987 tifer does not have the same effect in this example; although it would
4988 suppress backtracking during the first match attempt, the second
4989 attempt would start at the second character instead of skipping on to
4990 "c".
4991
4992 (*THEN)
4993
4994 This verb causes a skip to the next alternation if the rest of the pat-
4995 tern does not match. That is, it cancels pending backtracking, but only
4996 within the current alternation. Its name comes from the observation
4997 that it can be used for a pattern-based if-then-else block:
4998
4999 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5000
5001 If the COND1 pattern matches, FOO is tried (and possibly further items
5002 after the end of the group if FOO succeeds); on failure the matcher
5003 skips to the second alternative and tries COND2, without backtracking
5004 into COND1. If (*THEN) is used outside of any alternation, it acts
5005 exactly like (*PRUNE).
5006
5007
5008 SEE ALSO
5009
5010 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5011
5012
5013 AUTHOR
5014
5015 Philip Hazel
5016 University Computing Service
5017 Cambridge CB2 3QH, England.
5018
5019
5020 REVISION
5021
5022 Last updated: 08 March 2009
5023 Copyright (c) 1997-2009 University of Cambridge.
5024 ------------------------------------------------------------------------------
5025
5026
5027 PCRESYNTAX(3) PCRESYNTAX(3)
5028
5029
5030 NAME
5031 PCRE - Perl-compatible regular expressions
5032
5033
5034 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5035
5036 The full syntax and semantics of the regular expressions that are sup-
5037 ported by PCRE are described in the pcrepattern documentation. This
5038 document contains just a quick-reference summary of the syntax.
5039
5040
5041 QUOTING
5042
5043 \x where x is non-alphanumeric is a literal x
5044 \Q...\E treat enclosed characters as literal
5045
5046
5047 CHARACTERS
5048
5049 \a alarm, that is, the BEL character (hex 07)
5050 \cx "control-x", where x is any character
5051 \e escape (hex 1B)
5052 \f formfeed (hex 0C)
5053 \n newline (hex 0A)
5054 \r carriage return (hex 0D)
5055 \t tab (hex 09)
5056 \ddd character with octal code ddd, or backreference
5057 \xhh character with hex code hh
5058 \x{hhh..} character with hex code hhh..
5059
5060
5061 CHARACTER TYPES
5062
5063 . any character except newline;
5064 in dotall mode, any character whatsoever
5065 \C one byte, even in UTF-8 mode (best avoided)
5066 \d a decimal digit
5067 \D a character that is not a decimal digit
5068 \h a horizontal whitespace character
5069 \H a character that is not a horizontal whitespace character
5070 \p{xx} a character with the xx property
5071 \P{xx} a character without the xx property
5072 \R a newline sequence
5073 \s a whitespace character
5074 \S a character that is not a whitespace character
5075 \v a vertical whitespace character
5076 \V a character that is not a vertical whitespace character
5077 \w a "word" character
5078 \W a "non-word" character
5079 \X an extended Unicode sequence
5080
5081 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
5082
5083
5084 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
5085
5086 C Other
5087 Cc Control
5088 Cf Format
5089 Cn Unassigned
5090 Co Private use
5091 Cs Surrogate
5092
5093 L Letter
5094 Ll Lower case letter
5095 Lm Modifier letter
5096 Lo Other letter
5097 Lt Title case letter
5098 Lu Upper case letter
5099 L& Ll, Lu, or Lt
5100
5101 M Mark
5102 Mc Spacing mark
5103 Me Enclosing mark
5104 Mn Non-spacing mark
5105
5106 N Number
5107 Nd Decimal number
5108 Nl Letter number
5109 No Other number
5110
5111 P Punctuation
5112 Pc Connector punctuation
5113 Pd Dash punctuation
5114 Pe Close punctuation
5115 Pf Final punctuation
5116 Pi Initial punctuation
5117 Po Other punctuation
5118 Ps Open punctuation
5119
5120 S Symbol
5121 Sc Currency symbol
5122 Sk Modifier symbol
5123 Sm Mathematical symbol
5124 So Other symbol
5125
5126 Z Separator
5127 Zl Line separator
5128 Zp Paragraph separator
5129 Zs Space separator
5130
5131
5132 SCRIPT NAMES FOR \p AND \P
5133
5134 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
5135 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
5136 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
5137 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
5138 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
5139 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
5140 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
5141 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
5142 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
5143
5144
5145 CHARACTER CLASSES
5146
5147 [...] positive character class
5148 [^...] negative character class
5149 [x-y] range (can be used for hex characters)
5150 [[:xxx:]] positive POSIX named set
5151 [[:^xxx:]] negative POSIX named set
5152
5153 alnum alphanumeric
5154 alpha alphabetic
5155 ascii 0-127
5156 blank space or tab
5157 cntrl control character
5158 digit decimal digit
5159 graph printing, excluding space
5160 lower lower case letter
5161 print printing, including space
5162 punct printing, excluding alphanumeric
5163 space whitespace
5164 upper upper case letter
5165 word same as \w
5166 xdigit hexadecimal digit
5167
5168 In PCRE, POSIX character set names recognize only ASCII characters. You
5169 can use \Q...\E inside a character class.
5170
5171
5172 QUANTIFIERS
5173
5174 ? 0 or 1, greedy
5175 ?+ 0 or 1, possessive
5176 ?? 0 or 1, lazy
5177 * 0 or more, greedy
5178 *+ 0 or more, possessive
5179 *? 0 or more, lazy
5180 + 1 or more, greedy
5181 ++ 1 or more, possessive
5182 +? 1 or more, lazy
5183 {n} exactly n
5184 {n,m} at least n, no more than m, greedy
5185 {n,m}+ at least n, no more than m, possessive
5186 {n,m}? at least n, no more than m, lazy
5187 {n,} n or more, greedy
5188 {n,}+ n or more, possessive
5189 {n,}? n or more, lazy
5190
5191
5192 ANCHORS AND SIMPLE ASSERTIONS
5193
5194 \b word boundary
5195 \B not a word boundary
5196 ^ start of subject
5197 also after internal newline in multiline mode
5198 \A start of subject
5199 $ end of subject
5200 also before newline at end of subject
5201 also before internal newline in multiline mode
5202 \Z end of subject
5203 also before newline at end of subject
5204 \z end of subject
5205 \G first matching position in subject
5206
5207
5208 MATCH POINT RESET
5209
5210 \K reset start of match
5211
5212
5213 ALTERNATION
5214
5215 expr|expr|expr...
5216
5217
5218 CAPTURING
5219
5220 (...) capturing group
5221 (?<name>...) named capturing group (Perl)
5222 (?'name'...) named capturing group (Perl)
5223 (?P<name>...) named capturing group (Python)
5224 (?:...) non-capturing group
5225 (?|...) non-capturing group; reset group numbers for
5226 capturing groups in each alternative
5227
5228
5229 ATOMIC GROUPS
5230
5231 (?>...) atomic, non-capturing group
5232
5233
5234 COMMENT
5235
5236 (?#....) comment (not nestable)
5237
5238
5239 OPTION SETTING
5240
5241 (?i) caseless
5242 (?J) allow duplicate names
5243 (?m) multiline
5244 (?s) single line (dotall)
5245 (?U) default ungreedy (lazy)
5246 (?x) extended (ignore white space)
5247 (?-...) unset option(s)
5248
5249
5250 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5251
5252 (?=...) positive look ahead
5253 (?!...) negative look ahead
5254 (?<=...) positive look behind
5255 (?<!...) negative look behind
5256
5257 Each top-level branch of a look behind must be of a fixed length.
5258
5259
5260 BACKREFERENCES
5261
5262 \n reference by number (can be ambiguous)
5263 \gn reference by number
5264 \g{n} reference by number
5265 \g{-n} relative reference by number
5266 \k<name> reference by name (Perl)
5267 \k'name' reference by name (Perl)
5268 \g{name} reference by name (Perl)
5269 \k{name} reference by name (.NET)
5270 (?P=name) reference by name (Python)
5271
5272
5273 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5274
5275 (?R) recurse whole pattern
5276 (?n) call subpattern by absolute number
5277 (?+n) call subpattern by relative number
5278 (?-n) call subpattern by relative number
5279 (?&name) call subpattern by name (Perl)
5280 (?P>name) call subpattern by name (Python)
5281 \g<name> call subpattern by name (Oniguruma)
5282 \g'name' call subpattern by name (Oniguruma)
5283 \g<n> call subpattern by absolute number (Oniguruma)
5284 \g'n' call subpattern by absolute number (Oniguruma)
5285 \g<+n> call subpattern by relative number (PCRE extension)
5286 \g'+n' call subpattern by relative number (PCRE extension)
5287 \g<-n> call subpattern by relative number (PCRE extension)
5288 \g'-n' call subpattern by relative number (PCRE extension)
5289
5290
5291 CONDITIONAL PATTERNS
5292
5293 (?(condition)yes-pattern)
5294 (?(condition)yes-pattern|no-pattern)
5295
5296 (?(n)... absolute reference condition
5297 (?(+n)... relative reference condition
5298 (?(-n)... relative reference condition
5299 (?(<name>)... named reference condition (Perl)
5300 (?('name')... named reference condition (Perl)
5301 (?(name)... named reference condition (PCRE)
5302 (?(R)... overall recursion condition
5303 (?(Rn)... specific group recursion condition
5304 (?(R&name)... specific recursion condition
5305 (?(DEFINE)... define subpattern for reference
5306 (?(assert)... assertion condition
5307
5308
5309 BACKTRACKING CONTROL
5310
5311 The following act immediately they are reached:
5312
5313 (*ACCEPT) force successful match
5314 (*FAIL) force backtrack; synonym (*F)
5315
5316 The following act only when a subsequent match failure causes a back-
5317 track to reach them. They all force a match failure, but they differ in
5318 what happens afterwards. Those that advance the start-of-match point do
5319 so only if the pattern is not anchored.
5320
5321 (*COMMIT) overall failure, no advance of starting point
5322 (*PRUNE) advance to next starting character
5323 (*SKIP) advance start to current matching position
5324 (*THEN) local failure, backtrack to next alternation
5325
5326
5327 NEWLINE CONVENTIONS
5328
5329 These are recognized only at the very start of the pattern or after a
5330 (*BSR_...) option.
5331
5332 (*CR)
5333 (*LF)
5334 (*CRLF)
5335 (*ANYCRLF)
5336 (*ANY)
5337
5338
5339 WHAT \R MATCHES
5340
5341 These are recognized only at the very start of the pattern or after a
5342 (*...) option that sets the newline convention.
5343
5344 (*BSR_ANYCRLF)
5345 (*BSR_UNICODE)
5346
5347
5348 CALLOUTS
5349
5350 (?C) callout
5351 (?Cn) callout with data n
5352
5353
5354 SEE ALSO
5355
5356 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5357
5358
5359 AUTHOR
5360
5361 Philip Hazel
5362 University Computing Service
5363 Cambridge CB2 3QH, England.
5364
5365
5366 REVISION
5367
5368 Last updated: 09 April 2008
5369 Copyright (c) 1997-2008 University of Cambridge.
5370 ------------------------------------------------------------------------------
5371
5372
5373 PCREPARTIAL(3) PCREPARTIAL(3)
5374
5375
5376 NAME
5377 PCRE - Perl-compatible regular expressions
5378
5379
5380 PARTIAL MATCHING IN PCRE
5381
5382 In normal use of PCRE, if the subject string that is passed to
5383 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5384 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5385 There are circumstances where it might be helpful to distinguish this
5386 case from other cases in which there is no match.
5387
5388 Consider, for example, an application where a human is required to type
5389 in data for a field with specific formatting requirements. An example
5390 might be a date in the form ddmmmyy, defined by this pattern:
5391
5392 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5393
5394 If the application sees the user's keystrokes one by one, and can check
5395 that what has been typed so far is potentially valid, it is able to
5396 raise an error as soon as a mistake is made, possibly beeping and not
5397 reflecting the character that has been typed. This immediate feedback
5398 is likely to be a better user interface than a check that is delayed
5399 until the entire string has been entered.
5400
5401 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5402 TIAL option, which can be set when calling pcre_exec() or
5403 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5404 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5405 during the matching process the last part of the subject string matched
5406 part of the pattern. Unfortunately, for non-anchored matching, it is
5407 not possible to obtain the position of the start of the partial match.
5408 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5409
5410 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5411 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5412 the subject is reached, there have been no complete matches, but there
5413 is still at least one matching possibility. The portion of the string
5414 that provided the partial match is set as the first matching string.
5415
5416 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5417 the last literal byte in a pattern, and abandons matching immediately
5418 if such a byte is not present in the subject string. This optimization
5419 cannot be used for a subject string that might match only partially.
5420
5421
5422 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5423
5424 Because of the way certain internal optimizations are implemented in
5425 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5426 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5427 used. For pcre_exec(), repeated single characters such as
5428
5429 a{2,4}
5430
5431 and repeated single metasequences such as
5432
5433 \d+
5434
5435 are not permitted if the maximum number of occurrences is greater than
5436 one. Optional items such as \d? (where the maximum is one) are permit-
5437 ted. Quantifiers with any values are permitted after parentheses, so
5438 the invalid examples above can be coded thus:
5439
5440 (a){2,4}
5441 (\d)+
5442
5443 These constructions run more slowly, but for the kinds of application
5444 that are envisaged for this facility, this is not felt to be a major
5445 restriction.
5446
5447 If PCRE_PARTIAL is set for a pattern that does not conform to the
5448 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5449 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5450 find out if a compiled pattern can be used for partial matching.
5451
5452
5453 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5454
5455 If the escape sequence \P is present in a pcretest data line, the
5456 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5457 uses the date example quoted above:
5458
5459 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5460 data> 25jun04\P
5461 0: 25jun04
5462 1: jun
5463 data> 25dec3\P
5464 Partial match
5465 data> 3ju\P
5466 Partial match
5467 data> 3juj\P
5468 No match
5469 data> j\P
5470 No match
5471
5472 The first data string is matched completely, so pcretest shows the
5473 matched substrings. The remaining four strings do not match the com-
5474 plete pattern, but the first two are partial matches. The same test,
5475 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5476 produces the following output:
5477
5478 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5479 data> 25jun04\P\D
5480 0: 25jun04
5481 data> 23dec3\P\D
5482 Partial match: 23dec3
5483 data> 3ju\P\D
5484 Partial match: 3ju
5485 data> 3juj\P\D
5486 No match
5487 data> j\P\D
5488 No match
5489
5490 Notice that in this case the portion of the string that was matched is
5491 made available.
5492
5493
5494 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5495
5496 When a partial match has been found using pcre_dfa_exec(), it is possi-
5497 ble to continue the match by providing additional subject data and
5498 calling pcre_dfa_exec() again with the same compiled regular expres-
5499 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5500 the same working space as before, because this is where details of the
5501 previous partial match are stored. Here is an example using pcretest,
5502 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5503 \D are as above):
5504
5505 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5506 data> 23ja\P\D
5507 Partial match: 23ja
5508 data> n05\R\D
5509 0: n05
5510
5511 The first call has "23ja" as the subject, and requests partial match-
5512 ing; the second call has "n05" as the subject for the continued
5513 (restarted) match. Notice that when the match is complete, only the
5514 last part is shown; PCRE does not retain the previously partially-
5515 matched string. It is up to the calling program to do that if it needs
5516 to.
5517
5518 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5519 matching over multiple segments. This facility can be used to pass very
5520 long subject strings to pcre_dfa_exec(). However, some care is needed
5521 for certain types of pattern.
5522
5523 1. If the pattern contains tests for the beginning or end of a line,
5524 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5525 ate, when the subject string for any call does not contain the begin-
5526 ning or end of a line.
5527
5528 2. If the pattern contains backward assertions (including \b or \B),
5529 you need to arrange for some overlap in the subject strings to allow
5530 for this. For example, you could pass the subject in chunks that are
5531 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5532 set to 200 and the previous 200 bytes at the start of the buffer.
5533
5534 3. Matching a subject string that is split into multiple segments does
5535 not always produce exactly the same result as matching over one single
5536 long string. The difference arises when there are multiple matching
5537 possibilities, because a partial match result is given only when there
5538 are no completed matches in a call to pcre_dfa_exec(). This means that
5539 as soon as the shortest match has been found, continuation to a new
5540 subject segment is no longer possible. Consider this pcretest example:
5541
5542 re> /dog(sbody)?/
5543 data> do\P\D
5544 Partial match: do
5545 data> gsb\R\P\D
5546 0: g
5547 data> dogsbody\D
5548 0: dogsbody
5549 1: dog
5550
5551 The pattern matches the words "dog" or "dogsbody". When the subject is
5552 presented in several parts ("do" and "gsb" being the first two) the
5553 match stops when "dog" has been found, and it is not possible to con-
5554 tinue. On the other hand, if "dogsbody" is presented as a single
5555 string, both matches are found.
5556
5557 Because of this phenomenon, it does not usually make sense to end a
5558 pattern that is going to be matched in this way with a variable repeat.
5559
5560 4. Patterns that contain alternatives at the top level which do not all
5561 start with the same pattern item may not work as expected. For example,
5562 consider this pattern:
5563
5564 1234|3789
5565
5566 If the first part of the subject is "ABC123", a partial match of the
5567 first alternative is found at offset 3. There is no partial match for
5568 the second alternative, because such a match does not start at the same
5569 point in the subject string. Attempting to continue with the string
5570 "789" does not yield a match because only those alternatives that match
5571 at one point in the subject are remembered. The problem arises because
5572 the start of the second alternative matches within the first alterna-
5573 tive. There is no problem with anchored patterns or patterns such as:
5574
5575 1234|ABCD
5576
5577 where no string can be a partial match for both alternatives.
5578
5579
5580 AUTHOR
5581
5582 Philip Hazel
5583 University Computing Service
5584 Cambridge CB2 3QH, England.
5585
5586
5587 REVISION
5588
5589 Last updated: 04 June 2007
5590 Copyright (c) 1997-2007 University of Cambridge.
5591 ------------------------------------------------------------------------------
5592
5593
5594 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5595
5596
5597 NAME
5598 PCRE - Perl-compatible regular expressions
5599
5600
5601 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5602
5603 If you are running an application that uses a large number of regular
5604 expression patterns, it may be useful to store them in a precompiled
5605 form instead of having to compile them every time the application is
5606 run. If you are not using any private character tables (see the
5607 pcre_maketables() documentation), this is relatively straightforward.
5608 If you are using private tables, it is a little bit more complicated.
5609
5610 If you save compiled patterns to a file, you can copy them to a differ-
5611 ent host and run them there. This works even if the new host has the
5612 opposite endianness to the one on which the patterns were compiled.
5613 There may be a small performance penalty, but it should be insignifi-
5614 cant. However, compiling regular expressions with one version of PCRE
5615 for use with a different version is not guaranteed to work and may
5616 cause crashes.
5617
5618
5619 SAVING A COMPILED PATTERN
5620 The value returned by pcre_compile() points to a single block of memory
5621 that holds the compiled pattern and associated data. You can find the
5622 length of this block in bytes by calling pcre_fullinfo() with an argu-
5623 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5624 manner. Here is sample code that compiles a pattern and writes it to a
5625 file. It assumes that the variable fd refers to a file that is open for
5626 output:
5627
5628 int erroroffset, rc, size;
5629 char *error;
5630 pcre *re;
5631
5632 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5633 if (re == NULL) { ... handle errors ... }
5634 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5635 if (rc < 0) { ... handle errors ... }
5636 rc = fwrite(re, 1, size, fd);
5637 if (rc != size) { ... handle errors ... }
5638
5639 In this example, the bytes that comprise the compiled pattern are
5640 copied exactly. Note that this is binary data that may contain any of
5641 the 256 possible byte values. On systems that make a distinction
5642 between binary and non-binary data, be sure that the file is opened for
5643 binary output.
5644
5645 If you want to write more than one pattern to a file, you will have to
5646 devise a way of separating them. For binary data, preceding each pat-
5647 tern with its length is probably the most straightforward approach.
5648 Another possibility is to write out the data in hexadecimal instead of
5649 binary, one pattern to a line.
5650
5651 Saving compiled patterns in a file is only one possible way of storing
5652 them for later use. They could equally well be saved in a database, or
5653 in the memory of some daemon process that passes them via sockets to
5654 the processes that want them.
5655
5656 If the pattern has been studied, it is also possible to save the study
5657 data in a similar way to the compiled pattern itself. When studying
5658 generates additional information, pcre_study() returns a pointer to a
5659 pcre_extra data block. Its format is defined in the section on matching
5660 a pattern in the pcreapi documentation. The study_data field points to
5661 the binary study data, and this is what you must save (not the
5662 pcre_extra block itself). The length of the study data can be obtained
5663 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5664 Remember to check that pcre_study() did return a non-NULL value before
5665 trying to save the study data.
5666
5667
5668 RE-USING A PRECOMPILED PATTERN
5669
5670 Re-using a precompiled pattern is straightforward. Having reloaded it
5671 into main memory, you pass its pointer to pcre_exec() or
5672 pcre_dfa_exec() in the usual way. This should work even on another
5673 host, and even if that host has the opposite endianness to the one
5674 where the pattern was compiled.
5675
5676 However, if you passed a pointer to custom character tables when the
5677 pattern was compiled (the tableptr argument of pcre_compile()), you
5678 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5679 because the value saved with the compiled pattern will obviously be
5680 nonsense. A field in a pcre_extra() block is used to pass this data, as
5681 described in the section on matching a pattern in the pcreapi documen-
5682 tation.
5683
5684 If you did not provide custom character tables when the pattern was
5685 compiled, the pointer in the compiled pattern is NULL, which causes
5686 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5687 take any special action at run time in this case.
5688
5689 If you saved study data with the compiled pattern, you need to create
5690 your own pcre_extra data block and set the study_data field to point to
5691 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5692 bit in the flags field to indicate that study data is present. Then
5693 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5694 usual way.
5695
5696
5697 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5698
5699 In general, it is safest to recompile all saved patterns when you
5700 update to a new PCRE release, though not all updates actually require
5701 this. Recompiling is definitely needed for release 7.2.
5702
5703
5704 AUTHOR
5705
5706 Philip Hazel
5707 University Computing Service
5708 Cambridge CB2 3QH, England.
5709
5710
5711 REVISION
5712
5713 Last updated: 13 June 2007
5714 Copyright (c) 1997-2007 University of Cambridge.
5715 ------------------------------------------------------------------------------
5716
5717
5718 PCREPERFORM(3) PCREPERFORM(3)
5719
5720
5721 NAME
5722 PCRE - Perl-compatible regular expressions
5723
5724
5725 PCRE PERFORMANCE
5726
5727 Two aspects of performance are discussed below: memory usage and pro-
5728 cessing time. The way you express your pattern as a regular expression
5729 can affect both of them.
5730
5731
5732 MEMORY USAGE
5733
5734 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5735 that most simple patterns do not use much memory. However, there is one
5736 case where memory usage can be unexpectedly large. When a parenthesized
5737 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5738 ited maximum, the whole subpattern is repeated in the compiled code.
5739 For example, the pattern
5740
5741 (abc|def){2,4}
5742
5743 is compiled as if it were
5744
5745 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5746
5747 (Technical aside: It is done this way so that backtrack points within
5748 each of the repetitions can be independently maintained.)
5749
5750 For regular expressions whose quantifiers use only small numbers, this
5751 is not usually a problem. However, if the numbers are large, and par-
5752 ticularly if such repetitions are nested, the memory usage can become
5753 an embarrassment. For example, the very simple pattern
5754
5755 ((ab){1,1000}c){1,3}
5756
5757 uses 51K bytes when compiled. When PCRE is compiled with its default
5758 internal pointer size of two bytes, the size limit on a compiled pat-
5759 tern is 64K, and this is reached with the above pattern if the outer
5760 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5761 internal pointers and thus handle larger compiled patterns, but it is
5762 better to try to rewrite your pattern to use less memory if you can.
5763
5764 One way of reducing the memory usage for such patterns is to make use
5765 of PCRE's "subroutine" facility. Re-writing the above pattern as
5766
5767 ((ab)(?2){0,999}c)(?1){0,2}
5768
5769 reduces the memory requirements to 18K, and indeed it remains under 20K
5770 even with the outer repetition increased to 100. However, this pattern
5771 is not exactly equivalent, because the "subroutine" calls are treated
5772 as atomic groups into which there can be no backtracking if there is a
5773 subsequent matching failure. Therefore, PCRE cannot do this kind of
5774 rewriting automatically. Furthermore, there is a noticeable loss of
5775 speed when executing the modified pattern. Nevertheless, if the atomic
5776 grouping is not a problem and the loss of speed is acceptable, this
5777 kind of rewriting will allow you to process patterns that PCRE cannot
5778 otherwise handle.
5779
5780
5781 PROCESSING TIME
5782
5783 Certain items in regular expression patterns are processed more effi-
5784 ciently than others. It is more efficient to use a character class like
5785 [aeiou] than a set of single-character alternatives such as
5786 (a|e|i|o|u). In general, the simplest construction that provides the
5787 required behaviour is usually the most efficient. Jeffrey Friedl's book
5788 contains a lot of useful general discussion about optimizing regular
5789 expressions for efficient performance. This document contains a few
5790 observations about PCRE.
5791
5792 Using Unicode character properties (the \p, \P, and \X escapes) is
5793 slow, because PCRE has to scan a structure that contains data for over
5794 fifteen thousand characters whenever it needs a character's property.
5795 If you can find an alternative pattern that does not use character
5796 properties, it will probably be faster.
5797
5798 When a pattern begins with .* not in parentheses, or in parentheses
5799 that are not the subject of a backreference, and the PCRE_DOTALL option
5800 is set, the pattern is implicitly anchored by PCRE, since it can match
5801 only at the start of a subject string. However, if PCRE_DOTALL is not
5802 set, PCRE cannot make this optimization, because the . metacharacter
5803 does not then match a newline, and if the subject string contains new-
5804 lines, the pattern may match from the character immediately following
5805 one of them instead of from the very start. For example, the pattern
5806
5807 .*second
5808
5809 matches the subject "first\nand second" (where \n stands for a newline
5810 character), with the match starting at the seventh character. In order
5811 to do this, PCRE has to retry the match starting after every newline in
5812 the subject.
5813
5814 If you are using such a pattern with subject strings that do not con-
5815 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5816 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5817 ing. That saves PCRE from having to scan along the subject looking for
5818 a newline to restart at.
5819
5820 Beware of patterns that contain nested indefinite repeats. These can
5821 take a long time to run when applied to a string that does not match.
5822 Consider the pattern fragment
5823
5824 ^(a+)*
5825
5826 This can match "aaaa" in 16 different ways, and this number increases
5827 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5828 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5829 repeats can match different numbers of times.) When the remainder of
5830 the pattern is such that the entire match is going to fail, PCRE has in
5831 principle to try every possible variation, and this can take an
5832 extremely long time, even for relatively short strings.
5833
5834 An optimization catches some of the more simple cases such as
5835
5836 (a+)*b
5837
5838 where a literal character follows. Before embarking on the standard
5839 matching procedure, PCRE checks that there is a "b" later in the sub-
5840 ject string, and if there is not, it fails the match immediately. How-
5841 ever, when there is no following literal this optimization cannot be
5842 used. You can see the difference by comparing the behaviour of
5843
5844 (a+)*\d
5845
5846 with the pattern above. The former gives a failure almost instantly
5847 when applied to a whole line of "a" characters, whereas the latter
5848 takes an appreciable time with strings longer than about 20 characters.
5849
5850 In many cases, the solution to this kind of performance issue is to use
5851 an atomic group or a possessive quantifier.
5852
5853
5854 AUTHOR
5855
5856 Philip Hazel
5857 University Computing Service
5858 Cambridge CB2 3QH, England.
5859
5860
5861 REVISION
5862
5863 Last updated: 06 March 2007
5864 Copyright (c) 1997-2007 University of Cambridge.
5865 ------------------------------------------------------------------------------
5866
5867
5868 PCREPOSIX(3) PCREPOSIX(3)
5869
5870
5871 NAME
5872 PCRE - Perl-compatible regular expressions.
5873
5874
5875 SYNOPSIS OF POSIX API
5876
5877 #include <pcreposix.h>
5878
5879 int regcomp(regex_t *preg, const char *pattern,
5880 int cflags);
5881
5882 int regexec(regex_t *preg, const char *string,
5883 size_t nmatch, regmatch_t pmatch[], int eflags);
5884
5885 size_t regerror(int errcode, const regex_t *preg,
5886 char *errbuf, size_t errbuf_size);
5887
5888 void regfree(regex_t *preg);
5889
5890
5891 DESCRIPTION
5892
5893 This set of functions provides a POSIX-style API to the PCRE regular
5894 expression package. See the pcreapi documentation for a description of
5895 PCRE's native API, which contains much additional functionality.
5896
5897 The functions described here are just wrapper functions that ultimately
5898 call the PCRE native API. Their prototypes are defined in the
5899 pcreposix.h header file, and on Unix systems the library itself is
5900 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5901 command for linking an application that uses them. Because the POSIX
5902 functions call the native ones, it is also necessary to add -lpcre.
5903
5904 I have implemented only those POSIX option bits that can be reasonably
5905 mapped to PCRE native options. In addition, the option REG_EXTENDED is
5906 defined with the value zero. This has no effect, but since programs
5907 that are written to the POSIX interface often use it, this makes it
5908 easier to slot in PCRE as a replacement library. Other POSIX options
5909 are not even defined.
5910
5911 When PCRE is called via these functions, it is only the API that is
5912 POSIX-like in style. The syntax and semantics of the regular expres-
5913 sions themselves are still those of Perl, subject to the setting of
5914 various PCRE options, as described below. "POSIX-like in style" means
5915 that the API approximates to the POSIX definition; it is not fully
5916 POSIX-compatible, and in multi-byte encoding domains it is probably
5917 even less compatible.
5918
5919 The header for these functions is supplied as pcreposix.h to avoid any
5920 potential clash with other POSIX libraries. It can, of course, be
5921 renamed or aliased as regex.h, which is the "correct" name. It provides
5922 two structure types, regex_t for compiled internal forms, and reg-
5923 match_t for returning captured substrings. It also defines some con-
5924 stants whose names start with "REG_"; these are used for setting
5925 options and identifying error codes.
5926
5927
5928 COMPILING A PATTERN
5929
5930 The function regcomp() is called to compile a pattern into an internal
5931 form. The pattern is a C string terminated by a binary zero, and is
5932 passed in the argument pattern. The preg argument is a pointer to a
5933 regex_t structure that is used as a base for storing information about
5934 the compiled regular expression.
5935
5936 The argument cflags is either zero, or contains one or more of the bits
5937 defined by the following macros:
5938
5939 REG_DOTALL
5940
5941 The PCRE_DOTALL option is set when the regular expression is passed for
5942 compilation to the native function. Note that REG_DOTALL is not part of
5943 the POSIX standard.
5944
5945 REG_ICASE
5946
5947 The PCRE_CASELESS option is set when the regular expression is passed
5948 for compilation to the native function.
5949
5950 REG_NEWLINE
5951
5952 The PCRE_MULTILINE option is set when the regular expression is passed
5953 for compilation to the native function. Note that this does not mimic
5954 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5955 tion).
5956
5957 REG_NOSUB
5958
5959 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5960 passed for compilation to the native function. In addition, when a pat-
5961 tern that is compiled with this flag is passed to regexec() for match-
5962 ing, the nmatch and pmatch arguments are ignored, and no captured
5963 strings are returned.
5964
5965 REG_UTF8
5966
5967 The PCRE_UTF8 option is set when the regular expression is passed for
5968 compilation to the native function. This causes the pattern itself and
5969 all data strings used for matching it to be treated as UTF-8 strings.
5970 Note that REG_UTF8 is not part of the POSIX standard.
5971
5972 In the absence of these flags, no options are passed to the native
5973 function. This means the the regex is compiled with PCRE default
5974 semantics. In particular, the way it handles newline characters in the
5975 subject string is the Perl way, not the POSIX way. Note that setting
5976 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5977 It does not affect the way newlines are matched by . (they aren't) or
5978 by a negative class such as [^a] (they are).
5979
5980 The yield of regcomp() is zero on success, and non-zero otherwise. The
5981 preg structure is filled in on success, and one member of the structure
5982 is public: re_nsub contains the number of capturing subpatterns in the
5983 regular expression. Various error codes are defined in the header file.
5984
5985
5986 MATCHING NEWLINE CHARACTERS
5987
5988 This area is not simple, because POSIX and Perl take different views of
5989 things. It is not possible to get PCRE to obey POSIX semantics, but
5990 then PCRE was never intended to be a POSIX engine. The following table
5991 lists the different possibilities for matching newline characters in
5992 PCRE:
5993
5994 Default Change with
5995
5996 . matches newline no PCRE_DOTALL
5997 newline matches [^a] yes not changeable
5998 $ matches \n at end yes PCRE_DOLLARENDONLY
5999 $ matches \n in middle no PCRE_MULTILINE
6000 ^ matches \n in middle no PCRE_MULTILINE
6001
6002 This is the equivalent table for POSIX:
6003
6004 Default Change with
6005
6006 . matches newline yes REG_NEWLINE
6007 newline matches [^a] yes REG_NEWLINE
6008 $ matches \n at end no REG_NEWLINE
6009 $ matches \n in middle no REG_NEWLINE
6010 ^ matches \n in middle no REG_NEWLINE
6011
6012 PCRE's behaviour is the same as Perl's, except that there is no equiva-
6013 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
6014 no way to stop newline from matching [^a].
6015
6016 The default POSIX newline handling can be obtained by setting
6017 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
6018 behave exactly as for the REG_NEWLINE action.
6019
6020
6021 MATCHING A PATTERN
6022
6023 The function regexec() is called to match a compiled pattern preg
6024 against a given string, which is by default terminated by a zero byte
6025 (but see REG_STARTEND below), subject to the options in eflags. These
6026 can be:
6027
6028 REG_NOTBOL
6029
6030 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
6031 function.
6032
6033 REG_NOTEMPTY
6034
6035 The PCRE_NOTEMPTY option is set when calling the underlying PCRE match-
6036 ing function. Note that REG_NOTEMPTY is not part of the POSIX standard.
6037 However, setting this option can give more POSIX-like behaviour in some
6038 situations.
6039
6040 REG_NOTEOL
6041
6042 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
6043 function.
6044
6045 REG_STARTEND
6046
6047 The string is considered to start at string + pmatch[0].rm_so and to
6048 have a terminating NUL located at string + pmatch[0].rm_eo (there need
6049 not actually be a NUL at that location), regardless of the value of
6050 nmatch. This is a BSD extension, compatible with but not specified by
6051 IEEE Standard 1003.2 (POSIX.2), and should be used with caution in
6052 software intended to be portable to other systems. Note that a non-zero
6053 rm_so does not imply REG_NOTBOL; REG_STARTEND affects only the location
6054 of the string, not how it is matched.
6055
6056 If the pattern was compiled with the REG_NOSUB flag, no data about any
6057 matched strings is returned. The nmatch and pmatch arguments of
6058 regexec() are ignored.
6059
6060 Otherwise,the portion of the string that was matched, and also any cap-
6061 tured substrings, are returned via the pmatch argument, which points to
6062 an array of nmatch structures of type regmatch_t, containing the mem-
6063 bers rm_so and rm_eo. These contain the offset to the first character
6064 of each substring and the offset to the first character after the end
6065 of each substring, respectively. The 0th element of the vector relates
6066 to the entire portion of string that was matched; subsequent elements
6067 relate to the capturing subpatterns of the regular expression. Unused
6068 entries in the array have both structure members set to -1.
6069
6070 A successful match yields a zero return; various error codes are
6071 defined in the header file, of which REG_NOMATCH is the "expected"
6072 failure code.
6073
6074
6075 ERROR MESSAGES
6076
6077 The regerror() function maps a non-zero errorcode from either regcomp()
6078 or regexec() to a printable message. If preg is not NULL, the error
6079 should have arisen from the use of that structure. A message terminated
6080 by a binary zero is placed in errbuf. The length of the message,
6081 including the zero, is limited to errbuf_size. The yield of the func-
6082 tion is the size of buffer needed to hold the whole message.
6083
6084
6085 MEMORY USAGE
6086
6087 Compiling a regular expression causes memory to be allocated and asso-
6088 ciated with the preg structure. The function regfree() frees all such
6089 memory, after which preg may no longer be used as a compiled expres-
6090 sion.
6091
6092
6093 AUTHOR
6094
6095 Philip Hazel
6096 University Computing Service
6097 Cambridge CB2 3QH, England.
6098
6099