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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 247 - (show annotations)
Mon Sep 17 09:38:32 2007 UTC (7 years, 8 months ago) by ph10
File MIME type: text/plain
File size: 301740 byte(s)
Error occurred while calculating annotation data.
(1) Update tests 2,5,7,8 to run when --enable-bsr-anycrlf is used.
(2) Updates files changed by building 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.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
49 page.
50
51 Some features of PCRE can be included, excluded, or changed when the
52 library is built. The pcre_config() function makes it possible for a
53 client to discover which features are available. The features them-
54 selves are described in the pcrebuild page. Documentation about build-
55 ing PCRE for various operating systems can be found in the README file
56 in the source distribution.
57
58 The library contains a number of undocumented internal functions and
59 data tables that are used by more than one of the exported external
60 functions, but which are not intended for use by external callers.
61 Their names all begin with "_pcre_", which hopefully will not provoke
62 any name clashes. In some environments, it is possible to control which
63 external symbols are exported when a shared library is built, and in
64 these cases the undocumented symbols are not exported.
65
66
67 USER DOCUMENTATION
68
69 The user documentation for PCRE comprises a number of different sec-
70 tions. In the "man" format, each of these is a separate "man page". In
71 the HTML format, each is a separate page, linked from the index page.
72 In the plain text format, all the sections are concatenated, for ease
73 of searching. The sections are as follows:
74
75 pcre this document
76 pcre-config show PCRE installation configuration information
77 pcreapi details of PCRE's native C API
78 pcrebuild options for building PCRE
79 pcrecallout details of the callout feature
80 pcrecompat discussion of Perl compatibility
81 pcrecpp details of the C++ wrapper
82 pcregrep description of the pcregrep command
83 pcrematching discussion of the two matching algorithms
84 pcrepartial details of the partial matching facility
85 pcrepattern syntax and semantics of supported
86 regular expressions
87 pcresyntax quick syntax reference
88 pcreperform discussion of performance issues
89 pcreposix the POSIX-compatible C API
90 pcreprecompile details of saving and re-using precompiled patterns
91 pcresample discussion of the sample program
92 pcrestack discussion of stack usage
93 pcretest description of the pcretest testing command
94
95 In addition, in the "man" and HTML formats, there is a short page for
96 each C library function, listing its arguments and results.
97
98
99 LIMITATIONS
100
101 There are some size limitations in PCRE but it is hoped that they will
102 never in practice be relevant.
103
104 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
105 is compiled with the default internal linkage size of 2. If you want to
106 process regular expressions that are truly enormous, you can compile
107 PCRE with an internal linkage size of 3 or 4 (see the README file in
108 the source distribution and the pcrebuild documentation for details).
109 In these cases the limit is substantially larger. However, the speed
110 of execution is slower.
111
112 All values in repeating quantifiers must be less than 65536.
113
114 There is no limit to the number of parenthesized subpatterns, but there
115 can be no more than 65535 capturing subpatterns.
116
117 The maximum length of name for a named subpattern is 32 characters, and
118 the maximum number of named subpatterns is 10000.
119
120 The maximum length of a subject string is the largest positive number
121 that an integer variable can hold. However, when using the traditional
122 matching function, PCRE uses recursion to handle subpatterns and indef-
123 inite repetition. This means that the available stack space may limit
124 the size of a subject string that can be processed by certain patterns.
125 For a discussion of stack issues, see the pcrestack documentation.
126
127
128 UTF-8 AND UNICODE PROPERTY SUPPORT
129
130 From release 3.3, PCRE has had some support for character strings
131 encoded in the UTF-8 format. For release 4.0 this was greatly extended
132 to cover most common requirements, and in release 5.0 additional sup-
133 port for Unicode general category properties was added.
134
135 In order process UTF-8 strings, you must build PCRE to include UTF-8
136 support in the code, and, in addition, you must call pcre_compile()
137 with the PCRE_UTF8 option flag. When you do this, both the pattern and
138 any subject strings that are matched against it are treated as UTF-8
139 strings instead of just strings of bytes.
140
141 If you compile PCRE with UTF-8 support, but do not use it at run time,
142 the library will be a bit bigger, but the additional run time overhead
143 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
144 very big.
145
146 If PCRE is built with Unicode character property support (which implies
147 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
148 ported. The available properties that can be tested are limited to the
149 general category properties such as Lu for an upper case letter or Nd
150 for a decimal number, the Unicode script names such as Arabic or Han,
151 and the derived properties Any and L&. A full list is given in the
152 pcrepattern documentation. Only the short names for properties are sup-
153 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
154 ter}, is not supported. Furthermore, in Perl, many properties may
155 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
156 does not support this.
157
158 Validity of UTF-8 strings
159
160 When you set the PCRE_UTF8 flag, the strings passed as patterns and
161 subjects are (by default) checked for validity on entry to the relevant
162 functions. From release 7.3 of PCRE, the check is according the rules
163 of RFC 3629, which are themselves derived from the Unicode specifica-
164 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
165 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
166 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
167 to U+DFFF.
168
169 The excluded code points are the "Low Surrogate Area" of Unicode, of
170 which the Unicode Standard says this: "The Low Surrogate Area does not
171 contain any character assignments, consequently no character code
172 charts or namelists are provided for this area. Surrogates are reserved
173 for use with UTF-16 and then must be used in pairs." The code points
174 that are encoded by UTF-16 pairs are available as independent code
175 points in the UTF-8 encoding. (In other words, the whole surrogate
176 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
177
178 If an invalid UTF-8 string is passed to PCRE, an error return
179 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
180 that your strings are valid, and therefore want to skip these checks in
181 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
182 compile time or at run time, PCRE assumes that the pattern or subject
183 it is given (respectively) contains only valid UTF-8 codes. In this
184 case, it does not diagnose an invalid UTF-8 string.
185
186 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
187 what happens depends on why the string is invalid. If the string con-
188 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
189 string of characters in the range 0 to 0x7FFFFFFF. In other words,
190 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
191 strings according to the more liberal rules of RFC 2279. However, if
192 the string does not even conform to RFC 2279, the result is undefined.
193 Your program may crash.
194
195 If you want to process strings of values in the full range 0 to
196 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
197 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
198 this situation, you will have to apply your own validity check.
199
200 General comments about UTF-8 mode
201
202 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
203 two-byte UTF-8 character if the value is greater than 127.
204
205 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
206 characters for values greater than \177.
207
208 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
209 vidual bytes, for example: \x{100}{3}.
210
211 4. The dot metacharacter matches one UTF-8 character instead of a sin-
212 gle byte.
213
214 5. The escape sequence \C can be used to match a single byte in UTF-8
215 mode, but its use can lead to some strange effects. This facility is
216 not available in the alternative matching function, pcre_dfa_exec().
217
218 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
219 test characters of any code value, but the characters that PCRE recog-
220 nizes as digits, spaces, or word characters remain the same set as
221 before, all with values less than 256. This remains true even when PCRE
222 includes Unicode property support, because to do otherwise would slow
223 down PCRE in many common cases. If you really want to test for a wider
224 sense of, say, "digit", you must use Unicode property tests such as
225 \p{Nd}.
226
227 7. Similarly, characters that match the POSIX named character classes
228 are all low-valued characters.
229
230 8. However, the Perl 5.10 horizontal and vertical whitespace matching
231 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
232 acters.
233
234 9. Case-insensitive matching applies only to characters whose values
235 are less than 128, unless PCRE is built with Unicode property support.
236 Even when Unicode property support is available, PCRE still uses its
237 own character tables when checking the case of low-valued characters,
238 so as not to degrade performance. The Unicode property information is
239 used only for characters with higher values. Even when Unicode property
240 support is available, PCRE supports case-insensitive matching only when
241 there is a one-to-one mapping between a letter's cases. There are a
242 small number of many-to-one mappings in Unicode; these are not sup-
243 ported by PCRE.
244
245
246 AUTHOR
247
248 Philip Hazel
249 University Computing Service
250 Cambridge CB2 3QH, England.
251
252 Putting an actual email address here seems to have been a spam magnet,
253 so I've taken it away. If you want to email me, use my two initials,
254 followed by the two digits 10, at the domain cam.ac.uk.
255
256
257 REVISION
258
259 Last updated: 09 August 2007
260 Copyright (c) 1997-2007 University of Cambridge.
261 ------------------------------------------------------------------------------
262
263
264 PCREBUILD(3) PCREBUILD(3)
265
266
267 NAME
268 PCRE - Perl-compatible regular expressions
269
270
271 PCRE BUILD-TIME OPTIONS
272
273 This document describes the optional features of PCRE that can be
274 selected when the library is compiled. They are all selected, or dese-
275 lected, by providing options to the configure script that is run before
276 the make command. The complete list of options for configure (which
277 includes the standard ones such as the selection of the installation
278 directory) can be obtained by running
279
280 ./configure --help
281
282 The following sections include descriptions of options whose names
283 begin with --enable or --disable. These settings specify changes to the
284 defaults for the configure command. Because of the way that configure
285 works, --enable and --disable always come in pairs, so the complemen-
286 tary option always exists as well, but as it specifies the default, it
287 is not described.
288
289
290 C++ SUPPORT
291
292 By default, the configure script will search for a C++ compiler and C++
293 header files. If it finds them, it automatically builds the C++ wrapper
294 library for PCRE. You can disable this by adding
295
296 --disable-cpp
297
298 to the configure command.
299
300
301 UTF-8 SUPPORT
302
303 To build PCRE with support for UTF-8 character strings, add
304
305 --enable-utf8
306
307 to the configure command. Of itself, this does not make PCRE treat
308 strings as UTF-8. As well as compiling PCRE with this option, you also
309 have have to set the PCRE_UTF8 option when you call the pcre_compile()
310 function.
311
312
313 UNICODE CHARACTER PROPERTY SUPPORT
314
315 UTF-8 support allows PCRE to process character values greater than 255
316 in the strings that it handles. On its own, however, it does not pro-
317 vide any facilities for accessing the properties of such characters. If
318 you want to be able to use the pattern escapes \P, \p, and \X, which
319 refer to Unicode character properties, you must add
320
321 --enable-unicode-properties
322
323 to the configure command. This implies UTF-8 support, even if you have
324 not explicitly requested it.
325
326 Including Unicode property support adds around 30K of tables to the
327 PCRE library. Only the general category properties such as Lu and Nd
328 are supported. Details are given in the pcrepattern documentation.
329
330
331 CODE VALUE OF NEWLINE
332
333 By default, PCRE interprets character 10 (linefeed, LF) as indicating
334 the end of a line. This is the normal newline character on Unix-like
335 systems. You can compile PCRE to use character 13 (carriage return, CR)
336 instead, by adding
337
338 --enable-newline-is-cr
339
340 to the configure command. There is also a --enable-newline-is-lf
341 option, which explicitly specifies linefeed as the newline character.
342
343 Alternatively, you can specify that line endings are to be indicated by
344 the two character sequence CRLF. If you want this, add
345
346 --enable-newline-is-crlf
347
348 to the configure command. There is a fourth option, specified by
349
350 --enable-newline-is-anycrlf
351
352 which causes PCRE to recognize any of the three sequences CR, LF, or
353 CRLF as indicating a line ending. Finally, a fifth option, specified by
354
355 --enable-newline-is-any
356
357 causes PCRE to recognize any Unicode newline sequence.
358
359 Whatever line ending convention is selected when PCRE is built can be
360 overridden when the library functions are called. At build time it is
361 conventional to use the standard for your operating system.
362
363
364 WHAT \R MATCHES
365
366 By default, the sequence \R in a pattern matches any Unicode newline
367 sequence, whatever has been selected as the line ending sequence. If
368 you specify
369
370 --enable-bsr-anycrlf
371
372 the default is changed so that \R matches only CR, LF, or CRLF. What-
373 ever is selected when PCRE is built can be overridden when the library
374 functions are called.
375
376
377 BUILDING SHARED AND STATIC LIBRARIES
378
379 The PCRE building process uses libtool to build both shared and static
380 Unix libraries by default. You can suppress one of these by adding one
381 of
382
383 --disable-shared
384 --disable-static
385
386 to the configure command, as required.
387
388
389 POSIX MALLOC USAGE
390
391 When PCRE is called through the POSIX interface (see the pcreposix doc-
392 umentation), additional working storage is required for holding the
393 pointers to capturing substrings, because PCRE requires three integers
394 per substring, whereas the POSIX interface provides only two. If the
395 number of expected substrings is small, the wrapper function uses space
396 on the stack, because this is faster than using malloc() for each call.
397 The default threshold above which the stack is no longer used is 10; it
398 can be changed by adding a setting such as
399
400 --with-posix-malloc-threshold=20
401
402 to the configure command.
403
404
405 HANDLING VERY LARGE PATTERNS
406
407 Within a compiled pattern, offset values are used to point from one
408 part to another (for example, from an opening parenthesis to an alter-
409 nation metacharacter). By default, two-byte values are used for these
410 offsets, leading to a maximum size for a compiled pattern of around
411 64K. This is sufficient to handle all but the most gigantic patterns.
412 Nevertheless, some people do want to process enormous patterns, so it
413 is possible to compile PCRE to use three-byte or four-byte offsets by
414 adding a setting such as
415
416 --with-link-size=3
417
418 to the configure command. The value given must be 2, 3, or 4. Using
419 longer offsets slows down the operation of PCRE because it has to load
420 additional bytes when handling them.
421
422
423 AVOIDING EXCESSIVE STACK USAGE
424
425 When matching with the pcre_exec() function, PCRE implements backtrack-
426 ing by making recursive calls to an internal function called match().
427 In environments where the size of the stack is limited, this can se-
428 verely limit PCRE's operation. (The Unix environment does not usually
429 suffer from this problem, but it may sometimes be necessary to increase
430 the maximum stack size. There is a discussion in the pcrestack docu-
431 mentation.) An alternative approach to recursion that uses memory from
432 the heap to remember data, instead of using recursive function calls,
433 has been implemented to work round the problem of limited stack size.
434 If you want to build a version of PCRE that works this way, add
435
436 --disable-stack-for-recursion
437
438 to the configure command. With this configuration, PCRE will use the
439 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
440 ment functions. By default these point to malloc() and free(), but you
441 can replace the pointers so that your own functions are used.
442
443 Separate functions are provided rather than using pcre_malloc and
444 pcre_free because the usage is very predictable: the block sizes
445 requested are always the same, and the blocks are always freed in
446 reverse order. A calling program might be able to implement optimized
447 functions that perform better than malloc() and free(). PCRE runs
448 noticeably more slowly when built in this way. This option affects only
449 the pcre_exec() function; it is not relevant for the the
450 pcre_dfa_exec() function.
451
452
453 LIMITING PCRE RESOURCE USAGE
454
455 Internally, PCRE has a function called match(), which it calls repeat-
456 edly (sometimes recursively) when matching a pattern with the
457 pcre_exec() function. By controlling the maximum number of times this
458 function may be called during a single matching operation, a limit can
459 be placed on the resources used by a single call to pcre_exec(). The
460 limit can be changed at run time, as described in the pcreapi documen-
461 tation. The default is 10 million, but this can be changed by adding a
462 setting such as
463
464 --with-match-limit=500000
465
466 to the configure command. This setting has no effect on the
467 pcre_dfa_exec() matching function.
468
469 In some environments it is desirable to limit the depth of recursive
470 calls of match() more strictly than the total number of calls, in order
471 to restrict the maximum amount of stack (or heap, if --disable-stack-
472 for-recursion is specified) that is used. A second limit controls this;
473 it defaults to the value that is set for --with-match-limit, which
474 imposes no additional constraints. However, you can set a lower limit
475 by adding, for example,
476
477 --with-match-limit-recursion=10000
478
479 to the configure command. This value can also be overridden at run
480 time.
481
482
483 CREATING CHARACTER TABLES AT BUILD TIME
484
485 PCRE uses fixed tables for processing characters whose code values are
486 less than 256. By default, PCRE is built with a set of tables that are
487 distributed in the file pcre_chartables.c.dist. These tables are for
488 ASCII codes only. If you add
489
490 --enable-rebuild-chartables
491
492 to the configure command, the distributed tables are no longer used.
493 Instead, a program called dftables is compiled and run. This outputs
494 the source for new set of tables, created in the default locale of your
495 C runtime system. (This method of replacing the tables does not work if
496 you are cross compiling, because dftables is run on the local host. If
497 you need to create alternative tables when cross compiling, you will
498 have to do so "by hand".)
499
500
501 USING EBCDIC CODE
502
503 PCRE assumes by default that it will run in an environment where the
504 character code is ASCII (or Unicode, which is a superset of ASCII).
505 This is the case for most computer operating systems. PCRE can, how-
506 ever, be compiled to run in an EBCDIC environment by adding
507
508 --enable-ebcdic
509
510 to the configure command. This setting implies --enable-rebuild-charta-
511 bles. You should only use it if you know that you are in an EBCDIC
512 environment (for example, an IBM mainframe operating system).
513
514
515 SEE ALSO
516
517 pcreapi(3), pcre_config(3).
518
519
520 AUTHOR
521
522 Philip Hazel
523 University Computing Service
524 Cambridge CB2 3QH, England.
525
526
527 REVISION
528
529 Last updated: 11 September 2007
530 Copyright (c) 1997-2007 University of Cambridge.
531 ------------------------------------------------------------------------------
532
533
534 PCREMATCHING(3) PCREMATCHING(3)
535
536
537 NAME
538 PCRE - Perl-compatible regular expressions
539
540
541 PCRE MATCHING ALGORITHMS
542
543 This document describes the two different algorithms that are available
544 in PCRE for matching a compiled regular expression against a given sub-
545 ject string. The "standard" algorithm is the one provided by the
546 pcre_exec() function. This works in the same was as Perl's matching
547 function, and provides a Perl-compatible matching operation.
548
549 An alternative algorithm is provided by the pcre_dfa_exec() function;
550 this operates in a different way, and is not Perl-compatible. It has
551 advantages and disadvantages compared with the standard algorithm, and
552 these are described below.
553
554 When there is only one possible way in which a given subject string can
555 match a pattern, the two algorithms give the same answer. A difference
556 arises, however, when there are multiple possibilities. For example, if
557 the pattern
558
559 ^<.*>
560
561 is matched against the string
562
563 <something> <something else> <something further>
564
565 there are three possible answers. The standard algorithm finds only one
566 of them, whereas the alternative algorithm finds all three.
567
568
569 REGULAR EXPRESSIONS AS TREES
570
571 The set of strings that are matched by a regular expression can be rep-
572 resented as a tree structure. An unlimited repetition in the pattern
573 makes the tree of infinite size, but it is still a tree. Matching the
574 pattern to a given subject string (from a given starting point) can be
575 thought of as a search of the tree. There are two ways to search a
576 tree: depth-first and breadth-first, and these correspond to the two
577 matching algorithms provided by PCRE.
578
579
580 THE STANDARD MATCHING ALGORITHM
581
582 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
583 sions", the standard algorithm is an "NFA algorithm". It conducts a
584 depth-first search of the pattern tree. That is, it proceeds along a
585 single path through the tree, checking that the subject matches what is
586 required. When there is a mismatch, the algorithm tries any alterna-
587 tives at the current point, and if they all fail, it backs up to the
588 previous branch point in the tree, and tries the next alternative
589 branch at that level. This often involves backing up (moving to the
590 left) in the subject string as well. The order in which repetition
591 branches are tried is controlled by the greedy or ungreedy nature of
592 the quantifier.
593
594 If a leaf node is reached, a matching string has been found, and at
595 that point the algorithm stops. Thus, if there is more than one possi-
596 ble match, this algorithm returns the first one that it finds. Whether
597 this is the shortest, the longest, or some intermediate length depends
598 on the way the greedy and ungreedy repetition quantifiers are specified
599 in the pattern.
600
601 Because it ends up with a single path through the tree, it is rela-
602 tively straightforward for this algorithm to keep track of the sub-
603 strings that are matched by portions of the pattern in parentheses.
604 This provides support for capturing parentheses and back references.
605
606
607 THE ALTERNATIVE MATCHING ALGORITHM
608
609 This algorithm conducts a breadth-first search of the tree. Starting
610 from the first matching point in the subject, it scans the subject
611 string from left to right, once, character by character, and as it does
612 this, it remembers all the paths through the tree that represent valid
613 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
614 though it is not implemented as a traditional finite state machine (it
615 keeps multiple states active simultaneously).
616
617 The scan continues until either the end of the subject is reached, or
618 there are no more unterminated paths. At this point, terminated paths
619 represent the different matching possibilities (if there are none, the
620 match has failed). Thus, if there is more than one possible match,
621 this algorithm finds all of them, and in particular, it finds the long-
622 est. In PCRE, there is an option to stop the algorithm after the first
623 match (which is necessarily the shortest) has been found.
624
625 Note that all the matches that are found start at the same point in the
626 subject. If the pattern
627
628 cat(er(pillar)?)
629
630 is matched against the string "the caterpillar catchment", the result
631 will be the three strings "cat", "cater", and "caterpillar" that start
632 at the fourth character of the subject. The algorithm does not automat-
633 ically move on to find matches that start at later positions.
634
635 There are a number of features of PCRE regular expressions that are not
636 supported by the alternative matching algorithm. They are as follows:
637
638 1. Because the algorithm finds all possible matches, the greedy or
639 ungreedy nature of repetition quantifiers is not relevant. Greedy and
640 ungreedy quantifiers are treated in exactly the same way. However, pos-
641 sessive quantifiers can make a difference when what follows could also
642 match what is quantified, for example in a pattern like this:
643
644 ^a++\w!
645
646 This pattern matches "aaab!" but not "aaa!", which would be matched by
647 a non-possessive quantifier. Similarly, if an atomic group is present,
648 it is matched as if it were a standalone pattern at the current point,
649 and the longest match is then "locked in" for the rest of the overall
650 pattern.
651
652 2. When dealing with multiple paths through the tree simultaneously, it
653 is not straightforward to keep track of captured substrings for the
654 different matching possibilities, and PCRE's implementation of this
655 algorithm does not attempt to do this. This means that no captured sub-
656 strings are available.
657
658 3. Because no substrings are captured, back references within the pat-
659 tern are not supported, and cause errors if encountered.
660
661 4. For the same reason, conditional expressions that use a backrefer-
662 ence as the condition or test for a specific group recursion are not
663 supported.
664
665 5. Because many paths through the tree may be active, the \K escape
666 sequence, which resets the start of the match when encountered (but may
667 be on some paths and not on others), is not supported. It causes an
668 error if encountered.
669
670 6. Callouts are supported, but the value of the capture_top field is
671 always 1, and the value of the capture_last field is always -1.
672
673 7. The \C escape sequence, which (in the standard algorithm) matches a
674 single byte, even in UTF-8 mode, is not supported because the alterna-
675 tive algorithm moves through the subject string one character at a
676 time, for all active paths through the tree.
677
678 8. None of the backtracking control verbs such as (*PRUNE) are sup-
679 ported.
680
681
682 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
683
684 Using the alternative matching algorithm provides the following advan-
685 tages:
686
687 1. All possible matches (at a single point in the subject) are automat-
688 ically found, and in particular, the longest match is found. To find
689 more than one match using the standard algorithm, you have to do kludgy
690 things with callouts.
691
692 2. There is much better support for partial matching. The restrictions
693 on the content of the pattern that apply when using the standard algo-
694 rithm for partial matching do not apply to the alternative algorithm.
695 For non-anchored patterns, the starting position of a partial match is
696 available.
697
698 3. Because the alternative algorithm scans the subject string just
699 once, and never needs to backtrack, it is possible to pass very long
700 subject strings to the matching function in several pieces, checking
701 for partial matching each time.
702
703
704 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
705
706 The alternative algorithm suffers from a number of disadvantages:
707
708 1. It is substantially slower than the standard algorithm. This is
709 partly because it has to search for all possible matches, but is also
710 because it is less susceptible to optimization.
711
712 2. Capturing parentheses and back references are not supported.
713
714 3. Although atomic groups are supported, their use does not provide the
715 performance advantage that it does for the standard algorithm.
716
717
718 AUTHOR
719
720 Philip Hazel
721 University Computing Service
722 Cambridge CB2 3QH, England.
723
724
725 REVISION
726
727 Last updated: 08 August 2007
728 Copyright (c) 1997-2007 University of Cambridge.
729 ------------------------------------------------------------------------------
730
731
732 PCREAPI(3) PCREAPI(3)
733
734
735 NAME
736 PCRE - Perl-compatible regular expressions
737
738
739 PCRE NATIVE API
740
741 #include <pcre.h>
742
743 pcre *pcre_compile(const char *pattern, int options,
744 const char **errptr, int *erroffset,
745 const unsigned char *tableptr);
746
747 pcre *pcre_compile2(const char *pattern, int options,
748 int *errorcodeptr,
749 const char **errptr, int *erroffset,
750 const unsigned char *tableptr);
751
752 pcre_extra *pcre_study(const pcre *code, int options,
753 const char **errptr);
754
755 int pcre_exec(const pcre *code, const pcre_extra *extra,
756 const char *subject, int length, int startoffset,
757 int options, int *ovector, int ovecsize);
758
759 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
760 const char *subject, int length, int startoffset,
761 int options, int *ovector, int ovecsize,
762 int *workspace, int wscount);
763
764 int pcre_copy_named_substring(const pcre *code,
765 const char *subject, int *ovector,
766 int stringcount, const char *stringname,
767 char *buffer, int buffersize);
768
769 int pcre_copy_substring(const char *subject, int *ovector,
770 int stringcount, int stringnumber, char *buffer,
771 int buffersize);
772
773 int pcre_get_named_substring(const pcre *code,
774 const char *subject, int *ovector,
775 int stringcount, const char *stringname,
776 const char **stringptr);
777
778 int pcre_get_stringnumber(const pcre *code,
779 const char *name);
780
781 int pcre_get_stringtable_entries(const pcre *code,
782 const char *name, char **first, char **last);
783
784 int pcre_get_substring(const char *subject, int *ovector,
785 int stringcount, int stringnumber,
786 const char **stringptr);
787
788 int pcre_get_substring_list(const char *subject,
789 int *ovector, int stringcount, const char ***listptr);
790
791 void pcre_free_substring(const char *stringptr);
792
793 void pcre_free_substring_list(const char **stringptr);
794
795 const unsigned char *pcre_maketables(void);
796
797 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
798 int what, void *where);
799
800 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
801
802 int pcre_refcount(pcre *code, int adjust);
803
804 int pcre_config(int what, void *where);
805
806 char *pcre_version(void);
807
808 void *(*pcre_malloc)(size_t);
809
810 void (*pcre_free)(void *);
811
812 void *(*pcre_stack_malloc)(size_t);
813
814 void (*pcre_stack_free)(void *);
815
816 int (*pcre_callout)(pcre_callout_block *);
817
818
819 PCRE API OVERVIEW
820
821 PCRE has its own native API, which is described in this document. There
822 are also some wrapper functions that correspond to the POSIX regular
823 expression API. These are described in the pcreposix documentation.
824 Both of these APIs define a set of C function calls. A C++ wrapper is
825 distributed with PCRE. It is documented in the pcrecpp page.
826
827 The native API C function prototypes are defined in the header file
828 pcre.h, and on Unix systems the library itself is called libpcre. It
829 can normally be accessed by adding -lpcre to the command for linking an
830 application that uses PCRE. The header file defines the macros
831 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
832 bers for the library. Applications can use these to include support
833 for different releases of PCRE.
834
835 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
836 pcre_exec() are used for compiling and matching regular expressions in
837 a Perl-compatible manner. A sample program that demonstrates the sim-
838 plest way of using them is provided in the file called pcredemo.c in
839 the source distribution. The pcresample documentation describes how to
840 run it.
841
842 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
843 ble, is also provided. This uses a different algorithm for the match-
844 ing. The alternative algorithm finds all possible matches (at a given
845 point in the subject), and scans the subject just once. However, this
846 algorithm does not return captured substrings. A description of the two
847 matching algorithms and their advantages and disadvantages is given in
848 the pcrematching documentation.
849
850 In addition to the main compiling and matching functions, there are
851 convenience functions for extracting captured substrings from a subject
852 string that is matched by pcre_exec(). They are:
853
854 pcre_copy_substring()
855 pcre_copy_named_substring()
856 pcre_get_substring()
857 pcre_get_named_substring()
858 pcre_get_substring_list()
859 pcre_get_stringnumber()
860 pcre_get_stringtable_entries()
861
862 pcre_free_substring() and pcre_free_substring_list() are also provided,
863 to free the memory used for extracted strings.
864
865 The function pcre_maketables() is used to build a set of character
866 tables in the current locale for passing to pcre_compile(),
867 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
868 provided for specialist use. Most commonly, no special tables are
869 passed, in which case internal tables that are generated when PCRE is
870 built are used.
871
872 The function pcre_fullinfo() is used to find out information about a
873 compiled pattern; pcre_info() is an obsolete version that returns only
874 some of the available information, but is retained for backwards com-
875 patibility. The function pcre_version() returns a pointer to a string
876 containing the version of PCRE and its date of release.
877
878 The function pcre_refcount() maintains a reference count in a data
879 block containing a compiled pattern. This is provided for the benefit
880 of object-oriented applications.
881
882 The global variables pcre_malloc and pcre_free initially contain the
883 entry points of the standard malloc() and free() functions, respec-
884 tively. PCRE calls the memory management functions via these variables,
885 so a calling program can replace them if it wishes to intercept the
886 calls. This should be done before calling any PCRE functions.
887
888 The global variables pcre_stack_malloc and pcre_stack_free are also
889 indirections to memory management functions. These special functions
890 are used only when PCRE is compiled to use the heap for remembering
891 data, instead of recursive function calls, when running the pcre_exec()
892 function. See the pcrebuild documentation for details of how to do
893 this. It is a non-standard way of building PCRE, for use in environ-
894 ments that have limited stacks. Because of the greater use of memory
895 management, it runs more slowly. Separate functions are provided so
896 that special-purpose external code can be used for this case. When
897 used, these functions are always called in a stack-like manner (last
898 obtained, first freed), and always for memory blocks of the same size.
899 There is a discussion about PCRE's stack usage in the pcrestack docu-
900 mentation.
901
902 The global variable pcre_callout initially contains NULL. It can be set
903 by the caller to a "callout" function, which PCRE will then call at
904 specified points during a matching operation. Details are given in the
905 pcrecallout documentation.
906
907
908 NEWLINES
909
910 PCRE supports five different conventions for indicating line breaks in
911 strings: a single CR (carriage return) character, a single LF (line-
912 feed) character, the two-character sequence CRLF, any of the three pre-
913 ceding, or any Unicode newline sequence. The Unicode newline sequences
914 are the three just mentioned, plus the single characters VT (vertical
915 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
916 separator, U+2028), and PS (paragraph separator, U+2029).
917
918 Each of the first three conventions is used by at least one operating
919 system as its standard newline sequence. When PCRE is built, a default
920 can be specified. The default default is LF, which is the Unix stan-
921 dard. When PCRE is run, the default can be overridden, either when a
922 pattern is compiled, or when it is matched.
923
924 At compile time, the newline convention can be specified by the options
925 argument of pcre_compile(), or it can be specified by special text at
926 the start of the pattern itself; this overrides any other settings. See
927 the pcrepattern page for details of the special character sequences.
928
929 In the PCRE documentation the word "newline" is used to mean "the char-
930 acter or pair of characters that indicate a line break". The choice of
931 newline convention affects the handling of the dot, circumflex, and
932 dollar metacharacters, the handling of #-comments in /x mode, and, when
933 CRLF is a recognized line ending sequence, the match position advance-
934 ment for a non-anchored pattern. There is more detail about this in the
935 section on pcre_exec() options below.
936
937 The choice of newline convention does not affect the interpretation of
938 the \n or \r escape sequences, nor does it affect what \R matches,
939 which is controlled in a similar way, but by separate options.
940
941
942 MULTITHREADING
943
944 The PCRE functions can be used in multi-threading applications, with
945 the proviso that the memory management functions pointed to by
946 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
947 callout function pointed to by pcre_callout, are shared by all threads.
948
949 The compiled form of a regular expression is not altered during match-
950 ing, so the same compiled pattern can safely be used by several threads
951 at once.
952
953
954 SAVING PRECOMPILED PATTERNS FOR LATER USE
955
956 The compiled form of a regular expression can be saved and re-used at a
957 later time, possibly by a different program, and even on a host other
958 than the one on which it was compiled. Details are given in the
959 pcreprecompile documentation. However, compiling a regular expression
960 with one version of PCRE for use with a different version is not guar-
961 anteed to work and may cause crashes.
962
963
964 CHECKING BUILD-TIME OPTIONS
965
966 int pcre_config(int what, void *where);
967
968 The function pcre_config() makes it possible for a PCRE client to dis-
969 cover which optional features have been compiled into the PCRE library.
970 The pcrebuild documentation has more details about these optional fea-
971 tures.
972
973 The first argument for pcre_config() is an integer, specifying which
974 information is required; the second argument is a pointer to a variable
975 into which the information is placed. The following information is
976 available:
977
978 PCRE_CONFIG_UTF8
979
980 The output is an integer that is set to one if UTF-8 support is avail-
981 able; otherwise it is set to zero.
982
983 PCRE_CONFIG_UNICODE_PROPERTIES
984
985 The output is an integer that is set to one if support for Unicode
986 character properties is available; otherwise it is set to zero.
987
988 PCRE_CONFIG_NEWLINE
989
990 The output is an integer whose value specifies the default character
991 sequence that is recognized as meaning "newline". The four values that
992 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
993 and -1 for ANY. The default should normally be the standard sequence
994 for your operating system.
995
996 PCRE_CONFIG_BSR
997
998 The output is an integer whose value indicates what character sequences
999 the \R escape sequence matches by default. A value of 0 means that \R
1000 matches any Unicode line ending sequence; a value of 1 means that \R
1001 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1002 tern is compiled or matched.
1003
1004 PCRE_CONFIG_LINK_SIZE
1005
1006 The output is an integer that contains the number of bytes used for
1007 internal linkage in compiled regular expressions. The value is 2, 3, or
1008 4. Larger values allow larger regular expressions to be compiled, at
1009 the expense of slower matching. The default value of 2 is sufficient
1010 for all but the most massive patterns, since it allows the compiled
1011 pattern to be up to 64K in size.
1012
1013 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1014
1015 The output is an integer that contains the threshold above which the
1016 POSIX interface uses malloc() for output vectors. Further details are
1017 given in the pcreposix documentation.
1018
1019 PCRE_CONFIG_MATCH_LIMIT
1020
1021 The output is an integer that gives the default limit for the number of
1022 internal matching function calls in a pcre_exec() execution. Further
1023 details are given with pcre_exec() below.
1024
1025 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1026
1027 The output is an integer that gives the default limit for the depth of
1028 recursion when calling the internal matching function in a pcre_exec()
1029 execution. Further details are given with pcre_exec() below.
1030
1031 PCRE_CONFIG_STACKRECURSE
1032
1033 The output is an integer that is set to one if internal recursion when
1034 running pcre_exec() is implemented by recursive function calls that use
1035 the stack to remember their state. This is the usual way that PCRE is
1036 compiled. The output is zero if PCRE was compiled to use blocks of data
1037 on the heap instead of recursive function calls. In this case,
1038 pcre_stack_malloc and pcre_stack_free are called to manage memory
1039 blocks on the heap, thus avoiding the use of the stack.
1040
1041
1042 COMPILING A PATTERN
1043
1044 pcre *pcre_compile(const char *pattern, int options,
1045 const char **errptr, int *erroffset,
1046 const unsigned char *tableptr);
1047
1048 pcre *pcre_compile2(const char *pattern, int options,
1049 int *errorcodeptr,
1050 const char **errptr, int *erroffset,
1051 const unsigned char *tableptr);
1052
1053 Either of the functions pcre_compile() or pcre_compile2() can be called
1054 to compile a pattern into an internal form. The only difference between
1055 the two interfaces is that pcre_compile2() has an additional argument,
1056 errorcodeptr, via which a numerical error code can be returned.
1057
1058 The pattern is a C string terminated by a binary zero, and is passed in
1059 the pattern argument. A pointer to a single block of memory that is
1060 obtained via pcre_malloc is returned. This contains the compiled code
1061 and related data. The pcre type is defined for the returned block; this
1062 is a typedef for a structure whose contents are not externally defined.
1063 It is up to the caller to free the memory (via pcre_free) when it is no
1064 longer required.
1065
1066 Although the compiled code of a PCRE regex is relocatable, that is, it
1067 does not depend on memory location, the complete pcre data block is not
1068 fully relocatable, because it may contain a copy of the tableptr argu-
1069 ment, which is an address (see below).
1070
1071 The options argument contains various bit settings that affect the com-
1072 pilation. It should be zero if no options are required. The available
1073 options are described below. Some of them, in particular, those that
1074 are compatible with Perl, can also be set and unset from within the
1075 pattern (see the detailed description in the pcrepattern documenta-
1076 tion). For these options, the contents of the options argument speci-
1077 fies their initial settings at the start of compilation and execution.
1078 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1079 of matching as well as at compile time.
1080
1081 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1082 if compilation of a pattern fails, pcre_compile() returns NULL, and
1083 sets the variable pointed to by errptr to point to a textual error mes-
1084 sage. This is a static string that is part of the library. You must not
1085 try to free it. The offset from the start of the pattern to the charac-
1086 ter where the error was discovered is placed in the variable pointed to
1087 by erroffset, which must not be NULL. If it is, an immediate error is
1088 given.
1089
1090 If pcre_compile2() is used instead of pcre_compile(), and the error-
1091 codeptr argument is not NULL, a non-zero error code number is returned
1092 via this argument in the event of an error. This is in addition to the
1093 textual error message. Error codes and messages are listed below.
1094
1095 If the final argument, tableptr, is NULL, PCRE uses a default set of
1096 character tables that are built when PCRE is compiled, using the
1097 default C locale. Otherwise, tableptr must be an address that is the
1098 result of a call to pcre_maketables(). This value is stored with the
1099 compiled pattern, and used again by pcre_exec(), unless another table
1100 pointer is passed to it. For more discussion, see the section on locale
1101 support below.
1102
1103 This code fragment shows a typical straightforward call to pcre_com-
1104 pile():
1105
1106 pcre *re;
1107 const char *error;
1108 int erroffset;
1109 re = pcre_compile(
1110 "^A.*Z", /* the pattern */
1111 0, /* default options */
1112 &error, /* for error message */
1113 &erroffset, /* for error offset */
1114 NULL); /* use default character tables */
1115
1116 The following names for option bits are defined in the pcre.h header
1117 file:
1118
1119 PCRE_ANCHORED
1120
1121 If this bit is set, the pattern is forced to be "anchored", that is, it
1122 is constrained to match only at the first matching point in the string
1123 that is being searched (the "subject string"). This effect can also be
1124 achieved by appropriate constructs in the pattern itself, which is the
1125 only way to do it in Perl.
1126
1127 PCRE_AUTO_CALLOUT
1128
1129 If this bit is set, pcre_compile() automatically inserts callout items,
1130 all with number 255, before each pattern item. For discussion of the
1131 callout facility, see the pcrecallout documentation.
1132
1133 PCRE_BSR_ANYCRLF
1134 PCRE_BSR_UNICODE
1135
1136 These options (which are mutually exclusive) control what the \R escape
1137 sequence matches. The choice is either to match only CR, LF, or CRLF,
1138 or to match any Unicode newline sequence. The default is specified when
1139 PCRE is built. It can be overridden from within the pattern, or by set-
1140 ting an option when a compiled pattern is matched.
1141
1142 PCRE_CASELESS
1143
1144 If this bit is set, letters in the pattern match both upper and lower
1145 case letters. It is equivalent to Perl's /i option, and it can be
1146 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1147 always understands the concept of case for characters whose values are
1148 less than 128, so caseless matching is always possible. For characters
1149 with higher values, the concept of case is supported if PCRE is com-
1150 piled with Unicode property support, but not otherwise. If you want to
1151 use caseless matching for characters 128 and above, you must ensure
1152 that PCRE is compiled with Unicode property support as well as with
1153 UTF-8 support.
1154
1155 PCRE_DOLLAR_ENDONLY
1156
1157 If this bit is set, a dollar metacharacter in the pattern matches only
1158 at the end of the subject string. Without this option, a dollar also
1159 matches immediately before a newline at the end of the string (but not
1160 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1161 if PCRE_MULTILINE is set. There is no equivalent to this option in
1162 Perl, and no way to set it within a pattern.
1163
1164 PCRE_DOTALL
1165
1166 If this bit is set, a dot metacharater in the pattern matches all char-
1167 acters, including those that indicate newline. Without it, a dot does
1168 not match when the current position is at a newline. This option is
1169 equivalent to Perl's /s option, and it can be changed within a pattern
1170 by a (?s) option setting. A negative class such as [^a] always matches
1171 newline characters, independent of the setting of this option.
1172
1173 PCRE_DUPNAMES
1174
1175 If this bit is set, names used to identify capturing subpatterns need
1176 not be unique. This can be helpful for certain types of pattern when it
1177 is known that only one instance of the named subpattern can ever be
1178 matched. There are more details of named subpatterns below; see also
1179 the pcrepattern documentation.
1180
1181 PCRE_EXTENDED
1182
1183 If this bit is set, whitespace data characters in the pattern are
1184 totally ignored except when escaped or inside a character class. White-
1185 space does not include the VT character (code 11). In addition, charac-
1186 ters between an unescaped # outside a character class and the next new-
1187 line, inclusive, are also ignored. This is equivalent to Perl's /x
1188 option, and it can be changed within a pattern by a (?x) option set-
1189 ting.
1190
1191 This option makes it possible to include comments inside complicated
1192 patterns. Note, however, that this applies only to data characters.
1193 Whitespace characters may never appear within special character
1194 sequences in a pattern, for example within the sequence (?( which
1195 introduces a conditional subpattern.
1196
1197 PCRE_EXTRA
1198
1199 This option was invented in order to turn on additional functionality
1200 of PCRE that is incompatible with Perl, but it is currently of very
1201 little use. When set, any backslash in a pattern that is followed by a
1202 letter that has no special meaning causes an error, thus reserving
1203 these combinations for future expansion. By default, as in Perl, a
1204 backslash followed by a letter with no special meaning is treated as a
1205 literal. (Perl can, however, be persuaded to give a warning for this.)
1206 There are at present no other features controlled by this option. It
1207 can also be set by a (?X) option setting within a pattern.
1208
1209 PCRE_FIRSTLINE
1210
1211 If this option is set, an unanchored pattern is required to match
1212 before or at the first newline in the subject string, though the
1213 matched text may continue over the newline.
1214
1215 PCRE_MULTILINE
1216
1217 By default, PCRE treats the subject string as consisting of a single
1218 line of characters (even if it actually contains newlines). The "start
1219 of line" metacharacter (^) matches only at the start of the string,
1220 while the "end of line" metacharacter ($) matches only at the end of
1221 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1222 is set). This is the same as Perl.
1223
1224 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1225 constructs match immediately following or immediately before internal
1226 newlines in the subject string, respectively, as well as at the very
1227 start and end. This is equivalent to Perl's /m option, and it can be
1228 changed within a pattern by a (?m) option setting. If there are no new-
1229 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1230 setting PCRE_MULTILINE has no effect.
1231
1232 PCRE_NEWLINE_CR
1233 PCRE_NEWLINE_LF
1234 PCRE_NEWLINE_CRLF
1235 PCRE_NEWLINE_ANYCRLF
1236 PCRE_NEWLINE_ANY
1237
1238 These options override the default newline definition that was chosen
1239 when PCRE was built. Setting the first or the second specifies that a
1240 newline is indicated by a single character (CR or LF, respectively).
1241 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1242 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1243 that any of the three preceding sequences should be recognized. Setting
1244 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1245 recognized. The Unicode newline sequences are the three just mentioned,
1246 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1247 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1248 (paragraph separator, U+2029). The last two are recognized only in
1249 UTF-8 mode.
1250
1251 The newline setting in the options word uses three bits that are
1252 treated as a number, giving eight possibilities. Currently only six are
1253 used (default plus the five values above). This means that if you set
1254 more than one newline option, the combination may or may not be sensi-
1255 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1256 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1257 cause an error.
1258
1259 The only time that a line break is specially recognized when compiling
1260 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1261 character class is encountered. This indicates a comment that lasts
1262 until after the next line break sequence. In other circumstances, line
1263 break sequences are treated as literal data, except that in
1264 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1265 and are therefore ignored.
1266
1267 The newline option that is set at compile time becomes the default that
1268 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1269
1270 PCRE_NO_AUTO_CAPTURE
1271
1272 If this option is set, it disables the use of numbered capturing paren-
1273 theses in the pattern. Any opening parenthesis that is not followed by
1274 ? behaves as if it were followed by ?: but named parentheses can still
1275 be used for capturing (and they acquire numbers in the usual way).
1276 There is no equivalent of this option in Perl.
1277
1278 PCRE_UNGREEDY
1279
1280 This option inverts the "greediness" of the quantifiers so that they
1281 are not greedy by default, but become greedy if followed by "?". It is
1282 not compatible with Perl. It can also be set by a (?U) option setting
1283 within the pattern.
1284
1285 PCRE_UTF8
1286
1287 This option causes PCRE to regard both the pattern and the subject as
1288 strings of UTF-8 characters instead of single-byte character strings.
1289 However, it is available only when PCRE is built to include UTF-8 sup-
1290 port. If not, the use of this option provokes an error. Details of how
1291 this option changes the behaviour of PCRE are given in the section on
1292 UTF-8 support in the main pcre page.
1293
1294 PCRE_NO_UTF8_CHECK
1295
1296 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1297 automatically checked. There is a discussion about the validity of
1298 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1299 bytes is found, pcre_compile() returns an error. If you already know
1300 that your pattern is valid, and you want to skip this check for perfor-
1301 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1302 set, the effect of passing an invalid UTF-8 string as a pattern is
1303 undefined. It may cause your program to crash. Note that this option
1304 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1305 UTF-8 validity checking of subject strings.
1306
1307
1308 COMPILATION ERROR CODES
1309
1310 The following table lists the error codes than may be returned by
1311 pcre_compile2(), along with the error messages that may be returned by
1312 both compiling functions. As PCRE has developed, some error codes have
1313 fallen out of use. To avoid confusion, they have not been re-used.
1314
1315 0 no error
1316 1 \ at end of pattern
1317 2 \c at end of pattern
1318 3 unrecognized character follows \
1319 4 numbers out of order in {} quantifier
1320 5 number too big in {} quantifier
1321 6 missing terminating ] for character class
1322 7 invalid escape sequence in character class
1323 8 range out of order in character class
1324 9 nothing to repeat
1325 10 [this code is not in use]
1326 11 internal error: unexpected repeat
1327 12 unrecognized character after (?
1328 13 POSIX named classes are supported only within a class
1329 14 missing )
1330 15 reference to non-existent subpattern
1331 16 erroffset passed as NULL
1332 17 unknown option bit(s) set
1333 18 missing ) after comment
1334 19 [this code is not in use]
1335 20 regular expression too large
1336 21 failed to get memory
1337 22 unmatched parentheses
1338 23 internal error: code overflow
1339 24 unrecognized character after (?<
1340 25 lookbehind assertion is not fixed length
1341 26 malformed number or name after (?(
1342 27 conditional group contains more than two branches
1343 28 assertion expected after (?(
1344 29 (?R or (?[+-]digits must be followed by )
1345 30 unknown POSIX class name
1346 31 POSIX collating elements are not supported
1347 32 this version of PCRE is not compiled with PCRE_UTF8 support
1348 33 [this code is not in use]
1349 34 character value in \x{...} sequence is too large
1350 35 invalid condition (?(0)
1351 36 \C not allowed in lookbehind assertion
1352 37 PCRE does not support \L, \l, \N, \U, or \u
1353 38 number after (?C is > 255
1354 39 closing ) for (?C expected
1355 40 recursive call could loop indefinitely
1356 41 unrecognized character after (?P
1357 42 syntax error in subpattern name (missing terminator)
1358 43 two named subpatterns have the same name
1359 44 invalid UTF-8 string
1360 45 support for \P, \p, and \X has not been compiled
1361 46 malformed \P or \p sequence
1362 47 unknown property name after \P or \p
1363 48 subpattern name is too long (maximum 32 characters)
1364 49 too many named subpatterns (maximum 10,000)
1365 50 [this code is not in use]
1366 51 octal value is greater than \377 (not in UTF-8 mode)
1367 52 internal error: overran compiling workspace
1368 53 internal error: previously-checked referenced subpattern not
1369 found
1370 54 DEFINE group contains more than one branch
1371 55 repeating a DEFINE group is not allowed
1372 56 inconsistent NEWLINE options
1373 57 \g is not followed by a braced name or an optionally braced
1374 non-zero number
1375 58 (?+ or (?- or (?(+ or (?(- must be followed by a non-zero number
1376
1377
1378 STUDYING A PATTERN
1379
1380 pcre_extra *pcre_study(const pcre *code, int options
1381 const char **errptr);
1382
1383 If a compiled pattern is going to be used several times, it is worth
1384 spending more time analyzing it in order to speed up the time taken for
1385 matching. The function pcre_study() takes a pointer to a compiled pat-
1386 tern as its first argument. If studying the pattern produces additional
1387 information that will help speed up matching, pcre_study() returns a
1388 pointer to a pcre_extra block, in which the study_data field points to
1389 the results of the study.
1390
1391 The returned value from pcre_study() can be passed directly to
1392 pcre_exec(). However, a pcre_extra block also contains other fields
1393 that can be set by the caller before the block is passed; these are
1394 described below in the section on matching a pattern.
1395
1396 If studying the pattern does not produce any additional information
1397 pcre_study() returns NULL. In that circumstance, if the calling program
1398 wants to pass any of the other fields to pcre_exec(), it must set up
1399 its own pcre_extra block.
1400
1401 The second argument of pcre_study() contains option bits. At present,
1402 no options are defined, and this argument should always be zero.
1403
1404 The third argument for pcre_study() is a pointer for an error message.
1405 If studying succeeds (even if no data is returned), the variable it
1406 points to is set to NULL. Otherwise it is set to point to a textual
1407 error message. This is a static string that is part of the library. You
1408 must not try to free it. You should test the error pointer for NULL
1409 after calling pcre_study(), to be sure that it has run successfully.
1410
1411 This is a typical call to pcre_study():
1412
1413 pcre_extra *pe;
1414 pe = pcre_study(
1415 re, /* result of pcre_compile() */
1416 0, /* no options exist */
1417 &error); /* set to NULL or points to a message */
1418
1419 At present, studying a pattern is useful only for non-anchored patterns
1420 that do not have a single fixed starting character. A bitmap of possi-
1421 ble starting bytes is created.
1422
1423
1424 LOCALE SUPPORT
1425
1426 PCRE handles caseless matching, and determines whether characters are
1427 letters, digits, or whatever, by reference to a set of tables, indexed
1428 by character value. When running in UTF-8 mode, this applies only to
1429 characters with codes less than 128. Higher-valued codes never match
1430 escapes such as \w or \d, but can be tested with \p if PCRE is built
1431 with Unicode character property support. The use of locales with Uni-
1432 code is discouraged. If you are handling characters with codes greater
1433 than 128, you should either use UTF-8 and Unicode, or use locales, but
1434 not try to mix the two.
1435
1436 PCRE contains an internal set of tables that are used when the final
1437 argument of pcre_compile() is NULL. These are sufficient for many
1438 applications. Normally, the internal tables recognize only ASCII char-
1439 acters. However, when PCRE is built, it is possible to cause the inter-
1440 nal tables to be rebuilt in the default "C" locale of the local system,
1441 which may cause them to be different.
1442
1443 The internal tables can always be overridden by tables supplied by the
1444 application that calls PCRE. These may be created in a different locale
1445 from the default. As more and more applications change to using Uni-
1446 code, the need for this locale support is expected to die away.
1447
1448 External tables are built by calling the pcre_maketables() function,
1449 which has no arguments, in the relevant locale. The result can then be
1450 passed to pcre_compile() or pcre_exec() as often as necessary. For
1451 example, to build and use tables that are appropriate for the French
1452 locale (where accented characters with values greater than 128 are
1453 treated as letters), the following code could be used:
1454
1455 setlocale(LC_CTYPE, "fr_FR");
1456 tables = pcre_maketables();
1457 re = pcre_compile(..., tables);
1458
1459 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1460 if you are using Windows, the name for the French locale is "french".
1461
1462 When pcre_maketables() runs, the tables are built in memory that is
1463 obtained via pcre_malloc. It is the caller's responsibility to ensure
1464 that the memory containing the tables remains available for as long as
1465 it is needed.
1466
1467 The pointer that is passed to pcre_compile() is saved with the compiled
1468 pattern, and the same tables are used via this pointer by pcre_study()
1469 and normally also by pcre_exec(). Thus, by default, for any single pat-
1470 tern, compilation, studying and matching all happen in the same locale,
1471 but different patterns can be compiled in different locales.
1472
1473 It is possible to pass a table pointer or NULL (indicating the use of
1474 the internal tables) to pcre_exec(). Although not intended for this
1475 purpose, this facility could be used to match a pattern in a different
1476 locale from the one in which it was compiled. Passing table pointers at
1477 run time is discussed below in the section on matching a pattern.
1478
1479
1480 INFORMATION ABOUT A PATTERN
1481
1482 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1483 int what, void *where);
1484
1485 The pcre_fullinfo() function returns information about a compiled pat-
1486 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1487 less retained for backwards compability (and is documented below).
1488
1489 The first argument for pcre_fullinfo() is a pointer to the compiled
1490 pattern. The second argument is the result of pcre_study(), or NULL if
1491 the pattern was not studied. The third argument specifies which piece
1492 of information is required, and the fourth argument is a pointer to a
1493 variable to receive the data. The yield of the function is zero for
1494 success, or one of the following negative numbers:
1495
1496 PCRE_ERROR_NULL the argument code was NULL
1497 the argument where was NULL
1498 PCRE_ERROR_BADMAGIC the "magic number" was not found
1499 PCRE_ERROR_BADOPTION the value of what was invalid
1500
1501 The "magic number" is placed at the start of each compiled pattern as
1502 an simple check against passing an arbitrary memory pointer. Here is a
1503 typical call of pcre_fullinfo(), to obtain the length of the compiled
1504 pattern:
1505
1506 int rc;
1507 size_t length;
1508 rc = pcre_fullinfo(
1509 re, /* result of pcre_compile() */
1510 pe, /* result of pcre_study(), or NULL */
1511 PCRE_INFO_SIZE, /* what is required */
1512 &length); /* where to put the data */
1513
1514 The possible values for the third argument are defined in pcre.h, and
1515 are as follows:
1516
1517 PCRE_INFO_BACKREFMAX
1518
1519 Return the number of the highest back reference in the pattern. The
1520 fourth argument should point to an int variable. Zero is returned if
1521 there are no back references.
1522
1523 PCRE_INFO_CAPTURECOUNT
1524
1525 Return the number of capturing subpatterns in the pattern. The fourth
1526 argument should point to an int variable.
1527
1528 PCRE_INFO_DEFAULT_TABLES
1529
1530 Return a pointer to the internal default character tables within PCRE.
1531 The fourth argument should point to an unsigned char * variable. This
1532 information call is provided for internal use by the pcre_study() func-
1533 tion. External callers can cause PCRE to use its internal tables by
1534 passing a NULL table pointer.
1535
1536 PCRE_INFO_FIRSTBYTE
1537
1538 Return information about the first byte of any matched string, for a
1539 non-anchored pattern. The fourth argument should point to an int vari-
1540 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1541 is still recognized for backwards compatibility.)
1542
1543 If there is a fixed first byte, for example, from a pattern such as
1544 (cat|cow|coyote), its value is returned. Otherwise, if either
1545
1546 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1547 branch starts with "^", or
1548
1549 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1550 set (if it were set, the pattern would be anchored),
1551
1552 -1 is returned, indicating that the pattern matches only at the start
1553 of a subject string or after any newline within the string. Otherwise
1554 -2 is returned. For anchored patterns, -2 is returned.
1555
1556 PCRE_INFO_FIRSTTABLE
1557
1558 If the pattern was studied, and this resulted in the construction of a
1559 256-bit table indicating a fixed set of bytes for the first byte in any
1560 matching string, a pointer to the table is returned. Otherwise NULL is
1561 returned. The fourth argument should point to an unsigned char * vari-
1562 able.
1563
1564 PCRE_INFO_HASCRORLF
1565
1566 Return 1 if the pattern contains any explicit matches for CR or LF
1567 characters, otherwise 0. The fourth argument should point to an int
1568 variable. An explicit match is either a literal CR or LF character, or
1569 \r or \n.
1570
1571 PCRE_INFO_JCHANGED
1572
1573 Return 1 if the (?J) option setting is used in the pattern, otherwise
1574 0. The fourth argument should point to an int variable. The (?J) inter-
1575 nal option setting changes the local PCRE_DUPNAMES option.
1576
1577 PCRE_INFO_LASTLITERAL
1578
1579 Return the value of the rightmost literal byte that must exist in any
1580 matched string, other than at its start, if such a byte has been
1581 recorded. The fourth argument should point to an int variable. If there
1582 is no such byte, -1 is returned. For anchored patterns, a last literal
1583 byte is recorded only if it follows something of variable length. For
1584 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1585 /^a\dz\d/ the returned value is -1.
1586
1587 PCRE_INFO_NAMECOUNT
1588 PCRE_INFO_NAMEENTRYSIZE
1589 PCRE_INFO_NAMETABLE
1590
1591 PCRE supports the use of named as well as numbered capturing parenthe-
1592 ses. The names are just an additional way of identifying the parenthe-
1593 ses, which still acquire numbers. Several convenience functions such as
1594 pcre_get_named_substring() are provided for extracting captured sub-
1595 strings by name. It is also possible to extract the data directly, by
1596 first converting the name to a number in order to access the correct
1597 pointers in the output vector (described with pcre_exec() below). To do
1598 the conversion, you need to use the name-to-number map, which is
1599 described by these three values.
1600
1601 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1602 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1603 of each entry; both of these return an int value. The entry size
1604 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1605 a pointer to the first entry of the table (a pointer to char). The
1606 first two bytes of each entry are the number of the capturing parenthe-
1607 sis, most significant byte first. The rest of the entry is the corre-
1608 sponding name, zero terminated. The names are in alphabetical order.
1609 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1610 theses numbers. For example, consider the following pattern (assume
1611 PCRE_EXTENDED is set, so white space - including newlines - is
1612 ignored):
1613
1614 (?<date> (?<year>(\d\d)?\d\d) -
1615 (?<month>\d\d) - (?<day>\d\d) )
1616
1617 There are four named subpatterns, so the table has four entries, and
1618 each entry in the table is eight bytes long. The table is as follows,
1619 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1620 as ??:
1621
1622 00 01 d a t e 00 ??
1623 00 05 d a y 00 ?? ??
1624 00 04 m o n t h 00
1625 00 02 y e a r 00 ??
1626
1627 When writing code to extract data from named subpatterns using the
1628 name-to-number map, remember that the length of the entries is likely
1629 to be different for each compiled pattern.
1630
1631 PCRE_INFO_OKPARTIAL
1632
1633 Return 1 if the pattern can be used for partial matching, otherwise 0.
1634 The fourth argument should point to an int variable. The pcrepartial
1635 documentation lists the restrictions that apply to patterns when par-
1636 tial matching is used.
1637
1638 PCRE_INFO_OPTIONS
1639
1640 Return a copy of the options with which the pattern was compiled. The
1641 fourth argument should point to an unsigned long int variable. These
1642 option bits are those specified in the call to pcre_compile(), modified
1643 by any top-level option settings at the start of the pattern itself. In
1644 other words, they are the options that will be in force when matching
1645 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1646 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1647 and PCRE_EXTENDED.
1648
1649 A pattern is automatically anchored by PCRE if all of its top-level
1650 alternatives begin with one of the following:
1651
1652 ^ unless PCRE_MULTILINE is set
1653 \A always
1654 \G always
1655 .* if PCRE_DOTALL is set and there are no back
1656 references to the subpattern in which .* appears
1657
1658 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1659 by pcre_fullinfo().
1660
1661 PCRE_INFO_SIZE
1662
1663 Return the size of the compiled pattern, that is, the value that was
1664 passed as the argument to pcre_malloc() when PCRE was getting memory in
1665 which to place the compiled data. The fourth argument should point to a
1666 size_t variable.
1667
1668 PCRE_INFO_STUDYSIZE
1669
1670 Return the size of the data block pointed to by the study_data field in
1671 a pcre_extra block. That is, it is the value that was passed to
1672 pcre_malloc() when PCRE was getting memory into which to place the data
1673 created by pcre_study(). The fourth argument should point to a size_t
1674 variable.
1675
1676
1677 OBSOLETE INFO FUNCTION
1678
1679 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1680
1681 The pcre_info() function is now obsolete because its interface is too
1682 restrictive to return all the available data about a compiled pattern.
1683 New programs should use pcre_fullinfo() instead. The yield of
1684 pcre_info() is the number of capturing subpatterns, or one of the fol-
1685 lowing negative numbers:
1686
1687 PCRE_ERROR_NULL the argument code was NULL
1688 PCRE_ERROR_BADMAGIC the "magic number" was not found
1689
1690 If the optptr argument is not NULL, a copy of the options with which
1691 the pattern was compiled is placed in the integer it points to (see
1692 PCRE_INFO_OPTIONS above).
1693
1694 If the pattern is not anchored and the firstcharptr argument is not
1695 NULL, it is used to pass back information about the first character of
1696 any matched string (see PCRE_INFO_FIRSTBYTE above).
1697
1698
1699 REFERENCE COUNTS
1700
1701 int pcre_refcount(pcre *code, int adjust);
1702
1703 The pcre_refcount() function is used to maintain a reference count in
1704 the data block that contains a compiled pattern. It is provided for the
1705 benefit of applications that operate in an object-oriented manner,
1706 where different parts of the application may be using the same compiled
1707 pattern, but you want to free the block when they are all done.
1708
1709 When a pattern is compiled, the reference count field is initialized to
1710 zero. It is changed only by calling this function, whose action is to
1711 add the adjust value (which may be positive or negative) to it. The
1712 yield of the function is the new value. However, the value of the count
1713 is constrained to lie between 0 and 65535, inclusive. If the new value
1714 is outside these limits, it is forced to the appropriate limit value.
1715
1716 Except when it is zero, the reference count is not correctly preserved
1717 if a pattern is compiled on one host and then transferred to a host
1718 whose byte-order is different. (This seems a highly unlikely scenario.)
1719
1720
1721 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1722
1723 int pcre_exec(const pcre *code, const pcre_extra *extra,
1724 const char *subject, int length, int startoffset,
1725 int options, int *ovector, int ovecsize);
1726
1727 The function pcre_exec() is called to match a subject string against a
1728 compiled pattern, which is passed in the code argument. If the pattern
1729 has been studied, the result of the study should be passed in the extra
1730 argument. This function is the main matching facility of the library,
1731 and it operates in a Perl-like manner. For specialist use there is also
1732 an alternative matching function, which is described below in the sec-
1733 tion about the pcre_dfa_exec() function.
1734
1735 In most applications, the pattern will have been compiled (and option-
1736 ally studied) in the same process that calls pcre_exec(). However, it
1737 is possible to save compiled patterns and study data, and then use them
1738 later in different processes, possibly even on different hosts. For a
1739 discussion about this, see the pcreprecompile documentation.
1740
1741 Here is an example of a simple call to pcre_exec():
1742
1743 int rc;
1744 int ovector[30];
1745 rc = pcre_exec(
1746 re, /* result of pcre_compile() */
1747 NULL, /* we didn't study the pattern */
1748 "some string", /* the subject string */
1749 11, /* the length of the subject string */
1750 0, /* start at offset 0 in the subject */
1751 0, /* default options */
1752 ovector, /* vector of integers for substring information */
1753 30); /* number of elements (NOT size in bytes) */
1754
1755 Extra data for pcre_exec()
1756
1757 If the extra argument is not NULL, it must point to a pcre_extra data
1758 block. The pcre_study() function returns such a block (when it doesn't
1759 return NULL), but you can also create one for yourself, and pass addi-
1760 tional information in it. The pcre_extra block contains the following
1761 fields (not necessarily in this order):
1762
1763 unsigned long int flags;
1764 void *study_data;
1765 unsigned long int match_limit;
1766 unsigned long int match_limit_recursion;
1767 void *callout_data;
1768 const unsigned char *tables;
1769
1770 The flags field is a bitmap that specifies which of the other fields
1771 are set. The flag bits are:
1772
1773 PCRE_EXTRA_STUDY_DATA
1774 PCRE_EXTRA_MATCH_LIMIT
1775 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1776 PCRE_EXTRA_CALLOUT_DATA
1777 PCRE_EXTRA_TABLES
1778
1779 Other flag bits should be set to zero. The study_data field is set in
1780 the pcre_extra block that is returned by pcre_study(), together with
1781 the appropriate flag bit. You should not set this yourself, but you may
1782 add to the block by setting the other fields and their corresponding
1783 flag bits.
1784
1785 The match_limit field provides a means of preventing PCRE from using up
1786 a vast amount of resources when running patterns that are not going to
1787 match, but which have a very large number of possibilities in their
1788 search trees. The classic example is the use of nested unlimited
1789 repeats.
1790
1791 Internally, PCRE uses a function called match() which it calls repeat-
1792 edly (sometimes recursively). The limit set by match_limit is imposed
1793 on the number of times this function is called during a match, which
1794 has the effect of limiting the amount of backtracking that can take
1795 place. For patterns that are not anchored, the count restarts from zero
1796 for each position in the subject string.
1797
1798 The default value for the limit can be set when PCRE is built; the
1799 default default is 10 million, which handles all but the most extreme
1800 cases. You can override the default by suppling pcre_exec() with a
1801 pcre_extra block in which match_limit is set, and
1802 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1803 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1804
1805 The match_limit_recursion field is similar to match_limit, but instead
1806 of limiting the total number of times that match() is called, it limits
1807 the depth of recursion. The recursion depth is a smaller number than
1808 the total number of calls, because not all calls to match() are recur-
1809 sive. This limit is of use only if it is set smaller than match_limit.
1810
1811 Limiting the recursion depth limits the amount of stack that can be
1812 used, or, when PCRE has been compiled to use memory on the heap instead
1813 of the stack, the amount of heap memory that can be used.
1814
1815 The default value for match_limit_recursion can be set when PCRE is
1816 built; the default default is the same value as the default for
1817 match_limit. You can override the default by suppling pcre_exec() with
1818 a pcre_extra block in which match_limit_recursion is set, and
1819 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1820 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1821
1822 The pcre_callout field is used in conjunction with the "callout" fea-
1823 ture, which is described in the pcrecallout documentation.
1824
1825 The tables field is used to pass a character tables pointer to
1826 pcre_exec(); this overrides the value that is stored with the compiled
1827 pattern. A non-NULL value is stored with the compiled pattern only if
1828 custom tables were supplied to pcre_compile() via its tableptr argu-
1829 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1830 PCRE's internal tables to be used. This facility is helpful when re-
1831 using patterns that have been saved after compiling with an external
1832 set of tables, because the external tables might be at a different
1833 address when pcre_exec() is called. See the pcreprecompile documenta-
1834 tion for a discussion of saving compiled patterns for later use.
1835
1836 Option bits for pcre_exec()
1837
1838 The unused bits of the options argument for pcre_exec() must be zero.
1839 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1840 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1841 PCRE_PARTIAL.
1842
1843 PCRE_ANCHORED
1844
1845 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1846 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1847 turned out to be anchored by virtue of its contents, it cannot be made
1848 unachored at matching time.
1849
1850 PCRE_BSR_ANYCRLF
1851 PCRE_BSR_UNICODE
1852
1853 These options (which are mutually exclusive) control what the \R escape
1854 sequence matches. The choice is either to match only CR, LF, or CRLF,
1855 or to match any Unicode newline sequence. These options override the
1856 choice that was made or defaulted when the pattern was compiled.
1857
1858 PCRE_NEWLINE_CR
1859 PCRE_NEWLINE_LF
1860 PCRE_NEWLINE_CRLF
1861 PCRE_NEWLINE_ANYCRLF
1862 PCRE_NEWLINE_ANY
1863
1864 These options override the newline definition that was chosen or
1865 defaulted when the pattern was compiled. For details, see the descrip-
1866 tion of pcre_compile() above. During matching, the newline choice
1867 affects the behaviour of the dot, circumflex, and dollar metacharac-
1868 ters. It may also alter the way the match position is advanced after a
1869 match failure for an unanchored pattern.
1870
1871 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1872 set, and a match attempt for an unanchored pattern fails when the cur-
1873 rent position is at a CRLF sequence, and the pattern contains no
1874 explicit matches for CR or LF characters, the match position is
1875 advanced by two characters instead of one, in other words, to after the
1876 CRLF.
1877
1878 The above rule is a compromise that makes the most common cases work as
1879 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1880 option is not set), it does not match the string "\r\nA" because, after
1881 failing at the start, it skips both the CR and the LF before retrying.
1882 However, the pattern [\r\n]A does match that string, because it con-
1883 tains an explicit CR or LF reference, and so advances only by one char-
1884 acter after the first failure.
1885
1886 An explicit match for CR of LF is either a literal appearance of one of
1887 those characters, or one of the \r or \n escape sequences. Implicit
1888 matches such as [^X] do not count, nor does \s (which includes CR and
1889 LF in the characters that it matches).
1890
1891 Notwithstanding the above, anomalous effects may still occur when CRLF
1892 is a valid newline sequence and explicit \r or \n escapes appear in the
1893 pattern.
1894
1895 PCRE_NOTBOL
1896
1897 This option specifies that first character of the subject string is not
1898 the beginning of a line, so the circumflex metacharacter should not
1899 match before it. Setting this without PCRE_MULTILINE (at compile time)
1900 causes circumflex never to match. This option affects only the behav-
1901 iour of the circumflex metacharacter. It does not affect \A.
1902
1903 PCRE_NOTEOL
1904
1905 This option specifies that the end of the subject string is not the end
1906 of a line, so the dollar metacharacter should not match it nor (except
1907 in multiline mode) a newline immediately before it. Setting this with-
1908 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1909 option affects only the behaviour of the dollar metacharacter. It does
1910 not affect \Z or \z.
1911
1912 PCRE_NOTEMPTY
1913
1914 An empty string is not considered to be a valid match if this option is
1915 set. If there are alternatives in the pattern, they are tried. If all
1916 the alternatives match the empty string, the entire match fails. For
1917 example, if the pattern
1918
1919 a?b?
1920
1921 is applied to a string not beginning with "a" or "b", it matches the
1922 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1923 match is not valid, so PCRE searches further into the string for occur-
1924 rences of "a" or "b".
1925
1926 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1927 cial case of a pattern match of the empty string within its split()
1928 function, and when using the /g modifier. It is possible to emulate
1929 Perl's behaviour after matching a null string by first trying the match
1930 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1931 if that fails by advancing the starting offset (see below) and trying
1932 an ordinary match again. There is some code that demonstrates how to do
1933 this in the pcredemo.c sample program.
1934
1935 PCRE_NO_UTF8_CHECK
1936
1937 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1938 UTF-8 string is automatically checked when pcre_exec() is subsequently
1939 called. The value of startoffset is also checked to ensure that it
1940 points to the start of a UTF-8 character. There is a discussion about
1941 the validity of UTF-8 strings in the section on UTF-8 support in the
1942 main pcre page. If an invalid UTF-8 sequence of bytes is found,
1943 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
1944 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
1945
1946 If you already know that your subject is valid, and you want to skip
1947 these checks for performance reasons, you can set the
1948 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1949 do this for the second and subsequent calls to pcre_exec() if you are
1950 making repeated calls to find all the matches in a single subject
1951 string. However, you should be sure that the value of startoffset
1952 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1953 set, the effect of passing an invalid UTF-8 string as a subject, or a
1954 value of startoffset that does not point to the start of a UTF-8 char-
1955 acter, is undefined. Your program may crash.
1956
1957 PCRE_PARTIAL
1958
1959 This option turns on the partial matching feature. If the subject
1960 string fails to match the pattern, but at some point during the match-
1961 ing process the end of the subject was reached (that is, the subject
1962 partially matches the pattern and the failure to match occurred only
1963 because there were not enough subject characters), pcre_exec() returns
1964 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1965 used, there are restrictions on what may appear in the pattern. These
1966 are discussed in the pcrepartial documentation.
1967
1968 The string to be matched by pcre_exec()
1969
1970 The subject string is passed to pcre_exec() as a pointer in subject, a
1971 length in length, and a starting byte offset in startoffset. In UTF-8
1972 mode, the byte offset must point to the start of a UTF-8 character.
1973 Unlike the pattern string, the subject may contain binary zero bytes.
1974 When the starting offset is zero, the search for a match starts at the
1975 beginning of the subject, and this is by far the most common case.
1976
1977 A non-zero starting offset is useful when searching for another match
1978 in the same subject by calling pcre_exec() again after a previous suc-
1979 cess. Setting startoffset differs from just passing over a shortened
1980 string and setting PCRE_NOTBOL in the case of a pattern that begins
1981 with any kind of lookbehind. For example, consider the pattern
1982
1983 \Biss\B
1984
1985 which finds occurrences of "iss" in the middle of words. (\B matches
1986 only if the current position in the subject is not a word boundary.)
1987 When applied to the string "Mississipi" the first call to pcre_exec()
1988 finds the first occurrence. If pcre_exec() is called again with just
1989 the remainder of the subject, namely "issipi", it does not match,
1990 because \B is always false at the start of the subject, which is deemed
1991 to be a word boundary. However, if pcre_exec() is passed the entire
1992 string again, but with startoffset set to 4, it finds the second occur-
1993 rence of "iss" because it is able to look behind the starting point to
1994 discover that it is preceded by a letter.
1995
1996 If a non-zero starting offset is passed when the pattern is anchored,
1997 one attempt to match at the given offset is made. This can only succeed
1998 if the pattern does not require the match to be at the start of the
1999 subject.
2000
2001 How pcre_exec() returns captured substrings
2002
2003 In general, a pattern matches a certain portion of the subject, and in
2004 addition, further substrings from the subject may be picked out by
2005 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2006 this is called "capturing" in what follows, and the phrase "capturing
2007 subpattern" is used for a fragment of a pattern that picks out a sub-
2008 string. PCRE supports several other kinds of parenthesized subpattern
2009 that do not cause substrings to be captured.
2010
2011 Captured substrings are returned to the caller via a vector of integer
2012 offsets whose address is passed in ovector. The number of elements in
2013 the vector is passed in ovecsize, which must be a non-negative number.
2014 Note: this argument is NOT the size of ovector in bytes.
2015
2016 The first two-thirds of the vector is used to pass back captured sub-
2017 strings, each substring using a pair of integers. The remaining third
2018 of the vector is used as workspace by pcre_exec() while matching cap-
2019 turing subpatterns, and is not available for passing back information.
2020 The length passed in ovecsize should always be a multiple of three. If
2021 it is not, it is rounded down.
2022
2023 When a match is successful, information about captured substrings is
2024 returned in pairs of integers, starting at the beginning of ovector,
2025 and continuing up to two-thirds of its length at the most. The first
2026 element of a pair is set to the offset of the first character in a sub-
2027 string, and the second is set to the offset of the first character
2028 after the end of a substring. The first pair, ovector[0] and ovec-
2029 tor[1], identify the portion of the subject string matched by the
2030 entire pattern. The next pair is used for the first capturing subpat-
2031 tern, and so on. The value returned by pcre_exec() is one more than the
2032 highest numbered pair that has been set. For example, if two substrings
2033 have been captured, the returned value is 3. If there are no capturing
2034 subpatterns, the return value from a successful match is 1, indicating
2035 that just the first pair of offsets has been set.
2036
2037 If a capturing subpattern is matched repeatedly, it is the last portion
2038 of the string that it matched that is returned.
2039
2040 If the vector is too small to hold all the captured substring offsets,
2041 it is used as far as possible (up to two-thirds of its length), and the
2042 function returns a value of zero. In particular, if the substring off-
2043 sets are not of interest, pcre_exec() may be called with ovector passed
2044 as NULL and ovecsize as zero. However, if the pattern contains back
2045 references and the ovector is not big enough to remember the related
2046 substrings, PCRE has to get additional memory for use during matching.
2047 Thus it is usually advisable to supply an ovector.
2048
2049 The pcre_info() function can be used to find out how many capturing
2050 subpatterns there are in a compiled pattern. The smallest size for
2051 ovector that will allow for n captured substrings, in addition to the
2052 offsets of the substring matched by the whole pattern, is (n+1)*3.
2053
2054 It is possible for capturing subpattern number n+1 to match some part
2055 of the subject when subpattern n has not been used at all. For example,
2056 if the string "abc" is matched against the pattern (a|(z))(bc) the
2057 return from the function is 4, and subpatterns 1 and 3 are matched, but
2058 2 is not. When this happens, both values in the offset pairs corre-
2059 sponding to unused subpatterns are set to -1.
2060
2061 Offset values that correspond to unused subpatterns at the end of the
2062 expression are also set to -1. For example, if the string "abc" is
2063 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2064 matched. The return from the function is 2, because the highest used
2065 capturing subpattern number is 1. However, you can refer to the offsets
2066 for the second and third capturing subpatterns if you wish (assuming
2067 the vector is large enough, of course).
2068
2069 Some convenience functions are provided for extracting the captured
2070 substrings as separate strings. These are described below.
2071
2072 Error return values from pcre_exec()
2073
2074 If pcre_exec() fails, it returns a negative number. The following are
2075 defined in the header file:
2076
2077 PCRE_ERROR_NOMATCH (-1)
2078
2079 The subject string did not match the pattern.
2080
2081 PCRE_ERROR_NULL (-2)
2082
2083 Either code or subject was passed as NULL, or ovector was NULL and
2084 ovecsize was not zero.
2085
2086 PCRE_ERROR_BADOPTION (-3)
2087
2088 An unrecognized bit was set in the options argument.
2089
2090 PCRE_ERROR_BADMAGIC (-4)
2091
2092 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2093 to catch the case when it is passed a junk pointer and to detect when a
2094 pattern that was compiled in an environment of one endianness is run in
2095 an environment with the other endianness. This is the error that PCRE
2096 gives when the magic number is not present.
2097
2098 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2099
2100 While running the pattern match, an unknown item was encountered in the
2101 compiled pattern. This error could be caused by a bug in PCRE or by
2102 overwriting of the compiled pattern.
2103
2104 PCRE_ERROR_NOMEMORY (-6)
2105
2106 If a pattern contains back references, but the ovector that is passed
2107 to pcre_exec() is not big enough to remember the referenced substrings,
2108 PCRE gets a block of memory at the start of matching to use for this
2109 purpose. If the call via pcre_malloc() fails, this error is given. The
2110 memory is automatically freed at the end of matching.
2111
2112 PCRE_ERROR_NOSUBSTRING (-7)
2113
2114 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2115 and pcre_get_substring_list() functions (see below). It is never
2116 returned by pcre_exec().
2117
2118 PCRE_ERROR_MATCHLIMIT (-8)
2119
2120 The backtracking limit, as specified by the match_limit field in a
2121 pcre_extra structure (or defaulted) was reached. See the description
2122 above.
2123
2124 PCRE_ERROR_CALLOUT (-9)
2125
2126 This error is never generated by pcre_exec() itself. It is provided for
2127 use by callout functions that want to yield a distinctive error code.
2128 See the pcrecallout documentation for details.
2129
2130 PCRE_ERROR_BADUTF8 (-10)
2131
2132 A string that contains an invalid UTF-8 byte sequence was passed as a
2133 subject.
2134
2135 PCRE_ERROR_BADUTF8_OFFSET (-11)
2136
2137 The UTF-8 byte sequence that was passed as a subject was valid, but the
2138 value of startoffset did not point to the beginning of a UTF-8 charac-
2139 ter.
2140
2141 PCRE_ERROR_PARTIAL (-12)
2142
2143 The subject string did not match, but it did match partially. See the
2144 pcrepartial documentation for details of partial matching.
2145
2146 PCRE_ERROR_BADPARTIAL (-13)
2147
2148 The PCRE_PARTIAL option was used with a compiled pattern containing
2149 items that are not supported for partial matching. See the pcrepartial
2150 documentation for details of partial matching.
2151
2152 PCRE_ERROR_INTERNAL (-14)
2153
2154 An unexpected internal error has occurred. This error could be caused
2155 by a bug in PCRE or by overwriting of the compiled pattern.
2156
2157 PCRE_ERROR_BADCOUNT (-15)
2158
2159 This error is given if the value of the ovecsize argument is negative.
2160
2161 PCRE_ERROR_RECURSIONLIMIT (-21)
2162
2163 The internal recursion limit, as specified by the match_limit_recursion
2164 field in a pcre_extra structure (or defaulted) was reached. See the
2165 description above.
2166
2167 PCRE_ERROR_BADNEWLINE (-23)
2168
2169 An invalid combination of PCRE_NEWLINE_xxx options was given.
2170
2171 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2172
2173
2174 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2175
2176 int pcre_copy_substring(const char *subject, int *ovector,
2177 int stringcount, int stringnumber, char *buffer,
2178 int buffersize);
2179
2180 int pcre_get_substring(const char *subject, int *ovector,
2181 int stringcount, int stringnumber,
2182 const char **stringptr);
2183
2184 int pcre_get_substring_list(const char *subject,
2185 int *ovector, int stringcount, const char ***listptr);
2186
2187 Captured substrings can be accessed directly by using the offsets
2188 returned by pcre_exec() in ovector. For convenience, the functions
2189 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2190 string_list() are provided for extracting captured substrings as new,
2191 separate, zero-terminated strings. These functions identify substrings
2192 by number. The next section describes functions for extracting named
2193 substrings.
2194
2195 A substring that contains a binary zero is correctly extracted and has
2196 a further zero added on the end, but the result is not, of course, a C
2197 string. However, you can process such a string by referring to the
2198 length that is returned by pcre_copy_substring() and pcre_get_sub-
2199 string(). Unfortunately, the interface to pcre_get_substring_list() is
2200 not adequate for handling strings containing binary zeros, because the
2201 end of the final string is not independently indicated.
2202
2203 The first three arguments are the same for all three of these func-
2204 tions: subject is the subject string that has just been successfully
2205 matched, ovector is a pointer to the vector of integer offsets that was
2206 passed to pcre_exec(), and stringcount is the number of substrings that
2207 were captured by the match, including the substring that matched the
2208 entire regular expression. This is the value returned by pcre_exec() if
2209 it is greater than zero. If pcre_exec() returned zero, indicating that
2210 it ran out of space in ovector, the value passed as stringcount should
2211 be the number of elements in the vector divided by three.
2212
2213 The functions pcre_copy_substring() and pcre_get_substring() extract a
2214 single substring, whose number is given as stringnumber. A value of
2215 zero extracts the substring that matched the entire pattern, whereas
2216 higher values extract the captured substrings. For pcre_copy_sub-
2217 string(), the string is placed in buffer, whose length is given by
2218 buffersize, while for pcre_get_substring() a new block of memory is
2219 obtained via pcre_malloc, and its address is returned via stringptr.
2220 The yield of the function is the length of the string, not including
2221 the terminating zero, or one of these error codes:
2222
2223 PCRE_ERROR_NOMEMORY (-6)
2224
2225 The buffer was too small for pcre_copy_substring(), or the attempt to
2226 get memory failed for pcre_get_substring().
2227
2228 PCRE_ERROR_NOSUBSTRING (-7)
2229
2230 There is no substring whose number is stringnumber.
2231
2232 The pcre_get_substring_list() function extracts all available sub-
2233 strings and builds a list of pointers to them. All this is done in a
2234 single block of memory that is obtained via pcre_malloc. The address of
2235 the memory block is returned via listptr, which is also the start of
2236 the list of string pointers. The end of the list is marked by a NULL
2237 pointer. The yield of the function is zero if all went well, or the
2238 error code
2239
2240 PCRE_ERROR_NOMEMORY (-6)
2241
2242 if the attempt to get the memory block failed.
2243
2244 When any of these functions encounter a substring that is unset, which
2245 can happen when capturing subpattern number n+1 matches some part of
2246 the subject, but subpattern n has not been used at all, they return an
2247 empty string. This can be distinguished from a genuine zero-length sub-
2248 string by inspecting the appropriate offset in ovector, which is nega-
2249 tive for unset substrings.
2250
2251 The two convenience functions pcre_free_substring() and pcre_free_sub-
2252 string_list() can be used to free the memory returned by a previous
2253 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2254 tively. They do nothing more than call the function pointed to by
2255 pcre_free, which of course could be called directly from a C program.
2256 However, PCRE is used in some situations where it is linked via a spe-
2257 cial interface to another programming language that cannot use
2258 pcre_free directly; it is for these cases that the functions are pro-
2259 vided.
2260
2261
2262 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2263
2264 int pcre_get_stringnumber(const pcre *code,
2265 const char *name);
2266
2267 int pcre_copy_named_substring(const pcre *code,
2268 const char *subject, int *ovector,
2269 int stringcount, const char *stringname,
2270 char *buffer, int buffersize);
2271
2272 int pcre_get_named_substring(const pcre *code,
2273 const char *subject, int *ovector,
2274 int stringcount, const char *stringname,
2275 const char **stringptr);
2276
2277 To extract a substring by name, you first have to find associated num-
2278 ber. For example, for this pattern
2279
2280 (a+)b(?<xxx>\d+)...
2281
2282 the number of the subpattern called "xxx" is 2. If the name is known to
2283 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2284 name by calling pcre_get_stringnumber(). The first argument is the com-
2285 piled pattern, and the second is the name. The yield of the function is
2286 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2287 subpattern of that name.
2288
2289 Given the number, you can extract the substring directly, or use one of
2290 the functions described in the previous section. For convenience, there
2291 are also two functions that do the whole job.
2292
2293 Most of the arguments of pcre_copy_named_substring() and
2294 pcre_get_named_substring() are the same as those for the similarly
2295 named functions that extract by number. As these are described in the
2296 previous section, they are not re-described here. There are just two
2297 differences:
2298
2299 First, instead of a substring number, a substring name is given. Sec-
2300 ond, there is an extra argument, given at the start, which is a pointer
2301 to the compiled pattern. This is needed in order to gain access to the
2302 name-to-number translation table.
2303
2304 These functions call pcre_get_stringnumber(), and if it succeeds, they
2305 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2306 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2307 behaviour may not be what you want (see the next section).
2308
2309
2310 DUPLICATE SUBPATTERN NAMES
2311
2312 int pcre_get_stringtable_entries(const pcre *code,
2313 const char *name, char **first, char **last);
2314
2315 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2316 subpatterns are not required to be unique. Normally, patterns with
2317 duplicate names are such that in any one match, only one of the named
2318 subpatterns participates. An example is shown in the pcrepattern docu-
2319 mentation.
2320
2321 When duplicates are present, pcre_copy_named_substring() and
2322 pcre_get_named_substring() return the first substring corresponding to
2323 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2324 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2325 function returns one of the numbers that are associated with the name,
2326 but it is not defined which it is.
2327
2328 If you want to get full details of all captured substrings for a given
2329 name, you must use the pcre_get_stringtable_entries() function. The
2330 first argument is the compiled pattern, and the second is the name. The
2331 third and fourth are pointers to variables which are updated by the
2332 function. After it has run, they point to the first and last entries in
2333 the name-to-number table for the given name. The function itself
2334 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2335 there are none. The format of the table is described above in the sec-
2336 tion entitled Information about a pattern. Given all the relevant
2337 entries for the name, you can extract each of their numbers, and hence
2338 the captured data, if any.
2339
2340
2341 FINDING ALL POSSIBLE MATCHES
2342
2343 The traditional matching function uses a similar algorithm to Perl,
2344 which stops when it finds the first match, starting at a given point in
2345 the subject. If you want to find all possible matches, or the longest
2346 possible match, consider using the alternative matching function (see
2347 below) instead. If you cannot use the alternative function, but still
2348 need to find all possible matches, you can kludge it up by making use
2349 of the callout facility, which is described in the pcrecallout documen-
2350 tation.
2351
2352 What you have to do is to insert a callout right at the end of the pat-
2353 tern. When your callout function is called, extract and save the cur-
2354 rent matched substring. Then return 1, which forces pcre_exec() to
2355 backtrack and try other alternatives. Ultimately, when it runs out of
2356 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2357
2358
2359 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2360
2361 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2362 const char *subject, int length, int startoffset,
2363 int options, int *ovector, int ovecsize,
2364 int *workspace, int wscount);
2365
2366 The function pcre_dfa_exec() is called to match a subject string
2367 against a compiled pattern, using a matching algorithm that scans the
2368 subject string just once, and does not backtrack. This has different
2369 characteristics to the normal algorithm, and is not compatible with
2370 Perl. Some of the features of PCRE patterns are not supported. Never-
2371 theless, there are times when this kind of matching can be useful. For
2372 a discussion of the two matching algorithms, see the pcrematching docu-
2373 mentation.
2374
2375 The arguments for the pcre_dfa_exec() function are the same as for
2376 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2377 ent way, and this is described below. The other common arguments are
2378 used in the same way as for pcre_exec(), so their description is not
2379 repeated here.
2380
2381 The two additional arguments provide workspace for the function. The
2382 workspace vector should contain at least 20 elements. It is used for
2383 keeping track of multiple paths through the pattern tree. More
2384 workspace will be needed for patterns and subjects where there are a
2385 lot of potential matches.
2386
2387 Here is an example of a simple call to pcre_dfa_exec():
2388
2389 int rc;
2390 int ovector[10];
2391 int wspace[20];
2392 rc = pcre_dfa_exec(
2393 re, /* result of pcre_compile() */
2394 NULL, /* we didn't study the pattern */
2395 "some string", /* the subject string */
2396 11, /* the length of the subject string */
2397 0, /* start at offset 0 in the subject */
2398 0, /* default options */
2399 ovector, /* vector of integers for substring information */
2400 10, /* number of elements (NOT size in bytes) */
2401 wspace, /* working space vector */
2402 20); /* number of elements (NOT size in bytes) */
2403
2404 Option bits for pcre_dfa_exec()
2405
2406 The unused bits of the options argument for pcre_dfa_exec() must be
2407 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2408 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2409 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2410 three of these are the same as for pcre_exec(), so their description is
2411 not repeated here.
2412
2413 PCRE_PARTIAL
2414
2415 This has the same general effect as it does for pcre_exec(), but the
2416 details are slightly different. When PCRE_PARTIAL is set for
2417 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2418 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2419 been no complete matches, but there is still at least one matching pos-
2420 sibility. The portion of the string that provided the partial match is
2421 set as the first matching string.
2422
2423 PCRE_DFA_SHORTEST
2424
2425 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2426 stop as soon as it has found one match. Because of the way the alterna-
2427 tive algorithm works, this is necessarily the shortest possible match
2428 at the first possible matching point in the subject string.
2429
2430 PCRE_DFA_RESTART
2431
2432 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2433 returns a partial match, it is possible to call it again, with addi-
2434 tional subject characters, and have it continue with the same match.
2435 The PCRE_DFA_RESTART option requests this action; when it is set, the
2436 workspace and wscount options must reference the same vector as before
2437 because data about the match so far is left in them after a partial
2438 match. There is more discussion of this facility in the pcrepartial
2439 documentation.
2440
2441 Successful returns from pcre_dfa_exec()
2442
2443 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2444 string in the subject. Note, however, that all the matches from one run
2445 of the function start at the same point in the subject. The shorter
2446 matches are all initial substrings of the longer matches. For example,
2447 if the pattern
2448
2449 <.*>
2450
2451 is matched against the string
2452
2453 This is <something> <something else> <something further> no more
2454
2455 the three matched strings are
2456
2457 <something>
2458 <something> <something else>
2459 <something> <something else> <something further>
2460
2461 On success, the yield of the function is a number greater than zero,
2462 which is the number of matched substrings. The substrings themselves
2463 are returned in ovector. Each string uses two elements; the first is
2464 the offset to the start, and the second is the offset to the end. In
2465 fact, all the strings have the same start offset. (Space could have
2466 been saved by giving this only once, but it was decided to retain some
2467 compatibility with the way pcre_exec() returns data, even though the
2468 meaning of the strings is different.)
2469
2470 The strings are returned in reverse order of length; that is, the long-
2471 est matching string is given first. If there were too many matches to
2472 fit into ovector, the yield of the function is zero, and the vector is
2473 filled with the longest matches.
2474
2475 Error returns from pcre_dfa_exec()
2476
2477 The pcre_dfa_exec() function returns a negative number when it fails.
2478 Many of the errors are the same as for pcre_exec(), and these are
2479 described above. There are in addition the following errors that are
2480 specific to pcre_dfa_exec():
2481
2482 PCRE_ERROR_DFA_UITEM (-16)
2483
2484 This return is given if pcre_dfa_exec() encounters an item in the pat-
2485 tern that it does not support, for instance, the use of \C or a back
2486 reference.
2487
2488 PCRE_ERROR_DFA_UCOND (-17)
2489
2490 This return is given if pcre_dfa_exec() encounters a condition item
2491 that uses a back reference for the condition, or a test for recursion
2492 in a specific group. These are not supported.
2493
2494 PCRE_ERROR_DFA_UMLIMIT (-18)
2495
2496 This return is given if pcre_dfa_exec() is called with an extra block
2497 that contains a setting of the match_limit field. This is not supported
2498 (it is meaningless).
2499
2500 PCRE_ERROR_DFA_WSSIZE (-19)
2501
2502 This return is given if pcre_dfa_exec() runs out of space in the
2503 workspace vector.
2504
2505 PCRE_ERROR_DFA_RECURSE (-20)
2506
2507 When a recursive subpattern is processed, the matching function calls
2508 itself recursively, using private vectors for ovector and workspace.
2509 This error is given if the output vector is not large enough. This
2510 should be extremely rare, as a vector of size 1000 is used.
2511
2512
2513 SEE ALSO
2514
2515 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2516 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2517
2518
2519 AUTHOR
2520
2521 Philip Hazel
2522 University Computing Service
2523 Cambridge CB2 3QH, England.
2524
2525
2526 REVISION
2527
2528 Last updated: 11 September 2007
2529 Copyright (c) 1997-2007 University of Cambridge.
2530 ------------------------------------------------------------------------------
2531
2532
2533 PCRECALLOUT(3) PCRECALLOUT(3)
2534
2535
2536 NAME
2537 PCRE - Perl-compatible regular expressions
2538
2539
2540 PCRE CALLOUTS
2541
2542 int (*pcre_callout)(pcre_callout_block *);
2543
2544 PCRE provides a feature called "callout", which is a means of temporar-
2545 ily passing control to the caller of PCRE in the middle of pattern
2546 matching. The caller of PCRE provides an external function by putting
2547 its entry point in the global variable pcre_callout. By default, this
2548 variable contains NULL, which disables all calling out.
2549
2550 Within a regular expression, (?C) indicates the points at which the
2551 external function is to be called. Different callout points can be
2552 identified by putting a number less than 256 after the letter C. The
2553 default value is zero. For example, this pattern has two callout
2554 points:
2555
2556 (?C1)abc(?C2)def
2557
2558 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2559 called, PCRE automatically inserts callouts, all with number 255,
2560 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2561 used with the pattern
2562
2563 A(\d{2}|--)
2564
2565 it is processed as if it were
2566
2567 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2568
2569 Notice that there is a callout before and after each parenthesis and
2570 alternation bar. Automatic callouts can be used for tracking the
2571 progress of pattern matching. The pcretest command has an option that
2572 sets automatic callouts; when it is used, the output indicates how the
2573 pattern is matched. This is useful information when you are trying to
2574 optimize the performance of a particular pattern.
2575
2576
2577 MISSING CALLOUTS
2578
2579 You should be aware that, because of optimizations in the way PCRE
2580 matches patterns, callouts sometimes do not happen. For example, if the
2581 pattern is
2582
2583 ab(?C4)cd
2584
2585 PCRE knows that any matching string must contain the letter "d". If the
2586 subject string is "abyz", the lack of "d" means that matching doesn't
2587 ever start, and the callout is never reached. However, with "abyd",
2588 though the result is still no match, the callout is obeyed.
2589
2590
2591 THE CALLOUT INTERFACE
2592
2593 During matching, when PCRE reaches a callout point, the external func-
2594 tion defined by pcre_callout is called (if it is set). This applies to
2595 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2596 only argument to the callout function is a pointer to a pcre_callout
2597 block. This structure contains the following fields:
2598
2599 int version;
2600 int callout_number;
2601 int *offset_vector;
2602 const char *subject;
2603 int subject_length;
2604 int start_match;
2605 int current_position;
2606 int capture_top;
2607 int capture_last;
2608 void *callout_data;
2609 int pattern_position;
2610 int next_item_length;
2611
2612 The version field is an integer containing the version number of the
2613 block format. The initial version was 0; the current version is 1. The
2614 version number will change again in future if additional fields are
2615 added, but the intention is never to remove any of the existing fields.
2616
2617 The callout_number field contains the number of the callout, as com-
2618 piled into the pattern (that is, the number after ?C for manual call-
2619 outs, and 255 for automatically generated callouts).
2620
2621 The offset_vector field is a pointer to the vector of offsets that was
2622 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2623 pcre_exec() is used, the contents can be inspected in order to extract
2624 substrings that have been matched so far, in the same way as for
2625 extracting substrings after a match has completed. For pcre_dfa_exec()
2626 this field is not useful.
2627
2628 The subject and subject_length fields contain copies of the values that
2629 were passed to pcre_exec().
2630
2631 The start_match field normally contains the offset within the subject
2632 at which the current match attempt started. However, if the escape
2633 sequence \K has been encountered, this value is changed to reflect the
2634 modified starting point. If the pattern is not anchored, the callout
2635 function may be called several times from the same point in the pattern
2636 for different starting points in the subject.
2637
2638 The current_position field contains the offset within the subject of
2639 the current match pointer.
2640
2641 When the pcre_exec() function is used, the capture_top field contains
2642 one more than the number of the highest numbered captured substring so
2643 far. If no substrings have been captured, the value of capture_top is
2644 one. This is always the case when pcre_dfa_exec() is used, because it
2645 does not support captured substrings.
2646
2647 The capture_last field contains the number of the most recently cap-
2648 tured substring. If no substrings have been captured, its value is -1.
2649 This is always the case when pcre_dfa_exec() is used.
2650
2651 The callout_data field contains a value that is passed to pcre_exec()
2652 or pcre_dfa_exec() specifically so that it can be passed back in call-
2653 outs. It is passed in the pcre_callout field of the pcre_extra data
2654 structure. If no such data was passed, the value of callout_data in a
2655 pcre_callout block is NULL. There is a description of the pcre_extra
2656 structure in the pcreapi documentation.
2657
2658 The pattern_position field is present from version 1 of the pcre_call-
2659 out structure. It contains the offset to the next item to be matched in
2660 the pattern string.
2661
2662 The next_item_length field is present from version 1 of the pcre_call-
2663 out structure. It contains the length of the next item to be matched in
2664 the pattern string. When the callout immediately precedes an alterna-
2665 tion bar, a closing parenthesis, or the end of the pattern, the length
2666 is zero. When the callout precedes an opening parenthesis, the length
2667 is that of the entire subpattern.
2668
2669 The pattern_position and next_item_length fields are intended to help
2670 in distinguishing between different automatic callouts, which all have
2671 the same callout number. However, they are set for all callouts.
2672
2673
2674 RETURN VALUES
2675
2676 The external callout function returns an integer to PCRE. If the value
2677 is zero, matching proceeds as normal. If the value is greater than
2678 zero, matching fails at the current point, but the testing of other
2679 matching possibilities goes ahead, just as if a lookahead assertion had
2680 failed. If the value is less than zero, the match is abandoned, and
2681 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2682
2683 Negative values should normally be chosen from the set of
2684 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2685 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2686 reserved for use by callout functions; it will never be used by PCRE
2687 itself.
2688
2689
2690 AUTHOR
2691
2692 Philip Hazel
2693 University Computing Service
2694 Cambridge CB2 3QH, England.
2695
2696
2697 REVISION
2698
2699 Last updated: 29 May 2007
2700 Copyright (c) 1997-2007 University of Cambridge.
2701 ------------------------------------------------------------------------------
2702
2703
2704 PCRECOMPAT(3) PCRECOMPAT(3)
2705
2706
2707 NAME
2708 PCRE - Perl-compatible regular expressions
2709
2710
2711 DIFFERENCES BETWEEN PCRE AND PERL
2712
2713 This document describes the differences in the ways that PCRE and Perl
2714 handle regular expressions. The differences described here are mainly
2715 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2716 some features that are expected to be in the forthcoming Perl 5.10.
2717
2718 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2719 of what it does have are given in the section on UTF-8 support in the
2720 main pcre page.
2721
2722 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2723 permits them, but they do not mean what you might think. For example,
2724 (?!a){3} does not assert that the next three characters are not "a". It
2725 just asserts that the next character is not "a" three times.
2726
2727 3. Capturing subpatterns that occur inside negative lookahead asser-
2728 tions are counted, but their entries in the offsets vector are never
2729 set. Perl sets its numerical variables from any such patterns that are
2730 matched before the assertion fails to match something (thereby succeed-
2731 ing), but only if the negative lookahead assertion contains just one
2732 branch.
2733
2734 4. Though binary zero characters are supported in the subject string,
2735 they are not allowed in a pattern string because it is passed as a nor-
2736 mal C string, terminated by zero. The escape sequence \0 can be used in
2737 the pattern to represent a binary zero.
2738
2739 5. The following Perl escape sequences are not supported: \l, \u, \L,
2740 \U, and \N. In fact these are implemented by Perl's general string-han-
2741 dling and are not part of its pattern matching engine. If any of these
2742 are encountered by PCRE, an error is generated.
2743
2744 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2745 is built with Unicode character property support. The properties that
2746 can be tested with \p and \P are limited to the general category prop-
2747 erties such as Lu and Nd, script names such as Greek or Han, and the
2748 derived properties Any and L&.
2749
2750 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2751 ters in between are treated as literals. This is slightly different
2752 from Perl in that $ and @ are also handled as literals inside the
2753 quotes. In Perl, they cause variable interpolation (but of course PCRE
2754 does not have variables). Note the following examples:
2755
2756 Pattern PCRE matches Perl matches
2757
2758 \Qabc$xyz\E abc$xyz abc followed by the
2759 contents of $xyz
2760 \Qabc\$xyz\E abc\$xyz abc\$xyz
2761 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2762
2763 The \Q...\E sequence is recognized both inside and outside character
2764 classes.
2765
2766 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2767 constructions. However, there is support for recursive patterns. This
2768 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2769 "callout" feature allows an external function to be called during pat-
2770 tern matching. See the pcrecallout documentation for details.
2771
2772 9. Subpatterns that are called recursively or as "subroutines" are
2773 always treated as atomic groups in PCRE. This is like Python, but
2774 unlike Perl.
2775
2776 10. There are some differences that are concerned with the settings of
2777 captured strings when part of a pattern is repeated. For example,
2778 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2779 unset, but in PCRE it is set to "b".
2780
2781 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2782 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2783 the forms without an argument. PCRE does not support (*MARK). If
2784 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2785 ture group; this is different to Perl.
2786
2787 12. PCRE provides some extensions to the Perl regular expression facil-
2788 ities. Perl 5.10 will include new features that are not in earlier
2789 versions, some of which (such as named parentheses) have been in PCRE
2790 for some time. This list is with respect to Perl 5.10:
2791
2792 (a) Although lookbehind assertions must match fixed length strings,
2793 each alternative branch of a lookbehind assertion can match a different
2794 length of string. Perl requires them all to have the same length.
2795
2796 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2797 meta-character matches only at the very end of the string.
2798
2799 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2800 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2801 ignored. (Perl can be made to issue a warning.)
2802
2803 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2804 fiers is inverted, that is, by default they are not greedy, but if fol-
2805 lowed by a question mark they are.
2806
2807 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2808 tried only at the first matching position in the subject string.
2809
2810 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2811 TURE options for pcre_exec() have no Perl equivalents.
2812
2813 (g) The \R escape sequence can be restricted to match only CR, LF, or
2814 CRLF by the PCRE_BSR_ANYCRLF option.
2815
2816 (h) The callout facility is PCRE-specific.
2817
2818 (i) The partial matching facility is PCRE-specific.
2819
2820 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2821 even on different hosts that have the other endianness.
2822
2823 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2824 different way and is not Perl-compatible.
2825
2826 (l) PCRE recognizes some special sequences such as (*CR) at the start
2827 of a pattern that set overall options that cannot be changed within the
2828 pattern.
2829
2830
2831 AUTHOR
2832
2833 Philip Hazel
2834 University Computing Service
2835 Cambridge CB2 3QH, England.
2836
2837
2838 REVISION
2839
2840 Last updated: 11 September 2007
2841 Copyright (c) 1997-2007 University of Cambridge.
2842 ------------------------------------------------------------------------------
2843
2844
2845 PCREPATTERN(3) PCREPATTERN(3)
2846
2847
2848 NAME
2849 PCRE - Perl-compatible regular expressions
2850
2851
2852 PCRE REGULAR EXPRESSION DETAILS
2853
2854 The syntax and semantics of the regular expressions that are supported
2855 by PCRE are described in detail below. There is a quick-reference syn-
2856 tax summary in the pcresyntax page. Perl's regular expressions are
2857 described in its own documentation, and regular expressions in general
2858 are covered in a number of books, some of which have copious examples.
2859 Jeffrey Friedl's "Mastering Regular Expressions", published by
2860 O'Reilly, covers regular expressions in great detail. This description
2861 of PCRE's regular expressions is intended as reference material.
2862
2863 The original operation of PCRE was on strings of one-byte characters.
2864 However, there is now also support for UTF-8 character strings. To use
2865 this, you must build PCRE to include UTF-8 support, and then call
2866 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2867 matching is mentioned in several places below. There is also a summary
2868 of UTF-8 features in the section on UTF-8 support in the main pcre
2869 page.
2870
2871 The remainder of this document discusses the patterns that are sup-
2872 ported by PCRE when its main matching function, pcre_exec(), is used.
2873 From release 6.0, PCRE offers a second matching function,
2874 pcre_dfa_exec(), which matches using a different algorithm that is not
2875 Perl-compatible. Some of the features discussed below are not available
2876 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2877 alternative function, and how it differs from the normal function, are
2878 discussed in the pcrematching page.
2879
2880
2881 NEWLINE CONVENTIONS
2882
2883 PCRE supports five different conventions for indicating line breaks in
2884 strings: a single CR (carriage return) character, a single LF (line-
2885 feed) character, the two-character sequence CRLF, any of the three pre-
2886 ceding, or any Unicode newline sequence. The pcreapi page has further
2887 discussion about newlines, and shows how to set the newline convention
2888 in the options arguments for the compiling and matching functions.
2889
2890 It is also possible to specify a newline convention by starting a pat-
2891 tern string with one of the following five sequences:
2892
2893 (*CR) carriage return
2894 (*LF) linefeed
2895 (*CRLF) carriage return, followed by linefeed
2896 (*ANYCRLF) any of the three above
2897 (*ANY) all Unicode newline sequences
2898
2899 These override the default and the options given to pcre_compile(). For
2900 example, on a Unix system where LF is the default newline sequence, the
2901 pattern
2902
2903 (*CR)a.b
2904
2905 changes the convention to CR. That pattern matches "a\nb" because LF is
2906 no longer a newline. Note that these special settings, which are not
2907 Perl-compatible, are recognized only at the very start of a pattern,
2908 and that they must be in upper case. If more than one of them is
2909 present, the last one is used.
2910
2911 The newline convention does not affect what the \R escape sequence
2912 matches. By default, this is any Unicode newline sequence, for Perl
2913 compatibility. However, this can be changed; see the description of \R
2914 in the section entitled "Newline sequences" below. A change of \R set-
2915 ting can be combined with a change of newline convention.
2916
2917
2918 CHARACTERS AND METACHARACTERS
2919
2920 A regular expression is a pattern that is matched against a subject
2921 string from left to right. Most characters stand for themselves in a
2922 pattern, and match the corresponding characters in the subject. As a
2923 trivial example, the pattern
2924
2925 The quick brown fox
2926
2927 matches a portion of a subject string that is identical to itself. When
2928 caseless matching is specified (the PCRE_CASELESS option), letters are
2929 matched independently of case. In UTF-8 mode, PCRE always understands
2930 the concept of case for characters whose values are less than 128, so
2931 caseless matching is always possible. For characters with higher val-
2932 ues, the concept of case is supported if PCRE is compiled with Unicode
2933 property support, but not otherwise. If you want to use caseless
2934 matching for characters 128 and above, you must ensure that PCRE is
2935 compiled with Unicode property support as well as with UTF-8 support.
2936
2937 The power of regular expressions comes from the ability to include
2938 alternatives and repetitions in the pattern. These are encoded in the
2939 pattern by the use of metacharacters, which do not stand for themselves
2940 but instead are interpreted in some special way.
2941
2942 There are two different sets of metacharacters: those that are recog-
2943 nized anywhere in the pattern except within square brackets, and those
2944 that are recognized within square brackets. Outside square brackets,
2945 the metacharacters are as follows:
2946
2947 \ general escape character with several uses
2948 ^ assert start of string (or line, in multiline mode)
2949 $ assert end of string (or line, in multiline mode)
2950 . match any character except newline (by default)
2951 [ start character class definition
2952 | start of alternative branch
2953 ( start subpattern
2954 ) end subpattern
2955 ? extends the meaning of (
2956 also 0 or 1 quantifier
2957 also quantifier minimizer
2958 * 0 or more quantifier
2959 + 1 or more quantifier
2960 also "possessive quantifier"
2961 { start min/max quantifier
2962
2963 Part of a pattern that is in square brackets is called a "character
2964 class". In a character class the only metacharacters are:
2965
2966 \ general escape character
2967 ^ negate the class, but only if the first character
2968 - indicates character range
2969 [ POSIX character class (only if followed by POSIX
2970 syntax)
2971 ] terminates the character class
2972
2973 The following sections describe the use of each of the metacharacters.
2974
2975
2976 BACKSLASH
2977
2978 The backslash character has several uses. Firstly, if it is followed by
2979 a non-alphanumeric character, it takes away any special meaning that
2980 character may have. This use of backslash as an escape character
2981 applies both inside and outside character classes.
2982
2983 For example, if you want to match a * character, you write \* in the
2984 pattern. This escaping action applies whether or not the following
2985 character would otherwise be interpreted as a metacharacter, so it is
2986 always safe to precede a non-alphanumeric with backslash to specify
2987 that it stands for itself. In particular, if you want to match a back-
2988 slash, you write \\.
2989
2990 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2991 the pattern (other than in a character class) and characters between a
2992 # outside a character class and the next newline are ignored. An escap-
2993 ing backslash can be used to include a whitespace or # character as
2994 part of the pattern.
2995
2996 If you want to remove the special meaning from a sequence of charac-
2997 ters, you can do so by putting them between \Q and \E. This is differ-
2998 ent from Perl in that $ and @ are handled as literals in \Q...\E
2999 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3000 tion. Note the following examples:
3001
3002 Pattern PCRE matches Perl matches
3003
3004 \Qabc$xyz\E abc$xyz abc followed by the
3005 contents of $xyz
3006 \Qabc\$xyz\E abc\$xyz abc\$xyz
3007 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3008
3009 The \Q...\E sequence is recognized both inside and outside character
3010 classes.
3011
3012 Non-printing characters
3013
3014 A second use of backslash provides a way of encoding non-printing char-
3015 acters in patterns in a visible manner. There is no restriction on the
3016 appearance of non-printing characters, apart from the binary zero that
3017 terminates a pattern, but when a pattern is being prepared by text
3018 editing, it is usually easier to use one of the following escape
3019 sequences than the binary character it represents:
3020
3021 \a alarm, that is, the BEL character (hex 07)
3022 \cx "control-x", where x is any character
3023 \e escape (hex 1B)
3024 \f formfeed (hex 0C)
3025 \n linefeed (hex 0A)
3026 \r carriage return (hex 0D)
3027 \t tab (hex 09)
3028 \ddd character with octal code ddd, or backreference
3029 \xhh character with hex code hh
3030 \x{hhh..} character with hex code hhh..
3031
3032 The precise effect of \cx is as follows: if x is a lower case letter,
3033 it is converted to upper case. Then bit 6 of the character (hex 40) is
3034 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3035 becomes hex 7B.
3036
3037 After \x, from zero to two hexadecimal digits are read (letters can be
3038 in upper or lower case). Any number of hexadecimal digits may appear
3039 between \x{ and }, but the value of the character code must be less
3040 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3041 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3042 than the largest Unicode code point, which is 10FFFF.
3043
3044 If characters other than hexadecimal digits appear between \x{ and },
3045 or if there is no terminating }, this form of escape is not recognized.
3046 Instead, the initial \x will be interpreted as a basic hexadecimal
3047 escape, with no following digits, giving a character whose value is
3048 zero.
3049
3050 Characters whose value is less than 256 can be defined by either of the
3051 two syntaxes for \x. There is no difference in the way they are han-
3052 dled. For example, \xdc is exactly the same as \x{dc}.
3053
3054 After \0 up to two further octal digits are read. If there are fewer
3055 than two digits, just those that are present are used. Thus the
3056 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3057 (code value 7). Make sure you supply two digits after the initial zero
3058 if the pattern character that follows is itself an octal digit.
3059
3060 The handling of a backslash followed by a digit other than 0 is compli-
3061 cated. Outside a character class, PCRE reads it and any following dig-
3062 its as a decimal number. If the number is less than 10, or if there
3063 have been at least that many previous capturing left parentheses in the
3064 expression, the entire sequence is taken as a back reference. A
3065 description of how this works is given later, following the discussion
3066 of parenthesized subpatterns.
3067
3068 Inside a character class, or if the decimal number is greater than 9
3069 and there have not been that many capturing subpatterns, PCRE re-reads
3070 up to three octal digits following the backslash, and uses them to gen-
3071 erate a data character. Any subsequent digits stand for themselves. In
3072 non-UTF-8 mode, the value of a character specified in octal must be
3073 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3074 example:
3075
3076 \040 is another way of writing a space
3077 \40 is the same, provided there are fewer than 40
3078 previous capturing subpatterns
3079 \7 is always a back reference
3080 \11 might be a back reference, or another way of
3081 writing a tab
3082 \011 is always a tab
3083 \0113 is a tab followed by the character "3"
3084 \113 might be a back reference, otherwise the
3085 character with octal code 113
3086 \377 might be a back reference, otherwise
3087 the byte consisting entirely of 1 bits
3088 \81 is either a back reference, or a binary zero
3089 followed by the two characters "8" and "1"
3090
3091 Note that octal values of 100 or greater must not be introduced by a
3092 leading zero, because no more than three octal digits are ever read.
3093
3094 All the sequences that define a single character value can be used both
3095 inside and outside character classes. In addition, inside a character
3096 class, the sequence \b is interpreted as the backspace character (hex
3097 08), and the sequences \R and \X are interpreted as the characters "R"
3098 and "X", respectively. Outside a character class, these sequences have
3099 different meanings (see below).
3100
3101 Absolute and relative back references
3102
3103 The sequence \g followed by an unsigned or a negative number, option-
3104 ally enclosed in braces, is an absolute or relative back reference. A
3105 named back reference can be coded as \g{name}. Back references are dis-
3106 cussed later, following the discussion of parenthesized subpatterns.
3107
3108 Generic character types
3109
3110 Another use of backslash is for specifying generic character types. The
3111 following are always recognized:
3112
3113 \d any decimal digit
3114 \D any character that is not a decimal digit
3115 \h any horizontal whitespace character
3116 \H any character that is not a horizontal whitespace character
3117 \s any whitespace character
3118 \S any character that is not a whitespace character
3119 \v any vertical whitespace character
3120 \V any character that is not a vertical whitespace character
3121 \w any "word" character
3122 \W any "non-word" character
3123
3124 Each pair of escape sequences partitions the complete set of characters
3125 into two disjoint sets. Any given character matches one, and only one,
3126 of each pair.
3127
3128 These character type sequences can appear both inside and outside char-
3129 acter classes. They each match one character of the appropriate type.
3130 If the current matching point is at the end of the subject string, all
3131 of them fail, since there is no character to match.
3132
3133 For compatibility with Perl, \s does not match the VT character (code
3134 11). This makes it different from the the POSIX "space" class. The \s
3135 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3136 "use locale;" is included in a Perl script, \s may match the VT charac-
3137 ter. In PCRE, it never does.
3138
3139 In UTF-8 mode, characters with values greater than 128 never match \d,
3140 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3141 code character property support is available. These sequences retain
3142 their original meanings from before UTF-8 support was available, mainly
3143 for efficiency reasons.
3144
3145 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3146 the other sequences, these do match certain high-valued codepoints in
3147 UTF-8 mode. The horizontal space characters are:
3148
3149 U+0009 Horizontal tab
3150 U+0020 Space
3151 U+00A0 Non-break space
3152 U+1680 Ogham space mark
3153 U+180E Mongolian vowel separator
3154 U+2000 En quad
3155 U+2001 Em quad
3156 U+2002 En space
3157 U+2003 Em space
3158 U+2004 Three-per-em space
3159 U+2005 Four-per-em space
3160 U+2006 Six-per-em space
3161 U+2007 Figure space
3162 U+2008 Punctuation space
3163 U+2009 Thin space
3164 U+200A Hair space
3165 U+202F Narrow no-break space
3166 U+205F Medium mathematical space
3167 U+3000 Ideographic space
3168
3169 The vertical space characters are:
3170
3171 U+000A Linefeed
3172 U+000B Vertical tab
3173 U+000C Formfeed
3174 U+000D Carriage return
3175 U+0085 Next line
3176 U+2028 Line separator
3177 U+2029 Paragraph separator
3178
3179 A "word" character is an underscore or any character less than 256 that
3180 is a letter or digit. The definition of letters and digits is con-
3181 trolled by PCRE's low-valued character tables, and may vary if locale-
3182 specific matching is taking place (see "Locale support" in the pcreapi
3183 page). For example, in a French locale such as "fr_FR" in Unix-like
3184 systems, or "french" in Windows, some character codes greater than 128
3185 are used for accented letters, and these are matched by \w. The use of
3186 locales with Unicode is discouraged.
3187
3188 Newline sequences
3189
3190 Outside a character class, by default, the escape sequence \R matches
3191 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3192 mode \R is equivalent to the following:
3193
3194 (?>\r\n|\n|\x0b|\f|\r|\x85)
3195
3196 This is an example of an "atomic group", details of which are given
3197 below. This particular group matches either the two-character sequence
3198 CR followed by LF, or one of the single characters LF (linefeed,
3199 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3200 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3201 is treated as a single unit that cannot be split.
3202
3203 In UTF-8 mode, two additional characters whose codepoints are greater
3204 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3205 rator, U+2029). Unicode character property support is not needed for
3206 these characters to be recognized.
3207
3208 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3209 the complete set of Unicode line endings) by setting the option
3210 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3211 (BSR is an abbrevation for "backslash R".) This can be made the default
3212 when PCRE is built; if this is the case, the other behaviour can be
3213 requested via the PCRE_BSR_UNICODE option. It is also possible to
3214 specify these settings by starting a pattern string with one of the
3215 following sequences:
3216
3217 (*BSR_ANYCRLF) CR, LF, or CRLF only
3218 (*BSR_UNICODE) any Unicode newline sequence
3219
3220 These override the default and the options given to pcre_compile(), but
3221 they can be overridden by options given to pcre_exec(). Note that these
3222 special settings, which are not Perl-compatible, are recognized only at
3223 the very start of a pattern, and that they must be in upper case. If
3224 more than one of them is present, the last one is used. They can be
3225 combined with a change of newline convention, for example, a pattern
3226 can start with:
3227
3228 (*ANY)(*BSR_ANYCRLF)
3229
3230 Inside a character class, \R matches the letter "R".
3231
3232 Unicode character properties
3233
3234 When PCRE is built with Unicode character property support, three addi-
3235 tional escape sequences that match characters with specific properties
3236 are available. When not in UTF-8 mode, these sequences are of course
3237 limited to testing characters whose codepoints are less than 256, but
3238 they do work in this mode. The extra escape sequences are:
3239
3240 \p{xx} a character with the xx property
3241 \P{xx} a character without the xx property
3242 \X an extended Unicode sequence
3243
3244 The property names represented by xx above are limited to the Unicode
3245 script names, the general category properties, and "Any", which matches
3246 any character (including newline). Other properties such as "InMusical-
3247 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3248 not match any characters, so always causes a match failure.
3249
3250 Sets of Unicode characters are defined as belonging to certain scripts.
3251 A character from one of these sets can be matched using a script name.
3252 For example:
3253
3254 \p{Greek}
3255 \P{Han}
3256
3257 Those that are not part of an identified script are lumped together as
3258 "Common". The current list of scripts is:
3259
3260 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3261 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3262 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3263 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3264 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3265 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3266 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3267 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3268 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3269
3270 Each character has exactly one general category property, specified by
3271 a two-letter abbreviation. For compatibility with Perl, negation can be
3272 specified by including a circumflex between the opening brace and the
3273 property name. For example, \p{^Lu} is the same as \P{Lu}.
3274
3275 If only one letter is specified with \p or \P, it includes all the gen-
3276 eral category properties that start with that letter. In this case, in
3277 the absence of negation, the curly brackets in the escape sequence are
3278 optional; these two examples have the same effect:
3279
3280 \p{L}
3281 \pL
3282
3283 The following general category property codes are supported:
3284
3285 C Other
3286 Cc Control
3287 Cf Format
3288 Cn Unassigned
3289 Co Private use
3290 Cs Surrogate
3291
3292 L Letter
3293 Ll Lower case letter
3294 Lm Modifier letter
3295 Lo Other letter
3296 Lt Title case letter
3297 Lu Upper case letter
3298
3299 M Mark
3300 Mc Spacing mark
3301 Me Enclosing mark
3302 Mn Non-spacing mark
3303
3304 N Number
3305 Nd Decimal number
3306 Nl Letter number
3307 No Other number
3308
3309 P Punctuation
3310 Pc Connector punctuation
3311 Pd Dash punctuation
3312 Pe Close punctuation
3313 Pf Final punctuation
3314 Pi Initial punctuation
3315 Po Other punctuation
3316 Ps Open punctuation
3317
3318 S Symbol
3319 Sc Currency symbol
3320 Sk Modifier symbol
3321 Sm Mathematical symbol
3322 So Other symbol
3323
3324 Z Separator
3325 Zl Line separator
3326 Zp Paragraph separator
3327 Zs Space separator
3328
3329 The special property L& is also supported: it matches a character that
3330 has the Lu, Ll, or Lt property, in other words, a letter that is not
3331 classified as a modifier or "other".
3332
3333 The Cs (Surrogate) property applies only to characters in the range
3334 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3335 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3336 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3337 the pcreapi page).
3338
3339 The long synonyms for these properties that Perl supports (such as
3340 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3341 any of these properties with "Is".
3342
3343 No character that is in the Unicode table has the Cn (unassigned) prop-
3344 erty. Instead, this property is assumed for any code point that is not
3345 in the Unicode table.
3346
3347 Specifying caseless matching does not affect these escape sequences.
3348 For example, \p{Lu} always matches only upper case letters.
3349
3350 The \X escape matches any number of Unicode characters that form an
3351 extended Unicode sequence. \X is equivalent to
3352
3353 (?>\PM\pM*)
3354
3355 That is, it matches a character without the "mark" property, followed
3356 by zero or more characters with the "mark" property, and treats the
3357 sequence as an atomic group (see below). Characters with the "mark"
3358 property are typically accents that affect the preceding character.
3359 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3360 matches any one character.
3361
3362 Matching characters by Unicode property is not fast, because PCRE has
3363 to search a structure that contains data for over fifteen thousand
3364 characters. That is why the traditional escape sequences such as \d and
3365 \w do not use Unicode properties in PCRE.
3366
3367 Resetting the match start
3368
3369 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3370 ously matched characters not to be included in the final matched
3371 sequence. For example, the pattern:
3372
3373 foo\Kbar
3374
3375 matches "foobar", but reports that it has matched "bar". This feature
3376 is similar to a lookbehind assertion (described below). However, in
3377 this case, the part of the subject before the real match does not have
3378 to be of fixed length, as lookbehind assertions do. The use of \K does
3379 not interfere with the setting of captured substrings. For example,
3380 when the pattern
3381
3382 (foo)\Kbar
3383
3384 matches "foobar", the first substring is still set to "foo".
3385
3386 Simple assertions
3387
3388 The final use of backslash is for certain simple assertions. An asser-
3389 tion specifies a condition that has to be met at a particular point in
3390 a match, without consuming any characters from the subject string. The
3391 use of subpatterns for more complicated assertions is described below.
3392 The backslashed assertions are:
3393
3394 \b matches at a word boundary
3395 \B matches when not at a word boundary
3396 \A matches at the start of the subject
3397 \Z matches at the end of the subject
3398 also matches before a newline at the end of the subject
3399 \z matches only at the end of the subject
3400 \G matches at the first matching position in the subject
3401
3402 These assertions may not appear in character classes (but note that \b
3403 has a different meaning, namely the backspace character, inside a char-
3404 acter class).
3405
3406 A word boundary is a position in the subject string where the current
3407 character and the previous character do not both match \w or \W (i.e.
3408 one matches \w and the other matches \W), or the start or end of the
3409 string if the first or last character matches \w, respectively.
3410
3411 The \A, \Z, and \z assertions differ from the traditional circumflex
3412 and dollar (described in the next section) in that they only ever match
3413 at the very start and end of the subject string, whatever options are
3414 set. Thus, they are independent of multiline mode. These three asser-
3415 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3416 affect only the behaviour of the circumflex and dollar metacharacters.
3417 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3418 cating that matching is to start at a point other than the beginning of
3419 the subject, \A can never match. The difference between \Z and \z is
3420 that \Z matches before a newline at the end of the string as well as at
3421 the very end, whereas \z matches only at the end.
3422
3423 The \G assertion is true only when the current matching position is at
3424 the start point of the match, as specified by the startoffset argument
3425 of pcre_exec(). It differs from \A when the value of startoffset is
3426 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3427 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3428 mentation where \G can be useful.
3429
3430 Note, however, that PCRE's interpretation of \G, as the start of the
3431 current match, is subtly different from Perl's, which defines it as the
3432 end of the previous match. In Perl, these can be different when the
3433 previously matched string was empty. Because PCRE does just one match
3434 at a time, it cannot reproduce this behaviour.
3435
3436 If all the alternatives of a pattern begin with \G, the expression is
3437 anchored to the starting match position, and the "anchored" flag is set
3438 in the compiled regular expression.
3439
3440
3441 CIRCUMFLEX AND DOLLAR
3442
3443 Outside a character class, in the default matching mode, the circumflex
3444 character is an assertion that is true only if the current matching
3445 point is at the start of the subject string. If the startoffset argu-
3446 ment of pcre_exec() is non-zero, circumflex can never match if the
3447 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3448 has an entirely different meaning (see below).
3449
3450 Circumflex need not be the first character of the pattern if a number
3451 of alternatives are involved, but it should be the first thing in each
3452 alternative in which it appears if the pattern is ever to match that
3453 branch. If all possible alternatives start with a circumflex, that is,
3454 if the pattern is constrained to match only at the start of the sub-
3455 ject, it is said to be an "anchored" pattern. (There are also other
3456 constructs that can cause a pattern to be anchored.)
3457
3458 A dollar character is an assertion that is true only if the current
3459 matching point is at the end of the subject string, or immediately
3460 before a newline at the end of the string (by default). Dollar need not
3461 be the last character of the pattern if a number of alternatives are
3462 involved, but it should be the last item in any branch in which it
3463 appears. Dollar has no special meaning in a character class.
3464
3465 The meaning of dollar can be changed so that it matches only at the
3466 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3467 compile time. This does not affect the \Z assertion.
3468
3469 The meanings of the circumflex and dollar characters are changed if the
3470 PCRE_MULTILINE option is set. When this is the case, a circumflex
3471 matches immediately after internal newlines as well as at the start of
3472 the subject string. It does not match after a newline that ends the
3473 string. A dollar matches before any newlines in the string, as well as
3474 at the very end, when PCRE_MULTILINE is set. When newline is specified
3475 as the two-character sequence CRLF, isolated CR and LF characters do
3476 not indicate newlines.
3477
3478 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3479 (where \n represents a newline) in multiline mode, but not otherwise.
3480 Consequently, patterns that are anchored in single line mode because
3481 all branches start with ^ are not anchored in multiline mode, and a
3482 match for circumflex is possible when the startoffset argument of
3483 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3484 PCRE_MULTILINE is set.
3485
3486 Note that the sequences \A, \Z, and \z can be used to match the start
3487 and end of the subject in both modes, and if all branches of a pattern
3488 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3489 set.
3490
3491
3492 FULL STOP (PERIOD, DOT)
3493
3494 Outside a character class, a dot in the pattern matches any one charac-
3495 ter in the subject string except (by default) a character that signi-
3496 fies the end of a line. In UTF-8 mode, the matched character may be
3497 more than one byte long.
3498
3499 When a line ending is defined as a single character, dot never matches
3500 that character; when the two-character sequence CRLF is used, dot does
3501 not match CR if it is immediately followed by LF, but otherwise it
3502 matches all characters (including isolated CRs and LFs). When any Uni-
3503 code line endings are being recognized, dot does not match CR or LF or
3504 any of the other line ending characters.
3505
3506 The behaviour of dot with regard to newlines can be changed. If the
3507 PCRE_DOTALL option is set, a dot matches any one character, without
3508 exception. If the two-character sequence CRLF is present in the subject
3509 string, it takes two dots to match it.
3510
3511 The handling of dot is entirely independent of the handling of circum-
3512 flex and dollar, the only relationship being that they both involve
3513 newlines. Dot has no special meaning in a character class.
3514
3515
3516 MATCHING A SINGLE BYTE
3517
3518 Outside a character class, the escape sequence \C matches any one byte,
3519 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3520 line-ending characters. The feature is provided in Perl in order to
3521 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3522 acters into individual bytes, what remains in the string may be a mal-
3523 formed UTF-8 string. For this reason, the \C escape sequence is best
3524 avoided.
3525
3526 PCRE does not allow \C to appear in lookbehind assertions (described
3527 below), because in UTF-8 mode this would make it impossible to calcu-
3528 late the length of the lookbehind.
3529
3530
3531 SQUARE BRACKETS AND CHARACTER CLASSES
3532
3533 An opening square bracket introduces a character class, terminated by a
3534 closing square bracket. A closing square bracket on its own is not spe-
3535 cial. If a closing square bracket is required as a member of the class,
3536 it should be the first data character in the class (after an initial
3537 circumflex, if present) or escaped with a backslash.
3538
3539 A character class matches a single character in the subject. In UTF-8
3540 mode, the character may occupy more than one byte. A matched character
3541 must be in the set of characters defined by the class, unless the first
3542 character in the class definition is a circumflex, in which case the
3543 subject character must not be in the set defined by the class. If a
3544 circumflex is actually required as a member of the class, ensure it is
3545 not the first character, or escape it with a backslash.
3546
3547 For example, the character class [aeiou] matches any lower case vowel,
3548 while [^aeiou] matches any character that is not a lower case vowel.
3549 Note that a circumflex is just a convenient notation for specifying the
3550 characters that are in the class by enumerating those that are not. A
3551 class that starts with a circumflex is not an assertion: it still con-
3552 sumes a character from the subject string, and therefore it fails if
3553 the current pointer is at the end of the string.
3554
3555 In UTF-8 mode, characters with values greater than 255 can be included
3556 in a class as a literal string of bytes, or by using the \x{ escaping
3557 mechanism.
3558
3559 When caseless matching is set, any letters in a class represent both
3560 their upper case and lower case versions, so for example, a caseless
3561 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3562 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3563 understands the concept of case for characters whose values are less
3564 than 128, so caseless matching is always possible. For characters with
3565 higher values, the concept of case is supported if PCRE is compiled
3566 with Unicode property support, but not otherwise. If you want to use
3567 caseless matching for characters 128 and above, you must ensure that
3568 PCRE is compiled with Unicode property support as well as with UTF-8
3569 support.
3570
3571 Characters that might indicate line breaks are never treated in any
3572 special way when matching character classes, whatever line-ending
3573 sequence is in use, and whatever setting of the PCRE_DOTALL and
3574 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3575 of these characters.
3576
3577 The minus (hyphen) character can be used to specify a range of charac-
3578 ters in a character class. For example, [d-m] matches any letter
3579 between d and m, inclusive. If a minus character is required in a
3580 class, it must be escaped with a backslash or appear in a position
3581 where it cannot be interpreted as indicating a range, typically as the
3582 first or last character in the class.
3583
3584 It is not possible to have the literal character "]" as the end charac-
3585 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3586 two characters ("W" and "-") followed by a literal string "46]", so it
3587 would match "W46]" or "-46]". However, if the "]" is escaped with a
3588 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3589 preted as a class containing a range followed by two other characters.
3590 The octal or hexadecimal representation of "]" can also be used to end
3591 a range.
3592
3593 Ranges operate in the collating sequence of character values. They can
3594 also be used for characters specified numerically, for example
3595 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3596 are greater than 255, for example [\x{100}-\x{2ff}].
3597
3598 If a range that includes letters is used when caseless matching is set,
3599 it matches the letters in either case. For example, [W-c] is equivalent
3600 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3601 character tables for a French locale are in use, [\xc8-\xcb] matches
3602 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3603 concept of case for characters with values greater than 128 only when
3604 it is compiled with Unicode property support.
3605
3606 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3607 in a character class, and add the characters that they match to the
3608 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3609 flex can conveniently be used with the upper case character types to
3610 specify a more restricted set of characters than the matching lower
3611 case type. For example, the class [^\W_] matches any letter or digit,
3612 but not underscore.
3613
3614 The only metacharacters that are recognized in character classes are
3615 backslash, hyphen (only where it can be interpreted as specifying a
3616 range), circumflex (only at the start), opening square bracket (only
3617 when it can be interpreted as introducing a POSIX class name - see the
3618 next section), and the terminating closing square bracket. However,
3619 escaping other non-alphanumeric characters does no harm.
3620
3621
3622 POSIX CHARACTER CLASSES
3623
3624 Perl supports the POSIX notation for character classes. This uses names
3625 enclosed by [: and :] within the enclosing square brackets. PCRE also
3626 supports this notation. For example,
3627
3628 [01[:alpha:]%]
3629
3630 matches "0", "1", any alphabetic character, or "%". The supported class
3631 names are
3632
3633 alnum letters and digits
3634 alpha letters
3635 ascii character codes 0 - 127
3636 blank space or tab only
3637 cntrl control characters
3638 digit decimal digits (same as \d)
3639 graph printing characters, excluding space
3640 lower lower case letters
3641 print printing characters, including space
3642 punct printing characters, excluding letters and digits
3643 space white space (not quite the same as \s)
3644 upper upper case letters
3645 word "word" characters (same as \w)
3646 xdigit hexadecimal digits
3647
3648 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3649 and space (32). Notice that this list includes the VT character (code
3650 11). This makes "space" different to \s, which does not include VT (for
3651 Perl compatibility).
3652
3653 The name "word" is a Perl extension, and "blank" is a GNU extension
3654 from Perl 5.8. Another Perl extension is negation, which is indicated
3655 by a ^ character after the colon. For example,
3656
3657 [12[:^digit:]]
3658
3659 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3660 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3661 these are not supported, and an error is given if they are encountered.
3662
3663 In UTF-8 mode, characters with values greater than 128 do not match any
3664 of the POSIX character classes.
3665
3666
3667 VERTICAL BAR
3668
3669 Vertical bar characters are used to separate alternative patterns. For
3670 example, the pattern
3671
3672 gilbert|sullivan
3673
3674 matches either "gilbert" or "sullivan". Any number of alternatives may
3675 appear, and an empty alternative is permitted (matching the empty
3676 string). The matching process tries each alternative in turn, from left
3677 to right, and the first one that succeeds is used. If the alternatives
3678 are within a subpattern (defined below), "succeeds" means matching the
3679 rest of the main pattern as well as the alternative in the subpattern.
3680
3681
3682 INTERNAL OPTION SETTING
3683
3684 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3685 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3686 within the pattern by a sequence of Perl option letters enclosed
3687 between "(?" and ")". The option letters are
3688
3689 i for PCRE_CASELESS
3690 m for PCRE_MULTILINE
3691 s for PCRE_DOTALL
3692 x for PCRE_EXTENDED
3693
3694 For example, (?im) sets caseless, multiline matching. It is also possi-
3695 ble to unset these options by preceding the letter with a hyphen, and a
3696 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3697 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3698 is also permitted. If a letter appears both before and after the
3699 hyphen, the option is unset.
3700
3701 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3702 can be changed in the same way as the Perl-compatible options by using
3703 the characters J, U and X respectively.
3704
3705 When an option change occurs at top level (that is, not inside subpat-
3706 tern parentheses), the change applies to the remainder of the pattern
3707 that follows. If the change is placed right at the start of a pattern,
3708 PCRE extracts it into the global options (and it will therefore show up
3709 in data extracted by the pcre_fullinfo() function).
3710
3711 An option change within a subpattern (see below for a description of
3712 subpatterns) affects only that part of the current pattern that follows
3713 it, so
3714
3715 (a(?i)b)c
3716
3717 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3718 used). By this means, options can be made to have different settings
3719 in different parts of the pattern. Any changes made in one alternative
3720 do carry on into subsequent branches within the same subpattern. For
3721 example,
3722
3723 (a(?i)b|c)
3724
3725 matches "ab", "aB", "c", and "C", even though when matching "C" the
3726 first branch is abandoned before the option setting. This is because
3727 the effects of option settings happen at compile time. There would be
3728 some very weird behaviour otherwise.
3729
3730
3731 SUBPATTERNS
3732
3733 Subpatterns are delimited by parentheses (round brackets), which can be
3734 nested. Turning part of a pattern into a subpattern does two things:
3735
3736 1. It localizes a set of alternatives. For example, the pattern
3737
3738 cat(aract|erpillar|)
3739
3740 matches one of the words "cat", "cataract", or "caterpillar". Without
3741 the parentheses, it would match "cataract", "erpillar" or an empty
3742 string.
3743
3744 2. It sets up the subpattern as a capturing subpattern. This means
3745 that, when the whole pattern matches, that portion of the subject
3746 string that matched the subpattern is passed back to the caller via the
3747 ovector argument of pcre_exec(). Opening parentheses are counted from
3748 left to right (starting from 1) to obtain numbers for the capturing
3749 subpatterns.
3750
3751 For example, if the string "the red king" is matched against the pat-
3752 tern
3753
3754 the ((red|white) (king|queen))
3755
3756 the captured substrings are "red king", "red", and "king", and are num-
3757 bered 1, 2, and 3, respectively.
3758
3759 The fact that plain parentheses fulfil two functions is not always
3760 helpful. There are often times when a grouping subpattern is required
3761 without a capturing requirement. If an opening parenthesis is followed
3762 by a question mark and a colon, the subpattern does not do any captur-
3763 ing, and is not counted when computing the number of any subsequent
3764 capturing subpatterns. For example, if the string "the white queen" is
3765 matched against the pattern
3766
3767 the ((?:red|white) (king|queen))
3768
3769 the captured substrings are "white queen" and "queen", and are numbered
3770 1 and 2. The maximum number of capturing subpatterns is 65535.
3771
3772 As a convenient shorthand, if any option settings are required at the
3773 start of a non-capturing subpattern, the option letters may appear
3774 between the "?" and the ":". Thus the two patterns
3775
3776 (?i:saturday|sunday)
3777 (?:(?i)saturday|sunday)
3778
3779 match exactly the same set of strings. Because alternative branches are
3780 tried from left to right, and options are not reset until the end of
3781 the subpattern is reached, an option setting in one branch does affect
3782 subsequent branches, so the above patterns match "SUNDAY" as well as
3783 "Saturday".
3784
3785
3786 DUPLICATE SUBPATTERN NUMBERS
3787
3788 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3789 uses the same numbers for its capturing parentheses. Such a subpattern
3790 starts with (?| and is itself a non-capturing subpattern. For example,
3791 consider this pattern:
3792
3793 (?|(Sat)ur|(Sun))day
3794
3795 Because the two alternatives are inside a (?| group, both sets of cap-
3796 turing parentheses are numbered one. Thus, when the pattern matches,
3797 you can look at captured substring number one, whichever alternative
3798 matched. This construct is useful when you want to capture part, but
3799 not all, of one of a number of alternatives. Inside a (?| group, paren-
3800 theses are numbered as usual, but the number is reset at the start of
3801 each branch. The numbers of any capturing buffers that follow the sub-
3802 pattern start after the highest number used in any branch. The follow-
3803 ing example is taken from the Perl documentation. The numbers under-
3804 neath show in which buffer the captured content will be stored.
3805
3806 # before ---------------branch-reset----------- after
3807 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3808 # 1 2 2 3 2 3 4
3809
3810 A backreference or a recursive call to a numbered subpattern always
3811 refers to the first one in the pattern with the given number.
3812
3813 An alternative approach to using this "branch reset" feature is to use
3814 duplicate named subpatterns, as described in the next section.
3815
3816
3817 NAMED SUBPATTERNS
3818
3819 Identifying capturing parentheses by number is simple, but it can be
3820 very hard to keep track of the numbers in complicated regular expres-
3821 sions. Furthermore, if an expression is modified, the numbers may
3822 change. To help with this difficulty, PCRE supports the naming of sub-
3823 patterns. This feature was not added to Perl until release 5.10. Python
3824 had the feature earlier, and PCRE introduced it at release 4.0, using
3825 the Python syntax. PCRE now supports both the Perl and the Python syn-
3826 tax.
3827
3828 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3829 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3830 to capturing parentheses from other parts of the pattern, such as back-
3831 references, recursion, and conditions, can be made by name as well as
3832 by number.
3833
3834 Names consist of up to 32 alphanumeric characters and underscores.
3835 Named capturing parentheses are still allocated numbers as well as
3836 names, exactly as if the names were not present. The PCRE API provides
3837 function calls for extracting the name-to-number translation table from
3838 a compiled pattern. There is also a convenience function for extracting
3839 a captured substring by name.
3840
3841 By default, a name must be unique within a pattern, but it is possible
3842 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3843 time. This can be useful for patterns where only one instance of the
3844 named parentheses can match. Suppose you want to match the name of a
3845 weekday, either as a 3-letter abbreviation or as the full name, and in
3846 both cases you want to extract the abbreviation. This pattern (ignoring
3847 the line breaks) does the job:
3848
3849 (?<DN>Mon|Fri|Sun)(?:day)?|
3850 (?<DN>Tue)(?:sday)?|
3851 (?<DN>Wed)(?:nesday)?|
3852 (?<DN>Thu)(?:rsday)?|
3853 (?<DN>Sat)(?:urday)?
3854
3855 There are five capturing substrings, but only one is ever set after a
3856 match. (An alternative way of solving this problem is to use a "branch
3857 reset" subpattern, as described in the previous section.)
3858
3859 The convenience function for extracting the data by name returns the
3860 substring for the first (and in this example, the only) subpattern of
3861 that name that matched. This saves searching to find which numbered
3862 subpattern it was. If you make a reference to a non-unique named sub-
3863 pattern from elsewhere in the pattern, the one that corresponds to the
3864 lowest number is used. For further details of the interfaces for han-
3865 dling named subpatterns, see the pcreapi documentation.
3866
3867
3868 REPETITION
3869
3870 Repetition is specified by quantifiers, which can follow any of the
3871 following items:
3872
3873 a literal data character
3874 the dot metacharacter
3875 the \C escape sequence
3876 the \X escape sequence (in UTF-8 mode with Unicode properties)
3877 the \R escape sequence
3878 an escape such as \d that matches a single character
3879 a character class
3880 a back reference (see next section)
3881 a parenthesized subpattern (unless it is an assertion)
3882
3883 The general repetition quantifier specifies a minimum and maximum num-
3884 ber of permitted matches, by giving the two numbers in curly brackets
3885 (braces), separated by a comma. The numbers must be less than 65536,
3886 and the first must be less than or equal to the second. For example:
3887
3888 z{2,4}
3889
3890 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3891 special character. If the second number is omitted, but the comma is
3892 present, there is no upper limit; if the second number and the comma
3893 are both omitted, the quantifier specifies an exact number of required
3894 matches. Thus
3895
3896 [aeiou]{3,}
3897
3898 matches at least 3 successive vowels, but may match many more, while
3899
3900 \d{8}
3901
3902 matches exactly 8 digits. An opening curly bracket that appears in a
3903 position where a quantifier is not allowed, or one that does not match
3904 the syntax of a quantifier, is taken as a literal character. For exam-
3905 ple, {,6} is not a quantifier, but a literal string of four characters.
3906
3907 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3908 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3909 acters, each of which is represented by a two-byte sequence. Similarly,
3910 when Unicode property support is available, \X{3} matches three Unicode
3911 extended sequences, each of which may be several bytes long (and they
3912 may be of different lengths).
3913
3914 The quantifier {0} is permitted, causing the expression to behave as if
3915 the previous item and the quantifier were not present.
3916
3917 For convenience, the three most common quantifiers have single-charac-
3918 ter abbreviations:
3919
3920 * is equivalent to {0,}
3921 + is equivalent to {1,}
3922 ? is equivalent to {0,1}
3923
3924 It is possible to construct infinite loops by following a subpattern
3925 that can match no characters with a quantifier that has no upper limit,
3926 for example:
3927
3928 (a?)*
3929
3930 Earlier versions of Perl and PCRE used to give an error at compile time
3931 for such patterns. However, because there are cases where this can be
3932 useful, such patterns are now accepted, but if any repetition of the
3933 subpattern does in fact match no characters, the loop is forcibly bro-
3934 ken.
3935
3936 By default, the quantifiers are "greedy", that is, they match as much
3937 as possible (up to the maximum number of permitted times), without
3938 causing the rest of the pattern to fail. The classic example of where
3939 this gives problems is in trying to match comments in C programs. These
3940 appear between /* and */ and within the comment, individual * and /
3941 characters may appear. An attempt to match C comments by applying the
3942 pattern
3943
3944 /\*.*\*/
3945
3946 to the string
3947
3948 /* first comment */ not comment /* second comment */
3949
3950 fails, because it matches the entire string owing to the greediness of
3951 the .* item.
3952
3953 However, if a quantifier is followed by a question mark, it ceases to
3954 be greedy, and instead matches the minimum number of times possible, so
3955 the pattern
3956
3957 /\*.*?\*/
3958
3959 does the right thing with the C comments. The meaning of the various
3960 quantifiers is not otherwise changed, just the preferred number of
3961 matches. Do not confuse this use of question mark with its use as a
3962 quantifier in its own right. Because it has two uses, it can sometimes
3963 appear doubled, as in
3964
3965 \d??\d
3966
3967 which matches one digit by preference, but can match two if that is the
3968 only way the rest of the pattern matches.
3969
3970 If the PCRE_UNGREEDY option is set (an option that is not available in
3971 Perl), the quantifiers are not greedy by default, but individual ones
3972 can be made greedy by following them with a question mark. In other
3973 words, it inverts the default behaviour.
3974
3975 When a parenthesized subpattern is quantified with a minimum repeat
3976 count that is greater than 1 or with a limited maximum, more memory is
3977 required for the compiled pattern, in proportion to the size of the
3978 minimum or maximum.
3979
3980 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3981 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3982 the pattern is implicitly anchored, because whatever follows will be
3983 tried against every character position in the subject string, so there
3984 is no point in retrying the overall match at any position after the
3985 first. PCRE normally treats such a pattern as though it were preceded
3986 by \A.
3987
3988 In cases where it is known that the subject string contains no new-
3989 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3990 mization, or alternatively using ^ to indicate anchoring explicitly.
3991
3992 However, there is one situation where the optimization cannot be used.
3993 When .* is inside capturing parentheses that are the subject of a
3994 backreference elsewhere in the pattern, a match at the start may fail
3995 where a later one succeeds. Consider, for example:
3996
3997 (.*)abc\1
3998
3999 If the subject is "xyz123abc123" the match point is the fourth charac-
4000 ter. For this reason, such a pattern is not implicitly anchored.
4001
4002 When a capturing subpattern is repeated, the value captured is the sub-
4003 string that matched the final iteration. For example, after
4004
4005 (tweedle[dume]{3}\s*)+
4006
4007 has matched "tweedledum tweedledee" the value of the captured substring
4008 is "tweedledee". However, if there are nested capturing subpatterns,
4009 the corresponding captured values may have been set in previous itera-
4010 tions. For example, after
4011
4012 /(a|(b))+/
4013
4014 matches "aba" the value of the second captured substring is "b".
4015
4016
4017 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4018
4019 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4020 repetition, failure of what follows normally causes the repeated item
4021 to be re-evaluated to see if a different number of repeats allows the
4022 rest of the pattern to match. Sometimes it is useful to prevent this,
4023 either to change the nature of the match, or to cause it fail earlier
4024 than it otherwise might, when the author of the pattern knows there is
4025 no point in carrying on.
4026
4027 Consider, for example, the pattern \d+foo when applied to the subject
4028 line
4029
4030 123456bar
4031
4032 After matching all 6 digits and then failing to match "foo", the normal
4033 action of the matcher is to try again with only 5 digits matching the
4034 \d+ item, and then with 4, and so on, before ultimately failing.
4035 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4036 the means for specifying that once a subpattern has matched, it is not
4037 to be re-evaluated in this way.
4038
4039 If we use atomic grouping for the previous example, the matcher gives
4040 up immediately on failing to match "foo" the first time. The notation
4041 is a kind of special parenthesis, starting with (?> as in this example:
4042
4043 (?>\d+)foo
4044
4045 This kind of parenthesis "locks up" the part of the pattern it con-
4046 tains once it has matched, and a failure further into the pattern is
4047 prevented from backtracking into it. Backtracking past it to previous
4048 items, however, works as normal.
4049
4050 An alternative description is that a subpattern of this type matches
4051 the string of characters that an identical standalone pattern would
4052 match, if anchored at the current point in the subject string.
4053
4054 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4055 such as the above example can be thought of as a maximizing repeat that
4056 must swallow everything it can. So, while both \d+ and \d+? are pre-
4057 pared to adjust the number of digits they match in order to make the
4058 rest of the pattern match, (?>\d+) can only match an entire sequence of
4059 digits.
4060
4061 Atomic groups in general can of course contain arbitrarily complicated
4062 subpatterns, and can be nested. However, when the subpattern for an
4063 atomic group is just a single repeated item, as in the example above, a
4064 simpler notation, called a "possessive quantifier" can be used. This
4065 consists of an additional + character following a quantifier. Using
4066 this notation, the previous example can be rewritten as
4067
4068 \d++foo
4069
4070 Note that a possessive quantifier can be used with an entire group, for
4071 example:
4072
4073 (abc|xyz){2,3}+
4074
4075 Possessive quantifiers are always greedy; the setting of the
4076 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4077 simpler forms of atomic group. However, there is no difference in the
4078 meaning of a possessive quantifier and the equivalent atomic group,
4079 though there may be a performance difference; possessive quantifiers
4080 should be slightly faster.
4081
4082 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4083 tax. Jeffrey Friedl originated the idea (and the name) in the first
4084 edition of his book. Mike McCloskey liked it, so implemented it when he
4085 built Sun's Java package, and PCRE copied it from there. It ultimately
4086 found its way into Perl at release 5.10.
4087
4088 PCRE has an optimization that automatically "possessifies" certain sim-
4089 ple pattern constructs. For example, the sequence A+B is treated as
4090 A++B because there is no point in backtracking into a sequence of A's
4091 when B must follow.
4092
4093 When a pattern contains an unlimited repeat inside a subpattern that
4094 can itself be repeated an unlimited number of times, the use of an
4095 atomic group is the only way to avoid some failing matches taking a
4096 very long time indeed. The pattern
4097
4098 (\D+|<\d+>)*[!?]
4099
4100 matches an unlimited number of substrings that either consist of non-
4101 digits, or digits enclosed in <>, followed by either ! or ?. When it
4102 matches, it runs quickly. However, if it is applied to
4103
4104 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4105
4106 it takes a long time before reporting failure. This is because the
4107 string can be divided between the internal \D+ repeat and the external
4108 * repeat in a large number of ways, and all have to be tried. (The
4109 example uses [!?] rather than a single character at the end, because
4110 both PCRE and Perl have an optimization that allows for fast failure
4111 when a single character is used. They remember the last single charac-
4112 ter that is required for a match, and fail early if it is not present
4113 in the string.) If the pattern is changed so that it uses an atomic
4114 group, like this:
4115
4116 ((?>\D+)|<\d+>)*[!?]
4117
4118 sequences of non-digits cannot be broken, and failure happens quickly.
4119
4120
4121 BACK REFERENCES
4122
4123 Outside a character class, a backslash followed by a digit greater than
4124 0 (and possibly further digits) is a back reference to a capturing sub-
4125 pattern earlier (that is, to its left) in the pattern, provided there
4126 have been that many previous capturing left parentheses.
4127
4128 However, if the decimal number following the backslash is less than 10,
4129 it is always taken as a back reference, and causes an error only if
4130 there are not that many capturing left parentheses in the entire pat-
4131 tern. In other words, the parentheses that are referenced need not be
4132 to the left of the reference for numbers less than 10. A "forward back
4133 reference" of this type can make sense when a repetition is involved
4134 and the subpattern to the right has participated in an earlier itera-
4135 tion.
4136
4137 It is not possible to have a numerical "forward back reference" to a
4138 subpattern whose number is 10 or more using this syntax because a
4139 sequence such as \50 is interpreted as a character defined in octal.
4140 See the subsection entitled "Non-printing characters" above for further
4141 details of the handling of digits following a backslash. There is no
4142 such problem when named parentheses are used. A back reference to any
4143 subpattern is possible using named parentheses (see below).
4144
4145 Another way of avoiding the ambiguity inherent in the use of digits
4146 following a backslash is to use the \g escape sequence, which is a fea-
4147 ture introduced in Perl 5.10. This escape must be followed by an
4148 unsigned number or a negative number, optionally enclosed in braces.
4149 These examples are all identical:
4150
4151 (ring), \1
4152 (ring), \g1
4153 (ring), \g{1}
4154
4155 An unsigned number specifies an absolute reference without the ambigu-
4156 ity that is present in the older syntax. It is also useful when literal
4157 digits follow the reference. A negative number is a relative reference.
4158 Consider this example:
4159
4160 (abc(def)ghi)\g{-1}
4161
4162 The sequence \g{-1} is a reference to the most recently started captur-
4163 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4164 \g{-2} would be equivalent to \1. The use of relative references can be
4165 helpful in long patterns, and also in patterns that are created by
4166 joining together fragments that contain references within themselves.
4167
4168 A back reference matches whatever actually matched the capturing sub-
4169 pattern in the current subject string, rather than anything matching
4170 the subpattern itself (see "Subpatterns as subroutines" below for a way
4171 of doing that). So the pattern
4172
4173 (sens|respons)e and \1ibility
4174
4175 matches "sense and sensibility" and "response and responsibility", but
4176 not "sense and responsibility". If caseful matching is in force at the
4177 time of the back reference, the case of letters is relevant. For exam-
4178 ple,
4179
4180 ((?i)rah)\s+\1
4181
4182 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4183 original capturing subpattern is matched caselessly.
4184
4185 There are several different ways of writing back references to named
4186 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4187 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4188 unified back reference syntax, in which \g can be used for both numeric
4189 and named references, is also supported. We could rewrite the above
4190 example in any of the following ways:
4191
4192 (?<p1>(?i)rah)\s+\k<p1>
4193 (?'p1'(?i)rah)\s+\k{p1}
4194 (?P<p1>(?i)rah)\s+(?P=p1)
4195 (?<p1>(?i)rah)\s+\g{p1}
4196
4197 A subpattern that is referenced by name may appear in the pattern
4198 before or after the reference.
4199
4200 There may be more than one back reference to the same subpattern. If a
4201 subpattern has not actually been used in a particular match, any back
4202 references to it always fail. For example, the pattern
4203
4204 (a|(bc))\2
4205
4206 always fails if it starts to match "a" rather than "bc". Because there
4207 may be many capturing parentheses in a pattern, all digits following
4208 the backslash are taken as part of a potential back reference number.
4209 If the pattern continues with a digit character, some delimiter must be
4210 used to terminate the back reference. If the PCRE_EXTENDED option is
4211 set, this can be whitespace. Otherwise an empty comment (see "Com-
4212 ments" below) can be used.
4213
4214 A back reference that occurs inside the parentheses to which it refers
4215 fails when the subpattern is first used, so, for example, (a\1) never
4216 matches. However, such references can be useful inside repeated sub-
4217 patterns. For example, the pattern
4218
4219 (a|b\1)+
4220
4221 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4222 ation of the subpattern, the back reference matches the character
4223 string corresponding to the previous iteration. In order for this to
4224 work, the pattern must be such that the first iteration does not need
4225 to match the back reference. This can be done using alternation, as in
4226 the example above, or by a quantifier with a minimum of zero.
4227
4228
4229 ASSERTIONS
4230
4231 An assertion is a test on the characters following or preceding the
4232 current matching point that does not actually consume any characters.
4233 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4234 described above.
4235
4236 More complicated assertions are coded as subpatterns. There are two
4237 kinds: those that look ahead of the current position in the subject
4238 string, and those that look behind it. An assertion subpattern is
4239 matched in the normal way, except that it does not cause the current
4240 matching position to be changed.
4241
4242 Assertion subpatterns are not capturing subpatterns, and may not be
4243 repeated, because it makes no sense to assert the same thing several
4244 times. If any kind of assertion contains capturing subpatterns within
4245 it, these are counted for the purposes of numbering the capturing sub-
4246 patterns in the whole pattern. However, substring capturing is carried
4247 out only for positive assertions, because it does not make sense for
4248 negative assertions.
4249
4250 Lookahead assertions
4251
4252 Lookahead assertions start with (?= for positive assertions and (?! for
4253 negative assertions. For example,
4254
4255 \w+(?=;)
4256
4257 matches a word followed by a semicolon, but does not include the semi-
4258 colon in the match, and
4259
4260 foo(?!bar)
4261
4262 matches any occurrence of "foo" that is not followed by "bar". Note
4263 that the apparently similar pattern
4264
4265 (?!foo)bar
4266
4267 does not find an occurrence of "bar" that is preceded by something
4268 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4269 the assertion (?!foo) is always true when the next three characters are
4270 "bar". A lookbehind assertion is needed to achieve the other effect.
4271
4272 If you want to force a matching failure at some point in a pattern, the
4273 most convenient way to do it is with (?!) because an empty string
4274 always matches, so an assertion that requires there not to be an empty
4275 string must always fail.
4276
4277 Lookbehind assertions
4278
4279 Lookbehind assertions start with (?<= for positive assertions and (?<!
4280 for negative assertions. For example,
4281
4282 (?<!foo)bar
4283
4284 does find an occurrence of "bar" that is not preceded by "foo". The
4285 contents of a lookbehind assertion are restricted such that all the
4286 strings it matches must have a fixed length. However, if there are sev-
4287 eral top-level alternatives, they do not all have to have the same
4288 fixed length. Thus
4289
4290 (?<=bullock|donkey)
4291
4292 is permitted, but
4293
4294 (?<!dogs?|cats?)
4295
4296 causes an error at compile time. Branches that match different length
4297 strings are permitted only at the top level of a lookbehind assertion.
4298 This is an extension compared with Perl (at least for 5.8), which
4299 requires all branches to match the same length of string. An assertion
4300 such as
4301
4302 (?<=ab(c|de))
4303
4304 is not permitted, because its single top-level branch can match two
4305 different lengths, but it is acceptable if rewritten to use two top-
4306 level branches:
4307
4308 (?<=abc|abde)
4309
4310 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4311 instead of a lookbehind assertion; this is not restricted to a fixed-
4312 length.
4313
4314 The implementation of lookbehind assertions is, for each alternative,
4315 to temporarily move the current position back by the fixed length and
4316 then try to match. If there are insufficient characters before the cur-
4317 rent position, the assertion fails.
4318
4319 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4320 mode) to appear in lookbehind assertions, because it makes it impossi-
4321 ble to calculate the length of the lookbehind. The \X and \R escapes,
4322 which can match different numbers of bytes, are also not permitted.
4323
4324 Possessive quantifiers can be used in conjunction with lookbehind
4325 assertions to specify efficient matching at the end of the subject
4326 string. Consider a simple pattern such as
4327
4328 abcd$
4329
4330 when applied to a long string that does not match. Because matching
4331 proceeds from left to right, PCRE will look for each "a" in the subject
4332 and then see if what follows matches the rest of the pattern. If the
4333 pattern is specified as
4334
4335 ^.*abcd$
4336
4337 the initial .* matches the entire string at first, but when this fails
4338 (because there is no following "a"), it backtracks to match all but the
4339 last character, then all but the last two characters, and so on. Once
4340 again the search for "a" covers the entire string, from right to left,
4341 so we are no better off. However, if the pattern is written as
4342
4343 ^.*+(?<=abcd)
4344
4345 there can be no backtracking for the .*+ item; it can match only the
4346 entire string. The subsequent lookbehind assertion does a single test
4347 on the last four characters. If it fails, the match fails immediately.
4348 For long strings, this approach makes a significant difference to the
4349 processing time.
4350
4351 Using multiple assertions
4352
4353 Several assertions (of any sort) may occur in succession. For example,
4354
4355 (?<=\d{3})(?<!999)foo
4356
4357 matches "foo" preceded by three digits that are not "999". Notice that
4358 each of the assertions is applied independently at the same point in
4359 the subject string. First there is a check that the previous three
4360 characters are all digits, and then there is a check that the same
4361 three characters are not "999". This pattern does not match "foo" pre-
4362 ceded by six characters, the first of which are digits and the last
4363 three of which are not "999". For example, it doesn't match "123abc-
4364 foo". A pattern to do that is
4365
4366 (?<=\d{3}...)(?<!999)foo
4367
4368 This time the first assertion looks at the preceding six characters,
4369 checking that the first three are digits, and then the second assertion
4370 checks that the preceding three characters are not "999".
4371
4372 Assertions can be nested in any combination. For example,
4373
4374 (?<=(?<!foo)bar)baz
4375
4376 matches an occurrence of "baz" that is preceded by "bar" which in turn
4377 is not preceded by "foo", while
4378
4379 (?<=\d{3}(?!999)...)foo
4380
4381 is another pattern that matches "foo" preceded by three digits and any
4382 three characters that are not "999".
4383
4384
4385 CONDITIONAL SUBPATTERNS
4386
4387 It is possible to cause the matching process to obey a subpattern con-
4388 ditionally or to choose between two alternative subpatterns, depending
4389 on the result of an assertion, or whether a previous capturing subpat-
4390 tern matched or not. The two possible forms of conditional subpattern
4391 are
4392
4393 (?(condition)yes-pattern)
4394 (?(condition)yes-pattern|no-pattern)
4395
4396 If the condition is satisfied, the yes-pattern is used; otherwise the
4397 no-pattern (if present) is used. If there are more than two alterna-
4398 tives in the subpattern, a compile-time error occurs.
4399
4400 There are four kinds of condition: references to subpatterns, refer-
4401 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4402
4403 Checking for a used subpattern by number
4404
4405 If the text between the parentheses consists of a sequence of digits,
4406 the condition is true if the capturing subpattern of that number has
4407 previously matched. An alternative notation is to precede the digits
4408 with a plus or minus sign. In this case, the subpattern number is rela-
4409 tive rather than absolute. The most recently opened parentheses can be
4410 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4411 looping constructs it can also make sense to refer to subsequent groups
4412 with constructs such as (?(+2).
4413
4414 Consider the following pattern, which contains non-significant white
4415 space to make it more readable (assume the PCRE_EXTENDED option) and to
4416 divide it into three parts for ease of discussion:
4417
4418 ( \( )? [^()]+ (?(1) \) )
4419
4420 The first part matches an optional opening parenthesis, and if that
4421 character is present, sets it as the first captured substring. The sec-
4422 ond part matches one or more characters that are not parentheses. The
4423 third part is a conditional subpattern that tests whether the first set
4424 of parentheses matched or not. If they did, that is, if subject started
4425 with an opening parenthesis, the condition is true, and so the yes-pat-
4426 tern is executed and a closing parenthesis is required. Otherwise,
4427 since no-pattern is not present, the subpattern matches nothing. In
4428 other words, this pattern matches a sequence of non-parentheses,
4429 optionally enclosed in parentheses.
4430
4431 If you were embedding this pattern in a larger one, you could use a
4432 relative reference:
4433
4434 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4435
4436 This makes the fragment independent of the parentheses in the larger
4437 pattern.
4438
4439 Checking for a used subpattern by name
4440
4441 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4442 used subpattern by name. For compatibility with earlier versions of
4443 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4444 also recognized. However, there is a possible ambiguity with this syn-
4445 tax, because subpattern names may consist entirely of digits. PCRE
4446 looks first for a named subpattern; if it cannot find one and the name
4447 consists entirely of digits, PCRE looks for a subpattern of that num-
4448 ber, which must be greater than zero. Using subpattern names that con-
4449 sist entirely of digits is not recommended.
4450
4451 Rewriting the above example to use a named subpattern gives this:
4452
4453 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4454
4455
4456 Checking for pattern recursion
4457
4458 If the condition is the string (R), and there is no subpattern with the
4459 name R, the condition is true if a recursive call to the whole pattern
4460 or any subpattern has been made. If digits or a name preceded by amper-
4461 sand follow the letter R, for example:
4462
4463 (?(R3)...) or (?(R&name)...)
4464
4465 the condition is true if the most recent recursion is into the subpat-
4466 tern whose number or name is given. This condition does not check the
4467 entire recursion stack.
4468
4469 At "top level", all these recursion test conditions are false. Recur-
4470 sive patterns are described below.
4471
4472 Defining subpatterns for use by reference only
4473
4474 If the condition is the string (DEFINE), and there is no subpattern
4475 with the name DEFINE, the condition is always false. In this case,
4476 there may be only one alternative in the subpattern. It is always
4477 skipped if control reaches this point in the pattern; the idea of
4478 DEFINE is that it can be used to define "subroutines" that can be ref-
4479 erenced from elsewhere. (The use of "subroutines" is described below.)
4480 For example, a pattern to match an IPv4 address could be written like
4481 this (ignore whitespace and line breaks):
4482
4483 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4484 \b (?&byte) (\.(?&byte)){3} \b
4485
4486 The first part of the pattern is a DEFINE group inside which a another
4487 group named "byte" is defined. This matches an individual component of
4488 an IPv4 address (a number less than 256). When matching takes place,
4489 this part of the pattern is skipped because DEFINE acts like a false
4490 condition.
4491
4492 The rest of the pattern uses references to the named group to match the
4493 four dot-separated components of an IPv4 address, insisting on a word
4494 boundary at each end.
4495
4496 Assertion conditions
4497
4498 If the condition is not in any of the above formats, it must be an
4499 assertion. This may be a positive or negative lookahead or lookbehind
4500 assertion. Consider this pattern, again containing non-significant
4501 white space, and with the two alternatives on the second line:
4502
4503 (?(?=[^a-z]*[a-z])
4504 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4505
4506 The condition is a positive lookahead assertion that matches an
4507 optional sequence of non-letters followed by a letter. In other words,
4508 it tests for the presence of at least one letter in the subject. If a
4509 letter is found, the subject is matched against the first alternative;
4510 otherwise it is matched against the second. This pattern matches
4511 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4512 letters and dd are digits.
4513
4514
4515 COMMENTS
4516
4517 The sequence (?# marks the start of a comment that continues up to the
4518 next closing parenthesis. Nested parentheses are not permitted. The
4519 characters that make up a comment play no part in the pattern matching
4520 at all.
4521
4522 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4523 character class introduces a comment that continues to immediately
4524 after the next newline in the pattern.
4525
4526
4527 RECURSIVE PATTERNS
4528
4529 Consider the problem of matching a string in parentheses, allowing for
4530 unlimited nested parentheses. Without the use of recursion, the best
4531 that can be done is to use a pattern that matches up to some fixed
4532 depth of nesting. It is not possible to handle an arbitrary nesting
4533 depth.
4534
4535 For some time, Perl has provided a facility that allows regular expres-
4536 sions to recurse (amongst other things). It does this by interpolating
4537 Perl code in the expression at run time, and the code can refer to the
4538 expression itself. A Perl pattern using code interpolation to solve the
4539 parentheses problem can be created like this:
4540
4541 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4542
4543 The (?p{...}) item interpolates Perl code at run time, and in this case
4544 refers recursively to the pattern in which it appears.
4545
4546 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4547 it supports special syntax for recursion of the entire pattern, and
4548 also for individual subpattern recursion. After its introduction in
4549 PCRE and Python, this kind of recursion was introduced into Perl at
4550 release 5.10.
4551
4552 A special item that consists of (? followed by a number greater than
4553 zero and a closing parenthesis is a recursive call of the subpattern of
4554 the given number, provided that it occurs inside that subpattern. (If
4555 not, it is a "subroutine" call, which is described in the next sec-
4556 tion.) The special item (?R) or (?0) is a recursive call of the entire
4557 regular expression.
4558
4559 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4560 always treated as an atomic group. That is, once it has matched some of
4561 the subject string, it is never re-entered, even if it contains untried
4562 alternatives and there is a subsequent matching failure.
4563
4564 This PCRE pattern solves the nested parentheses problem (assume the
4565 PCRE_EXTENDED option is set so that white space is ignored):
4566
4567 \( ( (?>[^()]+) | (?R) )* \)
4568
4569 First it matches an opening parenthesis. Then it matches any number of
4570 substrings which can either be a sequence of non-parentheses, or a
4571 recursive match of the pattern itself (that is, a correctly parenthe-
4572 sized substring). Finally there is a closing parenthesis.
4573
4574 If this were part of a larger pattern, you would not want to recurse
4575 the entire pattern, so instead you could use this:
4576
4577 ( \( ( (?>[^()]+) | (?1) )* \) )
4578
4579 We have put the pattern into parentheses, and caused the recursion to
4580 refer to them instead of the whole pattern.
4581
4582 In a larger pattern, keeping track of parenthesis numbers can be
4583 tricky. This is made easier by the use of relative references. (A Perl
4584 5.10 feature.) Instead of (?1) in the pattern above you can write
4585 (?-2) to refer to the second most recently opened parentheses preceding
4586 the recursion. In other words, a negative number counts capturing
4587 parentheses leftwards from the point at which it is encountered.
4588
4589 It is also possible to refer to subsequently opened parentheses, by
4590 writing references such as (?+2). However, these cannot be recursive
4591 because the reference is not inside the parentheses that are refer-
4592 enced. They are always "subroutine" calls, as described in the next
4593 section.
4594
4595 An alternative approach is to use named parentheses instead. The Perl
4596 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4597 supported. We could rewrite the above example as follows:
4598
4599 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4600
4601 If there is more than one subpattern with the same name, the earliest
4602 one is used.
4603
4604 This particular example pattern that we have been looking at contains
4605 nested unlimited repeats, and so the use of atomic grouping for match-
4606 ing strings of non-parentheses is important when applying the pattern
4607 to strings that do not match. For example, when this pattern is applied
4608 to
4609
4610 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4611
4612 it yields "no match" quickly. However, if atomic grouping is not used,
4613 the match runs for a very long time indeed because there are so many
4614 different ways the + and * repeats can carve up the subject, and all
4615 have to be tested before failure can be reported.
4616
4617 At the end of a match, the values set for any capturing subpatterns are
4618 those from the outermost level of the recursion at which the subpattern
4619 value is set. If you want to obtain intermediate values, a callout
4620 function can be used (see below and the pcrecallout documentation). If
4621 the pattern above is matched against
4622
4623 (ab(cd)ef)
4624
4625 the value for the capturing parentheses is "ef", which is the last
4626 value taken on at the top level. If additional parentheses are added,
4627 giving
4628
4629 \( ( ( (?>[^()]+) | (?R) )* ) \)
4630 ^ ^
4631 ^ ^
4632
4633 the string they capture is "ab(cd)ef", the contents of the top level
4634 parentheses. If there are more than 15 capturing parentheses in a pat-
4635 tern, PCRE has to obtain extra memory to store data during a recursion,
4636 which it does by using pcre_malloc, freeing it via pcre_free after-
4637 wards. If no memory can be obtained, the match fails with the
4638 PCRE_ERROR_NOMEMORY error.
4639
4640 Do not confuse the (?R) item with the condition (R), which tests for
4641 recursion. Consider this pattern, which matches text in angle brack-
4642 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4643 brackets (that is, when recursing), whereas any characters are permit-
4644 ted at the outer level.
4645
4646 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4647
4648 In this pattern, (?(R) is the start of a conditional subpattern, with
4649 two different alternatives for the recursive and non-recursive cases.
4650 The (?R) item is the actual recursive call.
4651
4652
4653 SUBPATTERNS AS SUBROUTINES
4654
4655 If the syntax for a recursive subpattern reference (either by number or
4656 by name) is used outside the parentheses to which it refers, it oper-
4657 ates like a subroutine in a programming language. The "called" subpat-
4658 tern may be defined before or after the reference. A numbered reference
4659 can be absolute or relative, as in these examples:
4660
4661 (...(absolute)...)...(?2)...
4662 (...(relative)...)...(?-1)...
4663 (...(?+1)...(relative)...
4664
4665 An earlier example pointed out that the pattern
4666
4667 (sens|respons)e and \1ibility
4668
4669 matches "sense and sensibility" and "response and responsibility", but
4670 not "sense and responsibility". If instead the pattern
4671
4672 (sens|respons)e and (?1)ibility
4673
4674 is used, it does match "sense and responsibility" as well as the other
4675 two strings. Another example is given in the discussion of DEFINE
4676 above.
4677
4678 Like recursive subpatterns, a "subroutine" call is always treated as an
4679 atomic group. That is, once it has matched some of the subject string,
4680 it is never re-entered, even if it contains untried alternatives and
4681 there is a subsequent matching failure.
4682
4683 When a subpattern is used as a subroutine, processing options such as
4684 case-independence are fixed when the subpattern is defined. They cannot
4685 be changed for different calls. For example, consider this pattern:
4686
4687 (abc)(?i:(?-1))
4688
4689 It matches "abcabc". It does not match "abcABC" because the change of
4690 processing option does not affect the called subpattern.
4691
4692
4693 CALLOUTS
4694
4695 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4696 Perl code to be obeyed in the middle of matching a regular expression.
4697 This makes it possible, amongst other things, to extract different sub-
4698 strings that match the same pair of parentheses when there is a repeti-
4699 tion.
4700
4701 PCRE provides a similar feature, but of course it cannot obey arbitrary
4702 Perl code. The feature is called "callout". The caller of PCRE provides
4703 an external function by putting its entry point in the global variable
4704 pcre_callout. By default, this variable contains NULL, which disables
4705 all calling out.
4706
4707 Within a regular expression, (?C) indicates the points at which the
4708 external function is to be called. If you want to identify different
4709 callout points, you can put a number less than 256 after the letter C.
4710 The default value is zero. For example, this pattern has two callout
4711 points:
4712
4713 (?C1)abc(?C2)def
4714
4715 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4716 automatically installed before each item in the pattern. They are all
4717 numbered 255.
4718
4719 During matching, when PCRE reaches a callout point (and pcre_callout is
4720 set), the external function is called. It is provided with the number
4721 of the callout, the position in the pattern, and, optionally, one item
4722 of data originally supplied by the caller of pcre_exec(). The callout
4723 function may cause matching to proceed, to backtrack, or to fail alto-
4724 gether. A complete description of the interface to the callout function
4725 is given in the pcrecallout documentation.
4726
4727
4728 BACKTRACKING CONTROL
4729
4730 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4731 which are described in the Perl documentation as "experimental and sub-
4732 ject to change or removal in a future version of Perl". It goes on to
4733 say: "Their usage in production code should be noted to avoid problems
4734 during upgrades." The same remarks apply to the PCRE features described
4735 in this section.
4736
4737 Since these verbs are specifically related to backtracking, they can be
4738 used only when the pattern is to be matched using pcre_exec(), which
4739 uses a backtracking algorithm. They cause an error if encountered by
4740 pcre_dfa_exec().
4741
4742 The new verbs make use of what was previously invalid syntax: an open-
4743 ing parenthesis followed by an asterisk. In Perl, they are generally of
4744 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4745 its general form is just (*VERB). Any number of these verbs may occur
4746 in a pattern. There are two kinds:
4747
4748 Verbs that act immediately
4749
4750 The following verbs act as soon as they are encountered:
4751
4752 (*ACCEPT)
4753
4754 This verb causes the match to end successfully, skipping the remainder
4755 of the pattern. When inside a recursion, only the innermost pattern is
4756 ended immediately. PCRE differs from Perl in what happens if the
4757 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4758 captured: in PCRE no data is captured. For example:
4759
4760 A(A|B(*ACCEPT)|C)D
4761
4762 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4763 is captured.
4764
4765 (*FAIL) or (*F)
4766
4767 This verb causes the match to fail, forcing backtracking to occur. It
4768 is equivalent to (?!) but easier to read. The Perl documentation notes
4769 that it is probably useful only when combined with (?{}) or (??{}).
4770 Those are, of course, Perl features that are not present in PCRE. The
4771 nearest equivalent is the callout feature, as for example in this pat-
4772 tern:
4773
4774 a+(?C)(*FAIL)
4775
4776 A match with the string "aaaa" always fails, but the callout is taken
4777 before each backtrack happens (in this example, 10 times).
4778
4779 Verbs that act after backtracking
4780
4781 The following verbs do nothing when they are encountered. Matching con-
4782 tinues with what follows, but if there is no subsequent match, a fail-
4783 ure is forced. The verbs differ in exactly what kind of failure
4784 occurs.
4785
4786 (*COMMIT)
4787
4788 This verb causes the whole match to fail outright if the rest of the
4789 pattern does not match. Even if the pattern is unanchored, no further
4790 attempts to find a match by advancing the start point take place. Once
4791 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4792 at the current starting point, or not at all. For example:
4793
4794 a+(*COMMIT)b
4795
4796 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4797 of dynamic anchor, or "I've started, so I must finish."
4798
4799 (*PRUNE)
4800
4801 This verb causes the match to fail at the current position if the rest
4802 of the pattern does not match. If the pattern is unanchored, the normal
4803 "bumpalong" advance to the next starting character then happens. Back-
4804 tracking can occur as usual to the left of (*PRUNE), or when matching
4805 to the right of (*PRUNE), but if there is no match to the right, back-
4806 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4807 is just an alternative to an atomic group or possessive quantifier, but
4808 there are some uses of (*PRUNE) that cannot be expressed in any other
4809 way.
4810
4811 (*SKIP)
4812
4813 This verb is like (*PRUNE), except that if the pattern is unanchored,
4814 the "bumpalong" advance is not to the next character, but to the posi-
4815 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4816 that whatever text was matched leading up to it cannot be part of a
4817 successful match. Consider:
4818
4819 a+(*SKIP)b
4820
4821 If the subject is "aaaac...", after the first match attempt fails
4822 (starting at the first character in the string), the starting point
4823 skips on to start the next attempt at "c". Note that a possessive quan-
4824 tifer does not have the same effect in this example; although it would
4825 suppress backtracking during the first match attempt, the second
4826 attempt would start at the second character instead of skipping on to
4827 "c".
4828
4829 (*THEN)
4830
4831 This verb causes a skip to the next alternation if the rest of the pat-
4832 tern does not match. That is, it cancels pending backtracking, but only
4833 within the current alternation. Its name comes from the observation
4834 that it can be used for a pattern-based if-then-else block:
4835
4836 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
4837
4838 If the COND1 pattern matches, FOO is tried (and possibly further items
4839 after the end of the group if FOO succeeds); on failure the matcher
4840 skips to the second alternative and tries COND2, without backtracking
4841 into COND1. If (*THEN) is used outside of any alternation, it acts
4842 exactly like (*PRUNE).
4843
4844
4845 SEE ALSO
4846
4847 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4848
4849
4850 AUTHOR
4851
4852 Philip Hazel
4853 University Computing Service
4854 Cambridge CB2 3QH, England.
4855
4856
4857 REVISION
4858
4859 Last updated: 14 September 2007
4860 Copyright (c) 1997-2007 University of Cambridge.
4861 ------------------------------------------------------------------------------
4862
4863
4864 PCRESYNTAX(3) PCRESYNTAX(3)
4865
4866
4867 NAME
4868 PCRE - Perl-compatible regular expressions
4869
4870
4871 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
4872
4873 The full syntax and semantics of the regular expressions that are sup-
4874 ported by PCRE are described in the pcrepattern documentation. This
4875 document contains just a quick-reference summary of the syntax.
4876
4877
4878 QUOTING
4879
4880 \x where x is non-alphanumeric is a literal x
4881 \Q...\E treat enclosed characters as literal
4882
4883
4884 CHARACTERS
4885
4886 \a alarm, that is, the BEL character (hex 07)
4887 \cx "control-x", where x is any character
4888 \e escape (hex 1B)
4889 \f formfeed (hex 0C)
4890 \n newline (hex 0A)
4891 \r carriage return (hex 0D)
4892 \t tab (hex 09)
4893 \ddd character with octal code ddd, or backreference
4894 \xhh character with hex code hh
4895 \x{hhh..} character with hex code hhh..
4896
4897
4898 CHARACTER TYPES
4899
4900 . any character except newline;
4901 in dotall mode, any character whatsoever
4902 \C one byte, even in UTF-8 mode (best avoided)
4903 \d a decimal digit
4904 \D a character that is not a decimal digit
4905 \h a horizontal whitespace character
4906 \H a character that is not a horizontal whitespace character
4907 \p{xx} a character with the xx property
4908 \P{xx} a character without the xx property
4909 \R a newline sequence
4910 \s a whitespace character
4911 \S a character that is not a whitespace character
4912 \v a vertical whitespace character
4913 \V a character that is not a vertical whitespace character
4914 \w a "word" character
4915 \W a "non-word" character
4916 \X an extended Unicode sequence
4917
4918 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
4919
4920
4921 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
4922
4923 C Other
4924 Cc Control
4925 Cf Format
4926 Cn Unassigned
4927 Co Private use
4928 Cs Surrogate
4929
4930 L Letter
4931 Ll Lower case letter
4932 Lm Modifier letter
4933 Lo Other letter
4934 Lt Title case letter
4935 Lu Upper case letter
4936 L& Ll, Lu, or Lt
4937
4938 M Mark
4939 Mc Spacing mark
4940 Me Enclosing mark
4941 Mn Non-spacing mark
4942
4943 N Number
4944 Nd Decimal number
4945 Nl Letter number
4946 No Other number
4947
4948 P Punctuation
4949 Pc Connector punctuation
4950 Pd Dash punctuation
4951 Pe Close punctuation
4952 Pf Final punctuation
4953 Pi Initial punctuation
4954 Po Other punctuation
4955 Ps Open punctuation
4956
4957 S Symbol
4958 Sc Currency symbol
4959 Sk Modifier symbol
4960 Sm Mathematical symbol
4961 So Other symbol
4962
4963 Z Separator
4964 Zl Line separator
4965 Zp Paragraph separator
4966 Zs Space separator
4967
4968
4969 SCRIPT NAMES FOR \p AND \P
4970
4971 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
4972 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
4973 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
4974 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
4975 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
4976 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
4977 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
4978 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
4979 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
4980
4981
4982 CHARACTER CLASSES
4983
4984 [...] positive character class
4985 [^...] negative character class
4986 [x-y] range (can be used for hex characters)
4987 [[:xxx:]] positive POSIX named set
4988 [[^:xxx:]] negative POSIX named set
4989
4990 alnum alphanumeric
4991 alpha alphabetic
4992 ascii 0-127
4993 blank space or tab
4994 cntrl control character
4995 digit decimal digit
4996 graph printing, excluding space
4997 lower lower case letter
4998 print printing, including space
4999 punct printing, excluding alphanumeric
5000 space whitespace
5001 upper upper case letter
5002 word same as \w
5003 xdigit hexadecimal digit
5004
5005 In PCRE, POSIX character set names recognize only ASCII characters. You
5006 can use \Q...\E inside a character class.
5007
5008
5009 QUANTIFIERS
5010
5011 ? 0 or 1, greedy
5012 ?+ 0 or 1, possessive
5013 ?? 0 or 1, lazy
5014 * 0 or more, greedy
5015 *+ 0 or more, possessive
5016 *? 0 or more, lazy
5017 + 1 or more, greedy
5018 ++ 1 or more, possessive
5019 +? 1 or more, lazy
5020 {n} exactly n
5021 {n,m} at least n, no more than m, greedy
5022 {n,m}+ at least n, no more than m, possessive
5023 {n,m}? at least n, no more than m, lazy
5024 {n,} n or more, greedy
5025 {n,}+ n or more, possessive
5026 {n,}? n or more, lazy
5027
5028
5029 ANCHORS AND SIMPLE ASSERTIONS
5030
5031 \b word boundary
5032 \B not a word boundary
5033 ^ start of subject
5034 also after internal newline in multiline mode
5035 \A start of subject
5036 $ end of subject
5037 also before newline at end of subject
5038 also before internal newline in multiline mode
5039 \Z end of subject
5040 also before newline at end of subject
5041 \z end of subject
5042 \G first matching position in subject
5043
5044
5045 MATCH POINT RESET
5046
5047 \K reset start of match
5048
5049
5050 ALTERNATION
5051
5052 expr|expr|expr...
5053
5054
5055 CAPTURING
5056
5057 (...) capturing group
5058 (?<name>...) named capturing group (Perl)
5059 (?'name'...) named capturing group (Perl)
5060 (?P<name>...) named capturing group (Python)
5061 (?:...) non-capturing group
5062 (?|...) non-capturing group; reset group numbers for
5063 capturing groups in each alternative
5064
5065
5066 ATOMIC GROUPS
5067
5068 (?>...) atomic, non-capturing group
5069
5070
5071 COMMENT
5072
5073 (?#....) comment (not nestable)
5074
5075
5076 OPTION SETTING
5077
5078 (?i) caseless
5079 (?J) allow duplicate names
5080 (?m) multiline
5081 (?s) single line (dotall)
5082 (?U) default ungreedy (lazy)
5083 (?x) extended (ignore white space)
5084 (?-...) unset option(s)
5085
5086
5087 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5088
5089 (?=...) positive look ahead
5090 (?!...) negative look ahead
5091 (?<=...) positive look behind
5092 (?<!...) negative look behind
5093
5094 Each top-level branch of a look behind must be of a fixed length.
5095
5096
5097 BACKREFERENCES
5098
5099 \n reference by number (can be ambiguous)
5100 \gn reference by number
5101 \g{n} reference by number
5102 \g{-n} relative reference by number
5103 \k<name> reference by name (Perl)
5104 \k'name' reference by name (Perl)
5105 \g{name} reference by name (Perl)
5106 \k{name} reference by name (.NET)
5107 (?P=name) reference by name (Python)
5108
5109
5110 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5111
5112 (?R) recurse whole pattern
5113 (?n) call subpattern by absolute number
5114 (?+n) call subpattern by relative number
5115 (?-n) call subpattern by relative number
5116 (?&name) call subpattern by name (Perl)
5117 (?P>name) call subpattern by name (Python)
5118
5119
5120 CONDITIONAL PATTERNS
5121
5122 (?(condition)yes-pattern)
5123 (?(condition)yes-pattern|no-pattern)
5124
5125 (?(n)... absolute reference condition
5126 (?(+n)... relative reference condition
5127 (?(-n)... relative reference condition
5128 (?(<name>)... named reference condition (Perl)
5129 (?('name')... named reference condition (Perl)
5130 (?(name)... named reference condition (PCRE)
5131 (?(R)... overall recursion condition
5132 (?(Rn)... specific group recursion condition
5133 (?(R&name)... specific recursion condition
5134 (?(DEFINE)... define subpattern for reference
5135 (?(assert)... assertion condition
5136
5137
5138 BACKTRACKING CONTROL
5139
5140 The following act immediately they are reached:
5141
5142 (*ACCEPT) force successful match
5143 (*FAIL) force backtrack; synonym (*F)
5144
5145 The following act only when a subsequent match failure causes a back-
5146 track to reach them. They all force a match failure, but they differ in
5147 what happens afterwards. Those that advance the start-of-match point do
5148 so only if the pattern is not anchored.
5149
5150 (*COMMIT) overall failure, no advance of starting point
5151 (*PRUNE) advance to next starting character
5152 (*SKIP) advance start to current matching position
5153 (*THEN) local failure, backtrack to next alternation
5154
5155
5156 NEWLINE CONVENTIONS
5157
5158 These are recognized only at the very start of a pattern.
5159
5160 (*CR)
5161 (*LF)
5162 (*CRLF)
5163 (*ANYCRLF)
5164 (*ANY)
5165
5166
5167 WHAT \R MATCHES
5168
5169 These are recognized only at the very start of a pattern.
5170
5171 (*BSR_ANYCRLF)
5172 (*BSR_UNICODE)
5173
5174
5175 CALLOUTS
5176
5177 (?C) callout
5178 (?Cn) callout with data n
5179
5180
5181 SEE ALSO
5182
5183 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5184
5185
5186 AUTHOR
5187
5188 Philip Hazel
5189 University Computing Service
5190 Cambridge CB2 3QH, England.
5191
5192
5193 REVISION
5194
5195 Last updated: 11 September 2007
5196 Copyright (c) 1997-2007 University of Cambridge.
5197 ------------------------------------------------------------------------------
5198
5199
5200 PCREPARTIAL(3) PCREPARTIAL(3)
5201
5202
5203 NAME
5204 PCRE - Perl-compatible regular expressions
5205
5206
5207 PARTIAL MATCHING IN PCRE
5208
5209 In normal use of PCRE, if the subject string that is passed to
5210 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5211 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5212 There are circumstances where it might be helpful to distinguish this
5213 case from other cases in which there is no match.
5214
5215 Consider, for example, an application where a human is required to type
5216 in data for a field with specific formatting requirements. An example
5217 might be a date in the form ddmmmyy, defined by this pattern:
5218
5219 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5220
5221 If the application sees the user's keystrokes one by one, and can check
5222 that what has been typed so far is potentially valid, it is able to
5223 raise an error as soon as a mistake is made, possibly beeping and not
5224 reflecting the character that has been typed. This immediate feedback
5225 is likely to be a better user interface than a check that is delayed
5226 until the entire string has been entered.
5227
5228 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5229 TIAL option, which can be set when calling pcre_exec() or
5230 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5231 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5232 during the matching process the last part of the subject string matched
5233 part of the pattern. Unfortunately, for non-anchored matching, it is
5234 not possible to obtain the position of the start of the partial match.
5235 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5236
5237 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5238 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5239 the subject is reached, there have been no complete matches, but there
5240 is still at least one matching possibility. The portion of the string
5241 that provided the partial match is set as the first matching string.
5242
5243 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5244 the last literal byte in a pattern, and abandons matching immediately
5245 if such a byte is not present in the subject string. This optimization
5246 cannot be used for a subject string that might match only partially.
5247
5248
5249 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5250
5251 Because of the way certain internal optimizations are implemented in
5252 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5253 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5254 used. For pcre_exec(), repeated single characters such as
5255
5256 a{2,4}
5257
5258 and repeated single metasequences such as
5259
5260 \d+
5261
5262 are not permitted if the maximum number of occurrences is greater than
5263 one. Optional items such as \d? (where the maximum is one) are permit-
5264 ted. Quantifiers with any values are permitted after parentheses, so
5265 the invalid examples above can be coded thus:
5266
5267 (a){2,4}
5268 (\d)+
5269
5270 These constructions run more slowly, but for the kinds of application
5271 that are envisaged for this facility, this is not felt to be a major
5272 restriction.
5273
5274 If PCRE_PARTIAL is set for a pattern that does not conform to the
5275 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5276 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5277 find out if a compiled pattern can be used for partial matching.
5278
5279
5280 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5281
5282 If the escape sequence \P is present in a pcretest data line, the
5283 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5284 uses the date example quoted above:
5285
5286 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5287 data> 25jun04\P
5288 0: 25jun04
5289 1: jun
5290 data> 25dec3\P
5291 Partial match
5292 data> 3ju\P
5293 Partial match
5294 data> 3juj\P
5295 No match
5296 data> j\P
5297 No match
5298
5299 The first data string is matched completely, so pcretest shows the
5300 matched substrings. The remaining four strings do not match the com-
5301 plete pattern, but the first two are partial matches. The same test,
5302 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5303 produces the following output:
5304
5305 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5306 data> 25jun04\P\D
5307 0: 25jun04
5308 data> 23dec3\P\D
5309 Partial match: 23dec3
5310 data> 3ju\P\D
5311 Partial match: 3ju
5312 data> 3juj\P\D
5313 No match
5314 data> j\P\D
5315 No match
5316
5317 Notice that in this case the portion of the string that was matched is
5318 made available.
5319
5320
5321 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5322
5323 When a partial match has been found using pcre_dfa_exec(), it is possi-
5324 ble to continue the match by providing additional subject data and
5325 calling pcre_dfa_exec() again with the same compiled regular expres-
5326 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5327 the same working space as before, because this is where details of the
5328 previous partial match are stored. Here is an example using pcretest,
5329 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5330 \D are as above):
5331
5332 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5333 data> 23ja\P\D
5334 Partial match: 23ja
5335 data> n05\R\D
5336 0: n05
5337
5338 The first call has "23ja" as the subject, and requests partial match-
5339 ing; the second call has "n05" as the subject for the continued
5340 (restarted) match. Notice that when the match is complete, only the
5341 last part is shown; PCRE does not retain the previously partially-
5342 matched string. It is up to the calling program to do that if it needs
5343 to.
5344
5345 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5346 matching over multiple segments. This facility can be used to pass very
5347 long subject strings to pcre_dfa_exec(). However, some care is needed
5348 for certain types of pattern.
5349
5350 1. If the pattern contains tests for the beginning or end of a line,
5351 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5352 ate, when the subject string for any call does not contain the begin-
5353 ning or end of a line.
5354
5355 2. If the pattern contains backward assertions (including \b or \B),
5356 you need to arrange for some overlap in the subject strings to allow
5357 for this. For example, you could pass the subject in chunks that are
5358 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5359 set to 200 and the previous 200 bytes at the start of the buffer.
5360
5361 3. Matching a subject string that is split into multiple segments does
5362 not always produce exactly the same result as matching over one single
5363 long string. The difference arises when there are multiple matching
5364 possibilities, because a partial match result is given only when there
5365 are no completed matches in a call to pcre_dfa_exec(). This means that
5366 as soon as the shortest match has been found, continuation to a new
5367 subject segment is no longer possible. Consider this pcretest example:
5368
5369 re> /dog(sbody)?/
5370 data> do\P\D
5371 Partial match: do
5372 data> gsb\R\P\D
5373 0: g
5374 data> dogsbody\D
5375 0: dogsbody
5376 1: dog
5377
5378 The pattern matches the words "dog" or "dogsbody". When the subject is
5379 presented in several parts ("do" and "gsb" being the first two) the
5380 match stops when "dog" has been found, and it is not possible to con-
5381 tinue. On the other hand, if "dogsbody" is presented as a single
5382 string, both matches are found.
5383
5384 Because of this phenomenon, it does not usually make sense to end a
5385 pattern that is going to be matched in this way with a variable repeat.
5386
5387 4. Patterns that contain alternatives at the top level which do not all
5388 start with the same pattern item may not work as expected. For example,
5389 consider this pattern:
5390
5391 1234|3789
5392
5393 If the first part of the subject is "ABC123", a partial match of the
5394 first alternative is found at offset 3. There is no partial match for
5395 the second alternative, because such a match does not start at the same
5396 point in the subject string. Attempting to continue with the string
5397 "789" does not yield a match because only those alternatives that match
5398 at one point in the subject are remembered. The problem arises because
5399 the start of the second alternative matches within the first alterna-
5400 tive. There is no problem with anchored patterns or patterns such as:
5401
5402 1234|ABCD
5403
5404 where no string can be a partial match for both alternatives.
5405
5406
5407 AUTHOR
5408
5409 Philip Hazel
5410 University Computing Service
5411 Cambridge CB2 3QH, England.
5412
5413
5414 REVISION
5415
5416 Last updated: 04 June 2007
5417 Copyright (c) 1997-2007 University of Cambridge.
5418 ------------------------------------------------------------------------------
5419
5420
5421 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5422
5423
5424 NAME
5425 PCRE - Perl-compatible regular expressions
5426
5427
5428 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5429
5430 If you are running an application that uses a large number of regular
5431 expression patterns, it may be useful to store them in a precompiled
5432 form instead of having to compile them every time the application is
5433 run. If you are not using any private character tables (see the
5434 pcre_maketables() documentation), this is relatively straightforward.
5435 If you are using private tables, it is a little bit more complicated.
5436
5437 If you save compiled patterns to a file, you can copy them to a differ-
5438 ent host and run them there. This works even if the new host has the
5439 opposite endianness to the one on which the patterns were compiled.
5440 There may be a small performance penalty, but it should be insignifi-
5441 cant. However, compiling regular expressions with one version of PCRE
5442 for use with a different version is not guaranteed to work and may
5443 cause crashes.
5444
5445
5446 SAVING A COMPILED PATTERN
5447 The value returned by pcre_compile() points to a single block of memory
5448 that holds the compiled pattern and associated data. You can find the
5449 length of this block in bytes by calling pcre_fullinfo() with an argu-
5450 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5451 manner. Here is sample code that compiles a pattern and writes it to a
5452 file. It assumes that the variable fd refers to a file that is open for
5453 output:
5454
5455 int erroroffset, rc, size;
5456 char *error;
5457 pcre *re;
5458
5459 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5460 if (re == NULL) { ... handle errors ... }
5461 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5462 if (rc < 0) { ... handle errors ... }
5463 rc = fwrite(re, 1, size, fd);
5464 if (rc != size) { ... handle errors ... }
5465
5466 In this example, the bytes that comprise the compiled pattern are
5467 copied exactly. Note that this is binary data that may contain any of
5468 the 256 possible byte values. On systems that make a distinction
5469 between binary and non-binary data, be sure that the file is opened for
5470 binary output.
5471
5472 If you want to write more than one pattern to a file, you will have to
5473 devise a way of separating them. For binary data, preceding each pat-
5474 tern with its length is probably the most straightforward approach.
5475 Another possibility is to write out the data in hexadecimal instead of
5476 binary, one pattern to a line.
5477
5478 Saving compiled patterns in a file is only one possible way of storing
5479 them for later use. They could equally well be saved in a database, or
5480 in the memory of some daemon process that passes them via sockets to
5481 the processes that want them.
5482
5483 If the pattern has been studied, it is also possible to save the study
5484 data in a similar way to the compiled pattern itself. When studying
5485 generates additional information, pcre_study() returns a pointer to a
5486 pcre_extra data block. Its format is defined in the section on matching
5487 a pattern in the pcreapi documentation. The study_data field points to
5488 the binary study data, and this is what you must save (not the
5489 pcre_extra block itself). The length of the study data can be obtained
5490 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5491 Remember to check that pcre_study() did return a non-NULL value before
5492 trying to save the study data.
5493
5494
5495 RE-USING A PRECOMPILED PATTERN
5496
5497 Re-using a precompiled pattern is straightforward. Having reloaded it
5498 into main memory, you pass its pointer to pcre_exec() or
5499 pcre_dfa_exec() in the usual way. This should work even on another
5500 host, and even if that host has the opposite endianness to the one
5501 where the pattern was compiled.
5502
5503 However, if you passed a pointer to custom character tables when the
5504 pattern was compiled (the tableptr argument of pcre_compile()), you
5505 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5506 because the value saved with the compiled pattern will obviously be
5507 nonsense. A field in a pcre_extra() block is used to pass this data, as
5508 described in the section on matching a pattern in the pcreapi documen-
5509 tation.
5510
5511 If you did not provide custom character tables when the pattern was
5512 compiled, the pointer in the compiled pattern is NULL, which causes
5513 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5514 take any special action at run time in this case.
5515
5516 If you saved study data with the compiled pattern, you need to create
5517 your own pcre_extra data block and set the study_data field to point to
5518 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5519 bit in the flags field to indicate that study data is present. Then
5520 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5521 usual way.
5522
5523
5524 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5525
5526 In general, it is safest to recompile all saved patterns when you
5527 update to a new PCRE release, though not all updates actually require
5528 this. Recompiling is definitely needed for release 7.2.
5529
5530
5531 AUTHOR
5532
5533 Philip Hazel
5534 University Computing Service
5535 Cambridge CB2 3QH, England.
5536
5537
5538 REVISION
5539
5540 Last updated: 13 June 2007
5541 Copyright (c) 1997-2007 University of Cambridge.
5542 ------------------------------------------------------------------------------
5543
5544
5545 PCREPERFORM(3) PCREPERFORM(3)
5546
5547
5548 NAME
5549 PCRE - Perl-compatible regular expressions
5550
5551
5552 PCRE PERFORMANCE
5553
5554 Two aspects of performance are discussed below: memory usage and pro-
5555 cessing time. The way you express your pattern as a regular expression
5556 can affect both of them.
5557
5558
5559 MEMORY USAGE
5560
5561 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5562 that most simple patterns do not use much memory. However, there is one
5563 case where memory usage can be unexpectedly large. When a parenthesized
5564 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5565 ited maximum, the whole subpattern is repeated in the compiled code.
5566 For example, the pattern
5567
5568 (abc|def){2,4}
5569
5570 is compiled as if it were
5571
5572 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5573
5574 (Technical aside: It is done this way so that backtrack points within
5575 each of the repetitions can be independently maintained.)
5576
5577 For regular expressions whose quantifiers use only small numbers, this
5578 is not usually a problem. However, if the numbers are large, and par-
5579 ticularly if such repetitions are nested, the memory usage can become
5580 an embarrassment. For example, the very simple pattern
5581
5582 ((ab){1,1000}c){1,3}
5583
5584 uses 51K bytes when compiled. When PCRE is compiled with its default
5585 internal pointer size of two bytes, the size limit on a compiled pat-
5586 tern is 64K, and this is reached with the above pattern if the outer
5587 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5588 internal pointers and thus handle larger compiled patterns, but it is
5589 better to try to rewrite your pattern to use less memory if you can.
5590
5591 One way of reducing the memory usage for such patterns is to make use
5592 of PCRE's "subroutine" facility. Re-writing the above pattern as
5593
5594 ((ab)(?2){0,999}c)(?1){0,2}
5595
5596 reduces the memory requirements to 18K, and indeed it remains under 20K
5597 even with the outer repetition increased to 100. However, this pattern
5598 is not exactly equivalent, because the "subroutine" calls are treated
5599 as atomic groups into which there can be no backtracking if there is a
5600 subsequent matching failure. Therefore, PCRE cannot do this kind of
5601 rewriting automatically. Furthermore, there is a noticeable loss of
5602 speed when executing the modified pattern. Nevertheless, if the atomic
5603 grouping is not a problem and the loss of speed is acceptable, this
5604 kind of rewriting will allow you to process patterns that PCRE cannot
5605 otherwise handle.
5606
5607
5608 PROCESSING TIME
5609
5610 Certain items in regular expression patterns are processed more effi-
5611 ciently than others. It is more efficient to use a character class like
5612 [aeiou] than a set of single-character alternatives such as
5613 (a|e|i|o|u). In general, the simplest construction that provides the
5614 required behaviour is usually the most efficient. Jeffrey Friedl's book
5615 contains a lot of useful general discussion about optimizing regular
5616 expressions for efficient performance. This document contains a few
5617 observations about PCRE.
5618
5619 Using Unicode character properties (the \p, \P, and \X escapes) is
5620 slow, because PCRE has to scan a structure that contains data for over
5621 fifteen thousand characters whenever it needs a character's property.
5622 If you can find an alternative pattern that does not use character
5623 properties, it will probably be faster.
5624
5625 When a pattern begins with .* not in parentheses, or in parentheses
5626 that are not the subject of a backreference, and the PCRE_DOTALL option
5627 is set, the pattern is implicitly anchored by PCRE, since it can match
5628 only at the start of a subject string. However, if PCRE_DOTALL is not
5629 set, PCRE cannot make this optimization, because the . metacharacter
5630 does not then match a newline, and if the subject string contains new-
5631 lines, the pattern may match from the character immediately following
5632 one of them instead of from the very start. For example, the pattern
5633
5634 .*second
5635
5636 matches the subject "first\nand second" (where \n stands for a newline
5637 character), with the match starting at the seventh character. In order
5638 to do this, PCRE has to retry the match starting after every newline in
5639 the subject.
5640
5641 If you are using such a pattern with subject strings that do not con-
5642 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5643 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5644 ing. That saves PCRE from having to scan along the subject looking for
5645 a newline to restart at.
5646
5647 Beware of patterns that contain nested indefinite repeats. These can
5648 take a long time to run when applied to a string that does not match.
5649 Consider the pattern fragment
5650
5651 ^(a+)*
5652
5653 This can match "aaaa" in 16 different ways, and this number increases
5654 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5655 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5656 repeats can match different numbers of times.) When the remainder of
5657 the pattern is such that the entire match is going to fail, PCRE has in
5658 principle to try every possible variation, and this can take an
5659 extremely long time, even for relatively short strings.
5660
5661 An optimization catches some of the more simple cases such as
5662
5663 (a+)*b
5664
5665 where a literal character follows. Before embarking on the standard
5666 matching procedure, PCRE checks that there is a "b" later in the sub-
5667 ject string, and if there is not, it fails the match immediately. How-
5668 ever, when there is no following literal this optimization cannot be
5669 used. You can see the difference by comparing the behaviour of
5670
5671 (a+)*\d
5672
5673 with the pattern above. The former gives a failure almost instantly
5674 when applied to a whole line of "a" characters, whereas the latter
5675 takes an appreciable time with strings longer than about 20 characters.
5676
5677 In many cases, the solution to this kind of performance issue is to use
5678 an atomic group or a possessive quantifier.
5679
5680
5681 AUTHOR
5682
5683 Philip Hazel
5684 University Computing Service
5685 Cambridge CB2 3QH, England.
5686
5687
5688 REVISION
5689
5690 Last updated: 06 March 2007
5691 Copyright (c) 1997-2007 University of Cambridge.
5692 ------------------------------------------------------------------------------
5693
5694
5695 PCREPOSIX(3) PCREPOSIX(3)
5696
5697
5698 NAME
5699 PCRE - Perl-compatible regular expressions.
5700
5701
5702 SYNOPSIS OF POSIX API
5703
5704 #include <pcreposix.h>
5705
5706 int regcomp(regex_t *preg, const char *pattern,
5707 int cflags);
5708
5709 int regexec(regex_t *preg, const char *string,
5710 size_t nmatch, regmatch_t pmatch[], int eflags);
5711
5712 size_t regerror(int errcode, const regex_t *preg,
5713 char *errbuf, size_t errbuf_size);
5714
5715 void regfree(regex_t *preg);
5716
5717
5718 DESCRIPTION
5719
5720 This set of functions provides a POSIX-style API to the PCRE regular
5721 expression package. See the pcreapi documentation for a description of
5722 PCRE's native API, which contains much additional functionality.
5723
5724 The functions described here are just wrapper functions that ultimately
5725 call the PCRE native API. Their prototypes are defined in the
5726 pcreposix.h header file, and on Unix systems the library itself is
5727 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5728 command for linking an application that uses them. Because the POSIX
5729 functions call the native ones, it is also necessary to add -lpcre.
5730
5731 I have implemented only those option bits that can be reasonably mapped
5732 to PCRE native options. In addition, the option REG_EXTENDED is defined
5733 with the value zero. This has no effect, but since programs that are
5734 written to the POSIX interface often use it, this makes it easier to
5735 slot in PCRE as a replacement library. Other POSIX options are not even
5736 defined.
5737
5738 When PCRE is called via these functions, it is only the API that is
5739 POSIX-like in style. The syntax and semantics of the regular expres-
5740 sions themselves are still those of Perl, subject to the setting of
5741 various PCRE options, as described below. "POSIX-like in style" means
5742 that the API approximates to the POSIX definition; it is not fully
5743 POSIX-compatible, and in multi-byte encoding domains it is probably
5744 even less compatible.
5745
5746 The header for these functions is supplied as pcreposix.h to avoid any
5747 potential clash with other POSIX libraries. It can, of course, be
5748 renamed or aliased as regex.h, which is the "correct" name. It provides
5749 two structure types, regex_t for compiled internal forms, and reg-
5750 match_t for returning captured substrings. It also defines some con-
5751 stants whose names start with "REG_"; these are used for setting
5752 options and identifying error codes.
5753
5754
5755 COMPILING A PATTERN
5756
5757 The function regcomp() is called to compile a pattern into an internal
5758 form. The pattern is a C string terminated by a binary zero, and is
5759 passed in the argument pattern. The preg argument is a pointer to a
5760 regex_t structure that is used as a base for storing information about
5761 the compiled regular expression.
5762
5763 The argument cflags is either zero, or contains one or more of the bits
5764 defined by the following macros:
5765
5766 REG_DOTALL
5767
5768 The PCRE_DOTALL option is set when the regular expression is passed for
5769 compilation to the native function. Note that REG_DOTALL is not part of
5770 the POSIX standard.
5771
5772 REG_ICASE
5773
5774 The PCRE_CASELESS option is set when the regular expression is passed
5775 for compilation to the native function.
5776
5777 REG_NEWLINE
5778
5779 The PCRE_MULTILINE option is set when the regular expression is passed
5780 for compilation to the native function. Note that this does not mimic
5781 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5782 tion).
5783
5784 REG_NOSUB
5785
5786 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5787 passed for compilation to the native function. In addition, when a pat-
5788 tern that is compiled with this flag is passed to regexec() for match-
5789 ing, the nmatch and pmatch arguments are ignored, and no captured
5790 strings are returned.
5791
5792 REG_UTF8
5793
5794 The PCRE_UTF8 option is set when the regular expression is passed for
5795 compilation to the native function. This causes the pattern itself and
5796 all data strings used for matching it to be treated as UTF-8 strings.
5797 Note that REG_UTF8 is not part of the POSIX standard.
5798
5799 In the absence of these flags, no options are passed to the native
5800 function. This means the the regex is compiled with PCRE default
5801 semantics. In particular, the way it handles newline characters in the
5802 subject string is the Perl way, not the POSIX way. Note that setting
5803 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5804 It does not affect the way newlines are matched by . (they aren't) or
5805 by a negative class such as [^a] (they are).
5806
5807 The yield of regcomp() is zero on success, and non-zero otherwise. The
5808 preg structure is filled in on success, and one member of the structure
5809 is public: re_nsub contains the number of capturing subpatterns in the
5810 regular expression. Various error codes are defined in the header file.
5811
5812
5813 MATCHING NEWLINE CHARACTERS
5814
5815 This area is not simple, because POSIX and Perl take different views of
5816 things. It is not possible to get PCRE to obey POSIX semantics, but
5817 then PCRE was never intended to be a POSIX engine. The following table
5818 lists the different possibilities for matching newline characters in
5819 PCRE:
5820
5821 Default Change with
5822
5823 . matches newline no PCRE_DOTALL
5824 newline matches [^a] yes not changeable
5825 $ matches \n at end yes PCRE_DOLLARENDONLY
5826 $ matches \n in middle no PCRE_MULTILINE
5827 ^ matches \n in middle no PCRE_MULTILINE
5828
5829 This is the equivalent table for POSIX:
5830
5831 Default Change with
5832
5833 . matches newline yes REG_NEWLINE
5834 newline matches [^a] yes REG_NEWLINE
5835 $ matches \n at end no REG_NEWLINE
5836 $ matches \n in middle no REG_NEWLINE
5837 ^ matches \n in middle no REG_NEWLINE
5838
5839 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5840 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5841 no way to stop newline from matching [^a].
5842
5843 The default POSIX newline handling can be obtained by setting
5844 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5845 behave exactly as for the REG_NEWLINE action.
5846
5847
5848 MATCHING A PATTERN
5849
5850 The function regexec() is called to match a compiled pattern preg
5851 against a given string, which is terminated by a zero byte, subject to
5852 the options in eflags. These can be:
5853
5854 REG_NOTBOL
5855
5856 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5857 function.
5858
5859 REG_NOTEOL
5860
5861 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5862 function.
5863
5864 If the pattern was compiled with the REG_NOSUB flag, no data about any
5865 matched strings is returned. The nmatch and pmatch arguments of
5866 regexec() are ignored.
5867
5868 Otherwise,the portion of the string that was matched, and also any cap-
5869 tured substrings, are returned via the pmatch argument, which points to
5870 an array of nmatch structures of type regmatch_t, containing the mem-
5871 bers rm_so and rm_eo. These contain the offset to the first character
5872 of each substring and the offset to the first character after the end
5873 of each substring, respectively. The 0th element of the vector relates
5874 to the entire portion of string that was matched; subsequent elements
5875 relate to the capturing subpatterns of the regular expression. Unused
5876 entries in the array have both structure members set to -1.
5877
5878 A successful match yields a zero return; various error codes are
5879 defined in the header file, of which REG_NOMATCH is the "expected"
5880 failure code.
5881
5882
5883 ERROR MESSAGES
5884
5885 The regerror() function maps a non-zero errorcode from either regcomp()
5886 or regexec() to a printable message. If preg is not NULL, the error
5887 should have arisen from the use of that structure. A message terminated
5888 by a binary zero is placed in errbuf. The length of the message,
5889 including the zero, is limited to errbuf_size. The yield of the func-
5890 tion is the size of buffer needed to hold the whole message.
5891
5892
5893 MEMORY USAGE
5894
5895 Compiling a regular expression causes memory to be allocated and asso-
5896 ciated with the preg structure. The function regfree() frees all such
5897 memory, after which preg may no longer be used as a compiled expres-
5898 sion.
5899
5900
5901 AUTHOR
5902
5903 Philip Hazel
5904 University Computing Service
5905 Cambridge CB2 3QH, England.
5906
5907
5908 REVISION
5909
5910 Last updated: 06 March 2007
5911 Copyright (c) 1997-2007 University of Cambridge.
5912 ------------------------------------------------------------------------------
5913
5914
5915 PCRECPP(3) PCRECPP(3)
5916
5917
5918 NAME
5919 PCRE - Perl-compatible regular expressions.
5920
5921
5922 SYNOPSIS OF C++ WRAPPER
5923
5924 #include <pcrecpp.h>
5925
5926
5927 DESCRIPTION
5928
5929 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5930 functionality was added by Giuseppe Maxia. This brief man page was con-
5931 structed from the notes in the pcrecpp.h file, which should be con-
5932 sulted for further details.
5933
5934
5935 MATCHING INTERFACE
5936
5937 The "FullMatch" operation checks that supplied text matches a supplied
5938 pattern exactly. If pointer arguments are supplied, it copies matched
5939 sub-strings that match sub-patterns into them.
5940
5941 Example: successful match
5942 pcrecpp::RE re("h.*o");
5943 re.FullMatch("hello");
5944
5945 Example: unsuccessful match (requires full match):
5946 pcrecpp::RE re("e");
5947 !re.FullMatch("hello");
5948
5949 Example: creating a temporary RE object:
5950 pcrecpp::RE("h.*o").FullMatch("hello");
5951
5952 You can pass in a "const char*" or a "string" for "text". The examples
5953 below tend to use a const char*. You can, as in the different examples
5954 above, store the RE object explicitly in a variable or use a temporary
5955 RE object. The examples below use one mode or the other arbitrarily.
5956 Either could correctly be used for any of these examples.
5957
5958 You must supply extra pointer arguments to extract matched subpieces.
5959
5960 Example: extracts "ruby" into "s" and 1234 into "i"
5961 int i;
5962 string s;
5963 pcrecpp::RE re("(\\w+):(\\d+)");
5964 re.FullMatch("ruby:1234", &s, &i);
5965
5966 Example: does not try to extract any extra sub-patterns
5967 re.FullMatch("ruby:1234", &s);
5968
5969 Example: does not try to extract into NULL
5970 re.FullMatch("ruby:1234", NULL, &i);
5971
5972 Example: integer overflow causes failure
5973 !re.FullMatch("ruby:1234567891234", NULL, &i);
5974
5975 Example: fails because there aren't enough sub-patterns:
5976 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5977
5978 Example: fails because string cannot be stored in integer
5979 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5980
5981 The provided pointer arguments can be pointers to any scalar numeric
5982 type, or one of:
5983
5984 string (matched piece is copied to string)
5985 StringPiece (StringPiece is mutated to point to matched piece)
5986 T (where "bool T::ParseFrom(const char*, int)" exists)
5987 NULL (the corresponding matched sub-pattern is not copied)
5988
5989 The function returns true iff all of the following conditions are sat-
5990 isfied:
5991
5992 a. "text" matches "pattern" exactly;
5993
5994 b. The number of matched sub-patterns is >= number of supplied
5995 pointers;
5996
5997 c. The "i"th argument has a suitable type for holding the
5998 string captured as the "i"th sub-pattern. If you pass in
5999 NULL for the "i"th argument, or pass fewer arguments than
6000 number of sub-patterns, "i"th captured sub-pattern is
6001 ignored.
6002
6003 CAVEAT: An optional sub-pattern that does not exist in the matched
6004 string is assigned the empty string. Therefore, the following will
6005 return false (because the empty string is not a valid number):
6006
6007 int number;
6008 pcrecpp::RE::FullMatch("abc", "[a-z]+(\\d+)?", &number);
6009
6010 The matching interface supports at most 16 arguments per call. If you
6011 need more, consider using the more general interface
6012 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
6013
6014
6015 QUOTING METACHARACTERS
6016
6017 You can use the "QuoteMeta" operation to insert backslashes before all
6018 potentially meaningful characters in a string. The returned string,
6019 used as a regular expression, will exactly match the original string.
6020
6021 Example:
6022 string quoted = RE::QuoteMeta(unquoted);
6023
6024 Note that it's legal to escape a character even if it has no special
6025 meaning in a regular expression -- so this function does that. (This
6026 also makes it identical to the perl function of the same name; see
6027 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
6028 "1\.5\-2\.0\?".
6029
6030
6031 PARTIAL MATCHES
6032
6033 You can use the "PartialMatch" operation when you want the pattern to
6034 match any substring of the text.
6035
6036 Example: simple search for a string:
6037 pcrecpp::RE("ell").PartialMatch("hello");
6038
6039 Example: find first number in a string:
6040 int number;
6041 pcrecpp::RE re("(\\d+)");
6042 re.PartialMatch("x*100 + 20", &number);
6043 assert(number == 100);
6044
6045
6046 UTF-8 AND THE MATCHING INTERFACE
6047
6048 By default, pattern and text are plain text, one byte per character.
6049 The UTF8 flag, passed to the constructor, causes both pattern and
6050 string to be treated as UTF-8 text, still a byte stream but potentially
6051 multiple bytes per character. In practice, the text is likelier to be
6052 UTF-8 than the pattern, but the match returned may depend on the UTF8
6053 flag, so always use it when matching UTF8 text. For example, "." will
6054 match one byte normally but with UTF8 set may match up to three bytes
6055 of a multi-byte character.
6056
6057 Example:
6058 pcrecpp::RE_Options options;
6059 options.set_utf8();
6060 pcrecpp::RE re(utf8_pattern, options);
6061 re.FullMatch(utf8_string);
6062
6063 Example: using the convenience function UTF8():
6064 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
6065 re.FullMatch(utf8_string);
6066
6067 NOTE: The UTF8 flag is ignored if pcre was not configured with the
6068 --enable-utf8 flag.
6069
6070
6071 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
6072
6073 PCRE defines some modifiers to change the behavior of the regular
6074 expression engine. The C++ wrapper defines an auxiliary class,
6075 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
6076 rently, the following modifiers are supported:
6077
6078 modifier description Perl corresponding
6079
6080 PCRE_CASELESS case insensitive match /i
6081 PCRE_MULTILINE multiple lines match /m
6082 PCRE_DOTALL dot matches newlines /s
6083 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
6084 PCRE_EXTRA strict escape parsing N/A
6085 PCRE_EXTENDED ignore whitespaces /x
6086 PCRE_UTF8 handles UTF8 chars built-in
6087 PCRE_UNGREEDY reverses * and *? N/A
6088 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
6089
6090 (*) Both Perl and PCRE allow non capturing parentheses by means of the
6091 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
6092 ture, while (ab|cd) does.
6093
6094 For a full account on how each modifier works, please check the PCRE
6095 API reference page.
6096
6097 For each modifier, there are two member functions whose