ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 535 - (show annotations)
Thu Jun 3 19:18:24 2010 UTC (10 years, 9 months ago) by ph10
File MIME type: text/plain
File size: 353677 byte(s)
Error occurred while calculating annotation data.
Prepare for release candidate.
1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. Neither has
6 the pcredemo program. There are separate text files for the pcregrep and
7 pcretest commands.
8 -----------------------------------------------------------------------------
11 PCRE(3) PCRE(3)
15 PCRE - Perl-compatible regular expressions
20 The PCRE library is a set of functions that implement regular expres-
21 sion pattern matching using the same syntax and semantics as Perl, with
22 just a few differences. Some features that appeared in Python and PCRE
23 before they appeared in Perl are also available using the Python syn-
24 tax, there is some support for one or two .NET and Oniguruma syntax
25 items, and there is an option for requesting some minor changes that
26 give better JavaScript compatibility.
28 The current implementation of PCRE corresponds approximately with Perl
29 5.10/5.11, including support for UTF-8 encoded strings and Unicode gen-
30 eral category properties. However, UTF-8 and Unicode support has to be
31 explicitly enabled; it is not the default. The Unicode tables corre-
32 spond to Unicode release 5.2.0.
34 In addition to the Perl-compatible matching function, PCRE contains an
35 alternative function that matches the same compiled patterns in a dif-
36 ferent way. In certain circumstances, the alternative function has some
37 advantages. For a discussion of the two matching algorithms, see the
38 pcrematching page.
40 PCRE is written in C and released as a C library. A number of people
41 have written wrappers and interfaces of various kinds. In particular,
42 Google Inc. have provided a comprehensive C++ wrapper. This is now
43 included as part of the PCRE distribution. The pcrecpp page has details
44 of this interface. Other people's contributions can be found in the
45 Contrib directory at the primary FTP site, which is:
47 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
49 Details of exactly which Perl regular expression features are and are
50 not supported by PCRE are given in separate documents. See the pcrepat-
51 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
52 page.
54 Some features of PCRE can be included, excluded, or changed when the
55 library is built. The pcre_config() function makes it possible for a
56 client to discover which features are available. The features them-
57 selves are described in the pcrebuild page. Documentation about build-
58 ing PCRE for various operating systems can be found in the README and
59 NON-UNIX-USE files in the source distribution.
61 The library contains a number of undocumented internal functions and
62 data tables that are used by more than one of the exported external
63 functions, but which are not intended for use by external callers.
64 Their names all begin with "_pcre_", which hopefully will not provoke
65 any name clashes. In some environments, it is possible to control which
66 external symbols are exported when a shared library is built, and in
67 these cases the undocumented symbols are not exported.
72 The user documentation for PCRE comprises a number of different sec-
73 tions. In the "man" format, each of these is a separate "man page". In
74 the HTML format, each is a separate page, linked from the index page.
75 In the plain text format, all the sections, except the pcredemo sec-
76 tion, are concatenated, for ease of searching. The sections are as fol-
77 lows:
79 pcre this document
80 pcre-config show PCRE installation configuration information
81 pcreapi details of PCRE's native C API
82 pcrebuild options for building PCRE
83 pcrecallout details of the callout feature
84 pcrecompat discussion of Perl compatibility
85 pcrecpp details of the C++ wrapper
86 pcredemo a demonstration C program that uses PCRE
87 pcregrep description of the pcregrep command
88 pcrematching discussion of the two matching algorithms
89 pcrepartial details of the partial matching facility
90 pcrepattern syntax and semantics of supported
91 regular expressions
92 pcreperform discussion of performance issues
93 pcreposix the POSIX-compatible C API
94 pcreprecompile details of saving and re-using precompiled patterns
95 pcresample discussion of the pcredemo program
96 pcrestack discussion of stack usage
97 pcresyntax quick syntax reference
98 pcretest description of the pcretest testing command
100 In addition, in the "man" and HTML formats, there is a short page for
101 each C library function, listing its arguments and results.
106 There are some size limitations in PCRE but it is hoped that they will
107 never in practice be relevant.
109 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
110 is compiled with the default internal linkage size of 2. If you want to
111 process regular expressions that are truly enormous, you can compile
112 PCRE with an internal linkage size of 3 or 4 (see the README file in
113 the source distribution and the pcrebuild documentation for details).
114 In these cases the limit is substantially larger. However, the speed
115 of execution is slower.
117 All values in repeating quantifiers must be less than 65536.
119 There is no limit to the number of parenthesized subpatterns, but there
120 can be no more than 65535 capturing subpatterns.
122 The maximum length of name for a named subpattern is 32 characters, and
123 the maximum number of named subpatterns is 10000.
125 The maximum length of a subject string is the largest positive number
126 that an integer variable can hold. However, when using the traditional
127 matching function, PCRE uses recursion to handle subpatterns and indef-
128 inite repetition. This means that the available stack space may limit
129 the size of a subject string that can be processed by certain patterns.
130 For a discussion of stack issues, see the pcrestack documentation.
135 From release 3.3, PCRE has had some support for character strings
136 encoded in the UTF-8 format. For release 4.0 this was greatly extended
137 to cover most common requirements, and in release 5.0 additional sup-
138 port for Unicode general category properties was added.
140 In order process UTF-8 strings, you must build PCRE to include UTF-8
141 support in the code, and, in addition, you must call pcre_compile()
142 with the PCRE_UTF8 option flag, or the pattern must start with the
143 sequence (*UTF8). When either of these is the case, both the pattern
144 and any subject strings that are matched against it are treated as
145 UTF-8 strings instead of strings of 1-byte characters.
147 If you compile PCRE with UTF-8 support, but do not use it at run time,
148 the library will be a bit bigger, but the additional run time overhead
149 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
150 very big.
152 If PCRE is built with Unicode character property support (which implies
153 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
154 ported. The available properties that can be tested are limited to the
155 general category properties such as Lu for an upper case letter or Nd
156 for a decimal number, the Unicode script names such as Arabic or Han,
157 and the derived properties Any and L&. A full list is given in the
158 pcrepattern documentation. Only the short names for properties are sup-
159 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
160 ter}, is not supported. Furthermore, in Perl, many properties may
161 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
162 does not support this.
164 Validity of UTF-8 strings
166 When you set the PCRE_UTF8 flag, the strings passed as patterns and
167 subjects are (by default) checked for validity on entry to the relevant
168 functions. From release 7.3 of PCRE, the check is according the rules
169 of RFC 3629, which are themselves derived from the Unicode specifica-
170 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
171 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
172 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
173 to U+DFFF.
175 The excluded code points are the "Low Surrogate Area" of Unicode, of
176 which the Unicode Standard says this: "The Low Surrogate Area does not
177 contain any character assignments, consequently no character code
178 charts or namelists are provided for this area. Surrogates are reserved
179 for use with UTF-16 and then must be used in pairs." The code points
180 that are encoded by UTF-16 pairs are available as independent code
181 points in the UTF-8 encoding. (In other words, the whole surrogate
182 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
184 If an invalid UTF-8 string is passed to PCRE, an error return
185 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
186 that your strings are valid, and therefore want to skip these checks in
187 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
188 compile time or at run time, PCRE assumes that the pattern or subject
189 it is given (respectively) contains only valid UTF-8 codes. In this
190 case, it does not diagnose an invalid UTF-8 string.
192 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
193 what happens depends on why the string is invalid. If the string con-
194 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
195 string of characters in the range 0 to 0x7FFFFFFF. In other words,
196 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
197 strings according to the more liberal rules of RFC 2279. However, if
198 the string does not even conform to RFC 2279, the result is undefined.
199 Your program may crash.
201 If you want to process strings of values in the full range 0 to
202 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
203 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
204 this situation, you will have to apply your own validity check.
206 General comments about UTF-8 mode
208 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
209 two-byte UTF-8 character if the value is greater than 127.
211 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
212 characters for values greater than \177.
214 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
215 vidual bytes, for example: \x{100}{3}.
217 4. The dot metacharacter matches one UTF-8 character instead of a sin-
218 gle byte.
220 5. The escape sequence \C can be used to match a single byte in UTF-8
221 mode, but its use can lead to some strange effects. This facility is
222 not available in the alternative matching function, pcre_dfa_exec().
224 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
225 test characters of any code value, but, by default, the characters that
226 PCRE recognizes as digits, spaces, or word characters remain the same
227 set as before, all with values less than 256. This remains true even
228 when PCRE is built to include Unicode property support, because to do
229 otherwise would slow down PCRE in many common cases. Note that this
230 also applies to \b, because it is defined in terms of \w and \W. If you
231 really want to test for a wider sense of, say, "digit", you can use
232 explicit Unicode property tests such as \p{Nd}. Alternatively, if you
233 set the PCRE_UCP option, the way that the character escapes work is
234 changed so that Unicode properties are used to determine which charac-
235 ters match. There are more details in the section on generic character
236 types in the pcrepattern documentation.
238 7. Similarly, characters that match the POSIX named character classes
239 are all low-valued characters, unless the PCRE_UCP option is set.
241 8. However, the Perl 5.10 horizontal and vertical whitespace matching
242 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
243 acters, whether or not PCRE_UCP is set.
245 9. Case-insensitive matching applies only to characters whose values
246 are less than 128, unless PCRE is built with Unicode property support.
247 Even when Unicode property support is available, PCRE still uses its
248 own character tables when checking the case of low-valued characters,
249 so as not to degrade performance. The Unicode property information is
250 used only for characters with higher values. Even when Unicode property
251 support is available, PCRE supports case-insensitive matching only when
252 there is a one-to-one mapping between a letter's cases. There are a
253 small number of many-to-one mappings in Unicode; these are not sup-
254 ported by PCRE.
259 Philip Hazel
260 University Computing Service
261 Cambridge CB2 3QH, England.
263 Putting an actual email address here seems to have been a spam magnet,
264 so I've taken it away. If you want to email me, use my two initials,
265 followed by the two digits 10, at the domain cam.ac.uk.
270 Last updated: 12 May 2010
271 Copyright (c) 1997-2010 University of Cambridge.
272 ------------------------------------------------------------------------------
278 NAME
279 PCRE - Perl-compatible regular expressions
284 This document describes the optional features of PCRE that can be
285 selected when the library is compiled. It assumes use of the configure
286 script, where the optional features are selected or deselected by pro-
287 viding options to configure before running the make command. However,
288 the same options can be selected in both Unix-like and non-Unix-like
289 environments using the GUI facility of cmake-gui if you are using CMake
290 instead of configure to build PCRE.
292 There is a lot more information about building PCRE in non-Unix-like
293 environments in the file called NON_UNIX_USE, which is part of the PCRE
294 distribution. You should consult this file as well as the README file
295 if you are building in a non-Unix-like environment.
297 The complete list of options for configure (which includes the standard
298 ones such as the selection of the installation directory) can be
299 obtained by running
301 ./configure --help
303 The following sections include descriptions of options whose names
304 begin with --enable or --disable. These settings specify changes to the
305 defaults for the configure command. Because of the way that configure
306 works, --enable and --disable always come in pairs, so the complemen-
307 tary option always exists as well, but as it specifies the default, it
308 is not described.
313 By default, the configure script will search for a C++ compiler and C++
314 header files. If it finds them, it automatically builds the C++ wrapper
315 library for PCRE. You can disable this by adding
317 --disable-cpp
319 to the configure command.
324 To build PCRE with support for UTF-8 Unicode character strings, add
326 --enable-utf8
328 to the configure command. Of itself, this does not make PCRE treat
329 strings as UTF-8. As well as compiling PCRE with this option, you also
330 have have to set the PCRE_UTF8 option when you call the pcre_compile()
331 or pcre_compile2() functions.
333 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
334 expects its input to be either ASCII or UTF-8 (depending on the runtime
335 option). It is not possible to support both EBCDIC and UTF-8 codes in
336 the same version of the library. Consequently, --enable-utf8 and
337 --enable-ebcdic are mutually exclusive.
342 UTF-8 support allows PCRE to process character values greater than 255
343 in the strings that it handles. On its own, however, it does not pro-
344 vide any facilities for accessing the properties of such characters. If
345 you want to be able to use the pattern escapes \P, \p, and \X, which
346 refer to Unicode character properties, you must add
348 --enable-unicode-properties
350 to the configure command. This implies UTF-8 support, even if you have
351 not explicitly requested it.
353 Including Unicode property support adds around 30K of tables to the
354 PCRE library. Only the general category properties such as Lu and Nd
355 are supported. Details are given in the pcrepattern documentation.
360 By default, PCRE interprets the linefeed (LF) character as indicating
361 the end of a line. This is the normal newline character on Unix-like
362 systems. You can compile PCRE to use carriage return (CR) instead, by
363 adding
365 --enable-newline-is-cr
367 to the configure command. There is also a --enable-newline-is-lf
368 option, which explicitly specifies linefeed as the newline character.
370 Alternatively, you can specify that line endings are to be indicated by
371 the two character sequence CRLF. If you want this, add
373 --enable-newline-is-crlf
375 to the configure command. There is a fourth option, specified by
377 --enable-newline-is-anycrlf
379 which causes PCRE to recognize any of the three sequences CR, LF, or
380 CRLF as indicating a line ending. Finally, a fifth option, specified by
382 --enable-newline-is-any
384 causes PCRE to recognize any Unicode newline sequence.
386 Whatever line ending convention is selected when PCRE is built can be
387 overridden when the library functions are called. At build time it is
388 conventional to use the standard for your operating system.
393 By default, the sequence \R in a pattern matches any Unicode newline
394 sequence, whatever has been selected as the line ending sequence. If
395 you specify
397 --enable-bsr-anycrlf
399 the default is changed so that \R matches only CR, LF, or CRLF. What-
400 ever is selected when PCRE is built can be overridden when the library
401 functions are called.
406 The PCRE building process uses libtool to build both shared and static
407 Unix libraries by default. You can suppress one of these by adding one
408 of
410 --disable-shared
411 --disable-static
413 to the configure command, as required.
418 When PCRE is called through the POSIX interface (see the pcreposix doc-
419 umentation), additional working storage is required for holding the
420 pointers to capturing substrings, because PCRE requires three integers
421 per substring, whereas the POSIX interface provides only two. If the
422 number of expected substrings is small, the wrapper function uses space
423 on the stack, because this is faster than using malloc() for each call.
424 The default threshold above which the stack is no longer used is 10; it
425 can be changed by adding a setting such as
427 --with-posix-malloc-threshold=20
429 to the configure command.
434 Within a compiled pattern, offset values are used to point from one
435 part to another (for example, from an opening parenthesis to an alter-
436 nation metacharacter). By default, two-byte values are used for these
437 offsets, leading to a maximum size for a compiled pattern of around
438 64K. This is sufficient to handle all but the most gigantic patterns.
439 Nevertheless, some people do want to process truyl enormous patterns,
440 so it is possible to compile PCRE to use three-byte or four-byte off-
441 sets by adding a setting such as
443 --with-link-size=3
445 to the configure command. The value given must be 2, 3, or 4. Using
446 longer offsets slows down the operation of PCRE because it has to load
447 additional bytes when handling them.
452 When matching with the pcre_exec() function, PCRE implements backtrack-
453 ing by making recursive calls to an internal function called match().
454 In environments where the size of the stack is limited, this can se-
455 verely limit PCRE's operation. (The Unix environment does not usually
456 suffer from this problem, but it may sometimes be necessary to increase
457 the maximum stack size. There is a discussion in the pcrestack docu-
458 mentation.) An alternative approach to recursion that uses memory from
459 the heap to remember data, instead of using recursive function calls,
460 has been implemented to work round the problem of limited stack size.
461 If you want to build a version of PCRE that works this way, add
463 --disable-stack-for-recursion
465 to the configure command. With this configuration, PCRE will use the
466 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
467 ment functions. By default these point to malloc() and free(), but you
468 can replace the pointers so that your own functions are used instead.
470 Separate functions are provided rather than using pcre_malloc and
471 pcre_free because the usage is very predictable: the block sizes
472 requested are always the same, and the blocks are always freed in
473 reverse order. A calling program might be able to implement optimized
474 functions that perform better than malloc() and free(). PCRE runs
475 noticeably more slowly when built in this way. This option affects only
476 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
481 Internally, PCRE has a function called match(), which it calls repeat-
482 edly (sometimes recursively) when matching a pattern with the
483 pcre_exec() function. By controlling the maximum number of times this
484 function may be called during a single matching operation, a limit can
485 be placed on the resources used by a single call to pcre_exec(). The
486 limit can be changed at run time, as described in the pcreapi documen-
487 tation. The default is 10 million, but this can be changed by adding a
488 setting such as
490 --with-match-limit=500000
492 to the configure command. This setting has no effect on the
493 pcre_dfa_exec() matching function.
495 In some environments it is desirable to limit the depth of recursive
496 calls of match() more strictly than the total number of calls, in order
497 to restrict the maximum amount of stack (or heap, if --disable-stack-
498 for-recursion is specified) that is used. A second limit controls this;
499 it defaults to the value that is set for --with-match-limit, which
500 imposes no additional constraints. However, you can set a lower limit
501 by adding, for example,
503 --with-match-limit-recursion=10000
505 to the configure command. This value can also be overridden at run
506 time.
511 PCRE uses fixed tables for processing characters whose code values are
512 less than 256. By default, PCRE is built with a set of tables that are
513 distributed in the file pcre_chartables.c.dist. These tables are for
514 ASCII codes only. If you add
516 --enable-rebuild-chartables
518 to the configure command, the distributed tables are no longer used.
519 Instead, a program called dftables is compiled and run. This outputs
520 the source for new set of tables, created in the default locale of your
521 C runtime system. (This method of replacing the tables does not work if
522 you are cross compiling, because dftables is run on the local host. If
523 you need to create alternative tables when cross compiling, you will
524 have to do so "by hand".)
529 PCRE assumes by default that it will run in an environment where the
530 character code is ASCII (or Unicode, which is a superset of ASCII).
531 This is the case for most computer operating systems. PCRE can, how-
532 ever, be compiled to run in an EBCDIC environment by adding
534 --enable-ebcdic
536 to the configure command. This setting implies --enable-rebuild-charta-
537 bles. You should only use it if you know that you are in an EBCDIC
538 environment (for example, an IBM mainframe operating system). The
539 --enable-ebcdic option is incompatible with --enable-utf8.
544 By default, pcregrep reads all files as plain text. You can build it so
545 that it recognizes files whose names end in .gz or .bz2, and reads them
546 with libz or libbz2, respectively, by adding one or both of
548 --enable-pcregrep-libz
549 --enable-pcregrep-libbz2
551 to the configure command. These options naturally require that the rel-
552 evant libraries are installed on your system. Configuration will fail
553 if they are not.
558 If you add
560 --enable-pcretest-libreadline
562 to the configure command, pcretest is linked with the libreadline
563 library, and when its input is from a terminal, it reads it using the
564 readline() function. This provides line-editing and history facilities.
565 Note that libreadline is GPL-licensed, so if you distribute a binary of
566 pcretest linked in this way, there may be licensing issues.
568 Setting this option causes the -lreadline option to be added to the
569 pcretest build. In many operating environments with a sytem-installed
570 libreadline this is sufficient. However, in some environments (e.g. if
571 an unmodified distribution version of readline is in use), some extra
572 configuration may be necessary. The INSTALL file for libreadline says
573 this:
575 "Readline uses the termcap functions, but does not link with the
576 termcap or curses library itself, allowing applications which link
577 with readline the to choose an appropriate library."
579 If your environment has not been set up so that an appropriate library
580 is automatically included, you may need to add something like
582 LIBS="-ncurses"
584 immediately before the configure command.
589 pcreapi(3), pcre_config(3).
594 Philip Hazel
595 University Computing Service
596 Cambridge CB2 3QH, England.
601 Last updated: 29 September 2009
602 Copyright (c) 1997-2009 University of Cambridge.
603 ------------------------------------------------------------------------------
609 NAME
610 PCRE - Perl-compatible regular expressions
615 This document describes the two different algorithms that are available
616 in PCRE for matching a compiled regular expression against a given sub-
617 ject string. The "standard" algorithm is the one provided by the
618 pcre_exec() function. This works in the same was as Perl's matching
619 function, and provides a Perl-compatible matching operation.
621 An alternative algorithm is provided by the pcre_dfa_exec() function;
622 this operates in a different way, and is not Perl-compatible. It has
623 advantages and disadvantages compared with the standard algorithm, and
624 these are described below.
626 When there is only one possible way in which a given subject string can
627 match a pattern, the two algorithms give the same answer. A difference
628 arises, however, when there are multiple possibilities. For example, if
629 the pattern
631 ^<.*>
633 is matched against the string
635 <something> <something else> <something further>
637 there are three possible answers. The standard algorithm finds only one
638 of them, whereas the alternative algorithm finds all three.
643 The set of strings that are matched by a regular expression can be rep-
644 resented as a tree structure. An unlimited repetition in the pattern
645 makes the tree of infinite size, but it is still a tree. Matching the
646 pattern to a given subject string (from a given starting point) can be
647 thought of as a search of the tree. There are two ways to search a
648 tree: depth-first and breadth-first, and these correspond to the two
649 matching algorithms provided by PCRE.
654 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
655 sions", the standard algorithm is an "NFA algorithm". It conducts a
656 depth-first search of the pattern tree. That is, it proceeds along a
657 single path through the tree, checking that the subject matches what is
658 required. When there is a mismatch, the algorithm tries any alterna-
659 tives at the current point, and if they all fail, it backs up to the
660 previous branch point in the tree, and tries the next alternative
661 branch at that level. This often involves backing up (moving to the
662 left) in the subject string as well. The order in which repetition
663 branches are tried is controlled by the greedy or ungreedy nature of
664 the quantifier.
666 If a leaf node is reached, a matching string has been found, and at
667 that point the algorithm stops. Thus, if there is more than one possi-
668 ble match, this algorithm returns the first one that it finds. Whether
669 this is the shortest, the longest, or some intermediate length depends
670 on the way the greedy and ungreedy repetition quantifiers are specified
671 in the pattern.
673 Because it ends up with a single path through the tree, it is rela-
674 tively straightforward for this algorithm to keep track of the sub-
675 strings that are matched by portions of the pattern in parentheses.
676 This provides support for capturing parentheses and back references.
681 This algorithm conducts a breadth-first search of the tree. Starting
682 from the first matching point in the subject, it scans the subject
683 string from left to right, once, character by character, and as it does
684 this, it remembers all the paths through the tree that represent valid
685 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
686 though it is not implemented as a traditional finite state machine (it
687 keeps multiple states active simultaneously).
689 Although the general principle of this matching algorithm is that it
690 scans the subject string only once, without backtracking, there is one
691 exception: when a lookaround assertion is encountered, the characters
692 following or preceding the current point have to be independently
693 inspected.
695 The scan continues until either the end of the subject is reached, or
696 there are no more unterminated paths. At this point, terminated paths
697 represent the different matching possibilities (if there are none, the
698 match has failed). Thus, if there is more than one possible match,
699 this algorithm finds all of them, and in particular, it finds the long-
700 est. There is an option to stop the algorithm after the first match
701 (which is necessarily the shortest) is found.
703 Note that all the matches that are found start at the same point in the
704 subject. If the pattern
706 cat(er(pillar)?)
708 is matched against the string "the caterpillar catchment", the result
709 will be the three strings "cat", "cater", and "caterpillar" that start
710 at the fourth character of the subject. The algorithm does not automat-
711 ically move on to find matches that start at later positions.
713 There are a number of features of PCRE regular expressions that are not
714 supported by the alternative matching algorithm. They are as follows:
716 1. Because the algorithm finds all possible matches, the greedy or
717 ungreedy nature of repetition quantifiers is not relevant. Greedy and
718 ungreedy quantifiers are treated in exactly the same way. However, pos-
719 sessive quantifiers can make a difference when what follows could also
720 match what is quantified, for example in a pattern like this:
722 ^a++\w!
724 This pattern matches "aaab!" but not "aaa!", which would be matched by
725 a non-possessive quantifier. Similarly, if an atomic group is present,
726 it is matched as if it were a standalone pattern at the current point,
727 and the longest match is then "locked in" for the rest of the overall
728 pattern.
730 2. When dealing with multiple paths through the tree simultaneously, it
731 is not straightforward to keep track of captured substrings for the
732 different matching possibilities, and PCRE's implementation of this
733 algorithm does not attempt to do this. This means that no captured sub-
734 strings are available.
736 3. Because no substrings are captured, back references within the pat-
737 tern are not supported, and cause errors if encountered.
739 4. For the same reason, conditional expressions that use a backrefer-
740 ence as the condition or test for a specific group recursion are not
741 supported.
743 5. Because many paths through the tree may be active, the \K escape
744 sequence, which resets the start of the match when encountered (but may
745 be on some paths and not on others), is not supported. It causes an
746 error if encountered.
748 6. Callouts are supported, but the value of the capture_top field is
749 always 1, and the value of the capture_last field is always -1.
751 7. The \C escape sequence, which (in the standard algorithm) matches a
752 single byte, even in UTF-8 mode, is not supported because the alterna-
753 tive algorithm moves through the subject string one character at a
754 time, for all active paths through the tree.
756 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
757 are not supported. (*FAIL) is supported, and behaves like a failing
758 negative assertion.
763 Using the alternative matching algorithm provides the following advan-
764 tages:
766 1. All possible matches (at a single point in the subject) are automat-
767 ically found, and in particular, the longest match is found. To find
768 more than one match using the standard algorithm, you have to do kludgy
769 things with callouts.
771 2. Because the alternative algorithm scans the subject string just
772 once, and never needs to backtrack, it is possible to pass very long
773 subject strings to the matching function in several pieces, checking
774 for partial matching each time. The pcrepartial documentation gives
775 details of partial matching.
780 The alternative algorithm suffers from a number of disadvantages:
782 1. It is substantially slower than the standard algorithm. This is
783 partly because it has to search for all possible matches, but is also
784 because it is less susceptible to optimization.
786 2. Capturing parentheses and back references are not supported.
788 3. Although atomic groups are supported, their use does not provide the
789 performance advantage that it does for the standard algorithm.
794 Philip Hazel
795 University Computing Service
796 Cambridge CB2 3QH, England.
801 Last updated: 29 September 2009
802 Copyright (c) 1997-2009 University of Cambridge.
803 ------------------------------------------------------------------------------
809 NAME
810 PCRE - Perl-compatible regular expressions
815 #include <pcre.h>
817 pcre *pcre_compile(const char *pattern, int options,
818 const char **errptr, int *erroffset,
819 const unsigned char *tableptr);
821 pcre *pcre_compile2(const char *pattern, int options,
822 int *errorcodeptr,
823 const char **errptr, int *erroffset,
824 const unsigned char *tableptr);
826 pcre_extra *pcre_study(const pcre *code, int options,
827 const char **errptr);
829 int pcre_exec(const pcre *code, const pcre_extra *extra,
830 const char *subject, int length, int startoffset,
831 int options, int *ovector, int ovecsize);
833 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
834 const char *subject, int length, int startoffset,
835 int options, int *ovector, int ovecsize,
836 int *workspace, int wscount);
838 int pcre_copy_named_substring(const pcre *code,
839 const char *subject, int *ovector,
840 int stringcount, const char *stringname,
841 char *buffer, int buffersize);
843 int pcre_copy_substring(const char *subject, int *ovector,
844 int stringcount, int stringnumber, char *buffer,
845 int buffersize);
847 int pcre_get_named_substring(const pcre *code,
848 const char *subject, int *ovector,
849 int stringcount, const char *stringname,
850 const char **stringptr);
852 int pcre_get_stringnumber(const pcre *code,
853 const char *name);
855 int pcre_get_stringtable_entries(const pcre *code,
856 const char *name, char **first, char **last);
858 int pcre_get_substring(const char *subject, int *ovector,
859 int stringcount, int stringnumber,
860 const char **stringptr);
862 int pcre_get_substring_list(const char *subject,
863 int *ovector, int stringcount, const char ***listptr);
865 void pcre_free_substring(const char *stringptr);
867 void pcre_free_substring_list(const char **stringptr);
869 const unsigned char *pcre_maketables(void);
871 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
872 int what, void *where);
874 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
876 int pcre_refcount(pcre *code, int adjust);
878 int pcre_config(int what, void *where);
880 char *pcre_version(void);
882 void *(*pcre_malloc)(size_t);
884 void (*pcre_free)(void *);
886 void *(*pcre_stack_malloc)(size_t);
888 void (*pcre_stack_free)(void *);
890 int (*pcre_callout)(pcre_callout_block *);
895 PCRE has its own native API, which is described in this document. There
896 are also some wrapper functions that correspond to the POSIX regular
897 expression API. These are described in the pcreposix documentation.
898 Both of these APIs define a set of C function calls. A C++ wrapper is
899 distributed with PCRE. It is documented in the pcrecpp page.
901 The native API C function prototypes are defined in the header file
902 pcre.h, and on Unix systems the library itself is called libpcre. It
903 can normally be accessed by adding -lpcre to the command for linking an
904 application that uses PCRE. The header file defines the macros
905 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
906 bers for the library. Applications can use these to include support
907 for different releases of PCRE.
909 In a Windows environment, if you want to statically link an application
910 program against a non-dll pcre.a file, you must define PCRE_STATIC
911 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
912 loc() and pcre_free() exported functions will be declared
913 __declspec(dllimport), with unwanted results.
915 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
916 pcre_exec() are used for compiling and matching regular expressions in
917 a Perl-compatible manner. A sample program that demonstrates the sim-
918 plest way of using them is provided in the file called pcredemo.c in
919 the PCRE source distribution. A listing of this program is given in the
920 pcredemo documentation, and the pcresample documentation describes how
921 to compile and run it.
923 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
924 ble, is also provided. This uses a different algorithm for the match-
925 ing. The alternative algorithm finds all possible matches (at a given
926 point in the subject), and scans the subject just once (unless there
927 are lookbehind assertions). However, this algorithm does not return
928 captured substrings. A description of the two matching algorithms and
929 their advantages and disadvantages is given in the pcrematching docu-
930 mentation.
932 In addition to the main compiling and matching functions, there are
933 convenience functions for extracting captured substrings from a subject
934 string that is matched by pcre_exec(). They are:
936 pcre_copy_substring()
937 pcre_copy_named_substring()
938 pcre_get_substring()
939 pcre_get_named_substring()
940 pcre_get_substring_list()
941 pcre_get_stringnumber()
942 pcre_get_stringtable_entries()
944 pcre_free_substring() and pcre_free_substring_list() are also provided,
945 to free the memory used for extracted strings.
947 The function pcre_maketables() is used to build a set of character
948 tables in the current locale for passing to pcre_compile(),
949 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
950 provided for specialist use. Most commonly, no special tables are
951 passed, in which case internal tables that are generated when PCRE is
952 built are used.
954 The function pcre_fullinfo() is used to find out information about a
955 compiled pattern; pcre_info() is an obsolete version that returns only
956 some of the available information, but is retained for backwards com-
957 patibility. The function pcre_version() returns a pointer to a string
958 containing the version of PCRE and its date of release.
960 The function pcre_refcount() maintains a reference count in a data
961 block containing a compiled pattern. This is provided for the benefit
962 of object-oriented applications.
964 The global variables pcre_malloc and pcre_free initially contain the
965 entry points of the standard malloc() and free() functions, respec-
966 tively. PCRE calls the memory management functions via these variables,
967 so a calling program can replace them if it wishes to intercept the
968 calls. This should be done before calling any PCRE functions.
970 The global variables pcre_stack_malloc and pcre_stack_free are also
971 indirections to memory management functions. These special functions
972 are used only when PCRE is compiled to use the heap for remembering
973 data, instead of recursive function calls, when running the pcre_exec()
974 function. See the pcrebuild documentation for details of how to do
975 this. It is a non-standard way of building PCRE, for use in environ-
976 ments that have limited stacks. Because of the greater use of memory
977 management, it runs more slowly. Separate functions are provided so
978 that special-purpose external code can be used for this case. When
979 used, these functions are always called in a stack-like manner (last
980 obtained, first freed), and always for memory blocks of the same size.
981 There is a discussion about PCRE's stack usage in the pcrestack docu-
982 mentation.
984 The global variable pcre_callout initially contains NULL. It can be set
985 by the caller to a "callout" function, which PCRE will then call at
986 specified points during a matching operation. Details are given in the
987 pcrecallout documentation.
992 PCRE supports five different conventions for indicating line breaks in
993 strings: a single CR (carriage return) character, a single LF (line-
994 feed) character, the two-character sequence CRLF, any of the three pre-
995 ceding, or any Unicode newline sequence. The Unicode newline sequences
996 are the three just mentioned, plus the single characters VT (vertical
997 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
998 separator, U+2028), and PS (paragraph separator, U+2029).
1000 Each of the first three conventions is used by at least one operating
1001 system as its standard newline sequence. When PCRE is built, a default
1002 can be specified. The default default is LF, which is the Unix stan-
1003 dard. When PCRE is run, the default can be overridden, either when a
1004 pattern is compiled, or when it is matched.
1006 At compile time, the newline convention can be specified by the options
1007 argument of pcre_compile(), or it can be specified by special text at
1008 the start of the pattern itself; this overrides any other settings. See
1009 the pcrepattern page for details of the special character sequences.
1011 In the PCRE documentation the word "newline" is used to mean "the char-
1012 acter or pair of characters that indicate a line break". The choice of
1013 newline convention affects the handling of the dot, circumflex, and
1014 dollar metacharacters, the handling of #-comments in /x mode, and, when
1015 CRLF is a recognized line ending sequence, the match position advance-
1016 ment for a non-anchored pattern. There is more detail about this in the
1017 section on pcre_exec() options below.
1019 The choice of newline convention does not affect the interpretation of
1020 the \n or \r escape sequences, nor does it affect what \R matches,
1021 which is controlled in a similar way, but by separate options.
1026 The PCRE functions can be used in multi-threading applications, with
1027 the proviso that the memory management functions pointed to by
1028 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1029 callout function pointed to by pcre_callout, are shared by all threads.
1031 The compiled form of a regular expression is not altered during match-
1032 ing, so the same compiled pattern can safely be used by several threads
1033 at once.
1038 The compiled form of a regular expression can be saved and re-used at a
1039 later time, possibly by a different program, and even on a host other
1040 than the one on which it was compiled. Details are given in the
1041 pcreprecompile documentation. However, compiling a regular expression
1042 with one version of PCRE for use with a different version is not guar-
1043 anteed to work and may cause crashes.
1048 int pcre_config(int what, void *where);
1050 The function pcre_config() makes it possible for a PCRE client to dis-
1051 cover which optional features have been compiled into the PCRE library.
1052 The pcrebuild documentation has more details about these optional fea-
1053 tures.
1055 The first argument for pcre_config() is an integer, specifying which
1056 information is required; the second argument is a pointer to a variable
1057 into which the information is placed. The following information is
1058 available:
1062 The output is an integer that is set to one if UTF-8 support is avail-
1063 able; otherwise it is set to zero.
1067 The output is an integer that is set to one if support for Unicode
1068 character properties is available; otherwise it is set to zero.
1072 The output is an integer whose value specifies the default character
1073 sequence that is recognized as meaning "newline". The four values that
1074 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1075 and -1 for ANY. Though they are derived from ASCII, the same values
1076 are returned in EBCDIC environments. The default should normally corre-
1077 spond to the standard sequence for your operating system.
1081 The output is an integer whose value indicates what character sequences
1082 the \R escape sequence matches by default. A value of 0 means that \R
1083 matches any Unicode line ending sequence; a value of 1 means that \R
1084 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1085 tern is compiled or matched.
1089 The output is an integer that contains the number of bytes used for
1090 internal linkage in compiled regular expressions. The value is 2, 3, or
1091 4. Larger values allow larger regular expressions to be compiled, at
1092 the expense of slower matching. The default value of 2 is sufficient
1093 for all but the most massive patterns, since it allows the compiled
1094 pattern to be up to 64K in size.
1098 The output is an integer that contains the threshold above which the
1099 POSIX interface uses malloc() for output vectors. Further details are
1100 given in the pcreposix documentation.
1104 The output is a long integer that gives the default limit for the num-
1105 ber of internal matching function calls in a pcre_exec() execution.
1106 Further details are given with pcre_exec() below.
1110 The output is a long integer that gives the default limit for the depth
1111 of recursion when calling the internal matching function in a
1112 pcre_exec() execution. Further details are given with pcre_exec()
1113 below.
1117 The output is an integer that is set to one if internal recursion when
1118 running pcre_exec() is implemented by recursive function calls that use
1119 the stack to remember their state. This is the usual way that PCRE is
1120 compiled. The output is zero if PCRE was compiled to use blocks of data
1121 on the heap instead of recursive function calls. In this case,
1122 pcre_stack_malloc and pcre_stack_free are called to manage memory
1123 blocks on the heap, thus avoiding the use of the stack.
1128 pcre *pcre_compile(const char *pattern, int options,
1129 const char **errptr, int *erroffset,
1130 const unsigned char *tableptr);
1132 pcre *pcre_compile2(const char *pattern, int options,
1133 int *errorcodeptr,
1134 const char **errptr, int *erroffset,
1135 const unsigned char *tableptr);
1137 Either of the functions pcre_compile() or pcre_compile2() can be called
1138 to compile a pattern into an internal form. The only difference between
1139 the two interfaces is that pcre_compile2() has an additional argument,
1140 errorcodeptr, via which a numerical error code can be returned. To
1141 avoid too much repetition, we refer just to pcre_compile() below, but
1142 the information applies equally to pcre_compile2().
1144 The pattern is a C string terminated by a binary zero, and is passed in
1145 the pattern argument. A pointer to a single block of memory that is
1146 obtained via pcre_malloc is returned. This contains the compiled code
1147 and related data. The pcre type is defined for the returned block; this
1148 is a typedef for a structure whose contents are not externally defined.
1149 It is up to the caller to free the memory (via pcre_free) when it is no
1150 longer required.
1152 Although the compiled code of a PCRE regex is relocatable, that is, it
1153 does not depend on memory location, the complete pcre data block is not
1154 fully relocatable, because it may contain a copy of the tableptr argu-
1155 ment, which is an address (see below).
1157 The options argument contains various bit settings that affect the com-
1158 pilation. It should be zero if no options are required. The available
1159 options are described below. Some of them (in particular, those that
1160 are compatible with Perl, but some others as well) can also be set and
1161 unset from within the pattern (see the detailed description in the
1162 pcrepattern documentation). For those options that can be different in
1163 different parts of the pattern, the contents of the options argument
1164 specifies their settings at the start of compilation and execution. The
1165 PCRE_ANCHORED, PCRE_BSR_xxx, and PCRE_NEWLINE_xxx options can be set at
1166 the time of matching as well as at compile time.
1168 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1169 if compilation of a pattern fails, pcre_compile() returns NULL, and
1170 sets the variable pointed to by errptr to point to a textual error mes-
1171 sage. This is a static string that is part of the library. You must not
1172 try to free it. The byte offset from the start of the pattern to the
1173 character that was being processed when the error was discovered is
1174 placed in the variable pointed to by erroffset, which must not be NULL.
1175 If it is, an immediate error is given. Some errors are not detected
1176 until checks are carried out when the whole pattern has been scanned;
1177 in this case the offset is set to the end of the pattern.
1179 If pcre_compile2() is used instead of pcre_compile(), and the error-
1180 codeptr argument is not NULL, a non-zero error code number is returned
1181 via this argument in the event of an error. This is in addition to the
1182 textual error message. Error codes and messages are listed below.
1184 If the final argument, tableptr, is NULL, PCRE uses a default set of
1185 character tables that are built when PCRE is compiled, using the
1186 default C locale. Otherwise, tableptr must be an address that is the
1187 result of a call to pcre_maketables(). This value is stored with the
1188 compiled pattern, and used again by pcre_exec(), unless another table
1189 pointer is passed to it. For more discussion, see the section on locale
1190 support below.
1192 This code fragment shows a typical straightforward call to pcre_com-
1193 pile():
1195 pcre *re;
1196 const char *error;
1197 int erroffset;
1198 re = pcre_compile(
1199 "^A.*Z", /* the pattern */
1200 0, /* default options */
1201 &error, /* for error message */
1202 &erroffset, /* for error offset */
1203 NULL); /* use default character tables */
1205 The following names for option bits are defined in the pcre.h header
1206 file:
1210 If this bit is set, the pattern is forced to be "anchored", that is, it
1211 is constrained to match only at the first matching point in the string
1212 that is being searched (the "subject string"). This effect can also be
1213 achieved by appropriate constructs in the pattern itself, which is the
1214 only way to do it in Perl.
1218 If this bit is set, pcre_compile() automatically inserts callout items,
1219 all with number 255, before each pattern item. For discussion of the
1220 callout facility, see the pcrecallout documentation.
1225 These options (which are mutually exclusive) control what the \R escape
1226 sequence matches. The choice is either to match only CR, LF, or CRLF,
1227 or to match any Unicode newline sequence. The default is specified when
1228 PCRE is built. It can be overridden from within the pattern, or by set-
1229 ting an option when a compiled pattern is matched.
1233 If this bit is set, letters in the pattern match both upper and lower
1234 case letters. It is equivalent to Perl's /i option, and it can be
1235 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1236 always understands the concept of case for characters whose values are
1237 less than 128, so caseless matching is always possible. For characters
1238 with higher values, the concept of case is supported if PCRE is com-
1239 piled with Unicode property support, but not otherwise. If you want to
1240 use caseless matching for characters 128 and above, you must ensure
1241 that PCRE is compiled with Unicode property support as well as with
1242 UTF-8 support.
1246 If this bit is set, a dollar metacharacter in the pattern matches only
1247 at the end of the subject string. Without this option, a dollar also
1248 matches immediately before a newline at the end of the string (but not
1249 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1250 if PCRE_MULTILINE is set. There is no equivalent to this option in
1251 Perl, and no way to set it within a pattern.
1255 If this bit is set, a dot metacharater in the pattern matches all char-
1256 acters, including those that indicate newline. Without it, a dot does
1257 not match when the current position is at a newline. This option is
1258 equivalent to Perl's /s option, and it can be changed within a pattern
1259 by a (?s) option setting. A negative class such as [^a] always matches
1260 newline characters, independent of the setting of this option.
1264 If this bit is set, names used to identify capturing subpatterns need
1265 not be unique. This can be helpful for certain types of pattern when it
1266 is known that only one instance of the named subpattern can ever be
1267 matched. There are more details of named subpatterns below; see also
1268 the pcrepattern documentation.
1272 If this bit is set, whitespace data characters in the pattern are
1273 totally ignored except when escaped or inside a character class. White-
1274 space does not include the VT character (code 11). In addition, charac-
1275 ters between an unescaped # outside a character class and the next new-
1276 line, inclusive, are also ignored. This is equivalent to Perl's /x
1277 option, and it can be changed within a pattern by a (?x) option set-
1278 ting.
1280 This option makes it possible to include comments inside complicated
1281 patterns. Note, however, that this applies only to data characters.
1282 Whitespace characters may never appear within special character
1283 sequences in a pattern, for example within the sequence (?( which
1284 introduces a conditional subpattern.
1288 This option was invented in order to turn on additional functionality
1289 of PCRE that is incompatible with Perl, but it is currently of very
1290 little use. When set, any backslash in a pattern that is followed by a
1291 letter that has no special meaning causes an error, thus reserving
1292 these combinations for future expansion. By default, as in Perl, a
1293 backslash followed by a letter with no special meaning is treated as a
1294 literal. (Perl can, however, be persuaded to give an error for this, by
1295 running it with the -w option.) There are at present no other features
1296 controlled by this option. It can also be set by a (?X) option setting
1297 within a pattern.
1301 If this option is set, an unanchored pattern is required to match
1302 before or at the first newline in the subject string, though the
1303 matched text may continue over the newline.
1307 If this option is set, PCRE's behaviour is changed in some ways so that
1308 it is compatible with JavaScript rather than Perl. The changes are as
1309 follows:
1311 (1) A lone closing square bracket in a pattern causes a compile-time
1312 error, because this is illegal in JavaScript (by default it is treated
1313 as a data character). Thus, the pattern AB]CD becomes illegal when this
1314 option is set.
1316 (2) At run time, a back reference to an unset subpattern group matches
1317 an empty string (by default this causes the current matching alterna-
1318 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1319 set (assuming it can find an "a" in the subject), whereas it fails by
1320 default, for Perl compatibility.
1324 By default, PCRE treats the subject string as consisting of a single
1325 line of characters (even if it actually contains newlines). The "start
1326 of line" metacharacter (^) matches only at the start of the string,
1327 while the "end of line" metacharacter ($) matches only at the end of
1328 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1329 is set). This is the same as Perl.
1331 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1332 constructs match immediately following or immediately before internal
1333 newlines in the subject string, respectively, as well as at the very
1334 start and end. This is equivalent to Perl's /m option, and it can be
1335 changed within a pattern by a (?m) option setting. If there are no new-
1336 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1337 setting PCRE_MULTILINE has no effect.
1345 These options override the default newline definition that was chosen
1346 when PCRE was built. Setting the first or the second specifies that a
1347 newline is indicated by a single character (CR or LF, respectively).
1348 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1349 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1350 that any of the three preceding sequences should be recognized. Setting
1351 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1352 recognized. The Unicode newline sequences are the three just mentioned,
1353 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1354 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1355 (paragraph separator, U+2029). The last two are recognized only in
1356 UTF-8 mode.
1358 The newline setting in the options word uses three bits that are
1359 treated as a number, giving eight possibilities. Currently only six are
1360 used (default plus the five values above). This means that if you set
1361 more than one newline option, the combination may or may not be sensi-
1362 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1363 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1364 cause an error.
1366 The only time that a line break is specially recognized when compiling
1367 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1368 character class is encountered. This indicates a comment that lasts
1369 until after the next line break sequence. In other circumstances, line
1370 break sequences are treated as literal data, except that in
1371 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1372 and are therefore ignored.
1374 The newline option that is set at compile time becomes the default that
1375 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1379 If this option is set, it disables the use of numbered capturing paren-
1380 theses in the pattern. Any opening parenthesis that is not followed by
1381 ? behaves as if it were followed by ?: but named parentheses can still
1382 be used for capturing (and they acquire numbers in the usual way).
1383 There is no equivalent of this option in Perl.
1387 This option changes the way PCRE processes \b, \d, \s, \w, and some of
1388 the POSIX character classes. By default, only ASCII characters are rec-
1389 ognized, but if PCRE_UCP is set, Unicode properties are used instead to
1390 classify characters. More details are given in the section on generic
1391 character types in the pcrepattern page. If you set PCRE_UCP, matching
1392 one of the items it affects takes much longer. The option is available
1393 only if PCRE has been compiled with Unicode property support.
1397 This option inverts the "greediness" of the quantifiers so that they
1398 are not greedy by default, but become greedy if followed by "?". It is
1399 not compatible with Perl. It can also be set by a (?U) option setting
1400 within the pattern.
1402 PCRE_UTF8
1404 This option causes PCRE to regard both the pattern and the subject as
1405 strings of UTF-8 characters instead of single-byte character strings.
1406 However, it is available only when PCRE is built to include UTF-8 sup-
1407 port. If not, the use of this option provokes an error. Details of how
1408 this option changes the behaviour of PCRE are given in the section on
1409 UTF-8 support in the main pcre page.
1413 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1414 automatically checked. There is a discussion about the validity of
1415 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1416 bytes is found, pcre_compile() returns an error. If you already know
1417 that your pattern is valid, and you want to skip this check for perfor-
1418 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1419 set, the effect of passing an invalid UTF-8 string as a pattern is
1420 undefined. It may cause your program to crash. Note that this option
1421 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1422 UTF-8 validity checking of subject strings.
1427 The following table lists the error codes than may be returned by
1428 pcre_compile2(), along with the error messages that may be returned by
1429 both compiling functions. As PCRE has developed, some error codes have
1430 fallen out of use. To avoid confusion, they have not been re-used.
1432 0 no error
1433 1 \ at end of pattern
1434 2 \c at end of pattern
1435 3 unrecognized character follows \
1436 4 numbers out of order in {} quantifier
1437 5 number too big in {} quantifier
1438 6 missing terminating ] for character class
1439 7 invalid escape sequence in character class
1440 8 range out of order in character class
1441 9 nothing to repeat
1442 10 [this code is not in use]
1443 11 internal error: unexpected repeat
1444 12 unrecognized character after (? or (?-
1445 13 POSIX named classes are supported only within a class
1446 14 missing )
1447 15 reference to non-existent subpattern
1448 16 erroffset passed as NULL
1449 17 unknown option bit(s) set
1450 18 missing ) after comment
1451 19 [this code is not in use]
1452 20 regular expression is too large
1453 21 failed to get memory
1454 22 unmatched parentheses
1455 23 internal error: code overflow
1456 24 unrecognized character after (?<
1457 25 lookbehind assertion is not fixed length
1458 26 malformed number or name after (?(
1459 27 conditional group contains more than two branches
1460 28 assertion expected after (?(
1461 29 (?R or (?[+-]digits must be followed by )
1462 30 unknown POSIX class name
1463 31 POSIX collating elements are not supported
1464 32 this version of PCRE is not compiled with PCRE_UTF8 support
1465 33 [this code is not in use]
1466 34 character value in \x{...} sequence is too large
1467 35 invalid condition (?(0)
1468 36 \C not allowed in lookbehind assertion
1469 37 PCRE does not support \L, \l, \N, \U, or \u
1470 38 number after (?C is > 255
1471 39 closing ) for (?C expected
1472 40 recursive call could loop indefinitely
1473 41 unrecognized character after (?P
1474 42 syntax error in subpattern name (missing terminator)
1475 43 two named subpatterns have the same name
1476 44 invalid UTF-8 string
1477 45 support for \P, \p, and \X has not been compiled
1478 46 malformed \P or \p sequence
1479 47 unknown property name after \P or \p
1480 48 subpattern name is too long (maximum 32 characters)
1481 49 too many named subpatterns (maximum 10000)
1482 50 [this code is not in use]
1483 51 octal value is greater than \377 (not in UTF-8 mode)
1484 52 internal error: overran compiling workspace
1485 53 internal error: previously-checked referenced subpattern not
1486 found
1487 54 DEFINE group contains more than one branch
1488 55 repeating a DEFINE group is not allowed
1489 56 inconsistent NEWLINE options
1490 57 \g is not followed by a braced, angle-bracketed, or quoted
1491 name/number or by a plain number
1492 58 a numbered reference must not be zero
1493 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1494 60 (*VERB) not recognized
1495 61 number is too big
1496 62 subpattern name expected
1497 63 digit expected after (?+
1498 64 ] is an invalid data character in JavaScript compatibility mode
1499 65 different names for subpatterns of the same number are not
1500 allowed
1501 66 (*MARK) must have an argument
1502 67 this version of PCRE is not compiled with PCRE_UCP support
1504 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1505 values may be used if the limits were changed when PCRE was built.
1510 pcre_extra *pcre_study(const pcre *code, int options
1511 const char **errptr);
1513 If a compiled pattern is going to be used several times, it is worth
1514 spending more time analyzing it in order to speed up the time taken for
1515 matching. The function pcre_study() takes a pointer to a compiled pat-
1516 tern as its first argument. If studying the pattern produces additional
1517 information that will help speed up matching, pcre_study() returns a
1518 pointer to a pcre_extra block, in which the study_data field points to
1519 the results of the study.
1521 The returned value from pcre_study() can be passed directly to
1522 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1523 tains other fields that can be set by the caller before the block is
1524 passed; these are described below in the section on matching a pattern.
1526 If studying the pattern does not produce any useful information,
1527 pcre_study() returns NULL. In that circumstance, if the calling program
1528 wants to pass any of the other fields to pcre_exec() or
1529 pcre_dfa_exec(), it must set up its own pcre_extra block.
1531 The second argument of pcre_study() contains option bits. At present,
1532 no options are defined, and this argument should always be zero.
1534 The third argument for pcre_study() is a pointer for an error message.
1535 If studying succeeds (even if no data is returned), the variable it
1536 points to is set to NULL. Otherwise it is set to point to a textual
1537 error message. This is a static string that is part of the library. You
1538 must not try to free it. You should test the error pointer for NULL
1539 after calling pcre_study(), to be sure that it has run successfully.
1541 This is a typical call to pcre_study():
1543 pcre_extra *pe;
1544 pe = pcre_study(
1545 re, /* result of pcre_compile() */
1546 0, /* no options exist */
1547 &error); /* set to NULL or points to a message */
1549 Studying a pattern does two things: first, a lower bound for the length
1550 of subject string that is needed to match the pattern is computed. This
1551 does not mean that there are any strings of that length that match, but
1552 it does guarantee that no shorter strings match. The value is used by
1553 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1554 match strings that are shorter than the lower bound. You can find out
1555 the value in a calling program via the pcre_fullinfo() function.
1557 Studying a pattern is also useful for non-anchored patterns that do not
1558 have a single fixed starting character. A bitmap of possible starting
1559 bytes is created. This speeds up finding a position in the subject at
1560 which to start matching.
1565 PCRE handles caseless matching, and determines whether characters are
1566 letters, digits, or whatever, by reference to a set of tables, indexed
1567 by character value. When running in UTF-8 mode, this applies only to
1568 characters with codes less than 128. By default, higher-valued codes
1569 never match escapes such as \w or \d, but they can be tested with \p if
1570 PCRE is built with Unicode character property support. Alternatively,
1571 the PCRE_UCP option can be set at compile time; this causes \w and
1572 friends to use Unicode property support instead of built-in tables. The
1573 use of locales with Unicode is discouraged. If you are handling charac-
1574 ters with codes greater than 128, you should either use UTF-8 and Uni-
1575 code, or use locales, but not try to mix the two.
1577 PCRE contains an internal set of tables that are used when the final
1578 argument of pcre_compile() is NULL. These are sufficient for many
1579 applications. Normally, the internal tables recognize only ASCII char-
1580 acters. However, when PCRE is built, it is possible to cause the inter-
1581 nal tables to be rebuilt in the default "C" locale of the local system,
1582 which may cause them to be different.
1584 The internal tables can always be overridden by tables supplied by the
1585 application that calls PCRE. These may be created in a different locale
1586 from the default. As more and more applications change to using Uni-
1587 code, the need for this locale support is expected to die away.
1589 External tables are built by calling the pcre_maketables() function,
1590 which has no arguments, in the relevant locale. The result can then be
1591 passed to pcre_compile() or pcre_exec() as often as necessary. For
1592 example, to build and use tables that are appropriate for the French
1593 locale (where accented characters with values greater than 128 are
1594 treated as letters), the following code could be used:
1596 setlocale(LC_CTYPE, "fr_FR");
1597 tables = pcre_maketables();
1598 re = pcre_compile(..., tables);
1600 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1601 if you are using Windows, the name for the French locale is "french".
1603 When pcre_maketables() runs, the tables are built in memory that is
1604 obtained via pcre_malloc. It is the caller's responsibility to ensure
1605 that the memory containing the tables remains available for as long as
1606 it is needed.
1608 The pointer that is passed to pcre_compile() is saved with the compiled
1609 pattern, and the same tables are used via this pointer by pcre_study()
1610 and normally also by pcre_exec(). Thus, by default, for any single pat-
1611 tern, compilation, studying and matching all happen in the same locale,
1612 but different patterns can be compiled in different locales.
1614 It is possible to pass a table pointer or NULL (indicating the use of
1615 the internal tables) to pcre_exec(). Although not intended for this
1616 purpose, this facility could be used to match a pattern in a different
1617 locale from the one in which it was compiled. Passing table pointers at
1618 run time is discussed below in the section on matching a pattern.
1623 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1624 int what, void *where);
1626 The pcre_fullinfo() function returns information about a compiled pat-
1627 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1628 less retained for backwards compability (and is documented below).
1630 The first argument for pcre_fullinfo() is a pointer to the compiled
1631 pattern. The second argument is the result of pcre_study(), or NULL if
1632 the pattern was not studied. The third argument specifies which piece
1633 of information is required, and the fourth argument is a pointer to a
1634 variable to receive the data. The yield of the function is zero for
1635 success, or one of the following negative numbers:
1637 PCRE_ERROR_NULL the argument code was NULL
1638 the argument where was NULL
1639 PCRE_ERROR_BADMAGIC the "magic number" was not found
1640 PCRE_ERROR_BADOPTION the value of what was invalid
1642 The "magic number" is placed at the start of each compiled pattern as
1643 an simple check against passing an arbitrary memory pointer. Here is a
1644 typical call of pcre_fullinfo(), to obtain the length of the compiled
1645 pattern:
1647 int rc;
1648 size_t length;
1649 rc = pcre_fullinfo(
1650 re, /* result of pcre_compile() */
1651 pe, /* result of pcre_study(), or NULL */
1652 PCRE_INFO_SIZE, /* what is required */
1653 &length); /* where to put the data */
1655 The possible values for the third argument are defined in pcre.h, and
1656 are as follows:
1660 Return the number of the highest back reference in the pattern. The
1661 fourth argument should point to an int variable. Zero is returned if
1662 there are no back references.
1666 Return the number of capturing subpatterns in the pattern. The fourth
1667 argument should point to an int variable.
1671 Return a pointer to the internal default character tables within PCRE.
1672 The fourth argument should point to an unsigned char * variable. This
1673 information call is provided for internal use by the pcre_study() func-
1674 tion. External callers can cause PCRE to use its internal tables by
1675 passing a NULL table pointer.
1679 Return information about the first byte of any matched string, for a
1680 non-anchored pattern. The fourth argument should point to an int vari-
1681 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1682 is still recognized for backwards compatibility.)
1684 If there is a fixed first byte, for example, from a pattern such as
1685 (cat|cow|coyote), its value is returned. Otherwise, if either
1687 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1688 branch starts with "^", or
1690 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1691 set (if it were set, the pattern would be anchored),
1693 -1 is returned, indicating that the pattern matches only at the start
1694 of a subject string or after any newline within the string. Otherwise
1695 -2 is returned. For anchored patterns, -2 is returned.
1699 If the pattern was studied, and this resulted in the construction of a
1700 256-bit table indicating a fixed set of bytes for the first byte in any
1701 matching string, a pointer to the table is returned. Otherwise NULL is
1702 returned. The fourth argument should point to an unsigned char * vari-
1703 able.
1707 Return 1 if the pattern contains any explicit matches for CR or LF
1708 characters, otherwise 0. The fourth argument should point to an int
1709 variable. An explicit match is either a literal CR or LF character, or
1710 \r or \n.
1714 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1715 otherwise 0. The fourth argument should point to an int variable. (?J)
1716 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1720 Return the value of the rightmost literal byte that must exist in any
1721 matched string, other than at its start, if such a byte has been
1722 recorded. The fourth argument should point to an int variable. If there
1723 is no such byte, -1 is returned. For anchored patterns, a last literal
1724 byte is recorded only if it follows something of variable length. For
1725 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1726 /^a\dz\d/ the returned value is -1.
1730 If the pattern was studied and a minimum length for matching subject
1731 strings was computed, its value is returned. Otherwise the returned
1732 value is -1. The value is a number of characters, not bytes (this may
1733 be relevant in UTF-8 mode). The fourth argument should point to an int
1734 variable. A non-negative value is a lower bound to the length of any
1735 matching string. There may not be any strings of that length that do
1736 actually match, but every string that does match is at least that long.
1742 PCRE supports the use of named as well as numbered capturing parenthe-
1743 ses. The names are just an additional way of identifying the parenthe-
1744 ses, which still acquire numbers. Several convenience functions such as
1745 pcre_get_named_substring() are provided for extracting captured sub-
1746 strings by name. It is also possible to extract the data directly, by
1747 first converting the name to a number in order to access the correct
1748 pointers in the output vector (described with pcre_exec() below). To do
1749 the conversion, you need to use the name-to-number map, which is
1750 described by these three values.
1752 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1753 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1754 of each entry; both of these return an int value. The entry size
1755 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1756 a pointer to the first entry of the table (a pointer to char). The
1757 first two bytes of each entry are the number of the capturing parenthe-
1758 sis, most significant byte first. The rest of the entry is the corre-
1759 sponding name, zero terminated.
1761 The names are in alphabetical order. Duplicate names may appear if (?|
1762 is used to create multiple groups with the same number, as described in
1763 the section on duplicate subpattern numbers in the pcrepattern page.
1764 Duplicate names for subpatterns with different numbers are permitted
1765 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
1766 appear in the table in the order in which they were found in the pat-
1767 tern. In the absence of (?| this is the order of increasing number;
1768 when (?| is used this is not necessarily the case because later subpat-
1769 terns may have lower numbers.
1771 As a simple example of the name/number table, consider the following
1772 pattern (assume PCRE_EXTENDED is set, so white space - including new-
1773 lines - is ignored):
1775 (?<date> (?<year>(\d\d)?\d\d) -
1776 (?<month>\d\d) - (?<day>\d\d) )
1778 There are four named subpatterns, so the table has four entries, and
1779 each entry in the table is eight bytes long. The table is as follows,
1780 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1781 as ??:
1783 00 01 d a t e 00 ??
1784 00 05 d a y 00 ?? ??
1785 00 04 m o n t h 00
1786 00 02 y e a r 00 ??
1788 When writing code to extract data from named subpatterns using the
1789 name-to-number map, remember that the length of the entries is likely
1790 to be different for each compiled pattern.
1794 Return 1 if the pattern can be used for partial matching with
1795 pcre_exec(), otherwise 0. The fourth argument should point to an int
1796 variable. From release 8.00, this always returns 1, because the
1797 restrictions that previously applied to partial matching have been
1798 lifted. The pcrepartial documentation gives details of partial match-
1799 ing.
1803 Return a copy of the options with which the pattern was compiled. The
1804 fourth argument should point to an unsigned long int variable. These
1805 option bits are those specified in the call to pcre_compile(), modified
1806 by any top-level option settings at the start of the pattern itself. In
1807 other words, they are the options that will be in force when matching
1808 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1809 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1812 A pattern is automatically anchored by PCRE if all of its top-level
1813 alternatives begin with one of the following:
1815 ^ unless PCRE_MULTILINE is set
1816 \A always
1817 \G always
1818 .* if PCRE_DOTALL is set and there are no back
1819 references to the subpattern in which .* appears
1821 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1822 by pcre_fullinfo().
1826 Return the size of the compiled pattern, that is, the value that was
1827 passed as the argument to pcre_malloc() when PCRE was getting memory in
1828 which to place the compiled data. The fourth argument should point to a
1829 size_t variable.
1833 Return the size of the data block pointed to by the study_data field in
1834 a pcre_extra block. That is, it is the value that was passed to
1835 pcre_malloc() when PCRE was getting memory into which to place the data
1836 created by pcre_study(). If pcre_extra is NULL, or there is no study
1837 data, zero is returned. The fourth argument should point to a size_t
1838 variable.
1843 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1845 The pcre_info() function is now obsolete because its interface is too
1846 restrictive to return all the available data about a compiled pattern.
1847 New programs should use pcre_fullinfo() instead. The yield of
1848 pcre_info() is the number of capturing subpatterns, or one of the fol-
1849 lowing negative numbers:
1851 PCRE_ERROR_NULL the argument code was NULL
1852 PCRE_ERROR_BADMAGIC the "magic number" was not found
1854 If the optptr argument is not NULL, a copy of the options with which
1855 the pattern was compiled is placed in the integer it points to (see
1856 PCRE_INFO_OPTIONS above).
1858 If the pattern is not anchored and the firstcharptr argument is not
1859 NULL, it is used to pass back information about the first character of
1860 any matched string (see PCRE_INFO_FIRSTBYTE above).
1865 int pcre_refcount(pcre *code, int adjust);
1867 The pcre_refcount() function is used to maintain a reference count in
1868 the data block that contains a compiled pattern. It is provided for the
1869 benefit of applications that operate in an object-oriented manner,
1870 where different parts of the application may be using the same compiled
1871 pattern, but you want to free the block when they are all done.
1873 When a pattern is compiled, the reference count field is initialized to
1874 zero. It is changed only by calling this function, whose action is to
1875 add the adjust value (which may be positive or negative) to it. The
1876 yield of the function is the new value. However, the value of the count
1877 is constrained to lie between 0 and 65535, inclusive. If the new value
1878 is outside these limits, it is forced to the appropriate limit value.
1880 Except when it is zero, the reference count is not correctly preserved
1881 if a pattern is compiled on one host and then transferred to a host
1882 whose byte-order is different. (This seems a highly unlikely scenario.)
1887 int pcre_exec(const pcre *code, const pcre_extra *extra,
1888 const char *subject, int length, int startoffset,
1889 int options, int *ovector, int ovecsize);
1891 The function pcre_exec() is called to match a subject string against a
1892 compiled pattern, which is passed in the code argument. If the pattern
1893 was studied, the result of the study should be passed in the extra
1894 argument. This function is the main matching facility of the library,
1895 and it operates in a Perl-like manner. For specialist use there is also
1896 an alternative matching function, which is described below in the sec-
1897 tion about the pcre_dfa_exec() function.
1899 In most applications, the pattern will have been compiled (and option-
1900 ally studied) in the same process that calls pcre_exec(). However, it
1901 is possible to save compiled patterns and study data, and then use them
1902 later in different processes, possibly even on different hosts. For a
1903 discussion about this, see the pcreprecompile documentation.
1905 Here is an example of a simple call to pcre_exec():
1907 int rc;
1908 int ovector[30];
1909 rc = pcre_exec(
1910 re, /* result of pcre_compile() */
1911 NULL, /* we didn't study the pattern */
1912 "some string", /* the subject string */
1913 11, /* the length of the subject string */
1914 0, /* start at offset 0 in the subject */
1915 0, /* default options */
1916 ovector, /* vector of integers for substring information */
1917 30); /* number of elements (NOT size in bytes) */
1919 Extra data for pcre_exec()
1921 If the extra argument is not NULL, it must point to a pcre_extra data
1922 block. The pcre_study() function returns such a block (when it doesn't
1923 return NULL), but you can also create one for yourself, and pass addi-
1924 tional information in it. The pcre_extra block contains the following
1925 fields (not necessarily in this order):
1927 unsigned long int flags;
1928 void *study_data;
1929 unsigned long int match_limit;
1930 unsigned long int match_limit_recursion;
1931 void *callout_data;
1932 const unsigned char *tables;
1933 unsigned char **mark;
1935 The flags field is a bitmap that specifies which of the other fields
1936 are set. The flag bits are:
1945 Other flag bits should be set to zero. The study_data field is set in
1946 the pcre_extra block that is returned by pcre_study(), together with
1947 the appropriate flag bit. You should not set this yourself, but you may
1948 add to the block by setting the other fields and their corresponding
1949 flag bits.
1951 The match_limit field provides a means of preventing PCRE from using up
1952 a vast amount of resources when running patterns that are not going to
1953 match, but which have a very large number of possibilities in their
1954 search trees. The classic example is a pattern that uses nested unlim-
1955 ited repeats.
1957 Internally, PCRE uses a function called match() which it calls repeat-
1958 edly (sometimes recursively). The limit set by match_limit is imposed
1959 on the number of times this function is called during a match, which
1960 has the effect of limiting the amount of backtracking that can take
1961 place. For patterns that are not anchored, the count restarts from zero
1962 for each position in the subject string.
1964 The default value for the limit can be set when PCRE is built; the
1965 default default is 10 million, which handles all but the most extreme
1966 cases. You can override the default by suppling pcre_exec() with a
1967 pcre_extra block in which match_limit is set, and
1968 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1969 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1971 The match_limit_recursion field is similar to match_limit, but instead
1972 of limiting the total number of times that match() is called, it limits
1973 the depth of recursion. The recursion depth is a smaller number than
1974 the total number of calls, because not all calls to match() are recur-
1975 sive. This limit is of use only if it is set smaller than match_limit.
1977 Limiting the recursion depth limits the amount of stack that can be
1978 used, or, when PCRE has been compiled to use memory on the heap instead
1979 of the stack, the amount of heap memory that can be used.
1981 The default value for match_limit_recursion can be set when PCRE is
1982 built; the default default is the same value as the default for
1983 match_limit. You can override the default by suppling pcre_exec() with
1984 a pcre_extra block in which match_limit_recursion is set, and
1985 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1986 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1988 The callout_data field is used in conjunction with the "callout" fea-
1989 ture, and is described in the pcrecallout documentation.
1991 The tables field is used to pass a character tables pointer to
1992 pcre_exec(); this overrides the value that is stored with the compiled
1993 pattern. A non-NULL value is stored with the compiled pattern only if
1994 custom tables were supplied to pcre_compile() via its tableptr argu-
1995 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1996 PCRE's internal tables to be used. This facility is helpful when re-
1997 using patterns that have been saved after compiling with an external
1998 set of tables, because the external tables might be at a different
1999 address when pcre_exec() is called. See the pcreprecompile documenta-
2000 tion for a discussion of saving compiled patterns for later use.
2002 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
2003 set to point to a char * variable. If the pattern contains any back-
2004 tracking control verbs such as (*MARK:NAME), and the execution ends up
2005 with a name to pass back, a pointer to the name string (zero termi-
2006 nated) is placed in the variable pointed to by the mark field. The
2007 names are within the compiled pattern; if you wish to retain such a
2008 name you must copy it before freeing the memory of a compiled pattern.
2009 If there is no name to pass back, the variable pointed to by the mark
2010 field set to NULL. For details of the backtracking control verbs, see
2011 the section entitled "Backtracking control" in the pcrepattern documen-
2012 tation.
2014 Option bits for pcre_exec()
2016 The unused bits of the options argument for pcre_exec() must be zero.
2017 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
2024 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
2025 matching position. If a pattern was compiled with PCRE_ANCHORED, or
2026 turned out to be anchored by virtue of its contents, it cannot be made
2027 unachored at matching time.
2032 These options (which are mutually exclusive) control what the \R escape
2033 sequence matches. The choice is either to match only CR, LF, or CRLF,
2034 or to match any Unicode newline sequence. These options override the
2035 choice that was made or defaulted when the pattern was compiled.
2043 These options override the newline definition that was chosen or
2044 defaulted when the pattern was compiled. For details, see the descrip-
2045 tion of pcre_compile() above. During matching, the newline choice
2046 affects the behaviour of the dot, circumflex, and dollar metacharac-
2047 ters. It may also alter the way the match position is advanced after a
2048 match failure for an unanchored pattern.
2051 set, and a match attempt for an unanchored pattern fails when the cur-
2052 rent position is at a CRLF sequence, and the pattern contains no
2053 explicit matches for CR or LF characters, the match position is
2054 advanced by two characters instead of one, in other words, to after the
2055 CRLF.
2057 The above rule is a compromise that makes the most common cases work as
2058 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
2059 option is not set), it does not match the string "\r\nA" because, after
2060 failing at the start, it skips both the CR and the LF before retrying.
2061 However, the pattern [\r\n]A does match that string, because it con-
2062 tains an explicit CR or LF reference, and so advances only by one char-
2063 acter after the first failure.
2065 An explicit match for CR of LF is either a literal appearance of one of
2066 those characters, or one of the \r or \n escape sequences. Implicit
2067 matches such as [^X] do not count, nor does \s (which includes CR and
2068 LF in the characters that it matches).
2070 Notwithstanding the above, anomalous effects may still occur when CRLF
2071 is a valid newline sequence and explicit \r or \n escapes appear in the
2072 pattern.
2076 This option specifies that first character of the subject string is not
2077 the beginning of a line, so the circumflex metacharacter should not
2078 match before it. Setting this without PCRE_MULTILINE (at compile time)
2079 causes circumflex never to match. This option affects only the behav-
2080 iour of the circumflex metacharacter. It does not affect \A.
2084 This option specifies that the end of the subject string is not the end
2085 of a line, so the dollar metacharacter should not match it nor (except
2086 in multiline mode) a newline immediately before it. Setting this with-
2087 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2088 option affects only the behaviour of the dollar metacharacter. It does
2089 not affect \Z or \z.
2093 An empty string is not considered to be a valid match if this option is
2094 set. If there are alternatives in the pattern, they are tried. If all
2095 the alternatives match the empty string, the entire match fails. For
2096 example, if the pattern
2098 a?b?
2100 is applied to a string not beginning with "a" or "b", it matches an
2101 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2102 match is not valid, so PCRE searches further into the string for occur-
2103 rences of "a" or "b".
2107 This is like PCRE_NOTEMPTY, except that an empty string match that is
2108 not at the start of the subject is permitted. If the pattern is
2109 anchored, such a match can occur only if the pattern contains \K.
2111 Perl has no direct equivalent of PCRE_NOTEMPTY or
2112 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2113 match of the empty string within its split() function, and when using
2114 the /g modifier. It is possible to emulate Perl's behaviour after
2115 matching a null string by first trying the match again at the same off-
2116 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2117 fails, by advancing the starting offset (see below) and trying an ordi-
2118 nary match again. There is some code that demonstrates how to do this
2119 in the pcredemo sample program.
2123 There are a number of optimizations that pcre_exec() uses at the start
2124 of a match, in order to speed up the process. For example, if it is
2125 known that a match must start with a specific character, it searches
2126 the subject for that character, and fails immediately if it cannot find
2127 it, without actually running the main matching function. When callouts
2128 are in use, these optimizations can cause them to be skipped. This
2129 option disables the "start-up" optimizations, causing performance to
2130 suffer, but ensuring that the callouts do occur.
2134 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2135 UTF-8 string is automatically checked when pcre_exec() is subsequently
2136 called. The value of startoffset is also checked to ensure that it
2137 points to the start of a UTF-8 character. There is a discussion about
2138 the validity of UTF-8 strings in the section on UTF-8 support in the
2139 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2140 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
2141 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
2143 If you already know that your subject is valid, and you want to skip
2144 these checks for performance reasons, you can set the
2145 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2146 do this for the second and subsequent calls to pcre_exec() if you are
2147 making repeated calls to find all the matches in a single subject
2148 string. However, you should be sure that the value of startoffset
2149 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
2150 set, the effect of passing an invalid UTF-8 string as a subject, or a
2151 value of startoffset that does not point to the start of a UTF-8 char-
2152 acter, is undefined. Your program may crash.
2157 These options turn on the partial matching feature. For backwards com-
2158 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2159 match occurs if the end of the subject string is reached successfully,
2160 but there are not enough subject characters to complete the match. If
2161 this happens when PCRE_PARTIAL_HARD is set, pcre_exec() immediately
2162 returns PCRE_ERROR_PARTIAL. Otherwise, if PCRE_PARTIAL_SOFT is set,
2163 matching continues by testing any other alternatives. Only if they all
2164 fail is PCRE_ERROR_PARTIAL returned (instead of PCRE_ERROR_NOMATCH).
2165 The portion of the string that was inspected when the partial match was
2166 found is set as the first matching string. There is a more detailed
2167 discussion in the pcrepartial documentation.
2169 The string to be matched by pcre_exec()
2171 The subject string is passed to pcre_exec() as a pointer in subject, a
2172 length (in bytes) in length, and a starting byte offset in startoffset.
2173 In UTF-8 mode, the byte offset must point to the start of a UTF-8 char-
2174 acter. Unlike the pattern string, the subject may contain binary zero
2175 bytes. When the starting offset is zero, the search for a match starts
2176 at the beginning of the subject, and this is by far the most common
2177 case.
2179 A non-zero starting offset is useful when searching for another match
2180 in the same subject by calling pcre_exec() again after a previous suc-
2181 cess. Setting startoffset differs from just passing over a shortened
2182 string and setting PCRE_NOTBOL in the case of a pattern that begins
2183 with any kind of lookbehind. For example, consider the pattern
2185 \Biss\B
2187 which finds occurrences of "iss" in the middle of words. (\B matches
2188 only if the current position in the subject is not a word boundary.)
2189 When applied to the string "Mississipi" the first call to pcre_exec()
2190 finds the first occurrence. If pcre_exec() is called again with just
2191 the remainder of the subject, namely "issipi", it does not match,
2192 because \B is always false at the start of the subject, which is deemed
2193 to be a word boundary. However, if pcre_exec() is passed the entire
2194 string again, but with startoffset set to 4, it finds the second occur-
2195 rence of "iss" because it is able to look behind the starting point to
2196 discover that it is preceded by a letter.
2198 If a non-zero starting offset is passed when the pattern is anchored,
2199 one attempt to match at the given offset is made. This can only succeed
2200 if the pattern does not require the match to be at the start of the
2201 subject.
2203 How pcre_exec() returns captured substrings
2205 In general, a pattern matches a certain portion of the subject, and in
2206 addition, further substrings from the subject may be picked out by
2207 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2208 this is called "capturing" in what follows, and the phrase "capturing
2209 subpattern" is used for a fragment of a pattern that picks out a sub-
2210 string. PCRE supports several other kinds of parenthesized subpattern
2211 that do not cause substrings to be captured.
2213 Captured substrings are returned to the caller via a vector of integers
2214 whose address is passed in ovector. The number of elements in the vec-
2215 tor is passed in ovecsize, which must be a non-negative number. Note:
2216 this argument is NOT the size of ovector in bytes.
2218 The first two-thirds of the vector is used to pass back captured sub-
2219 strings, each substring using a pair of integers. The remaining third
2220 of the vector is used as workspace by pcre_exec() while matching cap-
2221 turing subpatterns, and is not available for passing back information.
2222 The number passed in ovecsize should always be a multiple of three. If
2223 it is not, it is rounded down.
2225 When a match is successful, information about captured substrings is
2226 returned in pairs of integers, starting at the beginning of ovector,
2227 and continuing up to two-thirds of its length at the most. The first
2228 element of each pair is set to the byte offset of the first character
2229 in a substring, and the second is set to the byte offset of the first
2230 character after the end of a substring. Note: these values are always
2231 byte offsets, even in UTF-8 mode. They are not character counts.
2233 The first pair of integers, ovector[0] and ovector[1], identify the
2234 portion of the subject string matched by the entire pattern. The next
2235 pair is used for the first capturing subpattern, and so on. The value
2236 returned by pcre_exec() is one more than the highest numbered pair that
2237 has been set. For example, if two substrings have been captured, the
2238 returned value is 3. If there are no capturing subpatterns, the return
2239 value from a successful match is 1, indicating that just the first pair
2240 of offsets has been set.
2242 If a capturing subpattern is matched repeatedly, it is the last portion
2243 of the string that it matched that is returned.
2245 If the vector is too small to hold all the captured substring offsets,
2246 it is used as far as possible (up to two-thirds of its length), and the
2247 function returns a value of zero. If the substring offsets are not of
2248 interest, pcre_exec() may be called with ovector passed as NULL and
2249 ovecsize as zero. However, if the pattern contains back references and
2250 the ovector is not big enough to remember the related substrings, PCRE
2251 has to get additional memory for use during matching. Thus it is usu-
2252 ally advisable to supply an ovector.
2254 The pcre_fullinfo() function can be used to find out how many capturing
2255 subpatterns there are in a compiled pattern. The smallest size for
2256 ovector that will allow for n captured substrings, in addition to the
2257 offsets of the substring matched by the whole pattern, is (n+1)*3.
2259 It is possible for capturing subpattern number n+1 to match some part
2260 of the subject when subpattern n has not been used at all. For example,
2261 if the string "abc" is matched against the pattern (a|(z))(bc) the
2262 return from the function is 4, and subpatterns 1 and 3 are matched, but
2263 2 is not. When this happens, both values in the offset pairs corre-
2264 sponding to unused subpatterns are set to -1.
2266 Offset values that correspond to unused subpatterns at the end of the
2267 expression are also set to -1. For example, if the string "abc" is
2268 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2269 matched. The return from the function is 2, because the highest used
2270 capturing subpattern number is 1. However, you can refer to the offsets
2271 for the second and third capturing subpatterns if you wish (assuming
2272 the vector is large enough, of course).
2274 Some convenience functions are provided for extracting the captured
2275 substrings as separate strings. These are described below.
2277 Error return values from pcre_exec()
2279 If pcre_exec() fails, it returns a negative number. The following are
2280 defined in the header file:
2284 The subject string did not match the pattern.
2288 Either code or subject was passed as NULL, or ovector was NULL and
2289 ovecsize was not zero.
2293 An unrecognized bit was set in the options argument.
2297 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2298 to catch the case when it is passed a junk pointer and to detect when a
2299 pattern that was compiled in an environment of one endianness is run in
2300 an environment with the other endianness. This is the error that PCRE
2301 gives when the magic number is not present.
2305 While running the pattern match, an unknown item was encountered in the
2306 compiled pattern. This error could be caused by a bug in PCRE or by
2307 overwriting of the compiled pattern.
2311 If a pattern contains back references, but the ovector that is passed
2312 to pcre_exec() is not big enough to remember the referenced substrings,
2313 PCRE gets a block of memory at the start of matching to use for this
2314 purpose. If the call via pcre_malloc() fails, this error is given. The
2315 memory is automatically freed at the end of matching.
2317 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2318 This can happen only when PCRE has been compiled with --disable-stack-
2319 for-recursion.
2323 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2324 and pcre_get_substring_list() functions (see below). It is never
2325 returned by pcre_exec().
2329 The backtracking limit, as specified by the match_limit field in a
2330 pcre_extra structure (or defaulted) was reached. See the description
2331 above.
2335 This error is never generated by pcre_exec() itself. It is provided for
2336 use by callout functions that want to yield a distinctive error code.
2337 See the pcrecallout documentation for details.
2341 A string that contains an invalid UTF-8 byte sequence was passed as a
2342 subject.
2346 The UTF-8 byte sequence that was passed as a subject was valid, but the
2347 value of startoffset did not point to the beginning of a UTF-8 charac-
2348 ter.
2352 The subject string did not match, but it did match partially. See the
2353 pcrepartial documentation for details of partial matching.
2357 This code is no longer in use. It was formerly returned when the
2358 PCRE_PARTIAL option was used with a compiled pattern containing items
2359 that were not supported for partial matching. From release 8.00
2360 onwards, there are no restrictions on partial matching.
2364 An unexpected internal error has occurred. This error could be caused
2365 by a bug in PCRE or by overwriting of the compiled pattern.
2369 This error is given if the value of the ovecsize argument is negative.
2373 The internal recursion limit, as specified by the match_limit_recursion
2374 field in a pcre_extra structure (or defaulted) was reached. See the
2375 description above.
2379 An invalid combination of PCRE_NEWLINE_xxx options was given.
2381 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2386 int pcre_copy_substring(const char *subject, int *ovector,
2387 int stringcount, int stringnumber, char *buffer,
2388 int buffersize);
2390 int pcre_get_substring(const char *subject, int *ovector,
2391 int stringcount, int stringnumber,
2392 const char **stringptr);
2394 int pcre_get_substring_list(const char *subject,
2395 int *ovector, int stringcount, const char ***listptr);
2397 Captured substrings can be accessed directly by using the offsets
2398 returned by pcre_exec() in ovector. For convenience, the functions
2399 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2400 string_list() are provided for extracting captured substrings as new,
2401 separate, zero-terminated strings. These functions identify substrings
2402 by number. The next section describes functions for extracting named
2403 substrings.
2405 A substring that contains a binary zero is correctly extracted and has
2406 a further zero added on the end, but the result is not, of course, a C
2407 string. However, you can process such a string by referring to the
2408 length that is returned by pcre_copy_substring() and pcre_get_sub-
2409 string(). Unfortunately, the interface to pcre_get_substring_list() is
2410 not adequate for handling strings containing binary zeros, because the
2411 end of the final string is not independently indicated.
2413 The first three arguments are the same for all three of these func-
2414 tions: subject is the subject string that has just been successfully
2415 matched, ovector is a pointer to the vector of integer offsets that was
2416 passed to pcre_exec(), and stringcount is the number of substrings that
2417 were captured by the match, including the substring that matched the
2418 entire regular expression. This is the value returned by pcre_exec() if
2419 it is greater than zero. If pcre_exec() returned zero, indicating that
2420 it ran out of space in ovector, the value passed as stringcount should
2421 be the number of elements in the vector divided by three.
2423 The functions pcre_copy_substring() and pcre_get_substring() extract a
2424 single substring, whose number is given as stringnumber. A value of
2425 zero extracts the substring that matched the entire pattern, whereas
2426 higher values extract the captured substrings. For pcre_copy_sub-
2427 string(), the string is placed in buffer, whose length is given by
2428 buffersize, while for pcre_get_substring() a new block of memory is
2429 obtained via pcre_malloc, and its address is returned via stringptr.
2430 The yield of the function is the length of the string, not including
2431 the terminating zero, or one of these error codes:
2435 The buffer was too small for pcre_copy_substring(), or the attempt to
2436 get memory failed for pcre_get_substring().
2440 There is no substring whose number is stringnumber.
2442 The pcre_get_substring_list() function extracts all available sub-
2443 strings and builds a list of pointers to them. All this is done in a
2444 single block of memory that is obtained via pcre_malloc. The address of
2445 the memory block is returned via listptr, which is also the start of
2446 the list of string pointers. The end of the list is marked by a NULL
2447 pointer. The yield of the function is zero if all went well, or the
2448 error code
2452 if the attempt to get the memory block failed.
2454 When any of these functions encounter a substring that is unset, which
2455 can happen when capturing subpattern number n+1 matches some part of
2456 the subject, but subpattern n has not been used at all, they return an
2457 empty string. This can be distinguished from a genuine zero-length sub-
2458 string by inspecting the appropriate offset in ovector, which is nega-
2459 tive for unset substrings.
2461 The two convenience functions pcre_free_substring() and pcre_free_sub-
2462 string_list() can be used to free the memory returned by a previous
2463 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2464 tively. They do nothing more than call the function pointed to by
2465 pcre_free, which of course could be called directly from a C program.
2466 However, PCRE is used in some situations where it is linked via a spe-
2467 cial interface to another programming language that cannot use
2468 pcre_free directly; it is for these cases that the functions are pro-
2469 vided.
2474 int pcre_get_stringnumber(const pcre *code,
2475 const char *name);
2477 int pcre_copy_named_substring(const pcre *code,
2478 const char *subject, int *ovector,
2479 int stringcount, const char *stringname,
2480 char *buffer, int buffersize);
2482 int pcre_get_named_substring(const pcre *code,
2483 const char *subject, int *ovector,
2484 int stringcount, const char *stringname,
2485 const char **stringptr);
2487 To extract a substring by name, you first have to find associated num-
2488 ber. For example, for this pattern
2490 (a+)b(?<xxx>\d+)...
2492 the number of the subpattern called "xxx" is 2. If the name is known to
2493 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2494 name by calling pcre_get_stringnumber(). The first argument is the com-
2495 piled pattern, and the second is the name. The yield of the function is
2496 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2497 subpattern of that name.
2499 Given the number, you can extract the substring directly, or use one of
2500 the functions described in the previous section. For convenience, there
2501 are also two functions that do the whole job.
2503 Most of the arguments of pcre_copy_named_substring() and
2504 pcre_get_named_substring() are the same as those for the similarly
2505 named functions that extract by number. As these are described in the
2506 previous section, they are not re-described here. There are just two
2507 differences:
2509 First, instead of a substring number, a substring name is given. Sec-
2510 ond, there is an extra argument, given at the start, which is a pointer
2511 to the compiled pattern. This is needed in order to gain access to the
2512 name-to-number translation table.
2514 These functions call pcre_get_stringnumber(), and if it succeeds, they
2515 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2516 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2517 behaviour may not be what you want (see the next section).
2519 Warning: If the pattern uses the (?| feature to set up multiple subpat-
2520 terns with the same number, as described in the section on duplicate
2521 subpattern numbers in the pcrepattern page, you cannot use names to
2522 distinguish the different subpatterns, because names are not included
2523 in the compiled code. The matching process uses only numbers. For this
2524 reason, the use of different names for subpatterns of the same number
2525 causes an error at compile time.
2530 int pcre_get_stringtable_entries(const pcre *code,
2531 const char *name, char **first, char **last);
2533 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2534 subpatterns are not required to be unique. (Duplicate names are always
2535 allowed for subpatterns with the same number, created by using the (?|
2536 feature. Indeed, if such subpatterns are named, they are required to
2537 use the same names.)
2539 Normally, patterns with duplicate names are such that in any one match,
2540 only one of the named subpatterns participates. An example is shown in
2541 the pcrepattern documentation.
2543 When duplicates are present, pcre_copy_named_substring() and
2544 pcre_get_named_substring() return the first substring corresponding to
2545 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2546 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2547 function returns one of the numbers that are associated with the name,
2548 but it is not defined which it is.
2550 If you want to get full details of all captured substrings for a given
2551 name, you must use the pcre_get_stringtable_entries() function. The
2552 first argument is the compiled pattern, and the second is the name. The
2553 third and fourth are pointers to variables which are updated by the
2554 function. After it has run, they point to the first and last entries in
2555 the name-to-number table for the given name. The function itself
2556 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2557 there are none. The format of the table is described above in the sec-
2558 tion entitled Information about a pattern. Given all the relevant
2559 entries for the name, you can extract each of their numbers, and hence
2560 the captured data, if any.
2565 The traditional matching function uses a similar algorithm to Perl,
2566 which stops when it finds the first match, starting at a given point in
2567 the subject. If you want to find all possible matches, or the longest
2568 possible match, consider using the alternative matching function (see
2569 below) instead. If you cannot use the alternative function, but still
2570 need to find all possible matches, you can kludge it up by making use
2571 of the callout facility, which is described in the pcrecallout documen-
2572 tation.
2574 What you have to do is to insert a callout right at the end of the pat-
2575 tern. When your callout function is called, extract and save the cur-
2576 rent matched substring. Then return 1, which forces pcre_exec() to
2577 backtrack and try other alternatives. Ultimately, when it runs out of
2578 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2583 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2584 const char *subject, int length, int startoffset,
2585 int options, int *ovector, int ovecsize,
2586 int *workspace, int wscount);
2588 The function pcre_dfa_exec() is called to match a subject string
2589 against a compiled pattern, using a matching algorithm that scans the
2590 subject string just once, and does not backtrack. This has different
2591 characteristics to the normal algorithm, and is not compatible with
2592 Perl. Some of the features of PCRE patterns are not supported. Never-
2593 theless, there are times when this kind of matching can be useful. For
2594 a discussion of the two matching algorithms, and a list of features
2595 that pcre_dfa_exec() does not support, see the pcrematching documenta-
2596 tion.
2598 The arguments for the pcre_dfa_exec() function are the same as for
2599 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2600 ent way, and this is described below. The other common arguments are
2601 used in the same way as for pcre_exec(), so their description is not
2602 repeated here.
2604 The two additional arguments provide workspace for the function. The
2605 workspace vector should contain at least 20 elements. It is used for
2606 keeping track of multiple paths through the pattern tree. More
2607 workspace will be needed for patterns and subjects where there are a
2608 lot of potential matches.
2610 Here is an example of a simple call to pcre_dfa_exec():
2612 int rc;
2613 int ovector[10];
2614 int wspace[20];
2615 rc = pcre_dfa_exec(
2616 re, /* result of pcre_compile() */
2617 NULL, /* we didn't study the pattern */
2618 "some string", /* the subject string */
2619 11, /* the length of the subject string */
2620 0, /* start at offset 0 in the subject */
2621 0, /* default options */
2622 ovector, /* vector of integers for substring information */
2623 10, /* number of elements (NOT size in bytes) */
2624 wspace, /* working space vector */
2625 20); /* number of elements (NOT size in bytes) */
2627 Option bits for pcre_dfa_exec()
2629 The unused bits of the options argument for pcre_dfa_exec() must be
2630 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2634 four of these are exactly the same as for pcre_exec(), so their
2635 description is not repeated here.
2640 These have the same general effect as they do for pcre_exec(), but the
2641 details are slightly different. When PCRE_PARTIAL_HARD is set for
2642 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
2643 ject is reached and there is still at least one matching possibility
2644 that requires additional characters. This happens even if some complete
2645 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
2646 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
2647 of the subject is reached, there have been no complete matches, but
2648 there is still at least one matching possibility. The portion of the
2649 string that was inspected when the longest partial match was found is
2650 set as the first matching string in both cases.
2654 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2655 stop as soon as it has found one match. Because of the way the alterna-
2656 tive algorithm works, this is necessarily the shortest possible match
2657 at the first possible matching point in the subject string.
2661 When pcre_dfa_exec() returns a partial match, it is possible to call it
2662 again, with additional subject characters, and have it continue with
2663 the same match. The PCRE_DFA_RESTART option requests this action; when
2664 it is set, the workspace and wscount options must reference the same
2665 vector as before because data about the match so far is left in them
2666 after a partial match. There is more discussion of this facility in the
2667 pcrepartial documentation.
2669 Successful returns from pcre_dfa_exec()
2671 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2672 string in the subject. Note, however, that all the matches from one run
2673 of the function start at the same point in the subject. The shorter
2674 matches are all initial substrings of the longer matches. For example,
2675 if the pattern
2677 <.*>
2679 is matched against the string
2681 This is <something> <something else> <something further> no more
2683 the three matched strings are
2685 <something>
2686 <something> <something else>
2687 <something> <something else> <something further>
2689 On success, the yield of the function is a number greater than zero,
2690 which is the number of matched substrings. The substrings themselves
2691 are returned in ovector. Each string uses two elements; the first is
2692 the offset to the start, and the second is the offset to the end. In
2693 fact, all the strings have the same start offset. (Space could have
2694 been saved by giving this only once, but it was decided to retain some
2695 compatibility with the way pcre_exec() returns data, even though the
2696 meaning of the strings is different.)
2698 The strings are returned in reverse order of length; that is, the long-
2699 est matching string is given first. If there were too many matches to
2700 fit into ovector, the yield of the function is zero, and the vector is
2701 filled with the longest matches.
2703 Error returns from pcre_dfa_exec()
2705 The pcre_dfa_exec() function returns a negative number when it fails.
2706 Many of the errors are the same as for pcre_exec(), and these are
2707 described above. There are in addition the following errors that are
2708 specific to pcre_dfa_exec():
2712 This return is given if pcre_dfa_exec() encounters an item in the pat-
2713 tern that it does not support, for instance, the use of \C or a back
2714 reference.
2718 This return is given if pcre_dfa_exec() encounters a condition item
2719 that uses a back reference for the condition, or a test for recursion
2720 in a specific group. These are not supported.
2724 This return is given if pcre_dfa_exec() is called with an extra block
2725 that contains a setting of the match_limit field. This is not supported
2726 (it is meaningless).
2730 This return is given if pcre_dfa_exec() runs out of space in the
2731 workspace vector.
2735 When a recursive subpattern is processed, the matching function calls
2736 itself recursively, using private vectors for ovector and workspace.
2737 This error is given if the output vector is not large enough. This
2738 should be extremely rare, as a vector of size 1000 is used.
2743 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2744 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2749 Philip Hazel
2750 University Computing Service
2751 Cambridge CB2 3QH, England.
2756 Last updated: 01 June 2010
2757 Copyright (c) 1997-2010 University of Cambridge.
2758 ------------------------------------------------------------------------------
2764 NAME
2765 PCRE - Perl-compatible regular expressions
2770 int (*pcre_callout)(pcre_callout_block *);
2772 PCRE provides a feature called "callout", which is a means of temporar-
2773 ily passing control to the caller of PCRE in the middle of pattern
2774 matching. The caller of PCRE provides an external function by putting
2775 its entry point in the global variable pcre_callout. By default, this
2776 variable contains NULL, which disables all calling out.
2778 Within a regular expression, (?C) indicates the points at which the
2779 external function is to be called. Different callout points can be
2780 identified by putting a number less than 256 after the letter C. The
2781 default value is zero. For example, this pattern has two callout
2782 points:
2784 (?C1)abc(?C2)def
2786 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() or
2787 pcre_compile2() is called, PCRE automatically inserts callouts, all
2788 with number 255, before each item in the pattern. For example, if
2789 PCRE_AUTO_CALLOUT is used with the pattern
2791 A(\d{2}|--)
2793 it is processed as if it were
2795 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2797 Notice that there is a callout before and after each parenthesis and
2798 alternation bar. Automatic callouts can be used for tracking the
2799 progress of pattern matching. The pcretest command has an option that
2800 sets automatic callouts; when it is used, the output indicates how the
2801 pattern is matched. This is useful information when you are trying to
2802 optimize the performance of a particular pattern.
2807 You should be aware that, because of optimizations in the way PCRE
2808 matches patterns by default, callouts sometimes do not happen. For
2809 example, if the pattern is
2811 ab(?C4)cd
2813 PCRE knows that any matching string must contain the letter "d". If the
2814 subject string is "abyz", the lack of "d" means that matching doesn't
2815 ever start, and the callout is never reached. However, with "abyd",
2816 though the result is still no match, the callout is obeyed.
2818 If the pattern is studied, PCRE knows the minimum length of a matching
2819 string, and will immediately give a "no match" return without actually
2820 running a match if the subject is not long enough, or, for unanchored
2821 patterns, if it has been scanned far enough.
2823 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2824 MIZE option to pcre_exec() or pcre_dfa_exec(). This slows down the
2825 matching process, but does ensure that callouts such as the example
2826 above are obeyed.
2831 During matching, when PCRE reaches a callout point, the external func-
2832 tion defined by pcre_callout is called (if it is set). This applies to
2833 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2834 only argument to the callout function is a pointer to a pcre_callout
2835 block. This structure contains the following fields:
2837 int version;
2838 int callout_number;
2839 int *offset_vector;
2840 const char *subject;
2841 int subject_length;
2842 int start_match;
2843 int current_position;
2844 int capture_top;
2845 int capture_last;
2846 void *callout_data;
2847 int pattern_position;
2848 int next_item_length;
2850 The version field is an integer containing the version number of the
2851 block format. The initial version was 0; the current version is 1. The
2852 version number will change again in future if additional fields are
2853 added, but the intention is never to remove any of the existing fields.
2855 The callout_number field contains the number of the callout, as com-
2856 piled into the pattern (that is, the number after ?C for manual call-
2857 outs, and 255 for automatically generated callouts).
2859 The offset_vector field is a pointer to the vector of offsets that was
2860 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2861 pcre_exec() is used, the contents can be inspected in order to extract
2862 substrings that have been matched so far, in the same way as for
2863 extracting substrings after a match has completed. For pcre_dfa_exec()
2864 this field is not useful.
2866 The subject and subject_length fields contain copies of the values that
2867 were passed to pcre_exec().
2869 The start_match field normally contains the offset within the subject
2870 at which the current match attempt started. However, if the escape
2871 sequence \K has been encountered, this value is changed to reflect the
2872 modified starting point. If the pattern is not anchored, the callout
2873 function may be called several times from the same point in the pattern
2874 for different starting points in the subject.
2876 The current_position field contains the offset within the subject of
2877 the current match pointer.
2879 When the pcre_exec() function is used, the capture_top field contains
2880 one more than the number of the highest numbered captured substring so
2881 far. If no substrings have been captured, the value of capture_top is
2882 one. This is always the case when pcre_dfa_exec() is used, because it
2883 does not support captured substrings.
2885 The capture_last field contains the number of the most recently cap-
2886 tured substring. If no substrings have been captured, its value is -1.
2887 This is always the case when pcre_dfa_exec() is used.
2889 The callout_data field contains a value that is passed to pcre_exec()
2890 or pcre_dfa_exec() specifically so that it can be passed back in call-
2891 outs. It is passed in the pcre_callout field of the pcre_extra data
2892 structure. If no such data was passed, the value of callout_data in a
2893 pcre_callout block is NULL. There is a description of the pcre_extra
2894 structure in the pcreapi documentation.
2896 The pattern_position field is present from version 1 of the pcre_call-
2897 out structure. It contains the offset to the next item to be matched in
2898 the pattern string.
2900 The next_item_length field is present from version 1 of the pcre_call-
2901 out structure. It contains the length of the next item to be matched in
2902 the pattern string. When the callout immediately precedes an alterna-
2903 tion bar, a closing parenthesis, or the end of the pattern, the length
2904 is zero. When the callout precedes an opening parenthesis, the length
2905 is that of the entire subpattern.
2907 The pattern_position and next_item_length fields are intended to help
2908 in distinguishing between different automatic callouts, which all have
2909 the same callout number. However, they are set for all callouts.
2914 The external callout function returns an integer to PCRE. If the value
2915 is zero, matching proceeds as normal. If the value is greater than
2916 zero, matching fails at the current point, but the testing of other
2917 matching possibilities goes ahead, just as if a lookahead assertion had
2918 failed. If the value is less than zero, the match is abandoned, and
2919 pcre_exec() or pcre_dfa_exec() returns the negative value.
2921 Negative values should normally be chosen from the set of
2922 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2923 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2924 reserved for use by callout functions; it will never be used by PCRE
2925 itself.
2930 Philip Hazel
2931 University Computing Service
2932 Cambridge CB2 3QH, England.
2937 Last updated: 29 September 2009
2938 Copyright (c) 1997-2009 University of Cambridge.
2939 ------------------------------------------------------------------------------
2945 NAME
2946 PCRE - Perl-compatible regular expressions
2951 This document describes the differences in the ways that PCRE and Perl
2952 handle regular expressions. The differences described here are with
2953 respect to Perl 5.10/5.11.
2955 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2956 of what it does have are given in the section on UTF-8 support in the
2957 main pcre page.
2959 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2960 permits them, but they do not mean what you might think. For example,
2961 (?!a){3} does not assert that the next three characters are not "a". It
2962 just asserts that the next character is not "a" three times.
2964 3. Capturing subpatterns that occur inside negative lookahead asser-
2965 tions are counted, but their entries in the offsets vector are never
2966 set. Perl sets its numerical variables from any such patterns that are
2967 matched before the assertion fails to match something (thereby succeed-
2968 ing), but only if the negative lookahead assertion contains just one
2969 branch.
2971 4. Though binary zero characters are supported in the subject string,
2972 they are not allowed in a pattern string because it is passed as a nor-
2973 mal C string, terminated by zero. The escape sequence \0 can be used in
2974 the pattern to represent a binary zero.
2976 5. The following Perl escape sequences are not supported: \l, \u, \L,
2977 \U, and \N. In fact these are implemented by Perl's general string-han-
2978 dling and are not part of its pattern matching engine. If any of these
2979 are encountered by PCRE, an error is generated.
2981 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2982 is built with Unicode character property support. The properties that
2983 can be tested with \p and \P are limited to the general category prop-
2984 erties such as Lu and Nd, script names such as Greek or Han, and the
2985 derived properties Any and L&. PCRE does support the Cs (surrogate)
2986 property, which Perl does not; the Perl documentation says "Because
2987 Perl hides the need for the user to understand the internal representa-
2988 tion of Unicode characters, there is no need to implement the somewhat
2989 messy concept of surrogates."
2991 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2992 ters in between are treated as literals. This is slightly different
2993 from Perl in that $ and @ are also handled as literals inside the
2994 quotes. In Perl, they cause variable interpolation (but of course PCRE
2995 does not have variables). Note the following examples:
2997 Pattern PCRE matches Perl matches
2999 \Qabc$xyz\E abc$xyz abc followed by the
3000 contents of $xyz
3001 \Qabc\$xyz\E abc\$xyz abc\$xyz
3002 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3004 The \Q...\E sequence is recognized both inside and outside character
3005 classes.
3007 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3008 constructions. However, there is support for recursive patterns. This
3009 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3010 "callout" feature allows an external function to be called during pat-
3011 tern matching. See the pcrecallout documentation for details.
3013 9. Subpatterns that are called recursively or as "subroutines" are
3014 always treated as atomic groups in PCRE. This is like Python, but
3015 unlike Perl. There is a discussion of an example that explains this in
3016 more detail in the section on recursion differences from Perl in the
3017 pcrepattern page.
3019 10. There are some differences that are concerned with the settings of
3020 captured strings when part of a pattern is repeated. For example,
3021 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3022 unset, but in PCRE it is set to "b".
3024 11. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3025 pattern names is not as general as Perl's. This is a consequence of the
3026 fact the PCRE works internally just with numbers, using an external ta-
3027 ble to translate between numbers and names. In particular, a pattern
3028 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3029 the same number but different names, is not supported, and causes an
3030 error at compile time. If it were allowed, it would not be possible to
3031 distinguish which parentheses matched, because both names map to cap-
3032 turing subpattern number 1. To avoid this confusing situation, an error
3033 is given at compile time.
3035 12. PCRE provides some extensions to the Perl regular expression facil-
3036 ities. Perl 5.10 includes new features that are not in earlier ver-
3037 sions of Perl, some of which (such as named parentheses) have been in
3038 PCRE for some time. This list is with respect to Perl 5.10:
3040 (a) Although lookbehind assertions in PCRE must match fixed length
3041 strings, each alternative branch of a lookbehind assertion can match a
3042 different length of string. Perl requires them all to have the same
3043 length.
3045 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3046 meta-character matches only at the very end of the string.
3048 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3049 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3050 ignored. (Perl can be made to issue a warning.)
3052 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3053 fiers is inverted, that is, by default they are not greedy, but if fol-
3054 lowed by a question mark they are.
3056 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3057 tried only at the first matching position in the subject string.
3060 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3061 lents.
3063 (g) The \R escape sequence can be restricted to match only CR, LF, or
3064 CRLF by the PCRE_BSR_ANYCRLF option.
3066 (h) The callout facility is PCRE-specific.
3068 (i) The partial matching facility is PCRE-specific.
3070 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3071 even on different hosts that have the other endianness.
3073 (k) The alternative matching function (pcre_dfa_exec()) matches in a
3074 different way and is not Perl-compatible.
3076 (l) PCRE recognizes some special sequences such as (*CR) at the start
3077 of a pattern that set overall options that cannot be changed within the
3078 pattern.
3083 Philip Hazel
3084 University Computing Service
3085 Cambridge CB2 3QH, England.
3090 Last updated: 12 May 2010
3091 Copyright (c) 1997-2010 University of Cambridge.
3092 ------------------------------------------------------------------------------
3098 NAME
3099 PCRE - Perl-compatible regular expressions
3104 The syntax and semantics of the regular expressions that are supported
3105 by PCRE are described in detail below. There is a quick-reference syn-
3106 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3107 semantics as closely as it can. PCRE also supports some alternative
3108 regular expression syntax (which does not conflict with the Perl syn-
3109 tax) in order to provide some compatibility with regular expressions in
3110 Python, .NET, and Oniguruma.
3112 Perl's regular expressions are described in its own documentation, and
3113 regular expressions in general are covered in a number of books, some
3114 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3115 Expressions", published by O'Reilly, covers regular expressions in
3116 great detail. This description of PCRE's regular expressions is
3117 intended as reference material.
3119 The original operation of PCRE was on strings of one-byte characters.
3120 However, there is now also support for UTF-8 character strings. To use
3121 this, PCRE must be built to include UTF-8 support, and you must call
3122 pcre_compile() or pcre_compile2() with the PCRE_UTF8 option. There is
3123 also a special sequence that can be given at the start of a pattern:
3125 (*UTF8)
3127 Starting a pattern with this sequence is equivalent to setting the
3128 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
3129 UTF-8 mode affects pattern matching is mentioned in several places
3130 below. There is also a summary of UTF-8 features in the section on
3131 UTF-8 support in the main pcre page.
3133 Another special sequence that may appear at the start of a pattern or
3134 in combination with (*UTF8) is:
3136 (*UCP)
3138 This has the same effect as setting the PCRE_UCP option: it causes
3139 sequences such as \d and \w to use Unicode properties to determine
3140 character types, instead of recognizing only characters with codes less
3141 than 128 via a lookup table.
3143 The remainder of this document discusses the patterns that are sup-
3144 ported by PCRE when its main matching function, pcre_exec(), is used.
3145 From release 6.0, PCRE offers a second matching function,
3146 pcre_dfa_exec(), which matches using a different algorithm that is not
3147 Perl-compatible. Some of the features discussed below are not available
3148 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3149 alternative function, and how it differs from the normal function, are
3150 discussed in the pcrematching page.
3155 PCRE supports five different conventions for indicating line breaks in
3156 strings: a single CR (carriage return) character, a single LF (line-
3157 feed) character, the two-character sequence CRLF, any of the three pre-
3158 ceding, or any Unicode newline sequence. The pcreapi page has further
3159 discussion about newlines, and shows how to set the newline convention
3160 in the options arguments for the compiling and matching functions.
3162 It is also possible to specify a newline convention by starting a pat-
3163 tern string with one of the following five sequences:
3165 (*CR) carriage return
3166 (*LF) linefeed
3167 (*CRLF) carriage return, followed by linefeed
3168 (*ANYCRLF) any of the three above
3169 (*ANY) all Unicode newline sequences
3171 These override the default and the options given to pcre_compile() or
3172 pcre_compile2(). For example, on a Unix system where LF is the default
3173 newline sequence, the pattern
3175 (*CR)a.b
3177 changes the convention to CR. That pattern matches "a\nb" because LF is
3178 no longer a newline. Note that these special settings, which are not
3179 Perl-compatible, are recognized only at the very start of a pattern,
3180 and that they must be in upper case. If more than one of them is
3181 present, the last one is used.
3183 The newline convention affects the interpretation of the dot metachar-
3184 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3185 ever, it does not affect what the \R escape sequence matches. By
3186 default, this is any Unicode newline sequence, for Perl compatibility.
3187 However, this can be changed; see the description of \R in the section
3188 entitled "Newline sequences" below. A change of \R setting can be com-
3189 bined with a change of newline convention.
3194 A regular expression is a pattern that is matched against a subject
3195 string from left to right. Most characters stand for themselves in a
3196 pattern, and match the corresponding characters in the subject. As a
3197 trivial example, the pattern
3199 The quick brown fox
3201 matches a portion of a subject string that is identical to itself. When
3202 caseless matching is specified (the PCRE_CASELESS option), letters are
3203 matched independently of case. In UTF-8 mode, PCRE always understands
3204 the concept of case for characters whose values are less than 128, so
3205 caseless matching is always possible. For characters with higher val-
3206 ues, the concept of case is supported if PCRE is compiled with Unicode
3207 property support, but not otherwise. If you want to use caseless
3208 matching for characters 128 and above, you must ensure that PCRE is
3209 compiled with Unicode property support as well as with UTF-8 support.
3211 The power of regular expressions comes from the ability to include
3212 alternatives and repetitions in the pattern. These are encoded in the
3213 pattern by the use of metacharacters, which do not stand for themselves
3214 but instead are interpreted in some special way.
3216 There are two different sets of metacharacters: those that are recog-
3217 nized anywhere in the pattern except within square brackets, and those
3218 that are recognized within square brackets. Outside square brackets,
3219 the metacharacters are as follows:
3221 \ general escape character with several uses
3222 ^ assert start of string (or line, in multiline mode)
3223 $ assert end of string (or line, in multiline mode)
3224 . match any character except newline (by default)
3225 [ start character class definition
3226 | start of alternative branch
3227 ( start subpattern
3228 ) end subpattern
3229 ? extends the meaning of (
3230 also 0 or 1 quantifier
3231 also quantifier minimizer
3232 * 0 or more quantifier
3233 + 1 or more quantifier
3234 also "possessive quantifier"
3235 { start min/max quantifier
3237 Part of a pattern that is in square brackets is called a "character
3238 class". In a character class the only metacharacters are:
3240 \ general escape character
3241 ^ negate the class, but only if the first character
3242 - indicates character range
3243 [ POSIX character class (only if followed by POSIX
3244 syntax)
3245 ] terminates the character class
3247 The following sections describe the use of each of the metacharacters.
3252 The backslash character has several uses. Firstly, if it is followed by
3253 a non-alphanumeric character, it takes away any special meaning that
3254 character may have. This use of backslash as an escape character
3255 applies both inside and outside character classes.
3257 For example, if you want to match a * character, you write \* in the
3258 pattern. This escaping action applies whether or not the following
3259 character would otherwise be interpreted as a metacharacter, so it is
3260 always safe to precede a non-alphanumeric with backslash to specify
3261 that it stands for itself. In particular, if you want to match a back-
3262 slash, you write \\.
3264 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3265 the pattern (other than in a character class) and characters between a
3266 # outside a character class and the next newline are ignored. An escap-
3267 ing backslash can be used to include a whitespace or # character as
3268 part of the pattern.
3270 If you want to remove the special meaning from a sequence of charac-
3271 ters, you can do so by putting them between \Q and \E. This is differ-
3272 ent from Perl in that $ and @ are handled as literals in \Q...\E
3273 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3274 tion. Note the following examples:
3276 Pattern PCRE matches Perl matches
3278 \Qabc$xyz\E abc$xyz abc followed by the
3279 contents of $xyz
3280 \Qabc\$xyz\E abc\$xyz abc\$xyz
3281 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3283 The \Q...\E sequence is recognized both inside and outside character
3284 classes.
3286 Non-printing characters
3288 A second use of backslash provides a way of encoding non-printing char-
3289 acters in patterns in a visible manner. There is no restriction on the
3290 appearance of non-printing characters, apart from the binary zero that
3291 terminates a pattern, but when a pattern is being prepared by text
3292 editing, it is often easier to use one of the following escape
3293 sequences than the binary character it represents:
3295 \a alarm, that is, the BEL character (hex 07)
3296 \cx "control-x", where x is any character
3297 \e escape (hex 1B)
3298 \f formfeed (hex 0C)
3299 \n linefeed (hex 0A)
3300 \r carriage return (hex 0D)
3301 \t tab (hex 09)
3302 \ddd character with octal code ddd, or back reference
3303 \xhh character with hex code hh
3304 \x{hhh..} character with hex code hhh..
3306 The precise effect of \cx is as follows: if x is a lower case letter,
3307 it is converted to upper case. Then bit 6 of the character (hex 40) is
3308 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3309 becomes hex 7B.
3311 After \x, from zero to two hexadecimal digits are read (letters can be
3312 in upper or lower case). Any number of hexadecimal digits may appear
3313 between \x{ and }, but the value of the character code must be less
3314 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3315 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3316 than the largest Unicode code point, which is 10FFFF.
3318 If characters other than hexadecimal digits appear between \x{ and },
3319 or if there is no terminating }, this form of escape is not recognized.
3320 Instead, the initial \x will be interpreted as a basic hexadecimal
3321 escape, with no following digits, giving a character whose value is
3322 zero.
3324 Characters whose value is less than 256 can be defined by either of the
3325 two syntaxes for \x. There is no difference in the way they are han-
3326 dled. For example, \xdc is exactly the same as \x{dc}.
3328 After \0 up to two further octal digits are read. If there are fewer
3329 than two digits, just those that are present are used. Thus the
3330 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3331 (code value 7). Make sure you supply two digits after the initial zero
3332 if the pattern character that follows is itself an octal digit.
3334 The handling of a backslash followed by a digit other than 0 is compli-
3335 cated. Outside a character class, PCRE reads it and any following dig-
3336 its as a decimal number. If the number is less than 10, or if there
3337 have been at least that many previous capturing left parentheses in the
3338 expression, the entire sequence is taken as a back reference. A
3339 description of how this works is given later, following the discussion
3340 of parenthesized subpatterns.
3342 Inside a character class, or if the decimal number is greater than 9
3343 and there have not been that many capturing subpatterns, PCRE re-reads
3344 up to three octal digits following the backslash, and uses them to gen-
3345 erate a data character. Any subsequent digits stand for themselves. In
3346 non-UTF-8 mode, the value of a character specified in octal must be
3347 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3348 example:
3350 \040 is another way of writing a space
3351 \40 is the same, provided there are fewer than 40
3352 previous capturing subpatterns
3353 \7 is always a back reference
3354 \11 might be a back reference, or another way of
3355 writing a tab
3356 \011 is always a tab
3357 \0113 is a tab followed by the character "3"
3358 \113 might be a back reference, otherwise the
3359 character with octal code 113
3360 \377 might be a back reference, otherwise
3361 the byte consisting entirely of 1 bits
3362 \81 is either a back reference, or a binary zero
3363 followed by the two characters "8" and "1"
3365 Note that octal values of 100 or greater must not be introduced by a
3366 leading zero, because no more than three octal digits are ever read.
3368 All the sequences that define a single character value can be used both
3369 inside and outside character classes. In addition, inside a character
3370 class, the sequence \b is interpreted as the backspace character (hex
3371 08). The sequences \B, \N, \R, and \X are not special inside a charac-
3372 ter class. Like any other unrecognized escape sequences, they are
3373 treated as the literal characters "B", "N", "R", and "X" by default,
3374 but cause an error if the PCRE_EXTRA option is set. Outside a character
3375 class, these sequences have different meanings.
3377 Absolute and relative back references
3379 The sequence \g followed by an unsigned or a negative number, option-
3380 ally enclosed in braces, is an absolute or relative back reference. A
3381 named back reference can be coded as \g{name}. Back references are dis-
3382 cussed later, following the discussion of parenthesized subpatterns.
3384 Absolute and relative subroutine calls
3386 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3387 name or a number enclosed either in angle brackets or single quotes, is
3388 an alternative syntax for referencing a subpattern as a "subroutine".
3389 Details are discussed later. Note that \g{...} (Perl syntax) and
3390 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3391 reference; the latter is a subroutine call.
3393 Generic character types
3395 Another use of backslash is for specifying generic character types:
3397 \d any decimal digit
3398 \D any character that is not a decimal digit
3399 \h any horizontal whitespace character
3400 \H any character that is not a horizontal whitespace character
3401 \s any whitespace character
3402 \S any character that is not a whitespace character
3403 \v any vertical whitespace character
3404 \V any character that is not a vertical whitespace character
3405 \w any "word" character
3406 \W any "non-word" character
3408 There is also the single sequence \N, which matches a non-newline char-
3409 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
3410 not set.
3412 Each pair of lower and upper case escape sequences partitions the com-
3413 plete set of characters into two disjoint sets. Any given character
3414 matches one, and only one, of each pair. The sequences can appear both
3415 inside and outside character classes. They each match one character of
3416 the appropriate type. If the current matching point is at the end of
3417 the subject string, all of them fail, because there is no character to
3418 match.
3420 For compatibility with Perl, \s does not match the VT character (code
3421 11). This makes it different from the the POSIX "space" class. The \s
3422 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3423 "use locale;" is included in a Perl script, \s may match the VT charac-
3424 ter. In PCRE, it never does.
3426 A "word" character is an underscore or any character that is a letter
3427 or digit. By default, the definition of letters and digits is con-
3428 trolled by PCRE's low-valued character tables, and may vary if locale-
3429 specific matching is taking place (see "Locale support" in the pcreapi
3430 page). For example, in a French locale such as "fr_FR" in Unix-like
3431 systems, or "french" in Windows, some character codes greater than 128
3432 are used for accented letters, and these are then matched by \w. The
3433 use of locales with Unicode is discouraged.
3435 By default, in UTF-8 mode, characters with values greater than 128
3436 never match \d, \s, or \w, and always match \D, \S, and \W. These
3437 sequences retain their original meanings from before UTF-8 support was
3438 available, mainly for efficiency reasons. However, if PCRE is compiled
3439 with Unicode property support, and the PCRE_UCP option is set, the be-
3440 haviour is changed so that Unicode properties are used to determine
3441 character types, as follows:
3443 \d any character that \p{Nd} matches (decimal digit)
3444 \s any character that \p{Z} matches, plus HT, LF, FF, CR
3445 \w any character that \p{L} or \p{N} matches, plus underscore
3447 The upper case escapes match the inverse sets of characters. Note that
3448 \d matches only decimal digits, whereas \w matches any Unicode digit,
3449 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
3450 affects \b, and \B because they are defined in terms of \w and \W.
3451 Matching these sequences is noticeably slower when PCRE_UCP is set.
3453 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3454 the other sequences, which match only ASCII characters by default,
3455 these always match certain high-valued codepoints in UTF-8 mode,
3456 whether or not PCRE_UCP is set. The horizontal space characters are:
3458 U+0009 Horizontal tab
3459 U+0020 Space
3460 U+00A0 Non-break space
3461 U+1680 Ogham space mark
3462 U+180E Mongolian vowel separator
3463 U+2000 En quad
3464 U+2001 Em quad
3465 U+2002 En space
3466 U+2003 Em space
3467 U+2004 Three-per-em space
3468 U+2005 Four-per-em space
3469 U+2006 Six-per-em space
3470 U+2007 Figure space
3471 U+2008 Punctuation space
3472 U+2009 Thin space
3473 U+200A Hair space
3474 U+202F Narrow no-break space
3475 U+205F Medium mathematical space
3476 U+3000 Ideographic space
3478 The vertical space characters are:
3480 U+000A Linefeed
3481 U+000B Vertical tab
3482 U+000C Formfeed
3483 U+000D Carriage return
3484 U+0085 Next line
3485 U+2028 Line separator
3486 U+2029 Paragraph separator
3488 Newline sequences
3490 Outside a character class, by default, the escape sequence \R matches
3491 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3492 mode \R is equivalent to the following:
3494 (?>\r\n|\n|\x0b|\f|\r|\x85)
3496 This is an example of an "atomic group", details of which are given
3497 below. This particular group matches either the two-character sequence
3498 CR followed by LF, or one of the single characters LF (linefeed,
3499 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3500 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3501 is treated as a single unit that cannot be split.
3503 In UTF-8 mode, two additional characters whose codepoints are greater
3504 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3505 rator, U+2029). Unicode character property support is not needed for
3506 these characters to be recognized.
3508 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3509 the complete set of Unicode line endings) by setting the option
3510 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3511 (BSR is an abbrevation for "backslash R".) This can be made the default
3512 when PCRE is built; if this is the case, the other behaviour can be
3513 requested via the PCRE_BSR_UNICODE option. It is also possible to
3514 specify these settings by starting a pattern string with one of the
3515 following sequences:
3517 (*BSR_ANYCRLF) CR, LF, or CRLF only
3518 (*BSR_UNICODE) any Unicode newline sequence
3520 These override the default and the options given to pcre_compile() or
3521 pcre_compile2(), but they can be overridden by options given to
3522 pcre_exec() or pcre_dfa_exec(). Note that these special settings, which
3523 are not Perl-compatible, are recognized only at the very start of a
3524 pattern, and that they must be in upper case. If more than one of them
3525 is present, the last one is used. They can be combined with a change of
3526 newline convention; for example, a pattern can start with:
3530 They can also be combined with the (*UTF8) or (*UCP) special sequences.
3531 Inside a character class, \R is treated as an unrecognized escape
3532 sequence, and so matches the letter "R" by default, but causes an error
3533 if PCRE_EXTRA is set.
3535 Unicode character properties
3537 When PCRE is built with Unicode character property support, three addi-
3538 tional escape sequences that match characters with specific properties
3539 are available. When not in UTF-8 mode, these sequences are of course
3540 limited to testing characters whose codepoints are less than 256, but
3541 they do work in this mode. The extra escape sequences are:
3543 \p{xx} a character with the xx property
3544 \P{xx} a character without the xx property
3545 \X an extended Unicode sequence
3547 The property names represented by xx above are limited to the Unicode
3548 script names, the general category properties, "Any", which matches any
3549 character (including newline), and some special PCRE properties
3550 (described in the next section). Other Perl properties such as "InMu-
3551 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
3552 does not match any characters, so always causes a match failure.
3554 Sets of Unicode characters are defined as belonging to certain scripts.
3555 A character from one of these sets can be matched using a script name.
3556 For example:
3558 \p{Greek}
3559 \P{Han}
3561 Those that are not part of an identified script are lumped together as
3562 "Common". The current list of scripts is:
3564 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
3565 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
3566 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
3567 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
3568 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
3569 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
3570 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
3571 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
3572 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
3573 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
3574 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
3575 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
3576 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
3577 Ugaritic, Vai, Yi.
3579 Each character has exactly one Unicode general category property, spec-
3580 ified by a two-letter abbreviation. For compatibility with Perl, nega-
3581 tion can be specified by including a circumflex between the opening
3582 brace and the property name. For example, \p{^Lu} is the same as
3583 \P{Lu}.
3585 If only one letter is specified with \p or \P, it includes all the gen-
3586 eral category properties that start with that letter. In this case, in
3587 the absence of negation, the curly brackets in the escape sequence are
3588 optional; these two examples have the same effect:
3590 \p{L}
3591 \pL
3593 The following general category property codes are supported:
3595 C Other
3596 Cc Control
3597 Cf Format
3598 Cn Unassigned
3599 Co Private use
3600 Cs Surrogate
3602 L Letter
3603 Ll Lower case letter
3604 Lm Modifier letter
3605 Lo Other letter
3606 Lt Title case letter
3607 Lu Upper case letter
3609 M Mark
3610 Mc Spacing mark
3611 Me Enclosing mark
3612 Mn Non-spacing mark
3614 N Number
3615 Nd Decimal number
3616 Nl Letter number
3617 No Other number
3619 P Punctuation
3620 Pc Connector punctuation
3621 Pd Dash punctuation
3622 Pe Close punctuation
3623 Pf Final punctuation
3624 Pi Initial punctuation
3625 Po Other punctuation
3626 Ps Open punctuation
3628 S Symbol
3629 Sc Currency symbol
3630 Sk Modifier symbol
3631 Sm Mathematical symbol
3632 So Other symbol
3634 Z Separator
3635 Zl Line separator
3636 Zp Paragraph separator
3637 Zs Space separator
3639 The special property L& is also supported: it matches a character that
3640 has the Lu, Ll, or Lt property, in other words, a letter that is not
3641 classified as a modifier or "other".
3643 The Cs (Surrogate) property applies only to characters in the range
3644 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3645 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3646 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3647 the pcreapi page). Perl does not support the Cs property.
3649 The long synonyms for property names that Perl supports (such as
3650 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3651 any of these properties with "Is".
3653 No character that is in the Unicode table has the Cn (unassigned) prop-
3654 erty. Instead, this property is assumed for any code point that is not
3655 in the Unicode table.
3657 Specifying caseless matching does not affect these escape sequences.
3658 For example, \p{Lu} always matches only upper case letters.
3660 The \X escape matches any number of Unicode characters that form an
3661 extended Unicode sequence. \X is equivalent to
3663 (?>\PM\pM*)
3665 That is, it matches a character without the "mark" property, followed
3666 by zero or more characters with the "mark" property, and treats the
3667 sequence as an atomic group (see below). Characters with the "mark"
3668 property are typically accents that affect the preceding character.
3669 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3670 matches any one character.
3672 Matching characters by Unicode property is not fast, because PCRE has
3673 to search a structure that contains data for over fifteen thousand
3674 characters. That is why the traditional escape sequences such as \d and
3675 \w do not use Unicode properties in PCRE by default, though you can
3676 make them do so by setting the PCRE_UCP option for pcre_compile() or by
3677 starting the pattern with (*UCP).
3679 PCRE's additional properties
3681 As well as the standard Unicode properties described in the previous
3682 section, PCRE supports four more that make it possible to convert tra-
3683 ditional escape sequences such as \w and \s and POSIX character classes
3684 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
3685 erties internally when PCRE_UCP is set. They are:
3687 Xan Any alphanumeric character
3688 Xps Any POSIX space character
3689 Xsp Any Perl space character
3690 Xwd Any Perl "word" character
3692 Xan matches characters that have either the L (letter) or the N (num-
3693 ber) property. Xps matches the characters tab, linefeed, vertical tab,
3694 formfeed, or carriage return, and any other character that has the Z
3695 (separator) property. Xsp is the same as Xps, except that vertical tab
3696 is excluded. Xwd matches the same characters as Xan, plus underscore.
3698 Resetting the match start
3700 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3701 ously matched characters not to be included in the final matched
3702 sequence. For example, the pattern:
3704 foo\Kbar
3706 matches "foobar", but reports that it has matched "bar". This feature
3707 is similar to a lookbehind assertion (described below). However, in
3708 this case, the part of the subject before the real match does not have
3709 to be of fixed length, as lookbehind assertions do. The use of \K does
3710 not interfere with the setting of captured substrings. For example,
3711 when the pattern
3713 (foo)\Kbar
3715 matches "foobar", the first substring is still set to "foo".
3717 Perl documents that the use of \K within assertions is "not well
3718 defined". In PCRE, \K is acted upon when it occurs inside positive
3719 assertions, but is ignored in negative assertions.
3721 Simple assertions
3723 The final use of backslash is for certain simple assertions. An asser-
3724 tion specifies a condition that has to be met at a particular point in
3725 a match, without consuming any characters from the subject string. The
3726 use of subpatterns for more complicated assertions is described below.
3727 The backslashed assertions are:
3729 \b matches at a word boundary
3730 \B matches when not at a word boundary
3731 \A matches at the start of the subject
3732 \Z matches at the end of the subject
3733 also matches before a newline at the end of the subject
3734 \z matches only at the end of the subject
3735 \G matches at the first matching position in the subject
3737 Inside a character class, \b has a different meaning; it matches the
3738 backspace character. If any other of these assertions appears in a
3739 character class, by default it matches the corresponding literal char-
3740 acter (for example, \B matches the letter B). However, if the
3741 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
3742 ated instead.
3744 A word boundary is a position in the subject string where the current
3745 character and the previous character do not both match \w or \W (i.e.
3746 one matches \w and the other matches \W), or the start or end of the
3747 string if the first or last character matches \w, respectively. In
3748 UTF-8 mode, the meanings of \w and \W can be changed by setting the
3749 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
3750 PCRE nor Perl has a separate "start of word" or "end of word" metase-
3751 quence. However, whatever follows \b normally determines which it is.
3752 For example, the fragment \ba matches "a" at the start of a word.
3754 The \A, \Z, and \z assertions differ from the traditional circumflex
3755 and dollar (described in the next section) in that they only ever match
3756 at the very start and end of the subject string, whatever options are
3757 set. Thus, they are independent of multiline mode. These three asser-
3758 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3759 affect only the behaviour of the circumflex and dollar metacharacters.
3760 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3761 cating that matching is to start at a point other than the beginning of
3762 the subject, \A can never match. The difference between \Z and \z is
3763 that \Z matches before a newline at the end of the string as well as at
3764 the very end, whereas \z matches only at the end.
3766 The \G assertion is true only when the current matching position is at
3767 the start point of the match, as specified by the startoffset argument
3768 of pcre_exec(). It differs from \A when the value of startoffset is
3769 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3770 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3771 mentation where \G can be useful.
3773 Note, however, that PCRE's interpretation of \G, as the start of the
3774 current match, is subtly different from Perl's, which defines it as the
3775 end of the previous match. In Perl, these can be different when the
3776 previously matched string was empty. Because PCRE does just one match
3777 at a time, it cannot reproduce this behaviour.
3779 If all the alternatives of a pattern begin with \G, the expression is
3780 anchored to the starting match position, and the "anchored" flag is set
3781 in the compiled regular expression.
3786 Outside a character class, in the default matching mode, the circumflex
3787 character is an assertion that is true only if the current matching
3788 point is at the start of the subject string. If the startoffset argu-
3789 ment of pcre_exec() is non-zero, circumflex can never match if the
3790 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3791 has an entirely different meaning (see below).
3793 Circumflex need not be the first character of the pattern if a number
3794 of alternatives are involved, but it should be the first thing in each
3795 alternative in which it appears if the pattern is ever to match that
3796 branch. If all possible alternatives start with a circumflex, that is,
3797 if the pattern is constrained to match only at the start of the sub-
3798 ject, it is said to be an "anchored" pattern. (There are also other
3799 constructs that can cause a pattern to be anchored.)
3801 A dollar character is an assertion that is true only if the current
3802 matching point is at the end of the subject string, or immediately
3803 before a newline at the end of the string (by default). Dollar need not
3804 be the last character of the pattern if a number of alternatives are
3805 involved, but it should be the last item in any branch in which it
3806 appears. Dollar has no special meaning in a character class.
3808 The meaning of dollar can be changed so that it matches only at the
3809 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3810 compile time. This does not affect the \Z assertion.
3812 The meanings of the circumflex and dollar characters are changed if the
3813 PCRE_MULTILINE option is set. When this is the case, a circumflex
3814 matches immediately after internal newlines as well as at the start of
3815 the subject string. It does not match after a newline that ends the
3816 string. A dollar matches before any newlines in the string, as well as
3817 at the very end, when PCRE_MULTILINE is set. When newline is specified
3818 as the two-character sequence CRLF, isolated CR and LF characters do
3819 not indicate newlines.
3821 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3822 (where \n represents a newline) in multiline mode, but not otherwise.
3823 Consequently, patterns that are anchored in single line mode because
3824 all branches start with ^ are not anchored in multiline mode, and a
3825 match for circumflex is possible when the startoffset argument of
3826 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3827 PCRE_MULTILINE is set.
3829 Note that the sequences \A, \Z, and \z can be used to match the start
3830 and end of the subject in both modes, and if all branches of a pattern
3831 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3832 set.
3837 Outside a character class, a dot in the pattern matches any one charac-
3838 ter in the subject string except (by default) a character that signi-
3839 fies the end of a line. In UTF-8 mode, the matched character may be
3840 more than one byte long.
3842 When a line ending is defined as a single character, dot never matches
3843 that character; when the two-character sequence CRLF is used, dot does
3844 not match CR if it is immediately followed by LF, but otherwise it
3845 matches all characters (including isolated CRs and LFs). When any Uni-
3846 code line endings are being recognized, dot does not match CR or LF or
3847 any of the other line ending characters.
3849 The behaviour of dot with regard to newlines can be changed. If the
3850 PCRE_DOTALL option is set, a dot matches any one character, without
3851 exception. If the two-character sequence CRLF is present in the subject
3852 string, it takes two dots to match it.
3854 The handling of dot is entirely independent of the handling of circum-
3855 flex and dollar, the only relationship being that they both involve
3856 newlines. Dot has no special meaning in a character class.
3858 The escape sequence \N always behaves as a dot does when PCRE_DOTALL is
3859 not set. In other words, it matches any one character except one that
3860 signifies the end of a line.
3865 Outside a character class, the escape sequence \C matches any one byte,
3866 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3867 line-ending characters. The feature is provided in Perl in order to
3868 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3869 acters into individual bytes, what remains in the string may be a mal-
3870 formed UTF-8 string. For this reason, the \C escape sequence is best
3871 avoided.
3873 PCRE does not allow \C to appear in lookbehind assertions (described
3874 below), because in UTF-8 mode this would make it impossible to calcu-
3875 late the length of the lookbehind.
3880 An opening square bracket introduces a character class, terminated by a
3881 closing square bracket. A closing square bracket on its own is not spe-
3882 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
3883 a lone closing square bracket causes a compile-time error. If a closing
3884 square bracket is required as a member of the class, it should be the
3885 first data character in the class (after an initial circumflex, if
3886 present) or escaped with a backslash.
3888 A character class matches a single character in the subject. In UTF-8
3889 mode, the character may be more than one byte long. A matched character
3890 must be in the set of characters defined by the class, unless the first
3891 character in the class definition is a circumflex, in which case the
3892 subject character must not be in the set defined by the class. If a
3893 circumflex is actually required as a member of the class, ensure it is
3894 not the first character, or escape it with a backslash.
3896 For example, the character class [aeiou] matches any lower case vowel,
3897 while [^aeiou] matches any character that is not a lower case vowel.
3898 Note that a circumflex is just a convenient notation for specifying the
3899 characters that are in the class by enumerating those that are not. A
3900 class that starts with a circumflex is not an assertion; it still con-
3901 sumes a character from the subject string, and therefore it fails if
3902 the current pointer is at the end of the string.
3904 In UTF-8 mode, characters with values greater than 255 can be included
3905 in a class as a literal string of bytes, or by using the \x{ escaping
3906 mechanism.
3908 When caseless matching is set, any letters in a class represent both
3909 their upper case and lower case versions, so for example, a caseless
3910 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3911 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3912 understands the concept of case for characters whose values are less
3913 than 128, so caseless matching is always possible. For characters with
3914 higher values, the concept of case is supported if PCRE is compiled
3915 with Unicode property support, but not otherwise. If you want to use
3916 caseless matching in UTF8-mode for characters 128 and above, you must
3917 ensure that PCRE is compiled with Unicode property support as well as
3918 with UTF-8 support.
3920 Characters that might indicate line breaks are never treated in any
3921 special way when matching character classes, whatever line-ending
3922 sequence is in use, and whatever setting of the PCRE_DOTALL and
3923 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3924 of these characters.
3926 The minus (hyphen) character can be used to specify a range of charac-
3927 ters in a character class. For example, [d-m] matches any letter
3928 between d and m, inclusive. If a minus character is required in a
3929 class, it must be escaped with a backslash or appear in a position
3930 where it cannot be interpreted as indicating a range, typically as the
3931 first or last character in the class.
3933 It is not possible to have the literal character "]" as the end charac-
3934 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3935 two characters ("W" and "-") followed by a literal string "46]", so it
3936 would match "W46]" or "-46]". However, if the "]" is escaped with a
3937 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3938 preted as a class containing a range followed by two other characters.
3939 The octal or hexadecimal representation of "]" can also be used to end
3940 a range.
3942 Ranges operate in the collating sequence of character values. They can
3943 also be used for characters specified numerically, for example
3944 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3945 are greater than 255, for example [\x{100}-\x{2ff}].
3947 If a range that includes letters is used when caseless matching is set,
3948 it matches the letters in either case. For example, [W-c] is equivalent
3949 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3950 character tables for a French locale are in use, [\xc8-\xcb] matches
3951 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3952 concept of case for characters with values greater than 128 only when
3953 it is compiled with Unicode property support.
3955 The character types \d, \D, \h, \H, \p, \P, \s, \S, \v, \V, \w, and \W
3956 may also appear in a character class, and add the characters that they
3957 match to the class. For example, [\dABCDEF] matches any hexadecimal
3958 digit. A circumflex can conveniently be used with the upper case char-
3959 acter types to specify a more restricted set of characters than the
3960 matching lower case type. For example, the class [^\W_] matches any
3961 letter or digit, but not underscore.
3963 The only metacharacters that are recognized in character classes are
3964 backslash, hyphen (only where it can be interpreted as specifying a
3965 range), circumflex (only at the start), opening square bracket (only
3966 when it can be interpreted as introducing a POSIX class name - see the
3967 next section), and the terminating closing square bracket. However,
3968 escaping other non-alphanumeric characters does no harm.
3973 Perl supports the POSIX notation for character classes. This uses names
3974 enclosed by [: and :] within the enclosing square brackets. PCRE also
3975 supports this notation. For example,
3977 [01[:alpha:]%]
3979 matches "0", "1", any alphabetic character, or "%". The supported class
3980 names are:
3982 alnum letters and digits
3983 alpha letters
3984 ascii character codes 0 - 127
3985 blank space or tab only
3986 cntrl control characters
3987 digit decimal digits (same as \d)
3988 graph printing characters, excluding space
3989 lower lower case letters
3990 print printing characters, including space
3991 punct printing characters, excluding letters and digits and space
3992 space white space (not quite the same as \s)
3993 upper upper case letters
3994 word "word" characters (same as \w)
3995 xdigit hexadecimal digits
3997 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3998 and space (32). Notice that this list includes the VT character (code
3999 11). This makes "space" different to \s, which does not include VT (for
4000 Perl compatibility).
4002 The name "word" is a Perl extension, and "blank" is a GNU extension
4003 from Perl 5.8. Another Perl extension is negation, which is indicated
4004 by a ^ character after the colon. For example,
4006 [12[:^digit:]]
4008 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4009 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4010 these are not supported, and an error is given if they are encountered.
4012 By default, in UTF-8 mode, characters with values greater than 128 do
4013 not match any of the POSIX character classes. However, if the PCRE_UCP
4014 option is passed to pcre_compile(), some of the classes are changed so
4015 that Unicode character properties are used. This is achieved by replac-
4016 ing the POSIX classes by other sequences, as follows:
4018 [:alnum:] becomes \p{Xan}
4019 [:alpha:] becomes \p{L}
4020 [:blank:] becomes \h
4021 [:digit:] becomes \p{Nd}
4022 [:lower:] becomes \p{Ll}
4023 [:space:] becomes \p{Xps}
4024 [:upper:] becomes \p{Lu}
4025 [:word:] becomes \p{Xwd}
4027 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4028 POSIX classes are unchanged, and match only characters with code points
4029 less than 128.
4034 Vertical bar characters are used to separate alternative patterns. For
4035 example, the pattern
4037 gilbert|sullivan
4039 matches either "gilbert" or "sullivan". Any number of alternatives may
4040 appear, and an empty alternative is permitted (matching the empty
4041 string). The matching process tries each alternative in turn, from left
4042 to right, and the first one that succeeds is used. If the alternatives
4043 are within a subpattern (defined below), "succeeds" means matching the
4044 rest of the main pattern as well as the alternative in the subpattern.
4049 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4050 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4051 within the pattern by a sequence of Perl option letters enclosed
4052 between "(?" and ")". The option letters are
4054 i for PCRE_CASELESS
4056 s for PCRE_DOTALL
4057 x for PCRE_EXTENDED
4059 For example, (?im) sets caseless, multiline matching. It is also possi-
4060 ble to unset these options by preceding the letter with a hyphen, and a
4061 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4063 is also permitted. If a letter appears both before and after the
4064 hyphen, the option is unset.
4067 can be changed in the same way as the Perl-compatible options by using
4068 the characters J, U and X respectively.
4070 When one of these option changes occurs at top level (that is, not
4071 inside subpattern parentheses), the change applies to the remainder of
4072 the pattern that follows. If the change is placed right at the start of
4073 a pattern, PCRE extracts it into the global options (and it will there-
4074 fore show up in data extracted by the pcre_fullinfo() function).
4076 An option change within a subpattern (see below for a description of
4077 subpatterns) affects only that part of the current pattern that follows
4078 it, so
4080 (a(?i)b)c
4082 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4083 used). By this means, options can be made to have different settings
4084 in different parts of the pattern. Any changes made in one alternative
4085 do carry on into subsequent branches within the same subpattern. For
4086 example,
4088 (a(?i)b|c)
4090 matches "ab", "aB", "c", and "C", even though when matching "C" the
4091 first branch is abandoned before the option setting. This is because
4092 the effects of option settings happen at compile time. There would be
4093 some very weird behaviour otherwise.
4095 Note: There are other PCRE-specific options that can be set by the
4096 application when the compile or match functions are called. In some
4097 cases the pattern can contain special leading sequences such as (*CRLF)
4098 to override what the application has set or what has been defaulted.
4099 Details are given in the section entitled "Newline sequences" above.
4100 There are also the (*UTF8) and (*UCP) leading sequences that can be
4101 used to set UTF-8 and Unicode property modes; they are equivalent to
4102 setting the PCRE_UTF8 and the PCRE_UCP options, respectively.
4107 Subpatterns are delimited by parentheses (round brackets), which can be
4108 nested. Turning part of a pattern into a subpattern does two things:
4110 1. It localizes a set of alternatives. For example, the pattern
4112 cat(aract|erpillar|)
4114 matches one of the words "cat", "cataract", or "caterpillar". Without
4115 the parentheses, it would match "cataract", "erpillar" or an empty
4116 string.
4118 2. It sets up the subpattern as a capturing subpattern. This means
4119 that, when the whole pattern matches, that portion of the subject
4120 string that matched the subpattern is passed back to the caller via the
4121 ovector argument of pcre_exec(). Opening parentheses are counted from
4122 left to right (starting from 1) to obtain numbers for the capturing
4123 subpatterns.
4125 For example, if the string "the red king" is matched against the pat-
4126 tern
4128 the ((red|white) (king|queen))
4130 the captured substrings are "red king", "red", and "king", and are num-
4131 bered 1, 2, and 3, respectively.
4133 The fact that plain parentheses fulfil two functions is not always
4134 helpful. There are often times when a grouping subpattern is required
4135 without a capturing requirement. If an opening parenthesis is followed
4136 by a question mark and a colon, the subpattern does not do any captur-
4137 ing, and is not counted when computing the number of any subsequent
4138 capturing subpatterns. For example, if the string "the white queen" is
4139 matched against the pattern
4141 the ((?:red|white) (king|queen))
4143 the captured substrings are "white queen" and "queen", and are numbered
4144 1 and 2. The maximum number of capturing subpatterns is 65535.
4146 As a convenient shorthand, if any option settings are required at the
4147 start of a non-capturing subpattern, the option letters may appear
4148 between the "?" and the ":". Thus the two patterns
4150 (?i:saturday|sunday)
4151 (?:(?i)saturday|sunday)
4153 match exactly the same set of strings. Because alternative branches are
4154 tried from left to right, and options are not reset until the end of
4155 the subpattern is reached, an option setting in one branch does affect
4156 subsequent branches, so the above patterns match "SUNDAY" as well as
4157 "Saturday".
4162 Perl 5.10 introduced a feature whereby each alternative in a subpattern
4163 uses the same numbers for its capturing parentheses. Such a subpattern
4164 starts with (?| and is itself a non-capturing subpattern. For example,
4165 consider this pattern:
4167 (?|(Sat)ur|(Sun))day
4169 Because the two alternatives are inside a (?| group, both sets of cap-
4170 turing parentheses are numbered one. Thus, when the pattern matches,
4171 you can look at captured substring number one, whichever alternative
4172 matched. This construct is useful when you want to capture part, but
4173 not all, of one of a number of alternatives. Inside a (?| group, paren-
4174 theses are numbered as usual, but the number is reset at the start of
4175 each branch. The numbers of any capturing buffers that follow the sub-
4176 pattern start after the highest number used in any branch. The follow-
4177 ing example is taken from the Perl documentation. The numbers under-
4178 neath show in which buffer the captured content will be stored.
4180 # before ---------------branch-reset----------- after
4181 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
4182 # 1 2 2 3 2 3 4
4184 A back reference to a numbered subpattern uses the most recent value
4185 that is set for that number by any subpattern. The following pattern
4186 matches "abcabc" or "defdef":
4188 /(?|(abc)|(def))\1/
4190 In contrast, a recursive or "subroutine" call to a numbered subpattern
4191 always refers to the first one in the pattern with the given number.
4192 The following pattern matches "abcabc" or "defabc":
4194 /(?|(abc)|(def))(?1)/
4196 If a condition test for a subpattern's having matched refers to a non-
4197 unique number, the test is true if any of the subpatterns of that num-
4198 ber have matched.
4200 An alternative approach to using this "branch reset" feature is to use
4201 duplicate named subpatterns, as described in the next section.
4206 Identifying capturing parentheses by number is simple, but it can be
4207 very hard to keep track of the numbers in complicated regular expres-
4208 sions. Furthermore, if an expression is modified, the numbers may
4209 change. To help with this difficulty, PCRE supports the naming of sub-
4210 patterns. This feature was not added to Perl until release 5.10. Python
4211 had the feature earlier, and PCRE introduced it at release 4.0, using
4212 the Python syntax. PCRE now supports both the Perl and the Python syn-
4213 tax. Perl allows identically numbered subpatterns to have different
4214 names, but PCRE does not.
4216 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
4217 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
4218 to capturing parentheses from other parts of the pattern, such as back
4219 references, recursion, and conditions, can be made by name as well as
4220 by number.
4222 Names consist of up to 32 alphanumeric characters and underscores.
4223 Named capturing parentheses are still allocated numbers as well as
4224 names, exactly as if the names were not present. The PCRE API provides
4225 function calls for extracting the name-to-number translation table from
4226 a compiled pattern. There is also a convenience function for extracting
4227 a captured substring by name.
4229 By default, a name must be unique within a pattern, but it is possible
4230 to relax this constraint by setting the PCRE_DUPNAMES option at compile
4231 time. (Duplicate names are also always permitted for subpatterns with
4232 the same number, set up as described in the previous section.) Dupli-
4233 cate names can be useful for patterns where only one instance of the
4234 named parentheses can match. Suppose you want to match the name of a
4235 weekday, either as a 3-letter abbreviation or as the full name, and in
4236 both cases you want to extract the abbreviation. This pattern (ignoring
4237 the line breaks) does the job:
4239 (?<DN>Mon|Fri|Sun)(?:day)?|
4240 (?<DN>Tue)(?:sday)?|
4241 (?<DN>Wed)(?:nesday)?|
4242 (?<DN>Thu)(?:rsday)?|
4243 (?<DN>Sat)(?:urday)?
4245 There are five capturing substrings, but only one is ever set after a
4246 match. (An alternative way of solving this problem is to use a "branch
4247 reset" subpattern, as described in the previous section.)
4249 The convenience function for extracting the data by name returns the
4250 substring for the first (and in this example, the only) subpattern of
4251 that name that matched. This saves searching to find which numbered
4252 subpattern it was.
4254 If you make a back reference to a non-unique named subpattern from
4255 elsewhere in the pattern, the one that corresponds to the first occur-
4256 rence of the name is used. In the absence of duplicate numbers (see the
4257 previous section) this is the one with the lowest number. If you use a
4258 named reference in a condition test (see the section about conditions
4259 below), either to check whether a subpattern has matched, or to check
4260 for recursion, all subpatterns with the same name are tested. If the
4261 condition is true for any one of them, the overall condition is true.
4262 This is the same behaviour as testing by number. For further details of
4263 the interfaces for handling named subpatterns, see the pcreapi documen-
4264 tation.
4266 Warning: You cannot use different names to distinguish between two sub-
4267 patterns with the same number because PCRE uses only the numbers when
4268 matching. For this reason, an error is given at compile time if differ-
4269 ent names are given to subpatterns with the same number. However, you
4270 can give the same name to subpatterns with the same number, even when
4271 PCRE_DUPNAMES is not set.
4276 Repetition is specified by quantifiers, which can follow any of the
4277 following items:
4279 a literal data character
4280 the dot metacharacter
4281 the \C escape sequence
4282 the \X escape sequence (in UTF-8 mode with Unicode properties)
4283 the \R escape sequence
4284 an escape such as \d that matches a single character
4285 a character class
4286 a back reference (see next section)
4287 a parenthesized subpattern (unless it is an assertion)
4288 a recursive or "subroutine" call to a subpattern
4290 The general repetition quantifier specifies a minimum and maximum num-
4291 ber of permitted matches, by giving the two numbers in curly brackets
4292 (braces), separated by a comma. The numbers must be less than 65536,
4293 and the first must be less than or equal to the second. For example:
4295 z{2,4}
4297 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4298 special character. If the second number is omitted, but the comma is
4299 present, there is no upper limit; if the second number and the comma
4300 are both omitted, the quantifier specifies an exact number of required
4301 matches. Thus
4303 [aeiou]{3,}
4305 matches at least 3 successive vowels, but may match many more, while
4307 \d{8}
4309 matches exactly 8 digits. An opening curly bracket that appears in a
4310 position where a quantifier is not allowed, or one that does not match
4311 the syntax of a quantifier, is taken as a literal character. For exam-
4312 ple, {,6} is not a quantifier, but a literal string of four characters.
4314 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4315 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4316 acters, each of which is represented by a two-byte sequence. Similarly,
4317 when Unicode property support is available, \X{3} matches three Unicode
4318 extended sequences, each of which may be several bytes long (and they
4319 may be of different lengths).
4321 The quantifier {0} is permitted, causing the expression to behave as if
4322 the previous item and the quantifier were not present. This may be use-
4323 ful for subpatterns that are referenced as subroutines from elsewhere
4324 in the pattern. Items other than subpatterns that have a {0} quantifier
4325 are omitted from the compiled pattern.
4327 For convenience, the three most common quantifiers have single-charac-
4328 ter abbreviations:
4330 * is equivalent to {0,}
4331 + is equivalent to {1,}
4332 ? is equivalent to {0,1}
4334 It is possible to construct infinite loops by following a subpattern
4335 that can match no characters with a quantifier that has no upper limit,
4336 for example:
4338 (a?)*
4340 Earlier versions of Perl and PCRE used to give an error at compile time
4341 for such patterns. However, because there are cases where this can be
4342 useful, such patterns are now accepted, but if any repetition of the
4343 subpattern does in fact match no characters, the loop is forcibly bro-
4344 ken.
4346 By default, the quantifiers are "greedy", that is, they match as much
4347 as possible (up to the maximum number of permitted times), without
4348 causing the rest of the pattern to fail. The classic example of where
4349 this gives problems is in trying to match comments in C programs. These
4350 appear between /* and */ and within the comment, individual * and /
4351 characters may appear. An attempt to match C comments by applying the
4352 pattern
4354 /\*.*\*/
4356 to the string
4358 /* first comment */ not comment /* second comment */
4360 fails, because it matches the entire string owing to the greediness of
4361 the .* item.
4363 However, if a quantifier is followed by a question mark, it ceases to
4364 be greedy, and instead matches the minimum number of times possible, so
4365 the pattern
4367 /\*.*?\*/
4369 does the right thing with the C comments. The meaning of the various
4370 quantifiers is not otherwise changed, just the preferred number of
4371 matches. Do not confuse this use of question mark with its use as a
4372 quantifier in its own right. Because it has two uses, it can sometimes
4373 appear doubled, as in
4375 \d??\d
4377 which matches one digit by preference, but can match two if that is the
4378 only way the rest of the pattern matches.
4380 If the PCRE_UNGREEDY option is set (an option that is not available in
4381 Perl), the quantifiers are not greedy by default, but individual ones
4382 can be made greedy by following them with a question mark. In other
4383 words, it inverts the default behaviour.
4385 When a parenthesized subpattern is quantified with a minimum repeat
4386 count that is greater than 1 or with a limited maximum, more memory is
4387 required for the compiled pattern, in proportion to the size of the
4388 minimum or maximum.
4390 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4391 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4392 the pattern is implicitly anchored, because whatever follows will be
4393 tried against every character position in the subject string, so there
4394 is no point in retrying the overall match at any position after the
4395 first. PCRE normally treats such a pattern as though it were preceded
4396 by \A.
4398 In cases where it is known that the subject string contains no new-
4399 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4400 mization, or alternatively using ^ to indicate anchoring explicitly.
4402 However, there is one situation where the optimization cannot be used.
4403 When .* is inside capturing parentheses that are the subject of a back
4404 reference elsewhere in the pattern, a match at the start may fail where
4405 a later one succeeds. Consider, for example:
4407 (.*)abc\1
4409 If the subject is "xyz123abc123" the match point is the fourth charac-
4410 ter. For this reason, such a pattern is not implicitly anchored.
4412 When a capturing subpattern is repeated, the value captured is the sub-
4413 string that matched the final iteration. For example, after
4415 (tweedle[dume]{3}\s*)+
4417 has matched "tweedledum tweedledee" the value of the captured substring
4418 is "tweedledee". However, if there are nested capturing subpatterns,
4419 the corresponding captured values may have been set in previous itera-
4420 tions. For example, after
4422 /(a|(b))+/
4424 matches "aba" the value of the second captured substring is "b".
4429 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4430 repetition, failure of what follows normally causes the repeated item
4431 to be re-evaluated to see if a different number of repeats allows the
4432 rest of the pattern to match. Sometimes it is useful to prevent this,
4433 either to change the nature of the match, or to cause it fail earlier
4434 than it otherwise might, when the author of the pattern knows there is
4435 no point in carrying on.
4437 Consider, for example, the pattern \d+foo when applied to the subject
4438 line
4440 123456bar
4442 After matching all 6 digits and then failing to match "foo", the normal
4443 action of the matcher is to try again with only 5 digits matching the
4444 \d+ item, and then with 4, and so on, before ultimately failing.
4445 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4446 the means for specifying that once a subpattern has matched, it is not
4447 to be re-evaluated in this way.
4449 If we use atomic grouping for the previous example, the matcher gives
4450 up immediately on failing to match "foo" the first time. The notation
4451 is a kind of special parenthesis, starting with (?> as in this example:
4453 (?>\d+)foo
4455 This kind of parenthesis "locks up" the part of the pattern it con-
4456 tains once it has matched, and a failure further into the pattern is
4457 prevented from backtracking into it. Backtracking past it to previous
4458 items, however, works as normal.
4460 An alternative description is that a subpattern of this type matches
4461 the string of characters that an identical standalone pattern would
4462 match, if anchored at the current point in the subject string.
4464 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4465 such as the above example can be thought of as a maximizing repeat that
4466 must swallow everything it can. So, while both \d+ and \d+? are pre-
4467 pared to adjust the number of digits they match in order to make the
4468 rest of the pattern match, (?>\d+) can only match an entire sequence of
4469 digits.
4471 Atomic groups in general can of course contain arbitrarily complicated
4472 subpatterns, and can be nested. However, when the subpattern for an
4473 atomic group is just a single repeated item, as in the example above, a
4474 simpler notation, called a "possessive quantifier" can be used. This
4475 consists of an additional + character following a quantifier. Using
4476 this notation, the previous example can be rewritten as
4478 \d++foo
4480 Note that a possessive quantifier can be used with an entire group, for
4481 example:
4483 (abc|xyz){2,3}+
4485 Possessive quantifiers are always greedy; the setting of the
4486 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4487 simpler forms of atomic group. However, there is no difference in the
4488 meaning of a possessive quantifier and the equivalent atomic group,
4489 though there may be a performance difference; possessive quantifiers
4490 should be slightly faster.
4492 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4493 tax. Jeffrey Friedl originated the idea (and the name) in the first
4494 edition of his book. Mike McCloskey liked it, so implemented it when he
4495 built Sun's Java package, and PCRE copied it from there. It ultimately
4496 found its way into Perl at release 5.10.
4498 PCRE has an optimization that automatically "possessifies" certain sim-
4499 ple pattern constructs. For example, the sequence A+B is treated as
4500 A++B because there is no point in backtracking into a sequence of A's
4501 when B must follow.
4503 When a pattern contains an unlimited repeat inside a subpattern that
4504 can itself be repeated an unlimited number of times, the use of an
4505 atomic group is the only way to avoid some failing matches taking a
4506 very long time indeed. The pattern
4508 (\D+|<\d+>)*[!?]
4510 matches an unlimited number of substrings that either consist of non-
4511 digits, or digits enclosed in <>, followed by either ! or ?. When it
4512 matches, it runs quickly. However, if it is applied to
4514 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4516 it takes a long time before reporting failure. This is because the
4517 string can be divided between the internal \D+ repeat and the external
4518 * repeat in a large number of ways, and all have to be tried. (The
4519 example uses [!?] rather than a single character at the end, because
4520 both PCRE and Perl have an optimization that allows for fast failure
4521 when a single character is used. They remember the last single charac-
4522 ter that is required for a match, and fail early if it is not present
4523 in the string.) If the pattern is changed so that it uses an atomic
4524 group, like this:
4526 ((?>\D+)|<\d+>)*[!?]
4528 sequences of non-digits cannot be broken, and failure happens quickly.
4533 Outside a character class, a backslash followed by a digit greater than
4534 0 (and possibly further digits) is a back reference to a capturing sub-
4535 pattern earlier (that is, to its left) in the pattern, provided there
4536 have been that many previous capturing left parentheses.
4538 However, if the decimal number following the backslash is less than 10,
4539 it is always taken as a back reference, and causes an error only if
4540 there are not that many capturing left parentheses in the entire pat-
4541 tern. In other words, the parentheses that are referenced need not be
4542 to the left of the reference for numbers less than 10. A "forward back
4543 reference" of this type can make sense when a repetition is involved
4544 and the subpattern to the right has participated in an earlier itera-
4545 tion.
4547 It is not possible to have a numerical "forward back reference" to a
4548 subpattern whose number is 10 or more using this syntax because a
4549 sequence such as \50 is interpreted as a character defined in octal.
4550 See the subsection entitled "Non-printing characters" above for further
4551 details of the handling of digits following a backslash. There is no
4552 such problem when named parentheses are used. A back reference to any
4553 subpattern is possible using named parentheses (see below).
4555 Another way of avoiding the ambiguity inherent in the use of digits
4556 following a backslash is to use the \g escape sequence, which is a fea-
4557 ture introduced in Perl 5.10. This escape must be followed by an
4558 unsigned number or a negative number, optionally enclosed in braces.
4559 These examples are all identical:
4561 (ring), \1
4562 (ring), \g1
4563 (ring), \g{1}
4565 An unsigned number specifies an absolute reference without the ambigu-
4566 ity that is present in the older syntax. It is also useful when literal
4567 digits follow the reference. A negative number is a relative reference.
4568 Consider this example:
4570 (abc(def)ghi)\g{-1}
4572 The sequence \g{-1} is a reference to the most recently started captur-
4573 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4574 \g{-2} would be equivalent to \1. The use of relative references can be
4575 helpful in long patterns, and also in patterns that are created by
4576 joining together fragments that contain references within themselves.
4578 A back reference matches whatever actually matched the capturing sub-
4579 pattern in the current subject string, rather than anything matching
4580 the subpattern itself (see "Subpatterns as subroutines" below for a way
4581 of doing that). So the pattern
4583 (sens|respons)e and \1ibility
4585 matches "sense and sensibility" and "response and responsibility", but
4586 not "sense and responsibility". If caseful matching is in force at the
4587 time of the back reference, the case of letters is relevant. For exam-
4588 ple,
4590 ((?i)rah)\s+\1
4592 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4593 original capturing subpattern is matched caselessly.
4595 There are several different ways of writing back references to named
4596 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4597 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4598 unified back reference syntax, in which \g can be used for both numeric
4599 and named references, is also supported. We could rewrite the above
4600 example in any of the following ways:
4602 (?<p1>(?i)rah)\s+\k<p1>
4603 (?'p1'(?i)rah)\s+\k{p1}
4604 (?P<p1>(?i)rah)\s+(?P=p1)
4605 (?<p1>(?i)rah)\s+\g{p1}
4607 A subpattern that is referenced by name may appear in the pattern
4608 before or after the reference.
4610 There may be more than one back reference to the same subpattern. If a
4611 subpattern has not actually been used in a particular match, any back
4612 references to it always fail by default. For example, the pattern
4614 (a|(bc))\2
4616 always fails if it starts to match "a" rather than "bc". However, if
4617 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
4618 ence to an unset value matches an empty string.
4620 Because there may be many capturing parentheses in a pattern, all dig-
4621 its following a backslash are taken as part of a potential back refer-
4622 ence number. If the pattern continues with a digit character, some
4623 delimiter must be used to terminate the back reference. If the
4624 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
4625 syntax or an empty comment (see "Comments" below) can be used.
4627 Recursive back references
4629 A back reference that occurs inside the parentheses to which it refers
4630 fails when the subpattern is first used, so, for example, (a\1) never
4631 matches. However, such references can be useful inside repeated sub-
4632 patterns. For example, the pattern
4634 (a|b\1)+
4636 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4637 ation of the subpattern, the back reference matches the character
4638 string corresponding to the previous iteration. In order for this to
4639 work, the pattern must be such that the first iteration does not need
4640 to match the back reference. This can be done using alternation, as in
4641 the example above, or by a quantifier with a minimum of zero.
4643 Back references of this type cause the group that they reference to be
4644 treated as an atomic group. Once the whole group has been matched, a
4645 subsequent matching failure cannot cause backtracking into the middle
4646 of the group.
4651 An assertion is a test on the characters following or preceding the
4652 current matching point that does not actually consume any characters.
4653 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4654 described above.
4656 More complicated assertions are coded as subpatterns. There are two
4657 kinds: those that look ahead of the current position in the subject
4658 string, and those that look behind it. An assertion subpattern is
4659 matched in the normal way, except that it does not cause the current
4660 matching position to be changed.
4662 Assertion subpatterns are not capturing subpatterns, and may not be
4663 repeated, because it makes no sense to assert the same thing several
4664 times. If any kind of assertion contains capturing subpatterns within
4665 it, these are counted for the purposes of numbering the capturing sub-
4666 patterns in the whole pattern. However, substring capturing is carried
4667 out only for positive assertions, because it does not make sense for
4668 negative assertions.
4670 Lookahead assertions
4672 Lookahead assertions start with (?= for positive assertions and (?! for
4673 negative assertions. For example,
4675 \w+(?=;)
4677 matches a word followed by a semicolon, but does not include the semi-
4678 colon in the match, and
4680 foo(?!bar)
4682 matches any occurrence of "foo" that is not followed by "bar". Note
4683 that the apparently similar pattern
4685 (?!foo)bar
4687 does not find an occurrence of "bar" that is preceded by something
4688 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4689 the assertion (?!foo) is always true when the next three characters are
4690 "bar". A lookbehind assertion is needed to achieve the other effect.
4692 If you want to force a matching failure at some point in a pattern, the
4693 most convenient way to do it is with (?!) because an empty string
4694 always matches, so an assertion that requires there not to be an empty
4695 string must always fail. The Perl 5.10 backtracking control verb
4696 (*FAIL) or (*F) is essentially a synonym for (?!).
4698 Lookbehind assertions
4700 Lookbehind assertions start with (?<= for positive assertions and (?<!
4701 for negative assertions. For example,
4703 (?<!foo)bar
4705 does find an occurrence of "bar" that is not preceded by "foo". The
4706 contents of a lookbehind assertion are restricted such that all the
4707 strings it matches must have a fixed length. However, if there are sev-
4708 eral top-level alternatives, they do not all have to have the same
4709 fixed length. Thus
4711 (?<=bullock|donkey)
4713 is permitted, but
4715 (?<!dogs?|cats?)
4717 causes an error at compile time. Branches that match different length
4718 strings are permitted only at the top level of a lookbehind assertion.
4719 This is an extension compared with Perl (5.8 and 5.10), which requires
4720 all branches to match the same length of string. An assertion such as
4722 (?<=ab(c|de))
4724 is not permitted, because its single top-level branch can match two
4725 different lengths, but it is acceptable to PCRE if rewritten to use two
4726 top-level branches:
4728 (?<=abc|abde)
4730 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4731 instead of a lookbehind assertion to get round the fixed-length
4732 restriction.
4734 The implementation of lookbehind assertions is, for each alternative,
4735 to temporarily move the current position back by the fixed length and
4736 then try to match. If there are insufficient characters before the cur-
4737 rent position, the assertion fails.
4739 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4740 mode) to appear in lookbehind assertions, because it makes it impossi-
4741 ble to calculate the length of the lookbehind. The \X and \R escapes,
4742 which can match different numbers of bytes, are also not permitted.
4744 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
4745 lookbehinds, as long as the subpattern matches a fixed-length string.
4746 Recursion, however, is not supported.
4748 Possessive quantifiers can be used in conjunction with lookbehind
4749 assertions to specify efficient matching of fixed-length strings at the
4750 end of subject strings. Consider a simple pattern such as
4752 abcd$
4754 when applied to a long string that does not match. Because matching
4755 proceeds from left to right, PCRE will look for each "a" in the subject
4756 and then see if what follows matches the rest of the pattern. If the
4757 pattern is specified as
4759 ^.*abcd$
4761 the initial .* matches the entire string at first, but when this fails
4762 (because there is no following "a"), it backtracks to match all but the
4763 last character, then all but the last two characters, and so on. Once
4764 again the search for "a" covers the entire string, from right to left,
4765 so we are no better off. However, if the pattern is written as
4767 ^.*+(?<=abcd)
4769 there can be no backtracking for the .*+ item; it can match only the
4770 entire string. The subsequent lookbehind assertion does a single test
4771 on the last four characters. If it fails, the match fails immediately.
4772 For long strings, this approach makes a significant difference to the
4773 processing time.
4775 Using multiple assertions
4777 Several assertions (of any sort) may occur in succession. For example,
4779 (?<=\d{3})(?<!999)foo
4781 matches "foo" preceded by three digits that are not "999". Notice that
4782 each of the assertions is applied independently at the same point in
4783 the subject string. First there is a check that the previous three
4784 characters are all digits, and then there is a check that the same
4785 three characters are not "999". This pattern does not match "foo" pre-
4786 ceded by six characters, the first of which are digits and the last
4787 three of which are not "999". For example, it doesn't match "123abc-
4788 foo". A pattern to do that is
4790 (?<=\d{3}...)(?<!999)foo
4792 This time the first assertion looks at the preceding six characters,
4793 checking that the first three are digits, and then the second assertion
4794 checks that the preceding three characters are not "999".
4796 Assertions can be nested in any combination. For example,
4798 (?<=(?<!foo)bar)baz
4800 matches an occurrence of "baz" that is preceded by "bar" which in turn
4801 is not preceded by "foo", while
4803 (?<=\d{3}(?!999)...)foo
4805 is another pattern that matches "foo" preceded by three digits and any
4806 three characters that are not "999".
4811 It is possible to cause the matching process to obey a subpattern con-
4812 ditionally or to choose between two alternative subpatterns, depending
4813 on the result of an assertion, or whether a specific capturing subpat-
4814 tern has already been matched. The two possible forms of conditional
4815 subpattern are:
4817 (?(condition)yes-pattern)
4818 (?(condition)yes-pattern|no-pattern)
4820 If the condition is satisfied, the yes-pattern is used; otherwise the
4821 no-pattern (if present) is used. If there are more than two alterna-
4822 tives in the subpattern, a compile-time error occurs.
4824 There are four kinds of condition: references to subpatterns, refer-
4825 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4827 Checking for a used subpattern by number
4829 If the text between the parentheses consists of a sequence of digits,
4830 the condition is true if a capturing subpattern of that number has pre-
4831 viously matched. If there is more than one capturing subpattern with
4832 the same number (see the earlier section about duplicate subpattern
4833 numbers), the condition is true if any of them have been set. An alter-
4834 native notation is to precede the digits with a plus or minus sign. In
4835 this case, the subpattern number is relative rather than absolute. The
4836 most recently opened parentheses can be referenced by (?(-1), the next
4837 most recent by (?(-2), and so on. In looping constructs it can also
4838 make sense to refer to subsequent groups with constructs such as
4839 (?(+2).
4841 Consider the following pattern, which contains non-significant white
4842 space to make it more readable (assume the PCRE_EXTENDED option) and to
4843 divide it into three parts for ease of discussion:
4845 ( \( )? [^()]+ (?(1) \) )
4847 The first part matches an optional opening parenthesis, and if that
4848 character is present, sets it as the first captured substring. The sec-
4849 ond part matches one or more characters that are not parentheses. The
4850 third part is a conditional subpattern that tests whether the first set
4851 of parentheses matched or not. If they did, that is, if subject started
4852 with an opening parenthesis, the condition is true, and so the yes-pat-
4853 tern is executed and a closing parenthesis is required. Otherwise,
4854 since no-pattern is not present, the subpattern matches nothing. In
4855 other words, this pattern matches a sequence of non-parentheses,
4856 optionally enclosed in parentheses.
4858 If you were embedding this pattern in a larger one, you could use a
4859 relative reference:
4861 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4863 This makes the fragment independent of the parentheses in the larger
4864 pattern.
4866 Checking for a used subpattern by name
4868 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4869 used subpattern by name. For compatibility with earlier versions of
4870 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4871 also recognized. However, there is a possible ambiguity with this syn-
4872 tax, because subpattern names may consist entirely of digits. PCRE
4873 looks first for a named subpattern; if it cannot find one and the name
4874 consists entirely of digits, PCRE looks for a subpattern of that num-
4875 ber, which must be greater than zero. Using subpattern names that con-
4876 sist entirely of digits is not recommended.
4878 Rewriting the above example to use a named subpattern gives this:
4880 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4882 If the name used in a condition of this kind is a duplicate, the test
4883 is applied to all subpatterns of the same name, and is true if any one
4884 of them has matched.
4886 Checking for pattern recursion
4888 If the condition is the string (R), and there is no subpattern with the
4889 name R, the condition is true if a recursive call to the whole pattern
4890 or any subpattern has been made. If digits or a name preceded by amper-
4891 sand follow the letter R, for example:
4893 (?(R3)...) or (?(R&name)...)
4895 the condition is true if the most recent recursion is into a subpattern
4896 whose number or name is given. This condition does not check the entire
4897 recursion stack. If the name used in a condition of this kind is a
4898 duplicate, the test is applied to all subpatterns of the same name, and
4899 is true if any one of them is the most recent recursion.
4901 At "top level", all these recursion test conditions are false. The
4902 syntax for recursive patterns is described below.
4904 Defining subpatterns for use by reference only
4906 If the condition is the string (DEFINE), and there is no subpattern
4907 with the name DEFINE, the condition is always false. In this case,
4908 there may be only one alternative in the subpattern. It is always
4909 skipped if control reaches this point in the pattern; the idea of
4910 DEFINE is that it can be used to define "subroutines" that can be ref-
4911 erenced from elsewhere. (The use of "subroutines" is described below.)
4912 For example, a pattern to match an IPv4 address could be written like
4913 this (ignore whitespace and line breaks):
4915 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4916 \b (?&byte) (\.(?&byte)){3} \b
4918 The first part of the pattern is a DEFINE group inside which a another
4919 group named "byte" is defined. This matches an individual component of
4920 an IPv4 address (a number less than 256). When matching takes place,
4921 this part of the pattern is skipped because DEFINE acts like a false
4922 condition. The rest of the pattern uses references to the named group
4923 to match the four dot-separated components of an IPv4 address, insist-
4924 ing on a word boundary at each end.
4926 Assertion conditions
4928 If the condition is not in any of the above formats, it must be an
4929 assertion. This may be a positive or negative lookahead or lookbehind
4930 assertion. Consider this pattern, again containing non-significant
4931 white space, and with the two alternatives on the second line:
4933 (?(?=[^a-z]*[a-z])
4934 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4936 The condition is a positive lookahead assertion that matches an
4937 optional sequence of non-letters followed by a letter. In other words,
4938 it tests for the presence of at least one letter in the subject. If a
4939 letter is found, the subject is matched against the first alternative;
4940 otherwise it is matched against the second. This pattern matches
4941 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4942 letters and dd are digits.
4947 The sequence (?# marks the start of a comment that continues up to the
4948 next closing parenthesis. Nested parentheses are not permitted. The
4949 characters that make up a comment play no part in the pattern matching
4950 at all.
4952 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4953 character class introduces a comment that continues to immediately
4954 after the next newline in the pattern.
4959 Consider the problem of matching a string in parentheses, allowing for
4960 unlimited nested parentheses. Without the use of recursion, the best
4961 that can be done is to use a pattern that matches up to some fixed
4962 depth of nesting. It is not possible to handle an arbitrary nesting
4963 depth.
4965 For some time, Perl has provided a facility that allows regular expres-
4966 sions to recurse (amongst other things). It does this by interpolating
4967 Perl code in the expression at run time, and the code can refer to the
4968 expression itself. A Perl pattern using code interpolation to solve the
4969 parentheses problem can be created like this:
4971 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4973 The (?p{...}) item interpolates Perl code at run time, and in this case
4974 refers recursively to the pattern in which it appears.
4976 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4977 it supports special syntax for recursion of the entire pattern, and
4978 also for individual subpattern recursion. After its introduction in
4979 PCRE and Python, this kind of recursion was subsequently introduced
4980 into Perl at release 5.10.
4982 A special item that consists of (? followed by a number greater than
4983 zero and a closing parenthesis is a recursive call of the subpattern of
4984 the given number, provided that it occurs inside that subpattern. (If
4985 not, it is a "subroutine" call, which is described in the next sec-
4986 tion.) The special item (?R) or (?0) is a recursive call of the entire
4987 regular expression.
4989 This PCRE pattern solves the nested parentheses problem (assume the
4990 PCRE_EXTENDED option is set so that white space is ignored):
4992 \( ( [^()]++ | (?R) )* \)
4994 First it matches an opening parenthesis. Then it matches any number of
4995 substrings which can either be a sequence of non-parentheses, or a
4996 recursive match of the pattern itself (that is, a correctly parenthe-
4997 sized substring). Finally there is a closing parenthesis. Note the use
4998 of a possessive quantifier to avoid backtracking into sequences of non-
4999 parentheses.
5001 If this were part of a larger pattern, you would not want to recurse
5002 the entire pattern, so instead you could use this:
5004 ( \( ( [^()]++ | (?1) )* \) )
5006 We have put the pattern into parentheses, and caused the recursion to
5007 refer to them instead of the whole pattern.
5009 In a larger pattern, keeping track of parenthesis numbers can be
5010 tricky. This is made easier by the use of relative references (a Perl
5011 5.10 feature). Instead of (?1) in the pattern above you can write
5012 (?-2) to refer to the second most recently opened parentheses preceding
5013 the recursion. In other words, a negative number counts capturing
5014 parentheses leftwards from the point at which it is encountered.
5016 It is also possible to refer to subsequently opened parentheses, by
5017 writing references such as (?+2). However, these cannot be recursive
5018 because the reference is not inside the parentheses that are refer-
5019 enced. They are always "subroutine" calls, as described in the next
5020 section.
5022 An alternative approach is to use named parentheses instead. The Perl
5023 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5024 supported. We could rewrite the above example as follows:
5026 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5028 If there is more than one subpattern with the same name, the earliest
5029 one is used.
5031 This particular example pattern that we have been looking at contains
5032 nested unlimited repeats, and so the use of a possessive quantifier for
5033 matching strings of non-parentheses is important when applying the pat-
5034 tern to strings that do not match. For example, when this pattern is
5035 applied to
5037 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5039 it yields "no match" quickly. However, if a possessive quantifier is
5040 not used, the match runs for a very long time indeed because there are
5041 so many different ways the + and * repeats can carve up the subject,
5042 and all have to be tested before failure can be reported.
5044 At the end of a match, the values of capturing parentheses are those
5045 from the outermost level. If you want to obtain intermediate values, a
5046 callout function can be used (see below and the pcrecallout documenta-
5047 tion). If the pattern above is matched against
5049 (ab(cd)ef)
5051 the value for the inner capturing parentheses (numbered 2) is "ef",
5052 which is the last value taken on at the top level. If a capturing sub-
5053 pattern is not matched at the top level, its final value is unset, even
5054 if it is (temporarily) set at a deeper level.
5056 If there are more than 15 capturing parentheses in a pattern, PCRE has
5057 to obtain extra memory to store data during a recursion, which it does
5058 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5059 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5061 Do not confuse the (?R) item with the condition (R), which tests for
5062 recursion. Consider this pattern, which matches text in angle brack-
5063 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5064 brackets (that is, when recursing), whereas any characters are permit-
5065 ted at the outer level.
5067 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5069 In this pattern, (?(R) is the start of a conditional subpattern, with
5070 two different alternatives for the recursive and non-recursive cases.
5071 The (?R) item is the actual recursive call.
5073 Recursion difference from Perl
5075 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5076 always treated as an atomic group. That is, once it has matched some of
5077 the subject string, it is never re-entered, even if it contains untried
5078 alternatives and there is a subsequent matching failure. This can be
5079 illustrated by the following pattern, which purports to match a palin-
5080 dromic string that contains an odd number of characters (for example,
5081 "a", "aba", "abcba", "abcdcba"):
5083 ^(.|(.)(?1)\2)$
5085 The idea is that it either matches a single character, or two identical
5086 characters surrounding a sub-palindrome. In Perl, this pattern works;
5087 in PCRE it does not if the pattern is longer than three characters.
5088 Consider the subject string "abcba":
5090 At the top level, the first character is matched, but as it is not at
5091 the end of the string, the first alternative fails; the second alterna-
5092 tive is taken and the recursion kicks in. The recursive call to subpat-
5093 tern 1 successfully matches the next character ("b"). (Note that the
5094 beginning and end of line tests are not part of the recursion).
5096 Back at the top level, the next character ("c") is compared with what
5097 subpattern 2 matched, which was "a". This fails. Because the recursion
5098 is treated as an atomic group, there are now no backtracking points,
5099 and so the entire match fails. (Perl is able, at this point, to re-
5100 enter the recursion and try the second alternative.) However, if the
5101 pattern is written with the alternatives in the other order, things are
5102 different:
5104 ^((.)(?1)\2|.)$
5106 This time, the recursing alternative is tried first, and continues to
5107 recurse until it runs out of characters, at which point the recursion
5108 fails. But this time we do have another alternative to try at the
5109 higher level. That is the big difference: in the previous case the
5110 remaining alternative is at a deeper recursion level, which PCRE cannot
5111 use.
5113 To change the pattern so that matches all palindromic strings, not just
5114 those with an odd number of characters, it is tempting to change the
5115 pattern to this:
5117 ^((.)(?1)\2|.?)$
5119 Again, this works in Perl, but not in PCRE, and for the same reason.
5120 When a deeper recursion has matched a single character, it cannot be
5121 entered again in order to match an empty string. The solution is to
5122 separate the two cases, and write out the odd and even cases as alter-
5123 natives at the higher level:
5125 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
5127 If you want to match typical palindromic phrases, the pattern has to
5128 ignore all non-word characters, which can be done like this:
5130 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
5132 If run with the PCRE_CASELESS option, this pattern matches phrases such
5133 as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
5134 Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
5135 ing into sequences of non-word characters. Without this, PCRE takes a
5136 great deal longer (ten times or more) to match typical phrases, and
5137 Perl takes so long that you think it has gone into a loop.
5139 WARNING: The palindrome-matching patterns above work only if the sub-
5140 ject string does not start with a palindrome that is shorter than the
5141 entire string. For example, although "abcba" is correctly matched, if
5142 the subject is "ababa", PCRE finds the palindrome "aba" at the start,
5143 then fails at top level because the end of the string does not follow.
5144 Once again, it cannot jump back into the recursion to try other alter-
5145 natives, so the entire match fails.
5150 If the syntax for a recursive subpattern reference (either by number or
5151 by name) is used outside the parentheses to which it refers, it oper-
5152 ates like a subroutine in a programming language. The "called" subpat-
5153 tern may be defined before or after the reference. A numbered reference
5154 can be absolute or relative, as in these examples:
5156 (...(absolute)...)...(?2)...
5157 (...(relative)...)...(?-1)...
5158 (...(?+1)...(relative)...
5160 An earlier example pointed out that the pattern
5162 (sens|respons)e and \1ibility
5164 matches "sense and sensibility" and "response and responsibility", but
5165 not "sense and responsibility". If instead the pattern
5167 (sens|respons)e and (?1)ibility
5169 is used, it does match "sense and responsibility" as well as the other
5170 two strings. Another example is given in the discussion of DEFINE
5171 above.
5173 Like recursive subpatterns, a subroutine call is always treated as an
5174 atomic group. That is, once it has matched some of the subject string,
5175 it is never re-entered, even if it contains untried alternatives and
5176 there is a subsequent matching failure. Any capturing parentheses that
5177 are set during the subroutine call revert to their previous values
5178 afterwards.
5180 When a subpattern is used as a subroutine, processing options such as
5181 case-independence are fixed when the subpattern is defined. They cannot
5182 be changed for different calls. For example, consider this pattern:
5184 (abc)(?i:(?-1))
5186 It matches "abcabc". It does not match "abcABC" because the change of
5187 processing option does not affect the called subpattern.
5192 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
5193 name or a number enclosed either in angle brackets or single quotes, is
5194 an alternative syntax for referencing a subpattern as a subroutine,
5195 possibly recursively. Here are two of the examples used above, rewrit-
5196 ten using this syntax:
5198 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
5199 (sens|respons)e and \g'1'ibility
5201 PCRE supports an extension to Oniguruma: if a number is preceded by a
5202 plus or a minus sign it is taken as a relative reference. For example:
5204 (abc)(?i:\g<-1>)
5206 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
5207 synonymous. The former is a back reference; the latter is a subroutine
5208 call.
5213 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
5214 Perl code to be obeyed in the middle of matching a regular expression.
5215 This makes it possible, amongst other things, to extract different sub-
5216 strings that match the same pair of parentheses when there is a repeti-
5217 tion.
5219 PCRE provides a similar feature, but of course it cannot obey arbitrary
5220 Perl code. The feature is called "callout". The caller of PCRE provides
5221 an external function by putting its entry point in the global variable
5222 pcre_callout. By default, this variable contains NULL, which disables
5223 all calling out.
5225 Within a regular expression, (?C) indicates the points at which the
5226 external function is to be called. If you want to identify different
5227 callout points, you can put a number less than 256 after the letter C.
5228 The default value is zero. For example, this pattern has two callout
5229 points:
5231 (?C1)abc(?C2)def
5233 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
5234 automatically installed before each item in the pattern. They are all
5235 numbered 255.
5237 During matching, when PCRE reaches a callout point (and pcre_callout is
5238 set), the external function is called. It is provided with the number
5239 of the callout, the position in the pattern, and, optionally, one item
5240 of data originally supplied by the caller of pcre_exec(). The callout
5241 function may cause matching to proceed, to backtrack, or to fail alto-
5242 gether. A complete description of the interface to the callout function
5243 is given in the pcrecallout documentation.
5248 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
5249 which are described in the Perl documentation as "experimental and sub-
5250 ject to change or removal in a future version of Perl". It goes on to
5251 say: "Their usage in production code should be noted to avoid problems
5252 during upgrades." The same remarks apply to the PCRE features described
5253 in this section.
5255 Since these verbs are specifically related to backtracking, most of
5256 them can be used only when the pattern is to be matched using
5257 pcre_exec(), which uses a backtracking algorithm. With the exception of
5258 (*FAIL), which behaves like a failing negative assertion, they cause an
5259 error if encountered by pcre_dfa_exec().
5261 If any of these verbs are used in an assertion or subroutine subpattern
5262 (including recursive subpatterns), their effect is confined to that
5263 subpattern; it does not extend to the surrounding pattern. Note that
5264 such subpatterns are processed as anchored at the point where they are
5265 tested.
5267 The new verbs make use of what was previously invalid syntax: an open-
5268 ing parenthesis followed by an asterisk. They are generally of the form
5269 (*VERB) or (*VERB:NAME). Some may take either form, with differing be-
5270 haviour, depending on whether or not an argument is present. An name is
5271 a sequence of letters, digits, and underscores. If the name is empty,
5272 that is, if the closing parenthesis immediately follows the colon, the
5273 effect is as if the colon were not there. Any number of these verbs may
5274 occur in a pattern.
5276 PCRE contains some optimizations that are used to speed up matching by
5277 running some checks at the start of each match attempt. For example, it
5278 may know the minimum length of matching subject, or that a particular
5279 character must be present. When one of these optimizations suppresses
5280 the running of a match, any included backtracking verbs will not, of
5281 course, be processed. You can suppress the start-of-match optimizations
5282 by setting the PCRE_NO_START_OPTIMIZE option when calling pcre_exec().
5284 Verbs that act immediately
5286 The following verbs act as soon as they are encountered. They may not
5287 be followed by a name.
5289 (*ACCEPT)
5291 This verb causes the match to end successfully, skipping the remainder
5292 of the pattern. When inside a recursion, only the innermost pattern is
5293 ended immediately. If (*ACCEPT) is inside capturing parentheses, the
5294 data so far is captured. (This feature was added to PCRE at release
5295 8.00.) For example:
5297 A((?:A|B(*ACCEPT)|C)D)
5299 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
5300 tured by the outer parentheses.
5302 (*FAIL) or (*F)
5304 This verb causes the match to fail, forcing backtracking to occur. It
5305 is equivalent to (?!) but easier to read. The Perl documentation notes
5306 that it is probably useful only when combined with (?{}) or (??{}).
5307 Those are, of course, Perl features that are not present in PCRE. The
5308 nearest equivalent is the callout feature, as for example in this pat-
5309 tern:
5311 a+(?C)(*FAIL)
5313 A match with the string "aaaa" always fails, but the callout is taken
5314 before each backtrack happens (in this example, 10 times).
5316 Recording which path was taken
5318 There is one verb whose main purpose is to track how a match was
5319 arrived at, though it also has a secondary use in conjunction with
5320 advancing the match starting point (see (*SKIP) below).
5322 (*MARK:NAME) or (*:NAME)
5324 A name is always required with this verb. There may be as many
5325 instances of (*MARK) as you like in a pattern, and their names do not
5326 have to be unique.
5328 When a match succeeds, the name of the last-encountered (*MARK) is
5329 passed back to the caller via the pcre_extra data structure, as
5330 described in the section on pcre_extra in the pcreapi documentation. No
5331 data is returned for a partial match. Here is an example of pcretest
5332 output, where the /K modifier requests the retrieval and outputting of
5333 (*MARK) data:
5335 /X(*MARK:A)Y|X(*MARK:B)Z/K
5336 XY
5337 0: XY
5338 MK: A
5339 XZ
5340 0: XZ
5341 MK: B
5343 The (*MARK) name is tagged with "MK:" in this output, and in this exam-
5344 ple it indicates which of the two alternatives matched. This is a more
5345 efficient way of obtaining this information than putting each alterna-
5346 tive in its own capturing parentheses.
5348 A name may also be returned after a failed match if the final path
5349 through the pattern involves (*MARK). However, unless (*MARK) used in
5350 conjunction with (*COMMIT), this is unlikely to happen for an unan-
5351 chored pattern because, as the starting point for matching is advanced,
5352 the final check is often with an empty string, causing a failure before
5353 (*MARK) is reached. For example:
5355 /X(*MARK:A)Y|X(*MARK:B)Z/K
5356 XP
5357 No match
5359 There are three potential starting points for this match (starting with
5360 X, starting with P, and with an empty string). If the pattern is
5361 anchored, the result is different:
5363 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
5364 XP
5365 No match, mark = B
5367 PCRE's start-of-match optimizations can also interfere with this. For
5368 example, if, as a result of a call to pcre_study(), it knows the mini-
5369 mum subject length for a match, a shorter subject will not be scanned
5370 at all.
5372 Note that similar anomalies (though different in detail) exist in Perl,
5373 no doubt for the same reasons. The use of (*MARK) data after a failed
5374 match of an unanchored pattern is not recommended, unless (*COMMIT) is
5375 involved.
5377 Verbs that act after backtracking
5379 The following verbs do nothing when they are encountered. Matching con-
5380 tinues with what follows, but if there is no subsequent match, causing
5381 a backtrack to the verb, a failure is forced. That is, backtracking
5382 cannot pass to the left of the verb. However, when one of these verbs
5383 appears inside an atomic group, its effect is confined to that group,
5384 because once the group has been matched, there is never any backtrack-
5385 ing into it. In this situation, backtracking can "jump back" to the
5386 left of the entire atomic group. (Remember also, as stated above, that
5387 this localization also applies in subroutine calls and assertions.)
5389 These verbs differ in exactly what kind of failure occurs when back-
5390 tracking reaches them.
5392 (*COMMIT)
5394 This verb, which may not be followed by a name, causes the whole match
5395 to fail outright if the rest of the pattern does not match. Even if the
5396 pattern is unanchored, no further attempts to find a match by advancing
5397 the starting point take place. Once (*COMMIT) has been passed,
5398 pcre_exec() is committed to finding a match at the current starting
5399 point, or not at all. For example:
5401 a+(*COMMIT)b
5403 This matches "xxaab" but not "aacaab". It can be thought of as a kind
5404 of dynamic anchor, or "I've started, so I must finish." The name of the
5405 most recently passed (*MARK) in the path is passed back when (*COMMIT)
5406 forces a match failure.
5408 Note that (*COMMIT) at the start of a pattern is not the same as an
5409 anchor, unless PCRE's start-of-match optimizations are turned off, as
5410 shown in this pcretest example:
5412 /(*COMMIT)abc/
5413 xyzabc
5414 0: abc
5415 xyzabc\Y
5416 No match
5418 PCRE knows that any match must start with "a", so the optimization
5419 skips along the subject to "a" before running the first match attempt,
5420 which succeeds. When the optimization is disabled by the \Y escape in
5421 the second subject, the match starts at "x" and so the (*COMMIT) causes
5422 it to fail without trying any other starting points.
5424 (*PRUNE) or (*PRUNE:NAME)
5426 This verb causes the match to fail at the current starting position in
5427 the subject if the rest of the pattern does not match. If the pattern
5428 is unanchored, the normal "bumpalong" advance to the next starting
5429 character then happens. Backtracking can occur as usual to the left of
5430 (*PRUNE), before it is reached, or when matching to the right of
5431 (*PRUNE), but if there is no match to the right, backtracking cannot
5432 cross (*PRUNE). In simple cases, the use of (*PRUNE) is just an alter-
5433 native to an atomic group or possessive quantifier, but there are some
5434 uses of (*PRUNE) that cannot be expressed in any other way. The behav-
5435 iour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
5436 match fails completely; the name is passed back if this is the final
5437 attempt. (*PRUNE:NAME) does not pass back a name if the match suc-
5438 ceeds. In an anchored pattern (*PRUNE) has the same effect as (*COM-
5439 MIT).
5441 (*SKIP)
5443 This verb, when given without a name, is like (*PRUNE), except that if
5444 the pattern is unanchored, the "bumpalong" advance is not to the next
5445 character, but to the position in the subject where (*SKIP) was encoun-
5446 tered. (*SKIP) signifies that whatever text was matched leading up to
5447 it cannot be part of a successful match. Consider:
5449 a+(*SKIP)b
5451 If the subject is "aaaac...", after the first match attempt fails
5452 (starting at the first character in the string), the starting point
5453 skips on to start the next attempt at "c". Note that a possessive quan-
5454 tifer does not have the same effect as this example; although it would
5455 suppress backtracking during the first match attempt, the second
5456 attempt would start at the second character instead of skipping on to
5457 "c".
5459 (*SKIP:NAME)
5461 When (*SKIP) has an associated name, its behaviour is modified. If the
5462 following pattern fails to match, the previous path through the pattern
5463 is searched for the most recent (*MARK) that has the same name. If one
5464 is found, the "bumpalong" advance is to the subject position that cor-
5465 responds to that (*MARK) instead of to where (*SKIP) was encountered.
5466 If no (*MARK) with a matching name is found, normal "bumpalong" of one
5467 character happens (the (*SKIP) is ignored).
5469 (*THEN) or (*THEN:NAME)
5471 This verb causes a skip to the next alternation if the rest of the pat-
5472 tern does not match. That is, it cancels pending backtracking, but only
5473 within the current alternation. Its name comes from the observation
5474 that it can be used for a pattern-based if-then-else block:
5476 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5478 If the COND1 pattern matches, FOO is tried (and possibly further items
5479 after the end of the group if FOO succeeds); on failure the matcher
5480 skips to the second alternative and tries COND2, without backtracking
5481 into COND1. The behaviour of (*THEN:NAME) is exactly the same as
5482 (*MARK:NAME)(*THEN) if the overall match fails. If (*THEN) is not
5483 directly inside an alternation, it acts like (*PRUNE).
5488 pcreapi(3), pcrecallout(3), pcrematching(3), pcresyntax(3), pcre(3).
5493 Philip Hazel
5494 University Computing Service
5495 Cambridge CB2 3QH, England.
5500 Last updated: 18 May 2010
5501 Copyright (c) 1997-2010 University of Cambridge.
5502 ------------------------------------------------------------------------------
5508 NAME
5509 PCRE - Perl-compatible regular expressions
5514 The full syntax and semantics of the regular expressions that are sup-
5515 ported by PCRE are described in the pcrepattern documentation. This
5516 document contains just a quick-reference summary of the syntax.
5521 \x where x is non-alphanumeric is a literal x
5522 \Q...\E treat enclosed characters as literal
5527 \a alarm, that is, the BEL character (hex 07)
5528 \cx "control-x", where x is any character
5529 \e escape (hex 1B)
5530 \f formfeed (hex 0C)
5531 \n newline (hex 0A)
5532 \r carriage return (hex 0D)
5533 \t tab (hex 09)
5534 \ddd character with octal code ddd, or backreference
5535 \xhh character with hex code hh
5536 \x{hhh..} character with hex code hhh..
5541 . any character except newline;
5542 in dotall mode, any character whatsoever
5543 \C one byte, even in UTF-8 mode (best avoided)
5544 \d a decimal digit
5545 \D a character that is not a decimal digit
5546 \h a horizontal whitespace character
5547 \H a character that is not a horizontal whitespace character
5548 \N a character that is not a newline
5549 \p{xx} a character with the xx property
5550 \P{xx} a character without the xx property
5551 \R a newline sequence
5552 \s a whitespace character
5553 \S a character that is not a whitespace character
5554 \v a vertical whitespace character
5555 \V a character that is not a vertical whitespace character
5556 \w a "word" character
5557 \W a "non-word" character
5558 \X an extended Unicode sequence
5560 In PCRE, by default, \d, \D, \s, \S, \w, and \W recognize only ASCII
5561 characters, even in UTF-8 mode. However, this can be changed by setting
5562 the PCRE_UCP option.
5567 C Other
5568 Cc Control
5569 Cf Format
5570 Cn Unassigned
5571 Co Private use
5572 Cs Surrogate
5574 L Letter
5575 Ll Lower case letter
5576 Lm Modifier letter
5577 Lo Other letter
5578 Lt Title case letter
5579 Lu Upper case letter
5580 L& Ll, Lu, or Lt
5582 M Mark
5583 Mc Spacing mark
5584 Me Enclosing mark
5585 Mn Non-spacing mark
5587 N Number
5588 Nd Decimal number
5589 Nl Letter number
5590 No Other number
5592 P Punctuation
5593 Pc Connector punctuation
5594 Pd Dash punctuation
5595 Pe Close punctuation
5596 Pf Final punctuation
5597 Pi Initial punctuation
5598 Po Other punctuation
5599 Ps Open punctuation
5601 S Symbol
5602 Sc Currency symbol
5603 Sk Modifier symbol
5604 Sm Mathematical symbol
5605 So Other symbol
5607 Z Separator
5608 Zl Line separator
5609 Zp Paragraph separator
5610 Zs Space separator
5615 Xan Alphanumeric: union of properties L and N
5616 Xps POSIX space: property Z or tab, NL, VT, FF, CR
5617 Xsp Perl space: property Z or tab, NL, FF, CR
5618 Xwd Perl word: property Xan or underscore
5623 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
5624 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
5625 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
5626 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
5627 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
5628 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
5629 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
5630 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
5631 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
5632 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
5633 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
5634 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
5635 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5636 Ugaritic, Vai, Yi.
5641 [...] positive character class
5642 [^...] negative character class
5643 [x-y] range (can be used for hex characters)
5644 [[:xxx:]] positive POSIX named set
5645 [[:^xxx:]] negative POSIX named set
5647 alnum alphanumeric
5648 alpha alphabetic
5649 ascii 0-127
5650 blank space or tab
5651 cntrl control character
5652 digit decimal digit
5653 graph printing, excluding space
5654 lower lower case letter
5655 print printing, including space
5656 punct printing, excluding alphanumeric
5657 space whitespace
5658 upper upper case letter
5659 word same as \w
5660 xdigit hexadecimal digit
5662 In PCRE, POSIX character set names recognize only ASCII characters by
5663 default, but some of them use Unicode properties if PCRE_UCP is set.
5664 You can use \Q...\E inside a character class.
5669 ? 0 or 1, greedy
5670 ?+ 0 or 1, possessive
5671 ?? 0 or 1, lazy
5672 * 0 or more, greedy
5673 *+ 0 or more, possessive
5674 *? 0 or more, lazy
5675 + 1 or more, greedy
5676 ++ 1 or more, possessive
5677 +? 1 or more, lazy
5678 {n} exactly n
5679 {n,m} at least n, no more than m, greedy
5680 {n,m}+ at least n, no more than m, possessive
5681 {n,m}? at least n, no more than m, lazy
5682 {n,} n or more, greedy
5683 {n,}+ n or more, possessive
5684 {n,}? n or more, lazy
5689 \b word boundary
5690 \B not a word boundary
5691 ^ start of subject
5692 also after internal newline in multiline mode
5693 \A start of subject
5694 $ end of subject
5695 also before newline at end of subject
5696 also before internal newline in multiline mode
5697 \Z end of subject
5698 also before newline at end of subject
5699 \z end of subject
5700 \G first matching position in subject
5705 \K reset start of match
5710 expr|expr|expr...
5715 (...) capturing group
5716 (?<name>...) named capturing group (Perl)
5717 (?'name'...) named capturing group (Perl)
5718 (?P<name>...) named capturing group (Python)
5719 (?:...) non-capturing group
5720 (?|...) non-capturing group; reset group numbers for
5721 capturing groups in each alternative
5726 (?>...) atomic, non-capturing group
5731 (?#....) comment (not nestable)
5736 (?i) caseless
5737 (?J) allow duplicate names
5738 (?m) multiline
5739 (?s) single line (dotall)
5740 (?U) default ungreedy (lazy)
5741 (?x) extended (ignore white space)
5742 (?-...) unset option(s)
5744 The following are recognized only at the start of a pattern or after
5745 one of the newline-setting options with similar syntax:
5747 (*UTF8) set UTF-8 mode (PCRE_UTF8)
5748 (*UCP) set PCRE_UCP (use Unicode properties for \d etc)
5753 (?=...) positive look ahead
5754 (?!...) negative look ahead
5755 (?<=...) positive look behind
5756 (?<!...) negative look behind
5758 Each top-level branch of a look behind must be of a fixed length.
5763 \n reference by number (can be ambiguous)
5764 \gn reference by number
5765 \g{n} reference by number
5766 \g{-n} relative reference by number
5767 \k<name> reference by name (Perl)
5768 \k'name' reference by name (Perl)
5769 \g{name} reference by name (Perl)
5770 \k{name} reference by name (.NET)
5771 (?P=name) reference by name (Python)
5776 (?R) recurse whole pattern
5777 (?n) call subpattern by absolute number
5778 (?+n) call subpattern by relative number
5779 (?-n) call subpattern by relative number
5780 (?&name) call subpattern by name (Perl)
5781 (?P>name) call subpattern by name (Python)
5782 \g<name> call subpattern by name (Oniguruma)
5783 \g'name' call subpattern by name (Oniguruma)
5784 \g<n> call subpattern by absolute number (Oniguruma)
5785 \g'n' call subpattern by absolute number (Oniguruma)
5786 \g<+n> call subpattern by relative number (PCRE extension)
5787 \g'+n' call subpattern by relative number (PCRE extension)
5788 \g<-n> call subpattern by relative number (PCRE extension)
5789 \g'-n' call subpattern by relative number (PCRE extension)
5794 (?(condition)yes-pattern)
5795 (?(condition)yes-pattern|no-pattern)
5797 (?(n)... absolute reference condition
5798 (?(+n)... relative reference condition
5799 (?(-n)... relative reference condition
5800 (?(<name>)... named reference condition (Perl)
5801 (?('name')... named reference condition (Perl)
5802 (?(name)... named reference condition (PCRE)
5803 (?(R)... overall recursion condition
5804 (?(Rn)... specific group recursion condition
5805 (?(R&name)... specific recursion condition
5806 (?(DEFINE)... define subpattern for reference
5807 (?(assert)... assertion condition
5812 The following act immediately they are reached:
5814 (*ACCEPT) force successful match
5815 (*FAIL) force backtrack; synonym (*F)
5817 The following act only when a subsequent match failure causes a back-
5818 track to reach them. They all force a match failure, but they differ in
5819 what happens afterwards. Those that advance the start-of-match point do
5820 so only if the pattern is not anchored.
5822 (*COMMIT) overall failure, no advance of starting point
5823 (*PRUNE) advance to next starting character
5824 (*SKIP) advance start to current matching position
5825 (*THEN) local failure, backtrack to next alternation
5830 These are recognized only at the very start of the pattern or after a
5831 (*BSR_...) or (*UTF8) or (*UCP) option.
5833 (*CR) carriage return only
5834 (*LF) linefeed only
5835 (*CRLF) carriage return followed by linefeed
5836 (*ANYCRLF) all three of the above
5837 (*ANY) any Unicode newline sequence
5842 These are recognized only at the very start of the pattern or after a
5843 (*...) option that sets the newline convention or UTF-8 or UCP mode.
5846 (*BSR_UNICODE) any Unicode newline sequence
5851 (?C) callout
5852 (?Cn) callout with data n
5857 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5862 Philip Hazel
5863 University Computing Service
5864 Cambridge CB2 3QH, England.
5869 Last updated: 12 May 2010
5870 Copyright (c) 1997-2010 University of Cambridge.
5871 ------------------------------------------------------------------------------
5877 NAME
5878 PCRE - Perl-compatible regular expressions
5883 In normal use of PCRE, if the subject string that is passed to
5884 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5885 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5886 There are circumstances where it might be helpful to distinguish this
5887 case from other cases in which there is no match.
5889 Consider, for example, an application where a human is required to type
5890 in data for a field with specific formatting requirements. An example
5891 might be a date in the form ddmmmyy, defined by this pattern:
5893 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5895 If the application sees the user's keystrokes one by one, and can check
5896 that what has been typed so far is potentially valid, it is able to
5897 raise an error as soon as a mistake is made, by beeping and not
5898 reflecting the character that has been typed, for example. This immedi-
5899 ate feedback is likely to be a better user interface than a check that
5900 is delayed until the entire string has been entered. Partial matching
5901 can also sometimes be useful when the subject string is very long and
5902 is not all available at once.
5904 PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
5905 PCRE_PARTIAL_HARD options, which can be set when calling pcre_exec() or
5906 pcre_dfa_exec(). For backwards compatibility, PCRE_PARTIAL is a synonym
5907 for PCRE_PARTIAL_SOFT. The essential difference between the two options
5908 is whether or not a partial match is preferred to an alternative com-
5909 plete match, though the details differ between the two matching func-
5910 tions. If both options are set, PCRE_PARTIAL_HARD takes precedence.
5912 Setting a partial matching option disables two of PCRE's optimizations.
5913 PCRE remembers the last literal byte in a pattern, and abandons match-
5914 ing immediately if such a byte is not present in the subject string.
5915 This optimization cannot be used for a subject string that might match
5916 only partially. If the pattern was studied, PCRE knows the minimum
5917 length of a matching string, and does not bother to run the matching
5918 function on shorter strings. This optimization is also disabled for
5919 partial matching.
5924 A partial match occurs during a call to pcre_exec() whenever the end of
5925 the subject string is reached successfully, but matching cannot con-
5926 tinue because more characters are needed. However, at least one charac-
5927 ter must have been matched. (In other words, a partial match can never
5928 be an empty string.)
5930 If PCRE_PARTIAL_SOFT is set, the partial match is remembered, but
5931 matching continues as normal, and other alternatives in the pattern are
5932 tried. If no complete match can be found, pcre_exec() returns
5933 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. If there are at least
5934 two slots in the offsets vector, the first of them is set to the offset
5935 of the earliest character that was inspected when the partial match was
5936 found. For convenience, the second offset points to the end of the
5937 string so that a substring can easily be identified.
5939 For the majority of patterns, the first offset identifies the start of
5940 the partially matched string. However, for patterns that contain look-
5941 behind assertions, or \K, or begin with \b or \B, earlier characters
5942 have been inspected while carrying out the match. For example:
5944 /(?<=abc)123/
5946 This pattern matches "123", but only if it is preceded by "abc". If the
5947 subject string is "xyzabc12", the offsets after a partial match are for
5948 the substring "abc12", because all these characters are needed if
5949 another match is tried with extra characters added.
5951 If there is more than one partial match, the first one that was found
5952 provides the data that is returned. Consider this pattern:
5954 /123\w+X|dogY/
5956 If this is matched against the subject string "abc123dog", both alter-
5957 natives fail to match, but the end of the subject is reached during
5958 matching, so PCRE_ERROR_PARTIAL is returned instead of
5959 PCRE_ERROR_NOMATCH. The offsets are set to 3 and 9, identifying
5960 "123dog" as the first partial match that was found. (In this example,
5961 there are two partial matches, because "dog" on its own partially
5962 matches the second alternative.)
5964 If PCRE_PARTIAL_HARD is set for pcre_exec(), it returns PCRE_ERROR_PAR-
5965 TIAL as soon as a partial match is found, without continuing to search
5966 for possible complete matches. The difference between the two options
5967 can be illustrated by a pattern such as:
5969 /dog(sbody)?/
5971 This matches either "dog" or "dogsbody", greedily (that is, it prefers
5972 the longer string if possible). If it is matched against the string
5973 "dog" with PCRE_PARTIAL_SOFT, it yields a complete match for "dog".
5974 However, if PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL.
5975 On the other hand, if the pattern is made ungreedy the result is dif-
5976 ferent:
5978 /dog(sbody)??/
5980 In this case the result is always a complete match because pcre_exec()
5981 finds that first, and it never continues after finding a match. It
5982 might be easier to follow this explanation by thinking of the two pat-
5983 terns like this:
5985 /dog(sbody)?/ is the same as /dogsbody|dog/
5986 /dog(sbody)??/ is the same as /dog|dogsbody/
5988 The second pattern will never match "dogsbody" when pcre_exec() is
5989 used, because it will always find the shorter match first.
5992 PARTIAL MATCHING USING pcre_dfa_exec()
5994 The pcre_dfa_exec() function moves along the subject string character
5995 by character, without backtracking, searching for all possible matches
5996 simultaneously. If the end of the subject is reached before the end of
5997 the pattern, there is the possibility of a partial match, again pro-
5998 vided that at least one character has matched.
6000 When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if
6001 there have been no complete matches. Otherwise, the complete matches
6002 are returned. However, if PCRE_PARTIAL_HARD is set, a partial match
6003 takes precedence over any complete matches. The portion of the string
6004 that was inspected when the longest partial match was found is set as
6005 the first matching string, provided there are at least two slots in the
6006 offsets vector.
6008 Because pcre_dfa_exec() always searches for all possible matches, and
6009 there is no difference between greedy and ungreedy repetition, its be-
6010 haviour is different from pcre_exec when PCRE_PARTIAL_HARD is set. Con-
6011 sider the string "dog" matched against the ungreedy pattern shown
6012 above:
6014 /dog(sbody)??/
6016 Whereas pcre_exec() stops as soon as it finds the complete match for
6017 "dog", pcre_dfa_exec() also finds the partial match for "dogsbody", and
6018 so returns that when PCRE_PARTIAL_HARD is set.
6023 If a pattern ends with one of sequences \b or \B, which test for word
6024 boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-
6025 intuitive results. Consider this pattern:
6027 /\bcat\b/
6029 This matches "cat", provided there is a word boundary at either end. If
6030 the subject string is "the cat", the comparison of the final "t" with a
6031 following character cannot take place, so a partial match is found.
6032 However, pcre_exec() carries on with normal matching, which matches \b
6033 at the end of the subject when the last character is a letter, thus
6034 finding a complete match. The result, therefore, is not PCRE_ERROR_PAR-
6035 TIAL. The same thing happens with pcre_dfa_exec(), because it also
6036 finds the complete match.
6038 Using PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL,
6039 because then the partial match takes precedence.
6044 For releases of PCRE prior to 8.00, because of the way certain internal
6045 optimizations were implemented in the pcre_exec() function, the
6046 PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be
6047 used with all patterns. From release 8.00 onwards, the restrictions no
6048 longer apply, and partial matching with pcre_exec() can be requested
6049 for any pattern.
6051 Items that were formerly restricted were rep