ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 903 - (show annotations)
Sat Jan 21 16:37:17 2012 UTC (7 years, 7 months ago) by ph10
File MIME type: text/plain
File size: 436473 byte(s)
Source file tidies for 8.30-RC1 release; fix Makefile.am bugs for building 
symbolic links to man pages.
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 Starting with release 8.30, it is possible to compile two separate PCRE
29 libraries: the original, which supports 8-bit character strings
30 (including UTF-8 strings), and a second library that supports 16-bit
31 character strings (including UTF-16 strings). The build process allows
32 either one or both to be built. The majority of the work to make this
33 possible was done by Zoltan Herczeg.
35 The two libraries contain identical sets of functions, except that the
36 names in the 16-bit library start with pcre16_ instead of pcre_. To
37 avoid over-complication and reduce the documentation maintenance load,
38 most of the documentation describes the 8-bit library, with the differ-
39 ences for the 16-bit library described separately in the pcre16 page.
40 References to functions or structures of the form pcre[16]_xxx should
41 be read as meaning "pcre_xxx when using the 8-bit library and
42 pcre16_xxx when using the 16-bit library".
44 The current implementation of PCRE corresponds approximately with Perl
45 5.12, including support for UTF-8/16 encoded strings and Unicode gen-
46 eral category properties. However, UTF-8/16 and Unicode support has to
47 be explicitly enabled; it is not the default. The Unicode tables corre-
48 spond to Unicode release 6.0.0.
50 In addition to the Perl-compatible matching function, PCRE contains an
51 alternative function that matches the same compiled patterns in a dif-
52 ferent way. In certain circumstances, the alternative function has some
53 advantages. For a discussion of the two matching algorithms, see the
54 pcrematching page.
56 PCRE is written in C and released as a C library. A number of people
57 have written wrappers and interfaces of various kinds. In particular,
58 Google Inc. have provided a comprehensive C++ wrapper for the 8-bit
59 library. This is now included as part of the PCRE distribution. The
60 pcrecpp page has details of this interface. Other people's contribu-
61 tions can be found in the Contrib directory at the primary FTP site,
62 which is:
64 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
66 Details of exactly which Perl regular expression features are and are
67 not supported by PCRE are given in separate documents. See the pcrepat-
68 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
69 page.
71 Some features of PCRE can be included, excluded, or changed when the
72 library is built. The pcre_config() function makes it possible for a
73 client to discover which features are available. The features them-
74 selves are described in the pcrebuild page. Documentation about build-
75 ing PCRE for various operating systems can be found in the README and
76 NON-UNIX-USE files in the source distribution.
78 The libraries contains a number of undocumented internal functions and
79 data tables that are used by more than one of the exported external
80 functions, but which are not intended for use by external callers.
81 Their names all begin with "_pcre_" or "_pcre16_", which hopefully will
82 not provoke any name clashes. In some environments, it is possible to
83 control which external symbols are exported when a shared library is
84 built, and in these cases the undocumented symbols are not exported.
89 The user documentation for PCRE comprises a number of different sec-
90 tions. In the "man" format, each of these is a separate "man page". In
91 the HTML format, each is a separate page, linked from the index page.
92 In the plain text format, all the sections, except the pcredemo sec-
93 tion, are concatenated, for ease of searching. The sections are as fol-
94 lows:
96 pcre this document
97 pcre16 details of the 16-bit library
98 pcre-config show PCRE installation configuration information
99 pcreapi details of PCRE's native C API
100 pcrebuild options for building PCRE
101 pcrecallout details of the callout feature
102 pcrecompat discussion of Perl compatibility
103 pcrecpp details of the C++ wrapper for the 8-bit library
104 pcredemo a demonstration C program that uses PCRE
105 pcregrep description of the pcregrep command (8-bit only)
106 pcrejit discussion of the just-in-time optimization support
107 pcrelimits details of size and other limits
108 pcrematching discussion of the two matching algorithms
109 pcrepartial details of the partial matching facility
110 pcrepattern syntax and semantics of supported
111 regular expressions
112 pcreperform discussion of performance issues
113 pcreposix the POSIX-compatible C API for the 8-bit library
114 pcreprecompile details of saving and re-using precompiled patterns
115 pcresample discussion of the pcredemo program
116 pcrestack discussion of stack usage
117 pcresyntax quick syntax reference
118 pcretest description of the pcretest testing command
119 pcreunicode discussion of Unicode and UTF-8/16 support
121 In addition, in the "man" and HTML formats, there is a short page for
122 each 8-bit C library function, listing its arguments and results.
127 Philip Hazel
128 University Computing Service
129 Cambridge CB2 3QH, England.
131 Putting an actual email address here seems to have been a spam magnet,
132 so I've taken it away. If you want to email me, use my two initials,
133 followed by the two digits 10, at the domain cam.ac.uk.
138 Last updated: 10 January 2012
139 Copyright (c) 1997-2012 University of Cambridge.
140 ------------------------------------------------------------------------------
143 PCRE(3) PCRE(3)
146 NAME
147 PCRE - Perl-compatible regular expressions
149 #include <pcre.h>
154 pcre16 *pcre16_compile(PCRE_SPTR16 pattern, int options,
155 const char **errptr, int *erroffset,
156 const unsigned char *tableptr);
158 pcre16 *pcre16_compile2(PCRE_SPTR16 pattern, int options,
159 int *errorcodeptr,
160 const char **errptr, int *erroffset,
161 const unsigned char *tableptr);
163 pcre16_extra *pcre16_study(const pcre16 *code, int options,
164 const char **errptr);
166 void pcre16_free_study(pcre16_extra *extra);
168 int pcre16_exec(const pcre16 *code, const pcre16_extra *extra,
169 PCRE_SPTR16 subject, int length, int startoffset,
170 int options, int *ovector, int ovecsize);
172 int pcre16_dfa_exec(const pcre16 *code, const pcre16_extra *extra,
173 PCRE_SPTR16 subject, int length, int startoffset,
174 int options, int *ovector, int ovecsize,
175 int *workspace, int wscount);
180 int pcre16_copy_named_substring(const pcre16 *code,
181 PCRE_SPTR16 subject, int *ovector,
182 int stringcount, PCRE_SPTR16 stringname,
183 PCRE_UCHAR16 *buffer, int buffersize);
185 int pcre16_copy_substring(PCRE_SPTR16 subject, int *ovector,
186 int stringcount, int stringnumber, PCRE_UCHAR16 *buffer,
187 int buffersize);
189 int pcre16_get_named_substring(const pcre16 *code,
190 PCRE_SPTR16 subject, int *ovector,
191 int stringcount, PCRE_SPTR16 stringname,
192 PCRE_SPTR16 *stringptr);
194 int pcre16_get_stringnumber(const pcre16 *code,
195 PCRE_SPTR16 name);
197 int pcre16_get_stringtable_entries(const pcre16 *code,
198 PCRE_SPTR16 name, PCRE_UCHAR16 **first, PCRE_UCHAR16 **last);
200 int pcre16_get_substring(PCRE_SPTR16 subject, int *ovector,
201 int stringcount, int stringnumber,
202 PCRE_SPTR16 *stringptr);
204 int pcre16_get_substring_list(PCRE_SPTR16 subject,
205 int *ovector, int stringcount, PCRE_SPTR16 **listptr);
207 void pcre16_free_substring(PCRE_SPTR16 stringptr);
209 void pcre16_free_substring_list(PCRE_SPTR16 *stringptr);
214 pcre16_jit_stack *pcre16_jit_stack_alloc(int startsize, int maxsize);
216 void pcre16_jit_stack_free(pcre16_jit_stack *stack);
218 void pcre16_assign_jit_stack(pcre16_extra *extra,
219 pcre16_jit_callback callback, void *data);
221 const unsigned char *pcre16_maketables(void);
223 int pcre16_fullinfo(const pcre16 *code, const pcre16_extra *extra,
224 int what, void *where);
226 int pcre16_refcount(pcre16 *code, int adjust);
228 int pcre16_config(int what, void *where);
230 const char *pcre16_version(void);
232 int pcre16_pattern_to_host_byte_order(pcre16 *code,
233 pcre16_extra *extra, const unsigned char *tables);
238 void *(*pcre16_malloc)(size_t);
240 void (*pcre16_free)(void *);
242 void *(*pcre16_stack_malloc)(size_t);
244 void (*pcre16_stack_free)(void *);
246 int (*pcre16_callout)(pcre16_callout_block *);
251 int pcre16_utf16_to_host_byte_order(PCRE_UCHAR16 *output,
252 PCRE_SPTR16 input, int length, int *byte_order,
253 int keep_boms);
258 Starting with release 8.30, it is possible to compile a PCRE library
259 that supports 16-bit character strings, including UTF-16 strings, as
260 well as or instead of the original 8-bit library. The majority of the
261 work to make this possible was done by Zoltan Herczeg. The two
262 libraries contain identical sets of functions, used in exactly the same
263 way. Only the names of the functions and the data types of their argu-
264 ments and results are different. To avoid over-complication and reduce
265 the documentation maintenance load, most of the PCRE documentation
266 describes the 8-bit library, with only occasional references to the
267 16-bit library. This page describes what is different when you use the
268 16-bit library.
270 WARNING: A single application can be linked with both libraries, but
271 you must take care when processing any particular pattern to use func-
272 tions from just one library. For example, if you want to study a pat-
273 tern that was compiled with pcre16_compile(), you must do so with
274 pcre16_study(), not pcre_study(), and you must free the study data with
275 pcre16_free_study().
280 There is only one header file, pcre.h. It contains prototypes for all
281 the functions in both libraries, as well as definitions of flags,
282 structures, error codes, etc.
287 In Unix-like systems, the 16-bit library is called libpcre16, and can
288 normally be accesss by adding -lpcre16 to the command for linking an
289 application that uses PCRE.
294 In the 8-bit library, strings are passed to PCRE library functions as
295 vectors of bytes with the C type "char *". In the 16-bit library,
296 strings are passed as vectors of unsigned 16-bit quantities. The macro
297 PCRE_UCHAR16 specifies an appropriate data type, and PCRE_SPTR16 is
298 defined as "const PCRE_UCHAR16 *". In very many environments, "short
299 int" is a 16-bit data type. When PCRE is built, it defines PCRE_UCHAR16
300 as "short int", but checks that it really is a 16-bit data type. If it
301 is not, the build fails with an error message telling the maintainer to
302 modify the definition appropriately.
307 The types of the opaque structures that are used for compiled 16-bit
308 patterns and JIT stacks are pcre16 and pcre16_jit_stack respectively.
309 The type of the user-accessible structure that is returned by
310 pcre16_study() is pcre16_extra, and the type of the structure that is
311 used for passing data to a callout function is pcre16_callout_block.
312 These structures contain the same fields, with the same names, as their
313 8-bit counterparts. The only difference is that pointers to character
314 strings are 16-bit instead of 8-bit types.
319 For every function in the 8-bit library there is a corresponding func-
320 tion in the 16-bit library with a name that starts with pcre16_ instead
321 of pcre_. The prototypes are listed above. In addition, there is one
322 extra function, pcre16_utf16_to_host_byte_order(). This is a utility
323 function that converts a UTF-16 character string to host byte order if
324 necessary. The other 16-bit functions expect the strings they are
325 passed to be in host byte order.
327 The input and output arguments of pcre16_utf16_to_host_byte_order() may
328 point to the same address, that is, conversion in place is supported.
329 The output buffer must be at least as long as the input.
331 The length argument specifies the number of 16-bit data units in the
332 input string; a negative value specifies a zero-terminated string.
334 If byte_order is NULL, it is assumed that the string starts off in host
335 byte order. This may be changed by byte-order marks (BOMs) anywhere in
336 the string (commonly as the first character).
338 If byte_order is not NULL, a non-zero value of the integer to which it
339 points means that the input starts off in host byte order, otherwise
340 the opposite order is assumed. Again, BOMs in the string can change
341 this. The final byte order is passed back at the end of processing.
343 If keep_boms is not zero, byte-order mark characters (0xfeff) are
344 copied into the output string. Otherwise they are discarded.
346 The result of the function is the number of 16-bit units placed into
347 the output buffer, including the zero terminator if the string was
348 zero-terminated.
353 The offsets within subject strings that are returned by the matching
354 functions are in 16-bit units rather than bytes.
359 The name-to-number translation table that is maintained for named sub-
360 patterns uses 16-bit characters. The pcre16_get_stringtable_entries()
361 function returns the length of each entry in the table as the number of
362 16-bit data units.
367 There are two new general option names, PCRE_UTF16 and
368 PCRE_NO_UTF16_CHECK, which correspond to PCRE_UTF8 and
369 PCRE_NO_UTF8_CHECK in the 8-bit library. In fact, these new options
370 define the same bits in the options word.
372 For the pcre16_config() function there is an option PCRE_CONFIG_UTF16
373 that returns 1 if UTF-16 support is configured, otherwise 0. If this
374 option is given to pcre_config(), or if the PCRE_CONFIG_UTF8 option is
375 given to pcre16_config(), the result is the PCRE_ERROR_BADOPTION error.
380 In 16-bit mode, when PCRE_UTF16 is not set, character values are
381 treated in the same way as in 8-bit, non UTF-8 mode, except, of course,
382 that they can range from 0 to 0xffff instead of 0 to 0xff. Character
383 types for characters less than 0xff can therefore be influenced by the
384 locale in the same way as before. Characters greater than 0xff have
385 only one case, and no "type" (such as letter or digit).
387 In UTF-16 mode, the character code is Unicode, in the range 0 to
388 0x10ffff, with the exception of values in the range 0xd800 to 0xdfff
389 because those are "surrogate" values that are used in pairs to encode
390 values greater than 0xffff.
392 A UTF-16 string can indicate its endianness by special code knows as a
393 byte-order mark (BOM). The PCRE functions do not handle this, expecting
394 strings to be in host byte order. A utility function called
395 pcre16_utf16_to_host_byte_order() is provided to help with this (see
396 above).
402 spond to their 8-bit counterparts. The error PCRE_ERROR_BADMODE is
403 given when a compiled pattern is passed to a function that processes
404 patterns in the other mode, for example, if a pattern compiled with
405 pcre_compile() is passed to pcre16_exec().
407 There are new error codes whose names begin with PCRE_UTF16_ERR for
408 invalid UTF-16 strings, corresponding to the PCRE_UTF8_ERR codes for
409 UTF-8 strings that are described in the section entitled "Reason codes
410 for invalid UTF-8 strings" in the main pcreapi page. The UTF-16 errors
411 are:
413 PCRE_UTF16_ERR1 Missing low surrogate at end of string
414 PCRE_UTF16_ERR2 Invalid low surrogate follows high surrogate
415 PCRE_UTF16_ERR3 Isolated low surrogate
416 PCRE_UTF16_ERR4 Invalid character 0xfffe
421 If there is an error while compiling a pattern, the error text that is
422 passed back by pcre16_compile() or pcre16_compile2() is still an 8-bit
423 character string, zero-terminated.
428 The subject and mark fields in the callout block that is passed to a
429 callout function point to 16-bit vectors.
434 The pcretest program continues to operate with 8-bit input and output
435 files, but it can be used for testing the 16-bit library. If it is run
436 with the command line option -16, patterns and subject strings are con-
437 verted from 8-bit to 16-bit before being passed to PCRE, and the 16-bit
438 library functions are used instead of the 8-bit ones. Returned 16-bit
439 strings are converted to 8-bit for output. If the 8-bit library was not
440 compiled, pcretest defaults to 16-bit and the -16 option is ignored.
442 When PCRE is being built, the RunTest script that is called by "make
443 check" uses the pcretest -C option to discover which of the 8-bit and
444 16-bit libraries has been built, and runs the tests appropriately.
449 Not all the features of the 8-bit library are available with the 16-bit
450 library. The C++ and POSIX wrapper functions support only the 8-bit
451 library, and the pcregrep program is at present 8-bit only.
456 Philip Hazel
457 University Computing Service
458 Cambridge CB2 3QH, England.
463 Last updated: 08 January 2012
464 Copyright (c) 1997-2012 University of Cambridge.
465 ------------------------------------------------------------------------------
471 NAME
472 PCRE - Perl-compatible regular expressions
477 This document describes the optional features of PCRE that can be
478 selected when the library is compiled. It assumes use of the configure
479 script, where the optional features are selected or deselected by pro-
480 viding options to configure before running the make command. However,
481 the same options can be selected in both Unix-like and non-Unix-like
482 environments using the GUI facility of cmake-gui if you are using CMake
483 instead of configure to build PCRE.
485 There is a lot more information about building PCRE in non-Unix-like
486 environments in the file called NON_UNIX_USE, which is part of the PCRE
487 distribution. You should consult this file as well as the README file
488 if you are building in a non-Unix-like environment.
490 The complete list of options for configure (which includes the standard
491 ones such as the selection of the installation directory) can be
492 obtained by running
494 ./configure --help
496 The following sections include descriptions of options whose names
497 begin with --enable or --disable. These settings specify changes to the
498 defaults for the configure command. Because of the way that configure
499 works, --enable and --disable always come in pairs, so the complemen-
500 tary option always exists as well, but as it specifies the default, it
501 is not described.
506 By default, a library called libpcre is built, containing functions
507 that take string arguments contained in vectors of bytes, either as
508 single-byte characters, or interpreted as UTF-8 strings. You can also
509 build a separate library, called libpcre16, in which strings are con-
510 tained in vectors of 16-bit data units and interpreted either as sin-
511 gle-unit characters or UTF-16 strings, by adding
513 --enable-pcre16
515 to the configure command. If you do not want the 8-bit library, add
517 --disable-pcre8
519 as well. At least one of the two libraries must be built. Note that the
520 C++ and POSIX wrappers are for the 8-bit library only, and that pcre-
521 grep is an 8-bit program. None of these are built if you select only
522 the 16-bit library.
527 The PCRE building process uses libtool to build both shared and static
528 Unix libraries by default. You can suppress one of these by adding one
529 of
531 --disable-shared
532 --disable-static
534 to the configure command, as required.
539 By default, if the 8-bit library is being built, the configure script
540 will search for a C++ compiler and C++ header files. If it finds them,
541 it automatically builds the C++ wrapper library (which supports only
542 8-bit strings). You can disable this by adding
544 --disable-cpp
546 to the configure command.
549 UTF-8 and UTF-16 SUPPORT
551 To build PCRE with support for UTF Unicode character strings, add
553 --enable-utf
555 to the configure command. This setting applies to both libraries,
556 adding support for UTF-8 to the 8-bit library and support for UTF-16 to
557 the 16-bit library. There are no separate options for enabling UTF-8
558 and UTF-16 independently because that would allow ridiculous settings
559 such as requesting UTF-16 support while building only the 8-bit
560 library. It is not possible to build one library with UTF support and
561 the other without in the same configuration. (For backwards compatibil-
562 ity, --enable-utf8 is a synonym of --enable-utf.)
564 Of itself, this setting does not make PCRE treat strings as UTF-8 or
565 UTF-16. As well as compiling PCRE with this option, you also have have
566 to set the PCRE_UTF8 or PCRE_UTF16 option when you call one of the pat-
567 tern compiling functions.
569 If you set --enable-utf when compiling in an EBCDIC environment, PCRE
570 expects its input to be either ASCII or UTF-8 (depending on the runtime
571 option). It is not possible to support both EBCDIC and UTF-8 codes in
572 the same version of the library. Consequently, --enable-utf and
573 --enable-ebcdic are mutually exclusive.
578 UTF support allows the libraries to process character codepoints up to
579 0x10ffff in the strings that they handle. On its own, however, it does
580 not provide any facilities for accessing the properties of such charac-
581 ters. If you want to be able to use the pattern escapes \P, \p, and \X,
582 which refer to Unicode character properties, you must add
584 --enable-unicode-properties
586 to the configure command. This implies UTF support, even if you have
587 not explicitly requested it.
589 Including Unicode property support adds around 30K of tables to the
590 PCRE library. Only the general category properties such as Lu and Nd
591 are supported. Details are given in the pcrepattern documentation.
596 Just-in-time compiler support is included in the build by specifying
598 --enable-jit
600 This support is available only for certain hardware architectures. If
601 this option is set for an unsupported architecture, a compile time
602 error occurs. See the pcrejit documentation for a discussion of JIT
603 usage. When JIT support is enabled, pcregrep automatically makes use of
604 it, unless you add
606 --disable-pcregrep-jit
608 to the "configure" command.
613 By default, PCRE interprets the linefeed (LF) character as indicating
614 the end of a line. This is the normal newline character on Unix-like
615 systems. You can compile PCRE to use carriage return (CR) instead, by
616 adding
618 --enable-newline-is-cr
620 to the configure command. There is also a --enable-newline-is-lf
621 option, which explicitly specifies linefeed as the newline character.
623 Alternatively, you can specify that line endings are to be indicated by
624 the two character sequence CRLF. If you want this, add
626 --enable-newline-is-crlf
628 to the configure command. There is a fourth option, specified by
630 --enable-newline-is-anycrlf
632 which causes PCRE to recognize any of the three sequences CR, LF, or
633 CRLF as indicating a line ending. Finally, a fifth option, specified by
635 --enable-newline-is-any
637 causes PCRE to recognize any Unicode newline sequence.
639 Whatever line ending convention is selected when PCRE is built can be
640 overridden when the library functions are called. At build time it is
641 conventional to use the standard for your operating system.
646 By default, the sequence \R in a pattern matches any Unicode newline
647 sequence, whatever has been selected as the line ending sequence. If
648 you specify
650 --enable-bsr-anycrlf
652 the default is changed so that \R matches only CR, LF, or CRLF. What-
653 ever is selected when PCRE is built can be overridden when the library
654 functions are called.
659 When the 8-bit library is called through the POSIX interface (see the
660 pcreposix documentation), additional working storage is required for
661 holding the pointers to capturing substrings, because PCRE requires
662 three integers per substring, whereas the POSIX interface provides only
663 two. If the number of expected substrings is small, the wrapper func-
664 tion uses space on the stack, because this is faster than using mal-
665 loc() for each call. The default threshold above which the stack is no
666 longer used is 10; it can be changed by adding a setting such as
668 --with-posix-malloc-threshold=20
670 to the configure command.
675 Within a compiled pattern, offset values are used to point from one
676 part to another (for example, from an opening parenthesis to an alter-
677 nation metacharacter). By default, two-byte values are used for these
678 offsets, leading to a maximum size for a compiled pattern of around
679 64K. This is sufficient to handle all but the most gigantic patterns.
680 Nevertheless, some people do want to process truly enormous patterns,
681 so it is possible to compile PCRE to use three-byte or four-byte off-
682 sets by adding a setting such as
684 --with-link-size=3
686 to the configure command. The value given must be 2, 3, or 4. For the
687 16-bit library, a value of 3 is rounded up to 4. Using longer offsets
688 slows down the operation of PCRE because it has to load additional data
689 when handling them.
694 When matching with the pcre_exec() function, PCRE implements backtrack-
695 ing by making recursive calls to an internal function called match().
696 In environments where the size of the stack is limited, this can se-
697 verely limit PCRE's operation. (The Unix environment does not usually
698 suffer from this problem, but it may sometimes be necessary to increase
699 the maximum stack size. There is a discussion in the pcrestack docu-
700 mentation.) An alternative approach to recursion that uses memory from
701 the heap to remember data, instead of using recursive function calls,
702 has been implemented to work round the problem of limited stack size.
703 If you want to build a version of PCRE that works this way, add
705 --disable-stack-for-recursion
707 to the configure command. With this configuration, PCRE will use the
708 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
709 ment functions. By default these point to malloc() and free(), but you
710 can replace the pointers so that your own functions are used instead.
712 Separate functions are provided rather than using pcre_malloc and
713 pcre_free because the usage is very predictable: the block sizes
714 requested are always the same, and the blocks are always freed in
715 reverse order. A calling program might be able to implement optimized
716 functions that perform better than malloc() and free(). PCRE runs
717 noticeably more slowly when built in this way. This option affects only
718 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
723 Internally, PCRE has a function called match(), which it calls repeat-
724 edly (sometimes recursively) when matching a pattern with the
725 pcre_exec() function. By controlling the maximum number of times this
726 function may be called during a single matching operation, a limit can
727 be placed on the resources used by a single call to pcre_exec(). The
728 limit can be changed at run time, as described in the pcreapi documen-
729 tation. The default is 10 million, but this can be changed by adding a
730 setting such as
732 --with-match-limit=500000
734 to the configure command. This setting has no effect on the
735 pcre_dfa_exec() matching function.
737 In some environments it is desirable to limit the depth of recursive
738 calls of match() more strictly than the total number of calls, in order
739 to restrict the maximum amount of stack (or heap, if --disable-stack-
740 for-recursion is specified) that is used. A second limit controls this;
741 it defaults to the value that is set for --with-match-limit, which
742 imposes no additional constraints. However, you can set a lower limit
743 by adding, for example,
745 --with-match-limit-recursion=10000
747 to the configure command. This value can also be overridden at run
748 time.
753 PCRE uses fixed tables for processing characters whose code values are
754 less than 256. By default, PCRE is built with a set of tables that are
755 distributed in the file pcre_chartables.c.dist. These tables are for
756 ASCII codes only. If you add
758 --enable-rebuild-chartables
760 to the configure command, the distributed tables are no longer used.
761 Instead, a program called dftables is compiled and run. This outputs
762 the source for new set of tables, created in the default locale of your
763 C runtime system. (This method of replacing the tables does not work if
764 you are cross compiling, because dftables is run on the local host. If
765 you need to create alternative tables when cross compiling, you will
766 have to do so "by hand".)
771 PCRE assumes by default that it will run in an environment where the
772 character code is ASCII (or Unicode, which is a superset of ASCII).
773 This is the case for most computer operating systems. PCRE can, how-
774 ever, be compiled to run in an EBCDIC environment by adding
776 --enable-ebcdic
778 to the configure command. This setting implies --enable-rebuild-charta-
779 bles. You should only use it if you know that you are in an EBCDIC
780 environment (for example, an IBM mainframe operating system). The
781 --enable-ebcdic option is incompatible with --enable-utf.
786 By default, pcregrep reads all files as plain text. You can build it so
787 that it recognizes files whose names end in .gz or .bz2, and reads them
788 with libz or libbz2, respectively, by adding one or both of
790 --enable-pcregrep-libz
791 --enable-pcregrep-libbz2
793 to the configure command. These options naturally require that the rel-
794 evant libraries are installed on your system. Configuration will fail
795 if they are not.
800 pcregrep uses an internal buffer to hold a "window" on the file it is
801 scanning, in order to be able to output "before" and "after" lines when
802 it finds a match. The size of the buffer is controlled by a parameter
803 whose default value is 20K. The buffer itself is three times this size,
804 but because of the way it is used for holding "before" lines, the long-
805 est line that is guaranteed to be processable is the parameter size.
806 You can change the default parameter value by adding, for example,
808 --with-pcregrep-bufsize=50K
810 to the configure command. The caller of pcregrep can, however, override
811 this value by specifying a run-time option.
816 If you add
818 --enable-pcretest-libreadline
820 to the configure command, pcretest is linked with the libreadline
821 library, and when its input is from a terminal, it reads it using the
822 readline() function. This provides line-editing and history facilities.
823 Note that libreadline is GPL-licensed, so if you distribute a binary of
824 pcretest linked in this way, there may be licensing issues.
826 Setting this option causes the -lreadline option to be added to the
827 pcretest build. In many operating environments with a sytem-installed
828 libreadline this is sufficient. However, in some environments (e.g. if
829 an unmodified distribution version of readline is in use), some extra
830 configuration may be necessary. The INSTALL file for libreadline says
831 this:
833 "Readline uses the termcap functions, but does not link with the
834 termcap or curses library itself, allowing applications which link
835 with readline the to choose an appropriate library."
837 If your environment has not been set up so that an appropriate library
838 is automatically included, you may need to add something like
840 LIBS="-ncurses"
842 immediately before the configure command.
847 pcreapi(3), pcre16, pcre_config(3).
852 Philip Hazel
853 University Computing Service
854 Cambridge CB2 3QH, England.
859 Last updated: 07 January 2012
860 Copyright (c) 1997-2012 University of Cambridge.
861 ------------------------------------------------------------------------------
867 NAME
868 PCRE - Perl-compatible regular expressions
873 This document describes the two different algorithms that are available
874 in PCRE for matching a compiled regular expression against a given sub-
875 ject string. The "standard" algorithm is the one provided by the
876 pcre_exec() and pcre16_exec() functions. These work in the same was as
877 Perl's matching function, and provide a Perl-compatible matching opera-
878 tion. The just-in-time (JIT) optimization that is described in the
879 pcrejit documentation is compatible with these functions.
881 An alternative algorithm is provided by the pcre_dfa_exec() and
882 pcre16_dfa_exec() functions; they operate in a different way, and are
883 not Perl-compatible. This alternative has advantages and disadvantages
884 compared with the standard algorithm, and these are described below.
886 When there is only one possible way in which a given subject string can
887 match a pattern, the two algorithms give the same answer. A difference
888 arises, however, when there are multiple possibilities. For example, if
889 the pattern
891 ^<.*>
893 is matched against the string
895 <something> <something else> <something further>
897 there are three possible answers. The standard algorithm finds only one
898 of them, whereas the alternative algorithm finds all three.
903 The set of strings that are matched by a regular expression can be rep-
904 resented as a tree structure. An unlimited repetition in the pattern
905 makes the tree of infinite size, but it is still a tree. Matching the
906 pattern to a given subject string (from a given starting point) can be
907 thought of as a search of the tree. There are two ways to search a
908 tree: depth-first and breadth-first, and these correspond to the two
909 matching algorithms provided by PCRE.
914 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
915 sions", the standard algorithm is an "NFA algorithm". It conducts a
916 depth-first search of the pattern tree. That is, it proceeds along a
917 single path through the tree, checking that the subject matches what is
918 required. When there is a mismatch, the algorithm tries any alterna-
919 tives at the current point, and if they all fail, it backs up to the
920 previous branch point in the tree, and tries the next alternative
921 branch at that level. This often involves backing up (moving to the
922 left) in the subject string as well. The order in which repetition
923 branches are tried is controlled by the greedy or ungreedy nature of
924 the quantifier.
926 If a leaf node is reached, a matching string has been found, and at
927 that point the algorithm stops. Thus, if there is more than one possi-
928 ble match, this algorithm returns the first one that it finds. Whether
929 this is the shortest, the longest, or some intermediate length depends
930 on the way the greedy and ungreedy repetition quantifiers are specified
931 in the pattern.
933 Because it ends up with a single path through the tree, it is rela-
934 tively straightforward for this algorithm to keep track of the sub-
935 strings that are matched by portions of the pattern in parentheses.
936 This provides support for capturing parentheses and back references.
941 This algorithm conducts a breadth-first search of the tree. Starting
942 from the first matching point in the subject, it scans the subject
943 string from left to right, once, character by character, and as it does
944 this, it remembers all the paths through the tree that represent valid
945 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
946 though it is not implemented as a traditional finite state machine (it
947 keeps multiple states active simultaneously).
949 Although the general principle of this matching algorithm is that it
950 scans the subject string only once, without backtracking, there is one
951 exception: when a lookaround assertion is encountered, the characters
952 following or preceding the current point have to be independently
953 inspected.
955 The scan continues until either the end of the subject is reached, or
956 there are no more unterminated paths. At this point, terminated paths
957 represent the different matching possibilities (if there are none, the
958 match has failed). Thus, if there is more than one possible match,
959 this algorithm finds all of them, and in particular, it finds the long-
960 est. The matches are returned in decreasing order of length. There is
961 an option to stop the algorithm after the first match (which is neces-
962 sarily the shortest) is found.
964 Note that all the matches that are found start at the same point in the
965 subject. If the pattern
967 cat(er(pillar)?)?
969 is matched against the string "the caterpillar catchment", the result
970 will be the three strings "caterpillar", "cater", and "cat" that start
971 at the fifth character of the subject. The algorithm does not automati-
972 cally move on to find matches that start at later positions.
974 There are a number of features of PCRE regular expressions that are not
975 supported by the alternative matching algorithm. They are as follows:
977 1. Because the algorithm finds all possible matches, the greedy or
978 ungreedy nature of repetition quantifiers is not relevant. Greedy and
979 ungreedy quantifiers are treated in exactly the same way. However, pos-
980 sessive quantifiers can make a difference when what follows could also
981 match what is quantified, for example in a pattern like this:
983 ^a++\w!
985 This pattern matches "aaab!" but not "aaa!", which would be matched by
986 a non-possessive quantifier. Similarly, if an atomic group is present,
987 it is matched as if it were a standalone pattern at the current point,
988 and the longest match is then "locked in" for the rest of the overall
989 pattern.
991 2. When dealing with multiple paths through the tree simultaneously, it
992 is not straightforward to keep track of captured substrings for the
993 different matching possibilities, and PCRE's implementation of this
994 algorithm does not attempt to do this. This means that no captured sub-
995 strings are available.
997 3. Because no substrings are captured, back references within the pat-
998 tern are not supported, and cause errors if encountered.
1000 4. For the same reason, conditional expressions that use a backrefer-
1001 ence as the condition or test for a specific group recursion are not
1002 supported.
1004 5. Because many paths through the tree may be active, the \K escape
1005 sequence, which resets the start of the match when encountered (but may
1006 be on some paths and not on others), is not supported. It causes an
1007 error if encountered.
1009 6. Callouts are supported, but the value of the capture_top field is
1010 always 1, and the value of the capture_last field is always -1.
1012 7. The \C escape sequence, which (in the standard algorithm) always
1013 matches a single data unit, even in UTF-8 or UTF-16 modes, is not sup-
1014 ported in these modes, because the alternative algorithm moves through
1015 the subject string one character (not data unit) at a time, for all
1016 active paths through the tree.
1018 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
1019 are not supported. (*FAIL) is supported, and behaves like a failing
1020 negative assertion.
1025 Using the alternative matching algorithm provides the following advan-
1026 tages:
1028 1. All possible matches (at a single point in the subject) are automat-
1029 ically found, and in particular, the longest match is found. To find
1030 more than one match using the standard algorithm, you have to do kludgy
1031 things with callouts.
1033 2. Because the alternative algorithm scans the subject string just
1034 once, and never needs to backtrack (except for lookbehinds), it is pos-
1035 sible to pass very long subject strings to the matching function in
1036 several pieces, checking for partial matching each time. Although it is
1037 possible to do multi-segment matching using the standard algorithm by
1038 retaining partially matched substrings, it is more complicated. The
1039 pcrepartial documentation gives details of partial matching and dis-
1040 cusses multi-segment matching.
1045 The alternative algorithm suffers from a number of disadvantages:
1047 1. It is substantially slower than the standard algorithm. This is
1048 partly because it has to search for all possible matches, but is also
1049 because it is less susceptible to optimization.
1051 2. Capturing parentheses and back references are not supported.
1053 3. Although atomic groups are supported, their use does not provide the
1054 performance advantage that it does for the standard algorithm.
1059 Philip Hazel
1060 University Computing Service
1061 Cambridge CB2 3QH, England.
1066 Last updated: 08 January 2012
1067 Copyright (c) 1997-2012 University of Cambridge.
1068 ------------------------------------------------------------------------------
1074 NAME
1075 PCRE - Perl-compatible regular expressions
1077 #include <pcre.h>
1082 pcre *pcre_compile(const char *pattern, int options,
1083 const char **errptr, int *erroffset,
1084 const unsigned char *tableptr);
1086 pcre *pcre_compile2(const char *pattern, int options,
1087 int *errorcodeptr,
1088 const char **errptr, int *erroffset,
1089 const unsigned char *tableptr);
1091 pcre_extra *pcre_study(const pcre *code, int options,
1092 const char **errptr);
1094 void pcre_free_study(pcre_extra *extra);
1096 int pcre_exec(const pcre *code, const pcre_extra *extra,
1097 const char *subject, int length, int startoffset,
1098 int options, int *ovector, int ovecsize);
1100 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
1101 const char *subject, int length, int startoffset,
1102 int options, int *ovector, int ovecsize,
1103 int *workspace, int wscount);
1108 int pcre_copy_named_substring(const pcre *code,
1109 const char *subject, int *ovector,
1110 int stringcount, const char *stringname,
1111 char *buffer, int buffersize);
1113 int pcre_copy_substring(const char *subject, int *ovector,
1114 int stringcount, int stringnumber, char *buffer,
1115 int buffersize);
1117 int pcre_get_named_substring(const pcre *code,
1118 const char *subject, int *ovector,
1119 int stringcount, const char *stringname,
1120 const char **stringptr);
1122 int pcre_get_stringnumber(const pcre *code,
1123 const char *name);
1125 int pcre_get_stringtable_entries(const pcre *code,
1126 const char *name, char **first, char **last);
1128 int pcre_get_substring(const char *subject, int *ovector,
1129 int stringcount, int stringnumber,
1130 const char **stringptr);
1132 int pcre_get_substring_list(const char *subject,
1133 int *ovector, int stringcount, const char ***listptr);
1135 void pcre_free_substring(const char *stringptr);
1137 void pcre_free_substring_list(const char **stringptr);
1142 pcre_jit_stack *pcre_jit_stack_alloc(int startsize, int maxsize);
1144 void pcre_jit_stack_free(pcre_jit_stack *stack);
1146 void pcre_assign_jit_stack(pcre_extra *extra,
1147 pcre_jit_callback callback, void *data);
1149 const unsigned char *pcre_maketables(void);
1151 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1152 int what, void *where);
1154 int pcre_refcount(pcre *code, int adjust);
1156 int pcre_config(int what, void *where);
1158 const char *pcre_version(void);
1160 int pcre_pattern_to_host_byte_order(pcre *code,
1161 pcre_extra *extra, const unsigned char *tables);
1166 void *(*pcre_malloc)(size_t);
1168 void (*pcre_free)(void *);
1170 void *(*pcre_stack_malloc)(size_t);
1172 void (*pcre_stack_free)(void *);
1174 int (*pcre_callout)(pcre_callout_block *);
1179 From release 8.30, PCRE can be compiled as a library for handling
1180 16-bit character strings as well as, or instead of, the original
1181 library that handles 8-bit character strings. To avoid too much compli-
1182 cation, this document describes the 8-bit versions of the functions,
1183 with only occasional references to the 16-bit library.
1185 The 16-bit functions operate in the same way as their 8-bit counter-
1186 parts; they just use different data types for their arguments and
1187 results, and their names start with pcre16_ instead of pcre_. For every
1188 option that has UTF8 in its name (for example, PCRE_UTF8), there is a
1189 corresponding 16-bit name with UTF8 replaced by UTF16. This facility is
1190 in fact just cosmetic; the 16-bit option names define the same bit val-
1191 ues.
1193 References to bytes and UTF-8 in this document should be read as refer-
1194 ences to 16-bit data quantities and UTF-16 when using the 16-bit
1195 library, unless specified otherwise. More details of the specific dif-
1196 ferences for the 16-bit library are given in the pcre16 page.
1201 PCRE has its own native API, which is described in this document. There
1202 are also some wrapper functions (for the 8-bit library only) that cor-
1203 respond to the POSIX regular expression API, but they do not give
1204 access to all the functionality. They are described in the pcreposix
1205 documentation. Both of these APIs define a set of C function calls. A
1206 C++ wrapper (again for the 8-bit library only) is also distributed with
1207 PCRE. It is documented in the pcrecpp page.
1209 The native API C function prototypes are defined in the header file
1210 pcre.h, and on Unix-like systems the (8-bit) library itself is called
1211 libpcre. It can normally be accessed by adding -lpcre to the command
1212 for linking an application that uses PCRE. The header file defines the
1213 macros PCRE_MAJOR and PCRE_MINOR to contain the major and minor release
1214 numbers for the library. Applications can use these to include support
1215 for different releases of PCRE.
1217 In a Windows environment, if you want to statically link an application
1218 program against a non-dll pcre.a file, you must define PCRE_STATIC
1219 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
1220 loc() and pcre_free() exported functions will be declared
1221 __declspec(dllimport), with unwanted results.
1223 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
1224 pcre_exec() are used for compiling and matching regular expressions in
1225 a Perl-compatible manner. A sample program that demonstrates the sim-
1226 plest way of using them is provided in the file called pcredemo.c in
1227 the PCRE source distribution. A listing of this program is given in the
1228 pcredemo documentation, and the pcresample documentation describes how
1229 to compile and run it.
1231 Just-in-time compiler support is an optional feature of PCRE that can
1232 be built in appropriate hardware environments. It greatly speeds up the
1233 matching performance of many patterns. Simple programs can easily
1234 request that it be used if available, by setting an option that is
1235 ignored when it is not relevant. More complicated programs might need
1236 to make use of the functions pcre_jit_stack_alloc(),
1237 pcre_jit_stack_free(), and pcre_assign_jit_stack() in order to control
1238 the JIT code's memory usage. These functions are discussed in the
1239 pcrejit documentation.
1241 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
1242 ble, is also provided. This uses a different algorithm for the match-
1243 ing. The alternative algorithm finds all possible matches (at a given
1244 point in the subject), and scans the subject just once (unless there
1245 are lookbehind assertions). However, this algorithm does not return
1246 captured substrings. A description of the two matching algorithms and
1247 their advantages and disadvantages is given in the pcrematching docu-
1248 mentation.
1250 In addition to the main compiling and matching functions, there are
1251 convenience functions for extracting captured substrings from a subject
1252 string that is matched by pcre_exec(). They are:
1254 pcre_copy_substring()
1255 pcre_copy_named_substring()
1256 pcre_get_substring()
1257 pcre_get_named_substring()
1258 pcre_get_substring_list()
1259 pcre_get_stringnumber()
1260 pcre_get_stringtable_entries()
1262 pcre_free_substring() and pcre_free_substring_list() are also provided,
1263 to free the memory used for extracted strings.
1265 The function pcre_maketables() is used to build a set of character
1266 tables in the current locale for passing to pcre_compile(),
1267 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
1268 provided for specialist use. Most commonly, no special tables are
1269 passed, in which case internal tables that are generated when PCRE is
1270 built are used.
1272 The function pcre_fullinfo() is used to find out information about a
1273 compiled pattern. The function pcre_version() returns a pointer to a
1274 string containing the version of PCRE and its date of release.
1276 The function pcre_refcount() maintains a reference count in a data
1277 block containing a compiled pattern. This is provided for the benefit
1278 of object-oriented applications.
1280 The global variables pcre_malloc and pcre_free initially contain the
1281 entry points of the standard malloc() and free() functions, respec-
1282 tively. PCRE calls the memory management functions via these variables,
1283 so a calling program can replace them if it wishes to intercept the
1284 calls. This should be done before calling any PCRE functions.
1286 The global variables pcre_stack_malloc and pcre_stack_free are also
1287 indirections to memory management functions. These special functions
1288 are used only when PCRE is compiled to use the heap for remembering
1289 data, instead of recursive function calls, when running the pcre_exec()
1290 function. See the pcrebuild documentation for details of how to do
1291 this. It is a non-standard way of building PCRE, for use in environ-
1292 ments that have limited stacks. Because of the greater use of memory
1293 management, it runs more slowly. Separate functions are provided so
1294 that special-purpose external code can be used for this case. When
1295 used, these functions are always called in a stack-like manner (last
1296 obtained, first freed), and always for memory blocks of the same size.
1297 There is a discussion about PCRE's stack usage in the pcrestack docu-
1298 mentation.
1300 The global variable pcre_callout initially contains NULL. It can be set
1301 by the caller to a "callout" function, which PCRE will then call at
1302 specified points during a matching operation. Details are given in the
1303 pcrecallout documentation.
1308 PCRE supports five different conventions for indicating line breaks in
1309 strings: a single CR (carriage return) character, a single LF (line-
1310 feed) character, the two-character sequence CRLF, any of the three pre-
1311 ceding, or any Unicode newline sequence. The Unicode newline sequences
1312 are the three just mentioned, plus the single characters VT (vertical
1313 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
1314 separator, U+2028), and PS (paragraph separator, U+2029).
1316 Each of the first three conventions is used by at least one operating
1317 system as its standard newline sequence. When PCRE is built, a default
1318 can be specified. The default default is LF, which is the Unix stan-
1319 dard. When PCRE is run, the default can be overridden, either when a
1320 pattern is compiled, or when it is matched.
1322 At compile time, the newline convention can be specified by the options
1323 argument of pcre_compile(), or it can be specified by special text at
1324 the start of the pattern itself; this overrides any other settings. See
1325 the pcrepattern page for details of the special character sequences.
1327 In the PCRE documentation the word "newline" is used to mean "the char-
1328 acter or pair of characters that indicate a line break". The choice of
1329 newline convention affects the handling of the dot, circumflex, and
1330 dollar metacharacters, the handling of #-comments in /x mode, and, when
1331 CRLF is a recognized line ending sequence, the match position advance-
1332 ment for a non-anchored pattern. There is more detail about this in the
1333 section on pcre_exec() options below.
1335 The choice of newline convention does not affect the interpretation of
1336 the \n or \r escape sequences, nor does it affect what \R matches,
1337 which is controlled in a similar way, but by separate options.
1342 The PCRE functions can be used in multi-threading applications, with
1343 the proviso that the memory management functions pointed to by
1344 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1345 callout function pointed to by pcre_callout, are shared by all threads.
1347 The compiled form of a regular expression is not altered during match-
1348 ing, so the same compiled pattern can safely be used by several threads
1349 at once.
1351 If the just-in-time optimization feature is being used, it needs sepa-
1352 rate memory stack areas for each thread. See the pcrejit documentation
1353 for more details.
1358 The compiled form of a regular expression can be saved and re-used at a
1359 later time, possibly by a different program, and even on a host other
1360 than the one on which it was compiled. Details are given in the
1361 pcreprecompile documentation, which includes a description of the
1362 pcre_pattern_to_host_byte_order() function. However, compiling a regu-
1363 lar expression with one version of PCRE for use with a different ver-
1364 sion is not guaranteed to work and may cause crashes.
1369 int pcre_config(int what, void *where);
1371 The function pcre_config() makes it possible for a PCRE client to dis-
1372 cover which optional features have been compiled into the PCRE library.
1373 The pcrebuild documentation has more details about these optional fea-
1374 tures.
1376 The first argument for pcre_config() is an integer, specifying which
1377 information is required; the second argument is a pointer to a variable
1378 into which the information is placed. The returned value is zero on
1379 success, or the negative error code PCRE_ERROR_BADOPTION if the value
1380 in the first argument is not recognized. The following information is
1381 available:
1385 The output is an integer that is set to one if UTF-8 support is avail-
1386 able; otherwise it is set to zero. If this option is given to the
1387 16-bit version of this function, pcre16_config(), the result is
1392 The output is an integer that is set to one if UTF-16 support is avail-
1393 able; otherwise it is set to zero. This value should normally be given
1394 to the 16-bit version of this function, pcre16_config(). If it is given
1395 to the 8-bit version of this function, the result is PCRE_ERROR_BADOP-
1396 TION.
1400 The output is an integer that is set to one if support for Unicode
1401 character properties is available; otherwise it is set to zero.
1405 The output is an integer that is set to one if support for just-in-time
1406 compiling is available; otherwise it is set to zero.
1410 The output is a pointer to a zero-terminated "const char *" string. If
1411 JIT support is available, the string contains the name of the architec-
1412 ture for which the JIT compiler is configured, for example "x86 32bit
1413 (little endian + unaligned)". If JIT support is not available, the
1414 result is NULL.
1418 The output is an integer whose value specifies the default character
1419 sequence that is recognized as meaning "newline". The four values that
1420 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1421 and -1 for ANY. Though they are derived from ASCII, the same values
1422 are returned in EBCDIC environments. The default should normally corre-
1423 spond to the standard sequence for your operating system.
1427 The output is an integer whose value indicates what character sequences
1428 the \R escape sequence matches by default. A value of 0 means that \R
1429 matches any Unicode line ending sequence; a value of 1 means that \R
1430 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1431 tern is compiled or matched.
1435 The output is an integer that contains the number of bytes used for
1436 internal linkage in compiled regular expressions. For the 8-bit
1437 library, the value can be 2, 3, or 4. For the 16-bit library, the value
1438 is either 2 or 4 and is still a number of bytes. The default value of 2
1439 is sufficient for all but the most massive patterns, since it allows
1440 the compiled pattern to be up to 64K in size. Larger values allow
1441 larger regular expressions to be compiled, at the expense of slower
1442 matching.
1446 The output is an integer that contains the threshold above which the
1447 POSIX interface uses malloc() for output vectors. Further details are
1448 given in the pcreposix documentation.
1452 The output is a long integer that gives the default limit for the num-
1453 ber of internal matching function calls in a pcre_exec() execution.
1454 Further details are given with pcre_exec() below.
1458 The output is a long integer that gives the default limit for the depth
1459 of recursion when calling the internal matching function in a
1460 pcre_exec() execution. Further details are given with pcre_exec()
1461 below.
1465 The output is an integer that is set to one if internal recursion when
1466 running pcre_exec() is implemented by recursive function calls that use
1467 the stack to remember their state. This is the usual way that PCRE is
1468 compiled. The output is zero if PCRE was compiled to use blocks of data
1469 on the heap instead of recursive function calls. In this case,
1470 pcre_stack_malloc and pcre_stack_free are called to manage memory
1471 blocks on the heap, thus avoiding the use of the stack.
1476 pcre *pcre_compile(const char *pattern, int options,
1477 const char **errptr, int *erroffset,
1478 const unsigned char *tableptr);
1480 pcre *pcre_compile2(const char *pattern, int options,
1481 int *errorcodeptr,
1482 const char **errptr, int *erroffset,
1483 const unsigned char *tableptr);
1485 Either of the functions pcre_compile() or pcre_compile2() can be called
1486 to compile a pattern into an internal form. The only difference between
1487 the two interfaces is that pcre_compile2() has an additional argument,
1488 errorcodeptr, via which a numerical error code can be returned. To
1489 avoid too much repetition, we refer just to pcre_compile() below, but
1490 the information applies equally to pcre_compile2().
1492 The pattern is a C string terminated by a binary zero, and is passed in
1493 the pattern argument. A pointer to a single block of memory that is
1494 obtained via pcre_malloc is returned. This contains the compiled code
1495 and related data. The pcre type is defined for the returned block; this
1496 is a typedef for a structure whose contents are not externally defined.
1497 It is up to the caller to free the memory (via pcre_free) when it is no
1498 longer required.
1500 Although the compiled code of a PCRE regex is relocatable, that is, it
1501 does not depend on memory location, the complete pcre data block is not
1502 fully relocatable, because it may contain a copy of the tableptr argu-
1503 ment, which is an address (see below).
1505 The options argument contains various bit settings that affect the com-
1506 pilation. It should be zero if no options are required. The available
1507 options are described below. Some of them (in particular, those that
1508 are compatible with Perl, but some others as well) can also be set and
1509 unset from within the pattern (see the detailed description in the
1510 pcrepattern documentation). For those options that can be different in
1511 different parts of the pattern, the contents of the options argument
1512 specifies their settings at the start of compilation and execution. The
1514 PCRE_NO_START_OPT options can be set at the time of matching as well as
1515 at compile time.
1517 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1518 if compilation of a pattern fails, pcre_compile() returns NULL, and
1519 sets the variable pointed to by errptr to point to a textual error mes-
1520 sage. This is a static string that is part of the library. You must not
1521 try to free it. Normally, the offset from the start of the pattern to
1522 the byte that was being processed when the error was discovered is
1523 placed in the variable pointed to by erroffset, which must not be NULL
1524 (if it is, an immediate error is given). However, for an invalid UTF-8
1525 string, the offset is that of the first byte of the failing character.
1527 Some errors are not detected until the whole pattern has been scanned;
1528 in these cases, the offset passed back is the length of the pattern.
1529 Note that the offset is in bytes, not characters, even in UTF-8 mode.
1530 It may sometimes point into the middle of a UTF-8 character.
1532 If pcre_compile2() is used instead of pcre_compile(), and the error-
1533 codeptr argument is not NULL, a non-zero error code number is returned
1534 via this argument in the event of an error. This is in addition to the
1535 textual error message. Error codes and messages are listed below.
1537 If the final argument, tableptr, is NULL, PCRE uses a default set of
1538 character tables that are built when PCRE is compiled, using the
1539 default C locale. Otherwise, tableptr must be an address that is the
1540 result of a call to pcre_maketables(). This value is stored with the
1541 compiled pattern, and used again by pcre_exec(), unless another table
1542 pointer is passed to it. For more discussion, see the section on locale
1543 support below.
1545 This code fragment shows a typical straightforward call to pcre_com-
1546 pile():
1548 pcre *re;
1549 const char *error;
1550 int erroffset;
1551 re = pcre_compile(
1552 "^A.*Z", /* the pattern */
1553 0, /* default options */
1554 &error, /* for error message */
1555 &erroffset, /* for error offset */
1556 NULL); /* use default character tables */
1558 The following names for option bits are defined in the pcre.h header
1559 file:
1563 If this bit is set, the pattern is forced to be "anchored", that is, it
1564 is constrained to match only at the first matching point in the string
1565 that is being searched (the "subject string"). This effect can also be
1566 achieved by appropriate constructs in the pattern itself, which is the
1567 only way to do it in Perl.
1571 If this bit is set, pcre_compile() automatically inserts callout items,
1572 all with number 255, before each pattern item. For discussion of the
1573 callout facility, see the pcrecallout documentation.
1578 These options (which are mutually exclusive) control what the \R escape
1579 sequence matches. The choice is either to match only CR, LF, or CRLF,
1580 or to match any Unicode newline sequence. The default is specified when
1581 PCRE is built. It can be overridden from within the pattern, or by set-
1582 ting an option when a compiled pattern is matched.
1586 If this bit is set, letters in the pattern match both upper and lower
1587 case letters. It is equivalent to Perl's /i option, and it can be
1588 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1589 always understands the concept of case for characters whose values are
1590 less than 128, so caseless matching is always possible. For characters
1591 with higher values, the concept of case is supported if PCRE is com-
1592 piled with Unicode property support, but not otherwise. If you want to
1593 use caseless matching for characters 128 and above, you must ensure
1594 that PCRE is compiled with Unicode property support as well as with
1595 UTF-8 support.
1599 If this bit is set, a dollar metacharacter in the pattern matches only
1600 at the end of the subject string. Without this option, a dollar also
1601 matches immediately before a newline at the end of the string (but not
1602 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1603 if PCRE_MULTILINE is set. There is no equivalent to this option in
1604 Perl, and no way to set it within a pattern.
1608 If this bit is set, a dot metacharacter in the pattern matches a char-
1609 acter of any value, including one that indicates a newline. However, it
1610 only ever matches one character, even if newlines are coded as CRLF.
1611 Without this option, a dot does not match when the current position is
1612 at a newline. This option is equivalent to Perl's /s option, and it can
1613 be changed within a pattern by a (?s) option setting. A negative class
1614 such as [^a] always matches newline characters, independent of the set-
1615 ting of this option.
1619 If this bit is set, names used to identify capturing subpatterns need
1620 not be unique. This can be helpful for certain types of pattern when it
1621 is known that only one instance of the named subpattern can ever be
1622 matched. There are more details of named subpatterns below; see also
1623 the pcrepattern documentation.
1627 If this bit is set, whitespace data characters in the pattern are
1628 totally ignored except when escaped or inside a character class. White-
1629 space does not include the VT character (code 11). In addition, charac-
1630 ters between an unescaped # outside a character class and the next new-
1631 line, inclusive, are also ignored. This is equivalent to Perl's /x
1632 option, and it can be changed within a pattern by a (?x) option set-
1633 ting.
1635 Which characters are interpreted as newlines is controlled by the
1636 options passed to pcre_compile() or by a special sequence at the start
1637 of the pattern, as described in the section entitled "Newline conven-
1638 tions" in the pcrepattern documentation. Note that the end of this type
1639 of comment is a literal newline sequence in the pattern; escape
1640 sequences that happen to represent a newline do not count.
1642 This option makes it possible to include comments inside complicated
1643 patterns. Note, however, that this applies only to data characters.
1644 Whitespace characters may never appear within special character
1645 sequences in a pattern, for example within the sequence (?( that intro-
1646 duces a conditional subpattern.
1650 This option was invented in order to turn on additional functionality
1651 of PCRE that is incompatible with Perl, but it is currently of very
1652 little use. When set, any backslash in a pattern that is followed by a
1653 letter that has no special meaning causes an error, thus reserving
1654 these combinations for future expansion. By default, as in Perl, a
1655 backslash followed by a letter with no special meaning is treated as a
1656 literal. (Perl can, however, be persuaded to give an error for this, by
1657 running it with the -w option.) There are at present no other features
1658 controlled by this option. It can also be set by a (?X) option setting
1659 within a pattern.
1663 If this option is set, an unanchored pattern is required to match
1664 before or at the first newline in the subject string, though the
1665 matched text may continue over the newline.
1669 If this option is set, PCRE's behaviour is changed in some ways so that
1670 it is compatible with JavaScript rather than Perl. The changes are as
1671 follows:
1673 (1) A lone closing square bracket in a pattern causes a compile-time
1674 error, because this is illegal in JavaScript (by default it is treated
1675 as a data character). Thus, the pattern AB]CD becomes illegal when this
1676 option is set.
1678 (2) At run time, a back reference to an unset subpattern group matches
1679 an empty string (by default this causes the current matching alterna-
1680 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1681 set (assuming it can find an "a" in the subject), whereas it fails by
1682 default, for Perl compatibility.
1684 (3) \U matches an upper case "U" character; by default \U causes a com-
1685 pile time error (Perl uses \U to upper case subsequent characters).
1687 (4) \u matches a lower case "u" character unless it is followed by four
1688 hexadecimal digits, in which case the hexadecimal number defines the
1689 code point to match. By default, \u causes a compile time error (Perl
1690 uses it to upper case the following character).
1692 (5) \x matches a lower case "x" character unless it is followed by two
1693 hexadecimal digits, in which case the hexadecimal number defines the
1694 code point to match. By default, as in Perl, a hexadecimal number is
1695 always expected after \x, but it may have zero, one, or two digits (so,
1696 for example, \xz matches a binary zero character followed by z).
1700 By default, PCRE treats the subject string as consisting of a single
1701 line of characters (even if it actually contains newlines). The "start
1702 of line" metacharacter (^) matches only at the start of the string,
1703 while the "end of line" metacharacter ($) matches only at the end of
1704 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1705 is set). This is the same as Perl.
1707 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1708 constructs match immediately following or immediately before internal
1709 newlines in the subject string, respectively, as well as at the very
1710 start and end. This is equivalent to Perl's /m option, and it can be
1711 changed within a pattern by a (?m) option setting. If there are no new-
1712 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1713 setting PCRE_MULTILINE has no effect.
1721 These options override the default newline definition that was chosen
1722 when PCRE was built. Setting the first or the second specifies that a
1723 newline is indicated by a single character (CR or LF, respectively).
1724 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1725 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1726 that any of the three preceding sequences should be recognized. Setting
1727 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1728 recognized. The Unicode newline sequences are the three just mentioned,
1729 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1730 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1731 (paragraph separator, U+2029). For the 8-bit library, the last two are
1732 recognized only in UTF-8 mode.
1734 The newline setting in the options word uses three bits that are
1735 treated as a number, giving eight possibilities. Currently only six are
1736 used (default plus the five values above). This means that if you set
1737 more than one newline option, the combination may or may not be sensi-
1738 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1739 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1740 cause an error.
1742 The only time that a line break in a pattern is specially recognized
1743 when compiling is when PCRE_EXTENDED is set. CR and LF are whitespace
1744 characters, and so are ignored in this mode. Also, an unescaped # out-
1745 side a character class indicates a comment that lasts until after the
1746 next line break sequence. In other circumstances, line break sequences
1747 in patterns are treated as literal data.
1749 The newline option that is set at compile time becomes the default that
1750 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1754 If this option is set, it disables the use of numbered capturing paren-
1755 theses in the pattern. Any opening parenthesis that is not followed by
1756 ? behaves as if it were followed by ?: but named parentheses can still
1757 be used for capturing (and they acquire numbers in the usual way).
1758 There is no equivalent of this option in Perl.
1762 This is an option that acts at matching time; that is, it is really an
1763 option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
1764 time, it is remembered with the compiled pattern and assumed at match-
1765 ing time. For details see the discussion of PCRE_NO_START_OPTIMIZE
1766 below.
1770 This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
1771 \w, and some of the POSIX character classes. By default, only ASCII
1772 characters are recognized, but if PCRE_UCP is set, Unicode properties
1773 are used instead to classify characters. More details are given in the
1774 section on generic character types in the pcrepattern page. If you set
1775 PCRE_UCP, matching one of the items it affects takes much longer. The
1776 option is available only if PCRE has been compiled with Unicode prop-
1777 erty support.
1781 This option inverts the "greediness" of the quantifiers so that they
1782 are not greedy by default, but become greedy if followed by "?". It is
1783 not compatible with Perl. It can also be set by a (?U) option setting
1784 within the pattern.
1786 PCRE_UTF8
1788 This option causes PCRE to regard both the pattern and the subject as
1789 strings of UTF-8 characters instead of single-byte strings. However, it
1790 is available only when PCRE is built to include UTF support. If not,
1791 the use of this option provokes an error. Details of how this option
1792 changes the behaviour of PCRE are given in the pcreunicode page.
1796 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1797 automatically checked. There is a discussion about the validity of
1798 UTF-8 strings in the pcreunicode page. If an invalid UTF-8 sequence is
1799 found, pcre_compile() returns an error. If you already know that your
1800 pattern is valid, and you want to skip this check for performance rea-
1801 sons, you can set the PCRE_NO_UTF8_CHECK option. When it is set, the
1802 effect of passing an invalid UTF-8 string as a pattern is undefined. It
1803 may cause your program to crash. Note that this option can also be
1804 passed to pcre_exec() and pcre_dfa_exec(), to suppress the validity
1805 checking of subject strings.
1810 The following table lists the error codes than may be returned by
1811 pcre_compile2(), along with the error messages that may be returned by
1812 both compiling functions. Note that error messages are always 8-bit
1813 ASCII strings, even in 16-bit mode. As PCRE has developed, some error
1814 codes have fallen out of use. To avoid confusion, they have not been
1815 re-used.
1817 0 no error
1818 1 \ at end of pattern
1819 2 \c at end of pattern
1820 3 unrecognized character follows \
1821 4 numbers out of order in {} quantifier
1822 5 number too big in {} quantifier
1823 6 missing terminating ] for character class
1824 7 invalid escape sequence in character class
1825 8 range out of order in character class
1826 9 nothing to repeat
1827 10 [this code is not in use]
1828 11 internal error: unexpected repeat
1829 12 unrecognized character after (? or (?-
1830 13 POSIX named classes are supported only within a class
1831 14 missing )
1832 15 reference to non-existent subpattern
1833 16 erroffset passed as NULL
1834 17 unknown option bit(s) set
1835 18 missing ) after comment
1836 19 [this code is not in use]
1837 20 regular expression is too large
1838 21 failed to get memory
1839 22 unmatched parentheses
1840 23 internal error: code overflow
1841 24 unrecognized character after (?<
1842 25 lookbehind assertion is not fixed length
1843 26 malformed number or name after (?(
1844 27 conditional group contains more than two branches
1845 28 assertion expected after (?(
1846 29 (?R or (?[+-]digits must be followed by )
1847 30 unknown POSIX class name
1848 31 POSIX collating elements are not supported
1849 32 this version of PCRE is compiled without UTF support
1850 33 [this code is not in use]
1851 34 character value in \x{...} sequence is too large
1852 35 invalid condition (?(0)
1853 36 \C not allowed in lookbehind assertion
1854 37 PCRE does not support \L, \l, \N{name}, \U, or \u
1855 38 number after (?C is > 255
1856 39 closing ) for (?C expected
1857 40 recursive call could loop indefinitely
1858 41 unrecognized character after (?P
1859 42 syntax error in subpattern name (missing terminator)
1860 43 two named subpatterns have the same name
1861 44 invalid UTF-8 string (specifically UTF-8)
1862 45 support for \P, \p, and \X has not been compiled
1863 46 malformed \P or \p sequence
1864 47 unknown property name after \P or \p
1865 48 subpattern name is too long (maximum 32 characters)
1866 49 too many named subpatterns (maximum 10000)
1867 50 [this code is not in use]
1868 51 octal value is greater than \377 in 8-bit non-UTF-8 mode
1869 52 internal error: overran compiling workspace
1870 53 internal error: previously-checked referenced subpattern
1871 not found
1872 54 DEFINE group contains more than one branch
1873 55 repeating a DEFINE group is not allowed
1874 56 inconsistent NEWLINE options
1875 57 \g is not followed by a braced, angle-bracketed, or quoted
1876 name/number or by a plain number
1877 58 a numbered reference must not be zero
1878 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1879 60 (*VERB) not recognized
1880 61 number is too big
1881 62 subpattern name expected
1882 63 digit expected after (?+
1883 64 ] is an invalid data character in JavaScript compatibility mode
1884 65 different names for subpatterns of the same number are
1885 not allowed
1886 66 (*MARK) must have an argument
1887 67 this version of PCRE is not compiled with Unicode property
1888 support
1889 68 \c must be followed by an ASCII character
1890 69 \k is not followed by a braced, angle-bracketed, or quoted name
1891 70 internal error: unknown opcode in find_fixedlength()
1892 71 \N is not supported in a class
1893 72 too many forward references
1894 73 disallowed Unicode code point (>= 0xd800 && <= 0xdfff)
1895 74 invalid UTF-16 string (specifically UTF-16)
1897 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1898 values may be used if the limits were changed when PCRE was built.
1903 pcre_extra *pcre_study(const pcre *code, int options
1904 const char **errptr);
1906 If a compiled pattern is going to be used several times, it is worth
1907 spending more time analyzing it in order to speed up the time taken for
1908 matching. The function pcre_study() takes a pointer to a compiled pat-
1909 tern as its first argument. If studying the pattern produces additional
1910 information that will help speed up matching, pcre_study() returns a
1911 pointer to a pcre_extra block, in which the study_data field points to
1912 the results of the study.
1914 The returned value from pcre_study() can be passed directly to
1915 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1916 tains other fields that can be set by the caller before the block is
1917 passed; these are described below in the section on matching a pattern.
1919 If studying the pattern does not produce any useful information,
1920 pcre_study() returns NULL. In that circumstance, if the calling program
1921 wants to pass any of the other fields to pcre_exec() or
1922 pcre_dfa_exec(), it must set up its own pcre_extra block.
1924 The second argument of pcre_study() contains option bits. There is only
1925 one option: PCRE_STUDY_JIT_COMPILE. If this is set, and the just-in-
1926 time compiler is available, the pattern is further compiled into
1927 machine code that executes much faster than the pcre_exec() matching
1928 function. If the just-in-time compiler is not available, this option is
1929 ignored. All other bits in the options argument must be zero.
1931 JIT compilation is a heavyweight optimization. It can take some time
1932 for patterns to be analyzed, and for one-off matches and simple pat-
1933 terns the benefit of faster execution might be offset by a much slower
1934 study time. Not all patterns can be optimized by the JIT compiler. For
1935 those that cannot be handled, matching automatically falls back to the
1936 pcre_exec() interpreter. For more details, see the pcrejit documenta-
1937 tion.
1939 The third argument for pcre_study() is a pointer for an error message.
1940 If studying succeeds (even if no data is returned), the variable it
1941 points to is set to NULL. Otherwise it is set to point to a textual
1942 error message. This is a static string that is part of the library. You
1943 must not try to free it. You should test the error pointer for NULL
1944 after calling pcre_study(), to be sure that it has run successfully.
1946 When you are finished with a pattern, you can free the memory used for
1947 the study data by calling pcre_free_study(). This function was added to
1948 the API for release 8.20. For earlier versions, the memory could be
1949 freed with pcre_free(), just like the pattern itself. This will still
1950 work in cases where PCRE_STUDY_JIT_COMPILE is not used, but it is
1951 advisable to change to the new function when convenient.
1953 This is a typical way in which pcre_study() is used (except that in a
1954 real application there should be tests for errors):
1956 int rc;
1957 pcre *re;
1958 pcre_extra *sd;
1959 re = pcre_compile("pattern", 0, &error, &erroroffset, NULL);
1960 sd = pcre_study(
1961 re, /* result of pcre_compile() */
1962 0, /* no options */
1963 &error); /* set to NULL or points to a message */
1964 rc = pcre_exec( /* see below for details of pcre_exec() options */
1965 re, sd, "subject", 7, 0, 0, ovector, 30);
1966 ...
1967 pcre_free_study(sd);
1968 pcre_free(re);
1970 Studying a pattern does two things: first, a lower bound for the length
1971 of subject string that is needed to match the pattern is computed. This
1972 does not mean that there are any strings of that length that match, but
1973 it does guarantee that no shorter strings match. The value is used by
1974 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1975 match strings that are shorter than the lower bound. You can find out
1976 the value in a calling program via the pcre_fullinfo() function.
1978 Studying a pattern is also useful for non-anchored patterns that do not
1979 have a single fixed starting character. A bitmap of possible starting
1980 bytes is created. This speeds up finding a position in the subject at
1981 which to start matching. (In 16-bit mode, the bitmap is used for 16-bit
1982 values less than 256.)
1984 These two optimizations apply to both pcre_exec() and pcre_dfa_exec().
1985 However, they are not used by pcre_exec() if pcre_study() is called
1986 with the PCRE_STUDY_JIT_COMPILE option, and just-in-time compiling is
1987 successful. The optimizations can be disabled by setting the
1988 PCRE_NO_START_OPTIMIZE option when calling pcre_exec() or
1989 pcre_dfa_exec(). You might want to do this if your pattern contains
1990 callouts or (*MARK) (which cannot be handled by the JIT compiler), and
1991 you want to make use of these facilities in cases where matching fails.
1992 See the discussion of PCRE_NO_START_OPTIMIZE below.
1997 PCRE handles caseless matching, and determines whether characters are
1998 letters, digits, or whatever, by reference to a set of tables, indexed
1999 by character value. When running in UTF-8 mode, this applies only to
2000 characters with codes less than 128. By default, higher-valued codes
2001 never match escapes such as \w or \d, but they can be tested with \p if
2002 PCRE is built with Unicode character property support. Alternatively,
2003 the PCRE_UCP option can be set at compile time; this causes \w and
2004 friends to use Unicode property support instead of built-in tables. The
2005 use of locales with Unicode is discouraged. If you are handling charac-
2006 ters with codes greater than 128, you should either use UTF-8 and Uni-
2007 code, or use locales, but not try to mix the two.
2009 PCRE contains an internal set of tables that are used when the final
2010 argument of pcre_compile() is NULL. These are sufficient for many
2011 applications. Normally, the internal tables recognize only ASCII char-
2012 acters. However, when PCRE is built, it is possible to cause the inter-
2013 nal tables to be rebuilt in the default "C" locale of the local system,
2014 which may cause them to be different.
2016 The internal tables can always be overridden by tables supplied by the
2017 application that calls PCRE. These may be created in a different locale
2018 from the default. As more and more applications change to using Uni-
2019 code, the need for this locale support is expected to die away.
2021 External tables are built by calling the pcre_maketables() function,
2022 which has no arguments, in the relevant locale. The result can then be
2023 passed to pcre_compile() or pcre_exec() as often as necessary. For
2024 example, to build and use tables that are appropriate for the French
2025 locale (where accented characters with values greater than 128 are
2026 treated as letters), the following code could be used:
2028 setlocale(LC_CTYPE, "fr_FR");
2029 tables = pcre_maketables();
2030 re = pcre_compile(..., tables);
2032 The locale name "fr_FR" is used on Linux and other Unix-like systems;
2033 if you are using Windows, the name for the French locale is "french".
2035 When pcre_maketables() runs, the tables are built in memory that is
2036 obtained via pcre_malloc. It is the caller's responsibility to ensure
2037 that the memory containing the tables remains available for as long as
2038 it is needed.
2040 The pointer that is passed to pcre_compile() is saved with the compiled
2041 pattern, and the same tables are used via this pointer by pcre_study()
2042 and normally also by pcre_exec(). Thus, by default, for any single pat-
2043 tern, compilation, studying and matching all happen in the same locale,
2044 but different patterns can be compiled in different locales.
2046 It is possible to pass a table pointer or NULL (indicating the use of
2047 the internal tables) to pcre_exec(). Although not intended for this
2048 purpose, this facility could be used to match a pattern in a different
2049 locale from the one in which it was compiled. Passing table pointers at
2050 run time is discussed below in the section on matching a pattern.
2055 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
2056 int what, void *where);
2058 The pcre_fullinfo() function returns information about a compiled pat-
2059 tern. It replaces the pcre_info() function, which was removed from the
2060 library at version 8.30, after more than 10 years of obsolescence.
2062 The first argument for pcre_fullinfo() is a pointer to the compiled
2063 pattern. The second argument is the result of pcre_study(), or NULL if
2064 the pattern was not studied. The third argument specifies which piece
2065 of information is required, and the fourth argument is a pointer to a
2066 variable to receive the data. The yield of the function is zero for
2067 success, or one of the following negative numbers:
2069 PCRE_ERROR_NULL the argument code was NULL
2070 the argument where was NULL
2071 PCRE_ERROR_BADMAGIC the "magic number" was not found
2072 PCRE_ERROR_BADENDIANNESS the pattern was compiled with different
2073 endianness
2074 PCRE_ERROR_BADOPTION the value of what was invalid
2076 The "magic number" is placed at the start of each compiled pattern as
2077 an simple check against passing an arbitrary memory pointer. The endi-
2078 anness error can occur if a compiled pattern is saved and reloaded on a
2079 different host. Here is a typical call of pcre_fullinfo(), to obtain
2080 the length of the compiled pattern:
2082 int rc;
2083 size_t length;
2084 rc = pcre_fullinfo(
2085 re, /* result of pcre_compile() */
2086 sd, /* result of pcre_study(), or NULL */
2087 PCRE_INFO_SIZE, /* what is required */
2088 &length); /* where to put the data */
2090 The possible values for the third argument are defined in pcre.h, and
2091 are as follows:
2095 Return the number of the highest back reference in the pattern. The
2096 fourth argument should point to an int variable. Zero is returned if
2097 there are no back references.
2101 Return the number of capturing subpatterns in the pattern. The fourth
2102 argument should point to an int variable.
2106 Return a pointer to the internal default character tables within PCRE.
2107 The fourth argument should point to an unsigned char * variable. This
2108 information call is provided for internal use by the pcre_study() func-
2109 tion. External callers can cause PCRE to use its internal tables by
2110 passing a NULL table pointer.
2114 Return information about the first data unit of any matched string, for
2115 a non-anchored pattern. (The name of this option refers to the 8-bit
2116 library, where data units are bytes.) The fourth argument should point
2117 to an int variable.
2119 If there is a fixed first value, for example, the letter "c" from a
2120 pattern such as (cat|cow|coyote), its value is returned. In the 8-bit
2121 library, the value is always less than 256; in the 16-bit library the
2122 value can be up to 0xffff.
2124 If there is no fixed first value, and if either
2126 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
2127 branch starts with "^", or
2129 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
2130 set (if it were set, the pattern would be anchored),
2132 -1 is returned, indicating that the pattern matches only at the start
2133 of a subject string or after any newline within the string. Otherwise
2134 -2 is returned. For anchored patterns, -2 is returned.
2138 If the pattern was studied, and this resulted in the construction of a
2139 256-bit table indicating a fixed set of values for the first data unit
2140 in any matching string, a pointer to the table is returned. Otherwise
2141 NULL is returned. The fourth argument should point to an unsigned char
2142 * variable.
2146 Return 1 if the pattern contains any explicit matches for CR or LF
2147 characters, otherwise 0. The fourth argument should point to an int
2148 variable. An explicit match is either a literal CR or LF character, or
2149 \r or \n.
2153 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
2154 otherwise 0. The fourth argument should point to an int variable. (?J)
2155 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
2159 Return 1 if the pattern was studied with the PCRE_STUDY_JIT_COMPILE
2160 option, and just-in-time compiling was successful. The fourth argument
2161 should point to an int variable. A return value of 0 means that JIT
2162 support is not available in this version of PCRE, or that the pattern
2163 was not studied with the PCRE_STUDY_JIT_COMPILE option, or that the JIT
2164 compiler could not handle this particular pattern. See the pcrejit doc-
2165 umentation for details of what can and cannot be handled.
2169 If the pattern was successfully studied with the PCRE_STUDY_JIT_COMPILE
2170 option, return the size of the JIT compiled code, otherwise return
2171 zero. The fourth argument should point to a size_t variable.
2175 Return the value of the rightmost literal data unit that must exist in
2176 any matched string, other than at its start, if such a value has been
2177 recorded. The fourth argument should point to an int variable. If there
2178 is no such value, -1 is returned. For anchored patterns, a last literal
2179 value is recorded only if it follows something of variable length. For
2180 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
2181 /^a\dz\d/ the returned value is -1.
2185 If the pattern was studied and a minimum length for matching subject
2186 strings was computed, its value is returned. Otherwise the returned
2187 value is -1. The value is a number of characters, which in UTF-8 mode
2188 may be different from the number of bytes. The fourth argument should
2189 point to an int variable. A non-negative value is a lower bound to the
2190 length of any matching string. There may not be any strings of that
2191 length that do actually match, but every string that does match is at
2192 least that long.
2198 PCRE supports the use of named as well as numbered capturing parenthe-
2199 ses. The names are just an additional way of identifying the parenthe-
2200 ses, which still acquire numbers. Several convenience functions such as
2201 pcre_get_named_substring() are provided for extracting captured sub-
2202 strings by name. It is also possible to extract the data directly, by
2203 first converting the name to a number in order to access the correct
2204 pointers in the output vector (described with pcre_exec() below). To do
2205 the conversion, you need to use the name-to-number map, which is
2206 described by these three values.
2208 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
2209 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
2210 of each entry; both of these return an int value. The entry size
2211 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
2212 a pointer to the first entry of the table. This is a pointer to char in
2213 the 8-bit library, where the first two bytes of each entry are the num-
2214 ber of the capturing parenthesis, most significant byte first. In the
2215 16-bit library, the pointer points to 16-bit data units, the first of
2216 which contains the parenthesis number. The rest of the entry is the
2217 corresponding name, zero terminated.
2219 The names are in alphabetical order. Duplicate names may appear if (?|
2220 is used to create multiple groups with the same number, as described in
2221 the section on duplicate subpattern numbers in the pcrepattern page.
2222 Duplicate names for subpatterns with different numbers are permitted
2223 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
2224 appear in the table in the order in which they were found in the pat-
2225 tern. In the absence of (?| this is the order of increasing number;
2226 when (?| is used this is not necessarily the case because later subpat-
2227 terns may have lower numbers.
2229 As a simple example of the name/number table, consider the following
2230 pattern after compilation by the 8-bit library (assume PCRE_EXTENDED is
2231 set, so white space - including newlines - is ignored):
2233 (?<date> (?<year>(\d\d)?\d\d) -
2234 (?<month>\d\d) - (?<day>\d\d) )
2236 There are four named subpatterns, so the table has four entries, and
2237 each entry in the table is eight bytes long. The table is as follows,
2238 with non-printing bytes shows in hexadecimal, and undefined bytes shown
2239 as ??:
2241 00 01 d a t e 00 ??
2242 00 05 d a y 00 ?? ??
2243 00 04 m o n t h 00
2244 00 02 y e a r 00 ??
2246 When writing code to extract data from named subpatterns using the
2247 name-to-number map, remember that the length of the entries is likely
2248 to be different for each compiled pattern.
2252 Return 1 if the pattern can be used for partial matching with
2253 pcre_exec(), otherwise 0. The fourth argument should point to an int
2254 variable. From release 8.00, this always returns 1, because the
2255 restrictions that previously applied to partial matching have been
2256 lifted. The pcrepartial documentation gives details of partial match-
2257 ing.
2261 Return a copy of the options with which the pattern was compiled. The
2262 fourth argument should point to an unsigned long int variable. These
2263 option bits are those specified in the call to pcre_compile(), modified
2264 by any top-level option settings at the start of the pattern itself. In
2265 other words, they are the options that will be in force when matching
2266 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
2267 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
2270 A pattern is automatically anchored by PCRE if all of its top-level
2271 alternatives begin with one of the following:
2273 ^ unless PCRE_MULTILINE is set
2274 \A always
2275 \G always
2276 .* if PCRE_DOTALL is set and there are no back
2277 references to the subpattern in which .* appears
2279 For such patterns, the PCRE_ANCHORED bit is set in the options returned
2280 by pcre_fullinfo().
2284 Return the size of the compiled pattern in bytes (for both libraries).
2285 The fourth argument should point to a size_t variable. This value does
2286 not include the size of the pcre structure that is returned by
2287 pcre_compile(). The value that is passed as the argument to pcre_mal-
2288 loc() when pcre_compile() is getting memory in which to place the com-
2289 piled data is the value returned by this option plus the size of the
2290 pcre structure. Studying a compiled pattern, with or without JIT, does
2291 not alter the value returned by this option.
2295 Return the size in bytes of the data block pointed to by the study_data
2296 field in a pcre_extra block. If pcre_extra is NULL, or there is no
2297 study data, zero is returned. The fourth argument should point to a
2298 size_t variable. The study_data field is set by pcre_study() to record
2299 information that will speed up matching (see the section entitled
2300 "Studying a pattern" above). The format of the study_data block is pri-
2301 vate, but its length is made available via this option so that it can
2302 be saved and restored (see the pcreprecompile documentation for
2303 details).
2308 int pcre_refcount(pcre *code, int adjust);
2310 The pcre_refcount() function is used to maintain a reference count in
2311 the data block that contains a compiled pattern. It is provided for the
2312 benefit of applications that operate in an object-oriented manner,
2313 where different parts of the application may be using the same compiled
2314 pattern, but you want to free the block when they are all done.
2316 When a pattern is compiled, the reference count field is initialized to
2317 zero. It is changed only by calling this function, whose action is to
2318 add the adjust value (which may be positive or negative) to it. The
2319 yield of the function is the new value. However, the value of the count
2320 is constrained to lie between 0 and 65535, inclusive. If the new value
2321 is outside these limits, it is forced to the appropriate limit value.
2323 Except when it is zero, the reference count is not correctly preserved
2324 if a pattern is compiled on one host and then transferred to a host
2325 whose byte-order is different. (This seems a highly unlikely scenario.)
2330 int pcre_exec(const pcre *code, const pcre_extra *extra,
2331 const char *subject, int length, int startoffset,
2332 int options, int *ovector, int ovecsize);
2334 The function pcre_exec() is called to match a subject string against a
2335 compiled pattern, which is passed in the code argument. If the pattern
2336 was studied, the result of the study should be passed in the extra
2337 argument. You can call pcre_exec() with the same code and extra argu-
2338 ments as many times as you like, in order to match different subject
2339 strings with the same pattern.
2341 This function is the main matching facility of the library, and it
2342 operates in a Perl-like manner. For specialist use there is also an
2343 alternative matching function, which is described below in the section
2344 about the pcre_dfa_exec() function.
2346 In most applications, the pattern will have been compiled (and option-
2347 ally studied) in the same process that calls pcre_exec(). However, it
2348 is possible to save compiled patterns and study data, and then use them
2349 later in different processes, possibly even on different hosts. For a
2350 discussion about this, see the pcreprecompile documentation.
2352 Here is an example of a simple call to pcre_exec():
2354 int rc;
2355 int ovector[30];
2356 rc = pcre_exec(
2357 re, /* result of pcre_compile() */
2358 NULL, /* we didn't study the pattern */
2359 "some string", /* the subject string */
2360 11, /* the length of the subject string */
2361 0, /* start at offset 0 in the subject */
2362 0, /* default options */
2363 ovector, /* vector of integers for substring information */
2364 30); /* number of elements (NOT size in bytes) */
2366 Extra data for pcre_exec()
2368 If the extra argument is not NULL, it must point to a pcre_extra data
2369 block. The pcre_study() function returns such a block (when it doesn't
2370 return NULL), but you can also create one for yourself, and pass addi-
2371 tional information in it. The pcre_extra block contains the following
2372 fields (not necessarily in this order):
2374 unsigned long int flags;
2375 void *study_data;
2376 void *executable_jit;
2377 unsigned long int match_limit;
2378 unsigned long int match_limit_recursion;
2379 void *callout_data;
2380 const unsigned char *tables;
2381 unsigned char **mark;
2383 In the 16-bit version of this structure, the mark field has type
2384 "PCRE_UCHAR16 **".
2386 The flags field is a bitmap that specifies which of the other fields
2387 are set. The flag bits are:
2397 Other flag bits should be set to zero. The study_data field and some-
2398 times the executable_jit field are set in the pcre_extra block that is
2399 returned by pcre_study(), together with the appropriate flag bits. You
2400 should not set these yourself, but you may add to the block by setting
2401 the other fields and their corresponding flag bits.
2403 The match_limit field provides a means of preventing PCRE from using up
2404 a vast amount of resources when running patterns that are not going to
2405 match, but which have a very large number of possibilities in their
2406 search trees. The classic example is a pattern that uses nested unlim-
2407 ited repeats.
2409 Internally, pcre_exec() uses a function called match(), which it calls
2410 repeatedly (sometimes recursively). The limit set by match_limit is
2411 imposed on the number of times this function is called during a match,
2412 which has the effect of limiting the amount of backtracking that can
2413 take place. For patterns that are not anchored, the count restarts from
2414 zero for each position in the subject string.
2416 When pcre_exec() is called with a pattern that was successfully studied
2417 with the PCRE_STUDY_JIT_COMPILE option, the way that the matching is
2418 executed is entirely different. However, there is still the possibility
2419 of runaway matching that goes on for a very long time, and so the
2420 match_limit value is also used in this case (but in a different way) to
2421 limit how long the matching can continue.
2423 The default value for the limit can be set when PCRE is built; the
2424 default default is 10 million, which handles all but the most extreme
2425 cases. You can override the default by suppling pcre_exec() with a
2426 pcre_extra block in which match_limit is set, and
2427 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
2428 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2430 The match_limit_recursion field is similar to match_limit, but instead
2431 of limiting the total number of times that match() is called, it limits
2432 the depth of recursion. The recursion depth is a smaller number than
2433 the total number of calls, because not all calls to match() are recur-
2434 sive. This limit is of use only if it is set smaller than match_limit.
2436 Limiting the recursion depth limits the amount of machine stack that
2437 can be used, or, when PCRE has been compiled to use memory on the heap
2438 instead of the stack, the amount of heap memory that can be used. This
2439 limit is not relevant, and is ignored, if the pattern was successfully
2440 studied with PCRE_STUDY_JIT_COMPILE.
2442 The default value for match_limit_recursion can be set when PCRE is
2443 built; the default default is the same value as the default for
2444 match_limit. You can override the default by suppling pcre_exec() with
2445 a pcre_extra block in which match_limit_recursion is set, and
2446 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
2447 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
2449 The callout_data field is used in conjunction with the "callout" fea-
2450 ture, and is described in the pcrecallout documentation.
2452 The tables field is used to pass a character tables pointer to
2453 pcre_exec(); this overrides the value that is stored with the compiled
2454 pattern. A non-NULL value is stored with the compiled pattern only if
2455 custom tables were supplied to pcre_compile() via its tableptr argu-
2456 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
2457 PCRE's internal tables to be used. This facility is helpful when re-
2458 using patterns that have been saved after compiling with an external
2459 set of tables, because the external tables might be at a different
2460 address when pcre_exec() is called. See the pcreprecompile documenta-
2461 tion for a discussion of saving compiled patterns for later use.
2463 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
2464 set to point to a suitable variable. If the pattern contains any back-
2465 tracking control verbs such as (*MARK:NAME), and the execution ends up
2466 with a name to pass back, a pointer to the name string (zero termi-
2467 nated) is placed in the variable pointed to by the mark field. The
2468 names are within the compiled pattern; if you wish to retain such a
2469 name you must copy it before freeing the memory of a compiled pattern.
2470 If there is no name to pass back, the variable pointed to by the mark
2471 field is set to NULL. For details of the backtracking control verbs,
2472 see the section entitled "Backtracking control" in the pcrepattern doc-
2473 umentation.
2475 Option bits for pcre_exec()
2477 The unused bits of the options argument for pcre_exec() must be zero.
2478 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
2483 If the pattern was successfully studied with the PCRE_STUDY_JIT_COMPILE
2484 option, the only supported options for JIT execution are
2486 PCRE_NOTEMPTY_ATSTART. Note in particular that partial matching is not
2487 supported. If an unsupported option is used, JIT execution is disabled
2488 and the normal interpretive code in pcre_exec() is run.
2492 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
2493 matching position. If a pattern was compiled with PCRE_ANCHORED, or
2494 turned out to be anchored by virtue of its contents, it cannot be made
2495 unachored at matching time.
2500 These options (which are mutually exclusive) control what the \R escape
2501 sequence matches. The choice is either to match only CR, LF, or CRLF,
2502 or to match any Unicode newline sequence. These options override the
2503 choice that was made or defaulted when the pattern was compiled.
2511 These options override the newline definition that was chosen or
2512 defaulted when the pattern was compiled. For details, see the descrip-
2513 tion of pcre_compile() above. During matching, the newline choice
2514 affects the behaviour of the dot, circumflex, and dollar metacharac-
2515 ters. It may also alter the way the match position is advanced after a
2516 match failure for an unanchored pattern.
2519 set, and a match attempt for an unanchored pattern fails when the cur-
2520 rent position is at a CRLF sequence, and the pattern contains no
2521 explicit matches for CR or LF characters, the match position is
2522 advanced by two characters instead of one, in other words, to after the
2523 CRLF.
2525 The above rule is a compromise that makes the most common cases work as
2526 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
2527 option is not set), it does not match the string "\r\nA" because, after
2528 failing at the start, it skips both the CR and the LF before retrying.
2529 However, the pattern [\r\n]A does match that string, because it con-
2530 tains an explicit CR or LF reference, and so advances only by one char-
2531 acter after the first failure.
2533 An explicit match for CR of LF is either a literal appearance of one of
2534 those characters, or one of the \r or \n escape sequences. Implicit
2535 matches such as [^X] do not count, nor does \s (which includes CR and
2536 LF in the characters that it matches).
2538 Notwithstanding the above, anomalous effects may still occur when CRLF
2539 is a valid newline sequence and explicit \r or \n escapes appear in the
2540 pattern.
2544 This option specifies that first character of the subject string is not
2545 the beginning of a line, so the circumflex metacharacter should not
2546 match before it. Setting this without PCRE_MULTILINE (at compile time)
2547 causes circumflex never to match. This option affects only the behav-
2548 iour of the circumflex metacharacter. It does not affect \A.
2552 This option specifies that the end of the subject string is not the end
2553 of a line, so the dollar metacharacter should not match it nor (except
2554 in multiline mode) a newline immediately before it. Setting this with-
2555 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2556 option affects only the behaviour of the dollar metacharacter. It does
2557 not affect \Z or \z.
2561 An empty string is not considered to be a valid match if this option is
2562 set. If there are alternatives in the pattern, they are tried. If all
2563 the alternatives match the empty string, the entire match fails. For
2564 example, if the pattern
2566 a?b?
2568 is applied to a string not beginning with "a" or "b", it matches an
2569 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2570 match is not valid, so PCRE searches further into the string for occur-
2571 rences of "a" or "b".
2575 This is like PCRE_NOTEMPTY, except that an empty string match that is
2576 not at the start of the subject is permitted. If the pattern is
2577 anchored, such a match can occur only if the pattern contains \K.
2579 Perl has no direct equivalent of PCRE_NOTEMPTY or
2580 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2581 match of the empty string within its split() function, and when using
2582 the /g modifier. It is possible to emulate Perl's behaviour after
2583 matching a null string by first trying the match again at the same off-
2584 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2585 fails, by advancing the starting offset (see below) and trying an ordi-
2586 nary match again. There is some code that demonstrates how to do this
2587 in the pcredemo sample program. In the most general case, you have to
2588 check to see if the newline convention recognizes CRLF as a newline,
2589 and if so, and the current character is CR followed by LF, advance the
2590 starting offset by two characters instead of one.
2594 There are a number of optimizations that pcre_exec() uses at the start
2595 of a match, in order to speed up the process. For example, if it is
2596 known that an unanchored match must start with a specific character, it
2597 searches the subject for that character, and fails immediately if it
2598 cannot find it, without actually running the main matching function.
2599 This means that a special item such as (*COMMIT) at the start of a pat-
2600 tern is not considered until after a suitable starting point for the
2601 match has been found. When callouts or (*MARK) items are in use, these
2602 "start-up" optimizations can cause them to be skipped if the pattern is
2603 never actually used. The start-up optimizations are in effect a pre-
2604 scan of the subject that takes place before the pattern is run.
2606 The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
2607 possibly causing performance to suffer, but ensuring that in cases
2608 where the result is "no match", the callouts do occur, and that items
2609 such as (*COMMIT) and (*MARK) are considered at every possible starting
2610 position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
2611 compile time, it cannot be unset at matching time.
2613 Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
2614 operation. Consider the pattern
2618 When this is compiled, PCRE records the fact that a match must start
2619 with the character "A". Suppose the subject string is "DEFABC". The
2620 start-up optimization scans along the subject, finds "A" and runs the
2621 first match attempt from there. The (*COMMIT) item means that the pat-
2622 tern must match the current starting position, which in this case, it
2623 does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
2624 set, the initial scan along the subject string does not happen. The
2625 first match attempt is run starting from "D" and when this fails,
2626 (*COMMIT) prevents any further matches being tried, so the overall
2627 result is "no match". If the pattern is studied, more start-up opti-
2628 mizations may be used. For example, a minimum length for the subject
2629 may be recorded. Consider the pattern
2631 (*MARK:A)(X|Y)
2633 The minimum length for a match is one character. If the subject is
2634 "ABC", there will be attempts to match "ABC", "BC", "C", and then
2635 finally an empty string. If the pattern is studied, the final attempt
2636 does not take place, because PCRE knows that the subject is too short,
2637 and so the (*MARK) is never encountered. In this case, studying the
2638 pattern does not affect the overall match result, which is still "no
2639 match", but it does affect the auxiliary information that is returned.
2643 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2644 UTF-8 string is automatically checked when pcre_exec() is subsequently
2645 called. The value of startoffset is also checked to ensure that it
2646 points to the start of a UTF-8 character. There is a discussion about
2647 the validity of UTF-8 strings in the pcreunicode page. If an invalid
2648 sequence of bytes is found, pcre_exec() returns the error
2649 PCRE_ERROR_BADUTF8 or, if PCRE_PARTIAL_HARD is set and the problem is a
2650 truncated character at the end of the subject, PCRE_ERROR_SHORTUTF8. In
2651 both cases, information about the precise nature of the error may also
2652 be returned (see the descriptions of these errors in the section enti-
2653 tled Error return values from pcre_exec() below). If startoffset con-
2654 tains a value that does not point to the start of a UTF-8 character (or
2655 to the end of the subject), PCRE_ERROR_BADUTF8_OFFSET is returned.
2657 If you already know that your subject is valid, and you want to skip
2658 these checks for performance reasons, you can set the
2659 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2660 do this for the second and subsequent calls to pcre_exec() if you are
2661 making repeated calls to find all the matches in a single subject
2662 string. However, you should be sure that the value of startoffset
2663 points to the start of a character (or the end of the subject). When
2664 PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid string as a
2665 subject or an invalid value of startoffset is undefined. Your program
2666 may crash.
2671 These options turn on the partial matching feature. For backwards com-
2672 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2673 match occurs if the end of the subject string is reached successfully,
2674 but there are not enough subject characters to complete the match. If
2675 this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
2676 matching continues by testing any remaining alternatives. Only if no
2677 complete match can be found is PCRE_ERROR_PARTIAL returned instead of
2678 PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
2679 caller is prepared to handle a partial match, but only if no complete
2680 match can be found.
2682 If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
2683 case, if a partial match is found, pcre_exec() immediately returns
2684 PCRE_ERROR_PARTIAL, without considering any other alternatives. In
2685 other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
2686 ered to be more important that an alternative complete match.
2688 In both cases, the portion of the string that was inspected when the
2689 partial match was found is set as the first matching string. There is a
2690 more detailed discussion of partial and multi-segment matching, with
2691 examples, in the pcrepartial documentation.
2693 The string to be matched by pcre_exec()
2695 The subject string is passed to pcre_exec() as a pointer in subject, a
2696 length in bytes in length, and a starting byte offset in startoffset.
2697 If this is negative or greater than the length of the subject,
2698 pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
2699 zero, the search for a match starts at the beginning of the subject,
2700 and this is by far the most common case. In UTF-8 mode, the byte offset
2701 must point to the start of a UTF-8 character (or the end of the sub-
2702 ject). Unlike the pattern string, the subject may contain binary zero
2703 bytes.
2705 A non-zero starting offset is useful when searching for another match
2706 in the same subject by calling pcre_exec() again after a previous suc-
2707 cess. Setting startoffset differs from just passing over a shortened
2708 string and setting PCRE_NOTBOL in the case of a pattern that begins
2709 with any kind of lookbehind. For example, consider the pattern
2711 \Biss\B
2713 which finds occurrences of "iss" in the middle of words. (\B matches
2714 only if the current position in the subject is not a word boundary.)
2715 When applied to the string "Mississipi" the first call to pcre_exec()
2716 finds the first occurrence. If pcre_exec() is called again with just
2717 the remainder of the subject, namely "issipi", it does not match,
2718 because \B is always false at the start of the subject, which is deemed
2719 to be a word boundary. However, if pcre_exec() is passed the entire
2720 string again, but with startoffset set to 4, it finds the second occur-
2721 rence of "iss" because it is able to look behind the starting point to
2722 discover that it is preceded by a letter.
2724 Finding all the matches in a subject is tricky when the pattern can
2725 match an empty string. It is possible to emulate Perl's /g behaviour by
2726 first trying the match again at the same offset, with the
2727 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
2728 fails, advancing the starting offset and trying an ordinary match
2729 again. There is some code that demonstrates how to do this in the pcre-
2730 demo sample program. In the most general case, you have to check to see
2731 if the newline convention recognizes CRLF as a newline, and if so, and
2732 the current character is CR followed by LF, advance the starting offset
2733 by two characters instead of one.
2735 If a non-zero starting offset is passed when the pattern is anchored,
2736 one attempt to match at the given offset is made. This can only succeed
2737 if the pattern does not require the match to be at the start of the
2738 subject.
2740 How pcre_exec() returns captured substrings
2742 In general, a pattern matches a certain portion of the subject, and in
2743 addition, further substrings from the subject may be picked out by
2744 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2745 this is called "capturing" in what follows, and the phrase "capturing
2746 subpattern" is used for a fragment of a pattern that picks out a sub-
2747 string. PCRE supports several other kinds of parenthesized subpattern
2748 that do not cause substrings to be captured.
2750 Captured substrings are returned to the caller via a vector of integers
2751 whose address is passed in ovector. The number of elements in the vec-
2752 tor is passed in ovecsize, which must be a non-negative number. Note:
2753 this argument is NOT the size of ovector in bytes.
2755 The first two-thirds of the vector is used to pass back captured sub-
2756 strings, each substring using a pair of integers. The remaining third
2757 of the vector is used as workspace by pcre_exec() while matching cap-
2758 turing subpatterns, and is not available for passing back information.
2759 The number passed in ovecsize should always be a multiple of three. If
2760 it is not, it is rounded down.
2762 When a match is successful, information about captured substrings is
2763 returned in pairs of integers, starting at the beginning of ovector,
2764 and continuing up to two-thirds of its length at the most. The first
2765 element of each pair is set to the byte offset of the first character
2766 in a substring, and the second is set to the byte offset of the first
2767 character after the end of a substring. Note: these values are always
2768 byte offsets, even in UTF-8 mode. They are not character counts.
2770 The first pair of integers, ovector[0] and ovector[1], identify the
2771 portion of the subject string matched by the entire pattern. The next
2772 pair is used for the first capturing subpattern, and so on. The value
2773 returned by pcre_exec() is one more than the highest numbered pair that
2774 has been set. For example, if two substrings have been captured, the
2775 returned value is 3. If there are no capturing subpatterns, the return
2776 value from a successful match is 1, indicating that just the first pair
2777 of offsets has been set.
2779 If a capturing subpattern is matched repeatedly, it is the last portion
2780 of the string that it matched that is returned.
2782 If the vector is too small to hold all the captured substring offsets,
2783 it is used as far as possible (up to two-thirds of its length), and the
2784 function returns a value of zero. If neither the actual string matched
2785 not any captured substrings are of interest, pcre_exec() may be called
2786 with ovector passed as NULL and ovecsize as zero. However, if the pat-
2787 tern contains back references and the ovector is not big enough to
2788 remember the related substrings, PCRE has to get additional memory for
2789 use during matching. Thus it is usually advisable to supply an ovector
2790 of reasonable size.
2792 There are some cases where zero is returned (indicating vector over-
2793 flow) when in fact the vector is exactly the right size for the final
2794 match. For example, consider the pattern
2796 (a)(?:(b)c|bd)
2798 If a vector of 6 elements (allowing for only 1 captured substring) is
2799 given with subject string "abd", pcre_exec() will try to set the second
2800 captured string, thereby recording a vector overflow, before failing to
2801 match "c" and backing up to try the second alternative. The zero
2802 return, however, does correctly indicate that the maximum number of
2803 slots (namely 2) have been filled. In similar cases where there is tem-
2804 porary overflow, but the final number of used slots is actually less
2805 than the maximum, a non-zero value is returned.
2807 The pcre_fullinfo() function can be used to find out how many capturing
2808 subpatterns there are in a compiled pattern. The smallest size for
2809 ovector that will allow for n captured substrings, in addition to the
2810 offsets of the substring matched by the whole pattern, is (n+1)*3.
2812 It is possible for capturing subpattern number n+1 to match some part
2813 of the subject when subpattern n has not been used at all. For example,
2814 if the string "abc" is matched against the pattern (a|(z))(bc) the
2815 return from the function is 4, and subpatterns 1 and 3 are matched, but
2816 2 is not. When this happens, both values in the offset pairs corre-
2817 sponding to unused subpatterns are set to -1.
2819 Offset values that correspond to unused subpatterns at the end of the
2820 expression are also set to -1. For example, if the string "abc" is
2821 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2822 matched. The return from the function is 2, because the highest used
2823 capturing subpattern number is 1, and the offsets for for the second
2824 and third capturing subpatterns (assuming the vector is large enough,
2825 of course) are set to -1.
2827 Note: Elements in the first two-thirds of ovector that do not corre-
2828 spond to capturing parentheses in the pattern are never changed. That
2829 is, if a pattern contains n capturing parentheses, no more than ovec-
2830 tor[0] to ovector[2n+1] are set by pcre_exec(). The other elements (in
2831 the first two-thirds) retain whatever values they previously had.
2833 Some convenience functions are provided for extracting the captured
2834 substrings as separate strings. These are described below.
2836 Error return values from pcre_exec()
2838 If pcre_exec() fails, it returns a negative number. The following are
2839 defined in the header file:
2843 The subject string did not match the pattern.
2847 Either code or subject was passed as NULL, or ovector was NULL and
2848 ovecsize was not zero.
2852 An unrecognized bit was set in the options argument.
2856 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2857 to catch the case when it is passed a junk pointer and to detect when a
2858 pattern that was compiled in an environment of one endianness is run in
2859 an environment with the other endianness. This is the error that PCRE
2860 gives when the magic number is not present.
2864 While running the pattern match, an unknown item was encountered in the
2865 compiled pattern. This error could be caused by a bug in PCRE or by
2866 overwriting of the compiled pattern.
2870 If a pattern contains back references, but the ovector that is passed
2871 to pcre_exec() is not big enough to remember the referenced substrings,
2872 PCRE gets a block of memory at the start of matching to use for this
2873 purpose. If the call via pcre_malloc() fails, this error is given. The
2874 memory is automatically freed at the end of matching.
2876 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2877 This can happen only when PCRE has been compiled with --disable-stack-
2878 for-recursion.
2882 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2883 and pcre_get_substring_list() functions (see below). It is never
2884 returned by pcre_exec().
2888 The backtracking limit, as specified by the match_limit field in a
2889 pcre_extra structure (or defaulted) was reached. See the description
2890 above.
2894 This error is never generated by pcre_exec() itself. It is provided for
2895 use by callout functions that want to yield a distinctive error code.
2896 See the pcrecallout documentation for details.
2900 A string that contains an invalid UTF-8 byte sequence was passed as a
2901 subject, and the PCRE_NO_UTF8_CHECK option was not set. If the size of
2902 the output vector (ovecsize) is at least 2, the byte offset to the
2903 start of the the invalid UTF-8 character is placed in the first ele-
2904 ment, and a reason code is placed in the second element. The reason
2905 codes are listed in the following section. For backward compatibility,
2906 if PCRE_PARTIAL_HARD is set and the problem is a truncated UTF-8 char-
2907 acter at the end of the subject (reason codes 1 to 5),
2908 PCRE_ERROR_SHORTUTF8 is returned instead of PCRE_ERROR_BADUTF8.
2912 The UTF-8 byte sequence that was passed as a subject was checked and
2913 found to be valid (the PCRE_NO_UTF8_CHECK option was not set), but the
2914 value of startoffset did not point to the beginning of a UTF-8 charac-
2915 ter or the end of the subject.
2919 The subject string did not match, but it did match partially. See the
2920 pcrepartial documentation for details of partial matching.
2924 This code is no longer in use. It was formerly returned when the
2925 PCRE_PARTIAL option was used with a compiled pattern containing items
2926 that were not supported for partial matching. From release 8.00
2927 onwards, there are no restrictions on partial matching.
2931 An unexpected internal error has occurred. This error could be caused
2932 by a bug in PCRE or by overwriting of the compiled pattern.
2936 This error is given if the value of the ovecsize argument is negative.
2940 The internal recursion limit, as specified by the match_limit_recursion
2941 field in a pcre_extra structure (or defaulted) was reached. See the
2942 description above.
2946 An invalid combination of PCRE_NEWLINE_xxx options was given.
2950 The value of startoffset was negative or greater than the length of the
2951 subject, that is, the value in length.
2955 This error is returned instead of PCRE_ERROR_BADUTF8 when the subject
2956 string ends with a truncated UTF-8 character and the PCRE_PARTIAL_HARD
2957 option is set. Information about the failure is returned as for
2958 PCRE_ERROR_BADUTF8. It is in fact sufficient to detect this case, but
2959 this special error code for PCRE_PARTIAL_HARD precedes the implementa-
2960 tion of returned information; it is retained for backwards compatibil-
2961 ity.
2965 This error is returned when pcre_exec() detects a recursion loop within
2966 the pattern. Specifically, it means that either the whole pattern or a
2967 subpattern has been called recursively for the second time at the same
2968 position in the subject string. Some simple patterns that might do this
2969 are detected and faulted at compile time, but more complicated cases,
2970 in particular mutual recursions between two different subpatterns, can-
2971 not be detected until run time.
2975 This error is returned when a pattern that was successfully studied
2976 using the PCRE_STUDY_JIT_COMPILE option is being matched, but the mem-
2977 ory available for the just-in-time processing stack is not large
2978 enough. See the pcrejit documentation for more details.
2982 This error is given if a pattern that was compiled by the 8-bit library
2983 is passed to a 16-bit library function, or vice versa.
2987 This error is given if a pattern that was compiled and saved is
2988 reloaded on a host with different endianness. The utility function
2989 pcre_pattern_to_host_byte_order() can be used to convert such a pattern
2990 so that it runs on the new host.
2992 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2994 Reason codes for invalid UTF-8 strings
2996 This section applies only to the 8-bit library. The corresponding
2997 information for the 16-bit library is given in the pcre16 page.
2999 When pcre_exec() returns either PCRE_ERROR_BADUTF8 or PCRE_ERROR_SHORT-
3000 UTF8, and the size of the output vector (ovecsize) is at least 2, the
3001 offset of the start of the invalid UTF-8 character is placed in the
3002 first output vector element (ovector[0]) and a reason code is placed in
3003 the second element (ovector[1]). The reason codes are given names in
3004 the pcre.h header file:
3012 The string ends with a truncated UTF-8 character; the code specifies
3013 how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8
3014 characters to be no longer than 4 bytes, the encoding scheme (origi-
3015 nally defined by RFC 2279) allows for up to 6 bytes, and this is
3016 checked first; hence the possibility of 4 or 5 missing bytes.
3022 PCRE_UTF8_ERR10
3024 The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of
3025 the character do not have the binary value 0b10 (that is, either the
3026 most significant bit is 0, or the next bit is 1).
3028 PCRE_UTF8_ERR11
3029 PCRE_UTF8_ERR12
3031 A character that is valid by the RFC 2279 rules is either 5 or 6 bytes
3032 long; these code points are excluded by RFC 3629.
3034 PCRE_UTF8_ERR13
3036 A 4-byte character has a value greater than 0x10fff; these code points
3037 are excluded by RFC 3629.
3039 PCRE_UTF8_ERR14
3041 A 3-byte character has a value in the range 0xd800 to 0xdfff; this
3042 range of code points are reserved by RFC 3629 for use with UTF-16, and
3043 so are excluded from UTF-8.
3045 PCRE_UTF8_ERR15
3046 PCRE_UTF8_ERR16
3047 PCRE_UTF8_ERR17
3048 PCRE_UTF8_ERR18
3049 PCRE_UTF8_ERR19
3051 A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes
3052 for a value that can be represented by fewer bytes, which is invalid.
3053 For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor-
3054 rect coding uses just one byte.
3056 PCRE_UTF8_ERR20
3058 The two most significant bits of the first byte of a character have the
3059 binary value 0b10 (that is, the most significant bit is 1 and the sec-
3060 ond is 0). Such a byte can only validly occur as the second or subse-
3061 quent byte of a multi-byte character.
3063 PCRE_UTF8_ERR21
3065 The first byte of a character has the value 0xfe or 0xff. These values
3066 can never occur in a valid UTF-8 string.
3071 int pcre_copy_substring(const char *subject, int *ovector,
3072 int stringcount, int stringnumber, char *buffer,
3073 int buffersize);
3075 int pcre_get_substring(const char *subject, int *ovector,
3076 int stringcount, int stringnumber,
3077 const char **stringptr);
3079 int pcre_get_substring_list(const char *subject,
3080 int *ovector, int stringcount, const char ***listptr);
3082 Captured substrings can be accessed directly by using the offsets
3083 returned by pcre_exec() in ovector. For convenience, the functions
3084 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
3085 string_list() are provided for extracting captured substrings as new,
3086 separate, zero-terminated strings. These functions identify substrings
3087 by number. The next section describes functions for extracting named
3088 substrings.
3090 A substring that contains a binary zero is correctly extracted and has
3091 a further zero added on the end, but the result is not, of course, a C
3092 string. However, you can process such a string by referring to the
3093 length that is returned by pcre_copy_substring() and pcre_get_sub-
3094 string(). Unfortunately, the interface to pcre_get_substring_list() is
3095 not adequate for handling strings containing binary zeros, because the
3096 end of the final string is not independently indicated.
3098 The first three arguments are the same for all three of these func-
3099 tions: subject is the subject string that has just been successfully
3100 matched, ovector is a pointer to the vector of integer offsets that was
3101 passed to pcre_exec(), and stringcount is the number of substrings that
3102 were captured by the match, including the substring that matched the
3103 entire regular expression. This is the value returned by pcre_exec() if
3104 it is greater than zero. If pcre_exec() returned zero, indicating that
3105 it ran out of space in ovector, the value passed as stringcount should
3106 be the number of elements in the vector divided by three.
3108 The functions pcre_copy_substring() and pcre_get_substring() extract a
3109 single substring, whose number is given as stringnumber. A value of
3110 zero extracts the substring that matched the entire pattern, whereas
3111 higher values extract the captured substrings. For pcre_copy_sub-
3112 string(), the string is placed in buffer, whose length is given by
3113 buffersize, while for pcre_get_substring() a new block of memory is
3114 obtained via pcre_malloc, and its address is returned via stringptr.
3115 The yield of the function is the length of the string, not including
3116 the terminating zero, or one of these error codes:
3120 The buffer was too small for pcre_copy_substring(), or the attempt to
3121 get memory failed for pcre_get_substring().
3125 There is no substring whose number is stringnumber.
3127 The pcre_get_substring_list() function extracts all available sub-
3128 strings and builds a list of pointers to them. All this is done in a
3129 single block of memory that is obtained via pcre_malloc. The address of
3130 the memory block is returned via listptr, which is also the start of
3131 the list of string pointers. The end of the list is marked by a NULL
3132 pointer. The yield of the function is zero if all went well, or the
3133 error code
3137 if the attempt to get the memory block failed.
3139 When any of these functions encounter a substring that is unset, which
3140 can happen when capturing subpattern number n+1 matches some part of
3141 the subject, but subpattern n has not been used at all, they return an
3142 empty string. This can be distinguished from a genuine zero-length sub-
3143 string by inspecting the appropriate offset in ovector, which is nega-
3144 tive for unset substrings.
3146 The two convenience functions pcre_free_substring() and pcre_free_sub-
3147 string_list() can be used to free the memory returned by a previous
3148 call of pcre_get_substring() or pcre_get_substring_list(), respec-
3149 tively. They do nothing more than call the function pointed to by
3150 pcre_free, which of course could be called directly from a C program.
3151 However, PCRE is used in some situations where it is linked via a spe-
3152 cial interface to another programming language that cannot use
3153 pcre_free directly; it is for these cases that the functions are pro-
3154 vided.
3159 int pcre_get_stringnumber(const pcre *code,
3160 const char *name);
3162 int pcre_copy_named_substring(const pcre *code,
3163 const char *subject, int *ovector,
3164 int stringcount, const char *stringname,
3165 char *buffer, int buffersize);
3167 int pcre_get_named_substring(const pcre *code,
3168 const char *subject, int *ovector,
3169 int stringcount, const char *stringname,
3170 const char **stringptr);
3172 To extract a substring by name, you first have to find associated num-
3173 ber. For example, for this pattern
3175 (a+)b(?<xxx>\d+)...
3177 the number of the subpattern called "xxx" is 2. If the name is known to
3178 be unique (PCRE_DUPNAMES was not set), you can find the number from the
3179 name by calling pcre_get_stringnumber(). The first argument is the com-
3180 piled pattern, and the second is the name. The yield of the function is
3181 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
3182 subpattern of that name.
3184 Given the number, you can extract the substring directly, or use one of
3185 the functions described in the previous section. For convenience, there
3186 are also two functions that do the whole job.
3188 Most of the arguments of pcre_copy_named_substring() and
3189 pcre_get_named_substring() are the same as those for the similarly
3190 named functions that extract by number. As these are described in the
3191 previous section, they are not re-described here. There are just two
3192 differences:
3194 First, instead of a substring number, a substring name is given. Sec-
3195 ond, there is an extra argument, given at the start, which is a pointer
3196 to the compiled pattern. This is needed in order to gain access to the
3197 name-to-number translation table.
3199 These functions call pcre_get_stringnumber(), and if it succeeds, they
3200 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
3201 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
3202 behaviour may not be what you want (see the next section).
3204 Warning: If the pattern uses the (?| feature to set up multiple subpat-
3205 terns with the same number, as described in the section on duplicate
3206 subpattern numbers in the pcrepattern page, you cannot use names to
3207 distinguish the different subpatterns, because names are not included
3208 in the compiled code. The matching process uses only numbers. For this
3209 reason, the use of different names for subpatterns of the same number
3210 causes an error at compile time.
3215 int pcre_get_stringtable_entries(const pcre *code,
3216 const char *name, char **first, char **last);
3218 When a pattern is compiled with the PCRE_DUPNAMES option, names for
3219 subpatterns are not required to be unique. (Duplicate names are always
3220 allowed for subpatterns with the same number, created by using the (?|
3221 feature. Indeed, if such subpatterns are named, they are required to
3222 use the same names.)
3224 Normally, patterns with duplicate names are such that in any one match,
3225 only one of the named subpatterns participates. An example is shown in
3226 the pcrepattern documentation.
3228 When duplicates are present, pcre_copy_named_substring() and
3229 pcre_get_named_substring() return the first substring corresponding to
3230 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
3231 (-7) is returned; no data is returned. The pcre_get_stringnumber()
3232 function returns one of the numbers that are associated with the name,
3233 but it is not defined which it is.
3235 If you want to get full details of all captured substrings for a given
3236 name, you must use the pcre_get_stringtable_entries() function. The
3237 first argument is the compiled pattern, and the second is the name. The
3238 third and fourth are pointers to variables which are updated by the
3239 function. After it has run, they point to the first and last entries in
3240 the name-to-number table for the given name. The function itself
3241 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
3242 there are none. The format of the table is described above in the sec-
3243 tion entitled Information about a pattern above. Given all the rele-
3244 vant entries for the name, you can extract each of their numbers, and
3245 hence the captured data, if any.
3250 The traditional matching function uses a similar algorithm to Perl,
3251 which stops when it finds the first match, starting at a given point in
3252 the subject. If you want to find all possible matches, or the longest
3253 possible match, consider using the alternative matching function (see
3254 below) instead. If you cannot use the alternative function, but still
3255 need to find all possible matches, you can kludge it up by making use
3256 of the callout facility, which is described in the pcrecallout documen-
3257 tation.
3259 What you have to do is to insert a callout right at the end of the pat-
3260 tern. When your callout function is called, extract and save the cur-
3261 rent matched substring. Then return 1, which forces pcre_exec() to
3262 backtrack and try other alternatives. Ultimately, when it runs out of
3263 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
3268 Matching certain patterns using pcre_exec() can use a lot of process
3269 stack, which in certain environments can be rather limited in size.
3270 Some users find it helpful to have an estimate of the amount of stack
3271 that is used by pcre_exec(), to help them set recursion limits, as
3272 described in the pcrestack documentation. The estimate that is output
3273 by pcretest when called with the -m and -C options is obtained by call-
3274 ing pcre_exec with the values NULL, NULL, NULL, -999, and -999 for its
3275 first five arguments.
3277 Normally, if its first argument is NULL, pcre_exec() immediately
3278 returns the negative error code PCRE_ERROR_NULL, but with this special
3279 combination of arguments, it returns instead a negative number whose
3280 absolute value is the approximate stack frame size in bytes. (A nega-
3281 tive number is used so that it is clear that no match has happened.)
3282 The value is approximate because in some cases, recursive calls to
3283 pcre_exec() occur when there are one or two additional variables on the
3284 stack.
3286 If PCRE has been compiled to use the heap instead of the stack for
3287 recursion, the value returned is the size of each block that is
3288 obtained from the heap.
3293 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
3294 const char *subject, int length, int startoffset,
3295 int options, int *ovector, int ovecsize,
3296 int *workspace, int wscount);
3298 The function pcre_dfa_exec() is called to match a subject string
3299 against a compiled pattern, using a matching algorithm that scans the
3300 subject string just once, and does not backtrack. This has different
3301 characteristics to the normal algorithm, and is not compatible with
3302 Perl. Some of the features of PCRE patterns are not supported. Never-
3303 theless, there are times when this kind of matching can be useful. For
3304 a discussion of the two matching algorithms, and a list of features
3305 that pcre_dfa_exec() does not support, see the pcrematching documenta-
3306 tion.
3308 The arguments for the pcre_dfa_exec() function are the same as for
3309 pcre_exec(), plus two extras. The ovector argument is used in a differ-
3310 ent way, and this is described below. The other common arguments are
3311 used in the same way as for pcre_exec(), so their description is not
3312 repeated here.
3314 The two additional arguments provide workspace for the function. The
3315 workspace vector should contain at least 20 elements. It is used for
3316 keeping track of multiple paths through the pattern tree. More
3317 workspace will be needed for patterns and subjects where there are a
3318 lot of potential matches.
3320 Here is an example of a simple call to pcre_dfa_exec():
3322 int rc;
3323 int ovector[10];
3324 int wspace[20];
3325 rc = pcre_dfa_exec(
3326 re, /* result of pcre_compile() */
3327 NULL, /* we didn't study the pattern */
3328 "some string", /* the subject string */
3329 11, /* the length of the subject string */
3330 0, /* start at offset 0 in the subject */
3331 0, /* default options */
3332 ovector, /* vector of integers for substring information */
3333 10, /* number of elements (NOT size in bytes) */
3334 wspace, /* working space vector */
3335 20); /* number of elements (NOT size in bytes) */
3337 Option bits for pcre_dfa_exec()
3339 The unused bits of the options argument for pcre_dfa_exec() must be
3340 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
3345 four of these are exactly the same as for pcre_exec(), so their
3346 description is not repeated here.
3351 These have the same general effect as they do for pcre_exec(), but the
3352 details are slightly different. When PCRE_PARTIAL_HARD is set for
3353 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
3354 ject is reached and there is still at least one matching possibility
3355 that requires additional characters. This happens even if some complete
3356 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
3357 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
3358 of the subject is reached, there have been no complete matches, but
3359 there is still at least one matching possibility. The portion of the
3360 string that was inspected when the longest partial match was found is
3361 set as the first matching string in both cases. There is a more
3362 detailed discussion of partial and multi-segment matching, with exam-
3363 ples, in the pcrepartial documentation.
3367 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
3368 stop as soon as it has found one match. Because of the way the alterna-
3369 tive algorithm works, this is necessarily the shortest possible match
3370 at the first possible matching point in the subject string.
3374 When pcre_dfa_exec() returns a partial match, it is possible to call it
3375 again, with additional subject characters, and have it continue with
3376 the same match. The PCRE_DFA_RESTART option requests this action; when
3377 it is set, the workspace and wscount options must reference the same
3378 vector as before because data about the match so far is left in them
3379 after a partial match. There is more discussion of this facility in the
3380 pcrepartial documentation.
3382 Successful returns from pcre_dfa_exec()
3384 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
3385 string in the subject. Note, however, that all the matches from one run
3386 of the function start at the same point in the subject. The shorter
3387 matches are all initial substrings of the longer matches. For example,
3388 if the pattern
3390 <.*>
3392 is matched against the string
3394 This is <something> <something else> <something further> no more
3396 the three matched strings are
3398 <something>
3399 <something> <something else>
3400 <something> <something else> <something further>
3402 On success, the yield of the function is a number greater than zero,
3403 which is the number of matched substrings. The substrings themselves
3404 are returned in ovector. Each string uses two elements; the first is
3405 the offset to the start, and the second is the offset to the end. In
3406 fact, all the strings have the same start offset. (Space could have
3407 been saved by giving this only once, but it was decided to retain some
3408 compatibility with the way pcre_exec() returns data, even though the
3409 meaning of the strings is different.)
3411 The strings are returned in reverse order of length; that is, the long-
3412 est matching string is given first. If there were too many matches to
3413 fit into ovector, the yield of the function is zero, and the vector is
3414 filled with the longest matches. Unlike pcre_exec(), pcre_dfa_exec()
3415 can use the entire ovector for returning matched strings.
3417 Error returns from pcre_dfa_exec()
3419 The pcre_dfa_exec() function returns a negative number when it fails.
3420 Many of the errors are the same as for pcre_exec(), and these are
3421 described above. There are in addition the following errors that are
3422 specific to pcre_dfa_exec():
3426 This return is given if pcre_dfa_exec() encounters an item in the pat-
3427 tern that it does not support, for instance, the use of \C or a back
3428 reference.
3432 This return is given if pcre_dfa_exec() encounters a condition item
3433 that uses a back reference for the condition, or a test for recursion
3434 in a specific group. These are not supported.
3438 This return is given if pcre_dfa_exec() is called with an extra block
3439 that contains a setting of the match_limit or match_limit_recursion
3440 fields. This is not supported (these fields are meaningless for DFA
3441 matching).
3445 This return is given if pcre_dfa_exec() runs out of space in the
3446 workspace vector.
3450 When a recursive subpattern is processed, the matching function calls
3451 itself recursively, using private vectors for ovector and workspace.
3452 This error is given if the output vector is not large enough. This
3453 should be extremely rare, as a vector of size 1000 is used.
3458 pcre16(3), pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematch-
3459 ing(3), pcrepartial(3), pcreposix(3), pcreprecompile(3), pcresample(3),
3460 pcrestack(3).
3465 Philip Hazel
3466 University Computing Service
3467 Cambridge CB2 3QH, England.
3472 Last updated: 21 January 2012
3473 Copyright (c) 1997-2012 University of Cambridge.
3474 ------------------------------------------------------------------------------
3480 NAME
3481 PCRE - Perl-compatible regular expressions
3486 int (*pcre_callout)(pcre_callout_block *);
3488 int (*pcre16_callout)(pcre16_callout_block *);
3490 PCRE provides a feature called "callout", which is a means of temporar-
3491 ily passing control to the caller of PCRE in the middle of pattern
3492 matching. The caller of PCRE provides an external function by putting
3493 its entry point in the global variable pcre_callout (pcre16_callout for
3494 the 16-bit library). By default, this variable contains NULL, which
3495 disables all calling out.
3497 Within a regular expression, (?C) indicates the points at which the
3498 external function is to be called. Different callout points can be
3499 identified by putting a number less than 256 after the letter C. The
3500 default value is zero. For example, this pattern has two callout
3501 points:
3503 (?C1)abc(?C2)def
3505 If the PCRE_AUTO_CALLOUT option bit is set when a pattern is compiled,
3506 PCRE automatically inserts callouts, all with number 255, before each
3507 item in the pattern. For example, if PCRE_AUTO_CALLOUT is used with the
3508 pattern
3510 A(\d{2}|--)
3512 it is processed as if it were
3514 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
3516 Notice that there is a callout before and after each parenthesis and
3517 alternation bar. Automatic callouts can be used for tracking the
3518 progress of pattern matching. The pcretest command has an option that
3519 sets automatic callouts; when it is used, the output indicates how the
3520 pattern is matched. This is useful information when you are trying to
3521 optimize the performance of a particular pattern.
3523 The use of callouts in a pattern makes it ineligible for optimization
3524 by the just-in-time compiler. Studying such a pattern with the
3525 PCRE_STUDY_JIT_COMPILE option always fails.
3530 You should be aware that, because of optimizations in the way PCRE
3531 matches patterns by default, callouts sometimes do not happen. For
3532 example, if the pattern is
3534 ab(?C4)cd
3536 PCRE knows that any matching string must contain the letter "d". If the
3537 subject string is "abyz", the lack of "d" means that matching doesn't
3538 ever start, and the callout is never reached. However, with "abyd",
3539 though the result is still no match, the callout is obeyed.
3541 If the pattern is studied, PCRE knows the minimum length of a matching
3542 string, and will immediately give a "no match" return without actually
3543 running a match if the subject is not long enough, or, for unanchored
3544 patterns, if it has been scanned far enough.
3546 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
3547 MIZE option to the matching function, or by starting the pattern with
3548 (*NO_START_OPT). This slows down the matching process, but does ensure
3549 that callouts such as the example above are obeyed.
3554 During matching, when PCRE reaches a callout point, the external func-
3555 tion defined by pcre_callout or pcre16_callout is called (if it is
3556 set). This applies to both normal and DFA matching. The only argument
3557 to the callout function is a pointer to a pcre_callout or pcre16_call-
3558 out block. These structures contains the following fields:
3560 int version;
3561 int callout_number;
3562 int *offset_vector;
3563 const char *subject; (8-bit version)
3564 PCRE_SPTR16 subject; (16-bit version)
3565 int subject_length;
3566 int start_match;
3567 int current_position;
3568 int capture_top;
3569 int capture_last;
3570 void *callout_data;
3571 int pattern_position;
3572 int next_item_length;
3573 const unsigned char *mark; (8-bit version)
3574 const PCRE_UCHAR16 *mark; (16-bit version)
3576 The version field is an integer containing the version number of the
3577 block format. The initial version was 0; the current version is 2. The
3578 version number will change again in future if additional fields are
3579 added, but the intention is never to remove any of the existing fields.
3581 The callout_number field contains the number of the callout, as com-
3582 piled into the pattern (that is, the number after ?C for manual call-
3583 outs, and 255 for automatically generated callouts).
3585 The offset_vector field is a pointer to the vector of offsets that was
3586 passed by the caller to the matching function. When pcre_exec() or
3587 pcre16_exec() is used, the contents can be inspected, in order to
3588 extract substrings that have been matched so far, in the same way as
3589 for extracting substrings after a match has completed. For the DFA
3590 matching functions, this field is not useful.
3592 The subject and subject_length fields contain copies of the values that
3593 were passed to the matching function.
3595 The start_match field normally contains the offset within the subject
3596 at which the current match attempt started. However, if the escape
3597 sequence \K has been encountered, this value is changed to reflect the
3598 modified starting point. If the pattern is not anchored, the callout
3599 function may be called several times from the same point in the pattern
3600 for different starting points in the subject.
3602 The current_position field contains the offset within the subject of
3603 the current match pointer.
3605 When the pcre_exec() or pcre16_exec() is used, the capture_top field
3606 contains one more than the number of the highest numbered captured sub-
3607 string so far. If no substrings have been captured, the value of cap-
3608 ture_top is one. This is always the case when the DFA functions are
3609 used, because they do not support captured substrings.
3611 The capture_last field contains the number of the most recently cap-
3612 tured substring. If no substrings have been captured, its value is -1.
3613 This is always the case for the DFA matching functions.
3615 The callout_data field contains a value that is passed to a matching
3616 function specifically so that it can be passed back in callouts. It is
3617 passed in the callout_data field of a pcre_extra or pcre16_extra data
3618 structure. If no such data was passed, the value of callout_data in a
3619 callout block is NULL. There is a description of the pcre_extra struc-
3620 ture in the pcreapi documentation.
3622 The pattern_position field is present from version 1 of the callout
3623 structure. It contains the offset to the next item to be matched in the
3624 pattern string.
3626 The next_item_length field is present from version 1 of the callout
3627 structure. It contains the length of the next item to be matched in the
3628 pattern string. When the callout immediately precedes an alternation
3629 bar, a closing parenthesis, or the end of the pattern, the length is
3630 zero. When the callout precedes an opening parenthesis, the length is
3631 that of the entire subpattern.
3633 The pattern_position and next_item_length fields are intended to help
3634 in distinguishing between different automatic callouts, which all have
3635 the same callout number. However, they are set for all callouts.
3637 The mark field is present from version 2 of the callout structure. In
3638 callouts from pcre_exec() or pcre16_exec() it contains a pointer to the
3639 zero-terminated name of the most recently passed (*MARK), (*PRUNE), or
3640 (*THEN) item in the match, or NULL if no such items have been passed.
3641 Instances of (*PRUNE) or (*THEN) without a name do not obliterate a
3642 previous (*MARK). In callouts from the DFA matching functions this
3643 field always contains NULL.
3648 The external callout function returns an integer to PCRE. If the value
3649 is zero, matching proceeds as normal. If the value is greater than
3650 zero, matching fails at the current point, but the testing of other
3651 matching possibilities goes ahead, just as if a lookahead assertion had
3652 failed. If the value is less than zero, the match is abandoned, the
3653 matching function returns the negative value.
3655 Negative values should normally be chosen from the set of
3656 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
3657 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
3658 reserved for use by callout functions; it will never be used by PCRE
3659 itself.
3664 Philip Hazel
3665 University Computing Service
3666 Cambridge CB2 3QH, England.
3671 Last updated: 08 Janurary 2012
3672 Copyright (c) 1997-2012 University of Cambridge.
3673 ------------------------------------------------------------------------------
3679 NAME
3680 PCRE - Perl-compatible regular expressions
3685 This document describes the differences in the ways that PCRE and Perl
3686 handle regular expressions. The differences described here are with
3687 respect to Perl versions 5.10 and above.
3689 1. PCRE has only a subset of Perl's Unicode support. Details of what it
3690 does have are given in the pcreunicode page.
3692 2. PCRE allows repeat quantifiers only on parenthesized assertions, but
3693 they do not mean what you might think. For example, (?!a){3} does not
3694 assert that the next three characters are not "a". It just asserts that
3695 the next character is not "a" three times (in principle: PCRE optimizes
3696 this to run the assertion just once). Perl allows repeat quantifiers on
3697 other assertions such as \b, but these do not seem to have any use.
3699 3. Capturing subpatterns that occur inside negative lookahead asser-
3700 tions are counted, but their entries in the offsets vector are never
3701 set. Perl sets its numerical variables from any such patterns that are
3702 matched before the assertion fails to match something (thereby succeed-
3703 ing), but only if the negative lookahead assertion contains just one
3704 branch.
3706 4. Though binary zero characters are supported in the subject string,
3707 they are not allowed in a pattern string because it is passed as a nor-
3708 mal C string, terminated by zero. The escape sequence \0 can be used in
3709 the pattern to represent a binary zero.
3711 5. The following Perl escape sequences are not supported: \l, \u, \L,
3712 \U, and \N when followed by a character name or Unicode value. (\N on
3713 its own, matching a non-newline character, is supported.) In fact these
3714 are implemented by Perl's general string-handling and are not part of
3715 its pattern matching engine. If any of these are encountered by PCRE,
3716 an error is generated by default. However, if the PCRE_JAVASCRIPT_COM-
3717 PAT option is set, \U and \u are interpreted as JavaScript interprets
3718 them.
3720 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
3721 is built with Unicode character property support. The properties that
3722 can be tested with \p and \P are limited to the general category prop-
3723 erties such as Lu and Nd, script names such as Greek or Han, and the
3724 derived properties Any and L&. PCRE does support the Cs (surrogate)
3725 property, which Perl does not; the Perl documentation says "Because
3726 Perl hides the need for the user to understand the internal representa-
3727 tion of Unicode characters, there is no need to implement the somewhat
3728 messy concept of surrogates."
3730 7. PCRE implements a simpler version of \X than Perl, which changed to
3731 make \X match what Unicode calls an "extended grapheme cluster". This
3732 is more complicated than an extended Unicode sequence, which is what
3733 PCRE matches.
3735 8. PCRE does support the \Q...\E escape for quoting substrings. Charac-
3736 ters in between are treated as literals. This is slightly different
3737 from Perl in that $ and @ are also handled as literals inside the
3738 quotes. In Perl, they cause variable interpolation (but of course PCRE
3739 does not have variables). Note the following examples:
3741 Pattern PCRE matches Perl matches
3743 \Qabc$xyz\E abc$xyz abc followed by the
3744 contents of $xyz
3745 \Qabc\$xyz\E abc\$xyz abc\$xyz
3746 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3748 The \Q...\E sequence is recognized both inside and outside character
3749 classes.
3751 9. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3752 constructions. However, there is support for recursive patterns. This
3753 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3754 "callout" feature allows an external function to be called during pat-
3755 tern matching. See the pcrecallout documentation for details.
3757 10. Subpatterns that are called as subroutines (whether or not recur-
3758 sively) are always treated as atomic groups in PCRE. This is like
3759 Python, but unlike Perl. Captured values that are set outside a sub-
3760 routine call can be reference from inside in PCRE, but not in Perl.
3761 There is a discussion that explains these differences in more detail in
3762 the section on recursion differences from Perl in the pcrepattern page.
3764 11. If (*THEN) is present in a group that is called as a subroutine,
3765 its action is limited to that group, even if the group does not contain
3766 any | characters.
3768 12. There are some differences that are concerned with the settings of
3769 captured strings when part of a pattern is repeated. For example,
3770 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3771 unset, but in PCRE it is set to "b".
3773 13. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3774 pattern names is not as general as Perl's. This is a consequence of the
3775 fact the PCRE works internally just with numbers, using an external ta-
3776 ble to translate between numbers and names. In particular, a pattern
3777 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3778 the same number but different names, is not supported, and causes an
3779 error at compile time. If it were allowed, it would not be possible to
3780 distinguish which parentheses matched, because both names map to cap-
3781 turing subpattern number 1. To avoid this confusing situation, an error
3782 is given at compile time.
3784 14. Perl recognizes comments in some places that PCRE does not, for
3785 example, between the ( and ? at the start of a subpattern. If the /x
3786 modifier is set, Perl allows whitespace between ( and ? but PCRE never
3787 does, even if the PCRE_EXTENDED option is set.
3789 15. PCRE provides some extensions to the Perl regular expression facil-
3790 ities. Perl 5.10 includes new features that are not in earlier ver-
3791 sions of Perl, some of which (such as named parentheses) have been in
3792 PCRE for some time. This list is with respect to Perl 5.10:
3794 (a) Although lookbehind assertions in PCRE must match fixed length
3795 strings, each alternative branch of a lookbehind assertion can match a
3796 different length of string. Perl requires them all to have the same
3797 length.
3799 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3800 meta-character matches only at the very end of the string.
3802 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3803 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3804 ignored. (Perl can be made to issue a warning.)
3806 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3807 fiers is inverted, that is, by default they are not greedy, but if fol-
3808 lowed by a question mark they are.
3810 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3811 tried only at the first matching position in the subject string.
3814 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3815 lents.
3817 (g) The \R escape sequence can be restricted to match only CR, LF, or
3818 CRLF by the PCRE_BSR_ANYCRLF option.
3820 (h) The callout facility is PCRE-specific.
3822 (i) The partial matching facility is PCRE-specific.
3824 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3825 even on different hosts that have the other endianness. However, this
3826 does not apply to optimized data created by the just-in-time compiler.
3828 (k) The alternative matching functions (pcre_dfa_exec() and
3829 pcre16_dfa_exec()) match in a different way and are not Perl-compati-
3830 ble.
3832 (l) PCRE recognizes some special sequences such as (*CR) at the start
3833 of a pattern that set overall options that cannot be changed within the
3834 pattern.
3839 Philip Hazel
3840 University Computing Service
3841 Cambridge CB2 3QH, England.
3846 Last updated: 08 Januray 2012
3847 Copyright (c) 1997-2012 University of Cambridge.
3848 ------------------------------------------------------------------------------
3854 NAME
3855 PCRE - Perl-compatible regular expressions
3860 The syntax and semantics of the regular expressions that are supported
3861 by PCRE are described in detail below. There is a quick-reference syn-
3862 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3863 semantics as closely as it can. PCRE also supports some alternative
3864 regular expression syntax (which does not conflict with the Perl syn-
3865 tax) in order to provide some compatibility with regular expressions in
3866 Python, .NET, and Oniguruma.
3868 Perl's regular expressions are described in its own documentation, and
3869 regular expressions in general are covered in a number of books, some
3870 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3871 Expressions", published by O'Reilly, covers regular expressions in
3872 great detail. This description of PCRE's regular expressions is
3873 intended as reference material.
3875 The original operation of PCRE was on strings of one-byte characters.
3876 However, there is now also support for UTF-8 strings in the original
3877 library, and a second library that supports 16-bit and UTF-16 character
3878 strings. To use these features, PCRE must be built to include appropri-
3879 ate support. When using UTF strings you must either call the compiling
3880 function with the PCRE_UTF8 or PCRE_UTF16 option, or the pattern must
3881 start with one of these special sequences:
3883 (*UTF8)
3884 (*UTF16)
3886 Starting a pattern with such a sequence is equivalent to setting the
3887 relevant option. This feature is not Perl-compatible. How setting a UTF
3888 mode affects pattern matching is mentioned in several places below.
3889 There is also a summary of features in the pcreunicode page.
3891 Another special sequence that may appear at the start of a pattern or
3892 in combination with (*UTF8) or (*UTF16) is:
3894 (*UCP)
3896 This has the same effect as setting the PCRE_UCP option: it causes
3897 sequences such as \d and \w to use Unicode properties to determine
3898 character types, instead of recognizing only characters with codes less
3899 than 128 via a lookup table.
3901 If a pattern starts with (*NO_START_OPT), it has the same effect as
3902 setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
3903 time. There are also some more of these special sequences that are con-
3904 cerned with the handling of newlines; they are described below.
3906 The remainder of this document discusses the patterns that are sup-
3907 ported by PCRE when one its main matching functions, pcre_exec()
3908 (8-bit) or pcre16_exec() (16-bit), is used. PCRE also has alternative
3909 matching functions, pcre_dfa_exec() and pcre16_dfa_exec(), which match
3910 using a different algorithm that is not Perl-compatible. Some of the
3911 features discussed below are not available when DFA matching is used.
3912 The advantages and disadvantages of the alternative functions, and how
3913 they differ from the normal functions, are discussed in the pcrematch-
3914 ing page.
3919 PCRE supports five different conventions for indicating line breaks in
3920 strings: a single CR (carriage return) character, a single LF (line-
3921 feed) character, the two-character sequence CRLF, any of the three pre-
3922 ceding, or any Unicode newline sequence. The pcreapi page has further
3923 discussion about newlines, and shows how to set the newline convention
3924 in the options arguments for the compiling and matching functions.
3926 It is also possible to specify a newline convention by starting a pat-
3927 tern string with one of the following five sequences:
3929 (*CR) carriage return
3930 (*LF) linefeed
3931 (*CRLF) carriage return, followed by linefeed
3932 (*ANYCRLF) any of the three above
3933 (*ANY) all Unicode newline sequences
3935 These override the default and the options given to the compiling func-
3936 tion. For example, on a Unix system where LF is the default newline
3937 sequence, the pattern
3939 (*CR)a.b
3941 changes the convention to CR. That pattern matches "a\nb" because LF is
3942 no longer a newline. Note that these special settings, which are not
3943 Perl-compatible, are recognized only at the very start of a pattern,
3944 and that they must be in upper case. If more than one of them is
3945 present, the last one is used.
3947 The newline convention affects the interpretation of the dot metachar-
3948 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3949 ever, it does not affect what the \R escape sequence matches. By
3950 default, this is any Unicode newline sequence, for Perl compatibility.
3951 However, this can be changed; see the description of \R in the section
3952 entitled "Newline sequences" below. A change of \R setting can be com-
3953 bined with a change of newline convention.
3958 A regular expression is a pattern that is matched against a subject
3959 string from left to right. Most characters stand for themselves in a
3960 pattern, and match the corresponding characters in the subject. As a
3961 trivial example, the pattern
3963 The quick brown fox
3965 matches a portion of a subject string that is identical to itself. When
3966 caseless matching is specified (the PCRE_CASELESS option), letters are
3967 matched independently of case. In a UTF mode, PCRE always understands
3968 the concept of case for characters whose values are less than 128, so
3969 caseless matching is always possible. For characters with higher val-
3970 ues, the concept of case is supported if PCRE is compiled with Unicode
3971 property support, but not otherwise. If you want to use caseless
3972 matching for characters 128 and above, you must ensure that PCRE is
3973 compiled with Unicode property support as well as with UTF support.
3975 The power of regular expressions comes from the ability to include
3976 alternatives and repetitions in the pattern. These are encoded in the
3977 pattern by the use of metacharacters, which do not stand for themselves
3978 but instead are interpreted in some special way.
3980 There are two different sets of metacharacters: those that are recog-
3981 nized anywhere in the pattern except within square brackets, and those
3982 that are recognized within square brackets. Outside square brackets,
3983 the metacharacters are as follows:
3985 \ general escape character with several uses
3986 ^ assert start of string (or line, in multiline mode)
3987 $ assert end of string (or line, in multiline mode)
3988 . match any character except newline (by default)
3989 [ start character class definition
3990 | start of alternative branch
3991 ( start subpattern
3992 ) end subpattern
3993 ? extends the meaning of (
3994 also 0 or 1 quantifier
3995 also quantifier minimizer
3996 * 0 or more quantifier
3997 + 1 or more quantifier
3998 also "possessive quantifier"
3999 { start min/max quantifier
4001 Part of a pattern that is in square brackets is called a "character
4002 class". In a character class the only metacharacters are:
4004 \ general escape character
4005 ^ negate the class, but only if the first character
4006 - indicates character range
4007 [ POSIX character class (only if followed by POSIX
4008 syntax)
4009 ] terminates the character class
4011 The following sections describe the use of each of the metacharacters.
4016 The backslash character has several uses. Firstly, if it is followed by
4017 a character that is not a number or a letter, it takes away any special
4018 meaning that character may have. This use of backslash as an escape
4019 character applies both inside and outside character classes.
4021 For example, if you want to match a * character, you write \* in the
4022 pattern. This escaping action applies whether or not the following
4023 character would otherwise be interpreted as a metacharacter, so it is
4024 always safe to precede a non-alphanumeric with backslash to specify
4025 that it stands for itself. In particular, if you want to match a back-
4026 slash, you write \\.
4028 In a UTF mode, only ASCII numbers and letters have any special meaning
4029 after a backslash. All other characters (in particular, those whose
4030 codepoints are greater than 127) are treated as literals.
4032 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
4033 the pattern (other than in a character class) and characters between a
4034 # outside a character class and the next newline are ignored. An escap-
4035 ing backslash can be used to include a whitespace or # character as
4036 part of the pattern.
4038 If you want to remove the special meaning from a sequence of charac-
4039 ters, you can do so by putting them between \Q and \E. This is differ-
4040 ent from Perl in that $ and @ are handled as literals in \Q...\E
4041 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
4042 tion. Note the following examples:
4044 Pattern PCRE matches Perl matches
4046 \Qabc$xyz\E abc$xyz abc followed by the
4047 contents of $xyz
4048 \Qabc\$xyz\E abc\$xyz abc\$xyz
4049 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4051 The \Q...\E sequence is recognized both inside and outside character
4052 classes. An isolated \E that is not preceded by \Q is ignored. If \Q
4053 is not followed by \E later in the pattern, the literal interpretation
4054 continues to the end of the pattern (that is, \E is assumed at the
4055 end). If the isolated \Q is inside a character class, this causes an
4056 error, because the character class is not terminated.
4058 Non-printing characters
4060 A second use of backslash provides a way of encoding non-printing char-
4061 acters in patterns in a visible manner. There is no restriction on the
4062 appearance of non-printing characters, apart from the binary zero that
4063 terminates a pattern, but when a pattern is being prepared by text
4064 editing, it is often easier to use one of the following escape
4065 sequences than the binary character it represents:
4067 \a alarm, that is, the BEL character (hex 07)
4068 \cx "control-x", where x is any ASCII character
4069 \e escape (hex 1B)
4070 \f formfeed (hex 0C)
4071 \n linefeed (hex 0A)
4072 \r carriage return (hex 0D)
4073 \t tab (hex 09)
4074 \ddd character with octal code ddd, or back reference
4075 \xhh character with hex code hh
4076 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
4077 \uhhhh character with hex code hhhh (JavaScript mode only)
4079 The precise effect of \cx is as follows: if x is a lower case letter,
4080 it is converted to upper case. Then bit 6 of the character (hex 40) is
4081 inverted. Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({
4082 is 7B), while \c; becomes hex 7B (; is 3B). If the byte following \c
4083 has a value greater than 127, a compile-time error occurs. This locks
4084 out non-ASCII characters in all modes. (When PCRE is compiled in EBCDIC
4085 mode, all byte values are valid. A lower case letter is converted to
4086 upper case, and then the 0xc0 bits are flipped.)
4088 By default, after \x, from zero to two hexadecimal digits are read
4089 (letters can be in upper or lower case). Any number of hexadecimal dig-
4090 its may appear between \x{ and }, but the character code is constrained
4091 as follows:
4093 8-bit non-UTF mode less than 0x100
4094 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
4095 16-bit non-UTF mode less than 0x10000
4096 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
4098 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-
4099 called "surrogate" codepoints).
4101 If characters other than hexadecimal digits appear between \x{ and },
4102 or if there is no terminating }, this form of escape is not recognized.
4103 Instead, the initial \x will be interpreted as a basic hexadecimal
4104 escape, with no following digits, giving a character whose value is
4105 zero.
4107 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x
4108 is as just described only when it is followed by two hexadecimal dig-
4109 its. Otherwise, it matches a literal "x" character. In JavaScript
4110 mode, support for code points greater than 256 is provided by \u, which
4111 must be followed by four hexadecimal digits; otherwise it matches a
4112 literal "u" character.
4114 Characters whose value is less than 256 can be defined by either of the
4115 two syntaxes for \x (or by \u in JavaScript mode). There is no differ-
4116 ence in the way they are handled. For example, \xdc is exactly the same
4117 as \x{dc} (or \u00dc in JavaScript mode).
4119 After \0 up to two further octal digits are read. If there are fewer
4120 than two digits, just those that are present are used. Thus the
4121 sequence \0\x\07 specifies two binary zeros followed by a BEL character
4122 (code value 7). Make sure you supply two digits after the initial zero
4123 if the pattern character that follows is itself an octal digit.
4125 The handling of a backslash followed by a digit other than 0 is compli-
4126 cated. Outside a character class, PCRE reads it and any following dig-
4127 its as a decimal number. If the number is less than 10, or if there
4128 have been at least that many previous capturing left parentheses in the
4129 expression, the entire sequence is taken as a back reference. A
4130 description of how this works is given later, following the discussion
4131 of parenthesized subpatterns.
4133 Inside a character class, or if the decimal number is greater than 9
4134 and there have not been that many capturing subpatterns, PCRE re-reads
4135 up to three octal digits following the backslash, and uses them to gen-
4136 erate a data character. Any subsequent digits stand for themselves. The
4137 value of the character is constrained in the same way as characters
4138 specified in hexadecimal. For example:
4140 \040 is another way of writing a space
4141 \40 is the same, provided there are fewer than 40
4142 previous capturing subpatterns
4143 \7 is always a back reference
4144 \11 might be a back reference, or another way of
4145 writing a tab
4146 \011 is always a tab
4147 \0113 is a tab followed by the character "3"
4148 \113 might be a back reference, otherwise the
4149 character with octal code 113
4150 \377 might be a back reference, otherwise
4151 the value 255 (decimal)
4152 \81 is either a back reference, or a binary zero
4153 followed by the two characters "8" and "1"
4155 Note that octal values of 100 or greater must not be introduced by a
4156 leading zero, because no more than three octal digits are ever read.
4158 All the sequences that define a single character value can be used both
4159 inside and outside character classes. In addition, inside a character
4160 class, \b is interpreted as the backspace character (hex 08).
4162 \N is not allowed in a character class. \B, \R, and \X are not special
4163 inside a character class. Like other unrecognized escape sequences,
4164 they are treated as the literal characters "B", "R", and "X" by
4165 default, but cause an error if the PCRE_EXTRA option is set. Outside a
4166 character class, these sequences have different meanings.
4168 Unsupported escape sequences
4170 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
4171 handler and used to modify the case of following characters. By
4172 default, PCRE does not support these escape sequences. However, if the
4173 PCRE_JAVASCRIPT_COMPAT option is set, \U matches a "U" character, and
4174 \u can be used to define a character by code point, as described in the
4175 previous section.
4177 Absolute and relative back references
4179 The sequence \g followed by an unsigned or a negative number, option-
4180 ally enclosed in braces, is an absolute or relative back reference. A
4181 named back reference can be coded as \g{name}. Back references are dis-
4182 cussed later, following the discussion of parenthesized subpatterns.
4184 Absolute and relative subroutine calls
4186 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4187 name or a number enclosed either in angle brackets or single quotes, is
4188 an alternative syntax for referencing a subpattern as a "subroutine".
4189 Details are discussed later. Note that \g{...} (Perl syntax) and
4190 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
4191 reference; the latter is a subroutine call.
4193 Generic character types
4195 Another use of backslash is for specifying generic character types:
4197 \d any decimal digit
4198 \D any character that is not a decimal digit
4199 \h any horizontal whitespace character
4200 \H any character that is not a horizontal whitespace character
4201 \s any whitespace character
4202 \S any character that is not a whitespace character
4203 \v any vertical whitespace character
4204 \V any character that is not a vertical whitespace character
4205 \w any "word" character
4206 \W any "non-word" character
4208 There is also the single sequence \N, which matches a non-newline char-
4209 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
4210 not set. Perl also uses \N to match characters by name; PCRE does not
4211 support this.
4213 Each pair of lower and upper case escape sequences partitions the com-
4214 plete set of characters into two disjoint sets. Any given character
4215 matches one, and only one, of each pair. The sequences can appear both
4216 inside and outside character classes. They each match one character of
4217 the appropriate type. If the current matching point is at the end of
4218 the subject string, all of them fail, because there is no character to
4219 match.
4221 For compatibility with Perl, \s does not match the VT character (code
4222 11). This makes it different from the the POSIX "space" class. The \s
4223 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
4224 "use locale;" is included in a Perl script, \s may match the VT charac-
4225 ter. In PCRE, it never does.
4227 A "word" character is an underscore or any character that is a letter
4228 or digit. By default, the definition of letters and digits is con-
4229 trolled by PCRE's low-valued character tables, and may vary if locale-
4230 specific matching is taking place (see "Locale support" in the pcreapi
4231 page). For example, in a French locale such as "fr_FR" in Unix-like
4232 systems, or "french" in Windows, some character codes greater than 128
4233 are used for accented letters, and these are then matched by \w. The
4234 use of locales with Unicode is discouraged.
4236 By default, in a UTF mode, characters with values greater than 128
4237 never match \d, \s, or \w, and always match \D, \S, and \W. These
4238 sequences retain their original meanings from before UTF support was
4239 available, mainly for efficiency reasons. However, if PCRE is compiled
4240 with Unicode property support, and the PCRE_UCP option is set, the be-
4241 haviour is changed so that Unicode properties are used to determine
4242 character types, as follows:
4244 \d any character that \p{Nd} matches (decimal digit)
4245 \s any character that \p{Z} matches, plus HT, LF, FF, CR
4246 \w any character that \p{L} or \p{N} matches, plus underscore
4248 The upper case escapes match the inverse sets of characters. Note that
4249 \d matches only decimal digits, whereas \w matches any Unicode digit,
4250 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
4251 affects \b, and \B because they are defined in terms of \w and \W.
4252 Matching these sequences is noticeably slower when PCRE_UCP is set.
4254 The sequences \h, \H, \v, and \V are features that were added to Perl
4255 at release 5.10. In contrast to the other sequences, which match only
4256 ASCII characters by default, these always match certain high-valued
4257 codepoints, whether or not PCRE_UCP is set. The horizontal space char-
4258 acters are:
4260 U+0009 Horizontal tab
4261 U+0020 Space
4262 U+00A0 Non-break space
4263 U+1680 Ogham space mark
4264 U+180E Mongolian vowel separator
4265 U+2000 En quad
4266 U+2001 Em quad
4267 U+2002 En space
4268 U+2003 Em space
4269 U+2004 Three-per-em space
4270 U+2005 Four-per-em space
4271 U+2006 Six-per-em space
4272 U+2007 Figure space
4273 U+2008 Punctuation space
4274 U+2009 Thin space
4275 U+200A Hair space
4276 U+202F Narrow no-break space
4277 U+205F Medium mathematical space
4278 U+3000 Ideographic space
4280 The vertical space characters are:
4282 U+000A Linefeed
4283 U+000B Vertical tab
4284 U+000C Formfeed
4285 U+000D Carriage return
4286 U+0085 Next line
4287 U+2028 Line separator
4288 U+2029 Paragraph separator
4290 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than
4291 256 are relevant.
4293 Newline sequences
4295 Outside a character class, by default, the escape sequence \R matches
4296 any Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent
4297 to the following:
4299 (?>\r\n|\n|\x0b|\f|\r|\x85)
4301 This is an example of an "atomic group", details of which are given
4302 below. This particular group matches either the two-character sequence
4303 CR followed by LF, or one of the single characters LF (linefeed,
4304 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
4305 return, U+000D), or NEL (next line, U+0085). The two-character sequence
4306 is treated as a single unit that cannot be split.
4308 In other modes, two additional characters whose codepoints are greater
4309 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
4310 rator, U+2029). Unicode character property support is not needed for
4311 these characters to be recognized.
4313 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
4314 the complete set of Unicode line endings) by setting the option
4315 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
4316 (BSR is an abbrevation for "backslash R".) This can be made the default
4317 when PCRE is built; if this is the case, the other behaviour can be
4318 requested via the PCRE_BSR_UNICODE option. It is also possible to
4319 specify these settings by starting a pattern string with one of the
4320 following sequences:
4322 (*BSR_ANYCRLF) CR, LF, or CRLF only
4323 (*BSR_UNICODE) any Unicode newline sequence
4325 These override the default and the options given to the compiling func-
4326 tion, but they can themselves be overridden by options given to a
4327 matching function. Note that these special settings, which are not
4328 Perl-compatible, are recognized only at the very start of a pattern,
4329 and that they must be in upper case. If more than one of them is
4330 present, the last one is used. They can be combined with a change of
4331 newline convention; for example, a pattern can start with:
4335 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
4336 sequences. Inside a character class, \R is treated as an unrecognized
4337 escape sequence, and so matches the letter "R" by default, but causes
4338 an error if PCRE_EXTRA is set.
4340 Unicode character properties
4342 When PCRE is built with Unicode character property support, three addi-
4343 tional escape sequences that match characters with specific properties
4344 are available. When in 8-bit non-UTF-8 mode, these sequences are of
4345 course limited to testing characters whose codepoints are less than
4346 256, but they do work in this mode. The extra escape sequences are:
4348 \p{xx} a character with the xx property
4349 \P{xx} a character without the xx property
4350 \X an extended Unicode sequence
4352 The property names represented by xx above are limited to the Unicode
4353 script names, the general category properties, "Any", which matches any
4354 character (including newline), and some special PCRE properties
4355 (described in the next section). Other Perl properties such as "InMu-
4356 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
4357 does not match any characters, so always causes a match failure.
4359 Sets of Unicode characters are defined as belonging to certain scripts.
4360 A character from one of these sets can be matched using a script name.
4361 For example:
4363 \p{Greek}
4364 \P{Han}
4366 Those that are not part of an identified script are lumped together as
4367 "Common". The current list of scripts is:
4369 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
4370 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
4371 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
4372 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
4373 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
4374 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
4375 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
4376 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
4377 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
4378 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
4379 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
4380 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
4381 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
4382 Ugaritic, Vai, Yi.
4384 Each character has exactly one Unicode general category property, spec-
4385 ified by a two-letter abbreviation. For compatibility with Perl, nega-
4386 tion can be specified by including a circumflex between the opening
4387 brace and the property name. For example, \p{^Lu} is the same as
4388 \P{Lu}.
4390 If only one letter is specified with \p or \P, it includes all the gen-
4391 eral category properties that start with that letter. In this case, in
4392 the absence of negation, the curly brackets in the escape sequence are
4393 optional; these two examples have the same effect:
4395 \p{L}
4396 \pL
4398 The following general category property codes are supported:
4400 C Other
4401 Cc Control
4402 Cf Format
4403 Cn Unassigned
4404 Co Private use
4405 Cs Surrogate
4407 L Letter
4408 Ll Lower case letter
4409 Lm Modifier letter
4410 Lo Other letter
4411 Lt Title case letter
4412 Lu Upper case letter
4414 M Mark
4415 Mc Spacing mark
4416 Me Enclosing mark
4417 Mn Non-spacing mark
4419 N Number
4420 Nd Decimal number
4421 Nl Letter number
4422 No Other number
4424 P Punctuation
4425 Pc Connector punctuation
4426 Pd Dash punctuation
4427 Pe Close punctuation
4428 Pf Final punctuation
4429 Pi Initial punctuation
4430 Po Other punctuation
4431 Ps Open punctuation
4433 S Symbol
4434 Sc Currency symbol
4435 Sk Modifier symbol
4436 Sm Mathematical symbol
4437 So Other symbol
4439 Z Separator
4440 Zl Line separator
4441 Zp Paragraph separator
4442 Zs Space separator
4444 The special property L& is also supported: it matches a character that
4445 has the Lu, Ll, or Lt property, in other words, a letter that is not
4446 classified as a modifier or "other".
4448 The Cs (Surrogate) property applies only to characters in the range
4449 U+D800 to U+DFFF. Such characters are not valid in Unicode strings and
4450 so cannot be tested by PCRE, unless UTF validity checking has been
4451 turned off (see the discussion of PCRE_NO_UTF8_CHECK and
4452 PCRE_NO_UTF16_CHECK in the pcreapi page). Perl does not support the Cs
4453 property.
4455 The long synonyms for property names that Perl supports (such as
4456 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
4457 any of these properties with "Is".
4459 No character that is in the Unicode table has the Cn (unassigned) prop-
4460 erty. Instead, this property is assumed for any code point that is not
4461 in the Unicode table.
4463 Specifying caseless matching does not affect these escape sequences.
4464 For example, \p{Lu} always matches only upper case letters.
4466 The \X escape matches any number of Unicode characters that form an
4467 extended Unicode sequence. \X is equivalent to
4469 (?>\PM\pM*)
4471 That is, it matches a character without the "mark" property, followed
4472 by zero or more characters with the "mark" property, and treats the
4473 sequence as an atomic group (see below). Characters with the "mark"
4474 property are typically accents that affect the preceding character.
4475 None of them have codepoints less than 256, so in 8-bit non-UTF-8 mode
4476 \X matches any one character.
4478 Note that recent versions of Perl have changed \X to match what Unicode
4479 calls an "extended grapheme cluster", which has a more complicated def-
4480 inition.
4482 Matching characters by Unicode property is not fast, because PCRE has
4483 to search a structure that contains data for over fifteen thousand
4484 characters. That is why the traditional escape sequences such as \d and
4485 \w do not use Unicode properties in PCRE by default, though you can
4486 make them do so by setting the PCRE_UCP option or by starting the pat-
4487 tern with (*UCP).
4489 PCRE's additional properties
4491 As well as the standard Unicode properties described in the previous
4492 section, PCRE supports four more that make it possible to convert tra-
4493 ditional escape sequences such as \w and \s and POSIX character classes
4494 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
4495 erties internally when PCRE_UCP is set. They are:
4497 Xan Any alphanumeric character
4498 Xps Any POSIX space character
4499 Xsp Any Perl space character
4500 Xwd Any Perl "word" character
4502 Xan matches characters that have either the L (letter) or the N (num-
4503 ber) property. Xps matches the characters tab, linefeed, vertical tab,
4504 formfeed, or carriage return, and any other character that has the Z
4505 (separator) property. Xsp is the same as Xps, except that vertical tab
4506 is excluded. Xwd matches the same characters as Xan, plus underscore.
4508 Resetting the match start
4510 The escape sequence \K causes any previously matched characters not to
4511 be included in the final matched sequence. For example, the pattern:
4513 foo\Kbar
4515 matches "foobar", but reports that it has matched "bar". This feature
4516 is similar to a lookbehind assertion (described below). However, in
4517 this case, the part of the subject before the real match does not have
4518 to be of fixed length, as lookbehind assertions do. The use of \K does
4519 not interfere with the setting of captured substrings. For example,
4520 when the pattern
4522 (foo)\Kbar
4524 matches "foobar", the first substring is still set to "foo".
4526 Perl documents that the use of \K within assertions is "not well
4527 defined". In PCRE, \K is acted upon when it occurs inside positive
4528 assertions, but is ignored in negative assertions.
4530 Simple assertions
4532 The final use of backslash is for certain simple assertions. An asser-
4533 tion specifies a condition that has to be met at a particular point in
4534 a match, without consuming any characters from the subject string. The
4535 use of subpatterns for more complicated assertions is described below.
4536 The backslashed assertions are:
4538 \b matches at a word boundary
4539 \B matches when not at a word boundary
4540 \A matches at the start of the subject
4541 \Z matches at the end of the subject
4542 also matches before a newline at the end of the subject
4543 \z matches only at the end of the subject
4544 \G matches at the first matching position in the subject
4546 Inside a character class, \b has a different meaning; it matches the
4547 backspace character. If any other of these assertions appears in a
4548 character class, by default it matches the corresponding literal char-
4549 acter (for example, \B matches the letter B). However, if the
4550 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
4551 ated instead.
4553 A word boundary is a position in the subject string where the current
4554 character and the previous character do not both match \w or \W (i.e.
4555 one matches \w and the other matches \W), or the start or end of the
4556 string if the first or last character matches \w, respectively. In a
4557 UTF mode, the meanings of \w and \W can be changed by setting the
4558 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
4559 PCRE nor Perl has a separate "start of word" or "end of word" metase-
4560 quence. However, whatever follows \b normally determines which it is.
4561 For example, the fragment \ba matches "a" at the start of a word.
4563 The \A, \Z, and \z assertions differ from the traditional circumflex
4564 and dollar (described in the next section) in that they only ever match
4565 at the very start and end of the subject string, whatever options are
4566 set. Thus, they are independent of multiline mode. These three asser-
4567 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
4568 affect only the behaviour of the circumflex and dollar metacharacters.
4569 However, if the startoffset argument of pcre_exec() is non-zero, indi-
4570 cating that matching is to start at a point other than the beginning of
4571 the subject, \A can never match. The difference between \Z and \z is
4572 that \Z matches before a newline at the end of the string as well as at
4573 the very end, whereas \z matches only at the end.
4575 The \G assertion is true only when the current matching position is at
4576 the start point of the match, as specified by the startoffset argument
4577 of pcre_exec(). It differs from \A when the value of startoffset is
4578 non-zero. By calling pcre_exec() multiple times with appropriate argu-
4579 ments, you can mimic Perl's /g option, and it is in this kind of imple-
4580 mentation where \G can be useful.
4582 Note, however, that PCRE's interpretation of \G, as the start of the
4583 current match, is subtly different from Perl's, which defines it as the
4584 end of the previous match. In Perl, these can be different when the
4585 previously matched string was empty. Because PCRE does just one match
4586 at a time, it cannot reproduce this behaviour.
4588 If all the alternatives of a pattern begin with \G, the expression is
4589 anchored to the starting match position, and the "anchored" flag is set
4590 in the compiled regular expression.
4595 Outside a character class, in the default matching mode, the circumflex
4596 character is an assertion that is true only if the current matching
4597 point is at the start of the subject string. If the startoffset argu-
4598 ment of pcre_exec() is non-zero, circumflex can never match if the
4599 PCRE_MULTILINE option is unset. Inside a character class, circumflex
4600 has an entirely different meaning (see below).
4602 Circumflex need not be the first character of the pattern if a number
4603 of alternatives are involved, but it should be the first thing in each
4604 alternative in which it appears if the pattern is ever to match that
4605 branch. If all possible alternatives start with a circumflex, that is,
4606 if the pattern is constrained to match only at the start of the sub-
4607 ject, it is said to be an "anchored" pattern. (There are also other
4608 constructs that can cause a pattern to be anchored.)
4610 A dollar character is an assertion that is true only if the current
4611 matching point is at the end of the subject string, or immediately
4612 before a newline at the end of the string (by default). Dollar need not
4613 be the last character of the pattern if a number of alternatives are
4614 involved, but it should be the last item in any branch in which it
4615 appears. Dollar has no special meaning in a character class.
4617 The meaning of dollar can be changed so that it matches only at the
4618 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
4619 compile time. This does not affect the \Z assertion.
4621 The meanings of the circumflex and dollar characters are changed if the
4622 PCRE_MULTILINE option is set. When this is the case, a circumflex
4623 matches immediately after internal newlines as well as at the start of
4624 the subject string. It does not match after a newline that ends the
4625 string. A dollar matches before any newlines in the string, as well as
4626 at the very end, when PCRE_MULTILINE is set. When newline is specified
4627 as the two-character sequence CRLF, isolated CR and LF characters do
4628 not indicate newlines.
4630 For example, the pattern /^abc$/ matches the subject string "def\nabc"
4631 (where \n represents a newline) in multiline mode, but not otherwise.
4632 Consequently, patterns that are anchored in single line mode because
4633 all branches start with ^ are not anchored in multiline mode, and a
4634 match for circumflex is possible when the startoffset argument of
4635 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
4636 PCRE_MULTILINE is set.
4638 Note that the sequences \A, \Z, and \z can be used to match the start
4639 and end of the subject in both modes, and if all branches of a pattern
4640 start with \A it is always anchored, whether or not PCRE_MULTILINE is
4641 set.
4646 Outside a character class, a dot in the pattern matches any one charac-
4647 ter in the subject string except (by default) a character that signi-
4648 fies the end of a line.
4650 When a line ending is defined as a single character, dot never matches
4651 that character; when the two-character sequence CRLF is used, dot does
4652 not match CR if it is immediately followed by LF, but otherwise it
4653 matches all characters (including isolated CRs and LFs). When any Uni-
4654 code line endings are being recognized, dot does not match CR or LF or
4655 any of the other line ending characters.
4657 The behaviour of dot with regard to newlines can be changed. If the
4658 PCRE_DOTALL option is set, a dot matches any one character, without
4659 exception. If the two-character sequence CRLF is present in the subject
4660 string, it takes two dots to match it.
4662 The handling of dot is entirely independent of the handling of circum-
4663 flex and dollar, the only relationship being that they both involve
4664 newlines. Dot has no special meaning in a character class.
4666 The escape sequence \N behaves like a dot, except that it is not
4667 affected by the PCRE_DOTALL option. In other words, it matches any
4668 character except one that signifies the end of a line. Perl also uses
4669 \N to match characters by name; PCRE does not support this.
4674 Outside a character class, the escape sequence \C matches any one data
4675 unit, whether or not a UTF mode is set. In the 8-bit library, one data
4676 unit is one byte; in the 16-bit library it is a 16-bit unit. Unlike a
4677 dot, \C always matches line-ending characters. The feature is provided
4678 in Perl in order to match individual bytes in UTF-8 mode, but it is
4679 unclear how it can usefully be used. Because \C breaks up characters
4680 into individual data units, matching one unit with \C in a UTF mode
4681 means that the rest of the string may start with a malformed UTF char-
4682 acter. This has undefined results, because PCRE assumes that it is
4683 dealing with valid UTF strings (and by default it checks this at the
4684 start of processing unless the PCRE_NO_UTF8_CHECK option is used).
4686 PCRE does not allow \C to appear in lookbehind assertions (described
4687 below) in a UTF mode, because this would make it impossible to calcu-
4688 late the length of the lookbehind.
4690 In general, the \C escape sequence is best avoided. However, one way of
4691 using it that avoids the problem of malformed UTF characters is to use
4692 a lookahead to check the length of the next character, as in this pat-
4693 tern, which could be used with a UTF-8 string (ignore white space and
4694 line breaks):
4696 (?| (?=[\x00-\x7f])(\C) |
4697 (?=[\x80-\x{7ff}])(\C)(\C) |
4698 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
4699 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
4701 A group that starts with (?| resets the capturing parentheses numbers
4702 in each alternative (see "Duplicate Subpattern Numbers" below). The
4703 assertions at the start of each branch check the next UTF-8 character
4704 for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
4705 character's individual bytes are then captured by the appropriate num-
4706 ber of groups.
4711 An opening square bracket introduces a character class, terminated by a
4712 closing square bracket. A closing square bracket on its own is not spe-
4713 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
4714 a lone closing square bracket causes a compile-time error. If a closing
4715 square bracket is required as a member of the class, it should be the
4716 first data character in the class (after an initial circumflex, if
4717 present) or escaped with a backslash.
4719 A character class matches a single character in the subject. In a UTF
4720 mode, the character may be more than one data unit long. A matched
4721 character must be in the set of characters defined by the class, unless
4722 the first character in the class definition is a circumflex, in which
4723 case the subject character must not be in the set defined by the class.
4724 If a circumflex is actually required as a member of the class, ensure
4725 it is not the first character, or escape it with a backslash.
4727 For example, the character class [aeiou] matches any lower case vowel,
4728 while [^aeiou] matches any character that is not a lower case vowel.
4729 Note that a circumflex is just a convenient notation for specifying the
4730 characters that are in the class by enumerating those that are not. A
4731 class that starts with a circumflex is not an assertion; it still con-
4732 sumes a character from the subject string, and therefore it fails if
4733 the current pointer is at the end of the string.
4735 In UTF-8 (UTF-16) mode, characters with values greater than 255
4736 (0xffff) can be included in a class as a literal string of data units,
4737 or by using the \x{ escaping mechanism.
4739 When caseless matching is set, any letters in a class represent both
4740 their upper case and lower case versions, so for example, a caseless
4741 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
4742 match "A", whereas a caseful version would. In a UTF mode, PCRE always
4743 understands the concept of case for characters whose values are less
4744 than 128, so caseless matching is always possible. For characters with
4745 higher values, the concept of case is supported if PCRE is compiled
4746 with Unicode property support, but not otherwise. If you want to use
4747 caseless matching in a UTF mode for characters 128 and above, you must
4748 ensure that PCRE is compiled with Unicode property support as well as
4749 with UTF support.
4751 Characters that might indicate line breaks are never treated in any
4752 special way when matching character classes, whatever line-ending
4753 sequence is in use, and whatever setting of the PCRE_DOTALL and
4754 PCRE_MULTILINE options is used. A class such as [^a] always matches one
4755 of these characters.
4757 The minus (hyphen) character can be used to specify a range of charac-
4758 ters in a character class. For example, [d-m] matches any letter
4759 between d and m, inclusive. If a minus character is required in a
4760 class, it must be escaped with a backslash or appear in a position
4761 where it cannot be interpreted as indicating a range, typically as the
4762 first or last character in the class.
4764 It is not possible to have the literal character "]" as the end charac-
4765 ter of a range. A pattern such as [W-]46] is interpreted as a class of
4766 two characters ("W" and "-") followed by a literal string "46]", so it
4767 would match "W46]" or "-46]". However, if the "]" is escaped with a
4768 backslash it is interpreted as the end of range, so [W-\]46] is inter-
4769 preted as a class containing a range followed by two other characters.
4770 The octal or hexadecimal representation of "]" can also be used to end
4771 a range.
4773 Ranges operate in the collating sequence of character values. They can
4774 also be used for characters specified numerically, for example
4775 [\000-\037]. Ranges can include any characters that are valid for the
4776 current mode.
4778 If a range that includes letters is used when caseless matching is set,
4779 it matches the letters in either case. For example, [W-c] is equivalent
4780 to [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if
4781 character tables for a French locale are in use, [\xc8-\xcb] matches
4782 accented E characters in both cases. In UTF modes, PCRE supports the
4783 concept of case for characters with values greater than 128 only when
4784 it is compiled with Unicode property support.
4786 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
4787 \w, and \W may appear in a character class, and add the characters that
4788 they match to the class. For example, [\dABCDEF] matches any hexadeci-
4789 mal digit. In UTF modes, the PCRE_UCP option affects the meanings of
4790 \d, \s, \w and their upper case partners, just as it does when they
4791 appear outside a character class, as described in the section entitled
4792 "Generic character types" above. The escape sequence \b has a different
4793 meaning inside a character class; it matches the backspace character.
4794 The sequences \B, \N, \R, and \X are not special inside a character
4795 class. Like any other unrecognized escape sequences, they are treated
4796 as the literal characters "B", "N", "R", and "X" by default, but cause
4797 an error if the PCRE_EXTRA option is set.
4799 A circumflex can conveniently be used with the upper case character
4800 types to specify a more restricted set of characters than the matching
4801 lower case type. For example, the class [^\W_] matches any letter or
4802 digit, but not underscore, whereas [\w] includes underscore. A positive
4803 character class should be read as "something OR something OR ..." and a
4804 negative class as "NOT something AND NOT something AND NOT ...".
4806 The only metacharacters that are recognized in character classes are
4807 backslash, hyphen (only where it can be interpreted as specifying a
4808 range), circumflex (only at the start), opening square bracket (only
4809 when it can be interpreted as introducing a POSIX class name - see the
4810 next section), and the terminating closing square bracket. However,
4811 escaping other non-alphanumeric characters does no harm.
4816 Perl supports the POSIX notation for character classes. This uses names
4817 enclosed by [: and :] within the enclosing square brackets. PCRE also
4818 supports this notation. For example,
4820 [01[:alpha:]%]
4822 matches "0", "1", any alphabetic character, or "%". The supported class
4823 names are:
4825 alnum letters and digits
4826 alpha letters
4827 ascii character codes 0 - 127
4828 blank space or tab only
4829 cntrl control characters
4830 digit decimal digits (same as \d)
4831 graph printing characters, excluding space
4832 lower lower case letters
4833 print printing characters, including space
4834 punct printing characters, excluding letters and digits and space
4835 space white space (not quite the same as \s)
4836 upper upper case letters
4837 word "word" characters (same as \w)
4838 xdigit hexadecimal digits
4840 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
4841 and space (32). Notice that this list includes the VT character (code
4842 11). This makes "space" different to \s, which does not include VT (for
4843 Perl compatibility).
4845 The name "word" is a Perl extension, and "blank" is a GNU extension
4846 from Perl 5.8. Another Perl extension is negation, which is indicated
4847 by a ^ character after the colon. For example,
4849 [12[:^digit:]]
4851 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4852 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4853 these are not supported, and an error is given if they are encountered.
4855 By default, in UTF modes, characters with values greater than 128 do
4856 not match any of the POSIX character classes. However, if the PCRE_UCP
4857 option is passed to pcre_compile(), some of the classes are changed so
4858 that Unicode character properties are used. This is achieved by replac-
4859 ing the POSIX classes by other sequences, as follows:
4861 [:alnum:] becomes \p{Xan}
4862 [:alpha:] becomes \p{L}
4863 [:blank:] becomes \h
4864 [:digit:] becomes \p{Nd}
4865 [:lower:] becomes \p{Ll}
4866 [:space:] becomes \p{Xps}
4867 [:upper:] becomes \p{Lu}
4868 [:word:] becomes \p{Xwd}
4870 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4871 POSIX classes are unchanged, and match only characters with code points
4872 less than 128.
4877 Vertical bar characters are used to separate alternative patterns. For
4878 example, the pattern
4880 gilbert|sullivan
4882 matches either "gilbert" or "sullivan". Any number of alternatives may
4883 appear, and an empty alternative is permitted (matching the empty
4884 string). The matching process tries each alternative in turn, from left
4885 to right, and the first one that succeeds is used. If the alternatives
4886 are within a subpattern (defined below), "succeeds" means matching the
4887 rest of the main pattern as well as the alternative in the subpattern.
4892 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4893 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4894 within the pattern by a sequence of Perl option letters enclosed
4895 between "(?" and ")". The option letters are
4897 i for PCRE_CASELESS
4899 s for PCRE_DOTALL
4900 x for PCRE_EXTENDED
4902 For example, (?im) sets caseless, multiline matching. It is also possi-
4903 ble to unset these options by preceding the letter with a hyphen, and a
4904 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4906 is also permitted. If a letter appears both before and after the
4907 hyphen, the option is unset.
4910 can be changed in the same way as the Perl-compatible options by using
4911 the characters J, U and X respectively.
4913 When one of these option changes occurs at top level (that is, not
4914 inside subpattern parentheses), the change applies to the remainder of
4915 the pattern that follows. If the change is placed right at the start of
4916 a pattern, PCRE extracts it into the global options (and it will there-
4917 fore show up in data extracted by the pcre_fullinfo() function).
4919 An option change within a subpattern (see below for a description of
4920 subpatterns) affects only that part of the subpattern that follows it,
4921 so
4923 (a(?i)b)c
4925 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4926 used). By this means, options can be made to have different settings
4927 in different parts of the pattern. Any changes made in one alternative
4928 do carry on into subsequent branches within the same subpattern. For
4929 example,
4931 (a(?i)b|c)
4933 matches "ab", "aB", "c", and "C", even though when matching "C" the
4934 first branch is abandoned before the option setting. This is because
4935 the effects of option settings happen at compile time. There would be
4936 some very weird behaviour otherwise.
4938 Note: There are other PCRE-specific options that can be set by the
4939 application when the compiling or matching functions are called. In
4940 some cases the pattern can contain special leading sequences such as
4941 (*CRLF) to override what the application has set or what has been
4942 defaulted. Details are given in the section entitled "Newline
4943 sequences" above. There are also the (*UTF8), (*UTF16), and (*UCP)
4944 leading sequences that can be used to set UTF and Unicode property
4945 modes; they are equivalent to setting the PCRE_UTF8, PCRE_UTF16, and
4946 the PCRE_UCP options, respectively.
4951 Subpatterns are delimited by parentheses (round brackets), which can be
4952 nested. Turning part of a pattern into a subpattern does two things:
4954 1. It localizes a set of alternatives. For example, the pattern
4956 cat(aract|erpillar|)
4958 matches "cataract", "caterpillar", or "cat". Without the parentheses,
4959 it would match "cataract", "erpillar" or an empty string.
4961 2. It sets up the subpattern as a capturing subpattern. This means
4962 that, when the whole pattern matches, that portion of the subject
4963 string that matched the subpattern is passed back to the caller via the
4964 ovector argument of the matching function. (This applies only to the
4965 traditional matching functions; the DFA matching functions do not sup-
4966 port capturing.)
4968 Opening parentheses are counted from left to right (starting from 1) to
4969 obtain numbers for the capturing subpatterns. For example, if the
4970 string "the red king" is matched against the pattern
4972 the ((red|white) (king|queen))
4974 the captured substrings are "red king", "red", and "king", and are num-
4975 bered 1, 2, and 3, respectively.
4977 The fact that plain parentheses fulfil two functions is not always
4978 helpful. There are often times when a grouping subpattern is required
4979 without a capturing requirement. If an opening parenthesis is followed
4980 by a question mark and a colon, the subpattern does not do any captur-
4981 ing, and is not counted when computing the number of any subsequent
4982 capturing subpatterns. For example, if the string "the white queen" is
4983 matched against the pattern
4985 the ((?:red|white) (king|queen))
4987 the captured substrings are "white queen" and "queen", and are numbered
4988 1 and 2. The maximum number of capturing subpatterns is 65535.
4990 As a convenient shorthand, if any option settings are required at the
4991 start of a non-capturing subpattern, the option letters may appear
4992 between the "?" and the ":". Thus the two patterns
4994 (?i:saturday|sunday)
4995 (?:(?i)saturday|sunday)
4997 match exactly the same set of strings. Because alternative branches are
4998 tried from left to right, and options are not reset until the end of
4999 the subpattern is reached, an option setting in one branch does affect
5000 subsequent branches, so the above patterns match "SUNDAY" as well as
5001 "Saturday".
5006 Perl 5.10 introduced a feature whereby each alternative in a subpattern
5007 uses the same numbers for its capturing parentheses. Such a subpattern
5008 starts with (?| and is itself a non-capturing subpattern. For example,
5009 consider this pattern:
5011 (?|(Sat)ur|(Sun))day
5013 Because the two alternatives are inside a (?| group, both sets of cap-
5014 turing parentheses are numbered one. Thus, when the pattern matches,
5015 you can look at captured substring number one, whichever alternative
5016 matched. This construct is useful when you want to capture part, but
5017 not all, of one of a number of alternatives. Inside a (?| group, paren-
5018 theses are numbered as usual, but the number is reset at the start of
5019 each branch. The numbers of any capturing parentheses that follow the
5020 subpattern start after the highest number used in any branch. The fol-
5021 lowing example is taken from the Perl documentation. The numbers under-
5022 neath show in which buffer the captured content will be stored.
5024 # before ---------------branch-reset----------- after
5025 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
5026 # 1 2 2 3 2 3 4
5028 A back reference to a numbered subpattern uses the most recent value
5029 that is set for that number by any subpattern. The following pattern
5030 matches "abcabc" or "defdef":
5032 /(?|(abc)|(def))\1/
5034 In contrast, a subroutine call to a numbered subpattern always refers
5035 to the first one in the pattern with the given number. The following
5036 pattern matches "abcabc" or "defabc":
5038 /(?|(abc)|(def))(?1)/
5040 If a condition test for a subpattern's having matched refers to a non-
5041 unique number, the test is true if any of the subpatterns of that num-
5042 ber have matched.
5044 An alternative approach to using this "branch reset" feature is to use
5045 duplicate named subpatterns, as described in the next section.
5050 Identifying capturing parentheses by number is simple, but it can be
5051 very hard to keep track of the numbers in complicated regular expres-
5052 sions. Furthermore, if an expression is modified, the numbers may
5053 change. To help with this difficulty, PCRE supports the naming of sub-
5054 patterns. This feature was not added to Perl until release 5.10. Python
5055 had the feature earlier, and PCRE introduced it at release 4.0, using
5056 the Python syntax. PCRE now supports both the Perl and the Python syn-
5057 tax. Perl allows identically numbered subpatterns to have different
5058 names, but PCRE does not.
5060 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
5061 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
5062 to capturing parentheses from other parts of the pattern, such as back
5063 references, recursion, and conditions, can be made by name as well as
5064 by number.
5066 Names consist of up to 32 alphanumeric characters and underscores.
5067 Named capturing parentheses are still allocated numbers as well as
5068 names, exactly as if the names were not present. The PCRE API provides
5069 function calls for extracting the name-to-number translation table from
5070 a compiled pattern. There is also a convenience function for extracting
5071 a captured substring by name.
5073 By default, a name must be unique within a pattern, but it is possible
5074 to relax this constraint by setting the PCRE_DUPNAMES option at compile
5075 time. (Duplicate names are also always permitted for subpatterns with
5076 the same number, set up as described in the previous section.) Dupli-
5077 cate names can be useful for patterns where only one instance of the
5078 named parentheses can match. Suppose you want to match the name of a
5079 weekday, either as a 3-letter abbreviation or as the full name, and in
5080 both cases you want to extract the abbreviation. This pattern (ignoring
5081 the line breaks) does the job:
5083 (?<DN>Mon|Fri|Sun)(?:day)?|
5084 (?<DN>Tue)(?:sday)?|
5085 (?<DN>Wed)(?:nesday)?|
5086 (?<DN>Thu)(?:rsday)?|
5087 (?<DN>Sat)(?:urday)?
5089 There are five capturing substrings, but only one is ever set after a
5090 match. (An alternative way of solving this problem is to use a "branch
5091 reset" subpattern, as described in the previous section.)
5093 The convenience function for extracting the data by name returns the
5094 substring for the first (and in this example, the only) subpattern of
5095 that name that matched. This saves searching to find which numbered
5096 subpattern it was.
5098 If you make a back reference to a non-unique named subpattern from
5099 elsewhere in the pattern, the one that corresponds to the first occur-
5100 rence of the name is used. In the absence of duplicate numbers (see the
5101 previous section) this is the one with the lowest number. If you use a
5102 named reference in a condition test (see the section about conditions
5103 below), either to check whether a subpattern has matched, or to check
5104 for recursion, all subpatterns with the same name are tested. If the
5105 condition is true for any one of them, the overall condition is true.
5106 This is the same behaviour as testing by number. For further details of
5107 the interfaces for handling named subpatterns, see the pcreapi documen-
5108 tation.
5110 Warning: You cannot use different names to distinguish between two sub-
5111 patterns with the same number because PCRE uses only the numbers when
5112 matching. For this reason, an error is given at compile time if differ-
5113 ent names are given to subpatterns with the same number. However, you
5114 can give the same name to subpatterns with the same number, even when
5115 PCRE_DUPNAMES is not set.
5120 Repetition is specified by quantifiers, which can follow any of the
5121 following items:
5123 a literal data character
5124 the dot metacharacter
5125 the \C escape sequence
5126 the \X escape sequence
5127 the \R escape sequence
5128 an escape such as \d or \pL that matches a single character
5129 a character class
5130 a back reference (see next section)
5131 a parenthesized subpattern (including assertions)
5132 a subroutine call to a subpattern (recursive or otherwise)
5134 The general repetition quantifier specifies a minimum and maximum num-
5135 ber of permitted matches, by giving the two numbers in curly brackets
5136 (braces), separated by a comma. The numbers must be less than 65536,
5137 and the first must be less than or equal to the second. For example:
5139 z{2,4}
5141 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
5142 special character. If the second number is omitted, but the comma is
5143 present, there is no upper limit; if the second number and the comma
5144 are both omitted, the quantifier specifies an exact number of required
5145 matches. Thus
5147 [aeiou]{3,}
5149 matches at least 3 successive vowels, but may match many more, while
5151 \d{8}
5153 matches exactly 8 digits. An opening curly bracket that appears in a
5154 position where a quantifier is not allowed, or one that does not match
5155 the syntax of a quantifier, is taken as a literal character. For exam-
5156 ple, {,6} is not a quantifier, but a literal string of four characters.
5158 In UTF modes, quantifiers apply to characters rather than to individual
5159 data units. Thus, for example, \x{100}{2} matches two characters, each
5160 of which is represented by a two-byte sequence in a UTF-8 string. Simi-
5161 larly, \X{3} matches three Unicode extended sequences, each of which
5162 may be several data units long (and they may be of different lengths).
5164 The quantifier {0} is permitted, causing the expression to behave as if
5165 the previous item and the quantifier were not present. This may be use-
5166 ful for subpatterns that are referenced as subroutines from elsewhere
5167 in the pattern (but see also the section entitled "Defining subpatterns
5168 for use by reference only" below). Items other than subpatterns that
5169 have a {0} quantifier are omitted from the compiled pattern.
5171 For convenience, the three most common quantifiers have single-charac-
5172 ter abbreviations:
5174 * is equivalent to {0,}
5175 + is equivalent to {1,}
5176 ? is equivalent to {0,1}
5178 It is possible to construct infinite loops by following a subpattern
5179 that can match no characters with a quantifier that has no upper limit,
5180 for example:
5182 (a?)*
5184 Earlier versions of Perl and PCRE used to give an error at compile time
5185 for such patterns. However, because there are cases where this can be
5186 useful, such patterns are now accepted, but if any repetition of the
5187 subpattern does in fact match no characters, the loop is forcibly bro-
5188 ken.
5190 By default, the quantifiers are "greedy", that is, they match as much
5191 as possible (up to the maximum number of permitted times), without
5192 causing the rest of the pattern to fail. The classic example of where
5193 this gives problems is in trying to match comments in C programs. These
5194 appear between /* and */ and within the comment, individual * and /
5195 characters may appear. An attempt to match C comments by applying the
5196 pattern
5198 /\*.*\*/
5200 to the string
5202 /* first comment */ not comment /* second comment */
5204 fails, because it matches the entire string owing to the greediness of
5205 the .* item.
5207 However, if a quantifier is followed by a question mark, it ceases to
5208 be greedy, and instead matches the minimum number of times possible, so
5209 the pattern
5211 /\*.*?\*/
5213 does the right thing with the C comments. The meaning of the various
5214 quantifiers is not otherwise changed, just the preferred number of
5215 matches. Do not confuse this use of question mark with its use as a
5216 quantifier in its own right. Because it has two uses, it can sometimes
5217 appear doubled, as in
5219 \d??\d
5221 which matches one digit by preference, but can match two if that is the
5222 only way the rest of the pattern matches.
5224 If the PCRE_UNGREEDY option is set (an option that is not available in
5225 Perl), the quantifiers are not greedy by default, but individual ones
5226 can be made greedy by following them with a question mark. In other
5227 words, it inverts the default behaviour.
5229 When a parenthesized subpattern is quantified with a minimum repeat
5230 count that is greater than 1 or with a limited maximum, more memory is
5231 required for the compiled pattern, in proportion to the size of the
5232 minimum or maximum.
5234 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
5235 alent to Perl's /s) is set, thus allowing the dot to match newlines,
5236 the pattern is implicitly anchored, because whatever follows will be
5237 tried against every character position in the subject string, so there
5238 is no point in retrying the overall match at any position after the
5239 first. PCRE normally treats such a pattern as though it were preceded
5240 by \A.
5242 In cases where it is known that the subject string contains no new-
5243 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
5244 mization, or alternatively using ^ to indicate anchoring explicitly.
5246 However, there is one situation where the optimization cannot be used.
5247 When .* is inside capturing parentheses that are the subject of a back
5248 reference elsewhere in the pattern, a match at the start may fail where
5249 a later one succeeds. Consider, for example:
5251 (.*)abc\1
5253 If the subject is "xyz123abc123" the match point is the fourth charac-
5254 ter. For this reason, such a pattern is not implicitly anchored.
5256 When a capturing subpattern is repeated, the value captured is the sub-
5257 string that matched the final iteration. For example, after
5259 (tweedle[dume]{3}\s*)+
5261 has matched "tweedledum tweedledee" the value of the captured substring
5262 is "tweedledee". However, if there are nested capturing subpatterns,
5263 the corresponding captured values may have been set in previous itera-
5264 tions. For example, after
5266 /(a|(b))+/
5268 matches "aba" the value of the second captured substring is "b".
5273 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
5274 repetition, failure of what follows normally causes the repeated item
5275 to be re-evaluated to see if a different number of repeats allows the
5276 rest of the pattern to match. Sometimes it is useful to prevent this,
5277 either to change the nature of the match, or to cause it fail earlier
5278 than it otherwise might, when the author of the pattern knows there is
5279 no point in carrying on.
5281 Consider, for example, the pattern \d+foo when applied to the subject
5282 line
5284 123456bar
5286 After matching all 6 digits and then failing to match "foo", the normal
5287 action of the matcher is to try again with only 5 digits matching the
5288 \d+ item, and then with 4, and so on, before ultimately failing.
5289 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
5290 the means for specifying that once a subpattern has matched, it is not
5291 to be re-evaluated in this way.
5293 If we use atomic grouping for the previous example, the matcher gives
5294 up immediately on failing to match "foo" the first time. The notation
5295 is a kind of special parenthesis, starting with (?> as in this example:
5297 (?>\d+)foo
5299 This kind of parenthesis "locks up" the part of the pattern it con-
5300 tains once it has matched, and a failure further into the pattern is
5301 prevented from backtracking into it. Backtracking past it to previous
5302 items, however, works as normal.
5304 An alternative description is that a subpattern of this type matches
5305 the string of characters that an identical standalone pattern would
5306 match, if anchored at the current point in the subject string.
5308 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
5309 such as the above example can be thought of as a maximizing repeat that
5310 must swallow everything it can. So, while both \d+ and \d+? are pre-
5311 pared to adjust the number of digits they match in order to make the
5312 rest of the pattern match, (?>\d+) can only match an entire sequence of
5313 digits.
5315 Atomic groups in general can of course contain arbitrarily complicated
5316 subpatterns, and can be nested. However, when the subpattern for an
5317 atomic group is just a single repeated item, as in the example above, a
5318 simpler notation, called a "possessive quantifier" can be used. This
5319 consists of an additional + character following a quantifier. Using
5320 this notation, the previous example can be rewritten as
5322 \d++foo
5324 Note that a possessive quantifier can be used with an entire group, for
5325 example:
5327 (abc|xyz){2,3}+
5329 Possessive quantifiers are always greedy; the setting of the
5330 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
5331 simpler forms of atomic group. However, there is no difference in the
5332 meaning of a possessive quantifier and the equivalent atomic group,
5333 though there may be a performance difference; possessive quantifiers
5334 should be slightly faster.
5336 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
5337 tax. Jeffrey Friedl originated the idea (and the name) in the first
5338 edition of his book. Mike McCloskey liked it, so implemented it when he
5339 built Sun's Java package, and PCRE copied it from there. It ultimately
5340 found its way into Perl at release 5.10.
5342 PCRE has an optimization that automatically "possessifies" certain sim-
5343 ple pattern constructs. For example, the sequence A+B is treated as
5344 A++B because there is no point in backtracking into a sequence of A's
5345 when B must follow.
5347 When a pattern contains an unlimited repeat inside a subpattern that
5348 can itself be repeated an unlimited number of times, the use of an
5349 atomic group is the only way to avoid some failing matches taking a
5350 very long time indeed. The pattern
5352 (\D+|<\d+>)*[!?]
5354 matches an unlimited number of substrings that either consist of non-
5355 digits, or digits enclosed in <>, followed by either ! or ?. When it
5356 matches, it runs quickly. However, if it is applied to
5358 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
5360 it takes a long time before reporting failure. This is because the
5361 string can be divided between the internal \D+ repeat and the external
5362 * repeat in a large number of ways, and all have to be tried. (The
5363 example uses [!?] rather than a single character at the end, because
5364 both PCRE and Perl have an optimization that allows for fast failure
5365 when a single character is used. They remember the last single charac-
5366 ter that is required for a match, and fail early if it is not present
5367 in the string.) If the pattern is changed so that it uses an atomic
5368 group, like this:
5370 ((?>\D+)|<\d+>)*[!?]
5372 sequences of non-digits cannot be broken, and failure happens quickly.
5377 Outside a character class, a backslash followed by a digit greater than
5378 0 (and possibly further digits) is a back reference to a capturing sub-
5379 pattern earlier (that is, to its left) in the pattern, provided there
5380 have been that many previous capturing left parentheses.
5382 However, if the decimal number following the backslash is less than 10,
5383 it is always taken as a back reference, and causes an error only if
5384 there are not that many capturing left parentheses in the entire pat-
5385 tern. In other words, the parentheses that are referenced need not be
5386 to the left of the reference for numbers less than 10. A "forward back
5387 reference" of this type can make sense when a repetition is involved
5388 and the subpattern to the right has participated in an earlier itera-
5389 tion.
5391 It is not possible to have a numerical "forward back reference" to a
5392 subpattern whose number is 10 or more using this syntax because a
5393 sequence such as \50 is interpreted as a character defined in octal.
5394 See the subsection entitled "Non-printing characters" above for further
5395 details of the handling of digits following a backslash. There is no
5396 such problem when named parentheses are used. A back reference to any
5397 subpattern is possible using named parentheses (see below).
5399 Another way of avoiding the ambiguity inherent in the use of digits
5400 following a backslash is to use the \g escape sequence. This escape
5401 must be followed by an unsigned number or a negative number, optionally
5402 enclosed in braces. These examples are all identical:
5404 (ring), \1
5405 (ring), \g1
5406 (ring), \g{1}
5408 An unsigned number specifies an absolute reference without the ambigu-
5409 ity that is present in the older syntax. It is also useful when literal
5410 digits follow the reference. A negative number is a relative reference.
5411 Consider this example:
5413 (abc(def)ghi)\g{-1}
5415 The sequence \g{-1} is a reference to the most recently started captur-
5416 ing subpattern before \g, that is, is it equivalent to \2 in this exam-
5417 ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
5418 references can be helpful in long patterns, and also in patterns that
5419 are created by joining together fragments that contain references
5420 within themselves.
5422 A back reference matches whatever actually matched the capturing sub-
5423 pattern in the current subject string, rather than anything matching
5424 the subpattern itself (see "Subpatterns as subroutines" below for a way
5425 of doing that). So the pattern
5427 (sens|respons)e and \1ibility
5429 matches "sense and sensibility" and "response and responsibility", but
5430 not "sense and responsibility". If caseful matching is in force at the
5431 time of the back reference, the case of letters is relevant. For exam-
5432 ple,
5434 ((?i)rah)\s+\1
5436 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
5437 original capturing subpattern is matched caselessly.
5439 There are several different ways of writing back references to named
5440 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
5441 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
5442 unified back reference syntax, in which \g can be used for both numeric
5443 and named references, is also supported. We could rewrite the above
5444 example in any of the following ways:
5446 (?<p1>(?i)rah)\s+\k<p1>
5447 (?'p1'(?i)rah)\s+\k{p1}
5448 (?P<p1>(?i)rah)\s+(?P=p1)
5449 (?<p1>(?i)rah)\s+\g{p1}
5451 A subpattern that is referenced by name may appear in the pattern
5452 before or after the reference.
5454 There may be more than one back reference to the same subpattern. If a
5455 subpattern has not actually been used in a particular match, any back
5456 references to it always fail by default. For example, the pattern
5458 (a|(bc))\2
5460 always fails if it starts to match "a" rather than "bc". However, if
5461 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
5462 ence to an unset value matches an empty string.
5464 Because there may be many capturing parentheses in a pattern, all dig-
5465 its following a backslash are taken as part of a potential back refer-
5466 ence number. If the pattern continues with a digit character, some
5467 delimiter must be used to terminate the back reference. If the
5468 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
5469 syntax or an empty comment (see "Comments" below) can be used.
5471 Recursive back references
5473 A back reference that occurs inside the parentheses to which it refers
5474 fails when the subpattern is first used, so, for example, (a\1) never
5475 matches. However, such references can be useful inside repeated sub-
5476 patterns. For example, the pattern
5478 (a|b\1)+
5480 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
5481 ation of the subpattern, the back reference matches the character
5482 string corresponding to the previous iteration. In order for this to
5483 work, the pattern must be such that the first iteration does not need
5484 to match the back reference. This can be done using alternation, as in
5485 the example above, or by a quantifier with a minimum of zero.
5487 Back references of this type cause the group that they reference to be
5488 treated as an atomic group. Once the whole group has been matched, a
5489 subsequent matching failure cannot cause backtracking into the middle
5490 of the group.
5495 An assertion is a test on the characters following or preceding the
5496 current matching point that does not actually consume any characters.
5497 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
5498 described above.
5500 More complicated assertions are coded as subpatterns. There are two
5501 kinds: those that look ahead of the current position in the subject
5502 string, and those that look behind it. An assertion subpattern is
5503 matched in the normal way, except that it does not cause the current
5504 matching position to be changed.
5506 Assertion subpatterns are not capturing subpatterns. If such an asser-
5507 tion contains capturing subpatterns within it, these are counted for
5508 the purposes of numbering the capturing subpatterns in the whole pat-
5509 tern. However, substring capturing is carried out only for positive
5510 assertions, because it does not make sense for negative assertions.
5512 For compatibility with Perl, assertion subpatterns may be repeated;
5513 though it makes no sense to assert the same thing several times, the
5514 side effect of capturing parentheses may occasionally be useful. In
5515 practice, there only three cases:
5517 (1) If the quantifier is {0}, the assertion is never obeyed during
5518 matching. However, it may contain internal capturing parenthesized
5519 groups that are called from elsewhere via the subroutine mechanism.
5521 (2) If quantifier is {0,n} where n is greater than zero, it is treated
5522 as if it were {0,1}. At run time, the rest of the pattern match is
5523 tried with and without the assertion, the order depending on the greed-
5524 iness of the quantifier.
5526 (3) If the minimum repetition is greater than zero, the quantifier is
5527 ignored. The assertion is obeyed just once when encountered during
5528 matching.
5530 Lookahead assertions
5532 Lookahead assertions start with (?= for positive assertions and (?! for
5533 negative assertions. For example,
5535 \w+(?=;)
5537 matches a word followed by a semicolon, but does not include the semi-
5538 colon in the match, and
5540 foo(?!bar)
5542 matches any occurrence of "foo" that is not followed by "bar". Note
5543 that the apparently similar pattern
5545 (?!foo)bar
5547 does not find an occurrence of "bar" that is preceded by something
5548 other than "foo"; it finds any occurrence of "bar" whatsoever, because
5549 the assertion (?!foo) is always true when the next three characters are
5550 "bar". A lookbehind assertion is needed to achieve the other effect.
5552 If you want to force a matching failure at some point in a pattern, the
5553 most convenient way to do it is with (?!) because an empty string
5554 always matches, so an assertion that requires there not to be an empty
5555 string must always fail. The backtracking control verb (*FAIL) or (*F)
5556 is a synonym for (?!).
5558 Lookbehind assertions
5560 Lookbehind assertions start with (?<= for positive assertions and (?<!
5561 for negative assertions. For example,
5563 (?<!foo)bar
5565 does find an occurrence of "bar" that is not preceded by "foo". The
5566 contents of a lookbehind assertion are restricted such that all the
5567 strings it matches must have a fixed length. However, if there are sev-
5568 eral top-level alternatives, they do not all have to have the same
5569 fixed length. Thus
5571 (?<=bullock|donkey)
5573 is permitted, but
5575 (?<!dogs?|cats?)
5577 causes an error at compile time. Branches that match different length
5578 strings are permitted only at the top level of a lookbehind assertion.
5579 This is an extension compared with Perl, which requires all branches to
5580 match the same length of string. An assertion such as
5582 (?<=ab(c|de))
5584 is not permitted, because its single top-level branch can match two
5585 different lengths, but it is acceptable to PCRE if rewritten to use two
5586 top-level branches:
5588 (?<=abc|abde)
5590 In some cases, the escape sequence \K (see above) can be used instead
5591 of a lookbehind assertion to get round the fixed-length restriction.
5593 The implementation of lookbehind assertions is, for each alternative,
5594 to temporarily move the current position back by the fixed length and
5595 then try to match. If there are insufficient characters before the cur-
5596 rent position, the assertion fails.
5598 In a UTF mode, PCRE does not allow the \C escape (which matches a sin-
5599 gle data unit even in a UTF mode) to appear in lookbehind assertions,
5600 because it makes it impossible to calculate the length of the lookbe-
5601 hind. The \X and \R escapes, which can match different numbers of data
5602 units, are also not permitted.
5604 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
5605 lookbehinds, as long as the subpattern matches a fixed-length string.
5606 Recursion, however, is not supported.
5608 Possessive quantifiers can be used in conjunction with lookbehind
5609 assertions to specify efficient matching of fixed-length strings at the
5610 end of subject strings. Consider a simple pattern such as
5612 abcd$
5614 when applied to a long string that does not match. Because matching
5615 proceeds from left to right, PCRE will look for each "a" in the subject
5616 and then see if what follows matches the rest of the pattern. If the
5617 pattern is specified as
5619 ^.*abcd$
5621 the initial .* matches the entire string at first, but when this fails
5622 (because there is no following "a"), it backtracks to match all but the
5623 last character, then all but the last two characters, and so on. Once
5624 again the search for "a" covers the entire string, from right to left,
5625 so we are no better off. However, if the pattern is written as
5627 ^.*+(?<=abcd)
5629 there can be no backtracking for the .*+ item; it can match only the
5630 entire string. The subsequent lookbehind assertion does a single test
5631 on the last four characters. If it fails, the match fails immediately.
5632 For long strings, this approach makes a significant difference to the
5633 processing time.
5635 Using multiple assertions
5637 Several assertions (of any sort) may occur in succession. For example,
5639 (?<=\d{3})(?<!999)foo
5641 matches "foo" preceded by three digits that are not "999". Notice that
5642 each of the assertions is applied independently at the same point in
5643 the subject string. First there is a check that the previous three
5644 characters are all digits, and then there is a check that the same
5645 three characters are not "999". This pattern does not match "foo" pre-
5646 ceded by six characters, the first of which are digits and the last
5647 three of which are not "999". For example, it doesn't match "123abc-
5648 foo". A pattern to do that is
5650 (?<=\d{3}...)(?<!999)foo
5652 This time the first assertion looks at the preceding six characters,
5653 checking that the first three are digits, and then the second assertion
5654 checks that the preceding three characters are not "999".
5656 Assertions can be nested in any combination. For example,
5658 (?<=(?<!foo)bar)baz
5660 matches an occurrence of "baz" that is preceded by "bar" which in turn
5661 is not preceded by "foo", while
5663 (?<=\d{3}(?!999)...)foo
5665 is another pattern that matches "foo" preceded by three digits and any
5666 three characters that are not "999".
5671 It is possible to cause the matching process to obey a subpattern con-
5672 ditionally or to choose between two alternative subpatterns, depending
5673 on the result of an assertion, or whether a specific capturing subpat-
5674 tern has already been matched. The two possible forms of conditional
5675 subpattern are:
5677 (?(condition)yes-pattern)
5678 (?(condition)yes-pattern|no-pattern)
5680 If the condition is satisfied, the yes-pattern is used; otherwise the
5681 no-pattern (if present) is used. If there are more than two alterna-
5682 tives in the subpattern, a compile-time error occurs. Each of the two
5683 alternatives may itself contain nested subpatterns of any form, includ-
5684 ing conditional subpatterns; the restriction to two alternatives
5685 applies only at the level of the condition. This pattern fragment is an
5686 example where the alternatives are complex:
5688 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
5691 There are four kinds of condition: references to subpatterns, refer-
5692 ences to recursion, a pseudo-condition called DEFINE, and assertions.
5694 Checking for a used subpattern by number
5696 If the text between the parentheses consists of a sequence of digits,
5697 the condition is true if a capturing subpattern of that number has pre-
5698 viously matched. If there is more than one capturing subpattern with
5699 the same number (see the earlier section about duplicate subpattern
5700 numbers), the condition is true if any of them have matched. An alter-
5701 native notation is to precede the digits with a plus or minus sign. In
5702 this case, the subpattern number is relative rather than absolute. The
5703 most recently opened parentheses can be referenced by (?(-1), the next
5704 most recent by (?(-2), and so on. Inside loops it can also make sense
5705 to refer to subsequent groups. The next parentheses to be opened can be
5706 referenced as (?(+1), and so on. (The value zero in any of these forms
5707 is not used; it provokes a compile-time error.)
5709 Consider the following pattern, which contains non-significant white
5710 space to make it more readable (assume the PCRE_EXTENDED option) and to
5711 divide it into three parts for ease of discussion:
5713 ( \( )? [^()]+ (?(1) \) )
5715 The first part matches an optional opening parenthesis, and if that
5716 character is present, sets it as the first captured substring. The sec-
5717 ond part matches one or more characters that are not parentheses. The
5718 third part is a conditional subpattern that tests whether or not the
5719 first set of parentheses matched. If they did, that is, if subject
5720 started with an opening parenthesis, the condition is true, and so the
5721 yes-pattern is executed and a closing parenthesis is required. Other-
5722 wise, since no-pattern is not present, the subpattern matches nothing.
5723 In other words, this pattern matches a sequence of non-parentheses,
5724 optionally enclosed in parentheses.
5726 If you were embedding this pattern in a larger one, you could use a
5727 relative reference:
5729 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
5731 This makes the fragment independent of the parentheses in the larger
5732 pattern.
5734 Checking for a used subpattern by name
5736 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
5737 used subpattern by name. For compatibility with earlier versions of
5738 PCRE, which had this facility before Perl, the syntax (?(name)...) is
5739 also recognized. However, there is a possible ambiguity with this syn-
5740 tax, because subpattern names may consist entirely of digits. PCRE
5741 looks first for a named subpattern; if it cannot find one and the name
5742 consists entirely of digits, PCRE looks for a subpattern of that num-
5743 ber, which must be greater than zero. Using subpattern names that con-
5744 sist entirely of digits is not recommended.
5746 Rewriting the above example to use a named subpattern gives this:
5748 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
5750 If the name used in a condition of this kind is a duplicate, the test
5751 is applied to all subpatterns of the same name, and is true if any one
5752 of them has matched.
5754 Checking for pattern recursion
5756 If the condition is the string (R), and there is no subpattern with the
5757 name R, the condition is true if a recursive call to the whole pattern
5758 or any subpattern has been made. If digits or a name preceded by amper-
5759 sand follow the letter R, for example:
5761 (?(R3)...) or (?(R&name)...)
5763 the condition is true if the most recent recursion is into a subpattern
5764 whose number or name is given. This condition does not check the entire
5765 recursion stack. If the name used in a condition of this kind is a
5766 duplicate, the test is applied to all subpatterns of the same name, and
5767 is true if any one of them is the most recent recursion.
5769 At "top level", all these recursion test conditions are false. The
5770 syntax for recursive patterns is described below.
5772 Defining subpatterns for use by reference only
5774 If the condition is the string (DEFINE), and there is no subpattern
5775 with the name DEFINE, the condition is always false. In this case,
5776 there may be only one alternative in the subpattern. It is always
5777 skipped if control reaches this point in the pattern; the idea of
5778 DEFINE is that it can be used to define subroutines that can be refer-
5779 enced from elsewhere. (The use of subroutines is described below.) For
5780 example, a pattern to match an IPv4 address such as ""
5781 could be written like this (ignore whitespace and line breaks):
5783 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
5784 \b (?&byte) (\.(?&byte)){3} \b
5786 The first part of the pattern is a DEFINE group inside which a another
5787 group named "byte" is defined. This matches an individual component of
5788 an IPv4 address (a number less than 256). When matching takes place,
5789 this part of the pattern is skipped because DEFINE acts like a false
5790 condition. The rest of the pattern uses references to the named group
5791 to match the four dot-separated components of an IPv4 address, insist-
5792 ing on a word boundary at each end.
5794 Assertion conditions
5796 If the condition is not in any of the above formats, it must be an
5797 assertion. This may be a positive or negative lookahead or lookbehind
5798 assertion. Consider this pattern, again containing non-significant
5799 white space, and with the two alternatives on the second line:
5801 (?(?=[^a-z]*[a-z])
5802 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
5804 The condition is a positive lookahead assertion that matches an
5805 optional sequence of non-letters followed by a letter. In other words,
5806 it tests for the presence of at least one letter in the subject. If a
5807 letter is found, the subject is matched against the first alternative;
5808 otherwise it is matched against the second. This pattern matches
5809 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
5810 letters and dd are digits.
5815 There are two ways of including comments in patterns that are processed
5816 by PCRE. In both cases, the start of the comment must not be in a char-
5817 acter class, nor in the middle of any other sequence of related charac-
5818 ters such as (?: or a subpattern name or number. The characters that
5819 make up a comment play no part in the pattern matching.
5821 The sequence (?# marks the start of a comment that continues up to the
5822 next closing parenthesis. Nested parentheses are not permitted. If the
5823 PCRE_EXTENDED option is set, an unescaped # character also introduces a
5824 comment, which in this case continues to immediately after the next
5825 newline character or character sequence in the pattern. Which charac-
5826 ters are interpreted as newlines is controlled by the options passed to
5827 a compiling function or by a special sequence at the start of the pat-
5828 tern, as described in the section entitled "Newline conventions" above.
5829 Note that the end of this type of comment is a literal newline sequence
5830 in the pattern; escape sequences that happen to represent a newline do
5831 not count. For example, consider this pattern when PCRE_EXTENDED is
5832 set, and the default newline convention is in force:
5834 abc #comment \n still comment
5836 On encountering the # character, pcre_compile() skips along, looking
5837 for a newline in the pattern. The sequence \n is still literal at this
5838 stage, so it does not terminate the comment. Only an actual character
5839 with the code value 0x0a (the default newline) does so.
5844 Consider the problem of matching a string in parentheses, allowing for
5845 unlimited nested parentheses. Without the use of recursion, the best
5846 that can be done is to use a pattern that matches up to some fixed
5847 depth of nesting. It is not possible to handle an arbitrary nesting
5848 depth.
5850 For some time, Perl has provided a facility that allows regular expres-
5851 sions to recurse (amongst other things). It does this by interpolating
5852 Perl code in the expression at run time, and the code can refer to the
5853 expression itself. A Perl pattern using code interpolation to solve the
5854 parentheses problem can be created like this:
5856 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
5858 The (?p{...}) item interpolates Perl code at run time, and in this case
5859 refers recursively to the pattern in which it appears.
5861 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
5862 it supports special syntax for recursion of the entire pattern, and
5863 also for individual subpattern recursion. After its introduction in
5864 PCRE and Python, this kind of recursion was subsequently introduced
5865 into Perl at release 5.10.
5867 A special item that consists of (? followed by a number greater than
5868 zero and a closing parenthesis is a recursive subroutine call of the
5869 subpattern of the given number, provided that it occurs inside that
5870 subpattern. (If not, it is a non-recursive subroutine call, which is
5871 described in the next section.) The special item (?R) or (?0) is a
5872 recursive call of the entire regular expression.
5874 This PCRE pattern solves the nested parentheses problem (assume the
5875 PCRE_EXTENDED option is set so that white space is ignored):
5877 \( ( [^()]++ | (?R) )* \)
5879 First it matches an opening parenthesis. Then it matches any number of
5880 substrings which can either be a sequence of non-parentheses, or a
5881 recursive match of the pattern itself (that is, a correctly parenthe-
5882 sized substring). Finally there is a closing parenthesis. Note the use
5883 of a possessive quantifier to avoid backtracking into sequences of non-
5884 parentheses.
5886 If this were part of a larger pattern, you would not want to recurse
5887 the entire pattern, so instead you could use this:
5889 ( \( ( [^()]++ | (?1) )* \) )
5891 We have put the pattern into parentheses, and caused the recursion to
5892 refer to them instead of the whole pattern.
5894 In a larger pattern, keeping track of parenthesis numbers can be
5895 tricky. This is made easier by the use of relative references. Instead
5896 of (?1) in the pattern above you can write (?-2) to refer to the second
5897 most recently opened parentheses preceding the recursion. In other
5898 words, a negative number counts capturing parentheses leftwards from
5899 the point at which it is encountered.
5901 It is also possible to refer to subsequently opened parentheses, by
5902 writing references such as (?+2). However, these cannot be recursive
5903 because the reference is not inside the parentheses that are refer-
5904 enced. They are always non-recursive subroutine calls, as described in
5905 the next section.
5907 An alternative approach is to use named parentheses instead. The Perl
5908 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5909 supported. We could rewrite the above example as follows:
5911 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5913 If there is more than one subpattern with the same name, the earliest
5914 one is used.
5916 This particular example pattern that we have been looking at contains
5917 nested unlimited repeats, and so the use of a possessive quantifier for
5918 matching strings of non-parentheses is important when applying the pat-
5919 tern to strings that do not match. For example, when this pattern is
5920 applied to
5922 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5924 it yields "no match" quickly. However, if a possessive quantifier is
5925 not used, the match runs for a very long time indeed because there are
5926 so many different ways the + and * repeats can carve up the subject,
5927 and all have to be tested before failure can be reported.
5929 At the end of a match, the values of capturing parentheses are those
5930 from the outermost level. If you want to obtain intermediate values, a
5931 callout function can be used (see below and the pcrecallout documenta-
5932 tion). If the pattern above is matched against
5934 (ab(cd)ef)
5936 the value for the inner capturing parentheses (numbered 2) is "ef",
5937 which is the last value taken on at the top level. If a capturing sub-
5938 pattern is not matched at the top level, its final captured value is
5939 unset, even if it was (temporarily) set at a deeper level during the
5940 matching process.
5942 If there are more than 15 capturing parentheses in a pattern, PCRE has
5943 to obtain extra memory to store data during a recursion, which it does
5944 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5945 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5947 Do not confuse the (?R) item with the condition (R), which tests for
5948 recursion. Consider this pattern, which matches text in angle brack-
5949 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5950 brackets (that is, when recursing), whereas any characters are permit-
5951 ted at the outer level.
5953 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5955 In this pattern, (?(R) is the start of a conditional subpattern, with
5956 two different alternatives for the recursive and non-recursive cases.
5957 The (?R) item is the actual recursive call.
5959 Differences in recursion processing between PCRE and Perl
5961 Recursion processing in PCRE differs from Perl in two important ways.
5962 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5963 always treated as an atomic group. That is, once it has matched some of
5964 the subject string, it is never re-entered, even if it contains untried
5965 alternatives and there is a subsequent matching failure. This can be
5966 illustrated by the following pattern, which purports to match a palin-
5967 dromic string that contains an odd number of characters (for example,
5968 "a", "aba", "abcba", "abcdcba"):
5970 ^(.|(.)(?1)\2)$
5972 The idea is that it either matches a single character, or two identical
5973 characters surrounding a sub-palindrome. In Perl, this pattern works;
5974 in PCRE it does not if the pattern is longer than three characters.
5975 Consider the subject string "abcba":
5977 At the top level, the first character is matched, but as it is not at
5978 the end of the string, the first alternative fails; the second alterna-
5979 tive is taken and the recursion kicks in. The recursive call to subpat-
5980 tern 1 successfully matches the next character ("b"). (Note that the
5981 beginning and end of line tests are not part of the recursion).
5983 Back at the top level, the next character ("c") is compared with what
5984 subpattern 2 matched, which was "a". This fails. Because the recursion
5985 is treated as an atomic group, there are now no backtracking points,
5986 and so the entire match fails. (Perl is able, at this point, to re-
5987 enter the recursion and try the second alternative.) However, if the
5988 pattern is written with the alternatives in the other order, things are
5989 different:
5991 ^((.)(?1)\2|.)$
5993 This time, the recursing alternative is tried first, and continues to
5994 recurse until it runs out of characters, at which point the recursion
5995 fails. But this time we do have another alternative to try at the
5996 higher level. That is the big difference: in the previous case the
5997 remaining alternative is at a deeper recursion level, which PCRE cannot
5998 use.
6000 To change the pattern so that it matches all palindromic strings, not
6001 just those with an odd number of characters, it is tempting to change
6002 the pattern to this:
6004 ^((.)(?1)\2|.?)$
6006 Again, this works in Perl, but not in PCRE, and for the same reason.
6007 When a deeper recursion has matched a single character, it cannot be
6008 entered again in order to match an empty string. The solution is to
6009 separate the two cases, and write out the odd and even cases as alter-
6010 natives at the higher level:
6012 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
6014 If you want to match typical palindromic phrases, the pattern has to
6015 ignore all non-word characters, which can be done like this: