ViewVC logotype

Contents of /code/trunk/doc/pcrepattern.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 517 - (show annotations)
Wed May 5 10:44:20 2010 UTC (10 years, 7 months ago) by ph10
File size: 107225 byte(s)
Add new special properties Xan, Xps, Xsp, Xwd to help with \w etc.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13 also supports some alternative regular expression syntax (which does not
14 conflict with the Perl syntax) in order to provide some compatibility with
15 regular expressions in Python, .NET, and Oniguruma.
16 .P
17 Perl's regular expressions are described in its own documentation, and
18 regular expressions in general are covered in a number of books, some of which
19 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20 published by O'Reilly, covers regular expressions in great detail. This
21 description of PCRE's regular expressions is intended as reference material.
22 .P
23 The original operation of PCRE was on strings of one-byte characters. However,
24 there is now also support for UTF-8 character strings. To use this,
25 PCRE must be built to include UTF-8 support, and you must call
26 \fBpcre_compile()\fP or \fBpcre_compile2()\fP with the PCRE_UTF8 option. There
27 is also a special sequence that can be given at the start of a pattern:
28 .sp
29 (*UTF8)
30 .sp
31 Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
32 option. This feature is not Perl-compatible. How setting UTF-8 mode affects
33 pattern matching is mentioned in several places below. There is also a summary
34 of UTF-8 features in the
35 .\" HTML <a href="pcre.html#utf8support">
36 .\" </a>
37 section on UTF-8 support
38 .\"
39 in the main
40 .\" HREF
41 \fBpcre\fP
42 .\"
43 page.
44 .P
45 The remainder of this document discusses the patterns that are supported by
46 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
47 From release 6.0, PCRE offers a second matching function,
48 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
49 Perl-compatible. Some of the features discussed below are not available when
50 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
51 alternative function, and how it differs from the normal function, are
52 discussed in the
53 .\" HREF
54 \fBpcrematching\fP
55 .\"
56 page.
57 .
58 .
60 .rs
61 .sp
62 PCRE supports five different conventions for indicating line breaks in
63 strings: a single CR (carriage return) character, a single LF (linefeed)
64 character, the two-character sequence CRLF, any of the three preceding, or any
65 Unicode newline sequence. The
66 .\" HREF
67 \fBpcreapi\fP
68 .\"
69 page has
70 .\" HTML <a href="pcreapi.html#newlines">
71 .\" </a>
72 further discussion
73 .\"
74 about newlines, and shows how to set the newline convention in the
75 \fIoptions\fP arguments for the compiling and matching functions.
76 .P
77 It is also possible to specify a newline convention by starting a pattern
78 string with one of the following five sequences:
79 .sp
80 (*CR) carriage return
81 (*LF) linefeed
82 (*CRLF) carriage return, followed by linefeed
83 (*ANYCRLF) any of the three above
84 (*ANY) all Unicode newline sequences
85 .sp
86 These override the default and the options given to \fBpcre_compile()\fP or
87 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
88 newline sequence, the pattern
89 .sp
90 (*CR)a.b
91 .sp
92 changes the convention to CR. That pattern matches "a\enb" because LF is no
93 longer a newline. Note that these special settings, which are not
94 Perl-compatible, are recognized only at the very start of a pattern, and that
95 they must be in upper case. If more than one of them is present, the last one
96 is used.
97 .P
98 The newline convention affects the interpretation of the dot metacharacter when
99 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
100 affect what the \eR escape sequence matches. By default, this is any Unicode
101 newline sequence, for Perl compatibility. However, this can be changed; see the
102 description of \eR in the section entitled
103 .\" HTML <a href="#newlineseq">
104 .\" </a>
105 "Newline sequences"
106 .\"
107 below. A change of \eR setting can be combined with a change of newline
108 convention.
109 .
110 .
112 .rs
113 .sp
114 A regular expression is a pattern that is matched against a subject string from
115 left to right. Most characters stand for themselves in a pattern, and match the
116 corresponding characters in the subject. As a trivial example, the pattern
117 .sp
118 The quick brown fox
119 .sp
120 matches a portion of a subject string that is identical to itself. When
121 caseless matching is specified (the PCRE_CASELESS option), letters are matched
122 independently of case. In UTF-8 mode, PCRE always understands the concept of
123 case for characters whose values are less than 128, so caseless matching is
124 always possible. For characters with higher values, the concept of case is
125 supported if PCRE is compiled with Unicode property support, but not otherwise.
126 If you want to use caseless matching for characters 128 and above, you must
127 ensure that PCRE is compiled with Unicode property support as well as with
128 UTF-8 support.
129 .P
130 The power of regular expressions comes from the ability to include alternatives
131 and repetitions in the pattern. These are encoded in the pattern by the use of
132 \fImetacharacters\fP, which do not stand for themselves but instead are
133 interpreted in some special way.
134 .P
135 There are two different sets of metacharacters: those that are recognized
136 anywhere in the pattern except within square brackets, and those that are
137 recognized within square brackets. Outside square brackets, the metacharacters
138 are as follows:
139 .sp
140 \e general escape character with several uses
141 ^ assert start of string (or line, in multiline mode)
142 $ assert end of string (or line, in multiline mode)
143 . match any character except newline (by default)
144 [ start character class definition
145 | start of alternative branch
146 ( start subpattern
147 ) end subpattern
148 ? extends the meaning of (
149 also 0 or 1 quantifier
150 also quantifier minimizer
151 * 0 or more quantifier
152 + 1 or more quantifier
153 also "possessive quantifier"
154 { start min/max quantifier
155 .sp
156 Part of a pattern that is in square brackets is called a "character class". In
157 a character class the only metacharacters are:
158 .sp
159 \e general escape character
160 ^ negate the class, but only if the first character
161 - indicates character range
162 .\" JOIN
163 [ POSIX character class (only if followed by POSIX
164 syntax)
165 ] terminates the character class
166 .sp
167 The following sections describe the use of each of the metacharacters.
168 .
169 .
171 .rs
172 .sp
173 The backslash character has several uses. Firstly, if it is followed by a
174 non-alphanumeric character, it takes away any special meaning that character
175 may have. This use of backslash as an escape character applies both inside and
176 outside character classes.
177 .P
178 For example, if you want to match a * character, you write \e* in the pattern.
179 This escaping action applies whether or not the following character would
180 otherwise be interpreted as a metacharacter, so it is always safe to precede a
181 non-alphanumeric with backslash to specify that it stands for itself. In
182 particular, if you want to match a backslash, you write \e\e.
183 .P
184 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
185 pattern (other than in a character class) and characters between a # outside
186 a character class and the next newline are ignored. An escaping backslash can
187 be used to include a whitespace or # character as part of the pattern.
188 .P
189 If you want to remove the special meaning from a sequence of characters, you
190 can do so by putting them between \eQ and \eE. This is different from Perl in
191 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
192 Perl, $ and @ cause variable interpolation. Note the following examples:
193 .sp
194 Pattern PCRE matches Perl matches
195 .sp
196 .\" JOIN
197 \eQabc$xyz\eE abc$xyz abc followed by the
198 contents of $xyz
199 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
200 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
201 .sp
202 The \eQ...\eE sequence is recognized both inside and outside character classes.
203 .
204 .
205 .\" HTML <a name="digitsafterbackslash"></a>
206 .SS "Non-printing characters"
207 .rs
208 .sp
209 A second use of backslash provides a way of encoding non-printing characters
210 in patterns in a visible manner. There is no restriction on the appearance of
211 non-printing characters, apart from the binary zero that terminates a pattern,
212 but when a pattern is being prepared by text editing, it is often easier to use
213 one of the following escape sequences than the binary character it represents:
214 .sp
215 \ea alarm, that is, the BEL character (hex 07)
216 \ecx "control-x", where x is any character
217 \ee escape (hex 1B)
218 \ef formfeed (hex 0C)
219 \en linefeed (hex 0A)
220 \er carriage return (hex 0D)
221 \et tab (hex 09)
222 \eddd character with octal code ddd, or back reference
223 \exhh character with hex code hh
224 \ex{hhh..} character with hex code hhh..
225 .sp
226 The precise effect of \ecx is as follows: if x is a lower case letter, it
227 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
228 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
229 7B.
230 .P
231 After \ex, from zero to two hexadecimal digits are read (letters can be in
232 upper or lower case). Any number of hexadecimal digits may appear between \ex{
233 and }, but the value of the character code must be less than 256 in non-UTF-8
234 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
235 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
236 point, which is 10FFFF.
237 .P
238 If characters other than hexadecimal digits appear between \ex{ and }, or if
239 there is no terminating }, this form of escape is not recognized. Instead, the
240 initial \ex will be interpreted as a basic hexadecimal escape, with no
241 following digits, giving a character whose value is zero.
242 .P
243 Characters whose value is less than 256 can be defined by either of the two
244 syntaxes for \ex. There is no difference in the way they are handled. For
245 example, \exdc is exactly the same as \ex{dc}.
246 .P
247 After \e0 up to two further octal digits are read. If there are fewer than two
248 digits, just those that are present are used. Thus the sequence \e0\ex\e07
249 specifies two binary zeros followed by a BEL character (code value 7). Make
250 sure you supply two digits after the initial zero if the pattern character that
251 follows is itself an octal digit.
252 .P
253 The handling of a backslash followed by a digit other than 0 is complicated.
254 Outside a character class, PCRE reads it and any following digits as a decimal
255 number. If the number is less than 10, or if there have been at least that many
256 previous capturing left parentheses in the expression, the entire sequence is
257 taken as a \fIback reference\fP. A description of how this works is given
258 .\" HTML <a href="#backreferences">
259 .\" </a>
260 later,
261 .\"
262 following the discussion of
263 .\" HTML <a href="#subpattern">
264 .\" </a>
265 parenthesized subpatterns.
266 .\"
267 .P
268 Inside a character class, or if the decimal number is greater than 9 and there
269 have not been that many capturing subpatterns, PCRE re-reads up to three octal
270 digits following the backslash, and uses them to generate a data character. Any
271 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
272 character specified in octal must be less than \e400. In UTF-8 mode, values up
273 to \e777 are permitted. For example:
274 .sp
275 \e040 is another way of writing a space
276 .\" JOIN
277 \e40 is the same, provided there are fewer than 40
278 previous capturing subpatterns
279 \e7 is always a back reference
280 .\" JOIN
281 \e11 might be a back reference, or another way of
282 writing a tab
283 \e011 is always a tab
284 \e0113 is a tab followed by the character "3"
285 .\" JOIN
286 \e113 might be a back reference, otherwise the
287 character with octal code 113
288 .\" JOIN
289 \e377 might be a back reference, otherwise
290 the byte consisting entirely of 1 bits
291 .\" JOIN
292 \e81 is either a back reference, or a binary zero
293 followed by the two characters "8" and "1"
294 .sp
295 Note that octal values of 100 or greater must not be introduced by a leading
296 zero, because no more than three octal digits are ever read.
297 .P
298 All the sequences that define a single character value can be used both inside
299 and outside character classes. In addition, inside a character class, the
300 sequence \eb is interpreted as the backspace character (hex 08). The sequences
301 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
302 unrecognized escape sequences, they are treated as the literal characters "B",
303 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
304 set. Outside a character class, these sequences have different meanings.
305 .
306 .
307 .SS "Absolute and relative back references"
308 .rs
309 .sp
310 The sequence \eg followed by an unsigned or a negative number, optionally
311 enclosed in braces, is an absolute or relative back reference. A named back
312 reference can be coded as \eg{name}. Back references are discussed
313 .\" HTML <a href="#backreferences">
314 .\" </a>
315 later,
316 .\"
317 following the discussion of
318 .\" HTML <a href="#subpattern">
319 .\" </a>
320 parenthesized subpatterns.
321 .\"
322 .
323 .
324 .SS "Absolute and relative subroutine calls"
325 .rs
326 .sp
327 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
328 a number enclosed either in angle brackets or single quotes, is an alternative
329 syntax for referencing a subpattern as a "subroutine". Details are discussed
330 .\" HTML <a href="#onigurumasubroutines">
331 .\" </a>
332 later.
333 .\"
334 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
335 synonymous. The former is a back reference; the latter is a
336 .\" HTML <a href="#subpatternsassubroutines">
337 .\" </a>
338 subroutine
339 .\"
340 call.
341 .
342 .
343 .SS "Generic character types"
344 .rs
345 .sp
346 Another use of backslash is for specifying generic character types:
347 .sp
348 \ed any decimal digit
349 \eD any character that is not a decimal digit
350 \eh any horizontal whitespace character
351 \eH any character that is not a horizontal whitespace character
352 \es any whitespace character
353 \eS any character that is not a whitespace character
354 \ev any vertical whitespace character
355 \eV any character that is not a vertical whitespace character
356 \ew any "word" character
357 \eW any "non-word" character
358 .sp
359 There is also the single sequence \eN, which matches a non-newline character.
360 This is the same as
361 .\" HTML <a href="#fullstopdot">
362 .\" </a>
363 the "." metacharacter
364 .\"
365 when PCRE_DOTALL is not set.
366 .P
367 Each pair of lower and upper case escape sequences partitions the complete set
368 of characters into two disjoint sets. Any given character matches one, and only
369 one, of each pair.
370 .P
371 These character type sequences can appear both inside and outside character
372 classes. They each match one character of the appropriate type. If the current
373 matching point is at the end of the subject string, all of them fail, since
374 there is no character to match.
375 .P
376 For compatibility with Perl, \es does not match the VT character (code 11).
377 This makes it different from the the POSIX "space" class. The \es characters
378 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
379 included in a Perl script, \es may match the VT character. In PCRE, it never
380 does.
381 .P
382 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
383 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
384 character property support is available. These sequences retain their original
385 meanings from before UTF-8 support was available, mainly for efficiency
386 reasons. Note that this also affects \eb, because it is defined in terms of \ew
387 and \eW.
388 .P
389 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
390 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
391 The horizontal space characters are:
392 .sp
393 U+0009 Horizontal tab
394 U+0020 Space
395 U+00A0 Non-break space
396 U+1680 Ogham space mark
397 U+180E Mongolian vowel separator
398 U+2000 En quad
399 U+2001 Em quad
400 U+2002 En space
401 U+2003 Em space
402 U+2004 Three-per-em space
403 U+2005 Four-per-em space
404 U+2006 Six-per-em space
405 U+2007 Figure space
406 U+2008 Punctuation space
407 U+2009 Thin space
408 U+200A Hair space
409 U+202F Narrow no-break space
410 U+205F Medium mathematical space
411 U+3000 Ideographic space
412 .sp
413 The vertical space characters are:
414 .sp
415 U+000A Linefeed
416 U+000B Vertical tab
417 U+000C Formfeed
418 U+000D Carriage return
419 U+0085 Next line
420 U+2028 Line separator
421 U+2029 Paragraph separator
422 .P
423 A "word" character is an underscore or any character less than 256 that is a
424 letter or digit. The definition of letters and digits is controlled by PCRE's
425 low-valued character tables, and may vary if locale-specific matching is taking
426 place (see
427 .\" HTML <a href="pcreapi.html#localesupport">
428 .\" </a>
429 "Locale support"
430 .\"
431 in the
432 .\" HREF
433 \fBpcreapi\fP
434 .\"
435 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
436 or "french" in Windows, some character codes greater than 128 are used for
437 accented letters, and these are matched by \ew. The use of locales with Unicode
438 is discouraged.
439 .
440 .
441 .\" HTML <a name="newlineseq"></a>
442 .SS "Newline sequences"
443 .rs
444 .sp
445 Outside a character class, by default, the escape sequence \eR matches any
446 Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
447 equivalent to the following:
448 .sp
449 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
450 .sp
451 This is an example of an "atomic group", details of which are given
452 .\" HTML <a href="#atomicgroup">
453 .\" </a>
454 below.
455 .\"
456 This particular group matches either the two-character sequence CR followed by
457 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
458 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
459 line, U+0085). The two-character sequence is treated as a single unit that
460 cannot be split.
461 .P
462 In UTF-8 mode, two additional characters whose codepoints are greater than 255
463 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
464 Unicode character property support is not needed for these characters to be
465 recognized.
466 .P
467 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
468 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
469 either at compile time or when the pattern is matched. (BSR is an abbrevation
470 for "backslash R".) This can be made the default when PCRE is built; if this is
471 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
472 It is also possible to specify these settings by starting a pattern string with
473 one of the following sequences:
474 .sp
475 (*BSR_ANYCRLF) CR, LF, or CRLF only
476 (*BSR_UNICODE) any Unicode newline sequence
477 .sp
478 These override the default and the options given to \fBpcre_compile()\fP or
479 \fBpcre_compile2()\fP, but they can be overridden by options given to
480 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
481 which are not Perl-compatible, are recognized only at the very start of a
482 pattern, and that they must be in upper case. If more than one of them is
483 present, the last one is used. They can be combined with a change of newline
484 convention, for example, a pattern can start with:
485 .sp
487 .sp
488 Inside a character class, \eR is treated as an unrecognized escape sequence,
489 and so matches the letter "R" by default, but causes an error if PCRE_EXTRA is
490 set.
491 .
492 .
493 .\" HTML <a name="uniextseq"></a>
494 .SS Unicode character properties
495 .rs
496 .sp
497 When PCRE is built with Unicode character property support, three additional
498 escape sequences that match characters with specific properties are available.
499 When not in UTF-8 mode, these sequences are of course limited to testing
500 characters whose codepoints are less than 256, but they do work in this mode.
501 The extra escape sequences are:
502 .sp
503 \ep{\fIxx\fP} a character with the \fIxx\fP property
504 \eP{\fIxx\fP} a character without the \fIxx\fP property
505 \eX an extended Unicode sequence
506 .sp
507 The property names represented by \fIxx\fP above are limited to the Unicode
508 script names, the general category properties, "Any", which matches any
509 character (including newline), and some special PCRE properties (described
510 in the
511 .\" HTML <a href="#extraprops">
512 .\" </a>
513 next section).
514 .\"
515 Other Perl properties such as "InMusicalSymbols" are not currently supported by
516 PCRE. Note that \eP{Any} does not match any characters, so always causes a
517 match failure.
518 .P
519 Sets of Unicode characters are defined as belonging to certain scripts. A
520 character from one of these sets can be matched using a script name. For
521 example:
522 .sp
523 \ep{Greek}
524 \eP{Han}
525 .sp
526 Those that are not part of an identified script are lumped together as
527 "Common". The current list of scripts is:
528 .P
529 Arabic,
530 Armenian,
531 Avestan,
532 Balinese,
533 Bamum,
534 Bengali,
535 Bopomofo,
536 Braille,
537 Buginese,
538 Buhid,
539 Canadian_Aboriginal,
540 Carian,
541 Cham,
542 Cherokee,
543 Common,
544 Coptic,
545 Cuneiform,
546 Cypriot,
547 Cyrillic,
548 Deseret,
549 Devanagari,
550 Egyptian_Hieroglyphs,
551 Ethiopic,
552 Georgian,
553 Glagolitic,
554 Gothic,
555 Greek,
556 Gujarati,
557 Gurmukhi,
558 Han,
559 Hangul,
560 Hanunoo,
561 Hebrew,
562 Hiragana,
563 Imperial_Aramaic,
564 Inherited,
565 Inscriptional_Pahlavi,
566 Inscriptional_Parthian,
567 Javanese,
568 Kaithi,
569 Kannada,
570 Katakana,
571 Kayah_Li,
572 Kharoshthi,
573 Khmer,
574 Lao,
575 Latin,
576 Lepcha,
577 Limbu,
578 Linear_B,
579 Lisu,
580 Lycian,
581 Lydian,
582 Malayalam,
583 Meetei_Mayek,
584 Mongolian,
585 Myanmar,
586 New_Tai_Lue,
587 Nko,
588 Ogham,
589 Old_Italic,
590 Old_Persian,
591 Old_South_Arabian,
592 Old_Turkic,
593 Ol_Chiki,
594 Oriya,
595 Osmanya,
596 Phags_Pa,
597 Phoenician,
598 Rejang,
599 Runic,
600 Samaritan,
601 Saurashtra,
602 Shavian,
603 Sinhala,
604 Sundanese,
605 Syloti_Nagri,
606 Syriac,
607 Tagalog,
608 Tagbanwa,
609 Tai_Le,
610 Tai_Tham,
611 Tai_Viet,
612 Tamil,
613 Telugu,
614 Thaana,
615 Thai,
616 Tibetan,
617 Tifinagh,
618 Ugaritic,
619 Vai,
620 Yi.
621 .P
622 Each character has exactly one Unicode general category property, specified by
623 a two-letter abbreviation. For compatibility with Perl, negation can be
624 specified by including a circumflex between the opening brace and the property
625 name. For example, \ep{^Lu} is the same as \eP{Lu}.
626 .P
627 If only one letter is specified with \ep or \eP, it includes all the general
628 category properties that start with that letter. In this case, in the absence
629 of negation, the curly brackets in the escape sequence are optional; these two
630 examples have the same effect:
631 .sp
632 \ep{L}
633 \epL
634 .sp
635 The following general category property codes are supported:
636 .sp
637 C Other
638 Cc Control
639 Cf Format
640 Cn Unassigned
641 Co Private use
642 Cs Surrogate
643 .sp
644 L Letter
645 Ll Lower case letter
646 Lm Modifier letter
647 Lo Other letter
648 Lt Title case letter
649 Lu Upper case letter
650 .sp
651 M Mark
652 Mc Spacing mark
653 Me Enclosing mark
654 Mn Non-spacing mark
655 .sp
656 N Number
657 Nd Decimal number
658 Nl Letter number
659 No Other number
660 .sp
661 P Punctuation
662 Pc Connector punctuation
663 Pd Dash punctuation
664 Pe Close punctuation
665 Pf Final punctuation
666 Pi Initial punctuation
667 Po Other punctuation
668 Ps Open punctuation
669 .sp
670 S Symbol
671 Sc Currency symbol
672 Sk Modifier symbol
673 Sm Mathematical symbol
674 So Other symbol
675 .sp
676 Z Separator
677 Zl Line separator
678 Zp Paragraph separator
679 Zs Space separator
680 .sp
681 The special property L& is also supported: it matches a character that has
682 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
683 a modifier or "other".
684 .P
685 The Cs (Surrogate) property applies only to characters in the range U+D800 to
686 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
687 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
688 (see the discussion of PCRE_NO_UTF8_CHECK in the
689 .\" HREF
690 \fBpcreapi\fP
691 .\"
692 page). Perl does not support the Cs property.
693 .P
694 The long synonyms for property names that Perl supports (such as \ep{Letter})
695 are not supported by PCRE, nor is it permitted to prefix any of these
696 properties with "Is".
697 .P
698 No character that is in the Unicode table has the Cn (unassigned) property.
699 Instead, this property is assumed for any code point that is not in the
700 Unicode table.
701 .P
702 Specifying caseless matching does not affect these escape sequences. For
703 example, \ep{Lu} always matches only upper case letters.
704 .P
705 The \eX escape matches any number of Unicode characters that form an extended
706 Unicode sequence. \eX is equivalent to
707 .sp
708 (?>\ePM\epM*)
709 .sp
710 That is, it matches a character without the "mark" property, followed by zero
711 or more characters with the "mark" property, and treats the sequence as an
712 atomic group
713 .\" HTML <a href="#atomicgroup">
714 .\" </a>
715 (see below).
716 .\"
717 Characters with the "mark" property are typically accents that affect the
718 preceding character. None of them have codepoints less than 256, so in
719 non-UTF-8 mode \eX matches any one character.
720 .P
721 Matching characters by Unicode property is not fast, because PCRE has to search
722 a structure that contains data for over fifteen thousand characters. That is
723 why the traditional escape sequences such as \ed and \ew do not use Unicode
724 properties in PCRE.
725 .
726 .
727 .\" HTML <a name="extraprops"></a>
728 .SS PCRE's additional properties
729 .rs
730 .sp
731 As well as the standard Unicode properties described in the previous
732 section, PCRE supports four more that make it possible to convert traditional
733 escape sequences such as \ew and \es and POSIX character classes to use Unicode
734 properties. These are:
735 .sp
736 Xan Any alphanumeric character
737 Xps Any POSIX space character
738 Xsp Any Perl space character
739 Xwd Any Perl "word" character
740 .sp
741 Xan matches characters that have either the L (letter) or the N (number)
742 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
743 carriage return, and any other character that has the Z (separator) property.
744 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
745 same characters as Xan, plus underscore.
746 .
747 .
748 .\" HTML <a name="resetmatchstart"></a>
749 .SS "Resetting the match start"
750 .rs
751 .sp
752 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
753 matched characters not to be included in the final matched sequence. For
754 example, the pattern:
755 .sp
756 foo\eKbar
757 .sp
758 matches "foobar", but reports that it has matched "bar". This feature is
759 similar to a lookbehind assertion
760 .\" HTML <a href="#lookbehind">
761 .\" </a>
762 (described below).
763 .\"
764 However, in this case, the part of the subject before the real match does not
765 have to be of fixed length, as lookbehind assertions do. The use of \eK does
766 not interfere with the setting of
767 .\" HTML <a href="#subpattern">
768 .\" </a>
769 captured substrings.
770 .\"
771 For example, when the pattern
772 .sp
773 (foo)\eKbar
774 .sp
775 matches "foobar", the first substring is still set to "foo".
776 .P
777 Perl documents that the use of \eK within assertions is "not well defined". In
778 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
779 ignored in negative assertions.
780 .
781 .
782 .\" HTML <a name="smallassertions"></a>
783 .SS "Simple assertions"
784 .rs
785 .sp
786 The final use of backslash is for certain simple assertions. An assertion
787 specifies a condition that has to be met at a particular point in a match,
788 without consuming any characters from the subject string. The use of
789 subpatterns for more complicated assertions is described
790 .\" HTML <a href="#bigassertions">
791 .\" </a>
792 below.
793 .\"
794 The backslashed assertions are:
795 .sp
796 \eb matches at a word boundary
797 \eB matches when not at a word boundary
798 \eA matches at the start of the subject
799 \eZ matches at the end of the subject
800 also matches before a newline at the end of the subject
801 \ez matches only at the end of the subject
802 \eG matches at the first matching position in the subject
803 .sp
804 Inside a character class, \eb has a different meaning; it matches the backspace
805 character. If any other of these assertions appears in a character class, by
806 default it matches the corresponding literal character (for example, \eB
807 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
808 escape sequence" error is generated instead.
809 .P
810 A word boundary is a position in the subject string where the current character
811 and the previous character do not both match \ew or \eW (i.e. one matches
812 \ew and the other matches \eW), or the start or end of the string if the
813 first or last character matches \ew, respectively. Neither PCRE nor Perl has a
814 separte "start of word" or "end of word" metasequence. However, whatever
815 follows \eb normally determines which it is. For example, the fragment
816 \eba matches "a" at the start of a word.
817 .P
818 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
819 dollar (described in the next section) in that they only ever match at the very
820 start and end of the subject string, whatever options are set. Thus, they are
821 independent of multiline mode. These three assertions are not affected by the
822 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
823 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
824 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
825 at a point other than the beginning of the subject, \eA can never match. The
826 difference between \eZ and \ez is that \eZ matches before a newline at the end
827 of the string as well as at the very end, whereas \ez matches only at the end.
828 .P
829 The \eG assertion is true only when the current matching position is at the
830 start point of the match, as specified by the \fIstartoffset\fP argument of
831 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
832 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
833 arguments, you can mimic Perl's /g option, and it is in this kind of
834 implementation where \eG can be useful.
835 .P
836 Note, however, that PCRE's interpretation of \eG, as the start of the current
837 match, is subtly different from Perl's, which defines it as the end of the
838 previous match. In Perl, these can be different when the previously matched
839 string was empty. Because PCRE does just one match at a time, it cannot
840 reproduce this behaviour.
841 .P
842 If all the alternatives of a pattern begin with \eG, the expression is anchored
843 to the starting match position, and the "anchored" flag is set in the compiled
844 regular expression.
845 .
846 .
848 .rs
849 .sp
850 Outside a character class, in the default matching mode, the circumflex
851 character is an assertion that is true only if the current matching point is
852 at the start of the subject string. If the \fIstartoffset\fP argument of
853 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
854 option is unset. Inside a character class, circumflex has an entirely different
855 meaning
856 .\" HTML <a href="#characterclass">
857 .\" </a>
858 (see below).
859 .\"
860 .P
861 Circumflex need not be the first character of the pattern if a number of
862 alternatives are involved, but it should be the first thing in each alternative
863 in which it appears if the pattern is ever to match that branch. If all
864 possible alternatives start with a circumflex, that is, if the pattern is
865 constrained to match only at the start of the subject, it is said to be an
866 "anchored" pattern. (There are also other constructs that can cause a pattern
867 to be anchored.)
868 .P
869 A dollar character is an assertion that is true only if the current matching
870 point is at the end of the subject string, or immediately before a newline
871 at the end of the string (by default). Dollar need not be the last character of
872 the pattern if a number of alternatives are involved, but it should be the last
873 item in any branch in which it appears. Dollar has no special meaning in a
874 character class.
875 .P
876 The meaning of dollar can be changed so that it matches only at the very end of
877 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
878 does not affect the \eZ assertion.
879 .P
880 The meanings of the circumflex and dollar characters are changed if the
881 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
882 immediately after internal newlines as well as at the start of the subject
883 string. It does not match after a newline that ends the string. A dollar
884 matches before any newlines in the string, as well as at the very end, when
885 PCRE_MULTILINE is set. When newline is specified as the two-character
886 sequence CRLF, isolated CR and LF characters do not indicate newlines.
887 .P
888 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
889 \en represents a newline) in multiline mode, but not otherwise. Consequently,
890 patterns that are anchored in single line mode because all branches start with
891 ^ are not anchored in multiline mode, and a match for circumflex is possible
892 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
893 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
894 .P
895 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
896 end of the subject in both modes, and if all branches of a pattern start with
897 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
898 .
899 .
900 .\" HTML <a name="fullstopdot"></a>
902 .rs
903 .sp
904 Outside a character class, a dot in the pattern matches any one character in
905 the subject string except (by default) a character that signifies the end of a
906 line. In UTF-8 mode, the matched character may be more than one byte long.
907 .P
908 When a line ending is defined as a single character, dot never matches that
909 character; when the two-character sequence CRLF is used, dot does not match CR
910 if it is immediately followed by LF, but otherwise it matches all characters
911 (including isolated CRs and LFs). When any Unicode line endings are being
912 recognized, dot does not match CR or LF or any of the other line ending
913 characters.
914 .P
915 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
916 option is set, a dot matches any one character, without exception. If the
917 two-character sequence CRLF is present in the subject string, it takes two dots
918 to match it.
919 .P
920 The handling of dot is entirely independent of the handling of circumflex and
921 dollar, the only relationship being that they both involve newlines. Dot has no
922 special meaning in a character class.
923 .P
924 The escape sequence \eN always behaves as a dot does when PCRE_DOTALL is not
925 set. In other words, it matches any one character except one that signifies the
926 end of a line.
927 .
928 .
930 .rs
931 .sp
932 Outside a character class, the escape sequence \eC matches any one byte, both
933 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
934 characters. The feature is provided in Perl in order to match individual bytes
935 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
936 what remains in the string may be a malformed UTF-8 string. For this reason,
937 the \eC escape sequence is best avoided.
938 .P
939 PCRE does not allow \eC to appear in lookbehind assertions
940 .\" HTML <a href="#lookbehind">
941 .\" </a>
942 (described below),
943 .\"
944 because in UTF-8 mode this would make it impossible to calculate the length of
945 the lookbehind.
946 .
947 .
948 .\" HTML <a name="characterclass"></a>
950 .rs
951 .sp
952 An opening square bracket introduces a character class, terminated by a closing
953 square bracket. A closing square bracket on its own is not special by default.
954 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
955 bracket causes a compile-time error. If a closing square bracket is required as
956 a member of the class, it should be the first data character in the class
957 (after an initial circumflex, if present) or escaped with a backslash.
958 .P
959 A character class matches a single character in the subject. In UTF-8 mode, the
960 character may be more than one byte long. A matched character must be in the
961 set of characters defined by the class, unless the first character in the class
962 definition is a circumflex, in which case the subject character must not be in
963 the set defined by the class. If a circumflex is actually required as a member
964 of the class, ensure it is not the first character, or escape it with a
965 backslash.
966 .P
967 For example, the character class [aeiou] matches any lower case vowel, while
968 [^aeiou] matches any character that is not a lower case vowel. Note that a
969 circumflex is just a convenient notation for specifying the characters that
970 are in the class by enumerating those that are not. A class that starts with a
971 circumflex is not an assertion; it still consumes a character from the subject
972 string, and therefore it fails if the current pointer is at the end of the
973 string.
974 .P
975 In UTF-8 mode, characters with values greater than 255 can be included in a
976 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
977 .P
978 When caseless matching is set, any letters in a class represent both their
979 upper case and lower case versions, so for example, a caseless [aeiou] matches
980 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
981 caseful version would. In UTF-8 mode, PCRE always understands the concept of
982 case for characters whose values are less than 128, so caseless matching is
983 always possible. For characters with higher values, the concept of case is
984 supported if PCRE is compiled with Unicode property support, but not otherwise.
985 If you want to use caseless matching in UTF8-mode for characters 128 and above,
986 you must ensure that PCRE is compiled with Unicode property support as well as
987 with UTF-8 support.
988 .P
989 Characters that might indicate line breaks are never treated in any special way
990 when matching character classes, whatever line-ending sequence is in use, and
991 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
992 such as [^a] always matches one of these characters.
993 .P
994 The minus (hyphen) character can be used to specify a range of characters in a
995 character class. For example, [d-m] matches any letter between d and m,
996 inclusive. If a minus character is required in a class, it must be escaped with
997 a backslash or appear in a position where it cannot be interpreted as
998 indicating a range, typically as the first or last character in the class.
999 .P
1000 It is not possible to have the literal character "]" as the end character of a
1001 range. A pattern such as [W-]46] is interpreted as a class of two characters
1002 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1003 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1004 the end of range, so [W-\e]46] is interpreted as a class containing a range
1005 followed by two other characters. The octal or hexadecimal representation of
1006 "]" can also be used to end a range.
1007 .P
1008 Ranges operate in the collating sequence of character values. They can also be
1009 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1010 mode, ranges can include characters whose values are greater than 255, for
1011 example [\ex{100}-\ex{2ff}].
1012 .P
1013 If a range that includes letters is used when caseless matching is set, it
1014 matches the letters in either case. For example, [W-c] is equivalent to
1015 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1016 tables for a French locale are in use, [\exc8-\excb] matches accented E
1017 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1018 characters with values greater than 128 only when it is compiled with Unicode
1019 property support.
1020 .P
1021 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
1022 in a character class, and add the characters that they match to the class. For
1023 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
1024 conveniently be used with the upper case character types to specify a more
1025 restricted set of characters than the matching lower case type. For example,
1026 the class [^\eW_] matches any letter or digit, but not underscore.
1027 .P
1028 The only metacharacters that are recognized in character classes are backslash,
1029 hyphen (only where it can be interpreted as specifying a range), circumflex
1030 (only at the start), opening square bracket (only when it can be interpreted as
1031 introducing a POSIX class name - see the next section), and the terminating
1032 closing square bracket. However, escaping other non-alphanumeric characters
1033 does no harm.
1034 .
1035 .
1037 .rs
1038 .sp
1039 Perl supports the POSIX notation for character classes. This uses names
1040 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1041 this notation. For example,
1042 .sp
1043 [01[:alpha:]%]
1044 .sp
1045 matches "0", "1", any alphabetic character, or "%". The supported class names
1046 are
1047 .sp
1048 alnum letters and digits
1049 alpha letters
1050 ascii character codes 0 - 127
1051 blank space or tab only
1052 cntrl control characters
1053 digit decimal digits (same as \ed)
1054 graph printing characters, excluding space
1055 lower lower case letters
1056 print printing characters, including space
1057 punct printing characters, excluding letters and digits
1058 space white space (not quite the same as \es)
1059 upper upper case letters
1060 word "word" characters (same as \ew)
1061 xdigit hexadecimal digits
1062 .sp
1063 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1064 space (32). Notice that this list includes the VT character (code 11). This
1065 makes "space" different to \es, which does not include VT (for Perl
1066 compatibility).
1067 .P
1068 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1069 5.8. Another Perl extension is negation, which is indicated by a ^ character
1070 after the colon. For example,
1071 .sp
1072 [12[:^digit:]]
1073 .sp
1074 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1075 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1076 supported, and an error is given if they are encountered.
1077 .P
1078 In UTF-8 mode, characters with values greater than 128 do not match any of
1079 the POSIX character classes.
1080 .
1081 .
1083 .rs
1084 .sp
1085 Vertical bar characters are used to separate alternative patterns. For example,
1086 the pattern
1087 .sp
1088 gilbert|sullivan
1089 .sp
1090 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1091 and an empty alternative is permitted (matching the empty string). The matching
1092 process tries each alternative in turn, from left to right, and the first one
1093 that succeeds is used. If the alternatives are within a subpattern
1094 .\" HTML <a href="#subpattern">
1095 .\" </a>
1096 (defined below),
1097 .\"
1098 "succeeds" means matching the rest of the main pattern as well as the
1099 alternative in the subpattern.
1100 .
1101 .
1103 .rs
1104 .sp
1105 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1106 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1107 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1108 The option letters are
1109 .sp
1110 i for PCRE_CASELESS
1112 s for PCRE_DOTALL
1113 x for PCRE_EXTENDED
1114 .sp
1115 For example, (?im) sets caseless, multiline matching. It is also possible to
1116 unset these options by preceding the letter with a hyphen, and a combined
1117 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1118 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1119 permitted. If a letter appears both before and after the hyphen, the option is
1120 unset.
1121 .P
1122 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1123 changed in the same way as the Perl-compatible options by using the characters
1124 J, U and X respectively.
1125 .P
1126 When one of these option changes occurs at top level (that is, not inside
1127 subpattern parentheses), the change applies to the remainder of the pattern
1128 that follows. If the change is placed right at the start of a pattern, PCRE
1129 extracts it into the global options (and it will therefore show up in data
1130 extracted by the \fBpcre_fullinfo()\fP function).
1131 .P
1132 An option change within a subpattern (see below for a description of
1133 subpatterns) affects only that part of the current pattern that follows it, so
1134 .sp
1135 (a(?i)b)c
1136 .sp
1137 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1138 By this means, options can be made to have different settings in different
1139 parts of the pattern. Any changes made in one alternative do carry on
1140 into subsequent branches within the same subpattern. For example,
1141 .sp
1142 (a(?i)b|c)
1143 .sp
1144 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1145 branch is abandoned before the option setting. This is because the effects of
1146 option settings happen at compile time. There would be some very weird
1147 behaviour otherwise.
1148 .P
1149 \fBNote:\fP There are other PCRE-specific options that can be set by the
1150 application when the compile or match functions are called. In some cases the
1151 pattern can contain special leading sequences such as (*CRLF) to override what
1152 the application has set or what has been defaulted. Details are given in the
1153 section entitled
1154 .\" HTML <a href="#newlineseq">
1155 .\" </a>
1156 "Newline sequences"
1157 .\"
1158 above. There is also the (*UTF8) leading sequence that can be used to set UTF-8
1159 mode; this is equivalent to setting the PCRE_UTF8 option.
1160 .
1161 .
1162 .\" HTML <a name="subpattern"></a>
1164 .rs
1165 .sp
1166 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1167 Turning part of a pattern into a subpattern does two things:
1168 .sp
1169 1. It localizes a set of alternatives. For example, the pattern
1170 .sp
1171 cat(aract|erpillar|)
1172 .sp
1173 matches one of the words "cat", "cataract", or "caterpillar". Without the
1174 parentheses, it would match "cataract", "erpillar" or an empty string.
1175 .sp
1176 2. It sets up the subpattern as a capturing subpattern. This means that, when
1177 the whole pattern matches, that portion of the subject string that matched the
1178 subpattern is passed back to the caller via the \fIovector\fP argument of
1179 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1180 from 1) to obtain numbers for the capturing subpatterns.
1181 .P
1182 For example, if the string "the red king" is matched against the pattern
1183 .sp
1184 the ((red|white) (king|queen))
1185 .sp
1186 the captured substrings are "red king", "red", and "king", and are numbered 1,
1187 2, and 3, respectively.
1188 .P
1189 The fact that plain parentheses fulfil two functions is not always helpful.
1190 There are often times when a grouping subpattern is required without a
1191 capturing requirement. If an opening parenthesis is followed by a question mark
1192 and a colon, the subpattern does not do any capturing, and is not counted when
1193 computing the number of any subsequent capturing subpatterns. For example, if
1194 the string "the white queen" is matched against the pattern
1195 .sp
1196 the ((?:red|white) (king|queen))
1197 .sp
1198 the captured substrings are "white queen" and "queen", and are numbered 1 and
1199 2. The maximum number of capturing subpatterns is 65535.
1200 .P
1201 As a convenient shorthand, if any option settings are required at the start of
1202 a non-capturing subpattern, the option letters may appear between the "?" and
1203 the ":". Thus the two patterns
1204 .sp
1205 (?i:saturday|sunday)
1206 (?:(?i)saturday|sunday)
1207 .sp
1208 match exactly the same set of strings. Because alternative branches are tried
1209 from left to right, and options are not reset until the end of the subpattern
1210 is reached, an option setting in one branch does affect subsequent branches, so
1211 the above patterns match "SUNDAY" as well as "Saturday".
1212 .
1213 .
1214 .\" HTML <a name="dupsubpatternnumber"></a>
1216 .rs
1217 .sp
1218 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1219 the same numbers for its capturing parentheses. Such a subpattern starts with
1220 (?| and is itself a non-capturing subpattern. For example, consider this
1221 pattern:
1222 .sp
1223 (?|(Sat)ur|(Sun))day
1224 .sp
1225 Because the two alternatives are inside a (?| group, both sets of capturing
1226 parentheses are numbered one. Thus, when the pattern matches, you can look
1227 at captured substring number one, whichever alternative matched. This construct
1228 is useful when you want to capture part, but not all, of one of a number of
1229 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1230 number is reset at the start of each branch. The numbers of any capturing
1231 buffers that follow the subpattern start after the highest number used in any
1232 branch. The following example is taken from the Perl documentation.
1233 The numbers underneath show in which buffer the captured content will be
1234 stored.
1235 .sp
1236 # before ---------------branch-reset----------- after
1237 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1238 # 1 2 2 3 2 3 4
1239 .sp
1240 A back reference to a numbered subpattern uses the most recent value that is
1241 set for that number by any subpattern. The following pattern matches "abcabc"
1242 or "defdef":
1243 .sp
1244 /(?|(abc)|(def))\e1/
1245 .sp
1246 In contrast, a recursive or "subroutine" call to a numbered subpattern always
1247 refers to the first one in the pattern with the given number. The following
1248 pattern matches "abcabc" or "defabc":
1249 .sp
1250 /(?|(abc)|(def))(?1)/
1251 .sp
1252 If a
1253 .\" HTML <a href="#conditions">
1254 .\" </a>
1255 condition test
1256 .\"
1257 for a subpattern's having matched refers to a non-unique number, the test is
1258 true if any of the subpatterns of that number have matched.
1259 .P
1260 An alternative approach to using this "branch reset" feature is to use
1261 duplicate named subpatterns, as described in the next section.
1262 .
1263 .
1265 .rs
1266 .sp
1267 Identifying capturing parentheses by number is simple, but it can be very hard
1268 to keep track of the numbers in complicated regular expressions. Furthermore,
1269 if an expression is modified, the numbers may change. To help with this
1270 difficulty, PCRE supports the naming of subpatterns. This feature was not
1271 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1272 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1273 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1274 have different names, but PCRE does not.
1275 .P
1276 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1277 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1278 parentheses from other parts of the pattern, such as
1279 .\" HTML <a href="#backreferences">
1280 .\" </a>
1281 back references,
1282 .\"
1283 .\" HTML <a href="#recursion">
1284 .\" </a>
1285 recursion,
1286 .\"
1287 and
1288 .\" HTML <a href="#conditions">
1289 .\" </a>
1290 conditions,
1291 .\"
1292 can be made by name as well as by number.
1293 .P
1294 Names consist of up to 32 alphanumeric characters and underscores. Named
1295 capturing parentheses are still allocated numbers as well as names, exactly as
1296 if the names were not present. The PCRE API provides function calls for
1297 extracting the name-to-number translation table from a compiled pattern. There
1298 is also a convenience function for extracting a captured substring by name.
1299 .P
1300 By default, a name must be unique within a pattern, but it is possible to relax
1301 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1302 names are also always permitted for subpatterns with the same number, set up as
1303 described in the previous section.) Duplicate names can be useful for patterns
1304 where only one instance of the named parentheses can match. Suppose you want to
1305 match the name of a weekday, either as a 3-letter abbreviation or as the full
1306 name, and in both cases you want to extract the abbreviation. This pattern
1307 (ignoring the line breaks) does the job:
1308 .sp
1309 (?<DN>Mon|Fri|Sun)(?:day)?|
1310 (?<DN>Tue)(?:sday)?|
1311 (?<DN>Wed)(?:nesday)?|
1312 (?<DN>Thu)(?:rsday)?|
1313 (?<DN>Sat)(?:urday)?
1314 .sp
1315 There are five capturing substrings, but only one is ever set after a match.
1316 (An alternative way of solving this problem is to use a "branch reset"
1317 subpattern, as described in the previous section.)
1318 .P
1319 The convenience function for extracting the data by name returns the substring
1320 for the first (and in this example, the only) subpattern of that name that
1321 matched. This saves searching to find which numbered subpattern it was.
1322 .P
1323 If you make a back reference to a non-unique named subpattern from elsewhere in
1324 the pattern, the one that corresponds to the first occurrence of the name is
1325 used. In the absence of duplicate numbers (see the previous section) this is
1326 the one with the lowest number. If you use a named reference in a condition
1327 test (see the
1328 .\"
1329 .\" HTML <a href="#conditions">
1330 .\" </a>
1331 section about conditions
1332 .\"
1333 below), either to check whether a subpattern has matched, or to check for
1334 recursion, all subpatterns with the same name are tested. If the condition is
1335 true for any one of them, the overall condition is true. This is the same
1336 behaviour as testing by number. For further details of the interfaces for
1337 handling named subpatterns, see the
1338 .\" HREF
1339 \fBpcreapi\fP
1340 .\"
1341 documentation.
1342 .P
1343 \fBWarning:\fP You cannot use different names to distinguish between two
1344 subpatterns with the same number because PCRE uses only the numbers when
1345 matching. For this reason, an error is given at compile time if different names
1346 are given to subpatterns with the same number. However, you can give the same
1347 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1348 .
1349 .
1351 .rs
1352 .sp
1353 Repetition is specified by quantifiers, which can follow any of the following
1354 items:
1355 .sp
1356 a literal data character
1357 the dot metacharacter
1358 the \eC escape sequence
1359 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1360 the \eR escape sequence
1361 an escape such as \ed that matches a single character
1362 a character class
1363 a back reference (see next section)
1364 a parenthesized subpattern (unless it is an assertion)
1365 a recursive or "subroutine" call to a subpattern
1366 .sp
1367 The general repetition quantifier specifies a minimum and maximum number of
1368 permitted matches, by giving the two numbers in curly brackets (braces),
1369 separated by a comma. The numbers must be less than 65536, and the first must
1370 be less than or equal to the second. For example:
1371 .sp
1372 z{2,4}
1373 .sp
1374 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1375 character. If the second number is omitted, but the comma is present, there is
1376 no upper limit; if the second number and the comma are both omitted, the
1377 quantifier specifies an exact number of required matches. Thus
1378 .sp
1379 [aeiou]{3,}
1380 .sp
1381 matches at least 3 successive vowels, but may match many more, while
1382 .sp
1383 \ed{8}
1384 .sp
1385 matches exactly 8 digits. An opening curly bracket that appears in a position
1386 where a quantifier is not allowed, or one that does not match the syntax of a
1387 quantifier, is taken as a literal character. For example, {,6} is not a
1388 quantifier, but a literal string of four characters.
1389 .P
1390 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1391 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1392 which is represented by a two-byte sequence. Similarly, when Unicode property
1393 support is available, \eX{3} matches three Unicode extended sequences, each of
1394 which may be several bytes long (and they may be of different lengths).
1395 .P
1396 The quantifier {0} is permitted, causing the expression to behave as if the
1397 previous item and the quantifier were not present. This may be useful for
1398 subpatterns that are referenced as
1399 .\" HTML <a href="#subpatternsassubroutines">
1400 .\" </a>
1401 subroutines
1402 .\"
1403 from elsewhere in the pattern. Items other than subpatterns that have a {0}
1404 quantifier are omitted from the compiled pattern.
1405 .P
1406 For convenience, the three most common quantifiers have single-character
1407 abbreviations:
1408 .sp
1409 * is equivalent to {0,}
1410 + is equivalent to {1,}
1411 ? is equivalent to {0,1}
1412 .sp
1413 It is possible to construct infinite loops by following a subpattern that can
1414 match no characters with a quantifier that has no upper limit, for example:
1415 .sp
1416 (a?)*
1417 .sp
1418 Earlier versions of Perl and PCRE used to give an error at compile time for
1419 such patterns. However, because there are cases where this can be useful, such
1420 patterns are now accepted, but if any repetition of the subpattern does in fact
1421 match no characters, the loop is forcibly broken.
1422 .P
1423 By default, the quantifiers are "greedy", that is, they match as much as
1424 possible (up to the maximum number of permitted times), without causing the
1425 rest of the pattern to fail. The classic example of where this gives problems
1426 is in trying to match comments in C programs. These appear between /* and */
1427 and within the comment, individual * and / characters may appear. An attempt to
1428 match C comments by applying the pattern
1429 .sp
1430 /\e*.*\e*/
1431 .sp
1432 to the string
1433 .sp
1434 /* first comment */ not comment /* second comment */
1435 .sp
1436 fails, because it matches the entire string owing to the greediness of the .*
1437 item.
1438 .P
1439 However, if a quantifier is followed by a question mark, it ceases to be
1440 greedy, and instead matches the minimum number of times possible, so the
1441 pattern
1442 .sp
1443 /\e*.*?\e*/
1444 .sp
1445 does the right thing with the C comments. The meaning of the various
1446 quantifiers is not otherwise changed, just the preferred number of matches.
1447 Do not confuse this use of question mark with its use as a quantifier in its
1448 own right. Because it has two uses, it can sometimes appear doubled, as in
1449 .sp
1450 \ed??\ed
1451 .sp
1452 which matches one digit by preference, but can match two if that is the only
1453 way the rest of the pattern matches.
1454 .P
1455 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1456 the quantifiers are not greedy by default, but individual ones can be made
1457 greedy by following them with a question mark. In other words, it inverts the
1458 default behaviour.
1459 .P
1460 When a parenthesized subpattern is quantified with a minimum repeat count that
1461 is greater than 1 or with a limited maximum, more memory is required for the
1462 compiled pattern, in proportion to the size of the minimum or maximum.
1463 .P
1464 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1465 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1466 implicitly anchored, because whatever follows will be tried against every
1467 character position in the subject string, so there is no point in retrying the
1468 overall match at any position after the first. PCRE normally treats such a
1469 pattern as though it were preceded by \eA.
1470 .P
1471 In cases where it is known that the subject string contains no newlines, it is
1472 worth setting PCRE_DOTALL in order to obtain this optimization, or
1473 alternatively using ^ to indicate anchoring explicitly.
1474 .P
1475 However, there is one situation where the optimization cannot be used. When .*
1476 is inside capturing parentheses that are the subject of a back reference
1477 elsewhere in the pattern, a match at the start may fail where a later one
1478 succeeds. Consider, for example:
1479 .sp
1480 (.*)abc\e1
1481 .sp
1482 If the subject is "xyz123abc123" the match point is the fourth character. For
1483 this reason, such a pattern is not implicitly anchored.
1484 .P
1485 When a capturing subpattern is repeated, the value captured is the substring
1486 that matched the final iteration. For example, after
1487 .sp
1488 (tweedle[dume]{3}\es*)+
1489 .sp
1490 has matched "tweedledum tweedledee" the value of the captured substring is
1491 "tweedledee". However, if there are nested capturing subpatterns, the
1492 corresponding captured values may have been set in previous iterations. For
1493 example, after
1494 .sp
1495 /(a|(b))+/
1496 .sp
1497 matches "aba" the value of the second captured substring is "b".
1498 .
1499 .
1500 .\" HTML <a name="atomicgroup"></a>
1502 .rs
1503 .sp
1504 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1505 repetition, failure of what follows normally causes the repeated item to be
1506 re-evaluated to see if a different number of repeats allows the rest of the
1507 pattern to match. Sometimes it is useful to prevent this, either to change the
1508 nature of the match, or to cause it fail earlier than it otherwise might, when
1509 the author of the pattern knows there is no point in carrying on.
1510 .P
1511 Consider, for example, the pattern \ed+foo when applied to the subject line
1512 .sp
1513 123456bar
1514 .sp
1515 After matching all 6 digits and then failing to match "foo", the normal
1516 action of the matcher is to try again with only 5 digits matching the \ed+
1517 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1518 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1519 that once a subpattern has matched, it is not to be re-evaluated in this way.
1520 .P
1521 If we use atomic grouping for the previous example, the matcher gives up
1522 immediately on failing to match "foo" the first time. The notation is a kind of
1523 special parenthesis, starting with (?> as in this example:
1524 .sp
1525 (?>\ed+)foo
1526 .sp
1527 This kind of parenthesis "locks up" the part of the pattern it contains once
1528 it has matched, and a failure further into the pattern is prevented from
1529 backtracking into it. Backtracking past it to previous items, however, works as
1530 normal.
1531 .P
1532 An alternative description is that a subpattern of this type matches the string
1533 of characters that an identical standalone pattern would match, if anchored at
1534 the current point in the subject string.
1535 .P
1536 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1537 the above example can be thought of as a maximizing repeat that must swallow
1538 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1539 number of digits they match in order to make the rest of the pattern match,
1540 (?>\ed+) can only match an entire sequence of digits.
1541 .P
1542 Atomic groups in general can of course contain arbitrarily complicated
1543 subpatterns, and can be nested. However, when the subpattern for an atomic
1544 group is just a single repeated item, as in the example above, a simpler
1545 notation, called a "possessive quantifier" can be used. This consists of an
1546 additional + character following a quantifier. Using this notation, the
1547 previous example can be rewritten as
1548 .sp
1549 \ed++foo
1550 .sp
1551 Note that a possessive quantifier can be used with an entire group, for
1552 example:
1553 .sp
1554 (abc|xyz){2,3}+
1555 .sp
1556 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1557 option is ignored. They are a convenient notation for the simpler forms of
1558 atomic group. However, there is no difference in the meaning of a possessive
1559 quantifier and the equivalent atomic group, though there may be a performance
1560 difference; possessive quantifiers should be slightly faster.
1561 .P
1562 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1563 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1564 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1565 package, and PCRE copied it from there. It ultimately found its way into Perl
1566 at release 5.10.
1567 .P
1568 PCRE has an optimization that automatically "possessifies" certain simple
1569 pattern constructs. For example, the sequence A+B is treated as A++B because
1570 there is no point in backtracking into a sequence of A's when B must follow.
1571 .P
1572 When a pattern contains an unlimited repeat inside a subpattern that can itself
1573 be repeated an unlimited number of times, the use of an atomic group is the
1574 only way to avoid some failing matches taking a very long time indeed. The
1575 pattern
1576 .sp
1577 (\eD+|<\ed+>)*[!?]
1578 .sp
1579 matches an unlimited number of substrings that either consist of non-digits, or
1580 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1581 quickly. However, if it is applied to
1582 .sp
1583 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1584 .sp
1585 it takes a long time before reporting failure. This is because the string can
1586 be divided between the internal \eD+ repeat and the external * repeat in a
1587 large number of ways, and all have to be tried. (The example uses [!?] rather
1588 than a single character at the end, because both PCRE and Perl have an
1589 optimization that allows for fast failure when a single character is used. They
1590 remember the last single character that is required for a match, and fail early
1591 if it is not present in the string.) If the pattern is changed so that it uses
1592 an atomic group, like this:
1593 .sp
1594 ((?>\eD+)|<\ed+>)*[!?]
1595 .sp
1596 sequences of non-digits cannot be broken, and failure happens quickly.
1597 .
1598 .
1599 .\" HTML <a name="backreferences"></a>
1601 .rs
1602 .sp
1603 Outside a character class, a backslash followed by a digit greater than 0 (and
1604 possibly further digits) is a back reference to a capturing subpattern earlier
1605 (that is, to its left) in the pattern, provided there have been that many
1606 previous capturing left parentheses.
1607 .P
1608 However, if the decimal number following the backslash is less than 10, it is
1609 always taken as a back reference, and causes an error only if there are not
1610 that many capturing left parentheses in the entire pattern. In other words, the
1611 parentheses that are referenced need not be to the left of the reference for
1612 numbers less than 10. A "forward back reference" of this type can make sense
1613 when a repetition is involved and the subpattern to the right has participated
1614 in an earlier iteration.
1615 .P
1616 It is not possible to have a numerical "forward back reference" to a subpattern
1617 whose number is 10 or more using this syntax because a sequence such as \e50 is
1618 interpreted as a character defined in octal. See the subsection entitled
1619 "Non-printing characters"
1620 .\" HTML <a href="#digitsafterbackslash">
1621 .\" </a>
1622 above
1623 .\"
1624 for further details of the handling of digits following a backslash. There is
1625 no such problem when named parentheses are used. A back reference to any
1626 subpattern is possible using named parentheses (see below).
1627 .P
1628 Another way of avoiding the ambiguity inherent in the use of digits following a
1629 backslash is to use the \eg escape sequence, which is a feature introduced in
1630 Perl 5.10. This escape must be followed by an unsigned number or a negative
1631 number, optionally enclosed in braces. These examples are all identical:
1632 .sp
1633 (ring), \e1
1634 (ring), \eg1
1635 (ring), \eg{1}
1636 .sp
1637 An unsigned number specifies an absolute reference without the ambiguity that
1638 is present in the older syntax. It is also useful when literal digits follow
1639 the reference. A negative number is a relative reference. Consider this
1640 example:
1641 .sp
1642 (abc(def)ghi)\eg{-1}
1643 .sp
1644 The sequence \eg{-1} is a reference to the most recently started capturing
1645 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1646 would be equivalent to \e1. The use of relative references can be helpful in
1647 long patterns, and also in patterns that are created by joining together
1648 fragments that contain references within themselves.
1649 .P
1650 A back reference matches whatever actually matched the capturing subpattern in
1651 the current subject string, rather than anything matching the subpattern
1652 itself (see
1653 .\" HTML <a href="#subpatternsassubroutines">
1654 .\" </a>
1655 "Subpatterns as subroutines"
1656 .\"
1657 below for a way of doing that). So the pattern
1658 .sp
1659 (sens|respons)e and \e1ibility
1660 .sp
1661 matches "sense and sensibility" and "response and responsibility", but not
1662 "sense and responsibility". If caseful matching is in force at the time of the
1663 back reference, the case of letters is relevant. For example,
1664 .sp
1665 ((?i)rah)\es+\e1
1666 .sp
1667 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1668 capturing subpattern is matched caselessly.
1669 .P
1670 There are several different ways of writing back references to named
1671 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1672 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1673 back reference syntax, in which \eg can be used for both numeric and named
1674 references, is also supported. We could rewrite the above example in any of
1675 the following ways:
1676 .sp
1677 (?<p1>(?i)rah)\es+\ek<p1>
1678 (?'p1'(?i)rah)\es+\ek{p1}
1679 (?P<p1>(?i)rah)\es+(?P=p1)
1680 (?<p1>(?i)rah)\es+\eg{p1}
1681 .sp
1682 A subpattern that is referenced by name may appear in the pattern before or
1683 after the reference.
1684 .P
1685 There may be more than one back reference to the same subpattern. If a
1686 subpattern has not actually been used in a particular match, any back
1687 references to it always fail by default. For example, the pattern
1688 .sp
1689 (a|(bc))\e2
1690 .sp
1691 always fails if it starts to match "a" rather than "bc". However, if the
1692 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1693 unset value matches an empty string.
1694 .P
1695 Because there may be many capturing parentheses in a pattern, all digits
1696 following a backslash are taken as part of a potential back reference number.
1697 If the pattern continues with a digit character, some delimiter must be used to
1698 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1699 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1700 .\" HTML <a href="#comments">
1701 .\" </a>
1702 "Comments"
1703 .\"
1704 below) can be used.
1705 .
1706 .SS "Recursive back references"
1707 .rs
1708 .sp
1709 A back reference that occurs inside the parentheses to which it refers fails
1710 when the subpattern is first used, so, for example, (a\e1) never matches.
1711 However, such references can be useful inside repeated subpatterns. For
1712 example, the pattern
1713 .sp
1714 (a|b\e1)+
1715 .sp
1716 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1717 the subpattern, the back reference matches the character string corresponding
1718 to the previous iteration. In order for this to work, the pattern must be such
1719 that the first iteration does not need to match the back reference. This can be
1720 done using alternation, as in the example above, or by a quantifier with a
1721 minimum of zero.
1722 .P
1723 Back references of this type cause the group that they reference to be treated
1724 as an
1725 .\" HTML <a href="#atomicgroup">
1726 .\" </a>
1727 atomic group.
1728 .\"
1729 Once the whole group has been matched, a subsequent matching failure cannot
1730 cause backtracking into the middle of the group.
1731 .
1732 .
1733 .\" HTML <a name="bigassertions"></a>
1735 .rs
1736 .sp
1737 An assertion is a test on the characters following or preceding the current
1738 matching point that does not actually consume any characters. The simple
1739 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1740 .\" HTML <a href="#smallassertions">
1741 .\" </a>
1742 above.
1743 .\"
1744 .P
1745 More complicated assertions are coded as subpatterns. There are two kinds:
1746 those that look ahead of the current position in the subject string, and those
1747 that look behind it. An assertion subpattern is matched in the normal way,
1748 except that it does not cause the current matching position to be changed.
1749 .P
1750 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1751 because it makes no sense to assert the same thing several times. If any kind
1752 of assertion contains capturing subpatterns within it, these are counted for
1753 the purposes of numbering the capturing subpatterns in the whole pattern.
1754 However, substring capturing is carried out only for positive assertions,
1755 because it does not make sense for negative assertions.
1756 .
1757 .
1758 .SS "Lookahead assertions"
1759 .rs
1760 .sp
1761 Lookahead assertions start with (?= for positive assertions and (?! for
1762 negative assertions. For example,
1763 .sp
1764 \ew+(?=;)
1765 .sp
1766 matches a word followed by a semicolon, but does not include the semicolon in
1767 the match, and
1768 .sp
1769 foo(?!bar)
1770 .sp
1771 matches any occurrence of "foo" that is not followed by "bar". Note that the
1772 apparently similar pattern
1773 .sp
1774 (?!foo)bar
1775 .sp
1776 does not find an occurrence of "bar" that is preceded by something other than
1777 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1778 (?!foo) is always true when the next three characters are "bar". A
1779 lookbehind assertion is needed to achieve the other effect.
1780 .P
1781 If you want to force a matching failure at some point in a pattern, the most
1782 convenient way to do it is with (?!) because an empty string always matches, so
1783 an assertion that requires there not to be an empty string must always fail.
1784 The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a
1785 synonym for (?!).
1786 .
1787 .
1788 .\" HTML <a name="lookbehind"></a>
1789 .SS "Lookbehind assertions"
1790 .rs
1791 .sp
1792 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1793 negative assertions. For example,
1794 .sp
1795 (?<!foo)bar
1796 .sp
1797 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1798 a lookbehind assertion are restricted such that all the strings it matches must
1799 have a fixed length. However, if there are several top-level alternatives, they
1800 do not all have to have the same fixed length. Thus
1801 .sp
1802 (?<=bullock|donkey)
1803 .sp
1804 is permitted, but
1805 .sp
1806 (?<!dogs?|cats?)
1807 .sp
1808 causes an error at compile time. Branches that match different length strings
1809 are permitted only at the top level of a lookbehind assertion. This is an
1810 extension compared with Perl (5.8 and 5.10), which requires all branches to
1811 match the same length of string. An assertion such as
1812 .sp
1813 (?<=ab(c|de))
1814 .sp
1815 is not permitted, because its single top-level branch can match two different
1816 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1817 branches:
1818 .sp
1819 (?<=abc|abde)
1820 .sp
1821 In some cases, the Perl 5.10 escape sequence \eK
1822 .\" HTML <a href="#resetmatchstart">
1823 .\" </a>
1824 (see above)
1825 .\"
1826 can be used instead of a lookbehind assertion to get round the fixed-length
1827 restriction.
1828 .P
1829 The implementation of lookbehind assertions is, for each alternative, to
1830 temporarily move the current position back by the fixed length and then try to
1831 match. If there are insufficient characters before the current position, the
1832 assertion fails.
1833 .P
1834 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1835 to appear in lookbehind assertions, because it makes it impossible to calculate
1836 the length of the lookbehind. The \eX and \eR escapes, which can match
1837 different numbers of bytes, are also not permitted.
1838 .P
1839 .\" HTML <a href="#subpatternsassubroutines">
1840 .\" </a>
1841 "Subroutine"
1842 .\"
1843 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1844 as the subpattern matches a fixed-length string.
1845 .\" HTML <a href="#recursion">
1846 .\" </a>
1847 Recursion,
1848 .\"
1849 however, is not supported.
1850 .P
1851 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1852 specify efficient matching of fixed-length strings at the end of subject
1853 strings. Consider a simple pattern such as
1854 .sp
1855 abcd$
1856 .sp
1857 when applied to a long string that does not match. Because matching proceeds
1858 from left to right, PCRE will look for each "a" in the subject and then see if
1859 what follows matches the rest of the pattern. If the pattern is specified as
1860 .sp
1861 ^.*abcd$
1862 .sp
1863 the initial .* matches the entire string at first, but when this fails (because
1864 there is no following "a"), it backtracks to match all but the last character,
1865 then all but the last two characters, and so on. Once again the search for "a"
1866 covers the entire string, from right to left, so we are no better off. However,
1867 if the pattern is written as
1868 .sp
1869 ^.*+(?<=abcd)
1870 .sp
1871 there can be no backtracking for the .*+ item; it can match only the entire
1872 string. The subsequent lookbehind assertion does a single test on the last four
1873 characters. If it fails, the match fails immediately. For long strings, this
1874 approach makes a significant difference to the processing time.
1875 .
1876 .
1877 .SS "Using multiple assertions"
1878 .rs
1879 .sp
1880 Several assertions (of any sort) may occur in succession. For example,
1881 .sp
1882 (?<=\ed{3})(?<!999)foo
1883 .sp
1884 matches "foo" preceded by three digits that are not "999". Notice that each of
1885 the assertions is applied independently at the same point in the subject
1886 string. First there is a check that the previous three characters are all
1887 digits, and then there is a check that the same three characters are not "999".
1888 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1889 of which are digits and the last three of which are not "999". For example, it
1890 doesn't match "123abcfoo". A pattern to do that is
1891 .sp
1892 (?<=\ed{3}...)(?<!999)foo
1893 .sp
1894 This time the first assertion looks at the preceding six characters, checking
1895 that the first three are digits, and then the second assertion checks that the
1896 preceding three characters are not "999".
1897 .P
1898 Assertions can be nested in any combination. For example,
1899 .sp
1900 (?<=(?<!foo)bar)baz
1901 .sp
1902 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1903 preceded by "foo", while
1904 .sp
1905 (?<=\ed{3}(?!999)...)foo
1906 .sp
1907 is another pattern that matches "foo" preceded by three digits and any three
1908 characters that are not "999".
1909 .
1910 .
1911 .\" HTML <a name="conditions"></a>
1913 .rs
1914 .sp
1915 It is possible to cause the matching process to obey a subpattern
1916 conditionally or to choose between two alternative subpatterns, depending on
1917 the result of an assertion, or whether a specific capturing subpattern has
1918 already been matched. The two possible forms of conditional subpattern are:
1919 .sp
1920 (?(condition)yes-pattern)
1921 (?(condition)yes-pattern|no-pattern)
1922 .sp
1923 If the condition is satisfied, the yes-pattern is used; otherwise the
1924 no-pattern (if present) is used. If there are more than two alternatives in the
1925 subpattern, a compile-time error occurs.
1926 .P
1927 There are four kinds of condition: references to subpatterns, references to
1928 recursion, a pseudo-condition called DEFINE, and assertions.
1929 .
1930 .SS "Checking for a used subpattern by number"
1931 .rs
1932 .sp
1933 If the text between the parentheses consists of a sequence of digits, the
1934 condition is true if a capturing subpattern of that number has previously
1935 matched. If there is more than one capturing subpattern with the same number
1936 (see the earlier
1937 .\"
1938 .\" HTML <a href="#recursion">
1939 .\" </a>
1940 section about duplicate subpattern numbers),
1941 .\"
1942 the condition is true if any of them have been set. An alternative notation is
1943 to precede the digits with a plus or minus sign. In this case, the subpattern
1944 number is relative rather than absolute. The most recently opened parentheses
1945 can be referenced by (?(-1), the next most recent by (?(-2), and so on. In
1946 looping constructs it can also make sense to refer to subsequent groups with
1947 constructs such as (?(+2).
1948 .P
1949 Consider the following pattern, which contains non-significant white space to
1950 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1951 three parts for ease of discussion:
1952 .sp
1953 ( \e( )? [^()]+ (?(1) \e) )
1954 .sp
1955 The first part matches an optional opening parenthesis, and if that
1956 character is present, sets it as the first captured substring. The second part
1957 matches one or more characters that are not parentheses. The third part is a
1958 conditional subpattern that tests whether the first set of parentheses matched
1959 or not. If they did, that is, if subject started with an opening parenthesis,
1960 the condition is true, and so the yes-pattern is executed and a closing
1961 parenthesis is required. Otherwise, since no-pattern is not present, the
1962 subpattern matches nothing. In other words, this pattern matches a sequence of
1963 non-parentheses, optionally enclosed in parentheses.
1964 .P
1965 If you were embedding this pattern in a larger one, you could use a relative
1966 reference:
1967 .sp
1968 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1969 .sp
1970 This makes the fragment independent of the parentheses in the larger pattern.
1971 .
1972 .SS "Checking for a used subpattern by name"
1973 .rs
1974 .sp
1975 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1976 subpattern by name. For compatibility with earlier versions of PCRE, which had
1977 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1978 there is a possible ambiguity with this syntax, because subpattern names may
1979 consist entirely of digits. PCRE looks first for a named subpattern; if it
1980 cannot find one and the name consists entirely of digits, PCRE looks for a
1981 subpattern of that number, which must be greater than zero. Using subpattern
1982 names that consist entirely of digits is not recommended.
1983 .P
1984 Rewriting the above example to use a named subpattern gives this:
1985 .sp
1986 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1987 .sp
1988 If the name used in a condition of this kind is a duplicate, the test is
1989 applied to all subpatterns of the same name, and is true if any one of them has
1990 matched.
1991 .
1992 .SS "Checking for pattern recursion"
1993 .rs
1994 .sp
1995 If the condition is the string (R), and there is no subpattern with the name R,
1996 the condition is true if a recursive call to the whole pattern or any
1997 subpattern has been made. If digits or a name preceded by ampersand follow the
1998 letter R, for example:
1999 .sp
2000 (?(R3)...) or (?(R&name)...)
2001 .sp
2002 the condition is true if the most recent recursion is into a subpattern whose
2003 number or name is given. This condition does not check the entire recursion
2004 stack. If the name used in a condition of this kind is a duplicate, the test is
2005 applied to all subpatterns of the same name, and is true if any one of them is
2006 the most recent recursion.
2007 .P
2008 At "top level", all these recursion test conditions are false.
2009 .\" HTML <a href="#recursion">
2010 .\" </a>
2011 The syntax for recursive patterns
2012 .\"
2013 is described below.
2014 .
2015 .SS "Defining subpatterns for use by reference only"
2016 .rs
2017 .sp
2018 If the condition is the string (DEFINE), and there is no subpattern with the
2019 name DEFINE, the condition is always false. In this case, there may be only one
2020 alternative in the subpattern. It is always skipped if control reaches this
2021 point in the pattern; the idea of DEFINE is that it can be used to define
2022 "subroutines" that can be referenced from elsewhere. (The use of
2023 .\" HTML <a href="#subpatternsassubroutines">
2024 .\" </a>
2025 "subroutines"
2026 .\"
2027 is described below.) For example, a pattern to match an IPv4 address could be
2028 written like this (ignore whitespace and line breaks):
2029 .sp
2030 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2031 \eb (?&byte) (\e.(?&byte)){3} \eb
2032 .sp
2033 The first part of the pattern is a DEFINE group inside which a another group
2034 named "byte" is defined. This matches an individual component of an IPv4
2035 address (a number less than 256). When matching takes place, this part of the
2036 pattern is skipped because DEFINE acts like a false condition. The rest of the
2037 pattern uses references to the named group to match the four dot-separated
2038 components of an IPv4 address, insisting on a word boundary at each end.
2039 .
2040 .SS "Assertion conditions"
2041 .rs
2042 .sp
2043 If the condition is not in any of the above formats, it must be an assertion.
2044 This may be a positive or negative lookahead or lookbehind assertion. Consider
2045 this pattern, again containing non-significant white space, and with the two
2046 alternatives on the second line:
2047 .sp
2048 (?(?=[^a-z]*[a-z])
2049 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2050 .sp
2051 The condition is a positive lookahead assertion that matches an optional
2052 sequence of non-letters followed by a letter. In other words, it tests for the
2053 presence of at least one letter in the subject. If a letter is found, the
2054 subject is matched against the first alternative; otherwise it is matched
2055 against the second. This pattern matches strings in one of the two forms
2056 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2057 .
2058 .
2059 .\" HTML <a name="comments"></a>
2061 .rs
2062 .sp
2063 The sequence (?# marks the start of a comment that continues up to the next
2064 closing parenthesis. Nested parentheses are not permitted. The characters
2065 that make up a comment play no part in the pattern matching at all.
2066 .P
2067 If the PCRE_EXTENDED option is set, an unescaped # character outside a
2068 character class introduces a comment that continues to immediately after the
2069 next newline in the pattern.
2070 .
2071 .
2072 .\" HTML <a name="recursion"></a>
2074 .rs
2075 .sp
2076 Consider the problem of matching a string in parentheses, allowing for
2077 unlimited nested parentheses. Without the use of recursion, the best that can
2078 be done is to use a pattern that matches up to some fixed depth of nesting. It
2079 is not possible to handle an arbitrary nesting depth.
2080 .P
2081 For some time, Perl has provided a facility that allows regular expressions to
2082 recurse (amongst other things). It does this by interpolating Perl code in the
2083 expression at run time, and the code can refer to the expression itself. A Perl
2084 pattern using code interpolation to solve the parentheses problem can be
2085 created like this:
2086 .sp
2087 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2088 .sp
2089 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2090 recursively to the pattern in which it appears.
2091 .P
2092 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2093 supports special syntax for recursion of the entire pattern, and also for
2094 individual subpattern recursion. After its introduction in PCRE and Python,
2095 this kind of recursion was subsequently introduced into Perl at release 5.10.
2096 .P
2097 A special item that consists of (? followed by a number greater than zero and a
2098 closing parenthesis is a recursive call of the subpattern of the given number,
2099 provided that it occurs inside that subpattern. (If not, it is a
2100 .\" HTML <a href="#subpatternsassubroutines">
2101 .\" </a>
2102 "subroutine"
2103 .\"
2104 call, which is described in the next section.) The special item (?R) or (?0) is
2105 a recursive call of the entire regular expression.
2106 .P
2107 This PCRE pattern solves the nested parentheses problem (assume the
2108 PCRE_EXTENDED option is set so that white space is ignored):
2109 .sp
2110 \e( ( [^()]++ | (?R) )* \e)
2111 .sp
2112 First it matches an opening parenthesis. Then it matches any number of
2113 substrings which can either be a sequence of non-parentheses, or a recursive
2114 match of the pattern itself (that is, a correctly parenthesized substring).
2115 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2116 to avoid backtracking into sequences of non-parentheses.
2117 .P
2118 If this were part of a larger pattern, you would not want to recurse the entire
2119 pattern, so instead you could use this:
2120 .sp
2121 ( \e( ( [^()]++ | (?1) )* \e) )
2122 .sp
2123 We have put the pattern into parentheses, and caused the recursion to refer to
2124 them instead of the whole pattern.
2125 .P
2126 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2127 is made easier by the use of relative references (a Perl 5.10 feature).
2128 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
2129 most recently opened parentheses preceding the recursion. In other words, a
2130 negative number counts capturing parentheses leftwards from the point at which
2131 it is encountered.
2132 .P
2133 It is also possible to refer to subsequently opened parentheses, by writing
2134 references such as (?+2). However, these cannot be recursive because the
2135 reference is not inside the parentheses that are referenced. They are always
2136 .\" HTML <a href="#subpatternsassubroutines">
2137 .\" </a>
2138 "subroutine"
2139 .\"
2140 calls, as described in the next section.
2141 .P
2142 An alternative approach is to use named parentheses instead. The Perl syntax
2143 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2144 could rewrite the above example as follows:
2145 .sp
2146 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2147 .sp
2148 If there is more than one subpattern with the same name, the earliest one is
2149 used.
2150 .P
2151 This particular example pattern that we have been looking at contains nested
2152 unlimited repeats, and so the use of a possessive quantifier for matching
2153 strings of non-parentheses is important when applying the pattern to strings
2154 that do not match. For example, when this pattern is applied to
2155 .sp
2156 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2157 .sp
2158 it yields "no match" quickly. However, if a possessive quantifier is not used,
2159 the match runs for a very long time indeed because there are so many different
2160 ways the + and * repeats can carve up the subject, and all have to be tested
2161 before failure can be reported.
2162 .P
2163 At the end of a match, the values of capturing parentheses are those from
2164 the outermost level. If you want to obtain intermediate values, a callout
2165 function can be used (see below and the
2166 .\" HREF
2167 \fBpcrecallout\fP
2168 .\"
2169 documentation). If the pattern above is matched against
2170 .sp
2171 (ab(cd)ef)
2172 .sp
2173 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2174 the last value taken on at the top level. If a capturing subpattern is not
2175 matched at the top level, its final value is unset, even if it is (temporarily)
2176 set at a deeper level.
2177 .P
2178 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2179 obtain extra memory to store data during a recursion, which it does by using
2180 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2181 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2182 .P
2183 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2184 Consider this pattern, which matches text in angle brackets, allowing for
2185 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2186 recursing), whereas any characters are permitted at the outer level.
2187 .sp
2188 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2189 .sp
2190 In this pattern, (?(R) is the start of a conditional subpattern, with two
2191 different alternatives for the recursive and non-recursive cases. The (?R) item
2192 is the actual recursive call.
2193 .
2194 .
2195 .\" HTML <a name="recursiondifference"></a>
2196 .SS "Recursion difference from Perl"
2197 .rs
2198 .sp
2199 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2200 treated as an atomic group. That is, once it has matched some of the subject
2201 string, it is never re-entered, even if it contains untried alternatives and
2202 there is a subsequent matching failure. This can be illustrated by the
2203 following pattern, which purports to match a palindromic string that contains
2204 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2205 .sp
2206 ^(.|(.)(?1)\e2)$
2207 .sp
2208 The idea is that it either matches a single character, or two identical
2209 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2210 it does not if the pattern is longer than three characters. Consider the
2211 subject string "abcba":
2212 .P
2213 At the top level, the first character is matched, but as it is not at the end
2214 of the string, the first alternative fails; the second alternative is taken
2215 and the recursion kicks in. The recursive call to subpattern 1 successfully
2216 matches the next character ("b"). (Note that the beginning and end of line
2217 tests are not part of the recursion).
2218 .P
2219 Back at the top level, the next character ("c") is compared with what
2220 subpattern 2 matched, which was "a". This fails. Because the recursion is
2221 treated as an atomic group, there are now no backtracking points, and so the
2222 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2223 try the second alternative.) However, if the pattern is written with the
2224 alternatives in the other order, things are different:
2225 .sp
2226 ^((.)(?1)\e2|.)$
2227 .sp
2228 This time, the recursing alternative is tried first, and continues to recurse
2229 until it runs out of characters, at which point the recursion fails. But this
2230 time we do have another alternative to try at the higher level. That is the big
2231 difference: in the previous case the remaining alternative is at a deeper
2232 recursion level, which PCRE cannot use.
2233 .P
2234 To change the pattern so that matches all palindromic strings, not just those
2235 with an odd number of characters, it is tempting to change the pattern to this:
2236 .sp
2237 ^((.)(?1)\e2|.?)$
2238 .sp
2239 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2240 deeper recursion has matched a single character, it cannot be entered again in
2241 order to match an empty string. The solution is to separate the two cases, and
2242 write out the odd and even cases as alternatives at the higher level:
2243 .sp
2244 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2245 .sp
2246 If you want to match typical palindromic phrases, the pattern has to ignore all
2247 non-word characters, which can be done like this:
2248 .sp
2249 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2250 .sp
2251 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2252 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2253 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2254 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2255 more) to match typical phrases, and Perl takes so long that you think it has
2256 gone into a loop.
2257 .P
2258 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2259 string does not start with a palindrome that is shorter than the entire string.
2260 For example, although "abcba" is correctly matched, if the subject is "ababa",
2261 PCRE finds the palindrome "aba" at the start, then fails at top level because
2262 the end of the string does not follow. Once again, it cannot jump back into the
2263 recursion to try other alternatives, so the entire match fails.
2264 .
2265 .
2266 .\" HTML <a name="subpatternsassubroutines"></a>
2268 .rs
2269 .sp
2270 If the syntax for a recursive subpattern reference (either by number or by
2271 name) is used outside the parentheses to which it refers, it operates like a
2272 subroutine in a programming language. The "called" subpattern may be defined
2273 before or after the reference. A numbered reference can be absolute or
2274 relative, as in these examples:
2275 .sp
2276 (...(absolute)...)...(?2)...
2277 (...(relative)...)...(?-1)...
2278 (...(?+1)...(relative)...
2279 .sp
2280 An earlier example pointed out that the pattern
2281 .sp
2282 (sens|respons)e and \e1ibility
2283 .sp
2284 matches "sense and sensibility" and "response and responsibility", but not
2285 "sense and responsibility". If instead the pattern
2286 .sp
2287 (sens|respons)e and (?1)ibility
2288 .sp
2289 is used, it does match "sense and responsibility" as well as the other two
2290 strings. Another example is given in the discussion of DEFINE above.
2291 .P
2292 Like recursive subpatterns, a subroutine call is always treated as an atomic
2293 group. That is, once it has matched some of the subject string, it is never
2294 re-entered, even if it contains untried alternatives and there is a subsequent
2295 matching failure. Any capturing parentheses that are set during the subroutine
2296 call revert to their previous values afterwards.
2297 .P
2298 When a subpattern is used as a subroutine, processing options such as
2299 case-independence are fixed when the subpattern is defined. They cannot be
2300 changed for different calls. For example, consider this pattern:
2301 .sp
2302 (abc)(?i:(?-1))
2303 .sp
2304 It matches "abcabc". It does not match "abcABC" because the change of
2305 processing option does not affect the called subpattern.
2306 .
2307 .
2308 .\" HTML <a name="onigurumasubroutines"></a>
2310 .rs
2311 .sp
2312 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2313 a number enclosed either in angle brackets or single quotes, is an alternative
2314 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2315 are two of the examples used above, rewritten using this syntax:
2316 .sp
2317 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2318 (sens|respons)e and \eg'1'ibility
2319 .sp
2320 PCRE supports an extension to Oniguruma: if a number is preceded by a
2321 plus or a minus sign it is taken as a relative reference. For example:
2322 .sp
2323 (abc)(?i:\eg<-1>)
2324 .sp
2325 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2326 synonymous. The former is a back reference; the latter is a subroutine call.
2327 .
2328 .
2330 .rs
2331 .sp
2332 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2333 code to be obeyed in the middle of matching a regular expression. This makes it
2334 possible, amongst other things, to extract different substrings that match the
2335 same pair of parentheses when there is a repetition.
2336 .P
2337 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2338 code. The feature is called "callout". The caller of PCRE provides an external
2339 function by putting its entry point in the global variable \fIpcre_callout\fP.
2340 By default, this variable contains NULL, which disables all calling out.
2341 .P
2342 Within a regular expression, (?C) indicates the points at which the external
2343 function is to be called. If you want to identify different callout points, you
2344 can put a number less than 256 after the letter C. The default value is zero.
2345 For example, this pattern has two callout points:
2346 .sp
2347 (?C1)abc(?C2)def
2348 .sp
2349 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2350 automatically installed before each item in the pattern. They are all numbered
2351 255.
2352 .P
2353 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2354 set), the external function is called. It is provided with the number of the
2355 callout, the position in the pattern, and, optionally, one item of data
2356 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2357 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2358 description of the interface to the callout function is given in the
2359 .\" HREF
2360 \fBpcrecallout\fP
2361 .\"
2362 documentation.
2363 .
2364 .
2365 .\" HTML <a name="backtrackcontrol"></a>
2367 .rs
2368 .sp
2369 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2370 are described in the Perl documentation as "experimental and subject to change
2371 or removal in a future version of Perl". It goes on to say: "Their usage in
2372 production code should be noted to avoid problems during upgrades." The same
2373 remarks apply to the PCRE features described in this section.
2374 .P
2375 Since these verbs are specifically related to backtracking, most of them can be
2376 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2377 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2378 failing negative assertion, they cause an error if encountered by
2379 \fBpcre_dfa_exec()\fP.
2380 .P
2381 If any of these verbs are used in an assertion or subroutine subpattern
2382 (including recursive subpatterns), their effect is confined to that subpattern;
2383 it does not extend to the surrounding pattern. Note that such subpatterns are
2384 processed as anchored at the point where they are tested.
2385 .P
2386 The new verbs make use of what was previously invalid syntax: an opening
2387 parenthesis followed by an asterisk. They are generally of the form
2388 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2389 depending on whether or not an argument is present. An name is a sequence of
2390 letters, digits, and underscores. If the name is empty, that is, if the closing
2391 parenthesis immediately follows the colon, the effect is as if the colon were
2392 not there. Any number of these verbs may occur in a pattern.
2393 .P
2394 PCRE contains some optimizations that are used to speed up matching by running
2395 some checks at the start of each match attempt. For example, it may know the
2396 minimum length of matching subject, or that a particular character must be
2397 present. When one of these optimizations suppresses the running of a match, any
2398 included backtracking verbs will not, of course, be processed. You can suppress
2399 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2400 when calling \fBpcre_exec()\fP.
2401 .
2402 .
2403 .SS "Verbs that act immediately"
2404 .rs
2405 .sp
2406 The following verbs act as soon as they are encountered. They may not be
2407 followed by a name.
2408 .sp
2409 (*ACCEPT)
2410 .sp
2411 This verb causes the match to end successfully, skipping the remainder of the
2412 pattern. When inside a recursion, only the innermost pattern is ended
2413 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2414 captured. (This feature was added to PCRE at release 8.00.) For example:
2415 .sp
2416 A((?:A|B(*ACCEPT)|C)D)
2417 .sp
2418 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2419 the outer parentheses.
2420 .sp
2421 (*FAIL) or (*F)
2422 .sp
2423 This verb causes the match to fail, forcing backtracking to occur. It is
2424 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2425 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2426 Perl features that are not present in PCRE. The nearest equivalent is the
2427 callout feature, as for example in this pattern:
2428 .sp
2429 a+(?C)(*FAIL)
2430 .sp
2431 A match with the string "aaaa" always fails, but the callout is taken before
2432 each backtrack happens (in this example, 10 times).
2433 .
2434 .
2435 .SS "Recording which path was taken"
2436 .rs
2437 .sp
2438 There is one verb whose main purpose is to track how a match was arrived at,
2439 though it also has a secondary use in conjunction with advancing the match
2440 starting point (see (*SKIP) below).
2441 .sp
2442 (*MARK:NAME) or (*:NAME)
2443 .sp
2444 A name is always required with this verb. There may be as many instances of
2445 (*MARK) as you like in a pattern, and their names do not have to be unique.
2446 .P
2447 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2448 to the caller via the \fIpcre_extra\fP data structure, as described in the
2449 .\" HTML <a href="pcreapi.html#extradata">
2450 .\" </a>
2451 section on \fIpcre_extra\fP
2452 .\"
2453 in the
2454 .\" HREF
2455 \fBpcreapi\fP
2456 .\"
2457 documentation. No data is returned for a partial match. Here is an example of
2458 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2459 outputting of (*MARK) data:
2460 .sp
2461 /X(*MARK:A)Y|X(*MARK:B)Z/K
2462 XY
2463 0: XY
2464 MK: A
2465 XZ
2466 0: XZ
2467 MK: B
2468 .sp
2469 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2470 indicates which of the two alternatives matched. This is a more efficient way
2471 of obtaining this information than putting each alternative in its own
2472 capturing parentheses.
2473 .P
2474 A name may also be returned after a failed match if the final path through the
2475 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2476 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2477 starting point for matching is advanced, the final check is often with an empty
2478 string, causing a failure before (*MARK) is reached. For example:
2479 .sp
2480 /X(*MARK:A)Y|X(*MARK:B)Z/K
2481 XP
2482 No match
2483 .sp
2484 There are three potential starting points for this match (starting with X,
2485 starting with P, and with an empty string). If the pattern is anchored, the
2486 result is different:
2487 .sp
2488 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2489 XP
2490 No match, mark = B
2491 .sp
2492 PCRE's start-of-match optimizations can also interfere with this. For example,
2493 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2494 subject length for a match, a shorter subject will not be scanned at all.
2495 .P
2496 Note that similar anomalies (though different in detail) exist in Perl, no
2497 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2498 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2499 .
2500 .
2501 .SS "Verbs that act after backtracking"
2502 .rs
2503 .sp
2504 The following verbs do nothing when they are encountered. Matching continues
2505 with what follows, but if there is no subsequent match, causing a backtrack to
2506 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2507 the verb. However, when one of these verbs appears inside an atomic group, its
2508 effect is confined to that group, because once the group has been matched,
2509 there is never any backtracking into it. In this situation, backtracking can
2510 "jump back" to the left of the entire atomic group. (Remember also, as stated
2511 above, that this localization also applies in subroutine calls and assertions.)
2512 .P
2513 These verbs differ in exactly what kind of failure occurs when backtracking
2514 reaches them.
2515 .sp
2516 (*COMMIT)
2517 .sp
2518 This verb, which may not be followed by a name, causes the whole match to fail
2519 outright if the rest of the pattern does not match. Even if the pattern is
2520 unanchored, no further attempts to find a match by advancing the starting point
2521 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2522 finding a match at the current starting point, or not at all. For example:
2523 .sp
2524 a+(*COMMIT)b
2525 .sp
2526 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2527 dynamic anchor, or "I've started, so I must finish." The name of the most
2528 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2529 match failure.
2530 .P
2531 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2532 unless PCRE's start-of-match optimizations are turned off, as shown in this
2533 \fBpcretest\fP example:
2534 .sp
2535 /(*COMMIT)abc/
2536 xyzabc
2537 0: abc
2538 xyzabc\eY
2539 No match
2540 .sp
2541 PCRE knows that any match must start with "a", so the optimization skips along
2542 the subject to "a" before running the first match attempt, which succeeds. When
2543 the optimization is disabled by the \eY escape in the second subject, the match
2544 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2545 starting points.
2546 .sp
2547 (*PRUNE) or (*PRUNE:NAME)
2548 .sp
2549 This verb causes the match to fail at the current starting position in the
2550 subject if the rest of the pattern does not match. If the pattern is
2551 unanchored, the normal "bumpalong" advance to the next starting character then
2552 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2553 reached, or when matching to the right of (*PRUNE), but if there is no match to
2554 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2555 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2556 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2557 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2558 match fails completely; the name is passed back if this is the final attempt.
2559 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2560 pattern (*PRUNE) has the same effect as (*COMMIT).
2561 .sp
2562 (*SKIP)
2563 .sp
2564 This verb, when given without a name, is like (*PRUNE), except that if the
2565 pattern is unanchored, the "bumpalong" advance is not to the next character,
2566 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2567 signifies that whatever text was matched leading up to it cannot be part of a
2568 successful match. Consider:
2569 .sp
2570 a+(*SKIP)b
2571 .sp
2572 If the subject is "aaaac...", after the first match attempt fails (starting at
2573 the first character in the string), the starting point skips on to start the
2574 next attempt at "c". Note that a possessive quantifer does not have the same
2575 effect as this example; although it would suppress backtracking during the
2576 first match attempt, the second attempt would start at the second character
2577 instead of skipping on to "c".
2578 .sp
2579 (*SKIP:NAME)
2580 .sp
2581 When (*SKIP) has an associated name, its behaviour is modified. If the
2582 following pattern fails to match, the previous path through the pattern is
2583 searched for the most recent (*MARK) that has the same name. If one is found,
2584 the "bumpalong" advance is to the subject position that corresponds to that
2585 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2586 matching name is found, normal "bumpalong" of one character happens (the
2587 (*SKIP) is ignored).
2588 .sp
2589 (*THEN) or (*THEN:NAME)
2590 .sp
2591 This verb causes a skip to the next alternation if the rest of the pattern does
2592 not match. That is, it cancels pending backtracking, but only within the
2593 current alternation. Its name comes from the observation that it can be used
2594 for a pattern-based if-then-else block:
2595 .sp
2596 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2597 .sp
2598 If the COND1 pattern matches, FOO is tried (and possibly further items after
2599 the end of the group if FOO succeeds); on failure the matcher skips to the
2600 second alternative and tries COND2, without backtracking into COND1. The
2601 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2602 overall match fails. If (*THEN) is not directly inside an alternation, it acts
2603 like (*PRUNE).
2604 .
2605 .
2606 .SH "SEE ALSO"
2607 .rs
2608 .sp
2609 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2610 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2611 .
2612 .
2614 .rs
2615 .sp
2616 .nf
2617 Philip Hazel
2618 University Computing Service
2619 Cambridge CB2 3QH, England.
2620 .fi
2621 .
2622 .
2624 .rs
2625 .sp
2626 .nf
2627 Last updated: 05 May 2010
2628 Copyright (c) 1997-2010 University of Cambridge.
2629 .fi


Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

  ViewVC Help
Powered by ViewVC 1.1.5