/[pcre]/code/trunk/doc/pcrepattern.3
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 513 - (show annotations)
Mon May 3 11:13:37 2010 UTC (5 years, 4 months ago) by ph10
File size: 105791 byte(s)
Error occurred while calculating annotation data.
Make \R and \X in a character class behave more like Perl
1 .TH PCREPATTERN 3
2 .SH NAME
3 PCRE - Perl-compatible regular expressions
4 .SH "PCRE REGULAR EXPRESSION DETAILS"
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 .
59 .SH "NEWLINE CONVENTIONS"
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 does not affect what the \eR escape sequence matches. By
99 default, this is any Unicode newline sequence, for Perl compatibility. However,
100 this can be changed; see the description of \eR in the section entitled
101 .\" HTML <a href="#newlineseq">
102 .\" </a>
103 "Newline sequences"
104 .\"
105 below. A change of \eR setting can be combined with a change of newline
106 convention.
107 .
108 .
109 .SH "CHARACTERS AND METACHARACTERS"
110 .rs
111 .sp
112 A regular expression is a pattern that is matched against a subject string from
113 left to right. Most characters stand for themselves in a pattern, and match the
114 corresponding characters in the subject. As a trivial example, the pattern
115 .sp
116 The quick brown fox
117 .sp
118 matches a portion of a subject string that is identical to itself. When
119 caseless matching is specified (the PCRE_CASELESS option), letters are matched
120 independently of case. In UTF-8 mode, PCRE always understands the concept of
121 case for characters whose values are less than 128, so caseless matching is
122 always possible. For characters with higher values, the concept of case is
123 supported if PCRE is compiled with Unicode property support, but not otherwise.
124 If you want to use caseless matching for characters 128 and above, you must
125 ensure that PCRE is compiled with Unicode property support as well as with
126 UTF-8 support.
127 .P
128 The power of regular expressions comes from the ability to include alternatives
129 and repetitions in the pattern. These are encoded in the pattern by the use of
130 \fImetacharacters\fP, which do not stand for themselves but instead are
131 interpreted in some special way.
132 .P
133 There are two different sets of metacharacters: those that are recognized
134 anywhere in the pattern except within square brackets, and those that are
135 recognized within square brackets. Outside square brackets, the metacharacters
136 are as follows:
137 .sp
138 \e general escape character with several uses
139 ^ assert start of string (or line, in multiline mode)
140 $ assert end of string (or line, in multiline mode)
141 . match any character except newline (by default)
142 [ start character class definition
143 | start of alternative branch
144 ( start subpattern
145 ) end subpattern
146 ? extends the meaning of (
147 also 0 or 1 quantifier
148 also quantifier minimizer
149 * 0 or more quantifier
150 + 1 or more quantifier
151 also "possessive quantifier"
152 { start min/max quantifier
153 .sp
154 Part of a pattern that is in square brackets is called a "character class". In
155 a character class the only metacharacters are:
156 .sp
157 \e general escape character
158 ^ negate the class, but only if the first character
159 - indicates character range
160 .\" JOIN
161 [ POSIX character class (only if followed by POSIX
162 syntax)
163 ] terminates the character class
164 .sp
165 The following sections describe the use of each of the metacharacters.
166 .
167 .
168 .SH BACKSLASH
169 .rs
170 .sp
171 The backslash character has several uses. Firstly, if it is followed by a
172 non-alphanumeric character, it takes away any special meaning that character
173 may have. This use of backslash as an escape character applies both inside and
174 outside character classes.
175 .P
176 For example, if you want to match a * character, you write \e* in the pattern.
177 This escaping action applies whether or not the following character would
178 otherwise be interpreted as a metacharacter, so it is always safe to precede a
179 non-alphanumeric with backslash to specify that it stands for itself. In
180 particular, if you want to match a backslash, you write \e\e.
181 .P
182 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
183 pattern (other than in a character class) and characters between a # outside
184 a character class and the next newline are ignored. An escaping backslash can
185 be used to include a whitespace or # character as part of the pattern.
186 .P
187 If you want to remove the special meaning from a sequence of characters, you
188 can do so by putting them between \eQ and \eE. This is different from Perl in
189 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
190 Perl, $ and @ cause variable interpolation. Note the following examples:
191 .sp
192 Pattern PCRE matches Perl matches
193 .sp
194 .\" JOIN
195 \eQabc$xyz\eE abc$xyz abc followed by the
196 contents of $xyz
197 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
198 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
199 .sp
200 The \eQ...\eE sequence is recognized both inside and outside character classes.
201 .
202 .
203 .\" HTML <a name="digitsafterbackslash"></a>
204 .SS "Non-printing characters"
205 .rs
206 .sp
207 A second use of backslash provides a way of encoding non-printing characters
208 in patterns in a visible manner. There is no restriction on the appearance of
209 non-printing characters, apart from the binary zero that terminates a pattern,
210 but when a pattern is being prepared by text editing, it is often easier to use
211 one of the following escape sequences than the binary character it represents:
212 .sp
213 \ea alarm, that is, the BEL character (hex 07)
214 \ecx "control-x", where x is any character
215 \ee escape (hex 1B)
216 \ef formfeed (hex 0C)
217 \en linefeed (hex 0A)
218 \er carriage return (hex 0D)
219 \et tab (hex 09)
220 \eddd character with octal code ddd, or back reference
221 \exhh character with hex code hh
222 \ex{hhh..} character with hex code hhh..
223 .sp
224 The precise effect of \ecx is as follows: if x is a lower case letter, it
225 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
226 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
227 7B.
228 .P
229 After \ex, from zero to two hexadecimal digits are read (letters can be in
230 upper or lower case). Any number of hexadecimal digits may appear between \ex{
231 and }, but the value of the character code must be less than 256 in non-UTF-8
232 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
233 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
234 point, which is 10FFFF.
235 .P
236 If characters other than hexadecimal digits appear between \ex{ and }, or if
237 there is no terminating }, this form of escape is not recognized. Instead, the
238 initial \ex will be interpreted as a basic hexadecimal escape, with no
239 following digits, giving a character whose value is zero.
240 .P
241 Characters whose value is less than 256 can be defined by either of the two
242 syntaxes for \ex. There is no difference in the way they are handled. For
243 example, \exdc is exactly the same as \ex{dc}.
244 .P
245 After \e0 up to two further octal digits are read. If there are fewer than two
246 digits, just those that are present are used. Thus the sequence \e0\ex\e07
247 specifies two binary zeros followed by a BEL character (code value 7). Make
248 sure you supply two digits after the initial zero if the pattern character that
249 follows is itself an octal digit.
250 .P
251 The handling of a backslash followed by a digit other than 0 is complicated.
252 Outside a character class, PCRE reads it and any following digits as a decimal
253 number. If the number is less than 10, or if there have been at least that many
254 previous capturing left parentheses in the expression, the entire sequence is
255 taken as a \fIback reference\fP. A description of how this works is given
256 .\" HTML <a href="#backreferences">
257 .\" </a>
258 later,
259 .\"
260 following the discussion of
261 .\" HTML <a href="#subpattern">
262 .\" </a>
263 parenthesized subpatterns.
264 .\"
265 .P
266 Inside a character class, or if the decimal number is greater than 9 and there
267 have not been that many capturing subpatterns, PCRE re-reads up to three octal
268 digits following the backslash, and uses them to generate a data character. Any
269 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
270 character specified in octal must be less than \e400. In UTF-8 mode, values up
271 to \e777 are permitted. For example:
272 .sp
273 \e040 is another way of writing a space
274 .\" JOIN
275 \e40 is the same, provided there are fewer than 40
276 previous capturing subpatterns
277 \e7 is always a back reference
278 .\" JOIN
279 \e11 might be a back reference, or another way of
280 writing a tab
281 \e011 is always a tab
282 \e0113 is a tab followed by the character "3"
283 .\" JOIN
284 \e113 might be a back reference, otherwise the
285 character with octal code 113
286 .\" JOIN
287 \e377 might be a back reference, otherwise
288 the byte consisting entirely of 1 bits
289 .\" JOIN
290 \e81 is either a back reference, or a binary zero
291 followed by the two characters "8" and "1"
292 .sp
293 Note that octal values of 100 or greater must not be introduced by a leading
294 zero, because no more than three octal digits are ever read.
295 .P
296 All the sequences that define a single character value can be used both inside
297 and outside character classes. In addition, inside a character class, the
298 sequence \eb is interpreted as the backspace character (hex 08). The sequences
299 \eB, \eR, and \eX are not special inside a character class. Like any other
300 unrecognized escape sequences, they are treated as the literal characters "B",
301 "R", and "X" by default, but cause an error if the PCRE_EXTRA option is set.
302 Outside a character class, these sequences have different meanings
303 .\" HTML <a href="#uniextseq">
304 .\" </a>
305 (see below).
306 .\"
307 .
308 .
309 .SS "Absolute and relative back references"
310 .rs
311 .sp
312 The sequence \eg followed by an unsigned or a negative number, optionally
313 enclosed in braces, is an absolute or relative back reference. A named back
314 reference can be coded as \eg{name}. Back references are discussed
315 .\" HTML <a href="#backreferences">
316 .\" </a>
317 later,
318 .\"
319 following the discussion of
320 .\" HTML <a href="#subpattern">
321 .\" </a>
322 parenthesized subpatterns.
323 .\"
324 .
325 .
326 .SS "Absolute and relative subroutine calls"
327 .rs
328 .sp
329 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
330 a number enclosed either in angle brackets or single quotes, is an alternative
331 syntax for referencing a subpattern as a "subroutine". Details are discussed
332 .\" HTML <a href="#onigurumasubroutines">
333 .\" </a>
334 later.
335 .\"
336 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
337 synonymous. The former is a back reference; the latter is a
338 .\" HTML <a href="#subpatternsassubroutines">
339 .\" </a>
340 subroutine
341 .\"
342 call.
343 .
344 .
345 .SS "Generic character types"
346 .rs
347 .sp
348 Another use of backslash is for specifying generic character types. The
349 following are always recognized:
350 .sp
351 \ed any decimal digit
352 \eD any character that is not a decimal digit
353 \eh any horizontal whitespace character
354 \eH any character that is not a horizontal whitespace character
355 \es any whitespace character
356 \eS any character that is not a whitespace character
357 \ev any vertical whitespace character
358 \eV any character that is not a vertical whitespace character
359 \ew any "word" character
360 \eW any "non-word" character
361 .sp
362 Each pair of escape sequences partitions the complete set of characters into
363 two disjoint sets. Any given character matches one, and only one, of each pair.
364 .P
365 These character type sequences can appear both inside and outside character
366 classes. They each match one character of the appropriate type. If the current
367 matching point is at the end of the subject string, all of them fail, since
368 there is no character to match.
369 .P
370 For compatibility with Perl, \es does not match the VT character (code 11).
371 This makes it different from the the POSIX "space" class. The \es characters
372 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
373 included in a Perl script, \es may match the VT character. In PCRE, it never
374 does.
375 .P
376 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
377 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
378 character property support is available. These sequences retain their original
379 meanings from before UTF-8 support was available, mainly for efficiency
380 reasons. Note that this also affects \eb, because it is defined in terms of \ew
381 and \eW.
382 .P
383 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
384 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
385 The horizontal space characters are:
386 .sp
387 U+0009 Horizontal tab
388 U+0020 Space
389 U+00A0 Non-break space
390 U+1680 Ogham space mark
391 U+180E Mongolian vowel separator
392 U+2000 En quad
393 U+2001 Em quad
394 U+2002 En space
395 U+2003 Em space
396 U+2004 Three-per-em space
397 U+2005 Four-per-em space
398 U+2006 Six-per-em space
399 U+2007 Figure space
400 U+2008 Punctuation space
401 U+2009 Thin space
402 U+200A Hair space
403 U+202F Narrow no-break space
404 U+205F Medium mathematical space
405 U+3000 Ideographic space
406 .sp
407 The vertical space characters are:
408 .sp
409 U+000A Linefeed
410 U+000B Vertical tab
411 U+000C Formfeed
412 U+000D Carriage return
413 U+0085 Next line
414 U+2028 Line separator
415 U+2029 Paragraph separator
416 .P
417 A "word" character is an underscore or any character less than 256 that is a
418 letter or digit. The definition of letters and digits is controlled by PCRE's
419 low-valued character tables, and may vary if locale-specific matching is taking
420 place (see
421 .\" HTML <a href="pcreapi.html#localesupport">
422 .\" </a>
423 "Locale support"
424 .\"
425 in the
426 .\" HREF
427 \fBpcreapi\fP
428 .\"
429 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
430 or "french" in Windows, some character codes greater than 128 are used for
431 accented letters, and these are matched by \ew. The use of locales with Unicode
432 is discouraged.
433 .
434 .
435 .\" HTML <a name="newlineseq"></a>
436 .SS "Newline sequences"
437 .rs
438 .sp
439 Outside a character class, by default, the escape sequence \eR matches any
440 Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
441 equivalent to the following:
442 .sp
443 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
444 .sp
445 This is an example of an "atomic group", details of which are given
446 .\" HTML <a href="#atomicgroup">
447 .\" </a>
448 below.
449 .\"
450 This particular group matches either the two-character sequence CR followed by
451 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
452 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
453 line, U+0085). The two-character sequence is treated as a single unit that
454 cannot be split.
455 .P
456 In UTF-8 mode, two additional characters whose codepoints are greater than 255
457 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
458 Unicode character property support is not needed for these characters to be
459 recognized.
460 .P
461 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
462 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
463 either at compile time or when the pattern is matched. (BSR is an abbrevation
464 for "backslash R".) This can be made the default when PCRE is built; if this is
465 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
466 It is also possible to specify these settings by starting a pattern string with
467 one of the following sequences:
468 .sp
469 (*BSR_ANYCRLF) CR, LF, or CRLF only
470 (*BSR_UNICODE) any Unicode newline sequence
471 .sp
472 These override the default and the options given to \fBpcre_compile()\fP or
473 \fBpcre_compile2()\fP, but they can be overridden by options given to
474 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
475 which are not Perl-compatible, are recognized only at the very start of a
476 pattern, and that they must be in upper case. If more than one of them is
477 present, the last one is used. They can be combined with a change of newline
478 convention, for example, a pattern can start with:
479 .sp
480 (*ANY)(*BSR_ANYCRLF)
481 .sp
482 Inside a character class, \eR is treated as an unrecognized escape sequence,
483 and so matches the letter "R" by default, but causes an error if PCRE_EXTRA is
484 set.
485 .
486 .
487 .\" HTML <a name="uniextseq"></a>
488 .SS Unicode character properties
489 .rs
490 .sp
491 When PCRE is built with Unicode character property support, three additional
492 escape sequences that match characters with specific properties are available.
493 When not in UTF-8 mode, these sequences are of course limited to testing
494 characters whose codepoints are less than 256, but they do work in this mode.
495 The extra escape sequences are:
496 .sp
497 \ep{\fIxx\fP} a character with the \fIxx\fP property
498 \eP{\fIxx\fP} a character without the \fIxx\fP property
499 \eX an extended Unicode sequence
500 .sp
501 The property names represented by \fIxx\fP above are limited to the Unicode
502 script names, the general category properties, and "Any", which matches any
503 character (including newline). Other properties such as "InMusicalSymbols" are
504 not currently supported by PCRE. Note that \eP{Any} does not match any
505 characters, so always causes a match failure.
506 .P
507 Sets of Unicode characters are defined as belonging to certain scripts. A
508 character from one of these sets can be matched using a script name. For
509 example:
510 .sp
511 \ep{Greek}
512 \eP{Han}
513 .sp
514 Those that are not part of an identified script are lumped together as
515 "Common". The current list of scripts is:
516 .P
517 Arabic,
518 Armenian,
519 Avestan,
520 Balinese,
521 Bamum,
522 Bengali,
523 Bopomofo,
524 Braille,
525 Buginese,
526 Buhid,
527 Canadian_Aboriginal,
528 Carian,
529 Cham,
530 Cherokee,
531 Common,
532 Coptic,
533 Cuneiform,
534 Cypriot,
535 Cyrillic,
536 Deseret,
537 Devanagari,
538 Egyptian_Hieroglyphs,
539 Ethiopic,
540 Georgian,
541 Glagolitic,
542 Gothic,
543 Greek,
544 Gujarati,
545 Gurmukhi,
546 Han,
547 Hangul,
548 Hanunoo,
549 Hebrew,
550 Hiragana,
551 Imperial_Aramaic,
552 Inherited,
553 Inscriptional_Pahlavi,
554 Inscriptional_Parthian,
555 Javanese,
556 Kaithi,
557 Kannada,
558 Katakana,
559 Kayah_Li,
560 Kharoshthi,
561 Khmer,
562 Lao,
563 Latin,
564 Lepcha,
565 Limbu,
566 Linear_B,
567 Lisu,
568 Lycian,
569 Lydian,
570 Malayalam,
571 Meetei_Mayek,
572 Mongolian,
573 Myanmar,
574 New_Tai_Lue,
575 Nko,
576 Ogham,
577 Old_Italic,
578 Old_Persian,
579 Old_South_Arabian,
580 Old_Turkic,
581 Ol_Chiki,
582 Oriya,
583 Osmanya,
584 Phags_Pa,
585 Phoenician,
586 Rejang,
587 Runic,
588 Samaritan,
589 Saurashtra,
590 Shavian,
591 Sinhala,
592 Sundanese,
593 Syloti_Nagri,
594 Syriac,
595 Tagalog,
596 Tagbanwa,
597 Tai_Le,
598 Tai_Tham,
599 Tai_Viet,
600 Tamil,
601 Telugu,
602 Thaana,
603 Thai,
604 Tibetan,
605 Tifinagh,
606 Ugaritic,
607 Vai,
608 Yi.
609 .P
610 Each character has exactly one general category property, specified by a
611 two-letter abbreviation. For compatibility with Perl, negation can be specified
612 by including a circumflex between the opening brace and the property name. For
613 example, \ep{^Lu} is the same as \eP{Lu}.
614 .P
615 If only one letter is specified with \ep or \eP, it includes all the general
616 category properties that start with that letter. In this case, in the absence
617 of negation, the curly brackets in the escape sequence are optional; these two
618 examples have the same effect:
619 .sp
620 \ep{L}
621 \epL
622 .sp
623 The following general category property codes are supported:
624 .sp
625 C Other
626 Cc Control
627 Cf Format
628 Cn Unassigned
629 Co Private use
630 Cs Surrogate
631 .sp
632 L Letter
633 Ll Lower case letter
634 Lm Modifier letter
635 Lo Other letter
636 Lt Title case letter
637 Lu Upper case letter
638 .sp
639 M Mark
640 Mc Spacing mark
641 Me Enclosing mark
642 Mn Non-spacing mark
643 .sp
644 N Number
645 Nd Decimal number
646 Nl Letter number
647 No Other number
648 .sp
649 P Punctuation
650 Pc Connector punctuation
651 Pd Dash punctuation
652 Pe Close punctuation
653 Pf Final punctuation
654 Pi Initial punctuation
655 Po Other punctuation
656 Ps Open punctuation
657 .sp
658 S Symbol
659 Sc Currency symbol
660 Sk Modifier symbol
661 Sm Mathematical symbol
662 So Other symbol
663 .sp
664 Z Separator
665 Zl Line separator
666 Zp Paragraph separator
667 Zs Space separator
668 .sp
669 The special property L& is also supported: it matches a character that has
670 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
671 a modifier or "other".
672 .P
673 The Cs (Surrogate) property applies only to characters in the range U+D800 to
674 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
675 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
676 (see the discussion of PCRE_NO_UTF8_CHECK in the
677 .\" HREF
678 \fBpcreapi\fP
679 .\"
680 page). Perl does not support the Cs property.
681 .P
682 The long synonyms for property names that Perl supports (such as \ep{Letter})
683 are not supported by PCRE, nor is it permitted to prefix any of these
684 properties with "Is".
685 .P
686 No character that is in the Unicode table has the Cn (unassigned) property.
687 Instead, this property is assumed for any code point that is not in the
688 Unicode table.
689 .P
690 Specifying caseless matching does not affect these escape sequences. For
691 example, \ep{Lu} always matches only upper case letters.
692 .P
693 The \eX escape matches any number of Unicode characters that form an extended
694 Unicode sequence. \eX is equivalent to
695 .sp
696 (?>\ePM\epM*)
697 .sp
698 That is, it matches a character without the "mark" property, followed by zero
699 or more characters with the "mark" property, and treats the sequence as an
700 atomic group
701 .\" HTML <a href="#atomicgroup">
702 .\" </a>
703 (see below).
704 .\"
705 Characters with the "mark" property are typically accents that affect the
706 preceding character. None of them have codepoints less than 256, so in
707 non-UTF-8 mode \eX matches any one character.
708 .P
709 Matching characters by Unicode property is not fast, because PCRE has to search
710 a structure that contains data for over fifteen thousand characters. That is
711 why the traditional escape sequences such as \ed and \ew do not use Unicode
712 properties in PCRE.
713 .
714 .
715 .\" HTML <a name="resetmatchstart"></a>
716 .SS "Resetting the match start"
717 .rs
718 .sp
719 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
720 matched characters not to be included in the final matched sequence. For
721 example, the pattern:
722 .sp
723 foo\eKbar
724 .sp
725 matches "foobar", but reports that it has matched "bar". This feature is
726 similar to a lookbehind assertion
727 .\" HTML <a href="#lookbehind">
728 .\" </a>
729 (described below).
730 .\"
731 However, in this case, the part of the subject before the real match does not
732 have to be of fixed length, as lookbehind assertions do. The use of \eK does
733 not interfere with the setting of
734 .\" HTML <a href="#subpattern">
735 .\" </a>
736 captured substrings.
737 .\"
738 For example, when the pattern
739 .sp
740 (foo)\eKbar
741 .sp
742 matches "foobar", the first substring is still set to "foo".
743 .P
744 Perl documents that the use of \eK within assertions is "not well defined". In
745 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
746 ignored in negative assertions.
747 .
748 .
749 .\" HTML <a name="smallassertions"></a>
750 .SS "Simple assertions"
751 .rs
752 .sp
753 The final use of backslash is for certain simple assertions. An assertion
754 specifies a condition that has to be met at a particular point in a match,
755 without consuming any characters from the subject string. The use of
756 subpatterns for more complicated assertions is described
757 .\" HTML <a href="#bigassertions">
758 .\" </a>
759 below.
760 .\"
761 The backslashed assertions are:
762 .sp
763 \eb matches at a word boundary
764 \eB matches when not at a word boundary
765 \eA matches at the start of the subject
766 \eZ matches at the end of the subject
767 also matches before a newline at the end of the subject
768 \ez matches only at the end of the subject
769 \eG matches at the first matching position in the subject
770 .sp
771 Inside a character class, \eb has a different meaning; it matches the backspace
772 character. If any other of these assertions appears in a character class, by
773 default it matches the corresponding literal character (for example, \eB
774 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
775 escape sequence" error is generated instead.
776 .P
777 A word boundary is a position in the subject string where the current character
778 and the previous character do not both match \ew or \eW (i.e. one matches
779 \ew and the other matches \eW), or the start or end of the string if the
780 first or last character matches \ew, respectively. Neither PCRE nor Perl has a
781 separte "start of word" or "end of word" metasequence. However, whatever
782 follows \eb normally determines which it is. For example, the fragment
783 \eba matches "a" at the start of a word.
784 .P
785 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
786 dollar (described in the next section) in that they only ever match at the very
787 start and end of the subject string, whatever options are set. Thus, they are
788 independent of multiline mode. These three assertions are not affected by the
789 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
790 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
791 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
792 at a point other than the beginning of the subject, \eA can never match. The
793 difference between \eZ and \ez is that \eZ matches before a newline at the end
794 of the string as well as at the very end, whereas \ez matches only at the end.
795 .P
796 The \eG assertion is true only when the current matching position is at the
797 start point of the match, as specified by the \fIstartoffset\fP argument of
798 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
799 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
800 arguments, you can mimic Perl's /g option, and it is in this kind of
801 implementation where \eG can be useful.
802 .P
803 Note, however, that PCRE's interpretation of \eG, as the start of the current
804 match, is subtly different from Perl's, which defines it as the end of the
805 previous match. In Perl, these can be different when the previously matched
806 string was empty. Because PCRE does just one match at a time, it cannot
807 reproduce this behaviour.
808 .P
809 If all the alternatives of a pattern begin with \eG, the expression is anchored
810 to the starting match position, and the "anchored" flag is set in the compiled
811 regular expression.
812 .
813 .
814 .SH "CIRCUMFLEX AND DOLLAR"
815 .rs
816 .sp
817 Outside a character class, in the default matching mode, the circumflex
818 character is an assertion that is true only if the current matching point is
819 at the start of the subject string. If the \fIstartoffset\fP argument of
820 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
821 option is unset. Inside a character class, circumflex has an entirely different
822 meaning
823 .\" HTML <a href="#characterclass">
824 .\" </a>
825 (see below).
826 .\"
827 .P
828 Circumflex need not be the first character of the pattern if a number of
829 alternatives are involved, but it should be the first thing in each alternative
830 in which it appears if the pattern is ever to match that branch. If all
831 possible alternatives start with a circumflex, that is, if the pattern is
832 constrained to match only at the start of the subject, it is said to be an
833 "anchored" pattern. (There are also other constructs that can cause a pattern
834 to be anchored.)
835 .P
836 A dollar character is an assertion that is true only if the current matching
837 point is at the end of the subject string, or immediately before a newline
838 at the end of the string (by default). Dollar need not be the last character of
839 the pattern if a number of alternatives are involved, but it should be the last
840 item in any branch in which it appears. Dollar has no special meaning in a
841 character class.
842 .P
843 The meaning of dollar can be changed so that it matches only at the very end of
844 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
845 does not affect the \eZ assertion.
846 .P
847 The meanings of the circumflex and dollar characters are changed if the
848 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
849 immediately after internal newlines as well as at the start of the subject
850 string. It does not match after a newline that ends the string. A dollar
851 matches before any newlines in the string, as well as at the very end, when
852 PCRE_MULTILINE is set. When newline is specified as the two-character
853 sequence CRLF, isolated CR and LF characters do not indicate newlines.
854 .P
855 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
856 \en represents a newline) in multiline mode, but not otherwise. Consequently,
857 patterns that are anchored in single line mode because all branches start with
858 ^ are not anchored in multiline mode, and a match for circumflex is possible
859 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
860 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
861 .P
862 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
863 end of the subject in both modes, and if all branches of a pattern start with
864 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
865 .
866 .
867 .SH "FULL STOP (PERIOD, DOT)"
868 .rs
869 .sp
870 Outside a character class, a dot in the pattern matches any one character in
871 the subject string except (by default) a character that signifies the end of a
872 line. In UTF-8 mode, the matched character may be more than one byte long.
873 .P
874 When a line ending is defined as a single character, dot never matches that
875 character; when the two-character sequence CRLF is used, dot does not match CR
876 if it is immediately followed by LF, but otherwise it matches all characters
877 (including isolated CRs and LFs). When any Unicode line endings are being
878 recognized, dot does not match CR or LF or any of the other line ending
879 characters.
880 .P
881 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
882 option is set, a dot matches any one character, without exception. If the
883 two-character sequence CRLF is present in the subject string, it takes two dots
884 to match it.
885 .P
886 The handling of dot is entirely independent of the handling of circumflex and
887 dollar, the only relationship being that they both involve newlines. Dot has no
888 special meaning in a character class.
889 .
890 .
891 .SH "MATCHING A SINGLE BYTE"
892 .rs
893 .sp
894 Outside a character class, the escape sequence \eC matches any one byte, both
895 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
896 characters. The feature is provided in Perl in order to match individual bytes
897 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
898 what remains in the string may be a malformed UTF-8 string. For this reason,
899 the \eC escape sequence is best avoided.
900 .P
901 PCRE does not allow \eC to appear in lookbehind assertions
902 .\" HTML <a href="#lookbehind">
903 .\" </a>
904 (described below),
905 .\"
906 because in UTF-8 mode this would make it impossible to calculate the length of
907 the lookbehind.
908 .
909 .
910 .\" HTML <a name="characterclass"></a>
911 .SH "SQUARE BRACKETS AND CHARACTER CLASSES"
912 .rs
913 .sp
914 An opening square bracket introduces a character class, terminated by a closing
915 square bracket. A closing square bracket on its own is not special by default.
916 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
917 bracket causes a compile-time error. If a closing square bracket is required as
918 a member of the class, it should be the first data character in the class
919 (after an initial circumflex, if present) or escaped with a backslash.
920 .P
921 A character class matches a single character in the subject. In UTF-8 mode, the
922 character may be more than one byte long. A matched character must be in the
923 set of characters defined by the class, unless the first character in the class
924 definition is a circumflex, in which case the subject character must not be in
925 the set defined by the class. If a circumflex is actually required as a member
926 of the class, ensure it is not the first character, or escape it with a
927 backslash.
928 .P
929 For example, the character class [aeiou] matches any lower case vowel, while
930 [^aeiou] matches any character that is not a lower case vowel. Note that a
931 circumflex is just a convenient notation for specifying the characters that
932 are in the class by enumerating those that are not. A class that starts with a
933 circumflex is not an assertion; it still consumes a character from the subject
934 string, and therefore it fails if the current pointer is at the end of the
935 string.
936 .P
937 In UTF-8 mode, characters with values greater than 255 can be included in a
938 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
939 .P
940 When caseless matching is set, any letters in a class represent both their
941 upper case and lower case versions, so for example, a caseless [aeiou] matches
942 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
943 caseful version would. In UTF-8 mode, PCRE always understands the concept of
944 case for characters whose values are less than 128, so caseless matching is
945 always possible. For characters with higher values, the concept of case is
946 supported if PCRE is compiled with Unicode property support, but not otherwise.
947 If you want to use caseless matching in UTF8-mode for characters 128 and above,
948 you must ensure that PCRE is compiled with Unicode property support as well as
949 with UTF-8 support.
950 .P
951 Characters that might indicate line breaks are never treated in any special way
952 when matching character classes, whatever line-ending sequence is in use, and
953 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
954 such as [^a] always matches one of these characters.
955 .P
956 The minus (hyphen) character can be used to specify a range of characters in a
957 character class. For example, [d-m] matches any letter between d and m,
958 inclusive. If a minus character is required in a class, it must be escaped with
959 a backslash or appear in a position where it cannot be interpreted as
960 indicating a range, typically as the first or last character in the class.
961 .P
962 It is not possible to have the literal character "]" as the end character of a
963 range. A pattern such as [W-]46] is interpreted as a class of two characters
964 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
965 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
966 the end of range, so [W-\e]46] is interpreted as a class containing a range
967 followed by two other characters. The octal or hexadecimal representation of
968 "]" can also be used to end a range.
969 .P
970 Ranges operate in the collating sequence of character values. They can also be
971 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
972 mode, ranges can include characters whose values are greater than 255, for
973 example [\ex{100}-\ex{2ff}].
974 .P
975 If a range that includes letters is used when caseless matching is set, it
976 matches the letters in either case. For example, [W-c] is equivalent to
977 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
978 tables for a French locale are in use, [\exc8-\excb] matches accented E
979 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
980 characters with values greater than 128 only when it is compiled with Unicode
981 property support.
982 .P
983 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
984 in a character class, and add the characters that they match to the class. For
985 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
986 conveniently be used with the upper case character types to specify a more
987 restricted set of characters than the matching lower case type. For example,
988 the class [^\eW_] matches any letter or digit, but not underscore.
989 .P
990 The only metacharacters that are recognized in character classes are backslash,
991 hyphen (only where it can be interpreted as specifying a range), circumflex
992 (only at the start), opening square bracket (only when it can be interpreted as
993 introducing a POSIX class name - see the next section), and the terminating
994 closing square bracket. However, escaping other non-alphanumeric characters
995 does no harm.
996 .
997 .
998 .SH "POSIX CHARACTER CLASSES"
999 .rs
1000 .sp
1001 Perl supports the POSIX notation for character classes. This uses names
1002 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1003 this notation. For example,
1004 .sp
1005 [01[:alpha:]%]
1006 .sp
1007 matches "0", "1", any alphabetic character, or "%". The supported class names
1008 are
1009 .sp
1010 alnum letters and digits
1011 alpha letters
1012 ascii character codes 0 - 127
1013 blank space or tab only
1014 cntrl control characters
1015 digit decimal digits (same as \ed)
1016 graph printing characters, excluding space
1017 lower lower case letters
1018 print printing characters, including space
1019 punct printing characters, excluding letters and digits
1020 space white space (not quite the same as \es)
1021 upper upper case letters
1022 word "word" characters (same as \ew)
1023 xdigit hexadecimal digits
1024 .sp
1025 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1026 space (32). Notice that this list includes the VT character (code 11). This
1027 makes "space" different to \es, which does not include VT (for Perl
1028 compatibility).
1029 .P
1030 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1031 5.8. Another Perl extension is negation, which is indicated by a ^ character
1032 after the colon. For example,
1033 .sp
1034 [12[:^digit:]]
1035 .sp
1036 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1037 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1038 supported, and an error is given if they are encountered.
1039 .P
1040 In UTF-8 mode, characters with values greater than 128 do not match any of
1041 the POSIX character classes.
1042 .
1043 .
1044 .SH "VERTICAL BAR"
1045 .rs
1046 .sp
1047 Vertical bar characters are used to separate alternative patterns. For example,
1048 the pattern
1049 .sp
1050 gilbert|sullivan
1051 .sp
1052 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1053 and an empty alternative is permitted (matching the empty string). The matching
1054 process tries each alternative in turn, from left to right, and the first one
1055 that succeeds is used. If the alternatives are within a subpattern
1056 .\" HTML <a href="#subpattern">
1057 .\" </a>
1058 (defined below),
1059 .\"
1060 "succeeds" means matching the rest of the main pattern as well as the
1061 alternative in the subpattern.
1062 .
1063 .
1064 .SH "INTERNAL OPTION SETTING"
1065 .rs
1066 .sp
1067 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1068 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1069 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1070 The option letters are
1071 .sp
1072 i for PCRE_CASELESS
1073 m for PCRE_MULTILINE
1074 s for PCRE_DOTALL
1075 x for PCRE_EXTENDED
1076 .sp
1077 For example, (?im) sets caseless, multiline matching. It is also possible to
1078 unset these options by preceding the letter with a hyphen, and a combined
1079 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1080 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1081 permitted. If a letter appears both before and after the hyphen, the option is
1082 unset.
1083 .P
1084 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1085 changed in the same way as the Perl-compatible options by using the characters
1086 J, U and X respectively.
1087 .P
1088 When one of these option changes occurs at top level (that is, not inside
1089 subpattern parentheses), the change applies to the remainder of the pattern
1090 that follows. If the change is placed right at the start of a pattern, PCRE
1091 extracts it into the global options (and it will therefore show up in data
1092 extracted by the \fBpcre_fullinfo()\fP function).
1093 .P
1094 An option change within a subpattern (see below for a description of
1095 subpatterns) affects only that part of the current pattern that follows it, so
1096 .sp
1097 (a(?i)b)c
1098 .sp
1099 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1100 By this means, options can be made to have different settings in different
1101 parts of the pattern. Any changes made in one alternative do carry on
1102 into subsequent branches within the same subpattern. For example,
1103 .sp
1104 (a(?i)b|c)
1105 .sp
1106 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1107 branch is abandoned before the option setting. This is because the effects of
1108 option settings happen at compile time. There would be some very weird
1109 behaviour otherwise.
1110 .P
1111 \fBNote:\fP There are other PCRE-specific options that can be set by the
1112 application when the compile or match functions are called. In some cases the
1113 pattern can contain special leading sequences such as (*CRLF) to override what
1114 the application has set or what has been defaulted. Details are given in the
1115 section entitled
1116 .\" HTML <a href="#newlineseq">
1117 .\" </a>
1118 "Newline sequences"
1119 .\"
1120 above. There is also the (*UTF8) leading sequence that can be used to set UTF-8
1121 mode; this is equivalent to setting the PCRE_UTF8 option.
1122 .
1123 .
1124 .\" HTML <a name="subpattern"></a>
1125 .SH SUBPATTERNS
1126 .rs
1127 .sp
1128 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1129 Turning part of a pattern into a subpattern does two things:
1130 .sp
1131 1. It localizes a set of alternatives. For example, the pattern
1132 .sp
1133 cat(aract|erpillar|)
1134 .sp
1135 matches one of the words "cat", "cataract", or "caterpillar". Without the
1136 parentheses, it would match "cataract", "erpillar" or an empty string.
1137 .sp
1138 2. It sets up the subpattern as a capturing subpattern. This means that, when
1139 the whole pattern matches, that portion of the subject string that matched the
1140 subpattern is passed back to the caller via the \fIovector\fP argument of
1141 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1142 from 1) to obtain numbers for the capturing subpatterns.
1143 .P
1144 For example, if the string "the red king" is matched against the pattern
1145 .sp
1146 the ((red|white) (king|queen))
1147 .sp
1148 the captured substrings are "red king", "red", and "king", and are numbered 1,
1149 2, and 3, respectively.
1150 .P
1151 The fact that plain parentheses fulfil two functions is not always helpful.
1152 There are often times when a grouping subpattern is required without a
1153 capturing requirement. If an opening parenthesis is followed by a question mark
1154 and a colon, the subpattern does not do any capturing, and is not counted when
1155 computing the number of any subsequent capturing subpatterns. For example, if
1156 the string "the white queen" is matched against the pattern
1157 .sp
1158 the ((?:red|white) (king|queen))
1159 .sp
1160 the captured substrings are "white queen" and "queen", and are numbered 1 and
1161 2. The maximum number of capturing subpatterns is 65535.
1162 .P
1163 As a convenient shorthand, if any option settings are required at the start of
1164 a non-capturing subpattern, the option letters may appear between the "?" and
1165 the ":". Thus the two patterns
1166 .sp
1167 (?i:saturday|sunday)
1168 (?:(?i)saturday|sunday)
1169 .sp
1170 match exactly the same set of strings. Because alternative branches are tried
1171 from left to right, and options are not reset until the end of the subpattern
1172 is reached, an option setting in one branch does affect subsequent branches, so
1173 the above patterns match "SUNDAY" as well as "Saturday".
1174 .
1175 .
1176 .\" HTML <a name="dupsubpatternnumber"></a>
1177 .SH "DUPLICATE SUBPATTERN NUMBERS"
1178 .rs
1179 .sp
1180 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1181 the same numbers for its capturing parentheses. Such a subpattern starts with
1182 (?| and is itself a non-capturing subpattern. For example, consider this
1183 pattern:
1184 .sp
1185 (?|(Sat)ur|(Sun))day
1186 .sp
1187 Because the two alternatives are inside a (?| group, both sets of capturing
1188 parentheses are numbered one. Thus, when the pattern matches, you can look
1189 at captured substring number one, whichever alternative matched. This construct
1190 is useful when you want to capture part, but not all, of one of a number of
1191 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1192 number is reset at the start of each branch. The numbers of any capturing
1193 buffers that follow the subpattern start after the highest number used in any
1194 branch. The following example is taken from the Perl documentation.
1195 The numbers underneath show in which buffer the captured content will be
1196 stored.
1197 .sp
1198 # before ---------------branch-reset----------- after
1199 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1200 # 1 2 2 3 2 3 4
1201 .sp
1202 A back reference to a numbered subpattern uses the most recent value that is
1203 set for that number by any subpattern. The following pattern matches "abcabc"
1204 or "defdef":
1205 .sp
1206 /(?|(abc)|(def))\e1/
1207 .sp
1208 In contrast, a recursive or "subroutine" call to a numbered subpattern always
1209 refers to the first one in the pattern with the given number. The following
1210 pattern matches "abcabc" or "defabc":
1211 .sp
1212 /(?|(abc)|(def))(?1)/
1213 .sp
1214 If a
1215 .\" HTML <a href="#conditions">
1216 .\" </a>
1217 condition test
1218 .\"
1219 for a subpattern's having matched refers to a non-unique number, the test is
1220 true if any of the subpatterns of that number have matched.
1221 .P
1222 An alternative approach to using this "branch reset" feature is to use
1223 duplicate named subpatterns, as described in the next section.
1224 .
1225 .
1226 .SH "NAMED SUBPATTERNS"
1227 .rs
1228 .sp
1229 Identifying capturing parentheses by number is simple, but it can be very hard
1230 to keep track of the numbers in complicated regular expressions. Furthermore,
1231 if an expression is modified, the numbers may change. To help with this
1232 difficulty, PCRE supports the naming of subpatterns. This feature was not
1233 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1234 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1235 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1236 have different names, but PCRE does not.
1237 .P
1238 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1239 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1240 parentheses from other parts of the pattern, such as
1241 .\" HTML <a href="#backreferences">
1242 .\" </a>
1243 back references,
1244 .\"
1245 .\" HTML <a href="#recursion">
1246 .\" </a>
1247 recursion,
1248 .\"
1249 and
1250 .\" HTML <a href="#conditions">
1251 .\" </a>
1252 conditions,
1253 .\"
1254 can be made by name as well as by number.
1255 .P
1256 Names consist of up to 32 alphanumeric characters and underscores. Named
1257 capturing parentheses are still allocated numbers as well as names, exactly as
1258 if the names were not present. The PCRE API provides function calls for
1259 extracting the name-to-number translation table from a compiled pattern. There
1260 is also a convenience function for extracting a captured substring by name.
1261 .P
1262 By default, a name must be unique within a pattern, but it is possible to relax
1263 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1264 names are also always permitted for subpatterns with the same number, set up as
1265 described in the previous section.) Duplicate names can be useful for patterns
1266 where only one instance of the named parentheses can match. Suppose you want to
1267 match the name of a weekday, either as a 3-letter abbreviation or as the full
1268 name, and in both cases you want to extract the abbreviation. This pattern
1269 (ignoring the line breaks) does the job:
1270 .sp
1271 (?<DN>Mon|Fri|Sun)(?:day)?|
1272 (?<DN>Tue)(?:sday)?|
1273 (?<DN>Wed)(?:nesday)?|
1274 (?<DN>Thu)(?:rsday)?|
1275 (?<DN>Sat)(?:urday)?
1276 .sp
1277 There are five capturing substrings, but only one is ever set after a match.
1278 (An alternative way of solving this problem is to use a "branch reset"
1279 subpattern, as described in the previous section.)
1280 .P
1281 The convenience function for extracting the data by name returns the substring
1282 for the first (and in this example, the only) subpattern of that name that
1283 matched. This saves searching to find which numbered subpattern it was.
1284 .P
1285 If you make a back reference to a non-unique named subpattern from elsewhere in
1286 the pattern, the one that corresponds to the first occurrence of the name is
1287 used. In the absence of duplicate numbers (see the previous section) this is
1288 the one with the lowest number. If you use a named reference in a condition
1289 test (see the
1290 .\"
1291 .\" HTML <a href="#conditions">
1292 .\" </a>
1293 section about conditions
1294 .\"
1295 below), either to check whether a subpattern has matched, or to check for
1296 recursion, all subpatterns with the same name are tested. If the condition is
1297 true for any one of them, the overall condition is true. This is the same
1298 behaviour as testing by number. For further details of the interfaces for
1299 handling named subpatterns, see the
1300 .\" HREF
1301 \fBpcreapi\fP
1302 .\"
1303 documentation.
1304 .P
1305 \fBWarning:\fP You cannot use different names to distinguish between two
1306 subpatterns with the same number because PCRE uses only the numbers when
1307 matching. For this reason, an error is given at compile time if different names
1308 are given to subpatterns with the same number. However, you can give the same
1309 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1310 .
1311 .
1312 .SH REPETITION
1313 .rs
1314 .sp
1315 Repetition is specified by quantifiers, which can follow any of the following
1316 items:
1317 .sp
1318 a literal data character
1319 the dot metacharacter
1320 the \eC escape sequence
1321 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1322 the \eR escape sequence
1323 an escape such as \ed that matches a single character
1324 a character class
1325 a back reference (see next section)
1326 a parenthesized subpattern (unless it is an assertion)
1327 a recursive or "subroutine" call to a subpattern
1328 .sp
1329 The general repetition quantifier specifies a minimum and maximum number of
1330 permitted matches, by giving the two numbers in curly brackets (braces),
1331 separated by a comma. The numbers must be less than 65536, and the first must
1332 be less than or equal to the second. For example:
1333 .sp
1334 z{2,4}
1335 .sp
1336 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1337 character. If the second number is omitted, but the comma is present, there is
1338 no upper limit; if the second number and the comma are both omitted, the
1339 quantifier specifies an exact number of required matches. Thus
1340 .sp
1341 [aeiou]{3,}
1342 .sp
1343 matches at least 3 successive vowels, but may match many more, while
1344 .sp
1345 \ed{8}
1346 .sp
1347 matches exactly 8 digits. An opening curly bracket that appears in a position
1348 where a quantifier is not allowed, or one that does not match the syntax of a
1349 quantifier, is taken as a literal character. For example, {,6} is not a
1350 quantifier, but a literal string of four characters.
1351 .P
1352 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1353 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1354 which is represented by a two-byte sequence. Similarly, when Unicode property
1355 support is available, \eX{3} matches three Unicode extended sequences, each of
1356 which may be several bytes long (and they may be of different lengths).
1357 .P
1358 The quantifier {0} is permitted, causing the expression to behave as if the
1359 previous item and the quantifier were not present. This may be useful for
1360 subpatterns that are referenced as
1361 .\" HTML <a href="#subpatternsassubroutines">
1362 .\" </a>
1363 subroutines
1364 .\"
1365 from elsewhere in the pattern. Items other than subpatterns that have a {0}
1366 quantifier are omitted from the compiled pattern.
1367 .P
1368 For convenience, the three most common quantifiers have single-character
1369 abbreviations:
1370 .sp
1371 * is equivalent to {0,}
1372 + is equivalent to {1,}
1373 ? is equivalent to {0,1}
1374 .sp
1375 It is possible to construct infinite loops by following a subpattern that can
1376 match no characters with a quantifier that has no upper limit, for example:
1377 .sp
1378 (a?)*
1379 .sp
1380 Earlier versions of Perl and PCRE used to give an error at compile time for
1381 such patterns. However, because there are cases where this can be useful, such
1382 patterns are now accepted, but if any repetition of the subpattern does in fact
1383 match no characters, the loop is forcibly broken.
1384 .P
1385 By default, the quantifiers are "greedy", that is, they match as much as
1386 possible (up to the maximum number of permitted times), without causing the
1387 rest of the pattern to fail. The classic example of where this gives problems
1388 is in trying to match comments in C programs. These appear between /* and */
1389 and within the comment, individual * and / characters may appear. An attempt to
1390 match C comments by applying the pattern
1391 .sp
1392 /\e*.*\e*/
1393 .sp
1394 to the string
1395 .sp
1396 /* first comment */ not comment /* second comment */
1397 .sp
1398 fails, because it matches the entire string owing to the greediness of the .*
1399 item.
1400 .P
1401 However, if a quantifier is followed by a question mark, it ceases to be
1402 greedy, and instead matches the minimum number of times possible, so the
1403 pattern
1404 .sp
1405 /\e*.*?\e*/
1406 .sp
1407 does the right thing with the C comments. The meaning of the various
1408 quantifiers is not otherwise changed, just the preferred number of matches.
1409 Do not confuse this use of question mark with its use as a quantifier in its
1410 own right. Because it has two uses, it can sometimes appear doubled, as in
1411 .sp
1412 \ed??\ed
1413 .sp
1414 which matches one digit by preference, but can match two if that is the only
1415 way the rest of the pattern matches.
1416 .P
1417 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1418 the quantifiers are not greedy by default, but individual ones can be made
1419 greedy by following them with a question mark. In other words, it inverts the
1420 default behaviour.
1421 .P
1422 When a parenthesized subpattern is quantified with a minimum repeat count that
1423 is greater than 1 or with a limited maximum, more memory is required for the
1424 compiled pattern, in proportion to the size of the minimum or maximum.
1425 .P
1426 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1427 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1428 implicitly anchored, because whatever follows will be tried against every
1429 character position in the subject string, so there is no point in retrying the
1430 overall match at any position after the first. PCRE normally treats such a
1431 pattern as though it were preceded by \eA.
1432 .P
1433 In cases where it is known that the subject string contains no newlines, it is
1434 worth setting PCRE_DOTALL in order to obtain this optimization, or
1435 alternatively using ^ to indicate anchoring explicitly.
1436 .P
1437 However, there is one situation where the optimization cannot be used. When .*
1438 is inside capturing parentheses that are the subject of a back reference
1439 elsewhere in the pattern, a match at the start may fail where a later one
1440 succeeds. Consider, for example:
1441 .sp
1442 (.*)abc\e1
1443 .sp
1444 If the subject is "xyz123abc123" the match point is the fourth character. For
1445 this reason, such a pattern is not implicitly anchored.
1446 .P
1447 When a capturing subpattern is repeated, the value captured is the substring
1448 that matched the final iteration. For example, after
1449 .sp
1450 (tweedle[dume]{3}\es*)+
1451 .sp
1452 has matched "tweedledum tweedledee" the value of the captured substring is
1453 "tweedledee". However, if there are nested capturing subpatterns, the
1454 corresponding captured values may have been set in previous iterations. For
1455 example, after
1456 .sp
1457 /(a|(b))+/
1458 .sp
1459 matches "aba" the value of the second captured substring is "b".
1460 .
1461 .
1462 .\" HTML <a name="atomicgroup"></a>
1463 .SH "ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS"
1464 .rs
1465 .sp
1466 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1467 repetition, failure of what follows normally causes the repeated item to be
1468 re-evaluated to see if a different number of repeats allows the rest of the
1469 pattern to match. Sometimes it is useful to prevent this, either to change the
1470 nature of the match, or to cause it fail earlier than it otherwise might, when
1471 the author of the pattern knows there is no point in carrying on.
1472 .P
1473 Consider, for example, the pattern \ed+foo when applied to the subject line
1474 .sp
1475 123456bar
1476 .sp
1477 After matching all 6 digits and then failing to match "foo", the normal
1478 action of the matcher is to try again with only 5 digits matching the \ed+
1479 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1480 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1481 that once a subpattern has matched, it is not to be re-evaluated in this way.
1482 .P
1483 If we use atomic grouping for the previous example, the matcher gives up
1484 immediately on failing to match "foo" the first time. The notation is a kind of
1485 special parenthesis, starting with (?> as in this example:
1486 .sp
1487 (?>\ed+)foo
1488 .sp
1489 This kind of parenthesis "locks up" the part of the pattern it contains once
1490 it has matched, and a failure further into the pattern is prevented from
1491 backtracking into it. Backtracking past it to previous items, however, works as
1492 normal.
1493 .P
1494 An alternative description is that a subpattern of this type matches the string
1495 of characters that an identical standalone pattern would match, if anchored at
1496 the current point in the subject string.
1497 .P
1498 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1499 the above example can be thought of as a maximizing repeat that must swallow
1500 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1501 number of digits they match in order to make the rest of the pattern match,
1502 (?>\ed+) can only match an entire sequence of digits.
1503 .P
1504 Atomic groups in general can of course contain arbitrarily complicated
1505 subpatterns, and can be nested. However, when the subpattern for an atomic
1506 group is just a single repeated item, as in the example above, a simpler
1507 notation, called a "possessive quantifier" can be used. This consists of an
1508 additional + character following a quantifier. Using this notation, the
1509 previous example can be rewritten as
1510 .sp
1511 \ed++foo
1512 .sp
1513 Note that a possessive quantifier can be used with an entire group, for
1514 example:
1515 .sp
1516 (abc|xyz){2,3}+
1517 .sp
1518 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1519 option is ignored. They are a convenient notation for the simpler forms of
1520 atomic group. However, there is no difference in the meaning of a possessive
1521 quantifier and the equivalent atomic group, though there may be a performance
1522 difference; possessive quantifiers should be slightly faster.
1523 .P
1524 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1525 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1526 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1527 package, and PCRE copied it from there. It ultimately found its way into Perl
1528 at release 5.10.
1529 .P
1530 PCRE has an optimization that automatically "possessifies" certain simple
1531 pattern constructs. For example, the sequence A+B is treated as A++B because
1532 there is no point in backtracking into a sequence of A's when B must follow.
1533 .P
1534 When a pattern contains an unlimited repeat inside a subpattern that can itself
1535 be repeated an unlimited number of times, the use of an atomic group is the
1536 only way to avoid some failing matches taking a very long time indeed. The
1537 pattern
1538 .sp
1539 (\eD+|<\ed+>)*[!?]
1540 .sp
1541 matches an unlimited number of substrings that either consist of non-digits, or
1542 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1543 quickly. However, if it is applied to
1544 .sp
1545 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1546 .sp
1547 it takes a long time before reporting failure. This is because the string can
1548 be divided between the internal \eD+ repeat and the external * repeat in a
1549 large number of ways, and all have to be tried. (The example uses [!?] rather
1550 than a single character at the end, because both PCRE and Perl have an
1551 optimization that allows for fast failure when a single character is used. They
1552 remember the last single character that is required for a match, and fail early
1553 if it is not present in the string.) If the pattern is changed so that it uses
1554 an atomic group, like this:
1555 .sp
1556 ((?>\eD+)|<\ed+>)*[!?]
1557 .sp
1558 sequences of non-digits cannot be broken, and failure happens quickly.
1559 .
1560 .
1561 .\" HTML <a name="backreferences"></a>
1562 .SH "BACK REFERENCES"
1563 .rs
1564 .sp
1565 Outside a character class, a backslash followed by a digit greater than 0 (and
1566 possibly further digits) is a back reference to a capturing subpattern earlier
1567 (that is, to its left) in the pattern, provided there have been that many
1568 previous capturing left parentheses.
1569 .P
1570 However, if the decimal number following the backslash is less than 10, it is
1571 always taken as a back reference, and causes an error only if there are not
1572 that many capturing left parentheses in the entire pattern. In other words, the
1573 parentheses that are referenced need not be to the left of the reference for
1574 numbers less than 10. A "forward back reference" of this type can make sense
1575 when a repetition is involved and the subpattern to the right has participated
1576 in an earlier iteration.
1577 .P
1578 It is not possible to have a numerical "forward back reference" to a subpattern
1579 whose number is 10 or more using this syntax because a sequence such as \e50 is
1580 interpreted as a character defined in octal. See the subsection entitled
1581 "Non-printing characters"
1582 .\" HTML <a href="#digitsafterbackslash">
1583 .\" </a>
1584 above
1585 .\"
1586 for further details of the handling of digits following a backslash. There is
1587 no such problem when named parentheses are used. A back reference to any
1588 subpattern is possible using named parentheses (see below).
1589 .P
1590 Another way of avoiding the ambiguity inherent in the use of digits following a
1591 backslash is to use the \eg escape sequence, which is a feature introduced in
1592 Perl 5.10. This escape must be followed by an unsigned number or a negative
1593 number, optionally enclosed in braces. These examples are all identical:
1594 .sp
1595 (ring), \e1
1596 (ring), \eg1
1597 (ring), \eg{1}
1598 .sp
1599 An unsigned number specifies an absolute reference without the ambiguity that
1600 is present in the older syntax. It is also useful when literal digits follow
1601 the reference. A negative number is a relative reference. Consider this
1602 example:
1603 .sp
1604 (abc(def)ghi)\eg{-1}
1605 .sp
1606 The sequence \eg{-1} is a reference to the most recently started capturing
1607 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1608 would be equivalent to \e1. The use of relative references can be helpful in
1609 long patterns, and also in patterns that are created by joining together
1610 fragments that contain references within themselves.
1611 .P
1612 A back reference matches whatever actually matched the capturing subpattern in
1613 the current subject string, rather than anything matching the subpattern
1614 itself (see
1615 .\" HTML <a href="#subpatternsassubroutines">
1616 .\" </a>
1617 "Subpatterns as subroutines"
1618 .\"
1619 below for a way of doing that). So the pattern
1620 .sp
1621 (sens|respons)e and \e1ibility
1622 .sp
1623 matches "sense and sensibility" and "response and responsibility", but not
1624 "sense and responsibility". If caseful matching is in force at the time of the
1625 back reference, the case of letters is relevant. For example,
1626 .sp
1627 ((?i)rah)\es+\e1
1628 .sp
1629 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1630 capturing subpattern is matched caselessly.
1631 .P
1632 There are several different ways of writing back references to named
1633 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1634 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1635 back reference syntax, in which \eg can be used for both numeric and named
1636 references, is also supported. We could rewrite the above example in any of
1637 the following ways:
1638 .sp
1639 (?<p1>(?i)rah)\es+\ek<p1>
1640 (?'p1'(?i)rah)\es+\ek{p1}
1641 (?P<p1>(?i)rah)\es+(?P=p1)
1642 (?<p1>(?i)rah)\es+\eg{p1}
1643 .sp
1644 A subpattern that is referenced by name may appear in the pattern before or
1645 after the reference.
1646 .P
1647 There may be more than one back reference to the same subpattern. If a
1648 subpattern has not actually been used in a particular match, any back
1649 references to it always fail by default. For example, the pattern
1650 .sp
1651 (a|(bc))\e2
1652 .sp
1653 always fails if it starts to match "a" rather than "bc". However, if the
1654 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1655 unset value matches an empty string.
1656 .P
1657 Because there may be many capturing parentheses in a pattern, all digits
1658 following a backslash are taken as part of a potential back reference number.
1659 If the pattern continues with a digit character, some delimiter must be used to
1660 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1661 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1662 .\" HTML <a href="#comments">
1663 .\" </a>
1664 "Comments"
1665 .\"
1666 below) can be used.
1667 .
1668 .SS "Recursive back references"
1669 .rs
1670 .sp
1671 A back reference that occurs inside the parentheses to which it refers fails
1672 when the subpattern is first used, so, for example, (a\e1) never matches.
1673 However, such references can be useful inside repeated subpatterns. For
1674 example, the pattern
1675 .sp
1676 (a|b\e1)+
1677 .sp
1678 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1679 the subpattern, the back reference matches the character string corresponding
1680 to the previous iteration. In order for this to work, the pattern must be such
1681 that the first iteration does not need to match the back reference. This can be
1682 done using alternation, as in the example above, or by a quantifier with a
1683 minimum of zero.
1684 .P
1685 Back references of this type cause the group that they reference to be treated
1686 as an
1687 .\" HTML <a href="#atomicgroup">
1688 .\" </a>
1689 atomic group.
1690 .\"
1691 Once the whole group has been matched, a subsequent matching failure cannot
1692 cause backtracking into the middle of the group.
1693 .
1694 .
1695 .\" HTML <a name="bigassertions"></a>
1696 .SH ASSERTIONS
1697 .rs
1698 .sp
1699 An assertion is a test on the characters following or preceding the current
1700 matching point that does not actually consume any characters. The simple
1701 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1702 .\" HTML <a href="#smallassertions">
1703 .\" </a>
1704 above.
1705 .\"
1706 .P
1707 More complicated assertions are coded as subpatterns. There are two kinds:
1708 those that look ahead of the current position in the subject string, and those
1709 that look behind it. An assertion subpattern is matched in the normal way,
1710 except that it does not cause the current matching position to be changed.
1711 .P
1712 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1713 because it makes no sense to assert the same thing several times. If any kind
1714 of assertion contains capturing subpatterns within it, these are counted for
1715 the purposes of numbering the capturing subpatterns in the whole pattern.
1716 However, substring capturing is carried out only for positive assertions,
1717 because it does not make sense for negative assertions.
1718 .
1719 .
1720 .SS "Lookahead assertions"
1721 .rs
1722 .sp
1723 Lookahead assertions start with (?= for positive assertions and (?! for
1724 negative assertions. For example,
1725 .sp
1726 \ew+(?=;)
1727 .sp
1728 matches a word followed by a semicolon, but does not include the semicolon in
1729 the match, and
1730 .sp
1731 foo(?!bar)
1732 .sp
1733 matches any occurrence of "foo" that is not followed by "bar". Note that the
1734 apparently similar pattern
1735 .sp
1736 (?!foo)bar
1737 .sp
1738 does not find an occurrence of "bar" that is preceded by something other than
1739 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1740 (?!foo) is always true when the next three characters are "bar". A
1741 lookbehind assertion is needed to achieve the other effect.
1742 .P
1743 If you want to force a matching failure at some point in a pattern, the most
1744 convenient way to do it is with (?!) because an empty string always matches, so
1745 an assertion that requires there not to be an empty string must always fail.
1746 The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a
1747 synonym for (?!).
1748 .
1749 .
1750 .\" HTML <a name="lookbehind"></a>
1751 .SS "Lookbehind assertions"
1752 .rs
1753 .sp
1754 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1755 negative assertions. For example,
1756 .sp
1757 (?<!foo)bar
1758 .sp
1759 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1760 a lookbehind assertion are restricted such that all the strings it matches must
1761 have a fixed length. However, if there are several top-level alternatives, they
1762 do not all have to have the same fixed length. Thus
1763 .sp
1764 (?<=bullock|donkey)
1765 .sp
1766 is permitted, but
1767 .sp
1768 (?<!dogs?|cats?)
1769 .sp
1770 causes an error at compile time. Branches that match different length strings
1771 are permitted only at the top level of a lookbehind assertion. This is an
1772 extension compared with Perl (5.8 and 5.10), which requires all branches to
1773 match the same length of string. An assertion such as
1774 .sp
1775 (?<=ab(c|de))
1776 .sp
1777 is not permitted, because its single top-level branch can match two different
1778 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1779 branches:
1780 .sp
1781 (?<=abc|abde)
1782 .sp
1783 In some cases, the Perl 5.10 escape sequence \eK
1784 .\" HTML <a href="#resetmatchstart">
1785 .\" </a>
1786 (see above)
1787 .\"
1788 can be used instead of a lookbehind assertion to get round the fixed-length
1789 restriction.
1790 .P
1791 The implementation of lookbehind assertions is, for each alternative, to
1792 temporarily move the current position back by the fixed length and then try to
1793 match. If there are insufficient characters before the current position, the
1794 assertion fails.
1795 .P
1796 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1797 to appear in lookbehind assertions, because it makes it impossible to calculate
1798 the length of the lookbehind. The \eX and \eR escapes, which can match
1799 different numbers of bytes, are also not permitted.
1800 .P
1801 .\" HTML <a href="#subpatternsassubroutines">
1802 .\" </a>
1803 "Subroutine"
1804 .\"
1805 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1806 as the subpattern matches a fixed-length string.
1807 .\" HTML <a href="#recursion">
1808 .\" </a>
1809 Recursion,
1810 .\"
1811 however, is not supported.
1812 .P
1813 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1814 specify efficient matching of fixed-length strings at the end of subject
1815 strings. Consider a simple pattern such as
1816 .sp
1817 abcd$
1818 .sp
1819 when applied to a long string that does not match. Because matching proceeds
1820 from left to right, PCRE will look for each "a" in the subject and then see if
1821 what follows matches the rest of the pattern. If the pattern is specified as
1822 .sp
1823 ^.*abcd$
1824 .sp
1825 the initial .* matches the entire string at first, but when this fails (because
1826 there is no following "a"), it backtracks to match all but the last character,
1827 then all but the last two characters, and so on. Once again the search for "a"
1828 covers the entire string, from right to left, so we are no better off. However,
1829 if the pattern is written as
1830 .sp
1831 ^.*+(?<=abcd)
1832 .sp
1833 there can be no backtracking for the .*+ item; it can match only the entire
1834 string. The subsequent lookbehind assertion does a single test on the last four
1835 characters. If it fails, the match fails immediately. For long strings, this
1836 approach makes a significant difference to the processing time.
1837 .
1838 .
1839 .SS "Using multiple assertions"
1840 .rs
1841 .sp
1842 Several assertions (of any sort) may occur in succession. For example,
1843 .sp
1844 (?<=\ed{3})(?<!999)foo
1845 .sp
1846 matches "foo" preceded by three digits that are not "999". Notice that each of
1847 the assertions is applied independently at the same point in the subject
1848 string. First there is a check that the previous three characters are all
1849 digits, and then there is a check that the same three characters are not "999".
1850 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1851 of which are digits and the last three of which are not "999". For example, it
1852 doesn't match "123abcfoo". A pattern to do that is
1853 .sp
1854 (?<=\ed{3}...)(?<!999)foo
1855 .sp
1856 This time the first assertion looks at the preceding six characters, checking
1857 that the first three are digits, and then the second assertion checks that the
1858 preceding three characters are not "999".
1859 .P
1860 Assertions can be nested in any combination. For example,
1861 .sp
1862 (?<=(?<!foo)bar)baz
1863 .sp
1864 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1865 preceded by "foo", while
1866 .sp
1867 (?<=\ed{3}(?!999)...)foo
1868 .sp
1869 is another pattern that matches "foo" preceded by three digits and any three
1870 characters that are not "999".
1871 .
1872 .
1873 .\" HTML <a name="conditions"></a>
1874 .SH "CONDITIONAL SUBPATTERNS"
1875 .rs
1876 .sp
1877 It is possible to cause the matching process to obey a subpattern
1878 conditionally or to choose between two alternative subpatterns, depending on
1879 the result of an assertion, or whether a specific capturing subpattern has
1880 already been matched. The two possible forms of conditional subpattern are:
1881 .sp
1882 (?(condition)yes-pattern)
1883 (?(condition)yes-pattern|no-pattern)
1884 .sp
1885 If the condition is satisfied, the yes-pattern is used; otherwise the
1886 no-pattern (if present) is used. If there are more than two alternatives in the
1887 subpattern, a compile-time error occurs.
1888 .P
1889 There are four kinds of condition: references to subpatterns, references to
1890 recursion, a pseudo-condition called DEFINE, and assertions.
1891 .
1892 .SS "Checking for a used subpattern by number"
1893 .rs
1894 .sp
1895 If the text between the parentheses consists of a sequence of digits, the
1896 condition is true if a capturing subpattern of that number has previously
1897 matched. If there is more than one capturing subpattern with the same number
1898 (see the earlier
1899 .\"
1900 .\" HTML <a href="#recursion">
1901 .\" </a>
1902 section about duplicate subpattern numbers),
1903 .\"
1904 the condition is true if any of them have been set. An alternative notation is
1905 to precede the digits with a plus or minus sign. In this case, the subpattern
1906 number is relative rather than absolute. The most recently opened parentheses
1907 can be referenced by (?(-1), the next most recent by (?(-2), and so on. In
1908 looping constructs it can also make sense to refer to subsequent groups with
1909 constructs such as (?(+2).
1910 .P
1911 Consider the following pattern, which contains non-significant white space to
1912 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1913 three parts for ease of discussion:
1914 .sp
1915 ( \e( )? [^()]+ (?(1) \e) )
1916 .sp
1917 The first part matches an optional opening parenthesis, and if that
1918 character is present, sets it as the first captured substring. The second part
1919 matches one or more characters that are not parentheses. The third part is a
1920 conditional subpattern that tests whether the first set of parentheses matched
1921 or not. If they did, that is, if subject started with an opening parenthesis,
1922 the condition is true, and so the yes-pattern is executed and a closing
1923 parenthesis is required. Otherwise, since no-pattern is not present, the
1924 subpattern matches nothing. In other words, this pattern matches a sequence of
1925 non-parentheses, optionally enclosed in parentheses.
1926 .P
1927 If you were embedding this pattern in a larger one, you could use a relative
1928 reference:
1929 .sp
1930 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1931 .sp
1932 This makes the fragment independent of the parentheses in the larger pattern.
1933 .
1934 .SS "Checking for a used subpattern by name"
1935 .rs
1936 .sp
1937 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1938 subpattern by name. For compatibility with earlier versions of PCRE, which had
1939 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1940 there is a possible ambiguity with this syntax, because subpattern names may
1941 consist entirely of digits. PCRE looks first for a named subpattern; if it
1942 cannot find one and the name consists entirely of digits, PCRE looks for a
1943 subpattern of that number, which must be greater than zero. Using subpattern
1944 names that consist entirely of digits is not recommended.
1945 .P
1946 Rewriting the above example to use a named subpattern gives this:
1947 .sp
1948 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1949 .sp
1950 If the name used in a condition of this kind is a duplicate, the test is
1951 applied to all subpatterns of the same name, and is true if any one of them has
1952 matched.
1953 .
1954 .SS "Checking for pattern recursion"
1955 .rs
1956 .sp
1957 If the condition is the string (R), and there is no subpattern with the name R,
1958 the condition is true if a recursive call to the whole pattern or any
1959 subpattern has been made. If digits or a name preceded by ampersand follow the
1960 letter R, for example:
1961 .sp
1962 (?(R3)...) or (?(R&name)...)
1963 .sp
1964 the condition is true if the most recent recursion is into a subpattern whose
1965 number or name is given. This condition does not check the entire recursion
1966 stack. If the name used in a condition of this kind is a duplicate, the test is
1967 applied to all subpatterns of the same name, and is true if any one of them is
1968 the most recent recursion.
1969 .P
1970 At "top level", all these recursion test conditions are false.
1971 .\" HTML <a href="#recursion">
1972 .\" </a>
1973 The syntax for recursive patterns
1974 .\"
1975 is described below.
1976 .
1977 .SS "Defining subpatterns for use by reference only"
1978 .rs
1979 .sp
1980 If the condition is the string (DEFINE), and there is no subpattern with the
1981 name DEFINE, the condition is always false. In this case, there may be only one
1982 alternative in the subpattern. It is always skipped if control reaches this
1983 point in the pattern; the idea of DEFINE is that it can be used to define
1984 "subroutines" that can be referenced from elsewhere. (The use of
1985 .\" HTML <a href="#subpatternsassubroutines">
1986 .\" </a>
1987 "subroutines"
1988 .\"
1989 is described below.) For example, a pattern to match an IPv4 address could be
1990 written like this (ignore whitespace and line breaks):
1991 .sp
1992 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1993 \eb (?&byte) (\e.(?&byte)){3} \eb
1994 .sp
1995 The first part of the pattern is a DEFINE group inside which a another group
1996 named "byte" is defined. This matches an individual component of an IPv4
1997 address (a number less than 256). When matching takes place, this part of the
1998 pattern is skipped because DEFINE acts like a false condition. The rest of the
1999 pattern uses references to the named group to match the four dot-separated
2000 components of an IPv4 address, insisting on a word boundary at each end.
2001 .
2002 .SS "Assertion conditions"
2003 .rs
2004 .sp
2005 If the condition is not in any of the above formats, it must be an assertion.
2006 This may be a positive or negative lookahead or lookbehind assertion. Consider
2007 this pattern, again containing non-significant white space, and with the two
2008 alternatives on the second line:
2009 .sp
2010 (?(?=[^a-z]*[a-z])
2011 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2012 .sp
2013 The condition is a positive lookahead assertion that matches an optional
2014 sequence of non-letters followed by a letter. In other words, it tests for the
2015 presence of at least one letter in the subject. If a letter is found, the
2016 subject is matched against the first alternative; otherwise it is matched
2017 against the second. This pattern matches strings in one of the two forms
2018 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2019 .
2020 .
2021 .\" HTML <a name="comments"></a>
2022 .SH COMMENTS
2023 .rs
2024 .sp
2025 The sequence (?# marks the start of a comment that continues up to the next
2026 closing parenthesis. Nested parentheses are not permitted. The characters
2027 that make up a comment play no part in the pattern matching at all.
2028 .P
2029 If the PCRE_EXTENDED option is set, an unescaped # character outside a
2030 character class introduces a comment that continues to immediately after the
2031 next newline in the pattern.
2032 .
2033 .
2034 .\" HTML <a name="recursion"></a>
2035 .SH "RECURSIVE PATTERNS"
2036 .rs
2037 .sp
2038 Consider the problem of matching a string in parentheses, allowing for
2039 unlimited nested parentheses. Without the use of recursion, the best that can
2040 be done is to use a pattern that matches up to some fixed depth of nesting. It
2041 is not possible to handle an arbitrary nesting depth.
2042 .P
2043 For some time, Perl has provided a facility that allows regular expressions to
2044 recurse (amongst other things). It does this by interpolating Perl code in the
2045 expression at run time, and the code can refer to the expression itself. A Perl
2046 pattern using code interpolation to solve the parentheses problem can be
2047 created like this:
2048 .sp
2049 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2050 .sp
2051 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2052 recursively to the pattern in which it appears.
2053 .P
2054 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2055 supports special syntax for recursion of the entire pattern, and also for
2056 individual subpattern recursion. After its introduction in PCRE and Python,
2057 this kind of recursion was subsequently introduced into Perl at release 5.10.
2058 .P
2059 A special item that consists of (? followed by a number greater than zero and a
2060 closing parenthesis is a recursive call of the subpattern of the given number,
2061 provided that it occurs inside that subpattern. (If not, it is a
2062 .\" HTML <a href="#subpatternsassubroutines">
2063 .\" </a>
2064 "subroutine"
2065 .\"
2066 call, which is described in the next section.) The special item (?R) or (?0) is
2067 a recursive call of the entire regular expression.
2068 .P
2069 This PCRE pattern solves the nested parentheses problem (assume the
2070 PCRE_EXTENDED option is set so that white space is ignored):
2071 .sp
2072 \e( ( [^()]++ | (?R) )* \e)
2073 .sp
2074 First it matches an opening parenthesis. Then it matches any number of
2075 substrings which can either be a sequence of non-parentheses, or a recursive
2076 match of the pattern itself (that is, a correctly parenthesized substring).
2077 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2078 to avoid backtracking into sequences of non-parentheses.
2079 .P
2080 If this were part of a larger pattern, you would not want to recurse the entire
2081 pattern, so instead you could use this:
2082 .sp
2083 ( \e( ( [^()]++ | (?1) )* \e) )
2084 .sp
2085 We have put the pattern into parentheses, and caused the recursion to refer to
2086 them instead of the whole pattern.
2087 .P
2088 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2089 is made easier by the use of relative references (a Perl 5.10 feature).
2090 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
2091 most recently opened parentheses preceding the recursion. In other words, a
2092 negative number counts capturing parentheses leftwards from the point at which
2093 it is encountered.
2094 .P
2095 It is also possible to refer to subsequently opened parentheses, by writing
2096 references such as (?+2). However, these cannot be recursive because the
2097 reference is not inside the parentheses that are referenced. They are always
2098 .\" HTML <a href="#subpatternsassubroutines">
2099 .\" </a>
2100 "subroutine"
2101 .\"
2102 calls, as described in the next section.
2103 .P
2104 An alternative approach is to use named parentheses instead. The Perl syntax
2105 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2106 could rewrite the above example as follows:
2107 .sp
2108 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2109 .sp
2110 If there is more than one subpattern with the same name, the earliest one is
2111 used.
2112 .P
2113 This particular example pattern that we have been looking at contains nested
2114 unlimited repeats, and so the use of a possessive quantifier for matching
2115 strings of non-parentheses is important when applying the pattern to strings
2116 that do not match. For example, when this pattern is applied to
2117 .sp
2118 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2119 .sp
2120 it yields "no match" quickly. However, if a possessive quantifier is not used,
2121 the match runs for a very long time indeed because there are so many different
2122 ways the + and * repeats can carve up the subject, and all have to be tested
2123 before failure can be reported.
2124 .P
2125 At the end of a match, the values of capturing parentheses are those from
2126 the outermost level. If you want to obtain intermediate values, a callout
2127 function can be used (see below and the
2128 .\" HREF
2129 \fBpcrecallout\fP
2130 .\"
2131 documentation). If the pattern above is matched against
2132 .sp
2133 (ab(cd)ef)
2134 .sp
2135 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2136 the last value taken on at the top level. If a capturing subpattern is not
2137 matched at the top level, its final value is unset, even if it is (temporarily)
2138 set at a deeper level.
2139 .P
2140 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2141 obtain extra memory to store data during a recursion, which it does by using
2142 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2143 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2144 .P
2145 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2146 Consider this pattern, which matches text in angle brackets, allowing for
2147 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2148 recursing), whereas any characters are permitted at the outer level.
2149 .sp
2150 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2151 .sp
2152 In this pattern, (?(R) is the start of a conditional subpattern, with two
2153 different alternatives for the recursive and non-recursive cases. The (?R) item
2154 is the actual recursive call.
2155 .
2156 .
2157 .\" HTML <a name="recursiondifference"></a>
2158 .SS "Recursion difference from Perl"
2159 .rs
2160 .sp
2161 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2162 treated as an atomic group. That is, once it has matched some of the subject
2163 string, it is never re-entered, even if it contains untried alternatives and
2164 there is a subsequent matching failure. This can be illustrated by the
2165 following pattern, which purports to match a palindromic string that contains
2166 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2167 .sp
2168 ^(.|(.)(?1)\e2)$
2169 .sp
2170 The idea is that it either matches a single character, or two identical
2171 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2172 it does not if the pattern is longer than three characters. Consider the
2173 subject string "abcba":
2174 .P
2175 At the top level, the first character is matched, but as it is not at the end
2176 of the string, the first alternative fails; the second alternative is taken
2177 and the recursion kicks in. The recursive call to subpattern 1 successfully
2178 matches the next character ("b"). (Note that the beginning and end of line
2179 tests are not part of the recursion).
2180 .P
2181 Back at the top level, the next character ("c") is compared with what
2182 subpattern 2 matched, which was "a". This fails. Because the recursion is
2183 treated as an atomic group, there are now no backtracking points, and so the
2184 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2185 try the second alternative.) However, if the pattern is written with the
2186 alternatives in the other order, things are different:
2187 .sp
2188 ^((.)(?1)\e2|.)$
2189 .sp
2190 This time, the recursing alternative is tried first, and continues to recurse
2191 until it runs out of characters, at which point the recursion fails. But this
2192 time we do have another alternative to try at the higher level. That is the big
2193 difference: in the previous case the remaining alternative is at a deeper
2194 recursion level, which PCRE cannot use.
2195 .P
2196 To change the pattern so that matches all palindromic strings, not just those
2197 with an odd number of characters, it is tempting to change the pattern to this:
2198 .sp
2199 ^((.)(?1)\e2|.?)$
2200 .sp
2201 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2202 deeper recursion has matched a single character, it cannot be entered again in
2203 order to match an empty string. The solution is to separate the two cases, and
2204 write out the odd and even cases as alternatives at the higher level:
2205 .sp
2206 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2207 .sp
2208 If you want to match typical palindromic phrases, the pattern has to ignore all
2209 non-word characters, which can be done like this:
2210 .sp
2211 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2212 .sp
2213 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2214 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2215 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2216 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2217 more) to match typical phrases, and Perl takes so long that you think it has
2218 gone into a loop.
2219 .P
2220 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2221 string does not start with a palindrome that is shorter than the entire string.
2222 For example, although "abcba" is correctly matched, if the subject is "ababa",
2223 PCRE finds the palindrome "aba" at the start, then fails at top level because
2224 the end of the string does not follow. Once again, it cannot jump back into the
2225 recursion to try other alternatives, so the entire match fails.
2226 .
2227 .
2228 .\" HTML <a name="subpatternsassubroutines"></a>
2229 .SH "SUBPATTERNS AS SUBROUTINES"
2230 .rs
2231 .sp
2232 If the syntax for a recursive subpattern reference (either by number or by
2233 name) is used outside the parentheses to which it refers, it operates like a
2234 subroutine in a programming language. The "called" subpattern may be defined
2235 before or after the reference. A numbered reference can be absolute or
2236 relative, as in these examples:
2237 .sp
2238 (...(absolute)...)...(?2)...
2239 (...(relative)...)...(?-1)...
2240 (...(?+1)...(relative)...
2241 .sp
2242 An earlier example pointed out that the pattern
2243 .sp
2244 (sens|respons)e and \e1ibility
2245 .sp
2246 matches "sense and sensibility" and "response and responsibility", but not
2247 "sense and responsibility". If instead the pattern
2248 .sp
2249 (sens|respons)e and (?1)ibility
2250 .sp
2251 is used, it does match "sense and responsibility" as well as the other two
2252 strings. Another example is given in the discussion of DEFINE above.
2253 .P
2254 Like recursive subpatterns, a subroutine call is always treated as an atomic
2255 group. That is, once it has matched some of the subject string, it is never
2256 re-entered, even if it contains untried alternatives and there is a subsequent
2257 matching failure. Any capturing parentheses that are set during the subroutine
2258 call revert to their previous values afterwards.
2259 .P
2260 When a subpattern is used as a subroutine, processing options such as
2261 case-independence are fixed when the subpattern is defined. They cannot be
2262 changed for different calls. For example, consider this pattern:
2263 .sp
2264 (abc)(?i:(?-1))
2265 .sp
2266 It matches "abcabc". It does not match "abcABC" because the change of
2267 processing option does not affect the called subpattern.
2268 .
2269 .
2270 .\" HTML <a name="onigurumasubroutines"></a>
2271 .SH "ONIGURUMA SUBROUTINE SYNTAX"
2272 .rs
2273 .sp
2274 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2275 a number enclosed either in angle brackets or single quotes, is an alternative
2276 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2277 are two of the examples used above, rewritten using this syntax:
2278 .sp
2279 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2280 (sens|respons)e and \eg'1'ibility
2281 .sp
2282 PCRE supports an extension to Oniguruma: if a number is preceded by a
2283 plus or a minus sign it is taken as a relative reference. For example:
2284 .sp
2285 (abc)(?i:\eg<-1>)
2286 .sp
2287 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2288 synonymous. The former is a back reference; the latter is a subroutine call.
2289 .
2290 .
2291 .SH CALLOUTS
2292 .rs
2293 .sp
2294 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2295 code to be obeyed in the middle of matching a regular expression. This makes it
2296 possible, amongst other things, to extract different substrings that match the
2297 same pair of parentheses when there is a repetition.
2298 .P
2299 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2300 code. The feature is called "callout". The caller of PCRE provides an external
2301 function by putting its entry point in the global variable \fIpcre_callout\fP.
2302 By default, this variable contains NULL, which disables all calling out.
2303 .P
2304 Within a regular expression, (?C) indicates the points at which the external
2305 function is to be called. If you want to identify different callout points, you
2306 can put a number less than 256 after the letter C. The default value is zero.
2307 For example, this pattern has two callout points:
2308 .sp
2309 (?C1)abc(?C2)def
2310 .sp
2311 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2312 automatically installed before each item in the pattern. They are all numbered
2313 255.
2314 .P
2315 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2316 set), the external function is called. It is provided with the number of the
2317 callout, the position in the pattern, and, optionally, one item of data
2318 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2319 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2320 description of the interface to the callout function is given in the
2321 .\" HREF
2322 \fBpcrecallout\fP
2323 .\"
2324 documentation.
2325 .
2326 .
2327 .\" HTML <a name="backtrackcontrol"></a>
2328 .SH "BACKTRACKING CONTROL"
2329 .rs
2330 .sp
2331 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2332 are described in the Perl documentation as "experimental and subject to change
2333 or removal in a future version of Perl". It goes on to say: "Their usage in
2334 production code should be noted to avoid problems during upgrades." The same
2335 remarks apply to the PCRE features described in this section.
2336 .P
2337 Since these verbs are specifically related to backtracking, most of them can be
2338 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2339 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2340 failing negative assertion, they cause an error if encountered by
2341 \fBpcre_dfa_exec()\fP.
2342 .P
2343 If any of these verbs are used in an assertion or subroutine subpattern
2344 (including recursive subpatterns), their effect is confined to that subpattern;
2345 it does not extend to the surrounding pattern. Note that such subpatterns are
2346 processed as anchored at the point where they are tested.
2347 .P
2348 The new verbs make use of what was previously invalid syntax: an opening
2349 parenthesis followed by an asterisk. They are generally of the form
2350 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2351 depending on whether or not an argument is present. An name is a sequence of
2352 letters, digits, and underscores. If the name is empty, that is, if the closing
2353 parenthesis immediately follows the colon, the effect is as if the colon were
2354 not there. Any number of these verbs may occur in a pattern.
2355 .P
2356 PCRE contains some optimizations that are used to speed up matching by running
2357 some checks at the start of each match attempt. For example, it may know the
2358 minimum length of matching subject, or that a particular character must be
2359 present. When one of these optimizations suppresses the running of a match, any
2360 included backtracking verbs will not, of course, be processed. You can suppress
2361 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2362 when calling \fBpcre_exec()\fP.
2363 .
2364 .
2365 .SS "Verbs that act immediately"
2366 .rs
2367 .sp
2368 The following verbs act as soon as they are encountered. They may not be
2369 followed by a name.
2370 .sp
2371 (*ACCEPT)
2372 .sp
2373 This verb causes the match to end successfully, skipping the remainder of the
2374 pattern. When inside a recursion, only the innermost pattern is ended
2375 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2376 captured. (This feature was added to PCRE at release 8.00.) For example:
2377 .sp
2378 A((?:A|B(*ACCEPT)|C)D)
2379 .sp
2380 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2381 the outer parentheses.
2382 .sp
2383 (*FAIL) or (*F)
2384 .sp
2385 This verb causes the match to fail, forcing backtracking to occur. It is
2386 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2387 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2388 Perl features that are not present in PCRE. The nearest equivalent is the
2389 callout feature, as for example in this pattern:
2390 .sp
2391 a+(?C)(*FAIL)
2392 .sp
2393 A match with the string "aaaa" always fails, but the callout is taken before
2394 each backtrack happens (in this example, 10 times).
2395 .
2396 .
2397 .SS "Recording which path was taken"
2398 .rs
2399 .sp
2400 There is one verb whose main purpose is to track how a match was arrived at,
2401 though it also has a secondary use in conjunction with advancing the match
2402 starting point (see (*SKIP) below).
2403 .sp
2404 (*MARK:NAME) or (*:NAME)
2405 .sp
2406 A name is always required with this verb. There may be as many instances of
2407 (*MARK) as you like in a pattern, and their names do not have to be unique.
2408 .P
2409 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2410 to the caller via the \fIpcre_extra\fP data structure, as described in the
2411 .\" HTML <a href="pcreapi.html#extradata">
2412 .\" </a>
2413 section on \fIpcre_extra\fP
2414 .\"
2415 in the
2416 .\" HREF
2417 \fBpcreapi\fP
2418 .\"
2419 documentation. No data is returned for a partial match. Here is an example of
2420 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2421 outputting of (*MARK) data:
2422 .sp
2423 /X(*MARK:A)Y|X(*MARK:B)Z/K
2424 XY
2425 0: XY
2426 MK: A
2427 XZ
2428 0: XZ
2429 MK: B
2430 .sp
2431 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2432 indicates which of the two alternatives matched. This is a more efficient way
2433 of obtaining this information than putting each alternative in its own
2434 capturing parentheses.
2435 .P
2436 A name may also be returned after a failed match if the final path through the
2437 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2438 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2439 starting point for matching is advanced, the final check is often with an empty
2440 string, causing a failure before (*MARK) is reached. For example:
2441 .sp
2442 /X(*MARK:A)Y|X(*MARK:B)Z/K
2443 XP
2444 No match
2445 .sp
2446 There are three potential starting points for this match (starting with X,
2447 starting with P, and with an empty string). If the pattern is anchored, the
2448 result is different:
2449 .sp
2450 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2451 XP
2452 No match, mark = B
2453 .sp
2454 PCRE's start-of-match optimizations can also interfere with this. For example,
2455 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2456 subject length for a match, a shorter subject will not be scanned at all.
2457 .P
2458 Note that similar anomalies (though different in detail) exist in Perl, no
2459 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2460 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2461 .
2462 .
2463 .SS "Verbs that act after backtracking"
2464 .rs
2465 .sp
2466 The following verbs do nothing when they are encountered. Matching continues
2467 with what follows, but if there is no subsequent match, causing a backtrack to
2468 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2469 the verb. However, when one of these verbs appears inside an atomic group, its
2470 effect is confined to that group, because once the group has been matched,
2471 there is never any backtracking into it. In this situation, backtracking can
2472 "jump back" to the left of the entire atomic group. (Remember also, as stated
2473 above, that this localization also applies in subroutine calls and assertions.)
2474 .P
2475 These verbs differ in exactly what kind of failure occurs when backtracking
2476 reaches them.
2477 .sp
2478 (*COMMIT)
2479 .sp
2480 This verb, which may not be followed by a name, causes the whole match to fail
2481 outright if the rest of the pattern does not match. Even if the pattern is
2482 unanchored, no further attempts to find a match by advancing the starting point
2483 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2484 finding a match at the current starting point, or not at all. For example:
2485 .sp
2486 a+(*COMMIT)b
2487 .sp
2488 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2489 dynamic anchor, or "I've started, so I must finish." The name of the most
2490 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2491 match failure.
2492 .P
2493 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2494 unless PCRE's start-of-match optimizations are turned off, as shown in this
2495 \fBpcretest\fP example:
2496 .sp
2497 /(*COMMIT)abc/
2498 xyzabc
2499 0: abc
2500 xyzabc\eY
2501 No match
2502 .sp
2503 PCRE knows that any match must start with "a", so the optimization skips along
2504 the subject to "a" before running the first match attempt, which succeeds. When
2505 the optimization is disabled by the \eY escape in the second subject, the match
2506 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2507 starting points.
2508 .sp
2509 (*PRUNE) or (*PRUNE:NAME)
2510 .sp
2511 This verb causes the match to fail at the current starting position in the
2512 subject if the rest of the pattern does not match. If the pattern is
2513 unanchored, the normal "bumpalong" advance to the next starting character then
2514 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2515 reached, or when matching to the right of (*PRUNE), but if there is no match to
2516 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2517 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2518 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2519 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2520 match fails completely; the name is passed back if this is the final attempt.
2521 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2522 pattern (*PRUNE) has the same effect as (*COMMIT).
2523 .sp
2524 (*SKIP)
2525 .sp
2526 This verb, when given without a name, is like (*PRUNE), except that if the
2527 pattern is unanchored, the "bumpalong" advance is not to the next character,
2528 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2529 signifies that whatever text was matched leading up to it cannot be part of a
2530 successful match. Consider:
2531 .sp
2532 a+(*SKIP)b
2533 .sp
2534 If the subject is "aaaac...", after the first match attempt fails (starting at
2535 the first character in the string), the starting point skips on to start the
2536 next attempt at "c". Note that a possessive quantifer does not have the same
2537 effect as this example; although it would suppress backtracking during the
2538 first match attempt, the second attempt would start at the second character
2539 instead of skipping on to "c".
2540 .sp
2541 (*SKIP:NAME)
2542 .sp
2543 When (*SKIP) has an associated name, its behaviour is modified. If the
2544 following pattern fails to match, the previous path through the pattern is
2545 searched for the most recent (*MARK) that has the same name. If one is found,
2546 the "bumpalong" advance is to the subject position that corresponds to that
2547 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2548 matching name is found, normal "bumpalong" of one character happens (the
2549 (*SKIP) is ignored).
2550 .sp
2551 (*THEN) or (*THEN:NAME)
2552 .sp
2553 This verb causes a skip to the next alternation if the rest of the pattern does
2554 not match. That is, it cancels pending backtracking, but only within the
2555 current alternation. Its name comes from the observation that it can be used
2556 for a pattern-based if-then-else block:
2557 .sp
2558 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2559 .sp
2560 If the COND1 pattern matches, FOO is tried (and possibly further items after
2561 the end of the group if FOO succeeds); on failure the matcher skips to the
2562 second alternative and tries COND2, without backtracking into COND1. The
2563 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2564 overall match fails. If (*THEN) is not directly inside an alternation, it acts
2565 like (*PRUNE).
2566 .
2567 .
2568 .SH "SEE ALSO"
2569 .rs
2570 .sp
2571 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2572 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2573 .
2574 .
2575 .SH AUTHOR
2576 .rs
2577 .sp
2578 .nf
2579 Philip Hazel
2580 University Computing Service
2581 Cambridge CB2 3QH, England.
2582 .fi
2583 .
2584 .
2585 .SH REVISION
2586 .rs
2587 .sp
2588 .nf
2589 Last updated: 03 May 2010
2590 Copyright (c) 1997-2010 University of Cambridge.
2591 .fi

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5