ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 210 - (show annotations)
Wed Aug 8 14:24:50 2007 UTC (13 years, 6 months ago) by ph10
File size: 85817 byte(s)
Add Perl 5.10's backtracking verbs.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. Perl's regular expressions are described in its own documentation, and
13 regular expressions in general are covered in a number of books, some of which
14 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
15 published by O'Reilly, covers regular expressions in great detail. This
16 description of PCRE's regular expressions is intended as reference material.
17 .P
18 The original operation of PCRE was on strings of one-byte characters. However,
19 there is now also support for UTF-8 character strings. To use this, you must
20 build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
21 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
22 places below. There is also a summary of UTF-8 features in the
23 .\" HTML <a href="pcre.html#utf8support">
24 .\" </a>
25 section on UTF-8 support
26 .\"
27 in the main
28 .\" HREF
29 \fBpcre\fP
30 .\"
31 page.
32 .P
33 The remainder of this document discusses the patterns that are supported by
34 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
35 From release 6.0, PCRE offers a second matching function,
36 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
37 Perl-compatible. Some of the features discussed below are not available when
38 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
39 alternative function, and how it differs from the normal function, are
40 discussed in the
41 .\" HREF
42 \fBpcrematching\fP
43 .\"
44 page.
45 .
46 .
48 .rs
49 .sp
50 A regular expression is a pattern that is matched against a subject string from
51 left to right. Most characters stand for themselves in a pattern, and match the
52 corresponding characters in the subject. As a trivial example, the pattern
53 .sp
54 The quick brown fox
55 .sp
56 matches a portion of a subject string that is identical to itself. When
57 caseless matching is specified (the PCRE_CASELESS option), letters are matched
58 independently of case. In UTF-8 mode, PCRE always understands the concept of
59 case for characters whose values are less than 128, so caseless matching is
60 always possible. For characters with higher values, the concept of case is
61 supported if PCRE is compiled with Unicode property support, but not otherwise.
62 If you want to use caseless matching for characters 128 and above, you must
63 ensure that PCRE is compiled with Unicode property support as well as with
64 UTF-8 support.
65 .P
66 The power of regular expressions comes from the ability to include alternatives
67 and repetitions in the pattern. These are encoded in the pattern by the use of
68 \fImetacharacters\fP, which do not stand for themselves but instead are
69 interpreted in some special way.
70 .P
71 There are two different sets of metacharacters: those that are recognized
72 anywhere in the pattern except within square brackets, and those that are
73 recognized within square brackets. Outside square brackets, the metacharacters
74 are as follows:
75 .sp
76 \e general escape character with several uses
77 ^ assert start of string (or line, in multiline mode)
78 $ assert end of string (or line, in multiline mode)
79 . match any character except newline (by default)
80 [ start character class definition
81 | start of alternative branch
82 ( start subpattern
83 ) end subpattern
84 ? extends the meaning of (
85 also 0 or 1 quantifier
86 also quantifier minimizer
87 * 0 or more quantifier
88 + 1 or more quantifier
89 also "possessive quantifier"
90 { start min/max quantifier
91 .sp
92 Part of a pattern that is in square brackets is called a "character class". In
93 a character class the only metacharacters are:
94 .sp
95 \e general escape character
96 ^ negate the class, but only if the first character
97 - indicates character range
98 .\" JOIN
99 [ POSIX character class (only if followed by POSIX
100 syntax)
101 ] terminates the character class
102 .sp
103 The following sections describe the use of each of the metacharacters.
104 .
105 .
107 .rs
108 .sp
109 The backslash character has several uses. Firstly, if it is followed by a
110 non-alphanumeric character, it takes away any special meaning that character
111 may have. This use of backslash as an escape character applies both inside and
112 outside character classes.
113 .P
114 For example, if you want to match a * character, you write \e* in the pattern.
115 This escaping action applies whether or not the following character would
116 otherwise be interpreted as a metacharacter, so it is always safe to precede a
117 non-alphanumeric with backslash to specify that it stands for itself. In
118 particular, if you want to match a backslash, you write \e\e.
119 .P
120 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
121 pattern (other than in a character class) and characters between a # outside
122 a character class and the next newline are ignored. An escaping backslash can
123 be used to include a whitespace or # character as part of the pattern.
124 .P
125 If you want to remove the special meaning from a sequence of characters, you
126 can do so by putting them between \eQ and \eE. This is different from Perl in
127 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
128 Perl, $ and @ cause variable interpolation. Note the following examples:
129 .sp
130 Pattern PCRE matches Perl matches
131 .sp
132 .\" JOIN
133 \eQabc$xyz\eE abc$xyz abc followed by the
134 contents of $xyz
135 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
136 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
137 .sp
138 The \eQ...\eE sequence is recognized both inside and outside character classes.
139 .
140 .
141 .\" HTML <a name="digitsafterbackslash"></a>
142 .SS "Non-printing characters"
143 .rs
144 .sp
145 A second use of backslash provides a way of encoding non-printing characters
146 in patterns in a visible manner. There is no restriction on the appearance of
147 non-printing characters, apart from the binary zero that terminates a pattern,
148 but when a pattern is being prepared by text editing, it is usually easier to
149 use one of the following escape sequences than the binary character it
150 represents:
151 .sp
152 \ea alarm, that is, the BEL character (hex 07)
153 \ecx "control-x", where x is any character
154 \ee escape (hex 1B)
155 \ef formfeed (hex 0C)
156 \en newline (hex 0A)
157 \er carriage return (hex 0D)
158 \et tab (hex 09)
159 \eddd character with octal code ddd, or backreference
160 \exhh character with hex code hh
161 \ex{hhh..} character with hex code hhh..
162 .sp
163 The precise effect of \ecx is as follows: if x is a lower case letter, it
164 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
165 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
166 7B.
167 .P
168 After \ex, from zero to two hexadecimal digits are read (letters can be in
169 upper or lower case). Any number of hexadecimal digits may appear between \ex{
170 and }, but the value of the character code must be less than 256 in non-UTF-8
171 mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
172 is 7FFFFFFF). If characters other than hexadecimal digits appear between \ex{
173 and }, or if there is no terminating }, this form of escape is not recognized.
174 Instead, the initial \ex will be interpreted as a basic hexadecimal escape,
175 with no following digits, giving a character whose value is zero.
176 .P
177 Characters whose value is less than 256 can be defined by either of the two
178 syntaxes for \ex. There is no difference in the way they are handled. For
179 example, \exdc is exactly the same as \ex{dc}.
180 .P
181 After \e0 up to two further octal digits are read. If there are fewer than two
182 digits, just those that are present are used. Thus the sequence \e0\ex\e07
183 specifies two binary zeros followed by a BEL character (code value 7). Make
184 sure you supply two digits after the initial zero if the pattern character that
185 follows is itself an octal digit.
186 .P
187 The handling of a backslash followed by a digit other than 0 is complicated.
188 Outside a character class, PCRE reads it and any following digits as a decimal
189 number. If the number is less than 10, or if there have been at least that many
190 previous capturing left parentheses in the expression, the entire sequence is
191 taken as a \fIback reference\fP. A description of how this works is given
192 .\" HTML <a href="#backreferences">
193 .\" </a>
194 later,
195 .\"
196 following the discussion of
197 .\" HTML <a href="#subpattern">
198 .\" </a>
199 parenthesized subpatterns.
200 .\"
201 .P
202 Inside a character class, or if the decimal number is greater than 9 and there
203 have not been that many capturing subpatterns, PCRE re-reads up to three octal
204 digits following the backslash, and uses them to generate a data character. Any
205 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
206 character specified in octal must be less than \e400. In UTF-8 mode, values up
207 to \e777 are permitted. For example:
208 .sp
209 \e040 is another way of writing a space
210 .\" JOIN
211 \e40 is the same, provided there are fewer than 40
212 previous capturing subpatterns
213 \e7 is always a back reference
214 .\" JOIN
215 \e11 might be a back reference, or another way of
216 writing a tab
217 \e011 is always a tab
218 \e0113 is a tab followed by the character "3"
219 .\" JOIN
220 \e113 might be a back reference, otherwise the
221 character with octal code 113
222 .\" JOIN
223 \e377 might be a back reference, otherwise
224 the byte consisting entirely of 1 bits
225 .\" JOIN
226 \e81 is either a back reference, or a binary zero
227 followed by the two characters "8" and "1"
228 .sp
229 Note that octal values of 100 or greater must not be introduced by a leading
230 zero, because no more than three octal digits are ever read.
231 .P
232 All the sequences that define a single character value can be used both inside
233 and outside character classes. In addition, inside a character class, the
234 sequence \eb is interpreted as the backspace character (hex 08), and the
235 sequences \eR and \eX are interpreted as the characters "R" and "X",
236 respectively. Outside a character class, these sequences have different
237 meanings
238 .\" HTML <a href="#uniextseq">
239 .\" </a>
240 (see below).
241 .\"
242 .
243 .
244 .SS "Absolute and relative back references"
245 .rs
246 .sp
247 The sequence \eg followed by an unsigned or a negative number, optionally
248 enclosed in braces, is an absolute or relative back reference. A named back
249 reference can be coded as \eg{name}. Back references are discussed
250 .\" HTML <a href="#backreferences">
251 .\" </a>
252 later,
253 .\"
254 following the discussion of
255 .\" HTML <a href="#subpattern">
256 .\" </a>
257 parenthesized subpatterns.
258 .\"
259 .
260 .
261 .SS "Generic character types"
262 .rs
263 .sp
264 Another use of backslash is for specifying generic character types. The
265 following are always recognized:
266 .sp
267 \ed any decimal digit
268 \eD any character that is not a decimal digit
269 \eh any horizontal whitespace character
270 \eH any character that is not a horizontal whitespace character
271 \es any whitespace character
272 \eS any character that is not a whitespace character
273 \ev any vertical whitespace character
274 \eV any character that is not a vertical whitespace character
275 \ew any "word" character
276 \eW any "non-word" character
277 .sp
278 Each pair of escape sequences partitions the complete set of characters into
279 two disjoint sets. Any given character matches one, and only one, of each pair.
280 .P
281 These character type sequences can appear both inside and outside character
282 classes. They each match one character of the appropriate type. If the current
283 matching point is at the end of the subject string, all of them fail, since
284 there is no character to match.
285 .P
286 For compatibility with Perl, \es does not match the VT character (code 11).
287 This makes it different from the the POSIX "space" class. The \es characters
288 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
289 included in a Perl script, \es may match the VT character. In PCRE, it never
290 does.
291 .P
292 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
293 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
294 character property support is available. These sequences retain their original
295 meanings from before UTF-8 support was available, mainly for efficiency
296 reasons.
297 .P
298 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
299 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
300 The horizontal space characters are:
301 .sp
302 U+0009 Horizontal tab
303 U+0020 Space
304 U+00A0 Non-break space
305 U+1680 Ogham space mark
306 U+180E Mongolian vowel separator
307 U+2000 En quad
308 U+2001 Em quad
309 U+2002 En space
310 U+2003 Em space
311 U+2004 Three-per-em space
312 U+2005 Four-per-em space
313 U+2006 Six-per-em space
314 U+2007 Figure space
315 U+2008 Punctuation space
316 U+2009 Thin space
317 U+200A Hair space
318 U+202F Narrow no-break space
319 U+205F Medium mathematical space
320 U+3000 Ideographic space
321 .sp
322 The vertical space characters are:
323 .sp
324 U+000A Linefeed
325 U+000B Vertical tab
326 U+000C Formfeed
327 U+000D Carriage return
328 U+0085 Next line
329 U+2028 Line separator
330 U+2029 Paragraph separator
331 .P
332 A "word" character is an underscore or any character less than 256 that is a
333 letter or digit. The definition of letters and digits is controlled by PCRE's
334 low-valued character tables, and may vary if locale-specific matching is taking
335 place (see
336 .\" HTML <a href="pcreapi.html#localesupport">
337 .\" </a>
338 "Locale support"
339 .\"
340 in the
341 .\" HREF
342 \fBpcreapi\fP
343 .\"
344 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
345 or "french" in Windows, some character codes greater than 128 are used for
346 accented letters, and these are matched by \ew. The use of locales with Unicode
347 is discouraged.
348 .
349 .
350 .SS "Newline sequences"
351 .rs
352 .sp
353 Outside a character class, the escape sequence \eR matches any Unicode newline
354 sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is equivalent to
355 the following:
356 .sp
357 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
358 .sp
359 This is an example of an "atomic group", details of which are given
360 .\" HTML <a href="#atomicgroup">
361 .\" </a>
362 below.
363 .\"
364 This particular group matches either the two-character sequence CR followed by
365 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
366 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
367 line, U+0085). The two-character sequence is treated as a single unit that
368 cannot be split.
369 .P
370 In UTF-8 mode, two additional characters whose codepoints are greater than 255
371 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
372 Unicode character property support is not needed for these characters to be
373 recognized.
374 .P
375 Inside a character class, \eR matches the letter "R".
376 .
377 .
378 .\" HTML <a name="uniextseq"></a>
379 .SS Unicode character properties
380 .rs
381 .sp
382 When PCRE is built with Unicode character property support, three additional
383 escape sequences that match characters with specific properties are available.
384 When not in UTF-8 mode, these sequences are of course limited to testing
385 characters whose codepoints are less than 256, but they do work in this mode.
386 The extra escape sequences are:
387 .sp
388 \ep{\fIxx\fP} a character with the \fIxx\fP property
389 \eP{\fIxx\fP} a character without the \fIxx\fP property
390 \eX an extended Unicode sequence
391 .sp
392 The property names represented by \fIxx\fP above are limited to the Unicode
393 script names, the general category properties, and "Any", which matches any
394 character (including newline). Other properties such as "InMusicalSymbols" are
395 not currently supported by PCRE. Note that \eP{Any} does not match any
396 characters, so always causes a match failure.
397 .P
398 Sets of Unicode characters are defined as belonging to certain scripts. A
399 character from one of these sets can be matched using a script name. For
400 example:
401 .sp
402 \ep{Greek}
403 \eP{Han}
404 .sp
405 Those that are not part of an identified script are lumped together as
406 "Common". The current list of scripts is:
407 .P
408 Arabic,
409 Armenian,
410 Balinese,
411 Bengali,
412 Bopomofo,
413 Braille,
414 Buginese,
415 Buhid,
416 Canadian_Aboriginal,
417 Cherokee,
418 Common,
419 Coptic,
420 Cuneiform,
421 Cypriot,
422 Cyrillic,
423 Deseret,
424 Devanagari,
425 Ethiopic,
426 Georgian,
427 Glagolitic,
428 Gothic,
429 Greek,
430 Gujarati,
431 Gurmukhi,
432 Han,
433 Hangul,
434 Hanunoo,
435 Hebrew,
436 Hiragana,
437 Inherited,
438 Kannada,
439 Katakana,
440 Kharoshthi,
441 Khmer,
442 Lao,
443 Latin,
444 Limbu,
445 Linear_B,
446 Malayalam,
447 Mongolian,
448 Myanmar,
449 New_Tai_Lue,
450 Nko,
451 Ogham,
452 Old_Italic,
453 Old_Persian,
454 Oriya,
455 Osmanya,
456 Phags_Pa,
457 Phoenician,
458 Runic,
459 Shavian,
460 Sinhala,
461 Syloti_Nagri,
462 Syriac,
463 Tagalog,
464 Tagbanwa,
465 Tai_Le,
466 Tamil,
467 Telugu,
468 Thaana,
469 Thai,
470 Tibetan,
471 Tifinagh,
472 Ugaritic,
473 Yi.
474 .P
475 Each character has exactly one general category property, specified by a
476 two-letter abbreviation. For compatibility with Perl, negation can be specified
477 by including a circumflex between the opening brace and the property name. For
478 example, \ep{^Lu} is the same as \eP{Lu}.
479 .P
480 If only one letter is specified with \ep or \eP, it includes all the general
481 category properties that start with that letter. In this case, in the absence
482 of negation, the curly brackets in the escape sequence are optional; these two
483 examples have the same effect:
484 .sp
485 \ep{L}
486 \epL
487 .sp
488 The following general category property codes are supported:
489 .sp
490 C Other
491 Cc Control
492 Cf Format
493 Cn Unassigned
494 Co Private use
495 Cs Surrogate
496 .sp
497 L Letter
498 Ll Lower case letter
499 Lm Modifier letter
500 Lo Other letter
501 Lt Title case letter
502 Lu Upper case letter
503 .sp
504 M Mark
505 Mc Spacing mark
506 Me Enclosing mark
507 Mn Non-spacing mark
508 .sp
509 N Number
510 Nd Decimal number
511 Nl Letter number
512 No Other number
513 .sp
514 P Punctuation
515 Pc Connector punctuation
516 Pd Dash punctuation
517 Pe Close punctuation
518 Pf Final punctuation
519 Pi Initial punctuation
520 Po Other punctuation
521 Ps Open punctuation
522 .sp
523 S Symbol
524 Sc Currency symbol
525 Sk Modifier symbol
526 Sm Mathematical symbol
527 So Other symbol
528 .sp
529 Z Separator
530 Zl Line separator
531 Zp Paragraph separator
532 Zs Space separator
533 .sp
534 The special property L& is also supported: it matches a character that has
535 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
536 a modifier or "other".
537 .P
538 The long synonyms for these properties that Perl supports (such as \ep{Letter})
539 are not supported by PCRE, nor is it permitted to prefix any of these
540 properties with "Is".
541 .P
542 No character that is in the Unicode table has the Cn (unassigned) property.
543 Instead, this property is assumed for any code point that is not in the
544 Unicode table.
545 .P
546 Specifying caseless matching does not affect these escape sequences. For
547 example, \ep{Lu} always matches only upper case letters.
548 .P
549 The \eX escape matches any number of Unicode characters that form an extended
550 Unicode sequence. \eX is equivalent to
551 .sp
552 (?>\ePM\epM*)
553 .sp
554 That is, it matches a character without the "mark" property, followed by zero
555 or more characters with the "mark" property, and treats the sequence as an
556 atomic group
557 .\" HTML <a href="#atomicgroup">
558 .\" </a>
559 (see below).
560 .\"
561 Characters with the "mark" property are typically accents that affect the
562 preceding character. None of them have codepoints less than 256, so in
563 non-UTF-8 mode \eX matches any one character.
564 .P
565 Matching characters by Unicode property is not fast, because PCRE has to search
566 a structure that contains data for over fifteen thousand characters. That is
567 why the traditional escape sequences such as \ed and \ew do not use Unicode
568 properties in PCRE.
569 .
570 .
571 .\" HTML <a name="resetmatchstart"></a>
572 .SS "Resetting the match start"
573 .rs
574 .sp
575 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
576 matched characters not to be included in the final matched sequence. For
577 example, the pattern:
578 .sp
579 foo\eKbar
580 .sp
581 matches "foobar", but reports that it has matched "bar". This feature is
582 similar to a lookbehind assertion
583 .\" HTML <a href="#lookbehind">
584 .\" </a>
585 (described below).
586 .\"
587 However, in this case, the part of the subject before the real match does not
588 have to be of fixed length, as lookbehind assertions do. The use of \eK does
589 not interfere with the setting of
590 .\" HTML <a href="#subpattern">
591 .\" </a>
592 captured substrings.
593 .\"
594 For example, when the pattern
595 .sp
596 (foo)\eKbar
597 .sp
598 matches "foobar", the first substring is still set to "foo".
599 .
600 .
601 .\" HTML <a name="smallassertions"></a>
602 .SS "Simple assertions"
603 .rs
604 .sp
605 The final use of backslash is for certain simple assertions. An assertion
606 specifies a condition that has to be met at a particular point in a match,
607 without consuming any characters from the subject string. The use of
608 subpatterns for more complicated assertions is described
609 .\" HTML <a href="#bigassertions">
610 .\" </a>
611 below.
612 .\"
613 The backslashed assertions are:
614 .sp
615 \eb matches at a word boundary
616 \eB matches when not at a word boundary
617 \eA matches at the start of the subject
618 \eZ matches at the end of the subject
619 also matches before a newline at the end of the subject
620 \ez matches only at the end of the subject
621 \eG matches at the first matching position in the subject
622 .sp
623 These assertions may not appear in character classes (but note that \eb has a
624 different meaning, namely the backspace character, inside a character class).
625 .P
626 A word boundary is a position in the subject string where the current character
627 and the previous character do not both match \ew or \eW (i.e. one matches
628 \ew and the other matches \eW), or the start or end of the string if the
629 first or last character matches \ew, respectively.
630 .P
631 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
632 dollar (described in the next section) in that they only ever match at the very
633 start and end of the subject string, whatever options are set. Thus, they are
634 independent of multiline mode. These three assertions are not affected by the
635 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
636 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
637 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
638 at a point other than the beginning of the subject, \eA can never match. The
639 difference between \eZ and \ez is that \eZ matches before a newline at the end
640 of the string as well as at the very end, whereas \ez matches only at the end.
641 .P
642 The \eG assertion is true only when the current matching position is at the
643 start point of the match, as specified by the \fIstartoffset\fP argument of
644 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
645 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
646 arguments, you can mimic Perl's /g option, and it is in this kind of
647 implementation where \eG can be useful.
648 .P
649 Note, however, that PCRE's interpretation of \eG, as the start of the current
650 match, is subtly different from Perl's, which defines it as the end of the
651 previous match. In Perl, these can be different when the previously matched
652 string was empty. Because PCRE does just one match at a time, it cannot
653 reproduce this behaviour.
654 .P
655 If all the alternatives of a pattern begin with \eG, the expression is anchored
656 to the starting match position, and the "anchored" flag is set in the compiled
657 regular expression.
658 .
659 .
661 .rs
662 .sp
663 Outside a character class, in the default matching mode, the circumflex
664 character is an assertion that is true only if the current matching point is
665 at the start of the subject string. If the \fIstartoffset\fP argument of
666 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
667 option is unset. Inside a character class, circumflex has an entirely different
668 meaning
669 .\" HTML <a href="#characterclass">
670 .\" </a>
671 (see below).
672 .\"
673 .P
674 Circumflex need not be the first character of the pattern if a number of
675 alternatives are involved, but it should be the first thing in each alternative
676 in which it appears if the pattern is ever to match that branch. If all
677 possible alternatives start with a circumflex, that is, if the pattern is
678 constrained to match only at the start of the subject, it is said to be an
679 "anchored" pattern. (There are also other constructs that can cause a pattern
680 to be anchored.)
681 .P
682 A dollar character is an assertion that is true only if the current matching
683 point is at the end of the subject string, or immediately before a newline
684 at the end of the string (by default). Dollar need not be the last character of
685 the pattern if a number of alternatives are involved, but it should be the last
686 item in any branch in which it appears. Dollar has no special meaning in a
687 character class.
688 .P
689 The meaning of dollar can be changed so that it matches only at the very end of
690 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
691 does not affect the \eZ assertion.
692 .P
693 The meanings of the circumflex and dollar characters are changed if the
694 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
695 immediately after internal newlines as well as at the start of the subject
696 string. It does not match after a newline that ends the string. A dollar
697 matches before any newlines in the string, as well as at the very end, when
698 PCRE_MULTILINE is set. When newline is specified as the two-character
699 sequence CRLF, isolated CR and LF characters do not indicate newlines.
700 .P
701 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
702 \en represents a newline) in multiline mode, but not otherwise. Consequently,
703 patterns that are anchored in single line mode because all branches start with
704 ^ are not anchored in multiline mode, and a match for circumflex is possible
705 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
706 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
707 .P
708 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
709 end of the subject in both modes, and if all branches of a pattern start with
710 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
711 .
712 .
714 .rs
715 .sp
716 Outside a character class, a dot in the pattern matches any one character in
717 the subject string except (by default) a character that signifies the end of a
718 line. In UTF-8 mode, the matched character may be more than one byte long.
719 .P
720 When a line ending is defined as a single character, dot never matches that
721 character; when the two-character sequence CRLF is used, dot does not match CR
722 if it is immediately followed by LF, but otherwise it matches all characters
723 (including isolated CRs and LFs). When any Unicode line endings are being
724 recognized, dot does not match CR or LF or any of the other line ending
725 characters.
726 .P
727 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
728 option is set, a dot matches any one character, without exception. If the
729 two-character sequence CRLF is present in the subject string, it takes two dots
730 to match it.
731 .P
732 The handling of dot is entirely independent of the handling of circumflex and
733 dollar, the only relationship being that they both involve newlines. Dot has no
734 special meaning in a character class.
735 .
736 .
738 .rs
739 .sp
740 Outside a character class, the escape sequence \eC matches any one byte, both
741 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
742 characters. The feature is provided in Perl in order to match individual bytes
743 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
744 what remains in the string may be a malformed UTF-8 string. For this reason,
745 the \eC escape sequence is best avoided.
746 .P
747 PCRE does not allow \eC to appear in lookbehind assertions
748 .\" HTML <a href="#lookbehind">
749 .\" </a>
750 (described below),
751 .\"
752 because in UTF-8 mode this would make it impossible to calculate the length of
753 the lookbehind.
754 .
755 .
756 .\" HTML <a name="characterclass"></a>
758 .rs
759 .sp
760 An opening square bracket introduces a character class, terminated by a closing
761 square bracket. A closing square bracket on its own is not special. If a
762 closing square bracket is required as a member of the class, it should be the
763 first data character in the class (after an initial circumflex, if present) or
764 escaped with a backslash.
765 .P
766 A character class matches a single character in the subject. In UTF-8 mode, the
767 character may occupy more than one byte. A matched character must be in the set
768 of characters defined by the class, unless the first character in the class
769 definition is a circumflex, in which case the subject character must not be in
770 the set defined by the class. If a circumflex is actually required as a member
771 of the class, ensure it is not the first character, or escape it with a
772 backslash.
773 .P
774 For example, the character class [aeiou] matches any lower case vowel, while
775 [^aeiou] matches any character that is not a lower case vowel. Note that a
776 circumflex is just a convenient notation for specifying the characters that
777 are in the class by enumerating those that are not. A class that starts with a
778 circumflex is not an assertion: it still consumes a character from the subject
779 string, and therefore it fails if the current pointer is at the end of the
780 string.
781 .P
782 In UTF-8 mode, characters with values greater than 255 can be included in a
783 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
784 .P
785 When caseless matching is set, any letters in a class represent both their
786 upper case and lower case versions, so for example, a caseless [aeiou] matches
787 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
788 caseful version would. In UTF-8 mode, PCRE always understands the concept of
789 case for characters whose values are less than 128, so caseless matching is
790 always possible. For characters with higher values, the concept of case is
791 supported if PCRE is compiled with Unicode property support, but not otherwise.
792 If you want to use caseless matching for characters 128 and above, you must
793 ensure that PCRE is compiled with Unicode property support as well as with
794 UTF-8 support.
795 .P
796 Characters that might indicate line breaks are never treated in any special way
797 when matching character classes, whatever line-ending sequence is in use, and
798 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
799 such as [^a] always matches one of these characters.
800 .P
801 The minus (hyphen) character can be used to specify a range of characters in a
802 character class. For example, [d-m] matches any letter between d and m,
803 inclusive. If a minus character is required in a class, it must be escaped with
804 a backslash or appear in a position where it cannot be interpreted as
805 indicating a range, typically as the first or last character in the class.
806 .P
807 It is not possible to have the literal character "]" as the end character of a
808 range. A pattern such as [W-]46] is interpreted as a class of two characters
809 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
810 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
811 the end of range, so [W-\e]46] is interpreted as a class containing a range
812 followed by two other characters. The octal or hexadecimal representation of
813 "]" can also be used to end a range.
814 .P
815 Ranges operate in the collating sequence of character values. They can also be
816 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
817 mode, ranges can include characters whose values are greater than 255, for
818 example [\ex{100}-\ex{2ff}].
819 .P
820 If a range that includes letters is used when caseless matching is set, it
821 matches the letters in either case. For example, [W-c] is equivalent to
822 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
823 tables for a French locale are in use, [\exc8-\excb] matches accented E
824 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
825 characters with values greater than 128 only when it is compiled with Unicode
826 property support.
827 .P
828 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
829 in a character class, and add the characters that they match to the class. For
830 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
831 conveniently be used with the upper case character types to specify a more
832 restricted set of characters than the matching lower case type. For example,
833 the class [^\eW_] matches any letter or digit, but not underscore.
834 .P
835 The only metacharacters that are recognized in character classes are backslash,
836 hyphen (only where it can be interpreted as specifying a range), circumflex
837 (only at the start), opening square bracket (only when it can be interpreted as
838 introducing a POSIX class name - see the next section), and the terminating
839 closing square bracket. However, escaping other non-alphanumeric characters
840 does no harm.
841 .
842 .
844 .rs
845 .sp
846 Perl supports the POSIX notation for character classes. This uses names
847 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
848 this notation. For example,
849 .sp
850 [01[:alpha:]%]
851 .sp
852 matches "0", "1", any alphabetic character, or "%". The supported class names
853 are
854 .sp
855 alnum letters and digits
856 alpha letters
857 ascii character codes 0 - 127
858 blank space or tab only
859 cntrl control characters
860 digit decimal digits (same as \ed)
861 graph printing characters, excluding space
862 lower lower case letters
863 print printing characters, including space
864 punct printing characters, excluding letters and digits
865 space white space (not quite the same as \es)
866 upper upper case letters
867 word "word" characters (same as \ew)
868 xdigit hexadecimal digits
869 .sp
870 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
871 space (32). Notice that this list includes the VT character (code 11). This
872 makes "space" different to \es, which does not include VT (for Perl
873 compatibility).
874 .P
875 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
876 5.8. Another Perl extension is negation, which is indicated by a ^ character
877 after the colon. For example,
878 .sp
879 [12[:^digit:]]
880 .sp
881 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
882 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
883 supported, and an error is given if they are encountered.
884 .P
885 In UTF-8 mode, characters with values greater than 128 do not match any of
886 the POSIX character classes.
887 .
888 .
890 .rs
891 .sp
892 Vertical bar characters are used to separate alternative patterns. For example,
893 the pattern
894 .sp
895 gilbert|sullivan
896 .sp
897 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
898 and an empty alternative is permitted (matching the empty string). The matching
899 process tries each alternative in turn, from left to right, and the first one
900 that succeeds is used. If the alternatives are within a subpattern
901 .\" HTML <a href="#subpattern">
902 .\" </a>
903 (defined below),
904 .\"
905 "succeeds" means matching the rest of the main pattern as well as the
906 alternative in the subpattern.
907 .
908 .
910 .rs
911 .sp
913 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
914 Perl option letters enclosed between "(?" and ")". The option letters are
915 .sp
918 s for PCRE_DOTALL
920 .sp
921 For example, (?im) sets caseless, multiline matching. It is also possible to
922 unset these options by preceding the letter with a hyphen, and a combined
923 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
924 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
925 permitted. If a letter appears both before and after the hyphen, the option is
926 unset.
927 .P
928 When an option change occurs at top level (that is, not inside subpattern
929 parentheses), the change applies to the remainder of the pattern that follows.
930 If the change is placed right at the start of a pattern, PCRE extracts it into
931 the global options (and it will therefore show up in data extracted by the
932 \fBpcre_fullinfo()\fP function).
933 .P
934 An option change within a subpattern (see below for a description of
935 subpatterns) affects only that part of the current pattern that follows it, so
936 .sp
937 (a(?i)b)c
938 .sp
939 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
940 By this means, options can be made to have different settings in different
941 parts of the pattern. Any changes made in one alternative do carry on
942 into subsequent branches within the same subpattern. For example,
943 .sp
944 (a(?i)b|c)
945 .sp
946 matches "ab", "aB", "c", and "C", even though when matching "C" the first
947 branch is abandoned before the option setting. This is because the effects of
948 option settings happen at compile time. There would be some very weird
949 behaviour otherwise.
950 .P
951 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
952 changed in the same way as the Perl-compatible options by using the characters
953 J, U and X respectively.
954 .
955 .
956 .\" HTML <a name="subpattern"></a>
958 .rs
959 .sp
960 Subpatterns are delimited by parentheses (round brackets), which can be nested.
961 Turning part of a pattern into a subpattern does two things:
962 .sp
963 1. It localizes a set of alternatives. For example, the pattern
964 .sp
965 cat(aract|erpillar|)
966 .sp
967 matches one of the words "cat", "cataract", or "caterpillar". Without the
968 parentheses, it would match "cataract", "erpillar" or an empty string.
969 .sp
970 2. It sets up the subpattern as a capturing subpattern. This means that, when
971 the whole pattern matches, that portion of the subject string that matched the
972 subpattern is passed back to the caller via the \fIovector\fP argument of
973 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
974 from 1) to obtain numbers for the capturing subpatterns.
975 .P
976 For example, if the string "the red king" is matched against the pattern
977 .sp
978 the ((red|white) (king|queen))
979 .sp
980 the captured substrings are "red king", "red", and "king", and are numbered 1,
981 2, and 3, respectively.
982 .P
983 The fact that plain parentheses fulfil two functions is not always helpful.
984 There are often times when a grouping subpattern is required without a
985 capturing requirement. If an opening parenthesis is followed by a question mark
986 and a colon, the subpattern does not do any capturing, and is not counted when
987 computing the number of any subsequent capturing subpatterns. For example, if
988 the string "the white queen" is matched against the pattern
989 .sp
990 the ((?:red|white) (king|queen))
991 .sp
992 the captured substrings are "white queen" and "queen", and are numbered 1 and
993 2. The maximum number of capturing subpatterns is 65535.
994 .P
995 As a convenient shorthand, if any option settings are required at the start of
996 a non-capturing subpattern, the option letters may appear between the "?" and
997 the ":". Thus the two patterns
998 .sp
999 (?i:saturday|sunday)
1000 (?:(?i)saturday|sunday)
1001 .sp
1002 match exactly the same set of strings. Because alternative branches are tried
1003 from left to right, and options are not reset until the end of the subpattern
1004 is reached, an option setting in one branch does affect subsequent branches, so
1005 the above patterns match "SUNDAY" as well as "Saturday".
1006 .
1007 .
1009 .rs
1010 .sp
1011 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1012 the same numbers for its capturing parentheses. Such a subpattern starts with
1013 (?| and is itself a non-capturing subpattern. For example, consider this
1014 pattern:
1015 .sp
1016 (?|(Sat)ur|(Sun))day
1017 .sp
1018 Because the two alternatives are inside a (?| group, both sets of capturing
1019 parentheses are numbered one. Thus, when the pattern matches, you can look
1020 at captured substring number one, whichever alternative matched. This construct
1021 is useful when you want to capture part, but not all, of one of a number of
1022 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1023 number is reset at the start of each branch. The numbers of any capturing
1024 buffers that follow the subpattern start after the highest number used in any
1025 branch. The following example is taken from the Perl documentation.
1026 The numbers underneath show in which buffer the captured content will be
1027 stored.
1028 .sp
1029 # before ---------------branch-reset----------- after
1030 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1031 # 1 2 2 3 2 3 4
1032 .sp
1033 A backreference or a recursive call to a numbered subpattern always refers to
1034 the first one in the pattern with the given number.
1035 .P
1036 An alternative approach to using this "branch reset" feature is to use
1037 duplicate named subpatterns, as described in the next section.
1038 .
1039 .
1041 .rs
1042 .sp
1043 Identifying capturing parentheses by number is simple, but it can be very hard
1044 to keep track of the numbers in complicated regular expressions. Furthermore,
1045 if an expression is modified, the numbers may change. To help with this
1046 difficulty, PCRE supports the naming of subpatterns. This feature was not
1047 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1048 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1049 the Perl and the Python syntax.
1050 .P
1051 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1052 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1053 parentheses from other parts of the pattern, such as
1054 .\" HTML <a href="#backreferences">
1055 .\" </a>
1056 backreferences,
1057 .\"
1058 .\" HTML <a href="#recursion">
1059 .\" </a>
1060 recursion,
1061 .\"
1062 and
1063 .\" HTML <a href="#conditions">
1064 .\" </a>
1065 conditions,
1066 .\"
1067 can be made by name as well as by number.
1068 .P
1069 Names consist of up to 32 alphanumeric characters and underscores. Named
1070 capturing parentheses are still allocated numbers as well as names, exactly as
1071 if the names were not present. The PCRE API provides function calls for
1072 extracting the name-to-number translation table from a compiled pattern. There
1073 is also a convenience function for extracting a captured substring by name.
1074 .P
1075 By default, a name must be unique within a pattern, but it is possible to relax
1076 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1077 be useful for patterns where only one instance of the named parentheses can
1078 match. Suppose you want to match the name of a weekday, either as a 3-letter
1079 abbreviation or as the full name, and in both cases you want to extract the
1080 abbreviation. This pattern (ignoring the line breaks) does the job:
1081 .sp
1082 (?<DN>Mon|Fri|Sun)(?:day)?|
1083 (?<DN>Tue)(?:sday)?|
1084 (?<DN>Wed)(?:nesday)?|
1085 (?<DN>Thu)(?:rsday)?|
1086 (?<DN>Sat)(?:urday)?
1087 .sp
1088 There are five capturing substrings, but only one is ever set after a match.
1089 (An alternative way of solving this problem is to use a "branch reset"
1090 subpattern, as described in the previous section.)
1091 .P
1092 The convenience function for extracting the data by name returns the substring
1093 for the first (and in this example, the only) subpattern of that name that
1094 matched. This saves searching to find which numbered subpattern it was. If you
1095 make a reference to a non-unique named subpattern from elsewhere in the
1096 pattern, the one that corresponds to the lowest number is used. For further
1097 details of the interfaces for handling named subpatterns, see the
1098 .\" HREF
1099 \fBpcreapi\fP
1100 .\"
1101 documentation.
1102 .
1103 .
1105 .rs
1106 .sp
1107 Repetition is specified by quantifiers, which can follow any of the following
1108 items:
1109 .sp
1110 a literal data character
1111 the dot metacharacter
1112 the \eC escape sequence
1113 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1114 the \eR escape sequence
1115 an escape such as \ed that matches a single character
1116 a character class
1117 a back reference (see next section)
1118 a parenthesized subpattern (unless it is an assertion)
1119 .sp
1120 The general repetition quantifier specifies a minimum and maximum number of
1121 permitted matches, by giving the two numbers in curly brackets (braces),
1122 separated by a comma. The numbers must be less than 65536, and the first must
1123 be less than or equal to the second. For example:
1124 .sp
1125 z{2,4}
1126 .sp
1127 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1128 character. If the second number is omitted, but the comma is present, there is
1129 no upper limit; if the second number and the comma are both omitted, the
1130 quantifier specifies an exact number of required matches. Thus
1131 .sp
1132 [aeiou]{3,}
1133 .sp
1134 matches at least 3 successive vowels, but may match many more, while
1135 .sp
1136 \ed{8}
1137 .sp
1138 matches exactly 8 digits. An opening curly bracket that appears in a position
1139 where a quantifier is not allowed, or one that does not match the syntax of a
1140 quantifier, is taken as a literal character. For example, {,6} is not a
1141 quantifier, but a literal string of four characters.
1142 .P
1143 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1144 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1145 which is represented by a two-byte sequence. Similarly, when Unicode property
1146 support is available, \eX{3} matches three Unicode extended sequences, each of
1147 which may be several bytes long (and they may be of different lengths).
1148 .P
1149 The quantifier {0} is permitted, causing the expression to behave as if the
1150 previous item and the quantifier were not present.
1151 .P
1152 For convenience, the three most common quantifiers have single-character
1153 abbreviations:
1154 .sp
1155 * is equivalent to {0,}
1156 + is equivalent to {1,}
1157 ? is equivalent to {0,1}
1158 .sp
1159 It is possible to construct infinite loops by following a subpattern that can
1160 match no characters with a quantifier that has no upper limit, for example:
1161 .sp
1162 (a?)*
1163 .sp
1164 Earlier versions of Perl and PCRE used to give an error at compile time for
1165 such patterns. However, because there are cases where this can be useful, such
1166 patterns are now accepted, but if any repetition of the subpattern does in fact
1167 match no characters, the loop is forcibly broken.
1168 .P
1169 By default, the quantifiers are "greedy", that is, they match as much as
1170 possible (up to the maximum number of permitted times), without causing the
1171 rest of the pattern to fail. The classic example of where this gives problems
1172 is in trying to match comments in C programs. These appear between /* and */
1173 and within the comment, individual * and / characters may appear. An attempt to
1174 match C comments by applying the pattern
1175 .sp
1176 /\e*.*\e*/
1177 .sp
1178 to the string
1179 .sp
1180 /* first comment */ not comment /* second comment */
1181 .sp
1182 fails, because it matches the entire string owing to the greediness of the .*
1183 item.
1184 .P
1185 However, if a quantifier is followed by a question mark, it ceases to be
1186 greedy, and instead matches the minimum number of times possible, so the
1187 pattern
1188 .sp
1189 /\e*.*?\e*/
1190 .sp
1191 does the right thing with the C comments. The meaning of the various
1192 quantifiers is not otherwise changed, just the preferred number of matches.
1193 Do not confuse this use of question mark with its use as a quantifier in its
1194 own right. Because it has two uses, it can sometimes appear doubled, as in
1195 .sp
1196 \ed??\ed
1197 .sp
1198 which matches one digit by preference, but can match two if that is the only
1199 way the rest of the pattern matches.
1200 .P
1201 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1202 the quantifiers are not greedy by default, but individual ones can be made
1203 greedy by following them with a question mark. In other words, it inverts the
1204 default behaviour.
1205 .P
1206 When a parenthesized subpattern is quantified with a minimum repeat count that
1207 is greater than 1 or with a limited maximum, more memory is required for the
1208 compiled pattern, in proportion to the size of the minimum or maximum.
1209 .P
1210 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1211 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1212 implicitly anchored, because whatever follows will be tried against every
1213 character position in the subject string, so there is no point in retrying the
1214 overall match at any position after the first. PCRE normally treats such a
1215 pattern as though it were preceded by \eA.
1216 .P
1217 In cases where it is known that the subject string contains no newlines, it is
1218 worth setting PCRE_DOTALL in order to obtain this optimization, or
1219 alternatively using ^ to indicate anchoring explicitly.
1220 .P
1221 However, there is one situation where the optimization cannot be used. When .*
1222 is inside capturing parentheses that are the subject of a backreference
1223 elsewhere in the pattern, a match at the start may fail where a later one
1224 succeeds. Consider, for example:
1225 .sp
1226 (.*)abc\e1
1227 .sp
1228 If the subject is "xyz123abc123" the match point is the fourth character. For
1229 this reason, such a pattern is not implicitly anchored.
1230 .P
1231 When a capturing subpattern is repeated, the value captured is the substring
1232 that matched the final iteration. For example, after
1233 .sp
1234 (tweedle[dume]{3}\es*)+
1235 .sp
1236 has matched "tweedledum tweedledee" the value of the captured substring is
1237 "tweedledee". However, if there are nested capturing subpatterns, the
1238 corresponding captured values may have been set in previous iterations. For
1239 example, after
1240 .sp
1241 /(a|(b))+/
1242 .sp
1243 matches "aba" the value of the second captured substring is "b".
1244 .
1245 .
1246 .\" HTML <a name="atomicgroup"></a>
1248 .rs
1249 .sp
1250 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1251 repetition, failure of what follows normally causes the repeated item to be
1252 re-evaluated to see if a different number of repeats allows the rest of the
1253 pattern to match. Sometimes it is useful to prevent this, either to change the
1254 nature of the match, or to cause it fail earlier than it otherwise might, when
1255 the author of the pattern knows there is no point in carrying on.
1256 .P
1257 Consider, for example, the pattern \ed+foo when applied to the subject line
1258 .sp
1259 123456bar
1260 .sp
1261 After matching all 6 digits and then failing to match "foo", the normal
1262 action of the matcher is to try again with only 5 digits matching the \ed+
1263 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1264 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1265 that once a subpattern has matched, it is not to be re-evaluated in this way.
1266 .P
1267 If we use atomic grouping for the previous example, the matcher gives up
1268 immediately on failing to match "foo" the first time. The notation is a kind of
1269 special parenthesis, starting with (?> as in this example:
1270 .sp
1271 (?>\ed+)foo
1272 .sp
1273 This kind of parenthesis "locks up" the part of the pattern it contains once
1274 it has matched, and a failure further into the pattern is prevented from
1275 backtracking into it. Backtracking past it to previous items, however, works as
1276 normal.
1277 .P
1278 An alternative description is that a subpattern of this type matches the string
1279 of characters that an identical standalone pattern would match, if anchored at
1280 the current point in the subject string.
1281 .P
1282 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1283 the above example can be thought of as a maximizing repeat that must swallow
1284 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1285 number of digits they match in order to make the rest of the pattern match,
1286 (?>\ed+) can only match an entire sequence of digits.
1287 .P
1288 Atomic groups in general can of course contain arbitrarily complicated
1289 subpatterns, and can be nested. However, when the subpattern for an atomic
1290 group is just a single repeated item, as in the example above, a simpler
1291 notation, called a "possessive quantifier" can be used. This consists of an
1292 additional + character following a quantifier. Using this notation, the
1293 previous example can be rewritten as
1294 .sp
1295 \ed++foo
1296 .sp
1297 Note that a possessive quantifier can be used with an entire group, for
1298 example:
1299 .sp
1300 (abc|xyz){2,3}+
1301 .sp
1302 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1303 option is ignored. They are a convenient notation for the simpler forms of
1304 atomic group. However, there is no difference in the meaning of a possessive
1305 quantifier and the equivalent atomic group, though there may be a performance
1306 difference; possessive quantifiers should be slightly faster.
1307 .P
1308 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1309 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1310 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1311 package, and PCRE copied it from there. It ultimately found its way into Perl
1312 at release 5.10.
1313 .P
1314 PCRE has an optimization that automatically "possessifies" certain simple
1315 pattern constructs. For example, the sequence A+B is treated as A++B because
1316 there is no point in backtracking into a sequence of A's when B must follow.
1317 .P
1318 When a pattern contains an unlimited repeat inside a subpattern that can itself
1319 be repeated an unlimited number of times, the use of an atomic group is the
1320 only way to avoid some failing matches taking a very long time indeed. The
1321 pattern
1322 .sp
1323 (\eD+|<\ed+>)*[!?]
1324 .sp
1325 matches an unlimited number of substrings that either consist of non-digits, or
1326 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1327 quickly. However, if it is applied to
1328 .sp
1329 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1330 .sp
1331 it takes a long time before reporting failure. This is because the string can
1332 be divided between the internal \eD+ repeat and the external * repeat in a
1333 large number of ways, and all have to be tried. (The example uses [!?] rather
1334 than a single character at the end, because both PCRE and Perl have an
1335 optimization that allows for fast failure when a single character is used. They
1336 remember the last single character that is required for a match, and fail early
1337 if it is not present in the string.) If the pattern is changed so that it uses
1338 an atomic group, like this:
1339 .sp
1340 ((?>\eD+)|<\ed+>)*[!?]
1341 .sp
1342 sequences of non-digits cannot be broken, and failure happens quickly.
1343 .
1344 .
1345 .\" HTML <a name="backreferences"></a>
1347 .rs
1348 .sp
1349 Outside a character class, a backslash followed by a digit greater than 0 (and
1350 possibly further digits) is a back reference to a capturing subpattern earlier
1351 (that is, to its left) in the pattern, provided there have been that many
1352 previous capturing left parentheses.
1353 .P
1354 However, if the decimal number following the backslash is less than 10, it is
1355 always taken as a back reference, and causes an error only if there are not
1356 that many capturing left parentheses in the entire pattern. In other words, the
1357 parentheses that are referenced need not be to the left of the reference for
1358 numbers less than 10. A "forward back reference" of this type can make sense
1359 when a repetition is involved and the subpattern to the right has participated
1360 in an earlier iteration.
1361 .P
1362 It is not possible to have a numerical "forward back reference" to a subpattern
1363 whose number is 10 or more using this syntax because a sequence such as \e50 is
1364 interpreted as a character defined in octal. See the subsection entitled
1365 "Non-printing characters"
1366 .\" HTML <a href="#digitsafterbackslash">
1367 .\" </a>
1368 above
1369 .\"
1370 for further details of the handling of digits following a backslash. There is
1371 no such problem when named parentheses are used. A back reference to any
1372 subpattern is possible using named parentheses (see below).
1373 .P
1374 Another way of avoiding the ambiguity inherent in the use of digits following a
1375 backslash is to use the \eg escape sequence, which is a feature introduced in
1376 Perl 5.10. This escape must be followed by an unsigned number or a negative
1377 number, optionally enclosed in braces. These examples are all identical:
1378 .sp
1379 (ring), \e1
1380 (ring), \eg1
1381 (ring), \eg{1}
1382 .sp
1383 An unsigned number specifies an absolute reference without the ambiguity that
1384 is present in the older syntax. It is also useful when literal digits follow
1385 the reference. A negative number is a relative reference. Consider this
1386 example:
1387 .sp
1388 (abc(def)ghi)\eg{-1}
1389 .sp
1390 The sequence \eg{-1} is a reference to the most recently started capturing
1391 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1392 would be equivalent to \e1. The use of relative references can be helpful in
1393 long patterns, and also in patterns that are created by joining together
1394 fragments that contain references within themselves.
1395 .P
1396 A back reference matches whatever actually matched the capturing subpattern in
1397 the current subject string, rather than anything matching the subpattern
1398 itself (see
1399 .\" HTML <a href="#subpatternsassubroutines">
1400 .\" </a>
1401 "Subpatterns as subroutines"
1402 .\"
1403 below for a way of doing that). So the pattern
1404 .sp
1405 (sens|respons)e and \e1ibility
1406 .sp
1407 matches "sense and sensibility" and "response and responsibility", but not
1408 "sense and responsibility". If caseful matching is in force at the time of the
1409 back reference, the case of letters is relevant. For example,
1410 .sp
1411 ((?i)rah)\es+\e1
1412 .sp
1413 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1414 capturing subpattern is matched caselessly.
1415 .P
1416 There are several different ways of writing back references to named
1417 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1418 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1419 back reference syntax, in which \eg can be used for both numeric and named
1420 references, is also supported. We could rewrite the above example in any of
1421 the following ways:
1422 .sp
1423 (?<p1>(?i)rah)\es+\ek<p1>
1424 (?'p1'(?i)rah)\es+\ek{p1}
1425 (?P<p1>(?i)rah)\es+(?P=p1)
1426 (?<p1>(?i)rah)\es+\eg{p1}
1427 .sp
1428 A subpattern that is referenced by name may appear in the pattern before or
1429 after the reference.
1430 .P
1431 There may be more than one back reference to the same subpattern. If a
1432 subpattern has not actually been used in a particular match, any back
1433 references to it always fail. For example, the pattern
1434 .sp
1435 (a|(bc))\e2
1436 .sp
1437 always fails if it starts to match "a" rather than "bc". Because there may be
1438 many capturing parentheses in a pattern, all digits following the backslash are
1439 taken as part of a potential back reference number. If the pattern continues
1440 with a digit character, some delimiter must be used to terminate the back
1441 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1442 Otherwise an empty comment (see
1443 .\" HTML <a href="#comments">
1444 .\" </a>
1445 "Comments"
1446 .\"
1447 below) can be used.
1448 .P
1449 A back reference that occurs inside the parentheses to which it refers fails
1450 when the subpattern is first used, so, for example, (a\e1) never matches.
1451 However, such references can be useful inside repeated subpatterns. For
1452 example, the pattern
1453 .sp
1454 (a|b\e1)+
1455 .sp
1456 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1457 the subpattern, the back reference matches the character string corresponding
1458 to the previous iteration. In order for this to work, the pattern must be such
1459 that the first iteration does not need to match the back reference. This can be
1460 done using alternation, as in the example above, or by a quantifier with a
1461 minimum of zero.
1462 .
1463 .
1464 .\" HTML <a name="bigassertions"></a>
1466 .rs
1467 .sp
1468 An assertion is a test on the characters following or preceding the current
1469 matching point that does not actually consume any characters. The simple
1470 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1471 .\" HTML <a href="#smallassertions">
1472 .\" </a>
1473 above.
1474 .\"
1475 .P
1476 More complicated assertions are coded as subpatterns. There are two kinds:
1477 those that look ahead of the current position in the subject string, and those
1478 that look behind it. An assertion subpattern is matched in the normal way,
1479 except that it does not cause the current matching position to be changed.
1480 .P
1481 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1482 because it makes no sense to assert the same thing several times. If any kind
1483 of assertion contains capturing subpatterns within it, these are counted for
1484 the purposes of numbering the capturing subpatterns in the whole pattern.
1485 However, substring capturing is carried out only for positive assertions,
1486 because it does not make sense for negative assertions.
1487 .
1488 .
1489 .SS "Lookahead assertions"
1490 .rs
1491 .sp
1492 Lookahead assertions start with (?= for positive assertions and (?! for
1493 negative assertions. For example,
1494 .sp
1495 \ew+(?=;)
1496 .sp
1497 matches a word followed by a semicolon, but does not include the semicolon in
1498 the match, and
1499 .sp
1500 foo(?!bar)
1501 .sp
1502 matches any occurrence of "foo" that is not followed by "bar". Note that the
1503 apparently similar pattern
1504 .sp
1505 (?!foo)bar
1506 .sp
1507 does not find an occurrence of "bar" that is preceded by something other than
1508 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1509 (?!foo) is always true when the next three characters are "bar". A
1510 lookbehind assertion is needed to achieve the other effect.
1511 .P
1512 If you want to force a matching failure at some point in a pattern, the most
1513 convenient way to do it is with (?!) because an empty string always matches, so
1514 an assertion that requires there not to be an empty string must always fail.
1515 .
1516 .
1517 .\" HTML <a name="lookbehind"></a>
1518 .SS "Lookbehind assertions"
1519 .rs
1520 .sp
1521 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1522 negative assertions. For example,
1523 .sp
1524 (?<!foo)bar
1525 .sp
1526 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1527 a lookbehind assertion are restricted such that all the strings it matches must
1528 have a fixed length. However, if there are several top-level alternatives, they
1529 do not all have to have the same fixed length. Thus
1530 .sp
1531 (?<=bullock|donkey)
1532 .sp
1533 is permitted, but
1534 .sp
1535 (?<!dogs?|cats?)
1536 .sp
1537 causes an error at compile time. Branches that match different length strings
1538 are permitted only at the top level of a lookbehind assertion. This is an
1539 extension compared with Perl (at least for 5.8), which requires all branches to
1540 match the same length of string. An assertion such as
1541 .sp
1542 (?<=ab(c|de))
1543 .sp
1544 is not permitted, because its single top-level branch can match two different
1545 lengths, but it is acceptable if rewritten to use two top-level branches:
1546 .sp
1547 (?<=abc|abde)
1548 .sp
1549 In some cases, the Perl 5.10 escape sequence \eK
1550 .\" HTML <a href="#resetmatchstart">
1551 .\" </a>
1552 (see above)
1553 .\"
1554 can be used instead of a lookbehind assertion; this is not restricted to a
1555 fixed-length.
1556 .P
1557 The implementation of lookbehind assertions is, for each alternative, to
1558 temporarily move the current position back by the fixed length and then try to
1559 match. If there are insufficient characters before the current position, the
1560 assertion fails.
1561 .P
1562 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1563 to appear in lookbehind assertions, because it makes it impossible to calculate
1564 the length of the lookbehind. The \eX and \eR escapes, which can match
1565 different numbers of bytes, are also not permitted.
1566 .P
1567 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1568 specify efficient matching at the end of the subject string. Consider a simple
1569 pattern such as
1570 .sp
1571 abcd$
1572 .sp
1573 when applied to a long string that does not match. Because matching proceeds
1574 from left to right, PCRE will look for each "a" in the subject and then see if
1575 what follows matches the rest of the pattern. If the pattern is specified as
1576 .sp
1577 ^.*abcd$
1578 .sp
1579 the initial .* matches the entire string at first, but when this fails (because
1580 there is no following "a"), it backtracks to match all but the last character,
1581 then all but the last two characters, and so on. Once again the search for "a"
1582 covers the entire string, from right to left, so we are no better off. However,
1583 if the pattern is written as
1584 .sp
1585 ^.*+(?<=abcd)
1586 .sp
1587 there can be no backtracking for the .*+ item; it can match only the entire
1588 string. The subsequent lookbehind assertion does a single test on the last four
1589 characters. If it fails, the match fails immediately. For long strings, this
1590 approach makes a significant difference to the processing time.
1591 .
1592 .
1593 .SS "Using multiple assertions"
1594 .rs
1595 .sp
1596 Several assertions (of any sort) may occur in succession. For example,
1597 .sp
1598 (?<=\ed{3})(?<!999)foo
1599 .sp
1600 matches "foo" preceded by three digits that are not "999". Notice that each of
1601 the assertions is applied independently at the same point in the subject
1602 string. First there is a check that the previous three characters are all
1603 digits, and then there is a check that the same three characters are not "999".
1604 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1605 of which are digits and the last three of which are not "999". For example, it
1606 doesn't match "123abcfoo". A pattern to do that is
1607 .sp
1608 (?<=\ed{3}...)(?<!999)foo
1609 .sp
1610 This time the first assertion looks at the preceding six characters, checking
1611 that the first three are digits, and then the second assertion checks that the
1612 preceding three characters are not "999".
1613 .P
1614 Assertions can be nested in any combination. For example,
1615 .sp
1616 (?<=(?<!foo)bar)baz
1617 .sp
1618 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1619 preceded by "foo", while
1620 .sp
1621 (?<=\ed{3}(?!999)...)foo
1622 .sp
1623 is another pattern that matches "foo" preceded by three digits and any three
1624 characters that are not "999".
1625 .
1626 .
1627 .\" HTML <a name="conditions"></a>
1629 .rs
1630 .sp
1631 It is possible to cause the matching process to obey a subpattern
1632 conditionally or to choose between two alternative subpatterns, depending on
1633 the result of an assertion, or whether a previous capturing subpattern matched
1634 or not. The two possible forms of conditional subpattern are
1635 .sp
1636 (?(condition)yes-pattern)
1637 (?(condition)yes-pattern|no-pattern)
1638 .sp
1639 If the condition is satisfied, the yes-pattern is used; otherwise the
1640 no-pattern (if present) is used. If there are more than two alternatives in the
1641 subpattern, a compile-time error occurs.
1642 .P
1643 There are four kinds of condition: references to subpatterns, references to
1644 recursion, a pseudo-condition called DEFINE, and assertions.
1645 .
1646 .SS "Checking for a used subpattern by number"
1647 .rs
1648 .sp
1649 If the text between the parentheses consists of a sequence of digits, the
1650 condition is true if the capturing subpattern of that number has previously
1651 matched. An alternative notation is to precede the digits with a plus or minus
1652 sign. In this case, the subpattern number is relative rather than absolute.
1653 The most recently opened parentheses can be referenced by (?(-1), the next most
1654 recent by (?(-2), and so on. In looping constructs it can also make sense to
1655 refer to subsequent groups with constructs such as (?(+2).
1656 .P
1657 Consider the following pattern, which contains non-significant white space to
1658 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1659 three parts for ease of discussion:
1660 .sp
1661 ( \e( )? [^()]+ (?(1) \e) )
1662 .sp
1663 The first part matches an optional opening parenthesis, and if that
1664 character is present, sets it as the first captured substring. The second part
1665 matches one or more characters that are not parentheses. The third part is a
1666 conditional subpattern that tests whether the first set of parentheses matched
1667 or not. If they did, that is, if subject started with an opening parenthesis,
1668 the condition is true, and so the yes-pattern is executed and a closing
1669 parenthesis is required. Otherwise, since no-pattern is not present, the
1670 subpattern matches nothing. In other words, this pattern matches a sequence of
1671 non-parentheses, optionally enclosed in parentheses.
1672 .P
1673 If you were embedding this pattern in a larger one, you could use a relative
1674 reference:
1675 .sp
1676 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1677 .sp
1678 This makes the fragment independent of the parentheses in the larger pattern.
1679 .
1680 .SS "Checking for a used subpattern by name"
1681 .rs
1682 .sp
1683 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1684 subpattern by name. For compatibility with earlier versions of PCRE, which had
1685 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1686 there is a possible ambiguity with this syntax, because subpattern names may
1687 consist entirely of digits. PCRE looks first for a named subpattern; if it
1688 cannot find one and the name consists entirely of digits, PCRE looks for a
1689 subpattern of that number, which must be greater than zero. Using subpattern
1690 names that consist entirely of digits is not recommended.
1691 .P
1692 Rewriting the above example to use a named subpattern gives this:
1693 .sp
1694 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1695 .sp
1696 .
1697 .SS "Checking for pattern recursion"
1698 .rs
1699 .sp
1700 If the condition is the string (R), and there is no subpattern with the name R,
1701 the condition is true if a recursive call to the whole pattern or any
1702 subpattern has been made. If digits or a name preceded by ampersand follow the
1703 letter R, for example:
1704 .sp
1705 (?(R3)...) or (?(R&name)...)
1706 .sp
1707 the condition is true if the most recent recursion is into the subpattern whose
1708 number or name is given. This condition does not check the entire recursion
1709 stack.
1710 .P
1711 At "top level", all these recursion test conditions are false. Recursive
1712 patterns are described below.
1713 .
1714 .SS "Defining subpatterns for use by reference only"
1715 .rs
1716 .sp
1717 If the condition is the string (DEFINE), and there is no subpattern with the
1718 name DEFINE, the condition is always false. In this case, there may be only one
1719 alternative in the subpattern. It is always skipped if control reaches this
1720 point in the pattern; the idea of DEFINE is that it can be used to define
1721 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1722 is described below.) For example, a pattern to match an IPv4 address could be
1723 written like this (ignore whitespace and line breaks):
1724 .sp
1725 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1726 \eb (?&byte) (\e.(?&byte)){3} \eb
1727 .sp
1728 The first part of the pattern is a DEFINE group inside which a another group
1729 named "byte" is defined. This matches an individual component of an IPv4
1730 address (a number less than 256). When matching takes place, this part of the
1731 pattern is skipped because DEFINE acts like a false condition.
1732 .P
1733 The rest of the pattern uses references to the named group to match the four
1734 dot-separated components of an IPv4 address, insisting on a word boundary at
1735 each end.
1736 .
1737 .SS "Assertion conditions"
1738 .rs
1739 .sp
1740 If the condition is not in any of the above formats, it must be an assertion.
1741 This may be a positive or negative lookahead or lookbehind assertion. Consider
1742 this pattern, again containing non-significant white space, and with the two
1743 alternatives on the second line:
1744 .sp
1745 (?(?=[^a-z]*[a-z])
1746 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1747 .sp
1748 The condition is a positive lookahead assertion that matches an optional
1749 sequence of non-letters followed by a letter. In other words, it tests for the
1750 presence of at least one letter in the subject. If a letter is found, the
1751 subject is matched against the first alternative; otherwise it is matched
1752 against the second. This pattern matches strings in one of the two forms
1753 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1754 .
1755 .
1756 .\" HTML <a name="comments"></a>
1758 .rs
1759 .sp
1760 The sequence (?# marks the start of a comment that continues up to the next
1761 closing parenthesis. Nested parentheses are not permitted. The characters
1762 that make up a comment play no part in the pattern matching at all.
1763 .P
1764 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1765 character class introduces a comment that continues to immediately after the
1766 next newline in the pattern.
1767 .
1768 .
1769 .\" HTML <a name="recursion"></a>
1771 .rs
1772 .sp
1773 Consider the problem of matching a string in parentheses, allowing for
1774 unlimited nested parentheses. Without the use of recursion, the best that can
1775 be done is to use a pattern that matches up to some fixed depth of nesting. It
1776 is not possible to handle an arbitrary nesting depth.
1777 .P
1778 For some time, Perl has provided a facility that allows regular expressions to
1779 recurse (amongst other things). It does this by interpolating Perl code in the
1780 expression at run time, and the code can refer to the expression itself. A Perl
1781 pattern using code interpolation to solve the parentheses problem can be
1782 created like this:
1783 .sp
1784 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1785 .sp
1786 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1787 recursively to the pattern in which it appears.
1788 .P
1789 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1790 supports special syntax for recursion of the entire pattern, and also for
1791 individual subpattern recursion. After its introduction in PCRE and Python,
1792 this kind of recursion was introduced into Perl at release 5.10.
1793 .P
1794 A special item that consists of (? followed by a number greater than zero and a
1795 closing parenthesis is a recursive call of the subpattern of the given number,
1796 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1797 call, which is described in the next section.) The special item (?R) or (?0) is
1798 a recursive call of the entire regular expression.
1799 .P
1800 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1801 treated as an atomic group. That is, once it has matched some of the subject
1802 string, it is never re-entered, even if it contains untried alternatives and
1803 there is a subsequent matching failure.
1804 .P
1805 This PCRE pattern solves the nested parentheses problem (assume the
1806 PCRE_EXTENDED option is set so that white space is ignored):
1807 .sp
1808 \e( ( (?>[^()]+) | (?R) )* \e)
1809 .sp
1810 First it matches an opening parenthesis. Then it matches any number of
1811 substrings which can either be a sequence of non-parentheses, or a recursive
1812 match of the pattern itself (that is, a correctly parenthesized substring).
1813 Finally there is a closing parenthesis.
1814 .P
1815 If this were part of a larger pattern, you would not want to recurse the entire
1816 pattern, so instead you could use this:
1817 .sp
1818 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1819 .sp
1820 We have put the pattern into parentheses, and caused the recursion to refer to
1821 them instead of the whole pattern.
1822 .P
1823 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1824 is made easier by the use of relative references. (A Perl 5.10 feature.)
1825 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1826 most recently opened parentheses preceding the recursion. In other words, a
1827 negative number counts capturing parentheses leftwards from the point at which
1828 it is encountered.
1829 .P
1830 It is also possible to refer to subsequently opened parentheses, by writing
1831 references such as (?+2). However, these cannot be recursive because the
1832 reference is not inside the parentheses that are referenced. They are always
1833 "subroutine" calls, as described in the next section.
1834 .P
1835 An alternative approach is to use named parentheses instead. The Perl syntax
1836 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
1837 could rewrite the above example as follows:
1838 .sp
1839 (?<pn> \e( ( (?>[^()]+) | (?&pn) )* \e) )
1840 .sp
1841 If there is more than one subpattern with the same name, the earliest one is
1842 used.
1843 .P
1844 This particular example pattern that we have been looking at contains nested
1845 unlimited repeats, and so the use of atomic grouping for matching strings of
1846 non-parentheses is important when applying the pattern to strings that do not
1847 match. For example, when this pattern is applied to
1848 .sp
1849 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1850 .sp
1851 it yields "no match" quickly. However, if atomic grouping is not used,
1852 the match runs for a very long time indeed because there are so many different
1853 ways the + and * repeats can carve up the subject, and all have to be tested
1854 before failure can be reported.
1855 .P
1856 At the end of a match, the values set for any capturing subpatterns are those
1857 from the outermost level of the recursion at which the subpattern value is set.
1858 If you want to obtain intermediate values, a callout function can be used (see
1859 below and the
1860 .\" HREF
1861 \fBpcrecallout\fP
1862 .\"
1863 documentation). If the pattern above is matched against
1864 .sp
1865 (ab(cd)ef)
1866 .sp
1867 the value for the capturing parentheses is "ef", which is the last value taken
1868 on at the top level. If additional parentheses are added, giving
1869 .sp
1870 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1871 ^ ^
1872 ^ ^
1873 .sp
1874 the string they capture is "ab(cd)ef", the contents of the top level
1875 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1876 has to obtain extra memory to store data during a recursion, which it does by
1877 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1878 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1879 .P
1880 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1881 Consider this pattern, which matches text in angle brackets, allowing for
1882 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1883 recursing), whereas any characters are permitted at the outer level.
1884 .sp
1885 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1886 .sp
1887 In this pattern, (?(R) is the start of a conditional subpattern, with two
1888 different alternatives for the recursive and non-recursive cases. The (?R) item
1889 is the actual recursive call.
1890 .
1891 .
1892 .\" HTML <a name="subpatternsassubroutines"></a>
1894 .rs
1895 .sp
1896 If the syntax for a recursive subpattern reference (either by number or by
1897 name) is used outside the parentheses to which it refers, it operates like a
1898 subroutine in a programming language. The "called" subpattern may be defined
1899 before or after the reference. A numbered reference can be absolute or
1900 relative, as in these examples:
1901 .sp
1902 (...(absolute)...)...(?2)...
1903 (...(relative)...)...(?-1)...
1904 (...(?+1)...(relative)...
1905 .sp
1906 An earlier example pointed out that the pattern
1907 .sp
1908 (sens|respons)e and \e1ibility
1909 .sp
1910 matches "sense and sensibility" and "response and responsibility", but not
1911 "sense and responsibility". If instead the pattern
1912 .sp
1913 (sens|respons)e and (?1)ibility
1914 .sp
1915 is used, it does match "sense and responsibility" as well as the other two
1916 strings. Another example is given in the discussion of DEFINE above.
1917 .P
1918 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1919 group. That is, once it has matched some of the subject string, it is never
1920 re-entered, even if it contains untried alternatives and there is a subsequent
1921 matching failure.
1922 .P
1923 When a subpattern is used as a subroutine, processing options such as
1924 case-independence are fixed when the subpattern is defined. They cannot be
1925 changed for different calls. For example, consider this pattern:
1926 .sp
1927 (abc)(?i:(?-1))
1928 .sp
1929 It matches "abcabc". It does not match "abcABC" because the change of
1930 processing option does not affect the called subpattern.
1931 .
1932 .
1934 .rs
1935 .sp
1936 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1937 code to be obeyed in the middle of matching a regular expression. This makes it
1938 possible, amongst other things, to extract different substrings that match the
1939 same pair of parentheses when there is a repetition.
1940 .P
1941 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1942 code. The feature is called "callout". The caller of PCRE provides an external
1943 function by putting its entry point in the global variable \fIpcre_callout\fP.
1944 By default, this variable contains NULL, which disables all calling out.
1945 .P
1946 Within a regular expression, (?C) indicates the points at which the external
1947 function is to be called. If you want to identify different callout points, you
1948 can put a number less than 256 after the letter C. The default value is zero.
1949 For example, this pattern has two callout points:
1950 .sp
1951 (?C1)abc(?C2)def
1952 .sp
1953 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
1954 automatically installed before each item in the pattern. They are all numbered
1955 255.
1956 .P
1957 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
1958 set), the external function is called. It is provided with the number of the
1959 callout, the position in the pattern, and, optionally, one item of data
1960 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
1961 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1962 description of the interface to the callout function is given in the
1963 .\" HREF
1964 \fBpcrecallout\fP
1965 .\"
1966 documentation.
1967 .
1968 .
1970 .rs
1971 .sp
1972 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
1973 are described in the Perl documentation as "experimental and subject to change
1974 or removal in a future version of Perl". It goes on to say: "Their usage in
1975 production code should be noted to avoid problems during upgrades." The same
1976 remarks apply to the PCRE features described in this section.
1977 .P
1978 Since these verbs are specifically related to backtracking, they can be used
1979 only when the pattern is to be matched using \fBpcre_exec()\fP, which uses a
1980 backtracking algorithm. They cause an error if encountered by
1981 \fBpcre_dfa_exec()\fP.
1982 .P
1983 The new verbs make use of what was previously invalid syntax: an opening
1984 parenthesis followed by an asterisk. In Perl, they are generally of the form
1985 (*VERB:ARG) but PCRE does not support the use of arguments, so its general
1986 form is just (*VERB). Any number of these verbs may occur in a pattern. There
1987 are two kinds:
1988 .
1989 .SS "Verbs that act immediately"
1990 .rs
1991 .sp
1992 The following verbs act as soon as they are encountered:
1993 .sp
1994 (*ACCEPT)
1995 .sp
1996 This verb causes the match to end successfully, skipping the remainder of the
1997 pattern. When inside a recursion, only the innermost pattern is ended
1998 immediately. PCRE differs from Perl in what happens if the (*ACCEPT) is inside
1999 capturing parentheses. In Perl, the data so far is captured: in PCRE no data is
2000 captured. For example:
2001 .sp
2002 A(A|B(*ACCEPT)|C)D
2003 .sp
2004 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data is
2005 captured.
2006 .sp
2007 (*FAIL) or (*F)
2008 .sp
2009 This verb causes the match to fail, forcing backtracking to occur. It is
2010 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2011 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2012 Perl features that are not present in PCRE. The nearest equivalent is the
2013 callout feature, as for example in this pattern:
2014 .sp
2015 a+(?C)(*FAIL)
2016 .sp
2017 A match with the string "aaaa" always fails, but the callout is taken before
2018 each backtrack happens (in this example, 10 times).
2019 .
2020 .SS "Verbs that act after backtracking"
2021 .rs
2022 .sp
2023 The following verbs do nothing when they are encountered. Matching continues
2024 with what follows, but if there is no subsequent match, a failure is forced.
2025 The verbs differ in exactly what kind of failure occurs.
2026 .sp
2027 (*COMMIT)
2028 .sp
2029 This verb causes the whole match to fail outright if the rest of the pattern
2030 does not match. Even if the pattern is unanchored, no further attempts to find
2031 a match by advancing the start point take place. Once (*COMMIT) has been
2032 passed, \fBpcre_exec()\fP is committed to finding a match at the current
2033 starting point, or not at all. For example:
2034 .sp
2035 a+(*COMMIT)b
2036 .sp
2037 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2038 dynamic anchor, or "I've started, so I must finish."
2039 .sp
2040 (*PRUNE)
2041 .sp
2042 This verb causes the match to fail at the current position if the rest of the
2043 pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2044 advance to the next starting character then happens. Backtracking can occur as
2045 usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2046 if there is no match to the right, backtracking cannot cross (*PRUNE).
2047 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2048 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2049 be expressed in any other way.
2050 .sp
2051 (*SKIP)
2052 .sp
2053 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2054 "bumpalong" advance is not to the next character, but to the position in the
2055 subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2056 was matched leading up to it cannot be part of a successful match. Consider:
2057 .sp
2058 a+(*SKIP)b
2059 .sp
2060 If the subject is "aaaac...", after the first match attempt fails (starting at
2061 the first character in the string), the starting point skips on to start the
2062 next attempt at "c". Note that a possessive quantifer does not have the same
2063 effect in this example; although it would suppress backtracking during the
2064 first match attempt, the second attempt would start at the second character
2065 instead of skipping on to "c".
2066 .sp
2067 (*THEN)
2068 .sp
2069 This verb causes a skip to the next alternation if the rest of the pattern does
2070 not match. That is, it cancels pending backtracking, but only within the
2071 current alternation. Its name comes from the observation that it can be used
2072 for a pattern-based if-then-else block:
2073 .sp
2074 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2075 .sp
2076 If the COND1 pattern matches, FOO is tried (and possibly further items after
2077 the end of the group if FOO succeeds); on failure the matcher skips to the
2078 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2079 is used outside of any alternation, it acts exactly like (*PRUNE).
2080 .
2081 .
2082 .SH "SEE ALSO"
2083 .rs
2084 .sp
2085 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3), \fBpcre\fP(3).
2086 .
2087 .
2089 .rs
2090 .sp
2091 .nf
2092 Philip Hazel
2093 University Computing Service
2094 Cambridge CB2 3QH, England.
2095 .fi
2096 .
2097 .
2099 .rs
2100 .sp
2101 .nf
2102 Last updated: 08 August 2007
2103 Copyright (c) 1997-2007 University of Cambridge.
2104 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5