ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 208 - (show annotations)
Mon Aug 6 15:23:29 2007 UTC (13 years, 7 months ago) by ph10
File MIME type: text/html
File size: 84388 byte(s)
Error occurred while calculating annotation data.
Added a pcresyntax man page; tidied some others.
1 <html>
2 <head>
3 <title>pcrepattern specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 <h1>pcrepattern man page</h1>
7 <p>
8 Return to the <a href="index.html">PCRE index page</a>.
9 </p>
10 <p>
11 This page is part of the PCRE HTML documentation. It was generated automatically
12 from the original man page. If there is any nonsense in it, please consult the
13 man page, in case the conversion went wrong.
14 <br>
15 <ul>
16 <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 <li><a name="TOC2" href="#SEC2">CHARACTERS AND METACHARACTERS</a>
18 <li><a name="TOC3" href="#SEC3">BACKSLASH</a>
19 <li><a name="TOC4" href="#SEC4">CIRCUMFLEX AND DOLLAR</a>
20 <li><a name="TOC5" href="#SEC5">FULL STOP (PERIOD, DOT)</a>
21 <li><a name="TOC6" href="#SEC6">MATCHING A SINGLE BYTE</a>
23 <li><a name="TOC8" href="#SEC8">POSIX CHARACTER CLASSES</a>
24 <li><a name="TOC9" href="#SEC9">VERTICAL BAR</a>
25 <li><a name="TOC10" href="#SEC10">INTERNAL OPTION SETTING</a>
26 <li><a name="TOC11" href="#SEC11">SUBPATTERNS</a>
27 <li><a name="TOC12" href="#SEC12">DUPLICATE SUBPATTERN NUMBERS</a>
28 <li><a name="TOC13" href="#SEC13">NAMED SUBPATTERNS</a>
29 <li><a name="TOC14" href="#SEC14">REPETITION</a>
31 <li><a name="TOC16" href="#SEC16">BACK REFERENCES</a>
32 <li><a name="TOC17" href="#SEC17">ASSERTIONS</a>
33 <li><a name="TOC18" href="#SEC18">CONDITIONAL SUBPATTERNS</a>
34 <li><a name="TOC19" href="#SEC19">COMMENTS</a>
35 <li><a name="TOC20" href="#SEC20">RECURSIVE PATTERNS</a>
36 <li><a name="TOC21" href="#SEC21">SUBPATTERNS AS SUBROUTINES</a>
37 <li><a name="TOC22" href="#SEC22">CALLOUTS</a>
38 <li><a name="TOC23" href="#SEC23">SEE ALSO</a>
39 <li><a name="TOC24" href="#SEC24">AUTHOR</a>
40 <li><a name="TOC25" href="#SEC25">REVISION</a>
41 </ul>
42 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
43 <P>
44 The syntax and semantics of the regular expressions that are supported by PCRE
45 are described in detail below. There is a quick-reference syntax summary in the
46 <a href="pcresyntax.html"><b>pcresyntax</b></a>
47 page. Perl's regular expressions are described in its own documentation, and
48 regular expressions in general are covered in a number of books, some of which
49 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
50 published by O'Reilly, covers regular expressions in great detail. This
51 description of PCRE's regular expressions is intended as reference material.
52 </P>
53 <P>
54 The original operation of PCRE was on strings of one-byte characters. However,
55 there is now also support for UTF-8 character strings. To use this, you must
56 build PCRE to include UTF-8 support, and then call <b>pcre_compile()</b> with
57 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
58 places below. There is also a summary of UTF-8 features in the
59 <a href="pcre.html#utf8support">section on UTF-8 support</a>
60 in the main
61 <a href="pcre.html"><b>pcre</b></a>
62 page.
63 </P>
64 <P>
65 The remainder of this document discusses the patterns that are supported by
66 PCRE when its main matching function, <b>pcre_exec()</b>, is used.
67 From release 6.0, PCRE offers a second matching function,
68 <b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not
69 Perl-compatible. Some of the features discussed below are not available when
70 <b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the
71 alternative function, and how it differs from the normal function, are
72 discussed in the
73 <a href="pcrematching.html"><b>pcrematching</b></a>
74 page.
75 </P>
76 <br><a name="SEC2" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
77 <P>
78 A regular expression is a pattern that is matched against a subject string from
79 left to right. Most characters stand for themselves in a pattern, and match the
80 corresponding characters in the subject. As a trivial example, the pattern
81 <pre>
82 The quick brown fox
83 </pre>
84 matches a portion of a subject string that is identical to itself. When
85 caseless matching is specified (the PCRE_CASELESS option), letters are matched
86 independently of case. In UTF-8 mode, PCRE always understands the concept of
87 case for characters whose values are less than 128, so caseless matching is
88 always possible. For characters with higher values, the concept of case is
89 supported if PCRE is compiled with Unicode property support, but not otherwise.
90 If you want to use caseless matching for characters 128 and above, you must
91 ensure that PCRE is compiled with Unicode property support as well as with
92 UTF-8 support.
93 </P>
94 <P>
95 The power of regular expressions comes from the ability to include alternatives
96 and repetitions in the pattern. These are encoded in the pattern by the use of
97 <i>metacharacters</i>, which do not stand for themselves but instead are
98 interpreted in some special way.
99 </P>
100 <P>
101 There are two different sets of metacharacters: those that are recognized
102 anywhere in the pattern except within square brackets, and those that are
103 recognized within square brackets. Outside square brackets, the metacharacters
104 are as follows:
105 <pre>
106 \ general escape character with several uses
107 ^ assert start of string (or line, in multiline mode)
108 $ assert end of string (or line, in multiline mode)
109 . match any character except newline (by default)
110 [ start character class definition
111 | start of alternative branch
112 ( start subpattern
113 ) end subpattern
114 ? extends the meaning of (
115 also 0 or 1 quantifier
116 also quantifier minimizer
117 * 0 or more quantifier
118 + 1 or more quantifier
119 also "possessive quantifier"
120 { start min/max quantifier
121 </pre>
122 Part of a pattern that is in square brackets is called a "character class". In
123 a character class the only metacharacters are:
124 <pre>
125 \ general escape character
126 ^ negate the class, but only if the first character
127 - indicates character range
128 [ POSIX character class (only if followed by POSIX syntax)
129 ] terminates the character class
130 </pre>
131 The following sections describe the use of each of the metacharacters.
132 </P>
133 <br><a name="SEC3" href="#TOC1">BACKSLASH</a><br>
134 <P>
135 The backslash character has several uses. Firstly, if it is followed by a
136 non-alphanumeric character, it takes away any special meaning that character
137 may have. This use of backslash as an escape character applies both inside and
138 outside character classes.
139 </P>
140 <P>
141 For example, if you want to match a * character, you write \* in the pattern.
142 This escaping action applies whether or not the following character would
143 otherwise be interpreted as a metacharacter, so it is always safe to precede a
144 non-alphanumeric with backslash to specify that it stands for itself. In
145 particular, if you want to match a backslash, you write \\.
146 </P>
147 <P>
148 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
149 pattern (other than in a character class) and characters between a # outside
150 a character class and the next newline are ignored. An escaping backslash can
151 be used to include a whitespace or # character as part of the pattern.
152 </P>
153 <P>
154 If you want to remove the special meaning from a sequence of characters, you
155 can do so by putting them between \Q and \E. This is different from Perl in
156 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
157 Perl, $ and @ cause variable interpolation. Note the following examples:
158 <pre>
159 Pattern PCRE matches Perl matches
161 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
162 \Qabc\$xyz\E abc\$xyz abc\$xyz
163 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
164 </pre>
165 The \Q...\E sequence is recognized both inside and outside character classes.
166 <a name="digitsafterbackslash"></a></P>
167 <br><b>
168 Non-printing characters
169 </b><br>
170 <P>
171 A second use of backslash provides a way of encoding non-printing characters
172 in patterns in a visible manner. There is no restriction on the appearance of
173 non-printing characters, apart from the binary zero that terminates a pattern,
174 but when a pattern is being prepared by text editing, it is usually easier to
175 use one of the following escape sequences than the binary character it
176 represents:
177 <pre>
178 \a alarm, that is, the BEL character (hex 07)
179 \cx "control-x", where x is any character
180 \e escape (hex 1B)
181 \f formfeed (hex 0C)
182 \n newline (hex 0A)
183 \r carriage return (hex 0D)
184 \t tab (hex 09)
185 \ddd character with octal code ddd, or backreference
186 \xhh character with hex code hh
187 \x{hhh..} character with hex code hhh..
188 </pre>
189 The precise effect of \cx is as follows: if x is a lower case letter, it
190 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
191 Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
192 7B.
193 </P>
194 <P>
195 After \x, from zero to two hexadecimal digits are read (letters can be in
196 upper or lower case). Any number of hexadecimal digits may appear between \x{
197 and }, but the value of the character code must be less than 256 in non-UTF-8
198 mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
199 is 7FFFFFFF). If characters other than hexadecimal digits appear between \x{
200 and }, or if there is no terminating }, this form of escape is not recognized.
201 Instead, the initial \x will be interpreted as a basic hexadecimal escape,
202 with no following digits, giving a character whose value is zero.
203 </P>
204 <P>
205 Characters whose value is less than 256 can be defined by either of the two
206 syntaxes for \x. There is no difference in the way they are handled. For
207 example, \xdc is exactly the same as \x{dc}.
208 </P>
209 <P>
210 After \0 up to two further octal digits are read. If there are fewer than two
211 digits, just those that are present are used. Thus the sequence \0\x\07
212 specifies two binary zeros followed by a BEL character (code value 7). Make
213 sure you supply two digits after the initial zero if the pattern character that
214 follows is itself an octal digit.
215 </P>
216 <P>
217 The handling of a backslash followed by a digit other than 0 is complicated.
218 Outside a character class, PCRE reads it and any following digits as a decimal
219 number. If the number is less than 10, or if there have been at least that many
220 previous capturing left parentheses in the expression, the entire sequence is
221 taken as a <i>back reference</i>. A description of how this works is given
222 <a href="#backreferences">later,</a>
223 following the discussion of
224 <a href="#subpattern">parenthesized subpatterns.</a>
225 </P>
226 <P>
227 Inside a character class, or if the decimal number is greater than 9 and there
228 have not been that many capturing subpatterns, PCRE re-reads up to three octal
229 digits following the backslash, and uses them to generate a data character. Any
230 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
231 character specified in octal must be less than \400. In UTF-8 mode, values up
232 to \777 are permitted. For example:
233 <pre>
234 \040 is another way of writing a space
235 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
236 \7 is always a back reference
237 \11 might be a back reference, or another way of writing a tab
238 \011 is always a tab
239 \0113 is a tab followed by the character "3"
240 \113 might be a back reference, otherwise the character with octal code 113
241 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
242 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
243 </pre>
244 Note that octal values of 100 or greater must not be introduced by a leading
245 zero, because no more than three octal digits are ever read.
246 </P>
247 <P>
248 All the sequences that define a single character value can be used both inside
249 and outside character classes. In addition, inside a character class, the
250 sequence \b is interpreted as the backspace character (hex 08), and the
251 sequences \R and \X are interpreted as the characters "R" and "X",
252 respectively. Outside a character class, these sequences have different
253 meanings
254 <a href="#uniextseq">(see below).</a>
255 </P>
256 <br><b>
257 Absolute and relative back references
258 </b><br>
259 <P>
260 The sequence \g followed by an unsigned or a negative number, optionally
261 enclosed in braces, is an absolute or relative back reference. A named back
262 reference can be coded as \g{name}. Back references are discussed
263 <a href="#backreferences">later,</a>
264 following the discussion of
265 <a href="#subpattern">parenthesized subpatterns.</a>
266 </P>
267 <br><b>
268 Generic character types
269 </b><br>
270 <P>
271 Another use of backslash is for specifying generic character types. The
272 following are always recognized:
273 <pre>
274 \d any decimal digit
275 \D any character that is not a decimal digit
276 \h any horizontal whitespace character
277 \H any character that is not a horizontal whitespace character
278 \s any whitespace character
279 \S any character that is not a whitespace character
280 \v any vertical whitespace character
281 \V any character that is not a vertical whitespace character
282 \w any "word" character
283 \W any "non-word" character
284 </pre>
285 Each pair of escape sequences partitions the complete set of characters into
286 two disjoint sets. Any given character matches one, and only one, of each pair.
287 </P>
288 <P>
289 These character type sequences can appear both inside and outside character
290 classes. They each match one character of the appropriate type. If the current
291 matching point is at the end of the subject string, all of them fail, since
292 there is no character to match.
293 </P>
294 <P>
295 For compatibility with Perl, \s does not match the VT character (code 11).
296 This makes it different from the the POSIX "space" class. The \s characters
297 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
298 included in a Perl script, \s may match the VT character. In PCRE, it never
299 does.
300 </P>
301 <P>
302 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
303 \w, and always match \D, \S, and \W. This is true even when Unicode
304 character property support is available. These sequences retain their original
305 meanings from before UTF-8 support was available, mainly for efficiency
306 reasons.
307 </P>
308 <P>
309 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to the
310 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
311 The horizontal space characters are:
312 <pre>
313 U+0009 Horizontal tab
314 U+0020 Space
315 U+00A0 Non-break space
316 U+1680 Ogham space mark
317 U+180E Mongolian vowel separator
318 U+2000 En quad
319 U+2001 Em quad
320 U+2002 En space
321 U+2003 Em space
322 U+2004 Three-per-em space
323 U+2005 Four-per-em space
324 U+2006 Six-per-em space
325 U+2007 Figure space
326 U+2008 Punctuation space
327 U+2009 Thin space
328 U+200A Hair space
329 U+202F Narrow no-break space
330 U+205F Medium mathematical space
331 U+3000 Ideographic space
332 </pre>
333 The vertical space characters are:
334 <pre>
335 U+000A Linefeed
336 U+000B Vertical tab
337 U+000C Formfeed
338 U+000D Carriage return
339 U+0085 Next line
340 U+2028 Line separator
341 U+2029 Paragraph separator
342 </PRE>
343 </P>
344 <P>
345 A "word" character is an underscore or any character less than 256 that is a
346 letter or digit. The definition of letters and digits is controlled by PCRE's
347 low-valued character tables, and may vary if locale-specific matching is taking
348 place (see
349 <a href="pcreapi.html#localesupport">"Locale support"</a>
350 in the
351 <a href="pcreapi.html"><b>pcreapi</b></a>
352 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
353 or "french" in Windows, some character codes greater than 128 are used for
354 accented letters, and these are matched by \w. The use of locales with Unicode
355 is discouraged.
356 </P>
357 <br><b>
358 Newline sequences
359 </b><br>
360 <P>
361 Outside a character class, the escape sequence \R matches any Unicode newline
362 sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is equivalent to
363 the following:
364 <pre>
365 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
366 </pre>
367 This is an example of an "atomic group", details of which are given
368 <a href="#atomicgroup">below.</a>
369 This particular group matches either the two-character sequence CR followed by
370 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
371 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
372 line, U+0085). The two-character sequence is treated as a single unit that
373 cannot be split.
374 </P>
375 <P>
376 In UTF-8 mode, two additional characters whose codepoints are greater than 255
377 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
378 Unicode character property support is not needed for these characters to be
379 recognized.
380 </P>
381 <P>
382 Inside a character class, \R matches the letter "R".
383 <a name="uniextseq"></a></P>
384 <br><b>
385 Unicode character properties
386 </b><br>
387 <P>
388 When PCRE is built with Unicode character property support, three additional
389 escape sequences that match characters with specific properties are available.
390 When not in UTF-8 mode, these sequences are of course limited to testing
391 characters whose codepoints are less than 256, but they do work in this mode.
392 The extra escape sequences are:
393 <pre>
394 \p{<i>xx</i>} a character with the <i>xx</i> property
395 \P{<i>xx</i>} a character without the <i>xx</i> property
396 \X an extended Unicode sequence
397 </pre>
398 The property names represented by <i>xx</i> above are limited to the Unicode
399 script names, the general category properties, and "Any", which matches any
400 character (including newline). Other properties such as "InMusicalSymbols" are
401 not currently supported by PCRE. Note that \P{Any} does not match any
402 characters, so always causes a match failure.
403 </P>
404 <P>
405 Sets of Unicode characters are defined as belonging to certain scripts. A
406 character from one of these sets can be matched using a script name. For
407 example:
408 <pre>
409 \p{Greek}
410 \P{Han}
411 </pre>
412 Those that are not part of an identified script are lumped together as
413 "Common". The current list of scripts is:
414 </P>
415 <P>
416 Arabic,
417 Armenian,
418 Balinese,
419 Bengali,
420 Bopomofo,
421 Braille,
422 Buginese,
423 Buhid,
424 Canadian_Aboriginal,
425 Cherokee,
426 Common,
427 Coptic,
428 Cuneiform,
429 Cypriot,
430 Cyrillic,
431 Deseret,
432 Devanagari,
433 Ethiopic,
434 Georgian,
435 Glagolitic,
436 Gothic,
437 Greek,
438 Gujarati,
439 Gurmukhi,
440 Han,
441 Hangul,
442 Hanunoo,
443 Hebrew,
444 Hiragana,
445 Inherited,
446 Kannada,
447 Katakana,
448 Kharoshthi,
449 Khmer,
450 Lao,
451 Latin,
452 Limbu,
453 Linear_B,
454 Malayalam,
455 Mongolian,
456 Myanmar,
457 New_Tai_Lue,
458 Nko,
459 Ogham,
460 Old_Italic,
461 Old_Persian,
462 Oriya,
463 Osmanya,
464 Phags_Pa,
465 Phoenician,
466 Runic,
467 Shavian,
468 Sinhala,
469 Syloti_Nagri,
470 Syriac,
471 Tagalog,
472 Tagbanwa,
473 Tai_Le,
474 Tamil,
475 Telugu,
476 Thaana,
477 Thai,
478 Tibetan,
479 Tifinagh,
480 Ugaritic,
481 Yi.
482 </P>
483 <P>
484 Each character has exactly one general category property, specified by a
485 two-letter abbreviation. For compatibility with Perl, negation can be specified
486 by including a circumflex between the opening brace and the property name. For
487 example, \p{^Lu} is the same as \P{Lu}.
488 </P>
489 <P>
490 If only one letter is specified with \p or \P, it includes all the general
491 category properties that start with that letter. In this case, in the absence
492 of negation, the curly brackets in the escape sequence are optional; these two
493 examples have the same effect:
494 <pre>
495 \p{L}
496 \pL
497 </pre>
498 The following general category property codes are supported:
499 <pre>
500 C Other
501 Cc Control
502 Cf Format
503 Cn Unassigned
504 Co Private use
505 Cs Surrogate
507 L Letter
508 Ll Lower case letter
509 Lm Modifier letter
510 Lo Other letter
511 Lt Title case letter
512 Lu Upper case letter
514 M Mark
515 Mc Spacing mark
516 Me Enclosing mark
517 Mn Non-spacing mark
519 N Number
520 Nd Decimal number
521 Nl Letter number
522 No Other number
524 P Punctuation
525 Pc Connector punctuation
526 Pd Dash punctuation
527 Pe Close punctuation
528 Pf Final punctuation
529 Pi Initial punctuation
530 Po Other punctuation
531 Ps Open punctuation
533 S Symbol
534 Sc Currency symbol
535 Sk Modifier symbol
536 Sm Mathematical symbol
537 So Other symbol
539 Z Separator
540 Zl Line separator
541 Zp Paragraph separator
542 Zs Space separator
543 </pre>
544 The special property L& is also supported: it matches a character that has
545 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
546 a modifier or "other".
547 </P>
548 <P>
549 The long synonyms for these properties that Perl supports (such as \p{Letter})
550 are not supported by PCRE, nor is it permitted to prefix any of these
551 properties with "Is".
552 </P>
553 <P>
554 No character that is in the Unicode table has the Cn (unassigned) property.
555 Instead, this property is assumed for any code point that is not in the
556 Unicode table.
557 </P>
558 <P>
559 Specifying caseless matching does not affect these escape sequences. For
560 example, \p{Lu} always matches only upper case letters.
561 </P>
562 <P>
563 The \X escape matches any number of Unicode characters that form an extended
564 Unicode sequence. \X is equivalent to
565 <pre>
566 (?&#62;\PM\pM*)
567 </pre>
568 That is, it matches a character without the "mark" property, followed by zero
569 or more characters with the "mark" property, and treats the sequence as an
570 atomic group
571 <a href="#atomicgroup">(see below).</a>
572 Characters with the "mark" property are typically accents that affect the
573 preceding character. None of them have codepoints less than 256, so in
574 non-UTF-8 mode \X matches any one character.
575 </P>
576 <P>
577 Matching characters by Unicode property is not fast, because PCRE has to search
578 a structure that contains data for over fifteen thousand characters. That is
579 why the traditional escape sequences such as \d and \w do not use Unicode
580 properties in PCRE.
581 <a name="resetmatchstart"></a></P>
582 <br><b>
583 Resetting the match start
584 </b><br>
585 <P>
586 The escape sequence \K, which is a Perl 5.10 feature, causes any previously
587 matched characters not to be included in the final matched sequence. For
588 example, the pattern:
589 <pre>
590 foo\Kbar
591 </pre>
592 matches "foobar", but reports that it has matched "bar". This feature is
593 similar to a lookbehind assertion
594 <a href="#lookbehind">(described below).</a>
595 However, in this case, the part of the subject before the real match does not
596 have to be of fixed length, as lookbehind assertions do. The use of \K does
597 not interfere with the setting of
598 <a href="#subpattern">captured substrings.</a>
599 For example, when the pattern
600 <pre>
601 (foo)\Kbar
602 </pre>
603 matches "foobar", the first substring is still set to "foo".
604 <a name="smallassertions"></a></P>
605 <br><b>
606 Simple assertions
607 </b><br>
608 <P>
609 The final use of backslash is for certain simple assertions. An assertion
610 specifies a condition that has to be met at a particular point in a match,
611 without consuming any characters from the subject string. The use of
612 subpatterns for more complicated assertions is described
613 <a href="#bigassertions">below.</a>
614 The backslashed assertions are:
615 <pre>
616 \b matches at a word boundary
617 \B matches when not at a word boundary
618 \A matches at the start of the subject
619 \Z matches at the end of the subject
620 also matches before a newline at the end of the subject
621 \z matches only at the end of the subject
622 \G matches at the first matching position in the subject
623 </pre>
624 These assertions may not appear in character classes (but note that \b has a
625 different meaning, namely the backspace character, inside a character class).
626 </P>
627 <P>
628 A word boundary is a position in the subject string where the current character
629 and the previous character do not both match \w or \W (i.e. one matches
630 \w and the other matches \W), or the start or end of the string if the
631 first or last character matches \w, respectively.
632 </P>
633 <P>
634 The \A, \Z, and \z assertions differ from the traditional circumflex and
635 dollar (described in the next section) in that they only ever match at the very
636 start and end of the subject string, whatever options are set. Thus, they are
637 independent of multiline mode. These three assertions are not affected by the
638 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
639 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
640 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
641 at a point other than the beginning of the subject, \A can never match. The
642 difference between \Z and \z is that \Z matches before a newline at the end
643 of the string as well as at the very end, whereas \z matches only at the end.
644 </P>
645 <P>
646 The \G assertion is true only when the current matching position is at the
647 start point of the match, as specified by the <i>startoffset</i> argument of
648 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
649 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
650 arguments, you can mimic Perl's /g option, and it is in this kind of
651 implementation where \G can be useful.
652 </P>
653 <P>
654 Note, however, that PCRE's interpretation of \G, as the start of the current
655 match, is subtly different from Perl's, which defines it as the end of the
656 previous match. In Perl, these can be different when the previously matched
657 string was empty. Because PCRE does just one match at a time, it cannot
658 reproduce this behaviour.
659 </P>
660 <P>
661 If all the alternatives of a pattern begin with \G, the expression is anchored
662 to the starting match position, and the "anchored" flag is set in the compiled
663 regular expression.
664 </P>
665 <br><a name="SEC4" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
666 <P>
667 Outside a character class, in the default matching mode, the circumflex
668 character is an assertion that is true only if the current matching point is
669 at the start of the subject string. If the <i>startoffset</i> argument of
670 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
671 option is unset. Inside a character class, circumflex has an entirely different
672 meaning
673 <a href="#characterclass">(see below).</a>
674 </P>
675 <P>
676 Circumflex need not be the first character of the pattern if a number of
677 alternatives are involved, but it should be the first thing in each alternative
678 in which it appears if the pattern is ever to match that branch. If all
679 possible alternatives start with a circumflex, that is, if the pattern is
680 constrained to match only at the start of the subject, it is said to be an
681 "anchored" pattern. (There are also other constructs that can cause a pattern
682 to be anchored.)
683 </P>
684 <P>
685 A dollar character is an assertion that is true only if the current matching
686 point is at the end of the subject string, or immediately before a newline
687 at the end of the string (by default). Dollar need not be the last character of
688 the pattern if a number of alternatives are involved, but it should be the last
689 item in any branch in which it appears. Dollar has no special meaning in a
690 character class.
691 </P>
692 <P>
693 The meaning of dollar can be changed so that it matches only at the very end of
694 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
695 does not affect the \Z assertion.
696 </P>
697 <P>
698 The meanings of the circumflex and dollar characters are changed if the
699 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
700 immediately after internal newlines as well as at the start of the subject
701 string. It does not match after a newline that ends the string. A dollar
702 matches before any newlines in the string, as well as at the very end, when
703 PCRE_MULTILINE is set. When newline is specified as the two-character
704 sequence CRLF, isolated CR and LF characters do not indicate newlines.
705 </P>
706 <P>
707 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
708 \n represents a newline) in multiline mode, but not otherwise. Consequently,
709 patterns that are anchored in single line mode because all branches start with
710 ^ are not anchored in multiline mode, and a match for circumflex is possible
711 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
712 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
713 </P>
714 <P>
715 Note that the sequences \A, \Z, and \z can be used to match the start and
716 end of the subject in both modes, and if all branches of a pattern start with
717 \A it is always anchored, whether or not PCRE_MULTILINE is set.
718 </P>
719 <br><a name="SEC5" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
720 <P>
721 Outside a character class, a dot in the pattern matches any one character in
722 the subject string except (by default) a character that signifies the end of a
723 line. In UTF-8 mode, the matched character may be more than one byte long.
724 </P>
725 <P>
726 When a line ending is defined as a single character, dot never matches that
727 character; when the two-character sequence CRLF is used, dot does not match CR
728 if it is immediately followed by LF, but otherwise it matches all characters
729 (including isolated CRs and LFs). When any Unicode line endings are being
730 recognized, dot does not match CR or LF or any of the other line ending
731 characters.
732 </P>
733 <P>
734 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
735 option is set, a dot matches any one character, without exception. If the
736 two-character sequence CRLF is present in the subject string, it takes two dots
737 to match it.
738 </P>
739 <P>
740 The handling of dot is entirely independent of the handling of circumflex and
741 dollar, the only relationship being that they both involve newlines. Dot has no
742 special meaning in a character class.
743 </P>
744 <br><a name="SEC6" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
745 <P>
746 Outside a character class, the escape sequence \C matches any one byte, both
747 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
748 characters. The feature is provided in Perl in order to match individual bytes
749 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
750 what remains in the string may be a malformed UTF-8 string. For this reason,
751 the \C escape sequence is best avoided.
752 </P>
753 <P>
754 PCRE does not allow \C to appear in lookbehind assertions
755 <a href="#lookbehind">(described below),</a>
756 because in UTF-8 mode this would make it impossible to calculate the length of
757 the lookbehind.
758 <a name="characterclass"></a></P>
759 <br><a name="SEC7" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
760 <P>
761 An opening square bracket introduces a character class, terminated by a closing
762 square bracket. A closing square bracket on its own is not special. If a
763 closing square bracket is required as a member of the class, it should be the
764 first data character in the class (after an initial circumflex, if present) or
765 escaped with a backslash.
766 </P>
767 <P>
768 A character class matches a single character in the subject. In UTF-8 mode, the
769 character may occupy more than one byte. A matched character must be in the set
770 of characters defined by the class, unless the first character in the class
771 definition is a circumflex, in which case the subject character must not be in
772 the set defined by the class. If a circumflex is actually required as a member
773 of the class, ensure it is not the first character, or escape it with a
774 backslash.
775 </P>
776 <P>
777 For example, the character class [aeiou] matches any lower case vowel, while
778 [^aeiou] matches any character that is not a lower case vowel. Note that a
779 circumflex is just a convenient notation for specifying the characters that
780 are in the class by enumerating those that are not. A class that starts with a
781 circumflex is not an assertion: it still consumes a character from the subject
782 string, and therefore it fails if the current pointer is at the end of the
783 string.
784 </P>
785 <P>
786 In UTF-8 mode, characters with values greater than 255 can be included in a
787 class as a literal string of bytes, or by using the \x{ escaping mechanism.
788 </P>
789 <P>
790 When caseless matching is set, any letters in a class represent both their
791 upper case and lower case versions, so for example, a caseless [aeiou] matches
792 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
793 caseful version would. In UTF-8 mode, PCRE always understands the concept of
794 case for characters whose values are less than 128, so caseless matching is
795 always possible. For characters with higher values, the concept of case is
796 supported if PCRE is compiled with Unicode property support, but not otherwise.
797 If you want to use caseless matching for characters 128 and above, you must
798 ensure that PCRE is compiled with Unicode property support as well as with
799 UTF-8 support.
800 </P>
801 <P>
802 Characters that might indicate line breaks are never treated in any special way
803 when matching character classes, whatever line-ending sequence is in use, and
804 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
805 such as [^a] always matches one of these characters.
806 </P>
807 <P>
808 The minus (hyphen) character can be used to specify a range of characters in a
809 character class. For example, [d-m] matches any letter between d and m,
810 inclusive. If a minus character is required in a class, it must be escaped with
811 a backslash or appear in a position where it cannot be interpreted as
812 indicating a range, typically as the first or last character in the class.
813 </P>
814 <P>
815 It is not possible to have the literal character "]" as the end character of a
816 range. A pattern such as [W-]46] is interpreted as a class of two characters
817 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
818 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
819 the end of range, so [W-\]46] is interpreted as a class containing a range
820 followed by two other characters. The octal or hexadecimal representation of
821 "]" can also be used to end a range.
822 </P>
823 <P>
824 Ranges operate in the collating sequence of character values. They can also be
825 used for characters specified numerically, for example [\000-\037]. In UTF-8
826 mode, ranges can include characters whose values are greater than 255, for
827 example [\x{100}-\x{2ff}].
828 </P>
829 <P>
830 If a range that includes letters is used when caseless matching is set, it
831 matches the letters in either case. For example, [W-c] is equivalent to
832 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
833 tables for a French locale are in use, [\xc8-\xcb] matches accented E
834 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
835 characters with values greater than 128 only when it is compiled with Unicode
836 property support.
837 </P>
838 <P>
839 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
840 in a character class, and add the characters that they match to the class. For
841 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
842 conveniently be used with the upper case character types to specify a more
843 restricted set of characters than the matching lower case type. For example,
844 the class [^\W_] matches any letter or digit, but not underscore.
845 </P>
846 <P>
847 The only metacharacters that are recognized in character classes are backslash,
848 hyphen (only where it can be interpreted as specifying a range), circumflex
849 (only at the start), opening square bracket (only when it can be interpreted as
850 introducing a POSIX class name - see the next section), and the terminating
851 closing square bracket. However, escaping other non-alphanumeric characters
852 does no harm.
853 </P>
854 <br><a name="SEC8" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
855 <P>
856 Perl supports the POSIX notation for character classes. This uses names
857 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
858 this notation. For example,
859 <pre>
860 [01[:alpha:]%]
861 </pre>
862 matches "0", "1", any alphabetic character, or "%". The supported class names
863 are
864 <pre>
865 alnum letters and digits
866 alpha letters
867 ascii character codes 0 - 127
868 blank space or tab only
869 cntrl control characters
870 digit decimal digits (same as \d)
871 graph printing characters, excluding space
872 lower lower case letters
873 print printing characters, including space
874 punct printing characters, excluding letters and digits
875 space white space (not quite the same as \s)
876 upper upper case letters
877 word "word" characters (same as \w)
878 xdigit hexadecimal digits
879 </pre>
880 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
881 space (32). Notice that this list includes the VT character (code 11). This
882 makes "space" different to \s, which does not include VT (for Perl
883 compatibility).
884 </P>
885 <P>
886 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
887 5.8. Another Perl extension is negation, which is indicated by a ^ character
888 after the colon. For example,
889 <pre>
890 [12[:^digit:]]
891 </pre>
892 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
893 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
894 supported, and an error is given if they are encountered.
895 </P>
896 <P>
897 In UTF-8 mode, characters with values greater than 128 do not match any of
898 the POSIX character classes.
899 </P>
900 <br><a name="SEC9" href="#TOC1">VERTICAL BAR</a><br>
901 <P>
902 Vertical bar characters are used to separate alternative patterns. For example,
903 the pattern
904 <pre>
905 gilbert|sullivan
906 </pre>
907 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
908 and an empty alternative is permitted (matching the empty string). The matching
909 process tries each alternative in turn, from left to right, and the first one
910 that succeeds is used. If the alternatives are within a subpattern
911 <a href="#subpattern">(defined below),</a>
912 "succeeds" means matching the rest of the main pattern as well as the
913 alternative in the subpattern.
914 </P>
915 <br><a name="SEC10" href="#TOC1">INTERNAL OPTION SETTING</a><br>
916 <P>
918 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
919 Perl option letters enclosed between "(?" and ")". The option letters are
920 <pre>
923 s for PCRE_DOTALL
925 </pre>
926 For example, (?im) sets caseless, multiline matching. It is also possible to
927 unset these options by preceding the letter with a hyphen, and a combined
928 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
929 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
930 permitted. If a letter appears both before and after the hyphen, the option is
931 unset.
932 </P>
933 <P>
934 When an option change occurs at top level (that is, not inside subpattern
935 parentheses), the change applies to the remainder of the pattern that follows.
936 If the change is placed right at the start of a pattern, PCRE extracts it into
937 the global options (and it will therefore show up in data extracted by the
938 <b>pcre_fullinfo()</b> function).
939 </P>
940 <P>
941 An option change within a subpattern (see below for a description of
942 subpatterns) affects only that part of the current pattern that follows it, so
943 <pre>
944 (a(?i)b)c
945 </pre>
946 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
947 By this means, options can be made to have different settings in different
948 parts of the pattern. Any changes made in one alternative do carry on
949 into subsequent branches within the same subpattern. For example,
950 <pre>
951 (a(?i)b|c)
952 </pre>
953 matches "ab", "aB", "c", and "C", even though when matching "C" the first
954 branch is abandoned before the option setting. This is because the effects of
955 option settings happen at compile time. There would be some very weird
956 behaviour otherwise.
957 </P>
958 <P>
959 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
960 changed in the same way as the Perl-compatible options by using the characters
961 J, U and X respectively.
962 <a name="subpattern"></a></P>
963 <br><a name="SEC11" href="#TOC1">SUBPATTERNS</a><br>
964 <P>
965 Subpatterns are delimited by parentheses (round brackets), which can be nested.
966 Turning part of a pattern into a subpattern does two things:
967 <br>
968 <br>
969 1. It localizes a set of alternatives. For example, the pattern
970 <pre>
971 cat(aract|erpillar|)
972 </pre>
973 matches one of the words "cat", "cataract", or "caterpillar". Without the
974 parentheses, it would match "cataract", "erpillar" or an empty string.
975 <br>
976 <br>
977 2. It sets up the subpattern as a capturing subpattern. This means that, when
978 the whole pattern matches, that portion of the subject string that matched the
979 subpattern is passed back to the caller via the <i>ovector</i> argument of
980 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
981 from 1) to obtain numbers for the capturing subpatterns.
982 </P>
983 <P>
984 For example, if the string "the red king" is matched against the pattern
985 <pre>
986 the ((red|white) (king|queen))
987 </pre>
988 the captured substrings are "red king", "red", and "king", and are numbered 1,
989 2, and 3, respectively.
990 </P>
991 <P>
992 The fact that plain parentheses fulfil two functions is not always helpful.
993 There are often times when a grouping subpattern is required without a
994 capturing requirement. If an opening parenthesis is followed by a question mark
995 and a colon, the subpattern does not do any capturing, and is not counted when
996 computing the number of any subsequent capturing subpatterns. For example, if
997 the string "the white queen" is matched against the pattern
998 <pre>
999 the ((?:red|white) (king|queen))
1000 </pre>
1001 the captured substrings are "white queen" and "queen", and are numbered 1 and
1002 2. The maximum number of capturing subpatterns is 65535.
1003 </P>
1004 <P>
1005 As a convenient shorthand, if any option settings are required at the start of
1006 a non-capturing subpattern, the option letters may appear between the "?" and
1007 the ":". Thus the two patterns
1008 <pre>
1009 (?i:saturday|sunday)
1010 (?:(?i)saturday|sunday)
1011 </pre>
1012 match exactly the same set of strings. Because alternative branches are tried
1013 from left to right, and options are not reset until the end of the subpattern
1014 is reached, an option setting in one branch does affect subsequent branches, so
1015 the above patterns match "SUNDAY" as well as "Saturday".
1016 </P>
1017 <br><a name="SEC12" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1018 <P>
1019 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1020 the same numbers for its capturing parentheses. Such a subpattern starts with
1021 (?| and is itself a non-capturing subpattern. For example, consider this
1022 pattern:
1023 <pre>
1024 (?|(Sat)ur|(Sun))day
1025 </pre>
1026 Because the two alternatives are inside a (?| group, both sets of capturing
1027 parentheses are numbered one. Thus, when the pattern matches, you can look
1028 at captured substring number one, whichever alternative matched. This construct
1029 is useful when you want to capture part, but not all, of one of a number of
1030 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1031 number is reset at the start of each branch. The numbers of any capturing
1032 buffers that follow the subpattern start after the highest number used in any
1033 branch. The following example is taken from the Perl documentation.
1034 The numbers underneath show in which buffer the captured content will be
1035 stored.
1036 <pre>
1037 # before ---------------branch-reset----------- after
1038 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1039 # 1 2 2 3 2 3 4
1040 </pre>
1041 A backreference or a recursive call to a numbered subpattern always refers to
1042 the first one in the pattern with the given number.
1043 </P>
1044 <P>
1045 An alternative approach to using this "branch reset" feature is to use
1046 duplicate named subpatterns, as described in the next section.
1047 </P>
1048 <br><a name="SEC13" href="#TOC1">NAMED SUBPATTERNS</a><br>
1049 <P>
1050 Identifying capturing parentheses by number is simple, but it can be very hard
1051 to keep track of the numbers in complicated regular expressions. Furthermore,
1052 if an expression is modified, the numbers may change. To help with this
1053 difficulty, PCRE supports the naming of subpatterns. This feature was not
1054 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1055 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1056 the Perl and the Python syntax.
1057 </P>
1058 <P>
1059 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1060 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1061 parentheses from other parts of the pattern, such as
1062 <a href="#backreferences">backreferences,</a>
1063 <a href="#recursion">recursion,</a>
1064 and
1065 <a href="#conditions">conditions,</a>
1066 can be made by name as well as by number.
1067 </P>
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 <P>
1076 By default, a name must be unique within a pattern, but it is possible to relax
1077 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1078 be useful for patterns where only one instance of the named parentheses can
1079 match. Suppose you want to match the name of a weekday, either as a 3-letter
1080 abbreviation or as the full name, and in both cases you want to extract the
1081 abbreviation. This pattern (ignoring the line breaks) does the job:
1082 <pre>
1083 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1084 (?&#60;DN&#62;Tue)(?:sday)?|
1085 (?&#60;DN&#62;Wed)(?:nesday)?|
1086 (?&#60;DN&#62;Thu)(?:rsday)?|
1087 (?&#60;DN&#62;Sat)(?:urday)?
1088 </pre>
1089 There are five capturing substrings, but only one is ever set after a match.
1090 (An alternative way of solving this problem is to use a "branch reset"
1091 subpattern, as described in the previous section.)
1092 </P>
1093 <P>
1094 The convenience function for extracting the data by name returns the substring
1095 for the first (and in this example, the only) subpattern of that name that
1096 matched. This saves searching to find which numbered subpattern it was. If you
1097 make a reference to a non-unique named subpattern from elsewhere in the
1098 pattern, the one that corresponds to the lowest number is used. For further
1099 details of the interfaces for handling named subpatterns, see the
1100 <a href="pcreapi.html"><b>pcreapi</b></a>
1101 documentation.
1102 </P>
1103 <br><a name="SEC14" href="#TOC1">REPETITION</a><br>
1104 <P>
1105 Repetition is specified by quantifiers, which can follow any of the following
1106 items:
1107 <pre>
1108 a literal data character
1109 the dot metacharacter
1110 the \C escape sequence
1111 the \X escape sequence (in UTF-8 mode with Unicode properties)
1112 the \R escape sequence
1113 an escape such as \d that matches a single character
1114 a character class
1115 a back reference (see next section)
1116 a parenthesized subpattern (unless it is an assertion)
1117 </pre>
1118 The general repetition quantifier specifies a minimum and maximum number of
1119 permitted matches, by giving the two numbers in curly brackets (braces),
1120 separated by a comma. The numbers must be less than 65536, and the first must
1121 be less than or equal to the second. For example:
1122 <pre>
1123 z{2,4}
1124 </pre>
1125 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1126 character. If the second number is omitted, but the comma is present, there is
1127 no upper limit; if the second number and the comma are both omitted, the
1128 quantifier specifies an exact number of required matches. Thus
1129 <pre>
1130 [aeiou]{3,}
1131 </pre>
1132 matches at least 3 successive vowels, but may match many more, while
1133 <pre>
1134 \d{8}
1135 </pre>
1136 matches exactly 8 digits. An opening curly bracket that appears in a position
1137 where a quantifier is not allowed, or one that does not match the syntax of a
1138 quantifier, is taken as a literal character. For example, {,6} is not a
1139 quantifier, but a literal string of four characters.
1140 </P>
1141 <P>
1142 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1143 bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
1144 which is represented by a two-byte sequence. Similarly, when Unicode property
1145 support is available, \X{3} matches three Unicode extended sequences, each of
1146 which may be several bytes long (and they may be of different lengths).
1147 </P>
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 <P>
1153 For convenience, the three most common quantifiers have single-character
1154 abbreviations:
1155 <pre>
1156 * is equivalent to {0,}
1157 + is equivalent to {1,}
1158 ? is equivalent to {0,1}
1159 </pre>
1160 It is possible to construct infinite loops by following a subpattern that can
1161 match no characters with a quantifier that has no upper limit, for example:
1162 <pre>
1163 (a?)*
1164 </pre>
1165 Earlier versions of Perl and PCRE used to give an error at compile time for
1166 such patterns. However, because there are cases where this can be useful, such
1167 patterns are now accepted, but if any repetition of the subpattern does in fact
1168 match no characters, the loop is forcibly broken.
1169 </P>
1170 <P>
1171 By default, the quantifiers are "greedy", that is, they match as much as
1172 possible (up to the maximum number of permitted times), without causing the
1173 rest of the pattern to fail. The classic example of where this gives problems
1174 is in trying to match comments in C programs. These appear between /* and */
1175 and within the comment, individual * and / characters may appear. An attempt to
1176 match C comments by applying the pattern
1177 <pre>
1178 /\*.*\*/
1179 </pre>
1180 to the string
1181 <pre>
1182 /* first comment */ not comment /* second comment */
1183 </pre>
1184 fails, because it matches the entire string owing to the greediness of the .*
1185 item.
1186 </P>
1187 <P>
1188 However, if a quantifier is followed by a question mark, it ceases to be
1189 greedy, and instead matches the minimum number of times possible, so the
1190 pattern
1191 <pre>
1192 /\*.*?\*/
1193 </pre>
1194 does the right thing with the C comments. The meaning of the various
1195 quantifiers is not otherwise changed, just the preferred number of matches.
1196 Do not confuse this use of question mark with its use as a quantifier in its
1197 own right. Because it has two uses, it can sometimes appear doubled, as in
1198 <pre>
1199 \d??\d
1200 </pre>
1201 which matches one digit by preference, but can match two if that is the only
1202 way the rest of the pattern matches.
1203 </P>
1204 <P>
1205 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1206 the quantifiers are not greedy by default, but individual ones can be made
1207 greedy by following them with a question mark. In other words, it inverts the
1208 default behaviour.
1209 </P>
1210 <P>
1211 When a parenthesized subpattern is quantified with a minimum repeat count that
1212 is greater than 1 or with a limited maximum, more memory is required for the
1213 compiled pattern, in proportion to the size of the minimum or maximum.
1214 </P>
1215 <P>
1216 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1217 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1218 implicitly anchored, because whatever follows will be tried against every
1219 character position in the subject string, so there is no point in retrying the
1220 overall match at any position after the first. PCRE normally treats such a
1221 pattern as though it were preceded by \A.
1222 </P>
1223 <P>
1224 In cases where it is known that the subject string contains no newlines, it is
1225 worth setting PCRE_DOTALL in order to obtain this optimization, or
1226 alternatively using ^ to indicate anchoring explicitly.
1227 </P>
1228 <P>
1229 However, there is one situation where the optimization cannot be used. When .*
1230 is inside capturing parentheses that are the subject of a backreference
1231 elsewhere in the pattern, a match at the start may fail where a later one
1232 succeeds. Consider, for example:
1233 <pre>
1234 (.*)abc\1
1235 </pre>
1236 If the subject is "xyz123abc123" the match point is the fourth character. For
1237 this reason, such a pattern is not implicitly anchored.
1238 </P>
1239 <P>
1240 When a capturing subpattern is repeated, the value captured is the substring
1241 that matched the final iteration. For example, after
1242 <pre>
1243 (tweedle[dume]{3}\s*)+
1244 </pre>
1245 has matched "tweedledum tweedledee" the value of the captured substring is
1246 "tweedledee". However, if there are nested capturing subpatterns, the
1247 corresponding captured values may have been set in previous iterations. For
1248 example, after
1249 <pre>
1250 /(a|(b))+/
1251 </pre>
1252 matches "aba" the value of the second captured substring is "b".
1253 <a name="atomicgroup"></a></P>
1254 <br><a name="SEC15" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1255 <P>
1256 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1257 repetition, failure of what follows normally causes the repeated item to be
1258 re-evaluated to see if a different number of repeats allows the rest of the
1259 pattern to match. Sometimes it is useful to prevent this, either to change the
1260 nature of the match, or to cause it fail earlier than it otherwise might, when
1261 the author of the pattern knows there is no point in carrying on.
1262 </P>
1263 <P>
1264 Consider, for example, the pattern \d+foo when applied to the subject line
1265 <pre>
1266 123456bar
1267 </pre>
1268 After matching all 6 digits and then failing to match "foo", the normal
1269 action of the matcher is to try again with only 5 digits matching the \d+
1270 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1271 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1272 that once a subpattern has matched, it is not to be re-evaluated in this way.
1273 </P>
1274 <P>
1275 If we use atomic grouping for the previous example, the matcher gives up
1276 immediately on failing to match "foo" the first time. The notation is a kind of
1277 special parenthesis, starting with (?&#62; as in this example:
1278 <pre>
1279 (?&#62;\d+)foo
1280 </pre>
1281 This kind of parenthesis "locks up" the part of the pattern it contains once
1282 it has matched, and a failure further into the pattern is prevented from
1283 backtracking into it. Backtracking past it to previous items, however, works as
1284 normal.
1285 </P>
1286 <P>
1287 An alternative description is that a subpattern of this type matches the string
1288 of characters that an identical standalone pattern would match, if anchored at
1289 the current point in the subject string.
1290 </P>
1291 <P>
1292 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1293 the above example can be thought of as a maximizing repeat that must swallow
1294 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1295 number of digits they match in order to make the rest of the pattern match,
1296 (?&#62;\d+) can only match an entire sequence of digits.
1297 </P>
1298 <P>
1299 Atomic groups in general can of course contain arbitrarily complicated
1300 subpatterns, and can be nested. However, when the subpattern for an atomic
1301 group is just a single repeated item, as in the example above, a simpler
1302 notation, called a "possessive quantifier" can be used. This consists of an
1303 additional + character following a quantifier. Using this notation, the
1304 previous example can be rewritten as
1305 <pre>
1306 \d++foo
1307 </pre>
1308 Note that a possessive quantifier can be used with an entire group, for
1309 example:
1310 <pre>
1311 (abc|xyz){2,3}+
1312 </pre>
1313 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1314 option is ignored. They are a convenient notation for the simpler forms of
1315 atomic group. However, there is no difference in the meaning of a possessive
1316 quantifier and the equivalent atomic group, though there may be a performance
1317 difference; possessive quantifiers should be slightly faster.
1318 </P>
1319 <P>
1320 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1321 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1322 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1323 package, and PCRE copied it from there. It ultimately found its way into Perl
1324 at release 5.10.
1325 </P>
1326 <P>
1327 PCRE has an optimization that automatically "possessifies" certain simple
1328 pattern constructs. For example, the sequence A+B is treated as A++B because
1329 there is no point in backtracking into a sequence of A's when B must follow.
1330 </P>
1331 <P>
1332 When a pattern contains an unlimited repeat inside a subpattern that can itself
1333 be repeated an unlimited number of times, the use of an atomic group is the
1334 only way to avoid some failing matches taking a very long time indeed. The
1335 pattern
1336 <pre>
1337 (\D+|&#60;\d+&#62;)*[!?]
1338 </pre>
1339 matches an unlimited number of substrings that either consist of non-digits, or
1340 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1341 quickly. However, if it is applied to
1342 <pre>
1343 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1344 </pre>
1345 it takes a long time before reporting failure. This is because the string can
1346 be divided between the internal \D+ repeat and the external * repeat in a
1347 large number of ways, and all have to be tried. (The example uses [!?] rather
1348 than a single character at the end, because both PCRE and Perl have an
1349 optimization that allows for fast failure when a single character is used. They
1350 remember the last single character that is required for a match, and fail early
1351 if it is not present in the string.) If the pattern is changed so that it uses
1352 an atomic group, like this:
1353 <pre>
1354 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1355 </pre>
1356 sequences of non-digits cannot be broken, and failure happens quickly.
1357 <a name="backreferences"></a></P>
1358 <br><a name="SEC16" href="#TOC1">BACK REFERENCES</a><br>
1359 <P>
1360 Outside a character class, a backslash followed by a digit greater than 0 (and
1361 possibly further digits) is a back reference to a capturing subpattern earlier
1362 (that is, to its left) in the pattern, provided there have been that many
1363 previous capturing left parentheses.
1364 </P>
1365 <P>
1366 However, if the decimal number following the backslash is less than 10, it is
1367 always taken as a back reference, and causes an error only if there are not
1368 that many capturing left parentheses in the entire pattern. In other words, the
1369 parentheses that are referenced need not be to the left of the reference for
1370 numbers less than 10. A "forward back reference" of this type can make sense
1371 when a repetition is involved and the subpattern to the right has participated
1372 in an earlier iteration.
1373 </P>
1374 <P>
1375 It is not possible to have a numerical "forward back reference" to a subpattern
1376 whose number is 10 or more using this syntax because a sequence such as \50 is
1377 interpreted as a character defined in octal. See the subsection entitled
1378 "Non-printing characters"
1379 <a href="#digitsafterbackslash">above</a>
1380 for further details of the handling of digits following a backslash. There is
1381 no such problem when named parentheses are used. A back reference to any
1382 subpattern is possible using named parentheses (see below).
1383 </P>
1384 <P>
1385 Another way of avoiding the ambiguity inherent in the use of digits following a
1386 backslash is to use the \g escape sequence, which is a feature introduced in
1387 Perl 5.10. This escape must be followed by an unsigned number or a negative
1388 number, optionally enclosed in braces. These examples are all identical:
1389 <pre>
1390 (ring), \1
1391 (ring), \g1
1392 (ring), \g{1}
1393 </pre>
1394 An unsigned number specifies an absolute reference without the ambiguity that
1395 is present in the older syntax. It is also useful when literal digits follow
1396 the reference. A negative number is a relative reference. Consider this
1397 example:
1398 <pre>
1399 (abc(def)ghi)\g{-1}
1400 </pre>
1401 The sequence \g{-1} is a reference to the most recently started capturing
1402 subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2}
1403 would be equivalent to \1. The use of relative references can be helpful in
1404 long patterns, and also in patterns that are created by joining together
1405 fragments that contain references within themselves.
1406 </P>
1407 <P>
1408 A back reference matches whatever actually matched the capturing subpattern in
1409 the current subject string, rather than anything matching the subpattern
1410 itself (see
1411 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1412 below for a way of doing that). So the pattern
1413 <pre>
1414 (sens|respons)e and \1ibility
1415 </pre>
1416 matches "sense and sensibility" and "response and responsibility", but not
1417 "sense and responsibility". If caseful matching is in force at the time of the
1418 back reference, the case of letters is relevant. For example,
1419 <pre>
1420 ((?i)rah)\s+\1
1421 </pre>
1422 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1423 capturing subpattern is matched caselessly.
1424 </P>
1425 <P>
1426 There are several different ways of writing back references to named
1427 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1428 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1429 back reference syntax, in which \g can be used for both numeric and named
1430 references, is also supported. We could rewrite the above example in any of
1431 the following ways:
1432 <pre>
1433 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1434 (?'p1'(?i)rah)\s+\k{p1}
1435 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1436 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1437 </pre>
1438 A subpattern that is referenced by name may appear in the pattern before or
1439 after the reference.
1440 </P>
1441 <P>
1442 There may be more than one back reference to the same subpattern. If a
1443 subpattern has not actually been used in a particular match, any back
1444 references to it always fail. For example, the pattern
1445 <pre>
1446 (a|(bc))\2
1447 </pre>
1448 always fails if it starts to match "a" rather than "bc". Because there may be
1449 many capturing parentheses in a pattern, all digits following the backslash are
1450 taken as part of a potential back reference number. If the pattern continues
1451 with a digit character, some delimiter must be used to terminate the back
1452 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1453 Otherwise an empty comment (see
1454 <a href="#comments">"Comments"</a>
1455 below) can be used.
1456 </P>
1457 <P>
1458 A back reference that occurs inside the parentheses to which it refers fails
1459 when the subpattern is first used, so, for example, (a\1) never matches.
1460 However, such references can be useful inside repeated subpatterns. For
1461 example, the pattern
1462 <pre>
1463 (a|b\1)+
1464 </pre>
1465 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1466 the subpattern, the back reference matches the character string corresponding
1467 to the previous iteration. In order for this to work, the pattern must be such
1468 that the first iteration does not need to match the back reference. This can be
1469 done using alternation, as in the example above, or by a quantifier with a
1470 minimum of zero.
1471 <a name="bigassertions"></a></P>
1472 <br><a name="SEC17" href="#TOC1">ASSERTIONS</a><br>
1473 <P>
1474 An assertion is a test on the characters following or preceding the current
1475 matching point that does not actually consume any characters. The simple
1476 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1477 <a href="#smallassertions">above.</a>
1478 </P>
1479 <P>
1480 More complicated assertions are coded as subpatterns. There are two kinds:
1481 those that look ahead of the current position in the subject string, and those
1482 that look behind it. An assertion subpattern is matched in the normal way,
1483 except that it does not cause the current matching position to be changed.
1484 </P>
1485 <P>
1486 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1487 because it makes no sense to assert the same thing several times. If any kind
1488 of assertion contains capturing subpatterns within it, these are counted for
1489 the purposes of numbering the capturing subpatterns in the whole pattern.
1490 However, substring capturing is carried out only for positive assertions,
1491 because it does not make sense for negative assertions.
1492 </P>
1493 <br><b>
1494 Lookahead assertions
1495 </b><br>
1496 <P>
1497 Lookahead assertions start with (?= for positive assertions and (?! for
1498 negative assertions. For example,
1499 <pre>
1500 \w+(?=;)
1501 </pre>
1502 matches a word followed by a semicolon, but does not include the semicolon in
1503 the match, and
1504 <pre>
1505 foo(?!bar)
1506 </pre>
1507 matches any occurrence of "foo" that is not followed by "bar". Note that the
1508 apparently similar pattern
1509 <pre>
1510 (?!foo)bar
1511 </pre>
1512 does not find an occurrence of "bar" that is preceded by something other than
1513 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1514 (?!foo) is always true when the next three characters are "bar". A
1515 lookbehind assertion is needed to achieve the other effect.
1516 </P>
1517 <P>
1518 If you want to force a matching failure at some point in a pattern, the most
1519 convenient way to do it is with (?!) because an empty string always matches, so
1520 an assertion that requires there not to be an empty string must always fail.
1521 <a name="lookbehind"></a></P>
1522 <br><b>
1523 Lookbehind assertions
1524 </b><br>
1525 <P>
1526 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1527 negative assertions. For example,
1528 <pre>
1529 (?&#60;!foo)bar
1530 </pre>
1531 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1532 a lookbehind assertion are restricted such that all the strings it matches must
1533 have a fixed length. However, if there are several top-level alternatives, they
1534 do not all have to have the same fixed length. Thus
1535 <pre>
1536 (?&#60;=bullock|donkey)
1537 </pre>
1538 is permitted, but
1539 <pre>
1540 (?&#60;!dogs?|cats?)
1541 </pre>
1542 causes an error at compile time. Branches that match different length strings
1543 are permitted only at the top level of a lookbehind assertion. This is an
1544 extension compared with Perl (at least for 5.8), which requires all branches to
1545 match the same length of string. An assertion such as
1546 <pre>
1547 (?&#60;=ab(c|de))
1548 </pre>
1549 is not permitted, because its single top-level branch can match two different
1550 lengths, but it is acceptable if rewritten to use two top-level branches:
1551 <pre>
1552 (?&#60;=abc|abde)
1553 </pre>
1554 In some cases, the Perl 5.10 escape sequence \K
1555 <a href="#resetmatchstart">(see above)</a>
1556 can be used instead of a lookbehind assertion; this is not restricted to a
1557 fixed-length.
1558 </P>
1559 <P>
1560 The implementation of lookbehind assertions is, for each alternative, to
1561 temporarily move the current position back by the fixed length and then try to
1562 match. If there are insufficient characters before the current position, the
1563 assertion fails.
1564 </P>
1565 <P>
1566 PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1567 to appear in lookbehind assertions, because it makes it impossible to calculate
1568 the length of the lookbehind. The \X and \R escapes, which can match
1569 different numbers of bytes, are also not permitted.
1570 </P>
1571 <P>
1572 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1573 specify efficient matching at the end of the subject string. Consider a simple
1574 pattern such as
1575 <pre>
1576 abcd$
1577 </pre>
1578 when applied to a long string that does not match. Because matching proceeds
1579 from left to right, PCRE will look for each "a" in the subject and then see if
1580 what follows matches the rest of the pattern. If the pattern is specified as
1581 <pre>
1582 ^.*abcd$
1583 </pre>
1584 the initial .* matches the entire string at first, but when this fails (because
1585 there is no following "a"), it backtracks to match all but the last character,
1586 then all but the last two characters, and so on. Once again the search for "a"
1587 covers the entire string, from right to left, so we are no better off. However,
1588 if the pattern is written as
1589 <pre>
1590 ^.*+(?&#60;=abcd)
1591 </pre>
1592 there can be no backtracking for the .*+ item; it can match only the entire
1593 string. The subsequent lookbehind assertion does a single test on the last four
1594 characters. If it fails, the match fails immediately. For long strings, this
1595 approach makes a significant difference to the processing time.
1596 </P>
1597 <br><b>
1598 Using multiple assertions
1599 </b><br>
1600 <P>
1601 Several assertions (of any sort) may occur in succession. For example,
1602 <pre>
1603 (?&#60;=\d{3})(?&#60;!999)foo
1604 </pre>
1605 matches "foo" preceded by three digits that are not "999". Notice that each of
1606 the assertions is applied independently at the same point in the subject
1607 string. First there is a check that the previous three characters are all
1608 digits, and then there is a check that the same three characters are not "999".
1609 This pattern does <i>not</i> match "foo" preceded by six characters, the first
1610 of which are digits and the last three of which are not "999". For example, it
1611 doesn't match "123abcfoo". A pattern to do that is
1612 <pre>
1613 (?&#60;=\d{3}...)(?&#60;!999)foo
1614 </pre>
1615 This time the first assertion looks at the preceding six characters, checking
1616 that the first three are digits, and then the second assertion checks that the
1617 preceding three characters are not "999".
1618 </P>
1619 <P>
1620 Assertions can be nested in any combination. For example,
1621 <pre>
1622 (?&#60;=(?&#60;!foo)bar)baz
1623 </pre>
1624 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1625 preceded by "foo", while
1626 <pre>
1627 (?&#60;=\d{3}(?!999)...)foo
1628 </pre>
1629 is another pattern that matches "foo" preceded by three digits and any three
1630 characters that are not "999".
1631 <a name="conditions"></a></P>
1632 <br><a name="SEC18" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1633 <P>
1634 It is possible to cause the matching process to obey a subpattern
1635 conditionally or to choose between two alternative subpatterns, depending on
1636 the result of an assertion, or whether a previous capturing subpattern matched
1637 or not. The two possible forms of conditional subpattern are
1638 <pre>
1639 (?(condition)yes-pattern)
1640 (?(condition)yes-pattern|no-pattern)
1641 </pre>
1642 If the condition is satisfied, the yes-pattern is used; otherwise the
1643 no-pattern (if present) is used. If there are more than two alternatives in the
1644 subpattern, a compile-time error occurs.
1645 </P>
1646 <P>
1647 There are four kinds of condition: references to subpatterns, references to
1648 recursion, a pseudo-condition called DEFINE, and assertions.
1649 </P>
1650 <br><b>
1651 Checking for a used subpattern by number
1652 </b><br>
1653 <P>
1654 If the text between the parentheses consists of a sequence of digits, the
1655 condition is true if the capturing subpattern of that number has previously
1656 matched. An alternative notation is to precede the digits with a plus or minus
1657 sign. In this case, the subpattern number is relative rather than absolute.
1658 The most recently opened parentheses can be referenced by (?(-1), the next most
1659 recent by (?(-2), and so on. In looping constructs it can also make sense to
1660 refer to subsequent groups with constructs such as (?(+2).
1661 </P>
1662 <P>
1663 Consider the following pattern, which contains non-significant white space to
1664 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1665 three parts for ease of discussion:
1666 <pre>
1667 ( \( )? [^()]+ (?(1) \) )
1668 </pre>
1669 The first part matches an optional opening parenthesis, and if that
1670 character is present, sets it as the first captured substring. The second part
1671 matches one or more characters that are not parentheses. The third part is a
1672 conditional subpattern that tests whether the first set of parentheses matched
1673 or not. If they did, that is, if subject started with an opening parenthesis,
1674 the condition is true, and so the yes-pattern is executed and a closing
1675 parenthesis is required. Otherwise, since no-pattern is not present, the
1676 subpattern matches nothing. In other words, this pattern matches a sequence of
1677 non-parentheses, optionally enclosed in parentheses.
1678 </P>
1679 <P>
1680 If you were embedding this pattern in a larger one, you could use a relative
1681 reference:
1682 <pre>
1683 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
1684 </pre>
1685 This makes the fragment independent of the parentheses in the larger pattern.
1686 </P>
1687 <br><b>
1688 Checking for a used subpattern by name
1689 </b><br>
1690 <P>
1691 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
1692 subpattern by name. For compatibility with earlier versions of PCRE, which had
1693 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1694 there is a possible ambiguity with this syntax, because subpattern names may
1695 consist entirely of digits. PCRE looks first for a named subpattern; if it
1696 cannot find one and the name consists entirely of digits, PCRE looks for a
1697 subpattern of that number, which must be greater than zero. Using subpattern
1698 names that consist entirely of digits is not recommended.
1699 </P>
1700 <P>
1701 Rewriting the above example to use a named subpattern gives this:
1702 <pre>
1703 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
1705 </PRE>
1706 </P>
1707 <br><b>
1708 Checking for pattern recursion
1709 </b><br>
1710 <P>
1711 If the condition is the string (R), and there is no subpattern with the name R,
1712 the condition is true if a recursive call to the whole pattern or any
1713 subpattern has been made. If digits or a name preceded by ampersand follow the
1714 letter R, for example:
1715 <pre>
1716 (?(R3)...) or (?(R&name)...)
1717 </pre>
1718 the condition is true if the most recent recursion is into the subpattern whose
1719 number or name is given. This condition does not check the entire recursion
1720 stack.
1721 </P>
1722 <P>
1723 At "top level", all these recursion test conditions are false. Recursive
1724 patterns are described below.
1725 </P>
1726 <br><b>
1727 Defining subpatterns for use by reference only
1728 </b><br>
1729 <P>
1730 If the condition is the string (DEFINE), and there is no subpattern with the
1731 name DEFINE, the condition is always false. In this case, there may be only one
1732 alternative in the subpattern. It is always skipped if control reaches this
1733 point in the pattern; the idea of DEFINE is that it can be used to define
1734 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1735 is described below.) For example, a pattern to match an IPv4 address could be
1736 written like this (ignore whitespace and line breaks):
1737 <pre>
1738 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
1739 \b (?&byte) (\.(?&byte)){3} \b
1740 </pre>
1741 The first part of the pattern is a DEFINE group inside which a another group
1742 named "byte" is defined. This matches an individual component of an IPv4
1743 address (a number less than 256). When matching takes place, this part of the
1744 pattern is skipped because DEFINE acts like a false condition.
1745 </P>
1746 <P>
1747 The rest of the pattern uses references to the named group to match the four
1748 dot-separated components of an IPv4 address, insisting on a word boundary at
1749 each end.
1750 </P>
1751 <br><b>
1752 Assertion conditions
1753 </b><br>
1754 <P>
1755 If the condition is not in any of the above formats, it must be an assertion.
1756 This may be a positive or negative lookahead or lookbehind assertion. Consider
1757 this pattern, again containing non-significant white space, and with the two
1758 alternatives on the second line:
1759 <pre>
1760 (?(?=[^a-z]*[a-z])
1761 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1762 </pre>
1763 The condition is a positive lookahead assertion that matches an optional
1764 sequence of non-letters followed by a letter. In other words, it tests for the
1765 presence of at least one letter in the subject. If a letter is found, the
1766 subject is matched against the first alternative; otherwise it is matched
1767 against the second. This pattern matches strings in one of the two forms
1768 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1769 <a name="comments"></a></P>
1770 <br><a name="SEC19" href="#TOC1">COMMENTS</a><br>
1771 <P>
1772 The sequence (?# marks the start of a comment that continues up to the next
1773 closing parenthesis. Nested parentheses are not permitted. The characters
1774 that make up a comment play no part in the pattern matching at all.
1775 </P>
1776 <P>
1777 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1778 character class introduces a comment that continues to immediately after the
1779 next newline in the pattern.
1780 <a name="recursion"></a></P>
1781 <br><a name="SEC20" href="#TOC1">RECURSIVE PATTERNS</a><br>
1782 <P>
1783 Consider the problem of matching a string in parentheses, allowing for
1784 unlimited nested parentheses. Without the use of recursion, the best that can
1785 be done is to use a pattern that matches up to some fixed depth of nesting. It
1786 is not possible to handle an arbitrary nesting depth.
1787 </P>
1788 <P>
1789 For some time, Perl has provided a facility that allows regular expressions to
1790 recurse (amongst other things). It does this by interpolating Perl code in the
1791 expression at run time, and the code can refer to the expression itself. A Perl
1792 pattern using code interpolation to solve the parentheses problem can be
1793 created like this:
1794 <pre>
1795 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
1796 </pre>
1797 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1798 recursively to the pattern in which it appears.
1799 </P>
1800 <P>
1801 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1802 supports special syntax for recursion of the entire pattern, and also for
1803 individual subpattern recursion. After its introduction in PCRE and Python,
1804 this kind of recursion was introduced into Perl at release 5.10.
1805 </P>
1806 <P>
1807 A special item that consists of (? followed by a number greater than zero and a
1808 closing parenthesis is a recursive call of the subpattern of the given number,
1809 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1810 call, which is described in the next section.) The special item (?R) or (?0) is
1811 a recursive call of the entire regular expression.
1812 </P>
1813 <P>
1814 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1815 treated as an atomic group. That is, once it has matched some of the subject
1816 string, it is never re-entered, even if it contains untried alternatives and
1817 there is a subsequent matching failure.
1818 </P>
1819 <P>
1820 This PCRE pattern solves the nested parentheses problem (assume the
1821 PCRE_EXTENDED option is set so that white space is ignored):
1822 <pre>
1823 \( ( (?&#62;[^()]+) | (?R) )* \)
1824 </pre>
1825 First it matches an opening parenthesis. Then it matches any number of
1826 substrings which can either be a sequence of non-parentheses, or a recursive
1827 match of the pattern itself (that is, a correctly parenthesized substring).
1828 Finally there is a closing parenthesis.
1829 </P>
1830 <P>
1831 If this were part of a larger pattern, you would not want to recurse the entire
1832 pattern, so instead you could use this:
1833 <pre>
1834 ( \( ( (?&#62;[^()]+) | (?1) )* \) )
1835 </pre>
1836 We have put the pattern into parentheses, and caused the recursion to refer to
1837 them instead of the whole pattern.
1838 </P>
1839 <P>
1840 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1841 is made easier by the use of relative references. (A Perl 5.10 feature.)
1842 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1843 most recently opened parentheses preceding the recursion. In other words, a
1844 negative number counts capturing parentheses leftwards from the point at which
1845 it is encountered.
1846 </P>
1847 <P>
1848 It is also possible to refer to subsequently opened parentheses, by writing
1849 references such as (?+2). However, these cannot be recursive because the
1850 reference is not inside the parentheses that are referenced. They are always
1851 "subroutine" calls, as described in the next section.
1852 </P>
1853 <P>
1854 An alternative approach is to use named parentheses instead. The Perl syntax
1855 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
1856 could rewrite the above example as follows:
1857 <pre>
1858 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | (?&pn) )* \) )
1859 </pre>
1860 If there is more than one subpattern with the same name, the earliest one is
1861 used.
1862 </P>
1863 <P>
1864 This particular example pattern that we have been looking at contains nested
1865 unlimited repeats, and so the use of atomic grouping for matching strings of
1866 non-parentheses is important when applying the pattern to strings that do not
1867 match. For example, when this pattern is applied to
1868 <pre>
1869 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1870 </pre>
1871 it yields "no match" quickly. However, if atomic grouping is not used,
1872 the match runs for a very long time indeed because there are so many different
1873 ways the + and * repeats can carve up the subject, and all have to be tested
1874 before failure can be reported.
1875 </P>
1876 <P>
1877 At the end of a match, the values set for any capturing subpatterns are those
1878 from the outermost level of the recursion at which the subpattern value is set.
1879 If you want to obtain intermediate values, a callout function can be used (see
1880 below and the
1881 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1882 documentation). If the pattern above is matched against
1883 <pre>
1884 (ab(cd)ef)
1885 </pre>
1886 the value for the capturing parentheses is "ef", which is the last value taken
1887 on at the top level. If additional parentheses are added, giving
1888 <pre>
1889 \( ( ( (?&#62;[^()]+) | (?R) )* ) \)
1890 ^ ^
1891 ^ ^
1892 </pre>
1893 the string they capture is "ab(cd)ef", the contents of the top level
1894 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1895 has to obtain extra memory to store data during a recursion, which it does by
1896 using <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no
1897 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1898 </P>
1899 <P>
1900 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1901 Consider this pattern, which matches text in angle brackets, allowing for
1902 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1903 recursing), whereas any characters are permitted at the outer level.
1904 <pre>
1905 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
1906 </pre>
1907 In this pattern, (?(R) is the start of a conditional subpattern, with two
1908 different alternatives for the recursive and non-recursive cases. The (?R) item
1909 is the actual recursive call.
1910 <a name="subpatternsassubroutines"></a></P>
1911 <br><a name="SEC21" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
1912 <P>
1913 If the syntax for a recursive subpattern reference (either by number or by
1914 name) is used outside the parentheses to which it refers, it operates like a
1915 subroutine in a programming language. The "called" subpattern may be defined
1916 before or after the reference. A numbered reference can be absolute or
1917 relative, as in these examples:
1918 <pre>
1919 (...(absolute)...)...(?2)...
1920 (...(relative)...)...(?-1)...
1921 (...(?+1)...(relative)...
1922 </pre>
1923 An earlier example pointed out that the pattern
1924 <pre>
1925 (sens|respons)e and \1ibility
1926 </pre>
1927 matches "sense and sensibility" and "response and responsibility", but not
1928 "sense and responsibility". If instead the pattern
1929 <pre>
1930 (sens|respons)e and (?1)ibility
1931 </pre>
1932 is used, it does match "sense and responsibility" as well as the other two
1933 strings. Another example is given in the discussion of DEFINE above.
1934 </P>
1935 <P>
1936 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1937 group. That is, once it has matched some of the subject string, it is never
1938 re-entered, even if it contains untried alternatives and there is a subsequent
1939 matching failure.
1940 </P>
1941 <P>
1942 When a subpattern is used as a subroutine, processing options such as
1943 case-independence are fixed when the subpattern is defined. They cannot be
1944 changed for different calls. For example, consider this pattern:
1945 <pre>
1946 (abc)(?i:(?-1))
1947 </pre>
1948 It matches "abcabc". It does not match "abcABC" because the change of
1949 processing option does not affect the called subpattern.
1950 </P>
1951 <br><a name="SEC22" href="#TOC1">CALLOUTS</a><br>
1952 <P>
1953 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1954 code to be obeyed in the middle of matching a regular expression. This makes it
1955 possible, amongst other things, to extract different substrings that match the
1956 same pair of parentheses when there is a repetition.
1957 </P>
1958 <P>
1959 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1960 code. The feature is called "callout". The caller of PCRE provides an external
1961 function by putting its entry point in the global variable <i>pcre_callout</i>.
1962 By default, this variable contains NULL, which disables all calling out.
1963 </P>
1964 <P>
1965 Within a regular expression, (?C) indicates the points at which the external
1966 function is to be called. If you want to identify different callout points, you
1967 can put a number less than 256 after the letter C. The default value is zero.
1968 For example, this pattern has two callout points:
1969 <pre>
1970 (?C1)abc(?C2)def
1971 </pre>
1972 If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
1973 automatically installed before each item in the pattern. They are all numbered
1974 255.
1975 </P>
1976 <P>
1977 During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
1978 set), the external function is called. It is provided with the number of the
1979 callout, the position in the pattern, and, optionally, one item of data
1980 originally supplied by the caller of <b>pcre_exec()</b>. The callout function
1981 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1982 description of the interface to the callout function is given in the
1983 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1984 documentation.
1985 </P>
1986 <br><a name="SEC23" href="#TOC1">SEE ALSO</a><br>
1987 <P>
1988 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), <b>pcre</b>(3).
1989 </P>
1990 <br><a name="SEC24" href="#TOC1">AUTHOR</a><br>
1991 <P>
1992 Philip Hazel
1993 <br>
1994 University Computing Service
1995 <br>
1996 Cambridge CB2 3QH, England.
1997 <br>
1998 </P>
1999 <br><a name="SEC25" href="#TOC1">REVISION</a><br>
2000 <P>
2001 Last updated: 06 August 2007
2002 <br>
2003 Copyright &copy; 1997-2007 University of Cambridge.
2004 <br>
2005 <p>
2006 Return to the <a href="index.html">PCRE index page</a>.
2007 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5