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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 172 - (show annotations)
Tue Jun 5 10:40:13 2007 UTC (8 years, 2 months ago) by ph10
File MIME type: text/html
File size: 80828 byte(s)
Error occurred while calculating annotation data.
Drastically reduce workspace used for alternatives in groups; also some 
trailing space removals for a test release.
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>
22 <li><a name="TOC7" href="#SEC7">SQUARE BRACKETS AND CHARACTER CLASSES</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">NAMED SUBPATTERNS</a>
28 <li><a name="TOC13" href="#SEC13">REPETITION</a>
29 <li><a name="TOC14" href="#SEC14">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a>
30 <li><a name="TOC15" href="#SEC15">BACK REFERENCES</a>
31 <li><a name="TOC16" href="#SEC16">ASSERTIONS</a>
32 <li><a name="TOC17" href="#SEC17">CONDITIONAL SUBPATTERNS</a>
33 <li><a name="TOC18" href="#SEC18">COMMENTS</a>
34 <li><a name="TOC19" href="#SEC19">RECURSIVE PATTERNS</a>
35 <li><a name="TOC20" href="#SEC20">SUBPATTERNS AS SUBROUTINES</a>
36 <li><a name="TOC21" href="#SEC21">CALLOUTS</a>
37 <li><a name="TOC22" href="#SEC22">SEE ALSO</a>
38 <li><a name="TOC23" href="#SEC23">AUTHOR</a>
39 <li><a name="TOC24" href="#SEC24">REVISION</a>
40 </ul>
41 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
42 <P>
43 The syntax and semantics of the regular expressions supported by PCRE are
44 described below. Regular expressions are also described in the Perl
45 documentation and in a number of books, some of which have copious examples.
46 Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers
47 regular expressions in great detail. This description of PCRE's regular
48 expressions is intended as reference material.
49 </P>
50 <P>
51 The original operation of PCRE was on strings of one-byte characters. However,
52 there is now also support for UTF-8 character strings. To use this, you must
53 build PCRE to include UTF-8 support, and then call <b>pcre_compile()</b> with
54 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
55 places below. There is also a summary of UTF-8 features in the
56 <a href="pcre.html#utf8support">section on UTF-8 support</a>
57 in the main
58 <a href="pcre.html"><b>pcre</b></a>
59 page.
60 </P>
61 <P>
62 The remainder of this document discusses the patterns that are supported by
63 PCRE when its main matching function, <b>pcre_exec()</b>, is used.
64 From release 6.0, PCRE offers a second matching function,
65 <b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not
66 Perl-compatible. Some of the features discussed below are not available when
67 <b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the
68 alternative function, and how it differs from the normal function, are
69 discussed in the
70 <a href="pcrematching.html"><b>pcrematching</b></a>
71 page.
72 </P>
73 <br><a name="SEC2" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
74 <P>
75 A regular expression is a pattern that is matched against a subject string from
76 left to right. Most characters stand for themselves in a pattern, and match the
77 corresponding characters in the subject. As a trivial example, the pattern
78 <pre>
79 The quick brown fox
80 </pre>
81 matches a portion of a subject string that is identical to itself. When
82 caseless matching is specified (the PCRE_CASELESS option), letters are matched
83 independently of case. In UTF-8 mode, PCRE always understands the concept of
84 case for characters whose values are less than 128, so caseless matching is
85 always possible. For characters with higher values, the concept of case is
86 supported if PCRE is compiled with Unicode property support, but not otherwise.
87 If you want to use caseless matching for characters 128 and above, you must
88 ensure that PCRE is compiled with Unicode property support as well as with
89 UTF-8 support.
90 </P>
91 <P>
92 The power of regular expressions comes from the ability to include alternatives
93 and repetitions in the pattern. These are encoded in the pattern by the use of
94 <i>metacharacters</i>, which do not stand for themselves but instead are
95 interpreted in some special way.
96 </P>
97 <P>
98 There are two different sets of metacharacters: those that are recognized
99 anywhere in the pattern except within square brackets, and those that are
100 recognized within square brackets. Outside square brackets, the metacharacters
101 are as follows:
102 <pre>
103 \ general escape character with several uses
104 ^ assert start of string (or line, in multiline mode)
105 $ assert end of string (or line, in multiline mode)
106 . match any character except newline (by default)
107 [ start character class definition
108 | start of alternative branch
109 ( start subpattern
110 ) end subpattern
111 ? extends the meaning of (
112 also 0 or 1 quantifier
113 also quantifier minimizer
114 * 0 or more quantifier
115 + 1 or more quantifier
116 also "possessive quantifier"
117 { start min/max quantifier
118 </pre>
119 Part of a pattern that is in square brackets is called a "character class". In
120 a character class the only metacharacters are:
121 <pre>
122 \ general escape character
123 ^ negate the class, but only if the first character
124 - indicates character range
125 [ POSIX character class (only if followed by POSIX syntax)
126 ] terminates the character class
127 </pre>
128 The following sections describe the use of each of the metacharacters.
129 </P>
130 <br><a name="SEC3" href="#TOC1">BACKSLASH</a><br>
131 <P>
132 The backslash character has several uses. Firstly, if it is followed by a
133 non-alphanumeric character, it takes away any special meaning that character
134 may have. This use of backslash as an escape character applies both inside and
135 outside character classes.
136 </P>
137 <P>
138 For example, if you want to match a * character, you write \* in the pattern.
139 This escaping action applies whether or not the following character would
140 otherwise be interpreted as a metacharacter, so it is always safe to precede a
141 non-alphanumeric with backslash to specify that it stands for itself. In
142 particular, if you want to match a backslash, you write \\.
143 </P>
144 <P>
145 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
146 pattern (other than in a character class) and characters between a # outside
147 a character class and the next newline are ignored. An escaping backslash can
148 be used to include a whitespace or # character as part of the pattern.
149 </P>
150 <P>
151 If you want to remove the special meaning from a sequence of characters, you
152 can do so by putting them between \Q and \E. This is different from Perl in
153 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
154 Perl, $ and @ cause variable interpolation. Note the following examples:
155 <pre>
156 Pattern PCRE matches Perl matches
157
158 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
159 \Qabc\$xyz\E abc\$xyz abc\$xyz
160 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
161 </pre>
162 The \Q...\E sequence is recognized both inside and outside character classes.
163 <a name="digitsafterbackslash"></a></P>
164 <br><b>
165 Non-printing characters
166 </b><br>
167 <P>
168 A second use of backslash provides a way of encoding non-printing characters
169 in patterns in a visible manner. There is no restriction on the appearance of
170 non-printing characters, apart from the binary zero that terminates a pattern,
171 but when a pattern is being prepared by text editing, it is usually easier to
172 use one of the following escape sequences than the binary character it
173 represents:
174 <pre>
175 \a alarm, that is, the BEL character (hex 07)
176 \cx "control-x", where x is any character
177 \e escape (hex 1B)
178 \f formfeed (hex 0C)
179 \n newline (hex 0A)
180 \r carriage return (hex 0D)
181 \t tab (hex 09)
182 \ddd character with octal code ddd, or backreference
183 \xhh character with hex code hh
184 \x{hhh..} character with hex code hhh..
185 </pre>
186 The precise effect of \cx is as follows: if x is a lower case letter, it
187 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
188 Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
189 7B.
190 </P>
191 <P>
192 After \x, from zero to two hexadecimal digits are read (letters can be in
193 upper or lower case). Any number of hexadecimal digits may appear between \x{
194 and }, but the value of the character code must be less than 256 in non-UTF-8
195 mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
196 is 7FFFFFFF). If characters other than hexadecimal digits appear between \x{
197 and }, or if there is no terminating }, this form of escape is not recognized.
198 Instead, the initial \x will be interpreted as a basic hexadecimal escape,
199 with no following digits, giving a character whose value is zero.
200 </P>
201 <P>
202 Characters whose value is less than 256 can be defined by either of the two
203 syntaxes for \x. There is no difference in the way they are handled. For
204 example, \xdc is exactly the same as \x{dc}.
205 </P>
206 <P>
207 After \0 up to two further octal digits are read. If there are fewer than two
208 digits, just those that are present are used. Thus the sequence \0\x\07
209 specifies two binary zeros followed by a BEL character (code value 7). Make
210 sure you supply two digits after the initial zero if the pattern character that
211 follows is itself an octal digit.
212 </P>
213 <P>
214 The handling of a backslash followed by a digit other than 0 is complicated.
215 Outside a character class, PCRE reads it and any following digits as a decimal
216 number. If the number is less than 10, or if there have been at least that many
217 previous capturing left parentheses in the expression, the entire sequence is
218 taken as a <i>back reference</i>. A description of how this works is given
219 <a href="#backreferences">later,</a>
220 following the discussion of
221 <a href="#subpattern">parenthesized subpatterns.</a>
222 </P>
223 <P>
224 Inside a character class, or if the decimal number is greater than 9 and there
225 have not been that many capturing subpatterns, PCRE re-reads up to three octal
226 digits following the backslash, and uses them to generate a data character. Any
227 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
228 character specified in octal must be less than \400. In UTF-8 mode, values up
229 to \777 are permitted. For example:
230 <pre>
231 \040 is another way of writing a space
232 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
233 \7 is always a back reference
234 \11 might be a back reference, or another way of writing a tab
235 \011 is always a tab
236 \0113 is a tab followed by the character "3"
237 \113 might be a back reference, otherwise the character with octal code 113
238 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
239 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
240 </pre>
241 Note that octal values of 100 or greater must not be introduced by a leading
242 zero, because no more than three octal digits are ever read.
243 </P>
244 <P>
245 All the sequences that define a single character value can be used both inside
246 and outside character classes. In addition, inside a character class, the
247 sequence \b is interpreted as the backspace character (hex 08), and the
248 sequences \R and \X are interpreted as the characters "R" and "X",
249 respectively. Outside a character class, these sequences have different
250 meanings
251 <a href="#uniextseq">(see below).</a>
252 </P>
253 <br><b>
254 Absolute and relative back references
255 </b><br>
256 <P>
257 The sequence \g followed by a positive or negative number, optionally enclosed
258 in braces, is an absolute or relative back reference. A named back reference
259 can be coded as \g{name}. Back references are discussed
260 <a href="#backreferences">later,</a>
261 following the discussion of
262 <a href="#subpattern">parenthesized subpatterns.</a>
263 </P>
264 <br><b>
265 Generic character types
266 </b><br>
267 <P>
268 Another use of backslash is for specifying generic character types. The
269 following are always recognized:
270 <pre>
271 \d any decimal digit
272 \D any character that is not a decimal digit
273 \s any whitespace character
274 \S any character that is not a whitespace character
275 \w any "word" character
276 \W any "non-word" character
277 </pre>
278 Each pair of escape sequences partitions the complete set of characters into
279 two disjoint sets. Any given character matches one, and only one, of each pair.
280 </P>
281 <P>
282 These character type sequences can appear both inside and outside character
283 classes. They each match one character of the appropriate type. If the current
284 matching point is at the end of the subject string, all of them fail, since
285 there is no character to match.
286 </P>
287 <P>
288 For compatibility with Perl, \s does not match the VT character (code 11).
289 This makes it different from the the POSIX "space" class. The \s characters
290 are HT (9), LF (10), FF (12), CR (13), and space (32). (If "use locale;" is
291 included in a Perl script, \s may match the VT character. In PCRE, it never
292 does.)
293 </P>
294 <P>
295 A "word" character is an underscore or any character less than 256 that is a
296 letter or digit. The definition of letters and digits is controlled by PCRE's
297 low-valued character tables, and may vary if locale-specific matching is taking
298 place (see
299 <a href="pcreapi.html#localesupport">"Locale support"</a>
300 in the
301 <a href="pcreapi.html"><b>pcreapi</b></a>
302 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
303 or "french" in Windows, some character codes greater than 128 are used for
304 accented letters, and these are matched by \w.
305 </P>
306 <P>
307 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
308 \w, and always match \D, \S, and \W. This is true even when Unicode
309 character property support is available. The use of locales with Unicode is
310 discouraged.
311 </P>
312 <br><b>
313 Newline sequences
314 </b><br>
315 <P>
316 Outside a character class, the escape sequence \R matches any Unicode newline
317 sequence. This is an extension to Perl. In non-UTF-8 mode \R is equivalent to
318 the following:
319 <pre>
320 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
321 </pre>
322 This is an example of an "atomic group", details of which are given
323 <a href="#atomicgroup">below.</a>
324 This particular group matches either the two-character sequence CR followed by
325 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
326 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
327 line, U+0085). The two-character sequence is treated as a single unit that
328 cannot be split.
329 </P>
330 <P>
331 In UTF-8 mode, two additional characters whose codepoints are greater than 255
332 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
333 Unicode character property support is not needed for these characters to be
334 recognized.
335 </P>
336 <P>
337 Inside a character class, \R matches the letter "R".
338 <a name="uniextseq"></a></P>
339 <br><b>
340 Unicode character properties
341 </b><br>
342 <P>
343 When PCRE is built with Unicode character property support, three additional
344 escape sequences to match character properties are available when UTF-8 mode
345 is selected. They are:
346 <pre>
347 \p{<i>xx</i>} a character with the <i>xx</i> property
348 \P{<i>xx</i>} a character without the <i>xx</i> property
349 \X an extended Unicode sequence
350 </pre>
351 The property names represented by <i>xx</i> above are limited to the Unicode
352 script names, the general category properties, and "Any", which matches any
353 character (including newline). Other properties such as "InMusicalSymbols" are
354 not currently supported by PCRE. Note that \P{Any} does not match any
355 characters, so always causes a match failure.
356 </P>
357 <P>
358 Sets of Unicode characters are defined as belonging to certain scripts. A
359 character from one of these sets can be matched using a script name. For
360 example:
361 <pre>
362 \p{Greek}
363 \P{Han}
364 </pre>
365 Those that are not part of an identified script are lumped together as
366 "Common". The current list of scripts is:
367 </P>
368 <P>
369 Arabic,
370 Armenian,
371 Balinese,
372 Bengali,
373 Bopomofo,
374 Braille,
375 Buginese,
376 Buhid,
377 Canadian_Aboriginal,
378 Cherokee,
379 Common,
380 Coptic,
381 Cuneiform,
382 Cypriot,
383 Cyrillic,
384 Deseret,
385 Devanagari,
386 Ethiopic,
387 Georgian,
388 Glagolitic,
389 Gothic,
390 Greek,
391 Gujarati,
392 Gurmukhi,
393 Han,
394 Hangul,
395 Hanunoo,
396 Hebrew,
397 Hiragana,
398 Inherited,
399 Kannada,
400 Katakana,
401 Kharoshthi,
402 Khmer,
403 Lao,
404 Latin,
405 Limbu,
406 Linear_B,
407 Malayalam,
408 Mongolian,
409 Myanmar,
410 New_Tai_Lue,
411 Nko,
412 Ogham,
413 Old_Italic,
414 Old_Persian,
415 Oriya,
416 Osmanya,
417 Phags_Pa,
418 Phoenician,
419 Runic,
420 Shavian,
421 Sinhala,
422 Syloti_Nagri,
423 Syriac,
424 Tagalog,
425 Tagbanwa,
426 Tai_Le,
427 Tamil,
428 Telugu,
429 Thaana,
430 Thai,
431 Tibetan,
432 Tifinagh,
433 Ugaritic,
434 Yi.
435 </P>
436 <P>
437 Each character has exactly one general category property, specified by a
438 two-letter abbreviation. For compatibility with Perl, negation can be specified
439 by including a circumflex between the opening brace and the property name. For
440 example, \p{^Lu} is the same as \P{Lu}.
441 </P>
442 <P>
443 If only one letter is specified with \p or \P, it includes all the general
444 category properties that start with that letter. In this case, in the absence
445 of negation, the curly brackets in the escape sequence are optional; these two
446 examples have the same effect:
447 <pre>
448 \p{L}
449 \pL
450 </pre>
451 The following general category property codes are supported:
452 <pre>
453 C Other
454 Cc Control
455 Cf Format
456 Cn Unassigned
457 Co Private use
458 Cs Surrogate
459
460 L Letter
461 Ll Lower case letter
462 Lm Modifier letter
463 Lo Other letter
464 Lt Title case letter
465 Lu Upper case letter
466
467 M Mark
468 Mc Spacing mark
469 Me Enclosing mark
470 Mn Non-spacing mark
471
472 N Number
473 Nd Decimal number
474 Nl Letter number
475 No Other number
476
477 P Punctuation
478 Pc Connector punctuation
479 Pd Dash punctuation
480 Pe Close punctuation
481 Pf Final punctuation
482 Pi Initial punctuation
483 Po Other punctuation
484 Ps Open punctuation
485
486 S Symbol
487 Sc Currency symbol
488 Sk Modifier symbol
489 Sm Mathematical symbol
490 So Other symbol
491
492 Z Separator
493 Zl Line separator
494 Zp Paragraph separator
495 Zs Space separator
496 </pre>
497 The special property L& is also supported: it matches a character that has
498 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
499 a modifier or "other".
500 </P>
501 <P>
502 The long synonyms for these properties that Perl supports (such as \p{Letter})
503 are not supported by PCRE, nor is it permitted to prefix any of these
504 properties with "Is".
505 </P>
506 <P>
507 No character that is in the Unicode table has the Cn (unassigned) property.
508 Instead, this property is assumed for any code point that is not in the
509 Unicode table.
510 </P>
511 <P>
512 Specifying caseless matching does not affect these escape sequences. For
513 example, \p{Lu} always matches only upper case letters.
514 </P>
515 <P>
516 The \X escape matches any number of Unicode characters that form an extended
517 Unicode sequence. \X is equivalent to
518 <pre>
519 (?&#62;\PM\pM*)
520 </pre>
521 That is, it matches a character without the "mark" property, followed by zero
522 or more characters with the "mark" property, and treats the sequence as an
523 atomic group
524 <a href="#atomicgroup">(see below).</a>
525 Characters with the "mark" property are typically accents that affect the
526 preceding character.
527 </P>
528 <P>
529 Matching characters by Unicode property is not fast, because PCRE has to search
530 a structure that contains data for over fifteen thousand characters. That is
531 why the traditional escape sequences such as \d and \w do not use Unicode
532 properties in PCRE.
533 <a name="resetmatchstart"></a></P>
534 <br><b>
535 Resetting the match start
536 </b><br>
537 <P>
538 The escape sequence \K, which is a Perl 5.10 feature, causes any previously
539 matched characters not to be included in the final matched sequence. For
540 example, the pattern:
541 <pre>
542 foo\Kbar
543 </pre>
544 matches "foobar", but reports that it has matched "bar". This feature is
545 similar to a lookbehind assertion
546 <a href="#lookbehind">(described below).</a>
547 However, in this case, the part of the subject before the real match does not
548 have to be of fixed length, as lookbehind assertions do. The use of \K does
549 not interfere with the setting of
550 <a href="#subpattern">captured substrings.</a>
551 For example, when the pattern
552 <pre>
553 (foo)\Kbar
554 </pre>
555 matches "foobar", the first substring is still set to "foo".
556 <a name="smallassertions"></a></P>
557 <br><b>
558 Simple assertions
559 </b><br>
560 <P>
561 The final use of backslash is for certain simple assertions. An assertion
562 specifies a condition that has to be met at a particular point in a match,
563 without consuming any characters from the subject string. The use of
564 subpatterns for more complicated assertions is described
565 <a href="#bigassertions">below.</a>
566 The backslashed assertions are:
567 <pre>
568 \b matches at a word boundary
569 \B matches when not at a word boundary
570 \A matches at the start of the subject
571 \Z matches at the end of the subject
572 also matches before a newline at the end of the subject
573 \z matches only at the end of the subject
574 \G matches at the first matching position in the subject
575 </pre>
576 These assertions may not appear in character classes (but note that \b has a
577 different meaning, namely the backspace character, inside a character class).
578 </P>
579 <P>
580 A word boundary is a position in the subject string where the current character
581 and the previous character do not both match \w or \W (i.e. one matches
582 \w and the other matches \W), or the start or end of the string if the
583 first or last character matches \w, respectively.
584 </P>
585 <P>
586 The \A, \Z, and \z assertions differ from the traditional circumflex and
587 dollar (described in the next section) in that they only ever match at the very
588 start and end of the subject string, whatever options are set. Thus, they are
589 independent of multiline mode. These three assertions are not affected by the
590 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
591 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
592 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
593 at a point other than the beginning of the subject, \A can never match. The
594 difference between \Z and \z is that \Z matches before a newline at the end
595 of the string as well as at the very end, whereas \z matches only at the end.
596 </P>
597 <P>
598 The \G assertion is true only when the current matching position is at the
599 start point of the match, as specified by the <i>startoffset</i> argument of
600 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
601 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
602 arguments, you can mimic Perl's /g option, and it is in this kind of
603 implementation where \G can be useful.
604 </P>
605 <P>
606 Note, however, that PCRE's interpretation of \G, as the start of the current
607 match, is subtly different from Perl's, which defines it as the end of the
608 previous match. In Perl, these can be different when the previously matched
609 string was empty. Because PCRE does just one match at a time, it cannot
610 reproduce this behaviour.
611 </P>
612 <P>
613 If all the alternatives of a pattern begin with \G, the expression is anchored
614 to the starting match position, and the "anchored" flag is set in the compiled
615 regular expression.
616 </P>
617 <br><a name="SEC4" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
618 <P>
619 Outside a character class, in the default matching mode, the circumflex
620 character is an assertion that is true only if the current matching point is
621 at the start of the subject string. If the <i>startoffset</i> argument of
622 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
623 option is unset. Inside a character class, circumflex has an entirely different
624 meaning
625 <a href="#characterclass">(see below).</a>
626 </P>
627 <P>
628 Circumflex need not be the first character of the pattern if a number of
629 alternatives are involved, but it should be the first thing in each alternative
630 in which it appears if the pattern is ever to match that branch. If all
631 possible alternatives start with a circumflex, that is, if the pattern is
632 constrained to match only at the start of the subject, it is said to be an
633 "anchored" pattern. (There are also other constructs that can cause a pattern
634 to be anchored.)
635 </P>
636 <P>
637 A dollar character is an assertion that is true only if the current matching
638 point is at the end of the subject string, or immediately before a newline
639 at the end of the string (by default). Dollar need not be the last character of
640 the pattern if a number of alternatives are involved, but it should be the last
641 item in any branch in which it appears. Dollar has no special meaning in a
642 character class.
643 </P>
644 <P>
645 The meaning of dollar can be changed so that it matches only at the very end of
646 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
647 does not affect the \Z assertion.
648 </P>
649 <P>
650 The meanings of the circumflex and dollar characters are changed if the
651 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
652 immediately after internal newlines as well as at the start of the subject
653 string. It does not match after a newline that ends the string. A dollar
654 matches before any newlines in the string, as well as at the very end, when
655 PCRE_MULTILINE is set. When newline is specified as the two-character
656 sequence CRLF, isolated CR and LF characters do not indicate newlines.
657 </P>
658 <P>
659 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
660 \n represents a newline) in multiline mode, but not otherwise. Consequently,
661 patterns that are anchored in single line mode because all branches start with
662 ^ are not anchored in multiline mode, and a match for circumflex is possible
663 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
664 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
665 </P>
666 <P>
667 Note that the sequences \A, \Z, and \z can be used to match the start and
668 end of the subject in both modes, and if all branches of a pattern start with
669 \A it is always anchored, whether or not PCRE_MULTILINE is set.
670 </P>
671 <br><a name="SEC5" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
672 <P>
673 Outside a character class, a dot in the pattern matches any one character in
674 the subject string except (by default) a character that signifies the end of a
675 line. In UTF-8 mode, the matched character may be more than one byte long.
676 </P>
677 <P>
678 When a line ending is defined as a single character, dot never matches that
679 character; when the two-character sequence CRLF is used, dot does not match CR
680 if it is immediately followed by LF, but otherwise it matches all characters
681 (including isolated CRs and LFs). When any Unicode line endings are being
682 recognized, dot does not match CR or LF or any of the other line ending
683 characters.
684 </P>
685 <P>
686 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
687 option is set, a dot matches any one character, without exception. If the
688 two-character sequence CRLF is present in the subject string, it takes two dots
689 to match it.
690 </P>
691 <P>
692 The handling of dot is entirely independent of the handling of circumflex and
693 dollar, the only relationship being that they both involve newlines. Dot has no
694 special meaning in a character class.
695 </P>
696 <br><a name="SEC6" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
697 <P>
698 Outside a character class, the escape sequence \C matches any one byte, both
699 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
700 characters. The feature is provided in Perl in order to match individual bytes
701 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
702 what remains in the string may be a malformed UTF-8 string. For this reason,
703 the \C escape sequence is best avoided.
704 </P>
705 <P>
706 PCRE does not allow \C to appear in lookbehind assertions
707 <a href="#lookbehind">(described below),</a>
708 because in UTF-8 mode this would make it impossible to calculate the length of
709 the lookbehind.
710 <a name="characterclass"></a></P>
711 <br><a name="SEC7" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
712 <P>
713 An opening square bracket introduces a character class, terminated by a closing
714 square bracket. A closing square bracket on its own is not special. If a
715 closing square bracket is required as a member of the class, it should be the
716 first data character in the class (after an initial circumflex, if present) or
717 escaped with a backslash.
718 </P>
719 <P>
720 A character class matches a single character in the subject. In UTF-8 mode, the
721 character may occupy more than one byte. A matched character must be in the set
722 of characters defined by the class, unless the first character in the class
723 definition is a circumflex, in which case the subject character must not be in
724 the set defined by the class. If a circumflex is actually required as a member
725 of the class, ensure it is not the first character, or escape it with a
726 backslash.
727 </P>
728 <P>
729 For example, the character class [aeiou] matches any lower case vowel, while
730 [^aeiou] matches any character that is not a lower case vowel. Note that a
731 circumflex is just a convenient notation for specifying the characters that
732 are in the class by enumerating those that are not. A class that starts with a
733 circumflex is not an assertion: it still consumes a character from the subject
734 string, and therefore it fails if the current pointer is at the end of the
735 string.
736 </P>
737 <P>
738 In UTF-8 mode, characters with values greater than 255 can be included in a
739 class as a literal string of bytes, or by using the \x{ escaping mechanism.
740 </P>
741 <P>
742 When caseless matching is set, any letters in a class represent both their
743 upper case and lower case versions, so for example, a caseless [aeiou] matches
744 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
745 caseful version would. In UTF-8 mode, PCRE always understands the concept of
746 case for characters whose values are less than 128, so caseless matching is
747 always possible. For characters with higher values, the concept of case is
748 supported if PCRE is compiled with Unicode property support, but not otherwise.
749 If you want to use caseless matching for characters 128 and above, you must
750 ensure that PCRE is compiled with Unicode property support as well as with
751 UTF-8 support.
752 </P>
753 <P>
754 Characters that might indicate line breaks are never treated in any special way
755 when matching character classes, whatever line-ending sequence is in use, and
756 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
757 such as [^a] always matches one of these characters.
758 </P>
759 <P>
760 The minus (hyphen) character can be used to specify a range of characters in a
761 character class. For example, [d-m] matches any letter between d and m,
762 inclusive. If a minus character is required in a class, it must be escaped with
763 a backslash or appear in a position where it cannot be interpreted as
764 indicating a range, typically as the first or last character in the class.
765 </P>
766 <P>
767 It is not possible to have the literal character "]" as the end character of a
768 range. A pattern such as [W-]46] is interpreted as a class of two characters
769 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
770 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
771 the end of range, so [W-\]46] is interpreted as a class containing a range
772 followed by two other characters. The octal or hexadecimal representation of
773 "]" can also be used to end a range.
774 </P>
775 <P>
776 Ranges operate in the collating sequence of character values. They can also be
777 used for characters specified numerically, for example [\000-\037]. In UTF-8
778 mode, ranges can include characters whose values are greater than 255, for
779 example [\x{100}-\x{2ff}].
780 </P>
781 <P>
782 If a range that includes letters is used when caseless matching is set, it
783 matches the letters in either case. For example, [W-c] is equivalent to
784 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
785 tables for a French locale are in use, [\xc8-\xcb] matches accented E
786 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
787 characters with values greater than 128 only when it is compiled with Unicode
788 property support.
789 </P>
790 <P>
791 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
792 in a character class, and add the characters that they match to the class. For
793 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
794 conveniently be used with the upper case character types to specify a more
795 restricted set of characters than the matching lower case type. For example,
796 the class [^\W_] matches any letter or digit, but not underscore.
797 </P>
798 <P>
799 The only metacharacters that are recognized in character classes are backslash,
800 hyphen (only where it can be interpreted as specifying a range), circumflex
801 (only at the start), opening square bracket (only when it can be interpreted as
802 introducing a POSIX class name - see the next section), and the terminating
803 closing square bracket. However, escaping other non-alphanumeric characters
804 does no harm.
805 </P>
806 <br><a name="SEC8" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
807 <P>
808 Perl supports the POSIX notation for character classes. This uses names
809 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
810 this notation. For example,
811 <pre>
812 [01[:alpha:]%]
813 </pre>
814 matches "0", "1", any alphabetic character, or "%". The supported class names
815 are
816 <pre>
817 alnum letters and digits
818 alpha letters
819 ascii character codes 0 - 127
820 blank space or tab only
821 cntrl control characters
822 digit decimal digits (same as \d)
823 graph printing characters, excluding space
824 lower lower case letters
825 print printing characters, including space
826 punct printing characters, excluding letters and digits
827 space white space (not quite the same as \s)
828 upper upper case letters
829 word "word" characters (same as \w)
830 xdigit hexadecimal digits
831 </pre>
832 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
833 space (32). Notice that this list includes the VT character (code 11). This
834 makes "space" different to \s, which does not include VT (for Perl
835 compatibility).
836 </P>
837 <P>
838 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
839 5.8. Another Perl extension is negation, which is indicated by a ^ character
840 after the colon. For example,
841 <pre>
842 [12[:^digit:]]
843 </pre>
844 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
845 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
846 supported, and an error is given if they are encountered.
847 </P>
848 <P>
849 In UTF-8 mode, characters with values greater than 128 do not match any of
850 the POSIX character classes.
851 </P>
852 <br><a name="SEC9" href="#TOC1">VERTICAL BAR</a><br>
853 <P>
854 Vertical bar characters are used to separate alternative patterns. For example,
855 the pattern
856 <pre>
857 gilbert|sullivan
858 </pre>
859 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
860 and an empty alternative is permitted (matching the empty string). The matching
861 process tries each alternative in turn, from left to right, and the first one
862 that succeeds is used. If the alternatives are within a subpattern
863 <a href="#subpattern">(defined below),</a>
864 "succeeds" means matching the rest of the main pattern as well as the
865 alternative in the subpattern.
866 </P>
867 <br><a name="SEC10" href="#TOC1">INTERNAL OPTION SETTING</a><br>
868 <P>
869 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
870 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
871 Perl option letters enclosed between "(?" and ")". The option letters are
872 <pre>
873 i for PCRE_CASELESS
874 m for PCRE_MULTILINE
875 s for PCRE_DOTALL
876 x for PCRE_EXTENDED
877 </pre>
878 For example, (?im) sets caseless, multiline matching. It is also possible to
879 unset these options by preceding the letter with a hyphen, and a combined
880 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
881 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
882 permitted. If a letter appears both before and after the hyphen, the option is
883 unset.
884 </P>
885 <P>
886 When an option change occurs at top level (that is, not inside subpattern
887 parentheses), the change applies to the remainder of the pattern that follows.
888 If the change is placed right at the start of a pattern, PCRE extracts it into
889 the global options (and it will therefore show up in data extracted by the
890 <b>pcre_fullinfo()</b> function).
891 </P>
892 <P>
893 An option change within a subpattern (see below for a description of
894 subpatterns) affects only that part of the current pattern that follows it, so
895 <pre>
896 (a(?i)b)c
897 </pre>
898 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
899 By this means, options can be made to have different settings in different
900 parts of the pattern. Any changes made in one alternative do carry on
901 into subsequent branches within the same subpattern. For example,
902 <pre>
903 (a(?i)b|c)
904 </pre>
905 matches "ab", "aB", "c", and "C", even though when matching "C" the first
906 branch is abandoned before the option setting. This is because the effects of
907 option settings happen at compile time. There would be some very weird
908 behaviour otherwise.
909 </P>
910 <P>
911 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
912 changed in the same way as the Perl-compatible options by using the characters
913 J, U and X respectively.
914 <a name="subpattern"></a></P>
915 <br><a name="SEC11" href="#TOC1">SUBPATTERNS</a><br>
916 <P>
917 Subpatterns are delimited by parentheses (round brackets), which can be nested.
918 Turning part of a pattern into a subpattern does two things:
919 <br>
920 <br>
921 1. It localizes a set of alternatives. For example, the pattern
922 <pre>
923 cat(aract|erpillar|)
924 </pre>
925 matches one of the words "cat", "cataract", or "caterpillar". Without the
926 parentheses, it would match "cataract", "erpillar" or an empty string.
927 <br>
928 <br>
929 2. It sets up the subpattern as a capturing subpattern. This means that, when
930 the whole pattern matches, that portion of the subject string that matched the
931 subpattern is passed back to the caller via the <i>ovector</i> argument of
932 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
933 from 1) to obtain numbers for the capturing subpatterns.
934 </P>
935 <P>
936 For example, if the string "the red king" is matched against the pattern
937 <pre>
938 the ((red|white) (king|queen))
939 </pre>
940 the captured substrings are "red king", "red", and "king", and are numbered 1,
941 2, and 3, respectively.
942 </P>
943 <P>
944 The fact that plain parentheses fulfil two functions is not always helpful.
945 There are often times when a grouping subpattern is required without a
946 capturing requirement. If an opening parenthesis is followed by a question mark
947 and a colon, the subpattern does not do any capturing, and is not counted when
948 computing the number of any subsequent capturing subpatterns. For example, if
949 the string "the white queen" is matched against the pattern
950 <pre>
951 the ((?:red|white) (king|queen))
952 </pre>
953 the captured substrings are "white queen" and "queen", and are numbered 1 and
954 2. The maximum number of capturing subpatterns is 65535.
955 </P>
956 <P>
957 As a convenient shorthand, if any option settings are required at the start of
958 a non-capturing subpattern, the option letters may appear between the "?" and
959 the ":". Thus the two patterns
960 <pre>
961 (?i:saturday|sunday)
962 (?:(?i)saturday|sunday)
963 </pre>
964 match exactly the same set of strings. Because alternative branches are tried
965 from left to right, and options are not reset until the end of the subpattern
966 is reached, an option setting in one branch does affect subsequent branches, so
967 the above patterns match "SUNDAY" as well as "Saturday".
968 </P>
969 <br><a name="SEC12" href="#TOC1">NAMED SUBPATTERNS</a><br>
970 <P>
971 Identifying capturing parentheses by number is simple, but it can be very hard
972 to keep track of the numbers in complicated regular expressions. Furthermore,
973 if an expression is modified, the numbers may change. To help with this
974 difficulty, PCRE supports the naming of subpatterns. This feature was not
975 added to Perl until release 5.10. Python had the feature earlier, and PCRE
976 introduced it at release 4.0, using the Python syntax. PCRE now supports both
977 the Perl and the Python syntax.
978 </P>
979 <P>
980 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
981 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
982 parentheses from other parts of the pattern, such as
983 <a href="#backreferences">backreferences,</a>
984 <a href="#recursion">recursion,</a>
985 and
986 <a href="#conditions">conditions,</a>
987 can be made by name as well as by number.
988 </P>
989 <P>
990 Names consist of up to 32 alphanumeric characters and underscores. Named
991 capturing parentheses are still allocated numbers as well as names, exactly as
992 if the names were not present. The PCRE API provides function calls for
993 extracting the name-to-number translation table from a compiled pattern. There
994 is also a convenience function for extracting a captured substring by name.
995 </P>
996 <P>
997 By default, a name must be unique within a pattern, but it is possible to relax
998 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
999 be useful for patterns where only one instance of the named parentheses can
1000 match. Suppose you want to match the name of a weekday, either as a 3-letter
1001 abbreviation or as the full name, and in both cases you want to extract the
1002 abbreviation. This pattern (ignoring the line breaks) does the job:
1003 <pre>
1004 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1005 (?&#60;DN&#62;Tue)(?:sday)?|
1006 (?&#60;DN&#62;Wed)(?:nesday)?|
1007 (?&#60;DN&#62;Thu)(?:rsday)?|
1008 (?&#60;DN&#62;Sat)(?:urday)?
1009 </pre>
1010 There are five capturing substrings, but only one is ever set after a match.
1011 The convenience function for extracting the data by name returns the substring
1012 for the first (and in this example, the only) subpattern of that name that
1013 matched. This saves searching to find which numbered subpattern it was. If you
1014 make a reference to a non-unique named subpattern from elsewhere in the
1015 pattern, the one that corresponds to the lowest number is used. For further
1016 details of the interfaces for handling named subpatterns, see the
1017 <a href="pcreapi.html"><b>pcreapi</b></a>
1018 documentation.
1019 </P>
1020 <br><a name="SEC13" href="#TOC1">REPETITION</a><br>
1021 <P>
1022 Repetition is specified by quantifiers, which can follow any of the following
1023 items:
1024 <pre>
1025 a literal data character
1026 the dot metacharacter
1027 the \C escape sequence
1028 the \X escape sequence (in UTF-8 mode with Unicode properties)
1029 the \R escape sequence
1030 an escape such as \d that matches a single character
1031 a character class
1032 a back reference (see next section)
1033 a parenthesized subpattern (unless it is an assertion)
1034 </pre>
1035 The general repetition quantifier specifies a minimum and maximum number of
1036 permitted matches, by giving the two numbers in curly brackets (braces),
1037 separated by a comma. The numbers must be less than 65536, and the first must
1038 be less than or equal to the second. For example:
1039 <pre>
1040 z{2,4}
1041 </pre>
1042 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1043 character. If the second number is omitted, but the comma is present, there is
1044 no upper limit; if the second number and the comma are both omitted, the
1045 quantifier specifies an exact number of required matches. Thus
1046 <pre>
1047 [aeiou]{3,}
1048 </pre>
1049 matches at least 3 successive vowels, but may match many more, while
1050 <pre>
1051 \d{8}
1052 </pre>
1053 matches exactly 8 digits. An opening curly bracket that appears in a position
1054 where a quantifier is not allowed, or one that does not match the syntax of a
1055 quantifier, is taken as a literal character. For example, {,6} is not a
1056 quantifier, but a literal string of four characters.
1057 </P>
1058 <P>
1059 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1060 bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
1061 which is represented by a two-byte sequence. Similarly, when Unicode property
1062 support is available, \X{3} matches three Unicode extended sequences, each of
1063 which may be several bytes long (and they may be of different lengths).
1064 </P>
1065 <P>
1066 The quantifier {0} is permitted, causing the expression to behave as if the
1067 previous item and the quantifier were not present.
1068 </P>
1069 <P>
1070 For convenience, the three most common quantifiers have single-character
1071 abbreviations:
1072 <pre>
1073 * is equivalent to {0,}
1074 + is equivalent to {1,}
1075 ? is equivalent to {0,1}
1076 </pre>
1077 It is possible to construct infinite loops by following a subpattern that can
1078 match no characters with a quantifier that has no upper limit, for example:
1079 <pre>
1080 (a?)*
1081 </pre>
1082 Earlier versions of Perl and PCRE used to give an error at compile time for
1083 such patterns. However, because there are cases where this can be useful, such
1084 patterns are now accepted, but if any repetition of the subpattern does in fact
1085 match no characters, the loop is forcibly broken.
1086 </P>
1087 <P>
1088 By default, the quantifiers are "greedy", that is, they match as much as
1089 possible (up to the maximum number of permitted times), without causing the
1090 rest of the pattern to fail. The classic example of where this gives problems
1091 is in trying to match comments in C programs. These appear between /* and */
1092 and within the comment, individual * and / characters may appear. An attempt to
1093 match C comments by applying the pattern
1094 <pre>
1095 /\*.*\*/
1096 </pre>
1097 to the string
1098 <pre>
1099 /* first comment */ not comment /* second comment */
1100 </pre>
1101 fails, because it matches the entire string owing to the greediness of the .*
1102 item.
1103 </P>
1104 <P>
1105 However, if a quantifier is followed by a question mark, it ceases to be
1106 greedy, and instead matches the minimum number of times possible, so the
1107 pattern
1108 <pre>
1109 /\*.*?\*/
1110 </pre>
1111 does the right thing with the C comments. The meaning of the various
1112 quantifiers is not otherwise changed, just the preferred number of matches.
1113 Do not confuse this use of question mark with its use as a quantifier in its
1114 own right. Because it has two uses, it can sometimes appear doubled, as in
1115 <pre>
1116 \d??\d
1117 </pre>
1118 which matches one digit by preference, but can match two if that is the only
1119 way the rest of the pattern matches.
1120 </P>
1121 <P>
1122 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1123 the quantifiers are not greedy by default, but individual ones can be made
1124 greedy by following them with a question mark. In other words, it inverts the
1125 default behaviour.
1126 </P>
1127 <P>
1128 When a parenthesized subpattern is quantified with a minimum repeat count that
1129 is greater than 1 or with a limited maximum, more memory is required for the
1130 compiled pattern, in proportion to the size of the minimum or maximum.
1131 </P>
1132 <P>
1133 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1134 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1135 implicitly anchored, because whatever follows will be tried against every
1136 character position in the subject string, so there is no point in retrying the
1137 overall match at any position after the first. PCRE normally treats such a
1138 pattern as though it were preceded by \A.
1139 </P>
1140 <P>
1141 In cases where it is known that the subject string contains no newlines, it is
1142 worth setting PCRE_DOTALL in order to obtain this optimization, or
1143 alternatively using ^ to indicate anchoring explicitly.
1144 </P>
1145 <P>
1146 However, there is one situation where the optimization cannot be used. When .*
1147 is inside capturing parentheses that are the subject of a backreference
1148 elsewhere in the pattern, a match at the start may fail where a later one
1149 succeeds. Consider, for example:
1150 <pre>
1151 (.*)abc\1
1152 </pre>
1153 If the subject is "xyz123abc123" the match point is the fourth character. For
1154 this reason, such a pattern is not implicitly anchored.
1155 </P>
1156 <P>
1157 When a capturing subpattern is repeated, the value captured is the substring
1158 that matched the final iteration. For example, after
1159 <pre>
1160 (tweedle[dume]{3}\s*)+
1161 </pre>
1162 has matched "tweedledum tweedledee" the value of the captured substring is
1163 "tweedledee". However, if there are nested capturing subpatterns, the
1164 corresponding captured values may have been set in previous iterations. For
1165 example, after
1166 <pre>
1167 /(a|(b))+/
1168 </pre>
1169 matches "aba" the value of the second captured substring is "b".
1170 <a name="atomicgroup"></a></P>
1171 <br><a name="SEC14" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1172 <P>
1173 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1174 repetition, failure of what follows normally causes the repeated item to be
1175 re-evaluated to see if a different number of repeats allows the rest of the
1176 pattern to match. Sometimes it is useful to prevent this, either to change the
1177 nature of the match, or to cause it fail earlier than it otherwise might, when
1178 the author of the pattern knows there is no point in carrying on.
1179 </P>
1180 <P>
1181 Consider, for example, the pattern \d+foo when applied to the subject line
1182 <pre>
1183 123456bar
1184 </pre>
1185 After matching all 6 digits and then failing to match "foo", the normal
1186 action of the matcher is to try again with only 5 digits matching the \d+
1187 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1188 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1189 that once a subpattern has matched, it is not to be re-evaluated in this way.
1190 </P>
1191 <P>
1192 If we use atomic grouping for the previous example, the matcher gives up
1193 immediately on failing to match "foo" the first time. The notation is a kind of
1194 special parenthesis, starting with (?&#62; as in this example:
1195 <pre>
1196 (?&#62;\d+)foo
1197 </pre>
1198 This kind of parenthesis "locks up" the part of the pattern it contains once
1199 it has matched, and a failure further into the pattern is prevented from
1200 backtracking into it. Backtracking past it to previous items, however, works as
1201 normal.
1202 </P>
1203 <P>
1204 An alternative description is that a subpattern of this type matches the string
1205 of characters that an identical standalone pattern would match, if anchored at
1206 the current point in the subject string.
1207 </P>
1208 <P>
1209 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1210 the above example can be thought of as a maximizing repeat that must swallow
1211 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1212 number of digits they match in order to make the rest of the pattern match,
1213 (?&#62;\d+) can only match an entire sequence of digits.
1214 </P>
1215 <P>
1216 Atomic groups in general can of course contain arbitrarily complicated
1217 subpatterns, and can be nested. However, when the subpattern for an atomic
1218 group is just a single repeated item, as in the example above, a simpler
1219 notation, called a "possessive quantifier" can be used. This consists of an
1220 additional + character following a quantifier. Using this notation, the
1221 previous example can be rewritten as
1222 <pre>
1223 \d++foo
1224 </pre>
1225 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1226 option is ignored. They are a convenient notation for the simpler forms of
1227 atomic group. However, there is no difference in the meaning of a possessive
1228 quantifier and the equivalent atomic group, though there may be a performance
1229 difference; possessive quantifiers should be slightly faster.
1230 </P>
1231 <P>
1232 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1233 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1234 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1235 package, and PCRE copied it from there. It ultimately found its way into Perl
1236 at release 5.10.
1237 </P>
1238 <P>
1239 PCRE has an optimization that automatically "possessifies" certain simple
1240 pattern constructs. For example, the sequence A+B is treated as A++B because
1241 there is no point in backtracking into a sequence of A's when B must follow.
1242 </P>
1243 <P>
1244 When a pattern contains an unlimited repeat inside a subpattern that can itself
1245 be repeated an unlimited number of times, the use of an atomic group is the
1246 only way to avoid some failing matches taking a very long time indeed. The
1247 pattern
1248 <pre>
1249 (\D+|&#60;\d+&#62;)*[!?]
1250 </pre>
1251 matches an unlimited number of substrings that either consist of non-digits, or
1252 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1253 quickly. However, if it is applied to
1254 <pre>
1255 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1256 </pre>
1257 it takes a long time before reporting failure. This is because the string can
1258 be divided between the internal \D+ repeat and the external * repeat in a
1259 large number of ways, and all have to be tried. (The example uses [!?] rather
1260 than a single character at the end, because both PCRE and Perl have an
1261 optimization that allows for fast failure when a single character is used. They
1262 remember the last single character that is required for a match, and fail early
1263 if it is not present in the string.) If the pattern is changed so that it uses
1264 an atomic group, like this:
1265 <pre>
1266 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1267 </pre>
1268 sequences of non-digits cannot be broken, and failure happens quickly.
1269 <a name="backreferences"></a></P>
1270 <br><a name="SEC15" href="#TOC1">BACK REFERENCES</a><br>
1271 <P>
1272 Outside a character class, a backslash followed by a digit greater than 0 (and
1273 possibly further digits) is a back reference to a capturing subpattern earlier
1274 (that is, to its left) in the pattern, provided there have been that many
1275 previous capturing left parentheses.
1276 </P>
1277 <P>
1278 However, if the decimal number following the backslash is less than 10, it is
1279 always taken as a back reference, and causes an error only if there are not
1280 that many capturing left parentheses in the entire pattern. In other words, the
1281 parentheses that are referenced need not be to the left of the reference for
1282 numbers less than 10. A "forward back reference" of this type can make sense
1283 when a repetition is involved and the subpattern to the right has participated
1284 in an earlier iteration.
1285 </P>
1286 <P>
1287 It is not possible to have a numerical "forward back reference" to a subpattern
1288 whose number is 10 or more using this syntax because a sequence such as \50 is
1289 interpreted as a character defined in octal. See the subsection entitled
1290 "Non-printing characters"
1291 <a href="#digitsafterbackslash">above</a>
1292 for further details of the handling of digits following a backslash. There is
1293 no such problem when named parentheses are used. A back reference to any
1294 subpattern is possible using named parentheses (see below).
1295 </P>
1296 <P>
1297 Another way of avoiding the ambiguity inherent in the use of digits following a
1298 backslash is to use the \g escape sequence, which is a feature introduced in
1299 Perl 5.10. This escape must be followed by a positive or a negative number,
1300 optionally enclosed in braces. These examples are all identical:
1301 <pre>
1302 (ring), \1
1303 (ring), \g1
1304 (ring), \g{1}
1305 </pre>
1306 A positive number specifies an absolute reference without the ambiguity that is
1307 present in the older syntax. It is also useful when literal digits follow the
1308 reference. A negative number is a relative reference. Consider this example:
1309 <pre>
1310 (abc(def)ghi)\g{-1}
1311 </pre>
1312 The sequence \g{-1} is a reference to the most recently started capturing
1313 subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2}
1314 would be equivalent to \1. The use of relative references can be helpful in
1315 long patterns, and also in patterns that are created by joining together
1316 fragments that contain references within themselves.
1317 </P>
1318 <P>
1319 A back reference matches whatever actually matched the capturing subpattern in
1320 the current subject string, rather than anything matching the subpattern
1321 itself (see
1322 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1323 below for a way of doing that). So the pattern
1324 <pre>
1325 (sens|respons)e and \1ibility
1326 </pre>
1327 matches "sense and sensibility" and "response and responsibility", but not
1328 "sense and responsibility". If caseful matching is in force at the time of the
1329 back reference, the case of letters is relevant. For example,
1330 <pre>
1331 ((?i)rah)\s+\1
1332 </pre>
1333 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1334 capturing subpattern is matched caselessly.
1335 </P>
1336 <P>
1337 There are several different ways of writing back references to named
1338 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1339 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1340 back reference syntax, in which \g can be used for both numeric and named
1341 references, is also supported. We could rewrite the above example in any of
1342 the following ways:
1343 <pre>
1344 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1345 (?'p1'(?i)rah)\s+\k{p1}
1346 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1347 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1348 </pre>
1349 A subpattern that is referenced by name may appear in the pattern before or
1350 after the reference.
1351 </P>
1352 <P>
1353 There may be more than one back reference to the same subpattern. If a
1354 subpattern has not actually been used in a particular match, any back
1355 references to it always fail. For example, the pattern
1356 <pre>
1357 (a|(bc))\2
1358 </pre>
1359 always fails if it starts to match "a" rather than "bc". Because there may be
1360 many capturing parentheses in a pattern, all digits following the backslash are
1361 taken as part of a potential back reference number. If the pattern continues
1362 with a digit character, some delimiter must be used to terminate the back
1363 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1364 Otherwise an empty comment (see
1365 <a href="#comments">"Comments"</a>
1366 below) can be used.
1367 </P>
1368 <P>
1369 A back reference that occurs inside the parentheses to which it refers fails
1370 when the subpattern is first used, so, for example, (a\1) never matches.
1371 However, such references can be useful inside repeated subpatterns. For
1372 example, the pattern
1373 <pre>
1374 (a|b\1)+
1375 </pre>
1376 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1377 the subpattern, the back reference matches the character string corresponding
1378 to the previous iteration. In order for this to work, the pattern must be such
1379 that the first iteration does not need to match the back reference. This can be
1380 done using alternation, as in the example above, or by a quantifier with a
1381 minimum of zero.
1382 <a name="bigassertions"></a></P>
1383 <br><a name="SEC16" href="#TOC1">ASSERTIONS</a><br>
1384 <P>
1385 An assertion is a test on the characters following or preceding the current
1386 matching point that does not actually consume any characters. The simple
1387 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1388 <a href="#smallassertions">above.</a>
1389 </P>
1390 <P>
1391 More complicated assertions are coded as subpatterns. There are two kinds:
1392 those that look ahead of the current position in the subject string, and those
1393 that look behind it. An assertion subpattern is matched in the normal way,
1394 except that it does not cause the current matching position to be changed.
1395 </P>
1396 <P>
1397 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1398 because it makes no sense to assert the same thing several times. If any kind
1399 of assertion contains capturing subpatterns within it, these are counted for
1400 the purposes of numbering the capturing subpatterns in the whole pattern.
1401 However, substring capturing is carried out only for positive assertions,
1402 because it does not make sense for negative assertions.
1403 </P>
1404 <br><b>
1405 Lookahead assertions
1406 </b><br>
1407 <P>
1408 Lookahead assertions start with (?= for positive assertions and (?! for
1409 negative assertions. For example,
1410 <pre>
1411 \w+(?=;)
1412 </pre>
1413 matches a word followed by a semicolon, but does not include the semicolon in
1414 the match, and
1415 <pre>
1416 foo(?!bar)
1417 </pre>
1418 matches any occurrence of "foo" that is not followed by "bar". Note that the
1419 apparently similar pattern
1420 <pre>
1421 (?!foo)bar
1422 </pre>
1423 does not find an occurrence of "bar" that is preceded by something other than
1424 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1425 (?!foo) is always true when the next three characters are "bar". A
1426 lookbehind assertion is needed to achieve the other effect.
1427 </P>
1428 <P>
1429 If you want to force a matching failure at some point in a pattern, the most
1430 convenient way to do it is with (?!) because an empty string always matches, so
1431 an assertion that requires there not to be an empty string must always fail.
1432 <a name="lookbehind"></a></P>
1433 <br><b>
1434 Lookbehind assertions
1435 </b><br>
1436 <P>
1437 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1438 negative assertions. For example,
1439 <pre>
1440 (?&#60;!foo)bar
1441 </pre>
1442 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1443 a lookbehind assertion are restricted such that all the strings it matches must
1444 have a fixed length. However, if there are several top-level alternatives, they
1445 do not all have to have the same fixed length. Thus
1446 <pre>
1447 (?&#60;=bullock|donkey)
1448 </pre>
1449 is permitted, but
1450 <pre>
1451 (?&#60;!dogs?|cats?)
1452 </pre>
1453 causes an error at compile time. Branches that match different length strings
1454 are permitted only at the top level of a lookbehind assertion. This is an
1455 extension compared with Perl (at least for 5.8), which requires all branches to
1456 match the same length of string. An assertion such as
1457 <pre>
1458 (?&#60;=ab(c|de))
1459 </pre>
1460 is not permitted, because its single top-level branch can match two different
1461 lengths, but it is acceptable if rewritten to use two top-level branches:
1462 <pre>
1463 (?&#60;=abc|abde)
1464 </pre>
1465 In some cases, the Perl 5.10 escape sequence \K
1466 <a href="#resetmatchstart">(see above)</a>
1467 can be used instead of a lookbehind assertion; this is not restricted to a
1468 fixed-length.
1469 </P>
1470 <P>
1471 The implementation of lookbehind assertions is, for each alternative, to
1472 temporarily move the current position back by the fixed length and then try to
1473 match. If there are insufficient characters before the current position, the
1474 assertion fails.
1475 </P>
1476 <P>
1477 PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1478 to appear in lookbehind assertions, because it makes it impossible to calculate
1479 the length of the lookbehind. The \X and \R escapes, which can match
1480 different numbers of bytes, are also not permitted.
1481 </P>
1482 <P>
1483 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1484 specify efficient matching at the end of the subject string. Consider a simple
1485 pattern such as
1486 <pre>
1487 abcd$
1488 </pre>
1489 when applied to a long string that does not match. Because matching proceeds
1490 from left to right, PCRE will look for each "a" in the subject and then see if
1491 what follows matches the rest of the pattern. If the pattern is specified as
1492 <pre>
1493 ^.*abcd$
1494 </pre>
1495 the initial .* matches the entire string at first, but when this fails (because
1496 there is no following "a"), it backtracks to match all but the last character,
1497 then all but the last two characters, and so on. Once again the search for "a"
1498 covers the entire string, from right to left, so we are no better off. However,
1499 if the pattern is written as
1500 <pre>
1501 ^.*+(?&#60;=abcd)
1502 </pre>
1503 there can be no backtracking for the .*+ item; it can match only the entire
1504 string. The subsequent lookbehind assertion does a single test on the last four
1505 characters. If it fails, the match fails immediately. For long strings, this
1506 approach makes a significant difference to the processing time.
1507 </P>
1508 <br><b>
1509 Using multiple assertions
1510 </b><br>
1511 <P>
1512 Several assertions (of any sort) may occur in succession. For example,
1513 <pre>
1514 (?&#60;=\d{3})(?&#60;!999)foo
1515 </pre>
1516 matches "foo" preceded by three digits that are not "999". Notice that each of
1517 the assertions is applied independently at the same point in the subject
1518 string. First there is a check that the previous three characters are all
1519 digits, and then there is a check that the same three characters are not "999".
1520 This pattern does <i>not</i> match "foo" preceded by six characters, the first
1521 of which are digits and the last three of which are not "999". For example, it
1522 doesn't match "123abcfoo". A pattern to do that is
1523 <pre>
1524 (?&#60;=\d{3}...)(?&#60;!999)foo
1525 </pre>
1526 This time the first assertion looks at the preceding six characters, checking
1527 that the first three are digits, and then the second assertion checks that the
1528 preceding three characters are not "999".
1529 </P>
1530 <P>
1531 Assertions can be nested in any combination. For example,
1532 <pre>
1533 (?&#60;=(?&#60;!foo)bar)baz
1534 </pre>
1535 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1536 preceded by "foo", while
1537 <pre>
1538 (?&#60;=\d{3}(?!999)...)foo
1539 </pre>
1540 is another pattern that matches "foo" preceded by three digits and any three
1541 characters that are not "999".
1542 <a name="conditions"></a></P>
1543 <br><a name="SEC17" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1544 <P>
1545 It is possible to cause the matching process to obey a subpattern
1546 conditionally or to choose between two alternative subpatterns, depending on
1547 the result of an assertion, or whether a previous capturing subpattern matched
1548 or not. The two possible forms of conditional subpattern are
1549 <pre>
1550 (?(condition)yes-pattern)
1551 (?(condition)yes-pattern|no-pattern)
1552 </pre>
1553 If the condition is satisfied, the yes-pattern is used; otherwise the
1554 no-pattern (if present) is used. If there are more than two alternatives in the
1555 subpattern, a compile-time error occurs.
1556 </P>
1557 <P>
1558 There are four kinds of condition: references to subpatterns, references to
1559 recursion, a pseudo-condition called DEFINE, and assertions.
1560 </P>
1561 <br><b>
1562 Checking for a used subpattern by number
1563 </b><br>
1564 <P>
1565 If the text between the parentheses consists of a sequence of digits, the
1566 condition is true if the capturing subpattern of that number has previously
1567 matched. An alternative notation is to precede the digits with a plus or minus
1568 sign. In this case, the subpattern number is relative rather than absolute.
1569 The most recently opened parentheses can be referenced by (?(-1), the next most
1570 recent by (?(-2), and so on. In looping constructs it can also make sense to
1571 refer to subsequent groups with constructs such as (?(+2).
1572 </P>
1573 <P>
1574 Consider the following pattern, which contains non-significant white space to
1575 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1576 three parts for ease of discussion:
1577 <pre>
1578 ( \( )? [^()]+ (?(1) \) )
1579 </pre>
1580 The first part matches an optional opening parenthesis, and if that
1581 character is present, sets it as the first captured substring. The second part
1582 matches one or more characters that are not parentheses. The third part is a
1583 conditional subpattern that tests whether the first set of parentheses matched
1584 or not. If they did, that is, if subject started with an opening parenthesis,
1585 the condition is true, and so the yes-pattern is executed and a closing
1586 parenthesis is required. Otherwise, since no-pattern is not present, the
1587 subpattern matches nothing. In other words, this pattern matches a sequence of
1588 non-parentheses, optionally enclosed in parentheses.
1589 </P>
1590 <P>
1591 If you were embedding this pattern in a larger one, you could use a relative
1592 reference:
1593 <pre>
1594 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
1595 </pre>
1596 This makes the fragment independent of the parentheses in the larger pattern.
1597 </P>
1598 <br><b>
1599 Checking for a used subpattern by name
1600 </b><br>
1601 <P>
1602 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
1603 subpattern by name. For compatibility with earlier versions of PCRE, which had
1604 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1605 there is a possible ambiguity with this syntax, because subpattern names may
1606 consist entirely of digits. PCRE looks first for a named subpattern; if it
1607 cannot find one and the name consists entirely of digits, PCRE looks for a
1608 subpattern of that number, which must be greater than zero. Using subpattern
1609 names that consist entirely of digits is not recommended.
1610 </P>
1611 <P>
1612 Rewriting the above example to use a named subpattern gives this:
1613 <pre>
1614 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
1615
1616 </PRE>
1617 </P>
1618 <br><b>
1619 Checking for pattern recursion
1620 </b><br>
1621 <P>
1622 If the condition is the string (R), and there is no subpattern with the name R,
1623 the condition is true if a recursive call to the whole pattern or any
1624 subpattern has been made. If digits or a name preceded by ampersand follow the
1625 letter R, for example:
1626 <pre>
1627 (?(R3)...) or (?(R&name)...)
1628 </pre>
1629 the condition is true if the most recent recursion is into the subpattern whose
1630 number or name is given. This condition does not check the entire recursion
1631 stack.
1632 </P>
1633 <P>
1634 At "top level", all these recursion test conditions are false. Recursive
1635 patterns are described below.
1636 </P>
1637 <br><b>
1638 Defining subpatterns for use by reference only
1639 </b><br>
1640 <P>
1641 If the condition is the string (DEFINE), and there is no subpattern with the
1642 name DEFINE, the condition is always false. In this case, there may be only one
1643 alternative in the subpattern. It is always skipped if control reaches this
1644 point in the pattern; the idea of DEFINE is that it can be used to define
1645 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1646 is described below.) For example, a pattern to match an IPv4 address could be
1647 written like this (ignore whitespace and line breaks):
1648 <pre>
1649 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
1650 \b (?&byte) (\.(?&byte)){3} \b
1651 </pre>
1652 The first part of the pattern is a DEFINE group inside which a another group
1653 named "byte" is defined. This matches an individual component of an IPv4
1654 address (a number less than 256). When matching takes place, this part of the
1655 pattern is skipped because DEFINE acts like a false condition.
1656 </P>
1657 <P>
1658 The rest of the pattern uses references to the named group to match the four
1659 dot-separated components of an IPv4 address, insisting on a word boundary at
1660 each end.
1661 </P>
1662 <br><b>
1663 Assertion conditions
1664 </b><br>
1665 <P>
1666 If the condition is not in any of the above formats, it must be an assertion.
1667 This may be a positive or negative lookahead or lookbehind assertion. Consider
1668 this pattern, again containing non-significant white space, and with the two
1669 alternatives on the second line:
1670 <pre>
1671 (?(?=[^a-z]*[a-z])
1672 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1673 </pre>
1674 The condition is a positive lookahead assertion that matches an optional
1675 sequence of non-letters followed by a letter. In other words, it tests for the
1676 presence of at least one letter in the subject. If a letter is found, the
1677 subject is matched against the first alternative; otherwise it is matched
1678 against the second. This pattern matches strings in one of the two forms
1679 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1680 <a name="comments"></a></P>
1681 <br><a name="SEC18" href="#TOC1">COMMENTS</a><br>
1682 <P>
1683 The sequence (?# marks the start of a comment that continues up to the next
1684 closing parenthesis. Nested parentheses are not permitted. The characters
1685 that make up a comment play no part in the pattern matching at all.
1686 </P>
1687 <P>
1688 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1689 character class introduces a comment that continues to immediately after the
1690 next newline in the pattern.
1691 <a name="recursion"></a></P>
1692 <br><a name="SEC19" href="#TOC1">RECURSIVE PATTERNS</a><br>
1693 <P>
1694 Consider the problem of matching a string in parentheses, allowing for
1695 unlimited nested parentheses. Without the use of recursion, the best that can
1696 be done is to use a pattern that matches up to some fixed depth of nesting. It
1697 is not possible to handle an arbitrary nesting depth.
1698 </P>
1699 <P>
1700 For some time, Perl has provided a facility that allows regular expressions to
1701 recurse (amongst other things). It does this by interpolating Perl code in the
1702 expression at run time, and the code can refer to the expression itself. A Perl
1703 pattern using code interpolation to solve the parentheses problem can be
1704 created like this:
1705 <pre>
1706 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
1707 </pre>
1708 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1709 recursively to the pattern in which it appears.
1710 </P>
1711 <P>
1712 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1713 supports special syntax for recursion of the entire pattern, and also for
1714 individual subpattern recursion. After its introduction in PCRE and Python,
1715 this kind of recursion was introduced into Perl at release 5.10.
1716 </P>
1717 <P>
1718 A special item that consists of (? followed by a number greater than zero and a
1719 closing parenthesis is a recursive call of the subpattern of the given number,
1720 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1721 call, which is described in the next section.) The special item (?R) or (?0) is
1722 a recursive call of the entire regular expression.
1723 </P>
1724 <P>
1725 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1726 treated as an atomic group. That is, once it has matched some of the subject
1727 string, it is never re-entered, even if it contains untried alternatives and
1728 there is a subsequent matching failure.
1729 </P>
1730 <P>
1731 This PCRE pattern solves the nested parentheses problem (assume the
1732 PCRE_EXTENDED option is set so that white space is ignored):
1733 <pre>
1734 \( ( (?&#62;[^()]+) | (?R) )* \)
1735 </pre>
1736 First it matches an opening parenthesis. Then it matches any number of
1737 substrings which can either be a sequence of non-parentheses, or a recursive
1738 match of the pattern itself (that is, a correctly parenthesized substring).
1739 Finally there is a closing parenthesis.
1740 </P>
1741 <P>
1742 If this were part of a larger pattern, you would not want to recurse the entire
1743 pattern, so instead you could use this:
1744 <pre>
1745 ( \( ( (?&#62;[^()]+) | (?1) )* \) )
1746 </pre>
1747 We have put the pattern into parentheses, and caused the recursion to refer to
1748 them instead of the whole pattern.
1749 </P>
1750 <P>
1751 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1752 is made easier by the use of relative references. (A Perl 5.10 feature.)
1753 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1754 most recently opened parentheses preceding the recursion. In other words, a
1755 negative number counts capturing parentheses leftwards from the point at which
1756 it is encountered.
1757 </P>
1758 <P>
1759 It is also possible to refer to subsequently opened parentheses, by writing
1760 references such as (?+2). However, these cannot be recursive because the
1761 reference is not inside the parentheses that are referenced. They are always
1762 "subroutine" calls, as described in the next section.
1763 </P>
1764 <P>
1765 An alternative approach is to use named parentheses instead. The Perl syntax
1766 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
1767 could rewrite the above example as follows:
1768 <pre>
1769 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | (?&pn) )* \) )
1770 </pre>
1771 If there is more than one subpattern with the same name, the earliest one is
1772 used.
1773 </P>
1774 <P>
1775 This particular example pattern that we have been looking at contains nested
1776 unlimited repeats, and so the use of atomic grouping for matching strings of
1777 non-parentheses is important when applying the pattern to strings that do not
1778 match. For example, when this pattern is applied to
1779 <pre>
1780 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1781 </pre>
1782 it yields "no match" quickly. However, if atomic grouping is not used,
1783 the match runs for a very long time indeed because there are so many different
1784 ways the + and * repeats can carve up the subject, and all have to be tested
1785 before failure can be reported.
1786 </P>
1787 <P>
1788 At the end of a match, the values set for any capturing subpatterns are those
1789 from the outermost level of the recursion at which the subpattern value is set.
1790 If you want to obtain intermediate values, a callout function can be used (see
1791 below and the
1792 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1793 documentation). If the pattern above is matched against
1794 <pre>
1795 (ab(cd)ef)
1796 </pre>
1797 the value for the capturing parentheses is "ef", which is the last value taken
1798 on at the top level. If additional parentheses are added, giving
1799 <pre>
1800 \( ( ( (?&#62;[^()]+) | (?R) )* ) \)
1801 ^ ^
1802 ^ ^
1803 </pre>
1804 the string they capture is "ab(cd)ef", the contents of the top level
1805 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1806 has to obtain extra memory to store data during a recursion, which it does by
1807 using <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no
1808 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1809 </P>
1810 <P>
1811 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1812 Consider this pattern, which matches text in angle brackets, allowing for
1813 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1814 recursing), whereas any characters are permitted at the outer level.
1815 <pre>
1816 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
1817 </pre>
1818 In this pattern, (?(R) is the start of a conditional subpattern, with two
1819 different alternatives for the recursive and non-recursive cases. The (?R) item
1820 is the actual recursive call.
1821 <a name="subpatternsassubroutines"></a></P>
1822 <br><a name="SEC20" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
1823 <P>
1824 If the syntax for a recursive subpattern reference (either by number or by
1825 name) is used outside the parentheses to which it refers, it operates like a
1826 subroutine in a programming language. The "called" subpattern may be defined
1827 before or after the reference. A numbered reference can be absolute or
1828 relative, as in these examples:
1829 <pre>
1830 (...(absolute)...)...(?2)...
1831 (...(relative)...)...(?-1)...
1832 (...(?+1)...(relative)...
1833 </pre>
1834 An earlier example pointed out that the pattern
1835 <pre>
1836 (sens|respons)e and \1ibility
1837 </pre>
1838 matches "sense and sensibility" and "response and responsibility", but not
1839 "sense and responsibility". If instead the pattern
1840 <pre>
1841 (sens|respons)e and (?1)ibility
1842 </pre>
1843 is used, it does match "sense and responsibility" as well as the other two
1844 strings. Another example is given in the discussion of DEFINE above.
1845 </P>
1846 <P>
1847 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1848 group. That is, once it has matched some of the subject string, it is never
1849 re-entered, even if it contains untried alternatives and there is a subsequent
1850 matching failure.
1851 </P>
1852 <P>
1853 When a subpattern is used as a subroutine, processing options such as
1854 case-independence are fixed when the subpattern is defined. They cannot be
1855 changed for different calls. For example, consider this pattern:
1856 <pre>
1857 (abc)(?i:(?-1))
1858 </pre>
1859 It matches "abcabc". It does not match "abcABC" because the change of
1860 processing option does not affect the called subpattern.
1861 </P>
1862 <br><a name="SEC21" href="#TOC1">CALLOUTS</a><br>
1863 <P>
1864 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1865 code to be obeyed in the middle of matching a regular expression. This makes it
1866 possible, amongst other things, to extract different substrings that match the
1867 same pair of parentheses when there is a repetition.
1868 </P>
1869 <P>
1870 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1871 code. The feature is called "callout". The caller of PCRE provides an external
1872 function by putting its entry point in the global variable <i>pcre_callout</i>.
1873 By default, this variable contains NULL, which disables all calling out.
1874 </P>
1875 <P>
1876 Within a regular expression, (?C) indicates the points at which the external
1877 function is to be called. If you want to identify different callout points, you
1878 can put a number less than 256 after the letter C. The default value is zero.
1879 For example, this pattern has two callout points:
1880 <pre>
1881 (?C1)abc(?C2)def
1882 </pre>
1883 If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
1884 automatically installed before each item in the pattern. They are all numbered
1885 255.
1886 </P>
1887 <P>
1888 During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
1889 set), the external function is called. It is provided with the number of the
1890 callout, the position in the pattern, and, optionally, one item of data
1891 originally supplied by the caller of <b>pcre_exec()</b>. The callout function
1892 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1893 description of the interface to the callout function is given in the
1894 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1895 documentation.
1896 </P>
1897 <br><a name="SEC22" href="#TOC1">SEE ALSO</a><br>
1898 <P>
1899 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), <b>pcre</b>(3).
1900 </P>
1901 <br><a name="SEC23" href="#TOC1">AUTHOR</a><br>
1902 <P>
1903 Philip Hazel
1904 <br>
1905 University Computing Service
1906 <br>
1907 Cambridge CB2 3QH, England.
1908 <br>
1909 </P>
1910 <br><a name="SEC24" href="#TOC1">REVISION</a><br>
1911 <P>
1912 Last updated: 29 May 2007
1913 <br>
1914 Copyright &copy; 1997-2007 University of Cambridge.
1915 <br>
1916 <p>
1917 Return to the <a href="index.html">PCRE index page</a>.
1918 </p>

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5