/[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 75 - (show annotations)
Sat Feb 24 21:40:37 2007 UTC (12 years, 6 months ago) by nigel
File MIME type: text/html
File size: 64471 byte(s)
Load pcre-5.0 into code/trunk.
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">BACKSLASH</a>
18 <li><a name="TOC3" href="#SEC3">CIRCUMFLEX AND DOLLAR</a>
19 <li><a name="TOC4" href="#SEC4">FULL STOP (PERIOD, DOT)</a>
20 <li><a name="TOC5" href="#SEC5">MATCHING A SINGLE BYTE</a>
21 <li><a name="TOC6" href="#SEC6">SQUARE BRACKETS AND CHARACTER CLASSES</a>
22 <li><a name="TOC7" href="#SEC7">POSIX CHARACTER CLASSES</a>
23 <li><a name="TOC8" href="#SEC8">VERTICAL BAR</a>
24 <li><a name="TOC9" href="#SEC9">INTERNAL OPTION SETTING</a>
25 <li><a name="TOC10" href="#SEC10">SUBPATTERNS</a>
26 <li><a name="TOC11" href="#SEC11">NAMED SUBPATTERNS</a>
27 <li><a name="TOC12" href="#SEC12">REPETITION</a>
28 <li><a name="TOC13" href="#SEC13">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a>
29 <li><a name="TOC14" href="#SEC14">BACK REFERENCES</a>
30 <li><a name="TOC15" href="#SEC15">ASSERTIONS</a>
31 <li><a name="TOC16" href="#SEC16">CONDITIONAL SUBPATTERNS</a>
32 <li><a name="TOC17" href="#SEC17">COMMENTS</a>
33 <li><a name="TOC18" href="#SEC18">RECURSIVE PATTERNS</a>
34 <li><a name="TOC19" href="#SEC19">SUBPATTERNS AS SUBROUTINES</a>
35 <li><a name="TOC20" href="#SEC20">CALLOUTS</a>
36 </ul>
37 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
38 <P>
39 The syntax and semantics of the regular expressions supported by PCRE are
40 described below. Regular expressions are also described in the Perl
41 documentation and in a number of books, some of which have copious examples.
42 Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers
43 regular expressions in great detail. This description of PCRE's regular
44 expressions is intended as reference material.
45 </P>
46 <P>
47 The original operation of PCRE was on strings of one-byte characters. However,
48 there is now also support for UTF-8 character strings. To use this, you must
49 build PCRE to include UTF-8 support, and then call <b>pcre_compile()</b> with
50 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
51 places below. There is also a summary of UTF-8 features in the
52 <a href="pcre.html#utf8support">section on UTF-8 support</a>
53 in the main
54 <a href="pcre.html"><b>pcre</b></a>
55 page.
56 </P>
57 <P>
58 A regular expression is a pattern that is matched against a subject string from
59 left to right. Most characters stand for themselves in a pattern, and match the
60 corresponding characters in the subject. As a trivial example, the pattern
61 <pre>
62 The quick brown fox
63 </pre>
64 matches a portion of a subject string that is identical to itself. The power of
65 regular expressions comes from the ability to include alternatives and
66 repetitions in the pattern. These are encoded in the pattern by the use of
67 <i>metacharacters</i>, which do not stand for themselves but instead are
68 interpreted in some special way.
69 </P>
70 <P>
71 There are two different sets of metacharacters: those that are recognized
72 anywhere in the pattern except within square brackets, and those that are
73 recognized in square brackets. Outside square brackets, the metacharacters are
74 as follows:
75 <pre>
76 \ general escape character with several uses
77 ^ assert start of string (or line, in multiline mode)
78 $ assert end of string (or line, in multiline mode)
79 . match any character except newline (by default)
80 [ start character class definition
81 | start of alternative branch
82 ( start subpattern
83 ) end subpattern
84 ? extends the meaning of (
85 also 0 or 1 quantifier
86 also quantifier minimizer
87 * 0 or more quantifier
88 + 1 or more quantifier
89 also "possessive quantifier"
90 { start min/max quantifier
91 </pre>
92 Part of a pattern that is in square brackets is called a "character class". In
93 a character class the only metacharacters are:
94 <pre>
95 \ general escape character
96 ^ negate the class, but only if the first character
97 - indicates character range
98 [ POSIX character class (only if followed by POSIX syntax)
99 ] terminates the character class
100 </pre>
101 The following sections describe the use of each of the metacharacters.
102 </P>
103 <br><a name="SEC2" href="#TOC1">BACKSLASH</a><br>
104 <P>
105 The backslash character has several uses. Firstly, if it is followed by a
106 non-alphanumeric character, it takes away any special meaning that character may
107 have. This use of backslash as an escape character applies both inside and
108 outside character classes.
109 </P>
110 <P>
111 For example, if you want to match a * character, you write \* in the pattern.
112 This escaping action applies whether or not the following character would
113 otherwise be interpreted as a metacharacter, so it is always safe to precede a
114 non-alphanumeric with backslash to specify that it stands for itself. In
115 particular, if you want to match a backslash, you write \\.
116 </P>
117 <P>
118 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
119 pattern (other than in a character class) and characters between a # outside
120 a character class and the next newline character are ignored. An escaping
121 backslash can be used to include a whitespace or # character as part of the
122 pattern.
123 </P>
124 <P>
125 If you want to remove the special meaning from a sequence of characters, you
126 can do so by putting them between \Q and \E. This is different from Perl in
127 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
128 Perl, $ and @ cause variable interpolation. Note the following examples:
129 <pre>
130 Pattern PCRE matches Perl matches
131
132 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
133 \Qabc\$xyz\E abc\$xyz abc\$xyz
134 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
135 </pre>
136 The \Q...\E sequence is recognized both inside and outside character classes.
137 <a name="digitsafterbackslash"></a></P>
138 <br><b>
139 Non-printing characters
140 </b><br>
141 <P>
142 A second use of backslash provides a way of encoding non-printing characters
143 in patterns in a visible manner. There is no restriction on the appearance of
144 non-printing characters, apart from the binary zero that terminates a pattern,
145 but when a pattern is being prepared by text editing, it is usually easier to
146 use one of the following escape sequences than the binary character it
147 represents:
148 <pre>
149 \a alarm, that is, the BEL character (hex 07)
150 \cx "control-x", where x is any character
151 \e escape (hex 1B)
152 \f formfeed (hex 0C)
153 \n newline (hex 0A)
154 \r carriage return (hex 0D)
155 \t tab (hex 09)
156 \ddd character with octal code ddd, or backreference
157 \xhh character with hex code hh
158 \x{hhh..} character with hex code hhh... (UTF-8 mode only)
159 </pre>
160 The precise effect of \cx is as follows: if x is a lower case letter, it
161 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
162 Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
163 7B.
164 </P>
165 <P>
166 After \x, from zero to two hexadecimal digits are read (letters can be in
167 upper or lower case). In UTF-8 mode, any number of hexadecimal digits may
168 appear between \x{ and }, but the value of the character code must be less
169 than 2**31 (that is, the maximum hexadecimal value is 7FFFFFFF). If characters
170 other than hexadecimal digits appear between \x{ and }, or if there is no
171 terminating }, this form of escape is not recognized. Instead, the initial
172 \x will be interpreted as a basic hexadecimal escape, with no following
173 digits, giving a character whose value is zero.
174 </P>
175 <P>
176 Characters whose value is less than 256 can be defined by either of the two
177 syntaxes for \x when PCRE is in UTF-8 mode. There is no difference in the
178 way they are handled. For example, \xdc is exactly the same as \x{dc}.
179 </P>
180 <P>
181 After \0 up to two further octal digits are read. In both cases, if there
182 are fewer than two digits, just those that are present are used. Thus the
183 sequence \0\x\07 specifies two binary zeros followed by a BEL character
184 (code value 7). Make sure you supply two digits after the initial zero if the
185 pattern character that follows is itself an octal digit.
186 </P>
187 <P>
188 The handling of a backslash followed by a digit other than 0 is complicated.
189 Outside a character class, PCRE reads it and any following digits as a decimal
190 number. If the number is less than 10, or if there have been at least that many
191 previous capturing left parentheses in the expression, the entire sequence is
192 taken as a <i>back reference</i>. A description of how this works is given
193 <a href="#backreferences">later,</a>
194 following the discussion of
195 <a href="#subpattern">parenthesized subpatterns.</a>
196 </P>
197 <P>
198 Inside a character class, or if the decimal number is greater than 9 and there
199 have not been that many capturing subpatterns, PCRE re-reads up to three octal
200 digits following the backslash, and generates a single byte from the least
201 significant 8 bits of the value. Any subsequent digits stand for themselves.
202 For example:
203 <pre>
204 \040 is another way of writing a space
205 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
206 \7 is always a back reference
207 \11 might be a back reference, or another way of writing a tab
208 \011 is always a tab
209 \0113 is a tab followed by the character "3"
210 \113 might be a back reference, otherwise the character with octal code 113
211 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
212 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
213 </pre>
214 Note that octal values of 100 or greater must not be introduced by a leading
215 zero, because no more than three octal digits are ever read.
216 </P>
217 <P>
218 All the sequences that define a single byte value or a single UTF-8 character
219 (in UTF-8 mode) can be used both inside and outside character classes. In
220 addition, inside a character class, the sequence \b is interpreted as the
221 backspace character (hex 08), and the sequence \X is interpreted as the
222 character "X". Outside a character class, these sequences have different
223 meanings
224 <a href="#uniextseq">(see below).</a>
225 </P>
226 <br><b>
227 Generic character types
228 </b><br>
229 <P>
230 The third use of backslash is for specifying generic character types. The
231 following are always recognized:
232 <pre>
233 \d any decimal digit
234 \D any character that is not a decimal digit
235 \s any whitespace character
236 \S any character that is not a whitespace character
237 \w any "word" character
238 \W any "non-word" character
239 </pre>
240 Each pair of escape sequences partitions the complete set of characters into
241 two disjoint sets. Any given character matches one, and only one, of each pair.
242 </P>
243 <P>
244 These character type sequences can appear both inside and outside character
245 classes. They each match one character of the appropriate type. If the current
246 matching point is at the end of the subject string, all of them fail, since
247 there is no character to match.
248 </P>
249 <P>
250 For compatibility with Perl, \s does not match the VT character (code 11).
251 This makes it different from the the POSIX "space" class. The \s characters
252 are HT (9), LF (10), FF (12), CR (13), and space (32).
253 </P>
254 <P>
255 A "word" character is an underscore or any character less than 256 that is a
256 letter or digit. The definition of letters and digits is controlled by PCRE's
257 low-valued character tables, and may vary if locale-specific matching is taking
258 place (see
259 <a href="pcreapi.html#localesupport">"Locale support"</a>
260 in the
261 <a href="pcreapi.html"><b>pcreapi</b></a>
262 page). For example, in the "fr_FR" (French) locale, some character codes
263 greater than 128 are used for accented letters, and these are matched by \w.
264 </P>
265 <P>
266 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
267 \w, and always match \D, \S, and \W. This is true even when Unicode
268 character property support is available.
269 <a name="uniextseq"></a></P>
270 <br><b>
271 Unicode character properties
272 </b><br>
273 <P>
274 When PCRE is built with Unicode character property support, three additional
275 escape sequences to match generic character types are available when UTF-8 mode
276 is selected. They are:
277 <pre>
278 \p{<i>xx</i>} a character with the <i>xx</i> property
279 \P{<i>xx</i>} a character without the <i>xx</i> property
280 \X an extended Unicode sequence
281 </pre>
282 The property names represented by <i>xx</i> above are limited to the
283 Unicode general category properties. Each character has exactly one such
284 property, specified by a two-letter abbreviation. For compatibility with Perl,
285 negation can be specified by including a circumflex between the opening brace
286 and the property name. For example, \p{^Lu} is the same as \P{Lu}.
287 </P>
288 <P>
289 If only one letter is specified with \p or \P, it includes all the properties
290 that start with that letter. In this case, in the absence of negation, the
291 curly brackets in the escape sequence are optional; these two examples have
292 the same effect:
293 <pre>
294 \p{L}
295 \pL
296 </pre>
297 The following property codes are supported:
298 <pre>
299 C Other
300 Cc Control
301 Cf Format
302 Cn Unassigned
303 Co Private use
304 Cs Surrogate
305
306 L Letter
307 Ll Lower case letter
308 Lm Modifier letter
309 Lo Other letter
310 Lt Title case letter
311 Lu Upper case letter
312
313 M Mark
314 Mc Spacing mark
315 Me Enclosing mark
316 Mn Non-spacing mark
317
318 N Number
319 Nd Decimal number
320 Nl Letter number
321 No Other number
322
323 P Punctuation
324 Pc Connector punctuation
325 Pd Dash punctuation
326 Pe Close punctuation
327 Pf Final punctuation
328 Pi Initial punctuation
329 Po Other punctuation
330 Ps Open punctuation
331
332 S Symbol
333 Sc Currency symbol
334 Sk Modifier symbol
335 Sm Mathematical symbol
336 So Other symbol
337
338 Z Separator
339 Zl Line separator
340 Zp Paragraph separator
341 Zs Space separator
342 </pre>
343 Extended properties such as "Greek" or "InMusicalSymbols" are not supported by
344 PCRE.
345 </P>
346 <P>
347 Specifying caseless matching does not affect these escape sequences. For
348 example, \p{Lu} always matches only upper case letters.
349 </P>
350 <P>
351 The \X escape matches any number of Unicode characters that form an extended
352 Unicode sequence. \X is equivalent to
353 <pre>
354 (?&#62;\PM\pM*)
355 </pre>
356 That is, it matches a character without the "mark" property, followed by zero
357 or more characters with the "mark" property, and treats the sequence as an
358 atomic group
359 <a href="#atomicgroup">(see below).</a>
360 Characters with the "mark" property are typically accents that affect the
361 preceding character.
362 </P>
363 <P>
364 Matching characters by Unicode property is not fast, because PCRE has to search
365 a structure that contains data for over fifteen thousand characters. That is
366 why the traditional escape sequences such as \d and \w do not use Unicode
367 properties in PCRE.
368 <a name="smallassertions"></a></P>
369 <br><b>
370 Simple assertions
371 </b><br>
372 <P>
373 The fourth use of backslash is for certain simple assertions. An assertion
374 specifies a condition that has to be met at a particular point in a match,
375 without consuming any characters from the subject string. The use of
376 subpatterns for more complicated assertions is described
377 <a href="#bigassertions">below.</a>
378 The backslashed
379 assertions are:
380 <pre>
381 \b matches at a word boundary
382 \B matches when not at a word boundary
383 \A matches at start of subject
384 \Z matches at end of subject or before newline at end
385 \z matches at end of subject
386 \G matches at first matching position in subject
387 </pre>
388 These assertions may not appear in character classes (but note that \b has a
389 different meaning, namely the backspace character, inside a character class).
390 </P>
391 <P>
392 A word boundary is a position in the subject string where the current character
393 and the previous character do not both match \w or \W (i.e. one matches
394 \w and the other matches \W), or the start or end of the string if the
395 first or last character matches \w, respectively.
396 </P>
397 <P>
398 The \A, \Z, and \z assertions differ from the traditional circumflex and
399 dollar (described in the next section) in that they only ever match at the very
400 start and end of the subject string, whatever options are set. Thus, they are
401 independent of multiline mode. These three assertions are not affected by the
402 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
403 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
404 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
405 at a point other than the beginning of the subject, \A can never match. The
406 difference between \Z and \z is that \Z matches before a newline that is the
407 last character of the string as well as at the end of the string, whereas \z
408 matches only at the end.
409 </P>
410 <P>
411 The \G assertion is true only when the current matching position is at the
412 start point of the match, as specified by the <i>startoffset</i> argument of
413 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
414 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
415 arguments, you can mimic Perl's /g option, and it is in this kind of
416 implementation where \G can be useful.
417 </P>
418 <P>
419 Note, however, that PCRE's interpretation of \G, as the start of the current
420 match, is subtly different from Perl's, which defines it as the end of the
421 previous match. In Perl, these can be different when the previously matched
422 string was empty. Because PCRE does just one match at a time, it cannot
423 reproduce this behaviour.
424 </P>
425 <P>
426 If all the alternatives of a pattern begin with \G, the expression is anchored
427 to the starting match position, and the "anchored" flag is set in the compiled
428 regular expression.
429 </P>
430 <br><a name="SEC3" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
431 <P>
432 Outside a character class, in the default matching mode, the circumflex
433 character is an assertion that is true only if the current matching point is
434 at the start of the subject string. If the <i>startoffset</i> argument of
435 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
436 option is unset. Inside a character class, circumflex has an entirely different
437 meaning
438 <a href="#characterclass">(see below).</a>
439 </P>
440 <P>
441 Circumflex need not be the first character of the pattern if a number of
442 alternatives are involved, but it should be the first thing in each alternative
443 in which it appears if the pattern is ever to match that branch. If all
444 possible alternatives start with a circumflex, that is, if the pattern is
445 constrained to match only at the start of the subject, it is said to be an
446 "anchored" pattern. (There are also other constructs that can cause a pattern
447 to be anchored.)
448 </P>
449 <P>
450 A dollar character is an assertion that is true only if the current matching
451 point is at the end of the subject string, or immediately before a newline
452 character that is the last character in the string (by default). Dollar need
453 not be the last character of the pattern if a number of alternatives are
454 involved, but it should be the last item in any branch in which it appears.
455 Dollar has no special meaning in a character class.
456 </P>
457 <P>
458 The meaning of dollar can be changed so that it matches only at the very end of
459 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
460 does not affect the \Z assertion.
461 </P>
462 <P>
463 The meanings of the circumflex and dollar characters are changed if the
464 PCRE_MULTILINE option is set. When this is the case, they match immediately
465 after and immediately before an internal newline character, respectively, in
466 addition to matching at the start and end of the subject string. For example,
467 the pattern /^abc$/ matches the subject string "def\nabc" (where \n
468 represents a newline character) in multiline mode, but not otherwise.
469 Consequently, patterns that are anchored in single line mode because all
470 branches start with ^ are not anchored in multiline mode, and a match for
471 circumflex is possible when the <i>startoffset</i> argument of <b>pcre_exec()</b>
472 is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
473 set.
474 </P>
475 <P>
476 Note that the sequences \A, \Z, and \z can be used to match the start and
477 end of the subject in both modes, and if all branches of a pattern start with
478 \A it is always anchored, whether PCRE_MULTILINE is set or not.
479 </P>
480 <br><a name="SEC4" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
481 <P>
482 Outside a character class, a dot in the pattern matches any one character in
483 the subject, including a non-printing character, but not (by default) newline.
484 In UTF-8 mode, a dot matches any UTF-8 character, which might be more than one
485 byte long, except (by default) newline. If the PCRE_DOTALL option is set,
486 dots match newlines as well. The handling of dot is entirely independent of the
487 handling of circumflex and dollar, the only relationship being that they both
488 involve newline characters. Dot has no special meaning in a character class.
489 </P>
490 <br><a name="SEC5" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
491 <P>
492 Outside a character class, the escape sequence \C matches any one byte, both
493 in and out of UTF-8 mode. Unlike a dot, it can match a newline. The feature is
494 provided in Perl in order to match individual bytes in UTF-8 mode. Because it
495 breaks up UTF-8 characters into individual bytes, what remains in the string
496 may be a malformed UTF-8 string. For this reason, the \C escape sequence is
497 best avoided.
498 </P>
499 <P>
500 PCRE does not allow \C to appear in lookbehind assertions
501 <a href="#lookbehind">(described below),</a>
502 because in UTF-8 mode this would make it impossible to calculate the length of
503 the lookbehind.
504 <a name="characterclass"></a></P>
505 <br><a name="SEC6" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
506 <P>
507 An opening square bracket introduces a character class, terminated by a closing
508 square bracket. A closing square bracket on its own is not special. If a
509 closing square bracket is required as a member of the class, it should be the
510 first data character in the class (after an initial circumflex, if present) or
511 escaped with a backslash.
512 </P>
513 <P>
514 A character class matches a single character in the subject. In UTF-8 mode, the
515 character may occupy more than one byte. A matched character must be in the set
516 of characters defined by the class, unless the first character in the class
517 definition is a circumflex, in which case the subject character must not be in
518 the set defined by the class. If a circumflex is actually required as a member
519 of the class, ensure it is not the first character, or escape it with a
520 backslash.
521 </P>
522 <P>
523 For example, the character class [aeiou] matches any lower case vowel, while
524 [^aeiou] matches any character that is not a lower case vowel. Note that a
525 circumflex is just a convenient notation for specifying the characters that
526 are in the class by enumerating those that are not. A class that starts with a
527 circumflex is not an assertion: it still consumes a character from the subject
528 string, and therefore it fails if the current pointer is at the end of the
529 string.
530 </P>
531 <P>
532 In UTF-8 mode, characters with values greater than 255 can be included in a
533 class as a literal string of bytes, or by using the \x{ escaping mechanism.
534 </P>
535 <P>
536 When caseless matching is set, any letters in a class represent both their
537 upper case and lower case versions, so for example, a caseless [aeiou] matches
538 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
539 caseful version would. When running in UTF-8 mode, PCRE supports the concept of
540 case for characters with values greater than 128 only when it is compiled with
541 Unicode property support.
542 </P>
543 <P>
544 The newline character is never treated in any special way in character classes,
545 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
546 such as [^a] will always match a newline.
547 </P>
548 <P>
549 The minus (hyphen) character can be used to specify a range of characters in a
550 character class. For example, [d-m] matches any letter between d and m,
551 inclusive. If a minus character is required in a class, it must be escaped with
552 a backslash or appear in a position where it cannot be interpreted as
553 indicating a range, typically as the first or last character in the class.
554 </P>
555 <P>
556 It is not possible to have the literal character "]" as the end character of a
557 range. A pattern such as [W-]46] is interpreted as a class of two characters
558 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
559 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
560 the end of range, so [W-\]46] is interpreted as a class containing a range
561 followed by two other characters. The octal or hexadecimal representation of
562 "]" can also be used to end a range.
563 </P>
564 <P>
565 Ranges operate in the collating sequence of character values. They can also be
566 used for characters specified numerically, for example [\000-\037]. In UTF-8
567 mode, ranges can include characters whose values are greater than 255, for
568 example [\x{100}-\x{2ff}].
569 </P>
570 <P>
571 If a range that includes letters is used when caseless matching is set, it
572 matches the letters in either case. For example, [W-c] is equivalent to
573 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
574 tables for the "fr_FR" locale are in use, [\xc8-\xcb] matches accented E
575 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
576 characters with values greater than 128 only when it is compiled with Unicode
577 property support.
578 </P>
579 <P>
580 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
581 in a character class, and add the characters that they match to the class. For
582 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
583 conveniently be used with the upper case character types to specify a more
584 restricted set of characters than the matching lower case type. For example,
585 the class [^\W_] matches any letter or digit, but not underscore.
586 </P>
587 <P>
588 The only metacharacters that are recognized in character classes are backslash,
589 hyphen (only where it can be interpreted as specifying a range), circumflex
590 (only at the start), opening square bracket (only when it can be interpreted as
591 introducing a POSIX class name - see the next section), and the terminating
592 closing square bracket. However, escaping other non-alphanumeric characters
593 does no harm.
594 </P>
595 <br><a name="SEC7" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
596 <P>
597 Perl supports the POSIX notation for character classes. This uses names
598 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
599 this notation. For example,
600 <pre>
601 [01[:alpha:]%]
602 </pre>
603 matches "0", "1", any alphabetic character, or "%". The supported class names
604 are
605 <pre>
606 alnum letters and digits
607 alpha letters
608 ascii character codes 0 - 127
609 blank space or tab only
610 cntrl control characters
611 digit decimal digits (same as \d)
612 graph printing characters, excluding space
613 lower lower case letters
614 print printing characters, including space
615 punct printing characters, excluding letters and digits
616 space white space (not quite the same as \s)
617 upper upper case letters
618 word "word" characters (same as \w)
619 xdigit hexadecimal digits
620 </pre>
621 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
622 space (32). Notice that this list includes the VT character (code 11). This
623 makes "space" different to \s, which does not include VT (for Perl
624 compatibility).
625 </P>
626 <P>
627 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
628 5.8. Another Perl extension is negation, which is indicated by a ^ character
629 after the colon. For example,
630 <pre>
631 [12[:^digit:]]
632 </pre>
633 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
634 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
635 supported, and an error is given if they are encountered.
636 </P>
637 <P>
638 In UTF-8 mode, characters with values greater than 128 do not match any of
639 the POSIX character classes.
640 </P>
641 <br><a name="SEC8" href="#TOC1">VERTICAL BAR</a><br>
642 <P>
643 Vertical bar characters are used to separate alternative patterns. For example,
644 the pattern
645 <pre>
646 gilbert|sullivan
647 </pre>
648 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
649 and an empty alternative is permitted (matching the empty string).
650 The matching process tries each alternative in turn, from left to right,
651 and the first one that succeeds is used. If the alternatives are within a
652 subpattern
653 <a href="#subpattern">(defined below),</a>
654 "succeeds" means matching the rest of the main pattern as well as the
655 alternative in the subpattern.
656 </P>
657 <br><a name="SEC9" href="#TOC1">INTERNAL OPTION SETTING</a><br>
658 <P>
659 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
660 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
661 Perl option letters enclosed between "(?" and ")". The option letters are
662 <pre>
663 i for PCRE_CASELESS
664 m for PCRE_MULTILINE
665 s for PCRE_DOTALL
666 x for PCRE_EXTENDED
667 </pre>
668 For example, (?im) sets caseless, multiline matching. It is also possible to
669 unset these options by preceding the letter with a hyphen, and a combined
670 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
671 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
672 permitted. If a letter appears both before and after the hyphen, the option is
673 unset.
674 </P>
675 <P>
676 When an option change occurs at top level (that is, not inside subpattern
677 parentheses), the change applies to the remainder of the pattern that follows.
678 If the change is placed right at the start of a pattern, PCRE extracts it into
679 the global options (and it will therefore show up in data extracted by the
680 <b>pcre_fullinfo()</b> function).
681 </P>
682 <P>
683 An option change within a subpattern affects only that part of the current
684 pattern that follows it, so
685 <pre>
686 (a(?i)b)c
687 </pre>
688 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
689 By this means, options can be made to have different settings in different
690 parts of the pattern. Any changes made in one alternative do carry on
691 into subsequent branches within the same subpattern. For example,
692 <pre>
693 (a(?i)b|c)
694 </pre>
695 matches "ab", "aB", "c", and "C", even though when matching "C" the first
696 branch is abandoned before the option setting. This is because the effects of
697 option settings happen at compile time. There would be some very weird
698 behaviour otherwise.
699 </P>
700 <P>
701 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
702 same way as the Perl-compatible options by using the characters U and X
703 respectively. The (?X) flag setting is special in that it must always occur
704 earlier in the pattern than any of the additional features it turns on, even
705 when it is at top level. It is best to put it at the start.
706 <a name="subpattern"></a></P>
707 <br><a name="SEC10" href="#TOC1">SUBPATTERNS</a><br>
708 <P>
709 Subpatterns are delimited by parentheses (round brackets), which can be nested.
710 Turning part of a pattern into a subpattern does two things:
711 <br>
712 <br>
713 1. It localizes a set of alternatives. For example, the pattern
714 <pre>
715 cat(aract|erpillar|)
716 </pre>
717 matches one of the words "cat", "cataract", or "caterpillar". Without the
718 parentheses, it would match "cataract", "erpillar" or the empty string.
719 <br>
720 <br>
721 2. It sets up the subpattern as a capturing subpattern. This means that, when
722 the whole pattern matches, that portion of the subject string that matched the
723 subpattern is passed back to the caller via the <i>ovector</i> argument of
724 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
725 from 1) to obtain numbers for the capturing subpatterns.
726 </P>
727 <P>
728 For example, if the string "the red king" is matched against the pattern
729 <pre>
730 the ((red|white) (king|queen))
731 </pre>
732 the captured substrings are "red king", "red", and "king", and are numbered 1,
733 2, and 3, respectively.
734 </P>
735 <P>
736 The fact that plain parentheses fulfil two functions is not always helpful.
737 There are often times when a grouping subpattern is required without a
738 capturing requirement. If an opening parenthesis is followed by a question mark
739 and a colon, the subpattern does not do any capturing, and is not counted when
740 computing the number of any subsequent capturing subpatterns. For example, if
741 the string "the white queen" is matched against the pattern
742 <pre>
743 the ((?:red|white) (king|queen))
744 </pre>
745 the captured substrings are "white queen" and "queen", and are numbered 1 and
746 2. The maximum number of capturing subpatterns is 65535, and the maximum depth
747 of nesting of all subpatterns, both capturing and non-capturing, is 200.
748 </P>
749 <P>
750 As a convenient shorthand, if any option settings are required at the start of
751 a non-capturing subpattern, the option letters may appear between the "?" and
752 the ":". Thus the two patterns
753 <pre>
754 (?i:saturday|sunday)
755 (?:(?i)saturday|sunday)
756 </pre>
757 match exactly the same set of strings. Because alternative branches are tried
758 from left to right, and options are not reset until the end of the subpattern
759 is reached, an option setting in one branch does affect subsequent branches, so
760 the above patterns match "SUNDAY" as well as "Saturday".
761 </P>
762 <br><a name="SEC11" href="#TOC1">NAMED SUBPATTERNS</a><br>
763 <P>
764 Identifying capturing parentheses by number is simple, but it can be very hard
765 to keep track of the numbers in complicated regular expressions. Furthermore,
766 if an expression is modified, the numbers may change. To help with this
767 difficulty, PCRE supports the naming of subpatterns, something that Perl does
768 not provide. The Python syntax (?P&#60;name&#62;...) is used. Names consist of
769 alphanumeric characters and underscores, and must be unique within a pattern.
770 </P>
771 <P>
772 Named capturing parentheses are still allocated numbers as well as names. The
773 PCRE API provides function calls for extracting the name-to-number translation
774 table from a compiled pattern. There is also a convenience function for
775 extracting a captured substring by name. For further details see the
776 <a href="pcreapi.html"><b>pcreapi</b></a>
777 documentation.
778 </P>
779 <br><a name="SEC12" href="#TOC1">REPETITION</a><br>
780 <P>
781 Repetition is specified by quantifiers, which can follow any of the following
782 items:
783 <pre>
784 a literal data character
785 the . metacharacter
786 the \C escape sequence
787 the \X escape sequence (in UTF-8 mode with Unicode properties)
788 an escape such as \d that matches a single character
789 a character class
790 a back reference (see next section)
791 a parenthesized subpattern (unless it is an assertion)
792 </pre>
793 The general repetition quantifier specifies a minimum and maximum number of
794 permitted matches, by giving the two numbers in curly brackets (braces),
795 separated by a comma. The numbers must be less than 65536, and the first must
796 be less than or equal to the second. For example:
797 <pre>
798 z{2,4}
799 </pre>
800 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
801 character. If the second number is omitted, but the comma is present, there is
802 no upper limit; if the second number and the comma are both omitted, the
803 quantifier specifies an exact number of required matches. Thus
804 <pre>
805 [aeiou]{3,}
806 </pre>
807 matches at least 3 successive vowels, but may match many more, while
808 <pre>
809 \d{8}
810 </pre>
811 matches exactly 8 digits. An opening curly bracket that appears in a position
812 where a quantifier is not allowed, or one that does not match the syntax of a
813 quantifier, is taken as a literal character. For example, {,6} is not a
814 quantifier, but a literal string of four characters.
815 </P>
816 <P>
817 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
818 bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
819 which is represented by a two-byte sequence. Similarly, when Unicode property
820 support is available, \X{3} matches three Unicode extended sequences, each of
821 which may be several bytes long (and they may be of different lengths).
822 </P>
823 <P>
824 The quantifier {0} is permitted, causing the expression to behave as if the
825 previous item and the quantifier were not present.
826 </P>
827 <P>
828 For convenience (and historical compatibility) the three most common
829 quantifiers have single-character abbreviations:
830 <pre>
831 * is equivalent to {0,}
832 + is equivalent to {1,}
833 ? is equivalent to {0,1}
834 </pre>
835 It is possible to construct infinite loops by following a subpattern that can
836 match no characters with a quantifier that has no upper limit, for example:
837 <pre>
838 (a?)*
839 </pre>
840 Earlier versions of Perl and PCRE used to give an error at compile time for
841 such patterns. However, because there are cases where this can be useful, such
842 patterns are now accepted, but if any repetition of the subpattern does in fact
843 match no characters, the loop is forcibly broken.
844 </P>
845 <P>
846 By default, the quantifiers are "greedy", that is, they match as much as
847 possible (up to the maximum number of permitted times), without causing the
848 rest of the pattern to fail. The classic example of where this gives problems
849 is in trying to match comments in C programs. These appear between /* and */
850 and within the comment, individual * and / characters may appear. An attempt to
851 match C comments by applying the pattern
852 <pre>
853 /\*.*\*/
854 </pre>
855 to the string
856 <pre>
857 /* first comment */ not comment /* second comment */
858 </pre>
859 fails, because it matches the entire string owing to the greediness of the .*
860 item.
861 </P>
862 <P>
863 However, if a quantifier is followed by a question mark, it ceases to be
864 greedy, and instead matches the minimum number of times possible, so the
865 pattern
866 <pre>
867 /\*.*?\*/
868 </pre>
869 does the right thing with the C comments. The meaning of the various
870 quantifiers is not otherwise changed, just the preferred number of matches.
871 Do not confuse this use of question mark with its use as a quantifier in its
872 own right. Because it has two uses, it can sometimes appear doubled, as in
873 <pre>
874 \d??\d
875 </pre>
876 which matches one digit by preference, but can match two if that is the only
877 way the rest of the pattern matches.
878 </P>
879 <P>
880 If the PCRE_UNGREEDY option is set (an option which is not available in Perl),
881 the quantifiers are not greedy by default, but individual ones can be made
882 greedy by following them with a question mark. In other words, it inverts the
883 default behaviour.
884 </P>
885 <P>
886 When a parenthesized subpattern is quantified with a minimum repeat count that
887 is greater than 1 or with a limited maximum, more memory is required for the
888 compiled pattern, in proportion to the size of the minimum or maximum.
889 </P>
890 <P>
891 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
892 to Perl's /s) is set, thus allowing the . to match newlines, the pattern is
893 implicitly anchored, because whatever follows will be tried against every
894 character position in the subject string, so there is no point in retrying the
895 overall match at any position after the first. PCRE normally treats such a
896 pattern as though it were preceded by \A.
897 </P>
898 <P>
899 In cases where it is known that the subject string contains no newlines, it is
900 worth setting PCRE_DOTALL in order to obtain this optimization, or
901 alternatively using ^ to indicate anchoring explicitly.
902 </P>
903 <P>
904 However, there is one situation where the optimization cannot be used. When .*
905 is inside capturing parentheses that are the subject of a backreference
906 elsewhere in the pattern, a match at the start may fail, and a later one
907 succeed. Consider, for example:
908 <pre>
909 (.*)abc\1
910 </pre>
911 If the subject is "xyz123abc123" the match point is the fourth character. For
912 this reason, such a pattern is not implicitly anchored.
913 </P>
914 <P>
915 When a capturing subpattern is repeated, the value captured is the substring
916 that matched the final iteration. For example, after
917 <pre>
918 (tweedle[dume]{3}\s*)+
919 </pre>
920 has matched "tweedledum tweedledee" the value of the captured substring is
921 "tweedledee". However, if there are nested capturing subpatterns, the
922 corresponding captured values may have been set in previous iterations. For
923 example, after
924 <pre>
925 /(a|(b))+/
926 </pre>
927 matches "aba" the value of the second captured substring is "b".
928 <a name="atomicgroup"></a></P>
929 <br><a name="SEC13" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
930 <P>
931 With both maximizing and minimizing repetition, failure of what follows
932 normally causes the repeated item to be re-evaluated to see if a different
933 number of repeats allows the rest of the pattern to match. Sometimes it is
934 useful to prevent this, either to change the nature of the match, or to cause
935 it fail earlier than it otherwise might, when the author of the pattern knows
936 there is no point in carrying on.
937 </P>
938 <P>
939 Consider, for example, the pattern \d+foo when applied to the subject line
940 <pre>
941 123456bar
942 </pre>
943 After matching all 6 digits and then failing to match "foo", the normal
944 action of the matcher is to try again with only 5 digits matching the \d+
945 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
946 (a term taken from Jeffrey Friedl's book) provides the means for specifying
947 that once a subpattern has matched, it is not to be re-evaluated in this way.
948 </P>
949 <P>
950 If we use atomic grouping for the previous example, the matcher would give up
951 immediately on failing to match "foo" the first time. The notation is a kind of
952 special parenthesis, starting with (?&#62; as in this example:
953 <pre>
954 (?&#62;\d+)foo
955 </pre>
956 This kind of parenthesis "locks up" the part of the pattern it contains once
957 it has matched, and a failure further into the pattern is prevented from
958 backtracking into it. Backtracking past it to previous items, however, works as
959 normal.
960 </P>
961 <P>
962 An alternative description is that a subpattern of this type matches the string
963 of characters that an identical standalone pattern would match, if anchored at
964 the current point in the subject string.
965 </P>
966 <P>
967 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
968 the above example can be thought of as a maximizing repeat that must swallow
969 everything it can. So, while both \d+ and \d+? are prepared to adjust the
970 number of digits they match in order to make the rest of the pattern match,
971 (?&#62;\d+) can only match an entire sequence of digits.
972 </P>
973 <P>
974 Atomic groups in general can of course contain arbitrarily complicated
975 subpatterns, and can be nested. However, when the subpattern for an atomic
976 group is just a single repeated item, as in the example above, a simpler
977 notation, called a "possessive quantifier" can be used. This consists of an
978 additional + character following a quantifier. Using this notation, the
979 previous example can be rewritten as
980 <pre>
981 \d++foo
982 </pre>
983 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
984 option is ignored. They are a convenient notation for the simpler forms of
985 atomic group. However, there is no difference in the meaning or processing of a
986 possessive quantifier and the equivalent atomic group.
987 </P>
988 <P>
989 The possessive quantifier syntax is an extension to the Perl syntax. It
990 originates in Sun's Java package.
991 </P>
992 <P>
993 When a pattern contains an unlimited repeat inside a subpattern that can itself
994 be repeated an unlimited number of times, the use of an atomic group is the
995 only way to avoid some failing matches taking a very long time indeed. The
996 pattern
997 <pre>
998 (\D+|&#60;\d+&#62;)*[!?]
999 </pre>
1000 matches an unlimited number of substrings that either consist of non-digits, or
1001 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1002 quickly. However, if it is applied to
1003 <pre>
1004 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1005 </pre>
1006 it takes a long time before reporting failure. This is because the string can
1007 be divided between the internal \D+ repeat and the external * repeat in a
1008 large number of ways, and all have to be tried. (The example uses [!?] rather
1009 than a single character at the end, because both PCRE and Perl have an
1010 optimization that allows for fast failure when a single character is used. They
1011 remember the last single character that is required for a match, and fail early
1012 if it is not present in the string.) If the pattern is changed so that it uses
1013 an atomic group, like this:
1014 <pre>
1015 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1016 </pre>
1017 sequences of non-digits cannot be broken, and failure happens quickly.
1018 <a name="backreferences"></a></P>
1019 <br><a name="SEC14" href="#TOC1">BACK REFERENCES</a><br>
1020 <P>
1021 Outside a character class, a backslash followed by a digit greater than 0 (and
1022 possibly further digits) is a back reference to a capturing subpattern earlier
1023 (that is, to its left) in the pattern, provided there have been that many
1024 previous capturing left parentheses.
1025 </P>
1026 <P>
1027 However, if the decimal number following the backslash is less than 10, it is
1028 always taken as a back reference, and causes an error only if there are not
1029 that many capturing left parentheses in the entire pattern. In other words, the
1030 parentheses that are referenced need not be to the left of the reference for
1031 numbers less than 10. See the subsection entitled "Non-printing characters"
1032 <a href="#digitsafterbackslash">above</a>
1033 for further details of the handling of digits following a backslash.
1034 </P>
1035 <P>
1036 A back reference matches whatever actually matched the capturing subpattern in
1037 the current subject string, rather than anything matching the subpattern
1038 itself (see
1039 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1040 below for a way of doing that). So the pattern
1041 <pre>
1042 (sens|respons)e and \1ibility
1043 </pre>
1044 matches "sense and sensibility" and "response and responsibility", but not
1045 "sense and responsibility". If caseful matching is in force at the time of the
1046 back reference, the case of letters is relevant. For example,
1047 <pre>
1048 ((?i)rah)\s+\1
1049 </pre>
1050 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1051 capturing subpattern is matched caselessly.
1052 </P>
1053 <P>
1054 Back references to named subpatterns use the Python syntax (?P=name). We could
1055 rewrite the above example as follows:
1056 <pre>
1057 (?&#60;p1&#62;(?i)rah)\s+(?P=p1)
1058 </pre>
1059 There may be more than one back reference to the same subpattern. If a
1060 subpattern has not actually been used in a particular match, any back
1061 references to it always fail. For example, the pattern
1062 <pre>
1063 (a|(bc))\2
1064 </pre>
1065 always fails if it starts to match "a" rather than "bc". Because there may be
1066 many capturing parentheses in a pattern, all digits following the backslash are
1067 taken as part of a potential back reference number. If the pattern continues
1068 with a digit character, some delimiter must be used to terminate the back
1069 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1070 Otherwise an empty comment (see
1071 <a href="#comments">"Comments"</a>
1072 below) can be used.
1073 </P>
1074 <P>
1075 A back reference that occurs inside the parentheses to which it refers fails
1076 when the subpattern is first used, so, for example, (a\1) never matches.
1077 However, such references can be useful inside repeated subpatterns. For
1078 example, the pattern
1079 <pre>
1080 (a|b\1)+
1081 </pre>
1082 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1083 the subpattern, the back reference matches the character string corresponding
1084 to the previous iteration. In order for this to work, the pattern must be such
1085 that the first iteration does not need to match the back reference. This can be
1086 done using alternation, as in the example above, or by a quantifier with a
1087 minimum of zero.
1088 <a name="bigassertions"></a></P>
1089 <br><a name="SEC15" href="#TOC1">ASSERTIONS</a><br>
1090 <P>
1091 An assertion is a test on the characters following or preceding the current
1092 matching point that does not actually consume any characters. The simple
1093 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1094 <a href="#smallassertions">above.</a>
1095 </P>
1096 <P>
1097 More complicated assertions are coded as subpatterns. There are two kinds:
1098 those that look ahead of the current position in the subject string, and those
1099 that look behind it. An assertion subpattern is matched in the normal way,
1100 except that it does not cause the current matching position to be changed.
1101 </P>
1102 <P>
1103 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1104 because it makes no sense to assert the same thing several times. If any kind
1105 of assertion contains capturing subpatterns within it, these are counted for
1106 the purposes of numbering the capturing subpatterns in the whole pattern.
1107 However, substring capturing is carried out only for positive assertions,
1108 because it does not make sense for negative assertions.
1109 </P>
1110 <br><b>
1111 Lookahead assertions
1112 </b><br>
1113 <P>
1114 Lookahead assertions start
1115 with (?= for positive assertions and (?! for negative assertions. For example,
1116 <pre>
1117 \w+(?=;)
1118 </pre>
1119 matches a word followed by a semicolon, but does not include the semicolon in
1120 the match, and
1121 <pre>
1122 foo(?!bar)
1123 </pre>
1124 matches any occurrence of "foo" that is not followed by "bar". Note that the
1125 apparently similar pattern
1126 <pre>
1127 (?!foo)bar
1128 </pre>
1129 does not find an occurrence of "bar" that is preceded by something other than
1130 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1131 (?!foo) is always true when the next three characters are "bar". A
1132 lookbehind assertion is needed to achieve the other effect.
1133 </P>
1134 <P>
1135 If you want to force a matching failure at some point in a pattern, the most
1136 convenient way to do it is with (?!) because an empty string always matches, so
1137 an assertion that requires there not to be an empty string must always fail.
1138 <a name="lookbehind"></a></P>
1139 <br><b>
1140 Lookbehind assertions
1141 </b><br>
1142 <P>
1143 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1144 negative assertions. For example,
1145 <pre>
1146 (?&#60;!foo)bar
1147 </pre>
1148 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1149 a lookbehind assertion are restricted such that all the strings it matches must
1150 have a fixed length. However, if there are several alternatives, they do not
1151 all have to have the same fixed length. Thus
1152 <pre>
1153 (?&#60;=bullock|donkey)
1154 </pre>
1155 is permitted, but
1156 <pre>
1157 (?&#60;!dogs?|cats?)
1158 </pre>
1159 causes an error at compile time. Branches that match different length strings
1160 are permitted only at the top level of a lookbehind assertion. This is an
1161 extension compared with Perl (at least for 5.8), which requires all branches to
1162 match the same length of string. An assertion such as
1163 <pre>
1164 (?&#60;=ab(c|de))
1165 </pre>
1166 is not permitted, because its single top-level branch can match two different
1167 lengths, but it is acceptable if rewritten to use two top-level branches:
1168 <pre>
1169 (?&#60;=abc|abde)
1170 </pre>
1171 The implementation of lookbehind assertions is, for each alternative, to
1172 temporarily move the current position back by the fixed width and then try to
1173 match. If there are insufficient characters before the current position, the
1174 match is deemed to fail.
1175 </P>
1176 <P>
1177 PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1178 to appear in lookbehind assertions, because it makes it impossible to calculate
1179 the length of the lookbehind. The \X escape, which can match different numbers
1180 of bytes, is also not permitted.
1181 </P>
1182 <P>
1183 Atomic groups can be used in conjunction with lookbehind assertions to specify
1184 efficient matching at the end of the subject string. Consider a simple pattern
1185 such as
1186 <pre>
1187 abcd$
1188 </pre>
1189 when applied to a long string that does not match. Because matching proceeds
1190 from left to right, PCRE will look for each "a" in the subject and then see if
1191 what follows matches the rest of the pattern. If the pattern is specified as
1192 <pre>
1193 ^.*abcd$
1194 </pre>
1195 the initial .* matches the entire string at first, but when this fails (because
1196 there is no following "a"), it backtracks to match all but the last character,
1197 then all but the last two characters, and so on. Once again the search for "a"
1198 covers the entire string, from right to left, so we are no better off. However,
1199 if the pattern is written as
1200 <pre>
1201 ^(?&#62;.*)(?&#60;=abcd)
1202 </pre>
1203 or, equivalently, using the possessive quantifier syntax,
1204 <pre>
1205 ^.*+(?&#60;=abcd)
1206 </pre>
1207 there can be no backtracking for the .* item; it can match only the entire
1208 string. The subsequent lookbehind assertion does a single test on the last four
1209 characters. If it fails, the match fails immediately. For long strings, this
1210 approach makes a significant difference to the processing time.
1211 </P>
1212 <br><b>
1213 Using multiple assertions
1214 </b><br>
1215 <P>
1216 Several assertions (of any sort) may occur in succession. For example,
1217 <pre>
1218 (?&#60;=\d{3})(?&#60;!999)foo
1219 </pre>
1220 matches "foo" preceded by three digits that are not "999". Notice that each of
1221 the assertions is applied independently at the same point in the subject
1222 string. First there is a check that the previous three characters are all
1223 digits, and then there is a check that the same three characters are not "999".
1224 This pattern does <i>not</i> match "foo" preceded by six characters, the first
1225 of which are digits and the last three of which are not "999". For example, it
1226 doesn't match "123abcfoo". A pattern to do that is
1227 <pre>
1228 (?&#60;=\d{3}...)(?&#60;!999)foo
1229 </pre>
1230 This time the first assertion looks at the preceding six characters, checking
1231 that the first three are digits, and then the second assertion checks that the
1232 preceding three characters are not "999".
1233 </P>
1234 <P>
1235 Assertions can be nested in any combination. For example,
1236 <pre>
1237 (?&#60;=(?&#60;!foo)bar)baz
1238 </pre>
1239 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1240 preceded by "foo", while
1241 <pre>
1242 (?&#60;=\d{3}(?!999)...)foo
1243 </pre>
1244 is another pattern that matches "foo" preceded by three digits and any three
1245 characters that are not "999".
1246 </P>
1247 <br><a name="SEC16" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1248 <P>
1249 It is possible to cause the matching process to obey a subpattern
1250 conditionally or to choose between two alternative subpatterns, depending on
1251 the result of an assertion, or whether a previous capturing subpattern matched
1252 or not. The two possible forms of conditional subpattern are
1253 <pre>
1254 (?(condition)yes-pattern)
1255 (?(condition)yes-pattern|no-pattern)
1256 </pre>
1257 If the condition is satisfied, the yes-pattern is used; otherwise the
1258 no-pattern (if present) is used. If there are more than two alternatives in the
1259 subpattern, a compile-time error occurs.
1260 </P>
1261 <P>
1262 There are three kinds of condition. If the text between the parentheses
1263 consists of a sequence of digits, the condition is satisfied if the capturing
1264 subpattern of that number has previously matched. The number must be greater
1265 than zero. Consider the following pattern, which contains non-significant white
1266 space to make it more readable (assume the PCRE_EXTENDED option) and to divide
1267 it into three parts for ease of discussion:
1268 <pre>
1269 ( \( )? [^()]+ (?(1) \) )
1270 </pre>
1271 The first part matches an optional opening parenthesis, and if that
1272 character is present, sets it as the first captured substring. The second part
1273 matches one or more characters that are not parentheses. The third part is a
1274 conditional subpattern that tests whether the first set of parentheses matched
1275 or not. If they did, that is, if subject started with an opening parenthesis,
1276 the condition is true, and so the yes-pattern is executed and a closing
1277 parenthesis is required. Otherwise, since no-pattern is not present, the
1278 subpattern matches nothing. In other words, this pattern matches a sequence of
1279 non-parentheses, optionally enclosed in parentheses.
1280 </P>
1281 <P>
1282 If the condition is the string (R), it is satisfied if a recursive call to the
1283 pattern or subpattern has been made. At "top level", the condition is false.
1284 This is a PCRE extension. Recursive patterns are described in the next section.
1285 </P>
1286 <P>
1287 If the condition is not a sequence of digits or (R), it must be an assertion.
1288 This may be a positive or negative lookahead or lookbehind assertion. Consider
1289 this pattern, again containing non-significant white space, and with the two
1290 alternatives on the second line:
1291 <pre>
1292 (?(?=[^a-z]*[a-z])
1293 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1294 </pre>
1295 The condition is a positive lookahead assertion that matches an optional
1296 sequence of non-letters followed by a letter. In other words, it tests for the
1297 presence of at least one letter in the subject. If a letter is found, the
1298 subject is matched against the first alternative; otherwise it is matched
1299 against the second. This pattern matches strings in one of the two forms
1300 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1301 <a name="comments"></a></P>
1302 <br><a name="SEC17" href="#TOC1">COMMENTS</a><br>
1303 <P>
1304 The sequence (?# marks the start of a comment that continues up to the next
1305 closing parenthesis. Nested parentheses are not permitted. The characters
1306 that make up a comment play no part in the pattern matching at all.
1307 </P>
1308 <P>
1309 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1310 character class introduces a comment that continues up to the next newline
1311 character in the pattern.
1312 </P>
1313 <br><a name="SEC18" href="#TOC1">RECURSIVE PATTERNS</a><br>
1314 <P>
1315 Consider the problem of matching a string in parentheses, allowing for
1316 unlimited nested parentheses. Without the use of recursion, the best that can
1317 be done is to use a pattern that matches up to some fixed depth of nesting. It
1318 is not possible to handle an arbitrary nesting depth. Perl provides a facility
1319 that allows regular expressions to recurse (amongst other things). It does this
1320 by interpolating Perl code in the expression at run time, and the code can
1321 refer to the expression itself. A Perl pattern to solve the parentheses problem
1322 can be created like this:
1323 <pre>
1324 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
1325 </pre>
1326 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1327 recursively to the pattern in which it appears. Obviously, PCRE cannot support
1328 the interpolation of Perl code. Instead, it supports some special syntax for
1329 recursion of the entire pattern, and also for individual subpattern recursion.
1330 </P>
1331 <P>
1332 The special item that consists of (? followed by a number greater than zero and
1333 a closing parenthesis is a recursive call of the subpattern of the given
1334 number, provided that it occurs inside that subpattern. (If not, it is a
1335 "subroutine" call, which is described in the next section.) The special item
1336 (?R) is a recursive call of the entire regular expression.
1337 </P>
1338 <P>
1339 For example, this PCRE pattern solves the nested parentheses problem (assume
1340 the PCRE_EXTENDED option is set so that white space is ignored):
1341 <pre>
1342 \( ( (?&#62;[^()]+) | (?R) )* \)
1343 </pre>
1344 First it matches an opening parenthesis. Then it matches any number of
1345 substrings which can either be a sequence of non-parentheses, or a recursive
1346 match of the pattern itself (that is a correctly parenthesized substring).
1347 Finally there is a closing parenthesis.
1348 </P>
1349 <P>
1350 If this were part of a larger pattern, you would not want to recurse the entire
1351 pattern, so instead you could use this:
1352 <pre>
1353 ( \( ( (?&#62;[^()]+) | (?1) )* \) )
1354 </pre>
1355 We have put the pattern into parentheses, and caused the recursion to refer to
1356 them instead of the whole pattern. In a larger pattern, keeping track of
1357 parenthesis numbers can be tricky. It may be more convenient to use named
1358 parentheses instead. For this, PCRE uses (?P&#62;name), which is an extension to
1359 the Python syntax that PCRE uses for named parentheses (Perl does not provide
1360 named parentheses). We could rewrite the above example as follows:
1361 <pre>
1362 (?P&#60;pn&#62; \( ( (?&#62;[^()]+) | (?P&#62;pn) )* \) )
1363 </pre>
1364 This particular example pattern contains nested unlimited repeats, and so the
1365 use of atomic grouping for matching strings of non-parentheses is important
1366 when applying the pattern to strings that do not match. For example, when this
1367 pattern is applied to
1368 <pre>
1369 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1370 </pre>
1371 it yields "no match" quickly. However, if atomic grouping is not used,
1372 the match runs for a very long time indeed because there are so many different
1373 ways the + and * repeats can carve up the subject, and all have to be tested
1374 before failure can be reported.
1375 </P>
1376 <P>
1377 At the end of a match, the values set for any capturing subpatterns are those
1378 from the outermost level of the recursion at which the subpattern value is set.
1379 If you want to obtain intermediate values, a callout function can be used (see
1380 the next section and the
1381 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1382 documentation). If the pattern above is matched against
1383 <pre>
1384 (ab(cd)ef)
1385 </pre>
1386 the value for the capturing parentheses is "ef", which is the last value taken
1387 on at the top level. If additional parentheses are added, giving
1388 <pre>
1389 \( ( ( (?&#62;[^()]+) | (?R) )* ) \)
1390 ^ ^
1391 ^ ^
1392 </pre>
1393 the string they capture is "ab(cd)ef", the contents of the top level
1394 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1395 has to obtain extra memory to store data during a recursion, which it does by
1396 using <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no
1397 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1398 </P>
1399 <P>
1400 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1401 Consider this pattern, which matches text in angle brackets, allowing for
1402 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1403 recursing), whereas any characters are permitted at the outer level.
1404 <pre>
1405 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
1406 </pre>
1407 In this pattern, (?(R) is the start of a conditional subpattern, with two
1408 different alternatives for the recursive and non-recursive cases. The (?R) item
1409 is the actual recursive call.
1410 <a name="subpatternsassubroutines"></a></P>
1411 <br><a name="SEC19" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
1412 <P>
1413 If the syntax for a recursive subpattern reference (either by number or by
1414 name) is used outside the parentheses to which it refers, it operates like a
1415 subroutine in a programming language. An earlier example pointed out that the
1416 pattern
1417 <pre>
1418 (sens|respons)e and \1ibility
1419 </pre>
1420 matches "sense and sensibility" and "response and responsibility", but not
1421 "sense and responsibility". If instead the pattern
1422 <pre>
1423 (sens|respons)e and (?1)ibility
1424 </pre>
1425 is used, it does match "sense and responsibility" as well as the other two
1426 strings. Such references must, however, follow the subpattern to which they
1427 refer.
1428 </P>
1429 <br><a name="SEC20" href="#TOC1">CALLOUTS</a><br>
1430 <P>
1431 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1432 code to be obeyed in the middle of matching a regular expression. This makes it
1433 possible, amongst other things, to extract different substrings that match the
1434 same pair of parentheses when there is a repetition.
1435 </P>
1436 <P>
1437 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1438 code. The feature is called "callout". The caller of PCRE provides an external
1439 function by putting its entry point in the global variable <i>pcre_callout</i>.
1440 By default, this variable contains NULL, which disables all calling out.
1441 </P>
1442 <P>
1443 Within a regular expression, (?C) indicates the points at which the external
1444 function is to be called. If you want to identify different callout points, you
1445 can put a number less than 256 after the letter C. The default value is zero.
1446 For example, this pattern has two callout points:
1447 <pre>
1448 (?C1)\dabc(?C2)def
1449 </pre>
1450 If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
1451 automatically installed before each item in the pattern. They are all numbered
1452 255.
1453 </P>
1454 <P>
1455 During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
1456 set), the external function is called. It is provided with the number of the
1457 callout, the position in the pattern, and, optionally, one item of data
1458 originally supplied by the caller of <b>pcre_exec()</b>. The callout function
1459 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1460 description of the interface to the callout function is given in the
1461 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1462 documentation.
1463 </P>
1464 <P>
1465 Last updated: 09 September 2004
1466 <br>
1467 Copyright &copy; 1997-2004 University of Cambridge.
1468 <p>
1469 Return to the <a href="index.html">PCRE index page</a>.
1470 </p>

  ViewVC Help
Powered by ViewVC 1.1.5