ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 87 - (show annotations)
Sat Feb 24 21:41:21 2007 UTC (14 years ago) by nigel
File size: 65123 byte(s)
Error occurred while calculating annotation data.
Load pcre-6.5 into code/trunk.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions supported by PCRE are
8 described below. Regular expressions are also described in the Perl
9 documentation and in a number of books, some of which have copious examples.
10 Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers
11 regular expressions in great detail. This description of PCRE's regular
12 expressions is intended as reference material.
13 .P
14 The original operation of PCRE was on strings of one-byte characters. However,
15 there is now also support for UTF-8 character strings. To use this, you must
16 build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
17 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
18 places below. There is also a summary of UTF-8 features in the
19 .\" HTML <a href="pcre.html#utf8support">
20 .\" </a>
21 section on UTF-8 support
22 .\"
23 in the main
24 .\" HREF
25 \fBpcre\fP
26 .\"
27 page.
28 .P
29 The remainder of this document discusses the patterns that are supported by
30 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
31 From release 6.0, PCRE offers a second matching function,
32 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
33 Perl-compatible. The advantages and disadvantages of the alternative function,
34 and how it differs from the normal function, are discussed in the
35 .\" HREF
36 \fBpcrematching\fP
37 .\"
38 page.
39 .P
40 A regular expression is a pattern that is matched against a subject string from
41 left to right. Most characters stand for themselves in a pattern, and match the
42 corresponding characters in the subject. As a trivial example, the pattern
43 .sp
44 The quick brown fox
45 .sp
46 matches a portion of a subject string that is identical to itself. When
47 caseless matching is specified (the PCRE_CASELESS option), letters are matched
48 independently of case. In UTF-8 mode, PCRE always understands the concept of
49 case for characters whose values are less than 128, so caseless matching is
50 always possible. For characters with higher values, the concept of case is
51 supported if PCRE is compiled with Unicode property support, but not otherwise.
52 If you want to use caseless matching for characters 128 and above, you must
53 ensure that PCRE is compiled with Unicode property support as well as with
54 UTF-8 support.
55 .P
56 The power of regular expressions comes from the ability to include alternatives
57 and repetitions in the pattern. These are encoded in the pattern by the use of
58 \fImetacharacters\fP, which do not stand for themselves but instead are
59 interpreted in some special way.
60 .P
61 There are two different sets of metacharacters: those that are recognized
62 anywhere in the pattern except within square brackets, and those that are
63 recognized in square brackets. Outside square brackets, the metacharacters are
64 as follows:
65 .sp
66 \e general escape character with several uses
67 ^ assert start of string (or line, in multiline mode)
68 $ assert end of string (or line, in multiline mode)
69 . match any character except newline (by default)
70 [ start character class definition
71 | start of alternative branch
72 ( start subpattern
73 ) end subpattern
74 ? extends the meaning of (
75 also 0 or 1 quantifier
76 also quantifier minimizer
77 * 0 or more quantifier
78 + 1 or more quantifier
79 also "possessive quantifier"
80 { start min/max quantifier
81 .sp
82 Part of a pattern that is in square brackets is called a "character class". In
83 a character class the only metacharacters are:
84 .sp
85 \e general escape character
86 ^ negate the class, but only if the first character
87 - indicates character range
88 .\" JOIN
89 [ POSIX character class (only if followed by POSIX
90 syntax)
91 ] terminates the character class
92 .sp
93 The following sections describe the use of each of the metacharacters.
94 .
96 .rs
97 .sp
98 The backslash character has several uses. Firstly, if it is followed by a
99 non-alphanumeric character, it takes away any special meaning that character may
100 have. This use of backslash as an escape character applies both inside and
101 outside character classes.
102 .P
103 For example, if you want to match a * character, you write \e* in the pattern.
104 This escaping action applies whether or not the following character would
105 otherwise be interpreted as a metacharacter, so it is always safe to precede a
106 non-alphanumeric with backslash to specify that it stands for itself. In
107 particular, if you want to match a backslash, you write \e\e.
108 .P
109 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
110 pattern (other than in a character class) and characters between a # outside
111 a character class and the next newline character are ignored. An escaping
112 backslash can be used to include a whitespace or # character as part of the
113 pattern.
114 .P
115 If you want to remove the special meaning from a sequence of characters, you
116 can do so by putting them between \eQ and \eE. This is different from Perl in
117 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
118 Perl, $ and @ cause variable interpolation. Note the following examples:
119 .sp
120 Pattern PCRE matches Perl matches
121 .sp
122 .\" JOIN
123 \eQabc$xyz\eE abc$xyz abc followed by the
124 contents of $xyz
125 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
126 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
127 .sp
128 The \eQ...\eE sequence is recognized both inside and outside character classes.
129 .
130 .
131 .\" HTML <a name="digitsafterbackslash"></a>
132 .SS "Non-printing characters"
133 .rs
134 .sp
135 A second use of backslash provides a way of encoding non-printing characters
136 in patterns in a visible manner. There is no restriction on the appearance of
137 non-printing characters, apart from the binary zero that terminates a pattern,
138 but when a pattern is being prepared by text editing, it is usually easier to
139 use one of the following escape sequences than the binary character it
140 represents:
141 .sp
142 \ea alarm, that is, the BEL character (hex 07)
143 \ecx "control-x", where x is any character
144 \ee escape (hex 1B)
145 \ef formfeed (hex 0C)
146 \en newline (hex 0A)
147 \er carriage return (hex 0D)
148 \et tab (hex 09)
149 \eddd character with octal code ddd, or backreference
150 \exhh character with hex code hh
151 \ex{hhh..} character with hex code hhh..
152 .sp
153 The precise effect of \ecx is as follows: if x is a lower case letter, it
154 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
155 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
156 7B.
157 .P
158 After \ex, from zero to two hexadecimal digits are read (letters can be in
159 upper or lower case). Any number of hexadecimal digits may appear between \ex{
160 and }, but the value of the character code must be less than 256 in non-UTF-8
161 mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
162 is 7FFFFFFF). If characters other than hexadecimal digits appear between \ex{
163 and }, or if there is no terminating }, this form of escape is not recognized.
164 Instead, the initial \ex will be interpreted as a basic hexadecimal escape,
165 with no following digits, giving a character whose value is zero.
166 .P
167 Characters whose value is less than 256 can be defined by either of the two
168 syntaxes for \ex. There is no difference in the way they are handled. For
169 example, \exdc is exactly the same as \ex{dc}.
170 .P
171 After \e0 up to two further octal digits are read. In both cases, if there
172 are fewer than two digits, just those that are present are used. Thus the
173 sequence \e0\ex\e07 specifies two binary zeros followed by a BEL character
174 (code value 7). Make sure you supply two digits after the initial zero if the
175 pattern character that follows is itself an octal digit.
176 .P
177 The handling of a backslash followed by a digit other than 0 is complicated.
178 Outside a character class, PCRE reads it and any following digits as a decimal
179 number. If the number is less than 10, or if there have been at least that many
180 previous capturing left parentheses in the expression, the entire sequence is
181 taken as a \fIback reference\fP. A description of how this works is given
182 .\" HTML <a href="#backreferences">
183 .\" </a>
184 later,
185 .\"
186 following the discussion of
187 .\" HTML <a href="#subpattern">
188 .\" </a>
189 parenthesized subpatterns.
190 .\"
191 .P
192 Inside a character class, or if the decimal number is greater than 9 and there
193 have not been that many capturing subpatterns, PCRE re-reads up to three octal
194 digits following the backslash, and generates a single byte from the least
195 significant 8 bits of the value. Any subsequent digits stand for themselves.
196 For example:
197 .sp
198 \e040 is another way of writing a space
199 .\" JOIN
200 \e40 is the same, provided there are fewer than 40
201 previous capturing subpatterns
202 \e7 is always a back reference
203 .\" JOIN
204 \e11 might be a back reference, or another way of
205 writing a tab
206 \e011 is always a tab
207 \e0113 is a tab followed by the character "3"
208 .\" JOIN
209 \e113 might be a back reference, otherwise the
210 character with octal code 113
211 .\" JOIN
212 \e377 might be a back reference, otherwise
213 the byte consisting entirely of 1 bits
214 .\" JOIN
215 \e81 is either a back reference, or a binary zero
216 followed by the two characters "8" and "1"
217 .sp
218 Note that octal values of 100 or greater must not be introduced by a leading
219 zero, because no more than three octal digits are ever read.
220 .P
221 All the sequences that define a single byte value or a single UTF-8 character
222 (in UTF-8 mode) can be used both inside and outside character classes. In
223 addition, inside a character class, the sequence \eb is interpreted as the
224 backspace character (hex 08), and the sequence \eX is interpreted as the
225 character "X". Outside a character class, these sequences have different
226 meanings
227 .\" HTML <a href="#uniextseq">
228 .\" </a>
229 (see below).
230 .\"
231 .
232 .
233 .SS "Generic character types"
234 .rs
235 .sp
236 The third use of backslash is for specifying generic character types. The
237 following are always recognized:
238 .sp
239 \ed any decimal digit
240 \eD any character that is not a decimal digit
241 \es any whitespace character
242 \eS any character that is not a whitespace character
243 \ew any "word" character
244 \eW any "non-word" character
245 .sp
246 Each pair of escape sequences partitions the complete set of characters into
247 two disjoint sets. Any given character matches one, and only one, of each pair.
248 .P
249 These character type sequences can appear both inside and outside character
250 classes. They each match one character of the appropriate type. If the current
251 matching point is at the end of the subject string, all of them fail, since
252 there is no character to match.
253 .P
254 For compatibility with Perl, \es does not match the VT character (code 11).
255 This makes it different from the the POSIX "space" class. The \es characters
256 are HT (9), LF (10), FF (12), CR (13), and space (32).
257 .P
258 A "word" character is an underscore or any character less than 256 that is a
259 letter or digit. The definition of letters and digits is controlled by PCRE's
260 low-valued character tables, and may vary if locale-specific matching is taking
261 place (see
262 .\" HTML <a href="pcreapi.html#localesupport">
263 .\" </a>
264 "Locale support"
265 .\"
266 in the
267 .\" HREF
268 \fBpcreapi\fP
269 .\"
270 page). For example, in the "fr_FR" (French) locale, some character codes
271 greater than 128 are used for accented letters, and these are matched by \ew.
272 .P
273 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
274 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
275 character property support is available. The use of locales with Unicode is
276 discouraged.
277 .
278 .
279 .\" HTML <a name="uniextseq"></a>
280 .SS Unicode character properties
281 .rs
282 .sp
283 When PCRE is built with Unicode character property support, three additional
284 escape sequences to match character properties are available when UTF-8 mode
285 is selected. They are:
286 .sp
287 \ep{\fIxx\fP} a character with the \fIxx\fP property
288 \eP{\fIxx\fP} a character without the \fIxx\fP property
289 \eX an extended Unicode sequence
290 .sp
291 The property names represented by \fIxx\fP above are limited to the Unicode
292 script names, the general category properties, and "Any", which matches any
293 character (including newline). Other properties such as "InMusicalSymbols" are
294 not currently supported by PCRE. Note that \eP{Any} does not match any
295 characters, so always causes a match failure.
296 .P
297 Sets of Unicode characters are defined as belonging to certain scripts. A
298 character from one of these sets can be matched using a script name. For
299 example:
300 .sp
301 \ep{Greek}
302 \eP{Han}
303 .sp
304 Those that are not part of an identified script are lumped together as
305 "Common". The current list of scripts is:
306 .P
307 Arabic,
308 Armenian,
309 Bengali,
310 Bopomofo,
311 Braille,
312 Buginese,
313 Buhid,
314 Canadian_Aboriginal,
315 Cherokee,
316 Common,
317 Coptic,
318 Cypriot,
319 Cyrillic,
320 Deseret,
321 Devanagari,
322 Ethiopic,
323 Georgian,
324 Glagolitic,
325 Gothic,
326 Greek,
327 Gujarati,
328 Gurmukhi,
329 Han,
330 Hangul,
331 Hanunoo,
332 Hebrew,
333 Hiragana,
334 Inherited,
335 Kannada,
336 Katakana,
337 Kharoshthi,
338 Khmer,
339 Lao,
340 Latin,
341 Limbu,
342 Linear_B,
343 Malayalam,
344 Mongolian,
345 Myanmar,
346 New_Tai_Lue,
347 Ogham,
348 Old_Italic,
349 Old_Persian,
350 Oriya,
351 Osmanya,
352 Runic,
353 Shavian,
354 Sinhala,
355 Syloti_Nagri,
356 Syriac,
357 Tagalog,
358 Tagbanwa,
359 Tai_Le,
360 Tamil,
361 Telugu,
362 Thaana,
363 Thai,
364 Tibetan,
365 Tifinagh,
366 Ugaritic,
367 Yi.
368 .P
369 Each character has exactly one general category property, specified by a
370 two-letter abbreviation. For compatibility with Perl, negation can be specified
371 by including a circumflex between the opening brace and the property name. For
372 example, \ep{^Lu} is the same as \eP{Lu}.
373 .P
374 If only one letter is specified with \ep or \eP, it includes all the general
375 category properties that start with that letter. In this case, in the absence
376 of negation, the curly brackets in the escape sequence are optional; these two
377 examples have the same effect:
378 .sp
379 \ep{L}
380 \epL
381 .sp
382 The following general category property codes are supported:
383 .sp
384 C Other
385 Cc Control
386 Cf Format
387 Cn Unassigned
388 Co Private use
389 Cs Surrogate
390 .sp
391 L Letter
392 Ll Lower case letter
393 Lm Modifier letter
394 Lo Other letter
395 Lt Title case letter
396 Lu Upper case letter
397 .sp
398 M Mark
399 Mc Spacing mark
400 Me Enclosing mark
401 Mn Non-spacing mark
402 .sp
403 N Number
404 Nd Decimal number
405 Nl Letter number
406 No Other number
407 .sp
408 P Punctuation
409 Pc Connector punctuation
410 Pd Dash punctuation
411 Pe Close punctuation
412 Pf Final punctuation
413 Pi Initial punctuation
414 Po Other punctuation
415 Ps Open punctuation
416 .sp
417 S Symbol
418 Sc Currency symbol
419 Sk Modifier symbol
420 Sm Mathematical symbol
421 So Other symbol
422 .sp
423 Z Separator
424 Zl Line separator
425 Zp Paragraph separator
426 Zs Space separator
427 .sp
428 The special property L& is also supported: it matches a character that has
429 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
430 a modifier or "other".
431 .P
432 The long synonyms for these properties that Perl supports (such as \ep{Letter})
433 are not supported by PCRE. Nor is is permitted to prefix any of these
434 properties with "Is".
435 .P
436 No character that is in the Unicode table has the Cn (unassigned) property.
437 Instead, this property is assumed for any code point that is not in the
438 Unicode table.
439 .P
440 Specifying caseless matching does not affect these escape sequences. For
441 example, \ep{Lu} always matches only upper case letters.
442 .P
443 The \eX escape matches any number of Unicode characters that form an extended
444 Unicode sequence. \eX is equivalent to
445 .sp
446 (?>\ePM\epM*)
447 .sp
448 That is, it matches a character without the "mark" property, followed by zero
449 or more characters with the "mark" property, and treats the sequence as an
450 atomic group
451 .\" HTML <a href="#atomicgroup">
452 .\" </a>
453 (see below).
454 .\"
455 Characters with the "mark" property are typically accents that affect the
456 preceding character.
457 .P
458 Matching characters by Unicode property is not fast, because PCRE has to search
459 a structure that contains data for over fifteen thousand characters. That is
460 why the traditional escape sequences such as \ed and \ew do not use Unicode
461 properties in PCRE.
462 .
463 .
464 .\" HTML <a name="smallassertions"></a>
465 .SS "Simple assertions"
466 .rs
467 .sp
468 The fourth use of backslash is for certain simple assertions. An assertion
469 specifies a condition that has to be met at a particular point in a match,
470 without consuming any characters from the subject string. The use of
471 subpatterns for more complicated assertions is described
472 .\" HTML <a href="#bigassertions">
473 .\" </a>
474 below.
475 .\"
476 The backslashed
477 assertions are:
478 .sp
479 \eb matches at a word boundary
480 \eB matches when not at a word boundary
481 \eA matches at start of subject
482 \eZ matches at end of subject or before newline at end
483 \ez matches at end of subject
484 \eG matches at first matching position in subject
485 .sp
486 These assertions may not appear in character classes (but note that \eb has a
487 different meaning, namely the backspace character, inside a character class).
488 .P
489 A word boundary is a position in the subject string where the current character
490 and the previous character do not both match \ew or \eW (i.e. one matches
491 \ew and the other matches \eW), or the start or end of the string if the
492 first or last character matches \ew, respectively.
493 .P
494 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
495 dollar (described in the next section) in that they only ever match at the very
496 start and end of the subject string, whatever options are set. Thus, they are
497 independent of multiline mode. These three assertions are not affected by the
498 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
499 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
500 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
501 at a point other than the beginning of the subject, \eA can never match. The
502 difference between \eZ and \ez is that \eZ matches before a newline that is the
503 last character of the string as well as at the end of the string, whereas \ez
504 matches only at the end.
505 .P
506 The \eG assertion is true only when the current matching position is at the
507 start point of the match, as specified by the \fIstartoffset\fP argument of
508 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
509 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
510 arguments, you can mimic Perl's /g option, and it is in this kind of
511 implementation where \eG can be useful.
512 .P
513 Note, however, that PCRE's interpretation of \eG, as the start of the current
514 match, is subtly different from Perl's, which defines it as the end of the
515 previous match. In Perl, these can be different when the previously matched
516 string was empty. Because PCRE does just one match at a time, it cannot
517 reproduce this behaviour.
518 .P
519 If all the alternatives of a pattern begin with \eG, the expression is anchored
520 to the starting match position, and the "anchored" flag is set in the compiled
521 regular expression.
522 .
523 .
525 .rs
526 .sp
527 Outside a character class, in the default matching mode, the circumflex
528 character is an assertion that is true only if the current matching point is
529 at the start of the subject string. If the \fIstartoffset\fP argument of
530 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
531 option is unset. Inside a character class, circumflex has an entirely different
532 meaning
533 .\" HTML <a href="#characterclass">
534 .\" </a>
535 (see below).
536 .\"
537 .P
538 Circumflex need not be the first character of the pattern if a number of
539 alternatives are involved, but it should be the first thing in each alternative
540 in which it appears if the pattern is ever to match that branch. If all
541 possible alternatives start with a circumflex, that is, if the pattern is
542 constrained to match only at the start of the subject, it is said to be an
543 "anchored" pattern. (There are also other constructs that can cause a pattern
544 to be anchored.)
545 .P
546 A dollar character is an assertion that is true only if the current matching
547 point is at the end of the subject string, or immediately before a newline
548 character that is the last character in the string (by default). Dollar need
549 not be the last character of the pattern if a number of alternatives are
550 involved, but it should be the last item in any branch in which it appears.
551 Dollar has no special meaning in a character class.
552 .P
553 The meaning of dollar can be changed so that it matches only at the very end of
554 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
555 does not affect the \eZ assertion.
556 .P
557 The meanings of the circumflex and dollar characters are changed if the
558 PCRE_MULTILINE option is set. When this is the case, they match immediately
559 after and immediately before an internal newline character, respectively, in
560 addition to matching at the start and end of the subject string. For example,
561 the pattern /^abc$/ matches the subject string "def\enabc" (where \en
562 represents a newline character) in multiline mode, but not otherwise.
563 Consequently, patterns that are anchored in single line mode because all
564 branches start with ^ are not anchored in multiline mode, and a match for
565 circumflex is possible when the \fIstartoffset\fP argument of \fBpcre_exec()\fP
566 is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
567 set.
568 .P
569 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
570 end of the subject in both modes, and if all branches of a pattern start with
571 \eA it is always anchored, whether PCRE_MULTILINE is set or not.
572 .
573 .
575 .rs
576 .sp
577 Outside a character class, a dot in the pattern matches any one character in
578 the subject, including a non-printing character, but not (by default) newline.
579 In UTF-8 mode, a dot matches any UTF-8 character, which might be more than one
580 byte long, except (by default) newline. If the PCRE_DOTALL option is set,
581 dots match newlines as well. The handling of dot is entirely independent of the
582 handling of circumflex and dollar, the only relationship being that they both
583 involve newline characters. Dot has no special meaning in a character class.
584 .
585 .
587 .rs
588 .sp
589 Outside a character class, the escape sequence \eC matches any one byte, both
590 in and out of UTF-8 mode. Unlike a dot, it can match a newline. The feature is
591 provided in Perl in order to match individual bytes in UTF-8 mode. Because it
592 breaks up UTF-8 characters into individual bytes, what remains in the string
593 may be a malformed UTF-8 string. For this reason, the \eC escape sequence is
594 best avoided.
595 .P
596 PCRE does not allow \eC to appear in lookbehind assertions
597 .\" HTML <a href="#lookbehind">
598 .\" </a>
599 (described below),
600 .\"
601 because in UTF-8 mode this would make it impossible to calculate the length of
602 the lookbehind.
603 .
604 .
605 .\" HTML <a name="characterclass"></a>
607 .rs
608 .sp
609 An opening square bracket introduces a character class, terminated by a closing
610 square bracket. A closing square bracket on its own is not special. If a
611 closing square bracket is required as a member of the class, it should be the
612 first data character in the class (after an initial circumflex, if present) or
613 escaped with a backslash.
614 .P
615 A character class matches a single character in the subject. In UTF-8 mode, the
616 character may occupy more than one byte. A matched character must be in the set
617 of characters defined by the class, unless the first character in the class
618 definition is a circumflex, in which case the subject character must not be in
619 the set defined by the class. If a circumflex is actually required as a member
620 of the class, ensure it is not the first character, or escape it with a
621 backslash.
622 .P
623 For example, the character class [aeiou] matches any lower case vowel, while
624 [^aeiou] matches any character that is not a lower case vowel. Note that a
625 circumflex is just a convenient notation for specifying the characters that
626 are in the class by enumerating those that are not. A class that starts with a
627 circumflex is not an assertion: it still consumes a character from the subject
628 string, and therefore it fails if the current pointer is at the end of the
629 string.
630 .P
631 In UTF-8 mode, characters with values greater than 255 can be included in a
632 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
633 .P
634 When caseless matching is set, any letters in a class represent both their
635 upper case and lower case versions, so for example, a caseless [aeiou] matches
636 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
637 caseful version would. In UTF-8 mode, PCRE always understands the concept of
638 case for characters whose values are less than 128, so caseless matching is
639 always possible. For characters with higher values, the concept of case is
640 supported if PCRE is compiled with Unicode property support, but not otherwise.
641 If you want to use caseless matching for characters 128 and above, you must
642 ensure that PCRE is compiled with Unicode property support as well as with
643 UTF-8 support.
644 .P
645 The newline character is never treated in any special way in character classes,
646 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
647 such as [^a] will always match a newline.
648 .P
649 The minus (hyphen) character can be used to specify a range of characters in a
650 character class. For example, [d-m] matches any letter between d and m,
651 inclusive. If a minus character is required in a class, it must be escaped with
652 a backslash or appear in a position where it cannot be interpreted as
653 indicating a range, typically as the first or last character in the class.
654 .P
655 It is not possible to have the literal character "]" as the end character of a
656 range. A pattern such as [W-]46] is interpreted as a class of two characters
657 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
658 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
659 the end of range, so [W-\e]46] is interpreted as a class containing a range
660 followed by two other characters. The octal or hexadecimal representation of
661 "]" can also be used to end a range.
662 .P
663 Ranges operate in the collating sequence of character values. They can also be
664 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
665 mode, ranges can include characters whose values are greater than 255, for
666 example [\ex{100}-\ex{2ff}].
667 .P
668 If a range that includes letters is used when caseless matching is set, it
669 matches the letters in either case. For example, [W-c] is equivalent to
670 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
671 tables for the "fr_FR" locale are in use, [\exc8-\excb] matches accented E
672 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
673 characters with values greater than 128 only when it is compiled with Unicode
674 property support.
675 .P
676 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
677 in a character class, and add the characters that they match to the class. For
678 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
679 conveniently be used with the upper case character types to specify a more
680 restricted set of characters than the matching lower case type. For example,
681 the class [^\eW_] matches any letter or digit, but not underscore.
682 .P
683 The only metacharacters that are recognized in character classes are backslash,
684 hyphen (only where it can be interpreted as specifying a range), circumflex
685 (only at the start), opening square bracket (only when it can be interpreted as
686 introducing a POSIX class name - see the next section), and the terminating
687 closing square bracket. However, escaping other non-alphanumeric characters
688 does no harm.
689 .
690 .
692 .rs
693 .sp
694 Perl supports the POSIX notation for character classes. This uses names
695 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
696 this notation. For example,
697 .sp
698 [01[:alpha:]%]
699 .sp
700 matches "0", "1", any alphabetic character, or "%". The supported class names
701 are
702 .sp
703 alnum letters and digits
704 alpha letters
705 ascii character codes 0 - 127
706 blank space or tab only
707 cntrl control characters
708 digit decimal digits (same as \ed)
709 graph printing characters, excluding space
710 lower lower case letters
711 print printing characters, including space
712 punct printing characters, excluding letters and digits
713 space white space (not quite the same as \es)
714 upper upper case letters
715 word "word" characters (same as \ew)
716 xdigit hexadecimal digits
717 .sp
718 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
719 space (32). Notice that this list includes the VT character (code 11). This
720 makes "space" different to \es, which does not include VT (for Perl
721 compatibility).
722 .P
723 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
724 5.8. Another Perl extension is negation, which is indicated by a ^ character
725 after the colon. For example,
726 .sp
727 [12[:^digit:]]
728 .sp
729 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
730 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
731 supported, and an error is given if they are encountered.
732 .P
733 In UTF-8 mode, characters with values greater than 128 do not match any of
734 the POSIX character classes.
735 .
736 .
738 .rs
739 .sp
740 Vertical bar characters are used to separate alternative patterns. For example,
741 the pattern
742 .sp
743 gilbert|sullivan
744 .sp
745 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
746 and an empty alternative is permitted (matching the empty string).
747 The matching process tries each alternative in turn, from left to right,
748 and the first one that succeeds is used. If the alternatives are within a
749 subpattern
750 .\" HTML <a href="#subpattern">
751 .\" </a>
752 (defined below),
753 .\"
754 "succeeds" means matching the rest of the main pattern as well as the
755 alternative in the subpattern.
756 .
757 .
759 .rs
760 .sp
762 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
763 Perl option letters enclosed between "(?" and ")". The option letters are
764 .sp
767 s for PCRE_DOTALL
769 .sp
770 For example, (?im) sets caseless, multiline matching. It is also possible to
771 unset these options by preceding the letter with a hyphen, and a combined
772 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
773 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
774 permitted. If a letter appears both before and after the hyphen, the option is
775 unset.
776 .P
777 When an option change occurs at top level (that is, not inside subpattern
778 parentheses), the change applies to the remainder of the pattern that follows.
779 If the change is placed right at the start of a pattern, PCRE extracts it into
780 the global options (and it will therefore show up in data extracted by the
781 \fBpcre_fullinfo()\fP function).
782 .P
783 An option change within a subpattern affects only that part of the current
784 pattern that follows it, so
785 .sp
786 (a(?i)b)c
787 .sp
788 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
789 By this means, options can be made to have different settings in different
790 parts of the pattern. Any changes made in one alternative do carry on
791 into subsequent branches within the same subpattern. For example,
792 .sp
793 (a(?i)b|c)
794 .sp
795 matches "ab", "aB", "c", and "C", even though when matching "C" the first
796 branch is abandoned before the option setting. This is because the effects of
797 option settings happen at compile time. There would be some very weird
798 behaviour otherwise.
799 .P
800 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
801 same way as the Perl-compatible options by using the characters U and X
802 respectively. The (?X) flag setting is special in that it must always occur
803 earlier in the pattern than any of the additional features it turns on, even
804 when it is at top level. It is best to put it at the start.
805 .
806 .
807 .\" HTML <a name="subpattern"></a>
809 .rs
810 .sp
811 Subpatterns are delimited by parentheses (round brackets), which can be nested.
812 Turning part of a pattern into a subpattern does two things:
813 .sp
814 1. It localizes a set of alternatives. For example, the pattern
815 .sp
816 cat(aract|erpillar|)
817 .sp
818 matches one of the words "cat", "cataract", or "caterpillar". Without the
819 parentheses, it would match "cataract", "erpillar" or the empty string.
820 .sp
821 2. It sets up the subpattern as a capturing subpattern. This means that, when
822 the whole pattern matches, that portion of the subject string that matched the
823 subpattern is passed back to the caller via the \fIovector\fP argument of
824 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
825 from 1) to obtain numbers for the capturing subpatterns.
826 .P
827 For example, if the string "the red king" is matched against the pattern
828 .sp
829 the ((red|white) (king|queen))
830 .sp
831 the captured substrings are "red king", "red", and "king", and are numbered 1,
832 2, and 3, respectively.
833 .P
834 The fact that plain parentheses fulfil two functions is not always helpful.
835 There are often times when a grouping subpattern is required without a
836 capturing requirement. If an opening parenthesis is followed by a question mark
837 and a colon, the subpattern does not do any capturing, and is not counted when
838 computing the number of any subsequent capturing subpatterns. For example, if
839 the string "the white queen" is matched against the pattern
840 .sp
841 the ((?:red|white) (king|queen))
842 .sp
843 the captured substrings are "white queen" and "queen", and are numbered 1 and
844 2. The maximum number of capturing subpatterns is 65535, and the maximum depth
845 of nesting of all subpatterns, both capturing and non-capturing, is 200.
846 .P
847 As a convenient shorthand, if any option settings are required at the start of
848 a non-capturing subpattern, the option letters may appear between the "?" and
849 the ":". Thus the two patterns
850 .sp
851 (?i:saturday|sunday)
852 (?:(?i)saturday|sunday)
853 .sp
854 match exactly the same set of strings. Because alternative branches are tried
855 from left to right, and options are not reset until the end of the subpattern
856 is reached, an option setting in one branch does affect subsequent branches, so
857 the above patterns match "SUNDAY" as well as "Saturday".
858 .
859 .
861 .rs
862 .sp
863 Identifying capturing parentheses by number is simple, but it can be very hard
864 to keep track of the numbers in complicated regular expressions. Furthermore,
865 if an expression is modified, the numbers may change. To help with this
866 difficulty, PCRE supports the naming of subpatterns, something that Perl does
867 not provide. The Python syntax (?P<name>...) is used. Names consist of
868 alphanumeric characters and underscores, and must be unique within a pattern.
869 .P
870 Named capturing parentheses are still allocated numbers as well as names. The
871 PCRE API provides function calls for extracting the name-to-number translation
872 table from a compiled pattern. There is also a convenience function for
873 extracting a captured substring by name. For further details see the
874 .\" HREF
875 \fBpcreapi\fP
876 .\"
877 documentation.
878 .
879 .
881 .rs
882 .sp
883 Repetition is specified by quantifiers, which can follow any of the following
884 items:
885 .sp
886 a literal data character
887 the . metacharacter
888 the \eC escape sequence
889 the \eX escape sequence (in UTF-8 mode with Unicode properties)
890 an escape such as \ed that matches a single character
891 a character class
892 a back reference (see next section)
893 a parenthesized subpattern (unless it is an assertion)
894 .sp
895 The general repetition quantifier specifies a minimum and maximum number of
896 permitted matches, by giving the two numbers in curly brackets (braces),
897 separated by a comma. The numbers must be less than 65536, and the first must
898 be less than or equal to the second. For example:
899 .sp
900 z{2,4}
901 .sp
902 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
903 character. If the second number is omitted, but the comma is present, there is
904 no upper limit; if the second number and the comma are both omitted, the
905 quantifier specifies an exact number of required matches. Thus
906 .sp
907 [aeiou]{3,}
908 .sp
909 matches at least 3 successive vowels, but may match many more, while
910 .sp
911 \ed{8}
912 .sp
913 matches exactly 8 digits. An opening curly bracket that appears in a position
914 where a quantifier is not allowed, or one that does not match the syntax of a
915 quantifier, is taken as a literal character. For example, {,6} is not a
916 quantifier, but a literal string of four characters.
917 .P
918 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
919 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
920 which is represented by a two-byte sequence. Similarly, when Unicode property
921 support is available, \eX{3} matches three Unicode extended sequences, each of
922 which may be several bytes long (and they may be of different lengths).
923 .P
924 The quantifier {0} is permitted, causing the expression to behave as if the
925 previous item and the quantifier were not present.
926 .P
927 For convenience (and historical compatibility) the three most common
928 quantifiers have single-character abbreviations:
929 .sp
930 * is equivalent to {0,}
931 + is equivalent to {1,}
932 ? is equivalent to {0,1}
933 .sp
934 It is possible to construct infinite loops by following a subpattern that can
935 match no characters with a quantifier that has no upper limit, for example:
936 .sp
937 (a?)*
938 .sp
939 Earlier versions of Perl and PCRE used to give an error at compile time for
940 such patterns. However, because there are cases where this can be useful, such
941 patterns are now accepted, but if any repetition of the subpattern does in fact
942 match no characters, the loop is forcibly broken.
943 .P
944 By default, the quantifiers are "greedy", that is, they match as much as
945 possible (up to the maximum number of permitted times), without causing the
946 rest of the pattern to fail. The classic example of where this gives problems
947 is in trying to match comments in C programs. These appear between /* and */
948 and within the comment, individual * and / characters may appear. An attempt to
949 match C comments by applying the pattern
950 .sp
951 /\e*.*\e*/
952 .sp
953 to the string
954 .sp
955 /* first comment */ not comment /* second comment */
956 .sp
957 fails, because it matches the entire string owing to the greediness of the .*
958 item.
959 .P
960 However, if a quantifier is followed by a question mark, it ceases to be
961 greedy, and instead matches the minimum number of times possible, so the
962 pattern
963 .sp
964 /\e*.*?\e*/
965 .sp
966 does the right thing with the C comments. The meaning of the various
967 quantifiers is not otherwise changed, just the preferred number of matches.
968 Do not confuse this use of question mark with its use as a quantifier in its
969 own right. Because it has two uses, it can sometimes appear doubled, as in
970 .sp
971 \ed??\ed
972 .sp
973 which matches one digit by preference, but can match two if that is the only
974 way the rest of the pattern matches.
975 .P
976 If the PCRE_UNGREEDY option is set (an option which is not available in Perl),
977 the quantifiers are not greedy by default, but individual ones can be made
978 greedy by following them with a question mark. In other words, it inverts the
979 default behaviour.
980 .P
981 When a parenthesized subpattern is quantified with a minimum repeat count that
982 is greater than 1 or with a limited maximum, more memory is required for the
983 compiled pattern, in proportion to the size of the minimum or maximum.
984 .P
985 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
986 to Perl's /s) is set, thus allowing the . to match newlines, the pattern is
987 implicitly anchored, because whatever follows will be tried against every
988 character position in the subject string, so there is no point in retrying the
989 overall match at any position after the first. PCRE normally treats such a
990 pattern as though it were preceded by \eA.
991 .P
992 In cases where it is known that the subject string contains no newlines, it is
993 worth setting PCRE_DOTALL in order to obtain this optimization, or
994 alternatively using ^ to indicate anchoring explicitly.
995 .P
996 However, there is one situation where the optimization cannot be used. When .*
997 is inside capturing parentheses that are the subject of a backreference
998 elsewhere in the pattern, a match at the start may fail, and a later one
999 succeed. Consider, for example:
1000 .sp
1001 (.*)abc\e1
1002 .sp
1003 If the subject is "xyz123abc123" the match point is the fourth character. For
1004 this reason, such a pattern is not implicitly anchored.
1005 .P
1006 When a capturing subpattern is repeated, the value captured is the substring
1007 that matched the final iteration. For example, after
1008 .sp
1009 (tweedle[dume]{3}\es*)+
1010 .sp
1011 has matched "tweedledum tweedledee" the value of the captured substring is
1012 "tweedledee". However, if there are nested capturing subpatterns, the
1013 corresponding captured values may have been set in previous iterations. For
1014 example, after
1015 .sp
1016 /(a|(b))+/
1017 .sp
1018 matches "aba" the value of the second captured substring is "b".
1019 .
1020 .
1021 .\" HTML <a name="atomicgroup"></a>
1023 .rs
1024 .sp
1025 With both maximizing and minimizing repetition, failure of what follows
1026 normally causes the repeated item to be re-evaluated to see if a different
1027 number of repeats allows the rest of the pattern to match. Sometimes it is
1028 useful to prevent this, either to change the nature of the match, or to cause
1029 it fail earlier than it otherwise might, when the author of the pattern knows
1030 there is no point in carrying on.
1031 .P
1032 Consider, for example, the pattern \ed+foo when applied to the subject line
1033 .sp
1034 123456bar
1035 .sp
1036 After matching all 6 digits and then failing to match "foo", the normal
1037 action of the matcher is to try again with only 5 digits matching the \ed+
1038 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1039 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1040 that once a subpattern has matched, it is not to be re-evaluated in this way.
1041 .P
1042 If we use atomic grouping for the previous example, the matcher would give up
1043 immediately on failing to match "foo" the first time. The notation is a kind of
1044 special parenthesis, starting with (?> as in this example:
1045 .sp
1046 (?>\ed+)foo
1047 .sp
1048 This kind of parenthesis "locks up" the part of the pattern it contains once
1049 it has matched, and a failure further into the pattern is prevented from
1050 backtracking into it. Backtracking past it to previous items, however, works as
1051 normal.
1052 .P
1053 An alternative description is that a subpattern of this type matches the string
1054 of characters that an identical standalone pattern would match, if anchored at
1055 the current point in the subject string.
1056 .P
1057 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1058 the above example can be thought of as a maximizing repeat that must swallow
1059 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1060 number of digits they match in order to make the rest of the pattern match,
1061 (?>\ed+) can only match an entire sequence of digits.
1062 .P
1063 Atomic groups in general can of course contain arbitrarily complicated
1064 subpatterns, and can be nested. However, when the subpattern for an atomic
1065 group is just a single repeated item, as in the example above, a simpler
1066 notation, called a "possessive quantifier" can be used. This consists of an
1067 additional + character following a quantifier. Using this notation, the
1068 previous example can be rewritten as
1069 .sp
1070 \ed++foo
1071 .sp
1072 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1073 option is ignored. They are a convenient notation for the simpler forms of
1074 atomic group. However, there is no difference in the meaning or processing of a
1075 possessive quantifier and the equivalent atomic group.
1076 .P
1077 The possessive quantifier syntax is an extension to the Perl syntax. It
1078 originates in Sun's Java package.
1079 .P
1080 When a pattern contains an unlimited repeat inside a subpattern that can itself
1081 be repeated an unlimited number of times, the use of an atomic group is the
1082 only way to avoid some failing matches taking a very long time indeed. The
1083 pattern
1084 .sp
1085 (\eD+|<\ed+>)*[!?]
1086 .sp
1087 matches an unlimited number of substrings that either consist of non-digits, or
1088 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1089 quickly. However, if it is applied to
1090 .sp
1091 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1092 .sp
1093 it takes a long time before reporting failure. This is because the string can
1094 be divided between the internal \eD+ repeat and the external * repeat in a
1095 large number of ways, and all have to be tried. (The example uses [!?] rather
1096 than a single character at the end, because both PCRE and Perl have an
1097 optimization that allows for fast failure when a single character is used. They
1098 remember the last single character that is required for a match, and fail early
1099 if it is not present in the string.) If the pattern is changed so that it uses
1100 an atomic group, like this:
1101 .sp
1102 ((?>\eD+)|<\ed+>)*[!?]
1103 .sp
1104 sequences of non-digits cannot be broken, and failure happens quickly.
1105 .
1106 .
1107 .\" HTML <a name="backreferences"></a>
1109 .rs
1110 .sp
1111 Outside a character class, a backslash followed by a digit greater than 0 (and
1112 possibly further digits) is a back reference to a capturing subpattern earlier
1113 (that is, to its left) in the pattern, provided there have been that many
1114 previous capturing left parentheses.
1115 .P
1116 However, if the decimal number following the backslash is less than 10, it is
1117 always taken as a back reference, and causes an error only if there are not
1118 that many capturing left parentheses in the entire pattern. In other words, the
1119 parentheses that are referenced need not be to the left of the reference for
1120 numbers less than 10. See the subsection entitled "Non-printing characters"
1121 .\" HTML <a href="#digitsafterbackslash">
1122 .\" </a>
1123 above
1124 .\"
1125 for further details of the handling of digits following a backslash.
1126 .P
1127 A back reference matches whatever actually matched the capturing subpattern in
1128 the current subject string, rather than anything matching the subpattern
1129 itself (see
1130 .\" HTML <a href="#subpatternsassubroutines">
1131 .\" </a>
1132 "Subpatterns as subroutines"
1133 .\"
1134 below for a way of doing that). So the pattern
1135 .sp
1136 (sens|respons)e and \e1ibility
1137 .sp
1138 matches "sense and sensibility" and "response and responsibility", but not
1139 "sense and responsibility". If caseful matching is in force at the time of the
1140 back reference, the case of letters is relevant. For example,
1141 .sp
1142 ((?i)rah)\es+\e1
1143 .sp
1144 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1145 capturing subpattern is matched caselessly.
1146 .P
1147 Back references to named subpatterns use the Python syntax (?P=name). We could
1148 rewrite the above example as follows:
1149 .sp
1150 (?<p1>(?i)rah)\es+(?P=p1)
1151 .sp
1152 There may be more than one back reference to the same subpattern. If a
1153 subpattern has not actually been used in a particular match, any back
1154 references to it always fail. For example, the pattern
1155 .sp
1156 (a|(bc))\e2
1157 .sp
1158 always fails if it starts to match "a" rather than "bc". Because there may be
1159 many capturing parentheses in a pattern, all digits following the backslash are
1160 taken as part of a potential back reference number. If the pattern continues
1161 with a digit character, some delimiter must be used to terminate the back
1162 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1163 Otherwise an empty comment (see
1164 .\" HTML <a href="#comments">
1165 .\" </a>
1166 "Comments"
1167 .\"
1168 below) can be used.
1169 .P
1170 A back reference that occurs inside the parentheses to which it refers fails
1171 when the subpattern is first used, so, for example, (a\e1) never matches.
1172 However, such references can be useful inside repeated subpatterns. For
1173 example, the pattern
1174 .sp
1175 (a|b\e1)+
1176 .sp
1177 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1178 the subpattern, the back reference matches the character string corresponding
1179 to the previous iteration. In order for this to work, the pattern must be such
1180 that the first iteration does not need to match the back reference. This can be
1181 done using alternation, as in the example above, or by a quantifier with a
1182 minimum of zero.
1183 .
1184 .
1185 .\" HTML <a name="bigassertions"></a>
1187 .rs
1188 .sp
1189 An assertion is a test on the characters following or preceding the current
1190 matching point that does not actually consume any characters. The simple
1191 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1192 .\" HTML <a href="#smallassertions">
1193 .\" </a>
1194 above.
1195 .\"
1196 .P
1197 More complicated assertions are coded as subpatterns. There are two kinds:
1198 those that look ahead of the current position in the subject string, and those
1199 that look behind it. An assertion subpattern is matched in the normal way,
1200 except that it does not cause the current matching position to be changed.
1201 .P
1202 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1203 because it makes no sense to assert the same thing several times. If any kind
1204 of assertion contains capturing subpatterns within it, these are counted for
1205 the purposes of numbering the capturing subpatterns in the whole pattern.
1206 However, substring capturing is carried out only for positive assertions,
1207 because it does not make sense for negative assertions.
1208 .
1209 .
1210 .SS "Lookahead assertions"
1211 .rs
1212 .sp
1213 Lookahead assertions start
1214 with (?= for positive assertions and (?! for negative assertions. For example,
1215 .sp
1216 \ew+(?=;)
1217 .sp
1218 matches a word followed by a semicolon, but does not include the semicolon in
1219 the match, and
1220 .sp
1221 foo(?!bar)
1222 .sp
1223 matches any occurrence of "foo" that is not followed by "bar". Note that the
1224 apparently similar pattern
1225 .sp
1226 (?!foo)bar
1227 .sp
1228 does not find an occurrence of "bar" that is preceded by something other than
1229 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1230 (?!foo) is always true when the next three characters are "bar". A
1231 lookbehind assertion is needed to achieve the other effect.
1232 .P
1233 If you want to force a matching failure at some point in a pattern, the most
1234 convenient way to do it is with (?!) because an empty string always matches, so
1235 an assertion that requires there not to be an empty string must always fail.
1236 .
1237 .
1238 .\" HTML <a name="lookbehind"></a>
1239 .SS "Lookbehind assertions"
1240 .rs
1241 .sp
1242 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1243 negative assertions. For example,
1244 .sp
1245 (?<!foo)bar
1246 .sp
1247 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1248 a lookbehind assertion are restricted such that all the strings it matches must
1249 have a fixed length. However, if there are several alternatives, they do not
1250 all have to have the same fixed length. Thus
1251 .sp
1252 (?<=bullock|donkey)
1253 .sp
1254 is permitted, but
1255 .sp
1256 (?<!dogs?|cats?)
1257 .sp
1258 causes an error at compile time. Branches that match different length strings
1259 are permitted only at the top level of a lookbehind assertion. This is an
1260 extension compared with Perl (at least for 5.8), which requires all branches to
1261 match the same length of string. An assertion such as
1262 .sp
1263 (?<=ab(c|de))
1264 .sp
1265 is not permitted, because its single top-level branch can match two different
1266 lengths, but it is acceptable if rewritten to use two top-level branches:
1267 .sp
1268 (?<=abc|abde)
1269 .sp
1270 The implementation of lookbehind assertions is, for each alternative, to
1271 temporarily move the current position back by the fixed width and then try to
1272 match. If there are insufficient characters before the current position, the
1273 match is deemed to fail.
1274 .P
1275 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1276 to appear in lookbehind assertions, because it makes it impossible to calculate
1277 the length of the lookbehind. The \eX escape, which can match different numbers
1278 of bytes, is also not permitted.
1279 .P
1280 Atomic groups can be used in conjunction with lookbehind assertions to specify
1281 efficient matching at the end of the subject string. Consider a simple pattern
1282 such as
1283 .sp
1284 abcd$
1285 .sp
1286 when applied to a long string that does not match. Because matching proceeds
1287 from left to right, PCRE will look for each "a" in the subject and then see if
1288 what follows matches the rest of the pattern. If the pattern is specified as
1289 .sp
1290 ^.*abcd$
1291 .sp
1292 the initial .* matches the entire string at first, but when this fails (because
1293 there is no following "a"), it backtracks to match all but the last character,
1294 then all but the last two characters, and so on. Once again the search for "a"
1295 covers the entire string, from right to left, so we are no better off. However,
1296 if the pattern is written as
1297 .sp
1298 ^(?>.*)(?<=abcd)
1299 .sp
1300 or, equivalently, using the possessive quantifier syntax,
1301 .sp
1302 ^.*+(?<=abcd)
1303 .sp
1304 there can be no backtracking for the .* item; it can match only the entire
1305 string. The subsequent lookbehind assertion does a single test on the last four
1306 characters. If it fails, the match fails immediately. For long strings, this
1307 approach makes a significant difference to the processing time.
1308 .
1309 .
1310 .SS "Using multiple assertions"
1311 .rs
1312 .sp
1313 Several assertions (of any sort) may occur in succession. For example,
1314 .sp
1315 (?<=\ed{3})(?<!999)foo
1316 .sp
1317 matches "foo" preceded by three digits that are not "999". Notice that each of
1318 the assertions is applied independently at the same point in the subject
1319 string. First there is a check that the previous three characters are all
1320 digits, and then there is a check that the same three characters are not "999".
1321 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1322 of which are digits and the last three of which are not "999". For example, it
1323 doesn't match "123abcfoo". A pattern to do that is
1324 .sp
1325 (?<=\ed{3}...)(?<!999)foo
1326 .sp
1327 This time the first assertion looks at the preceding six characters, checking
1328 that the first three are digits, and then the second assertion checks that the
1329 preceding three characters are not "999".
1330 .P
1331 Assertions can be nested in any combination. For example,
1332 .sp
1333 (?<=(?<!foo)bar)baz
1334 .sp
1335 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1336 preceded by "foo", while
1337 .sp
1338 (?<=\ed{3}(?!999)...)foo
1339 .sp
1340 is another pattern that matches "foo" preceded by three digits and any three
1341 characters that are not "999".
1342 .
1343 .
1345 .rs
1346 .sp
1347 It is possible to cause the matching process to obey a subpattern
1348 conditionally or to choose between two alternative subpatterns, depending on
1349 the result of an assertion, or whether a previous capturing subpattern matched
1350 or not. The two possible forms of conditional subpattern are
1351 .sp
1352 (?(condition)yes-pattern)
1353 (?(condition)yes-pattern|no-pattern)
1354 .sp
1355 If the condition is satisfied, the yes-pattern is used; otherwise the
1356 no-pattern (if present) is used. If there are more than two alternatives in the
1357 subpattern, a compile-time error occurs.
1358 .P
1359 There are three kinds of condition. If the text between the parentheses
1360 consists of a sequence of digits, the condition is satisfied if the capturing
1361 subpattern of that number has previously matched. The number must be greater
1362 than zero. Consider the following pattern, which contains non-significant white
1363 space to make it more readable (assume the PCRE_EXTENDED option) and to divide
1364 it into three parts for ease of discussion:
1365 .sp
1366 ( \e( )? [^()]+ (?(1) \e) )
1367 .sp
1368 The first part matches an optional opening parenthesis, and if that
1369 character is present, sets it as the first captured substring. The second part
1370 matches one or more characters that are not parentheses. The third part is a
1371 conditional subpattern that tests whether the first set of parentheses matched
1372 or not. If they did, that is, if subject started with an opening parenthesis,
1373 the condition is true, and so the yes-pattern is executed and a closing
1374 parenthesis is required. Otherwise, since no-pattern is not present, the
1375 subpattern matches nothing. In other words, this pattern matches a sequence of
1376 non-parentheses, optionally enclosed in parentheses.
1377 .P
1378 If the condition is the string (R), it is satisfied if a recursive call to the
1379 pattern or subpattern has been made. At "top level", the condition is false.
1380 This is a PCRE extension. Recursive patterns are described in the next section.
1381 .P
1382 If the condition is not a sequence of digits or (R), it must be an assertion.
1383 This may be a positive or negative lookahead or lookbehind assertion. Consider
1384 this pattern, again containing non-significant white space, and with the two
1385 alternatives on the second line:
1386 .sp
1387 (?(?=[^a-z]*[a-z])
1388 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1389 .sp
1390 The condition is a positive lookahead assertion that matches an optional
1391 sequence of non-letters followed by a letter. In other words, it tests for the
1392 presence of at least one letter in the subject. If a letter is found, the
1393 subject is matched against the first alternative; otherwise it is matched
1394 against the second. This pattern matches strings in one of the two forms
1395 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1396 .
1397 .
1398 .\" HTML <a name="comments"></a>
1400 .rs
1401 .sp
1402 The sequence (?# marks the start of a comment that continues up to the next
1403 closing parenthesis. Nested parentheses are not permitted. The characters
1404 that make up a comment play no part in the pattern matching at all.
1405 .P
1406 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1407 character class introduces a comment that continues up to the next newline
1408 character in the pattern.
1409 .
1410 .
1412 .rs
1413 .sp
1414 Consider the problem of matching a string in parentheses, allowing for
1415 unlimited nested parentheses. Without the use of recursion, the best that can
1416 be done is to use a pattern that matches up to some fixed depth of nesting. It
1417 is not possible to handle an arbitrary nesting depth. Perl provides a facility
1418 that allows regular expressions to recurse (amongst other things). It does this
1419 by interpolating Perl code in the expression at run time, and the code can
1420 refer to the expression itself. A Perl pattern to solve the parentheses problem
1421 can be created like this:
1422 .sp
1423 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1424 .sp
1425 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1426 recursively to the pattern in which it appears. Obviously, PCRE cannot support
1427 the interpolation of Perl code. Instead, it supports some special syntax for
1428 recursion of the entire pattern, and also for individual subpattern recursion.
1429 .P
1430 The special item that consists of (? followed by a number greater than zero and
1431 a closing parenthesis is a recursive call of the subpattern of the given
1432 number, provided that it occurs inside that subpattern. (If not, it is a
1433 "subroutine" call, which is described in the next section.) The special item
1434 (?R) is a recursive call of the entire regular expression.
1435 .P
1436 A recursive subpattern call is always treated as an atomic group. That is, once
1437 it has matched some of the subject string, it is never re-entered, even if
1438 it contains untried alternatives and there is a subsequent matching failure.
1439 .P
1440 This PCRE pattern solves the nested parentheses problem (assume the
1441 PCRE_EXTENDED option is set so that white space is ignored):
1442 .sp
1443 \e( ( (?>[^()]+) | (?R) )* \e)
1444 .sp
1445 First it matches an opening parenthesis. Then it matches any number of
1446 substrings which can either be a sequence of non-parentheses, or a recursive
1447 match of the pattern itself (that is, a correctly parenthesized substring).
1448 Finally there is a closing parenthesis.
1449 .P
1450 If this were part of a larger pattern, you would not want to recurse the entire
1451 pattern, so instead you could use this:
1452 .sp
1453 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1454 .sp
1455 We have put the pattern into parentheses, and caused the recursion to refer to
1456 them instead of the whole pattern. In a larger pattern, keeping track of
1457 parenthesis numbers can be tricky. It may be more convenient to use named
1458 parentheses instead. For this, PCRE uses (?P>name), which is an extension to
1459 the Python syntax that PCRE uses for named parentheses (Perl does not provide
1460 named parentheses). We could rewrite the above example as follows:
1461 .sp
1462 (?P<pn> \e( ( (?>[^()]+) | (?P>pn) )* \e) )
1463 .sp
1464 This particular example pattern contains nested unlimited repeats, and so the
1465 use of atomic grouping for matching strings of non-parentheses is important
1466 when applying the pattern to strings that do not match. For example, when this
1467 pattern is applied to
1468 .sp
1469 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1470 .sp
1471 it yields "no match" quickly. However, if atomic grouping is not used,
1472 the match runs for a very long time indeed because there are so many different
1473 ways the + and * repeats can carve up the subject, and all have to be tested
1474 before failure can be reported.
1475 .P
1476 At the end of a match, the values set for any capturing subpatterns are those
1477 from the outermost level of the recursion at which the subpattern value is set.
1478 If you want to obtain intermediate values, a callout function can be used (see
1479 the next section and the
1480 .\" HREF
1481 \fBpcrecallout\fP
1482 .\"
1483 documentation). If the pattern above is matched against
1484 .sp
1485 (ab(cd)ef)
1486 .sp
1487 the value for the capturing parentheses is "ef", which is the last value taken
1488 on at the top level. If additional parentheses are added, giving
1489 .sp
1490 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1491 ^ ^
1492 ^ ^
1493 .sp
1494 the string they capture is "ab(cd)ef", the contents of the top level
1495 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1496 has to obtain extra memory to store data during a recursion, which it does by
1497 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1498 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1499 .P
1500 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1501 Consider this pattern, which matches text in angle brackets, allowing for
1502 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1503 recursing), whereas any characters are permitted at the outer level.
1504 .sp
1505 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1506 .sp
1507 In this pattern, (?(R) is the start of a conditional subpattern, with two
1508 different alternatives for the recursive and non-recursive cases. The (?R) item
1509 is the actual recursive call.
1510 .
1511 .
1512 .\" HTML <a name="subpatternsassubroutines"></a>
1514 .rs
1515 .sp
1516 If the syntax for a recursive subpattern reference (either by number or by
1517 name) is used outside the parentheses to which it refers, it operates like a
1518 subroutine in a programming language. An earlier example pointed out that the
1519 pattern
1520 .sp
1521 (sens|respons)e and \e1ibility
1522 .sp
1523 matches "sense and sensibility" and "response and responsibility", but not
1524 "sense and responsibility". If instead the pattern
1525 .sp
1526 (sens|respons)e and (?1)ibility
1527 .sp
1528 is used, it does match "sense and responsibility" as well as the other two
1529 strings. Such references must, however, follow the subpattern to which they
1530 refer.
1531 .P
1532 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1533 group. That is, once it has matched some of the subject string, it is never
1534 re-entered, even if it contains untried alternatives and there is a subsequent
1535 matching failure.
1536 .
1537 .
1539 .rs
1540 .sp
1541 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1542 code to be obeyed in the middle of matching a regular expression. This makes it
1543 possible, amongst other things, to extract different substrings that match the
1544 same pair of parentheses when there is a repetition.
1545 .P
1546 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1547 code. The feature is called "callout". The caller of PCRE provides an external
1548 function by putting its entry point in the global variable \fIpcre_callout\fP.
1549 By default, this variable contains NULL, which disables all calling out.
1550 .P
1551 Within a regular expression, (?C) indicates the points at which the external
1552 function is to be called. If you want to identify different callout points, you
1553 can put a number less than 256 after the letter C. The default value is zero.
1554 For example, this pattern has two callout points:
1555 .sp
1556 (?C1)\dabc(?C2)def
1557 .sp
1558 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
1559 automatically installed before each item in the pattern. They are all numbered
1560 255.
1561 .P
1562 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
1563 set), the external function is called. It is provided with the number of the
1564 callout, the position in the pattern, and, optionally, one item of data
1565 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
1566 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1567 description of the interface to the callout function is given in the
1568 .\" HREF
1569 \fBpcrecallout\fP
1570 .\"
1571 documentation.
1572 .P
1573 .in 0
1574 Last updated: 24 January 2006
1575 .br
1576 Copyright (c) 1997-2006 University of Cambridge.

  ViewVC Help
Powered by ViewVC 1.1.5