ViewVC logotype

Contents of /code/tags/pcre-5.0/doc/pcrepattern.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 76 - (show annotations)
Sat Feb 24 21:40:39 2007 UTC (14 years, 1 month ago) by nigel
File size: 61694 byte(s)
Tag code/trunk as code/tags/pcre-5.0.
1 .TH PCRE 3
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 A regular expression is a pattern that is matched against a subject string from
30 left to right. Most characters stand for themselves in a pattern, and match the
31 corresponding characters in the subject. As a trivial example, the pattern
32 .sp
33 The quick brown fox
34 .sp
35 matches a portion of a subject string that is identical to itself. The power of
36 regular expressions comes from the ability to include alternatives and
37 repetitions in the pattern. These are encoded in the pattern by the use of
38 \fImetacharacters\fP, which do not stand for themselves but instead are
39 interpreted in some special way.
40 .P
41 There are two different sets of metacharacters: those that are recognized
42 anywhere in the pattern except within square brackets, and those that are
43 recognized in square brackets. Outside square brackets, the metacharacters are
44 as follows:
45 .sp
46 \e general escape character with several uses
47 ^ assert start of string (or line, in multiline mode)
48 $ assert end of string (or line, in multiline mode)
49 . match any character except newline (by default)
50 [ start character class definition
51 | start of alternative branch
52 ( start subpattern
53 ) end subpattern
54 ? extends the meaning of (
55 also 0 or 1 quantifier
56 also quantifier minimizer
57 * 0 or more quantifier
58 + 1 or more quantifier
59 also "possessive quantifier"
60 { start min/max quantifier
61 .sp
62 Part of a pattern that is in square brackets is called a "character class". In
63 a character class the only metacharacters are:
64 .sp
65 \e general escape character
66 ^ negate the class, but only if the first character
67 - indicates character range
68 .\" JOIN
69 [ POSIX character class (only if followed by POSIX
70 syntax)
71 ] terminates the character class
72 .sp
73 The following sections describe the use of each of the metacharacters.
74 .
76 .rs
77 .sp
78 The backslash character has several uses. Firstly, if it is followed by a
79 non-alphanumeric character, it takes away any special meaning that character may
80 have. This use of backslash as an escape character applies both inside and
81 outside character classes.
82 .P
83 For example, if you want to match a * character, you write \e* in the pattern.
84 This escaping action applies whether or not the following character would
85 otherwise be interpreted as a metacharacter, so it is always safe to precede a
86 non-alphanumeric with backslash to specify that it stands for itself. In
87 particular, if you want to match a backslash, you write \e\e.
88 .P
89 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
90 pattern (other than in a character class) and characters between a # outside
91 a character class and the next newline character are ignored. An escaping
92 backslash can be used to include a whitespace or # character as part of the
93 pattern.
94 .P
95 If you want to remove the special meaning from a sequence of characters, you
96 can do so by putting them between \eQ and \eE. This is different from Perl in
97 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
98 Perl, $ and @ cause variable interpolation. Note the following examples:
99 .sp
100 Pattern PCRE matches Perl matches
101 .sp
102 .\" JOIN
103 \eQabc$xyz\eE abc$xyz abc followed by the
104 contents of $xyz
105 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
106 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
107 .sp
108 The \eQ...\eE sequence is recognized both inside and outside character classes.
109 .
110 .
111 .\" HTML <a name="digitsafterbackslash"></a>
112 .SS "Non-printing characters"
113 .rs
114 .sp
115 A second use of backslash provides a way of encoding non-printing characters
116 in patterns in a visible manner. There is no restriction on the appearance of
117 non-printing characters, apart from the binary zero that terminates a pattern,
118 but when a pattern is being prepared by text editing, it is usually easier to
119 use one of the following escape sequences than the binary character it
120 represents:
121 .sp
122 \ea alarm, that is, the BEL character (hex 07)
123 \ecx "control-x", where x is any character
124 \ee escape (hex 1B)
125 \ef formfeed (hex 0C)
126 \en newline (hex 0A)
127 \er carriage return (hex 0D)
128 \et tab (hex 09)
129 \eddd character with octal code ddd, or backreference
130 \exhh character with hex code hh
131 \ex{hhh..} character with hex code hhh... (UTF-8 mode only)
132 .sp
133 The precise effect of \ecx is as follows: if x is a lower case letter, it
134 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
135 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
136 7B.
137 .P
138 After \ex, from zero to two hexadecimal digits are read (letters can be in
139 upper or lower case). In UTF-8 mode, any number of hexadecimal digits may
140 appear between \ex{ and }, but the value of the character code must be less
141 than 2**31 (that is, the maximum hexadecimal value is 7FFFFFFF). If characters
142 other than hexadecimal digits appear between \ex{ and }, or if there is no
143 terminating }, this form of escape is not recognized. Instead, the initial
144 \ex will be interpreted as a basic hexadecimal escape, with no following
145 digits, giving a character whose value is zero.
146 .P
147 Characters whose value is less than 256 can be defined by either of the two
148 syntaxes for \ex when PCRE is in UTF-8 mode. There is no difference in the
149 way they are handled. For example, \exdc is exactly the same as \ex{dc}.
150 .P
151 After \e0 up to two further octal digits are read. In both cases, if there
152 are fewer than two digits, just those that are present are used. Thus the
153 sequence \e0\ex\e07 specifies two binary zeros followed by a BEL character
154 (code value 7). Make sure you supply two digits after the initial zero if the
155 pattern character that follows is itself an octal digit.
156 .P
157 The handling of a backslash followed by a digit other than 0 is complicated.
158 Outside a character class, PCRE reads it and any following digits as a decimal
159 number. If the number is less than 10, or if there have been at least that many
160 previous capturing left parentheses in the expression, the entire sequence is
161 taken as a \fIback reference\fP. A description of how this works is given
162 .\" HTML <a href="#backreferences">
163 .\" </a>
164 later,
165 .\"
166 following the discussion of
167 .\" HTML <a href="#subpattern">
168 .\" </a>
169 parenthesized subpatterns.
170 .\"
171 .P
172 Inside a character class, or if the decimal number is greater than 9 and there
173 have not been that many capturing subpatterns, PCRE re-reads up to three octal
174 digits following the backslash, and generates a single byte from the least
175 significant 8 bits of the value. Any subsequent digits stand for themselves.
176 For example:
177 .sp
178 \e040 is another way of writing a space
179 .\" JOIN
180 \e40 is the same, provided there are fewer than 40
181 previous capturing subpatterns
182 \e7 is always a back reference
183 .\" JOIN
184 \e11 might be a back reference, or another way of
185 writing a tab
186 \e011 is always a tab
187 \e0113 is a tab followed by the character "3"
188 .\" JOIN
189 \e113 might be a back reference, otherwise the
190 character with octal code 113
191 .\" JOIN
192 \e377 might be a back reference, otherwise
193 the byte consisting entirely of 1 bits
194 .\" JOIN
195 \e81 is either a back reference, or a binary zero
196 followed by the two characters "8" and "1"
197 .sp
198 Note that octal values of 100 or greater must not be introduced by a leading
199 zero, because no more than three octal digits are ever read.
200 .P
201 All the sequences that define a single byte value or a single UTF-8 character
202 (in UTF-8 mode) can be used both inside and outside character classes. In
203 addition, inside a character class, the sequence \eb is interpreted as the
204 backspace character (hex 08), and the sequence \eX is interpreted as the
205 character "X". Outside a character class, these sequences have different
206 meanings
207 .\" HTML <a href="#uniextseq">
208 .\" </a>
209 (see below).
210 .\"
211 .
212 .
213 .SS "Generic character types"
214 .rs
215 .sp
216 The third use of backslash is for specifying generic character types. The
217 following are always recognized:
218 .sp
219 \ed any decimal digit
220 \eD any character that is not a decimal digit
221 \es any whitespace character
222 \eS any character that is not a whitespace character
223 \ew any "word" character
224 \eW any "non-word" character
225 .sp
226 Each pair of escape sequences partitions the complete set of characters into
227 two disjoint sets. Any given character matches one, and only one, of each pair.
228 .P
229 These character type sequences can appear both inside and outside character
230 classes. They each match one character of the appropriate type. If the current
231 matching point is at the end of the subject string, all of them fail, since
232 there is no character to match.
233 .P
234 For compatibility with Perl, \es does not match the VT character (code 11).
235 This makes it different from the the POSIX "space" class. The \es characters
236 are HT (9), LF (10), FF (12), CR (13), and space (32).
237 .P
238 A "word" character is an underscore or any character less than 256 that is a
239 letter or digit. The definition of letters and digits is controlled by PCRE's
240 low-valued character tables, and may vary if locale-specific matching is taking
241 place (see
242 .\" HTML <a href="pcreapi.html#localesupport">
243 .\" </a>
244 "Locale support"
245 .\"
246 in the
247 .\" HREF
248 \fBpcreapi\fP
249 .\"
250 page). For example, in the "fr_FR" (French) locale, some character codes
251 greater than 128 are used for accented letters, and these are matched by \ew.
252 .P
253 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
254 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
255 character property support is available.
256 .
257 .
258 .\" HTML <a name="uniextseq"></a>
259 .SS Unicode character properties
260 .rs
261 .sp
262 When PCRE is built with Unicode character property support, three additional
263 escape sequences to match generic character types are available when UTF-8 mode
264 is selected. They are:
265 .sp
266 \ep{\fIxx\fP} a character with the \fIxx\fP property
267 \eP{\fIxx\fP} a character without the \fIxx\fP property
268 \eX an extended Unicode sequence
269 .sp
270 The property names represented by \fIxx\fP above are limited to the
271 Unicode general category properties. Each character has exactly one such
272 property, specified by a two-letter abbreviation. For compatibility with Perl,
273 negation can be specified by including a circumflex between the opening brace
274 and the property name. For example, \ep{^Lu} is the same as \eP{Lu}.
275 .P
276 If only one letter is specified with \ep or \eP, it includes all the properties
277 that start with that letter. In this case, in the absence of negation, the
278 curly brackets in the escape sequence are optional; these two examples have
279 the same effect:
280 .sp
281 \ep{L}
282 \epL
283 .sp
284 The following property codes are supported:
285 .sp
286 C Other
287 Cc Control
288 Cf Format
289 Cn Unassigned
290 Co Private use
291 Cs Surrogate
292 .sp
293 L Letter
294 Ll Lower case letter
295 Lm Modifier letter
296 Lo Other letter
297 Lt Title case letter
298 Lu Upper case letter
299 .sp
300 M Mark
301 Mc Spacing mark
302 Me Enclosing mark
303 Mn Non-spacing mark
304 .sp
305 N Number
306 Nd Decimal number
307 Nl Letter number
308 No Other number
309 .sp
310 P Punctuation
311 Pc Connector punctuation
312 Pd Dash punctuation
313 Pe Close punctuation
314 Pf Final punctuation
315 Pi Initial punctuation
316 Po Other punctuation
317 Ps Open punctuation
318 .sp
319 S Symbol
320 Sc Currency symbol
321 Sk Modifier symbol
322 Sm Mathematical symbol
323 So Other symbol
324 .sp
325 Z Separator
326 Zl Line separator
327 Zp Paragraph separator
328 Zs Space separator
329 .sp
330 Extended properties such as "Greek" or "InMusicalSymbols" are not supported by
331 PCRE.
332 .P
333 Specifying caseless matching does not affect these escape sequences. For
334 example, \ep{Lu} always matches only upper case letters.
335 .P
336 The \eX escape matches any number of Unicode characters that form an extended
337 Unicode sequence. \eX is equivalent to
338 .sp
339 (?>\ePM\epM*)
340 .sp
341 That is, it matches a character without the "mark" property, followed by zero
342 or more characters with the "mark" property, and treats the sequence as an
343 atomic group
344 .\" HTML <a href="#atomicgroup">
345 .\" </a>
346 (see below).
347 .\"
348 Characters with the "mark" property are typically accents that affect the
349 preceding character.
350 .P
351 Matching characters by Unicode property is not fast, because PCRE has to search
352 a structure that contains data for over fifteen thousand characters. That is
353 why the traditional escape sequences such as \ed and \ew do not use Unicode
354 properties in PCRE.
355 .
356 .
357 .\" HTML <a name="smallassertions"></a>
358 .SS "Simple assertions"
359 .rs
360 .sp
361 The fourth use of backslash is for certain simple assertions. An assertion
362 specifies a condition that has to be met at a particular point in a match,
363 without consuming any characters from the subject string. The use of
364 subpatterns for more complicated assertions is described
365 .\" HTML <a href="#bigassertions">
366 .\" </a>
367 below.
368 .\"
369 The backslashed
370 assertions are:
371 .sp
372 \eb matches at a word boundary
373 \eB matches when not at a word boundary
374 \eA matches at start of subject
375 \eZ matches at end of subject or before newline at end
376 \ez matches at end of subject
377 \eG matches at first matching position in subject
378 .sp
379 These assertions may not appear in character classes (but note that \eb has a
380 different meaning, namely the backspace character, inside a character class).
381 .P
382 A word boundary is a position in the subject string where the current character
383 and the previous character do not both match \ew or \eW (i.e. one matches
384 \ew and the other matches \eW), or the start or end of the string if the
385 first or last character matches \ew, respectively.
386 .P
387 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
388 dollar (described in the next section) in that they only ever match at the very
389 start and end of the subject string, whatever options are set. Thus, they are
390 independent of multiline mode. These three assertions are not affected by the
391 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
392 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
393 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
394 at a point other than the beginning of the subject, \eA can never match. The
395 difference between \eZ and \ez is that \eZ matches before a newline that is the
396 last character of the string as well as at the end of the string, whereas \ez
397 matches only at the end.
398 .P
399 The \eG assertion is true only when the current matching position is at the
400 start point of the match, as specified by the \fIstartoffset\fP argument of
401 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
402 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
403 arguments, you can mimic Perl's /g option, and it is in this kind of
404 implementation where \eG can be useful.
405 .P
406 Note, however, that PCRE's interpretation of \eG, as the start of the current
407 match, is subtly different from Perl's, which defines it as the end of the
408 previous match. In Perl, these can be different when the previously matched
409 string was empty. Because PCRE does just one match at a time, it cannot
410 reproduce this behaviour.
411 .P
412 If all the alternatives of a pattern begin with \eG, the expression is anchored
413 to the starting match position, and the "anchored" flag is set in the compiled
414 regular expression.
415 .
416 .
418 .rs
419 .sp
420 Outside a character class, in the default matching mode, the circumflex
421 character is an assertion that is true only if the current matching point is
422 at the start of the subject string. If the \fIstartoffset\fP argument of
423 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
424 option is unset. Inside a character class, circumflex has an entirely different
425 meaning
426 .\" HTML <a href="#characterclass">
427 .\" </a>
428 (see below).
429 .\"
430 .P
431 Circumflex need not be the first character of the pattern if a number of
432 alternatives are involved, but it should be the first thing in each alternative
433 in which it appears if the pattern is ever to match that branch. If all
434 possible alternatives start with a circumflex, that is, if the pattern is
435 constrained to match only at the start of the subject, it is said to be an
436 "anchored" pattern. (There are also other constructs that can cause a pattern
437 to be anchored.)
438 .P
439 A dollar character is an assertion that is true only if the current matching
440 point is at the end of the subject string, or immediately before a newline
441 character that is the last character in the string (by default). Dollar need
442 not be the last character of the pattern if a number of alternatives are
443 involved, but it should be the last item in any branch in which it appears.
444 Dollar has no special meaning in a character class.
445 .P
446 The meaning of dollar can be changed so that it matches only at the very end of
447 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
448 does not affect the \eZ assertion.
449 .P
450 The meanings of the circumflex and dollar characters are changed if the
451 PCRE_MULTILINE option is set. When this is the case, they match immediately
452 after and immediately before an internal newline character, respectively, in
453 addition to matching at the start and end of the subject string. For example,
454 the pattern /^abc$/ matches the subject string "def\enabc" (where \en
455 represents a newline character) in multiline mode, but not otherwise.
456 Consequently, patterns that are anchored in single line mode because all
457 branches start with ^ are not anchored in multiline mode, and a match for
458 circumflex is possible when the \fIstartoffset\fP argument of \fBpcre_exec()\fP
459 is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
460 set.
461 .P
462 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
463 end of the subject in both modes, and if all branches of a pattern start with
464 \eA it is always anchored, whether PCRE_MULTILINE is set or not.
465 .
466 .
468 .rs
469 .sp
470 Outside a character class, a dot in the pattern matches any one character in
471 the subject, including a non-printing character, but not (by default) newline.
472 In UTF-8 mode, a dot matches any UTF-8 character, which might be more than one
473 byte long, except (by default) newline. If the PCRE_DOTALL option is set,
474 dots match newlines as well. The handling of dot is entirely independent of the
475 handling of circumflex and dollar, the only relationship being that they both
476 involve newline characters. Dot has no special meaning in a character class.
477 .
478 .
480 .rs
481 .sp
482 Outside a character class, the escape sequence \eC matches any one byte, both
483 in and out of UTF-8 mode. Unlike a dot, it can match a newline. The feature is
484 provided in Perl in order to match individual bytes in UTF-8 mode. Because it
485 breaks up UTF-8 characters into individual bytes, what remains in the string
486 may be a malformed UTF-8 string. For this reason, the \eC escape sequence is
487 best avoided.
488 .P
489 PCRE does not allow \eC to appear in lookbehind assertions
490 .\" HTML <a href="#lookbehind">
491 .\" </a>
492 (described below),
493 .\"
494 because in UTF-8 mode this would make it impossible to calculate the length of
495 the lookbehind.
496 .
497 .
498 .\" HTML <a name="characterclass"></a>
500 .rs
501 .sp
502 An opening square bracket introduces a character class, terminated by a closing
503 square bracket. A closing square bracket on its own is not special. If a
504 closing square bracket is required as a member of the class, it should be the
505 first data character in the class (after an initial circumflex, if present) or
506 escaped with a backslash.
507 .P
508 A character class matches a single character in the subject. In UTF-8 mode, the
509 character may occupy more than one byte. A matched character must be in the set
510 of characters defined by the class, unless the first character in the class
511 definition is a circumflex, in which case the subject character must not be in
512 the set defined by the class. If a circumflex is actually required as a member
513 of the class, ensure it is not the first character, or escape it with a
514 backslash.
515 .P
516 For example, the character class [aeiou] matches any lower case vowel, while
517 [^aeiou] matches any character that is not a lower case vowel. Note that a
518 circumflex is just a convenient notation for specifying the characters that
519 are in the class by enumerating those that are not. A class that starts with a
520 circumflex is not an assertion: it still consumes a character from the subject
521 string, and therefore it fails if the current pointer is at the end of the
522 string.
523 .P
524 In UTF-8 mode, characters with values greater than 255 can be included in a
525 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
526 .P
527 When caseless matching is set, any letters in a class represent both their
528 upper case and lower case versions, so for example, a caseless [aeiou] matches
529 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
530 caseful version would. When running in UTF-8 mode, PCRE supports the concept of
531 case for characters with values greater than 128 only when it is compiled with
532 Unicode property support.
533 .P
534 The newline character is never treated in any special way in character classes,
535 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
536 such as [^a] will always match a newline.
537 .P
538 The minus (hyphen) character can be used to specify a range of characters in a
539 character class. For example, [d-m] matches any letter between d and m,
540 inclusive. If a minus character is required in a class, it must be escaped with
541 a backslash or appear in a position where it cannot be interpreted as
542 indicating a range, typically as the first or last character in the class.
543 .P
544 It is not possible to have the literal character "]" as the end character of a
545 range. A pattern such as [W-]46] is interpreted as a class of two characters
546 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
547 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
548 the end of range, so [W-\e]46] is interpreted as a class containing a range
549 followed by two other characters. The octal or hexadecimal representation of
550 "]" can also be used to end a range.
551 .P
552 Ranges operate in the collating sequence of character values. They can also be
553 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
554 mode, ranges can include characters whose values are greater than 255, for
555 example [\ex{100}-\ex{2ff}].
556 .P
557 If a range that includes letters is used when caseless matching is set, it
558 matches the letters in either case. For example, [W-c] is equivalent to
559 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
560 tables for the "fr_FR" locale are in use, [\exc8-\excb] matches accented E
561 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
562 characters with values greater than 128 only when it is compiled with Unicode
563 property support.
564 .P
565 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
566 in a character class, and add the characters that they match to the class. For
567 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
568 conveniently be used with the upper case character types to specify a more
569 restricted set of characters than the matching lower case type. For example,
570 the class [^\eW_] matches any letter or digit, but not underscore.
571 .P
572 The only metacharacters that are recognized in character classes are backslash,
573 hyphen (only where it can be interpreted as specifying a range), circumflex
574 (only at the start), opening square bracket (only when it can be interpreted as
575 introducing a POSIX class name - see the next section), and the terminating
576 closing square bracket. However, escaping other non-alphanumeric characters
577 does no harm.
578 .
579 .
581 .rs
582 .sp
583 Perl supports the POSIX notation for character classes. This uses names
584 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
585 this notation. For example,
586 .sp
587 [01[:alpha:]%]
588 .sp
589 matches "0", "1", any alphabetic character, or "%". The supported class names
590 are
591 .sp
592 alnum letters and digits
593 alpha letters
594 ascii character codes 0 - 127
595 blank space or tab only
596 cntrl control characters
597 digit decimal digits (same as \ed)
598 graph printing characters, excluding space
599 lower lower case letters
600 print printing characters, including space
601 punct printing characters, excluding letters and digits
602 space white space (not quite the same as \es)
603 upper upper case letters
604 word "word" characters (same as \ew)
605 xdigit hexadecimal digits
606 .sp
607 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
608 space (32). Notice that this list includes the VT character (code 11). This
609 makes "space" different to \es, which does not include VT (for Perl
610 compatibility).
611 .P
612 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
613 5.8. Another Perl extension is negation, which is indicated by a ^ character
614 after the colon. For example,
615 .sp
616 [12[:^digit:]]
617 .sp
618 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
619 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
620 supported, and an error is given if they are encountered.
621 .P
622 In UTF-8 mode, characters with values greater than 128 do not match any of
623 the POSIX character classes.
624 .
625 .
627 .rs
628 .sp
629 Vertical bar characters are used to separate alternative patterns. For example,
630 the pattern
631 .sp
632 gilbert|sullivan
633 .sp
634 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
635 and an empty alternative is permitted (matching the empty string).
636 The matching process tries each alternative in turn, from left to right,
637 and the first one that succeeds is used. If the alternatives are within a
638 subpattern
639 .\" HTML <a href="#subpattern">
640 .\" </a>
641 (defined below),
642 .\"
643 "succeeds" means matching the rest of the main pattern as well as the
644 alternative in the subpattern.
645 .
646 .
648 .rs
649 .sp
651 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
652 Perl option letters enclosed between "(?" and ")". The option letters are
653 .sp
656 s for PCRE_DOTALL
658 .sp
659 For example, (?im) sets caseless, multiline matching. It is also possible to
660 unset these options by preceding the letter with a hyphen, and a combined
661 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
662 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
663 permitted. If a letter appears both before and after the hyphen, the option is
664 unset.
665 .P
666 When an option change occurs at top level (that is, not inside subpattern
667 parentheses), the change applies to the remainder of the pattern that follows.
668 If the change is placed right at the start of a pattern, PCRE extracts it into
669 the global options (and it will therefore show up in data extracted by the
670 \fBpcre_fullinfo()\fP function).
671 .P
672 An option change within a subpattern affects only that part of the current
673 pattern that follows it, so
674 .sp
675 (a(?i)b)c
676 .sp
677 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
678 By this means, options can be made to have different settings in different
679 parts of the pattern. Any changes made in one alternative do carry on
680 into subsequent branches within the same subpattern. For example,
681 .sp
682 (a(?i)b|c)
683 .sp
684 matches "ab", "aB", "c", and "C", even though when matching "C" the first
685 branch is abandoned before the option setting. This is because the effects of
686 option settings happen at compile time. There would be some very weird
687 behaviour otherwise.
688 .P
689 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
690 same way as the Perl-compatible options by using the characters U and X
691 respectively. The (?X) flag setting is special in that it must always occur
692 earlier in the pattern than any of the additional features it turns on, even
693 when it is at top level. It is best to put it at the start.
694 .
695 .
696 .\" HTML <a name="subpattern"></a>
698 .rs
699 .sp
700 Subpatterns are delimited by parentheses (round brackets), which can be nested.
701 Turning part of a pattern into a subpattern does two things:
702 .sp
703 1. It localizes a set of alternatives. For example, the pattern
704 .sp
705 cat(aract|erpillar|)
706 .sp
707 matches one of the words "cat", "cataract", or "caterpillar". Without the
708 parentheses, it would match "cataract", "erpillar" or the empty string.
709 .sp
710 2. It sets up the subpattern as a capturing subpattern. This means that, when
711 the whole pattern matches, that portion of the subject string that matched the
712 subpattern is passed back to the caller via the \fIovector\fP argument of
713 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
714 from 1) to obtain numbers for the capturing subpatterns.
715 .P
716 For example, if the string "the red king" is matched against the pattern
717 .sp
718 the ((red|white) (king|queen))
719 .sp
720 the captured substrings are "red king", "red", and "king", and are numbered 1,
721 2, and 3, respectively.
722 .P
723 The fact that plain parentheses fulfil two functions is not always helpful.
724 There are often times when a grouping subpattern is required without a
725 capturing requirement. If an opening parenthesis is followed by a question mark
726 and a colon, the subpattern does not do any capturing, and is not counted when
727 computing the number of any subsequent capturing subpatterns. For example, if
728 the string "the white queen" is matched against the pattern
729 .sp
730 the ((?:red|white) (king|queen))
731 .sp
732 the captured substrings are "white queen" and "queen", and are numbered 1 and
733 2. The maximum number of capturing subpatterns is 65535, and the maximum depth
734 of nesting of all subpatterns, both capturing and non-capturing, is 200.
735 .P
736 As a convenient shorthand, if any option settings are required at the start of
737 a non-capturing subpattern, the option letters may appear between the "?" and
738 the ":". Thus the two patterns
739 .sp
740 (?i:saturday|sunday)
741 (?:(?i)saturday|sunday)
742 .sp
743 match exactly the same set of strings. Because alternative branches are tried
744 from left to right, and options are not reset until the end of the subpattern
745 is reached, an option setting in one branch does affect subsequent branches, so
746 the above patterns match "SUNDAY" as well as "Saturday".
747 .
748 .
750 .rs
751 .sp
752 Identifying capturing parentheses by number is simple, but it can be very hard
753 to keep track of the numbers in complicated regular expressions. Furthermore,
754 if an expression is modified, the numbers may change. To help with this
755 difficulty, PCRE supports the naming of subpatterns, something that Perl does
756 not provide. The Python syntax (?P<name>...) is used. Names consist of
757 alphanumeric characters and underscores, and must be unique within a pattern.
758 .P
759 Named capturing parentheses are still allocated numbers as well as names. The
760 PCRE API provides function calls for extracting the name-to-number translation
761 table from a compiled pattern. There is also a convenience function for
762 extracting a captured substring by name. For further details see the
763 .\" HREF
764 \fBpcreapi\fP
765 .\"
766 documentation.
767 .
768 .
770 .rs
771 .sp
772 Repetition is specified by quantifiers, which can follow any of the following
773 items:
774 .sp
775 a literal data character
776 the . metacharacter
777 the \eC escape sequence
778 the \eX escape sequence (in UTF-8 mode with Unicode properties)
779 an escape such as \ed that matches a single character
780 a character class
781 a back reference (see next section)
782 a parenthesized subpattern (unless it is an assertion)
783 .sp
784 The general repetition quantifier specifies a minimum and maximum number of
785 permitted matches, by giving the two numbers in curly brackets (braces),
786 separated by a comma. The numbers must be less than 65536, and the first must
787 be less than or equal to the second. For example:
788 .sp
789 z{2,4}
790 .sp
791 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
792 character. If the second number is omitted, but the comma is present, there is
793 no upper limit; if the second number and the comma are both omitted, the
794 quantifier specifies an exact number of required matches. Thus
795 .sp
796 [aeiou]{3,}
797 .sp
798 matches at least 3 successive vowels, but may match many more, while
799 .sp
800 \ed{8}
801 .sp
802 matches exactly 8 digits. An opening curly bracket that appears in a position
803 where a quantifier is not allowed, or one that does not match the syntax of a
804 quantifier, is taken as a literal character. For example, {,6} is not a
805 quantifier, but a literal string of four characters.
806 .P
807 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
808 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
809 which is represented by a two-byte sequence. Similarly, when Unicode property
810 support is available, \eX{3} matches three Unicode extended sequences, each of
811 which may be several bytes long (and they may be of different lengths).
812 .P
813 The quantifier {0} is permitted, causing the expression to behave as if the
814 previous item and the quantifier were not present.
815 .P
816 For convenience (and historical compatibility) the three most common
817 quantifiers have single-character abbreviations:
818 .sp
819 * is equivalent to {0,}
820 + is equivalent to {1,}
821 ? is equivalent to {0,1}
822 .sp
823 It is possible to construct infinite loops by following a subpattern that can
824 match no characters with a quantifier that has no upper limit, for example:
825 .sp
826 (a?)*
827 .sp
828 Earlier versions of Perl and PCRE used to give an error at compile time for
829 such patterns. However, because there are cases where this can be useful, such
830 patterns are now accepted, but if any repetition of the subpattern does in fact
831 match no characters, the loop is forcibly broken.
832 .P
833 By default, the quantifiers are "greedy", that is, they match as much as
834 possible (up to the maximum number of permitted times), without causing the
835 rest of the pattern to fail. The classic example of where this gives problems
836 is in trying to match comments in C programs. These appear between /* and */
837 and within the comment, individual * and / characters may appear. An attempt to
838 match C comments by applying the pattern
839 .sp
840 /\e*.*\e*/
841 .sp
842 to the string
843 .sp
844 /* first comment */ not comment /* second comment */
845 .sp
846 fails, because it matches the entire string owing to the greediness of the .*
847 item.
848 .P
849 However, if a quantifier is followed by a question mark, it ceases to be
850 greedy, and instead matches the minimum number of times possible, so the
851 pattern
852 .sp
853 /\e*.*?\e*/
854 .sp
855 does the right thing with the C comments. The meaning of the various
856 quantifiers is not otherwise changed, just the preferred number of matches.
857 Do not confuse this use of question mark with its use as a quantifier in its
858 own right. Because it has two uses, it can sometimes appear doubled, as in
859 .sp
860 \ed??\ed
861 .sp
862 which matches one digit by preference, but can match two if that is the only
863 way the rest of the pattern matches.
864 .P
865 If the PCRE_UNGREEDY option is set (an option which is not available in Perl),
866 the quantifiers are not greedy by default, but individual ones can be made
867 greedy by following them with a question mark. In other words, it inverts the
868 default behaviour.
869 .P
870 When a parenthesized subpattern is quantified with a minimum repeat count that
871 is greater than 1 or with a limited maximum, more memory is required for the
872 compiled pattern, in proportion to the size of the minimum or maximum.
873 .P
874 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
875 to Perl's /s) is set, thus allowing the . to match newlines, the pattern is
876 implicitly anchored, because whatever follows will be tried against every
877 character position in the subject string, so there is no point in retrying the
878 overall match at any position after the first. PCRE normally treats such a
879 pattern as though it were preceded by \eA.
880 .P
881 In cases where it is known that the subject string contains no newlines, it is
882 worth setting PCRE_DOTALL in order to obtain this optimization, or
883 alternatively using ^ to indicate anchoring explicitly.
884 .P
885 However, there is one situation where the optimization cannot be used. When .*
886 is inside capturing parentheses that are the subject of a backreference
887 elsewhere in the pattern, a match at the start may fail, and a later one
888 succeed. Consider, for example:
889 .sp
890 (.*)abc\e1
891 .sp
892 If the subject is "xyz123abc123" the match point is the fourth character. For
893 this reason, such a pattern is not implicitly anchored.
894 .P
895 When a capturing subpattern is repeated, the value captured is the substring
896 that matched the final iteration. For example, after
897 .sp
898 (tweedle[dume]{3}\es*)+
899 .sp
900 has matched "tweedledum tweedledee" the value of the captured substring is
901 "tweedledee". However, if there are nested capturing subpatterns, the
902 corresponding captured values may have been set in previous iterations. For
903 example, after
904 .sp
905 /(a|(b))+/
906 .sp
907 matches "aba" the value of the second captured substring is "b".
908 .
909 .
910 .\" HTML <a name="atomicgroup"></a>
912 .rs
913 .sp
914 With both maximizing and minimizing repetition, failure of what follows
915 normally causes the repeated item to be re-evaluated to see if a different
916 number of repeats allows the rest of the pattern to match. Sometimes it is
917 useful to prevent this, either to change the nature of the match, or to cause
918 it fail earlier than it otherwise might, when the author of the pattern knows
919 there is no point in carrying on.
920 .P
921 Consider, for example, the pattern \ed+foo when applied to the subject line
922 .sp
923 123456bar
924 .sp
925 After matching all 6 digits and then failing to match "foo", the normal
926 action of the matcher is to try again with only 5 digits matching the \ed+
927 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
928 (a term taken from Jeffrey Friedl's book) provides the means for specifying
929 that once a subpattern has matched, it is not to be re-evaluated in this way.
930 .P
931 If we use atomic grouping for the previous example, the matcher would give up
932 immediately on failing to match "foo" the first time. The notation is a kind of
933 special parenthesis, starting with (?> as in this example:
934 .sp
935 (?>\ed+)foo
936 .sp
937 This kind of parenthesis "locks up" the part of the pattern it contains once
938 it has matched, and a failure further into the pattern is prevented from
939 backtracking into it. Backtracking past it to previous items, however, works as
940 normal.
941 .P
942 An alternative description is that a subpattern of this type matches the string
943 of characters that an identical standalone pattern would match, if anchored at
944 the current point in the subject string.
945 .P
946 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
947 the above example can be thought of as a maximizing repeat that must swallow
948 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
949 number of digits they match in order to make the rest of the pattern match,
950 (?>\ed+) can only match an entire sequence of digits.
951 .P
952 Atomic groups in general can of course contain arbitrarily complicated
953 subpatterns, and can be nested. However, when the subpattern for an atomic
954 group is just a single repeated item, as in the example above, a simpler
955 notation, called a "possessive quantifier" can be used. This consists of an
956 additional + character following a quantifier. Using this notation, the
957 previous example can be rewritten as
958 .sp
959 \ed++foo
960 .sp
961 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
962 option is ignored. They are a convenient notation for the simpler forms of
963 atomic group. However, there is no difference in the meaning or processing of a
964 possessive quantifier and the equivalent atomic group.
965 .P
966 The possessive quantifier syntax is an extension to the Perl syntax. It
967 originates in Sun's Java package.
968 .P
969 When a pattern contains an unlimited repeat inside a subpattern that can itself
970 be repeated an unlimited number of times, the use of an atomic group is the
971 only way to avoid some failing matches taking a very long time indeed. The
972 pattern
973 .sp
974 (\eD+|<\ed+>)*[!?]
975 .sp
976 matches an unlimited number of substrings that either consist of non-digits, or
977 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
978 quickly. However, if it is applied to
979 .sp
980 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
981 .sp
982 it takes a long time before reporting failure. This is because the string can
983 be divided between the internal \eD+ repeat and the external * repeat in a
984 large number of ways, and all have to be tried. (The example uses [!?] rather
985 than a single character at the end, because both PCRE and Perl have an
986 optimization that allows for fast failure when a single character is used. They
987 remember the last single character that is required for a match, and fail early
988 if it is not present in the string.) If the pattern is changed so that it uses
989 an atomic group, like this:
990 .sp
991 ((?>\eD+)|<\ed+>)*[!?]
992 .sp
993 sequences of non-digits cannot be broken, and failure happens quickly.
994 .
995 .
996 .\" HTML <a name="backreferences"></a>
998 .rs
999 .sp
1000 Outside a character class, a backslash followed by a digit greater than 0 (and
1001 possibly further digits) is a back reference to a capturing subpattern earlier
1002 (that is, to its left) in the pattern, provided there have been that many
1003 previous capturing left parentheses.
1004 .P
1005 However, if the decimal number following the backslash is less than 10, it is
1006 always taken as a back reference, and causes an error only if there are not
1007 that many capturing left parentheses in the entire pattern. In other words, the
1008 parentheses that are referenced need not be to the left of the reference for
1009 numbers less than 10. See the subsection entitled "Non-printing characters"
1010 .\" HTML <a href="#digitsafterbackslash">
1011 .\" </a>
1012 above
1013 .\"
1014 for further details of the handling of digits following a backslash.
1015 .P
1016 A back reference matches whatever actually matched the capturing subpattern in
1017 the current subject string, rather than anything matching the subpattern
1018 itself (see
1019 .\" HTML <a href="#subpatternsassubroutines">
1020 .\" </a>
1021 "Subpatterns as subroutines"
1022 .\"
1023 below for a way of doing that). So the pattern
1024 .sp
1025 (sens|respons)e and \e1ibility
1026 .sp
1027 matches "sense and sensibility" and "response and responsibility", but not
1028 "sense and responsibility". If caseful matching is in force at the time of the
1029 back reference, the case of letters is relevant. For example,
1030 .sp
1031 ((?i)rah)\es+\e1
1032 .sp
1033 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1034 capturing subpattern is matched caselessly.
1035 .P
1036 Back references to named subpatterns use the Python syntax (?P=name). We could
1037 rewrite the above example as follows:
1038 .sp
1039 (?<p1>(?i)rah)\es+(?P=p1)
1040 .sp
1041 There may be more than one back reference to the same subpattern. If a
1042 subpattern has not actually been used in a particular match, any back
1043 references to it always fail. For example, the pattern
1044 .sp
1045 (a|(bc))\e2
1046 .sp
1047 always fails if it starts to match "a" rather than "bc". Because there may be
1048 many capturing parentheses in a pattern, all digits following the backslash are
1049 taken as part of a potential back reference number. If the pattern continues
1050 with a digit character, some delimiter must be used to terminate the back
1051 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1052 Otherwise an empty comment (see
1053 .\" HTML <a href="#comments">
1054 .\" </a>
1055 "Comments"
1056 .\"
1057 below) can be used.
1058 .P
1059 A back reference that occurs inside the parentheses to which it refers fails
1060 when the subpattern is first used, so, for example, (a\e1) never matches.
1061 However, such references can be useful inside repeated subpatterns. For
1062 example, the pattern
1063 .sp
1064 (a|b\e1)+
1065 .sp
1066 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1067 the subpattern, the back reference matches the character string corresponding
1068 to the previous iteration. In order for this to work, the pattern must be such
1069 that the first iteration does not need to match the back reference. This can be
1070 done using alternation, as in the example above, or by a quantifier with a
1071 minimum of zero.
1072 .
1073 .
1074 .\" HTML <a name="bigassertions"></a>
1076 .rs
1077 .sp
1078 An assertion is a test on the characters following or preceding the current
1079 matching point that does not actually consume any characters. The simple
1080 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1081 .\" HTML <a href="#smallassertions">
1082 .\" </a>
1083 above.
1084 .\"
1085 .P
1086 More complicated assertions are coded as subpatterns. There are two kinds:
1087 those that look ahead of the current position in the subject string, and those
1088 that look behind it. An assertion subpattern is matched in the normal way,
1089 except that it does not cause the current matching position to be changed.
1090 .P
1091 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1092 because it makes no sense to assert the same thing several times. If any kind
1093 of assertion contains capturing subpatterns within it, these are counted for
1094 the purposes of numbering the capturing subpatterns in the whole pattern.
1095 However, substring capturing is carried out only for positive assertions,
1096 because it does not make sense for negative assertions.
1097 .
1098 .
1099 .SS "Lookahead assertions"
1100 .rs
1101 .sp
1102 Lookahead assertions start
1103 with (?= for positive assertions and (?! for negative assertions. For example,
1104 .sp
1105 \ew+(?=;)
1106 .sp
1107 matches a word followed by a semicolon, but does not include the semicolon in
1108 the match, and
1109 .sp
1110 foo(?!bar)
1111 .sp
1112 matches any occurrence of "foo" that is not followed by "bar". Note that the
1113 apparently similar pattern
1114 .sp
1115 (?!foo)bar
1116 .sp
1117 does not find an occurrence of "bar" that is preceded by something other than
1118 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1119 (?!foo) is always true when the next three characters are "bar". A
1120 lookbehind assertion is needed to achieve the other effect.
1121 .P
1122 If you want to force a matching failure at some point in a pattern, the most
1123 convenient way to do it is with (?!) because an empty string always matches, so
1124 an assertion that requires there not to be an empty string must always fail.
1125 .
1126 .
1127 .\" HTML <a name="lookbehind"></a>
1128 .SS "Lookbehind assertions"
1129 .rs
1130 .sp
1131 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1132 negative assertions. For example,
1133 .sp
1134 (?<!foo)bar
1135 .sp
1136 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1137 a lookbehind assertion are restricted such that all the strings it matches must
1138 have a fixed length. However, if there are several alternatives, they do not
1139 all have to have the same fixed length. Thus
1140 .sp
1141 (?<=bullock|donkey)
1142 .sp
1143 is permitted, but
1144 .sp
1145 (?<!dogs?|cats?)
1146 .sp
1147 causes an error at compile time. Branches that match different length strings
1148 are permitted only at the top level of a lookbehind assertion. This is an
1149 extension compared with Perl (at least for 5.8), which requires all branches to
1150 match the same length of string. An assertion such as
1151 .sp
1152 (?<=ab(c|de))
1153 .sp
1154 is not permitted, because its single top-level branch can match two different
1155 lengths, but it is acceptable if rewritten to use two top-level branches:
1156 .sp
1157 (?<=abc|abde)
1158 .sp
1159 The implementation of lookbehind assertions is, for each alternative, to
1160 temporarily move the current position back by the fixed width and then try to
1161 match. If there are insufficient characters before the current position, the
1162 match is deemed to fail.
1163 .P
1164 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1165 to appear in lookbehind assertions, because it makes it impossible to calculate
1166 the length of the lookbehind. The \eX escape, which can match different numbers
1167 of bytes, is also not permitted.
1168 .P
1169 Atomic groups can be used in conjunction with lookbehind assertions to specify
1170 efficient matching at the end of the subject string. Consider a simple pattern
1171 such as
1172 .sp
1173 abcd$
1174 .sp
1175 when applied to a long string that does not match. Because matching proceeds
1176 from left to right, PCRE will look for each "a" in the subject and then see if
1177 what follows matches the rest of the pattern. If the pattern is specified as
1178 .sp
1179 ^.*abcd$
1180 .sp
1181 the initial .* matches the entire string at first, but when this fails (because
1182 there is no following "a"), it backtracks to match all but the last character,
1183 then all but the last two characters, and so on. Once again the search for "a"
1184 covers the entire string, from right to left, so we are no better off. However,
1185 if the pattern is written as
1186 .sp
1187 ^(?>.*)(?<=abcd)
1188 .sp
1189 or, equivalently, using the possessive quantifier syntax,
1190 .sp
1191 ^.*+(?<=abcd)
1192 .sp
1193 there can be no backtracking for the .* item; it can match only the entire
1194 string. The subsequent lookbehind assertion does a single test on the last four
1195 characters. If it fails, the match fails immediately. For long strings, this
1196 approach makes a significant difference to the processing time.
1197 .
1198 .
1199 .SS "Using multiple assertions"
1200 .rs
1201 .sp
1202 Several assertions (of any sort) may occur in succession. For example,
1203 .sp
1204 (?<=\ed{3})(?<!999)foo
1205 .sp
1206 matches "foo" preceded by three digits that are not "999". Notice that each of
1207 the assertions is applied independently at the same point in the subject
1208 string. First there is a check that the previous three characters are all
1209 digits, and then there is a check that the same three characters are not "999".
1210 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1211 of which are digits and the last three of which are not "999". For example, it
1212 doesn't match "123abcfoo". A pattern to do that is
1213 .sp
1214 (?<=\ed{3}...)(?<!999)foo
1215 .sp
1216 This time the first assertion looks at the preceding six characters, checking
1217 that the first three are digits, and then the second assertion checks that the
1218 preceding three characters are not "999".
1219 .P
1220 Assertions can be nested in any combination. For example,
1221 .sp
1222 (?<=(?<!foo)bar)baz
1223 .sp
1224 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1225 preceded by "foo", while
1226 .sp
1227 (?<=\ed{3}(?!999)...)foo
1228 .sp
1229 is another pattern that matches "foo" preceded by three digits and any three
1230 characters that are not "999".
1231 .
1232 .
1234 .rs
1235 .sp
1236 It is possible to cause the matching process to obey a subpattern
1237 conditionally or to choose between two alternative subpatterns, depending on
1238 the result of an assertion, or whether a previous capturing subpattern matched
1239 or not. The two possible forms of conditional subpattern are
1240 .sp
1241 (?(condition)yes-pattern)
1242 (?(condition)yes-pattern|no-pattern)
1243 .sp
1244 If the condition is satisfied, the yes-pattern is used; otherwise the
1245 no-pattern (if present) is used. If there are more than two alternatives in the
1246 subpattern, a compile-time error occurs.
1247 .P
1248 There are three kinds of condition. If the text between the parentheses
1249 consists of a sequence of digits, the condition is satisfied if the capturing
1250 subpattern of that number has previously matched. The number must be greater
1251 than zero. Consider the following pattern, which contains non-significant white
1252 space to make it more readable (assume the PCRE_EXTENDED option) and to divide
1253 it into three parts for ease of discussion:
1254 .sp
1255 ( \e( )? [^()]+ (?(1) \e) )
1256 .sp
1257 The first part matches an optional opening parenthesis, and if that
1258 character is present, sets it as the first captured substring. The second part
1259 matches one or more characters that are not parentheses. The third part is a
1260 conditional subpattern that tests whether the first set of parentheses matched
1261 or not. If they did, that is, if subject started with an opening parenthesis,
1262 the condition is true, and so the yes-pattern is executed and a closing
1263 parenthesis is required. Otherwise, since no-pattern is not present, the
1264 subpattern matches nothing. In other words, this pattern matches a sequence of
1265 non-parentheses, optionally enclosed in parentheses.
1266 .P
1267 If the condition is the string (R), it is satisfied if a recursive call to the
1268 pattern or subpattern has been made. At "top level", the condition is false.
1269 This is a PCRE extension. Recursive patterns are described in the next section.
1270 .P
1271 If the condition is not a sequence of digits or (R), it must be an assertion.
1272 This may be a positive or negative lookahead or lookbehind assertion. Consider
1273 this pattern, again containing non-significant white space, and with the two
1274 alternatives on the second line:
1275 .sp
1276 (?(?=[^a-z]*[a-z])
1277 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1278 .sp
1279 The condition is a positive lookahead assertion that matches an optional
1280 sequence of non-letters followed by a letter. In other words, it tests for the
1281 presence of at least one letter in the subject. If a letter is found, the
1282 subject is matched against the first alternative; otherwise it is matched
1283 against the second. This pattern matches strings in one of the two forms
1284 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1285 .
1286 .
1287 .\" HTML <a name="comments"></a>
1289 .rs
1290 .sp
1291 The sequence (?# marks the start of a comment that continues up to the next
1292 closing parenthesis. Nested parentheses are not permitted. The characters
1293 that make up a comment play no part in the pattern matching at all.
1294 .P
1295 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1296 character class introduces a comment that continues up to the next newline
1297 character in the pattern.
1298 .
1299 .
1301 .rs
1302 .sp
1303 Consider the problem of matching a string in parentheses, allowing for
1304 unlimited nested parentheses. Without the use of recursion, the best that can
1305 be done is to use a pattern that matches up to some fixed depth of nesting. It
1306 is not possible to handle an arbitrary nesting depth. Perl provides a facility
1307 that allows regular expressions to recurse (amongst other things). It does this
1308 by interpolating Perl code in the expression at run time, and the code can
1309 refer to the expression itself. A Perl pattern to solve the parentheses problem
1310 can be created like this:
1311 .sp
1312 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1313 .sp
1314 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1315 recursively to the pattern in which it appears. Obviously, PCRE cannot support
1316 the interpolation of Perl code. Instead, it supports some special syntax for
1317 recursion of the entire pattern, and also for individual subpattern recursion.
1318 .P
1319 The special item that consists of (? followed by a number greater than zero and
1320 a closing parenthesis is a recursive call of the subpattern of the given
1321 number, provided that it occurs inside that subpattern. (If not, it is a
1322 "subroutine" call, which is described in the next section.) The special item
1323 (?R) is a recursive call of the entire regular expression.
1324 .P
1325 For example, this PCRE pattern solves the nested parentheses problem (assume
1326 the PCRE_EXTENDED option is set so that white space is ignored):
1327 .sp
1328 \e( ( (?>[^()]+) | (?R) )* \e)
1329 .sp
1330 First it matches an opening parenthesis. Then it matches any number of
1331 substrings which can either be a sequence of non-parentheses, or a recursive
1332 match of the pattern itself (that is a correctly parenthesized substring).
1333 Finally there is a closing parenthesis.
1334 .P
1335 If this were part of a larger pattern, you would not want to recurse the entire
1336 pattern, so instead you could use this:
1337 .sp
1338 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1339 .sp
1340 We have put the pattern into parentheses, and caused the recursion to refer to
1341 them instead of the whole pattern. In a larger pattern, keeping track of
1342 parenthesis numbers can be tricky. It may be more convenient to use named
1343 parentheses instead. For this, PCRE uses (?P>name), which is an extension to
1344 the Python syntax that PCRE uses for named parentheses (Perl does not provide
1345 named parentheses). We could rewrite the above example as follows:
1346 .sp
1347 (?P<pn> \e( ( (?>[^()]+) | (?P>pn) )* \e) )
1348 .sp
1349 This particular example pattern contains nested unlimited repeats, and so the
1350 use of atomic grouping for matching strings of non-parentheses is important
1351 when applying the pattern to strings that do not match. For example, when this
1352 pattern is applied to
1353 .sp
1354 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1355 .sp
1356 it yields "no match" quickly. However, if atomic grouping is not used,
1357 the match runs for a very long time indeed because there are so many different
1358 ways the + and * repeats can carve up the subject, and all have to be tested
1359 before failure can be reported.
1360 .P
1361 At the end of a match, the values set for any capturing subpatterns are those
1362 from the outermost level of the recursion at which the subpattern value is set.
1363 If you want to obtain intermediate values, a callout function can be used (see
1364 the next section and the
1365 .\" HREF
1366 \fBpcrecallout\fP
1367 .\"
1368 documentation). If the pattern above is matched against
1369 .sp
1370 (ab(cd)ef)
1371 .sp
1372 the value for the capturing parentheses is "ef", which is the last value taken
1373 on at the top level. If additional parentheses are added, giving
1374 .sp
1375 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1376 ^ ^
1377 ^ ^
1378 .sp
1379 the string they capture is "ab(cd)ef", the contents of the top level
1380 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1381 has to obtain extra memory to store data during a recursion, which it does by
1382 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1383 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1384 .P
1385 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1386 Consider this pattern, which matches text in angle brackets, allowing for
1387 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1388 recursing), whereas any characters are permitted at the outer level.
1389 .sp
1390 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1391 .sp
1392 In this pattern, (?(R) is the start of a conditional subpattern, with two
1393 different alternatives for the recursive and non-recursive cases. The (?R) item
1394 is the actual recursive call.
1395 .
1396 .
1397 .\" HTML <a name="subpatternsassubroutines"></a>
1399 .rs
1400 .sp
1401 If the syntax for a recursive subpattern reference (either by number or by
1402 name) is used outside the parentheses to which it refers, it operates like a
1403 subroutine in a programming language. An earlier example pointed out that the
1404 pattern
1405 .sp
1406 (sens|respons)e and \e1ibility
1407 .sp
1408 matches "sense and sensibility" and "response and responsibility", but not
1409 "sense and responsibility". If instead the pattern
1410 .sp
1411 (sens|respons)e and (?1)ibility
1412 .sp
1413 is used, it does match "sense and responsibility" as well as the other two
1414 strings. Such references must, however, follow the subpattern to which they
1415 refer.
1416 .
1417 .
1419 .rs
1420 .sp
1421 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1422 code to be obeyed in the middle of matching a regular expression. This makes it
1423 possible, amongst other things, to extract different substrings that match the
1424 same pair of parentheses when there is a repetition.
1425 .P
1426 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1427 code. The feature is called "callout". The caller of PCRE provides an external
1428 function by putting its entry point in the global variable \fIpcre_callout\fP.
1429 By default, this variable contains NULL, which disables all calling out.
1430 .P
1431 Within a regular expression, (?C) indicates the points at which the external
1432 function is to be called. If you want to identify different callout points, you
1433 can put a number less than 256 after the letter C. The default value is zero.
1434 For example, this pattern has two callout points:
1435 .sp
1436 (?C1)\dabc(?C2)def
1437 .sp
1438 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
1439 automatically installed before each item in the pattern. They are all numbered
1440 255.
1441 .P
1442 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
1443 set), the external function is called. It is provided with the number of the
1444 callout, the position in the pattern, and, optionally, one item of data
1445 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
1446 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1447 description of the interface to the callout function is given in the
1448 .\" HREF
1449 \fBpcrecallout\fP
1450 .\"
1451 documentation.
1452 .P
1453 .in 0
1454 Last updated: 09 September 2004
1455 .br
1456 Copyright (c) 1997-2004 University of Cambridge.

  ViewVC Help
Powered by ViewVC 1.1.5