ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 738 - (show annotations)
Fri Oct 21 09:04:01 2011 UTC (4 years, 1 month ago) by ph10
File size: 119215 byte(s)
Error occurred while calculating annotation data.
Source tidies for 8.20 release.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13 also supports some alternative regular expression syntax (which does not
14 conflict with the Perl syntax) in order to provide some compatibility with
15 regular expressions in Python, .NET, and Oniguruma.
16 .P
17 Perl's regular expressions are described in its own documentation, and
18 regular expressions in general are covered in a number of books, some of which
19 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20 published by O'Reilly, covers regular expressions in great detail. This
21 description of PCRE's regular expressions is intended as reference material.
22 .P
23 The original operation of PCRE was on strings of one-byte characters. However,
24 there is now also support for UTF-8 character strings. To use this,
25 PCRE must be built to include UTF-8 support, and you must call
26 \fBpcre_compile()\fP or \fBpcre_compile2()\fP with the PCRE_UTF8 option. There
27 is also a special sequence that can be given at the start of a pattern:
28 .sp
29 (*UTF8)
30 .sp
31 Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
32 option. This feature is not Perl-compatible. How setting UTF-8 mode affects
33 pattern matching is mentioned in several places below. There is also a summary
34 of UTF-8 features in the
35 .\" HREF
36 \fBpcreunicode\fP
37 .\"
38 page.
39 .P
40 Another special sequence that may appear at the start of a pattern or in
41 combination with (*UTF8) is:
42 .sp
43 (*UCP)
44 .sp
45 This has the same effect as setting the PCRE_UCP option: it causes sequences
46 such as \ed and \ew to use Unicode properties to determine character types,
47 instead of recognizing only characters with codes less than 128 via a lookup
48 table.
49 .P
50 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
51 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
52 also some more of these special sequences that are concerned with the handling
53 of newlines; they are described below.
54 .P
55 The remainder of this document discusses the patterns that are supported by
56 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
57 From release 6.0, PCRE offers a second matching function,
58 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
59 Perl-compatible. Some of the features discussed below are not available when
60 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
61 alternative function, and how it differs from the normal function, are
62 discussed in the
63 .\" HREF
64 \fBpcrematching\fP
65 .\"
66 page.
67 .
68 .
69 .\" HTML <a name="newlines"></a>
71 .rs
72 .sp
73 PCRE supports five different conventions for indicating line breaks in
74 strings: a single CR (carriage return) character, a single LF (linefeed)
75 character, the two-character sequence CRLF, any of the three preceding, or any
76 Unicode newline sequence. The
77 .\" HREF
78 \fBpcreapi\fP
79 .\"
80 page has
81 .\" HTML <a href="pcreapi.html#newlines">
82 .\" </a>
83 further discussion
84 .\"
85 about newlines, and shows how to set the newline convention in the
86 \fIoptions\fP arguments for the compiling and matching functions.
87 .P
88 It is also possible to specify a newline convention by starting a pattern
89 string with one of the following five sequences:
90 .sp
91 (*CR) carriage return
92 (*LF) linefeed
93 (*CRLF) carriage return, followed by linefeed
94 (*ANYCRLF) any of the three above
95 (*ANY) all Unicode newline sequences
96 .sp
97 These override the default and the options given to \fBpcre_compile()\fP or
98 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
99 newline sequence, the pattern
100 .sp
101 (*CR)a.b
102 .sp
103 changes the convention to CR. That pattern matches "a\enb" because LF is no
104 longer a newline. Note that these special settings, which are not
105 Perl-compatible, are recognized only at the very start of a pattern, and that
106 they must be in upper case. If more than one of them is present, the last one
107 is used.
108 .P
109 The newline convention affects the interpretation of the dot metacharacter when
110 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
111 affect what the \eR escape sequence matches. By default, this is any Unicode
112 newline sequence, for Perl compatibility. However, this can be changed; see the
113 description of \eR in the section entitled
114 .\" HTML <a href="#newlineseq">
115 .\" </a>
116 "Newline sequences"
117 .\"
118 below. A change of \eR setting can be combined with a change of newline
119 convention.
120 .
121 .
123 .rs
124 .sp
125 A regular expression is a pattern that is matched against a subject string from
126 left to right. Most characters stand for themselves in a pattern, and match the
127 corresponding characters in the subject. As a trivial example, the pattern
128 .sp
129 The quick brown fox
130 .sp
131 matches a portion of a subject string that is identical to itself. When
132 caseless matching is specified (the PCRE_CASELESS option), letters are matched
133 independently of case. In UTF-8 mode, PCRE always understands the concept of
134 case for characters whose values are less than 128, so caseless matching is
135 always possible. For characters with higher values, the concept of case is
136 supported if PCRE is compiled with Unicode property support, but not otherwise.
137 If you want to use caseless matching for characters 128 and above, you must
138 ensure that PCRE is compiled with Unicode property support as well as with
139 UTF-8 support.
140 .P
141 The power of regular expressions comes from the ability to include alternatives
142 and repetitions in the pattern. These are encoded in the pattern by the use of
143 \fImetacharacters\fP, which do not stand for themselves but instead are
144 interpreted in some special way.
145 .P
146 There are two different sets of metacharacters: those that are recognized
147 anywhere in the pattern except within square brackets, and those that are
148 recognized within square brackets. Outside square brackets, the metacharacters
149 are as follows:
150 .sp
151 \e general escape character with several uses
152 ^ assert start of string (or line, in multiline mode)
153 $ assert end of string (or line, in multiline mode)
154 . match any character except newline (by default)
155 [ start character class definition
156 | start of alternative branch
157 ( start subpattern
158 ) end subpattern
159 ? extends the meaning of (
160 also 0 or 1 quantifier
161 also quantifier minimizer
162 * 0 or more quantifier
163 + 1 or more quantifier
164 also "possessive quantifier"
165 { start min/max quantifier
166 .sp
167 Part of a pattern that is in square brackets is called a "character class". In
168 a character class the only metacharacters are:
169 .sp
170 \e general escape character
171 ^ negate the class, but only if the first character
172 - indicates character range
173 .\" JOIN
174 [ POSIX character class (only if followed by POSIX
175 syntax)
176 ] terminates the character class
177 .sp
178 The following sections describe the use of each of the metacharacters.
179 .
180 .
182 .rs
183 .sp
184 The backslash character has several uses. Firstly, if it is followed by a
185 character that is not a number or a letter, it takes away any special meaning
186 that character may have. This use of backslash as an escape character applies
187 both inside and outside character classes.
188 .P
189 For example, if you want to match a * character, you write \e* in the pattern.
190 This escaping action applies whether or not the following character would
191 otherwise be interpreted as a metacharacter, so it is always safe to precede a
192 non-alphanumeric with backslash to specify that it stands for itself. In
193 particular, if you want to match a backslash, you write \e\e.
194 .P
195 In UTF-8 mode, only ASCII numbers and letters have any special meaning after a
196 backslash. All other characters (in particular, those whose codepoints are
197 greater than 127) are treated as literals.
198 .P
199 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
200 pattern (other than in a character class) and characters between a # outside
201 a character class and the next newline are ignored. An escaping backslash can
202 be used to include a whitespace or # character as part of the pattern.
203 .P
204 If you want to remove the special meaning from a sequence of characters, you
205 can do so by putting them between \eQ and \eE. This is different from Perl in
206 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
207 Perl, $ and @ cause variable interpolation. Note the following examples:
208 .sp
209 Pattern PCRE matches Perl matches
210 .sp
211 .\" JOIN
212 \eQabc$xyz\eE abc$xyz abc followed by the
213 contents of $xyz
214 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
215 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
216 .sp
217 The \eQ...\eE sequence is recognized both inside and outside character classes.
218 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
219 by \eE later in the pattern, the literal interpretation continues to the end of
220 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
221 a character class, this causes an error, because the character class is not
222 terminated.
223 .
224 .
225 .\" HTML <a name="digitsafterbackslash"></a>
226 .SS "Non-printing characters"
227 .rs
228 .sp
229 A second use of backslash provides a way of encoding non-printing characters
230 in patterns in a visible manner. There is no restriction on the appearance of
231 non-printing characters, apart from the binary zero that terminates a pattern,
232 but when a pattern is being prepared by text editing, it is often easier to use
233 one of the following escape sequences than the binary character it represents:
234 .sp
235 \ea alarm, that is, the BEL character (hex 07)
236 \ecx "control-x", where x is any ASCII character
237 \ee escape (hex 1B)
238 \ef formfeed (hex 0C)
239 \en linefeed (hex 0A)
240 \er carriage return (hex 0D)
241 \et tab (hex 09)
242 \eddd character with octal code ddd, or back reference
243 \exhh character with hex code hh
244 \ex{hhh..} character with hex code hhh..
245 .sp
246 The precise effect of \ecx is as follows: if x is a lower case letter, it
247 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
248 Thus \ecz becomes hex 1A (z is 7A), but \ec{ becomes hex 3B ({ is 7B), while
249 \ec; becomes hex 7B (; is 3B). If the byte following \ec has a value greater
250 than 127, a compile-time error occurs. This locks out non-ASCII characters in
251 both byte mode and UTF-8 mode. (When PCRE is compiled in EBCDIC mode, all byte
252 values are valid. A lower case letter is converted to upper case, and then the
253 0xc0 bits are flipped.)
254 .P
255 After \ex, from zero to two hexadecimal digits are read (letters can be in
256 upper or lower case). Any number of hexadecimal digits may appear between \ex{
257 and }, but the value of the character code must be less than 256 in non-UTF-8
258 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
259 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
260 point, which is 10FFFF.
261 .P
262 If characters other than hexadecimal digits appear between \ex{ and }, or if
263 there is no terminating }, this form of escape is not recognized. Instead, the
264 initial \ex will be interpreted as a basic hexadecimal escape, with no
265 following digits, giving a character whose value is zero.
266 .P
267 Characters whose value is less than 256 can be defined by either of the two
268 syntaxes for \ex. There is no difference in the way they are handled. For
269 example, \exdc is exactly the same as \ex{dc}.
270 .P
271 After \e0 up to two further octal digits are read. If there are fewer than two
272 digits, just those that are present are used. Thus the sequence \e0\ex\e07
273 specifies two binary zeros followed by a BEL character (code value 7). Make
274 sure you supply two digits after the initial zero if the pattern character that
275 follows is itself an octal digit.
276 .P
277 The handling of a backslash followed by a digit other than 0 is complicated.
278 Outside a character class, PCRE reads it and any following digits as a decimal
279 number. If the number is less than 10, or if there have been at least that many
280 previous capturing left parentheses in the expression, the entire sequence is
281 taken as a \fIback reference\fP. A description of how this works is given
282 .\" HTML <a href="#backreferences">
283 .\" </a>
284 later,
285 .\"
286 following the discussion of
287 .\" HTML <a href="#subpattern">
288 .\" </a>
289 parenthesized subpatterns.
290 .\"
291 .P
292 Inside a character class, or if the decimal number is greater than 9 and there
293 have not been that many capturing subpatterns, PCRE re-reads up to three octal
294 digits following the backslash, and uses them to generate a data character. Any
295 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
296 character specified in octal must be less than \e400. In UTF-8 mode, values up
297 to \e777 are permitted. For example:
298 .sp
299 \e040 is another way of writing a space
300 .\" JOIN
301 \e40 is the same, provided there are fewer than 40
302 previous capturing subpatterns
303 \e7 is always a back reference
304 .\" JOIN
305 \e11 might be a back reference, or another way of
306 writing a tab
307 \e011 is always a tab
308 \e0113 is a tab followed by the character "3"
309 .\" JOIN
310 \e113 might be a back reference, otherwise the
311 character with octal code 113
312 .\" JOIN
313 \e377 might be a back reference, otherwise
314 the byte consisting entirely of 1 bits
315 .\" JOIN
316 \e81 is either a back reference, or a binary zero
317 followed by the two characters "8" and "1"
318 .sp
319 Note that octal values of 100 or greater must not be introduced by a leading
320 zero, because no more than three octal digits are ever read.
321 .P
322 All the sequences that define a single character value can be used both inside
323 and outside character classes. In addition, inside a character class, the
324 sequence \eb is interpreted as the backspace character (hex 08). The sequences
325 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
326 unrecognized escape sequences, they are treated as the literal characters "B",
327 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
328 set. Outside a character class, these sequences have different meanings.
329 .
330 .
331 .SS "Absolute and relative back references"
332 .rs
333 .sp
334 The sequence \eg followed by an unsigned or a negative number, optionally
335 enclosed in braces, is an absolute or relative back reference. A named back
336 reference can be coded as \eg{name}. Back references are discussed
337 .\" HTML <a href="#backreferences">
338 .\" </a>
339 later,
340 .\"
341 following the discussion of
342 .\" HTML <a href="#subpattern">
343 .\" </a>
344 parenthesized subpatterns.
345 .\"
346 .
347 .
348 .SS "Absolute and relative subroutine calls"
349 .rs
350 .sp
351 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
352 a number enclosed either in angle brackets or single quotes, is an alternative
353 syntax for referencing a subpattern as a "subroutine". Details are discussed
354 .\" HTML <a href="#onigurumasubroutines">
355 .\" </a>
356 later.
357 .\"
358 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
359 synonymous. The former is a back reference; the latter is a
360 .\" HTML <a href="#subpatternsassubroutines">
361 .\" </a>
362 subroutine
363 .\"
364 call.
365 .
366 .
367 .\" HTML <a name="genericchartypes"></a>
368 .SS "Generic character types"
369 .rs
370 .sp
371 Another use of backslash is for specifying generic character types:
372 .sp
373 \ed any decimal digit
374 \eD any character that is not a decimal digit
375 \eh any horizontal whitespace character
376 \eH any character that is not a horizontal whitespace character
377 \es any whitespace character
378 \eS any character that is not a whitespace character
379 \ev any vertical whitespace character
380 \eV any character that is not a vertical whitespace character
381 \ew any "word" character
382 \eW any "non-word" character
383 .sp
384 There is also the single sequence \eN, which matches a non-newline character.
385 This is the same as
386 .\" HTML <a href="#fullstopdot">
387 .\" </a>
388 the "." metacharacter
389 .\"
390 when PCRE_DOTALL is not set.
391 .P
392 Each pair of lower and upper case escape sequences partitions the complete set
393 of characters into two disjoint sets. Any given character matches one, and only
394 one, of each pair. The sequences can appear both inside and outside character
395 classes. They each match one character of the appropriate type. If the current
396 matching point is at the end of the subject string, all of them fail, because
397 there is no character to match.
398 .P
399 For compatibility with Perl, \es does not match the VT character (code 11).
400 This makes it different from the the POSIX "space" class. The \es characters
401 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
402 included in a Perl script, \es may match the VT character. In PCRE, it never
403 does.
404 .P
405 A "word" character is an underscore or any character that is a letter or digit.
406 By default, the definition of letters and digits is controlled by PCRE's
407 low-valued character tables, and may vary if locale-specific matching is taking
408 place (see
409 .\" HTML <a href="pcreapi.html#localesupport">
410 .\" </a>
411 "Locale support"
412 .\"
413 in the
414 .\" HREF
415 \fBpcreapi\fP
416 .\"
417 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
418 or "french" in Windows, some character codes greater than 128 are used for
419 accented letters, and these are then matched by \ew. The use of locales with
420 Unicode is discouraged.
421 .P
422 By default, in UTF-8 mode, characters with values greater than 128 never match
423 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
424 their original meanings from before UTF-8 support was available, mainly for
425 efficiency reasons. However, if PCRE is compiled with Unicode property support,
426 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
427 properties are used to determine character types, as follows:
428 .sp
429 \ed any character that \ep{Nd} matches (decimal digit)
430 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
431 \ew any character that \ep{L} or \ep{N} matches, plus underscore
432 .sp
433 The upper case escapes match the inverse sets of characters. Note that \ed
434 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
435 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
436 \eB because they are defined in terms of \ew and \eW. Matching these sequences
437 is noticeably slower when PCRE_UCP is set.
438 .P
439 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
440 release 5.10. In contrast to the other sequences, which match only ASCII
441 characters by default, these always match certain high-valued codepoints in
442 UTF-8 mode, whether or not PCRE_UCP is set. The horizontal space characters
443 are:
444 .sp
445 U+0009 Horizontal tab
446 U+0020 Space
447 U+00A0 Non-break space
448 U+1680 Ogham space mark
449 U+180E Mongolian vowel separator
450 U+2000 En quad
451 U+2001 Em quad
452 U+2002 En space
453 U+2003 Em space
454 U+2004 Three-per-em space
455 U+2005 Four-per-em space
456 U+2006 Six-per-em space
457 U+2007 Figure space
458 U+2008 Punctuation space
459 U+2009 Thin space
460 U+200A Hair space
461 U+202F Narrow no-break space
462 U+205F Medium mathematical space
463 U+3000 Ideographic space
464 .sp
465 The vertical space characters are:
466 .sp
467 U+000A Linefeed
468 U+000B Vertical tab
469 U+000C Formfeed
470 U+000D Carriage return
471 U+0085 Next line
472 U+2028 Line separator
473 U+2029 Paragraph separator
474 .
475 .
476 .\" HTML <a name="newlineseq"></a>
477 .SS "Newline sequences"
478 .rs
479 .sp
480 Outside a character class, by default, the escape sequence \eR matches any
481 Unicode newline sequence. In non-UTF-8 mode \eR is equivalent to the following:
482 .sp
483 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
484 .sp
485 This is an example of an "atomic group", details of which are given
486 .\" HTML <a href="#atomicgroup">
487 .\" </a>
488 below.
489 .\"
490 This particular group matches either the two-character sequence CR followed by
491 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
492 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
493 line, U+0085). The two-character sequence is treated as a single unit that
494 cannot be split.
495 .P
496 In UTF-8 mode, two additional characters whose codepoints are greater than 255
497 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
498 Unicode character property support is not needed for these characters to be
499 recognized.
500 .P
501 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
502 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
503 either at compile time or when the pattern is matched. (BSR is an abbrevation
504 for "backslash R".) This can be made the default when PCRE is built; if this is
505 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
506 It is also possible to specify these settings by starting a pattern string with
507 one of the following sequences:
508 .sp
509 (*BSR_ANYCRLF) CR, LF, or CRLF only
510 (*BSR_UNICODE) any Unicode newline sequence
511 .sp
512 These override the default and the options given to \fBpcre_compile()\fP or
513 \fBpcre_compile2()\fP, but they can be overridden by options given to
514 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
515 which are not Perl-compatible, are recognized only at the very start of a
516 pattern, and that they must be in upper case. If more than one of them is
517 present, the last one is used. They can be combined with a change of newline
518 convention; for example, a pattern can start with:
519 .sp
521 .sp
522 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
523 a character class, \eR is treated as an unrecognized escape sequence, and so
524 matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
525 .
526 .
527 .\" HTML <a name="uniextseq"></a>
528 .SS Unicode character properties
529 .rs
530 .sp
531 When PCRE is built with Unicode character property support, three additional
532 escape sequences that match characters with specific properties are available.
533 When not in UTF-8 mode, these sequences are of course limited to testing
534 characters whose codepoints are less than 256, but they do work in this mode.
535 The extra escape sequences are:
536 .sp
537 \ep{\fIxx\fP} a character with the \fIxx\fP property
538 \eP{\fIxx\fP} a character without the \fIxx\fP property
539 \eX an extended Unicode sequence
540 .sp
541 The property names represented by \fIxx\fP above are limited to the Unicode
542 script names, the general category properties, "Any", which matches any
543 character (including newline), and some special PCRE properties (described
544 in the
545 .\" HTML <a href="#extraprops">
546 .\" </a>
547 next section).
548 .\"
549 Other Perl properties such as "InMusicalSymbols" are not currently supported by
550 PCRE. Note that \eP{Any} does not match any characters, so always causes a
551 match failure.
552 .P
553 Sets of Unicode characters are defined as belonging to certain scripts. A
554 character from one of these sets can be matched using a script name. For
555 example:
556 .sp
557 \ep{Greek}
558 \eP{Han}
559 .sp
560 Those that are not part of an identified script are lumped together as
561 "Common". The current list of scripts is:
562 .P
563 Arabic,
564 Armenian,
565 Avestan,
566 Balinese,
567 Bamum,
568 Bengali,
569 Bopomofo,
570 Braille,
571 Buginese,
572 Buhid,
573 Canadian_Aboriginal,
574 Carian,
575 Cham,
576 Cherokee,
577 Common,
578 Coptic,
579 Cuneiform,
580 Cypriot,
581 Cyrillic,
582 Deseret,
583 Devanagari,
584 Egyptian_Hieroglyphs,
585 Ethiopic,
586 Georgian,
587 Glagolitic,
588 Gothic,
589 Greek,
590 Gujarati,
591 Gurmukhi,
592 Han,
593 Hangul,
594 Hanunoo,
595 Hebrew,
596 Hiragana,
597 Imperial_Aramaic,
598 Inherited,
599 Inscriptional_Pahlavi,
600 Inscriptional_Parthian,
601 Javanese,
602 Kaithi,
603 Kannada,
604 Katakana,
605 Kayah_Li,
606 Kharoshthi,
607 Khmer,
608 Lao,
609 Latin,
610 Lepcha,
611 Limbu,
612 Linear_B,
613 Lisu,
614 Lycian,
615 Lydian,
616 Malayalam,
617 Meetei_Mayek,
618 Mongolian,
619 Myanmar,
620 New_Tai_Lue,
621 Nko,
622 Ogham,
623 Old_Italic,
624 Old_Persian,
625 Old_South_Arabian,
626 Old_Turkic,
627 Ol_Chiki,
628 Oriya,
629 Osmanya,
630 Phags_Pa,
631 Phoenician,
632 Rejang,
633 Runic,
634 Samaritan,
635 Saurashtra,
636 Shavian,
637 Sinhala,
638 Sundanese,
639 Syloti_Nagri,
640 Syriac,
641 Tagalog,
642 Tagbanwa,
643 Tai_Le,
644 Tai_Tham,
645 Tai_Viet,
646 Tamil,
647 Telugu,
648 Thaana,
649 Thai,
650 Tibetan,
651 Tifinagh,
652 Ugaritic,
653 Vai,
654 Yi.
655 .P
656 Each character has exactly one Unicode general category property, specified by
657 a two-letter abbreviation. For compatibility with Perl, negation can be
658 specified by including a circumflex between the opening brace and the property
659 name. For example, \ep{^Lu} is the same as \eP{Lu}.
660 .P
661 If only one letter is specified with \ep or \eP, it includes all the general
662 category properties that start with that letter. In this case, in the absence
663 of negation, the curly brackets in the escape sequence are optional; these two
664 examples have the same effect:
665 .sp
666 \ep{L}
667 \epL
668 .sp
669 The following general category property codes are supported:
670 .sp
671 C Other
672 Cc Control
673 Cf Format
674 Cn Unassigned
675 Co Private use
676 Cs Surrogate
677 .sp
678 L Letter
679 Ll Lower case letter
680 Lm Modifier letter
681 Lo Other letter
682 Lt Title case letter
683 Lu Upper case letter
684 .sp
685 M Mark
686 Mc Spacing mark
687 Me Enclosing mark
688 Mn Non-spacing mark
689 .sp
690 N Number
691 Nd Decimal number
692 Nl Letter number
693 No Other number
694 .sp
695 P Punctuation
696 Pc Connector punctuation
697 Pd Dash punctuation
698 Pe Close punctuation
699 Pf Final punctuation
700 Pi Initial punctuation
701 Po Other punctuation
702 Ps Open punctuation
703 .sp
704 S Symbol
705 Sc Currency symbol
706 Sk Modifier symbol
707 Sm Mathematical symbol
708 So Other symbol
709 .sp
710 Z Separator
711 Zl Line separator
712 Zp Paragraph separator
713 Zs Space separator
714 .sp
715 The special property L& is also supported: it matches a character that has
716 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
717 a modifier or "other".
718 .P
719 The Cs (Surrogate) property applies only to characters in the range U+D800 to
720 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
721 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
722 (see the discussion of PCRE_NO_UTF8_CHECK in the
723 .\" HREF
724 \fBpcreapi\fP
725 .\"
726 page). Perl does not support the Cs property.
727 .P
728 The long synonyms for property names that Perl supports (such as \ep{Letter})
729 are not supported by PCRE, nor is it permitted to prefix any of these
730 properties with "Is".
731 .P
732 No character that is in the Unicode table has the Cn (unassigned) property.
733 Instead, this property is assumed for any code point that is not in the
734 Unicode table.
735 .P
736 Specifying caseless matching does not affect these escape sequences. For
737 example, \ep{Lu} always matches only upper case letters.
738 .P
739 The \eX escape matches any number of Unicode characters that form an extended
740 Unicode sequence. \eX is equivalent to
741 .sp
742 (?>\ePM\epM*)
743 .sp
744 That is, it matches a character without the "mark" property, followed by zero
745 or more characters with the "mark" property, and treats the sequence as an
746 atomic group
747 .\" HTML <a href="#atomicgroup">
748 .\" </a>
749 (see below).
750 .\"
751 Characters with the "mark" property are typically accents that affect the
752 preceding character. None of them have codepoints less than 256, so in
753 non-UTF-8 mode \eX matches any one character.
754 .P
755 Note that recent versions of Perl have changed \eX to match what Unicode calls
756 an "extended grapheme cluster", which has a more complicated definition.
757 .P
758 Matching characters by Unicode property is not fast, because PCRE has to search
759 a structure that contains data for over fifteen thousand characters. That is
760 why the traditional escape sequences such as \ed and \ew do not use Unicode
761 properties in PCRE by default, though you can make them do so by setting the
762 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
763 (*UCP).
764 .
765 .
766 .\" HTML <a name="extraprops"></a>
767 .SS PCRE's additional properties
768 .rs
769 .sp
770 As well as the standard Unicode properties described in the previous
771 section, PCRE supports four more that make it possible to convert traditional
772 escape sequences such as \ew and \es and POSIX character classes to use Unicode
773 properties. PCRE uses these non-standard, non-Perl properties internally when
774 PCRE_UCP is set. They are:
775 .sp
776 Xan Any alphanumeric character
777 Xps Any POSIX space character
778 Xsp Any Perl space character
779 Xwd Any Perl "word" character
780 .sp
781 Xan matches characters that have either the L (letter) or the N (number)
782 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
783 carriage return, and any other character that has the Z (separator) property.
784 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
785 same characters as Xan, plus underscore.
786 .
787 .
788 .\" HTML <a name="resetmatchstart"></a>
789 .SS "Resetting the match start"
790 .rs
791 .sp
792 The escape sequence \eK causes any previously matched characters not to be
793 included in the final matched sequence. For example, the pattern:
794 .sp
795 foo\eKbar
796 .sp
797 matches "foobar", but reports that it has matched "bar". This feature is
798 similar to a lookbehind assertion
799 .\" HTML <a href="#lookbehind">
800 .\" </a>
801 (described below).
802 .\"
803 However, in this case, the part of the subject before the real match does not
804 have to be of fixed length, as lookbehind assertions do. The use of \eK does
805 not interfere with the setting of
806 .\" HTML <a href="#subpattern">
807 .\" </a>
808 captured substrings.
809 .\"
810 For example, when the pattern
811 .sp
812 (foo)\eKbar
813 .sp
814 matches "foobar", the first substring is still set to "foo".
815 .P
816 Perl documents that the use of \eK within assertions is "not well defined". In
817 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
818 ignored in negative assertions.
819 .
820 .
821 .\" HTML <a name="smallassertions"></a>
822 .SS "Simple assertions"
823 .rs
824 .sp
825 The final use of backslash is for certain simple assertions. An assertion
826 specifies a condition that has to be met at a particular point in a match,
827 without consuming any characters from the subject string. The use of
828 subpatterns for more complicated assertions is described
829 .\" HTML <a href="#bigassertions">
830 .\" </a>
831 below.
832 .\"
833 The backslashed assertions are:
834 .sp
835 \eb matches at a word boundary
836 \eB matches when not at a word boundary
837 \eA matches at the start of the subject
838 \eZ matches at the end of the subject
839 also matches before a newline at the end of the subject
840 \ez matches only at the end of the subject
841 \eG matches at the first matching position in the subject
842 .sp
843 Inside a character class, \eb has a different meaning; it matches the backspace
844 character. If any other of these assertions appears in a character class, by
845 default it matches the corresponding literal character (for example, \eB
846 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
847 escape sequence" error is generated instead.
848 .P
849 A word boundary is a position in the subject string where the current character
850 and the previous character do not both match \ew or \eW (i.e. one matches
851 \ew and the other matches \eW), or the start or end of the string if the
852 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
853 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
854 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
855 of word" or "end of word" metasequence. However, whatever follows \eb normally
856 determines which it is. For example, the fragment \eba matches "a" at the start
857 of a word.
858 .P
859 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
860 dollar (described in the next section) in that they only ever match at the very
861 start and end of the subject string, whatever options are set. Thus, they are
862 independent of multiline mode. These three assertions are not affected by the
863 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
864 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
865 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
866 at a point other than the beginning of the subject, \eA can never match. The
867 difference between \eZ and \ez is that \eZ matches before a newline at the end
868 of the string as well as at the very end, whereas \ez matches only at the end.
869 .P
870 The \eG assertion is true only when the current matching position is at the
871 start point of the match, as specified by the \fIstartoffset\fP argument of
872 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
873 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
874 arguments, you can mimic Perl's /g option, and it is in this kind of
875 implementation where \eG can be useful.
876 .P
877 Note, however, that PCRE's interpretation of \eG, as the start of the current
878 match, is subtly different from Perl's, which defines it as the end of the
879 previous match. In Perl, these can be different when the previously matched
880 string was empty. Because PCRE does just one match at a time, it cannot
881 reproduce this behaviour.
882 .P
883 If all the alternatives of a pattern begin with \eG, the expression is anchored
884 to the starting match position, and the "anchored" flag is set in the compiled
885 regular expression.
886 .
887 .
889 .rs
890 .sp
891 Outside a character class, in the default matching mode, the circumflex
892 character is an assertion that is true only if the current matching point is
893 at the start of the subject string. If the \fIstartoffset\fP argument of
894 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
895 option is unset. Inside a character class, circumflex has an entirely different
896 meaning
897 .\" HTML <a href="#characterclass">
898 .\" </a>
899 (see below).
900 .\"
901 .P
902 Circumflex need not be the first character of the pattern if a number of
903 alternatives are involved, but it should be the first thing in each alternative
904 in which it appears if the pattern is ever to match that branch. If all
905 possible alternatives start with a circumflex, that is, if the pattern is
906 constrained to match only at the start of the subject, it is said to be an
907 "anchored" pattern. (There are also other constructs that can cause a pattern
908 to be anchored.)
909 .P
910 A dollar character is an assertion that is true only if the current matching
911 point is at the end of the subject string, or immediately before a newline
912 at the end of the string (by default). Dollar need not be the last character of
913 the pattern if a number of alternatives are involved, but it should be the last
914 item in any branch in which it appears. Dollar has no special meaning in a
915 character class.
916 .P
917 The meaning of dollar can be changed so that it matches only at the very end of
918 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
919 does not affect the \eZ assertion.
920 .P
921 The meanings of the circumflex and dollar characters are changed if the
922 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
923 immediately after internal newlines as well as at the start of the subject
924 string. It does not match after a newline that ends the string. A dollar
925 matches before any newlines in the string, as well as at the very end, when
926 PCRE_MULTILINE is set. When newline is specified as the two-character
927 sequence CRLF, isolated CR and LF characters do not indicate newlines.
928 .P
929 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
930 \en represents a newline) in multiline mode, but not otherwise. Consequently,
931 patterns that are anchored in single line mode because all branches start with
932 ^ are not anchored in multiline mode, and a match for circumflex is possible
933 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
934 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
935 .P
936 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
937 end of the subject in both modes, and if all branches of a pattern start with
938 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
939 .
940 .
941 .\" HTML <a name="fullstopdot"></a>
943 .rs
944 .sp
945 Outside a character class, a dot in the pattern matches any one character in
946 the subject string except (by default) a character that signifies the end of a
947 line. In UTF-8 mode, the matched character may be more than one byte long.
948 .P
949 When a line ending is defined as a single character, dot never matches that
950 character; when the two-character sequence CRLF is used, dot does not match CR
951 if it is immediately followed by LF, but otherwise it matches all characters
952 (including isolated CRs and LFs). When any Unicode line endings are being
953 recognized, dot does not match CR or LF or any of the other line ending
954 characters.
955 .P
956 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
957 option is set, a dot matches any one character, without exception. If the
958 two-character sequence CRLF is present in the subject string, it takes two dots
959 to match it.
960 .P
961 The handling of dot is entirely independent of the handling of circumflex and
962 dollar, the only relationship being that they both involve newlines. Dot has no
963 special meaning in a character class.
964 .P
965 The escape sequence \eN behaves like a dot, except that it is not affected by
966 the PCRE_DOTALL option. In other words, it matches any character except one
967 that signifies the end of a line.
968 .
969 .
971 .rs
972 .sp
973 Outside a character class, the escape sequence \eC matches any one byte, both
974 in and out of UTF-8 mode. Unlike a dot, it always matches line-ending
975 characters. The feature is provided in Perl in order to match individual bytes
976 in UTF-8 mode, but it is unclear how it can usefully be used. Because \eC
977 breaks up characters into individual bytes, matching one byte with \eC in UTF-8
978 mode means that the rest of the string may start with a malformed UTF-8
979 character. This has undefined results, because PCRE assumes that it is dealing
980 with valid UTF-8 strings (and by default it checks this at the start of
981 processing unless the PCRE_NO_UTF8_CHECK option is used).
982 .P
983 PCRE does not allow \eC to appear in lookbehind assertions
984 .\" HTML <a href="#lookbehind">
985 .\" </a>
986 (described below),
987 .\"
988 because in UTF-8 mode this would make it impossible to calculate the length of
989 the lookbehind.
990 .P
991 In general, the \eC escape sequence is best avoided in UTF-8 mode. However, one
992 way of using it that avoids the problem of malformed UTF-8 characters is to
993 use a lookahead to check the length of the next character, as in this pattern
994 (ignore white space and line breaks):
995 .sp
996 (?| (?=[\ex00-\ex7f])(\eC) |
997 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
998 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
999 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1000 .sp
1001 A group that starts with (?| resets the capturing parentheses numbers in each
1002 alternative (see
1003 .\" HTML <a href="#dupsubpatternnumber">
1004 .\" </a>
1005 "Duplicate Subpattern Numbers"
1006 .\"
1007 below). The assertions at the start of each branch check the next UTF-8
1008 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1009 character's individual bytes are then captured by the appropriate number of
1010 groups.
1011 .
1012 .
1013 .\" HTML <a name="characterclass"></a>
1015 .rs
1016 .sp
1017 An opening square bracket introduces a character class, terminated by a closing
1018 square bracket. A closing square bracket on its own is not special by default.
1019 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1020 bracket causes a compile-time error. If a closing square bracket is required as
1021 a member of the class, it should be the first data character in the class
1022 (after an initial circumflex, if present) or escaped with a backslash.
1023 .P
1024 A character class matches a single character in the subject. In UTF-8 mode, the
1025 character may be more than one byte long. A matched character must be in the
1026 set of characters defined by the class, unless the first character in the class
1027 definition is a circumflex, in which case the subject character must not be in
1028 the set defined by the class. If a circumflex is actually required as a member
1029 of the class, ensure it is not the first character, or escape it with a
1030 backslash.
1031 .P
1032 For example, the character class [aeiou] matches any lower case vowel, while
1033 [^aeiou] matches any character that is not a lower case vowel. Note that a
1034 circumflex is just a convenient notation for specifying the characters that
1035 are in the class by enumerating those that are not. A class that starts with a
1036 circumflex is not an assertion; it still consumes a character from the subject
1037 string, and therefore it fails if the current pointer is at the end of the
1038 string.
1039 .P
1040 In UTF-8 mode, characters with values greater than 255 can be included in a
1041 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1042 .P
1043 When caseless matching is set, any letters in a class represent both their
1044 upper case and lower case versions, so for example, a caseless [aeiou] matches
1045 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1046 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1047 case for characters whose values are less than 128, so caseless matching is
1048 always possible. For characters with higher values, the concept of case is
1049 supported if PCRE is compiled with Unicode property support, but not otherwise.
1050 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1051 you must ensure that PCRE is compiled with Unicode property support as well as
1052 with UTF-8 support.
1053 .P
1054 Characters that might indicate line breaks are never treated in any special way
1055 when matching character classes, whatever line-ending sequence is in use, and
1056 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1057 such as [^a] always matches one of these characters.
1058 .P
1059 The minus (hyphen) character can be used to specify a range of characters in a
1060 character class. For example, [d-m] matches any letter between d and m,
1061 inclusive. If a minus character is required in a class, it must be escaped with
1062 a backslash or appear in a position where it cannot be interpreted as
1063 indicating a range, typically as the first or last character in the class.
1064 .P
1065 It is not possible to have the literal character "]" as the end character of a
1066 range. A pattern such as [W-]46] is interpreted as a class of two characters
1067 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1068 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1069 the end of range, so [W-\e]46] is interpreted as a class containing a range
1070 followed by two other characters. The octal or hexadecimal representation of
1071 "]" can also be used to end a range.
1072 .P
1073 Ranges operate in the collating sequence of character values. They can also be
1074 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1075 mode, ranges can include characters whose values are greater than 255, for
1076 example [\ex{100}-\ex{2ff}].
1077 .P
1078 If a range that includes letters is used when caseless matching is set, it
1079 matches the letters in either case. For example, [W-c] is equivalent to
1080 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1081 tables for a French locale are in use, [\exc8-\excb] matches accented E
1082 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1083 characters with values greater than 128 only when it is compiled with Unicode
1084 property support.
1085 .P
1086 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1087 \eV, \ew, and \eW may appear in a character class, and add the characters that
1088 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1089 digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1090 and their upper case partners, just as it does when they appear outside a
1091 character class, as described in the section entitled
1092 .\" HTML <a href="#genericchartypes">
1093 .\" </a>
1094 "Generic character types"
1095 .\"
1096 above. The escape sequence \eb has a different meaning inside a character
1097 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1098 are not special inside a character class. Like any other unrecognized escape
1099 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1100 default, but cause an error if the PCRE_EXTRA option is set.
1101 .P
1102 A circumflex can conveniently be used with the upper case character types to
1103 specify a more restricted set of characters than the matching lower case type.
1104 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1105 whereas [\ew] includes underscore. A positive character class should be read as
1106 "something OR something OR ..." and a negative class as "NOT something AND NOT
1107 something AND NOT ...".
1108 .P
1109 The only metacharacters that are recognized in character classes are backslash,
1110 hyphen (only where it can be interpreted as specifying a range), circumflex
1111 (only at the start), opening square bracket (only when it can be interpreted as
1112 introducing a POSIX class name - see the next section), and the terminating
1113 closing square bracket. However, escaping other non-alphanumeric characters
1114 does no harm.
1115 .
1116 .
1118 .rs
1119 .sp
1120 Perl supports the POSIX notation for character classes. This uses names
1121 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1122 this notation. For example,
1123 .sp
1124 [01[:alpha:]%]
1125 .sp
1126 matches "0", "1", any alphabetic character, or "%". The supported class names
1127 are:
1128 .sp
1129 alnum letters and digits
1130 alpha letters
1131 ascii character codes 0 - 127
1132 blank space or tab only
1133 cntrl control characters
1134 digit decimal digits (same as \ed)
1135 graph printing characters, excluding space
1136 lower lower case letters
1137 print printing characters, including space
1138 punct printing characters, excluding letters and digits and space
1139 space white space (not quite the same as \es)
1140 upper upper case letters
1141 word "word" characters (same as \ew)
1142 xdigit hexadecimal digits
1143 .sp
1144 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1145 space (32). Notice that this list includes the VT character (code 11). This
1146 makes "space" different to \es, which does not include VT (for Perl
1147 compatibility).
1148 .P
1149 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1150 5.8. Another Perl extension is negation, which is indicated by a ^ character
1151 after the colon. For example,
1152 .sp
1153 [12[:^digit:]]
1154 .sp
1155 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1156 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1157 supported, and an error is given if they are encountered.
1158 .P
1159 By default, in UTF-8 mode, characters with values greater than 128 do not match
1160 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1161 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1162 character properties are used. This is achieved by replacing the POSIX classes
1163 by other sequences, as follows:
1164 .sp
1165 [:alnum:] becomes \ep{Xan}
1166 [:alpha:] becomes \ep{L}
1167 [:blank:] becomes \eh
1168 [:digit:] becomes \ep{Nd}
1169 [:lower:] becomes \ep{Ll}
1170 [:space:] becomes \ep{Xps}
1171 [:upper:] becomes \ep{Lu}
1172 [:word:] becomes \ep{Xwd}
1173 .sp
1174 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1175 classes are unchanged, and match only characters with code points less than
1176 128.
1177 .
1178 .
1180 .rs
1181 .sp
1182 Vertical bar characters are used to separate alternative patterns. For example,
1183 the pattern
1184 .sp
1185 gilbert|sullivan
1186 .sp
1187 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1188 and an empty alternative is permitted (matching the empty string). The matching
1189 process tries each alternative in turn, from left to right, and the first one
1190 that succeeds is used. If the alternatives are within a subpattern
1191 .\" HTML <a href="#subpattern">
1192 .\" </a>
1193 (defined below),
1194 .\"
1195 "succeeds" means matching the rest of the main pattern as well as the
1196 alternative in the subpattern.
1197 .
1198 .
1200 .rs
1201 .sp
1202 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1203 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1204 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1205 The option letters are
1206 .sp
1207 i for PCRE_CASELESS
1209 s for PCRE_DOTALL
1210 x for PCRE_EXTENDED
1211 .sp
1212 For example, (?im) sets caseless, multiline matching. It is also possible to
1213 unset these options by preceding the letter with a hyphen, and a combined
1214 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1215 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1216 permitted. If a letter appears both before and after the hyphen, the option is
1217 unset.
1218 .P
1219 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1220 changed in the same way as the Perl-compatible options by using the characters
1221 J, U and X respectively.
1222 .P
1223 When one of these option changes occurs at top level (that is, not inside
1224 subpattern parentheses), the change applies to the remainder of the pattern
1225 that follows. If the change is placed right at the start of a pattern, PCRE
1226 extracts it into the global options (and it will therefore show up in data
1227 extracted by the \fBpcre_fullinfo()\fP function).
1228 .P
1229 An option change within a subpattern (see below for a description of
1230 subpatterns) affects only that part of the subpattern that follows it, so
1231 .sp
1232 (a(?i)b)c
1233 .sp
1234 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1235 By this means, options can be made to have different settings in different
1236 parts of the pattern. Any changes made in one alternative do carry on
1237 into subsequent branches within the same subpattern. For example,
1238 .sp
1239 (a(?i)b|c)
1240 .sp
1241 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1242 branch is abandoned before the option setting. This is because the effects of
1243 option settings happen at compile time. There would be some very weird
1244 behaviour otherwise.
1245 .P
1246 \fBNote:\fP There are other PCRE-specific options that can be set by the
1247 application when the compile or match functions are called. In some cases the
1248 pattern can contain special leading sequences such as (*CRLF) to override what
1249 the application has set or what has been defaulted. Details are given in the
1250 section entitled
1251 .\" HTML <a href="#newlineseq">
1252 .\" </a>
1253 "Newline sequences"
1254 .\"
1255 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1256 to set UTF-8 and Unicode property modes; they are equivalent to setting the
1257 PCRE_UTF8 and the PCRE_UCP options, respectively.
1258 .
1259 .
1260 .\" HTML <a name="subpattern"></a>
1262 .rs
1263 .sp
1264 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1265 Turning part of a pattern into a subpattern does two things:
1266 .sp
1267 1. It localizes a set of alternatives. For example, the pattern
1268 .sp
1269 cat(aract|erpillar|)
1270 .sp
1271 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1272 match "cataract", "erpillar" or an empty string.
1273 .sp
1274 2. It sets up the subpattern as a capturing subpattern. This means that, when
1275 the whole pattern matches, that portion of the subject string that matched the
1276 subpattern is passed back to the caller via the \fIovector\fP argument of
1277 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1278 from 1) to obtain numbers for the capturing subpatterns. For example, if the
1279 string "the red king" is matched against the pattern
1280 .sp
1281 the ((red|white) (king|queen))
1282 .sp
1283 the captured substrings are "red king", "red", and "king", and are numbered 1,
1284 2, and 3, respectively.
1285 .P
1286 The fact that plain parentheses fulfil two functions is not always helpful.
1287 There are often times when a grouping subpattern is required without a
1288 capturing requirement. If an opening parenthesis is followed by a question mark
1289 and a colon, the subpattern does not do any capturing, and is not counted when
1290 computing the number of any subsequent capturing subpatterns. For example, if
1291 the string "the white queen" is matched against the pattern
1292 .sp
1293 the ((?:red|white) (king|queen))
1294 .sp
1295 the captured substrings are "white queen" and "queen", and are numbered 1 and
1296 2. The maximum number of capturing subpatterns is 65535.
1297 .P
1298 As a convenient shorthand, if any option settings are required at the start of
1299 a non-capturing subpattern, the option letters may appear between the "?" and
1300 the ":". Thus the two patterns
1301 .sp
1302 (?i:saturday|sunday)
1303 (?:(?i)saturday|sunday)
1304 .sp
1305 match exactly the same set of strings. Because alternative branches are tried
1306 from left to right, and options are not reset until the end of the subpattern
1307 is reached, an option setting in one branch does affect subsequent branches, so
1308 the above patterns match "SUNDAY" as well as "Saturday".
1309 .
1310 .
1311 .\" HTML <a name="dupsubpatternnumber"></a>
1313 .rs
1314 .sp
1315 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1316 the same numbers for its capturing parentheses. Such a subpattern starts with
1317 (?| and is itself a non-capturing subpattern. For example, consider this
1318 pattern:
1319 .sp
1320 (?|(Sat)ur|(Sun))day
1321 .sp
1322 Because the two alternatives are inside a (?| group, both sets of capturing
1323 parentheses are numbered one. Thus, when the pattern matches, you can look
1324 at captured substring number one, whichever alternative matched. This construct
1325 is useful when you want to capture part, but not all, of one of a number of
1326 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1327 number is reset at the start of each branch. The numbers of any capturing
1328 parentheses that follow the subpattern start after the highest number used in
1329 any branch. The following example is taken from the Perl documentation. The
1330 numbers underneath show in which buffer the captured content will be stored.
1331 .sp
1332 # before ---------------branch-reset----------- after
1333 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1334 # 1 2 2 3 2 3 4
1335 .sp
1336 A back reference to a numbered subpattern uses the most recent value that is
1337 set for that number by any subpattern. The following pattern matches "abcabc"
1338 or "defdef":
1339 .sp
1340 /(?|(abc)|(def))\e1/
1341 .sp
1342 In contrast, a subroutine call to a numbered subpattern always refers to the
1343 first one in the pattern with the given number. The following pattern matches
1344 "abcabc" or "defabc":
1345 .sp
1346 /(?|(abc)|(def))(?1)/
1347 .sp
1348 If a
1349 .\" HTML <a href="#conditions">
1350 .\" </a>
1351 condition test
1352 .\"
1353 for a subpattern's having matched refers to a non-unique number, the test is
1354 true if any of the subpatterns of that number have matched.
1355 .P
1356 An alternative approach to using this "branch reset" feature is to use
1357 duplicate named subpatterns, as described in the next section.
1358 .
1359 .
1361 .rs
1362 .sp
1363 Identifying capturing parentheses by number is simple, but it can be very hard
1364 to keep track of the numbers in complicated regular expressions. Furthermore,
1365 if an expression is modified, the numbers may change. To help with this
1366 difficulty, PCRE supports the naming of subpatterns. This feature was not
1367 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1368 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1369 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1370 have different names, but PCRE does not.
1371 .P
1372 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1373 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1374 parentheses from other parts of the pattern, such as
1375 .\" HTML <a href="#backreferences">
1376 .\" </a>
1377 back references,
1378 .\"
1379 .\" HTML <a href="#recursion">
1380 .\" </a>
1381 recursion,
1382 .\"
1383 and
1384 .\" HTML <a href="#conditions">
1385 .\" </a>
1386 conditions,
1387 .\"
1388 can be made by name as well as by number.
1389 .P
1390 Names consist of up to 32 alphanumeric characters and underscores. Named
1391 capturing parentheses are still allocated numbers as well as names, exactly as
1392 if the names were not present. The PCRE API provides function calls for
1393 extracting the name-to-number translation table from a compiled pattern. There
1394 is also a convenience function for extracting a captured substring by name.
1395 .P
1396 By default, a name must be unique within a pattern, but it is possible to relax
1397 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1398 names are also always permitted for subpatterns with the same number, set up as
1399 described in the previous section.) Duplicate names can be useful for patterns
1400 where only one instance of the named parentheses can match. Suppose you want to
1401 match the name of a weekday, either as a 3-letter abbreviation or as the full
1402 name, and in both cases you want to extract the abbreviation. This pattern
1403 (ignoring the line breaks) does the job:
1404 .sp
1405 (?<DN>Mon|Fri|Sun)(?:day)?|
1406 (?<DN>Tue)(?:sday)?|
1407 (?<DN>Wed)(?:nesday)?|
1408 (?<DN>Thu)(?:rsday)?|
1409 (?<DN>Sat)(?:urday)?
1410 .sp
1411 There are five capturing substrings, but only one is ever set after a match.
1412 (An alternative way of solving this problem is to use a "branch reset"
1413 subpattern, as described in the previous section.)
1414 .P
1415 The convenience function for extracting the data by name returns the substring
1416 for the first (and in this example, the only) subpattern of that name that
1417 matched. This saves searching to find which numbered subpattern it was.
1418 .P
1419 If you make a back reference to a non-unique named subpattern from elsewhere in
1420 the pattern, the one that corresponds to the first occurrence of the name is
1421 used. In the absence of duplicate numbers (see the previous section) this is
1422 the one with the lowest number. If you use a named reference in a condition
1423 test (see the
1424 .\"
1425 .\" HTML <a href="#conditions">
1426 .\" </a>
1427 section about conditions
1428 .\"
1429 below), either to check whether a subpattern has matched, or to check for
1430 recursion, all subpatterns with the same name are tested. If the condition is
1431 true for any one of them, the overall condition is true. This is the same
1432 behaviour as testing by number. For further details of the interfaces for
1433 handling named subpatterns, see the
1434 .\" HREF
1435 \fBpcreapi\fP
1436 .\"
1437 documentation.
1438 .P
1439 \fBWarning:\fP You cannot use different names to distinguish between two
1440 subpatterns with the same number because PCRE uses only the numbers when
1441 matching. For this reason, an error is given at compile time if different names
1442 are given to subpatterns with the same number. However, you can give the same
1443 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1444 .
1445 .
1447 .rs
1448 .sp
1449 Repetition is specified by quantifiers, which can follow any of the following
1450 items:
1451 .sp
1452 a literal data character
1453 the dot metacharacter
1454 the \eC escape sequence
1455 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1456 the \eR escape sequence
1457 an escape such as \ed or \epL that matches a single character
1458 a character class
1459 a back reference (see next section)
1460 a parenthesized subpattern (including assertions)
1461 a subroutine call to a subpattern (recursive or otherwise)
1462 .sp
1463 The general repetition quantifier specifies a minimum and maximum number of
1464 permitted matches, by giving the two numbers in curly brackets (braces),
1465 separated by a comma. The numbers must be less than 65536, and the first must
1466 be less than or equal to the second. For example:
1467 .sp
1468 z{2,4}
1469 .sp
1470 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1471 character. If the second number is omitted, but the comma is present, there is
1472 no upper limit; if the second number and the comma are both omitted, the
1473 quantifier specifies an exact number of required matches. Thus
1474 .sp
1475 [aeiou]{3,}
1476 .sp
1477 matches at least 3 successive vowels, but may match many more, while
1478 .sp
1479 \ed{8}
1480 .sp
1481 matches exactly 8 digits. An opening curly bracket that appears in a position
1482 where a quantifier is not allowed, or one that does not match the syntax of a
1483 quantifier, is taken as a literal character. For example, {,6} is not a
1484 quantifier, but a literal string of four characters.
1485 .P
1486 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1487 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1488 which is represented by a two-byte sequence. Similarly, when Unicode property
1489 support is available, \eX{3} matches three Unicode extended sequences, each of
1490 which may be several bytes long (and they may be of different lengths).
1491 .P
1492 The quantifier {0} is permitted, causing the expression to behave as if the
1493 previous item and the quantifier were not present. This may be useful for
1494 subpatterns that are referenced as
1495 .\" HTML <a href="#subpatternsassubroutines">
1496 .\" </a>
1497 subroutines
1498 .\"
1499 from elsewhere in the pattern (but see also the section entitled
1500 .\" HTML <a href="#subdefine">
1501 .\" </a>
1502 "Defining subpatterns for use by reference only"
1503 .\"
1504 below). Items other than subpatterns that have a {0} quantifier are omitted
1505 from the compiled pattern.
1506 .P
1507 For convenience, the three most common quantifiers have single-character
1508 abbreviations:
1509 .sp
1510 * is equivalent to {0,}
1511 + is equivalent to {1,}
1512 ? is equivalent to {0,1}
1513 .sp
1514 It is possible to construct infinite loops by following a subpattern that can
1515 match no characters with a quantifier that has no upper limit, for example:
1516 .sp
1517 (a?)*
1518 .sp
1519 Earlier versions of Perl and PCRE used to give an error at compile time for
1520 such patterns. However, because there are cases where this can be useful, such
1521 patterns are now accepted, but if any repetition of the subpattern does in fact
1522 match no characters, the loop is forcibly broken.
1523 .P
1524 By default, the quantifiers are "greedy", that is, they match as much as
1525 possible (up to the maximum number of permitted times), without causing the
1526 rest of the pattern to fail. The classic example of where this gives problems
1527 is in trying to match comments in C programs. These appear between /* and */
1528 and within the comment, individual * and / characters may appear. An attempt to
1529 match C comments by applying the pattern
1530 .sp
1531 /\e*.*\e*/
1532 .sp
1533 to the string
1534 .sp
1535 /* first comment */ not comment /* second comment */
1536 .sp
1537 fails, because it matches the entire string owing to the greediness of the .*
1538 item.
1539 .P
1540 However, if a quantifier is followed by a question mark, it ceases to be
1541 greedy, and instead matches the minimum number of times possible, so the
1542 pattern
1543 .sp
1544 /\e*.*?\e*/
1545 .sp
1546 does the right thing with the C comments. The meaning of the various
1547 quantifiers is not otherwise changed, just the preferred number of matches.
1548 Do not confuse this use of question mark with its use as a quantifier in its
1549 own right. Because it has two uses, it can sometimes appear doubled, as in
1550 .sp
1551 \ed??\ed
1552 .sp
1553 which matches one digit by preference, but can match two if that is the only
1554 way the rest of the pattern matches.
1555 .P
1556 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1557 the quantifiers are not greedy by default, but individual ones can be made
1558 greedy by following them with a question mark. In other words, it inverts the
1559 default behaviour.
1560 .P
1561 When a parenthesized subpattern is quantified with a minimum repeat count that
1562 is greater than 1 or with a limited maximum, more memory is required for the
1563 compiled pattern, in proportion to the size of the minimum or maximum.
1564 .P
1565 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1566 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1567 implicitly anchored, because whatever follows will be tried against every
1568 character position in the subject string, so there is no point in retrying the
1569 overall match at any position after the first. PCRE normally treats such a
1570 pattern as though it were preceded by \eA.
1571 .P
1572 In cases where it is known that the subject string contains no newlines, it is
1573 worth setting PCRE_DOTALL in order to obtain this optimization, or
1574 alternatively using ^ to indicate anchoring explicitly.
1575 .P
1576 However, there is one situation where the optimization cannot be used. When .*
1577 is inside capturing parentheses that are the subject of a back reference
1578 elsewhere in the pattern, a match at the start may fail where a later one
1579 succeeds. Consider, for example:
1580 .sp
1581 (.*)abc\e1
1582 .sp
1583 If the subject is "xyz123abc123" the match point is the fourth character. For
1584 this reason, such a pattern is not implicitly anchored.
1585 .P
1586 When a capturing subpattern is repeated, the value captured is the substring
1587 that matched the final iteration. For example, after
1588 .sp
1589 (tweedle[dume]{3}\es*)+
1590 .sp
1591 has matched "tweedledum tweedledee" the value of the captured substring is
1592 "tweedledee". However, if there are nested capturing subpatterns, the
1593 corresponding captured values may have been set in previous iterations. For
1594 example, after
1595 .sp
1596 /(a|(b))+/
1597 .sp
1598 matches "aba" the value of the second captured substring is "b".
1599 .
1600 .
1601 .\" HTML <a name="atomicgroup"></a>
1603 .rs
1604 .sp
1605 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1606 repetition, failure of what follows normally causes the repeated item to be
1607 re-evaluated to see if a different number of repeats allows the rest of the
1608 pattern to match. Sometimes it is useful to prevent this, either to change the
1609 nature of the match, or to cause it fail earlier than it otherwise might, when
1610 the author of the pattern knows there is no point in carrying on.
1611 .P
1612 Consider, for example, the pattern \ed+foo when applied to the subject line
1613 .sp
1614 123456bar
1615 .sp
1616 After matching all 6 digits and then failing to match "foo", the normal
1617 action of the matcher is to try again with only 5 digits matching the \ed+
1618 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1619 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1620 that once a subpattern has matched, it is not to be re-evaluated in this way.
1621 .P
1622 If we use atomic grouping for the previous example, the matcher gives up
1623 immediately on failing to match "foo" the first time. The notation is a kind of
1624 special parenthesis, starting with (?> as in this example:
1625 .sp
1626 (?>\ed+)foo
1627 .sp
1628 This kind of parenthesis "locks up" the part of the pattern it contains once
1629 it has matched, and a failure further into the pattern is prevented from
1630 backtracking into it. Backtracking past it to previous items, however, works as
1631 normal.
1632 .P
1633 An alternative description is that a subpattern of this type matches the string
1634 of characters that an identical standalone pattern would match, if anchored at
1635 the current point in the subject string.
1636 .P
1637 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1638 the above example can be thought of as a maximizing repeat that must swallow
1639 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1640 number of digits they match in order to make the rest of the pattern match,
1641 (?>\ed+) can only match an entire sequence of digits.
1642 .P
1643 Atomic groups in general can of course contain arbitrarily complicated
1644 subpatterns, and can be nested. However, when the subpattern for an atomic
1645 group is just a single repeated item, as in the example above, a simpler
1646 notation, called a "possessive quantifier" can be used. This consists of an
1647 additional + character following a quantifier. Using this notation, the
1648 previous example can be rewritten as
1649 .sp
1650 \ed++foo
1651 .sp
1652 Note that a possessive quantifier can be used with an entire group, for
1653 example:
1654 .sp
1655 (abc|xyz){2,3}+
1656 .sp
1657 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1658 option is ignored. They are a convenient notation for the simpler forms of
1659 atomic group. However, there is no difference in the meaning of a possessive
1660 quantifier and the equivalent atomic group, though there may be a performance
1661 difference; possessive quantifiers should be slightly faster.
1662 .P
1663 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1664 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1665 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1666 package, and PCRE copied it from there. It ultimately found its way into Perl
1667 at release 5.10.
1668 .P
1669 PCRE has an optimization that automatically "possessifies" certain simple
1670 pattern constructs. For example, the sequence A+B is treated as A++B because
1671 there is no point in backtracking into a sequence of A's when B must follow.
1672 .P
1673 When a pattern contains an unlimited repeat inside a subpattern that can itself
1674 be repeated an unlimited number of times, the use of an atomic group is the
1675 only way to avoid some failing matches taking a very long time indeed. The
1676 pattern
1677 .sp
1678 (\eD+|<\ed+>)*[!?]
1679 .sp
1680 matches an unlimited number of substrings that either consist of non-digits, or
1681 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1682 quickly. However, if it is applied to
1683 .sp
1684 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1685 .sp
1686 it takes a long time before reporting failure. This is because the string can
1687 be divided between the internal \eD+ repeat and the external * repeat in a
1688 large number of ways, and all have to be tried. (The example uses [!?] rather
1689 than a single character at the end, because both PCRE and Perl have an
1690 optimization that allows for fast failure when a single character is used. They
1691 remember the last single character that is required for a match, and fail early
1692 if it is not present in the string.) If the pattern is changed so that it uses
1693 an atomic group, like this:
1694 .sp
1695 ((?>\eD+)|<\ed+>)*[!?]
1696 .sp
1697 sequences of non-digits cannot be broken, and failure happens quickly.
1698 .
1699 .
1700 .\" HTML <a name="backreferences"></a>
1702 .rs
1703 .sp
1704 Outside a character class, a backslash followed by a digit greater than 0 (and
1705 possibly further digits) is a back reference to a capturing subpattern earlier
1706 (that is, to its left) in the pattern, provided there have been that many
1707 previous capturing left parentheses.
1708 .P
1709 However, if the decimal number following the backslash is less than 10, it is
1710 always taken as a back reference, and causes an error only if there are not
1711 that many capturing left parentheses in the entire pattern. In other words, the
1712 parentheses that are referenced need not be to the left of the reference for
1713 numbers less than 10. A "forward back reference" of this type can make sense
1714 when a repetition is involved and the subpattern to the right has participated
1715 in an earlier iteration.
1716 .P
1717 It is not possible to have a numerical "forward back reference" to a subpattern
1718 whose number is 10 or more using this syntax because a sequence such as \e50 is
1719 interpreted as a character defined in octal. See the subsection entitled
1720 "Non-printing characters"
1721 .\" HTML <a href="#digitsafterbackslash">
1722 .\" </a>
1723 above
1724 .\"
1725 for further details of the handling of digits following a backslash. There is
1726 no such problem when named parentheses are used. A back reference to any
1727 subpattern is possible using named parentheses (see below).
1728 .P
1729 Another way of avoiding the ambiguity inherent in the use of digits following a
1730 backslash is to use the \eg escape sequence. This escape must be followed by an
1731 unsigned number or a negative number, optionally enclosed in braces. These
1732 examples are all identical:
1733 .sp
1734 (ring), \e1
1735 (ring), \eg1
1736 (ring), \eg{1}
1737 .sp
1738 An unsigned number specifies an absolute reference without the ambiguity that
1739 is present in the older syntax. It is also useful when literal digits follow
1740 the reference. A negative number is a relative reference. Consider this
1741 example:
1742 .sp
1743 (abc(def)ghi)\eg{-1}
1744 .sp
1745 The sequence \eg{-1} is a reference to the most recently started capturing
1746 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1747 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1748 can be helpful in long patterns, and also in patterns that are created by
1749 joining together fragments that contain references within themselves.
1750 .P
1751 A back reference matches whatever actually matched the capturing subpattern in
1752 the current subject string, rather than anything matching the subpattern
1753 itself (see
1754 .\" HTML <a href="#subpatternsassubroutines">
1755 .\" </a>
1756 "Subpatterns as subroutines"
1757 .\"
1758 below for a way of doing that). So the pattern
1759 .sp
1760 (sens|respons)e and \e1ibility
1761 .sp
1762 matches "sense and sensibility" and "response and responsibility", but not
1763 "sense and responsibility". If caseful matching is in force at the time of the
1764 back reference, the case of letters is relevant. For example,
1765 .sp
1766 ((?i)rah)\es+\e1
1767 .sp
1768 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1769 capturing subpattern is matched caselessly.
1770 .P
1771 There are several different ways of writing back references to named
1772 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1773 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1774 back reference syntax, in which \eg can be used for both numeric and named
1775 references, is also supported. We could rewrite the above example in any of
1776 the following ways:
1777 .sp
1778 (?<p1>(?i)rah)\es+\ek<p1>
1779 (?'p1'(?i)rah)\es+\ek{p1}
1780 (?P<p1>(?i)rah)\es+(?P=p1)
1781 (?<p1>(?i)rah)\es+\eg{p1}
1782 .sp
1783 A subpattern that is referenced by name may appear in the pattern before or
1784 after the reference.
1785 .P
1786 There may be more than one back reference to the same subpattern. If a
1787 subpattern has not actually been used in a particular match, any back
1788 references to it always fail by default. For example, the pattern
1789 .sp
1790 (a|(bc))\e2
1791 .sp
1792 always fails if it starts to match "a" rather than "bc". However, if the
1793 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1794 unset value matches an empty string.
1795 .P
1796 Because there may be many capturing parentheses in a pattern, all digits
1797 following a backslash are taken as part of a potential back reference number.
1798 If the pattern continues with a digit character, some delimiter must be used to
1799 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1800 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1801 .\" HTML <a href="#comments">
1802 .\" </a>
1803 "Comments"
1804 .\"
1805 below) can be used.
1806 .
1807 .SS "Recursive back references"
1808 .rs
1809 .sp
1810 A back reference that occurs inside the parentheses to which it refers fails
1811 when the subpattern is first used, so, for example, (a\e1) never matches.
1812 However, such references can be useful inside repeated subpatterns. For
1813 example, the pattern
1814 .sp
1815 (a|b\e1)+
1816 .sp
1817 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1818 the subpattern, the back reference matches the character string corresponding
1819 to the previous iteration. In order for this to work, the pattern must be such
1820 that the first iteration does not need to match the back reference. This can be
1821 done using alternation, as in the example above, or by a quantifier with a
1822 minimum of zero.
1823 .P
1824 Back references of this type cause the group that they reference to be treated
1825 as an
1826 .\" HTML <a href="#atomicgroup">
1827 .\" </a>
1828 atomic group.
1829 .\"
1830 Once the whole group has been matched, a subsequent matching failure cannot
1831 cause backtracking into the middle of the group.
1832 .
1833 .
1834 .\" HTML <a name="bigassertions"></a>
1836 .rs
1837 .sp
1838 An assertion is a test on the characters following or preceding the current
1839 matching point that does not actually consume any characters. The simple
1840 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1841 .\" HTML <a href="#smallassertions">
1842 .\" </a>
1843 above.
1844 .\"
1845 .P
1846 More complicated assertions are coded as subpatterns. There are two kinds:
1847 those that look ahead of the current position in the subject string, and those
1848 that look behind it. An assertion subpattern is matched in the normal way,
1849 except that it does not cause the current matching position to be changed.
1850 .P
1851 Assertion subpatterns are not capturing subpatterns. If such an assertion
1852 contains capturing subpatterns within it, these are counted for the purposes of
1853 numbering the capturing subpatterns in the whole pattern. However, substring
1854 capturing is carried out only for positive assertions, because it does not make
1855 sense for negative assertions.
1856 .P
1857 For compatibility with Perl, assertion subpatterns may be repeated; though
1858 it makes no sense to assert the same thing several times, the side effect of
1859 capturing parentheses may occasionally be useful. In practice, there only three
1860 cases:
1861 .sp
1862 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1863 However, it may contain internal capturing parenthesized groups that are called
1864 from elsewhere via the
1865 .\" HTML <a href="#subpatternsassubroutines">
1866 .\" </a>
1867 subroutine mechanism.
1868 .\"
1869 .sp
1870 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1871 were {0,1}. At run time, the rest of the pattern match is tried with and
1872 without the assertion, the order depending on the greediness of the quantifier.
1873 .sp
1874 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1875 The assertion is obeyed just once when encountered during matching.
1876 .
1877 .
1878 .SS "Lookahead assertions"
1879 .rs
1880 .sp
1881 Lookahead assertions start with (?= for positive assertions and (?! for
1882 negative assertions. For example,
1883 .sp
1884 \ew+(?=;)
1885 .sp
1886 matches a word followed by a semicolon, but does not include the semicolon in
1887 the match, and
1888 .sp
1889 foo(?!bar)
1890 .sp
1891 matches any occurrence of "foo" that is not followed by "bar". Note that the
1892 apparently similar pattern
1893 .sp
1894 (?!foo)bar
1895 .sp
1896 does not find an occurrence of "bar" that is preceded by something other than
1897 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1898 (?!foo) is always true when the next three characters are "bar". A
1899 lookbehind assertion is needed to achieve the other effect.
1900 .P
1901 If you want to force a matching failure at some point in a pattern, the most
1902 convenient way to do it is with (?!) because an empty string always matches, so
1903 an assertion that requires there not to be an empty string must always fail.
1904 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1905 .
1906 .
1907 .\" HTML <a name="lookbehind"></a>
1908 .SS "Lookbehind assertions"
1909 .rs
1910 .sp
1911 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1912 negative assertions. For example,
1913 .sp
1914 (?<!foo)bar
1915 .sp
1916 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1917 a lookbehind assertion are restricted such that all the strings it matches must
1918 have a fixed length. However, if there are several top-level alternatives, they
1919 do not all have to have the same fixed length. Thus
1920 .sp
1921 (?<=bullock|donkey)
1922 .sp
1923 is permitted, but
1924 .sp
1925 (?<!dogs?|cats?)
1926 .sp
1927 causes an error at compile time. Branches that match different length strings
1928 are permitted only at the top level of a lookbehind assertion. This is an
1929 extension compared with Perl, which requires all branches to match the same
1930 length of string. An assertion such as
1931 .sp
1932 (?<=ab(c|de))
1933 .sp
1934 is not permitted, because its single top-level branch can match two different
1935 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1936 branches:
1937 .sp
1938 (?<=abc|abde)
1939 .sp
1940 In some cases, the escape sequence \eK
1941 .\" HTML <a href="#resetmatchstart">
1942 .\" </a>
1943 (see above)
1944 .\"
1945 can be used instead of a lookbehind assertion to get round the fixed-length
1946 restriction.
1947 .P
1948 The implementation of lookbehind assertions is, for each alternative, to
1949 temporarily move the current position back by the fixed length and then try to
1950 match. If there are insufficient characters before the current position, the
1951 assertion fails.
1952 .P
1953 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1954 to appear in lookbehind assertions, because it makes it impossible to calculate
1955 the length of the lookbehind. The \eX and \eR escapes, which can match
1956 different numbers of bytes, are also not permitted.
1957 .P
1958 .\" HTML <a href="#subpatternsassubroutines">
1959 .\" </a>
1960 "Subroutine"
1961 .\"
1962 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1963 as the subpattern matches a fixed-length string.
1964 .\" HTML <a href="#recursion">
1965 .\" </a>
1966 Recursion,
1967 .\"
1968 however, is not supported.
1969 .P
1970 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1971 specify efficient matching of fixed-length strings at the end of subject
1972 strings. Consider a simple pattern such as
1973 .sp
1974 abcd$
1975 .sp
1976 when applied to a long string that does not match. Because matching proceeds
1977 from left to right, PCRE will look for each "a" in the subject and then see if
1978 what follows matches the rest of the pattern. If the pattern is specified as
1979 .sp
1980 ^.*abcd$
1981 .sp
1982 the initial .* matches the entire string at first, but when this fails (because
1983 there is no following "a"), it backtracks to match all but the last character,
1984 then all but the last two characters, and so on. Once again the search for "a"
1985 covers the entire string, from right to left, so we are no better off. However,
1986 if the pattern is written as
1987 .sp
1988 ^.*+(?<=abcd)
1989 .sp
1990 there can be no backtracking for the .*+ item; it can match only the entire
1991 string. The subsequent lookbehind assertion does a single test on the last four
1992 characters. If it fails, the match fails immediately. For long strings, this
1993 approach makes a significant difference to the processing time.
1994 .
1995 .
1996 .SS "Using multiple assertions"
1997 .rs
1998 .sp
1999 Several assertions (of any sort) may occur in succession. For example,
2000 .sp
2001 (?<=\ed{3})(?<!999)foo
2002 .sp
2003 matches "foo" preceded by three digits that are not "999". Notice that each of
2004 the assertions is applied independently at the same point in the subject
2005 string. First there is a check that the previous three characters are all
2006 digits, and then there is a check that the same three characters are not "999".
2007 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2008 of which are digits and the last three of which are not "999". For example, it
2009 doesn't match "123abcfoo". A pattern to do that is
2010 .sp
2011 (?<=\ed{3}...)(?<!999)foo
2012 .sp
2013 This time the first assertion looks at the preceding six characters, checking
2014 that the first three are digits, and then the second assertion checks that the
2015 preceding three characters are not "999".
2016 .P
2017 Assertions can be nested in any combination. For example,
2018 .sp
2019 (?<=(?<!foo)bar)baz
2020 .sp
2021 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2022 preceded by "foo", while
2023 .sp
2024 (?<=\ed{3}(?!999)...)foo
2025 .sp
2026 is another pattern that matches "foo" preceded by three digits and any three
2027 characters that are not "999".
2028 .
2029 .
2030 .\" HTML <a name="conditions"></a>
2032 .rs
2033 .sp
2034 It is possible to cause the matching process to obey a subpattern
2035 conditionally or to choose between two alternative subpatterns, depending on
2036 the result of an assertion, or whether a specific capturing subpattern has
2037 already been matched. The two possible forms of conditional subpattern are:
2038 .sp
2039 (?(condition)yes-pattern)
2040 (?(condition)yes-pattern|no-pattern)
2041 .sp
2042 If the condition is satisfied, the yes-pattern is used; otherwise the
2043 no-pattern (if present) is used. If there are more than two alternatives in the
2044 subpattern, a compile-time error occurs. Each of the two alternatives may
2045 itself contain nested subpatterns of any form, including conditional
2046 subpatterns; the restriction to two alternatives applies only at the level of
2047 the condition. This pattern fragment is an example where the alternatives are
2048 complex:
2049 .sp
2050 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2051 .sp
2052 .P
2053 There are four kinds of condition: references to subpatterns, references to
2054 recursion, a pseudo-condition called DEFINE, and assertions.
2055 .
2056 .SS "Checking for a used subpattern by number"
2057 .rs
2058 .sp
2059 If the text between the parentheses consists of a sequence of digits, the
2060 condition is true if a capturing subpattern of that number has previously
2061 matched. If there is more than one capturing subpattern with the same number
2062 (see the earlier
2063 .\"
2064 .\" HTML <a href="#recursion">
2065 .\" </a>
2066 section about duplicate subpattern numbers),
2067 .\"
2068 the condition is true if any of them have matched. An alternative notation is
2069 to precede the digits with a plus or minus sign. In this case, the subpattern
2070 number is relative rather than absolute. The most recently opened parentheses
2071 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2072 loops it can also make sense to refer to subsequent groups. The next
2073 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2074 zero in any of these forms is not used; it provokes a compile-time error.)
2075 .P
2076 Consider the following pattern, which contains non-significant white space to
2077 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2078 three parts for ease of discussion:
2079 .sp
2080 ( \e( )? [^()]+ (?(1) \e) )
2081 .sp
2082 The first part matches an optional opening parenthesis, and if that
2083 character is present, sets it as the first captured substring. The second part
2084 matches one or more characters that are not parentheses. The third part is a
2085 conditional subpattern that tests whether or not the first set of parentheses
2086 matched. If they did, that is, if subject started with an opening parenthesis,
2087 the condition is true, and so the yes-pattern is executed and a closing
2088 parenthesis is required. Otherwise, since no-pattern is not present, the
2089 subpattern matches nothing. In other words, this pattern matches a sequence of
2090 non-parentheses, optionally enclosed in parentheses.
2091 .P
2092 If you were embedding this pattern in a larger one, you could use a relative
2093 reference:
2094 .sp
2095 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2096 .sp
2097 This makes the fragment independent of the parentheses in the larger pattern.
2098 .
2099 .SS "Checking for a used subpattern by name"
2100 .rs
2101 .sp
2102 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2103 subpattern by name. For compatibility with earlier versions of PCRE, which had
2104 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2105 there is a possible ambiguity with this syntax, because subpattern names may
2106 consist entirely of digits. PCRE looks first for a named subpattern; if it
2107 cannot find one and the name consists entirely of digits, PCRE looks for a
2108 subpattern of that number, which must be greater than zero. Using subpattern
2109 names that consist entirely of digits is not recommended.
2110 .P
2111 Rewriting the above example to use a named subpattern gives this:
2112 .sp
2113 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2114 .sp
2115 If the name used in a condition of this kind is a duplicate, the test is
2116 applied to all subpatterns of the same name, and is true if any one of them has
2117 matched.
2118 .
2119 .SS "Checking for pattern recursion"
2120 .rs
2121 .sp
2122 If the condition is the string (R), and there is no subpattern with the name R,
2123 the condition is true if a recursive call to the whole pattern or any
2124 subpattern has been made. If digits or a name preceded by ampersand follow the
2125 letter R, for example:
2126 .sp
2127 (?(R3)...) or (?(R&name)...)
2128 .sp
2129 the condition is true if the most recent recursion is into a subpattern whose
2130 number or name is given. This condition does not check the entire recursion
2131 stack. If the name used in a condition of this kind is a duplicate, the test is
2132 applied to all subpatterns of the same name, and is true if any one of them is
2133 the most recent recursion.
2134 .P
2135 At "top level", all these recursion test conditions are false.
2136 .\" HTML <a href="#recursion">
2137 .\" </a>
2138 The syntax for recursive patterns
2139 .\"
2140 is described below.
2141 .
2142 .\" HTML <a name="subdefine"></a>
2143 .SS "Defining subpatterns for use by reference only"
2144 .rs
2145 .sp
2146 If the condition is the string (DEFINE), and there is no subpattern with the
2147 name DEFINE, the condition is always false. In this case, there may be only one
2148 alternative in the subpattern. It is always skipped if control reaches this
2149 point in the pattern; the idea of DEFINE is that it can be used to define
2150 subroutines that can be referenced from elsewhere. (The use of
2151 .\" HTML <a href="#subpatternsassubroutines">
2152 .\" </a>
2153 subroutines
2154 .\"
2155 is described below.) For example, a pattern to match an IPv4 address such as
2156 "" could be written like this (ignore whitespace and line
2157 breaks):
2158 .sp
2159 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2160 \eb (?&byte) (\e.(?&byte)){3} \eb
2161 .sp
2162 The first part of the pattern is a DEFINE group inside which a another group
2163 named "byte" is defined. This matches an individual component of an IPv4
2164 address (a number less than 256). When matching takes place, this part of the
2165 pattern is skipped because DEFINE acts like a false condition. The rest of the
2166 pattern uses references to the named group to match the four dot-separated
2167 components of an IPv4 address, insisting on a word boundary at each end.
2168 .
2169 .SS "Assertion conditions"
2170 .rs
2171 .sp
2172 If the condition is not in any of the above formats, it must be an assertion.
2173 This may be a positive or negative lookahead or lookbehind assertion. Consider
2174 this pattern, again containing non-significant white space, and with the two
2175 alternatives on the second line:
2176 .sp
2177 (?(?=[^a-z]*[a-z])
2178 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2179 .sp
2180 The condition is a positive lookahead assertion that matches an optional
2181 sequence of non-letters followed by a letter. In other words, it tests for the
2182 presence of at least one letter in the subject. If a letter is found, the
2183 subject is matched against the first alternative; otherwise it is matched
2184 against the second. This pattern matches strings in one of the two forms
2185 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2186 .
2187 .
2188 .\" HTML <a name="comments"></a>
2190 .rs
2191 .sp
2192 There are two ways of including comments in patterns that are processed by
2193 PCRE. In both cases, the start of the comment must not be in a character class,
2194 nor in the middle of any other sequence of related characters such as (?: or a
2195 subpattern name or number. The characters that make up a comment play no part
2196 in the pattern matching.
2197 .P
2198 The sequence (?# marks the start of a comment that continues up to the next
2199 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2200 option is set, an unescaped # character also introduces a comment, which in
2201 this case continues to immediately after the next newline character or
2202 character sequence in the pattern. Which characters are interpreted as newlines
2203 is controlled by the options passed to \fBpcre_compile()\fP or by a special
2204 sequence at the start of the pattern, as described in the section entitled
2205 .\" HTML <a href="#newlines">
2206 .\" </a>
2207 "Newline conventions"
2208 .\"
2209 above. Note that the end of this type of comment is a literal newline sequence
2210 in the pattern; escape sequences that happen to represent a newline do not
2211 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2212 default newline convention is in force:
2213 .sp
2214 abc #comment \en still comment
2215 .sp
2216 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2217 a newline in the pattern. The sequence \en is still literal at this stage, so
2218 it does not terminate the comment. Only an actual character with the code value
2219 0x0a (the default newline) does so.
2220 .
2221 .
2222 .\" HTML <a name="recursion"></a>
2224 .rs
2225 .sp
2226 Consider the problem of matching a string in parentheses, allowing for
2227 unlimited nested parentheses. Without the use of recursion, the best that can
2228 be done is to use a pattern that matches up to some fixed depth of nesting. It
2229 is not possible to handle an arbitrary nesting depth.
2230 .P
2231 For some time, Perl has provided a facility that allows regular expressions to
2232 recurse (amongst other things). It does this by interpolating Perl code in the
2233 expression at run time, and the code can refer to the expression itself. A Perl
2234 pattern using code interpolation to solve the parentheses problem can be
2235 created like this:
2236 .sp
2237 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2238 .sp
2239 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2240 recursively to the pattern in which it appears.
2241 .P
2242 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2243 supports special syntax for recursion of the entire pattern, and also for
2244 individual subpattern recursion. After its introduction in PCRE and Python,
2245 this kind of recursion was subsequently introduced into Perl at release 5.10.
2246 .P
2247 A special item that consists of (? followed by a number greater than zero and a
2248 closing parenthesis is a recursive subroutine call of the subpattern of the
2249 given number, provided that it occurs inside that subpattern. (If not, it is a
2250 .\" HTML <a href="#subpatternsassubroutines">
2251 .\" </a>
2252 non-recursive subroutine
2253 .\"
2254 call, which is described in the next section.) The special item (?R) or (?0) is
2255 a recursive call of the entire regular expression.
2256 .P
2257 This PCRE pattern solves the nested parentheses problem (assume the
2258 PCRE_EXTENDED option is set so that white space is ignored):
2259 .sp
2260 \e( ( [^()]++ | (?R) )* \e)
2261 .sp
2262 First it matches an opening parenthesis. Then it matches any number of
2263 substrings which can either be a sequence of non-parentheses, or a recursive
2264 match of the pattern itself (that is, a correctly parenthesized substring).
2265 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2266 to avoid backtracking into sequences of non-parentheses.
2267 .P
2268 If this were part of a larger pattern, you would not want to recurse the entire
2269 pattern, so instead you could use this:
2270 .sp
2271 ( \e( ( [^()]++ | (?1) )* \e) )
2272 .sp
2273 We have put the pattern into parentheses, and caused the recursion to refer to
2274 them instead of the whole pattern.
2275 .P
2276 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2277 is made easier by the use of relative references. Instead of (?1) in the
2278 pattern above you can write (?-2) to refer to the second most recently opened
2279 parentheses preceding the recursion. In other words, a negative number counts
2280 capturing parentheses leftwards from the point at which it is encountered.
2281 .P
2282 It is also possible to refer to subsequently opened parentheses, by writing
2283 references such as (?+2). However, these cannot be recursive because the
2284 reference is not inside the parentheses that are referenced. They are always
2285 .\" HTML <a href="#subpatternsassubroutines">
2286 .\" </a>
2287 non-recursive subroutine
2288 .\"
2289 calls, as described in the next section.
2290 .P
2291 An alternative approach is to use named parentheses instead. The Perl syntax
2292 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2293 could rewrite the above example as follows:
2294 .sp
2295 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2296 .sp
2297 If there is more than one subpattern with the same name, the earliest one is
2298 used.
2299 .P
2300 This particular example pattern that we have been looking at contains nested
2301 unlimited repeats, and so the use of a possessive quantifier for matching
2302 strings of non-parentheses is important when applying the pattern to strings
2303 that do not match. For example, when this pattern is applied to
2304 .sp
2305 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2306 .sp
2307 it yields "no match" quickly. However, if a possessive quantifier is not used,
2308 the match runs for a very long time indeed because there are so many different
2309 ways the + and * repeats can carve up the subject, and all have to be tested
2310 before failure can be reported.
2311 .P
2312 At the end of a match, the values of capturing parentheses are those from
2313 the outermost level. If you want to obtain intermediate values, a callout
2314 function can be used (see below and the
2315 .\" HREF
2316 \fBpcrecallout\fP
2317 .\"
2318 documentation). If the pattern above is matched against
2319 .sp
2320 (ab(cd)ef)
2321 .sp
2322 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2323 the last value taken on at the top level. If a capturing subpattern is not
2324 matched at the top level, its final captured value is unset, even if it was
2325 (temporarily) set at a deeper level during the matching process.
2326 .P
2327 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2328 obtain extra memory to store data during a recursion, which it does by using
2329 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2330 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2331 .P
2332 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2333 Consider this pattern, which matches text in angle brackets, allowing for
2334 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2335 recursing), whereas any characters are permitted at the outer level.
2336 .sp
2337 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2338 .sp
2339 In this pattern, (?(R) is the start of a conditional subpattern, with two
2340 different alternatives for the recursive and non-recursive cases. The (?R) item
2341 is the actual recursive call.
2342 .
2343 .
2344 .\" HTML <a name="recursiondifference"></a>
2345 .SS "Differences in recursion processing between PCRE and Perl"
2346 .rs
2347 .sp
2348 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2349 (like Python, but unlike Perl), a recursive subpattern call is always treated
2350 as an atomic group. That is, once it has matched some of the subject string, it
2351 is never re-entered, even if it contains untried alternatives and there is a
2352 subsequent matching failure. This can be illustrated by the following pattern,
2353 which purports to match a palindromic string that contains an odd number of
2354 characters (for example, "a", "aba", "abcba", "abcdcba"):
2355 .sp
2356 ^(.|(.)(?1)\e2)$
2357 .sp
2358 The idea is that it either matches a single character, or two identical
2359 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2360 it does not if the pattern is longer than three characters. Consider the
2361 subject string "abcba":
2362 .P
2363 At the top level, the first character is matched, but as it is not at the end
2364 of the string, the first alternative fails; the second alternative is taken
2365 and the recursion kicks in. The recursive call to subpattern 1 successfully
2366 matches the next character ("b"). (Note that the beginning and end of line
2367 tests are not part of the recursion).
2368 .P
2369 Back at the top level, the next character ("c") is compared with what
2370 subpattern 2 matched, which was "a". This fails. Because the recursion is
2371 treated as an atomic group, there are now no backtracking points, and so the
2372 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2373 try the second alternative.) However, if the pattern is written with the
2374 alternatives in the other order, things are different:
2375 .sp
2376 ^((.)(?1)\e2|.)$
2377 .sp
2378 This time, the recursing alternative is tried first, and continues to recurse
2379 until it runs out of characters, at which point the recursion fails. But this
2380 time we do have another alternative to try at the higher level. That is the big
2381 difference: in the previous case the remaining alternative is at a deeper
2382 recursion level, which PCRE cannot use.
2383 .P
2384 To change the pattern so that it matches all palindromic strings, not just
2385 those with an odd number of characters, it is tempting to change the pattern to
2386 this:
2387 .sp
2388 ^((.)(?1)\e2|.?)$
2389 .sp
2390 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2391 deeper recursion has matched a single character, it cannot be entered again in
2392 order to match an empty string. The solution is to separate the two cases, and
2393 write out the odd and even cases as alternatives at the higher level:
2394 .sp
2395 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2396 .sp
2397 If you want to match typical palindromic phrases, the pattern has to ignore all
2398 non-word characters, which can be done like this:
2399 .sp
2400 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2401 .sp
2402 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2403 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2404 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2405 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2406 more) to match typical phrases, and Perl takes so long that you think it has
2407 gone into a loop.
2408 .P
2409 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2410 string does not start with a palindrome that is shorter than the entire string.
2411 For example, although "abcba" is correctly matched, if the subject is "ababa",
2412 PCRE finds the palindrome "aba" at the start, then fails at top level because
2413 the end of the string does not follow. Once again, it cannot jump back into the
2414 recursion to try other alternatives, so the entire match fails.
2415 .P
2416 The second way in which PCRE and Perl differ in their recursion processing is
2417 in the handling of captured values. In Perl, when a subpattern is called
2418 recursively or as a subpattern (see the next section), it has no access to any
2419 values that were captured outside the recursion, whereas in PCRE these values
2420 can be referenced. Consider this pattern:
2421 .sp
2422 ^(.)(\e1|a(?2))
2423 .sp
2424 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2425 then in the second group, when the back reference \e1 fails to match "b", the
2426 second alternative matches "a" and then recurses. In the recursion, \e1 does
2427 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2428 match because inside the recursive call \e1 cannot access the externally set
2429 value.
2430 .
2431 .
2432 .\" HTML <a name="subpatternsassubroutines"></a>
2434 .rs
2435 .sp
2436 If the syntax for a recursive subpattern call (either by number or by
2437 name) is used outside the parentheses to which it refers, it operates like a
2438 subroutine in a programming language. The called subpattern may be defined
2439 before or after the reference. A numbered reference can be absolute or
2440 relative, as in these examples:
2441 .sp
2442 (...(absolute)...)...(?2)...
2443 (...(relative)...)...(?-1)...
2444 (...(?+1)...(relative)...
2445 .sp
2446 An earlier example pointed out that the pattern
2447 .sp
2448 (sens|respons)e and \e1ibility
2449 .sp
2450 matches "sense and sensibility" and "response and responsibility", but not
2451 "sense and responsibility". If instead the pattern
2452 .sp
2453 (sens|respons)e and (?1)ibility
2454 .sp
2455 is used, it does match "sense and responsibility" as well as the other two
2456 strings. Another example is given in the discussion of DEFINE above.
2457 .P
2458 All subroutine calls, whether recursive or not, are always treated as atomic
2459 groups. That is, once a subroutine has matched some of the subject string, it
2460 is never re-entered, even if it contains untried alternatives and there is a
2461 subsequent matching failure. Any capturing parentheses that are set during the
2462 subroutine call revert to their previous values afterwards.
2463 .P
2464 Processing options such as case-independence are fixed when a subpattern is
2465 defined, so if it is used as a subroutine, such options cannot be changed for
2466 different calls. For example, consider this pattern:
2467 .sp
2468 (abc)(?i:(?-1))
2469 .sp
2470 It matches "abcabc". It does not match "abcABC" because the change of
2471 processing option does not affect the called subpattern.
2472 .
2473 .
2474 .\" HTML <a name="onigurumasubroutines"></a>
2476 .rs
2477 .sp
2478 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2479 a number enclosed either in angle brackets or single quotes, is an alternative
2480 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2481 are two of the examples used above, rewritten using this syntax:
2482 .sp
2483 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2484 (sens|respons)e and \eg'1'ibility
2485 .sp
2486 PCRE supports an extension to Oniguruma: if a number is preceded by a
2487 plus or a minus sign it is taken as a relative reference. For example:
2488 .sp
2489 (abc)(?i:\eg<-1>)
2490 .sp
2491 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2492 synonymous. The former is a back reference; the latter is a subroutine call.
2493 .
2494 .
2496 .rs
2497 .sp
2498 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2499 code to be obeyed in the middle of matching a regular expression. This makes it
2500 possible, amongst other things, to extract different substrings that match the
2501 same pair of parentheses when there is a repetition.
2502 .P
2503 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2504 code. The feature is called "callout". The caller of PCRE provides an external
2505 function by putting its entry point in the global variable \fIpcre_callout\fP.
2506 By default, this variable contains NULL, which disables all calling out.
2507 .P
2508 Within a regular expression, (?C) indicates the points at which the external
2509 function is to be called. If you want to identify different callout points, you
2510 can put a number less than 256 after the letter C. The default value is zero.
2511 For example, this pattern has two callout points:
2512 .sp
2513 (?C1)abc(?C2)def
2514 .sp
2515 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2516 automatically installed before each item in the pattern. They are all numbered
2517 255.
2518 .P
2519 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2520 set), the external function is called. It is provided with the number of the
2521 callout, the position in the pattern, and, optionally, one item of data
2522 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2523 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2524 description of the interface to the callout function is given in the
2525 .\" HREF
2526 \fBpcrecallout\fP
2527 .\"
2528 documentation.
2529 .
2530 .
2531 .\" HTML <a name="backtrackcontrol"></a>
2533 .rs
2534 .sp
2535 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2536 are described in the Perl documentation as "experimental and subject to change
2537 or removal in a future version of Perl". It goes on to say: "Their usage in
2538 production code should be noted to avoid problems during upgrades." The same
2539 remarks apply to the PCRE features described in this section.
2540 .P
2541 Since these verbs are specifically related to backtracking, most of them can be
2542 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2543 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2544 failing negative assertion, they cause an error if encountered by
2545 \fBpcre_dfa_exec()\fP.
2546 .P
2547 If any of these verbs are used in an assertion or in a subpattern that is
2548 called as a subroutine (whether or not recursively), their effect is confined
2549 to that subpattern; it does not extend to the surrounding pattern, with one
2550 exception: a *MARK that is encountered in a positive assertion \fIis\fP passed
2551 back (compare capturing parentheses in assertions). Note that such subpatterns
2552 are processed as anchored at the point where they are tested. Note also that
2553 Perl's treatment of subroutines is different in some cases.
2554 .P
2555 The new verbs make use of what was previously invalid syntax: an opening
2556 parenthesis followed by an asterisk. They are generally of the form
2557 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2558 depending on whether or not an argument is present. A name is any sequence of
2559 characters that does not include a closing parenthesis. If the name is empty,
2560 that is, if the closing parenthesis immediately follows the colon, the effect
2561 is as if the colon were not there. Any number of these verbs may occur in a
2562 pattern.
2563 .P
2564 PCRE contains some optimizations that are used to speed up matching by running
2565 some checks at the start of each match attempt. For example, it may know the
2566 minimum length of matching subject, or that a particular character must be
2567 present. When one of these optimizations suppresses the running of a match, any
2568 included backtracking verbs will not, of course, be processed. You can suppress
2569 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2570 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2571 pattern with (*NO_START_OPT).
2572 .
2573 .
2574 .SS "Verbs that act immediately"
2575 .rs
2576 .sp
2577 The following verbs act as soon as they are encountered. They may not be
2578 followed by a name.
2579 .sp
2580 (*ACCEPT)
2581 .sp
2582 This verb causes the match to end successfully, skipping the remainder of the
2583 pattern. However, when it is inside a subpattern that is called as a
2584 subroutine, only that subpattern is ended successfully. Matching then continues
2585 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2586 far is captured. For example:
2587 .sp
2588 A((?:A|B(*ACCEPT)|C)D)
2589 .sp
2590 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2591 the outer parentheses.
2592 .sp
2593 (*FAIL) or (*F)
2594 .sp
2595 This verb causes a matching failure, forcing backtracking to occur. It is
2596 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2597 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2598 Perl features that are not present in PCRE. The nearest equivalent is the
2599 callout feature, as for example in this pattern:
2600 .sp
2601 a+(?C)(*FAIL)
2602 .sp
2603 A match with the string "aaaa" always fails, but the callout is taken before
2604 each backtrack happens (in this example, 10 times).
2605 .
2606 .
2607 .SS "Recording which path was taken"
2608 .rs
2609 .sp
2610 There is one verb whose main purpose is to track how a match was arrived at,
2611 though it also has a secondary use in conjunction with advancing the match
2612 starting point (see (*SKIP) below).
2613 .sp
2614 (*MARK:NAME) or (*:NAME)
2615 .sp
2616 A name is always required with this verb. There may be as many instances of
2617 (*MARK) as you like in a pattern, and their names do not have to be unique.
2618 .P
2619 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2620 to the caller via the \fIpcre_extra\fP data structure, as described in the
2621 .\" HTML <a href="pcreapi.html#extradata">
2622 .\" </a>
2623 section on \fIpcre_extra\fP
2624 .\"
2625 in the
2626 .\" HREF
2627 \fBpcreapi\fP
2628 .\"
2629 documentation. No data is returned for a partial match. Here is an example of
2630 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2631 outputting of (*MARK) data:
2632 .sp
2633 /X(*MARK:A)Y|X(*MARK:B)Z/K
2634 XY
2635 0: XY
2636 MK: A
2637 XZ
2638 0: XZ
2639 MK: B
2640 .sp
2641 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2642 indicates which of the two alternatives matched. This is a more efficient way
2643 of obtaining this information than putting each alternative in its own
2644 capturing parentheses.
2645 .P
2646 If (*MARK) is encountered in a positive assertion, its name is recorded and
2647 passed back if it is the last-encountered. This does not happen for negative
2648 assertions.
2649 .P
2650 A name may also be returned after a failed match if the final path through the
2651 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2652 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2653 starting point for matching is advanced, the final check is often with an empty
2654 string, causing a failure before (*MARK) is reached. For example:
2655 .sp
2656 /X(*MARK:A)Y|X(*MARK:B)Z/K
2657 XP
2658 No match
2659 .sp
2660 There are three potential starting points for this match (starting with X,
2661 starting with P, and with an empty string). If the pattern is anchored, the
2662 result is different:
2663 .sp
2664 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2665 XP
2666 No match, mark = B
2667 .sp
2668 PCRE's start-of-match optimizations can also interfere with this. For example,
2669 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2670 subject length for a match, a shorter subject will not be scanned at all.
2671 .P
2672 Note that similar anomalies (though different in detail) exist in Perl, no
2673 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2674 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2675 .
2676 .
2677 .SS "Verbs that act after backtracking"
2678 .rs
2679 .sp
2680 The following verbs do nothing when they are encountered. Matching continues
2681 with what follows, but if there is no subsequent match, causing a backtrack to
2682 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2683 the verb. However, when one of these verbs appears inside an atomic group, its
2684 effect is confined to that group, because once the group has been matched,
2685 there is never any backtracking into it. In this situation, backtracking can
2686 "jump back" to the left of the entire atomic group. (Remember also, as stated
2687 above, that this localization also applies in subroutine calls and assertions.)
2688 .P
2689 These verbs differ in exactly what kind of failure occurs when backtracking
2690 reaches them.
2691 .sp
2692 (*COMMIT)
2693 .sp
2694 This verb, which may not be followed by a name, causes the whole match to fail
2695 outright if the rest of the pattern does not match. Even if the pattern is
2696 unanchored, no further attempts to find a match by advancing the starting point
2697 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2698 finding a match at the current starting point, or not at all. For example:
2699 .sp
2700 a+(*COMMIT)b
2701 .sp
2702 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2703 dynamic anchor, or "I've started, so I must finish." The name of the most
2704 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2705 match failure.
2706 .P
2707 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2708 unless PCRE's start-of-match optimizations are turned off, as shown in this
2709 \fBpcretest\fP example:
2710 .sp
2711 /(*COMMIT)abc/
2712 xyzabc
2713 0: abc
2714 xyzabc\eY
2715 No match
2716 .sp
2717 PCRE knows that any match must start with "a", so the optimization skips along
2718 the subject to "a" before running the first match attempt, which succeeds. When
2719 the optimization is disabled by the \eY escape in the second subject, the match
2720 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2721 starting points.
2722 .sp
2723 (*PRUNE) or (*PRUNE:NAME)
2724 .sp
2725 This verb causes the match to fail at the current starting position in the
2726 subject if the rest of the pattern does not match. If the pattern is
2727 unanchored, the normal "bumpalong" advance to the next starting character then
2728 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2729 reached, or when matching to the right of (*PRUNE), but if there is no match to
2730 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2731 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2732 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2733 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2734 match fails completely; the name is passed back if this is the final attempt.
2735 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2736 pattern (*PRUNE) has the same effect as (*COMMIT).
2737 .sp
2738 (*SKIP)
2739 .sp
2740 This verb, when given without a name, is like (*PRUNE), except that if the
2741 pattern is unanchored, the "bumpalong" advance is not to the next character,
2742 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2743 signifies that whatever text was matched leading up to it cannot be part of a
2744 successful match. Consider:
2745 .sp
2746 a+(*SKIP)b
2747 .sp
2748 If the subject is "aaaac...", after the first match attempt fails (starting at
2749 the first character in the string), the starting point skips on to start the
2750 next attempt at "c". Note that a possessive quantifer does not have the same
2751 effect as this example; although it would suppress backtracking during the
2752 first match attempt, the second attempt would start at the second character
2753 instead of skipping on to "c".
2754 .sp
2755 (*SKIP:NAME)
2756 .sp
2757 When (*SKIP) has an associated name, its behaviour is modified. If the
2758 following pattern fails to match, the previous path through the pattern is
2759 searched for the most recent (*MARK) that has the same name. If one is found,
2760 the "bumpalong" advance is to the subject position that corresponds to that
2761 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2762 matching name is found, normal "bumpalong" of one character happens (that is,
2763 the (*SKIP) is ignored).
2764 .sp
2765 (*THEN) or (*THEN:NAME)
2766 .sp
2767 This verb causes a skip to the next innermost alternative if the rest of the
2768 pattern does not match. That is, it cancels pending backtracking, but only
2769 within the current alternative. Its name comes from the observation that it can
2770 be used for a pattern-based if-then-else block:
2771 .sp
2772 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2773 .sp
2774 If the COND1 pattern matches, FOO is tried (and possibly further items after
2775 the end of the group if FOO succeeds); on failure, the matcher skips to the
2776 second alternative and tries COND2, without backtracking into COND1. The
2777 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2778 overall match fails. If (*THEN) is not inside an alternation, it acts like
2779 (*PRUNE).
2780 .P
2781 Note that a subpattern that does not contain a | character is just a part of
2782 the enclosing alternative; it is not a nested alternation with only one
2783 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2784 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2785 pattern fragments that do not contain any | characters at this level:
2786 .sp
2787 A (B(*THEN)C) | D
2788 .sp
2789 If A and B are matched, but there is a failure in C, matching does not
2790 backtrack into A; instead it moves to the next alternative, that is, D.
2791 However, if the subpattern containing (*THEN) is given an alternative, it
2792 behaves differently:
2793 .sp
2794 A (B(*THEN)C | (*FAIL)) | D
2795 .sp
2796 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2797 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2798 because there are no more alternatives to try. In this case, matching does now
2799 backtrack into A.
2800 .P
2801 Note also that a conditional subpattern is not considered as having two
2802 alternatives, because only one is ever used. In other words, the | character in
2803 a conditional subpattern has a different meaning. Ignoring white space,
2804 consider:
2805 .sp
2806 ^.*? (?(?=a) a | b(*THEN)c )
2807 .sp
2808 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2809 it initially matches zero characters. The condition (?=a) then fails, the
2810 character "b" is matched, but "c" is not. At this point, matching does not
2811 backtrack to .*? as might perhaps be expected from the presence of the |
2812 character. The conditional subpattern is part of the single alternative that
2813 comprises the whole pattern, and so the match fails. (If there was a backtrack
2814 into .*?, allowing it to match "b", the match would succeed.)
2815 .P
2816 The verbs just described provide four different "strengths" of control when
2817 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2818 next alternative. (*PRUNE) comes next, failing the match at the current
2819 starting position, but allowing an advance to the next character (for an
2820 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2821 than one character. (*COMMIT) is the strongest, causing the entire match to
2822 fail.
2823 .P
2824 If more than one such verb is present in a pattern, the "strongest" one wins.
2825 For example, consider this pattern, where A, B, etc. are complex pattern
2826 fragments:
2827 .sp
2828 (A(*COMMIT)B(*THEN)C|D)
2829 .sp
2830 Once A has matched, PCRE is committed to this match, at the current starting
2831 position. If subsequently B matches, but C does not, the normal (*THEN) action
2832 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2833 overrides.
2834 .
2835 .
2836 .SH "SEE ALSO"
2837 .rs
2838 .sp
2839 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2840 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2841 .
2842 .
2844 .rs
2845 .sp
2846 .nf
2847 Philip Hazel
2848 University Computing Service
2849 Cambridge CB2 3QH, England.
2850 .fi
2851 .
2852 .
2854 .rs
2855 .sp
2856 .nf
2857 Last updated: 19 October 2011
2858 Copyright (c) 1997-2011 University of Cambridge.
2859 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5