ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 577 - (show annotations)
Sun Nov 21 18:51:22 2010 UTC (10 years, 8 months ago) by ph10
File size: 113879 byte(s)
More documentation for (*NO_START_OPT).
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 .\" HTML <a href="pcre.html#utf8support">
36 .\" </a>
37 section on UTF-8 support
38 .\"
39 in the main
40 .\" HREF
41 \fBpcre\fP
42 .\"
43 page.
44 .P
45 Another special sequence that may appear at the start of a pattern or in
46 combination with (*UTF8) is:
47 .sp
48 (*UCP)
49 .sp
50 This has the same effect as setting the PCRE_UCP option: it causes sequences
51 such as \ed and \ew to use Unicode properties to determine character types,
52 instead of recognizing only characters with codes less than 128 via a lookup
53 table.
54 .P
55 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
56 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
57 also some more of these special sequences that are concerned with the handling
58 of newlines; they are described below.
59 .P
60 The remainder of this document discusses the patterns that are supported by
61 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
62 From release 6.0, PCRE offers a second matching function,
63 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
64 Perl-compatible. Some of the features discussed below are not available when
65 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
66 alternative function, and how it differs from the normal function, are
67 discussed in the
68 .\" HREF
69 \fBpcrematching\fP
70 .\"
71 page.
72 .
73 .
74 .\" HTML <a name="newlines"></a>
76 .rs
77 .sp
78 PCRE supports five different conventions for indicating line breaks in
79 strings: a single CR (carriage return) character, a single LF (linefeed)
80 character, the two-character sequence CRLF, any of the three preceding, or any
81 Unicode newline sequence. The
82 .\" HREF
83 \fBpcreapi\fP
84 .\"
85 page has
86 .\" HTML <a href="pcreapi.html#newlines">
87 .\" </a>
88 further discussion
89 .\"
90 about newlines, and shows how to set the newline convention in the
91 \fIoptions\fP arguments for the compiling and matching functions.
92 .P
93 It is also possible to specify a newline convention by starting a pattern
94 string with one of the following five sequences:
95 .sp
96 (*CR) carriage return
97 (*LF) linefeed
98 (*CRLF) carriage return, followed by linefeed
99 (*ANYCRLF) any of the three above
100 (*ANY) all Unicode newline sequences
101 .sp
102 These override the default and the options given to \fBpcre_compile()\fP or
103 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
104 newline sequence, the pattern
105 .sp
106 (*CR)a.b
107 .sp
108 changes the convention to CR. That pattern matches "a\enb" because LF is no
109 longer a newline. Note that these special settings, which are not
110 Perl-compatible, are recognized only at the very start of a pattern, and that
111 they must be in upper case. If more than one of them is present, the last one
112 is used.
113 .P
114 The newline convention affects the interpretation of the dot metacharacter when
115 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
116 affect what the \eR escape sequence matches. By default, this is any Unicode
117 newline sequence, for Perl compatibility. However, this can be changed; see the
118 description of \eR in the section entitled
119 .\" HTML <a href="#newlineseq">
120 .\" </a>
121 "Newline sequences"
122 .\"
123 below. A change of \eR setting can be combined with a change of newline
124 convention.
125 .
126 .
128 .rs
129 .sp
130 A regular expression is a pattern that is matched against a subject string from
131 left to right. Most characters stand for themselves in a pattern, and match the
132 corresponding characters in the subject. As a trivial example, the pattern
133 .sp
134 The quick brown fox
135 .sp
136 matches a portion of a subject string that is identical to itself. When
137 caseless matching is specified (the PCRE_CASELESS option), letters are matched
138 independently of case. In UTF-8 mode, PCRE always understands the concept of
139 case for characters whose values are less than 128, so caseless matching is
140 always possible. For characters with higher values, the concept of case is
141 supported if PCRE is compiled with Unicode property support, but not otherwise.
142 If you want to use caseless matching for characters 128 and above, you must
143 ensure that PCRE is compiled with Unicode property support as well as with
144 UTF-8 support.
145 .P
146 The power of regular expressions comes from the ability to include alternatives
147 and repetitions in the pattern. These are encoded in the pattern by the use of
148 \fImetacharacters\fP, which do not stand for themselves but instead are
149 interpreted in some special way.
150 .P
151 There are two different sets of metacharacters: those that are recognized
152 anywhere in the pattern except within square brackets, and those that are
153 recognized within square brackets. Outside square brackets, the metacharacters
154 are as follows:
155 .sp
156 \e general escape character with several uses
157 ^ assert start of string (or line, in multiline mode)
158 $ assert end of string (or line, in multiline mode)
159 . match any character except newline (by default)
160 [ start character class definition
161 | start of alternative branch
162 ( start subpattern
163 ) end subpattern
164 ? extends the meaning of (
165 also 0 or 1 quantifier
166 also quantifier minimizer
167 * 0 or more quantifier
168 + 1 or more quantifier
169 also "possessive quantifier"
170 { start min/max quantifier
171 .sp
172 Part of a pattern that is in square brackets is called a "character class". In
173 a character class the only metacharacters are:
174 .sp
175 \e general escape character
176 ^ negate the class, but only if the first character
177 - indicates character range
178 .\" JOIN
179 [ POSIX character class (only if followed by POSIX
180 syntax)
181 ] terminates the character class
182 .sp
183 The following sections describe the use of each of the metacharacters.
184 .
185 .
187 .rs
188 .sp
189 The backslash character has several uses. Firstly, if it is followed by a
190 character that is not a number or a letter, it takes away any special meaning
191 that character may have. This use of backslash as an escape character applies
192 both inside and outside character classes.
193 .P
194 For example, if you want to match a * character, you write \e* in the pattern.
195 This escaping action applies whether or not the following character would
196 otherwise be interpreted as a metacharacter, so it is always safe to precede a
197 non-alphanumeric with backslash to specify that it stands for itself. In
198 particular, if you want to match a backslash, you write \e\e.
199 .P
200 In UTF-8 mode, only ASCII numbers and letters have any special meaning after a
201 backslash. All other characters (in particular, those whose codepoints are
202 greater than 127) are treated as literals.
203 .P
204 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
205 pattern (other than in a character class) and characters between a # outside
206 a character class and the next newline are ignored. An escaping backslash can
207 be used to include a whitespace or # character as part of the pattern.
208 .P
209 If you want to remove the special meaning from a sequence of characters, you
210 can do so by putting them between \eQ and \eE. This is different from Perl in
211 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
212 Perl, $ and @ cause variable interpolation. Note the following examples:
213 .sp
214 Pattern PCRE matches Perl matches
215 .sp
216 .\" JOIN
217 \eQabc$xyz\eE abc$xyz abc followed by the
218 contents of $xyz
219 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
220 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
221 .sp
222 The \eQ...\eE sequence is recognized both inside and outside character classes.
223 An isolated \eE that is not preceded by \eQ is ignored.
224 .
225 .
226 .\" HTML <a name="digitsafterbackslash"></a>
227 .SS "Non-printing characters"
228 .rs
229 .sp
230 A second use of backslash provides a way of encoding non-printing characters
231 in patterns in a visible manner. There is no restriction on the appearance of
232 non-printing characters, apart from the binary zero that terminates a pattern,
233 but when a pattern is being prepared by text editing, it is often easier to use
234 one of the following escape sequences than the binary character it represents:
235 .sp
236 \ea alarm, that is, the BEL character (hex 07)
237 \ecx "control-x", where x is any ASCII character
238 \ee escape (hex 1B)
239 \ef formfeed (hex 0C)
240 \en linefeed (hex 0A)
241 \er carriage return (hex 0D)
242 \et tab (hex 09)
243 \eddd character with octal code ddd, or back reference
244 \exhh character with hex code hh
245 \ex{hhh..} character with hex code hhh..
246 .sp
247 The precise effect of \ecx is as follows: if x is a lower case letter, it
248 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
249 Thus \ecz becomes hex 1A (z is 7A), but \ec{ becomes hex 3B ({ is 7B), while
250 \ec; becomes hex 7B (; is 3B). If the byte following \ec has a value greater
251 than 127, a compile-time error occurs. This locks out non-ASCII characters in
252 both byte mode and UTF-8 mode. (When PCRE is compiled in EBCDIC mode, all byte
253 values are valid. A lower case letter is converted to upper case, and then the
254 0xc0 bits are flipped.)
255 .P
256 After \ex, from zero to two hexadecimal digits are read (letters can be in
257 upper or lower case). Any number of hexadecimal digits may appear between \ex{
258 and }, but the value of the character code must be less than 256 in non-UTF-8
259 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
260 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
261 point, which is 10FFFF.
262 .P
263 If characters other than hexadecimal digits appear between \ex{ and }, or if
264 there is no terminating }, this form of escape is not recognized. Instead, the
265 initial \ex will be interpreted as a basic hexadecimal escape, with no
266 following digits, giving a character whose value is zero.
267 .P
268 Characters whose value is less than 256 can be defined by either of the two
269 syntaxes for \ex. There is no difference in the way they are handled. For
270 example, \exdc is exactly the same as \ex{dc}.
271 .P
272 After \e0 up to two further octal digits are read. If there are fewer than two
273 digits, just those that are present are used. Thus the sequence \e0\ex\e07
274 specifies two binary zeros followed by a BEL character (code value 7). Make
275 sure you supply two digits after the initial zero if the pattern character that
276 follows is itself an octal digit.
277 .P
278 The handling of a backslash followed by a digit other than 0 is complicated.
279 Outside a character class, PCRE reads it and any following digits as a decimal
280 number. If the number is less than 10, or if there have been at least that many
281 previous capturing left parentheses in the expression, the entire sequence is
282 taken as a \fIback reference\fP. A description of how this works is given
283 .\" HTML <a href="#backreferences">
284 .\" </a>
285 later,
286 .\"
287 following the discussion of
288 .\" HTML <a href="#subpattern">
289 .\" </a>
290 parenthesized subpatterns.
291 .\"
292 .P
293 Inside a character class, or if the decimal number is greater than 9 and there
294 have not been that many capturing subpatterns, PCRE re-reads up to three octal
295 digits following the backslash, and uses them to generate a data character. Any
296 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
297 character specified in octal must be less than \e400. In UTF-8 mode, values up
298 to \e777 are permitted. For example:
299 .sp
300 \e040 is another way of writing a space
301 .\" JOIN
302 \e40 is the same, provided there are fewer than 40
303 previous capturing subpatterns
304 \e7 is always a back reference
305 .\" JOIN
306 \e11 might be a back reference, or another way of
307 writing a tab
308 \e011 is always a tab
309 \e0113 is a tab followed by the character "3"
310 .\" JOIN
311 \e113 might be a back reference, otherwise the
312 character with octal code 113
313 .\" JOIN
314 \e377 might be a back reference, otherwise
315 the byte consisting entirely of 1 bits
316 .\" JOIN
317 \e81 is either a back reference, or a binary zero
318 followed by the two characters "8" and "1"
319 .sp
320 Note that octal values of 100 or greater must not be introduced by a leading
321 zero, because no more than three octal digits are ever read.
322 .P
323 All the sequences that define a single character value can be used both inside
324 and outside character classes. In addition, inside a character class, the
325 sequence \eb is interpreted as the backspace character (hex 08). The sequences
326 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
327 unrecognized escape sequences, they are treated as the literal characters "B",
328 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
329 set. Outside a character class, these sequences have different meanings.
330 .
331 .
332 .SS "Absolute and relative back references"
333 .rs
334 .sp
335 The sequence \eg followed by an unsigned or a negative number, optionally
336 enclosed in braces, is an absolute or relative back reference. A named back
337 reference can be coded as \eg{name}. Back references are discussed
338 .\" HTML <a href="#backreferences">
339 .\" </a>
340 later,
341 .\"
342 following the discussion of
343 .\" HTML <a href="#subpattern">
344 .\" </a>
345 parenthesized subpatterns.
346 .\"
347 .
348 .
349 .SS "Absolute and relative subroutine calls"
350 .rs
351 .sp
352 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
353 a number enclosed either in angle brackets or single quotes, is an alternative
354 syntax for referencing a subpattern as a "subroutine". Details are discussed
355 .\" HTML <a href="#onigurumasubroutines">
356 .\" </a>
357 later.
358 .\"
359 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
360 synonymous. The former is a back reference; the latter is a
361 .\" HTML <a href="#subpatternsassubroutines">
362 .\" </a>
363 subroutine
364 .\"
365 call.
366 .
367 .
368 .\" HTML <a name="genericchartypes"></a>
369 .SS "Generic character types"
370 .rs
371 .sp
372 Another use of backslash is for specifying generic character types:
373 .sp
374 \ed any decimal digit
375 \eD any character that is not a decimal digit
376 \eh any horizontal whitespace character
377 \eH any character that is not a horizontal whitespace character
378 \es any whitespace character
379 \eS any character that is not a whitespace character
380 \ev any vertical whitespace character
381 \eV any character that is not a vertical whitespace character
382 \ew any "word" character
383 \eW any "non-word" character
384 .sp
385 There is also the single sequence \eN, which matches a non-newline character.
386 This is the same as
387 .\" HTML <a href="#fullstopdot">
388 .\" </a>
389 the "." metacharacter
390 .\"
391 when PCRE_DOTALL is not set.
392 .P
393 Each pair of lower and upper case escape sequences partitions the complete set
394 of characters into two disjoint sets. Any given character matches one, and only
395 one, of each pair. The sequences can appear both inside and outside character
396 classes. They each match one character of the appropriate type. If the current
397 matching point is at the end of the subject string, all of them fail, because
398 there is no character to match.
399 .P
400 For compatibility with Perl, \es does not match the VT character (code 11).
401 This makes it different from the the POSIX "space" class. The \es characters
402 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
403 included in a Perl script, \es may match the VT character. In PCRE, it never
404 does.
405 .P
406 A "word" character is an underscore or any character that is a letter or digit.
407 By default, the definition of letters and digits is controlled by PCRE's
408 low-valued character tables, and may vary if locale-specific matching is taking
409 place (see
410 .\" HTML <a href="pcreapi.html#localesupport">
411 .\" </a>
412 "Locale support"
413 .\"
414 in the
415 .\" HREF
416 \fBpcreapi\fP
417 .\"
418 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
419 or "french" in Windows, some character codes greater than 128 are used for
420 accented letters, and these are then matched by \ew. The use of locales with
421 Unicode is discouraged.
422 .P
423 By default, in UTF-8 mode, characters with values greater than 128 never match
424 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
425 their original meanings from before UTF-8 support was available, mainly for
426 efficiency reasons. However, if PCRE is compiled with Unicode property support,
427 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
428 properties are used to determine character types, as follows:
429 .sp
430 \ed any character that \ep{Nd} matches (decimal digit)
431 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
432 \ew any character that \ep{L} or \ep{N} matches, plus underscore
433 .sp
434 The upper case escapes match the inverse sets of characters. Note that \ed
435 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
436 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
437 \eB because they are defined in terms of \ew and \eW. Matching these sequences
438 is noticeably slower when PCRE_UCP is set.
439 .P
440 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
441 release 5.10. In contrast to the other sequences, which match only ASCII
442 characters by default, these always match certain high-valued codepoints in
443 UTF-8 mode, whether or not PCRE_UCP is set. The horizontal space characters
444 are:
445 .sp
446 U+0009 Horizontal tab
447 U+0020 Space
448 U+00A0 Non-break space
449 U+1680 Ogham space mark
450 U+180E Mongolian vowel separator
451 U+2000 En quad
452 U+2001 Em quad
453 U+2002 En space
454 U+2003 Em space
455 U+2004 Three-per-em space
456 U+2005 Four-per-em space
457 U+2006 Six-per-em space
458 U+2007 Figure space
459 U+2008 Punctuation space
460 U+2009 Thin space
461 U+200A Hair space
462 U+202F Narrow no-break space
463 U+205F Medium mathematical space
464 U+3000 Ideographic space
465 .sp
466 The vertical space characters are:
467 .sp
468 U+000A Linefeed
469 U+000B Vertical tab
470 U+000C Formfeed
471 U+000D Carriage return
472 U+0085 Next line
473 U+2028 Line separator
474 U+2029 Paragraph separator
475 .
476 .
477 .\" HTML <a name="newlineseq"></a>
478 .SS "Newline sequences"
479 .rs
480 .sp
481 Outside a character class, by default, the escape sequence \eR matches any
482 Unicode newline sequence. In non-UTF-8 mode \eR is equivalent to the following:
483 .sp
484 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
485 .sp
486 This is an example of an "atomic group", details of which are given
487 .\" HTML <a href="#atomicgroup">
488 .\" </a>
489 below.
490 .\"
491 This particular group matches either the two-character sequence CR followed by
492 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
493 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
494 line, U+0085). The two-character sequence is treated as a single unit that
495 cannot be split.
496 .P
497 In UTF-8 mode, two additional characters whose codepoints are greater than 255
498 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
499 Unicode character property support is not needed for these characters to be
500 recognized.
501 .P
502 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
503 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
504 either at compile time or when the pattern is matched. (BSR is an abbrevation
505 for "backslash R".) This can be made the default when PCRE is built; if this is
506 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
507 It is also possible to specify these settings by starting a pattern string with
508 one of the following sequences:
509 .sp
510 (*BSR_ANYCRLF) CR, LF, or CRLF only
511 (*BSR_UNICODE) any Unicode newline sequence
512 .sp
513 These override the default and the options given to \fBpcre_compile()\fP or
514 \fBpcre_compile2()\fP, but they can be overridden by options given to
515 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
516 which are not Perl-compatible, are recognized only at the very start of a
517 pattern, and that they must be in upper case. If more than one of them is
518 present, the last one is used. They can be combined with a change of newline
519 convention; for example, a pattern can start with:
520 .sp
522 .sp
523 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
524 a character class, \eR is treated as an unrecognized escape sequence, and so
525 matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
526 .
527 .
528 .\" HTML <a name="uniextseq"></a>
529 .SS Unicode character properties
530 .rs
531 .sp
532 When PCRE is built with Unicode character property support, three additional
533 escape sequences that match characters with specific properties are available.
534 When not in UTF-8 mode, these sequences are of course limited to testing
535 characters whose codepoints are less than 256, but they do work in this mode.
536 The extra escape sequences are:
537 .sp
538 \ep{\fIxx\fP} a character with the \fIxx\fP property
539 \eP{\fIxx\fP} a character without the \fIxx\fP property
540 \eX an extended Unicode sequence
541 .sp
542 The property names represented by \fIxx\fP above are limited to the Unicode
543 script names, the general category properties, "Any", which matches any
544 character (including newline), and some special PCRE properties (described
545 in the
546 .\" HTML <a href="#extraprops">
547 .\" </a>
548 next section).
549 .\"
550 Other Perl properties such as "InMusicalSymbols" are not currently supported by
551 PCRE. Note that \eP{Any} does not match any characters, so always causes a
552 match failure.
553 .P
554 Sets of Unicode characters are defined as belonging to certain scripts. A
555 character from one of these sets can be matched using a script name. For
556 example:
557 .sp
558 \ep{Greek}
559 \eP{Han}
560 .sp
561 Those that are not part of an identified script are lumped together as
562 "Common". The current list of scripts is:
563 .P
564 Arabic,
565 Armenian,
566 Avestan,
567 Balinese,
568 Bamum,
569 Bengali,
570 Bopomofo,
571 Braille,
572 Buginese,
573 Buhid,
574 Canadian_Aboriginal,
575 Carian,
576 Cham,
577 Cherokee,
578 Common,
579 Coptic,
580 Cuneiform,
581 Cypriot,
582 Cyrillic,
583 Deseret,
584 Devanagari,
585 Egyptian_Hieroglyphs,
586 Ethiopic,
587 Georgian,
588 Glagolitic,
589 Gothic,
590 Greek,
591 Gujarati,
592 Gurmukhi,
593 Han,
594 Hangul,
595 Hanunoo,
596 Hebrew,
597 Hiragana,
598 Imperial_Aramaic,
599 Inherited,
600 Inscriptional_Pahlavi,
601 Inscriptional_Parthian,
602 Javanese,
603 Kaithi,
604 Kannada,
605 Katakana,
606 Kayah_Li,
607 Kharoshthi,
608 Khmer,
609 Lao,
610 Latin,
611 Lepcha,
612 Limbu,
613 Linear_B,
614 Lisu,
615 Lycian,
616 Lydian,
617 Malayalam,
618 Meetei_Mayek,
619 Mongolian,
620 Myanmar,
621 New_Tai_Lue,
622 Nko,
623 Ogham,
624 Old_Italic,
625 Old_Persian,
626 Old_South_Arabian,
627 Old_Turkic,
628 Ol_Chiki,
629 Oriya,
630 Osmanya,
631 Phags_Pa,
632 Phoenician,
633 Rejang,
634 Runic,
635 Samaritan,
636 Saurashtra,
637 Shavian,
638 Sinhala,
639 Sundanese,
640 Syloti_Nagri,
641 Syriac,
642 Tagalog,
643 Tagbanwa,
644 Tai_Le,
645 Tai_Tham,
646 Tai_Viet,
647 Tamil,
648 Telugu,
649 Thaana,
650 Thai,
651 Tibetan,
652 Tifinagh,
653 Ugaritic,
654 Vai,
655 Yi.
656 .P
657 Each character has exactly one Unicode general category property, specified by
658 a two-letter abbreviation. For compatibility with Perl, negation can be
659 specified by including a circumflex between the opening brace and the property
660 name. For example, \ep{^Lu} is the same as \eP{Lu}.
661 .P
662 If only one letter is specified with \ep or \eP, it includes all the general
663 category properties that start with that letter. In this case, in the absence
664 of negation, the curly brackets in the escape sequence are optional; these two
665 examples have the same effect:
666 .sp
667 \ep{L}
668 \epL
669 .sp
670 The following general category property codes are supported:
671 .sp
672 C Other
673 Cc Control
674 Cf Format
675 Cn Unassigned
676 Co Private use
677 Cs Surrogate
678 .sp
679 L Letter
680 Ll Lower case letter
681 Lm Modifier letter
682 Lo Other letter
683 Lt Title case letter
684 Lu Upper case letter
685 .sp
686 M Mark
687 Mc Spacing mark
688 Me Enclosing mark
689 Mn Non-spacing mark
690 .sp
691 N Number
692 Nd Decimal number
693 Nl Letter number
694 No Other number
695 .sp
696 P Punctuation
697 Pc Connector punctuation
698 Pd Dash punctuation
699 Pe Close punctuation
700 Pf Final punctuation
701 Pi Initial punctuation
702 Po Other punctuation
703 Ps Open punctuation
704 .sp
705 S Symbol
706 Sc Currency symbol
707 Sk Modifier symbol
708 Sm Mathematical symbol
709 So Other symbol
710 .sp
711 Z Separator
712 Zl Line separator
713 Zp Paragraph separator
714 Zs Space separator
715 .sp
716 The special property L& is also supported: it matches a character that has
717 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
718 a modifier or "other".
719 .P
720 The Cs (Surrogate) property applies only to characters in the range U+D800 to
721 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
722 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
723 (see the discussion of PCRE_NO_UTF8_CHECK in the
724 .\" HREF
725 \fBpcreapi\fP
726 .\"
727 page). Perl does not support the Cs property.
728 .P
729 The long synonyms for property names that Perl supports (such as \ep{Letter})
730 are not supported by PCRE, nor is it permitted to prefix any of these
731 properties with "Is".
732 .P
733 No character that is in the Unicode table has the Cn (unassigned) property.
734 Instead, this property is assumed for any code point that is not in the
735 Unicode table.
736 .P
737 Specifying caseless matching does not affect these escape sequences. For
738 example, \ep{Lu} always matches only upper case letters.
739 .P
740 The \eX escape matches any number of Unicode characters that form an extended
741 Unicode sequence. \eX is equivalent to
742 .sp
743 (?>\ePM\epM*)
744 .sp
745 That is, it matches a character without the "mark" property, followed by zero
746 or more characters with the "mark" property, and treats the sequence as an
747 atomic group
748 .\" HTML <a href="#atomicgroup">
749 .\" </a>
750 (see below).
751 .\"
752 Characters with the "mark" property are typically accents that affect the
753 preceding character. None of them have codepoints less than 256, so in
754 non-UTF-8 mode \eX matches any one character.
755 .P
756 Matching characters by Unicode property is not fast, because PCRE has to search
757 a structure that contains data for over fifteen thousand characters. That is
758 why the traditional escape sequences such as \ed and \ew do not use Unicode
759 properties in PCRE by default, though you can make them do so by setting the
760 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
761 (*UCP).
762 .
763 .
764 .\" HTML <a name="extraprops"></a>
765 .SS PCRE's additional properties
766 .rs
767 .sp
768 As well as the standard Unicode properties described in the previous
769 section, PCRE supports four more that make it possible to convert traditional
770 escape sequences such as \ew and \es and POSIX character classes to use Unicode
771 properties. PCRE uses these non-standard, non-Perl properties internally when
772 PCRE_UCP is set. They are:
773 .sp
774 Xan Any alphanumeric character
775 Xps Any POSIX space character
776 Xsp Any Perl space character
777 Xwd Any Perl "word" character
778 .sp
779 Xan matches characters that have either the L (letter) or the N (number)
780 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
781 carriage return, and any other character that has the Z (separator) property.
782 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
783 same characters as Xan, plus underscore.
784 .
785 .
786 .\" HTML <a name="resetmatchstart"></a>
787 .SS "Resetting the match start"
788 .rs
789 .sp
790 The escape sequence \eK causes any previously matched characters not to be
791 included in the final matched sequence. For example, the pattern:
792 .sp
793 foo\eKbar
794 .sp
795 matches "foobar", but reports that it has matched "bar". This feature is
796 similar to a lookbehind assertion
797 .\" HTML <a href="#lookbehind">
798 .\" </a>
799 (described below).
800 .\"
801 However, in this case, the part of the subject before the real match does not
802 have to be of fixed length, as lookbehind assertions do. The use of \eK does
803 not interfere with the setting of
804 .\" HTML <a href="#subpattern">
805 .\" </a>
806 captured substrings.
807 .\"
808 For example, when the pattern
809 .sp
810 (foo)\eKbar
811 .sp
812 matches "foobar", the first substring is still set to "foo".
813 .P
814 Perl documents that the use of \eK within assertions is "not well defined". In
815 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
816 ignored in negative assertions.
817 .
818 .
819 .\" HTML <a name="smallassertions"></a>
820 .SS "Simple assertions"
821 .rs
822 .sp
823 The final use of backslash is for certain simple assertions. An assertion
824 specifies a condition that has to be met at a particular point in a match,
825 without consuming any characters from the subject string. The use of
826 subpatterns for more complicated assertions is described
827 .\" HTML <a href="#bigassertions">
828 .\" </a>
829 below.
830 .\"
831 The backslashed assertions are:
832 .sp
833 \eb matches at a word boundary
834 \eB matches when not at a word boundary
835 \eA matches at the start of the subject
836 \eZ matches at the end of the subject
837 also matches before a newline at the end of the subject
838 \ez matches only at the end of the subject
839 \eG matches at the first matching position in the subject
840 .sp
841 Inside a character class, \eb has a different meaning; it matches the backspace
842 character. If any other of these assertions appears in a character class, by
843 default it matches the corresponding literal character (for example, \eB
844 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
845 escape sequence" error is generated instead.
846 .P
847 A word boundary is a position in the subject string where the current character
848 and the previous character do not both match \ew or \eW (i.e. one matches
849 \ew and the other matches \eW), or the start or end of the string if the
850 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
851 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
852 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
853 of word" or "end of word" metasequence. However, whatever follows \eb normally
854 determines which it is. For example, the fragment \eba matches "a" at the start
855 of a word.
856 .P
857 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
858 dollar (described in the next section) in that they only ever match at the very
859 start and end of the subject string, whatever options are set. Thus, they are
860 independent of multiline mode. These three assertions are not affected by the
861 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
862 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
863 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
864 at a point other than the beginning of the subject, \eA can never match. The
865 difference between \eZ and \ez is that \eZ matches before a newline at the end
866 of the string as well as at the very end, whereas \ez matches only at the end.
867 .P
868 The \eG assertion is true only when the current matching position is at the
869 start point of the match, as specified by the \fIstartoffset\fP argument of
870 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
871 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
872 arguments, you can mimic Perl's /g option, and it is in this kind of
873 implementation where \eG can be useful.
874 .P
875 Note, however, that PCRE's interpretation of \eG, as the start of the current
876 match, is subtly different from Perl's, which defines it as the end of the
877 previous match. In Perl, these can be different when the previously matched
878 string was empty. Because PCRE does just one match at a time, it cannot
879 reproduce this behaviour.
880 .P
881 If all the alternatives of a pattern begin with \eG, the expression is anchored
882 to the starting match position, and the "anchored" flag is set in the compiled
883 regular expression.
884 .
885 .
887 .rs
888 .sp
889 Outside a character class, in the default matching mode, the circumflex
890 character is an assertion that is true only if the current matching point is
891 at the start of the subject string. If the \fIstartoffset\fP argument of
892 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
893 option is unset. Inside a character class, circumflex has an entirely different
894 meaning
895 .\" HTML <a href="#characterclass">
896 .\" </a>
897 (see below).
898 .\"
899 .P
900 Circumflex need not be the first character of the pattern if a number of
901 alternatives are involved, but it should be the first thing in each alternative
902 in which it appears if the pattern is ever to match that branch. If all
903 possible alternatives start with a circumflex, that is, if the pattern is
904 constrained to match only at the start of the subject, it is said to be an
905 "anchored" pattern. (There are also other constructs that can cause a pattern
906 to be anchored.)
907 .P
908 A dollar character is an assertion that is true only if the current matching
909 point is at the end of the subject string, or immediately before a newline
910 at the end of the string (by default). Dollar need not be the last character of
911 the pattern if a number of alternatives are involved, but it should be the last
912 item in any branch in which it appears. Dollar has no special meaning in a
913 character class.
914 .P
915 The meaning of dollar can be changed so that it matches only at the very end of
916 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
917 does not affect the \eZ assertion.
918 .P
919 The meanings of the circumflex and dollar characters are changed if the
920 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
921 immediately after internal newlines as well as at the start of the subject
922 string. It does not match after a newline that ends the string. A dollar
923 matches before any newlines in the string, as well as at the very end, when
924 PCRE_MULTILINE is set. When newline is specified as the two-character
925 sequence CRLF, isolated CR and LF characters do not indicate newlines.
926 .P
927 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
928 \en represents a newline) in multiline mode, but not otherwise. Consequently,
929 patterns that are anchored in single line mode because all branches start with
930 ^ are not anchored in multiline mode, and a match for circumflex is possible
931 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
932 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
933 .P
934 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
935 end of the subject in both modes, and if all branches of a pattern start with
936 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
937 .
938 .
939 .\" HTML <a name="fullstopdot"></a>
941 .rs
942 .sp
943 Outside a character class, a dot in the pattern matches any one character in
944 the subject string except (by default) a character that signifies the end of a
945 line. In UTF-8 mode, the matched character may be more than one byte long.
946 .P
947 When a line ending is defined as a single character, dot never matches that
948 character; when the two-character sequence CRLF is used, dot does not match CR
949 if it is immediately followed by LF, but otherwise it matches all characters
950 (including isolated CRs and LFs). When any Unicode line endings are being
951 recognized, dot does not match CR or LF or any of the other line ending
952 characters.
953 .P
954 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
955 option is set, a dot matches any one character, without exception. If the
956 two-character sequence CRLF is present in the subject string, it takes two dots
957 to match it.
958 .P
959 The handling of dot is entirely independent of the handling of circumflex and
960 dollar, the only relationship being that they both involve newlines. Dot has no
961 special meaning in a character class.
962 .P
963 The escape sequence \eN behaves like a dot, except that it is not affected by
964 the PCRE_DOTALL option. In other words, it matches any character except one
965 that signifies the end of a line.
966 .
967 .
969 .rs
970 .sp
971 Outside a character class, the escape sequence \eC matches any one byte, both
972 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
973 characters. The feature is provided in Perl in order to match individual bytes
974 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes, the
975 rest of the string may start with a malformed UTF-8 character. For this reason,
976 the \eC escape sequence is best avoided.
977 .P
978 PCRE does not allow \eC to appear in lookbehind assertions
979 .\" HTML <a href="#lookbehind">
980 .\" </a>
981 (described below),
982 .\"
983 because in UTF-8 mode this would make it impossible to calculate the length of
984 the lookbehind.
985 .
986 .
987 .\" HTML <a name="characterclass"></a>
989 .rs
990 .sp
991 An opening square bracket introduces a character class, terminated by a closing
992 square bracket. A closing square bracket on its own is not special by default.
993 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
994 bracket causes a compile-time error. If a closing square bracket is required as
995 a member of the class, it should be the first data character in the class
996 (after an initial circumflex, if present) or escaped with a backslash.
997 .P
998 A character class matches a single character in the subject. In UTF-8 mode, the
999 character may be more than one byte long. A matched character must be in the
1000 set of characters defined by the class, unless the first character in the class
1001 definition is a circumflex, in which case the subject character must not be in
1002 the set defined by the class. If a circumflex is actually required as a member
1003 of the class, ensure it is not the first character, or escape it with a
1004 backslash.
1005 .P
1006 For example, the character class [aeiou] matches any lower case vowel, while
1007 [^aeiou] matches any character that is not a lower case vowel. Note that a
1008 circumflex is just a convenient notation for specifying the characters that
1009 are in the class by enumerating those that are not. A class that starts with a
1010 circumflex is not an assertion; it still consumes a character from the subject
1011 string, and therefore it fails if the current pointer is at the end of the
1012 string.
1013 .P
1014 In UTF-8 mode, characters with values greater than 255 can be included in a
1015 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1016 .P
1017 When caseless matching is set, any letters in a class represent both their
1018 upper case and lower case versions, so for example, a caseless [aeiou] matches
1019 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1020 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1021 case for characters whose values are less than 128, so caseless matching is
1022 always possible. For characters with higher values, the concept of case is
1023 supported if PCRE is compiled with Unicode property support, but not otherwise.
1024 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1025 you must ensure that PCRE is compiled with Unicode property support as well as
1026 with UTF-8 support.
1027 .P
1028 Characters that might indicate line breaks are never treated in any special way
1029 when matching character classes, whatever line-ending sequence is in use, and
1030 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1031 such as [^a] always matches one of these characters.
1032 .P
1033 The minus (hyphen) character can be used to specify a range of characters in a
1034 character class. For example, [d-m] matches any letter between d and m,
1035 inclusive. If a minus character is required in a class, it must be escaped with
1036 a backslash or appear in a position where it cannot be interpreted as
1037 indicating a range, typically as the first or last character in the class.
1038 .P
1039 It is not possible to have the literal character "]" as the end character of a
1040 range. A pattern such as [W-]46] is interpreted as a class of two characters
1041 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1042 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1043 the end of range, so [W-\e]46] is interpreted as a class containing a range
1044 followed by two other characters. The octal or hexadecimal representation of
1045 "]" can also be used to end a range.
1046 .P
1047 Ranges operate in the collating sequence of character values. They can also be
1048 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1049 mode, ranges can include characters whose values are greater than 255, for
1050 example [\ex{100}-\ex{2ff}].
1051 .P
1052 If a range that includes letters is used when caseless matching is set, it
1053 matches the letters in either case. For example, [W-c] is equivalent to
1054 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1055 tables for a French locale are in use, [\exc8-\excb] matches accented E
1056 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1057 characters with values greater than 128 only when it is compiled with Unicode
1058 property support.
1059 .P
1060 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1061 \eV, \ew, and \eW may appear in a character class, and add the characters that
1062 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1063 digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1064 and their upper case partners, just as it does when they appear outside a
1065 character class, as described in the section entitled
1066 .\" HTML <a href="#genericchartypes">
1067 .\" </a>
1068 "Generic character types"
1069 .\"
1070 above. The escape sequence \eb has a different meaning inside a character
1071 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1072 are not special inside a character class. Like any other unrecognized escape
1073 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1074 default, but cause an error if the PCRE_EXTRA option is set.
1075 .P
1076 A circumflex can conveniently be used with the upper case character types to
1077 specify a more restricted set of characters than the matching lower case type.
1078 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1079 whereas [\ew] includes underscore. A positive character class should be read as
1080 "something OR something OR ..." and a negative class as "NOT something AND NOT
1081 something AND NOT ...".
1082 .P
1083 The only metacharacters that are recognized in character classes are backslash,
1084 hyphen (only where it can be interpreted as specifying a range), circumflex
1085 (only at the start), opening square bracket (only when it can be interpreted as
1086 introducing a POSIX class name - see the next section), and the terminating
1087 closing square bracket. However, escaping other non-alphanumeric characters
1088 does no harm.
1089 .
1090 .
1092 .rs
1093 .sp
1094 Perl supports the POSIX notation for character classes. This uses names
1095 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1096 this notation. For example,
1097 .sp
1098 [01[:alpha:]%]
1099 .sp
1100 matches "0", "1", any alphabetic character, or "%". The supported class names
1101 are:
1102 .sp
1103 alnum letters and digits
1104 alpha letters
1105 ascii character codes 0 - 127
1106 blank space or tab only
1107 cntrl control characters
1108 digit decimal digits (same as \ed)
1109 graph printing characters, excluding space
1110 lower lower case letters
1111 print printing characters, including space
1112 punct printing characters, excluding letters and digits and space
1113 space white space (not quite the same as \es)
1114 upper upper case letters
1115 word "word" characters (same as \ew)
1116 xdigit hexadecimal digits
1117 .sp
1118 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1119 space (32). Notice that this list includes the VT character (code 11). This
1120 makes "space" different to \es, which does not include VT (for Perl
1121 compatibility).
1122 .P
1123 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1124 5.8. Another Perl extension is negation, which is indicated by a ^ character
1125 after the colon. For example,
1126 .sp
1127 [12[:^digit:]]
1128 .sp
1129 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1130 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1131 supported, and an error is given if they are encountered.
1132 .P
1133 By default, in UTF-8 mode, characters with values greater than 128 do not match
1134 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1135 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1136 character properties are used. This is achieved by replacing the POSIX classes
1137 by other sequences, as follows:
1138 .sp
1139 [:alnum:] becomes \ep{Xan}
1140 [:alpha:] becomes \ep{L}
1141 [:blank:] becomes \eh
1142 [:digit:] becomes \ep{Nd}
1143 [:lower:] becomes \ep{Ll}
1144 [:space:] becomes \ep{Xps}
1145 [:upper:] becomes \ep{Lu}
1146 [:word:] becomes \ep{Xwd}
1147 .sp
1148 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1149 classes are unchanged, and match only characters with code points less than
1150 128.
1151 .
1152 .
1154 .rs
1155 .sp
1156 Vertical bar characters are used to separate alternative patterns. For example,
1157 the pattern
1158 .sp
1159 gilbert|sullivan
1160 .sp
1161 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1162 and an empty alternative is permitted (matching the empty string). The matching
1163 process tries each alternative in turn, from left to right, and the first one
1164 that succeeds is used. If the alternatives are within a subpattern
1165 .\" HTML <a href="#subpattern">
1166 .\" </a>
1167 (defined below),
1168 .\"
1169 "succeeds" means matching the rest of the main pattern as well as the
1170 alternative in the subpattern.
1171 .
1172 .
1174 .rs
1175 .sp
1176 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1177 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1178 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1179 The option letters are
1180 .sp
1181 i for PCRE_CASELESS
1183 s for PCRE_DOTALL
1184 x for PCRE_EXTENDED
1185 .sp
1186 For example, (?im) sets caseless, multiline matching. It is also possible to
1187 unset these options by preceding the letter with a hyphen, and a combined
1188 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1189 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1190 permitted. If a letter appears both before and after the hyphen, the option is
1191 unset.
1192 .P
1193 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1194 changed in the same way as the Perl-compatible options by using the characters
1195 J, U and X respectively.
1196 .P
1197 When one of these option changes occurs at top level (that is, not inside
1198 subpattern parentheses), the change applies to the remainder of the pattern
1199 that follows. If the change is placed right at the start of a pattern, PCRE
1200 extracts it into the global options (and it will therefore show up in data
1201 extracted by the \fBpcre_fullinfo()\fP function).
1202 .P
1203 An option change within a subpattern (see below for a description of
1204 subpatterns) affects only that part of the subpattern that follows it, so
1205 .sp
1206 (a(?i)b)c
1207 .sp
1208 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1209 By this means, options can be made to have different settings in different
1210 parts of the pattern. Any changes made in one alternative do carry on
1211 into subsequent branches within the same subpattern. For example,
1212 .sp
1213 (a(?i)b|c)
1214 .sp
1215 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1216 branch is abandoned before the option setting. This is because the effects of
1217 option settings happen at compile time. There would be some very weird
1218 behaviour otherwise.
1219 .P
1220 \fBNote:\fP There are other PCRE-specific options that can be set by the
1221 application when the compile or match functions are called. In some cases the
1222 pattern can contain special leading sequences such as (*CRLF) to override what
1223 the application has set or what has been defaulted. Details are given in the
1224 section entitled
1225 .\" HTML <a href="#newlineseq">
1226 .\" </a>
1227 "Newline sequences"
1228 .\"
1229 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1230 to set UTF-8 and Unicode property modes; they are equivalent to setting the
1231 PCRE_UTF8 and the PCRE_UCP options, respectively.
1232 .
1233 .
1234 .\" HTML <a name="subpattern"></a>
1236 .rs
1237 .sp
1238 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1239 Turning part of a pattern into a subpattern does two things:
1240 .sp
1241 1. It localizes a set of alternatives. For example, the pattern
1242 .sp
1243 cat(aract|erpillar|)
1244 .sp
1245 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1246 match "cataract", "erpillar" or an empty string.
1247 .sp
1248 2. It sets up the subpattern as a capturing subpattern. This means that, when
1249 the whole pattern matches, that portion of the subject string that matched the
1250 subpattern is passed back to the caller via the \fIovector\fP argument of
1251 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1252 from 1) to obtain numbers for the capturing subpatterns. For example, if the
1253 string "the red king" is matched against the pattern
1254 .sp
1255 the ((red|white) (king|queen))
1256 .sp
1257 the captured substrings are "red king", "red", and "king", and are numbered 1,
1258 2, and 3, respectively.
1259 .P
1260 The fact that plain parentheses fulfil two functions is not always helpful.
1261 There are often times when a grouping subpattern is required without a
1262 capturing requirement. If an opening parenthesis is followed by a question mark
1263 and a colon, the subpattern does not do any capturing, and is not counted when
1264 computing the number of any subsequent capturing subpatterns. For example, if
1265 the string "the white queen" is matched against the pattern
1266 .sp
1267 the ((?:red|white) (king|queen))
1268 .sp
1269 the captured substrings are "white queen" and "queen", and are numbered 1 and
1270 2. The maximum number of capturing subpatterns is 65535.
1271 .P
1272 As a convenient shorthand, if any option settings are required at the start of
1273 a non-capturing subpattern, the option letters may appear between the "?" and
1274 the ":". Thus the two patterns
1275 .sp
1276 (?i:saturday|sunday)
1277 (?:(?i)saturday|sunday)
1278 .sp
1279 match exactly the same set of strings. Because alternative branches are tried
1280 from left to right, and options are not reset until the end of the subpattern
1281 is reached, an option setting in one branch does affect subsequent branches, so
1282 the above patterns match "SUNDAY" as well as "Saturday".
1283 .
1284 .
1285 .\" HTML <a name="dupsubpatternnumber"></a>
1287 .rs
1288 .sp
1289 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1290 the same numbers for its capturing parentheses. Such a subpattern starts with
1291 (?| and is itself a non-capturing subpattern. For example, consider this
1292 pattern:
1293 .sp
1294 (?|(Sat)ur|(Sun))day
1295 .sp
1296 Because the two alternatives are inside a (?| group, both sets of capturing
1297 parentheses are numbered one. Thus, when the pattern matches, you can look
1298 at captured substring number one, whichever alternative matched. This construct
1299 is useful when you want to capture part, but not all, of one of a number of
1300 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1301 number is reset at the start of each branch. The numbers of any capturing
1302 parentheses that follow the subpattern start after the highest number used in
1303 any branch. The following example is taken from the Perl documentation. The
1304 numbers underneath show in which buffer the captured content will be stored.
1305 .sp
1306 # before ---------------branch-reset----------- after
1307 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1308 # 1 2 2 3 2 3 4
1309 .sp
1310 A back reference to a numbered subpattern uses the most recent value that is
1311 set for that number by any subpattern. The following pattern matches "abcabc"
1312 or "defdef":
1313 .sp
1314 /(?|(abc)|(def))\e1/
1315 .sp
1316 In contrast, a recursive or "subroutine" call to a numbered subpattern always
1317 refers to the first one in the pattern with the given number. The following
1318 pattern matches "abcabc" or "defabc":
1319 .sp
1320 /(?|(abc)|(def))(?1)/
1321 .sp
1322 If a
1323 .\" HTML <a href="#conditions">
1324 .\" </a>
1325 condition test
1326 .\"
1327 for a subpattern's having matched refers to a non-unique number, the test is
1328 true if any of the subpatterns of that number have matched.
1329 .P
1330 An alternative approach to using this "branch reset" feature is to use
1331 duplicate named subpatterns, as described in the next section.
1332 .
1333 .
1335 .rs
1336 .sp
1337 Identifying capturing parentheses by number is simple, but it can be very hard
1338 to keep track of the numbers in complicated regular expressions. Furthermore,
1339 if an expression is modified, the numbers may change. To help with this
1340 difficulty, PCRE supports the naming of subpatterns. This feature was not
1341 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1342 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1343 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1344 have different names, but PCRE does not.
1345 .P
1346 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1347 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1348 parentheses from other parts of the pattern, such as
1349 .\" HTML <a href="#backreferences">
1350 .\" </a>
1351 back references,
1352 .\"
1353 .\" HTML <a href="#recursion">
1354 .\" </a>
1355 recursion,
1356 .\"
1357 and
1358 .\" HTML <a href="#conditions">
1359 .\" </a>
1360 conditions,
1361 .\"
1362 can be made by name as well as by number.
1363 .P
1364 Names consist of up to 32 alphanumeric characters and underscores. Named
1365 capturing parentheses are still allocated numbers as well as names, exactly as
1366 if the names were not present. The PCRE API provides function calls for
1367 extracting the name-to-number translation table from a compiled pattern. There
1368 is also a convenience function for extracting a captured substring by name.
1369 .P
1370 By default, a name must be unique within a pattern, but it is possible to relax
1371 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1372 names are also always permitted for subpatterns with the same number, set up as
1373 described in the previous section.) Duplicate names can be useful for patterns
1374 where only one instance of the named parentheses can match. Suppose you want to
1375 match the name of a weekday, either as a 3-letter abbreviation or as the full
1376 name, and in both cases you want to extract the abbreviation. This pattern
1377 (ignoring the line breaks) does the job:
1378 .sp
1379 (?<DN>Mon|Fri|Sun)(?:day)?|
1380 (?<DN>Tue)(?:sday)?|
1381 (?<DN>Wed)(?:nesday)?|
1382 (?<DN>Thu)(?:rsday)?|
1383 (?<DN>Sat)(?:urday)?
1384 .sp
1385 There are five capturing substrings, but only one is ever set after a match.
1386 (An alternative way of solving this problem is to use a "branch reset"
1387 subpattern, as described in the previous section.)
1388 .P
1389 The convenience function for extracting the data by name returns the substring
1390 for the first (and in this example, the only) subpattern of that name that
1391 matched. This saves searching to find which numbered subpattern it was.
1392 .P
1393 If you make a back reference to a non-unique named subpattern from elsewhere in
1394 the pattern, the one that corresponds to the first occurrence of the name is
1395 used. In the absence of duplicate numbers (see the previous section) this is
1396 the one with the lowest number. If you use a named reference in a condition
1397 test (see the
1398 .\"
1399 .\" HTML <a href="#conditions">
1400 .\" </a>
1401 section about conditions
1402 .\"
1403 below), either to check whether a subpattern has matched, or to check for
1404 recursion, all subpatterns with the same name are tested. If the condition is
1405 true for any one of them, the overall condition is true. This is the same
1406 behaviour as testing by number. For further details of the interfaces for
1407 handling named subpatterns, see the
1408 .\" HREF
1409 \fBpcreapi\fP
1410 .\"
1411 documentation.
1412 .P
1413 \fBWarning:\fP You cannot use different names to distinguish between two
1414 subpatterns with the same number because PCRE uses only the numbers when
1415 matching. For this reason, an error is given at compile time if different names
1416 are given to subpatterns with the same number. However, you can give the same
1417 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1418 .
1419 .
1421 .rs
1422 .sp
1423 Repetition is specified by quantifiers, which can follow any of the following
1424 items:
1425 .sp
1426 a literal data character
1427 the dot metacharacter
1428 the \eC escape sequence
1429 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1430 the \eR escape sequence
1431 an escape such as \ed or \epL that matches a single character
1432 a character class
1433 a back reference (see next section)
1434 a parenthesized subpattern (unless it is an assertion)
1435 a recursive or "subroutine" call to a subpattern
1436 .sp
1437 The general repetition quantifier specifies a minimum and maximum number of
1438 permitted matches, by giving the two numbers in curly brackets (braces),
1439 separated by a comma. The numbers must be less than 65536, and the first must
1440 be less than or equal to the second. For example:
1441 .sp
1442 z{2,4}
1443 .sp
1444 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1445 character. If the second number is omitted, but the comma is present, there is
1446 no upper limit; if the second number and the comma are both omitted, the
1447 quantifier specifies an exact number of required matches. Thus
1448 .sp
1449 [aeiou]{3,}
1450 .sp
1451 matches at least 3 successive vowels, but may match many more, while
1452 .sp
1453 \ed{8}
1454 .sp
1455 matches exactly 8 digits. An opening curly bracket that appears in a position
1456 where a quantifier is not allowed, or one that does not match the syntax of a
1457 quantifier, is taken as a literal character. For example, {,6} is not a
1458 quantifier, but a literal string of four characters.
1459 .P
1460 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1461 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1462 which is represented by a two-byte sequence. Similarly, when Unicode property
1463 support is available, \eX{3} matches three Unicode extended sequences, each of
1464 which may be several bytes long (and they may be of different lengths).
1465 .P
1466 The quantifier {0} is permitted, causing the expression to behave as if the
1467 previous item and the quantifier were not present. This may be useful for
1468 subpatterns that are referenced as
1469 .\" HTML <a href="#subpatternsassubroutines">
1470 .\" </a>
1471 subroutines
1472 .\"
1473 from elsewhere in the pattern (but see also the section entitled
1474 .\" HTML <a href="#subdefine">
1475 .\" </a>
1476 "Defining subpatterns for use by reference only"
1477 .\"
1478 below). Items other than subpatterns that have a {0} quantifier are omitted
1479 from the compiled pattern.
1480 .P
1481 For convenience, the three most common quantifiers have single-character
1482 abbreviations:
1483 .sp
1484 * is equivalent to {0,}
1485 + is equivalent to {1,}
1486 ? is equivalent to {0,1}
1487 .sp
1488 It is possible to construct infinite loops by following a subpattern that can
1489 match no characters with a quantifier that has no upper limit, for example:
1490 .sp
1491 (a?)*
1492 .sp
1493 Earlier versions of Perl and PCRE used to give an error at compile time for
1494 such patterns. However, because there are cases where this can be useful, such
1495 patterns are now accepted, but if any repetition of the subpattern does in fact
1496 match no characters, the loop is forcibly broken.
1497 .P
1498 By default, the quantifiers are "greedy", that is, they match as much as
1499 possible (up to the maximum number of permitted times), without causing the
1500 rest of the pattern to fail. The classic example of where this gives problems
1501 is in trying to match comments in C programs. These appear between /* and */
1502 and within the comment, individual * and / characters may appear. An attempt to
1503 match C comments by applying the pattern
1504 .sp
1505 /\e*.*\e*/
1506 .sp
1507 to the string
1508 .sp
1509 /* first comment */ not comment /* second comment */
1510 .sp
1511 fails, because it matches the entire string owing to the greediness of the .*
1512 item.
1513 .P
1514 However, if a quantifier is followed by a question mark, it ceases to be
1515 greedy, and instead matches the minimum number of times possible, so the
1516 pattern
1517 .sp
1518 /\e*.*?\e*/
1519 .sp
1520 does the right thing with the C comments. The meaning of the various
1521 quantifiers is not otherwise changed, just the preferred number of matches.
1522 Do not confuse this use of question mark with its use as a quantifier in its
1523 own right. Because it has two uses, it can sometimes appear doubled, as in
1524 .sp
1525 \ed??\ed
1526 .sp
1527 which matches one digit by preference, but can match two if that is the only
1528 way the rest of the pattern matches.
1529 .P
1530 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1531 the quantifiers are not greedy by default, but individual ones can be made
1532 greedy by following them with a question mark. In other words, it inverts the
1533 default behaviour.
1534 .P
1535 When a parenthesized subpattern is quantified with a minimum repeat count that
1536 is greater than 1 or with a limited maximum, more memory is required for the
1537 compiled pattern, in proportion to the size of the minimum or maximum.
1538 .P
1539 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1540 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1541 implicitly anchored, because whatever follows will be tried against every
1542 character position in the subject string, so there is no point in retrying the
1543 overall match at any position after the first. PCRE normally treats such a
1544 pattern as though it were preceded by \eA.
1545 .P
1546 In cases where it is known that the subject string contains no newlines, it is
1547 worth setting PCRE_DOTALL in order to obtain this optimization, or
1548 alternatively using ^ to indicate anchoring explicitly.
1549 .P
1550 However, there is one situation where the optimization cannot be used. When .*
1551 is inside capturing parentheses that are the subject of a back reference
1552 elsewhere in the pattern, a match at the start may fail where a later one
1553 succeeds. Consider, for example:
1554 .sp
1555 (.*)abc\e1
1556 .sp
1557 If the subject is "xyz123abc123" the match point is the fourth character. For
1558 this reason, such a pattern is not implicitly anchored.
1559 .P
1560 When a capturing subpattern is repeated, the value captured is the substring
1561 that matched the final iteration. For example, after
1562 .sp
1563 (tweedle[dume]{3}\es*)+
1564 .sp
1565 has matched "tweedledum tweedledee" the value of the captured substring is
1566 "tweedledee". However, if there are nested capturing subpatterns, the
1567 corresponding captured values may have been set in previous iterations. For
1568 example, after
1569 .sp
1570 /(a|(b))+/
1571 .sp
1572 matches "aba" the value of the second captured substring is "b".
1573 .
1574 .
1575 .\" HTML <a name="atomicgroup"></a>
1577 .rs
1578 .sp
1579 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1580 repetition, failure of what follows normally causes the repeated item to be
1581 re-evaluated to see if a different number of repeats allows the rest of the
1582 pattern to match. Sometimes it is useful to prevent this, either to change the
1583 nature of the match, or to cause it fail earlier than it otherwise might, when
1584 the author of the pattern knows there is no point in carrying on.
1585 .P
1586 Consider, for example, the pattern \ed+foo when applied to the subject line
1587 .sp
1588 123456bar
1589 .sp
1590 After matching all 6 digits and then failing to match "foo", the normal
1591 action of the matcher is to try again with only 5 digits matching the \ed+
1592 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1593 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1594 that once a subpattern has matched, it is not to be re-evaluated in this way.
1595 .P
1596 If we use atomic grouping for the previous example, the matcher gives up
1597 immediately on failing to match "foo" the first time. The notation is a kind of
1598 special parenthesis, starting with (?> as in this example:
1599 .sp
1600 (?>\ed+)foo
1601 .sp
1602 This kind of parenthesis "locks up" the part of the pattern it contains once
1603 it has matched, and a failure further into the pattern is prevented from
1604 backtracking into it. Backtracking past it to previous items, however, works as
1605 normal.
1606 .P
1607 An alternative description is that a subpattern of this type matches the string
1608 of characters that an identical standalone pattern would match, if anchored at
1609 the current point in the subject string.
1610 .P
1611 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1612 the above example can be thought of as a maximizing repeat that must swallow
1613 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1614 number of digits they match in order to make the rest of the pattern match,
1615 (?>\ed+) can only match an entire sequence of digits.
1616 .P
1617 Atomic groups in general can of course contain arbitrarily complicated
1618 subpatterns, and can be nested. However, when the subpattern for an atomic
1619 group is just a single repeated item, as in the example above, a simpler
1620 notation, called a "possessive quantifier" can be used. This consists of an
1621 additional + character following a quantifier. Using this notation, the
1622 previous example can be rewritten as
1623 .sp
1624 \ed++foo
1625 .sp
1626 Note that a possessive quantifier can be used with an entire group, for
1627 example:
1628 .sp
1629 (abc|xyz){2,3}+
1630 .sp
1631 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1632 option is ignored. They are a convenient notation for the simpler forms of
1633 atomic group. However, there is no difference in the meaning of a possessive
1634 quantifier and the equivalent atomic group, though there may be a performance
1635 difference; possessive quantifiers should be slightly faster.
1636 .P
1637 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1638 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1639 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1640 package, and PCRE copied it from there. It ultimately found its way into Perl
1641 at release 5.10.
1642 .P
1643 PCRE has an optimization that automatically "possessifies" certain simple
1644 pattern constructs. For example, the sequence A+B is treated as A++B because
1645 there is no point in backtracking into a sequence of A's when B must follow.
1646 .P
1647 When a pattern contains an unlimited repeat inside a subpattern that can itself
1648 be repeated an unlimited number of times, the use of an atomic group is the
1649 only way to avoid some failing matches taking a very long time indeed. The
1650 pattern
1651 .sp
1652 (\eD+|<\ed+>)*[!?]
1653 .sp
1654 matches an unlimited number of substrings that either consist of non-digits, or
1655 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1656 quickly. However, if it is applied to
1657 .sp
1658 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1659 .sp
1660 it takes a long time before reporting failure. This is because the string can
1661 be divided between the internal \eD+ repeat and the external * repeat in a
1662 large number of ways, and all have to be tried. (The example uses [!?] rather
1663 than a single character at the end, because both PCRE and Perl have an
1664 optimization that allows for fast failure when a single character is used. They
1665 remember the last single character that is required for a match, and fail early
1666 if it is not present in the string.) If the pattern is changed so that it uses
1667 an atomic group, like this:
1668 .sp
1669 ((?>\eD+)|<\ed+>)*[!?]
1670 .sp
1671 sequences of non-digits cannot be broken, and failure happens quickly.
1672 .
1673 .
1674 .\" HTML <a name="backreferences"></a>
1676 .rs
1677 .sp
1678 Outside a character class, a backslash followed by a digit greater than 0 (and
1679 possibly further digits) is a back reference to a capturing subpattern earlier
1680 (that is, to its left) in the pattern, provided there have been that many
1681 previous capturing left parentheses.
1682 .P
1683 However, if the decimal number following the backslash is less than 10, it is
1684 always taken as a back reference, and causes an error only if there are not
1685 that many capturing left parentheses in the entire pattern. In other words, the
1686 parentheses that are referenced need not be to the left of the reference for
1687 numbers less than 10. A "forward back reference" of this type can make sense
1688 when a repetition is involved and the subpattern to the right has participated
1689 in an earlier iteration.
1690 .P
1691 It is not possible to have a numerical "forward back reference" to a subpattern
1692 whose number is 10 or more using this syntax because a sequence such as \e50 is
1693 interpreted as a character defined in octal. See the subsection entitled
1694 "Non-printing characters"
1695 .\" HTML <a href="#digitsafterbackslash">
1696 .\" </a>
1697 above
1698 .\"
1699 for further details of the handling of digits following a backslash. There is
1700 no such problem when named parentheses are used. A back reference to any
1701 subpattern is possible using named parentheses (see below).
1702 .P
1703 Another way of avoiding the ambiguity inherent in the use of digits following a
1704 backslash is to use the \eg escape sequence. This escape must be followed by an
1705 unsigned number or a negative number, optionally enclosed in braces. These
1706 examples are all identical:
1707 .sp
1708 (ring), \e1
1709 (ring), \eg1
1710 (ring), \eg{1}
1711 .sp
1712 An unsigned number specifies an absolute reference without the ambiguity that
1713 is present in the older syntax. It is also useful when literal digits follow
1714 the reference. A negative number is a relative reference. Consider this
1715 example:
1716 .sp
1717 (abc(def)ghi)\eg{-1}
1718 .sp
1719 The sequence \eg{-1} is a reference to the most recently started capturing
1720 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1721 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1722 can be helpful in long patterns, and also in patterns that are created by
1723 joining together fragments that contain references within themselves.
1724 .P
1725 A back reference matches whatever actually matched the capturing subpattern in
1726 the current subject string, rather than anything matching the subpattern
1727 itself (see
1728 .\" HTML <a href="#subpatternsassubroutines">
1729 .\" </a>
1730 "Subpatterns as subroutines"
1731 .\"
1732 below for a way of doing that). So the pattern
1733 .sp
1734 (sens|respons)e and \e1ibility
1735 .sp
1736 matches "sense and sensibility" and "response and responsibility", but not
1737 "sense and responsibility". If caseful matching is in force at the time of the
1738 back reference, the case of letters is relevant. For example,
1739 .sp
1740 ((?i)rah)\es+\e1
1741 .sp
1742 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1743 capturing subpattern is matched caselessly.
1744 .P
1745 There are several different ways of writing back references to named
1746 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1747 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1748 back reference syntax, in which \eg can be used for both numeric and named
1749 references, is also supported. We could rewrite the above example in any of
1750 the following ways:
1751 .sp
1752 (?<p1>(?i)rah)\es+\ek<p1>
1753 (?'p1'(?i)rah)\es+\ek{p1}
1754 (?P<p1>(?i)rah)\es+(?P=p1)
1755 (?<p1>(?i)rah)\es+\eg{p1}
1756 .sp
1757 A subpattern that is referenced by name may appear in the pattern before or
1758 after the reference.
1759 .P
1760 There may be more than one back reference to the same subpattern. If a
1761 subpattern has not actually been used in a particular match, any back
1762 references to it always fail by default. For example, the pattern
1763 .sp
1764 (a|(bc))\e2
1765 .sp
1766 always fails if it starts to match "a" rather than "bc". However, if the
1767 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1768 unset value matches an empty string.
1769 .P
1770 Because there may be many capturing parentheses in a pattern, all digits
1771 following a backslash are taken as part of a potential back reference number.
1772 If the pattern continues with a digit character, some delimiter must be used to
1773 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1774 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1775 .\" HTML <a href="#comments">
1776 .\" </a>
1777 "Comments"
1778 .\"
1779 below) can be used.
1780 .
1781 .SS "Recursive back references"
1782 .rs
1783 .sp
1784 A back reference that occurs inside the parentheses to which it refers fails
1785 when the subpattern is first used, so, for example, (a\e1) never matches.
1786 However, such references can be useful inside repeated subpatterns. For
1787 example, the pattern
1788 .sp
1789 (a|b\e1)+
1790 .sp
1791 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1792 the subpattern, the back reference matches the character string corresponding
1793 to the previous iteration. In order for this to work, the pattern must be such
1794 that the first iteration does not need to match the back reference. This can be
1795 done using alternation, as in the example above, or by a quantifier with a
1796 minimum of zero.
1797 .P
1798 Back references of this type cause the group that they reference to be treated
1799 as an
1800 .\" HTML <a href="#atomicgroup">
1801 .\" </a>
1802 atomic group.
1803 .\"
1804 Once the whole group has been matched, a subsequent matching failure cannot
1805 cause backtracking into the middle of the group.
1806 .
1807 .
1808 .\" HTML <a name="bigassertions"></a>
1810 .rs
1811 .sp
1812 An assertion is a test on the characters following or preceding the current
1813 matching point that does not actually consume any characters. The simple
1814 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1815 .\" HTML <a href="#smallassertions">
1816 .\" </a>
1817 above.
1818 .\"
1819 .P
1820 More complicated assertions are coded as subpatterns. There are two kinds:
1821 those that look ahead of the current position in the subject string, and those
1822 that look behind it. An assertion subpattern is matched in the normal way,
1823 except that it does not cause the current matching position to be changed.
1824 .P
1825 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1826 because it makes no sense to assert the same thing several times. If any kind
1827 of assertion contains capturing subpatterns within it, these are counted for
1828 the purposes of numbering the capturing subpatterns in the whole pattern.
1829 However, substring capturing is carried out only for positive assertions,
1830 because it does not make sense for negative assertions.
1831 .
1832 .
1833 .SS "Lookahead assertions"
1834 .rs
1835 .sp
1836 Lookahead assertions start with (?= for positive assertions and (?! for
1837 negative assertions. For example,
1838 .sp
1839 \ew+(?=;)
1840 .sp
1841 matches a word followed by a semicolon, but does not include the semicolon in
1842 the match, and
1843 .sp
1844 foo(?!bar)
1845 .sp
1846 matches any occurrence of "foo" that is not followed by "bar". Note that the
1847 apparently similar pattern
1848 .sp
1849 (?!foo)bar
1850 .sp
1851 does not find an occurrence of "bar" that is preceded by something other than
1852 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1853 (?!foo) is always true when the next three characters are "bar". A
1854 lookbehind assertion is needed to achieve the other effect.
1855 .P
1856 If you want to force a matching failure at some point in a pattern, the most
1857 convenient way to do it is with (?!) because an empty string always matches, so
1858 an assertion that requires there not to be an empty string must always fail.
1859 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1860 .
1861 .
1862 .\" HTML <a name="lookbehind"></a>
1863 .SS "Lookbehind assertions"
1864 .rs
1865 .sp
1866 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1867 negative assertions. For example,
1868 .sp
1869 (?<!foo)bar
1870 .sp
1871 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1872 a lookbehind assertion are restricted such that all the strings it matches must
1873 have a fixed length. However, if there are several top-level alternatives, they
1874 do not all have to have the same fixed length. Thus
1875 .sp
1876 (?<=bullock|donkey)
1877 .sp
1878 is permitted, but
1879 .sp
1880 (?<!dogs?|cats?)
1881 .sp
1882 causes an error at compile time. Branches that match different length strings
1883 are permitted only at the top level of a lookbehind assertion. This is an
1884 extension compared with Perl, which requires all branches to match the same
1885 length of string. An assertion such as
1886 .sp
1887 (?<=ab(c|de))
1888 .sp
1889 is not permitted, because its single top-level branch can match two different
1890 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1891 branches:
1892 .sp
1893 (?<=abc|abde)
1894 .sp
1895 In some cases, the escape sequence \eK
1896 .\" HTML <a href="#resetmatchstart">
1897 .\" </a>
1898 (see above)
1899 .\"
1900 can be used instead of a lookbehind assertion to get round the fixed-length
1901 restriction.
1902 .P
1903 The implementation of lookbehind assertions is, for each alternative, to
1904 temporarily move the current position back by the fixed length and then try to
1905 match. If there are insufficient characters before the current position, the
1906 assertion fails.
1907 .P
1908 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1909 to appear in lookbehind assertions, because it makes it impossible to calculate
1910 the length of the lookbehind. The \eX and \eR escapes, which can match
1911 different numbers of bytes, are also not permitted.
1912 .P
1913 .\" HTML <a href="#subpatternsassubroutines">
1914 .\" </a>
1915 "Subroutine"
1916 .\"
1917 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1918 as the subpattern matches a fixed-length string.
1919 .\" HTML <a href="#recursion">
1920 .\" </a>
1921 Recursion,
1922 .\"
1923 however, is not supported.
1924 .P
1925 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1926 specify efficient matching of fixed-length strings at the end of subject
1927 strings. Consider a simple pattern such as
1928 .sp
1929 abcd$
1930 .sp
1931 when applied to a long string that does not match. Because matching proceeds
1932 from left to right, PCRE will look for each "a" in the subject and then see if
1933 what follows matches the rest of the pattern. If the pattern is specified as
1934 .sp
1935 ^.*abcd$
1936 .sp
1937 the initial .* matches the entire string at first, but when this fails (because
1938 there is no following "a"), it backtracks to match all but the last character,
1939 then all but the last two characters, and so on. Once again the search for "a"
1940 covers the entire string, from right to left, so we are no better off. However,
1941 if the pattern is written as
1942 .sp
1943 ^.*+(?<=abcd)
1944 .sp
1945 there can be no backtracking for the .*+ item; it can match only the entire
1946 string. The subsequent lookbehind assertion does a single test on the last four
1947 characters. If it fails, the match fails immediately. For long strings, this
1948 approach makes a significant difference to the processing time.
1949 .
1950 .
1951 .SS "Using multiple assertions"
1952 .rs
1953 .sp
1954 Several assertions (of any sort) may occur in succession. For example,
1955 .sp
1956 (?<=\ed{3})(?<!999)foo
1957 .sp
1958 matches "foo" preceded by three digits that are not "999". Notice that each of
1959 the assertions is applied independently at the same point in the subject
1960 string. First there is a check that the previous three characters are all
1961 digits, and then there is a check that the same three characters are not "999".
1962 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1963 of which are digits and the last three of which are not "999". For example, it
1964 doesn't match "123abcfoo". A pattern to do that is
1965 .sp
1966 (?<=\ed{3}...)(?<!999)foo
1967 .sp
1968 This time the first assertion looks at the preceding six characters, checking
1969 that the first three are digits, and then the second assertion checks that the
1970 preceding three characters are not "999".
1971 .P
1972 Assertions can be nested in any combination. For example,
1973 .sp
1974 (?<=(?<!foo)bar)baz
1975 .sp
1976 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1977 preceded by "foo", while
1978 .sp
1979 (?<=\ed{3}(?!999)...)foo
1980 .sp
1981 is another pattern that matches "foo" preceded by three digits and any three
1982 characters that are not "999".
1983 .
1984 .
1985 .\" HTML <a name="conditions"></a>
1987 .rs
1988 .sp
1989 It is possible to cause the matching process to obey a subpattern
1990 conditionally or to choose between two alternative subpatterns, depending on
1991 the result of an assertion, or whether a specific capturing subpattern has
1992 already been matched. The two possible forms of conditional subpattern are:
1993 .sp
1994 (?(condition)yes-pattern)
1995 (?(condition)yes-pattern|no-pattern)
1996 .sp
1997 If the condition is satisfied, the yes-pattern is used; otherwise the
1998 no-pattern (if present) is used. If there are more than two alternatives in the
1999 subpattern, a compile-time error occurs. Each of the two alternatives may
2000 itself contain nested subpatterns of any form, including conditional
2001 subpatterns; the restriction to two alternatives applies only at the level of
2002 the condition. This pattern fragment is an example where the alternatives are
2003 complex:
2004 .sp
2005 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2006 .sp
2007 .P
2008 There are four kinds of condition: references to subpatterns, references to
2009 recursion, a pseudo-condition called DEFINE, and assertions.
2010 .
2011 .SS "Checking for a used subpattern by number"
2012 .rs
2013 .sp
2014 If the text between the parentheses consists of a sequence of digits, the
2015 condition is true if a capturing subpattern of that number has previously
2016 matched. If there is more than one capturing subpattern with the same number
2017 (see the earlier
2018 .\"
2019 .\" HTML <a href="#recursion">
2020 .\" </a>
2021 section about duplicate subpattern numbers),
2022 .\"
2023 the condition is true if any of them have matched. An alternative notation is
2024 to precede the digits with a plus or minus sign. In this case, the subpattern
2025 number is relative rather than absolute. The most recently opened parentheses
2026 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2027 loops it can also make sense to refer to subsequent groups. The next
2028 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2029 zero in any of these forms is not used; it provokes a compile-time error.)
2030 .P
2031 Consider the following pattern, which contains non-significant white space to
2032 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2033 three parts for ease of discussion:
2034 .sp
2035 ( \e( )? [^()]+ (?(1) \e) )
2036 .sp
2037 The first part matches an optional opening parenthesis, and if that
2038 character is present, sets it as the first captured substring. The second part
2039 matches one or more characters that are not parentheses. The third part is a
2040 conditional subpattern that tests whether or not the first set of parentheses
2041 matched. If they did, that is, if subject started with an opening parenthesis,
2042 the condition is true, and so the yes-pattern is executed and a closing
2043 parenthesis is required. Otherwise, since no-pattern is not present, the
2044 subpattern matches nothing. In other words, this pattern matches a sequence of
2045 non-parentheses, optionally enclosed in parentheses.
2046 .P
2047 If you were embedding this pattern in a larger one, you could use a relative
2048 reference:
2049 .sp
2050 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2051 .sp
2052 This makes the fragment independent of the parentheses in the larger pattern.
2053 .
2054 .SS "Checking for a used subpattern by name"
2055 .rs
2056 .sp
2057 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2058 subpattern by name. For compatibility with earlier versions of PCRE, which had
2059 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2060 there is a possible ambiguity with this syntax, because subpattern names may
2061 consist entirely of digits. PCRE looks first for a named subpattern; if it
2062 cannot find one and the name consists entirely of digits, PCRE looks for a
2063 subpattern of that number, which must be greater than zero. Using subpattern
2064 names that consist entirely of digits is not recommended.
2065 .P
2066 Rewriting the above example to use a named subpattern gives this:
2067 .sp
2068 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2069 .sp
2070 If the name used in a condition of this kind is a duplicate, the test is
2071 applied to all subpatterns of the same name, and is true if any one of them has
2072 matched.
2073 .
2074 .SS "Checking for pattern recursion"
2075 .rs
2076 .sp
2077 If the condition is the string (R), and there is no subpattern with the name R,
2078 the condition is true if a recursive call to the whole pattern or any
2079 subpattern has been made. If digits or a name preceded by ampersand follow the
2080 letter R, for example:
2081 .sp
2082 (?(R3)...) or (?(R&name)...)
2083 .sp
2084 the condition is true if the most recent recursion is into a subpattern whose
2085 number or name is given. This condition does not check the entire recursion
2086 stack. If the name used in a condition of this kind is a duplicate, the test is
2087 applied to all subpatterns of the same name, and is true if any one of them is
2088 the most recent recursion.
2089 .P
2090 At "top level", all these recursion test conditions are false.
2091 .\" HTML <a href="#recursion">
2092 .\" </a>
2093 The syntax for recursive patterns
2094 .\"
2095 is described below.
2096 .
2097 .\" HTML <a name="subdefine"></a>
2098 .SS "Defining subpatterns for use by reference only"
2099 .rs
2100 .sp
2101 If the condition is the string (DEFINE), and there is no subpattern with the
2102 name DEFINE, the condition is always false. In this case, there may be only one
2103 alternative in the subpattern. It is always skipped if control reaches this
2104 point in the pattern; the idea of DEFINE is that it can be used to define
2105 "subroutines" that can be referenced from elsewhere. (The use of
2106 .\" HTML <a href="#subpatternsassubroutines">
2107 .\" </a>
2108 "subroutines"
2109 .\"
2110 is described below.) For example, a pattern to match an IPv4 address such as
2111 "" could be written like this (ignore whitespace and line
2112 breaks):
2113 .sp
2114 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2115 \eb (?&byte) (\e.(?&byte)){3} \eb
2116 .sp
2117 The first part of the pattern is a DEFINE group inside which a another group
2118 named "byte" is defined. This matches an individual component of an IPv4
2119 address (a number less than 256). When matching takes place, this part of the
2120 pattern is skipped because DEFINE acts like a false condition. The rest of the
2121 pattern uses references to the named group to match the four dot-separated
2122 components of an IPv4 address, insisting on a word boundary at each end.
2123 .
2124 .SS "Assertion conditions"
2125 .rs
2126 .sp
2127 If the condition is not in any of the above formats, it must be an assertion.
2128 This may be a positive or negative lookahead or lookbehind assertion. Consider
2129 this pattern, again containing non-significant white space, and with the two
2130 alternatives on the second line:
2131 .sp
2132 (?(?=[^a-z]*[a-z])
2133 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2134 .sp
2135 The condition is a positive lookahead assertion that matches an optional
2136 sequence of non-letters followed by a letter. In other words, it tests for the
2137 presence of at least one letter in the subject. If a letter is found, the
2138 subject is matched against the first alternative; otherwise it is matched
2139 against the second. This pattern matches strings in one of the two forms
2140 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2141 .
2142 .
2143 .\" HTML <a name="comments"></a>
2145 .rs
2146 .sp
2147 There are two ways of including comments in patterns that are processed by
2148 PCRE. In both cases, the start of the comment must not be in a character class,
2149 nor in the middle of any other sequence of related characters such as (?: or a
2150 subpattern name or number. The characters that make up a comment play no part
2151 in the pattern matching.
2152 .P
2153 The sequence (?# marks the start of a comment that continues up to the next
2154 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2155 option is set, an unescaped # character also introduces a comment, which in
2156 this case continues to immediately after the next newline character or
2157 character sequence in the pattern. Which characters are interpreted as newlines
2158 is controlled by the options passed to \fBpcre_compile()\fP or by a special
2159 sequence at the start of the pattern, as described in the section entitled
2160 .\" HTML <a href="#newlines">
2161 .\" </a>
2162 "Newline conventions"
2163 .\"
2164 above. Note that the end of this type of comment is a literal newline sequence
2165 in the pattern; escape sequences that happen to represent a newline do not
2166 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2167 default newline convention is in force:
2168 .sp
2169 abc #comment \en still comment
2170 .sp
2171 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2172 a newline in the pattern. The sequence \en is still literal at this stage, so
2173 it does not terminate the comment. Only an actual character with the code value
2174 0x0a (the default newline) does so.
2175 .
2176 .
2177 .\" HTML <a name="recursion"></a>
2179 .rs
2180 .sp
2181 Consider the problem of matching a string in parentheses, allowing for
2182 unlimited nested parentheses. Without the use of recursion, the best that can
2183 be done is to use a pattern that matches up to some fixed depth of nesting. It
2184 is not possible to handle an arbitrary nesting depth.
2185 .P
2186 For some time, Perl has provided a facility that allows regular expressions to
2187 recurse (amongst other things). It does this by interpolating Perl code in the
2188 expression at run time, and the code can refer to the expression itself. A Perl
2189 pattern using code interpolation to solve the parentheses problem can be
2190 created like this:
2191 .sp
2192 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2193 .sp
2194 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2195 recursively to the pattern in which it appears.
2196 .P
2197 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2198 supports special syntax for recursion of the entire pattern, and also for
2199 individual subpattern recursion. After its introduction in PCRE and Python,
2200 this kind of recursion was subsequently introduced into Perl at release 5.10.
2201 .P
2202 A special item that consists of (? followed by a number greater than zero and a
2203 closing parenthesis is a recursive call of the subpattern of the given number,
2204 provided that it occurs inside that subpattern. (If not, it is a
2205 .\" HTML <a href="#subpatternsassubroutines">
2206 .\" </a>
2207 "subroutine"
2208 .\"
2209 call, which is described in the next section.) The special item (?R) or (?0) is
2210 a recursive call of the entire regular expression.
2211 .P
2212 This PCRE pattern solves the nested parentheses problem (assume the
2213 PCRE_EXTENDED option is set so that white space is ignored):
2214 .sp
2215 \e( ( [^()]++ | (?R) )* \e)
2216 .sp
2217 First it matches an opening parenthesis. Then it matches any number of
2218 substrings which can either be a sequence of non-parentheses, or a recursive
2219 match of the pattern itself (that is, a correctly parenthesized substring).
2220 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2221 to avoid backtracking into sequences of non-parentheses.
2222 .P
2223 If this were part of a larger pattern, you would not want to recurse the entire
2224 pattern, so instead you could use this:
2225 .sp
2226 ( \e( ( [^()]++ | (?1) )* \e) )
2227 .sp
2228 We have put the pattern into parentheses, and caused the recursion to refer to
2229 them instead of the whole pattern.
2230 .P
2231 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2232 is made easier by the use of relative references. Instead of (?1) in the
2233 pattern above you can write (?-2) to refer to the second most recently opened
2234 parentheses preceding the recursion. In other words, a negative number counts
2235 capturing parentheses leftwards from the point at which it is encountered.
2236 .P
2237 It is also possible to refer to subsequently opened parentheses, by writing
2238 references such as (?+2). However, these cannot be recursive because the
2239 reference is not inside the parentheses that are referenced. They are always
2240 .\" HTML <a href="#subpatternsassubroutines">
2241 .\" </a>
2242 "subroutine"
2243 .\"
2244 calls, as described in the next section.
2245 .P
2246 An alternative approach is to use named parentheses instead. The Perl syntax
2247 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2248 could rewrite the above example as follows:
2249 .sp
2250 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2251 .sp
2252 If there is more than one subpattern with the same name, the earliest one is
2253 used.
2254 .P
2255 This particular example pattern that we have been looking at contains nested
2256 unlimited repeats, and so the use of a possessive quantifier for matching
2257 strings of non-parentheses is important when applying the pattern to strings
2258 that do not match. For example, when this pattern is applied to
2259 .sp
2260 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2261 .sp
2262 it yields "no match" quickly. However, if a possessive quantifier is not used,
2263 the match runs for a very long time indeed because there are so many different
2264 ways the + and * repeats can carve up the subject, and all have to be tested
2265 before failure can be reported.
2266 .P
2267 At the end of a match, the values of capturing parentheses are those from
2268 the outermost level. If you want to obtain intermediate values, a callout
2269 function can be used (see below and the
2270 .\" HREF
2271 \fBpcrecallout\fP
2272 .\"
2273 documentation). If the pattern above is matched against
2274 .sp
2275 (ab(cd)ef)
2276 .sp
2277 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2278 the last value taken on at the top level. If a capturing subpattern is not
2279 matched at the top level, its final value is unset, even if it is (temporarily)
2280 set at a deeper level.
2281 .P
2282 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2283 obtain extra memory to store data during a recursion, which it does by using
2284 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2285 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2286 .P
2287 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2288 Consider this pattern, which matches text in angle brackets, allowing for
2289 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2290 recursing), whereas any characters are permitted at the outer level.
2291 .sp
2292 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2293 .sp
2294 In this pattern, (?(R) is the start of a conditional subpattern, with two
2295 different alternatives for the recursive and non-recursive cases. The (?R) item
2296 is the actual recursive call.
2297 .
2298 .
2299 .\" HTML <a name="recursiondifference"></a>
2300 .SS "Recursion difference from Perl"
2301 .rs
2302 .sp
2303 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2304 treated as an atomic group. That is, once it has matched some of the subject
2305 string, it is never re-entered, even if it contains untried alternatives and
2306 there is a subsequent matching failure. This can be illustrated by the
2307 following pattern, which purports to match a palindromic string that contains
2308 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2309 .sp
2310 ^(.|(.)(?1)\e2)$
2311 .sp
2312 The idea is that it either matches a single character, or two identical
2313 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2314 it does not if the pattern is longer than three characters. Consider the
2315 subject string "abcba":
2316 .P
2317 At the top level, the first character is matched, but as it is not at the end
2318 of the string, the first alternative fails; the second alternative is taken
2319 and the recursion kicks in. The recursive call to subpattern 1 successfully
2320 matches the next character ("b"). (Note that the beginning and end of line
2321 tests are not part of the recursion).
2322 .P
2323 Back at the top level, the next character ("c") is compared with what
2324 subpattern 2 matched, which was "a". This fails. Because the recursion is
2325 treated as an atomic group, there are now no backtracking points, and so the
2326 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2327 try the second alternative.) However, if the pattern is written with the
2328 alternatives in the other order, things are different:
2329 .sp
2330 ^((.)(?1)\e2|.)$
2331 .sp
2332 This time, the recursing alternative is tried first, and continues to recurse
2333 until it runs out of characters, at which point the recursion fails. But this
2334 time we do have another alternative to try at the higher level. That is the big
2335 difference: in the previous case the remaining alternative is at a deeper
2336 recursion level, which PCRE cannot use.
2337 .P
2338 To change the pattern so that it matches all palindromic strings, not just
2339 those with an odd number of characters, it is tempting to change the pattern to
2340 this:
2341 .sp
2342 ^((.)(?1)\e2|.?)$
2343 .sp
2344 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2345 deeper recursion has matched a single character, it cannot be entered again in
2346 order to match an empty string. The solution is to separate the two cases, and
2347 write out the odd and even cases as alternatives at the higher level:
2348 .sp
2349 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2350 .sp
2351 If you want to match typical palindromic phrases, the pattern has to ignore all
2352 non-word characters, which can be done like this:
2353 .sp
2354 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2355 .sp
2356 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2357 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2358 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2359 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2360 more) to match typical phrases, and Perl takes so long that you think it has
2361 gone into a loop.
2362 .P
2363 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2364 string does not start with a palindrome that is shorter than the entire string.
2365 For example, although "abcba" is correctly matched, if the subject is "ababa",
2366 PCRE finds the palindrome "aba" at the start, then fails at top level because
2367 the end of the string does not follow. Once again, it cannot jump back into the
2368 recursion to try other alternatives, so the entire match fails.
2369 .
2370 .
2371 .\" HTML <a name="subpatternsassubroutines"></a>
2373 .rs
2374 .sp
2375 If the syntax for a recursive subpattern reference (either by number or by
2376 name) is used outside the parentheses to which it refers, it operates like a
2377 subroutine in a programming language. The "called" subpattern may be defined
2378 before or after the reference. A numbered reference can be absolute or
2379 relative, as in these examples:
2380 .sp
2381 (...(absolute)...)...(?2)...
2382 (...(relative)...)...(?-1)...
2383 (...(?+1)...(relative)...
2384 .sp
2385 An earlier example pointed out that the pattern
2386 .sp
2387 (sens|respons)e and \e1ibility
2388 .sp
2389 matches "sense and sensibility" and "response and responsibility", but not
2390 "sense and responsibility". If instead the pattern
2391 .sp
2392 (sens|respons)e and (?1)ibility
2393 .sp
2394 is used, it does match "sense and responsibility" as well as the other two
2395 strings. Another example is given in the discussion of DEFINE above.
2396 .P
2397 Like recursive subpatterns, a subroutine call is always treated as an atomic
2398 group. That is, once it has matched some of the subject string, it is never
2399 re-entered, even if it contains untried alternatives and there is a subsequent
2400 matching failure. Any capturing parentheses that are set during the subroutine
2401 call revert to their previous values afterwards.
2402 .P
2403 When a subpattern is used as a subroutine, processing options such as
2404 case-independence are fixed when the subpattern is defined. They cannot be
2405 changed for different calls. For example, consider this pattern:
2406 .sp
2407 (abc)(?i:(?-1))
2408 .sp
2409 It matches "abcabc". It does not match "abcABC" because the change of
2410 processing option does not affect the called subpattern.
2411 .
2412 .
2413 .\" HTML <a name="onigurumasubroutines"></a>
2415 .rs
2416 .sp
2417 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2418 a number enclosed either in angle brackets or single quotes, is an alternative
2419 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2420 are two of the examples used above, rewritten using this syntax:
2421 .sp
2422 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2423 (sens|respons)e and \eg'1'ibility
2424 .sp
2425 PCRE supports an extension to Oniguruma: if a number is preceded by a
2426 plus or a minus sign it is taken as a relative reference. For example:
2427 .sp
2428 (abc)(?i:\eg<-1>)
2429 .sp
2430 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2431 synonymous. The former is a back reference; the latter is a subroutine call.
2432 .
2433 .
2435 .rs
2436 .sp
2437 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2438 code to be obeyed in the middle of matching a regular expression. This makes it
2439 possible, amongst other things, to extract different substrings that match the
2440 same pair of parentheses when there is a repetition.
2441 .P
2442 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2443 code. The feature is called "callout". The caller of PCRE provides an external
2444 function by putting its entry point in the global variable \fIpcre_callout\fP.
2445 By default, this variable contains NULL, which disables all calling out.
2446 .P
2447 Within a regular expression, (?C) indicates the points at which the external
2448 function is to be called. If you want to identify different callout points, you
2449 can put a number less than 256 after the letter C. The default value is zero.
2450 For example, this pattern has two callout points:
2451 .sp
2452 (?C1)abc(?C2)def
2453 .sp
2454 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2455 automatically installed before each item in the pattern. They are all numbered
2456 255.
2457 .P
2458 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2459 set), the external function is called. It is provided with the number of the
2460 callout, the position in the pattern, and, optionally, one item of data
2461 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2462 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2463 description of the interface to the callout function is given in the
2464 .\" HREF
2465 \fBpcrecallout\fP
2466 .\"
2467 documentation.
2468 .
2469 .
2470 .\" HTML <a name="backtrackcontrol"></a>
2472 .rs
2473 .sp
2474 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2475 are described in the Perl documentation as "experimental and subject to change
2476 or removal in a future version of Perl". It goes on to say: "Their usage in
2477 production code should be noted to avoid problems during upgrades." The same
2478 remarks apply to the PCRE features described in this section.
2479 .P
2480 Since these verbs are specifically related to backtracking, most of them can be
2481 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2482 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2483 failing negative assertion, they cause an error if encountered by
2484 \fBpcre_dfa_exec()\fP.
2485 .P
2486 If any of these verbs are used in an assertion or subroutine subpattern
2487 (including recursive subpatterns), their effect is confined to that subpattern;
2488 it does not extend to the surrounding pattern. Note that such subpatterns are
2489 processed as anchored at the point where they are tested.
2490 .P
2491 The new verbs make use of what was previously invalid syntax: an opening
2492 parenthesis followed by an asterisk. They are generally of the form
2493 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2494 depending on whether or not an argument is present. An name is a sequence of
2495 letters, digits, and underscores. If the name is empty, that is, if the closing
2496 parenthesis immediately follows the colon, the effect is as if the colon were
2497 not there. Any number of these verbs may occur in a pattern.
2498 .P
2499 PCRE contains some optimizations that are used to speed up matching by running
2500 some checks at the start of each match attempt. For example, it may know the
2501 minimum length of matching subject, or that a particular character must be
2502 present. When one of these optimizations suppresses the running of a match, any
2503 included backtracking verbs will not, of course, be processed. You can suppress
2504 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2505 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2506 pattern with (*NO_START_OPT).
2507 .
2508 .
2509 .SS "Verbs that act immediately"
2510 .rs
2511 .sp
2512 The following verbs act as soon as they are encountered. They may not be
2513 followed by a name.
2514 .sp
2515 (*ACCEPT)
2516 .sp
2517 This verb causes the match to end successfully, skipping the remainder of the
2518 pattern. When inside a recursion, only the innermost pattern is ended
2519 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2520 captured. (This feature was added to PCRE at release 8.00.) For example:
2521 .sp
2522 A((?:A|B(*ACCEPT)|C)D)
2523 .sp
2524 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2525 the outer parentheses.
2526 .sp
2527 (*FAIL) or (*F)
2528 .sp
2529 This verb causes the match to fail, forcing backtracking to occur. It is
2530 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2531 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2532 Perl features that are not present in PCRE. The nearest equivalent is the
2533 callout feature, as for example in this pattern:
2534 .sp
2535 a+(?C)(*FAIL)
2536 .sp
2537 A match with the string "aaaa" always fails, but the callout is taken before
2538 each backtrack happens (in this example, 10 times).
2539 .
2540 .
2541 .SS "Recording which path was taken"
2542 .rs
2543 .sp
2544 There is one verb whose main purpose is to track how a match was arrived at,
2545 though it also has a secondary use in conjunction with advancing the match
2546 starting point (see (*SKIP) below).
2547 .sp
2548 (*MARK:NAME) or (*:NAME)
2549 .sp
2550 A name is always required with this verb. There may be as many instances of
2551 (*MARK) as you like in a pattern, and their names do not have to be unique.
2552 .P
2553 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2554 to the caller via the \fIpcre_extra\fP data structure, as described in the
2555 .\" HTML <a href="pcreapi.html#extradata">
2556 .\" </a>
2557 section on \fIpcre_extra\fP
2558 .\"
2559 in the
2560 .\" HREF
2561 \fBpcreapi\fP
2562 .\"
2563 documentation. No data is returned for a partial match. Here is an example of
2564 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2565 outputting of (*MARK) data:
2566 .sp
2567 /X(*MARK:A)Y|X(*MARK:B)Z/K
2568 XY
2569 0: XY
2570 MK: A
2571 XZ
2572 0: XZ
2573 MK: B
2574 .sp
2575 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2576 indicates which of the two alternatives matched. This is a more efficient way
2577 of obtaining this information than putting each alternative in its own
2578 capturing parentheses.
2579 .P
2580 A name may also be returned after a failed match if the final path through the
2581 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2582 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2583 starting point for matching is advanced, the final check is often with an empty
2584 string, causing a failure before (*MARK) is reached. For example:
2585 .sp
2586 /X(*MARK:A)Y|X(*MARK:B)Z/K
2587 XP
2588 No match
2589 .sp
2590 There are three potential starting points for this match (starting with X,
2591 starting with P, and with an empty string). If the pattern is anchored, the
2592 result is different:
2593 .sp
2594 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2595 XP
2596 No match, mark = B
2597 .sp
2598 PCRE's start-of-match optimizations can also interfere with this. For example,
2599 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2600 subject length for a match, a shorter subject will not be scanned at all.
2601 .P
2602 Note that similar anomalies (though different in detail) exist in Perl, no
2603 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2604 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2605 .
2606 .
2607 .SS "Verbs that act after backtracking"
2608 .rs
2609 .sp
2610 The following verbs do nothing when they are encountered. Matching continues
2611 with what follows, but if there is no subsequent match, causing a backtrack to
2612 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2613 the verb. However, when one of these verbs appears inside an atomic group, its
2614 effect is confined to that group, because once the group has been matched,
2615 there is never any backtracking into it. In this situation, backtracking can
2616 "jump back" to the left of the entire atomic group. (Remember also, as stated
2617 above, that this localization also applies in subroutine calls and assertions.)
2618 .P
2619 These verbs differ in exactly what kind of failure occurs when backtracking
2620 reaches them.
2621 .sp
2622 (*COMMIT)
2623 .sp
2624 This verb, which may not be followed by a name, causes the whole match to fail
2625 outright if the rest of the pattern does not match. Even if the pattern is
2626 unanchored, no further attempts to find a match by advancing the starting point
2627 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2628 finding a match at the current starting point, or not at all. For example:
2629 .sp
2630 a+(*COMMIT)b
2631 .sp
2632 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2633 dynamic anchor, or "I've started, so I must finish." The name of the most
2634 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2635 match failure.
2636 .P
2637 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2638 unless PCRE's start-of-match optimizations are turned off, as shown in this
2639 \fBpcretest\fP example:
2640 .sp
2641 /(*COMMIT)abc/
2642 xyzabc
2643 0: abc
2644 xyzabc\eY
2645 No match
2646 .sp
2647 PCRE knows that any match must start with "a", so the optimization skips along
2648 the subject to "a" before running the first match attempt, which succeeds. When
2649 the optimization is disabled by the \eY escape in the second subject, the match
2650 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2651 starting points.
2652 .sp
2653 (*PRUNE) or (*PRUNE:NAME)
2654 .sp
2655 This verb causes the match to fail at the current starting position in the
2656 subject if the rest of the pattern does not match. If the pattern is
2657 unanchored, the normal "bumpalong" advance to the next starting character then
2658 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2659 reached, or when matching to the right of (*PRUNE), but if there is no match to
2660 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2661 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2662 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2663 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2664 match fails completely; the name is passed back if this is the final attempt.
2665 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2666 pattern (*PRUNE) has the same effect as (*COMMIT).
2667 .sp
2668 (*SKIP)
2669 .sp
2670 This verb, when given without a name, is like (*PRUNE), except that if the
2671 pattern is unanchored, the "bumpalong" advance is not to the next character,
2672 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2673 signifies that whatever text was matched leading up to it cannot be part of a
2674 successful match. Consider:
2675 .sp
2676 a+(*SKIP)b
2677 .sp
2678 If the subject is "aaaac...", after the first match attempt fails (starting at
2679 the first character in the string), the starting point skips on to start the
2680 next attempt at "c". Note that a possessive quantifer does not have the same
2681 effect as this example; although it would suppress backtracking during the
2682 first match attempt, the second attempt would start at the second character
2683 instead of skipping on to "c".
2684 .sp
2685 (*SKIP:NAME)
2686 .sp
2687 When (*SKIP) has an associated name, its behaviour is modified. If the
2688 following pattern fails to match, the previous path through the pattern is
2689 searched for the most recent (*MARK) that has the same name. If one is found,
2690 the "bumpalong" advance is to the subject position that corresponds to that
2691 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2692 matching name is found, normal "bumpalong" of one character happens (the
2693 (*SKIP) is ignored).
2694 .sp
2695 (*THEN) or (*THEN:NAME)
2696 .sp
2697 This verb causes a skip to the next alternation in the innermost enclosing
2698 group if the rest of the pattern does not match. That is, it cancels pending
2699 backtracking, but only within the current alternation. Its name comes from the
2700 observation that it can be used for a pattern-based if-then-else block:
2701 .sp
2702 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2703 .sp
2704 If the COND1 pattern matches, FOO is tried (and possibly further items after
2705 the end of the group if FOO succeeds); on failure the matcher skips to the
2706 second alternative and tries COND2, without backtracking into COND1. The
2707 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2708 overall match fails. If (*THEN) is not directly inside an alternation, it acts
2709 like (*PRUNE).
2710 .
2711 .P
2712 The above verbs provide four different "strengths" of control when subsequent
2713 matching fails. (*THEN) is the weakest, carrying on the match at the next
2714 alternation. (*PRUNE) comes next, failing the match at the current starting
2715 position, but allowing an advance to the next character (for an unanchored
2716 pattern). (*SKIP) is similar, except that the advance may be more than one
2717 character. (*COMMIT) is the strongest, causing the entire match to fail.
2718 .P
2719 If more than one is present in a pattern, the "stongest" one wins. For example,
2720 consider this pattern, where A, B, etc. are complex pattern fragments:
2721 .sp
2722 (A(*COMMIT)B(*THEN)C|D)
2723 .sp
2724 Once A has matched, PCRE is committed to this match, at the current starting
2725 position. If subsequently B matches, but C does not, the normal (*THEN) action
2726 of trying the next alternation (that is, D) does not happen because (*COMMIT)
2727 overrides.
2728 .
2729 .
2730 .SH "SEE ALSO"
2731 .rs
2732 .sp
2733 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2734 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2735 .
2736 .
2738 .rs
2739 .sp
2740 .nf
2741 Philip Hazel
2742 University Computing Service
2743 Cambridge CB2 3QH, England.
2744 .fi
2745 .
2746 .
2748 .rs
2749 .sp
2750 .nf
2751 Last updated: 21 November 2010
2752 Copyright (c) 1997-2010 University of Cambridge.
2753 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5