/[pcre]/code/trunk/doc/pcrepattern.3
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 975 - (show annotations)
Sat Jun 2 11:03:06 2012 UTC (7 years, 2 months ago) by ph10
File size: 120904 byte(s)
Document update for 8.31-RC1 test release.
1 .TH PCREPATTERN 3 "04 May 2012" "PCRE 8.31"
2 .SH NAME
3 PCRE - Perl-compatible regular expressions
4 .SH "PCRE REGULAR EXPRESSION DETAILS"
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 strings in the original library, and a
25 second library that supports 16-bit and UTF-16 character strings. To use these
26 features, PCRE must be built to include appropriate support. When using UTF
27 strings you must either call the compiling function with the PCRE_UTF8 or
28 PCRE_UTF16 option, or the pattern must start with one of these special
29 sequences:
30 .sp
31 (*UTF8)
32 (*UTF16)
33 .sp
34 Starting a pattern with such a sequence is equivalent to setting the relevant
35 option. This feature is not Perl-compatible. How setting a UTF mode affects
36 pattern matching is mentioned in several places below. There is also a summary
37 of features in the
38 .\" HREF
39 \fBpcreunicode\fP
40 .\"
41 page.
42 .P
43 Another special sequence that may appear at the start of a pattern or in
44 combination with (*UTF8) or (*UTF16) is:
45 .sp
46 (*UCP)
47 .sp
48 This has the same effect as setting the PCRE_UCP option: it causes sequences
49 such as \ed and \ew to use Unicode properties to determine character types,
50 instead of recognizing only characters with codes less than 128 via a lookup
51 table.
52 .P
53 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
54 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
55 also some more of these special sequences that are concerned with the handling
56 of newlines; they are described below.
57 .P
58 The remainder of this document discusses the patterns that are supported by
59 PCRE when one its main matching functions, \fBpcre_exec()\fP (8-bit) or
60 \fBpcre16_exec()\fP (16-bit), is used. PCRE also has alternative matching
61 functions, \fBpcre_dfa_exec()\fP and \fBpcre16_dfa_exec()\fP, which match using
62 a different algorithm that is not Perl-compatible. Some of the features
63 discussed below are not available when DFA matching is used. The advantages and
64 disadvantages of the alternative functions, and how they differ from the normal
65 functions, are discussed in the
66 .\" HREF
67 \fBpcrematching\fP
68 .\"
69 page.
70 .
71 .
72 .\" HTML <a name="newlines"></a>
73 .SH "NEWLINE CONVENTIONS"
74 .rs
75 .sp
76 PCRE supports five different conventions for indicating line breaks in
77 strings: a single CR (carriage return) character, a single LF (linefeed)
78 character, the two-character sequence CRLF, any of the three preceding, or any
79 Unicode newline sequence. The
80 .\" HREF
81 \fBpcreapi\fP
82 .\"
83 page has
84 .\" HTML <a href="pcreapi.html#newlines">
85 .\" </a>
86 further discussion
87 .\"
88 about newlines, and shows how to set the newline convention in the
89 \fIoptions\fP arguments for the compiling and matching functions.
90 .P
91 It is also possible to specify a newline convention by starting a pattern
92 string with one of the following five sequences:
93 .sp
94 (*CR) carriage return
95 (*LF) linefeed
96 (*CRLF) carriage return, followed by linefeed
97 (*ANYCRLF) any of the three above
98 (*ANY) all Unicode newline sequences
99 .sp
100 These override the default and the options given to the compiling function. For
101 example, on a Unix system where LF is the default newline sequence, the pattern
102 .sp
103 (*CR)a.b
104 .sp
105 changes the convention to CR. That pattern matches "a\enb" because LF is no
106 longer a newline. Note that these special settings, which are not
107 Perl-compatible, are recognized only at the very start of a pattern, and that
108 they must be in upper case. If more than one of them is present, the last one
109 is used.
110 .P
111 The newline convention affects the interpretation of the dot metacharacter when
112 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
113 affect what the \eR escape sequence matches. By default, this is any Unicode
114 newline sequence, for Perl compatibility. However, this can be changed; see the
115 description of \eR in the section entitled
116 .\" HTML <a href="#newlineseq">
117 .\" </a>
118 "Newline sequences"
119 .\"
120 below. A change of \eR setting can be combined with a change of newline
121 convention.
122 .
123 .
124 .SH "CHARACTERS AND METACHARACTERS"
125 .rs
126 .sp
127 A regular expression is a pattern that is matched against a subject string from
128 left to right. Most characters stand for themselves in a pattern, and match the
129 corresponding characters in the subject. As a trivial example, the pattern
130 .sp
131 The quick brown fox
132 .sp
133 matches a portion of a subject string that is identical to itself. When
134 caseless matching is specified (the PCRE_CASELESS option), letters are matched
135 independently of case. In a UTF mode, PCRE always understands the concept of
136 case for characters whose values are less than 128, so caseless matching is
137 always possible. For characters with higher values, the concept of case is
138 supported if PCRE is compiled with Unicode property support, but not otherwise.
139 If you want to use caseless matching for characters 128 and above, you must
140 ensure that PCRE is compiled with Unicode property support as well as with
141 UTF support.
142 .P
143 The power of regular expressions comes from the ability to include alternatives
144 and repetitions in the pattern. These are encoded in the pattern by the use of
145 \fImetacharacters\fP, which do not stand for themselves but instead are
146 interpreted in some special way.
147 .P
148 There are two different sets of metacharacters: those that are recognized
149 anywhere in the pattern except within square brackets, and those that are
150 recognized within square brackets. Outside square brackets, the metacharacters
151 are as follows:
152 .sp
153 \e general escape character with several uses
154 ^ assert start of string (or line, in multiline mode)
155 $ assert end of string (or line, in multiline mode)
156 . match any character except newline (by default)
157 [ start character class definition
158 | start of alternative branch
159 ( start subpattern
160 ) end subpattern
161 ? extends the meaning of (
162 also 0 or 1 quantifier
163 also quantifier minimizer
164 * 0 or more quantifier
165 + 1 or more quantifier
166 also "possessive quantifier"
167 { start min/max quantifier
168 .sp
169 Part of a pattern that is in square brackets is called a "character class". In
170 a character class the only metacharacters are:
171 .sp
172 \e general escape character
173 ^ negate the class, but only if the first character
174 - indicates character range
175 .\" JOIN
176 [ POSIX character class (only if followed by POSIX
177 syntax)
178 ] terminates the character class
179 .sp
180 The following sections describe the use of each of the metacharacters.
181 .
182 .
183 .SH BACKSLASH
184 .rs
185 .sp
186 The backslash character has several uses. Firstly, if it is followed by a
187 character that is not a number or a letter, it takes away any special meaning
188 that character may have. This use of backslash as an escape character applies
189 both inside and outside character classes.
190 .P
191 For example, if you want to match a * character, you write \e* in the pattern.
192 This escaping action applies whether or not the following character would
193 otherwise be interpreted as a metacharacter, so it is always safe to precede a
194 non-alphanumeric with backslash to specify that it stands for itself. In
195 particular, if you want to match a backslash, you write \e\e.
196 .P
197 In a UTF mode, only ASCII numbers and letters have any special meaning after a
198 backslash. All other characters (in particular, those whose codepoints are
199 greater than 127) are treated as literals.
200 .P
201 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
202 pattern (other than in a character class) and characters between a # outside
203 a character class and the next newline are ignored. An escaping backslash can
204 be used to include a white space or # character as part of the pattern.
205 .P
206 If you want to remove the special meaning from a sequence of characters, you
207 can do so by putting them between \eQ and \eE. This is different from Perl in
208 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
209 Perl, $ and @ cause variable interpolation. Note the following examples:
210 .sp
211 Pattern PCRE matches Perl matches
212 .sp
213 .\" JOIN
214 \eQabc$xyz\eE abc$xyz abc followed by the
215 contents of $xyz
216 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
217 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
218 .sp
219 The \eQ...\eE sequence is recognized both inside and outside character classes.
220 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
221 by \eE later in the pattern, the literal interpretation continues to the end of
222 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
223 a character class, this causes an error, because the character class is not
224 terminated.
225 .
226 .
227 .\" HTML <a name="digitsafterbackslash"></a>
228 .SS "Non-printing characters"
229 .rs
230 .sp
231 A second use of backslash provides a way of encoding non-printing characters
232 in patterns in a visible manner. There is no restriction on the appearance of
233 non-printing characters, apart from the binary zero that terminates a pattern,
234 but when a pattern is being prepared by text editing, it is often easier to use
235 one of the following escape sequences than the binary character it represents:
236 .sp
237 \ea alarm, that is, the BEL character (hex 07)
238 \ecx "control-x", where x is any ASCII character
239 \ee escape (hex 1B)
240 \ef form feed (hex 0C)
241 \en linefeed (hex 0A)
242 \er carriage return (hex 0D)
243 \et tab (hex 09)
244 \eddd character with octal code ddd, or back reference
245 \exhh character with hex code hh
246 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
247 \euhhhh character with hex code hhhh (JavaScript mode only)
248 .sp
249 The precise effect of \ecx is as follows: if x is a lower case letter, it
250 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
251 Thus \ecz becomes hex 1A (z is 7A), but \ec{ becomes hex 3B ({ is 7B), while
252 \ec; becomes hex 7B (; is 3B). If the byte following \ec has a value greater
253 than 127, a compile-time error occurs. This locks out non-ASCII characters in
254 all modes. (When PCRE is compiled in EBCDIC mode, all byte values are valid. A
255 lower case letter is converted to upper case, and then the 0xc0 bits are
256 flipped.)
257 .P
258 By default, after \ex, from zero to two hexadecimal digits are read (letters
259 can be in upper or lower case). Any number of hexadecimal digits may appear
260 between \ex{ and }, but the character code is constrained as follows:
261 .sp
262 8-bit non-UTF mode less than 0x100
263 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
264 16-bit non-UTF mode less than 0x10000
265 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
266 .sp
267 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
268 "surrogate" codepoints).
269 .P
270 If characters other than hexadecimal digits appear between \ex{ and }, or if
271 there is no terminating }, this form of escape is not recognized. Instead, the
272 initial \ex will be interpreted as a basic hexadecimal escape, with no
273 following digits, giving a character whose value is zero.
274 .P
275 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
276 as just described only when it is followed by two hexadecimal digits.
277 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
278 code points greater than 256 is provided by \eu, which must be followed by
279 four hexadecimal digits; otherwise it matches a literal "u" character.
280 .P
281 Characters whose value is less than 256 can be defined by either of the two
282 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
283 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
284 \eu00dc in JavaScript mode).
285 .P
286 After \e0 up to two further octal digits are read. If there are fewer than two
287 digits, just those that are present are used. Thus the sequence \e0\ex\e07
288 specifies two binary zeros followed by a BEL character (code value 7). Make
289 sure you supply two digits after the initial zero if the pattern character that
290 follows is itself an octal digit.
291 .P
292 The handling of a backslash followed by a digit other than 0 is complicated.
293 Outside a character class, PCRE reads it and any following digits as a decimal
294 number. If the number is less than 10, or if there have been at least that many
295 previous capturing left parentheses in the expression, the entire sequence is
296 taken as a \fIback reference\fP. A description of how this works is given
297 .\" HTML <a href="#backreferences">
298 .\" </a>
299 later,
300 .\"
301 following the discussion of
302 .\" HTML <a href="#subpattern">
303 .\" </a>
304 parenthesized subpatterns.
305 .\"
306 .P
307 Inside a character class, or if the decimal number is greater than 9 and there
308 have not been that many capturing subpatterns, PCRE re-reads up to three octal
309 digits following the backslash, and uses them to generate a data character. Any
310 subsequent digits stand for themselves. The value of the character is
311 constrained in the same way as characters specified in hexadecimal.
312 For example:
313 .sp
314 \e040 is another way of writing a space
315 .\" JOIN
316 \e40 is the same, provided there are fewer than 40
317 previous capturing subpatterns
318 \e7 is always a back reference
319 .\" JOIN
320 \e11 might be a back reference, or another way of
321 writing a tab
322 \e011 is always a tab
323 \e0113 is a tab followed by the character "3"
324 .\" JOIN
325 \e113 might be a back reference, otherwise the
326 character with octal code 113
327 .\" JOIN
328 \e377 might be a back reference, otherwise
329 the value 255 (decimal)
330 .\" JOIN
331 \e81 is either a back reference, or a binary zero
332 followed by the two characters "8" and "1"
333 .sp
334 Note that octal values of 100 or greater must not be introduced by a leading
335 zero, because no more than three octal digits are ever read.
336 .P
337 All the sequences that define a single character value can be used both inside
338 and outside character classes. In addition, inside a character class, \eb is
339 interpreted as the backspace character (hex 08).
340 .P
341 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
342 inside a character class. Like other unrecognized escape sequences, they are
343 treated as the literal characters "B", "R", and "X" by default, but cause an
344 error if the PCRE_EXTRA option is set. Outside a character class, these
345 sequences have different meanings.
346 .
347 .
348 .SS "Unsupported escape sequences"
349 .rs
350 .sp
351 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
352 handler and used to modify the case of following characters. By default, PCRE
353 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
354 option is set, \eU matches a "U" character, and \eu can be used to define a
355 character by code point, as described in the previous section.
356 .
357 .
358 .SS "Absolute and relative back references"
359 .rs
360 .sp
361 The sequence \eg followed by an unsigned or a negative number, optionally
362 enclosed in braces, is an absolute or relative back reference. A named back
363 reference can be coded as \eg{name}. Back references are discussed
364 .\" HTML <a href="#backreferences">
365 .\" </a>
366 later,
367 .\"
368 following the discussion of
369 .\" HTML <a href="#subpattern">
370 .\" </a>
371 parenthesized subpatterns.
372 .\"
373 .
374 .
375 .SS "Absolute and relative subroutine calls"
376 .rs
377 .sp
378 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
379 a number enclosed either in angle brackets or single quotes, is an alternative
380 syntax for referencing a subpattern as a "subroutine". Details are discussed
381 .\" HTML <a href="#onigurumasubroutines">
382 .\" </a>
383 later.
384 .\"
385 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
386 synonymous. The former is a back reference; the latter is a
387 .\" HTML <a href="#subpatternsassubroutines">
388 .\" </a>
389 subroutine
390 .\"
391 call.
392 .
393 .
394 .\" HTML <a name="genericchartypes"></a>
395 .SS "Generic character types"
396 .rs
397 .sp
398 Another use of backslash is for specifying generic character types:
399 .sp
400 \ed any decimal digit
401 \eD any character that is not a decimal digit
402 \eh any horizontal white space character
403 \eH any character that is not a horizontal white space character
404 \es any white space character
405 \eS any character that is not a white space character
406 \ev any vertical white space character
407 \eV any character that is not a vertical white space character
408 \ew any "word" character
409 \eW any "non-word" character
410 .sp
411 There is also the single sequence \eN, which matches a non-newline character.
412 This is the same as
413 .\" HTML <a href="#fullstopdot">
414 .\" </a>
415 the "." metacharacter
416 .\"
417 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
418 PCRE does not support this.
419 .P
420 Each pair of lower and upper case escape sequences partitions the complete set
421 of characters into two disjoint sets. Any given character matches one, and only
422 one, of each pair. The sequences can appear both inside and outside character
423 classes. They each match one character of the appropriate type. If the current
424 matching point is at the end of the subject string, all of them fail, because
425 there is no character to match.
426 .P
427 For compatibility with Perl, \es does not match the VT character (code 11).
428 This makes it different from the the POSIX "space" class. The \es characters
429 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
430 included in a Perl script, \es may match the VT character. In PCRE, it never
431 does.
432 .P
433 A "word" character is an underscore or any character that is a letter or digit.
434 By default, the definition of letters and digits is controlled by PCRE's
435 low-valued character tables, and may vary if locale-specific matching is taking
436 place (see
437 .\" HTML <a href="pcreapi.html#localesupport">
438 .\" </a>
439 "Locale support"
440 .\"
441 in the
442 .\" HREF
443 \fBpcreapi\fP
444 .\"
445 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
446 or "french" in Windows, some character codes greater than 128 are used for
447 accented letters, and these are then matched by \ew. The use of locales with
448 Unicode is discouraged.
449 .P
450 By default, in a UTF mode, characters with values greater than 128 never match
451 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
452 their original meanings from before UTF support was available, mainly for
453 efficiency reasons. However, if PCRE is compiled with Unicode property support,
454 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
455 properties are used to determine character types, as follows:
456 .sp
457 \ed any character that \ep{Nd} matches (decimal digit)
458 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
459 \ew any character that \ep{L} or \ep{N} matches, plus underscore
460 .sp
461 The upper case escapes match the inverse sets of characters. Note that \ed
462 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
463 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
464 \eB because they are defined in terms of \ew and \eW. Matching these sequences
465 is noticeably slower when PCRE_UCP is set.
466 .P
467 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
468 release 5.10. In contrast to the other sequences, which match only ASCII
469 characters by default, these always match certain high-valued codepoints,
470 whether or not PCRE_UCP is set. The horizontal space characters are:
471 .sp
472 U+0009 Horizontal tab
473 U+0020 Space
474 U+00A0 Non-break space
475 U+1680 Ogham space mark
476 U+180E Mongolian vowel separator
477 U+2000 En quad
478 U+2001 Em quad
479 U+2002 En space
480 U+2003 Em space
481 U+2004 Three-per-em space
482 U+2005 Four-per-em space
483 U+2006 Six-per-em space
484 U+2007 Figure space
485 U+2008 Punctuation space
486 U+2009 Thin space
487 U+200A Hair space
488 U+202F Narrow no-break space
489 U+205F Medium mathematical space
490 U+3000 Ideographic space
491 .sp
492 The vertical space characters are:
493 .sp
494 U+000A Linefeed
495 U+000B Vertical tab
496 U+000C Form feed
497 U+000D Carriage return
498 U+0085 Next line
499 U+2028 Line separator
500 U+2029 Paragraph separator
501 .sp
502 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
503 relevant.
504 .
505 .
506 .\" HTML <a name="newlineseq"></a>
507 .SS "Newline sequences"
508 .rs
509 .sp
510 Outside a character class, by default, the escape sequence \eR matches any
511 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
512 following:
513 .sp
514 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
515 .sp
516 This is an example of an "atomic group", details of which are given
517 .\" HTML <a href="#atomicgroup">
518 .\" </a>
519 below.
520 .\"
521 This particular group matches either the two-character sequence CR followed by
522 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
523 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
524 line, U+0085). The two-character sequence is treated as a single unit that
525 cannot be split.
526 .P
527 In other modes, two additional characters whose codepoints are greater than 255
528 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
529 Unicode character property support is not needed for these characters to be
530 recognized.
531 .P
532 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
533 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
534 either at compile time or when the pattern is matched. (BSR is an abbrevation
535 for "backslash R".) This can be made the default when PCRE is built; if this is
536 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
537 It is also possible to specify these settings by starting a pattern string with
538 one of the following sequences:
539 .sp
540 (*BSR_ANYCRLF) CR, LF, or CRLF only
541 (*BSR_UNICODE) any Unicode newline sequence
542 .sp
543 These override the default and the options given to the compiling function, but
544 they can themselves be overridden by options given to a matching function. Note
545 that these special settings, which are not Perl-compatible, are recognized only
546 at the very start of a pattern, and that they must be in upper case. If more
547 than one of them is present, the last one is used. They can be combined with a
548 change of newline convention; for example, a pattern can start with:
549 .sp
550 (*ANY)(*BSR_ANYCRLF)
551 .sp
552 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
553 sequences. Inside a character class, \eR is treated as an unrecognized escape
554 sequence, and so matches the letter "R" by default, but causes an error if
555 PCRE_EXTRA is set.
556 .
557 .
558 .\" HTML <a name="uniextseq"></a>
559 .SS Unicode character properties
560 .rs
561 .sp
562 When PCRE is built with Unicode character property support, three additional
563 escape sequences that match characters with specific properties are available.
564 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
565 characters whose codepoints are less than 256, but they do work in this mode.
566 The extra escape sequences are:
567 .sp
568 \ep{\fIxx\fP} a character with the \fIxx\fP property
569 \eP{\fIxx\fP} a character without the \fIxx\fP property
570 \eX an extended Unicode sequence
571 .sp
572 The property names represented by \fIxx\fP above are limited to the Unicode
573 script names, the general category properties, "Any", which matches any
574 character (including newline), and some special PCRE properties (described
575 in the
576 .\" HTML <a href="#extraprops">
577 .\" </a>
578 next section).
579 .\"
580 Other Perl properties such as "InMusicalSymbols" are not currently supported by
581 PCRE. Note that \eP{Any} does not match any characters, so always causes a
582 match failure.
583 .P
584 Sets of Unicode characters are defined as belonging to certain scripts. A
585 character from one of these sets can be matched using a script name. For
586 example:
587 .sp
588 \ep{Greek}
589 \eP{Han}
590 .sp
591 Those that are not part of an identified script are lumped together as
592 "Common". The current list of scripts is:
593 .P
594 Arabic,
595 Armenian,
596 Avestan,
597 Balinese,
598 Bamum,
599 Batak,
600 Bengali,
601 Bopomofo,
602 Brahmi,
603 Braille,
604 Buginese,
605 Buhid,
606 Canadian_Aboriginal,
607 Carian,
608 Chakma,
609 Cham,
610 Cherokee,
611 Common,
612 Coptic,
613 Cuneiform,
614 Cypriot,
615 Cyrillic,
616 Deseret,
617 Devanagari,
618 Egyptian_Hieroglyphs,
619 Ethiopic,
620 Georgian,
621 Glagolitic,
622 Gothic,
623 Greek,
624 Gujarati,
625 Gurmukhi,
626 Han,
627 Hangul,
628 Hanunoo,
629 Hebrew,
630 Hiragana,
631 Imperial_Aramaic,
632 Inherited,
633 Inscriptional_Pahlavi,
634 Inscriptional_Parthian,
635 Javanese,
636 Kaithi,
637 Kannada,
638 Katakana,
639 Kayah_Li,
640 Kharoshthi,
641 Khmer,
642 Lao,
643 Latin,
644 Lepcha,
645 Limbu,
646 Linear_B,
647 Lisu,
648 Lycian,
649 Lydian,
650 Malayalam,
651 Mandaic,
652 Meetei_Mayek,
653 Meroitic_Cursive,
654 Meroitic_Hieroglyphs,
655 Miao,
656 Mongolian,
657 Myanmar,
658 New_Tai_Lue,
659 Nko,
660 Ogham,
661 Old_Italic,
662 Old_Persian,
663 Old_South_Arabian,
664 Old_Turkic,
665 Ol_Chiki,
666 Oriya,
667 Osmanya,
668 Phags_Pa,
669 Phoenician,
670 Rejang,
671 Runic,
672 Samaritan,
673 Saurashtra,
674 Sharada,
675 Shavian,
676 Sinhala,
677 Sora_Sompeng,
678 Sundanese,
679 Syloti_Nagri,
680 Syriac,
681 Tagalog,
682 Tagbanwa,
683 Tai_Le,
684 Tai_Tham,
685 Tai_Viet,
686 Takri,
687 Tamil,
688 Telugu,
689 Thaana,
690 Thai,
691 Tibetan,
692 Tifinagh,
693 Ugaritic,
694 Vai,
695 Yi.
696 .P
697 Each character has exactly one Unicode general category property, specified by
698 a two-letter abbreviation. For compatibility with Perl, negation can be
699 specified by including a circumflex between the opening brace and the property
700 name. For example, \ep{^Lu} is the same as \eP{Lu}.
701 .P
702 If only one letter is specified with \ep or \eP, it includes all the general
703 category properties that start with that letter. In this case, in the absence
704 of negation, the curly brackets in the escape sequence are optional; these two
705 examples have the same effect:
706 .sp
707 \ep{L}
708 \epL
709 .sp
710 The following general category property codes are supported:
711 .sp
712 C Other
713 Cc Control
714 Cf Format
715 Cn Unassigned
716 Co Private use
717 Cs Surrogate
718 .sp
719 L Letter
720 Ll Lower case letter
721 Lm Modifier letter
722 Lo Other letter
723 Lt Title case letter
724 Lu Upper case letter
725 .sp
726 M Mark
727 Mc Spacing mark
728 Me Enclosing mark
729 Mn Non-spacing mark
730 .sp
731 N Number
732 Nd Decimal number
733 Nl Letter number
734 No Other number
735 .sp
736 P Punctuation
737 Pc Connector punctuation
738 Pd Dash punctuation
739 Pe Close punctuation
740 Pf Final punctuation
741 Pi Initial punctuation
742 Po Other punctuation
743 Ps Open punctuation
744 .sp
745 S Symbol
746 Sc Currency symbol
747 Sk Modifier symbol
748 Sm Mathematical symbol
749 So Other symbol
750 .sp
751 Z Separator
752 Zl Line separator
753 Zp Paragraph separator
754 Zs Space separator
755 .sp
756 The special property L& is also supported: it matches a character that has
757 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
758 a modifier or "other".
759 .P
760 The Cs (Surrogate) property applies only to characters in the range U+D800 to
761 U+DFFF. Such characters are not valid in Unicode strings and so
762 cannot be tested by PCRE, unless UTF validity checking has been turned off
763 (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
764 .\" HREF
765 \fBpcreapi\fP
766 .\"
767 page). Perl does not support the Cs property.
768 .P
769 The long synonyms for property names that Perl supports (such as \ep{Letter})
770 are not supported by PCRE, nor is it permitted to prefix any of these
771 properties with "Is".
772 .P
773 No character that is in the Unicode table has the Cn (unassigned) property.
774 Instead, this property is assumed for any code point that is not in the
775 Unicode table.
776 .P
777 Specifying caseless matching does not affect these escape sequences. For
778 example, \ep{Lu} always matches only upper case letters.
779 .P
780 The \eX escape matches any number of Unicode characters that form an extended
781 Unicode sequence. \eX is equivalent to
782 .sp
783 (?>\ePM\epM*)
784 .sp
785 That is, it matches a character without the "mark" property, followed by zero
786 or more characters with the "mark" property, and treats the sequence as an
787 atomic group
788 .\" HTML <a href="#atomicgroup">
789 .\" </a>
790 (see below).
791 .\"
792 Characters with the "mark" property are typically accents that affect the
793 preceding character. None of them have codepoints less than 256, so in
794 8-bit non-UTF-8 mode \eX matches any one character.
795 .P
796 Note that recent versions of Perl have changed \eX to match what Unicode calls
797 an "extended grapheme cluster", which has a more complicated definition.
798 .P
799 Matching characters by Unicode property is not fast, because PCRE has to search
800 a structure that contains data for over fifteen thousand characters. That is
801 why the traditional escape sequences such as \ed and \ew do not use Unicode
802 properties in PCRE by default, though you can make them do so by setting the
803 PCRE_UCP option or by starting the pattern with (*UCP).
804 .
805 .
806 .\" HTML <a name="extraprops"></a>
807 .SS PCRE's additional properties
808 .rs
809 .sp
810 As well as the standard Unicode properties described in the previous
811 section, PCRE supports four more that make it possible to convert traditional
812 escape sequences such as \ew and \es and POSIX character classes to use Unicode
813 properties. PCRE uses these non-standard, non-Perl properties internally when
814 PCRE_UCP is set. They are:
815 .sp
816 Xan Any alphanumeric character
817 Xps Any POSIX space character
818 Xsp Any Perl space character
819 Xwd Any Perl "word" character
820 .sp
821 Xan matches characters that have either the L (letter) or the N (number)
822 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
823 carriage return, and any other character that has the Z (separator) property.
824 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
825 same characters as Xan, plus underscore.
826 .
827 .
828 .\" HTML <a name="resetmatchstart"></a>
829 .SS "Resetting the match start"
830 .rs
831 .sp
832 The escape sequence \eK causes any previously matched characters not to be
833 included in the final matched sequence. For example, the pattern:
834 .sp
835 foo\eKbar
836 .sp
837 matches "foobar", but reports that it has matched "bar". This feature is
838 similar to a lookbehind assertion
839 .\" HTML <a href="#lookbehind">
840 .\" </a>
841 (described below).
842 .\"
843 However, in this case, the part of the subject before the real match does not
844 have to be of fixed length, as lookbehind assertions do. The use of \eK does
845 not interfere with the setting of
846 .\" HTML <a href="#subpattern">
847 .\" </a>
848 captured substrings.
849 .\"
850 For example, when the pattern
851 .sp
852 (foo)\eKbar
853 .sp
854 matches "foobar", the first substring is still set to "foo".
855 .P
856 Perl documents that the use of \eK within assertions is "not well defined". In
857 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
858 ignored in negative assertions.
859 .
860 .
861 .\" HTML <a name="smallassertions"></a>
862 .SS "Simple assertions"
863 .rs
864 .sp
865 The final use of backslash is for certain simple assertions. An assertion
866 specifies a condition that has to be met at a particular point in a match,
867 without consuming any characters from the subject string. The use of
868 subpatterns for more complicated assertions is described
869 .\" HTML <a href="#bigassertions">
870 .\" </a>
871 below.
872 .\"
873 The backslashed assertions are:
874 .sp
875 \eb matches at a word boundary
876 \eB matches when not at a word boundary
877 \eA matches at the start of the subject
878 \eZ matches at the end of the subject
879 also matches before a newline at the end of the subject
880 \ez matches only at the end of the subject
881 \eG matches at the first matching position in the subject
882 .sp
883 Inside a character class, \eb has a different meaning; it matches the backspace
884 character. If any other of these assertions appears in a character class, by
885 default it matches the corresponding literal character (for example, \eB
886 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
887 escape sequence" error is generated instead.
888 .P
889 A word boundary is a position in the subject string where the current character
890 and the previous character do not both match \ew or \eW (i.e. one matches
891 \ew and the other matches \eW), or the start or end of the string if the
892 first or last character matches \ew, respectively. In a UTF mode, the meanings
893 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
894 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
895 of word" or "end of word" metasequence. However, whatever follows \eb normally
896 determines which it is. For example, the fragment \eba matches "a" at the start
897 of a word.
898 .P
899 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
900 dollar (described in the next section) in that they only ever match at the very
901 start and end of the subject string, whatever options are set. Thus, they are
902 independent of multiline mode. These three assertions are not affected by the
903 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
904 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
905 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
906 at a point other than the beginning of the subject, \eA can never match. The
907 difference between \eZ and \ez is that \eZ matches before a newline at the end
908 of the string as well as at the very end, whereas \ez matches only at the end.
909 .P
910 The \eG assertion is true only when the current matching position is at the
911 start point of the match, as specified by the \fIstartoffset\fP argument of
912 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
913 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
914 arguments, you can mimic Perl's /g option, and it is in this kind of
915 implementation where \eG can be useful.
916 .P
917 Note, however, that PCRE's interpretation of \eG, as the start of the current
918 match, is subtly different from Perl's, which defines it as the end of the
919 previous match. In Perl, these can be different when the previously matched
920 string was empty. Because PCRE does just one match at a time, it cannot
921 reproduce this behaviour.
922 .P
923 If all the alternatives of a pattern begin with \eG, the expression is anchored
924 to the starting match position, and the "anchored" flag is set in the compiled
925 regular expression.
926 .
927 .
928 .SH "CIRCUMFLEX AND DOLLAR"
929 .rs
930 .sp
931 Outside a character class, in the default matching mode, the circumflex
932 character is an assertion that is true only if the current matching point is
933 at the start of the subject string. If the \fIstartoffset\fP argument of
934 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
935 option is unset. Inside a character class, circumflex has an entirely different
936 meaning
937 .\" HTML <a href="#characterclass">
938 .\" </a>
939 (see below).
940 .\"
941 .P
942 Circumflex need not be the first character of the pattern if a number of
943 alternatives are involved, but it should be the first thing in each alternative
944 in which it appears if the pattern is ever to match that branch. If all
945 possible alternatives start with a circumflex, that is, if the pattern is
946 constrained to match only at the start of the subject, it is said to be an
947 "anchored" pattern. (There are also other constructs that can cause a pattern
948 to be anchored.)
949 .P
950 A dollar character is an assertion that is true only if the current matching
951 point is at the end of the subject string, or immediately before a newline
952 at the end of the string (by default). Dollar need not be the last character of
953 the pattern if a number of alternatives are involved, but it should be the last
954 item in any branch in which it appears. Dollar has no special meaning in a
955 character class.
956 .P
957 The meaning of dollar can be changed so that it matches only at the very end of
958 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
959 does not affect the \eZ assertion.
960 .P
961 The meanings of the circumflex and dollar characters are changed if the
962 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
963 immediately after internal newlines as well as at the start of the subject
964 string. It does not match after a newline that ends the string. A dollar
965 matches before any newlines in the string, as well as at the very end, when
966 PCRE_MULTILINE is set. When newline is specified as the two-character
967 sequence CRLF, isolated CR and LF characters do not indicate newlines.
968 .P
969 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
970 \en represents a newline) in multiline mode, but not otherwise. Consequently,
971 patterns that are anchored in single line mode because all branches start with
972 ^ are not anchored in multiline mode, and a match for circumflex is possible
973 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
974 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
975 .P
976 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
977 end of the subject in both modes, and if all branches of a pattern start with
978 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
979 .
980 .
981 .\" HTML <a name="fullstopdot"></a>
982 .SH "FULL STOP (PERIOD, DOT) AND \eN"
983 .rs
984 .sp
985 Outside a character class, a dot in the pattern matches any one character in
986 the subject string except (by default) a character that signifies the end of a
987 line.
988 .P
989 When a line ending is defined as a single character, dot never matches that
990 character; when the two-character sequence CRLF is used, dot does not match CR
991 if it is immediately followed by LF, but otherwise it matches all characters
992 (including isolated CRs and LFs). When any Unicode line endings are being
993 recognized, dot does not match CR or LF or any of the other line ending
994 characters.
995 .P
996 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
997 option is set, a dot matches any one character, without exception. If the
998 two-character sequence CRLF is present in the subject string, it takes two dots
999 to match it.
1000 .P
1001 The handling of dot is entirely independent of the handling of circumflex and
1002 dollar, the only relationship being that they both involve newlines. Dot has no
1003 special meaning in a character class.
1004 .P
1005 The escape sequence \eN behaves like a dot, except that it is not affected by
1006 the PCRE_DOTALL option. In other words, it matches any character except one
1007 that signifies the end of a line. Perl also uses \eN to match characters by
1008 name; PCRE does not support this.
1009 .
1010 .
1011 .SH "MATCHING A SINGLE DATA UNIT"
1012 .rs
1013 .sp
1014 Outside a character class, the escape sequence \eC matches any one data unit,
1015 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1016 byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \eC always
1017 matches line-ending characters. The feature is provided in Perl in order to
1018 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1019 used. Because \eC breaks up characters into individual data units, matching one
1020 unit with \eC in a UTF mode means that the rest of the string may start with a
1021 malformed UTF character. This has undefined results, because PCRE assumes that
1022 it is dealing with valid UTF strings (and by default it checks this at the
1023 start of processing unless the PCRE_NO_UTF8_CHECK or PCRE_NO_UTF16_CHECK option
1024 is used).
1025 .P
1026 PCRE does not allow \eC to appear in lookbehind assertions
1027 .\" HTML <a href="#lookbehind">
1028 .\" </a>
1029 (described below)
1030 .\"
1031 in a UTF mode, because this would make it impossible to calculate the length of
1032 the lookbehind.
1033 .P
1034 In general, the \eC escape sequence is best avoided. However, one
1035 way of using it that avoids the problem of malformed UTF characters is to use a
1036 lookahead to check the length of the next character, as in this pattern, which
1037 could be used with a UTF-8 string (ignore white space and line breaks):
1038 .sp
1039 (?| (?=[\ex00-\ex7f])(\eC) |
1040 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1041 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1042 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1043 .sp
1044 A group that starts with (?| resets the capturing parentheses numbers in each
1045 alternative (see
1046 .\" HTML <a href="#dupsubpatternnumber">
1047 .\" </a>
1048 "Duplicate Subpattern Numbers"
1049 .\"
1050 below). The assertions at the start of each branch check the next UTF-8
1051 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1052 character's individual bytes are then captured by the appropriate number of
1053 groups.
1054 .
1055 .
1056 .\" HTML <a name="characterclass"></a>
1057 .SH "SQUARE BRACKETS AND CHARACTER CLASSES"
1058 .rs
1059 .sp
1060 An opening square bracket introduces a character class, terminated by a closing
1061 square bracket. A closing square bracket on its own is not special by default.
1062 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1063 bracket causes a compile-time error. If a closing square bracket is required as
1064 a member of the class, it should be the first data character in the class
1065 (after an initial circumflex, if present) or escaped with a backslash.
1066 .P
1067 A character class matches a single character in the subject. In a UTF mode, the
1068 character may be more than one data unit long. A matched character must be in
1069 the set of characters defined by the class, unless the first character in the
1070 class definition is a circumflex, in which case the subject character must not
1071 be in the set defined by the class. If a circumflex is actually required as a
1072 member of the class, ensure it is not the first character, or escape it with a
1073 backslash.
1074 .P
1075 For example, the character class [aeiou] matches any lower case vowel, while
1076 [^aeiou] matches any character that is not a lower case vowel. Note that a
1077 circumflex is just a convenient notation for specifying the characters that
1078 are in the class by enumerating those that are not. A class that starts with a
1079 circumflex is not an assertion; it still consumes a character from the subject
1080 string, and therefore it fails if the current pointer is at the end of the
1081 string.
1082 .P
1083 In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
1084 included in a class as a literal string of data units, or by using the \ex{
1085 escaping mechanism.
1086 .P
1087 When caseless matching is set, any letters in a class represent both their
1088 upper case and lower case versions, so for example, a caseless [aeiou] matches
1089 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1090 caseful version would. In a UTF mode, PCRE always understands the concept of
1091 case for characters whose values are less than 128, so caseless matching is
1092 always possible. For characters with higher values, the concept of case is
1093 supported if PCRE is compiled with Unicode property support, but not otherwise.
1094 If you want to use caseless matching in a UTF mode for characters 128 and
1095 above, you must ensure that PCRE is compiled with Unicode property support as
1096 well as with UTF support.
1097 .P
1098 Characters that might indicate line breaks are never treated in any special way
1099 when matching character classes, whatever line-ending sequence is in use, and
1100 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1101 such as [^a] always matches one of these characters.
1102 .P
1103 The minus (hyphen) character can be used to specify a range of characters in a
1104 character class. For example, [d-m] matches any letter between d and m,
1105 inclusive. If a minus character is required in a class, it must be escaped with
1106 a backslash or appear in a position where it cannot be interpreted as
1107 indicating a range, typically as the first or last character in the class.
1108 .P
1109 It is not possible to have the literal character "]" as the end character of a
1110 range. A pattern such as [W-]46] is interpreted as a class of two characters
1111 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1112 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1113 the end of range, so [W-\e]46] is interpreted as a class containing a range
1114 followed by two other characters. The octal or hexadecimal representation of
1115 "]" can also be used to end a range.
1116 .P
1117 Ranges operate in the collating sequence of character values. They can also be
1118 used for characters specified numerically, for example [\e000-\e037]. Ranges
1119 can include any characters that are valid for the current mode.
1120 .P
1121 If a range that includes letters is used when caseless matching is set, it
1122 matches the letters in either case. For example, [W-c] is equivalent to
1123 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1124 tables for a French locale are in use, [\exc8-\excb] matches accented E
1125 characters in both cases. In UTF modes, PCRE supports the concept of case for
1126 characters with values greater than 128 only when it is compiled with Unicode
1127 property support.
1128 .P
1129 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1130 \eV, \ew, and \eW may appear in a character class, and add the characters that
1131 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1132 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1133 and their upper case partners, just as it does when they appear outside a
1134 character class, as described in the section entitled
1135 .\" HTML <a href="#genericchartypes">
1136 .\" </a>
1137 "Generic character types"
1138 .\"
1139 above. The escape sequence \eb has a different meaning inside a character
1140 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1141 are not special inside a character class. Like any other unrecognized escape
1142 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1143 default, but cause an error if the PCRE_EXTRA option is set.
1144 .P
1145 A circumflex can conveniently be used with the upper case character types to
1146 specify a more restricted set of characters than the matching lower case type.
1147 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1148 whereas [\ew] includes underscore. A positive character class should be read as
1149 "something OR something OR ..." and a negative class as "NOT something AND NOT
1150 something AND NOT ...".
1151 .P
1152 The only metacharacters that are recognized in character classes are backslash,
1153 hyphen (only where it can be interpreted as specifying a range), circumflex
1154 (only at the start), opening square bracket (only when it can be interpreted as
1155 introducing a POSIX class name - see the next section), and the terminating
1156 closing square bracket. However, escaping other non-alphanumeric characters
1157 does no harm.
1158 .
1159 .
1160 .SH "POSIX CHARACTER CLASSES"
1161 .rs
1162 .sp
1163 Perl supports the POSIX notation for character classes. This uses names
1164 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1165 this notation. For example,
1166 .sp
1167 [01[:alpha:]%]
1168 .sp
1169 matches "0", "1", any alphabetic character, or "%". The supported class names
1170 are:
1171 .sp
1172 alnum letters and digits
1173 alpha letters
1174 ascii character codes 0 - 127
1175 blank space or tab only
1176 cntrl control characters
1177 digit decimal digits (same as \ed)
1178 graph printing characters, excluding space
1179 lower lower case letters
1180 print printing characters, including space
1181 punct printing characters, excluding letters and digits and space
1182 space white space (not quite the same as \es)
1183 upper upper case letters
1184 word "word" characters (same as \ew)
1185 xdigit hexadecimal digits
1186 .sp
1187 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1188 space (32). Notice that this list includes the VT character (code 11). This
1189 makes "space" different to \es, which does not include VT (for Perl
1190 compatibility).
1191 .P
1192 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1193 5.8. Another Perl extension is negation, which is indicated by a ^ character
1194 after the colon. For example,
1195 .sp
1196 [12[:^digit:]]
1197 .sp
1198 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1199 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1200 supported, and an error is given if they are encountered.
1201 .P
1202 By default, in UTF modes, characters with values greater than 128 do not match
1203 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1204 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1205 character properties are used. This is achieved by replacing the POSIX classes
1206 by other sequences, as follows:
1207 .sp
1208 [:alnum:] becomes \ep{Xan}
1209 [:alpha:] becomes \ep{L}
1210 [:blank:] becomes \eh
1211 [:digit:] becomes \ep{Nd}
1212 [:lower:] becomes \ep{Ll}
1213 [:space:] becomes \ep{Xps}
1214 [:upper:] becomes \ep{Lu}
1215 [:word:] becomes \ep{Xwd}
1216 .sp
1217 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1218 classes are unchanged, and match only characters with code points less than
1219 128.
1220 .
1221 .
1222 .SH "VERTICAL BAR"
1223 .rs
1224 .sp
1225 Vertical bar characters are used to separate alternative patterns. For example,
1226 the pattern
1227 .sp
1228 gilbert|sullivan
1229 .sp
1230 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1231 and an empty alternative is permitted (matching the empty string). The matching
1232 process tries each alternative in turn, from left to right, and the first one
1233 that succeeds is used. If the alternatives are within a subpattern
1234 .\" HTML <a href="#subpattern">
1235 .\" </a>
1236 (defined below),
1237 .\"
1238 "succeeds" means matching the rest of the main pattern as well as the
1239 alternative in the subpattern.
1240 .
1241 .
1242 .SH "INTERNAL OPTION SETTING"
1243 .rs
1244 .sp
1245 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1246 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1247 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1248 The option letters are
1249 .sp
1250 i for PCRE_CASELESS
1251 m for PCRE_MULTILINE
1252 s for PCRE_DOTALL
1253 x for PCRE_EXTENDED
1254 .sp
1255 For example, (?im) sets caseless, multiline matching. It is also possible to
1256 unset these options by preceding the letter with a hyphen, and a combined
1257 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1258 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1259 permitted. If a letter appears both before and after the hyphen, the option is
1260 unset.
1261 .P
1262 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1263 changed in the same way as the Perl-compatible options by using the characters
1264 J, U and X respectively.
1265 .P
1266 When one of these option changes occurs at top level (that is, not inside
1267 subpattern parentheses), the change applies to the remainder of the pattern
1268 that follows. If the change is placed right at the start of a pattern, PCRE
1269 extracts it into the global options (and it will therefore show up in data
1270 extracted by the \fBpcre_fullinfo()\fP function).
1271 .P
1272 An option change within a subpattern (see below for a description of
1273 subpatterns) affects only that part of the subpattern that follows it, so
1274 .sp
1275 (a(?i)b)c
1276 .sp
1277 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1278 By this means, options can be made to have different settings in different
1279 parts of the pattern. Any changes made in one alternative do carry on
1280 into subsequent branches within the same subpattern. For example,
1281 .sp
1282 (a(?i)b|c)
1283 .sp
1284 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1285 branch is abandoned before the option setting. This is because the effects of
1286 option settings happen at compile time. There would be some very weird
1287 behaviour otherwise.
1288 .P
1289 \fBNote:\fP There are other PCRE-specific options that can be set by the
1290 application when the compiling or matching functions are called. In some cases
1291 the pattern can contain special leading sequences such as (*CRLF) to override
1292 what the application has set or what has been defaulted. Details are given in
1293 the section entitled
1294 .\" HTML <a href="#newlineseq">
1295 .\" </a>
1296 "Newline sequences"
1297 .\"
1298 above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
1299 can be used to set UTF and Unicode property modes; they are equivalent to
1300 setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1301 .
1302 .
1303 .\" HTML <a name="subpattern"></a>
1304 .SH SUBPATTERNS
1305 .rs
1306 .sp
1307 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1308 Turning part of a pattern into a subpattern does two things:
1309 .sp
1310 1. It localizes a set of alternatives. For example, the pattern
1311 .sp
1312 cat(aract|erpillar|)
1313 .sp
1314 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1315 match "cataract", "erpillar" or an empty string.
1316 .sp
1317 2. It sets up the subpattern as a capturing subpattern. This means that, when
1318 the whole pattern matches, that portion of the subject string that matched the
1319 subpattern is passed back to the caller via the \fIovector\fP argument of the
1320 matching function. (This applies only to the traditional matching functions;
1321 the DFA matching functions do not support capturing.)
1322 .P
1323 Opening parentheses are counted from left to right (starting from 1) to obtain
1324 numbers for the capturing subpatterns. For example, if the string "the red
1325 king" is matched against the pattern
1326 .sp
1327 the ((red|white) (king|queen))
1328 .sp
1329 the captured substrings are "red king", "red", and "king", and are numbered 1,
1330 2, and 3, respectively.
1331 .P
1332 The fact that plain parentheses fulfil two functions is not always helpful.
1333 There are often times when a grouping subpattern is required without a
1334 capturing requirement. If an opening parenthesis is followed by a question mark
1335 and a colon, the subpattern does not do any capturing, and is not counted when
1336 computing the number of any subsequent capturing subpatterns. For example, if
1337 the string "the white queen" is matched against the pattern
1338 .sp
1339 the ((?:red|white) (king|queen))
1340 .sp
1341 the captured substrings are "white queen" and "queen", and are numbered 1 and
1342 2. The maximum number of capturing subpatterns is 65535.
1343 .P
1344 As a convenient shorthand, if any option settings are required at the start of
1345 a non-capturing subpattern, the option letters may appear between the "?" and
1346 the ":". Thus the two patterns
1347 .sp
1348 (?i:saturday|sunday)
1349 (?:(?i)saturday|sunday)
1350 .sp
1351 match exactly the same set of strings. Because alternative branches are tried
1352 from left to right, and options are not reset until the end of the subpattern
1353 is reached, an option setting in one branch does affect subsequent branches, so
1354 the above patterns match "SUNDAY" as well as "Saturday".
1355 .
1356 .
1357 .\" HTML <a name="dupsubpatternnumber"></a>
1358 .SH "DUPLICATE SUBPATTERN NUMBERS"
1359 .rs
1360 .sp
1361 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1362 the same numbers for its capturing parentheses. Such a subpattern starts with
1363 (?| and is itself a non-capturing subpattern. For example, consider this
1364 pattern:
1365 .sp
1366 (?|(Sat)ur|(Sun))day
1367 .sp
1368 Because the two alternatives are inside a (?| group, both sets of capturing
1369 parentheses are numbered one. Thus, when the pattern matches, you can look
1370 at captured substring number one, whichever alternative matched. This construct
1371 is useful when you want to capture part, but not all, of one of a number of
1372 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1373 number is reset at the start of each branch. The numbers of any capturing
1374 parentheses that follow the subpattern start after the highest number used in
1375 any branch. The following example is taken from the Perl documentation. The
1376 numbers underneath show in which buffer the captured content will be stored.
1377 .sp
1378 # before ---------------branch-reset----------- after
1379 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1380 # 1 2 2 3 2 3 4
1381 .sp
1382 A back reference to a numbered subpattern uses the most recent value that is
1383 set for that number by any subpattern. The following pattern matches "abcabc"
1384 or "defdef":
1385 .sp
1386 /(?|(abc)|(def))\e1/
1387 .sp
1388 In contrast, a subroutine call to a numbered subpattern always refers to the
1389 first one in the pattern with the given number. The following pattern matches
1390 "abcabc" or "defabc":
1391 .sp
1392 /(?|(abc)|(def))(?1)/
1393 .sp
1394 If a
1395 .\" HTML <a href="#conditions">
1396 .\" </a>
1397 condition test
1398 .\"
1399 for a subpattern's having matched refers to a non-unique number, the test is
1400 true if any of the subpatterns of that number have matched.
1401 .P
1402 An alternative approach to using this "branch reset" feature is to use
1403 duplicate named subpatterns, as described in the next section.
1404 .
1405 .
1406 .SH "NAMED SUBPATTERNS"
1407 .rs
1408 .sp
1409 Identifying capturing parentheses by number is simple, but it can be very hard
1410 to keep track of the numbers in complicated regular expressions. Furthermore,
1411 if an expression is modified, the numbers may change. To help with this
1412 difficulty, PCRE supports the naming of subpatterns. This feature was not
1413 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1414 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1415 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1416 have different names, but PCRE does not.
1417 .P
1418 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1419 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1420 parentheses from other parts of the pattern, such as
1421 .\" HTML <a href="#backreferences">
1422 .\" </a>
1423 back references,
1424 .\"
1425 .\" HTML <a href="#recursion">
1426 .\" </a>
1427 recursion,
1428 .\"
1429 and
1430 .\" HTML <a href="#conditions">
1431 .\" </a>
1432 conditions,
1433 .\"
1434 can be made by name as well as by number.
1435 .P
1436 Names consist of up to 32 alphanumeric characters and underscores. Named
1437 capturing parentheses are still allocated numbers as well as names, exactly as
1438 if the names were not present. The PCRE API provides function calls for
1439 extracting the name-to-number translation table from a compiled pattern. There
1440 is also a convenience function for extracting a captured substring by name.
1441 .P
1442 By default, a name must be unique within a pattern, but it is possible to relax
1443 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1444 names are also always permitted for subpatterns with the same number, set up as
1445 described in the previous section.) Duplicate names can be useful for patterns
1446 where only one instance of the named parentheses can match. Suppose you want to
1447 match the name of a weekday, either as a 3-letter abbreviation or as the full
1448 name, and in both cases you want to extract the abbreviation. This pattern
1449 (ignoring the line breaks) does the job:
1450 .sp
1451 (?<DN>Mon|Fri|Sun)(?:day)?|
1452 (?<DN>Tue)(?:sday)?|
1453 (?<DN>Wed)(?:nesday)?|
1454 (?<DN>Thu)(?:rsday)?|
1455 (?<DN>Sat)(?:urday)?
1456 .sp
1457 There are five capturing substrings, but only one is ever set after a match.
1458 (An alternative way of solving this problem is to use a "branch reset"
1459 subpattern, as described in the previous section.)
1460 .P
1461 The convenience function for extracting the data by name returns the substring
1462 for the first (and in this example, the only) subpattern of that name that
1463 matched. This saves searching to find which numbered subpattern it was.
1464 .P
1465 If you make a back reference to a non-unique named subpattern from elsewhere in
1466 the pattern, the one that corresponds to the first occurrence of the name is
1467 used. In the absence of duplicate numbers (see the previous section) this is
1468 the one with the lowest number. If you use a named reference in a condition
1469 test (see the
1470 .\"
1471 .\" HTML <a href="#conditions">
1472 .\" </a>
1473 section about conditions
1474 .\"
1475 below), either to check whether a subpattern has matched, or to check for
1476 recursion, all subpatterns with the same name are tested. If the condition is
1477 true for any one of them, the overall condition is true. This is the same
1478 behaviour as testing by number. For further details of the interfaces for
1479 handling named subpatterns, see the
1480 .\" HREF
1481 \fBpcreapi\fP
1482 .\"
1483 documentation.
1484 .P
1485 \fBWarning:\fP You cannot use different names to distinguish between two
1486 subpatterns with the same number because PCRE uses only the numbers when
1487 matching. For this reason, an error is given at compile time if different names
1488 are given to subpatterns with the same number. However, you can give the same
1489 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1490 .
1491 .
1492 .SH REPETITION
1493 .rs
1494 .sp
1495 Repetition is specified by quantifiers, which can follow any of the following
1496 items:
1497 .sp
1498 a literal data character
1499 the dot metacharacter
1500 the \eC escape sequence
1501 the \eX escape sequence
1502 the \eR escape sequence
1503 an escape such as \ed or \epL that matches a single character
1504 a character class
1505 a back reference (see next section)
1506 a parenthesized subpattern (including assertions)
1507 a subroutine call to a subpattern (recursive or otherwise)
1508 .sp
1509 The general repetition quantifier specifies a minimum and maximum number of
1510 permitted matches, by giving the two numbers in curly brackets (braces),
1511 separated by a comma. The numbers must be less than 65536, and the first must
1512 be less than or equal to the second. For example:
1513 .sp
1514 z{2,4}
1515 .sp
1516 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1517 character. If the second number is omitted, but the comma is present, there is
1518 no upper limit; if the second number and the comma are both omitted, the
1519 quantifier specifies an exact number of required matches. Thus
1520 .sp
1521 [aeiou]{3,}
1522 .sp
1523 matches at least 3 successive vowels, but may match many more, while
1524 .sp
1525 \ed{8}
1526 .sp
1527 matches exactly 8 digits. An opening curly bracket that appears in a position
1528 where a quantifier is not allowed, or one that does not match the syntax of a
1529 quantifier, is taken as a literal character. For example, {,6} is not a
1530 quantifier, but a literal string of four characters.
1531 .P
1532 In UTF modes, quantifiers apply to characters rather than to individual data
1533 units. Thus, for example, \ex{100}{2} matches two characters, each of
1534 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1535 \eX{3} matches three Unicode extended sequences, each of which may be several
1536 data units long (and they may be of different lengths).
1537 .P
1538 The quantifier {0} is permitted, causing the expression to behave as if the
1539 previous item and the quantifier were not present. This may be useful for
1540 subpatterns that are referenced as
1541 .\" HTML <a href="#subpatternsassubroutines">
1542 .\" </a>
1543 subroutines
1544 .\"
1545 from elsewhere in the pattern (but see also the section entitled
1546 .\" HTML <a href="#subdefine">
1547 .\" </a>
1548 "Defining subpatterns for use by reference only"
1549 .\"
1550 below). Items other than subpatterns that have a {0} quantifier are omitted
1551 from the compiled pattern.
1552 .P
1553 For convenience, the three most common quantifiers have single-character
1554 abbreviations:
1555 .sp
1556 * is equivalent to {0,}
1557 + is equivalent to {1,}
1558 ? is equivalent to {0,1}
1559 .sp
1560 It is possible to construct infinite loops by following a subpattern that can
1561 match no characters with a quantifier that has no upper limit, for example:
1562 .sp
1563 (a?)*
1564 .sp
1565 Earlier versions of Perl and PCRE used to give an error at compile time for
1566 such patterns. However, because there are cases where this can be useful, such
1567 patterns are now accepted, but if any repetition of the subpattern does in fact
1568 match no characters, the loop is forcibly broken.
1569 .P
1570 By default, the quantifiers are "greedy", that is, they match as much as
1571 possible (up to the maximum number of permitted times), without causing the
1572 rest of the pattern to fail. The classic example of where this gives problems
1573 is in trying to match comments in C programs. These appear between /* and */
1574 and within the comment, individual * and / characters may appear. An attempt to
1575 match C comments by applying the pattern
1576 .sp
1577 /\e*.*\e*/
1578 .sp
1579 to the string
1580 .sp
1581 /* first comment */ not comment /* second comment */
1582 .sp
1583 fails, because it matches the entire string owing to the greediness of the .*
1584 item.
1585 .P
1586 However, if a quantifier is followed by a question mark, it ceases to be
1587 greedy, and instead matches the minimum number of times possible, so the
1588 pattern
1589 .sp
1590 /\e*.*?\e*/
1591 .sp
1592 does the right thing with the C comments. The meaning of the various
1593 quantifiers is not otherwise changed, just the preferred number of matches.
1594 Do not confuse this use of question mark with its use as a quantifier in its
1595 own right. Because it has two uses, it can sometimes appear doubled, as in
1596 .sp
1597 \ed??\ed
1598 .sp
1599 which matches one digit by preference, but can match two if that is the only
1600 way the rest of the pattern matches.
1601 .P
1602 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1603 the quantifiers are not greedy by default, but individual ones can be made
1604 greedy by following them with a question mark. In other words, it inverts the
1605 default behaviour.
1606 .P
1607 When a parenthesized subpattern is quantified with a minimum repeat count that
1608 is greater than 1 or with a limited maximum, more memory is required for the
1609 compiled pattern, in proportion to the size of the minimum or maximum.
1610 .P
1611 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1612 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1613 implicitly anchored, because whatever follows will be tried against every
1614 character position in the subject string, so there is no point in retrying the
1615 overall match at any position after the first. PCRE normally treats such a
1616 pattern as though it were preceded by \eA.
1617 .P
1618 In cases where it is known that the subject string contains no newlines, it is
1619 worth setting PCRE_DOTALL in order to obtain this optimization, or
1620 alternatively using ^ to indicate anchoring explicitly.
1621 .P
1622 However, there is one situation where the optimization cannot be used. When .*
1623 is inside capturing parentheses that are the subject of a back reference
1624 elsewhere in the pattern, a match at the start may fail where a later one
1625 succeeds. Consider, for example:
1626 .sp
1627 (.*)abc\e1
1628 .sp
1629 If the subject is "xyz123abc123" the match point is the fourth character. For
1630 this reason, such a pattern is not implicitly anchored.
1631 .P
1632 When a capturing subpattern is repeated, the value captured is the substring
1633 that matched the final iteration. For example, after
1634 .sp
1635 (tweedle[dume]{3}\es*)+
1636 .sp
1637 has matched "tweedledum tweedledee" the value of the captured substring is
1638 "tweedledee". However, if there are nested capturing subpatterns, the
1639 corresponding captured values may have been set in previous iterations. For
1640 example, after
1641 .sp
1642 /(a|(b))+/
1643 .sp
1644 matches "aba" the value of the second captured substring is "b".
1645 .
1646 .
1647 .\" HTML <a name="atomicgroup"></a>
1648 .SH "ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS"
1649 .rs
1650 .sp
1651 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1652 repetition, failure of what follows normally causes the repeated item to be
1653 re-evaluated to see if a different number of repeats allows the rest of the
1654 pattern to match. Sometimes it is useful to prevent this, either to change the
1655 nature of the match, or to cause it fail earlier than it otherwise might, when
1656 the author of the pattern knows there is no point in carrying on.
1657 .P
1658 Consider, for example, the pattern \ed+foo when applied to the subject line
1659 .sp
1660 123456bar
1661 .sp
1662 After matching all 6 digits and then failing to match "foo", the normal
1663 action of the matcher is to try again with only 5 digits matching the \ed+
1664 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1665 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1666 that once a subpattern has matched, it is not to be re-evaluated in this way.
1667 .P
1668 If we use atomic grouping for the previous example, the matcher gives up
1669 immediately on failing to match "foo" the first time. The notation is a kind of
1670 special parenthesis, starting with (?> as in this example:
1671 .sp
1672 (?>\ed+)foo
1673 .sp
1674 This kind of parenthesis "locks up" the part of the pattern it contains once
1675 it has matched, and a failure further into the pattern is prevented from
1676 backtracking into it. Backtracking past it to previous items, however, works as
1677 normal.
1678 .P
1679 An alternative description is that a subpattern of this type matches the string
1680 of characters that an identical standalone pattern would match, if anchored at
1681 the current point in the subject string.
1682 .P
1683 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1684 the above example can be thought of as a maximizing repeat that must swallow
1685 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1686 number of digits they match in order to make the rest of the pattern match,
1687 (?>\ed+) can only match an entire sequence of digits.
1688 .P
1689 Atomic groups in general can of course contain arbitrarily complicated
1690 subpatterns, and can be nested. However, when the subpattern for an atomic
1691 group is just a single repeated item, as in the example above, a simpler
1692 notation, called a "possessive quantifier" can be used. This consists of an
1693 additional + character following a quantifier. Using this notation, the
1694 previous example can be rewritten as
1695 .sp
1696 \ed++foo
1697 .sp
1698 Note that a possessive quantifier can be used with an entire group, for
1699 example:
1700 .sp
1701 (abc|xyz){2,3}+
1702 .sp
1703 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1704 option is ignored. They are a convenient notation for the simpler forms of
1705 atomic group. However, there is no difference in the meaning of a possessive
1706 quantifier and the equivalent atomic group, though there may be a performance
1707 difference; possessive quantifiers should be slightly faster.
1708 .P
1709 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1710 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1711 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1712 package, and PCRE copied it from there. It ultimately found its way into Perl
1713 at release 5.10.
1714 .P
1715 PCRE has an optimization that automatically "possessifies" certain simple
1716 pattern constructs. For example, the sequence A+B is treated as A++B because
1717 there is no point in backtracking into a sequence of A's when B must follow.
1718 .P
1719 When a pattern contains an unlimited repeat inside a subpattern that can itself
1720 be repeated an unlimited number of times, the use of an atomic group is the
1721 only way to avoid some failing matches taking a very long time indeed. The
1722 pattern
1723 .sp
1724 (\eD+|<\ed+>)*[!?]
1725 .sp
1726 matches an unlimited number of substrings that either consist of non-digits, or
1727 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1728 quickly. However, if it is applied to
1729 .sp
1730 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1731 .sp
1732 it takes a long time before reporting failure. This is because the string can
1733 be divided between the internal \eD+ repeat and the external * repeat in a
1734 large number of ways, and all have to be tried. (The example uses [!?] rather
1735 than a single character at the end, because both PCRE and Perl have an
1736 optimization that allows for fast failure when a single character is used. They
1737 remember the last single character that is required for a match, and fail early
1738 if it is not present in the string.) If the pattern is changed so that it uses
1739 an atomic group, like this:
1740 .sp
1741 ((?>\eD+)|<\ed+>)*[!?]
1742 .sp
1743 sequences of non-digits cannot be broken, and failure happens quickly.
1744 .
1745 .
1746 .\" HTML <a name="backreferences"></a>
1747 .SH "BACK REFERENCES"
1748 .rs
1749 .sp
1750 Outside a character class, a backslash followed by a digit greater than 0 (and
1751 possibly further digits) is a back reference to a capturing subpattern earlier
1752 (that is, to its left) in the pattern, provided there have been that many
1753 previous capturing left parentheses.
1754 .P
1755 However, if the decimal number following the backslash is less than 10, it is
1756 always taken as a back reference, and causes an error only if there are not
1757 that many capturing left parentheses in the entire pattern. In other words, the
1758 parentheses that are referenced need not be to the left of the reference for
1759 numbers less than 10. A "forward back reference" of this type can make sense
1760 when a repetition is involved and the subpattern to the right has participated
1761 in an earlier iteration.
1762 .P
1763 It is not possible to have a numerical "forward back reference" to a subpattern
1764 whose number is 10 or more using this syntax because a sequence such as \e50 is
1765 interpreted as a character defined in octal. See the subsection entitled
1766 "Non-printing characters"
1767 .\" HTML <a href="#digitsafterbackslash">
1768 .\" </a>
1769 above
1770 .\"
1771 for further details of the handling of digits following a backslash. There is
1772 no such problem when named parentheses are used. A back reference to any
1773 subpattern is possible using named parentheses (see below).
1774 .P
1775 Another way of avoiding the ambiguity inherent in the use of digits following a
1776 backslash is to use the \eg escape sequence. This escape must be followed by an
1777 unsigned number or a negative number, optionally enclosed in braces. These
1778 examples are all identical:
1779 .sp
1780 (ring), \e1
1781 (ring), \eg1
1782 (ring), \eg{1}
1783 .sp
1784 An unsigned number specifies an absolute reference without the ambiguity that
1785 is present in the older syntax. It is also useful when literal digits follow
1786 the reference. A negative number is a relative reference. Consider this
1787 example:
1788 .sp
1789 (abc(def)ghi)\eg{-1}
1790 .sp
1791 The sequence \eg{-1} is a reference to the most recently started capturing
1792 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1793 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1794 can be helpful in long patterns, and also in patterns that are created by
1795 joining together fragments that contain references within themselves.
1796 .P
1797 A back reference matches whatever actually matched the capturing subpattern in
1798 the current subject string, rather than anything matching the subpattern
1799 itself (see
1800 .\" HTML <a href="#subpatternsassubroutines">
1801 .\" </a>
1802 "Subpatterns as subroutines"
1803 .\"
1804 below for a way of doing that). So the pattern
1805 .sp
1806 (sens|respons)e and \e1ibility
1807 .sp
1808 matches "sense and sensibility" and "response and responsibility", but not
1809 "sense and responsibility". If caseful matching is in force at the time of the
1810 back reference, the case of letters is relevant. For example,
1811 .sp
1812 ((?i)rah)\es+\e1
1813 .sp
1814 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1815 capturing subpattern is matched caselessly.
1816 .P
1817 There are several different ways of writing back references to named
1818 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1819 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1820 back reference syntax, in which \eg can be used for both numeric and named
1821 references, is also supported. We could rewrite the above example in any of
1822 the following ways:
1823 .sp
1824 (?<p1>(?i)rah)\es+\ek<p1>
1825 (?'p1'(?i)rah)\es+\ek{p1}
1826 (?P<p1>(?i)rah)\es+(?P=p1)
1827 (?<p1>(?i)rah)\es+\eg{p1}
1828 .sp
1829 A subpattern that is referenced by name may appear in the pattern before or
1830 after the reference.
1831 .P
1832 There may be more than one back reference to the same subpattern. If a
1833 subpattern has not actually been used in a particular match, any back
1834 references to it always fail by default. For example, the pattern
1835 .sp
1836 (a|(bc))\e2
1837 .sp
1838 always fails if it starts to match "a" rather than "bc". However, if the
1839 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1840 unset value matches an empty string.
1841 .P
1842 Because there may be many capturing parentheses in a pattern, all digits
1843 following a backslash are taken as part of a potential back reference number.
1844 If the pattern continues with a digit character, some delimiter must be used to
1845 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1846 white space. Otherwise, the \eg{ syntax or an empty comment (see
1847 .\" HTML <a href="#comments">
1848 .\" </a>
1849 "Comments"
1850 .\"
1851 below) can be used.
1852 .
1853 .SS "Recursive back references"
1854 .rs
1855 .sp
1856 A back reference that occurs inside the parentheses to which it refers fails
1857 when the subpattern is first used, so, for example, (a\e1) never matches.
1858 However, such references can be useful inside repeated subpatterns. For
1859 example, the pattern
1860 .sp
1861 (a|b\e1)+
1862 .sp
1863 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1864 the subpattern, the back reference matches the character string corresponding
1865 to the previous iteration. In order for this to work, the pattern must be such
1866 that the first iteration does not need to match the back reference. This can be
1867 done using alternation, as in the example above, or by a quantifier with a
1868 minimum of zero.
1869 .P
1870 Back references of this type cause the group that they reference to be treated
1871 as an
1872 .\" HTML <a href="#atomicgroup">
1873 .\" </a>
1874 atomic group.
1875 .\"
1876 Once the whole group has been matched, a subsequent matching failure cannot
1877 cause backtracking into the middle of the group.
1878 .
1879 .
1880 .\" HTML <a name="bigassertions"></a>
1881 .SH ASSERTIONS
1882 .rs
1883 .sp
1884 An assertion is a test on the characters following or preceding the current
1885 matching point that does not actually consume any characters. The simple
1886 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1887 .\" HTML <a href="#smallassertions">
1888 .\" </a>
1889 above.
1890 .\"
1891 .P
1892 More complicated assertions are coded as subpatterns. There are two kinds:
1893 those that look ahead of the current position in the subject string, and those
1894 that look behind it. An assertion subpattern is matched in the normal way,
1895 except that it does not cause the current matching position to be changed.
1896 .P
1897 Assertion subpatterns are not capturing subpatterns. If such an assertion
1898 contains capturing subpatterns within it, these are counted for the purposes of
1899 numbering the capturing subpatterns in the whole pattern. However, substring
1900 capturing is carried out only for positive assertions, because it does not make
1901 sense for negative assertions.
1902 .P
1903 For compatibility with Perl, assertion subpatterns may be repeated; though
1904 it makes no sense to assert the same thing several times, the side effect of
1905 capturing parentheses may occasionally be useful. In practice, there only three
1906 cases:
1907 .sp
1908 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1909 However, it may contain internal capturing parenthesized groups that are called
1910 from elsewhere via the
1911 .\" HTML <a href="#subpatternsassubroutines">
1912 .\" </a>
1913 subroutine mechanism.
1914 .\"
1915 .sp
1916 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1917 were {0,1}. At run time, the rest of the pattern match is tried with and
1918 without the assertion, the order depending on the greediness of the quantifier.
1919 .sp
1920 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1921 The assertion is obeyed just once when encountered during matching.
1922 .
1923 .
1924 .SS "Lookahead assertions"
1925 .rs
1926 .sp
1927 Lookahead assertions start with (?= for positive assertions and (?! for
1928 negative assertions. For example,
1929 .sp
1930 \ew+(?=;)
1931 .sp
1932 matches a word followed by a semicolon, but does not include the semicolon in
1933 the match, and
1934 .sp
1935 foo(?!bar)
1936 .sp
1937 matches any occurrence of "foo" that is not followed by "bar". Note that the
1938 apparently similar pattern
1939 .sp
1940 (?!foo)bar
1941 .sp
1942 does not find an occurrence of "bar" that is preceded by something other than
1943 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1944 (?!foo) is always true when the next three characters are "bar". A
1945 lookbehind assertion is needed to achieve the other effect.
1946 .P
1947 If you want to force a matching failure at some point in a pattern, the most
1948 convenient way to do it is with (?!) because an empty string always matches, so
1949 an assertion that requires there not to be an empty string must always fail.
1950 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1951 .
1952 .
1953 .\" HTML <a name="lookbehind"></a>
1954 .SS "Lookbehind assertions"
1955 .rs
1956 .sp
1957 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1958 negative assertions. For example,
1959 .sp
1960 (?<!foo)bar
1961 .sp
1962 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1963 a lookbehind assertion are restricted such that all the strings it matches must
1964 have a fixed length. However, if there are several top-level alternatives, they
1965 do not all have to have the same fixed length. Thus
1966 .sp
1967 (?<=bullock|donkey)
1968 .sp
1969 is permitted, but
1970 .sp
1971 (?<!dogs?|cats?)
1972 .sp
1973 causes an error at compile time. Branches that match different length strings
1974 are permitted only at the top level of a lookbehind assertion. This is an
1975 extension compared with Perl, which requires all branches to match the same
1976 length of string. An assertion such as
1977 .sp
1978 (?<=ab(c|de))
1979 .sp
1980 is not permitted, because its single top-level branch can match two different
1981 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1982 branches:
1983 .sp
1984 (?<=abc|abde)
1985 .sp
1986 In some cases, the escape sequence \eK
1987 .\" HTML <a href="#resetmatchstart">
1988 .\" </a>
1989 (see above)
1990 .\"
1991 can be used instead of a lookbehind assertion to get round the fixed-length
1992 restriction.
1993 .P
1994 The implementation of lookbehind assertions is, for each alternative, to
1995 temporarily move the current position back by the fixed length and then try to
1996 match. If there are insufficient characters before the current position, the
1997 assertion fails.
1998 .P
1999 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2000 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2001 it impossible to calculate the length of the lookbehind. The \eX and \eR
2002 escapes, which can match different numbers of data units, are also not
2003 permitted.
2004 .P
2005 .\" HTML <a href="#subpatternsassubroutines">
2006 .\" </a>
2007 "Subroutine"
2008 .\"
2009 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2010 as the subpattern matches a fixed-length string.
2011 .\" HTML <a href="#recursion">
2012 .\" </a>
2013 Recursion,
2014 .\"
2015 however, is not supported.
2016 .P
2017 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2018 specify efficient matching of fixed-length strings at the end of subject
2019 strings. Consider a simple pattern such as
2020 .sp
2021 abcd$
2022 .sp
2023 when applied to a long string that does not match. Because matching proceeds
2024 from left to right, PCRE will look for each "a" in the subject and then see if
2025 what follows matches the rest of the pattern. If the pattern is specified as
2026 .sp
2027 ^.*abcd$
2028 .sp
2029 the initial .* matches the entire string at first, but when this fails (because
2030 there is no following "a"), it backtracks to match all but the last character,
2031 then all but the last two characters, and so on. Once again the search for "a"
2032 covers the entire string, from right to left, so we are no better off. However,
2033 if the pattern is written as
2034 .sp
2035 ^.*+(?<=abcd)
2036 .sp
2037 there can be no backtracking for the .*+ item; it can match only the entire
2038 string. The subsequent lookbehind assertion does a single test on the last four
2039 characters. If it fails, the match fails immediately. For long strings, this
2040 approach makes a significant difference to the processing time.
2041 .
2042 .
2043 .SS "Using multiple assertions"
2044 .rs
2045 .sp
2046 Several assertions (of any sort) may occur in succession. For example,
2047 .sp
2048 (?<=\ed{3})(?<!999)foo
2049 .sp
2050 matches "foo" preceded by three digits that are not "999". Notice that each of
2051 the assertions is applied independently at the same point in the subject
2052 string. First there is a check that the previous three characters are all
2053 digits, and then there is a check that the same three characters are not "999".
2054 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2055 of which are digits and the last three of which are not "999". For example, it
2056 doesn't match "123abcfoo". A pattern to do that is
2057 .sp
2058 (?<=\ed{3}...)(?<!999)foo
2059 .sp
2060 This time the first assertion looks at the preceding six characters, checking
2061 that the first three are digits, and then the second assertion checks that the
2062 preceding three characters are not "999".
2063 .P
2064 Assertions can be nested in any combination. For example,
2065 .sp
2066 (?<=(?<!foo)bar)baz
2067 .sp
2068 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2069 preceded by "foo", while
2070 .sp
2071 (?<=\ed{3}(?!999)...)foo
2072 .sp
2073 is another pattern that matches "foo" preceded by three digits and any three
2074 characters that are not "999".
2075 .
2076 .
2077 .\" HTML <a name="conditions"></a>
2078 .SH "CONDITIONAL SUBPATTERNS"
2079 .rs
2080 .sp
2081 It is possible to cause the matching process to obey a subpattern
2082 conditionally or to choose between two alternative subpatterns, depending on
2083 the result of an assertion, or whether a specific capturing subpattern has
2084 already been matched. The two possible forms of conditional subpattern are:
2085 .sp
2086 (?(condition)yes-pattern)
2087 (?(condition)yes-pattern|no-pattern)
2088 .sp
2089 If the condition is satisfied, the yes-pattern is used; otherwise the
2090 no-pattern (if present) is used. If there are more than two alternatives in the
2091 subpattern, a compile-time error occurs. Each of the two alternatives may
2092 itself contain nested subpatterns of any form, including conditional
2093 subpatterns; the restriction to two alternatives applies only at the level of
2094 the condition. This pattern fragment is an example where the alternatives are
2095 complex:
2096 .sp
2097 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2098 .sp
2099 .P
2100 There are four kinds of condition: references to subpatterns, references to
2101 recursion, a pseudo-condition called DEFINE, and assertions.
2102 .
2103 .SS "Checking for a used subpattern by number"
2104 .rs
2105 .sp
2106 If the text between the parentheses consists of a sequence of digits, the
2107 condition is true if a capturing subpattern of that number has previously
2108 matched. If there is more than one capturing subpattern with the same number
2109 (see the earlier
2110 .\"
2111 .\" HTML <a href="#recursion">
2112 .\" </a>
2113 section about duplicate subpattern numbers),
2114 .\"
2115 the condition is true if any of them have matched. An alternative notation is
2116 to precede the digits with a plus or minus sign. In this case, the subpattern
2117 number is relative rather than absolute. The most recently opened parentheses
2118 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2119 loops it can also make sense to refer to subsequent groups. The next
2120 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2121 zero in any of these forms is not used; it provokes a compile-time error.)
2122 .P
2123 Consider the following pattern, which contains non-significant white space to
2124 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2125 three parts for ease of discussion:
2126 .sp
2127 ( \e( )? [^()]+ (?(1) \e) )
2128 .sp
2129 The first part matches an optional opening parenthesis, and if that
2130 character is present, sets it as the first captured substring. The second part
2131 matches one or more characters that are not parentheses. The third part is a
2132 conditional subpattern that tests whether or not the first set of parentheses
2133 matched. If they did, that is, if subject started with an opening parenthesis,
2134 the condition is true, and so the yes-pattern is executed and a closing
2135 parenthesis is required. Otherwise, since no-pattern is not present, the
2136 subpattern matches nothing. In other words, this pattern matches a sequence of
2137 non-parentheses, optionally enclosed in parentheses.
2138 .P
2139 If you were embedding this pattern in a larger one, you could use a relative
2140 reference:
2141 .sp
2142 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2143 .sp
2144 This makes the fragment independent of the parentheses in the larger pattern.
2145 .
2146 .SS "Checking for a used subpattern by name"
2147 .rs
2148 .sp
2149 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2150 subpattern by name. For compatibility with earlier versions of PCRE, which had
2151 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2152 there is a possible ambiguity with this syntax, because subpattern names may
2153 consist entirely of digits. PCRE looks first for a named subpattern; if it
2154 cannot find one and the name consists entirely of digits, PCRE looks for a
2155 subpattern of that number, which must be greater than zero. Using subpattern
2156 names that consist entirely of digits is not recommended.
2157 .P
2158 Rewriting the above example to use a named subpattern gives this:
2159 .sp
2160 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2161 .sp
2162 If the name used in a condition of this kind is a duplicate, the test is
2163 applied to all subpatterns of the same name, and is true if any one of them has
2164 matched.
2165 .
2166 .SS "Checking for pattern recursion"
2167 .rs
2168 .sp
2169 If the condition is the string (R), and there is no subpattern with the name R,
2170 the condition is true if a recursive call to the whole pattern or any
2171 subpattern has been made. If digits or a name preceded by ampersand follow the
2172 letter R, for example:
2173 .sp
2174 (?(R3)...) or (?(R&name)...)
2175 .sp
2176 the condition is true if the most recent recursion is into a subpattern whose
2177 number or name is given. This condition does not check the entire recursion
2178 stack. If the name used in a condition of this kind is a duplicate, the test is
2179 applied to all subpatterns of the same name, and is true if any one of them is
2180 the most recent recursion.
2181 .P
2182 At "top level", all these recursion test conditions are false.
2183 .\" HTML <a href="#recursion">
2184 .\" </a>
2185 The syntax for recursive patterns
2186 .\"
2187 is described below.
2188 .
2189 .\" HTML <a name="subdefine"></a>
2190 .SS "Defining subpatterns for use by reference only"
2191 .rs
2192 .sp
2193 If the condition is the string (DEFINE), and there is no subpattern with the
2194 name DEFINE, the condition is always false. In this case, there may be only one
2195 alternative in the subpattern. It is always skipped if control reaches this
2196 point in the pattern; the idea of DEFINE is that it can be used to define
2197 subroutines that can be referenced from elsewhere. (The use of
2198 .\" HTML <a href="#subpatternsassubroutines">
2199 .\" </a>
2200 subroutines
2201 .\"
2202 is described below.) For example, a pattern to match an IPv4 address such as
2203 "192.168.23.245" could be written like this (ignore white space and line
2204 breaks):
2205 .sp
2206 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2207 \eb (?&byte) (\e.(?&byte)){3} \eb
2208 .sp
2209 The first part of the pattern is a DEFINE group inside which a another group
2210 named "byte" is defined. This matches an individual component of an IPv4
2211 address (a number less than 256). When matching takes place, this part of the
2212 pattern is skipped because DEFINE acts like a false condition. The rest of the
2213 pattern uses references to the named group to match the four dot-separated
2214 components of an IPv4 address, insisting on a word boundary at each end.
2215 .
2216 .SS "Assertion conditions"
2217 .rs
2218 .sp
2219 If the condition is not in any of the above formats, it must be an assertion.
2220 This may be a positive or negative lookahead or lookbehind assertion. Consider
2221 this pattern, again containing non-significant white space, and with the two
2222 alternatives on the second line:
2223 .sp
2224 (?(?=[^a-z]*[a-z])
2225 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2226 .sp
2227 The condition is a positive lookahead assertion that matches an optional
2228 sequence of non-letters followed by a letter. In other words, it tests for the
2229 presence of at least one letter in the subject. If a letter is found, the
2230 subject is matched against the first alternative; otherwise it is matched
2231 against the second. This pattern matches strings in one of the two forms
2232 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2233 .
2234 .
2235 .\" HTML <a name="comments"></a>
2236 .SH COMMENTS
2237 .rs
2238 .sp
2239 There are two ways of including comments in patterns that are processed by
2240 PCRE. In both cases, the start of the comment must not be in a character class,
2241 nor in the middle of any other sequence of related characters such as (?: or a
2242 subpattern name or number. The characters that make up a comment play no part
2243 in the pattern matching.
2244 .P
2245 The sequence (?# marks the start of a comment that continues up to the next
2246 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2247 option is set, an unescaped # character also introduces a comment, which in
2248 this case continues to immediately after the next newline character or
2249 character sequence in the pattern. Which characters are interpreted as newlines
2250 is controlled by the options passed to a compiling function or by a special
2251 sequence at the start of the pattern, as described in the section entitled
2252 .\" HTML <a href="#newlines">
2253 .\" </a>
2254 "Newline conventions"
2255 .\"
2256 above. Note that the end of this type of comment is a literal newline sequence
2257 in the pattern; escape sequences that happen to represent a newline do not
2258 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2259 default newline convention is in force:
2260 .sp
2261 abc #comment \en still comment
2262 .sp
2263 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2264 a newline in the pattern. The sequence \en is still literal at this stage, so
2265 it does not terminate the comment. Only an actual character with the code value
2266 0x0a (the default newline) does so.
2267 .
2268 .
2269 .\" HTML <a name="recursion"></a>
2270 .SH "RECURSIVE PATTERNS"
2271 .rs
2272 .sp
2273 Consider the problem of matching a string in parentheses, allowing for
2274 unlimited nested parentheses. Without the use of recursion, the best that can
2275 be done is to use a pattern that matches up to some fixed depth of nesting. It
2276 is not possible to handle an arbitrary nesting depth.
2277 .P
2278 For some time, Perl has provided a facility that allows regular expressions to
2279 recurse (amongst other things). It does this by interpolating Perl code in the
2280 expression at run time, and the code can refer to the expression itself. A Perl
2281 pattern using code interpolation to solve the parentheses problem can be
2282 created like this:
2283 .sp
2284 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2285 .sp
2286 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2287 recursively to the pattern in which it appears.
2288 .P
2289 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2290 supports special syntax for recursion of the entire pattern, and also for
2291 individual subpattern recursion. After its introduction in PCRE and Python,
2292 this kind of recursion was subsequently introduced into Perl at release 5.10.
2293 .P
2294 A special item that consists of (? followed by a number greater than zero and a
2295 closing parenthesis is a recursive subroutine call of the subpattern of the
2296 given number, provided that it occurs inside that subpattern. (If not, it is a
2297 .\" HTML <a href="#subpatternsassubroutines">
2298 .\" </a>
2299 non-recursive subroutine
2300 .\"
2301 call, which is described in the next section.) The special item (?R) or (?0) is
2302 a recursive call of the entire regular expression.
2303 .P
2304 This PCRE pattern solves the nested parentheses problem (assume the
2305 PCRE_EXTENDED option is set so that white space is ignored):
2306 .sp
2307 \e( ( [^()]++ | (?R) )* \e)
2308 .sp
2309 First it matches an opening parenthesis. Then it matches any number of
2310 substrings which can either be a sequence of non-parentheses, or a recursive
2311 match of the pattern itself (that is, a correctly parenthesized substring).
2312 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2313 to avoid backtracking into sequences of non-parentheses.
2314 .P
2315 If this were part of a larger pattern, you would not want to recurse the entire
2316 pattern, so instead you could use this:
2317 .sp
2318 ( \e( ( [^()]++ | (?1) )* \e) )
2319 .sp
2320 We have put the pattern into parentheses, and caused the recursion to refer to
2321 them instead of the whole pattern.
2322 .P
2323 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2324 is made easier by the use of relative references. Instead of (?1) in the
2325 pattern above you can write (?-2) to refer to the second most recently opened
2326 parentheses preceding the recursion. In other words, a negative number counts
2327 capturing parentheses leftwards from the point at which it is encountered.
2328 .P
2329 It is also possible to refer to subsequently opened parentheses, by writing
2330 references such as (?+2). However, these cannot be recursive because the
2331 reference is not inside the parentheses that are referenced. They are always
2332 .\" HTML <a href="#subpatternsassubroutines">
2333 .\" </a>
2334 non-recursive subroutine
2335 .\"
2336 calls, as described in the next section.
2337 .P
2338 An alternative approach is to use named parentheses instead. The Perl syntax
2339 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2340 could rewrite the above example as follows:
2341 .sp
2342 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2343 .sp
2344 If there is more than one subpattern with the same name, the earliest one is
2345 used.
2346 .P
2347 This particular example pattern that we have been looking at contains nested
2348 unlimited repeats, and so the use of a possessive quantifier for matching
2349 strings of non-parentheses is important when applying the pattern to strings
2350 that do not match. For example, when this pattern is applied to
2351 .sp
2352 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2353 .sp
2354 it yields "no match" quickly. However, if a possessive quantifier is not used,
2355 the match runs for a very long time indeed because there are so many different
2356 ways the + and * repeats can carve up the subject, and all have to be tested
2357 before failure can be reported.
2358 .P
2359 At the end of a match, the values of capturing parentheses are those from
2360 the outermost level. If you want to obtain intermediate values, a callout
2361 function can be used (see below and the
2362 .\" HREF
2363 \fBpcrecallout\fP
2364 .\"
2365 documentation). If the pattern above is matched against
2366 .sp
2367 (ab(cd)ef)
2368 .sp
2369 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2370 the last value taken on at the top level. If a capturing subpattern is not
2371 matched at the top level, its final captured value is unset, even if it was
2372 (temporarily) set at a deeper level during the matching process.
2373 .P
2374 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2375 obtain extra memory to store data during a recursion, which it does by using
2376 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2377 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2378 .P
2379 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2380 Consider this pattern, which matches text in angle brackets, allowing for
2381 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2382 recursing), whereas any characters are permitted at the outer level.
2383 .sp
2384 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2385 .sp
2386 In this pattern, (?(R) is the start of a conditional subpattern, with two
2387 different alternatives for the recursive and non-recursive cases. The (?R) item
2388 is the actual recursive call.
2389 .
2390 .
2391 .\" HTML <a name="recursiondifference"></a>
2392 .SS "Differences in recursion processing between PCRE and Perl"
2393 .rs
2394 .sp
2395 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2396 (like Python, but unlike Perl), a recursive subpattern call is always treated
2397 as an atomic group. That is, once it has matched some of the subject string, it
2398 is never re-entered, even if it contains untried alternatives and there is a
2399 subsequent matching failure. This can be illustrated by the following pattern,
2400 which purports to match a palindromic string that contains an odd number of
2401 characters (for example, "a", "aba", "abcba", "abcdcba"):
2402 .sp
2403 ^(.|(.)(?1)\e2)$
2404 .sp
2405 The idea is that it either matches a single character, or two identical
2406 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2407 it does not if the pattern is longer than three characters. Consider the
2408 subject string "abcba":
2409 .P
2410 At the top level, the first character is matched, but as it is not at the end
2411 of the string, the first alternative fails; the second alternative is taken
2412 and the recursion kicks in. The recursive call to subpattern 1 successfully
2413 matches the next character ("b"). (Note that the beginning and end of line
2414 tests are not part of the recursion).
2415 .P
2416 Back at the top level, the next character ("c") is compared with what
2417 subpattern 2 matched, which was "a". This fails. Because the recursion is
2418 treated as an atomic group, there are now no backtracking points, and so the
2419 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2420 try the second alternative.) However, if the pattern is written with the
2421 alternatives in the other order, things are different:
2422 .sp
2423 ^((.)(?1)\e2|.)$
2424 .sp
2425 This time, the recursing alternative is tried first, and continues to recurse
2426 until it runs out of characters, at which point the recursion fails. But this
2427 time we do have another alternative to try at the higher level. That is the big
2428 difference: in the previous case the remaining alternative is at a deeper
2429 recursion level, which PCRE cannot use.
2430 .P
2431 To change the pattern so that it matches all palindromic strings, not just
2432 those with an odd number of characters, it is tempting to change the pattern to
2433 this:
2434 .sp
2435 ^((.)(?1)\e2|.?)$
2436 .sp
2437 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2438 deeper recursion has matched a single character, it cannot be entered again in
2439 order to match an empty string. The solution is to separate the two cases, and
2440 write out the odd and even cases as alternatives at the higher level:
2441 .sp
2442 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2443 .sp
2444 If you want to match typical palindromic phrases, the pattern has to ignore all
2445 non-word characters, which can be done like this:
2446 .sp
2447 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2448 .sp
2449 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2450 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2451 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2452 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2453 more) to match typical phrases, and Perl takes so long that you think it has
2454 gone into a loop.
2455 .P
2456 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2457 string does not start with a palindrome that is shorter than the entire string.
2458 For example, although "abcba" is correctly matched, if the subject is "ababa",
2459 PCRE finds the palindrome "aba" at the start, then fails at top level because
2460 the end of the string does not follow. Once again, it cannot jump back into the
2461 recursion to try other alternatives, so the entire match fails.
2462 .P
2463 The second way in which PCRE and Perl differ in their recursion processing is
2464 in the handling of captured values. In Perl, when a subpattern is called
2465 recursively or as a subpattern (see the next section), it has no access to any
2466 values that were captured outside the recursion, whereas in PCRE these values
2467 can be referenced. Consider this pattern:
2468 .sp
2469 ^(.)(\e1|a(?2))
2470 .sp
2471 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2472 then in the second group, when the back reference \e1 fails to match "b", the
2473 second alternative matches "a" and then recurses. In the recursion, \e1 does
2474 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2475 match because inside the recursive call \e1 cannot access the externally set
2476 value.
2477 .
2478 .
2479 .\" HTML <a name="subpatternsassubroutines"></a>
2480 .SH "SUBPATTERNS AS SUBROUTINES"
2481 .rs
2482 .sp
2483 If the syntax for a recursive subpattern call (either by number or by
2484 name) is used outside the parentheses to which it refers, it operates like a
2485 subroutine in a programming language. The called subpattern may be defined
2486 before or after the reference. A numbered reference can be absolute or
2487 relative, as in these examples:
2488 .sp
2489 (...(absolute)...)...(?2)...
2490 (...(relative)...)...(?-1)...
2491 (...(?+1)...(relative)...
2492 .sp
2493 An earlier example pointed out that the pattern
2494 .sp
2495 (sens|respons)e and \e1ibility
2496 .sp
2497 matches "sense and sensibility" and "response and responsibility", but not
2498 "sense and responsibility". If instead the pattern
2499 .sp
2500 (sens|respons)e and (?1)ibility
2501 .sp
2502 is used, it does match "sense and responsibility" as well as the other two
2503 strings. Another example is given in the discussion of DEFINE above.
2504 .P
2505 All subroutine calls, whether recursive or not, are always treated as atomic
2506 groups. That is, once a subroutine has matched some of the subject string, it
2507 is never re-entered, even if it contains untried alternatives and there is a
2508 subsequent matching failure. Any capturing parentheses that are set during the
2509 subroutine call revert to their previous values afterwards.
2510 .P
2511 Processing options such as case-independence are fixed when a subpattern is
2512 defined, so if it is used as a subroutine, such options cannot be changed for
2513 different calls. For example, consider this pattern:
2514 .sp
2515 (abc)(?i:(?-1))
2516 .sp
2517 It matches "abcabc". It does not match "abcABC" because the change of
2518 processing option does not affect the called subpattern.
2519 .
2520 .
2521 .\" HTML <a name="onigurumasubroutines"></a>
2522 .SH "ONIGURUMA SUBROUTINE SYNTAX"
2523 .rs
2524 .sp
2525 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2526 a number enclosed either in angle brackets or single quotes, is an alternative
2527 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2528 are two of the examples used above, rewritten using this syntax:
2529 .sp
2530 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2531 (sens|respons)e and \eg'1'ibility
2532 .sp
2533 PCRE supports an extension to Oniguruma: if a number is preceded by a
2534 plus or a minus sign it is taken as a relative reference. For example:
2535 .sp
2536 (abc)(?i:\eg<-1>)
2537 .sp
2538 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2539 synonymous. The former is a back reference; the latter is a subroutine call.
2540 .
2541 .
2542 .SH CALLOUTS
2543 .rs
2544 .sp
2545 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2546 code to be obeyed in the middle of matching a regular expression. This makes it
2547 possible, amongst other things, to extract different substrings that match the
2548 same pair of parentheses when there is a repetition.
2549 .P
2550 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2551 code. The feature is called "callout". The caller of PCRE provides an external
2552 function by putting its entry point in the global variable \fIpcre_callout\fP
2553 (8-bit library) or \fIpcre16_callout\fP (16-bit library). By default, this
2554 variable contains NULL, which disables all calling out.
2555 .P
2556 Within a regular expression, (?C) indicates the points at which the external
2557 function is to be called. If you want to identify different callout points, you
2558 can put a number less than 256 after the letter C. The default value is zero.
2559 For example, this pattern has two callout points:
2560 .sp
2561 (?C1)abc(?C2)def
2562 .sp
2563 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2564 automatically installed before each item in the pattern. They are all numbered
2565 255.
2566 .P
2567 During matching, when PCRE reaches a callout point, the external function is
2568 called. It is provided with the number of the callout, the position in the
2569 pattern, and, optionally, one item of data originally supplied by the caller of
2570 the matching function. The callout function may cause matching to proceed, to
2571 backtrack, or to fail altogether. A complete description of the interface to
2572 the callout function is given in the
2573 .\" HREF
2574 \fBpcrecallout\fP
2575 .\"
2576 documentation.
2577 .
2578 .
2579 .\" HTML <a name="backtrackcontrol"></a>
2580 .SH "BACKTRACKING CONTROL"
2581 .rs
2582 .sp
2583 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2584 are described in the Perl documentation as "experimental and subject to change
2585 or removal in a future version of Perl". It goes on to say: "Their usage in
2586 production code should be noted to avoid problems during upgrades." The same
2587 remarks apply to the PCRE features described in this section.
2588 .P
2589 Since these verbs are specifically related to backtracking, most of them can be
2590 used only when the pattern is to be matched using one of the traditional
2591 matching functions, which use a backtracking algorithm. With the exception of
2592 (*FAIL), which behaves like a failing negative assertion, they cause an error
2593 if encountered by a DFA matching function.
2594 .P
2595 If any of these verbs are used in an assertion or in a subpattern that is
2596 called as a subroutine (whether or not recursively), their effect is confined
2597 to that subpattern; it does not extend to the surrounding pattern, with one
2598 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2599 a successful positive assertion \fIis\fP passed back when a match succeeds
2600 (compare capturing parentheses in assertions). Note that such subpatterns are
2601 processed as anchored at the point where they are tested. Note also that Perl's
2602 treatment of subroutines and assertions is different in some cases.
2603 .P
2604 The new verbs make use of what was previously invalid syntax: an opening
2605 parenthesis followed by an asterisk. They are generally of the form
2606 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2607 depending on whether or not an argument is present. A name is any sequence of
2608 characters that does not include a closing parenthesis. The maximum length of
2609 name is 255 in the 8-bit library and 65535 in the 16-bit library. If the name
2610 is empty, that is, if the closing parenthesis immediately follows the colon,
2611 the effect is as if the colon were not there. Any number of these verbs may
2612 occur in a pattern.
2613 .
2614 .
2615 .\" HTML <a name="nooptimize"></a>
2616 .SS "Optimizations that affect backtracking verbs"
2617 .rs
2618 .sp
2619 PCRE contains some optimizations that are used to speed up matching by running
2620 some checks at the start of each match attempt. For example, it may know the
2621 minimum length of matching subject, or that a particular character must be
2622 present. When one of these optimizations suppresses the running of a match, any
2623 included backtracking verbs will not, of course, be processed. You can suppress
2624 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2625 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2626 pattern with (*NO_START_OPT). There is more discussion of this option in the
2627 section entitled
2628 .\" HTML <a href="pcreapi.html#execoptions">
2629 .\" </a>
2630 "Option bits for \fBpcre_exec()\fP"
2631 .\"
2632 in the
2633 .\" HREF
2634 \fBpcreapi\fP
2635 .\"
2636 documentation.
2637 .P
2638 Experiments with Perl suggest that it too has similar optimizations, sometimes
2639 leading to anomalous results.
2640 .
2641 .
2642 .SS "Verbs that act immediately"
2643 .rs
2644 .sp
2645 The following verbs act as soon as they are encountered. They may not be
2646 followed by a name.
2647 .sp
2648 (*ACCEPT)
2649 .sp
2650 This verb causes the match to end successfully, skipping the remainder of the
2651 pattern. However, when it is inside a subpattern that is called as a
2652 subroutine, only that subpattern is ended successfully. Matching then continues
2653 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2654 far is captured. For example:
2655 .sp
2656 A((?:A|B(*ACCEPT)|C)D)
2657 .sp
2658 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2659 the outer parentheses.
2660 .sp
2661 (*FAIL) or (*F)
2662 .sp
2663 This verb causes a matching failure, forcing backtracking to occur. It is
2664 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2665 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2666 Perl features that are not present in PCRE. The nearest equivalent is the
2667 callout feature, as for example in this pattern:
2668 .sp
2669 a+(?C)(*FAIL)
2670 .sp
2671 A match with the string "aaaa" always fails, but the callout is taken before
2672 each backtrack happens (in this example, 10 times).
2673 .
2674 .
2675 .SS "Recording which path was taken"
2676 .rs
2677 .sp
2678 There is one verb whose main purpose is to track how a match was arrived at,
2679 though it also has a secondary use in conjunction with advancing the match
2680 starting point (see (*SKIP) below).
2681 .sp
2682 (*MARK:NAME) or (*:NAME)
2683 .sp
2684 A name is always required with this verb. There may be as many instances of
2685 (*MARK) as you like in a pattern, and their names do not have to be unique.
2686 .P
2687 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2688 path is passed back to the caller as described in the section entitled
2689 .\" HTML <a href="pcreapi.html#extradata">
2690 .\" </a>
2691 "Extra data for \fBpcre_exec()\fP"
2692 .\"
2693 in the
2694 .\" HREF
2695 \fBpcreapi\fP
2696 .\"
2697 documentation. Here is an example of \fBpcretest\fP output, where the /K
2698 modifier requests the retrieval and outputting of (*MARK) data:
2699 .sp
2700 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2701 data> XY
2702 0: XY
2703 MK: A
2704 XZ
2705 0: XZ
2706 MK: B
2707 .sp
2708 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2709 indicates which of the two alternatives matched. This is a more efficient way
2710 of obtaining this information than putting each alternative in its own
2711 capturing parentheses.
2712 .P
2713 If (*MARK) is encountered in a positive assertion, its name is recorded and
2714 passed back if it is the last-encountered. This does not happen for negative
2715 assertions.
2716 .P
2717 After a partial match or a failed match, the name of the last encountered
2718 (*MARK) in the entire match process is returned. For example:
2719 .sp
2720 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2721 data> XP
2722 No match, mark = B
2723 .sp
2724 Note that in this unanchored example the mark is retained from the match
2725 attempt that started at the letter "X" in the subject. Subsequent match
2726 attempts starting at "P" and then with an empty string do not get as far as the
2727 (*MARK) item, but nevertheless do not reset it.
2728 .P
2729 If you are interested in (*MARK) values after failed matches, you should
2730 probably set the PCRE_NO_START_OPTIMIZE option
2731 .\" HTML <a href="#nooptimize">
2732 .\" </a>
2733 (see above)
2734 .\"
2735 to ensure that the match is always attempted.
2736 .
2737 .
2738 .SS "Verbs that act after backtracking"
2739 .rs
2740 .sp
2741 The following verbs do nothing when they are encountered. Matching continues
2742 with what follows, but if there is no subsequent match, causing a backtrack to
2743 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2744 the verb. However, when one of these verbs appears inside an atomic group, its
2745 effect is confined to that group, because once the group has been matched,
2746 there is never any backtracking into it. In this situation, backtracking can
2747 "jump back" to the left of the entire atomic group. (Remember also, as stated
2748 above, that this localization also applies in subroutine calls and assertions.)
2749 .P
2750 These verbs differ in exactly what kind of failure occurs when backtracking
2751 reaches them.
2752 .sp
2753 (*COMMIT)
2754 .sp
2755 This verb, which may not be followed by a name, causes the whole match to fail
2756 outright if the rest of the pattern does not match. Even if the pattern is
2757 unanchored, no further attempts to find a match by advancing the starting point
2758 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2759 finding a match at the current starting point, or not at all. For example:
2760 .sp
2761 a+(*COMMIT)b
2762 .sp
2763 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2764 dynamic anchor, or "I've started, so I must finish." The name of the most
2765 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2766 match failure.
2767 .P
2768 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2769 unless PCRE's start-of-match optimizations are turned off, as shown in this
2770 \fBpcretest\fP example:
2771 .sp
2772 re> /(*COMMIT)abc/
2773 data> xyzabc
2774 0: abc
2775 xyzabc\eY
2776 No match
2777 .sp
2778 PCRE knows that any match must start with "a", so the optimization skips along
2779 the subject to "a" before running the first match attempt, which succeeds. When
2780 the optimization is disabled by the \eY escape in the second subject, the match
2781 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2782 starting points.
2783 .sp
2784 (*PRUNE) or (*PRUNE:NAME)
2785 .sp
2786 This verb causes the match to fail at the current starting position in the
2787 subject if the rest of the pattern does not match. If the pattern is
2788 unanchored, the normal "bumpalong" advance to the next starting character then
2789 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2790 reached, or when matching to the right of (*PRUNE), but if there is no match to
2791 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2792 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2793 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2794 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2795 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2796 .sp
2797 (*SKIP)
2798 .sp
2799 This verb, when given without a name, is like (*PRUNE), except that if the
2800 pattern is unanchored, the "bumpalong" advance is not to the next character,
2801 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2802 signifies that whatever text was matched leading up to it cannot be part of a
2803 successful match. Consider:
2804 .sp
2805 a+(*SKIP)b
2806 .sp
2807 If the subject is "aaaac...", after the first match attempt fails (starting at
2808 the first character in the string), the starting point skips on to start the
2809 next attempt at "c". Note that a possessive quantifer does not have the same
2810 effect as this example; although it would suppress backtracking during the
2811 first match attempt, the second attempt would start at the second character
2812 instead of skipping on to "c".
2813 .sp
2814 (*SKIP:NAME)
2815 .sp
2816 When (*SKIP) has an associated name, its behaviour is modified. If the
2817 following pattern fails to match, the previous path through the pattern is
2818 searched for the most recent (*MARK) that has the same name. If one is found,
2819 the "bumpalong" advance is to the subject position that corresponds to that
2820 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2821 matching name is found, the (*SKIP) is ignored.
2822 .sp
2823 (*THEN) or (*THEN:NAME)
2824 .sp
2825 This verb causes a skip to the next innermost alternative if the rest of the
2826 pattern does not match. That is, it cancels pending backtracking, but only
2827 within the current alternative. Its name comes from the observation that it can
2828 be used for a pattern-based if-then-else block:
2829 .sp
2830 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2831 .sp
2832 If the COND1 pattern matches, FOO is tried (and possibly further items after
2833 the end of the group if FOO succeeds); on failure, the matcher skips to the
2834 second alternative and tries COND2, without backtracking into COND1. The
2835 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2836 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2837 .P
2838 Note that a subpattern that does not contain a | character is just a part of
2839 the enclosing alternative; it is not a nested alternation with only one
2840 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2841 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2842 pattern fragments that do not contain any | characters at this level:
2843 .sp
2844 A (B(*THEN)C) | D
2845 .sp
2846 If A and B are matched, but there is a failure in C, matching does not
2847 backtrack into A; instead it moves to the next alternative, that is, D.
2848 However, if the subpattern containing (*THEN) is given an alternative, it
2849 behaves differently:
2850 .sp
2851 A (B(*THEN)C | (*FAIL)) | D
2852 .sp
2853 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2854 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2855 because there are no more alternatives to try. In this case, matching does now
2856 backtrack into A.
2857 .P
2858 Note also that a conditional subpattern is not considered as having two
2859 alternatives, because only one is ever used. In other words, the | character in
2860 a conditional subpattern has a different meaning. Ignoring white space,
2861 consider:
2862 .sp
2863 ^.*? (?(?=a) a | b(*THEN)c )
2864 .sp
2865 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2866 it initially matches zero characters. The condition (?=a) then fails, the
2867 character "b" is matched, but "c" is not. At this point, matching does not
2868 backtrack to .*? as might perhaps be expected from the presence of the |
2869 character. The conditional subpattern is part of the single alternative that
2870 comprises the whole pattern, and so the match fails. (If there was a backtrack
2871 into .*?, allowing it to match "b", the match would succeed.)
2872 .P
2873 The verbs just described provide four different "strengths" of control when
2874 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2875 next alternative. (*PRUNE) comes next, failing the match at the current
2876 starting position, but allowing an advance to the next character (for an
2877 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2878 than one character. (*COMMIT) is the strongest, causing the entire match to
2879 fail.
2880 .P
2881 If more than one such verb is present in a pattern, the "strongest" one wins.
2882 For example, consider this pattern, where A, B, etc. are complex pattern
2883 fragments:
2884 .sp
2885 (A(*COMMIT)B(*THEN)C|D)
2886 .sp
2887 Once A has matched, PCRE is committed to this match, at the current starting
2888 position. If subsequently B matches, but C does not, the normal (*THEN) action
2889 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2890 overrides.
2891 .
2892 .
2893 .SH "SEE ALSO"
2894 .rs
2895 .sp
2896 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2897 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP.
2898 .
2899 .
2900 .SH AUTHOR
2901 .rs
2902 .sp
2903 .nf
2904 Philip Hazel
2905 University Computing Service
2906 Cambridge CB2 3QH, England.
2907 .fi
2908 .
2909 .
2910 .SH REVISION
2911 .rs
2912 .sp
2913 .nf
2914 Last updated: 01 June 2012
2915 Copyright (c) 1997-2012 University of Cambridge.
2916 .fi

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5