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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 982 - (show annotations)
Wed Jun 20 15:15:27 2012 UTC (7 years, 4 months ago) by ph10
File size: 121038 byte(s)
Source tidies for 8.31-RC2.
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 Character codes specified by \eu in JavaScript mode are constrained in the same
281 was as those specified by \ex in non-JavaScript mode.
282 .P
283 Characters whose value is less than 256 can be defined by either of the two
284 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
285 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
286 \eu00dc in JavaScript mode).
287 .P
288 After \e0 up to two further octal digits are read. If there are fewer than two
289 digits, just those that are present are used. Thus the sequence \e0\ex\e07
290 specifies two binary zeros followed by a BEL character (code value 7). Make
291 sure you supply two digits after the initial zero if the pattern character that
292 follows is itself an octal digit.
293 .P
294 The handling of a backslash followed by a digit other than 0 is complicated.
295 Outside a character class, PCRE reads it and any following digits as a decimal
296 number. If the number is less than 10, or if there have been at least that many
297 previous capturing left parentheses in the expression, the entire sequence is
298 taken as a \fIback reference\fP. A description of how this works is given
299 .\" HTML <a href="#backreferences">
300 .\" </a>
301 later,
302 .\"
303 following the discussion of
304 .\" HTML <a href="#subpattern">
305 .\" </a>
306 parenthesized subpatterns.
307 .\"
308 .P
309 Inside a character class, or if the decimal number is greater than 9 and there
310 have not been that many capturing subpatterns, PCRE re-reads up to three octal
311 digits following the backslash, and uses them to generate a data character. Any
312 subsequent digits stand for themselves. The value of the character is
313 constrained in the same way as characters specified in hexadecimal.
314 For example:
315 .sp
316 \e040 is another way of writing a space
317 .\" JOIN
318 \e40 is the same, provided there are fewer than 40
319 previous capturing subpatterns
320 \e7 is always a back reference
321 .\" JOIN
322 \e11 might be a back reference, or another way of
323 writing a tab
324 \e011 is always a tab
325 \e0113 is a tab followed by the character "3"
326 .\" JOIN
327 \e113 might be a back reference, otherwise the
328 character with octal code 113
329 .\" JOIN
330 \e377 might be a back reference, otherwise
331 the value 255 (decimal)
332 .\" JOIN
333 \e81 is either a back reference, or a binary zero
334 followed by the two characters "8" and "1"
335 .sp
336 Note that octal values of 100 or greater must not be introduced by a leading
337 zero, because no more than three octal digits are ever read.
338 .P
339 All the sequences that define a single character value can be used both inside
340 and outside character classes. In addition, inside a character class, \eb is
341 interpreted as the backspace character (hex 08).
342 .P
343 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
344 inside a character class. Like other unrecognized escape sequences, they are
345 treated as the literal characters "B", "R", and "X" by default, but cause an
346 error if the PCRE_EXTRA option is set. Outside a character class, these
347 sequences have different meanings.
348 .
349 .
350 .SS "Unsupported escape sequences"
351 .rs
352 .sp
353 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
354 handler and used to modify the case of following characters. By default, PCRE
355 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
356 option is set, \eU matches a "U" character, and \eu can be used to define a
357 character by code point, as described in the previous section.
358 .
359 .
360 .SS "Absolute and relative back references"
361 .rs
362 .sp
363 The sequence \eg followed by an unsigned or a negative number, optionally
364 enclosed in braces, is an absolute or relative back reference. A named back
365 reference can be coded as \eg{name}. Back references are discussed
366 .\" HTML <a href="#backreferences">
367 .\" </a>
368 later,
369 .\"
370 following the discussion of
371 .\" HTML <a href="#subpattern">
372 .\" </a>
373 parenthesized subpatterns.
374 .\"
375 .
376 .
377 .SS "Absolute and relative subroutine calls"
378 .rs
379 .sp
380 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
381 a number enclosed either in angle brackets or single quotes, is an alternative
382 syntax for referencing a subpattern as a "subroutine". Details are discussed
383 .\" HTML <a href="#onigurumasubroutines">
384 .\" </a>
385 later.
386 .\"
387 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
388 synonymous. The former is a back reference; the latter is a
389 .\" HTML <a href="#subpatternsassubroutines">
390 .\" </a>
391 subroutine
392 .\"
393 call.
394 .
395 .
396 .\" HTML <a name="genericchartypes"></a>
397 .SS "Generic character types"
398 .rs
399 .sp
400 Another use of backslash is for specifying generic character types:
401 .sp
402 \ed any decimal digit
403 \eD any character that is not a decimal digit
404 \eh any horizontal white space character
405 \eH any character that is not a horizontal white space character
406 \es any white space character
407 \eS any character that is not a white space character
408 \ev any vertical white space character
409 \eV any character that is not a vertical white space character
410 \ew any "word" character
411 \eW any "non-word" character
412 .sp
413 There is also the single sequence \eN, which matches a non-newline character.
414 This is the same as
415 .\" HTML <a href="#fullstopdot">
416 .\" </a>
417 the "." metacharacter
418 .\"
419 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
420 PCRE does not support this.
421 .P
422 Each pair of lower and upper case escape sequences partitions the complete set
423 of characters into two disjoint sets. Any given character matches one, and only
424 one, of each pair. The sequences can appear both inside and outside character
425 classes. They each match one character of the appropriate type. If the current
426 matching point is at the end of the subject string, all of them fail, because
427 there is no character to match.
428 .P
429 For compatibility with Perl, \es does not match the VT character (code 11).
430 This makes it different from the the POSIX "space" class. The \es characters
431 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
432 included in a Perl script, \es may match the VT character. In PCRE, it never
433 does.
434 .P
435 A "word" character is an underscore or any character that is a letter or digit.
436 By default, the definition of letters and digits is controlled by PCRE's
437 low-valued character tables, and may vary if locale-specific matching is taking
438 place (see
439 .\" HTML <a href="pcreapi.html#localesupport">
440 .\" </a>
441 "Locale support"
442 .\"
443 in the
444 .\" HREF
445 \fBpcreapi\fP
446 .\"
447 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
448 or "french" in Windows, some character codes greater than 128 are used for
449 accented letters, and these are then matched by \ew. The use of locales with
450 Unicode is discouraged.
451 .P
452 By default, in a UTF mode, characters with values greater than 128 never match
453 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
454 their original meanings from before UTF support was available, mainly for
455 efficiency reasons. However, if PCRE is compiled with Unicode property support,
456 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
457 properties are used to determine character types, as follows:
458 .sp
459 \ed any character that \ep{Nd} matches (decimal digit)
460 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
461 \ew any character that \ep{L} or \ep{N} matches, plus underscore
462 .sp
463 The upper case escapes match the inverse sets of characters. Note that \ed
464 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
465 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
466 \eB because they are defined in terms of \ew and \eW. Matching these sequences
467 is noticeably slower when PCRE_UCP is set.
468 .P
469 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
470 release 5.10. In contrast to the other sequences, which match only ASCII
471 characters by default, these always match certain high-valued codepoints,
472 whether or not PCRE_UCP is set. The horizontal space characters are:
473 .sp
474 U+0009 Horizontal tab
475 U+0020 Space
476 U+00A0 Non-break space
477 U+1680 Ogham space mark
478 U+180E Mongolian vowel separator
479 U+2000 En quad
480 U+2001 Em quad
481 U+2002 En space
482 U+2003 Em space
483 U+2004 Three-per-em space
484 U+2005 Four-per-em space
485 U+2006 Six-per-em space
486 U+2007 Figure space
487 U+2008 Punctuation space
488 U+2009 Thin space
489 U+200A Hair space
490 U+202F Narrow no-break space
491 U+205F Medium mathematical space
492 U+3000 Ideographic space
493 .sp
494 The vertical space characters are:
495 .sp
496 U+000A Linefeed
497 U+000B Vertical tab
498 U+000C Form feed
499 U+000D Carriage return
500 U+0085 Next line
501 U+2028 Line separator
502 U+2029 Paragraph separator
503 .sp
504 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
505 relevant.
506 .
507 .
508 .\" HTML <a name="newlineseq"></a>
509 .SS "Newline sequences"
510 .rs
511 .sp
512 Outside a character class, by default, the escape sequence \eR matches any
513 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
514 following:
515 .sp
516 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
517 .sp
518 This is an example of an "atomic group", details of which are given
519 .\" HTML <a href="#atomicgroup">
520 .\" </a>
521 below.
522 .\"
523 This particular group matches either the two-character sequence CR followed by
524 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
525 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
526 line, U+0085). The two-character sequence is treated as a single unit that
527 cannot be split.
528 .P
529 In other modes, two additional characters whose codepoints are greater than 255
530 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
531 Unicode character property support is not needed for these characters to be
532 recognized.
533 .P
534 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
535 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
536 either at compile time or when the pattern is matched. (BSR is an abbrevation
537 for "backslash R".) This can be made the default when PCRE is built; if this is
538 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
539 It is also possible to specify these settings by starting a pattern string with
540 one of the following sequences:
541 .sp
542 (*BSR_ANYCRLF) CR, LF, or CRLF only
543 (*BSR_UNICODE) any Unicode newline sequence
544 .sp
545 These override the default and the options given to the compiling function, but
546 they can themselves be overridden by options given to a matching function. Note
547 that these special settings, which are not Perl-compatible, are recognized only
548 at the very start of a pattern, and that they must be in upper case. If more
549 than one of them is present, the last one is used. They can be combined with a
550 change of newline convention; for example, a pattern can start with:
551 .sp
552 (*ANY)(*BSR_ANYCRLF)
553 .sp
554 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
555 sequences. Inside a character class, \eR is treated as an unrecognized escape
556 sequence, and so matches the letter "R" by default, but causes an error if
557 PCRE_EXTRA is set.
558 .
559 .
560 .\" HTML <a name="uniextseq"></a>
561 .SS Unicode character properties
562 .rs
563 .sp
564 When PCRE is built with Unicode character property support, three additional
565 escape sequences that match characters with specific properties are available.
566 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
567 characters whose codepoints are less than 256, but they do work in this mode.
568 The extra escape sequences are:
569 .sp
570 \ep{\fIxx\fP} a character with the \fIxx\fP property
571 \eP{\fIxx\fP} a character without the \fIxx\fP property
572 \eX an extended Unicode sequence
573 .sp
574 The property names represented by \fIxx\fP above are limited to the Unicode
575 script names, the general category properties, "Any", which matches any
576 character (including newline), and some special PCRE properties (described
577 in the
578 .\" HTML <a href="#extraprops">
579 .\" </a>
580 next section).
581 .\"
582 Other Perl properties such as "InMusicalSymbols" are not currently supported by
583 PCRE. Note that \eP{Any} does not match any characters, so always causes a
584 match failure.
585 .P
586 Sets of Unicode characters are defined as belonging to certain scripts. A
587 character from one of these sets can be matched using a script name. For
588 example:
589 .sp
590 \ep{Greek}
591 \eP{Han}
592 .sp
593 Those that are not part of an identified script are lumped together as
594 "Common". The current list of scripts is:
595 .P
596 Arabic,
597 Armenian,
598 Avestan,
599 Balinese,
600 Bamum,
601 Batak,
602 Bengali,
603 Bopomofo,
604 Brahmi,
605 Braille,
606 Buginese,
607 Buhid,
608 Canadian_Aboriginal,
609 Carian,
610 Chakma,
611 Cham,
612 Cherokee,
613 Common,
614 Coptic,
615 Cuneiform,
616 Cypriot,
617 Cyrillic,
618 Deseret,
619 Devanagari,
620 Egyptian_Hieroglyphs,
621 Ethiopic,
622 Georgian,
623 Glagolitic,
624 Gothic,
625 Greek,
626 Gujarati,
627 Gurmukhi,
628 Han,
629 Hangul,
630 Hanunoo,
631 Hebrew,
632 Hiragana,
633 Imperial_Aramaic,
634 Inherited,
635 Inscriptional_Pahlavi,
636 Inscriptional_Parthian,
637 Javanese,
638 Kaithi,
639 Kannada,
640 Katakana,
641 Kayah_Li,
642 Kharoshthi,
643 Khmer,
644 Lao,
645 Latin,
646 Lepcha,
647 Limbu,
648 Linear_B,
649 Lisu,
650 Lycian,
651 Lydian,
652 Malayalam,
653 Mandaic,
654 Meetei_Mayek,
655 Meroitic_Cursive,
656 Meroitic_Hieroglyphs,
657 Miao,
658 Mongolian,
659 Myanmar,
660 New_Tai_Lue,
661 Nko,
662 Ogham,
663 Old_Italic,
664 Old_Persian,
665 Old_South_Arabian,
666 Old_Turkic,
667 Ol_Chiki,
668 Oriya,
669 Osmanya,
670 Phags_Pa,
671 Phoenician,
672 Rejang,
673 Runic,
674 Samaritan,
675 Saurashtra,
676 Sharada,
677 Shavian,
678 Sinhala,
679 Sora_Sompeng,
680 Sundanese,
681 Syloti_Nagri,
682 Syriac,
683 Tagalog,
684 Tagbanwa,
685 Tai_Le,
686 Tai_Tham,
687 Tai_Viet,
688 Takri,
689 Tamil,
690 Telugu,
691 Thaana,
692 Thai,
693 Tibetan,
694 Tifinagh,
695 Ugaritic,
696 Vai,
697 Yi.
698 .P
699 Each character has exactly one Unicode general category property, specified by
700 a two-letter abbreviation. For compatibility with Perl, negation can be
701 specified by including a circumflex between the opening brace and the property
702 name. For example, \ep{^Lu} is the same as \eP{Lu}.
703 .P
704 If only one letter is specified with \ep or \eP, it includes all the general
705 category properties that start with that letter. In this case, in the absence
706 of negation, the curly brackets in the escape sequence are optional; these two
707 examples have the same effect:
708 .sp
709 \ep{L}
710 \epL
711 .sp
712 The following general category property codes are supported:
713 .sp
714 C Other
715 Cc Control
716 Cf Format
717 Cn Unassigned
718 Co Private use
719 Cs Surrogate
720 .sp
721 L Letter
722 Ll Lower case letter
723 Lm Modifier letter
724 Lo Other letter
725 Lt Title case letter
726 Lu Upper case letter
727 .sp
728 M Mark
729 Mc Spacing mark
730 Me Enclosing mark
731 Mn Non-spacing mark
732 .sp
733 N Number
734 Nd Decimal number
735 Nl Letter number
736 No Other number
737 .sp
738 P Punctuation
739 Pc Connector punctuation
740 Pd Dash punctuation
741 Pe Close punctuation
742 Pf Final punctuation
743 Pi Initial punctuation
744 Po Other punctuation
745 Ps Open punctuation
746 .sp
747 S Symbol
748 Sc Currency symbol
749 Sk Modifier symbol
750 Sm Mathematical symbol
751 So Other symbol
752 .sp
753 Z Separator
754 Zl Line separator
755 Zp Paragraph separator
756 Zs Space separator
757 .sp
758 The special property L& is also supported: it matches a character that has
759 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
760 a modifier or "other".
761 .P
762 The Cs (Surrogate) property applies only to characters in the range U+D800 to
763 U+DFFF. Such characters are not valid in Unicode strings and so
764 cannot be tested by PCRE, unless UTF validity checking has been turned off
765 (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
766 .\" HREF
767 \fBpcreapi\fP
768 .\"
769 page). Perl does not support the Cs property.
770 .P
771 The long synonyms for property names that Perl supports (such as \ep{Letter})
772 are not supported by PCRE, nor is it permitted to prefix any of these
773 properties with "Is".
774 .P
775 No character that is in the Unicode table has the Cn (unassigned) property.
776 Instead, this property is assumed for any code point that is not in the
777 Unicode table.
778 .P
779 Specifying caseless matching does not affect these escape sequences. For
780 example, \ep{Lu} always matches only upper case letters.
781 .P
782 The \eX escape matches any number of Unicode characters that form an extended
783 Unicode sequence. \eX is equivalent to
784 .sp
785 (?>\ePM\epM*)
786 .sp
787 That is, it matches a character without the "mark" property, followed by zero
788 or more characters with the "mark" property, and treats the sequence as an
789 atomic group
790 .\" HTML <a href="#atomicgroup">
791 .\" </a>
792 (see below).
793 .\"
794 Characters with the "mark" property are typically accents that affect the
795 preceding character. None of them have codepoints less than 256, so in
796 8-bit non-UTF-8 mode \eX matches any one character.
797 .P
798 Note that recent versions of Perl have changed \eX to match what Unicode calls
799 an "extended grapheme cluster", which has a more complicated definition.
800 .P
801 Matching characters by Unicode property is not fast, because PCRE has to search
802 a structure that contains data for over fifteen thousand characters. That is
803 why the traditional escape sequences such as \ed and \ew do not use Unicode
804 properties in PCRE by default, though you can make them do so by setting the
805 PCRE_UCP option or by starting the pattern with (*UCP).
806 .
807 .
808 .\" HTML <a name="extraprops"></a>
809 .SS PCRE's additional properties
810 .rs
811 .sp
812 As well as the standard Unicode properties described in the previous
813 section, PCRE supports four more that make it possible to convert traditional
814 escape sequences such as \ew and \es and POSIX character classes to use Unicode
815 properties. PCRE uses these non-standard, non-Perl properties internally when
816 PCRE_UCP is set. They are:
817 .sp
818 Xan Any alphanumeric character
819 Xps Any POSIX space character
820 Xsp Any Perl space character
821 Xwd Any Perl "word" character
822 .sp
823 Xan matches characters that have either the L (letter) or the N (number)
824 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
825 carriage return, and any other character that has the Z (separator) property.
826 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
827 same characters as Xan, plus underscore.
828 .
829 .
830 .\" HTML <a name="resetmatchstart"></a>
831 .SS "Resetting the match start"
832 .rs
833 .sp
834 The escape sequence \eK causes any previously matched characters not to be
835 included in the final matched sequence. For example, the pattern:
836 .sp
837 foo\eKbar
838 .sp
839 matches "foobar", but reports that it has matched "bar". This feature is
840 similar to a lookbehind assertion
841 .\" HTML <a href="#lookbehind">
842 .\" </a>
843 (described below).
844 .\"
845 However, in this case, the part of the subject before the real match does not
846 have to be of fixed length, as lookbehind assertions do. The use of \eK does
847 not interfere with the setting of
848 .\" HTML <a href="#subpattern">
849 .\" </a>
850 captured substrings.
851 .\"
852 For example, when the pattern
853 .sp
854 (foo)\eKbar
855 .sp
856 matches "foobar", the first substring is still set to "foo".
857 .P
858 Perl documents that the use of \eK within assertions is "not well defined". In
859 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
860 ignored in negative assertions.
861 .
862 .
863 .\" HTML <a name="smallassertions"></a>
864 .SS "Simple assertions"
865 .rs
866 .sp
867 The final use of backslash is for certain simple assertions. An assertion
868 specifies a condition that has to be met at a particular point in a match,
869 without consuming any characters from the subject string. The use of
870 subpatterns for more complicated assertions is described
871 .\" HTML <a href="#bigassertions">
872 .\" </a>
873 below.
874 .\"
875 The backslashed assertions are:
876 .sp
877 \eb matches at a word boundary
878 \eB matches when not at a word boundary
879 \eA matches at the start of the subject
880 \eZ matches at the end of the subject
881 also matches before a newline at the end of the subject
882 \ez matches only at the end of the subject
883 \eG matches at the first matching position in the subject
884 .sp
885 Inside a character class, \eb has a different meaning; it matches the backspace
886 character. If any other of these assertions appears in a character class, by
887 default it matches the corresponding literal character (for example, \eB
888 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
889 escape sequence" error is generated instead.
890 .P
891 A word boundary is a position in the subject string where the current character
892 and the previous character do not both match \ew or \eW (i.e. one matches
893 \ew and the other matches \eW), or the start or end of the string if the
894 first or last character matches \ew, respectively. In a UTF mode, the meanings
895 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
896 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
897 of word" or "end of word" metasequence. However, whatever follows \eb normally
898 determines which it is. For example, the fragment \eba matches "a" at the start
899 of a word.
900 .P
901 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
902 dollar (described in the next section) in that they only ever match at the very
903 start and end of the subject string, whatever options are set. Thus, they are
904 independent of multiline mode. These three assertions are not affected by the
905 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
906 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
907 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
908 at a point other than the beginning of the subject, \eA can never match. The
909 difference between \eZ and \ez is that \eZ matches before a newline at the end
910 of the string as well as at the very end, whereas \ez matches only at the end.
911 .P
912 The \eG assertion is true only when the current matching position is at the
913 start point of the match, as specified by the \fIstartoffset\fP argument of
914 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
915 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
916 arguments, you can mimic Perl's /g option, and it is in this kind of
917 implementation where \eG can be useful.
918 .P
919 Note, however, that PCRE's interpretation of \eG, as the start of the current
920 match, is subtly different from Perl's, which defines it as the end of the
921 previous match. In Perl, these can be different when the previously matched
922 string was empty. Because PCRE does just one match at a time, it cannot
923 reproduce this behaviour.
924 .P
925 If all the alternatives of a pattern begin with \eG, the expression is anchored
926 to the starting match position, and the "anchored" flag is set in the compiled
927 regular expression.
928 .
929 .
930 .SH "CIRCUMFLEX AND DOLLAR"
931 .rs
932 .sp
933 Outside a character class, in the default matching mode, the circumflex
934 character is an assertion that is true only if the current matching point is
935 at the start of the subject string. If the \fIstartoffset\fP argument of
936 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
937 option is unset. Inside a character class, circumflex has an entirely different
938 meaning
939 .\" HTML <a href="#characterclass">
940 .\" </a>
941 (see below).
942 .\"
943 .P
944 Circumflex need not be the first character of the pattern if a number of
945 alternatives are involved, but it should be the first thing in each alternative
946 in which it appears if the pattern is ever to match that branch. If all
947 possible alternatives start with a circumflex, that is, if the pattern is
948 constrained to match only at the start of the subject, it is said to be an
949 "anchored" pattern. (There are also other constructs that can cause a pattern
950 to be anchored.)
951 .P
952 A dollar character is an assertion that is true only if the current matching
953 point is at the end of the subject string, or immediately before a newline
954 at the end of the string (by default). Dollar need not be the last character of
955 the pattern if a number of alternatives are involved, but it should be the last
956 item in any branch in which it appears. Dollar has no special meaning in a
957 character class.
958 .P
959 The meaning of dollar can be changed so that it matches only at the very end of
960 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
961 does not affect the \eZ assertion.
962 .P
963 The meanings of the circumflex and dollar characters are changed if the
964 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
965 immediately after internal newlines as well as at the start of the subject
966 string. It does not match after a newline that ends the string. A dollar
967 matches before any newlines in the string, as well as at the very end, when
968 PCRE_MULTILINE is set. When newline is specified as the two-character
969 sequence CRLF, isolated CR and LF characters do not indicate newlines.
970 .P
971 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
972 \en represents a newline) in multiline mode, but not otherwise. Consequently,
973 patterns that are anchored in single line mode because all branches start with
974 ^ are not anchored in multiline mode, and a match for circumflex is possible
975 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
976 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
977 .P
978 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
979 end of the subject in both modes, and if all branches of a pattern start with
980 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
981 .
982 .
983 .\" HTML <a name="fullstopdot"></a>
984 .SH "FULL STOP (PERIOD, DOT) AND \eN"
985 .rs
986 .sp
987 Outside a character class, a dot in the pattern matches any one character in
988 the subject string except (by default) a character that signifies the end of a
989 line.
990 .P
991 When a line ending is defined as a single character, dot never matches that
992 character; when the two-character sequence CRLF is used, dot does not match CR
993 if it is immediately followed by LF, but otherwise it matches all characters
994 (including isolated CRs and LFs). When any Unicode line endings are being
995 recognized, dot does not match CR or LF or any of the other line ending
996 characters.
997 .P
998 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
999 option is set, a dot matches any one character, without exception. If the
1000 two-character sequence CRLF is present in the subject string, it takes two dots
1001 to match it.
1002 .P
1003 The handling of dot is entirely independent of the handling of circumflex and
1004 dollar, the only relationship being that they both involve newlines. Dot has no
1005 special meaning in a character class.
1006 .P
1007 The escape sequence \eN behaves like a dot, except that it is not affected by
1008 the PCRE_DOTALL option. In other words, it matches any character except one
1009 that signifies the end of a line. Perl also uses \eN to match characters by
1010 name; PCRE does not support this.
1011 .
1012 .
1013 .SH "MATCHING A SINGLE DATA UNIT"
1014 .rs
1015 .sp
1016 Outside a character class, the escape sequence \eC matches any one data unit,
1017 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1018 byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \eC always
1019 matches line-ending characters. The feature is provided in Perl in order to
1020 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1021 used. Because \eC breaks up characters into individual data units, matching one
1022 unit with \eC in a UTF mode means that the rest of the string may start with a
1023 malformed UTF character. This has undefined results, because PCRE assumes that
1024 it is dealing with valid UTF strings (and by default it checks this at the
1025 start of processing unless the PCRE_NO_UTF8_CHECK or PCRE_NO_UTF16_CHECK option
1026 is used).
1027 .P
1028 PCRE does not allow \eC to appear in lookbehind assertions
1029 .\" HTML <a href="#lookbehind">
1030 .\" </a>
1031 (described below)
1032 .\"
1033 in a UTF mode, because this would make it impossible to calculate the length of
1034 the lookbehind.
1035 .P
1036 In general, the \eC escape sequence is best avoided. However, one
1037 way of using it that avoids the problem of malformed UTF characters is to use a
1038 lookahead to check the length of the next character, as in this pattern, which
1039 could be used with a UTF-8 string (ignore white space and line breaks):
1040 .sp
1041 (?| (?=[\ex00-\ex7f])(\eC) |
1042 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1043 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1044 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1045 .sp
1046 A group that starts with (?| resets the capturing parentheses numbers in each
1047 alternative (see
1048 .\" HTML <a href="#dupsubpatternnumber">
1049 .\" </a>
1050 "Duplicate Subpattern Numbers"
1051 .\"
1052 below). The assertions at the start of each branch check the next UTF-8
1053 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1054 character's individual bytes are then captured by the appropriate number of
1055 groups.
1056 .
1057 .
1058 .\" HTML <a name="characterclass"></a>
1059 .SH "SQUARE BRACKETS AND CHARACTER CLASSES"
1060 .rs
1061 .sp
1062 An opening square bracket introduces a character class, terminated by a closing
1063 square bracket. A closing square bracket on its own is not special by default.
1064 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1065 bracket causes a compile-time error. If a closing square bracket is required as
1066 a member of the class, it should be the first data character in the class
1067 (after an initial circumflex, if present) or escaped with a backslash.
1068 .P
1069 A character class matches a single character in the subject. In a UTF mode, the
1070 character may be more than one data unit long. A matched character must be in
1071 the set of characters defined by the class, unless the first character in the
1072 class definition is a circumflex, in which case the subject character must not
1073 be in the set defined by the class. If a circumflex is actually required as a
1074 member of the class, ensure it is not the first character, or escape it with a
1075 backslash.
1076 .P
1077 For example, the character class [aeiou] matches any lower case vowel, while
1078 [^aeiou] matches any character that is not a lower case vowel. Note that a
1079 circumflex is just a convenient notation for specifying the characters that
1080 are in the class by enumerating those that are not. A class that starts with a
1081 circumflex is not an assertion; it still consumes a character from the subject
1082 string, and therefore it fails if the current pointer is at the end of the
1083 string.
1084 .P
1085 In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
1086 included in a class as a literal string of data units, or by using the \ex{
1087 escaping mechanism.
1088 .P
1089 When caseless matching is set, any letters in a class represent both their
1090 upper case and lower case versions, so for example, a caseless [aeiou] matches
1091 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1092 caseful version would. In a UTF mode, PCRE always understands the concept of
1093 case for characters whose values are less than 128, so caseless matching is
1094 always possible. For characters with higher values, the concept of case is
1095 supported if PCRE is compiled with Unicode property support, but not otherwise.
1096 If you want to use caseless matching in a UTF mode for characters 128 and
1097 above, you must ensure that PCRE is compiled with Unicode property support as
1098 well as with UTF support.
1099 .P
1100 Characters that might indicate line breaks are never treated in any special way
1101 when matching character classes, whatever line-ending sequence is in use, and
1102 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1103 such as [^a] always matches one of these characters.
1104 .P
1105 The minus (hyphen) character can be used to specify a range of characters in a
1106 character class. For example, [d-m] matches any letter between d and m,
1107 inclusive. If a minus character is required in a class, it must be escaped with
1108 a backslash or appear in a position where it cannot be interpreted as
1109 indicating a range, typically as the first or last character in the class.
1110 .P
1111 It is not possible to have the literal character "]" as the end character of a
1112 range. A pattern such as [W-]46] is interpreted as a class of two characters
1113 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1114 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1115 the end of range, so [W-\e]46] is interpreted as a class containing a range
1116 followed by two other characters. The octal or hexadecimal representation of
1117 "]" can also be used to end a range.
1118 .P
1119 Ranges operate in the collating sequence of character values. They can also be
1120 used for characters specified numerically, for example [\e000-\e037]. Ranges
1121 can include any characters that are valid for the current mode.
1122 .P
1123 If a range that includes letters is used when caseless matching is set, it
1124 matches the letters in either case. For example, [W-c] is equivalent to
1125 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1126 tables for a French locale are in use, [\exc8-\excb] matches accented E
1127 characters in both cases. In UTF modes, PCRE supports the concept of case for
1128 characters with values greater than 128 only when it is compiled with Unicode
1129 property support.
1130 .P
1131 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1132 \eV, \ew, and \eW may appear in a character class, and add the characters that
1133 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1134 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1135 and their upper case partners, just as it does when they appear outside a
1136 character class, as described in the section entitled
1137 .\" HTML <a href="#genericchartypes">
1138 .\" </a>
1139 "Generic character types"
1140 .\"
1141 above. The escape sequence \eb has a different meaning inside a character
1142 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1143 are not special inside a character class. Like any other unrecognized escape
1144 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1145 default, but cause an error if the PCRE_EXTRA option is set.
1146 .P
1147 A circumflex can conveniently be used with the upper case character types to
1148 specify a more restricted set of characters than the matching lower case type.
1149 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1150 whereas [\ew] includes underscore. A positive character class should be read as
1151 "something OR something OR ..." and a negative class as "NOT something AND NOT
1152 something AND NOT ...".
1153 .P
1154 The only metacharacters that are recognized in character classes are backslash,
1155 hyphen (only where it can be interpreted as specifying a range), circumflex
1156 (only at the start), opening square bracket (only when it can be interpreted as
1157 introducing a POSIX class name - see the next section), and the terminating
1158 closing square bracket. However, escaping other non-alphanumeric characters
1159 does no harm.
1160 .
1161 .
1162 .SH "POSIX CHARACTER CLASSES"
1163 .rs
1164 .sp
1165 Perl supports the POSIX notation for character classes. This uses names
1166 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1167 this notation. For example,
1168 .sp
1169 [01[:alpha:]%]
1170 .sp
1171 matches "0", "1", any alphabetic character, or "%". The supported class names
1172 are:
1173 .sp
1174 alnum letters and digits
1175 alpha letters
1176 ascii character codes 0 - 127
1177 blank space or tab only
1178 cntrl control characters
1179 digit decimal digits (same as \ed)
1180 graph printing characters, excluding space
1181 lower lower case letters
1182 print printing characters, including space
1183 punct printing characters, excluding letters and digits and space
1184 space white space (not quite the same as \es)
1185 upper upper case letters
1186 word "word" characters (same as \ew)
1187 xdigit hexadecimal digits
1188 .sp
1189 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1190 space (32). Notice that this list includes the VT character (code 11). This
1191 makes "space" different to \es, which does not include VT (for Perl
1192 compatibility).
1193 .P
1194 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1195 5.8. Another Perl extension is negation, which is indicated by a ^ character
1196 after the colon. For example,
1197 .sp
1198 [12[:^digit:]]
1199 .sp
1200 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1201 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1202 supported, and an error is given if they are encountered.
1203 .P
1204 By default, in UTF modes, characters with values greater than 128 do not match
1205 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1206 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1207 character properties are used. This is achieved by replacing the POSIX classes
1208 by other sequences, as follows:
1209 .sp
1210 [:alnum:] becomes \ep{Xan}
1211 [:alpha:] becomes \ep{L}
1212 [:blank:] becomes \eh
1213 [:digit:] becomes \ep{Nd}
1214 [:lower:] becomes \ep{Ll}
1215 [:space:] becomes \ep{Xps}
1216 [:upper:] becomes \ep{Lu}
1217 [:word:] becomes \ep{Xwd}
1218 .sp
1219 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1220 classes are unchanged, and match only characters with code points less than
1221 128.
1222 .
1223 .
1224 .SH "VERTICAL BAR"
1225 .rs
1226 .sp
1227 Vertical bar characters are used to separate alternative patterns. For example,
1228 the pattern
1229 .sp
1230 gilbert|sullivan
1231 .sp
1232 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1233 and an empty alternative is permitted (matching the empty string). The matching
1234 process tries each alternative in turn, from left to right, and the first one
1235 that succeeds is used. If the alternatives are within a subpattern
1236 .\" HTML <a href="#subpattern">
1237 .\" </a>
1238 (defined below),
1239 .\"
1240 "succeeds" means matching the rest of the main pattern as well as the
1241 alternative in the subpattern.
1242 .
1243 .
1244 .SH "INTERNAL OPTION SETTING"
1245 .rs
1246 .sp
1247 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1248 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1249 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1250 The option letters are
1251 .sp
1252 i for PCRE_CASELESS
1253 m for PCRE_MULTILINE
1254 s for PCRE_DOTALL
1255 x for PCRE_EXTENDED
1256 .sp
1257 For example, (?im) sets caseless, multiline matching. It is also possible to
1258 unset these options by preceding the letter with a hyphen, and a combined
1259 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1260 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1261 permitted. If a letter appears both before and after the hyphen, the option is
1262 unset.
1263 .P
1264 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1265 changed in the same way as the Perl-compatible options by using the characters
1266 J, U and X respectively.
1267 .P
1268 When one of these option changes occurs at top level (that is, not inside
1269 subpattern parentheses), the change applies to the remainder of the pattern
1270 that follows. If the change is placed right at the start of a pattern, PCRE
1271 extracts it into the global options (and it will therefore show up in data
1272 extracted by the \fBpcre_fullinfo()\fP function).
1273 .P
1274 An option change within a subpattern (see below for a description of
1275 subpatterns) affects only that part of the subpattern that follows it, so
1276 .sp
1277 (a(?i)b)c
1278 .sp
1279 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1280 By this means, options can be made to have different settings in different
1281 parts of the pattern. Any changes made in one alternative do carry on
1282 into subsequent branches within the same subpattern. For example,
1283 .sp
1284 (a(?i)b|c)
1285 .sp
1286 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1287 branch is abandoned before the option setting. This is because the effects of
1288 option settings happen at compile time. There would be some very weird
1289 behaviour otherwise.
1290 .P
1291 \fBNote:\fP There are other PCRE-specific options that can be set by the
1292 application when the compiling or matching functions are called. In some cases
1293 the pattern can contain special leading sequences such as (*CRLF) to override
1294 what the application has set or what has been defaulted. Details are given in
1295 the section entitled
1296 .\" HTML <a href="#newlineseq">
1297 .\" </a>
1298 "Newline sequences"
1299 .\"
1300 above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
1301 can be used to set UTF and Unicode property modes; they are equivalent to
1302 setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1303 .
1304 .
1305 .\" HTML <a name="subpattern"></a>
1306 .SH SUBPATTERNS
1307 .rs
1308 .sp
1309 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1310 Turning part of a pattern into a subpattern does two things:
1311 .sp
1312 1. It localizes a set of alternatives. For example, the pattern
1313 .sp
1314 cat(aract|erpillar|)
1315 .sp
1316 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1317 match "cataract", "erpillar" or an empty string.
1318 .sp
1319 2. It sets up the subpattern as a capturing subpattern. This means that, when
1320 the whole pattern matches, that portion of the subject string that matched the
1321 subpattern is passed back to the caller via the \fIovector\fP argument of the
1322 matching function. (This applies only to the traditional matching functions;
1323 the DFA matching functions do not support capturing.)
1324 .P
1325 Opening parentheses are counted from left to right (starting from 1) to obtain
1326 numbers for the capturing subpatterns. For example, if the string "the red
1327 king" is matched against the pattern
1328 .sp
1329 the ((red|white) (king|queen))
1330 .sp
1331 the captured substrings are "red king", "red", and "king", and are numbered 1,
1332 2, and 3, respectively.
1333 .P
1334 The fact that plain parentheses fulfil two functions is not always helpful.
1335 There are often times when a grouping subpattern is required without a
1336 capturing requirement. If an opening parenthesis is followed by a question mark
1337 and a colon, the subpattern does not do any capturing, and is not counted when
1338 computing the number of any subsequent capturing subpatterns. For example, if
1339 the string "the white queen" is matched against the pattern
1340 .sp
1341 the ((?:red|white) (king|queen))
1342 .sp
1343 the captured substrings are "white queen" and "queen", and are numbered 1 and
1344 2. The maximum number of capturing subpatterns is 65535.
1345 .P
1346 As a convenient shorthand, if any option settings are required at the start of
1347 a non-capturing subpattern, the option letters may appear between the "?" and
1348 the ":". Thus the two patterns
1349 .sp
1350 (?i:saturday|sunday)
1351 (?:(?i)saturday|sunday)
1352 .sp
1353 match exactly the same set of strings. Because alternative branches are tried
1354 from left to right, and options are not reset until the end of the subpattern
1355 is reached, an option setting in one branch does affect subsequent branches, so
1356 the above patterns match "SUNDAY" as well as "Saturday".
1357 .
1358 .
1359 .\" HTML <a name="dupsubpatternnumber"></a>
1360 .SH "DUPLICATE SUBPATTERN NUMBERS"
1361 .rs
1362 .sp
1363 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1364 the same numbers for its capturing parentheses. Such a subpattern starts with
1365 (?| and is itself a non-capturing subpattern. For example, consider this
1366 pattern:
1367 .sp
1368 (?|(Sat)ur|(Sun))day
1369 .sp
1370 Because the two alternatives are inside a (?| group, both sets of capturing
1371 parentheses are numbered one. Thus, when the pattern matches, you can look
1372 at captured substring number one, whichever alternative matched. This construct
1373 is useful when you want to capture part, but not all, of one of a number of
1374 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1375 number is reset at the start of each branch. The numbers of any capturing
1376 parentheses that follow the subpattern start after the highest number used in
1377 any branch. The following example is taken from the Perl documentation. The
1378 numbers underneath show in which buffer the captured content will be stored.
1379 .sp
1380 # before ---------------branch-reset----------- after
1381 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1382 # 1 2 2 3 2 3 4
1383 .sp
1384 A back reference to a numbered subpattern uses the most recent value that is
1385 set for that number by any subpattern. The following pattern matches "abcabc"
1386 or "defdef":
1387 .sp
1388 /(?|(abc)|(def))\e1/
1389 .sp
1390 In contrast, a subroutine call to a numbered subpattern always refers to the
1391 first one in the pattern with the given number. The following pattern matches
1392 "abcabc" or "defabc":
1393 .sp
1394 /(?|(abc)|(def))(?1)/
1395 .sp
1396 If a
1397 .\" HTML <a href="#conditions">
1398 .\" </a>
1399 condition test
1400 .\"
1401 for a subpattern's having matched refers to a non-unique number, the test is
1402 true if any of the subpatterns of that number have matched.
1403 .P
1404 An alternative approach to using this "branch reset" feature is to use
1405 duplicate named subpatterns, as described in the next section.
1406 .
1407 .
1408 .SH "NAMED SUBPATTERNS"
1409 .rs
1410 .sp
1411 Identifying capturing parentheses by number is simple, but it can be very hard
1412 to keep track of the numbers in complicated regular expressions. Furthermore,
1413 if an expression is modified, the numbers may change. To help with this
1414 difficulty, PCRE supports the naming of subpatterns. This feature was not
1415 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1416 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1417 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1418 have different names, but PCRE does not.
1419 .P
1420 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1421 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1422 parentheses from other parts of the pattern, such as
1423 .\" HTML <a href="#backreferences">
1424 .\" </a>
1425 back references,
1426 .\"
1427 .\" HTML <a href="#recursion">
1428 .\" </a>
1429 recursion,
1430 .\"
1431 and
1432 .\" HTML <a href="#conditions">
1433 .\" </a>
1434 conditions,
1435 .\"
1436 can be made by name as well as by number.
1437 .P
1438 Names consist of up to 32 alphanumeric characters and underscores. Named
1439 capturing parentheses are still allocated numbers as well as names, exactly as
1440 if the names were not present. The PCRE API provides function calls for
1441 extracting the name-to-number translation table from a compiled pattern. There
1442 is also a convenience function for extracting a captured substring by name.
1443 .P
1444 By default, a name must be unique within a pattern, but it is possible to relax
1445 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1446 names are also always permitted for subpatterns with the same number, set up as
1447 described in the previous section.) Duplicate names can be useful for patterns
1448 where only one instance of the named parentheses can match. Suppose you want to
1449 match the name of a weekday, either as a 3-letter abbreviation or as the full
1450 name, and in both cases you want to extract the abbreviation. This pattern
1451 (ignoring the line breaks) does the job:
1452 .sp
1453 (?<DN>Mon|Fri|Sun)(?:day)?|
1454 (?<DN>Tue)(?:sday)?|
1455 (?<DN>Wed)(?:nesday)?|
1456 (?<DN>Thu)(?:rsday)?|
1457 (?<DN>Sat)(?:urday)?
1458 .sp
1459 There are five capturing substrings, but only one is ever set after a match.
1460 (An alternative way of solving this problem is to use a "branch reset"
1461 subpattern, as described in the previous section.)
1462 .P
1463 The convenience function for extracting the data by name returns the substring
1464 for the first (and in this example, the only) subpattern of that name that
1465 matched. This saves searching to find which numbered subpattern it was.
1466 .P
1467 If you make a back reference to a non-unique named subpattern from elsewhere in
1468 the pattern, the one that corresponds to the first occurrence of the name is
1469 used. In the absence of duplicate numbers (see the previous section) this is
1470 the one with the lowest number. If you use a named reference in a condition
1471 test (see the
1472 .\"
1473 .\" HTML <a href="#conditions">
1474 .\" </a>
1475 section about conditions
1476 .\"
1477 below), either to check whether a subpattern has matched, or to check for
1478 recursion, all subpatterns with the same name are tested. If the condition is
1479 true for any one of them, the overall condition is true. This is the same
1480 behaviour as testing by number. For further details of the interfaces for
1481 handling named subpatterns, see the
1482 .\" HREF
1483 \fBpcreapi\fP
1484 .\"
1485 documentation.
1486 .P
1487 \fBWarning:\fP You cannot use different names to distinguish between two
1488 subpatterns with the same number because PCRE uses only the numbers when
1489 matching. For this reason, an error is given at compile time if different names
1490 are given to subpatterns with the same number. However, you can give the same
1491 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1492 .
1493 .
1494 .SH REPETITION
1495 .rs
1496 .sp
1497 Repetition is specified by quantifiers, which can follow any of the following
1498 items:
1499 .sp
1500 a literal data character
1501 the dot metacharacter
1502 the \eC escape sequence
1503 the \eX escape sequence
1504 the \eR escape sequence
1505 an escape such as \ed or \epL that matches a single character
1506 a character class
1507 a back reference (see next section)
1508 a parenthesized subpattern (including assertions)
1509 a subroutine call to a subpattern (recursive or otherwise)
1510 .sp
1511 The general repetition quantifier specifies a minimum and maximum number of
1512 permitted matches, by giving the two numbers in curly brackets (braces),
1513 separated by a comma. The numbers must be less than 65536, and the first must
1514 be less than or equal to the second. For example:
1515 .sp
1516 z{2,4}
1517 .sp
1518 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1519 character. If the second number is omitted, but the comma is present, there is
1520 no upper limit; if the second number and the comma are both omitted, the
1521 quantifier specifies an exact number of required matches. Thus
1522 .sp
1523 [aeiou]{3,}
1524 .sp
1525 matches at least 3 successive vowels, but may match many more, while
1526 .sp
1527 \ed{8}
1528 .sp
1529 matches exactly 8 digits. An opening curly bracket that appears in a position
1530 where a quantifier is not allowed, or one that does not match the syntax of a
1531 quantifier, is taken as a literal character. For example, {,6} is not a
1532 quantifier, but a literal string of four characters.
1533 .P
1534 In UTF modes, quantifiers apply to characters rather than to individual data
1535 units. Thus, for example, \ex{100}{2} matches two characters, each of
1536 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1537 \eX{3} matches three Unicode extended sequences, each of which may be several
1538 data units long (and they may be of different lengths).
1539 .P
1540 The quantifier {0} is permitted, causing the expression to behave as if the
1541 previous item and the quantifier were not present. This may be useful for
1542 subpatterns that are referenced as
1543 .\" HTML <a href="#subpatternsassubroutines">
1544 .\" </a>
1545 subroutines
1546 .\"
1547 from elsewhere in the pattern (but see also the section entitled
1548 .\" HTML <a href="#subdefine">
1549 .\" </a>
1550 "Defining subpatterns for use by reference only"
1551 .\"
1552 below). Items other than subpatterns that have a {0} quantifier are omitted
1553 from the compiled pattern.
1554 .P
1555 For convenience, the three most common quantifiers have single-character
1556 abbreviations:
1557 .sp
1558 * is equivalent to {0,}
1559 + is equivalent to {1,}
1560 ? is equivalent to {0,1}
1561 .sp
1562 It is possible to construct infinite loops by following a subpattern that can
1563 match no characters with a quantifier that has no upper limit, for example:
1564 .sp
1565 (a?)*
1566 .sp
1567 Earlier versions of Perl and PCRE used to give an error at compile time for
1568 such patterns. However, because there are cases where this can be useful, such
1569 patterns are now accepted, but if any repetition of the subpattern does in fact
1570 match no characters, the loop is forcibly broken.
1571 .P
1572 By default, the quantifiers are "greedy", that is, they match as much as
1573 possible (up to the maximum number of permitted times), without causing the
1574 rest of the pattern to fail. The classic example of where this gives problems
1575 is in trying to match comments in C programs. These appear between /* and */
1576 and within the comment, individual * and / characters may appear. An attempt to
1577 match C comments by applying the pattern
1578 .sp
1579 /\e*.*\e*/
1580 .sp
1581 to the string
1582 .sp
1583 /* first comment */ not comment /* second comment */
1584 .sp
1585 fails, because it matches the entire string owing to the greediness of the .*
1586 item.
1587 .P
1588 However, if a quantifier is followed by a question mark, it ceases to be
1589 greedy, and instead matches the minimum number of times possible, so the
1590 pattern
1591 .sp
1592 /\e*.*?\e*/
1593 .sp
1594 does the right thing with the C comments. The meaning of the various
1595 quantifiers is not otherwise changed, just the preferred number of matches.
1596 Do not confuse this use of question mark with its use as a quantifier in its
1597 own right. Because it has two uses, it can sometimes appear doubled, as in
1598 .sp
1599 \ed??\ed
1600 .sp
1601 which matches one digit by preference, but can match two if that is the only
1602 way the rest of the pattern matches.
1603 .P
1604 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1605 the quantifiers are not greedy by default, but individual ones can be made
1606 greedy by following them with a question mark. In other words, it inverts the
1607 default behaviour.
1608 .P
1609 When a parenthesized subpattern is quantified with a minimum repeat count that
1610 is greater than 1 or with a limited maximum, more memory is required for the
1611 compiled pattern, in proportion to the size of the minimum or maximum.
1612 .P
1613 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1614 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1615 implicitly anchored, because whatever follows will be tried against every
1616 character position in the subject string, so there is no point in retrying the
1617 overall match at any position after the first. PCRE normally treats such a
1618 pattern as though it were preceded by \eA.
1619 .P
1620 In cases where it is known that the subject string contains no newlines, it is
1621 worth setting PCRE_DOTALL in order to obtain this optimization, or
1622 alternatively using ^ to indicate anchoring explicitly.
1623 .P
1624 However, there is one situation where the optimization cannot be used. When .*
1625 is inside capturing parentheses that are the subject of a back reference
1626 elsewhere in the pattern, a match at the start may fail where a later one
1627 succeeds. Consider, for example:
1628 .sp
1629 (.*)abc\e1
1630 .sp
1631 If the subject is "xyz123abc123" the match point is the fourth character. For
1632 this reason, such a pattern is not implicitly anchored.
1633 .P
1634 When a capturing subpattern is repeated, the value captured is the substring
1635 that matched the final iteration. For example, after
1636 .sp
1637 (tweedle[dume]{3}\es*)+
1638 .sp
1639 has matched "tweedledum tweedledee" the value of the captured substring is
1640 "tweedledee". However, if there are nested capturing subpatterns, the
1641 corresponding captured values may have been set in previous iterations. For
1642 example, after
1643 .sp
1644 /(a|(b))+/
1645 .sp
1646 matches "aba" the value of the second captured substring is "b".
1647 .
1648 .
1649 .\" HTML <a name="atomicgroup"></a>
1650 .SH "ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS"
1651 .rs
1652 .sp
1653 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1654 repetition, failure of what follows normally causes the repeated item to be
1655 re-evaluated to see if a different number of repeats allows the rest of the
1656 pattern to match. Sometimes it is useful to prevent this, either to change the
1657 nature of the match, or to cause it fail earlier than it otherwise might, when
1658 the author of the pattern knows there is no point in carrying on.
1659 .P
1660 Consider, for example, the pattern \ed+foo when applied to the subject line
1661 .sp
1662 123456bar
1663 .sp
1664 After matching all 6 digits and then failing to match "foo", the normal
1665 action of the matcher is to try again with only 5 digits matching the \ed+
1666 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1667 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1668 that once a subpattern has matched, it is not to be re-evaluated in this way.
1669 .P
1670 If we use atomic grouping for the previous example, the matcher gives up
1671 immediately on failing to match "foo" the first time. The notation is a kind of
1672 special parenthesis, starting with (?> as in this example:
1673 .sp
1674 (?>\ed+)foo
1675 .sp
1676 This kind of parenthesis "locks up" the part of the pattern it contains once
1677 it has matched, and a failure further into the pattern is prevented from
1678 backtracking into it. Backtracking past it to previous items, however, works as
1679 normal.
1680 .P
1681 An alternative description is that a subpattern of this type matches the string
1682 of characters that an identical standalone pattern would match, if anchored at
1683 the current point in the subject string.
1684 .P
1685 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1686 the above example can be thought of as a maximizing repeat that must swallow
1687 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1688 number of digits they match in order to make the rest of the pattern match,
1689 (?>\ed+) can only match an entire sequence of digits.
1690 .P
1691 Atomic groups in general can of course contain arbitrarily complicated
1692 subpatterns, and can be nested. However, when the subpattern for an atomic
1693 group is just a single repeated item, as in the example above, a simpler
1694 notation, called a "possessive quantifier" can be used. This consists of an
1695 additional + character following a quantifier. Using this notation, the
1696 previous example can be rewritten as
1697 .sp
1698 \ed++foo
1699 .sp
1700 Note that a possessive quantifier can be used with an entire group, for
1701 example:
1702 .sp
1703 (abc|xyz){2,3}+
1704 .sp
1705 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1706 option is ignored. They are a convenient notation for the simpler forms of
1707 atomic group. However, there is no difference in the meaning of a possessive
1708 quantifier and the equivalent atomic group, though there may be a performance
1709 difference; possessive quantifiers should be slightly faster.
1710 .P
1711 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1712 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1713 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1714 package, and PCRE copied it from there. It ultimately found its way into Perl
1715 at release 5.10.
1716 .P
1717 PCRE has an optimization that automatically "possessifies" certain simple
1718 pattern constructs. For example, the sequence A+B is treated as A++B because
1719 there is no point in backtracking into a sequence of A's when B must follow.
1720 .P
1721 When a pattern contains an unlimited repeat inside a subpattern that can itself
1722 be repeated an unlimited number of times, the use of an atomic group is the
1723 only way to avoid some failing matches taking a very long time indeed. The
1724 pattern
1725 .sp
1726 (\eD+|<\ed+>)*[!?]
1727 .sp
1728 matches an unlimited number of substrings that either consist of non-digits, or
1729 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1730 quickly. However, if it is applied to
1731 .sp
1732 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1733 .sp
1734 it takes a long time before reporting failure. This is because the string can
1735 be divided between the internal \eD+ repeat and the external * repeat in a
1736 large number of ways, and all have to be tried. (The example uses [!?] rather
1737 than a single character at the end, because both PCRE and Perl have an
1738 optimization that allows for fast failure when a single character is used. They
1739 remember the last single character that is required for a match, and fail early
1740 if it is not present in the string.) If the pattern is changed so that it uses
1741 an atomic group, like this:
1742 .sp
1743 ((?>\eD+)|<\ed+>)*[!?]
1744 .sp
1745 sequences of non-digits cannot be broken, and failure happens quickly.
1746 .
1747 .
1748 .\" HTML <a name="backreferences"></a>
1749 .SH "BACK REFERENCES"
1750 .rs
1751 .sp
1752 Outside a character class, a backslash followed by a digit greater than 0 (and
1753 possibly further digits) is a back reference to a capturing subpattern earlier
1754 (that is, to its left) in the pattern, provided there have been that many
1755 previous capturing left parentheses.
1756 .P
1757 However, if the decimal number following the backslash is less than 10, it is
1758 always taken as a back reference, and causes an error only if there are not
1759 that many capturing left parentheses in the entire pattern. In other words, the
1760 parentheses that are referenced need not be to the left of the reference for
1761 numbers less than 10. A "forward back reference" of this type can make sense
1762 when a repetition is involved and the subpattern to the right has participated
1763 in an earlier iteration.
1764 .P
1765 It is not possible to have a numerical "forward back reference" to a subpattern
1766 whose number is 10 or more using this syntax because a sequence such as \e50 is
1767 interpreted as a character defined in octal. See the subsection entitled
1768 "Non-printing characters"
1769 .\" HTML <a href="#digitsafterbackslash">
1770 .\" </a>
1771 above
1772 .\"
1773 for further details of the handling of digits following a backslash. There is
1774 no such problem when named parentheses are used. A back reference to any
1775 subpattern is possible using named parentheses (see below).
1776 .P
1777 Another way of avoiding the ambiguity inherent in the use of digits following a
1778 backslash is to use the \eg escape sequence. This escape must be followed by an
1779 unsigned number or a negative number, optionally enclosed in braces. These
1780 examples are all identical:
1781 .sp
1782 (ring), \e1
1783 (ring), \eg1
1784 (ring), \eg{1}
1785 .sp
1786 An unsigned number specifies an absolute reference without the ambiguity that
1787 is present in the older syntax. It is also useful when literal digits follow
1788 the reference. A negative number is a relative reference. Consider this
1789 example:
1790 .sp
1791 (abc(def)ghi)\eg{-1}
1792 .sp
1793 The sequence \eg{-1} is a reference to the most recently started capturing
1794 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1795 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1796 can be helpful in long patterns, and also in patterns that are created by
1797 joining together fragments that contain references within themselves.
1798 .P
1799 A back reference matches whatever actually matched the capturing subpattern in
1800 the current subject string, rather than anything matching the subpattern
1801 itself (see
1802 .\" HTML <a href="#subpatternsassubroutines">
1803 .\" </a>
1804 "Subpatterns as subroutines"
1805 .\"
1806 below for a way of doing that). So the pattern
1807 .sp
1808 (sens|respons)e and \e1ibility
1809 .sp
1810 matches "sense and sensibility" and "response and responsibility", but not
1811 "sense and responsibility". If caseful matching is in force at the time of the
1812 back reference, the case of letters is relevant. For example,
1813 .sp
1814 ((?i)rah)\es+\e1
1815 .sp
1816 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1817 capturing subpattern is matched caselessly.
1818 .P
1819 There are several different ways of writing back references to named
1820 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1821 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1822 back reference syntax, in which \eg can be used for both numeric and named
1823 references, is also supported. We could rewrite the above example in any of
1824 the following ways:
1825 .sp
1826 (?<p1>(?i)rah)\es+\ek<p1>
1827 (?'p1'(?i)rah)\es+\ek{p1}
1828 (?P<p1>(?i)rah)\es+(?P=p1)
1829 (?<p1>(?i)rah)\es+\eg{p1}
1830 .sp
1831 A subpattern that is referenced by name may appear in the pattern before or
1832 after the reference.
1833 .P
1834 There may be more than one back reference to the same subpattern. If a
1835 subpattern has not actually been used in a particular match, any back
1836 references to it always fail by default. For example, the pattern
1837 .sp
1838 (a|(bc))\e2
1839 .sp
1840 always fails if it starts to match "a" rather than "bc". However, if the
1841 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1842 unset value matches an empty string.
1843 .P
1844 Because there may be many capturing parentheses in a pattern, all digits
1845 following a backslash are taken as part of a potential back reference number.
1846 If the pattern continues with a digit character, some delimiter must be used to
1847 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1848 white space. Otherwise, the \eg{ syntax or an empty comment (see
1849 .\" HTML <a href="#comments">
1850 .\" </a>
1851 "Comments"
1852 .\"
1853 below) can be used.
1854 .
1855 .SS "Recursive back references"
1856 .rs
1857 .sp
1858 A back reference that occurs inside the parentheses to which it refers fails
1859 when the subpattern is first used, so, for example, (a\e1) never matches.
1860 However, such references can be useful inside repeated subpatterns. For
1861 example, the pattern
1862 .sp
1863 (a|b\e1)+
1864 .sp
1865 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1866 the subpattern, the back reference matches the character string corresponding
1867 to the previous iteration. In order for this to work, the pattern must be such
1868 that the first iteration does not need to match the back reference. This can be
1869 done using alternation, as in the example above, or by a quantifier with a
1870 minimum of zero.
1871 .P
1872 Back references of this type cause the group that they reference to be treated
1873 as an
1874 .\" HTML <a href="#atomicgroup">
1875 .\" </a>
1876 atomic group.
1877 .\"
1878 Once the whole group has been matched, a subsequent matching failure cannot
1879 cause backtracking into the middle of the group.
1880 .
1881 .
1882 .\" HTML <a name="bigassertions"></a>
1883 .SH ASSERTIONS
1884 .rs
1885 .sp
1886 An assertion is a test on the characters following or preceding the current
1887 matching point that does not actually consume any characters. The simple
1888 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1889 .\" HTML <a href="#smallassertions">
1890 .\" </a>
1891 above.
1892 .\"
1893 .P
1894 More complicated assertions are coded as subpatterns. There are two kinds:
1895 those that look ahead of the current position in the subject string, and those
1896 that look behind it. An assertion subpattern is matched in the normal way,
1897 except that it does not cause the current matching position to be changed.
1898 .P
1899 Assertion subpatterns are not capturing subpatterns. If such an assertion
1900 contains capturing subpatterns within it, these are counted for the purposes of
1901 numbering the capturing subpatterns in the whole pattern. However, substring
1902 capturing is carried out only for positive assertions, because it does not make
1903 sense for negative assertions.
1904 .P
1905 For compatibility with Perl, assertion subpatterns may be repeated; though
1906 it makes no sense to assert the same thing several times, the side effect of
1907 capturing parentheses may occasionally be useful. In practice, there only three
1908 cases:
1909 .sp
1910 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1911 However, it may contain internal capturing parenthesized groups that are called
1912 from elsewhere via the
1913 .\" HTML <a href="#subpatternsassubroutines">
1914 .\" </a>
1915 subroutine mechanism.
1916 .\"
1917 .sp
1918 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1919 were {0,1}. At run time, the rest of the pattern match is tried with and
1920 without the assertion, the order depending on the greediness of the quantifier.
1921 .sp
1922 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1923 The assertion is obeyed just once when encountered during matching.
1924 .
1925 .
1926 .SS "Lookahead assertions"
1927 .rs
1928 .sp
1929 Lookahead assertions start with (?= for positive assertions and (?! for
1930 negative assertions. For example,
1931 .sp
1932 \ew+(?=;)
1933 .sp
1934 matches a word followed by a semicolon, but does not include the semicolon in
1935 the match, and
1936 .sp
1937 foo(?!bar)
1938 .sp
1939 matches any occurrence of "foo" that is not followed by "bar". Note that the
1940 apparently similar pattern
1941 .sp
1942 (?!foo)bar
1943 .sp
1944 does not find an occurrence of "bar" that is preceded by something other than
1945 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1946 (?!foo) is always true when the next three characters are "bar". A
1947 lookbehind assertion is needed to achieve the other effect.
1948 .P
1949 If you want to force a matching failure at some point in a pattern, the most
1950 convenient way to do it is with (?!) because an empty string always matches, so
1951 an assertion that requires there not to be an empty string must always fail.
1952 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1953 .
1954 .
1955 .\" HTML <a name="lookbehind"></a>
1956 .SS "Lookbehind assertions"
1957 .rs
1958 .sp
1959 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1960 negative assertions. For example,
1961 .sp
1962 (?<!foo)bar
1963 .sp
1964 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1965 a lookbehind assertion are restricted such that all the strings it matches must
1966 have a fixed length. However, if there are several top-level alternatives, they
1967 do not all have to have the same fixed length. Thus
1968 .sp
1969 (?<=bullock|donkey)
1970 .sp
1971 is permitted, but
1972 .sp
1973 (?<!dogs?|cats?)
1974 .sp
1975 causes an error at compile time. Branches that match different length strings
1976 are permitted only at the top level of a lookbehind assertion. This is an
1977 extension compared with Perl, which requires all branches to match the same
1978 length of string. An assertion such as
1979 .sp
1980 (?<=ab(c|de))
1981 .sp
1982 is not permitted, because its single top-level branch can match two different
1983 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1984 branches:
1985 .sp
1986 (?<=abc|abde)
1987 .sp
1988 In some cases, the escape sequence \eK
1989 .\" HTML <a href="#resetmatchstart">
1990 .\" </a>
1991 (see above)
1992 .\"
1993 can be used instead of a lookbehind assertion to get round the fixed-length
1994 restriction.
1995 .P
1996 The implementation of lookbehind assertions is, for each alternative, to
1997 temporarily move the current position back by the fixed length and then try to
1998 match. If there are insufficient characters before the current position, the
1999 assertion fails.
2000 .P
2001 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2002 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2003 it impossible to calculate the length of the lookbehind. The \eX and \eR
2004 escapes, which can match different numbers of data units, are also not
2005 permitted.
2006 .P
2007 .\" HTML <a href="#subpatternsassubroutines">
2008 .\" </a>
2009 "Subroutine"
2010 .\"
2011 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2012 as the subpattern matches a fixed-length string.
2013 .\" HTML <a href="#recursion">
2014 .\" </a>
2015 Recursion,
2016 .\"
2017 however, is not supported.
2018 .P
2019 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2020 specify efficient matching of fixed-length strings at the end of subject
2021 strings. Consider a simple pattern such as
2022 .sp
2023 abcd$
2024 .sp
2025 when applied to a long string that does not match. Because matching proceeds
2026 from left to right, PCRE will look for each "a" in the subject and then see if
2027 what follows matches the rest of the pattern. If the pattern is specified as
2028 .sp
2029 ^.*abcd$
2030 .sp
2031 the initial .* matches the entire string at first, but when this fails (because
2032 there is no following "a"), it backtracks to match all but the last character,
2033 then all but the last two characters, and so on. Once again the search for "a"
2034 covers the entire string, from right to left, so we are no better off. However,
2035 if the pattern is written as
2036 .sp
2037 ^.*+(?<=abcd)
2038 .sp
2039 there can be no backtracking for the .*+ item; it can match only the entire
2040 string. The subsequent lookbehind assertion does a single test on the last four
2041 characters. If it fails, the match fails immediately. For long strings, this
2042 approach makes a significant difference to the processing time.
2043 .
2044 .
2045 .SS "Using multiple assertions"
2046 .rs
2047 .sp
2048 Several assertions (of any sort) may occur in succession. For example,
2049 .sp
2050 (?<=\ed{3})(?<!999)foo
2051 .sp
2052 matches "foo" preceded by three digits that are not "999". Notice that each of
2053 the assertions is applied independently at the same point in the subject
2054 string. First there is a check that the previous three characters are all
2055 digits, and then there is a check that the same three characters are not "999".
2056 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2057 of which are digits and the last three of which are not "999". For example, it
2058 doesn't match "123abcfoo". A pattern to do that is
2059 .sp
2060 (?<=\ed{3}...)(?<!999)foo
2061 .sp
2062 This time the first assertion looks at the preceding six characters, checking
2063 that the first three are digits, and then the second assertion checks that the
2064 preceding three characters are not "999".
2065 .P
2066 Assertions can be nested in any combination. For example,
2067 .sp
2068 (?<=(?<!foo)bar)baz
2069 .sp
2070 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2071 preceded by "foo", while
2072 .sp
2073 (?<=\ed{3}(?!999)...)foo
2074 .sp
2075 is another pattern that matches "foo" preceded by three digits and any three
2076 characters that are not "999".
2077 .
2078 .
2079 .\" HTML <a name="conditions"></a>
2080 .SH "CONDITIONAL SUBPATTERNS"
2081 .rs
2082 .sp
2083 It is possible to cause the matching process to obey a subpattern
2084 conditionally or to choose between two alternative subpatterns, depending on
2085 the result of an assertion, or whether a specific capturing subpattern has
2086 already been matched. The two possible forms of conditional subpattern are:
2087 .sp
2088 (?(condition)yes-pattern)
2089 (?(condition)yes-pattern|no-pattern)
2090 .sp
2091 If the condition is satisfied, the yes-pattern is used; otherwise the
2092 no-pattern (if present) is used. If there are more than two alternatives in the
2093 subpattern, a compile-time error occurs. Each of the two alternatives may
2094 itself contain nested subpatterns of any form, including conditional
2095 subpatterns; the restriction to two alternatives applies only at the level of
2096 the condition. This pattern fragment is an example where the alternatives are
2097 complex:
2098 .sp
2099 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2100 .sp
2101 .P
2102 There are four kinds of condition: references to subpatterns, references to
2103 recursion, a pseudo-condition called DEFINE, and assertions.
2104 .
2105 .SS "Checking for a used subpattern by number"
2106 .rs
2107 .sp
2108 If the text between the parentheses consists of a sequence of digits, the
2109 condition is true if a capturing subpattern of that number has previously
2110 matched. If there is more than one capturing subpattern with the same number
2111 (see the earlier
2112 .\"
2113 .\" HTML <a href="#recursion">
2114 .\" </a>
2115 section about duplicate subpattern numbers),
2116 .\"
2117 the condition is true if any of them have matched. An alternative notation is
2118 to precede the digits with a plus or minus sign. In this case, the subpattern
2119 number is relative rather than absolute. The most recently opened parentheses
2120 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2121 loops it can also make sense to refer to subsequent groups. The next
2122 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2123 zero in any of these forms is not used; it provokes a compile-time error.)
2124 .P
2125 Consider the following pattern, which contains non-significant white space to
2126 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2127 three parts for ease of discussion:
2128 .sp
2129 ( \e( )? [^()]+ (?(1) \e) )
2130 .sp
2131 The first part matches an optional opening parenthesis, and if that
2132 character is present, sets it as the first captured substring. The second part
2133 matches one or more characters that are not parentheses. The third part is a
2134 conditional subpattern that tests whether or not the first set of parentheses
2135 matched. If they did, that is, if subject started with an opening parenthesis,
2136 the condition is true, and so the yes-pattern is executed and a closing
2137 parenthesis is required. Otherwise, since no-pattern is not present, the
2138 subpattern matches nothing. In other words, this pattern matches a sequence of
2139 non-parentheses, optionally enclosed in parentheses.
2140 .P
2141 If you were embedding this pattern in a larger one, you could use a relative
2142 reference:
2143 .sp
2144 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2145 .sp
2146 This makes the fragment independent of the parentheses in the larger pattern.
2147 .
2148 .SS "Checking for a used subpattern by name"
2149 .rs
2150 .sp
2151 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2152 subpattern by name. For compatibility with earlier versions of PCRE, which had
2153 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2154 there is a possible ambiguity with this syntax, because subpattern names may
2155 consist entirely of digits. PCRE looks first for a named subpattern; if it
2156 cannot find one and the name consists entirely of digits, PCRE looks for a
2157 subpattern of that number, which must be greater than zero. Using subpattern
2158 names that consist entirely of digits is not recommended.
2159 .P
2160 Rewriting the above example to use a named subpattern gives this:
2161 .sp
2162 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2163 .sp
2164 If the name used in a condition of this kind is a duplicate, the test is
2165 applied to all subpatterns of the same name, and is true if any one of them has
2166 matched.
2167 .
2168 .SS "Checking for pattern recursion"
2169 .rs
2170 .sp
2171 If the condition is the string (R), and there is no subpattern with the name R,
2172 the condition is true if a recursive call to the whole pattern or any
2173 subpattern has been made. If digits or a name preceded by ampersand follow the
2174 letter R, for example:
2175 .sp
2176 (?(R3)...) or (?(R&name)...)
2177 .sp
2178 the condition is true if the most recent recursion is into a subpattern whose
2179 number or name is given. This condition does not check the entire recursion
2180 stack. If the name used in a condition of this kind is a duplicate, the test is
2181 applied to all subpatterns of the same name, and is true if any one of them is
2182 the most recent recursion.
2183 .P
2184 At "top level", all these recursion test conditions are false.
2185 .\" HTML <a href="#recursion">
2186 .\" </a>
2187 The syntax for recursive patterns
2188 .\"
2189 is described below.
2190 .
2191 .\" HTML <a name="subdefine"></a>
2192 .SS "Defining subpatterns for use by reference only"
2193 .rs
2194 .sp
2195 If the condition is the string (DEFINE), and there is no subpattern with the
2196 name DEFINE, the condition is always false. In this case, there may be only one
2197 alternative in the subpattern. It is always skipped if control reaches this
2198 point in the pattern; the idea of DEFINE is that it can be used to define
2199 subroutines that can be referenced from elsewhere. (The use of
2200 .\" HTML <a href="#subpatternsassubroutines">
2201 .\" </a>
2202 subroutines
2203 .\"
2204 is described below.) For example, a pattern to match an IPv4 address such as
2205 "192.168.23.245" could be written like this (ignore white space and line
2206 breaks):
2207 .sp
2208 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2209 \eb (?&byte) (\e.(?&byte)){3} \eb
2210 .sp
2211 The first part of the pattern is a DEFINE group inside which a another group
2212 named "byte" is defined. This matches an individual component of an IPv4
2213 address (a number less than 256). When matching takes place, this part of the
2214 pattern is skipped because DEFINE acts like a false condition. The rest of the
2215 pattern uses references to the named group to match the four dot-separated
2216 components of an IPv4 address, insisting on a word boundary at each end.
2217 .
2218 .SS "Assertion conditions"
2219 .rs
2220 .sp
2221 If the condition is not in any of the above formats, it must be an assertion.
2222 This may be a positive or negative lookahead or lookbehind assertion. Consider
2223 this pattern, again containing non-significant white space, and with the two
2224 alternatives on the second line:
2225 .sp
2226 (?(?=[^a-z]*[a-z])
2227 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2228 .sp
2229 The condition is a positive lookahead assertion that matches an optional
2230 sequence of non-letters followed by a letter. In other words, it tests for the
2231 presence of at least one letter in the subject. If a letter is found, the
2232 subject is matched against the first alternative; otherwise it is matched
2233 against the second. This pattern matches strings in one of the two forms
2234 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2235 .
2236 .
2237 .\" HTML <a name="comments"></a>
2238 .SH COMMENTS
2239 .rs
2240 .sp
2241 There are two ways of including comments in patterns that are processed by
2242 PCRE. In both cases, the start of the comment must not be in a character class,
2243 nor in the middle of any other sequence of related characters such as (?: or a
2244 subpattern name or number. The characters that make up a comment play no part
2245 in the pattern matching.
2246 .P
2247 The sequence (?# marks the start of a comment that continues up to the next
2248 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2249 option is set, an unescaped # character also introduces a comment, which in
2250 this case continues to immediately after the next newline character or
2251 character sequence in the pattern. Which characters are interpreted as newlines
2252 is controlled by the options passed to a compiling function or by a special
2253 sequence at the start of the pattern, as described in the section entitled
2254 .\" HTML <a href="#newlines">
2255 .\" </a>
2256 "Newline conventions"
2257 .\"
2258 above. Note that the end of this type of comment is a literal newline sequence
2259 in the pattern; escape sequences that happen to represent a newline do not
2260 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2261 default newline convention is in force:
2262 .sp
2263 abc #comment \en still comment
2264 .sp
2265 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2266 a newline in the pattern. The sequence \en is still literal at this stage, so
2267 it does not terminate the comment. Only an actual character with the code value
2268 0x0a (the default newline) does so.
2269 .
2270 .
2271 .\" HTML <a name="recursion"></a>
2272 .SH "RECURSIVE PATTERNS"
2273 .rs
2274 .sp
2275 Consider the problem of matching a string in parentheses, allowing for
2276 unlimited nested parentheses. Without the use of recursion, the best that can
2277 be done is to use a pattern that matches up to some fixed depth of nesting. It
2278 is not possible to handle an arbitrary nesting depth.
2279 .P
2280 For some time, Perl has provided a facility that allows regular expressions to
2281 recurse (amongst other things). It does this by interpolating Perl code in the
2282 expression at run time, and the code can refer to the expression itself. A Perl
2283 pattern using code interpolation to solve the parentheses problem can be
2284 created like this:
2285 .sp
2286 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2287 .sp
2288 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2289 recursively to the pattern in which it appears.
2290 .P
2291 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2292 supports special syntax for recursion of the entire pattern, and also for
2293 individual subpattern recursion. After its introduction in PCRE and Python,
2294 this kind of recursion was subsequently introduced into Perl at release 5.10.
2295 .P
2296 A special item that consists of (? followed by a number greater than zero and a
2297 closing parenthesis is a recursive subroutine call of the subpattern of the
2298 given number, provided that it occurs inside that subpattern. (If not, it is a
2299 .\" HTML <a href="#subpatternsassubroutines">
2300 .\" </a>
2301 non-recursive subroutine
2302 .\"
2303 call, which is described in the next section.) The special item (?R) or (?0) is
2304 a recursive call of the entire regular expression.
2305 .P
2306 This PCRE pattern solves the nested parentheses problem (assume the
2307 PCRE_EXTENDED option is set so that white space is ignored):
2308 .sp
2309 \e( ( [^()]++ | (?R) )* \e)
2310 .sp
2311 First it matches an opening parenthesis. Then it matches any number of
2312 substrings which can either be a sequence of non-parentheses, or a recursive
2313 match of the pattern itself (that is, a correctly parenthesized substring).
2314 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2315 to avoid backtracking into sequences of non-parentheses.
2316 .P
2317 If this were part of a larger pattern, you would not want to recurse the entire
2318 pattern, so instead you could use this:
2319 .sp
2320 ( \e( ( [^()]++ | (?1) )* \e) )
2321 .sp
2322 We have put the pattern into parentheses, and caused the recursion to refer to
2323 them instead of the whole pattern.
2324 .P
2325 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2326 is made easier by the use of relative references. Instead of (?1) in the
2327 pattern above you can write (?-2) to refer to the second most recently opened
2328 parentheses preceding the recursion. In other words, a negative number counts
2329 capturing parentheses leftwards from the point at which it is encountered.
2330 .P
2331 It is also possible to refer to subsequently opened parentheses, by writing
2332 references such as (?+2). However, these cannot be recursive because the
2333 reference is not inside the parentheses that are referenced. They are always
2334 .\" HTML <a href="#subpatternsassubroutines">
2335 .\" </a>
2336 non-recursive subroutine
2337 .\"
2338 calls, as described in the next section.
2339 .P
2340 An alternative approach is to use named parentheses instead. The Perl syntax
2341 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2342 could rewrite the above example as follows:
2343 .sp
2344 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2345 .sp
2346 If there is more than one subpattern with the same name, the earliest one is
2347 used.
2348 .P
2349 This particular example pattern that we have been looking at contains nested
2350 unlimited repeats, and so the use of a possessive quantifier for matching
2351 strings of non-parentheses is important when applying the pattern to strings
2352 that do not match. For example, when this pattern is applied to
2353 .sp
2354 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2355 .sp
2356 it yields "no match" quickly. However, if a possessive quantifier is not used,
2357 the match runs for a very long time indeed because there are so many different
2358 ways the + and * repeats can carve up the subject, and all have to be tested
2359 before failure can be reported.
2360 .P
2361 At the end of a match, the values of capturing parentheses are those from
2362 the outermost level. If you want to obtain intermediate values, a callout
2363 function can be used (see below and the
2364 .\" HREF
2365 \fBpcrecallout\fP
2366 .\"
2367 documentation). If the pattern above is matched against
2368 .sp
2369 (ab(cd)ef)
2370 .sp
2371 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2372 the last value taken on at the top level. If a capturing subpattern is not
2373 matched at the top level, its final captured value is unset, even if it was
2374 (temporarily) set at a deeper level during the matching process.
2375 .P
2376 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2377 obtain extra memory to store data during a recursion, which it does by using
2378 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2379 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2380 .P
2381 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2382 Consider this pattern, which matches text in angle brackets, allowing for
2383 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2384 recursing), whereas any characters are permitted at the outer level.
2385 .sp
2386 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2387 .sp
2388 In this pattern, (?(R) is the start of a conditional subpattern, with two
2389 different alternatives for the recursive and non-recursive cases. The (?R) item
2390 is the actual recursive call.
2391 .
2392 .
2393 .\" HTML <a name="recursiondifference"></a>
2394 .SS "Differences in recursion processing between PCRE and Perl"
2395 .rs
2396 .sp
2397 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2398 (like Python, but unlike Perl), a recursive subpattern call is always treated
2399 as an atomic group. That is, once it has matched some of the subject string, it
2400 is never re-entered, even if it contains untried alternatives and there is a
2401 subsequent matching failure. This can be illustrated by the following pattern,
2402 which purports to match a palindromic string that contains an odd number of
2403 characters (for example, "a", "aba", "abcba", "abcdcba"):
2404 .sp
2405 ^(.|(.)(?1)\e2)$
2406 .sp
2407 The idea is that it either matches a single character, or two identical
2408 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2409 it does not if the pattern is longer than three characters. Consider the
2410 subject string "abcba":
2411 .P
2412 At the top level, the first character is matched, but as it is not at the end
2413 of the string, the first alternative fails; the second alternative is taken
2414 and the recursion kicks in. The recursive call to subpattern 1 successfully
2415 matches the next character ("b"). (Note that the beginning and end of line
2416 tests are not part of the recursion).
2417 .P
2418 Back at the top level, the next character ("c") is compared with what
2419 subpattern 2 matched, which was "a". This fails. Because the recursion is
2420 treated as an atomic group, there are now no backtracking points, and so the
2421 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2422 try the second alternative.) However, if the pattern is written with the
2423 alternatives in the other order, things are different:
2424 .sp
2425 ^((.)(?1)\e2|.)$
2426 .sp
2427 This time, the recursing alternative is tried first, and continues to recurse
2428 until it runs out of characters, at which point the recursion fails. But this
2429 time we do have another alternative to try at the higher level. That is the big
2430 difference: in the previous case the remaining alternative is at a deeper
2431 recursion level, which PCRE cannot use.
2432 .P
2433 To change the pattern so that it matches all palindromic strings, not just
2434 those with an odd number of characters, it is tempting to change the pattern to
2435 this:
2436 .sp
2437 ^((.)(?1)\e2|.?)$
2438 .sp
2439 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2440 deeper recursion has matched a single character, it cannot be entered again in
2441 order to match an empty string. The solution is to separate the two cases, and
2442 write out the odd and even cases as alternatives at the higher level:
2443 .sp
2444 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2445 .sp
2446 If you want to match typical palindromic phrases, the pattern has to ignore all
2447 non-word characters, which can be done like this:
2448 .sp
2449 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2450 .sp
2451 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2452 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2453 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2454 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2455 more) to match typical phrases, and Perl takes so long that you think it has
2456 gone into a loop.
2457 .P
2458 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2459 string does not start with a palindrome that is shorter than the entire string.
2460 For example, although "abcba" is correctly matched, if the subject is "ababa",
2461 PCRE finds the palindrome "aba" at the start, then fails at top level because
2462 the end of the string does not follow. Once again, it cannot jump back into the
2463 recursion to try other alternatives, so the entire match fails.
2464 .P
2465 The second way in which PCRE and Perl differ in their recursion processing is
2466 in the handling of captured values. In Perl, when a subpattern is called
2467 recursively or as a subpattern (see the next section), it has no access to any
2468 values that were captured outside the recursion, whereas in PCRE these values
2469 can be referenced. Consider this pattern:
2470 .sp
2471 ^(.)(\e1|a(?2))
2472 .sp
2473 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2474 then in the second group, when the back reference \e1 fails to match "b", the
2475 second alternative matches "a" and then recurses. In the recursion, \e1 does
2476 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2477 match because inside the recursive call \e1 cannot access the externally set
2478 value.
2479 .
2480 .
2481 .\" HTML <a name="subpatternsassubroutines"></a>
2482 .SH "SUBPATTERNS AS SUBROUTINES"
2483 .rs
2484 .sp
2485 If the syntax for a recursive subpattern call (either by number or by
2486 name) is used outside the parentheses to which it refers, it operates like a
2487 subroutine in a programming language. The called subpattern may be defined
2488 before or after the reference. A numbered reference can be absolute or
2489 relative, as in these examples:
2490 .sp
2491 (...(absolute)...)...(?2)...
2492 (...(relative)...)...(?-1)...
2493 (...(?+1)...(relative)...
2494 .sp
2495 An earlier example pointed out that the pattern
2496 .sp
2497 (sens|respons)e and \e1ibility
2498 .sp
2499 matches "sense and sensibility" and "response and responsibility", but not
2500 "sense and responsibility". If instead the pattern
2501 .sp
2502 (sens|respons)e and (?1)ibility
2503 .sp
2504 is used, it does match "sense and responsibility" as well as the other two
2505 strings. Another example is given in the discussion of DEFINE above.
2506 .P
2507 All subroutine calls, whether recursive or not, are always treated as atomic
2508 groups. That is, once a subroutine has matched some of the subject string, it
2509 is never re-entered, even if it contains untried alternatives and there is a
2510 subsequent matching failure. Any capturing parentheses that are set during the
2511 subroutine call revert to their previous values afterwards.
2512 .P
2513 Processing options such as case-independence are fixed when a subpattern is
2514 defined, so if it is used as a subroutine, such options cannot be changed for
2515 different calls. For example, consider this pattern:
2516 .sp
2517 (abc)(?i:(?-1))
2518 .sp
2519 It matches "abcabc". It does not match "abcABC" because the change of
2520 processing option does not affect the called subpattern.
2521 .
2522 .
2523 .\" HTML <a name="onigurumasubroutines"></a>
2524 .SH "ONIGURUMA SUBROUTINE SYNTAX"
2525 .rs
2526 .sp
2527 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2528 a number enclosed either in angle brackets or single quotes, is an alternative
2529 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2530 are two of the examples used above, rewritten using this syntax:
2531 .sp
2532 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2533 (sens|respons)e and \eg'1'ibility
2534 .sp
2535 PCRE supports an extension to Oniguruma: if a number is preceded by a
2536 plus or a minus sign it is taken as a relative reference. For example:
2537 .sp
2538 (abc)(?i:\eg<-1>)
2539 .sp
2540 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2541 synonymous. The former is a back reference; the latter is a subroutine call.
2542 .
2543 .
2544 .SH CALLOUTS
2545 .rs
2546 .sp
2547 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2548 code to be obeyed in the middle of matching a regular expression. This makes it
2549 possible, amongst other things, to extract different substrings that match the
2550 same pair of parentheses when there is a repetition.
2551 .P
2552 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2553 code. The feature is called "callout". The caller of PCRE provides an external
2554 function by putting its entry point in the global variable \fIpcre_callout\fP
2555 (8-bit library) or \fIpcre16_callout\fP (16-bit library). By default, this
2556 variable contains NULL, which disables all calling out.
2557 .P
2558 Within a regular expression, (?C) indicates the points at which the external
2559 function is to be called. If you want to identify different callout points, you
2560 can put a number less than 256 after the letter C. The default value is zero.
2561 For example, this pattern has two callout points:
2562 .sp
2563 (?C1)abc(?C2)def
2564 .sp
2565 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2566 automatically installed before each item in the pattern. They are all numbered
2567 255.
2568 .P
2569 During matching, when PCRE reaches a callout point, the external function is
2570 called. It is provided with the number of the callout, the position in the
2571 pattern, and, optionally, one item of data originally supplied by the caller of
2572 the matching function. The callout function may cause matching to proceed, to
2573 backtrack, or to fail altogether. A complete description of the interface to
2574 the callout function is given in the
2575 .\" HREF
2576 \fBpcrecallout\fP
2577 .\"
2578 documentation.
2579 .
2580 .
2581 .\" HTML <a name="backtrackcontrol"></a>
2582 .SH "BACKTRACKING CONTROL"
2583 .rs
2584 .sp
2585 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2586 are described in the Perl documentation as "experimental and subject to change
2587 or removal in a future version of Perl". It goes on to say: "Their usage in
2588 production code should be noted to avoid problems during upgrades." The same
2589 remarks apply to the PCRE features described in this section.
2590 .P
2591 Since these verbs are specifically related to backtracking, most of them can be
2592 used only when the pattern is to be matched using one of the traditional
2593 matching functions, which use a backtracking algorithm. With the exception of
2594 (*FAIL), which behaves like a failing negative assertion, they cause an error
2595 if encountered by a DFA matching function.
2596 .P
2597 If any of these verbs are used in an assertion or in a subpattern that is
2598 called as a subroutine (whether or not recursively), their effect is confined
2599 to that subpattern; it does not extend to the surrounding pattern, with one
2600 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2601 a successful positive assertion \fIis\fP passed back when a match succeeds
2602 (compare capturing parentheses in assertions). Note that such subpatterns are
2603 processed as anchored at the point where they are tested. Note also that Perl's
2604 treatment of subroutines and assertions is different in some cases.
2605 .P
2606 The new verbs make use of what was previously invalid syntax: an opening
2607 parenthesis followed by an asterisk. They are generally of the form
2608 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2609 depending on whether or not an argument is present. A name is any sequence of
2610 characters that does not include a closing parenthesis. The maximum length of
2611 name is 255 in the 8-bit library and 65535 in the 16-bit library. If the name
2612 is empty, that is, if the closing parenthesis immediately follows the colon,
2613 the effect is as if the colon were not there. Any number of these verbs may
2614 occur in a pattern.
2615 .
2616 .
2617 .\" HTML <a name="nooptimize"></a>
2618 .SS "Optimizations that affect backtracking verbs"
2619 .rs
2620 .sp
2621 PCRE contains some optimizations that are used to speed up matching by running
2622 some checks at the start of each match attempt. For example, it may know the
2623 minimum length of matching subject, or that a particular character must be
2624 present. When one of these optimizations suppresses the running of a match, any
2625 included backtracking verbs will not, of course, be processed. You can suppress
2626 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2627 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2628 pattern with (*NO_START_OPT). There is more discussion of this option in the
2629 section entitled
2630 .\" HTML <a href="pcreapi.html#execoptions">
2631 .\" </a>
2632 "Option bits for \fBpcre_exec()\fP"
2633 .\"
2634 in the
2635 .\" HREF
2636 \fBpcreapi\fP
2637 .\"
2638 documentation.
2639 .P
2640 Experiments with Perl suggest that it too has similar optimizations, sometimes
2641 leading to anomalous results.
2642 .
2643 .
2644 .SS "Verbs that act immediately"
2645 .rs
2646 .sp
2647 The following verbs act as soon as they are encountered. They may not be
2648 followed by a name.
2649 .sp
2650 (*ACCEPT)
2651 .sp
2652 This verb causes the match to end successfully, skipping the remainder of the
2653 pattern. However, when it is inside a subpattern that is called as a
2654 subroutine, only that subpattern is ended successfully. Matching then continues
2655 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2656 far is captured. For example:
2657 .sp
2658 A((?:A|B(*ACCEPT)|C)D)
2659 .sp
2660 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2661 the outer parentheses.
2662 .sp
2663 (*FAIL) or (*F)
2664 .sp
2665 This verb causes a matching failure, forcing backtracking to occur. It is
2666 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2667 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2668 Perl features that are not present in PCRE. The nearest equivalent is the
2669 callout feature, as for example in this pattern:
2670 .sp
2671 a+(?C)(*FAIL)
2672 .sp
2673 A match with the string "aaaa" always fails, but the callout is taken before
2674 each backtrack happens (in this example, 10 times).
2675 .
2676 .
2677 .SS "Recording which path was taken"
2678 .rs
2679 .sp
2680 There is one verb whose main purpose is to track how a match was arrived at,
2681 though it also has a secondary use in conjunction with advancing the match
2682 starting point (see (*SKIP) below).
2683 .sp
2684 (*MARK:NAME) or (*:NAME)
2685 .sp
2686 A name is always required with this verb. There may be as many instances of
2687 (*MARK) as you like in a pattern, and their names do not have to be unique.
2688 .P
2689 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2690 path is passed back to the caller as described in the section entitled
2691 .\" HTML <a href="pcreapi.html#extradata">
2692 .\" </a>
2693 "Extra data for \fBpcre_exec()\fP"
2694 .\"
2695 in the
2696 .\" HREF
2697 \fBpcreapi\fP
2698 .\"
2699 documentation. Here is an example of \fBpcretest\fP output, where the /K
2700 modifier requests the retrieval and outputting of (*MARK) data:
2701 .sp
2702 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2703 data> XY
2704 0: XY
2705 MK: A
2706 XZ
2707 0: XZ
2708 MK: B
2709 .sp
2710 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2711 indicates which of the two alternatives matched. This is a more efficient way
2712 of obtaining this information than putting each alternative in its own
2713 capturing parentheses.
2714 .P
2715 If (*MARK) is encountered in a positive assertion, its name is recorded and
2716 passed back if it is the last-encountered. This does not happen for negative
2717 assertions.
2718 .P
2719 After a partial match or a failed match, the name of the last encountered
2720 (*MARK) in the entire match process is returned. For example:
2721 .sp
2722 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2723 data> XP
2724 No match, mark = B
2725 .sp
2726 Note that in this unanchored example the mark is retained from the match
2727 attempt that started at the letter "X" in the subject. Subsequent match
2728 attempts starting at "P" and then with an empty string do not get as far as the
2729 (*MARK) item, but nevertheless do not reset it.
2730 .P
2731 If you are interested in (*MARK) values after failed matches, you should
2732 probably set the PCRE_NO_START_OPTIMIZE option
2733 .\" HTML <a href="#nooptimize">
2734 .\" </a>
2735 (see above)
2736 .\"
2737 to ensure that the match is always attempted.
2738 .
2739 .
2740 .SS "Verbs that act after backtracking"
2741 .rs
2742 .sp
2743 The following verbs do nothing when they are encountered. Matching continues
2744 with what follows, but if there is no subsequent match, causing a backtrack to
2745 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2746 the verb. However, when one of these verbs appears inside an atomic group, its
2747 effect is confined to that group, because once the group has been matched,
2748 there is never any backtracking into it. In this situation, backtracking can
2749 "jump back" to the left of the entire atomic group. (Remember also, as stated
2750 above, that this localization also applies in subroutine calls and assertions.)
2751 .P
2752 These verbs differ in exactly what kind of failure occurs when backtracking
2753 reaches them.
2754 .sp
2755 (*COMMIT)
2756 .sp
2757 This verb, which may not be followed by a name, causes the whole match to fail
2758 outright if the rest of the pattern does not match. Even if the pattern is
2759 unanchored, no further attempts to find a match by advancing the starting point
2760 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2761 finding a match at the current starting point, or not at all. For example:
2762 .sp
2763 a+(*COMMIT)b
2764 .sp
2765 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2766 dynamic anchor, or "I've started, so I must finish." The name of the most
2767 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2768 match failure.
2769 .P
2770 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2771 unless PCRE's start-of-match optimizations are turned off, as shown in this
2772 \fBpcretest\fP example:
2773 .sp
2774 re> /(*COMMIT)abc/
2775 data> xyzabc
2776 0: abc
2777 xyzabc\eY
2778 No match
2779 .sp
2780 PCRE knows that any match must start with "a", so the optimization skips along
2781 the subject to "a" before running the first match attempt, which succeeds. When
2782 the optimization is disabled by the \eY escape in the second subject, the match
2783 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2784 starting points.
2785 .sp
2786 (*PRUNE) or (*PRUNE:NAME)
2787 .sp
2788 This verb causes the match to fail at the current starting position in the
2789 subject if the rest of the pattern does not match. If the pattern is
2790 unanchored, the normal "bumpalong" advance to the next starting character then
2791 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2792 reached, or when matching to the right of (*PRUNE), but if there is no match to
2793 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2794 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2795 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2796 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2797 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2798 .sp
2799 (*SKIP)
2800 .sp
2801 This verb, when given without a name, is like (*PRUNE), except that if the
2802 pattern is unanchored, the "bumpalong" advance is not to the next character,
2803 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2804 signifies that whatever text was matched leading up to it cannot be part of a
2805 successful match. Consider:
2806 .sp
2807 a+(*SKIP)b
2808 .sp
2809 If the subject is "aaaac...", after the first match attempt fails (starting at
2810 the first character in the string), the starting point skips on to start the
2811 next attempt at "c". Note that a possessive quantifer does not have the same
2812 effect as this example; although it would suppress backtracking during the
2813 first match attempt, the second attempt would start at the second character
2814 instead of skipping on to "c".
2815 .sp
2816 (*SKIP:NAME)
2817 .sp
2818 When (*SKIP) has an associated name, its behaviour is modified. If the
2819 following pattern fails to match, the previous path through the pattern is
2820 searched for the most recent (*MARK) that has the same name. If one is found,
2821 the "bumpalong" advance is to the subject position that corresponds to that
2822 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2823 matching name is found, the (*SKIP) is ignored.
2824 .sp
2825 (*THEN) or (*THEN:NAME)
2826 .sp
2827 This verb causes a skip to the next innermost alternative if the rest of the
2828 pattern does not match. That is, it cancels pending backtracking, but only
2829 within the current alternative. Its name comes from the observation that it can
2830 be used for a pattern-based if-then-else block:
2831 .sp
2832 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2833 .sp
2834 If the COND1 pattern matches, FOO is tried (and possibly further items after
2835 the end of the group if FOO succeeds); on failure, the matcher skips to the
2836 second alternative and tries COND2, without backtracking into COND1. The
2837 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2838 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2839 .P
2840 Note that a subpattern that does not contain a | character is just a part of
2841 the enclosing alternative; it is not a nested alternation with only one
2842 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2843 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2844 pattern fragments that do not contain any | characters at this level:
2845 .sp
2846 A (B(*THEN)C) | D
2847 .sp
2848 If A and B are matched, but there is a failure in C, matching does not
2849 backtrack into A; instead it moves to the next alternative, that is, D.
2850 However, if the subpattern containing (*THEN) is given an alternative, it
2851 behaves differently:
2852 .sp
2853 A (B(*THEN)C | (*FAIL)) | D
2854 .sp
2855 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2856 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2857 because there are no more alternatives to try. In this case, matching does now
2858 backtrack into A.
2859 .P
2860 Note also that a conditional subpattern is not considered as having two
2861 alternatives, because only one is ever used. In other words, the | character in
2862 a conditional subpattern has a different meaning. Ignoring white space,
2863 consider:
2864 .sp
2865 ^.*? (?(?=a) a | b(*THEN)c )
2866 .sp
2867 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2868 it initially matches zero characters. The condition (?=a) then fails, the
2869 character "b" is matched, but "c" is not. At this point, matching does not
2870 backtrack to .*? as might perhaps be expected from the presence of the |
2871 character. The conditional subpattern is part of the single alternative that
2872 comprises the whole pattern, and so the match fails. (If there was a backtrack
2873 into .*?, allowing it to match "b", the match would succeed.)
2874 .P
2875 The verbs just described provide four different "strengths" of control when
2876 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2877 next alternative. (*PRUNE) comes next, failing the match at the current
2878 starting position, but allowing an advance to the next character (for an
2879 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2880 than one character. (*COMMIT) is the strongest, causing the entire match to
2881 fail.
2882 .P
2883 If more than one such verb is present in a pattern, the "strongest" one wins.
2884 For example, consider this pattern, where A, B, etc. are complex pattern
2885 fragments:
2886 .sp
2887 (A(*COMMIT)B(*THEN)C|D)
2888 .sp
2889 Once A has matched, PCRE is committed to this match, at the current starting
2890 position. If subsequently B matches, but C does not, the normal (*THEN) action
2891 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2892 overrides.
2893 .
2894 .
2895 .SH "SEE ALSO"
2896 .rs
2897 .sp
2898 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2899 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP.
2900 .
2901 .
2902 .SH AUTHOR
2903 .rs
2904 .sp
2905 .nf
2906 Philip Hazel
2907 University Computing Service
2908 Cambridge CB2 3QH, England.
2909 .fi
2910 .
2911 .
2912 .SH REVISION
2913 .rs
2914 .sp
2915 .nf
2916 Last updated: 17 June 2012
2917 Copyright (c) 1997-2012 University of Cambridge.
2918 .fi

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5