ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1033 - (show annotations)
Mon Sep 10 11:02:48 2012 UTC (9 years ago) by ph10
File size: 123344 byte(s)
Error occurred while calculating annotation data.
General spring-clean of EBCDIC-related issues in the code, which had decayed 
over time. Also the documentation. Added one test that can be run in an ASCII
world to do a little testing of EBCDIC-related things. 
1 .TH PCREPATTERN 3 "10 September 2012" "PCRE 8.31"
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13 also supports some alternative regular expression syntax (which does not
14 conflict with the Perl syntax) in order to provide some compatibility with
15 regular expressions in Python, .NET, and Oniguruma.
16 .P
17 Perl's regular expressions are described in its own documentation, and
18 regular expressions in general are covered in a number of books, some of which
19 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20 published by O'Reilly, covers regular expressions in great detail. This
21 description of PCRE's regular expressions is intended as reference material.
22 .P
23 The original operation of PCRE was on strings of one-byte characters. However,
24 there is now also support for UTF-8 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 .
73 .rs
74 .sp
75 PCRE can be compiled to run in an environment that uses EBCDIC as its character
76 code rather than ASCII or Unicode (typically a mainframe system). In the
77 sections below, character code values are ASCII or Unicode; in an EBCDIC
78 environment these characters may have different code values, and there are no
79 code points greater than 255.
80 .
81 .
82 .\" HTML <a name="newlines"></a>
84 .rs
85 .sp
86 PCRE supports five different conventions for indicating line breaks in
87 strings: a single CR (carriage return) character, a single LF (linefeed)
88 character, the two-character sequence CRLF, any of the three preceding, or any
89 Unicode newline sequence. The
90 .\" HREF
91 \fBpcreapi\fP
92 .\"
93 page has
94 .\" HTML <a href="pcreapi.html#newlines">
95 .\" </a>
96 further discussion
97 .\"
98 about newlines, and shows how to set the newline convention in the
99 \fIoptions\fP arguments for the compiling and matching functions.
100 .P
101 It is also possible to specify a newline convention by starting a pattern
102 string with one of the following five sequences:
103 .sp
104 (*CR) carriage return
105 (*LF) linefeed
106 (*CRLF) carriage return, followed by linefeed
107 (*ANYCRLF) any of the three above
108 (*ANY) all Unicode newline sequences
109 .sp
110 These override the default and the options given to the compiling function. For
111 example, on a Unix system where LF is the default newline sequence, the pattern
112 .sp
113 (*CR)a.b
114 .sp
115 changes the convention to CR. That pattern matches "a\enb" because LF is no
116 longer a newline. Note that these special settings, which are not
117 Perl-compatible, are recognized only at the very start of a pattern, and that
118 they must be in upper case. If more than one of them is present, the last one
119 is used.
120 .P
121 The newline convention affects the interpretation of the dot metacharacter when
122 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
123 affect what the \eR escape sequence matches. By default, this is any Unicode
124 newline sequence, for Perl compatibility. However, this can be changed; see the
125 description of \eR in the section entitled
126 .\" HTML <a href="#newlineseq">
127 .\" </a>
128 "Newline sequences"
129 .\"
130 below. A change of \eR setting can be combined with a change of newline
131 convention.
132 .
133 .
135 .rs
136 .sp
137 A regular expression is a pattern that is matched against a subject string from
138 left to right. Most characters stand for themselves in a pattern, and match the
139 corresponding characters in the subject. As a trivial example, the pattern
140 .sp
141 The quick brown fox
142 .sp
143 matches a portion of a subject string that is identical to itself. When
144 caseless matching is specified (the PCRE_CASELESS option), letters are matched
145 independently of case. In a UTF mode, PCRE always understands the concept of
146 case for characters whose values are less than 128, so caseless matching is
147 always possible. For characters with higher values, the concept of case is
148 supported if PCRE is compiled with Unicode property support, but not otherwise.
149 If you want to use caseless matching for characters 128 and above, you must
150 ensure that PCRE is compiled with Unicode property support as well as with
151 UTF support.
152 .P
153 The power of regular expressions comes from the ability to include alternatives
154 and repetitions in the pattern. These are encoded in the pattern by the use of
155 \fImetacharacters\fP, which do not stand for themselves but instead are
156 interpreted in some special way.
157 .P
158 There are two different sets of metacharacters: those that are recognized
159 anywhere in the pattern except within square brackets, and those that are
160 recognized within square brackets. Outside square brackets, the metacharacters
161 are as follows:
162 .sp
163 \e general escape character with several uses
164 ^ assert start of string (or line, in multiline mode)
165 $ assert end of string (or line, in multiline mode)
166 . match any character except newline (by default)
167 [ start character class definition
168 | start of alternative branch
169 ( start subpattern
170 ) end subpattern
171 ? extends the meaning of (
172 also 0 or 1 quantifier
173 also quantifier minimizer
174 * 0 or more quantifier
175 + 1 or more quantifier
176 also "possessive quantifier"
177 { start min/max quantifier
178 .sp
179 Part of a pattern that is in square brackets is called a "character class". In
180 a character class the only metacharacters are:
181 .sp
182 \e general escape character
183 ^ negate the class, but only if the first character
184 - indicates character range
185 .\" JOIN
186 [ POSIX character class (only if followed by POSIX
187 syntax)
188 ] terminates the character class
189 .sp
190 The following sections describe the use of each of the metacharacters.
191 .
192 .
194 .rs
195 .sp
196 The backslash character has several uses. Firstly, if it is followed by a
197 character that is not a number or a letter, it takes away any special meaning
198 that character may have. This use of backslash as an escape character applies
199 both inside and outside character classes.
200 .P
201 For example, if you want to match a * character, you write \e* in the pattern.
202 This escaping action applies whether or not the following character would
203 otherwise be interpreted as a metacharacter, so it is always safe to precede a
204 non-alphanumeric with backslash to specify that it stands for itself. In
205 particular, if you want to match a backslash, you write \e\e.
206 .P
207 In a UTF mode, only ASCII numbers and letters have any special meaning after a
208 backslash. All other characters (in particular, those whose codepoints are
209 greater than 127) are treated as literals.
210 .P
211 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
212 pattern (other than in a character class) and characters between a # outside
213 a character class and the next newline are ignored. An escaping backslash can
214 be used to include a white space or # character as part of the pattern.
215 .P
216 If you want to remove the special meaning from a sequence of characters, you
217 can do so by putting them between \eQ and \eE. This is different from Perl in
218 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
219 Perl, $ and @ cause variable interpolation. Note the following examples:
220 .sp
221 Pattern PCRE matches Perl matches
222 .sp
223 .\" JOIN
224 \eQabc$xyz\eE abc$xyz abc followed by the
225 contents of $xyz
226 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
227 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
228 .sp
229 The \eQ...\eE sequence is recognized both inside and outside character classes.
230 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
231 by \eE later in the pattern, the literal interpretation continues to the end of
232 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
233 a character class, this causes an error, because the character class is not
234 terminated.
235 .
236 .
237 .\" HTML <a name="digitsafterbackslash"></a>
238 .SS "Non-printing characters"
239 .rs
240 .sp
241 A second use of backslash provides a way of encoding non-printing characters
242 in patterns in a visible manner. There is no restriction on the appearance of
243 non-printing characters, apart from the binary zero that terminates a pattern,
244 but when a pattern is being prepared by text editing, it is often easier to use
245 one of the following escape sequences than the binary character it represents:
246 .sp
247 \ea alarm, that is, the BEL character (hex 07)
248 \ecx "control-x", where x is any ASCII character
249 \ee escape (hex 1B)
250 \ef form feed (hex 0C)
251 \en linefeed (hex 0A)
252 \er carriage return (hex 0D)
253 \et tab (hex 09)
254 \eddd character with octal code ddd, or back reference
255 \exhh character with hex code hh
256 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
257 \euhhhh character with hex code hhhh (JavaScript mode only)
258 .sp
259 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
260 case letter, it is converted to upper case. Then bit 6 of the character (hex
261 40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
262 but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
263 data item (byte or 16-bit value) following \ec has a value greater than 127, a
264 compile-time error occurs. This locks out non-ASCII characters in all modes.
265 .P
266 The \ec facility was designed for use with ASCII characters, but with the
267 extension to Unicode it is even less useful than it once was. It is, however,
268 recognized when PCRE is compiled in EBCDIC mode, where data items are always
269 bytes. In this mode, all values are valid after \ec. If the next character is a
270 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
271 byte are inverted. Thus \ecA becomes hex 01, as in ASCII (A is C1), but because
272 the EBCDIC letters are disjoint, \ecZ becomes hex 29 (Z is E9), and other
273 characters also generate different values.
274 .P
275 By default, after \ex, from zero to two hexadecimal digits are read (letters
276 can be in upper or lower case). Any number of hexadecimal digits may appear
277 between \ex{ and }, but the character code is constrained as follows:
278 .sp
279 8-bit non-UTF mode less than 0x100
280 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
281 16-bit non-UTF mode less than 0x10000
282 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
283 .sp
284 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
285 "surrogate" codepoints).
286 .P
287 If characters other than hexadecimal digits appear between \ex{ and }, or if
288 there is no terminating }, this form of escape is not recognized. Instead, the
289 initial \ex will be interpreted as a basic hexadecimal escape, with no
290 following digits, giving a character whose value is zero.
291 .P
292 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
293 as just described only when it is followed by two hexadecimal digits.
294 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
295 code points greater than 256 is provided by \eu, which must be followed by
296 four hexadecimal digits; otherwise it matches a literal "u" character.
297 Character codes specified by \eu in JavaScript mode are constrained in the same
298 was as those specified by \ex in non-JavaScript mode.
299 .P
300 Characters whose value is less than 256 can be defined by either of the two
301 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
302 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
303 \eu00dc in JavaScript mode).
304 .P
305 After \e0 up to two further octal digits are read. If there are fewer than two
306 digits, just those that are present are used. Thus the sequence \e0\ex\e07
307 specifies two binary zeros followed by a BEL character (code value 7). Make
308 sure you supply two digits after the initial zero if the pattern character that
309 follows is itself an octal digit.
310 .P
311 The handling of a backslash followed by a digit other than 0 is complicated.
312 Outside a character class, PCRE reads it and any following digits as a decimal
313 number. If the number is less than 10, or if there have been at least that many
314 previous capturing left parentheses in the expression, the entire sequence is
315 taken as a \fIback reference\fP. A description of how this works is given
316 .\" HTML <a href="#backreferences">
317 .\" </a>
318 later,
319 .\"
320 following the discussion of
321 .\" HTML <a href="#subpattern">
322 .\" </a>
323 parenthesized subpatterns.
324 .\"
325 .P
326 Inside a character class, or if the decimal number is greater than 9 and there
327 have not been that many capturing subpatterns, PCRE re-reads up to three octal
328 digits following the backslash, and uses them to generate a data character. Any
329 subsequent digits stand for themselves. The value of the character is
330 constrained in the same way as characters specified in hexadecimal.
331 For example:
332 .sp
333 \e040 is another way of writing an ASCII space
334 .\" JOIN
335 \e40 is the same, provided there are fewer than 40
336 previous capturing subpatterns
337 \e7 is always a back reference
338 .\" JOIN
339 \e11 might be a back reference, or another way of
340 writing a tab
341 \e011 is always a tab
342 \e0113 is a tab followed by the character "3"
343 .\" JOIN
344 \e113 might be a back reference, otherwise the
345 character with octal code 113
346 .\" JOIN
347 \e377 might be a back reference, otherwise
348 the value 255 (decimal)
349 .\" JOIN
350 \e81 is either a back reference, or a binary zero
351 followed by the two characters "8" and "1"
352 .sp
353 Note that octal values of 100 or greater must not be introduced by a leading
354 zero, because no more than three octal digits are ever read.
355 .P
356 All the sequences that define a single character value can be used both inside
357 and outside character classes. In addition, inside a character class, \eb is
358 interpreted as the backspace character (hex 08).
359 .P
360 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
361 inside a character class. Like other unrecognized escape sequences, they are
362 treated as the literal characters "B", "R", and "X" by default, but cause an
363 error if the PCRE_EXTRA option is set. Outside a character class, these
364 sequences have different meanings.
365 .
366 .
367 .SS "Unsupported escape sequences"
368 .rs
369 .sp
370 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
371 handler and used to modify the case of following characters. By default, PCRE
372 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
373 option is set, \eU matches a "U" character, and \eu can be used to define a
374 character by code point, as described in the previous section.
375 .
376 .
377 .SS "Absolute and relative back references"
378 .rs
379 .sp
380 The sequence \eg followed by an unsigned or a negative number, optionally
381 enclosed in braces, is an absolute or relative back reference. A named back
382 reference can be coded as \eg{name}. Back references are discussed
383 .\" HTML <a href="#backreferences">
384 .\" </a>
385 later,
386 .\"
387 following the discussion of
388 .\" HTML <a href="#subpattern">
389 .\" </a>
390 parenthesized subpatterns.
391 .\"
392 .
393 .
394 .SS "Absolute and relative subroutine calls"
395 .rs
396 .sp
397 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
398 a number enclosed either in angle brackets or single quotes, is an alternative
399 syntax for referencing a subpattern as a "subroutine". Details are discussed
400 .\" HTML <a href="#onigurumasubroutines">
401 .\" </a>
402 later.
403 .\"
404 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
405 synonymous. The former is a back reference; the latter is a
406 .\" HTML <a href="#subpatternsassubroutines">
407 .\" </a>
408 subroutine
409 .\"
410 call.
411 .
412 .
413 .\" HTML <a name="genericchartypes"></a>
414 .SS "Generic character types"
415 .rs
416 .sp
417 Another use of backslash is for specifying generic character types:
418 .sp
419 \ed any decimal digit
420 \eD any character that is not a decimal digit
421 \eh any horizontal white space character
422 \eH any character that is not a horizontal white space character
423 \es any white space character
424 \eS any character that is not a white space character
425 \ev any vertical white space character
426 \eV any character that is not a vertical white space character
427 \ew any "word" character
428 \eW any "non-word" character
429 .sp
430 There is also the single sequence \eN, which matches a non-newline character.
431 This is the same as
432 .\" HTML <a href="#fullstopdot">
433 .\" </a>
434 the "." metacharacter
435 .\"
436 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
437 PCRE does not support this.
438 .P
439 Each pair of lower and upper case escape sequences partitions the complete set
440 of characters into two disjoint sets. Any given character matches one, and only
441 one, of each pair. The sequences can appear both inside and outside character
442 classes. They each match one character of the appropriate type. If the current
443 matching point is at the end of the subject string, all of them fail, because
444 there is no character to match.
445 .P
446 For compatibility with Perl, \es does not match the VT character (code 11).
447 This makes it different from the the POSIX "space" class. The \es characters
448 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
449 included in a Perl script, \es may match the VT character. In PCRE, it never
450 does.
451 .P
452 A "word" character is an underscore or any character that is a letter or digit.
453 By default, the definition of letters and digits is controlled by PCRE's
454 low-valued character tables, and may vary if locale-specific matching is taking
455 place (see
456 .\" HTML <a href="pcreapi.html#localesupport">
457 .\" </a>
458 "Locale support"
459 .\"
460 in the
461 .\" HREF
462 \fBpcreapi\fP
463 .\"
464 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
465 or "french" in Windows, some character codes greater than 128 are used for
466 accented letters, and these are then matched by \ew. The use of locales with
467 Unicode is discouraged.
468 .P
469 By default, in a UTF mode, characters with values greater than 128 never match
470 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
471 their original meanings from before UTF support was available, mainly for
472 efficiency reasons. However, if PCRE is compiled with Unicode property support,
473 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
474 properties are used to determine character types, as follows:
475 .sp
476 \ed any character that \ep{Nd} matches (decimal digit)
477 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
478 \ew any character that \ep{L} or \ep{N} matches, plus underscore
479 .sp
480 The upper case escapes match the inverse sets of characters. Note that \ed
481 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
482 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
483 \eB because they are defined in terms of \ew and \eW. Matching these sequences
484 is noticeably slower when PCRE_UCP is set.
485 .P
486 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
487 release 5.10. In contrast to the other sequences, which match only ASCII
488 characters by default, these always match certain high-valued codepoints,
489 whether or not PCRE_UCP is set. The horizontal space characters are:
490 .sp
491 U+0009 Horizontal tab (HT)
492 U+0020 Space
493 U+00A0 Non-break space
494 U+1680 Ogham space mark
495 U+180E Mongolian vowel separator
496 U+2000 En quad
497 U+2001 Em quad
498 U+2002 En space
499 U+2003 Em space
500 U+2004 Three-per-em space
501 U+2005 Four-per-em space
502 U+2006 Six-per-em space
503 U+2007 Figure space
504 U+2008 Punctuation space
505 U+2009 Thin space
506 U+200A Hair space
507 U+202F Narrow no-break space
508 U+205F Medium mathematical space
509 U+3000 Ideographic space
510 .sp
511 The vertical space characters are:
512 .sp
513 U+000A Linefeed (LF)
514 U+000B Vertical tab (VT)
515 U+000C Form feed (FF)
516 U+000D Carriage return (CR)
517 U+0085 Next line (NEL)
518 U+2028 Line separator
519 U+2029 Paragraph separator
520 .sp
521 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
522 relevant.
523 .
524 .
525 .\" HTML <a name="newlineseq"></a>
526 .SS "Newline sequences"
527 .rs
528 .sp
529 Outside a character class, by default, the escape sequence \eR matches any
530 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
531 following:
532 .sp
533 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
534 .sp
535 This is an example of an "atomic group", details of which are given
536 .\" HTML <a href="#atomicgroup">
537 .\" </a>
538 below.
539 .\"
540 This particular group matches either the two-character sequence CR followed by
541 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
542 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
543 line, U+0085). The two-character sequence is treated as a single unit that
544 cannot be split.
545 .P
546 In other modes, two additional characters whose codepoints are greater than 255
547 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
548 Unicode character property support is not needed for these characters to be
549 recognized.
550 .P
551 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
552 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
553 either at compile time or when the pattern is matched. (BSR is an abbrevation
554 for "backslash R".) This can be made the default when PCRE is built; if this is
555 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
556 It is also possible to specify these settings by starting a pattern string with
557 one of the following sequences:
558 .sp
559 (*BSR_ANYCRLF) CR, LF, or CRLF only
560 (*BSR_UNICODE) any Unicode newline sequence
561 .sp
562 These override the default and the options given to the compiling function, but
563 they can themselves be overridden by options given to a matching function. Note
564 that these special settings, which are not Perl-compatible, are recognized only
565 at the very start of a pattern, and that they must be in upper case. If more
566 than one of them is present, the last one is used. They can be combined with a
567 change of newline convention; for example, a pattern can start with:
568 .sp
570 .sp
571 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
572 sequences. Inside a character class, \eR is treated as an unrecognized escape
573 sequence, and so matches the letter "R" by default, but causes an error if
574 PCRE_EXTRA is set.
575 .
576 .
577 .\" HTML <a name="uniextseq"></a>
578 .SS Unicode character properties
579 .rs
580 .sp
581 When PCRE is built with Unicode character property support, three additional
582 escape sequences that match characters with specific properties are available.
583 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
584 characters whose codepoints are less than 256, but they do work in this mode.
585 The extra escape sequences are:
586 .sp
587 \ep{\fIxx\fP} a character with the \fIxx\fP property
588 \eP{\fIxx\fP} a character without the \fIxx\fP property
589 \eX a Unicode extended grapheme cluster
590 .sp
591 The property names represented by \fIxx\fP above are limited to the Unicode
592 script names, the general category properties, "Any", which matches any
593 character (including newline), and some special PCRE properties (described
594 in the
595 .\" HTML <a href="#extraprops">
596 .\" </a>
597 next section).
598 .\"
599 Other Perl properties such as "InMusicalSymbols" are not currently supported by
600 PCRE. Note that \eP{Any} does not match any characters, so always causes a
601 match failure.
602 .P
603 Sets of Unicode characters are defined as belonging to certain scripts. A
604 character from one of these sets can be matched using a script name. For
605 example:
606 .sp
607 \ep{Greek}
608 \eP{Han}
609 .sp
610 Those that are not part of an identified script are lumped together as
611 "Common". The current list of scripts is:
612 .P
613 Arabic,
614 Armenian,
615 Avestan,
616 Balinese,
617 Bamum,
618 Batak,
619 Bengali,
620 Bopomofo,
621 Brahmi,
622 Braille,
623 Buginese,
624 Buhid,
625 Canadian_Aboriginal,
626 Carian,
627 Chakma,
628 Cham,
629 Cherokee,
630 Common,
631 Coptic,
632 Cuneiform,
633 Cypriot,
634 Cyrillic,
635 Deseret,
636 Devanagari,
637 Egyptian_Hieroglyphs,
638 Ethiopic,
639 Georgian,
640 Glagolitic,
641 Gothic,
642 Greek,
643 Gujarati,
644 Gurmukhi,
645 Han,
646 Hangul,
647 Hanunoo,
648 Hebrew,
649 Hiragana,
650 Imperial_Aramaic,
651 Inherited,
652 Inscriptional_Pahlavi,
653 Inscriptional_Parthian,
654 Javanese,
655 Kaithi,
656 Kannada,
657 Katakana,
658 Kayah_Li,
659 Kharoshthi,
660 Khmer,
661 Lao,
662 Latin,
663 Lepcha,
664 Limbu,
665 Linear_B,
666 Lisu,
667 Lycian,
668 Lydian,
669 Malayalam,
670 Mandaic,
671 Meetei_Mayek,
672 Meroitic_Cursive,
673 Meroitic_Hieroglyphs,
674 Miao,
675 Mongolian,
676 Myanmar,
677 New_Tai_Lue,
678 Nko,
679 Ogham,
680 Old_Italic,
681 Old_Persian,
682 Old_South_Arabian,
683 Old_Turkic,
684 Ol_Chiki,
685 Oriya,
686 Osmanya,
687 Phags_Pa,
688 Phoenician,
689 Rejang,
690 Runic,
691 Samaritan,
692 Saurashtra,
693 Sharada,
694 Shavian,
695 Sinhala,
696 Sora_Sompeng,
697 Sundanese,
698 Syloti_Nagri,
699 Syriac,
700 Tagalog,
701 Tagbanwa,
702 Tai_Le,
703 Tai_Tham,
704 Tai_Viet,
705 Takri,
706 Tamil,
707 Telugu,
708 Thaana,
709 Thai,
710 Tibetan,
711 Tifinagh,
712 Ugaritic,
713 Vai,
714 Yi.
715 .P
716 Each character has exactly one Unicode general category property, specified by
717 a two-letter abbreviation. For compatibility with Perl, negation can be
718 specified by including a circumflex between the opening brace and the property
719 name. For example, \ep{^Lu} is the same as \eP{Lu}.
720 .P
721 If only one letter is specified with \ep or \eP, it includes all the general
722 category properties that start with that letter. In this case, in the absence
723 of negation, the curly brackets in the escape sequence are optional; these two
724 examples have the same effect:
725 .sp
726 \ep{L}
727 \epL
728 .sp
729 The following general category property codes are supported:
730 .sp
731 C Other
732 Cc Control
733 Cf Format
734 Cn Unassigned
735 Co Private use
736 Cs Surrogate
737 .sp
738 L Letter
739 Ll Lower case letter
740 Lm Modifier letter
741 Lo Other letter
742 Lt Title case letter
743 Lu Upper case letter
744 .sp
745 M Mark
746 Mc Spacing mark
747 Me Enclosing mark
748 Mn Non-spacing mark
749 .sp
750 N Number
751 Nd Decimal number
752 Nl Letter number
753 No Other number
754 .sp
755 P Punctuation
756 Pc Connector punctuation
757 Pd Dash punctuation
758 Pe Close punctuation
759 Pf Final punctuation
760 Pi Initial punctuation
761 Po Other punctuation
762 Ps Open punctuation
763 .sp
764 S Symbol
765 Sc Currency symbol
766 Sk Modifier symbol
767 Sm Mathematical symbol
768 So Other symbol
769 .sp
770 Z Separator
771 Zl Line separator
772 Zp Paragraph separator
773 Zs Space separator
774 .sp
775 The special property L& is also supported: it matches a character that has
776 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
777 a modifier or "other".
778 .P
779 The Cs (Surrogate) property applies only to characters in the range U+D800 to
780 U+DFFF. Such characters are not valid in Unicode strings and so
781 cannot be tested by PCRE, unless UTF validity checking has been turned off
782 (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
783 .\" HREF
784 \fBpcreapi\fP
785 .\"
786 page). Perl does not support the Cs property.
787 .P
788 The long synonyms for property names that Perl supports (such as \ep{Letter})
789 are not supported by PCRE, nor is it permitted to prefix any of these
790 properties with "Is".
791 .P
792 No character that is in the Unicode table has the Cn (unassigned) property.
793 Instead, this property is assumed for any code point that is not in the
794 Unicode table.
795 .P
796 Specifying caseless matching does not affect these escape sequences. For
797 example, \ep{Lu} always matches only upper case letters.
798 .P
799 Matching characters by Unicode property is not fast, because PCRE has to do a
800 multistage table lookup in order to find a character's property. That is why
801 the traditional escape sequences such as \ed and \ew do not use Unicode
802 properties in PCRE by default, though you can make them do so by setting the
803 PCRE_UCP option or by starting the pattern with (*UCP).
804 .
805 .
806 .SS Extended grapheme clusters
807 .rs
808 .sp
809 The \eX escape matches any number of Unicode characters that form an "extended
810 grapheme cluster", and treats the sequence as an atomic group
811 .\" HTML <a href="#atomicgroup">
812 .\" </a>
813 (see below).
814 .\"
815 Up to and including release 8.31, PCRE matched an earlier, simpler definition
816 that was equivalent to
817 .sp
818 (?>\ePM\epM*)
819 .sp
820 That is, it matched a character without the "mark" property, followed by zero
821 or more characters with the "mark" property. Characters with the "mark"
822 property are typically non-spacing accents that affect the preceding character.
823 .P
824 This simple definition was extended in Unicode to include more complicated
825 kinds of composite character by giving each character a grapheme breaking
826 property, and creating rules that use these properties to define the boundaries
827 of extended grapheme clusters. In releases of PCRE later than 8.31, \eX matches
828 one of these clusters.
829 .P
830 \eX always matches at least one character. Then it decides whether to add
831 additional characters according to the following rules for ending a cluster:
832 .P
833 1. End at the end of the subject string.
834 .P
835 2. Do not end between CR and LF; otherwise end after any control character.
836 .P
837 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
838 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
839 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
840 character; an LVT or T character may be follwed only by a T character.
841 .P
842 4. Do not end before extending characters or spacing marks. Characters with
843 the "mark" property always have the "extend" grapheme breaking property.
844 .P
845 5. Do not end after prepend characters.
846 .P
847 6. Otherwise, end the cluster.
848 .
849 .
850 .\" HTML <a name="extraprops"></a>
851 .SS PCRE's additional properties
852 .rs
853 .sp
854 As well as the standard Unicode properties described above, PCRE supports four
855 more that make it possible to convert traditional escape sequences such as \ew
856 and \es and POSIX character classes to use Unicode properties. PCRE uses these
857 non-standard, non-Perl properties internally when PCRE_UCP is set. They are:
858 .sp
859 Xan Any alphanumeric character
860 Xps Any POSIX space character
861 Xsp Any Perl space character
862 Xwd Any Perl "word" character
863 .sp
864 Xan matches characters that have either the L (letter) or the N (number)
865 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
866 carriage return, and any other character that has the Z (separator) property.
867 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
868 same characters as Xan, plus underscore.
869 .
870 .
871 .\" HTML <a name="resetmatchstart"></a>
872 .SS "Resetting the match start"
873 .rs
874 .sp
875 The escape sequence \eK causes any previously matched characters not to be
876 included in the final matched sequence. For example, the pattern:
877 .sp
878 foo\eKbar
879 .sp
880 matches "foobar", but reports that it has matched "bar". This feature is
881 similar to a lookbehind assertion
882 .\" HTML <a href="#lookbehind">
883 .\" </a>
884 (described below).
885 .\"
886 However, in this case, the part of the subject before the real match does not
887 have to be of fixed length, as lookbehind assertions do. The use of \eK does
888 not interfere with the setting of
889 .\" HTML <a href="#subpattern">
890 .\" </a>
891 captured substrings.
892 .\"
893 For example, when the pattern
894 .sp
895 (foo)\eKbar
896 .sp
897 matches "foobar", the first substring is still set to "foo".
898 .P
899 Perl documents that the use of \eK within assertions is "not well defined". In
900 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
901 ignored in negative assertions.
902 .
903 .
904 .\" HTML <a name="smallassertions"></a>
905 .SS "Simple assertions"
906 .rs
907 .sp
908 The final use of backslash is for certain simple assertions. An assertion
909 specifies a condition that has to be met at a particular point in a match,
910 without consuming any characters from the subject string. The use of
911 subpatterns for more complicated assertions is described
912 .\" HTML <a href="#bigassertions">
913 .\" </a>
914 below.
915 .\"
916 The backslashed assertions are:
917 .sp
918 \eb matches at a word boundary
919 \eB matches when not at a word boundary
920 \eA matches at the start of the subject
921 \eZ matches at the end of the subject
922 also matches before a newline at the end of the subject
923 \ez matches only at the end of the subject
924 \eG matches at the first matching position in the subject
925 .sp
926 Inside a character class, \eb has a different meaning; it matches the backspace
927 character. If any other of these assertions appears in a character class, by
928 default it matches the corresponding literal character (for example, \eB
929 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
930 escape sequence" error is generated instead.
931 .P
932 A word boundary is a position in the subject string where the current character
933 and the previous character do not both match \ew or \eW (i.e. one matches
934 \ew and the other matches \eW), or the start or end of the string if the
935 first or last character matches \ew, respectively. In a UTF mode, the meanings
936 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
937 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
938 of word" or "end of word" metasequence. However, whatever follows \eb normally
939 determines which it is. For example, the fragment \eba matches "a" at the start
940 of a word.
941 .P
942 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
943 dollar (described in the next section) in that they only ever match at the very
944 start and end of the subject string, whatever options are set. Thus, they are
945 independent of multiline mode. These three assertions are not affected by the
946 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
947 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
948 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
949 at a point other than the beginning of the subject, \eA can never match. The
950 difference between \eZ and \ez is that \eZ matches before a newline at the end
951 of the string as well as at the very end, whereas \ez matches only at the end.
952 .P
953 The \eG assertion is true only when the current matching position is at the
954 start point of the match, as specified by the \fIstartoffset\fP argument of
955 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
956 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
957 arguments, you can mimic Perl's /g option, and it is in this kind of
958 implementation where \eG can be useful.
959 .P
960 Note, however, that PCRE's interpretation of \eG, as the start of the current
961 match, is subtly different from Perl's, which defines it as the end of the
962 previous match. In Perl, these can be different when the previously matched
963 string was empty. Because PCRE does just one match at a time, it cannot
964 reproduce this behaviour.
965 .P
966 If all the alternatives of a pattern begin with \eG, the expression is anchored
967 to the starting match position, and the "anchored" flag is set in the compiled
968 regular expression.
969 .
970 .
972 .rs
973 .sp
974 Outside a character class, in the default matching mode, the circumflex
975 character is an assertion that is true only if the current matching point is
976 at the start of the subject string. If the \fIstartoffset\fP argument of
977 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
978 option is unset. Inside a character class, circumflex has an entirely different
979 meaning
980 .\" HTML <a href="#characterclass">
981 .\" </a>
982 (see below).
983 .\"
984 .P
985 Circumflex need not be the first character of the pattern if a number of
986 alternatives are involved, but it should be the first thing in each alternative
987 in which it appears if the pattern is ever to match that branch. If all
988 possible alternatives start with a circumflex, that is, if the pattern is
989 constrained to match only at the start of the subject, it is said to be an
990 "anchored" pattern. (There are also other constructs that can cause a pattern
991 to be anchored.)
992 .P
993 A dollar character is an assertion that is true only if the current matching
994 point is at the end of the subject string, or immediately before a newline
995 at the end of the string (by default). Dollar need not be the last character of
996 the pattern if a number of alternatives are involved, but it should be the last
997 item in any branch in which it appears. Dollar has no special meaning in a
998 character class.
999 .P
1000 The meaning of dollar can be changed so that it matches only at the very end of
1001 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1002 does not affect the \eZ assertion.
1003 .P
1004 The meanings of the circumflex and dollar characters are changed if the
1005 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1006 immediately after internal newlines as well as at the start of the subject
1007 string. It does not match after a newline that ends the string. A dollar
1008 matches before any newlines in the string, as well as at the very end, when
1009 PCRE_MULTILINE is set. When newline is specified as the two-character
1010 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1011 .P
1012 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
1013 \en represents a newline) in multiline mode, but not otherwise. Consequently,
1014 patterns that are anchored in single line mode because all branches start with
1015 ^ are not anchored in multiline mode, and a match for circumflex is possible
1016 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
1017 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1018 .P
1019 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
1020 end of the subject in both modes, and if all branches of a pattern start with
1021 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
1022 .
1023 .
1024 .\" HTML <a name="fullstopdot"></a>
1026 .rs
1027 .sp
1028 Outside a character class, a dot in the pattern matches any one character in
1029 the subject string except (by default) a character that signifies the end of a
1030 line.
1031 .P
1032 When a line ending is defined as a single character, dot never matches that
1033 character; when the two-character sequence CRLF is used, dot does not match CR
1034 if it is immediately followed by LF, but otherwise it matches all characters
1035 (including isolated CRs and LFs). When any Unicode line endings are being
1036 recognized, dot does not match CR or LF or any of the other line ending
1037 characters.
1038 .P
1039 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1040 option is set, a dot matches any one character, without exception. If the
1041 two-character sequence CRLF is present in the subject string, it takes two dots
1042 to match it.
1043 .P
1044 The handling of dot is entirely independent of the handling of circumflex and
1045 dollar, the only relationship being that they both involve newlines. Dot has no
1046 special meaning in a character class.
1047 .P
1048 The escape sequence \eN behaves like a dot, except that it is not affected by
1049 the PCRE_DOTALL option. In other words, it matches any character except one
1050 that signifies the end of a line. Perl also uses \eN to match characters by
1051 name; PCRE does not support this.
1052 .
1053 .
1055 .rs
1056 .sp
1057 Outside a character class, the escape sequence \eC matches any one data unit,
1058 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1059 byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \eC always
1060 matches line-ending characters. The feature is provided in Perl in order to
1061 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1062 used. Because \eC breaks up characters into individual data units, matching one
1063 unit with \eC in a UTF mode means that the rest of the string may start with a
1064 malformed UTF character. This has undefined results, because PCRE assumes that
1065 it is dealing with valid UTF strings (and by default it checks this at the
1066 start of processing unless the PCRE_NO_UTF8_CHECK or PCRE_NO_UTF16_CHECK option
1067 is used).
1068 .P
1069 PCRE does not allow \eC to appear in lookbehind assertions
1070 .\" HTML <a href="#lookbehind">
1071 .\" </a>
1072 (described below)
1073 .\"
1074 in a UTF mode, because this would make it impossible to calculate the length of
1075 the lookbehind.
1076 .P
1077 In general, the \eC escape sequence is best avoided. However, one
1078 way of using it that avoids the problem of malformed UTF characters is to use a
1079 lookahead to check the length of the next character, as in this pattern, which
1080 could be used with a UTF-8 string (ignore white space and line breaks):
1081 .sp
1082 (?| (?=[\ex00-\ex7f])(\eC) |
1083 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1084 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1085 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1086 .sp
1087 A group that starts with (?| resets the capturing parentheses numbers in each
1088 alternative (see
1089 .\" HTML <a href="#dupsubpatternnumber">
1090 .\" </a>
1091 "Duplicate Subpattern Numbers"
1092 .\"
1093 below). The assertions at the start of each branch check the next UTF-8
1094 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1095 character's individual bytes are then captured by the appropriate number of
1096 groups.
1097 .
1098 .
1099 .\" HTML <a name="characterclass"></a>
1101 .rs
1102 .sp
1103 An opening square bracket introduces a character class, terminated by a closing
1104 square bracket. A closing square bracket on its own is not special by default.
1105 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1106 bracket causes a compile-time error. If a closing square bracket is required as
1107 a member of the class, it should be the first data character in the class
1108 (after an initial circumflex, if present) or escaped with a backslash.
1109 .P
1110 A character class matches a single character in the subject. In a UTF mode, the
1111 character may be more than one data unit long. A matched character must be in
1112 the set of characters defined by the class, unless the first character in the
1113 class definition is a circumflex, in which case the subject character must not
1114 be in the set defined by the class. If a circumflex is actually required as a
1115 member of the class, ensure it is not the first character, or escape it with a
1116 backslash.
1117 .P
1118 For example, the character class [aeiou] matches any lower case vowel, while
1119 [^aeiou] matches any character that is not a lower case vowel. Note that a
1120 circumflex is just a convenient notation for specifying the characters that
1121 are in the class by enumerating those that are not. A class that starts with a
1122 circumflex is not an assertion; it still consumes a character from the subject
1123 string, and therefore it fails if the current pointer is at the end of the
1124 string.
1125 .P
1126 In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
1127 included in a class as a literal string of data units, or by using the \ex{
1128 escaping mechanism.
1129 .P
1130 When caseless matching is set, any letters in a class represent both their
1131 upper case and lower case versions, so for example, a caseless [aeiou] matches
1132 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1133 caseful version would. In a UTF mode, PCRE always understands the concept of
1134 case for characters whose values are less than 128, so caseless matching is
1135 always possible. For characters with higher values, the concept of case is
1136 supported if PCRE is compiled with Unicode property support, but not otherwise.
1137 If you want to use caseless matching in a UTF mode for characters 128 and
1138 above, you must ensure that PCRE is compiled with Unicode property support as
1139 well as with UTF support.
1140 .P
1141 Characters that might indicate line breaks are never treated in any special way
1142 when matching character classes, whatever line-ending sequence is in use, and
1143 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1144 such as [^a] always matches one of these characters.
1145 .P
1146 The minus (hyphen) character can be used to specify a range of characters in a
1147 character class. For example, [d-m] matches any letter between d and m,
1148 inclusive. If a minus character is required in a class, it must be escaped with
1149 a backslash or appear in a position where it cannot be interpreted as
1150 indicating a range, typically as the first or last character in the class.
1151 .P
1152 It is not possible to have the literal character "]" as the end character of a
1153 range. A pattern such as [W-]46] is interpreted as a class of two characters
1154 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1155 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1156 the end of range, so [W-\e]46] is interpreted as a class containing a range
1157 followed by two other characters. The octal or hexadecimal representation of
1158 "]" can also be used to end a range.
1159 .P
1160 Ranges operate in the collating sequence of character values. They can also be
1161 used for characters specified numerically, for example [\e000-\e037]. Ranges
1162 can include any characters that are valid for the current mode.
1163 .P
1164 If a range that includes letters is used when caseless matching is set, it
1165 matches the letters in either case. For example, [W-c] is equivalent to
1166 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1167 tables for a French locale are in use, [\exc8-\excb] matches accented E
1168 characters in both cases. In UTF modes, PCRE supports the concept of case for
1169 characters with values greater than 128 only when it is compiled with Unicode
1170 property support.
1171 .P
1172 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1173 \eV, \ew, and \eW may appear in a character class, and add the characters that
1174 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1175 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1176 and their upper case partners, just as it does when they appear outside a
1177 character class, as described in the section entitled
1178 .\" HTML <a href="#genericchartypes">
1179 .\" </a>
1180 "Generic character types"
1181 .\"
1182 above. The escape sequence \eb has a different meaning inside a character
1183 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1184 are not special inside a character class. Like any other unrecognized escape
1185 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1186 default, but cause an error if the PCRE_EXTRA option is set.
1187 .P
1188 A circumflex can conveniently be used with the upper case character types to
1189 specify a more restricted set of characters than the matching lower case type.
1190 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1191 whereas [\ew] includes underscore. A positive character class should be read as
1192 "something OR something OR ..." and a negative class as "NOT something AND NOT
1193 something AND NOT ...".
1194 .P
1195 The only metacharacters that are recognized in character classes are backslash,
1196 hyphen (only where it can be interpreted as specifying a range), circumflex
1197 (only at the start), opening square bracket (only when it can be interpreted as
1198 introducing a POSIX class name - see the next section), and the terminating
1199 closing square bracket. However, escaping other non-alphanumeric characters
1200 does no harm.
1201 .
1202 .
1204 .rs
1205 .sp
1206 Perl supports the POSIX notation for character classes. This uses names
1207 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1208 this notation. For example,
1209 .sp
1210 [01[:alpha:]%]
1211 .sp
1212 matches "0", "1", any alphabetic character, or "%". The supported class names
1213 are:
1214 .sp
1215 alnum letters and digits
1216 alpha letters
1217 ascii character codes 0 - 127
1218 blank space or tab only
1219 cntrl control characters
1220 digit decimal digits (same as \ed)
1221 graph printing characters, excluding space
1222 lower lower case letters
1223 print printing characters, including space
1224 punct printing characters, excluding letters and digits and space
1225 space white space (not quite the same as \es)
1226 upper upper case letters
1227 word "word" characters (same as \ew)
1228 xdigit hexadecimal digits
1229 .sp
1230 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1231 space (32). Notice that this list includes the VT character (code 11). This
1232 makes "space" different to \es, which does not include VT (for Perl
1233 compatibility).
1234 .P
1235 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1236 5.8. Another Perl extension is negation, which is indicated by a ^ character
1237 after the colon. For example,
1238 .sp
1239 [12[:^digit:]]
1240 .sp
1241 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1242 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1243 supported, and an error is given if they are encountered.
1244 .P
1245 By default, in UTF modes, characters with values greater than 128 do not match
1246 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1247 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1248 character properties are used. This is achieved by replacing the POSIX classes
1249 by other sequences, as follows:
1250 .sp
1251 [:alnum:] becomes \ep{Xan}
1252 [:alpha:] becomes \ep{L}
1253 [:blank:] becomes \eh
1254 [:digit:] becomes \ep{Nd}
1255 [:lower:] becomes \ep{Ll}
1256 [:space:] becomes \ep{Xps}
1257 [:upper:] becomes \ep{Lu}
1258 [:word:] becomes \ep{Xwd}
1259 .sp
1260 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1261 classes are unchanged, and match only characters with code points less than
1262 128.
1263 .
1264 .
1266 .rs
1267 .sp
1268 Vertical bar characters are used to separate alternative patterns. For example,
1269 the pattern
1270 .sp
1271 gilbert|sullivan
1272 .sp
1273 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1274 and an empty alternative is permitted (matching the empty string). The matching
1275 process tries each alternative in turn, from left to right, and the first one
1276 that succeeds is used. If the alternatives are within a subpattern
1277 .\" HTML <a href="#subpattern">
1278 .\" </a>
1279 (defined below),
1280 .\"
1281 "succeeds" means matching the rest of the main pattern as well as the
1282 alternative in the subpattern.
1283 .
1284 .
1286 .rs
1287 .sp
1288 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1289 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1290 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1291 The option letters are
1292 .sp
1293 i for PCRE_CASELESS
1295 s for PCRE_DOTALL
1296 x for PCRE_EXTENDED
1297 .sp
1298 For example, (?im) sets caseless, multiline matching. It is also possible to
1299 unset these options by preceding the letter with a hyphen, and a combined
1300 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1301 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1302 permitted. If a letter appears both before and after the hyphen, the option is
1303 unset.
1304 .P
1305 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1306 changed in the same way as the Perl-compatible options by using the characters
1307 J, U and X respectively.
1308 .P
1309 When one of these option changes occurs at top level (that is, not inside
1310 subpattern parentheses), the change applies to the remainder of the pattern
1311 that follows. If the change is placed right at the start of a pattern, PCRE
1312 extracts it into the global options (and it will therefore show up in data
1313 extracted by the \fBpcre_fullinfo()\fP function).
1314 .P
1315 An option change within a subpattern (see below for a description of
1316 subpatterns) affects only that part of the subpattern that follows it, so
1317 .sp
1318 (a(?i)b)c
1319 .sp
1320 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1321 By this means, options can be made to have different settings in different
1322 parts of the pattern. Any changes made in one alternative do carry on
1323 into subsequent branches within the same subpattern. For example,
1324 .sp
1325 (a(?i)b|c)
1326 .sp
1327 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1328 branch is abandoned before the option setting. This is because the effects of
1329 option settings happen at compile time. There would be some very weird
1330 behaviour otherwise.
1331 .P
1332 \fBNote:\fP There are other PCRE-specific options that can be set by the
1333 application when the compiling or matching functions are called. In some cases
1334 the pattern can contain special leading sequences such as (*CRLF) to override
1335 what the application has set or what has been defaulted. Details are given in
1336 the section entitled
1337 .\" HTML <a href="#newlineseq">
1338 .\" </a>
1339 "Newline sequences"
1340 .\"
1341 above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
1342 can be used to set UTF and Unicode property modes; they are equivalent to
1343 setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1344 .
1345 .
1346 .\" HTML <a name="subpattern"></a>
1348 .rs
1349 .sp
1350 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1351 Turning part of a pattern into a subpattern does two things:
1352 .sp
1353 1. It localizes a set of alternatives. For example, the pattern
1354 .sp
1355 cat(aract|erpillar|)
1356 .sp
1357 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1358 match "cataract", "erpillar" or an empty string.
1359 .sp
1360 2. It sets up the subpattern as a capturing subpattern. This means that, when
1361 the whole pattern matches, that portion of the subject string that matched the
1362 subpattern is passed back to the caller via the \fIovector\fP argument of the
1363 matching function. (This applies only to the traditional matching functions;
1364 the DFA matching functions do not support capturing.)
1365 .P
1366 Opening parentheses are counted from left to right (starting from 1) to obtain
1367 numbers for the capturing subpatterns. For example, if the string "the red
1368 king" is matched against the pattern
1369 .sp
1370 the ((red|white) (king|queen))
1371 .sp
1372 the captured substrings are "red king", "red", and "king", and are numbered 1,
1373 2, and 3, respectively.
1374 .P
1375 The fact that plain parentheses fulfil two functions is not always helpful.
1376 There are often times when a grouping subpattern is required without a
1377 capturing requirement. If an opening parenthesis is followed by a question mark
1378 and a colon, the subpattern does not do any capturing, and is not counted when
1379 computing the number of any subsequent capturing subpatterns. For example, if
1380 the string "the white queen" is matched against the pattern
1381 .sp
1382 the ((?:red|white) (king|queen))
1383 .sp
1384 the captured substrings are "white queen" and "queen", and are numbered 1 and
1385 2. The maximum number of capturing subpatterns is 65535.
1386 .P
1387 As a convenient shorthand, if any option settings are required at the start of
1388 a non-capturing subpattern, the option letters may appear between the "?" and
1389 the ":". Thus the two patterns
1390 .sp
1391 (?i:saturday|sunday)
1392 (?:(?i)saturday|sunday)
1393 .sp
1394 match exactly the same set of strings. Because alternative branches are tried
1395 from left to right, and options are not reset until the end of the subpattern
1396 is reached, an option setting in one branch does affect subsequent branches, so
1397 the above patterns match "SUNDAY" as well as "Saturday".
1398 .
1399 .
1400 .\" HTML <a name="dupsubpatternnumber"></a>
1402 .rs
1403 .sp
1404 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1405 the same numbers for its capturing parentheses. Such a subpattern starts with
1406 (?| and is itself a non-capturing subpattern. For example, consider this
1407 pattern:
1408 .sp
1409 (?|(Sat)ur|(Sun))day
1410 .sp
1411 Because the two alternatives are inside a (?| group, both sets of capturing
1412 parentheses are numbered one. Thus, when the pattern matches, you can look
1413 at captured substring number one, whichever alternative matched. This construct
1414 is useful when you want to capture part, but not all, of one of a number of
1415 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1416 number is reset at the start of each branch. The numbers of any capturing
1417 parentheses that follow the subpattern start after the highest number used in
1418 any branch. The following example is taken from the Perl documentation. The
1419 numbers underneath show in which buffer the captured content will be stored.
1420 .sp
1421 # before ---------------branch-reset----------- after
1422 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1423 # 1 2 2 3 2 3 4
1424 .sp
1425 A back reference to a numbered subpattern uses the most recent value that is
1426 set for that number by any subpattern. The following pattern matches "abcabc"
1427 or "defdef":
1428 .sp
1429 /(?|(abc)|(def))\e1/
1430 .sp
1431 In contrast, a subroutine call to a numbered subpattern always refers to the
1432 first one in the pattern with the given number. The following pattern matches
1433 "abcabc" or "defabc":
1434 .sp
1435 /(?|(abc)|(def))(?1)/
1436 .sp
1437 If a
1438 .\" HTML <a href="#conditions">
1439 .\" </a>
1440 condition test
1441 .\"
1442 for a subpattern's having matched refers to a non-unique number, the test is
1443 true if any of the subpatterns of that number have matched.
1444 .P
1445 An alternative approach to using this "branch reset" feature is to use
1446 duplicate named subpatterns, as described in the next section.
1447 .
1448 .
1450 .rs
1451 .sp
1452 Identifying capturing parentheses by number is simple, but it can be very hard
1453 to keep track of the numbers in complicated regular expressions. Furthermore,
1454 if an expression is modified, the numbers may change. To help with this
1455 difficulty, PCRE supports the naming of subpatterns. This feature was not
1456 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1457 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1458 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1459 have different names, but PCRE does not.
1460 .P
1461 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1462 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1463 parentheses from other parts of the pattern, such as
1464 .\" HTML <a href="#backreferences">
1465 .\" </a>
1466 back references,
1467 .\"
1468 .\" HTML <a href="#recursion">
1469 .\" </a>
1470 recursion,
1471 .\"
1472 and
1473 .\" HTML <a href="#conditions">
1474 .\" </a>
1475 conditions,
1476 .\"
1477 can be made by name as well as by number.
1478 .P
1479 Names consist of up to 32 alphanumeric characters and underscores. Named
1480 capturing parentheses are still allocated numbers as well as names, exactly as
1481 if the names were not present. The PCRE API provides function calls for
1482 extracting the name-to-number translation table from a compiled pattern. There
1483 is also a convenience function for extracting a captured substring by name.
1484 .P
1485 By default, a name must be unique within a pattern, but it is possible to relax
1486 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1487 names are also always permitted for subpatterns with the same number, set up as
1488 described in the previous section.) Duplicate names can be useful for patterns
1489 where only one instance of the named parentheses can match. Suppose you want to
1490 match the name of a weekday, either as a 3-letter abbreviation or as the full
1491 name, and in both cases you want to extract the abbreviation. This pattern
1492 (ignoring the line breaks) does the job:
1493 .sp
1494 (?<DN>Mon|Fri|Sun)(?:day)?|
1495 (?<DN>Tue)(?:sday)?|
1496 (?<DN>Wed)(?:nesday)?|
1497 (?<DN>Thu)(?:rsday)?|
1498 (?<DN>Sat)(?:urday)?
1499 .sp
1500 There are five capturing substrings, but only one is ever set after a match.
1501 (An alternative way of solving this problem is to use a "branch reset"
1502 subpattern, as described in the previous section.)
1503 .P
1504 The convenience function for extracting the data by name returns the substring
1505 for the first (and in this example, the only) subpattern of that name that
1506 matched. This saves searching to find which numbered subpattern it was.
1507 .P
1508 If you make a back reference to a non-unique named subpattern from elsewhere in
1509 the pattern, the one that corresponds to the first occurrence of the name is
1510 used. In the absence of duplicate numbers (see the previous section) this is
1511 the one with the lowest number. If you use a named reference in a condition
1512 test (see the
1513 .\"
1514 .\" HTML <a href="#conditions">
1515 .\" </a>
1516 section about conditions
1517 .\"
1518 below), either to check whether a subpattern has matched, or to check for
1519 recursion, all subpatterns with the same name are tested. If the condition is
1520 true for any one of them, the overall condition is true. This is the same
1521 behaviour as testing by number. For further details of the interfaces for
1522 handling named subpatterns, see the
1523 .\" HREF
1524 \fBpcreapi\fP
1525 .\"
1526 documentation.
1527 .P
1528 \fBWarning:\fP You cannot use different names to distinguish between two
1529 subpatterns with the same number because PCRE uses only the numbers when
1530 matching. For this reason, an error is given at compile time if different names
1531 are given to subpatterns with the same number. However, you can give the same
1532 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1533 .
1534 .
1536 .rs
1537 .sp
1538 Repetition is specified by quantifiers, which can follow any of the following
1539 items:
1540 .sp
1541 a literal data character
1542 the dot metacharacter
1543 the \eC escape sequence
1544 the \eX escape sequence
1545 the \eR escape sequence
1546 an escape such as \ed or \epL that matches a single character
1547 a character class
1548 a back reference (see next section)
1549 a parenthesized subpattern (including assertions)
1550 a subroutine call to a subpattern (recursive or otherwise)
1551 .sp
1552 The general repetition quantifier specifies a minimum and maximum number of
1553 permitted matches, by giving the two numbers in curly brackets (braces),
1554 separated by a comma. The numbers must be less than 65536, and the first must
1555 be less than or equal to the second. For example:
1556 .sp
1557 z{2,4}
1558 .sp
1559 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1560 character. If the second number is omitted, but the comma is present, there is
1561 no upper limit; if the second number and the comma are both omitted, the
1562 quantifier specifies an exact number of required matches. Thus
1563 .sp
1564 [aeiou]{3,}
1565 .sp
1566 matches at least 3 successive vowels, but may match many more, while
1567 .sp
1568 \ed{8}
1569 .sp
1570 matches exactly 8 digits. An opening curly bracket that appears in a position
1571 where a quantifier is not allowed, or one that does not match the syntax of a
1572 quantifier, is taken as a literal character. For example, {,6} is not a
1573 quantifier, but a literal string of four characters.
1574 .P
1575 In UTF modes, quantifiers apply to characters rather than to individual data
1576 units. Thus, for example, \ex{100}{2} matches two characters, each of
1577 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1578 \eX{3} matches three Unicode extended grapheme clusters, each of which may be
1579 several data units long (and they may be of different lengths).
1580 .P
1581 The quantifier {0} is permitted, causing the expression to behave as if the
1582 previous item and the quantifier were not present. This may be useful for
1583 subpatterns that are referenced as
1584 .\" HTML <a href="#subpatternsassubroutines">
1585 .\" </a>
1586 subroutines
1587 .\"
1588 from elsewhere in the pattern (but see also the section entitled
1589 .\" HTML <a href="#subdefine">
1590 .\" </a>
1591 "Defining subpatterns for use by reference only"
1592 .\"
1593 below). Items other than subpatterns that have a {0} quantifier are omitted
1594 from the compiled pattern.
1595 .P
1596 For convenience, the three most common quantifiers have single-character
1597 abbreviations:
1598 .sp
1599 * is equivalent to {0,}
1600 + is equivalent to {1,}
1601 ? is equivalent to {0,1}
1602 .sp
1603 It is possible to construct infinite loops by following a subpattern that can
1604 match no characters with a quantifier that has no upper limit, for example:
1605 .sp
1606 (a?)*
1607 .sp
1608 Earlier versions of Perl and PCRE used to give an error at compile time for
1609 such patterns. However, because there are cases where this can be useful, such
1610 patterns are now accepted, but if any repetition of the subpattern does in fact
1611 match no characters, the loop is forcibly broken.
1612 .P
1613 By default, the quantifiers are "greedy", that is, they match as much as
1614 possible (up to the maximum number of permitted times), without causing the
1615 rest of the pattern to fail. The classic example of where this gives problems
1616 is in trying to match comments in C programs. These appear between /* and */
1617 and within the comment, individual * and / characters may appear. An attempt to
1618 match C comments by applying the pattern
1619 .sp
1620 /\e*.*\e*/
1621 .sp
1622 to the string
1623 .sp
1624 /* first comment */ not comment /* second comment */
1625 .sp
1626 fails, because it matches the entire string owing to the greediness of the .*
1627 item.
1628 .P
1629 However, if a quantifier is followed by a question mark, it ceases to be
1630 greedy, and instead matches the minimum number of times possible, so the
1631 pattern
1632 .sp
1633 /\e*.*?\e*/
1634 .sp
1635 does the right thing with the C comments. The meaning of the various
1636 quantifiers is not otherwise changed, just the preferred number of matches.
1637 Do not confuse this use of question mark with its use as a quantifier in its
1638 own right. Because it has two uses, it can sometimes appear doubled, as in
1639 .sp
1640 \ed??\ed
1641 .sp
1642 which matches one digit by preference, but can match two if that is the only
1643 way the rest of the pattern matches.
1644 .P
1645 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1646 the quantifiers are not greedy by default, but individual ones can be made
1647 greedy by following them with a question mark. In other words, it inverts the
1648 default behaviour.
1649 .P
1650 When a parenthesized subpattern is quantified with a minimum repeat count that
1651 is greater than 1 or with a limited maximum, more memory is required for the
1652 compiled pattern, in proportion to the size of the minimum or maximum.
1653 .P
1654 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1655 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1656 implicitly anchored, because whatever follows will be tried against every
1657 character position in the subject string, so there is no point in retrying the
1658 overall match at any position after the first. PCRE normally treats such a
1659 pattern as though it were preceded by \eA.
1660 .P
1661 In cases where it is known that the subject string contains no newlines, it is
1662 worth setting PCRE_DOTALL in order to obtain this optimization, or
1663 alternatively using ^ to indicate anchoring explicitly.
1664 .P
1665 However, there are some cases where the optimization cannot be used. When .*
1666 is inside capturing parentheses that are the subject of a back reference
1667 elsewhere in the pattern, a match at the start may fail where a later one
1668 succeeds. Consider, for example:
1669 .sp
1670 (.*)abc\e1
1671 .sp
1672 If the subject is "xyz123abc123" the match point is the fourth character. For
1673 this reason, such a pattern is not implicitly anchored.
1674 .P
1675 Another case where implicit anchoring is not applied is when the leading .* is
1676 inside an atomic group. Once again, a match at the start may fail where a later
1677 one succeeds. Consider this pattern:
1678 .sp
1679 (?>.*?a)b
1680 .sp
1681 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1682 (*PRUNE) and (*SKIP) also disable this optimization.
1683 .P
1684 When a capturing subpattern is repeated, the value captured is the substring
1685 that matched the final iteration. For example, after
1686 .sp
1687 (tweedle[dume]{3}\es*)+
1688 .sp
1689 has matched "tweedledum tweedledee" the value of the captured substring is
1690 "tweedledee". However, if there are nested capturing subpatterns, the
1691 corresponding captured values may have been set in previous iterations. For
1692 example, after
1693 .sp
1694 /(a|(b))+/
1695 .sp
1696 matches "aba" the value of the second captured substring is "b".
1697 .
1698 .
1699 .\" HTML <a name="atomicgroup"></a>
1701 .rs
1702 .sp
1703 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1704 repetition, failure of what follows normally causes the repeated item to be
1705 re-evaluated to see if a different number of repeats allows the rest of the
1706 pattern to match. Sometimes it is useful to prevent this, either to change the
1707 nature of the match, or to cause it fail earlier than it otherwise might, when
1708 the author of the pattern knows there is no point in carrying on.
1709 .P
1710 Consider, for example, the pattern \ed+foo when applied to the subject line
1711 .sp
1712 123456bar
1713 .sp
1714 After matching all 6 digits and then failing to match "foo", the normal
1715 action of the matcher is to try again with only 5 digits matching the \ed+
1716 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1717 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1718 that once a subpattern has matched, it is not to be re-evaluated in this way.
1719 .P
1720 If we use atomic grouping for the previous example, the matcher gives up
1721 immediately on failing to match "foo" the first time. The notation is a kind of
1722 special parenthesis, starting with (?> as in this example:
1723 .sp
1724 (?>\ed+)foo
1725 .sp
1726 This kind of parenthesis "locks up" the part of the pattern it contains once
1727 it has matched, and a failure further into the pattern is prevented from
1728 backtracking into it. Backtracking past it to previous items, however, works as
1729 normal.
1730 .P
1731 An alternative description is that a subpattern of this type matches the string
1732 of characters that an identical standalone pattern would match, if anchored at
1733 the current point in the subject string.
1734 .P
1735 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1736 the above example can be thought of as a maximizing repeat that must swallow
1737 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1738 number of digits they match in order to make the rest of the pattern match,
1739 (?>\ed+) can only match an entire sequence of digits.
1740 .P
1741 Atomic groups in general can of course contain arbitrarily complicated
1742 subpatterns, and can be nested. However, when the subpattern for an atomic
1743 group is just a single repeated item, as in the example above, a simpler
1744 notation, called a "possessive quantifier" can be used. This consists of an
1745 additional + character following a quantifier. Using this notation, the
1746 previous example can be rewritten as
1747 .sp
1748 \ed++foo
1749 .sp
1750 Note that a possessive quantifier can be used with an entire group, for
1751 example:
1752 .sp
1753 (abc|xyz){2,3}+
1754 .sp
1755 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1756 option is ignored. They are a convenient notation for the simpler forms of
1757 atomic group. However, there is no difference in the meaning of a possessive
1758 quantifier and the equivalent atomic group, though there may be a performance
1759 difference; possessive quantifiers should be slightly faster.
1760 .P
1761 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1762 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1763 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1764 package, and PCRE copied it from there. It ultimately found its way into Perl
1765 at release 5.10.
1766 .P
1767 PCRE has an optimization that automatically "possessifies" certain simple
1768 pattern constructs. For example, the sequence A+B is treated as A++B because
1769 there is no point in backtracking into a sequence of A's when B must follow.
1770 .P
1771 When a pattern contains an unlimited repeat inside a subpattern that can itself
1772 be repeated an unlimited number of times, the use of an atomic group is the
1773 only way to avoid some failing matches taking a very long time indeed. The
1774 pattern
1775 .sp
1776 (\eD+|<\ed+>)*[!?]
1777 .sp
1778 matches an unlimited number of substrings that either consist of non-digits, or
1779 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1780 quickly. However, if it is applied to
1781 .sp
1782 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1783 .sp
1784 it takes a long time before reporting failure. This is because the string can
1785 be divided between the internal \eD+ repeat and the external * repeat in a
1786 large number of ways, and all have to be tried. (The example uses [!?] rather
1787 than a single character at the end, because both PCRE and Perl have an
1788 optimization that allows for fast failure when a single character is used. They
1789 remember the last single character that is required for a match, and fail early
1790 if it is not present in the string.) If the pattern is changed so that it uses
1791 an atomic group, like this:
1792 .sp
1793 ((?>\eD+)|<\ed+>)*[!?]
1794 .sp
1795 sequences of non-digits cannot be broken, and failure happens quickly.
1796 .
1797 .
1798 .\" HTML <a name="backreferences"></a>
1800 .rs
1801 .sp
1802 Outside a character class, a backslash followed by a digit greater than 0 (and
1803 possibly further digits) is a back reference to a capturing subpattern earlier
1804 (that is, to its left) in the pattern, provided there have been that many
1805 previous capturing left parentheses.
1806 .P
1807 However, if the decimal number following the backslash is less than 10, it is
1808 always taken as a back reference, and causes an error only if there are not
1809 that many capturing left parentheses in the entire pattern. In other words, the
1810 parentheses that are referenced need not be to the left of the reference for
1811 numbers less than 10. A "forward back reference" of this type can make sense
1812 when a repetition is involved and the subpattern to the right has participated
1813 in an earlier iteration.
1814 .P
1815 It is not possible to have a numerical "forward back reference" to a subpattern
1816 whose number is 10 or more using this syntax because a sequence such as \e50 is
1817 interpreted as a character defined in octal. See the subsection entitled
1818 "Non-printing characters"
1819 .\" HTML <a href="#digitsafterbackslash">
1820 .\" </a>
1821 above
1822 .\"
1823 for further details of the handling of digits following a backslash. There is
1824 no such problem when named parentheses are used. A back reference to any
1825 subpattern is possible using named parentheses (see below).
1826 .P
1827 Another way of avoiding the ambiguity inherent in the use of digits following a
1828 backslash is to use the \eg escape sequence. This escape must be followed by an
1829 unsigned number or a negative number, optionally enclosed in braces. These
1830 examples are all identical:
1831 .sp
1832 (ring), \e1
1833 (ring), \eg1
1834 (ring), \eg{1}
1835 .sp
1836 An unsigned number specifies an absolute reference without the ambiguity that
1837 is present in the older syntax. It is also useful when literal digits follow
1838 the reference. A negative number is a relative reference. Consider this
1839 example:
1840 .sp
1841 (abc(def)ghi)\eg{-1}
1842 .sp
1843 The sequence \eg{-1} is a reference to the most recently started capturing
1844 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1845 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1846 can be helpful in long patterns, and also in patterns that are created by
1847 joining together fragments that contain references within themselves.
1848 .P
1849 A back reference matches whatever actually matched the capturing subpattern in
1850 the current subject string, rather than anything matching the subpattern
1851 itself (see
1852 .\" HTML <a href="#subpatternsassubroutines">
1853 .\" </a>
1854 "Subpatterns as subroutines"
1855 .\"
1856 below for a way of doing that). So the pattern
1857 .sp
1858 (sens|respons)e and \e1ibility
1859 .sp
1860 matches "sense and sensibility" and "response and responsibility", but not
1861 "sense and responsibility". If caseful matching is in force at the time of the
1862 back reference, the case of letters is relevant. For example,
1863 .sp
1864 ((?i)rah)\es+\e1
1865 .sp
1866 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1867 capturing subpattern is matched caselessly.
1868 .P
1869 There are several different ways of writing back references to named
1870 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1871 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1872 back reference syntax, in which \eg can be used for both numeric and named
1873 references, is also supported. We could rewrite the above example in any of
1874 the following ways:
1875 .sp
1876 (?<p1>(?i)rah)\es+\ek<p1>
1877 (?'p1'(?i)rah)\es+\ek{p1}
1878 (?P<p1>(?i)rah)\es+(?P=p1)
1879 (?<p1>(?i)rah)\es+\eg{p1}
1880 .sp
1881 A subpattern that is referenced by name may appear in the pattern before or
1882 after the reference.
1883 .P
1884 There may be more than one back reference to the same subpattern. If a
1885 subpattern has not actually been used in a particular match, any back
1886 references to it always fail by default. For example, the pattern
1887 .sp
1888 (a|(bc))\e2
1889 .sp
1890 always fails if it starts to match "a" rather than "bc". However, if the
1891 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1892 unset value matches an empty string.
1893 .P
1894 Because there may be many capturing parentheses in a pattern, all digits
1895 following a backslash are taken as part of a potential back reference number.
1896 If the pattern continues with a digit character, some delimiter must be used to
1897 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1898 white space. Otherwise, the \eg{ syntax or an empty comment (see
1899 .\" HTML <a href="#comments">
1900 .\" </a>
1901 "Comments"
1902 .\"
1903 below) can be used.
1904 .
1905 .SS "Recursive back references"
1906 .rs
1907 .sp
1908 A back reference that occurs inside the parentheses to which it refers fails
1909 when the subpattern is first used, so, for example, (a\e1) never matches.
1910 However, such references can be useful inside repeated subpatterns. For
1911 example, the pattern
1912 .sp
1913 (a|b\e1)+
1914 .sp
1915 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1916 the subpattern, the back reference matches the character string corresponding
1917 to the previous iteration. In order for this to work, the pattern must be such
1918 that the first iteration does not need to match the back reference. This can be
1919 done using alternation, as in the example above, or by a quantifier with a
1920 minimum of zero.
1921 .P
1922 Back references of this type cause the group that they reference to be treated
1923 as an
1924 .\" HTML <a href="#atomicgroup">
1925 .\" </a>
1926 atomic group.
1927 .\"
1928 Once the whole group has been matched, a subsequent matching failure cannot
1929 cause backtracking into the middle of the group.
1930 .
1931 .
1932 .\" HTML <a name="bigassertions"></a>
1934 .rs
1935 .sp
1936 An assertion is a test on the characters following or preceding the current
1937 matching point that does not actually consume any characters. The simple
1938 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1939 .\" HTML <a href="#smallassertions">
1940 .\" </a>
1941 above.
1942 .\"
1943 .P
1944 More complicated assertions are coded as subpatterns. There are two kinds:
1945 those that look ahead of the current position in the subject string, and those
1946 that look behind it. An assertion subpattern is matched in the normal way,
1947 except that it does not cause the current matching position to be changed.
1948 .P
1949 Assertion subpatterns are not capturing subpatterns. If such an assertion
1950 contains capturing subpatterns within it, these are counted for the purposes of
1951 numbering the capturing subpatterns in the whole pattern. However, substring
1952 capturing is carried out only for positive assertions, because it does not make
1953 sense for negative assertions.
1954 .P
1955 For compatibility with Perl, assertion subpatterns may be repeated; though
1956 it makes no sense to assert the same thing several times, the side effect of
1957 capturing parentheses may occasionally be useful. In practice, there only three
1958 cases:
1959 .sp
1960 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1961 However, it may contain internal capturing parenthesized groups that are called
1962 from elsewhere via the
1963 .\" HTML <a href="#subpatternsassubroutines">
1964 .\" </a>
1965 subroutine mechanism.
1966 .\"
1967 .sp
1968 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1969 were {0,1}. At run time, the rest of the pattern match is tried with and
1970 without the assertion, the order depending on the greediness of the quantifier.
1971 .sp
1972 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1973 The assertion is obeyed just once when encountered during matching.
1974 .
1975 .
1976 .SS "Lookahead assertions"
1977 .rs
1978 .sp
1979 Lookahead assertions start with (?= for positive assertions and (?! for
1980 negative assertions. For example,
1981 .sp
1982 \ew+(?=;)
1983 .sp
1984 matches a word followed by a semicolon, but does not include the semicolon in
1985 the match, and
1986 .sp
1987 foo(?!bar)
1988 .sp
1989 matches any occurrence of "foo" that is not followed by "bar". Note that the
1990 apparently similar pattern
1991 .sp
1992 (?!foo)bar
1993 .sp
1994 does not find an occurrence of "bar" that is preceded by something other than
1995 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1996 (?!foo) is always true when the next three characters are "bar". A
1997 lookbehind assertion is needed to achieve the other effect.
1998 .P
1999 If you want to force a matching failure at some point in a pattern, the most
2000 convenient way to do it is with (?!) because an empty string always matches, so
2001 an assertion that requires there not to be an empty string must always fail.
2002 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2003 .
2004 .
2005 .\" HTML <a name="lookbehind"></a>
2006 .SS "Lookbehind assertions"
2007 .rs
2008 .sp
2009 Lookbehind assertions start with (?<= for positive assertions and (?<! for
2010 negative assertions. For example,
2011 .sp
2012 (?<!foo)bar
2013 .sp
2014 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2015 a lookbehind assertion are restricted such that all the strings it matches must
2016 have a fixed length. However, if there are several top-level alternatives, they
2017 do not all have to have the same fixed length. Thus
2018 .sp
2019 (?<=bullock|donkey)
2020 .sp
2021 is permitted, but
2022 .sp
2023 (?<!dogs?|cats?)
2024 .sp
2025 causes an error at compile time. Branches that match different length strings
2026 are permitted only at the top level of a lookbehind assertion. This is an
2027 extension compared with Perl, which requires all branches to match the same
2028 length of string. An assertion such as
2029 .sp
2030 (?<=ab(c|de))
2031 .sp
2032 is not permitted, because its single top-level branch can match two different
2033 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2034 branches:
2035 .sp
2036 (?<=abc|abde)
2037 .sp
2038 In some cases, the escape sequence \eK
2039 .\" HTML <a href="#resetmatchstart">
2040 .\" </a>
2041 (see above)
2042 .\"
2043 can be used instead of a lookbehind assertion to get round the fixed-length
2044 restriction.
2045 .P
2046 The implementation of lookbehind assertions is, for each alternative, to
2047 temporarily move the current position back by the fixed length and then try to
2048 match. If there are insufficient characters before the current position, the
2049 assertion fails.
2050 .P
2051 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2052 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2053 it impossible to calculate the length of the lookbehind. The \eX and \eR
2054 escapes, which can match different numbers of data units, are also not
2055 permitted.
2056 .P
2057 .\" HTML <a href="#subpatternsassubroutines">
2058 .\" </a>
2059 "Subroutine"
2060 .\"
2061 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2062 as the subpattern matches a fixed-length string.
2063 .\" HTML <a href="#recursion">
2064 .\" </a>
2065 Recursion,
2066 .\"
2067 however, is not supported.
2068 .P
2069 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2070 specify efficient matching of fixed-length strings at the end of subject
2071 strings. Consider a simple pattern such as
2072 .sp
2073 abcd$
2074 .sp
2075 when applied to a long string that does not match. Because matching proceeds
2076 from left to right, PCRE will look for each "a" in the subject and then see if
2077 what follows matches the rest of the pattern. If the pattern is specified as
2078 .sp
2079 ^.*abcd$
2080 .sp
2081 the initial .* matches the entire string at first, but when this fails (because
2082 there is no following "a"), it backtracks to match all but the last character,
2083 then all but the last two characters, and so on. Once again the search for "a"
2084 covers the entire string, from right to left, so we are no better off. However,
2085 if the pattern is written as
2086 .sp
2087 ^.*+(?<=abcd)
2088 .sp
2089 there can be no backtracking for the .*+ item; it can match only the entire
2090 string. The subsequent lookbehind assertion does a single test on the last four
2091 characters. If it fails, the match fails immediately. For long strings, this
2092 approach makes a significant difference to the processing time.
2093 .
2094 .
2095 .SS "Using multiple assertions"
2096 .rs
2097 .sp
2098 Several assertions (of any sort) may occur in succession. For example,
2099 .sp
2100 (?<=\ed{3})(?<!999)foo
2101 .sp
2102 matches "foo" preceded by three digits that are not "999". Notice that each of
2103 the assertions is applied independently at the same point in the subject
2104 string. First there is a check that the previous three characters are all
2105 digits, and then there is a check that the same three characters are not "999".
2106 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2107 of which are digits and the last three of which are not "999". For example, it
2108 doesn't match "123abcfoo". A pattern to do that is
2109 .sp
2110 (?<=\ed{3}...)(?<!999)foo
2111 .sp
2112 This time the first assertion looks at the preceding six characters, checking
2113 that the first three are digits, and then the second assertion checks that the
2114 preceding three characters are not "999".
2115 .P
2116 Assertions can be nested in any combination. For example,
2117 .sp
2118 (?<=(?<!foo)bar)baz
2119 .sp
2120 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2121 preceded by "foo", while
2122 .sp
2123 (?<=\ed{3}(?!999)...)foo
2124 .sp
2125 is another pattern that matches "foo" preceded by three digits and any three
2126 characters that are not "999".
2127 .
2128 .
2129 .\" HTML <a name="conditions"></a>
2131 .rs
2132 .sp
2133 It is possible to cause the matching process to obey a subpattern
2134 conditionally or to choose between two alternative subpatterns, depending on
2135 the result of an assertion, or whether a specific capturing subpattern has
2136 already been matched. The two possible forms of conditional subpattern are:
2137 .sp
2138 (?(condition)yes-pattern)
2139 (?(condition)yes-pattern|no-pattern)
2140 .sp
2141 If the condition is satisfied, the yes-pattern is used; otherwise the
2142 no-pattern (if present) is used. If there are more than two alternatives in the
2143 subpattern, a compile-time error occurs. Each of the two alternatives may
2144 itself contain nested subpatterns of any form, including conditional
2145 subpatterns; the restriction to two alternatives applies only at the level of
2146 the condition. This pattern fragment is an example where the alternatives are
2147 complex:
2148 .sp
2149 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2150 .sp
2151 .P
2152 There are four kinds of condition: references to subpatterns, references to
2153 recursion, a pseudo-condition called DEFINE, and assertions.
2154 .
2155 .SS "Checking for a used subpattern by number"
2156 .rs
2157 .sp
2158 If the text between the parentheses consists of a sequence of digits, the
2159 condition is true if a capturing subpattern of that number has previously
2160 matched. If there is more than one capturing subpattern with the same number
2161 (see the earlier
2162 .\"
2163 .\" HTML <a href="#recursion">
2164 .\" </a>
2165 section about duplicate subpattern numbers),
2166 .\"
2167 the condition is true if any of them have matched. An alternative notation is
2168 to precede the digits with a plus or minus sign. In this case, the subpattern
2169 number is relative rather than absolute. The most recently opened parentheses
2170 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2171 loops it can also make sense to refer to subsequent groups. The next
2172 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2173 zero in any of these forms is not used; it provokes a compile-time error.)
2174 .P
2175 Consider the following pattern, which contains non-significant white space to
2176 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2177 three parts for ease of discussion:
2178 .sp
2179 ( \e( )? [^()]+ (?(1) \e) )
2180 .sp
2181 The first part matches an optional opening parenthesis, and if that
2182 character is present, sets it as the first captured substring. The second part
2183 matches one or more characters that are not parentheses. The third part is a
2184 conditional subpattern that tests whether or not the first set of parentheses
2185 matched. If they did, that is, if subject started with an opening parenthesis,
2186 the condition is true, and so the yes-pattern is executed and a closing
2187 parenthesis is required. Otherwise, since no-pattern is not present, the
2188 subpattern matches nothing. In other words, this pattern matches a sequence of
2189 non-parentheses, optionally enclosed in parentheses.
2190 .P
2191 If you were embedding this pattern in a larger one, you could use a relative
2192 reference:
2193 .sp
2194 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2195 .sp
2196 This makes the fragment independent of the parentheses in the larger pattern.
2197 .
2198 .SS "Checking for a used subpattern by name"
2199 .rs
2200 .sp
2201 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2202 subpattern by name. For compatibility with earlier versions of PCRE, which had
2203 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2204 there is a possible ambiguity with this syntax, because subpattern names may
2205 consist entirely of digits. PCRE looks first for a named subpattern; if it
2206 cannot find one and the name consists entirely of digits, PCRE looks for a
2207 subpattern of that number, which must be greater than zero. Using subpattern
2208 names that consist entirely of digits is not recommended.
2209 .P
2210 Rewriting the above example to use a named subpattern gives this:
2211 .sp
2212 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2213 .sp
2214 If the name used in a condition of this kind is a duplicate, the test is
2215 applied to all subpatterns of the same name, and is true if any one of them has
2216 matched.
2217 .
2218 .SS "Checking for pattern recursion"
2219 .rs
2220 .sp
2221 If the condition is the string (R), and there is no subpattern with the name R,
2222 the condition is true if a recursive call to the whole pattern or any
2223 subpattern has been made. If digits or a name preceded by ampersand follow the
2224 letter R, for example:
2225 .sp
2226 (?(R3)...) or (?(R&name)...)
2227 .sp
2228 the condition is true if the most recent recursion is into a subpattern whose
2229 number or name is given. This condition does not check the entire recursion
2230 stack. If the name used in a condition of this kind is a duplicate, the test is
2231 applied to all subpatterns of the same name, and is true if any one of them is
2232 the most recent recursion.
2233 .P
2234 At "top level", all these recursion test conditions are false.
2235 .\" HTML <a href="#recursion">
2236 .\" </a>
2237 The syntax for recursive patterns
2238 .\"
2239 is described below.
2240 .
2241 .\" HTML <a name="subdefine"></a>
2242 .SS "Defining subpatterns for use by reference only"
2243 .rs
2244 .sp
2245 If the condition is the string (DEFINE), and there is no subpattern with the
2246 name DEFINE, the condition is always false. In this case, there may be only one
2247 alternative in the subpattern. It is always skipped if control reaches this
2248 point in the pattern; the idea of DEFINE is that it can be used to define
2249 subroutines that can be referenced from elsewhere. (The use of
2250 .\" HTML <a href="#subpatternsassubroutines">
2251 .\" </a>
2252 subroutines
2253 .\"
2254 is described below.) For example, a pattern to match an IPv4 address such as
2255 "" could be written like this (ignore white space and line
2256 breaks):
2257 .sp
2258 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2259 \eb (?&byte) (\e.(?&byte)){3} \eb
2260 .sp
2261 The first part of the pattern is a DEFINE group inside which a another group
2262 named "byte" is defined. This matches an individual component of an IPv4
2263 address (a number less than 256). When matching takes place, this part of the
2264 pattern is skipped because DEFINE acts like a false condition. The rest of the
2265 pattern uses references to the named group to match the four dot-separated
2266 components of an IPv4 address, insisting on a word boundary at each end.
2267 .
2268 .SS "Assertion conditions"
2269 .rs
2270 .sp
2271 If the condition is not in any of the above formats, it must be an assertion.
2272 This may be a positive or negative lookahead or lookbehind assertion. Consider
2273 this pattern, again containing non-significant white space, and with the two
2274 alternatives on the second line:
2275 .sp
2276 (?(?=[^a-z]*[a-z])
2277 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2278 .sp
2279 The condition is a positive lookahead assertion that matches an optional
2280 sequence of non-letters followed by a letter. In other words, it tests for the
2281 presence of at least one letter in the subject. If a letter is found, the
2282 subject is matched against the first alternative; otherwise it is matched
2283 against the second. This pattern matches strings in one of the two forms
2284 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2285 .
2286 .
2287 .\" HTML <a name="comments"></a>
2289 .rs
2290 .sp
2291 There are two ways of including comments in patterns that are processed by
2292 PCRE. In both cases, the start of the comment must not be in a character class,
2293 nor in the middle of any other sequence of related characters such as (?: or a
2294 subpattern name or number. The characters that make up a comment play no part
2295 in the pattern matching.
2296 .P
2297 The sequence (?# marks the start of a comment that continues up to the next
2298 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2299 option is set, an unescaped # character also introduces a comment, which in
2300 this case continues to immediately after the next newline character or
2301 character sequence in the pattern. Which characters are interpreted as newlines
2302 is controlled by the options passed to a compiling function or by a special
2303 sequence at the start of the pattern, as described in the section entitled
2304 .\" HTML <a href="#newlines">
2305 .\" </a>
2306 "Newline conventions"
2307 .\"
2308 above. Note that the end of this type of comment is a literal newline sequence
2309 in the pattern; escape sequences that happen to represent a newline do not
2310 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2311 default newline convention is in force:
2312 .sp
2313 abc #comment \en still comment
2314 .sp
2315 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2316 a newline in the pattern. The sequence \en is still literal at this stage, so
2317 it does not terminate the comment. Only an actual character with the code value
2318 0x0a (the default newline) does so.
2319 .
2320 .
2321 .\" HTML <a name="recursion"></a>
2323 .rs
2324 .sp
2325 Consider the problem of matching a string in parentheses, allowing for
2326 unlimited nested parentheses. Without the use of recursion, the best that can
2327 be done is to use a pattern that matches up to some fixed depth of nesting. It
2328 is not possible to handle an arbitrary nesting depth.
2329 .P
2330 For some time, Perl has provided a facility that allows regular expressions to
2331 recurse (amongst other things). It does this by interpolating Perl code in the
2332 expression at run time, and the code can refer to the expression itself. A Perl
2333 pattern using code interpolation to solve the parentheses problem can be
2334 created like this:
2335 .sp
2336 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2337 .sp
2338 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2339 recursively to the pattern in which it appears.
2340 .P
2341 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2342 supports special syntax for recursion of the entire pattern, and also for
2343 individual subpattern recursion. After its introduction in PCRE and Python,
2344 this kind of recursion was subsequently introduced into Perl at release 5.10.
2345 .P
2346 A special item that consists of (? followed by a number greater than zero and a
2347 closing parenthesis is a recursive subroutine call of the subpattern of the
2348 given number, provided that it occurs inside that subpattern. (If not, it is a
2349 .\" HTML <a href="#subpatternsassubroutines">
2350 .\" </a>
2351 non-recursive subroutine
2352 .\"
2353 call, which is described in the next section.) The special item (?R) or (?0) is
2354 a recursive call of the entire regular expression.
2355 .P
2356 This PCRE pattern solves the nested parentheses problem (assume the
2357 PCRE_EXTENDED option is set so that white space is ignored):
2358 .sp
2359 \e( ( [^()]++ | (?R) )* \e)
2360 .sp
2361 First it matches an opening parenthesis. Then it matches any number of
2362 substrings which can either be a sequence of non-parentheses, or a recursive
2363 match of the pattern itself (that is, a correctly parenthesized substring).
2364 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2365 to avoid backtracking into sequences of non-parentheses.
2366 .P
2367 If this were part of a larger pattern, you would not want to recurse the entire
2368 pattern, so instead you could use this:
2369 .sp
2370 ( \e( ( [^()]++ | (?1) )* \e) )
2371 .sp
2372 We have put the pattern into parentheses, and caused the recursion to refer to
2373 them instead of the whole pattern.
2374 .P
2375 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2376 is made easier by the use of relative references. Instead of (?1) in the
2377 pattern above you can write (?-2) to refer to the second most recently opened
2378 parentheses preceding the recursion. In other words, a negative number counts
2379 capturing parentheses leftwards from the point at which it is encountered.
2380 .P
2381 It is also possible to refer to subsequently opened parentheses, by writing
2382 references such as (?+2). However, these cannot be recursive because the
2383 reference is not inside the parentheses that are referenced. They are always
2384 .\" HTML <a href="#subpatternsassubroutines">
2385 .\" </a>
2386 non-recursive subroutine
2387 .\"
2388 calls, as described in the next section.
2389 .P
2390 An alternative approach is to use named parentheses instead. The Perl syntax
2391 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2392 could rewrite the above example as follows:
2393 .sp
2394 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2395 .sp
2396 If there is more than one subpattern with the same name, the earliest one is
2397 used.
2398 .P
2399 This particular example pattern that we have been looking at contains nested
2400 unlimited repeats, and so the use of a possessive quantifier for matching
2401 strings of non-parentheses is important when applying the pattern to strings
2402 that do not match. For example, when this pattern is applied to
2403 .sp
2404 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2405 .sp
2406 it yields "no match" quickly. However, if a possessive quantifier is not used,
2407 the match runs for a very long time indeed because there are so many different
2408 ways the + and * repeats can carve up the subject, and all have to be tested
2409 before failure can be reported.
2410 .P
2411 At the end of a match, the values of capturing parentheses are those from
2412 the outermost level. If you want to obtain intermediate values, a callout
2413 function can be used (see below and the
2414 .\" HREF
2415 \fBpcrecallout\fP
2416 .\"
2417 documentation). If the pattern above is matched against
2418 .sp
2419 (ab(cd)ef)
2420 .sp
2421 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2422 the last value taken on at the top level. If a capturing subpattern is not
2423 matched at the top level, its final captured value is unset, even if it was
2424 (temporarily) set at a deeper level during the matching process.
2425 .P
2426 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2427 obtain extra memory to store data during a recursion, which it does by using
2428 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2429 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2430 .P
2431 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2432 Consider this pattern, which matches text in angle brackets, allowing for
2433 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2434 recursing), whereas any characters are permitted at the outer level.
2435 .sp
2436 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2437 .sp
2438 In this pattern, (?(R) is the start of a conditional subpattern, with two
2439 different alternatives for the recursive and non-recursive cases. The (?R) item
2440 is the actual recursive call.
2441 .
2442 .
2443 .\" HTML <a name="recursiondifference"></a>
2444 .SS "Differences in recursion processing between PCRE and Perl"
2445 .rs
2446 .sp
2447 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2448 (like Python, but unlike Perl), a recursive subpattern call is always treated
2449 as an atomic group. That is, once it has matched some of the subject string, it
2450 is never re-entered, even if it contains untried alternatives and there is a
2451 subsequent matching failure. This can be illustrated by the following pattern,
2452 which purports to match a palindromic string that contains an odd number of
2453 characters (for example, "a", "aba", "abcba", "abcdcba"):
2454 .sp
2455 ^(.|(.)(?1)\e2)$
2456 .sp
2457 The idea is that it either matches a single character, or two identical
2458 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2459 it does not if the pattern is longer than three characters. Consider the
2460 subject string "abcba":
2461 .P
2462 At the top level, the first character is matched, but as it is not at the end
2463 of the string, the first alternative fails; the second alternative is taken
2464 and the recursion kicks in. The recursive call to subpattern 1 successfully
2465 matches the next character ("b"). (Note that the beginning and end of line
2466 tests are not part of the recursion).
2467 .P
2468 Back at the top level, the next character ("c") is compared with what
2469 subpattern 2 matched, which was "a". This fails. Because the recursion is
2470 treated as an atomic group, there are now no backtracking points, and so the
2471 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2472 try the second alternative.) However, if the pattern is written with the
2473 alternatives in the other order, things are different:
2474 .sp
2475 ^((.)(?1)\e2|.)$
2476 .sp
2477 This time, the recursing alternative is tried first, and continues to recurse
2478 until it runs out of characters, at which point the recursion fails. But this
2479 time we do have another alternative to try at the higher level. That is the big
2480 difference: in the previous case the remaining alternative is at a deeper
2481 recursion level, which PCRE cannot use.
2482 .P
2483 To change the pattern so that it matches all palindromic strings, not just
2484 those with an odd number of characters, it is tempting to change the pattern to
2485 this:
2486 .sp
2487 ^((.)(?1)\e2|.?)$
2488 .sp
2489 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2490 deeper recursion has matched a single character, it cannot be entered again in
2491 order to match an empty string. The solution is to separate the two cases, and
2492 write out the odd and even cases as alternatives at the higher level:
2493 .sp
2494 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2495 .sp
2496 If you want to match typical palindromic phrases, the pattern has to ignore all
2497 non-word characters, which can be done like this:
2498 .sp
2499 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2500 .sp
2501 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2502 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2503 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2504 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2505 more) to match typical phrases, and Perl takes so long that you think it has
2506 gone into a loop.
2507 .P
2508 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2509 string does not start with a palindrome that is shorter than the entire string.
2510 For example, although "abcba" is correctly matched, if the subject is "ababa",
2511 PCRE finds the palindrome "aba" at the start, then fails at top level because
2512 the end of the string does not follow. Once again, it cannot jump back into the
2513 recursion to try other alternatives, so the entire match fails.
2514 .P
2515 The second way in which PCRE and Perl differ in their recursion processing is
2516 in the handling of captured values. In Perl, when a subpattern is called
2517 recursively or as a subpattern (see the next section), it has no access to any
2518 values that were captured outside the recursion, whereas in PCRE these values
2519 can be referenced. Consider this pattern:
2520 .sp
2521 ^(.)(\e1|a(?2))
2522 .sp
2523 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2524 then in the second group, when the back reference \e1 fails to match "b", the
2525 second alternative matches "a" and then recurses. In the recursion, \e1 does
2526 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2527 match because inside the recursive call \e1 cannot access the externally set
2528 value.
2529 .
2530 .
2531 .\" HTML <a name="subpatternsassubroutines"></a>
2533 .rs
2534 .sp
2535 If the syntax for a recursive subpattern call (either by number or by
2536 name) is used outside the parentheses to which it refers, it operates like a
2537 subroutine in a programming language. The called subpattern may be defined
2538 before or after the reference. A numbered reference can be absolute or
2539 relative, as in these examples:
2540 .sp
2541 (...(absolute)...)...(?2)...
2542 (...(relative)...)...(?-1)...
2543 (...(?+1)...(relative)...
2544 .sp
2545 An earlier example pointed out that the pattern
2546 .sp
2547 (sens|respons)e and \e1ibility
2548 .sp
2549 matches "sense and sensibility" and "response and responsibility", but not
2550 "sense and responsibility". If instead the pattern
2551 .sp
2552 (sens|respons)e and (?1)ibility
2553 .sp
2554 is used, it does match "sense and responsibility" as well as the other two
2555 strings. Another example is given in the discussion of DEFINE above.
2556 .P
2557 All subroutine calls, whether recursive or not, are always treated as atomic
2558 groups. That is, once a subroutine has matched some of the subject string, it
2559 is never re-entered, even if it contains untried alternatives and there is a
2560 subsequent matching failure. Any capturing parentheses that are set during the
2561 subroutine call revert to their previous values afterwards.
2562 .P
2563 Processing options such as case-independence are fixed when a subpattern is
2564 defined, so if it is used as a subroutine, such options cannot be changed for
2565 different calls. For example, consider this pattern:
2566 .sp
2567 (abc)(?i:(?-1))
2568 .sp
2569 It matches "abcabc". It does not match "abcABC" because the change of
2570 processing option does not affect the called subpattern.
2571 .
2572 .
2573 .\" HTML <a name="onigurumasubroutines"></a>
2575 .rs
2576 .sp
2577 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2578 a number enclosed either in angle brackets or single quotes, is an alternative
2579 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2580 are two of the examples used above, rewritten using this syntax:
2581 .sp
2582 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2583 (sens|respons)e and \eg'1'ibility
2584 .sp
2585 PCRE supports an extension to Oniguruma: if a number is preceded by a
2586 plus or a minus sign it is taken as a relative reference. For example:
2587 .sp
2588 (abc)(?i:\eg<-1>)
2589 .sp
2590 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2591 synonymous. The former is a back reference; the latter is a subroutine call.
2592 .
2593 .
2595 .rs
2596 .sp
2597 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2598 code to be obeyed in the middle of matching a regular expression. This makes it
2599 possible, amongst other things, to extract different substrings that match the
2600 same pair of parentheses when there is a repetition.
2601 .P
2602 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2603 code. The feature is called "callout". The caller of PCRE provides an external
2604 function by putting its entry point in the global variable \fIpcre_callout\fP
2605 (8-bit library) or \fIpcre16_callout\fP (16-bit library). By default, this
2606 variable contains NULL, which disables all calling out.
2607 .P
2608 Within a regular expression, (?C) indicates the points at which the external
2609 function is to be called. If you want to identify different callout points, you
2610 can put a number less than 256 after the letter C. The default value is zero.
2611 For example, this pattern has two callout points:
2612 .sp
2613 (?C1)abc(?C2)def
2614 .sp
2615 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2616 automatically installed before each item in the pattern. They are all numbered
2617 255.
2618 .P
2619 During matching, when PCRE reaches a callout point, the external function is
2620 called. It is provided with the number of the callout, the position in the
2621 pattern, and, optionally, one item of data originally supplied by the caller of
2622 the matching function. The callout function may cause matching to proceed, to
2623 backtrack, or to fail altogether. A complete description of the interface to
2624 the callout function is given in the
2625 .\" HREF
2626 \fBpcrecallout\fP
2627 .\"
2628 documentation.
2629 .
2630 .
2631 .\" HTML <a name="backtrackcontrol"></a>
2633 .rs
2634 .sp
2635 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2636 are described in the Perl documentation as "experimental and subject to change
2637 or removal in a future version of Perl". It goes on to say: "Their usage in
2638 production code should be noted to avoid problems during upgrades." The same
2639 remarks apply to the PCRE features described in this section.
2640 .P
2641 Since these verbs are specifically related to backtracking, most of them can be
2642 used only when the pattern is to be matched using one of the traditional
2643 matching functions, which use a backtracking algorithm. With the exception of
2644 (*FAIL), which behaves like a failing negative assertion, they cause an error
2645 if encountered by a DFA matching function.
2646 .P
2647 If any of these verbs are used in an assertion or in a subpattern that is
2648 called as a subroutine (whether or not recursively), their effect is confined
2649 to that subpattern; it does not extend to the surrounding pattern, with one
2650 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2651 a successful positive assertion \fIis\fP passed back when a match succeeds
2652 (compare capturing parentheses in assertions). Note that such subpatterns are
2653 processed as anchored at the point where they are tested. Note also that Perl's
2654 treatment of subroutines and assertions is different in some cases.
2655 .P
2656 The new verbs make use of what was previously invalid syntax: an opening
2657 parenthesis followed by an asterisk. They are generally of the form
2658 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2659 depending on whether or not an argument is present. A name is any sequence of
2660 characters that does not include a closing parenthesis. The maximum length of
2661 name is 255 in the 8-bit library and 65535 in the 16-bit library. If the name
2662 is empty, that is, if the closing parenthesis immediately follows the colon,
2663 the effect is as if the colon were not there. Any number of these verbs may
2664 occur in a pattern.
2665 .
2666 .
2667 .\" HTML <a name="nooptimize"></a>
2668 .SS "Optimizations that affect backtracking verbs"
2669 .rs
2670 .sp
2671 PCRE contains some optimizations that are used to speed up matching by running
2672 some checks at the start of each match attempt. For example, it may know the
2673 minimum length of matching subject, or that a particular character must be
2674 present. When one of these optimizations suppresses the running of a match, any
2675 included backtracking verbs will not, of course, be processed. You can suppress
2676 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2677 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2678 pattern with (*NO_START_OPT). There is more discussion of this option in the
2679 section entitled
2680 .\" HTML <a href="pcreapi.html#execoptions">
2681 .\" </a>
2682 "Option bits for \fBpcre_exec()\fP"
2683 .\"
2684 in the
2685 .\" HREF
2686 \fBpcreapi\fP
2687 .\"
2688 documentation.
2689 .P
2690 Experiments with Perl suggest that it too has similar optimizations, sometimes
2691 leading to anomalous results.
2692 .
2693 .
2694 .SS "Verbs that act immediately"
2695 .rs
2696 .sp
2697 The following verbs act as soon as they are encountered. They may not be
2698 followed by a name.
2699 .sp
2700 (*ACCEPT)
2701 .sp
2702 This verb causes the match to end successfully, skipping the remainder of the
2703 pattern. However, when it is inside a subpattern that is called as a
2704 subroutine, only that subpattern is ended successfully. Matching then continues
2705 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2706 far is captured. For example:
2707 .sp
2708 A((?:A|B(*ACCEPT)|C)D)
2709 .sp
2710 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2711 the outer parentheses.
2712 .sp
2713 (*FAIL) or (*F)
2714 .sp
2715 This verb causes a matching failure, forcing backtracking to occur. It is
2716 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2717 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2718 Perl features that are not present in PCRE. The nearest equivalent is the
2719 callout feature, as for example in this pattern:
2720 .sp
2721 a+(?C)(*FAIL)
2722 .sp
2723 A match with the string "aaaa" always fails, but the callout is taken before
2724 each backtrack happens (in this example, 10 times).
2725 .
2726 .
2727 .SS "Recording which path was taken"
2728 .rs
2729 .sp
2730 There is one verb whose main purpose is to track how a match was arrived at,
2731 though it also has a secondary use in conjunction with advancing the match
2732 starting point (see (*SKIP) below).
2733 .sp
2734 (*MARK:NAME) or (*:NAME)
2735 .sp
2736 A name is always required with this verb. There may be as many instances of
2737 (*MARK) as you like in a pattern, and their names do not have to be unique.
2738 .P
2739 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2740 path is passed back to the caller as described in the section entitled
2741 .\" HTML <a href="pcreapi.html#extradata">
2742 .\" </a>
2743 "Extra data for \fBpcre_exec()\fP"
2744 .\"
2745 in the
2746 .\" HREF
2747 \fBpcreapi\fP
2748 .\"
2749 documentation. Here is an example of \fBpcretest\fP output, where the /K
2750 modifier requests the retrieval and outputting of (*MARK) data:
2751 .sp
2752 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2753 data> XY
2754 0: XY
2755 MK: A
2756 XZ
2757 0: XZ
2758 MK: B
2759 .sp
2760 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2761 indicates which of the two alternatives matched. This is a more efficient way
2762 of obtaining this information than putting each alternative in its own
2763 capturing parentheses.
2764 .P
2765 If (*MARK) is encountered in a positive assertion, its name is recorded and
2766 passed back if it is the last-encountered. This does not happen for negative
2767 assertions.
2768 .P
2769 After a partial match or a failed match, the name of the last encountered
2770 (*MARK) in the entire match process is returned. For example:
2771 .sp
2772 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2773 data> XP
2774 No match, mark = B
2775 .sp
2776 Note that in this unanchored example the mark is retained from the match
2777 attempt that started at the letter "X" in the subject. Subsequent match
2778 attempts starting at "P" and then with an empty string do not get as far as the
2779 (*MARK) item, but nevertheless do not reset it.
2780 .P
2781 If you are interested in (*MARK) values after failed matches, you should
2782 probably set the PCRE_NO_START_OPTIMIZE option
2783 .\" HTML <a href="#nooptimize">
2784 .\" </a>
2785 (see above)
2786 .\"
2787 to ensure that the match is always attempted.
2788 .
2789 .
2790 .SS "Verbs that act after backtracking"
2791 .rs
2792 .sp
2793 The following verbs do nothing when they are encountered. Matching continues
2794 with what follows, but if there is no subsequent match, causing a backtrack to
2795 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2796 the verb. However, when one of these verbs appears inside an atomic group, its
2797 effect is confined to that group, because once the group has been matched,
2798 there is never any backtracking into it. In this situation, backtracking can
2799 "jump back" to the left of the entire atomic group. (Remember also, as stated
2800 above, that this localization also applies in subroutine calls and assertions.)
2801 .P
2802 These verbs differ in exactly what kind of failure occurs when backtracking
2803 reaches them.
2804 .sp
2805 (*COMMIT)
2806 .sp
2807 This verb, which may not be followed by a name, causes the whole match to fail
2808 outright if the rest of the pattern does not match. Even if the pattern is
2809 unanchored, no further attempts to find a match by advancing the starting point
2810 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2811 finding a match at the current starting point, or not at all. For example:
2812 .sp
2813 a+(*COMMIT)b
2814 .sp
2815 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2816 dynamic anchor, or "I've started, so I must finish." The name of the most
2817 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2818 match failure.
2819 .P
2820 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2821 unless PCRE's start-of-match optimizations are turned off, as shown in this
2822 \fBpcretest\fP example:
2823 .sp
2824 re> /(*COMMIT)abc/
2825 data> xyzabc
2826 0: abc
2827 xyzabc\eY
2828 No match
2829 .sp
2830 PCRE knows that any match must start with "a", so the optimization skips along
2831 the subject to "a" before running the first match attempt, which succeeds. When
2832 the optimization is disabled by the \eY escape in the second subject, the match
2833 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2834 starting points.
2835 .sp
2836 (*PRUNE) or (*PRUNE:NAME)
2837 .sp
2838 This verb causes the match to fail at the current starting position in the
2839 subject if the rest of the pattern does not match. If the pattern is
2840 unanchored, the normal "bumpalong" advance to the next starting character then
2841 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2842 reached, or when matching to the right of (*PRUNE), but if there is no match to
2843 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2844 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2845 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2846 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2847 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2848 .sp
2849 (*SKIP)
2850 .sp
2851 This verb, when given without a name, is like (*PRUNE), except that if the
2852 pattern is unanchored, the "bumpalong" advance is not to the next character,
2853 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2854 signifies that whatever text was matched leading up to it cannot be part of a
2855 successful match. Consider:
2856 .sp
2857 a+(*SKIP)b
2858 .sp
2859 If the subject is "aaaac...", after the first match attempt fails (starting at
2860 the first character in the string), the starting point skips on to start the
2861 next attempt at "c". Note that a possessive quantifer does not have the same
2862 effect as this example; although it would suppress backtracking during the
2863 first match attempt, the second attempt would start at the second character
2864 instead of skipping on to "c".
2865 .sp
2866 (*SKIP:NAME)
2867 .sp
2868 When (*SKIP) has an associated name, its behaviour is modified. If the
2869 following pattern fails to match, the previous path through the pattern is
2870 searched for the most recent (*MARK) that has the same name. If one is found,
2871 the "bumpalong" advance is to the subject position that corresponds to that
2872 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2873 matching name is found, the (*SKIP) is ignored.
2874 .sp
2875 (*THEN) or (*THEN:NAME)
2876 .sp
2877 This verb causes a skip to the next innermost alternative if the rest of the
2878 pattern does not match. That is, it cancels pending backtracking, but only
2879 within the current alternative. Its name comes from the observation that it can
2880 be used for a pattern-based if-then-else block:
2881 .sp
2882 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2883 .sp
2884 If the COND1 pattern matches, FOO is tried (and possibly further items after
2885 the end of the group if FOO succeeds); on failure, the matcher skips to the
2886 second alternative and tries COND2, without backtracking into COND1. The
2887 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2888 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2889 .P
2890 Note that a subpattern that does not contain a | character is just a part of
2891 the enclosing alternative; it is not a nested alternation with only one
2892 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2893 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2894 pattern fragments that do not contain any | characters at this level:
2895 .sp
2896 A (B(*THEN)C) | D
2897 .sp
2898 If A and B are matched, but there is a failure in C, matching does not
2899 backtrack into A; instead it moves to the next alternative, that is, D.
2900 However, if the subpattern containing (*THEN) is given an alternative, it
2901 behaves differently:
2902 .sp
2903 A (B(*THEN)C | (*FAIL)) | D
2904 .sp
2905 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2906 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2907 because there are no more alternatives to try. In this case, matching does now
2908 backtrack into A.
2909 .P
2910 Note also that a conditional subpattern is not considered as having two
2911 alternatives, because only one is ever used. In other words, the | character in
2912 a conditional subpattern has a different meaning. Ignoring white space,
2913 consider:
2914 .sp
2915 ^.*? (?(?=a) a | b(*THEN)c )
2916 .sp
2917 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2918 it initially matches zero characters. The condition (?=a) then fails, the
2919 character "b" is matched, but "c" is not. At this point, matching does not
2920 backtrack to .*? as might perhaps be expected from the presence of the |
2921 character. The conditional subpattern is part of the single alternative that
2922 comprises the whole pattern, and so the match fails. (If there was a backtrack
2923 into .*?, allowing it to match "b", the match would succeed.)
2924 .P
2925 The verbs just described provide four different "strengths" of control when
2926 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2927 next alternative. (*PRUNE) comes next, failing the match at the current
2928 starting position, but allowing an advance to the next character (for an
2929 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2930 than one character. (*COMMIT) is the strongest, causing the entire match to
2931 fail.
2932 .P
2933 If more than one such verb is present in a pattern, the "strongest" one wins.
2934 For example, consider this pattern, where A, B, etc. are complex pattern
2935 fragments:
2936 .sp
2937 (A(*COMMIT)B(*THEN)C|D)
2938 .sp
2939 Once A has matched, PCRE is committed to this match, at the current starting
2940 position. If subsequently B matches, but C does not, the normal (*THEN) action
2941 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2942 overrides.
2943 .
2944 .
2945 .SH "SEE ALSO"
2946 .rs
2947 .sp
2948 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2949 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP.
2950 .
2951 .
2953 .rs
2954 .sp
2955 .nf
2956 Philip Hazel
2957 University Computing Service
2958 Cambridge CB2 3QH, England.
2959 .fi
2960 .
2961 .
2963 .rs
2964 .sp
2965 .nf
2966 Last updated: 10 September 2012
2967 Copyright (c) 1997-2012 University of Cambridge.
2968 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5