ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1055 - (show annotations)
Tue Oct 16 15:53:30 2012 UTC (7 years, 4 months ago) by chpe
File size: 123749 byte(s)
pcre32: Add 32-bit library

Create libpcre32 that operates on 32-bit characters (UTF-32).

This turned out to be surprisingly simple after the UTF-16 support
was introduced; mostly just extra ifdefs and adjusting and adding
some tests.
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, an
25 extra library that supports 16-bit and UTF-16 character strings, and an
26 extra library that supports 32-bit and UTF-32 character strings. To use these
27 features, PCRE must be built to include appropriate support. When using UTF
28 strings you must either call the compiling function with the PCRE_UTF8,
29 PCRE_UTF16 or PCRE_UTF32 option, or the pattern must start with one of
30 these special sequences:
31 .sp
32 (*UTF8)
33 (*UTF16)
34 (*UTF32)
35 .sp
36 Starting a pattern with such a sequence is equivalent to setting the relevant
37 option. This feature is not Perl-compatible. How setting a UTF mode affects
38 pattern matching is mentioned in several places below. There is also a summary
39 of features in the
40 .\" HREF
41 \fBpcreunicode\fP
42 .\"
43 page.
44 .P
45 Another special sequence that may appear at the start of a pattern or in
46 combination with (*UTF8) or (*UTF16) or (*UTF32) is:
47 .sp
48 (*UCP)
49 .sp
50 This has the same effect as setting the PCRE_UCP option: it causes sequences
51 such as \ed and \ew to use Unicode properties to determine character types,
52 instead of recognizing only characters with codes less than 128 via a lookup
53 table.
54 .P
55 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
56 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
57 also some more of these special sequences that are concerned with the handling
58 of newlines; they are described below.
59 .P
60 The remainder of this document discusses the patterns that are supported by
61 PCRE when one its main matching functions, \fBpcre_exec()\fP (8-bit) or
62 \fBpcre[16|32]_exec()\fP (16- or 32-bit), is used. PCRE also has alternative
63 matching functions, \fBpcre_dfa_exec()\fP and \fBpcre[16|32_dfa_exec()\fP,
64 which match using a different algorithm that is not Perl-compatible. Some of
65 the features discussed below are not available when DFA matching is used. The
66 advantages and disadvantages of the alternative functions, and how they differ
67 from the normal functions, are discussed in the
68 .\" HREF
69 \fBpcrematching\fP
70 .\"
71 page.
72 .
73 .
75 .rs
76 .sp
77 PCRE can be compiled to run in an environment that uses EBCDIC as its character
78 code rather than ASCII or Unicode (typically a mainframe system). In the
79 sections below, character code values are ASCII or Unicode; in an EBCDIC
80 environment these characters may have different code values, and there are no
81 code points greater than 255.
82 .
83 .
84 .\" HTML <a name="newlines"></a>
86 .rs
87 .sp
88 PCRE supports five different conventions for indicating line breaks in
89 strings: a single CR (carriage return) character, a single LF (linefeed)
90 character, the two-character sequence CRLF, any of the three preceding, or any
91 Unicode newline sequence. The
92 .\" HREF
93 \fBpcreapi\fP
94 .\"
95 page has
96 .\" HTML <a href="pcreapi.html#newlines">
97 .\" </a>
98 further discussion
99 .\"
100 about newlines, and shows how to set the newline convention in the
101 \fIoptions\fP arguments for the compiling and matching functions.
102 .P
103 It is also possible to specify a newline convention by starting a pattern
104 string with one of the following five sequences:
105 .sp
106 (*CR) carriage return
107 (*LF) linefeed
108 (*CRLF) carriage return, followed by linefeed
109 (*ANYCRLF) any of the three above
110 (*ANY) all Unicode newline sequences
111 .sp
112 These override the default and the options given to the compiling function. For
113 example, on a Unix system where LF is the default newline sequence, the pattern
114 .sp
115 (*CR)a.b
116 .sp
117 changes the convention to CR. That pattern matches "a\enb" because LF is no
118 longer a newline. Note that these special settings, which are not
119 Perl-compatible, are recognized only at the very start of a pattern, and that
120 they must be in upper case. If more than one of them is present, the last one
121 is used.
122 .P
123 The newline convention affects the interpretation of the dot metacharacter when
124 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
125 affect what the \eR escape sequence matches. By default, this is any Unicode
126 newline sequence, for Perl compatibility. However, this can be changed; see the
127 description of \eR in the section entitled
128 .\" HTML <a href="#newlineseq">
129 .\" </a>
130 "Newline sequences"
131 .\"
132 below. A change of \eR setting can be combined with a change of newline
133 convention.
134 .
135 .
137 .rs
138 .sp
139 A regular expression is a pattern that is matched against a subject string from
140 left to right. Most characters stand for themselves in a pattern, and match the
141 corresponding characters in the subject. As a trivial example, the pattern
142 .sp
143 The quick brown fox
144 .sp
145 matches a portion of a subject string that is identical to itself. When
146 caseless matching is specified (the PCRE_CASELESS option), letters are matched
147 independently of case. In a UTF mode, PCRE always understands the concept of
148 case for characters whose values are less than 128, so caseless matching is
149 always possible. For characters with higher values, the concept of case is
150 supported if PCRE is compiled with Unicode property support, but not otherwise.
151 If you want to use caseless matching for characters 128 and above, you must
152 ensure that PCRE is compiled with Unicode property support as well as with
153 UTF support.
154 .P
155 The power of regular expressions comes from the ability to include alternatives
156 and repetitions in the pattern. These are encoded in the pattern by the use of
157 \fImetacharacters\fP, which do not stand for themselves but instead are
158 interpreted in some special way.
159 .P
160 There are two different sets of metacharacters: those that are recognized
161 anywhere in the pattern except within square brackets, and those that are
162 recognized within square brackets. Outside square brackets, the metacharacters
163 are as follows:
164 .sp
165 \e general escape character with several uses
166 ^ assert start of string (or line, in multiline mode)
167 $ assert end of string (or line, in multiline mode)
168 . match any character except newline (by default)
169 [ start character class definition
170 | start of alternative branch
171 ( start subpattern
172 ) end subpattern
173 ? extends the meaning of (
174 also 0 or 1 quantifier
175 also quantifier minimizer
176 * 0 or more quantifier
177 + 1 or more quantifier
178 also "possessive quantifier"
179 { start min/max quantifier
180 .sp
181 Part of a pattern that is in square brackets is called a "character class". In
182 a character class the only metacharacters are:
183 .sp
184 \e general escape character
185 ^ negate the class, but only if the first character
186 - indicates character range
187 .\" JOIN
188 [ POSIX character class (only if followed by POSIX
189 syntax)
190 ] terminates the character class
191 .sp
192 The following sections describe the use of each of the metacharacters.
193 .
194 .
196 .rs
197 .sp
198 The backslash character has several uses. Firstly, if it is followed by a
199 character that is not a number or a letter, it takes away any special meaning
200 that character may have. This use of backslash as an escape character applies
201 both inside and outside character classes.
202 .P
203 For example, if you want to match a * character, you write \e* in the pattern.
204 This escaping action applies whether or not the following character would
205 otherwise be interpreted as a metacharacter, so it is always safe to precede a
206 non-alphanumeric with backslash to specify that it stands for itself. In
207 particular, if you want to match a backslash, you write \e\e.
208 .P
209 In a UTF mode, only ASCII numbers and letters have any special meaning after a
210 backslash. All other characters (in particular, those whose codepoints are
211 greater than 127) are treated as literals.
212 .P
213 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
214 pattern (other than in a character class) and characters between a # outside
215 a character class and the next newline are ignored. An escaping backslash can
216 be used to include a white space or # character as part of the pattern.
217 .P
218 If you want to remove the special meaning from a sequence of characters, you
219 can do so by putting them between \eQ and \eE. This is different from Perl in
220 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
221 Perl, $ and @ cause variable interpolation. Note the following examples:
222 .sp
223 Pattern PCRE matches Perl matches
224 .sp
225 .\" JOIN
226 \eQabc$xyz\eE abc$xyz abc followed by the
227 contents of $xyz
228 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
229 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
230 .sp
231 The \eQ...\eE sequence is recognized both inside and outside character classes.
232 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
233 by \eE later in the pattern, the literal interpretation continues to the end of
234 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
235 a character class, this causes an error, because the character class is not
236 terminated.
237 .
238 .
239 .\" HTML <a name="digitsafterbackslash"></a>
240 .SS "Non-printing characters"
241 .rs
242 .sp
243 A second use of backslash provides a way of encoding non-printing characters
244 in patterns in a visible manner. There is no restriction on the appearance of
245 non-printing characters, apart from the binary zero that terminates a pattern,
246 but when a pattern is being prepared by text editing, it is often easier to use
247 one of the following escape sequences than the binary character it represents:
248 .sp
249 \ea alarm, that is, the BEL character (hex 07)
250 \ecx "control-x", where x is any ASCII character
251 \ee escape (hex 1B)
252 \ef form feed (hex 0C)
253 \en linefeed (hex 0A)
254 \er carriage return (hex 0D)
255 \et tab (hex 09)
256 \eddd character with octal code ddd, or back reference
257 \exhh character with hex code hh
258 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
259 \euhhhh character with hex code hhhh (JavaScript mode only)
260 .sp
261 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
262 case letter, it is converted to upper case. Then bit 6 of the character (hex
263 40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
264 but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
265 data item (byte or 16-bit value) following \ec has a value greater than 127, a
266 compile-time error occurs. This locks out non-ASCII characters in all modes.
267 .P
268 The \ec facility was designed for use with ASCII characters, but with the
269 extension to Unicode it is even less useful than it once was. It is, however,
270 recognized when PCRE is compiled in EBCDIC mode, where data items are always
271 bytes. In this mode, all values are valid after \ec. If the next character is a
272 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
273 byte are inverted. Thus \ecA becomes hex 01, as in ASCII (A is C1), but because
274 the EBCDIC letters are disjoint, \ecZ becomes hex 29 (Z is E9), and other
275 characters also generate different values.
276 .P
277 By default, after \ex, from zero to two hexadecimal digits are read (letters
278 can be in upper or lower case). Any number of hexadecimal digits may appear
279 between \ex{ and }, but the character code is constrained as follows:
280 .sp
281 8-bit non-UTF mode less than 0x100
282 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
283 16-bit non-UTF mode less than 0x10000
284 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
285 32-bit non-UTF mode less than 0x80000000
286 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
287 .sp
288 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
289 "surrogate" codepoints), and 0xffef.
290 .P
291 If characters other than hexadecimal digits appear between \ex{ and }, or if
292 there is no terminating }, this form of escape is not recognized. Instead, the
293 initial \ex will be interpreted as a basic hexadecimal escape, with no
294 following digits, giving a character whose value is zero.
295 .P
296 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
297 as just described only when it is followed by two hexadecimal digits.
298 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
299 code points greater than 256 is provided by \eu, which must be followed by
300 four hexadecimal digits; otherwise it matches a literal "u" character.
301 Character codes specified by \eu in JavaScript mode are constrained in the same
302 was as those specified by \ex in non-JavaScript mode.
303 .P
304 Characters whose value is less than 256 can be defined by either of the two
305 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
306 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
307 \eu00dc in JavaScript mode).
308 .P
309 After \e0 up to two further octal digits are read. If there are fewer than two
310 digits, just those that are present are used. Thus the sequence \e0\ex\e07
311 specifies two binary zeros followed by a BEL character (code value 7). Make
312 sure you supply two digits after the initial zero if the pattern character that
313 follows is itself an octal digit.
314 .P
315 The handling of a backslash followed by a digit other than 0 is complicated.
316 Outside a character class, PCRE reads it and any following digits as a decimal
317 number. If the number is less than 10, or if there have been at least that many
318 previous capturing left parentheses in the expression, the entire sequence is
319 taken as a \fIback reference\fP. A description of how this works is given
320 .\" HTML <a href="#backreferences">
321 .\" </a>
322 later,
323 .\"
324 following the discussion of
325 .\" HTML <a href="#subpattern">
326 .\" </a>
327 parenthesized subpatterns.
328 .\"
329 .P
330 Inside a character class, or if the decimal number is greater than 9 and there
331 have not been that many capturing subpatterns, PCRE re-reads up to three octal
332 digits following the backslash, and uses them to generate a data character. Any
333 subsequent digits stand for themselves. The value of the character is
334 constrained in the same way as characters specified in hexadecimal.
335 For example:
336 .sp
337 \e040 is another way of writing an ASCII space
338 .\" JOIN
339 \e40 is the same, provided there are fewer than 40
340 previous capturing subpatterns
341 \e7 is always a back reference
342 .\" JOIN
343 \e11 might be a back reference, or another way of
344 writing a tab
345 \e011 is always a tab
346 \e0113 is a tab followed by the character "3"
347 .\" JOIN
348 \e113 might be a back reference, otherwise the
349 character with octal code 113
350 .\" JOIN
351 \e377 might be a back reference, otherwise
352 the value 255 (decimal)
353 .\" JOIN
354 \e81 is either a back reference, or a binary zero
355 followed by the two characters "8" and "1"
356 .sp
357 Note that octal values of 100 or greater must not be introduced by a leading
358 zero, because no more than three octal digits are ever read.
359 .P
360 All the sequences that define a single character value can be used both inside
361 and outside character classes. In addition, inside a character class, \eb is
362 interpreted as the backspace character (hex 08).
363 .P
364 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
365 inside a character class. Like other unrecognized escape sequences, they are
366 treated as the literal characters "B", "R", and "X" by default, but cause an
367 error if the PCRE_EXTRA option is set. Outside a character class, these
368 sequences have different meanings.
369 .
370 .
371 .SS "Unsupported escape sequences"
372 .rs
373 .sp
374 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
375 handler and used to modify the case of following characters. By default, PCRE
376 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
377 option is set, \eU matches a "U" character, and \eu can be used to define a
378 character by code point, as described in the previous section.
379 .
380 .
381 .SS "Absolute and relative back references"
382 .rs
383 .sp
384 The sequence \eg followed by an unsigned or a negative number, optionally
385 enclosed in braces, is an absolute or relative back reference. A named back
386 reference can be coded as \eg{name}. Back references are discussed
387 .\" HTML <a href="#backreferences">
388 .\" </a>
389 later,
390 .\"
391 following the discussion of
392 .\" HTML <a href="#subpattern">
393 .\" </a>
394 parenthesized subpatterns.
395 .\"
396 .
397 .
398 .SS "Absolute and relative subroutine calls"
399 .rs
400 .sp
401 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
402 a number enclosed either in angle brackets or single quotes, is an alternative
403 syntax for referencing a subpattern as a "subroutine". Details are discussed
404 .\" HTML <a href="#onigurumasubroutines">
405 .\" </a>
406 later.
407 .\"
408 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
409 synonymous. The former is a back reference; the latter is a
410 .\" HTML <a href="#subpatternsassubroutines">
411 .\" </a>
412 subroutine
413 .\"
414 call.
415 .
416 .
417 .\" HTML <a name="genericchartypes"></a>
418 .SS "Generic character types"
419 .rs
420 .sp
421 Another use of backslash is for specifying generic character types:
422 .sp
423 \ed any decimal digit
424 \eD any character that is not a decimal digit
425 \eh any horizontal white space character
426 \eH any character that is not a horizontal white space character
427 \es any white space character
428 \eS any character that is not a white space character
429 \ev any vertical white space character
430 \eV any character that is not a vertical white space character
431 \ew any "word" character
432 \eW any "non-word" character
433 .sp
434 There is also the single sequence \eN, which matches a non-newline character.
435 This is the same as
436 .\" HTML <a href="#fullstopdot">
437 .\" </a>
438 the "." metacharacter
439 .\"
440 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
441 PCRE does not support this.
442 .P
443 Each pair of lower and upper case escape sequences partitions the complete set
444 of characters into two disjoint sets. Any given character matches one, and only
445 one, of each pair. The sequences can appear both inside and outside character
446 classes. They each match one character of the appropriate type. If the current
447 matching point is at the end of the subject string, all of them fail, because
448 there is no character to match.
449 .P
450 For compatibility with Perl, \es does not match the VT character (code 11).
451 This makes it different from the the POSIX "space" class. The \es characters
452 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
453 included in a Perl script, \es may match the VT character. In PCRE, it never
454 does.
455 .P
456 A "word" character is an underscore or any character that is a letter or digit.
457 By default, the definition of letters and digits is controlled by PCRE's
458 low-valued character tables, and may vary if locale-specific matching is taking
459 place (see
460 .\" HTML <a href="pcreapi.html#localesupport">
461 .\" </a>
462 "Locale support"
463 .\"
464 in the
465 .\" HREF
466 \fBpcreapi\fP
467 .\"
468 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
469 or "french" in Windows, some character codes greater than 128 are used for
470 accented letters, and these are then matched by \ew. The use of locales with
471 Unicode is discouraged.
472 .P
473 By default, in a UTF mode, characters with values greater than 128 never match
474 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
475 their original meanings from before UTF support was available, mainly for
476 efficiency reasons. However, if PCRE is compiled with Unicode property support,
477 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
478 properties are used to determine character types, as follows:
479 .sp
480 \ed any character that \ep{Nd} matches (decimal digit)
481 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
482 \ew any character that \ep{L} or \ep{N} matches, plus underscore
483 .sp
484 The upper case escapes match the inverse sets of characters. Note that \ed
485 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
486 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
487 \eB because they are defined in terms of \ew and \eW. Matching these sequences
488 is noticeably slower when PCRE_UCP is set.
489 .P
490 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
491 release 5.10. In contrast to the other sequences, which match only ASCII
492 characters by default, these always match certain high-valued codepoints,
493 whether or not PCRE_UCP is set. The horizontal space characters are:
494 .sp
495 U+0009 Horizontal tab (HT)
496 U+0020 Space
497 U+00A0 Non-break space
498 U+1680 Ogham space mark
499 U+180E Mongolian vowel separator
500 U+2000 En quad
501 U+2001 Em quad
502 U+2002 En space
503 U+2003 Em space
504 U+2004 Three-per-em space
505 U+2005 Four-per-em space
506 U+2006 Six-per-em space
507 U+2007 Figure space
508 U+2008 Punctuation space
509 U+2009 Thin space
510 U+200A Hair space
511 U+202F Narrow no-break space
512 U+205F Medium mathematical space
513 U+3000 Ideographic space
514 .sp
515 The vertical space characters are:
516 .sp
517 U+000A Linefeed (LF)
518 U+000B Vertical tab (VT)
519 U+000C Form feed (FF)
520 U+000D Carriage return (CR)
521 U+0085 Next line (NEL)
522 U+2028 Line separator
523 U+2029 Paragraph separator
524 .sp
525 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
526 relevant.
527 .
528 .
529 .\" HTML <a name="newlineseq"></a>
530 .SS "Newline sequences"
531 .rs
532 .sp
533 Outside a character class, by default, the escape sequence \eR matches any
534 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
535 following:
536 .sp
537 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
538 .sp
539 This is an example of an "atomic group", details of which are given
540 .\" HTML <a href="#atomicgroup">
541 .\" </a>
542 below.
543 .\"
544 This particular group matches either the two-character sequence CR followed by
545 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
546 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
547 line, U+0085). The two-character sequence is treated as a single unit that
548 cannot be split.
549 .P
550 In other modes, two additional characters whose codepoints are greater than 255
551 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
552 Unicode character property support is not needed for these characters to be
553 recognized.
554 .P
555 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
556 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
557 either at compile time or when the pattern is matched. (BSR is an abbrevation
558 for "backslash R".) This can be made the default when PCRE is built; if this is
559 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
560 It is also possible to specify these settings by starting a pattern string with
561 one of the following sequences:
562 .sp
563 (*BSR_ANYCRLF) CR, LF, or CRLF only
564 (*BSR_UNICODE) any Unicode newline sequence
565 .sp
566 These override the default and the options given to the compiling function, but
567 they can themselves be overridden by options given to a matching function. Note
568 that these special settings, which are not Perl-compatible, are recognized only
569 at the very start of a pattern, and that they must be in upper case. If more
570 than one of them is present, the last one is used. They can be combined with a
571 change of newline convention; for example, a pattern can start with:
572 .sp
574 .sp
575 They can also be combined with the (*UTF8), (*UTF16), (*UTF32) or (*UCP) special
576 sequences. Inside a character class, \eR is treated as an unrecognized escape
577 sequence, and so matches the letter "R" by default, but causes an error if
578 PCRE_EXTRA is set.
579 .
580 .
581 .\" HTML <a name="uniextseq"></a>
582 .SS Unicode character properties
583 .rs
584 .sp
585 When PCRE is built with Unicode character property support, three additional
586 escape sequences that match characters with specific properties are available.
587 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
588 characters whose codepoints are less than 256, but they do work in this mode.
589 The extra escape sequences are:
590 .sp
591 \ep{\fIxx\fP} a character with the \fIxx\fP property
592 \eP{\fIxx\fP} a character without the \fIxx\fP property
593 \eX a Unicode extended grapheme cluster
594 .sp
595 The property names represented by \fIxx\fP above are limited to the Unicode
596 script names, the general category properties, "Any", which matches any
597 character (including newline), and some special PCRE properties (described
598 in the
599 .\" HTML <a href="#extraprops">
600 .\" </a>
601 next section).
602 .\"
603 Other Perl properties such as "InMusicalSymbols" are not currently supported by
604 PCRE. Note that \eP{Any} does not match any characters, so always causes a
605 match failure.
606 .P
607 Sets of Unicode characters are defined as belonging to certain scripts. A
608 character from one of these sets can be matched using a script name. For
609 example:
610 .sp
611 \ep{Greek}
612 \eP{Han}
613 .sp
614 Those that are not part of an identified script are lumped together as
615 "Common". The current list of scripts is:
616 .P
617 Arabic,
618 Armenian,
619 Avestan,
620 Balinese,
621 Bamum,
622 Batak,
623 Bengali,
624 Bopomofo,
625 Brahmi,
626 Braille,
627 Buginese,
628 Buhid,
629 Canadian_Aboriginal,
630 Carian,
631 Chakma,
632 Cham,
633 Cherokee,
634 Common,
635 Coptic,
636 Cuneiform,
637 Cypriot,
638 Cyrillic,
639 Deseret,
640 Devanagari,
641 Egyptian_Hieroglyphs,
642 Ethiopic,
643 Georgian,
644 Glagolitic,
645 Gothic,
646 Greek,
647 Gujarati,
648 Gurmukhi,
649 Han,
650 Hangul,
651 Hanunoo,
652 Hebrew,
653 Hiragana,
654 Imperial_Aramaic,
655 Inherited,
656 Inscriptional_Pahlavi,
657 Inscriptional_Parthian,
658 Javanese,
659 Kaithi,
660 Kannada,
661 Katakana,
662 Kayah_Li,
663 Kharoshthi,
664 Khmer,
665 Lao,
666 Latin,
667 Lepcha,
668 Limbu,
669 Linear_B,
670 Lisu,
671 Lycian,
672 Lydian,
673 Malayalam,
674 Mandaic,
675 Meetei_Mayek,
676 Meroitic_Cursive,
677 Meroitic_Hieroglyphs,
678 Miao,
679 Mongolian,
680 Myanmar,
681 New_Tai_Lue,
682 Nko,
683 Ogham,
684 Old_Italic,
685 Old_Persian,
686 Old_South_Arabian,
687 Old_Turkic,
688 Ol_Chiki,
689 Oriya,
690 Osmanya,
691 Phags_Pa,
692 Phoenician,
693 Rejang,
694 Runic,
695 Samaritan,
696 Saurashtra,
697 Sharada,
698 Shavian,
699 Sinhala,
700 Sora_Sompeng,
701 Sundanese,
702 Syloti_Nagri,
703 Syriac,
704 Tagalog,
705 Tagbanwa,
706 Tai_Le,
707 Tai_Tham,
708 Tai_Viet,
709 Takri,
710 Tamil,
711 Telugu,
712 Thaana,
713 Thai,
714 Tibetan,
715 Tifinagh,
716 Ugaritic,
717 Vai,
718 Yi.
719 .P
720 Each character has exactly one Unicode general category property, specified by
721 a two-letter abbreviation. For compatibility with Perl, negation can be
722 specified by including a circumflex between the opening brace and the property
723 name. For example, \ep{^Lu} is the same as \eP{Lu}.
724 .P
725 If only one letter is specified with \ep or \eP, it includes all the general
726 category properties that start with that letter. In this case, in the absence
727 of negation, the curly brackets in the escape sequence are optional; these two
728 examples have the same effect:
729 .sp
730 \ep{L}
731 \epL
732 .sp
733 The following general category property codes are supported:
734 .sp
735 C Other
736 Cc Control
737 Cf Format
738 Cn Unassigned
739 Co Private use
740 Cs Surrogate
741 .sp
742 L Letter
743 Ll Lower case letter
744 Lm Modifier letter
745 Lo Other letter
746 Lt Title case letter
747 Lu Upper case letter
748 .sp
749 M Mark
750 Mc Spacing mark
751 Me Enclosing mark
752 Mn Non-spacing mark
753 .sp
754 N Number
755 Nd Decimal number
756 Nl Letter number
757 No Other number
758 .sp
759 P Punctuation
760 Pc Connector punctuation
761 Pd Dash punctuation
762 Pe Close punctuation
763 Pf Final punctuation
764 Pi Initial punctuation
765 Po Other punctuation
766 Ps Open punctuation
767 .sp
768 S Symbol
769 Sc Currency symbol
770 Sk Modifier symbol
771 Sm Mathematical symbol
772 So Other symbol
773 .sp
774 Z Separator
775 Zl Line separator
776 Zp Paragraph separator
777 Zs Space separator
778 .sp
779 The special property L& is also supported: it matches a character that has
780 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
781 a modifier or "other".
782 .P
783 The Cs (Surrogate) property applies only to characters in the range U+D800 to
784 U+DFFF. Such characters are not valid in Unicode strings and so
785 cannot be tested by PCRE, unless UTF validity checking has been turned off
786 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
787 PCRE_NO_UTF32_CHECK in the
788 .\" HREF
789 \fBpcreapi\fP
790 .\"
791 page). Perl does not support the Cs property.
792 .P
793 The long synonyms for property names that Perl supports (such as \ep{Letter})
794 are not supported by PCRE, nor is it permitted to prefix any of these
795 properties with "Is".
796 .P
797 No character that is in the Unicode table has the Cn (unassigned) property.
798 Instead, this property is assumed for any code point that is not in the
799 Unicode table.
800 .P
801 Specifying caseless matching does not affect these escape sequences. For
802 example, \ep{Lu} always matches only upper case letters.
803 .P
804 Matching characters by Unicode property is not fast, because PCRE has to do a
805 multistage table lookup in order to find a character's property. That is why
806 the traditional escape sequences such as \ed and \ew do not use Unicode
807 properties in PCRE by default, though you can make them do so by setting the
808 PCRE_UCP option or by starting the pattern with (*UCP).
809 .
810 .
811 .SS Extended grapheme clusters
812 .rs
813 .sp
814 The \eX escape matches any number of Unicode characters that form an "extended
815 grapheme cluster", and treats the sequence as an atomic group
816 .\" HTML <a href="#atomicgroup">
817 .\" </a>
818 (see below).
819 .\"
820 Up to and including release 8.31, PCRE matched an earlier, simpler definition
821 that was equivalent to
822 .sp
823 (?>\ePM\epM*)
824 .sp
825 That is, it matched a character without the "mark" property, followed by zero
826 or more characters with the "mark" property. Characters with the "mark"
827 property are typically non-spacing accents that affect the preceding character.
828 .P
829 This simple definition was extended in Unicode to include more complicated
830 kinds of composite character by giving each character a grapheme breaking
831 property, and creating rules that use these properties to define the boundaries
832 of extended grapheme clusters. In releases of PCRE later than 8.31, \eX matches
833 one of these clusters.
834 .P
835 \eX always matches at least one character. Then it decides whether to add
836 additional characters according to the following rules for ending a cluster:
837 .P
838 1. End at the end of the subject string.
839 .P
840 2. Do not end between CR and LF; otherwise end after any control character.
841 .P
842 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
843 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
844 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
845 character; an LVT or T character may be follwed only by a T character.
846 .P
847 4. Do not end before extending characters or spacing marks. Characters with
848 the "mark" property always have the "extend" grapheme breaking property.
849 .P
850 5. Do not end after prepend characters.
851 .P
852 6. Otherwise, end the cluster.
853 .
854 .
855 .\" HTML <a name="extraprops"></a>
856 .SS PCRE's additional properties
857 .rs
858 .sp
859 As well as the standard Unicode properties described above, PCRE supports four
860 more that make it possible to convert traditional escape sequences such as \ew
861 and \es and POSIX character classes to use Unicode properties. PCRE uses these
862 non-standard, non-Perl properties internally when PCRE_UCP is set. They are:
863 .sp
864 Xan Any alphanumeric character
865 Xps Any POSIX space character
866 Xsp Any Perl space character
867 Xwd Any Perl "word" character
868 .sp
869 Xan matches characters that have either the L (letter) or the N (number)
870 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
871 carriage return, and any other character that has the Z (separator) property.
872 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
873 same characters as Xan, plus underscore.
874 .
875 .
876 .\" HTML <a name="resetmatchstart"></a>
877 .SS "Resetting the match start"
878 .rs
879 .sp
880 The escape sequence \eK causes any previously matched characters not to be
881 included in the final matched sequence. For example, the pattern:
882 .sp
883 foo\eKbar
884 .sp
885 matches "foobar", but reports that it has matched "bar". This feature is
886 similar to a lookbehind assertion
887 .\" HTML <a href="#lookbehind">
888 .\" </a>
889 (described below).
890 .\"
891 However, in this case, the part of the subject before the real match does not
892 have to be of fixed length, as lookbehind assertions do. The use of \eK does
893 not interfere with the setting of
894 .\" HTML <a href="#subpattern">
895 .\" </a>
896 captured substrings.
897 .\"
898 For example, when the pattern
899 .sp
900 (foo)\eKbar
901 .sp
902 matches "foobar", the first substring is still set to "foo".
903 .P
904 Perl documents that the use of \eK within assertions is "not well defined". In
905 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
906 ignored in negative assertions.
907 .
908 .
909 .\" HTML <a name="smallassertions"></a>
910 .SS "Simple assertions"
911 .rs
912 .sp
913 The final use of backslash is for certain simple assertions. An assertion
914 specifies a condition that has to be met at a particular point in a match,
915 without consuming any characters from the subject string. The use of
916 subpatterns for more complicated assertions is described
917 .\" HTML <a href="#bigassertions">
918 .\" </a>
919 below.
920 .\"
921 The backslashed assertions are:
922 .sp
923 \eb matches at a word boundary
924 \eB matches when not at a word boundary
925 \eA matches at the start of the subject
926 \eZ matches at the end of the subject
927 also matches before a newline at the end of the subject
928 \ez matches only at the end of the subject
929 \eG matches at the first matching position in the subject
930 .sp
931 Inside a character class, \eb has a different meaning; it matches the backspace
932 character. If any other of these assertions appears in a character class, by
933 default it matches the corresponding literal character (for example, \eB
934 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
935 escape sequence" error is generated instead.
936 .P
937 A word boundary is a position in the subject string where the current character
938 and the previous character do not both match \ew or \eW (i.e. one matches
939 \ew and the other matches \eW), or the start or end of the string if the
940 first or last character matches \ew, respectively. In a UTF mode, the meanings
941 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
942 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
943 of word" or "end of word" metasequence. However, whatever follows \eb normally
944 determines which it is. For example, the fragment \eba matches "a" at the start
945 of a word.
946 .P
947 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
948 dollar (described in the next section) in that they only ever match at the very
949 start and end of the subject string, whatever options are set. Thus, they are
950 independent of multiline mode. These three assertions are not affected by the
951 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
952 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
953 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
954 at a point other than the beginning of the subject, \eA can never match. The
955 difference between \eZ and \ez is that \eZ matches before a newline at the end
956 of the string as well as at the very end, whereas \ez matches only at the end.
957 .P
958 The \eG assertion is true only when the current matching position is at the
959 start point of the match, as specified by the \fIstartoffset\fP argument of
960 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
961 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
962 arguments, you can mimic Perl's /g option, and it is in this kind of
963 implementation where \eG can be useful.
964 .P
965 Note, however, that PCRE's interpretation of \eG, as the start of the current
966 match, is subtly different from Perl's, which defines it as the end of the
967 previous match. In Perl, these can be different when the previously matched
968 string was empty. Because PCRE does just one match at a time, it cannot
969 reproduce this behaviour.
970 .P
971 If all the alternatives of a pattern begin with \eG, the expression is anchored
972 to the starting match position, and the "anchored" flag is set in the compiled
973 regular expression.
974 .
975 .
977 .rs
978 .sp
979 Outside a character class, in the default matching mode, the circumflex
980 character is an assertion that is true only if the current matching point is
981 at the start of the subject string. If the \fIstartoffset\fP argument of
982 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
983 option is unset. Inside a character class, circumflex has an entirely different
984 meaning
985 .\" HTML <a href="#characterclass">
986 .\" </a>
987 (see below).
988 .\"
989 .P
990 Circumflex need not be the first character of the pattern if a number of
991 alternatives are involved, but it should be the first thing in each alternative
992 in which it appears if the pattern is ever to match that branch. If all
993 possible alternatives start with a circumflex, that is, if the pattern is
994 constrained to match only at the start of the subject, it is said to be an
995 "anchored" pattern. (There are also other constructs that can cause a pattern
996 to be anchored.)
997 .P
998 A dollar character is an assertion that is true only if the current matching
999 point is at the end of the subject string, or immediately before a newline
1000 at the end of the string (by default). Dollar need not be the last character of
1001 the pattern if a number of alternatives are involved, but it should be the last
1002 item in any branch in which it appears. Dollar has no special meaning in a
1003 character class.
1004 .P
1005 The meaning of dollar can be changed so that it matches only at the very end of
1006 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1007 does not affect the \eZ assertion.
1008 .P
1009 The meanings of the circumflex and dollar characters are changed if the
1010 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1011 immediately after internal newlines as well as at the start of the subject
1012 string. It does not match after a newline that ends the string. A dollar
1013 matches before any newlines in the string, as well as at the very end, when
1014 PCRE_MULTILINE is set. When newline is specified as the two-character
1015 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1016 .P
1017 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
1018 \en represents a newline) in multiline mode, but not otherwise. Consequently,
1019 patterns that are anchored in single line mode because all branches start with
1020 ^ are not anchored in multiline mode, and a match for circumflex is possible
1021 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
1022 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1023 .P
1024 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
1025 end of the subject in both modes, and if all branches of a pattern start with
1026 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
1027 .
1028 .
1029 .\" HTML <a name="fullstopdot"></a>
1031 .rs
1032 .sp
1033 Outside a character class, a dot in the pattern matches any one character in
1034 the subject string except (by default) a character that signifies the end of a
1035 line.
1036 .P
1037 When a line ending is defined as a single character, dot never matches that
1038 character; when the two-character sequence CRLF is used, dot does not match CR
1039 if it is immediately followed by LF, but otherwise it matches all characters
1040 (including isolated CRs and LFs). When any Unicode line endings are being
1041 recognized, dot does not match CR or LF or any of the other line ending
1042 characters.
1043 .P
1044 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1045 option is set, a dot matches any one character, without exception. If the
1046 two-character sequence CRLF is present in the subject string, it takes two dots
1047 to match it.
1048 .P
1049 The handling of dot is entirely independent of the handling of circumflex and
1050 dollar, the only relationship being that they both involve newlines. Dot has no
1051 special meaning in a character class.
1052 .P
1053 The escape sequence \eN behaves like a dot, except that it is not affected by
1054 the PCRE_DOTALL option. In other words, it matches any character except one
1055 that signifies the end of a line. Perl also uses \eN to match characters by
1056 name; PCRE does not support this.
1057 .
1058 .
1060 .rs
1061 .sp
1062 Outside a character class, the escape sequence \eC matches any one data unit,
1063 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1064 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1065 a 32-bit unit. Unlike a dot, \eC always
1066 matches line-ending characters. The feature is provided in Perl in order to
1067 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1068 used. Because \eC breaks up characters into individual data units, matching one
1069 unit with \eC in a UTF mode means that the rest of the string may start with a
1070 malformed UTF character. This has undefined results, because PCRE assumes that
1071 it is dealing with valid UTF strings (and by default it checks this at the
1072 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1073 PCRE_NO_UTF32_CHECK option is used).
1074 .P
1075 PCRE does not allow \eC to appear in lookbehind assertions
1076 .\" HTML <a href="#lookbehind">
1077 .\" </a>
1078 (described below)
1079 .\"
1080 in a UTF mode, because this would make it impossible to calculate the length of
1081 the lookbehind.
1082 .P
1083 In general, the \eC escape sequence is best avoided. However, one
1084 way of using it that avoids the problem of malformed UTF characters is to use a
1085 lookahead to check the length of the next character, as in this pattern, which
1086 could be used with a UTF-8 string (ignore white space and line breaks):
1087 .sp
1088 (?| (?=[\ex00-\ex7f])(\eC) |
1089 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1090 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1091 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1092 .sp
1093 A group that starts with (?| resets the capturing parentheses numbers in each
1094 alternative (see
1095 .\" HTML <a href="#dupsubpatternnumber">
1096 .\" </a>
1097 "Duplicate Subpattern Numbers"
1098 .\"
1099 below). The assertions at the start of each branch check the next UTF-8
1100 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1101 character's individual bytes are then captured by the appropriate number of
1102 groups.
1103 .
1104 .
1105 .\" HTML <a name="characterclass"></a>
1107 .rs
1108 .sp
1109 An opening square bracket introduces a character class, terminated by a closing
1110 square bracket. A closing square bracket on its own is not special by default.
1111 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1112 bracket causes a compile-time error. If a closing square bracket is required as
1113 a member of the class, it should be the first data character in the class
1114 (after an initial circumflex, if present) or escaped with a backslash.
1115 .P
1116 A character class matches a single character in the subject. In a UTF mode, the
1117 character may be more than one data unit long. A matched character must be in
1118 the set of characters defined by the class, unless the first character in the
1119 class definition is a circumflex, in which case the subject character must not
1120 be in the set defined by the class. If a circumflex is actually required as a
1121 member of the class, ensure it is not the first character, or escape it with a
1122 backslash.
1123 .P
1124 For example, the character class [aeiou] matches any lower case vowel, while
1125 [^aeiou] matches any character that is not a lower case vowel. Note that a
1126 circumflex is just a convenient notation for specifying the characters that
1127 are in the class by enumerating those that are not. A class that starts with a
1128 circumflex is not an assertion; it still consumes a character from the subject
1129 string, and therefore it fails if the current pointer is at the end of the
1130 string.
1131 .P
1132 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1133 can be included in a class as a literal string of data units, or by using the
1134 \ex{ escaping mechanism.
1135 .P
1136 When caseless matching is set, any letters in a class represent both their
1137 upper case and lower case versions, so for example, a caseless [aeiou] matches
1138 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1139 caseful version would. In a UTF mode, PCRE always understands the concept of
1140 case for characters whose values are less than 128, so caseless matching is
1141 always possible. For characters with higher values, the concept of case is
1142 supported if PCRE is compiled with Unicode property support, but not otherwise.
1143 If you want to use caseless matching in a UTF mode for characters 128 and
1144 above, you must ensure that PCRE is compiled with Unicode property support as
1145 well as with UTF support.
1146 .P
1147 Characters that might indicate line breaks are never treated in any special way
1148 when matching character classes, whatever line-ending sequence is in use, and
1149 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1150 such as [^a] always matches one of these characters.
1151 .P
1152 The minus (hyphen) character can be used to specify a range of characters in a
1153 character class. For example, [d-m] matches any letter between d and m,
1154 inclusive. If a minus character is required in a class, it must be escaped with
1155 a backslash or appear in a position where it cannot be interpreted as
1156 indicating a range, typically as the first or last character in the class.
1157 .P
1158 It is not possible to have the literal character "]" as the end character of a
1159 range. A pattern such as [W-]46] is interpreted as a class of two characters
1160 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1161 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1162 the end of range, so [W-\e]46] is interpreted as a class containing a range
1163 followed by two other characters. The octal or hexadecimal representation of
1164 "]" can also be used to end a range.
1165 .P
1166 Ranges operate in the collating sequence of character values. They can also be
1167 used for characters specified numerically, for example [\e000-\e037]. Ranges
1168 can include any characters that are valid for the current mode.
1169 .P
1170 If a range that includes letters is used when caseless matching is set, it
1171 matches the letters in either case. For example, [W-c] is equivalent to
1172 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1173 tables for a French locale are in use, [\exc8-\excb] matches accented E
1174 characters in both cases. In UTF modes, PCRE supports the concept of case for
1175 characters with values greater than 128 only when it is compiled with Unicode
1176 property support.
1177 .P
1178 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1179 \eV, \ew, and \eW may appear in a character class, and add the characters that
1180 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1181 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1182 and their upper case partners, just as it does when they appear outside a
1183 character class, as described in the section entitled
1184 .\" HTML <a href="#genericchartypes">
1185 .\" </a>
1186 "Generic character types"
1187 .\"
1188 above. The escape sequence \eb has a different meaning inside a character
1189 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1190 are not special inside a character class. Like any other unrecognized escape
1191 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1192 default, but cause an error if the PCRE_EXTRA option is set.
1193 .P
1194 A circumflex can conveniently be used with the upper case character types to
1195 specify a more restricted set of characters than the matching lower case type.
1196 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1197 whereas [\ew] includes underscore. A positive character class should be read as
1198 "something OR something OR ..." and a negative class as "NOT something AND NOT
1199 something AND NOT ...".
1200 .P
1201 The only metacharacters that are recognized in character classes are backslash,
1202 hyphen (only where it can be interpreted as specifying a range), circumflex
1203 (only at the start), opening square bracket (only when it can be interpreted as
1204 introducing a POSIX class name - see the next section), and the terminating
1205 closing square bracket. However, escaping other non-alphanumeric characters
1206 does no harm.
1207 .
1208 .
1210 .rs
1211 .sp
1212 Perl supports the POSIX notation for character classes. This uses names
1213 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1214 this notation. For example,
1215 .sp
1216 [01[:alpha:]%]
1217 .sp
1218 matches "0", "1", any alphabetic character, or "%". The supported class names
1219 are:
1220 .sp
1221 alnum letters and digits
1222 alpha letters
1223 ascii character codes 0 - 127
1224 blank space or tab only
1225 cntrl control characters
1226 digit decimal digits (same as \ed)
1227 graph printing characters, excluding space
1228 lower lower case letters
1229 print printing characters, including space
1230 punct printing characters, excluding letters and digits and space
1231 space white space (not quite the same as \es)
1232 upper upper case letters
1233 word "word" characters (same as \ew)
1234 xdigit hexadecimal digits
1235 .sp
1236 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1237 space (32). Notice that this list includes the VT character (code 11). This
1238 makes "space" different to \es, which does not include VT (for Perl
1239 compatibility).
1240 .P
1241 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1242 5.8. Another Perl extension is negation, which is indicated by a ^ character
1243 after the colon. For example,
1244 .sp
1245 [12[:^digit:]]
1246 .sp
1247 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1248 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1249 supported, and an error is given if they are encountered.
1250 .P
1251 By default, in UTF modes, characters with values greater than 128 do not match
1252 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1253 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1254 character properties are used. This is achieved by replacing the POSIX classes
1255 by other sequences, as follows:
1256 .sp
1257 [:alnum:] becomes \ep{Xan}
1258 [:alpha:] becomes \ep{L}
1259 [:blank:] becomes \eh
1260 [:digit:] becomes \ep{Nd}
1261 [:lower:] becomes \ep{Ll}
1262 [:space:] becomes \ep{Xps}
1263 [:upper:] becomes \ep{Lu}
1264 [:word:] becomes \ep{Xwd}
1265 .sp
1266 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1267 classes are unchanged, and match only characters with code points less than
1268 128.
1269 .
1270 .
1272 .rs
1273 .sp
1274 Vertical bar characters are used to separate alternative patterns. For example,
1275 the pattern
1276 .sp
1277 gilbert|sullivan
1278 .sp
1279 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1280 and an empty alternative is permitted (matching the empty string). The matching
1281 process tries each alternative in turn, from left to right, and the first one
1282 that succeeds is used. If the alternatives are within a subpattern
1283 .\" HTML <a href="#subpattern">
1284 .\" </a>
1285 (defined below),
1286 .\"
1287 "succeeds" means matching the rest of the main pattern as well as the
1288 alternative in the subpattern.
1289 .
1290 .
1292 .rs
1293 .sp
1294 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1295 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1296 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1297 The option letters are
1298 .sp
1299 i for PCRE_CASELESS
1301 s for PCRE_DOTALL
1302 x for PCRE_EXTENDED
1303 .sp
1304 For example, (?im) sets caseless, multiline matching. It is also possible to
1305 unset these options by preceding the letter with a hyphen, and a combined
1306 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1307 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1308 permitted. If a letter appears both before and after the hyphen, the option is
1309 unset.
1310 .P
1311 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1312 changed in the same way as the Perl-compatible options by using the characters
1313 J, U and X respectively.
1314 .P
1315 When one of these option changes occurs at top level (that is, not inside
1316 subpattern parentheses), the change applies to the remainder of the pattern
1317 that follows. If the change is placed right at the start of a pattern, PCRE
1318 extracts it into the global options (and it will therefore show up in data
1319 extracted by the \fBpcre_fullinfo()\fP function).
1320 .P
1321 An option change within a subpattern (see below for a description of
1322 subpatterns) affects only that part of the subpattern that follows it, so
1323 .sp
1324 (a(?i)b)c
1325 .sp
1326 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1327 By this means, options can be made to have different settings in different
1328 parts of the pattern. Any changes made in one alternative do carry on
1329 into subsequent branches within the same subpattern. For example,
1330 .sp
1331 (a(?i)b|c)
1332 .sp
1333 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1334 branch is abandoned before the option setting. This is because the effects of
1335 option settings happen at compile time. There would be some very weird
1336 behaviour otherwise.
1337 .P
1338 \fBNote:\fP There are other PCRE-specific options that can be set by the
1339 application when the compiling or matching functions are called. In some cases
1340 the pattern can contain special leading sequences such as (*CRLF) to override
1341 what the application has set or what has been defaulted. Details are given in
1342 the section entitled
1343 .\" HTML <a href="#newlineseq">
1344 .\" </a>
1345 "Newline sequences"
1346 .\"
1347 above. There are also the (*UTF8), (*UTF16),(*UTF32) and (*UCP) leading
1348 sequences that can be used to set UTF and Unicode property modes; they are
1349 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1350 options, respectively.
1351 .
1352 .
1353 .\" HTML <a name="subpattern"></a>
1355 .rs
1356 .sp
1357 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1358 Turning part of a pattern into a subpattern does two things:
1359 .sp
1360 1. It localizes a set of alternatives. For example, the pattern
1361 .sp
1362 cat(aract|erpillar|)
1363 .sp
1364 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1365 match "cataract", "erpillar" or an empty string.
1366 .sp
1367 2. It sets up the subpattern as a capturing subpattern. This means that, when
1368 the whole pattern matches, that portion of the subject string that matched the
1369 subpattern is passed back to the caller via the \fIovector\fP argument of the
1370 matching function. (This applies only to the traditional matching functions;
1371 the DFA matching functions do not support capturing.)
1372 .P
1373 Opening parentheses are counted from left to right (starting from 1) to obtain
1374 numbers for the capturing subpatterns. For example, if the string "the red
1375 king" is matched against the pattern
1376 .sp
1377 the ((red|white) (king|queen))
1378 .sp
1379 the captured substrings are "red king", "red", and "king", and are numbered 1,
1380 2, and 3, respectively.
1381 .P
1382 The fact that plain parentheses fulfil two functions is not always helpful.
1383 There are often times when a grouping subpattern is required without a
1384 capturing requirement. If an opening parenthesis is followed by a question mark
1385 and a colon, the subpattern does not do any capturing, and is not counted when
1386 computing the number of any subsequent capturing subpatterns. For example, if
1387 the string "the white queen" is matched against the pattern
1388 .sp
1389 the ((?:red|white) (king|queen))
1390 .sp
1391 the captured substrings are "white queen" and "queen", and are numbered 1 and
1392 2. The maximum number of capturing subpatterns is 65535.
1393 .P
1394 As a convenient shorthand, if any option settings are required at the start of
1395 a non-capturing subpattern, the option letters may appear between the "?" and
1396 the ":". Thus the two patterns
1397 .sp
1398 (?i:saturday|sunday)
1399 (?:(?i)saturday|sunday)
1400 .sp
1401 match exactly the same set of strings. Because alternative branches are tried
1402 from left to right, and options are not reset until the end of the subpattern
1403 is reached, an option setting in one branch does affect subsequent branches, so
1404 the above patterns match "SUNDAY" as well as "Saturday".
1405 .
1406 .
1407 .\" HTML <a name="dupsubpatternnumber"></a>
1409 .rs
1410 .sp
1411 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1412 the same numbers for its capturing parentheses. Such a subpattern starts with
1413 (?| and is itself a non-capturing subpattern. For example, consider this
1414 pattern:
1415 .sp
1416 (?|(Sat)ur|(Sun))day
1417 .sp
1418 Because the two alternatives are inside a (?| group, both sets of capturing
1419 parentheses are numbered one. Thus, when the pattern matches, you can look
1420 at captured substring number one, whichever alternative matched. This construct
1421 is useful when you want to capture part, but not all, of one of a number of
1422 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1423 number is reset at the start of each branch. The numbers of any capturing
1424 parentheses that follow the subpattern start after the highest number used in
1425 any branch. The following example is taken from the Perl documentation. The
1426 numbers underneath show in which buffer the captured content will be stored.
1427 .sp
1428 # before ---------------branch-reset----------- after
1429 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1430 # 1 2 2 3 2 3 4
1431 .sp
1432 A back reference to a numbered subpattern uses the most recent value that is
1433 set for that number by any subpattern. The following pattern matches "abcabc"
1434 or "defdef":
1435 .sp
1436 /(?|(abc)|(def))\e1/
1437 .sp
1438 In contrast, a subroutine call to a numbered subpattern always refers to the
1439 first one in the pattern with the given number. The following pattern matches
1440 "abcabc" or "defabc":
1441 .sp
1442 /(?|(abc)|(def))(?1)/
1443 .sp
1444 If a
1445 .\" HTML <a href="#conditions">
1446 .\" </a>
1447 condition test
1448 .\"
1449 for a subpattern's having matched refers to a non-unique number, the test is
1450 true if any of the subpatterns of that number have matched.
1451 .P
1452 An alternative approach to using this "branch reset" feature is to use
1453 duplicate named subpatterns, as described in the next section.
1454 .
1455 .
1457 .rs
1458 .sp
1459 Identifying capturing parentheses by number is simple, but it can be very hard
1460 to keep track of the numbers in complicated regular expressions. Furthermore,
1461 if an expression is modified, the numbers may change. To help with this
1462 difficulty, PCRE supports the naming of subpatterns. This feature was not
1463 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1464 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1465 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1466 have different names, but PCRE does not.
1467 .P
1468 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1469 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1470 parentheses from other parts of the pattern, such as
1471 .\" HTML <a href="#backreferences">
1472 .\" </a>
1473 back references,
1474 .\"
1475 .\" HTML <a href="#recursion">
1476 .\" </a>
1477 recursion,
1478 .\"
1479 and
1480 .\" HTML <a href="#conditions">
1481 .\" </a>
1482 conditions,
1483 .\"
1484 can be made by name as well as by number.
1485 .P
1486 Names consist of up to 32 alphanumeric characters and underscores. Named
1487 capturing parentheses are still allocated numbers as well as names, exactly as
1488 if the names were not present. The PCRE API provides function calls for
1489 extracting the name-to-number translation table from a compiled pattern. There
1490 is also a convenience function for extracting a captured substring by name.
1491 .P
1492 By default, a name must be unique within a pattern, but it is possible to relax
1493 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1494 names are also always permitted for subpatterns with the same number, set up as
1495 described in the previous section.) Duplicate names can be useful for patterns
1496 where only one instance of the named parentheses can match. Suppose you want to
1497 match the name of a weekday, either as a 3-letter abbreviation or as the full
1498 name, and in both cases you want to extract the abbreviation. This pattern
1499 (ignoring the line breaks) does the job:
1500 .sp
1501 (?<DN>Mon|Fri|Sun)(?:day)?|
1502 (?<DN>Tue)(?:sday)?|
1503 (?<DN>Wed)(?:nesday)?|
1504 (?<DN>Thu)(?:rsday)?|
1505 (?<DN>Sat)(?:urday)?
1506 .sp
1507 There are five capturing substrings, but only one is ever set after a match.
1508 (An alternative way of solving this problem is to use a "branch reset"
1509 subpattern, as described in the previous section.)
1510 .P
1511 The convenience function for extracting the data by name returns the substring
1512 for the first (and in this example, the only) subpattern of that name that
1513 matched. This saves searching to find which numbered subpattern it was.
1514 .P
1515 If you make a back reference to a non-unique named subpattern from elsewhere in
1516 the pattern, the one that corresponds to the first occurrence of the name is
1517 used. In the absence of duplicate numbers (see the previous section) this is
1518 the one with the lowest number. If you use a named reference in a condition
1519 test (see the
1520 .\"
1521 .\" HTML <a href="#conditions">
1522 .\" </a>
1523 section about conditions
1524 .\"
1525 below), either to check whether a subpattern has matched, or to check for
1526 recursion, all subpatterns with the same name are tested. If the condition is
1527 true for any one of them, the overall condition is true. This is the same
1528 behaviour as testing by number. For further details of the interfaces for
1529 handling named subpatterns, see the
1530 .\" HREF
1531 \fBpcreapi\fP
1532 .\"
1533 documentation.
1534 .P
1535 \fBWarning:\fP You cannot use different names to distinguish between two
1536 subpatterns with the same number because PCRE uses only the numbers when
1537 matching. For this reason, an error is given at compile time if different names
1538 are given to subpatterns with the same number. However, you can give the same
1539 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1540 .
1541 .
1543 .rs
1544 .sp
1545 Repetition is specified by quantifiers, which can follow any of the following
1546 items:
1547 .sp
1548 a literal data character
1549 the dot metacharacter
1550 the \eC escape sequence
1551 the \eX escape sequence
1552 the \eR escape sequence
1553 an escape such as \ed or \epL that matches a single character
1554 a character class
1555 a back reference (see next section)
1556 a parenthesized subpattern (including assertions)
1557 a subroutine call to a subpattern (recursive or otherwise)
1558 .sp
1559 The general repetition quantifier specifies a minimum and maximum number of
1560 permitted matches, by giving the two numbers in curly brackets (braces),
1561 separated by a comma. The numbers must be less than 65536, and the first must
1562 be less than or equal to the second. For example:
1563 .sp
1564 z{2,4}
1565 .sp
1566 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1567 character. If the second number is omitted, but the comma is present, there is
1568 no upper limit; if the second number and the comma are both omitted, the
1569 quantifier specifies an exact number of required matches. Thus
1570 .sp
1571 [aeiou]{3,}
1572 .sp
1573 matches at least 3 successive vowels, but may match many more, while
1574 .sp
1575 \ed{8}
1576 .sp
1577 matches exactly 8 digits. An opening curly bracket that appears in a position
1578 where a quantifier is not allowed, or one that does not match the syntax of a
1579 quantifier, is taken as a literal character. For example, {,6} is not a
1580 quantifier, but a literal string of four characters.
1581 .P
1582 In UTF modes, quantifiers apply to characters rather than to individual data
1583 units. Thus, for example, \ex{100}{2} matches two characters, each of
1584 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1585 \eX{3} matches three Unicode extended grapheme clusters, each of which may be
1586 several data units long (and they may be of different lengths).
1587 .P
1588 The quantifier {0} is permitted, causing the expression to behave as if the
1589 previous item and the quantifier were not present. This may be useful for
1590 subpatterns that are referenced as
1591 .\" HTML <a href="#subpatternsassubroutines">
1592 .\" </a>
1593 subroutines
1594 .\"
1595 from elsewhere in the pattern (but see also the section entitled
1596 .\" HTML <a href="#subdefine">
1597 .\" </a>
1598 "Defining subpatterns for use by reference only"
1599 .\"
1600 below). Items other than subpatterns that have a {0} quantifier are omitted
1601 from the compiled pattern.
1602 .P
1603 For convenience, the three most common quantifiers have single-character
1604 abbreviations:
1605 .sp
1606 * is equivalent to {0,}
1607 + is equivalent to {1,}
1608 ? is equivalent to {0,1}
1609 .sp
1610 It is possible to construct infinite loops by following a subpattern that can
1611 match no characters with a quantifier that has no upper limit, for example:
1612 .sp
1613 (a?)*
1614 .sp
1615 Earlier versions of Perl and PCRE used to give an error at compile time for
1616 such patterns. However, because there are cases where this can be useful, such
1617 patterns are now accepted, but if any repetition of the subpattern does in fact
1618 match no characters, the loop is forcibly broken.
1619 .P
1620 By default, the quantifiers are "greedy", that is, they match as much as
1621 possible (up to the maximum number of permitted times), without causing the
1622 rest of the pattern to fail. The classic example of where this gives problems
1623 is in trying to match comments in C programs. These appear between /* and */
1624 and within the comment, individual * and / characters may appear. An attempt to
1625 match C comments by applying the pattern
1626 .sp
1627 /\e*.*\e*/
1628 .sp
1629 to the string
1630 .sp
1631 /* first comment */ not comment /* second comment */
1632 .sp
1633 fails, because it matches the entire string owing to the greediness of the .*
1634 item.
1635 .P
1636 However, if a quantifier is followed by a question mark, it ceases to be
1637 greedy, and instead matches the minimum number of times possible, so the
1638 pattern
1639 .sp
1640 /\e*.*?\e*/
1641 .sp
1642 does the right thing with the C comments. The meaning of the various
1643 quantifiers is not otherwise changed, just the preferred number of matches.
1644 Do not confuse this use of question mark with its use as a quantifier in its
1645 own right. Because it has two uses, it can sometimes appear doubled, as in
1646 .sp
1647 \ed??\ed
1648 .sp
1649 which matches one digit by preference, but can match two if that is the only
1650 way the rest of the pattern matches.
1651 .P
1652 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1653 the quantifiers are not greedy by default, but individual ones can be made
1654 greedy by following them with a question mark. In other words, it inverts the
1655 default behaviour.
1656 .P
1657 When a parenthesized subpattern is quantified with a minimum repeat count that
1658 is greater than 1 or with a limited maximum, more memory is required for the
1659 compiled pattern, in proportion to the size of the minimum or maximum.
1660 .P
1661 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1662 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1663 implicitly anchored, because whatever follows will be tried against every
1664 character position in the subject string, so there is no point in retrying the
1665 overall match at any position after the first. PCRE normally treats such a
1666 pattern as though it were preceded by \eA.
1667 .P
1668 In cases where it is known that the subject string contains no newlines, it is
1669 worth setting PCRE_DOTALL in order to obtain this optimization, or
1670 alternatively using ^ to indicate anchoring explicitly.
1671 .P
1672 However, there are some cases where the optimization cannot be used. When .*
1673 is inside capturing parentheses that are the subject of a back reference
1674 elsewhere in the pattern, a match at the start may fail where a later one
1675 succeeds. Consider, for example:
1676 .sp
1677 (.*)abc\e1
1678 .sp
1679 If the subject is "xyz123abc123" the match point is the fourth character. For
1680 this reason, such a pattern is not implicitly anchored.
1681 .P
1682 Another case where implicit anchoring is not applied is when the leading .* is
1683 inside an atomic group. Once again, a match at the start may fail where a later
1684 one succeeds. Consider this pattern:
1685 .sp
1686 (?>.*?a)b
1687 .sp
1688 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1689 (*PRUNE) and (*SKIP) also disable this optimization.
1690 .P
1691 When a capturing subpattern is repeated, the value captured is the substring
1692 that matched the final iteration. For example, after
1693 .sp
1694 (tweedle[dume]{3}\es*)+
1695 .sp
1696 has matched "tweedledum tweedledee" the value of the captured substring is
1697 "tweedledee". However, if there are nested capturing subpatterns, the
1698 corresponding captured values may have been set in previous iterations. For
1699 example, after
1700 .sp
1701 /(a|(b))+/
1702 .sp
1703 matches "aba" the value of the second captured substring is "b".
1704 .
1705 .
1706 .\" HTML <a name="atomicgroup"></a>
1708 .rs
1709 .sp
1710 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1711 repetition, failure of what follows normally causes the repeated item to be
1712 re-evaluated to see if a different number of repeats allows the rest of the
1713 pattern to match. Sometimes it is useful to prevent this, either to change the
1714 nature of the match, or to cause it fail earlier than it otherwise might, when
1715 the author of the pattern knows there is no point in carrying on.
1716 .P
1717 Consider, for example, the pattern \ed+foo when applied to the subject line
1718 .sp
1719 123456bar
1720 .sp
1721 After matching all 6 digits and then failing to match "foo", the normal
1722 action of the matcher is to try again with only 5 digits matching the \ed+
1723 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1724 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1725 that once a subpattern has matched, it is not to be re-evaluated in this way.
1726 .P
1727 If we use atomic grouping for the previous example, the matcher gives up
1728 immediately on failing to match "foo" the first time. The notation is a kind of
1729 special parenthesis, starting with (?> as in this example:
1730 .sp
1731 (?>\ed+)foo
1732 .sp
1733 This kind of parenthesis "locks up" the part of the pattern it contains once
1734 it has matched, and a failure further into the pattern is prevented from
1735 backtracking into it. Backtracking past it to previous items, however, works as
1736 normal.
1737 .P
1738 An alternative description is that a subpattern of this type matches the string
1739 of characters that an identical standalone pattern would match, if anchored at
1740 the current point in the subject string.
1741 .P
1742 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1743 the above example can be thought of as a maximizing repeat that must swallow
1744 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1745 number of digits they match in order to make the rest of the pattern match,
1746 (?>\ed+) can only match an entire sequence of digits.
1747 .P
1748 Atomic groups in general can of course contain arbitrarily complicated
1749 subpatterns, and can be nested. However, when the subpattern for an atomic
1750 group is just a single repeated item, as in the example above, a simpler
1751 notation, called a "possessive quantifier" can be used. This consists of an
1752 additional + character following a quantifier. Using this notation, the
1753 previous example can be rewritten as
1754 .sp
1755 \ed++foo
1756 .sp
1757 Note that a possessive quantifier can be used with an entire group, for
1758 example:
1759 .sp
1760 (abc|xyz){2,3}+
1761 .sp
1762 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1763 option is ignored. They are a convenient notation for the simpler forms of
1764 atomic group. However, there is no difference in the meaning of a possessive
1765 quantifier and the equivalent atomic group, though there may be a performance
1766 difference; possessive quantifiers should be slightly faster.
1767 .P
1768 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1769 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1770 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1771 package, and PCRE copied it from there. It ultimately found its way into Perl
1772 at release 5.10.
1773 .P
1774 PCRE has an optimization that automatically "possessifies" certain simple
1775 pattern constructs. For example, the sequence A+B is treated as A++B because
1776 there is no point in backtracking into a sequence of A's when B must follow.
1777 .P
1778 When a pattern contains an unlimited repeat inside a subpattern that can itself
1779 be repeated an unlimited number of times, the use of an atomic group is the
1780 only way to avoid some failing matches taking a very long time indeed. The
1781 pattern
1782 .sp
1783 (\eD+|<\ed+>)*[!?]
1784 .sp
1785 matches an unlimited number of substrings that either consist of non-digits, or
1786 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1787 quickly. However, if it is applied to
1788 .sp
1789 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1790 .sp
1791 it takes a long time before reporting failure. This is because the string can
1792 be divided between the internal \eD+ repeat and the external * repeat in a
1793 large number of ways, and all have to be tried. (The example uses [!?] rather
1794 than a single character at the end, because both PCRE and Perl have an
1795 optimization that allows for fast failure when a single character is used. They
1796 remember the last single character that is required for a match, and fail early
1797 if it is not present in the string.) If the pattern is changed so that it uses
1798 an atomic group, like this:
1799 .sp
1800 ((?>\eD+)|<\ed+>)*[!?]
1801 .sp
1802 sequences of non-digits cannot be broken, and failure happens quickly.
1803 .
1804 .
1805 .\" HTML <a name="backreferences"></a>
1807 .rs
1808 .sp
1809 Outside a character class, a backslash followed by a digit greater than 0 (and
1810 possibly further digits) is a back reference to a capturing subpattern earlier
1811 (that is, to its left) in the pattern, provided there have been that many
1812 previous capturing left parentheses.
1813 .P
1814 However, if the decimal number following the backslash is less than 10, it is
1815 always taken as a back reference, and causes an error only if there are not
1816 that many capturing left parentheses in the entire pattern. In other words, the
1817 parentheses that are referenced need not be to the left of the reference for
1818 numbers less than 10. A "forward back reference" of this type can make sense
1819 when a repetition is involved and the subpattern to the right has participated
1820 in an earlier iteration.
1821 .P
1822 It is not possible to have a numerical "forward back reference" to a subpattern
1823 whose number is 10 or more using this syntax because a sequence such as \e50 is
1824 interpreted as a character defined in octal. See the subsection entitled
1825 "Non-printing characters"
1826 .\" HTML <a href="#digitsafterbackslash">
1827 .\" </a>
1828 above
1829 .\"
1830 for further details of the handling of digits following a backslash. There is
1831 no such problem when named parentheses are used. A back reference to any
1832 subpattern is possible using named parentheses (see below).
1833 .P
1834 Another way of avoiding the ambiguity inherent in the use of digits following a
1835 backslash is to use the \eg escape sequence. This escape must be followed by an
1836 unsigned number or a negative number, optionally enclosed in braces. These
1837 examples are all identical:
1838 .sp
1839 (ring), \e1
1840 (ring), \eg1
1841 (ring), \eg{1}
1842 .sp
1843 An unsigned number specifies an absolute reference without the ambiguity that
1844 is present in the older syntax. It is also useful when literal digits follow
1845 the reference. A negative number is a relative reference. Consider this
1846 example:
1847 .sp
1848 (abc(def)ghi)\eg{-1}
1849 .sp
1850 The sequence \eg{-1} is a reference to the most recently started capturing
1851 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1852 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1853 can be helpful in long patterns, and also in patterns that are created by
1854 joining together fragments that contain references within themselves.
1855 .P
1856 A back reference matches whatever actually matched the capturing subpattern in
1857 the current subject string, rather than anything matching the subpattern
1858 itself (see
1859 .\" HTML <a href="#subpatternsassubroutines">
1860 .\" </a>
1861 "Subpatterns as subroutines"
1862 .\"
1863 below for a way of doing that). So the pattern
1864 .sp
1865 (sens|respons)e and \e1ibility
1866 .sp
1867 matches "sense and sensibility" and "response and responsibility", but not
1868 "sense and responsibility". If caseful matching is in force at the time of the
1869 back reference, the case of letters is relevant. For example,
1870 .sp
1871 ((?i)rah)\es+\e1
1872 .sp
1873 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1874 capturing subpattern is matched caselessly.
1875 .P
1876 There are several different ways of writing back references to named
1877 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1878 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1879 back reference syntax, in which \eg can be used for both numeric and named
1880 references, is also supported. We could rewrite the above example in any of
1881 the following ways:
1882 .sp
1883 (?<p1>(?i)rah)\es+\ek<p1>
1884 (?'p1'(?i)rah)\es+\ek{p1}
1885 (?P<p1>(?i)rah)\es+(?P=p1)
1886 (?<p1>(?i)rah)\es+\eg{p1}
1887 .sp
1888 A subpattern that is referenced by name may appear in the pattern before or
1889 after the reference.
1890 .P
1891 There may be more than one back reference to the same subpattern. If a
1892 subpattern has not actually been used in a particular match, any back
1893 references to it always fail by default. For example, the pattern
1894 .sp
1895 (a|(bc))\e2
1896 .sp
1897 always fails if it starts to match "a" rather than "bc". However, if the
1898 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1899 unset value matches an empty string.
1900 .P
1901 Because there may be many capturing parentheses in a pattern, all digits
1902 following a backslash are taken as part of a potential back reference number.
1903 If the pattern continues with a digit character, some delimiter must be used to
1904 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1905 white space. Otherwise, the \eg{ syntax or an empty comment (see
1906 .\" HTML <a href="#comments">
1907 .\" </a>
1908 "Comments"
1909 .\"
1910 below) can be used.
1911 .
1912 .SS "Recursive back references"
1913 .rs
1914 .sp
1915 A back reference that occurs inside the parentheses to which it refers fails
1916 when the subpattern is first used, so, for example, (a\e1) never matches.
1917 However, such references can be useful inside repeated subpatterns. For
1918 example, the pattern
1919 .sp
1920 (a|b\e1)+
1921 .sp
1922 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1923 the subpattern, the back reference matches the character string corresponding
1924 to the previous iteration. In order for this to work, the pattern must be such
1925 that the first iteration does not need to match the back reference. This can be
1926 done using alternation, as in the example above, or by a quantifier with a
1927 minimum of zero.
1928 .P
1929 Back references of this type cause the group that they reference to be treated
1930 as an
1931 .\" HTML <a href="#atomicgroup">
1932 .\" </a>
1933 atomic group.
1934 .\"
1935 Once the whole group has been matched, a subsequent matching failure cannot
1936 cause backtracking into the middle of the group.
1937 .
1938 .
1939 .\" HTML <a name="bigassertions"></a>
1941 .rs
1942 .sp
1943 An assertion is a test on the characters following or preceding the current
1944 matching point that does not actually consume any characters. The simple
1945 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1946 .\" HTML <a href="#smallassertions">
1947 .\" </a>
1948 above.
1949 .\"
1950 .P
1951 More complicated assertions are coded as subpatterns. There are two kinds:
1952 those that look ahead of the current position in the subject string, and those
1953 that look behind it. An assertion subpattern is matched in the normal way,
1954 except that it does not cause the current matching position to be changed.
1955 .P
1956 Assertion subpatterns are not capturing subpatterns. If such an assertion
1957 contains capturing subpatterns within it, these are counted for the purposes of
1958 numbering the capturing subpatterns in the whole pattern. However, substring
1959 capturing is carried out only for positive assertions, because it does not make
1960 sense for negative assertions.
1961 .P
1962 For compatibility with Perl, assertion subpatterns may be repeated; though
1963 it makes no sense to assert the same thing several times, the side effect of
1964 capturing parentheses may occasionally be useful. In practice, there only three
1965 cases:
1966 .sp
1967 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1968 However, it may contain internal capturing parenthesized groups that are called
1969 from elsewhere via the
1970 .\" HTML <a href="#subpatternsassubroutines">
1971 .\" </a>
1972 subroutine mechanism.
1973 .\"
1974 .sp
1975 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1976 were {0,1}. At run time, the rest of the pattern match is tried with and
1977 without the assertion, the order depending on the greediness of the quantifier.
1978 .sp
1979 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1980 The assertion is obeyed just once when encountered during matching.
1981 .
1982 .
1983 .SS "Lookahead assertions"
1984 .rs
1985 .sp
1986 Lookahead assertions start with (?= for positive assertions and (?! for
1987 negative assertions. For example,
1988 .sp
1989 \ew+(?=;)
1990 .sp
1991 matches a word followed by a semicolon, but does not include the semicolon in
1992 the match, and
1993 .sp
1994 foo(?!bar)
1995 .sp
1996 matches any occurrence of "foo" that is not followed by "bar". Note that the
1997 apparently similar pattern
1998 .sp
1999 (?!foo)bar
2000 .sp
2001 does not find an occurrence of "bar" that is preceded by something other than
2002 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2003 (?!foo) is always true when the next three characters are "bar". A
2004 lookbehind assertion is needed to achieve the other effect.
2005 .P
2006 If you want to force a matching failure at some point in a pattern, the most
2007 convenient way to do it is with (?!) because an empty string always matches, so
2008 an assertion that requires there not to be an empty string must always fail.
2009 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2010 .
2011 .
2012 .\" HTML <a name="lookbehind"></a>
2013 .SS "Lookbehind assertions"
2014 .rs
2015 .sp
2016 Lookbehind assertions start with (?<= for positive assertions and (?<! for
2017 negative assertions. For example,
2018 .sp
2019 (?<!foo)bar
2020 .sp
2021 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2022 a lookbehind assertion are restricted such that all the strings it matches must
2023 have a fixed length. However, if there are several top-level alternatives, they
2024 do not all have to have the same fixed length. Thus
2025 .sp
2026 (?<=bullock|donkey)
2027 .sp
2028 is permitted, but
2029 .sp
2030 (?<!dogs?|cats?)
2031 .sp
2032 causes an error at compile time. Branches that match different length strings
2033 are permitted only at the top level of a lookbehind assertion. This is an
2034 extension compared with Perl, which requires all branches to match the same
2035 length of string. An assertion such as
2036 .sp
2037 (?<=ab(c|de))
2038 .sp
2039 is not permitted, because its single top-level branch can match two different
2040 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2041 branches:
2042 .sp
2043 (?<=abc|abde)
2044 .sp
2045 In some cases, the escape sequence \eK
2046 .\" HTML <a href="#resetmatchstart">
2047 .\" </a>
2048 (see above)
2049 .\"
2050 can be used instead of a lookbehind assertion to get round the fixed-length
2051 restriction.
2052 .P
2053 The implementation of lookbehind assertions is, for each alternative, to
2054 temporarily move the current position back by the fixed length and then try to
2055 match. If there are insufficient characters before the current position, the
2056 assertion fails.
2057 .P
2058 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2059 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2060 it impossible to calculate the length of the lookbehind. The \eX and \eR
2061 escapes, which can match different numbers of data units, are also not
2062 permitted.
2063 .P
2064 .\" HTML <a href="#subpatternsassubroutines">
2065 .\" </a>
2066 "Subroutine"
2067 .\"
2068 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2069 as the subpattern matches a fixed-length string.
2070 .\" HTML <a href="#recursion">
2071 .\" </a>
2072 Recursion,
2073 .\"
2074 however, is not supported.
2075 .P
2076 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2077 specify efficient matching of fixed-length strings at the end of subject
2078 strings. Consider a simple pattern such as
2079 .sp
2080 abcd$
2081 .sp
2082 when applied to a long string that does not match. Because matching proceeds
2083 from left to right, PCRE will look for each "a" in the subject and then see if
2084 what follows matches the rest of the pattern. If the pattern is specified as
2085 .sp
2086 ^.*abcd$
2087 .sp
2088 the initial .* matches the entire string at first, but when this fails (because
2089 there is no following "a"), it backtracks to match all but the last character,
2090 then all but the last two characters, and so on. Once again the search for "a"
2091 covers the entire string, from right to left, so we are no better off. However,
2092 if the pattern is written as
2093 .sp
2094 ^.*+(?<=abcd)
2095 .sp
2096 there can be no backtracking for the .*+ item; it can match only the entire
2097 string. The subsequent lookbehind assertion does a single test on the last four
2098 characters. If it fails, the match fails immediately. For long strings, this
2099 approach makes a significant difference to the processing time.
2100 .
2101 .
2102 .SS "Using multiple assertions"
2103 .rs
2104 .sp
2105 Several assertions (of any sort) may occur in succession. For example,
2106 .sp
2107 (?<=\ed{3})(?<!999)foo
2108 .sp
2109 matches "foo" preceded by three digits that are not "999". Notice that each of
2110 the assertions is applied independently at the same point in the subject
2111 string. First there is a check that the previous three characters are all
2112 digits, and then there is a check that the same three characters are not "999".
2113 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2114 of which are digits and the last three of which are not "999". For example, it
2115 doesn't match "123abcfoo". A pattern to do that is
2116 .sp
2117 (?<=\ed{3}...)(?<!999)foo
2118 .sp
2119 This time the first assertion looks at the preceding six characters, checking
2120 that the first three are digits, and then the second assertion checks that the
2121 preceding three characters are not "999".
2122 .P
2123 Assertions can be nested in any combination. For example,
2124 .sp
2125 (?<=(?<!foo)bar)baz
2126 .sp
2127 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2128 preceded by "foo", while
2129 .sp
2130 (?<=\ed{3}(?!999)...)foo
2131 .sp
2132 is another pattern that matches "foo" preceded by three digits and any three
2133 characters that are not "999".
2134 .
2135 .
2136 .\" HTML <a name="conditions"></a>
2138 .rs
2139 .sp
2140 It is possible to cause the matching process to obey a subpattern
2141 conditionally or to choose between two alternative subpatterns, depending on
2142 the result of an assertion, or whether a specific capturing subpattern has
2143 already been matched. The two possible forms of conditional subpattern are:
2144 .sp
2145 (?(condition)yes-pattern)
2146 (?(condition)yes-pattern|no-pattern)
2147 .sp
2148 If the condition is satisfied, the yes-pattern is used; otherwise the
2149 no-pattern (if present) is used. If there are more than two alternatives in the
2150 subpattern, a compile-time error occurs. Each of the two alternatives may
2151 itself contain nested subpatterns of any form, including conditional
2152 subpatterns; the restriction to two alternatives applies only at the level of
2153 the condition. This pattern fragment is an example where the alternatives are
2154 complex:
2155 .sp
2156 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2157 .sp
2158 .P
2159 There are four kinds of condition: references to subpatterns, references to
2160 recursion, a pseudo-condition called DEFINE, and assertions.
2161 .
2162 .SS "Checking for a used subpattern by number"
2163 .rs
2164 .sp
2165 If the text between the parentheses consists of a sequence of digits, the
2166 condition is true if a capturing subpattern of that number has previously
2167 matched. If there is more than one capturing subpattern with the same number
2168 (see the earlier
2169 .\"
2170 .\" HTML <a href="#recursion">
2171 .\" </a>
2172 section about duplicate subpattern numbers),
2173 .\"
2174 the condition is true if any of them have matched. An alternative notation is
2175 to precede the digits with a plus or minus sign. In this case, the subpattern
2176 number is relative rather than absolute. The most recently opened parentheses
2177 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2178 loops it can also make sense to refer to subsequent groups. The next
2179 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2180 zero in any of these forms is not used; it provokes a compile-time error.)
2181 .P
2182 Consider the following pattern, which contains non-significant white space to
2183 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2184 three parts for ease of discussion:
2185 .sp
2186 ( \e( )? [^()]+ (?(1) \e) )
2187 .sp
2188 The first part matches an optional opening parenthesis, and if that
2189 character is present, sets it as the first captured substring. The second part
2190 matches one or more characters that are not parentheses. The third part is a
2191 conditional subpattern that tests whether or not the first set of parentheses
2192 matched. If they did, that is, if subject started with an opening parenthesis,
2193 the condition is true, and so the yes-pattern is executed and a closing
2194 parenthesis is required. Otherwise, since no-pattern is not present, the
2195 subpattern matches nothing. In other words, this pattern matches a sequence of
2196 non-parentheses, optionally enclosed in parentheses.
2197 .P
2198 If you were embedding this pattern in a larger one, you could use a relative
2199 reference:
2200 .sp
2201 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2202 .sp
2203 This makes the fragment independent of the parentheses in the larger pattern.
2204 .
2205 .SS "Checking for a used subpattern by name"
2206 .rs
2207 .sp
2208 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2209 subpattern by name. For compatibility with earlier versions of PCRE, which had
2210 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2211 there is a possible ambiguity with this syntax, because subpattern names may
2212 consist entirely of digits. PCRE looks first for a named subpattern; if it
2213 cannot find one and the name consists entirely of digits, PCRE looks for a
2214 subpattern of that number, which must be greater than zero. Using subpattern
2215 names that consist entirely of digits is not recommended.
2216 .P
2217 Rewriting the above example to use a named subpattern gives this:
2218 .sp
2219 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2220 .sp
2221 If the name used in a condition of this kind is a duplicate, the test is
2222 applied to all subpatterns of the same name, and is true if any one of them has
2223 matched.
2224 .
2225 .SS "Checking for pattern recursion"
2226 .rs
2227 .sp
2228 If the condition is the string (R), and there is no subpattern with the name R,
2229 the condition is true if a recursive call to the whole pattern or any
2230 subpattern has been made. If digits or a name preceded by ampersand follow the
2231 letter R, for example:
2232 .sp
2233 (?(R3)...) or (?(R&name)...)
2234 .sp
2235 the condition is true if the most recent recursion is into a subpattern whose
2236 number or name is given. This condition does not check the entire recursion
2237 stack. If the name used in a condition of this kind is a duplicate, the test is
2238 applied to all subpatterns of the same name, and is true if any one of them is
2239 the most recent recursion.
2240 .P
2241 At "top level", all these recursion test conditions are false.
2242 .\" HTML <a href="#recursion">
2243 .\" </a>
2244 The syntax for recursive patterns
2245 .\"
2246 is described below.
2247 .
2248 .\" HTML <a name="subdefine"></a>
2249 .SS "Defining subpatterns for use by reference only"
2250 .rs
2251 .sp
2252 If the condition is the string (DEFINE), and there is no subpattern with the
2253 name DEFINE, the condition is always false. In this case, there may be only one
2254 alternative in the subpattern. It is always skipped if control reaches this
2255 point in the pattern; the idea of DEFINE is that it can be used to define
2256 subroutines that can be referenced from elsewhere. (The use of
2257 .\" HTML <a href="#subpatternsassubroutines">
2258 .\" </a>
2259 subroutines
2260 .\"
2261 is described below.) For example, a pattern to match an IPv4 address such as
2262 "" could be written like this (ignore white space and line
2263 breaks):
2264 .sp
2265 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2266 \eb (?&byte) (\e.(?&byte)){3} \eb
2267 .sp
2268 The first part of the pattern is a DEFINE group inside which a another group
2269 named "byte" is defined. This matches an individual component of an IPv4
2270 address (a number less than 256). When matching takes place, this part of the
2271 pattern is skipped because DEFINE acts like a false condition. The rest of the
2272 pattern uses references to the named group to match the four dot-separated
2273 components of an IPv4 address, insisting on a word boundary at each end.
2274 .
2275 .SS "Assertion conditions"
2276 .rs
2277 .sp
2278 If the condition is not in any of the above formats, it must be an assertion.
2279 This may be a positive or negative lookahead or lookbehind assertion. Consider
2280 this pattern, again containing non-significant white space, and with the two
2281 alternatives on the second line:
2282 .sp
2283 (?(?=[^a-z]*[a-z])
2284 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2285 .sp
2286 The condition is a positive lookahead assertion that matches an optional
2287 sequence of non-letters followed by a letter. In other words, it tests for the
2288 presence of at least one letter in the subject. If a letter is found, the
2289 subject is matched against the first alternative; otherwise it is matched
2290 against the second. This pattern matches strings in one of the two forms
2291 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2292 .
2293 .
2294 .\" HTML <a name="comments"></a>
2296 .rs
2297 .sp
2298 There are two ways of including comments in patterns that are processed by
2299 PCRE. In both cases, the start of the comment must not be in a character class,
2300 nor in the middle of any other sequence of related characters such as (?: or a
2301 subpattern name or number. The characters that make up a comment play no part
2302 in the pattern matching.
2303 .P
2304 The sequence (?# marks the start of a comment that continues up to the next
2305 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2306 option is set, an unescaped # character also introduces a comment, which in
2307 this case continues to immediately after the next newline character or
2308 character sequence in the pattern. Which characters are interpreted as newlines
2309 is controlled by the options passed to a compiling function or by a special
2310 sequence at the start of the pattern, as described in the section entitled
2311 .\" HTML <a href="#newlines">
2312 .\" </a>
2313 "Newline conventions"
2314 .\"
2315 above. Note that the end of this type of comment is a literal newline sequence
2316 in the pattern; escape sequences that happen to represent a newline do not
2317 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2318 default newline convention is in force:
2319 .sp
2320 abc #comment \en still comment
2321 .sp
2322 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2323 a newline in the pattern. The sequence \en is still literal at this stage, so
2324 it does not terminate the comment. Only an actual character with the code value
2325 0x0a (the default newline) does so.
2326 .
2327 .
2328 .\" HTML <a name="recursion"></a>
2330 .rs
2331 .sp
2332 Consider the problem of matching a string in parentheses, allowing for
2333 unlimited nested parentheses. Without the use of recursion, the best that can
2334 be done is to use a pattern that matches up to some fixed depth of nesting. It
2335 is not possible to handle an arbitrary nesting depth.
2336 .P
2337 For some time, Perl has provided a facility that allows regular expressions to
2338 recurse (amongst other things). It does this by interpolating Perl code in the
2339 expression at run time, and the code can refer to the expression itself. A Perl
2340 pattern using code interpolation to solve the parentheses problem can be
2341 created like this:
2342 .sp
2343 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2344 .sp
2345 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2346 recursively to the pattern in which it appears.
2347 .P
2348 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2349 supports special syntax for recursion of the entire pattern, and also for
2350 individual subpattern recursion. After its introduction in PCRE and Python,
2351 this kind of recursion was subsequently introduced into Perl at release 5.10.
2352 .P
2353 A special item that consists of (? followed by a number greater than zero and a
2354 closing parenthesis is a recursive subroutine call of the subpattern of the
2355 given number, provided that it occurs inside that subpattern. (If not, it is a
2356 .\" HTML <a href="#subpatternsassubroutines">
2357 .\" </a>
2358 non-recursive subroutine
2359 .\"
2360 call, which is described in the next section.) The special item (?R) or (?0) is
2361 a recursive call of the entire regular expression.
2362 .P
2363 This PCRE pattern solves the nested parentheses problem (assume the
2364 PCRE_EXTENDED option is set so that white space is ignored):
2365 .sp
2366 \e( ( [^()]++ | (?R) )* \e)
2367 .sp
2368 First it matches an opening parenthesis. Then it matches any number of
2369 substrings which can either be a sequence of non-parentheses, or a recursive
2370 match of the pattern itself (that is, a correctly parenthesized substring).
2371 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2372 to avoid backtracking into sequences of non-parentheses.
2373 .P
2374 If this were part of a larger pattern, you would not want to recurse the entire
2375 pattern, so instead you could use this:
2376 .sp
2377 ( \e( ( [^()]++ | (?1) )* \e) )
2378 .sp
2379 We have put the pattern into parentheses, and caused the recursion to refer to
2380 them instead of the whole pattern.
2381 .P
2382 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2383 is made easier by the use of relative references. Instead of (?1) in the
2384 pattern above you can write (?-2) to refer to the second most recently opened
2385 parentheses preceding the recursion. In other words, a negative number counts
2386 capturing parentheses leftwards from the point at which it is encountered.
2387 .P
2388 It is also possible to refer to subsequently opened parentheses, by writing
2389 references such as (?+2). However, these cannot be recursive because the
2390 reference is not inside the parentheses that are referenced. They are always
2391 .\" HTML <a href="#subpatternsassubroutines">
2392 .\" </a>
2393 non-recursive subroutine
2394 .\"
2395 calls, as described in the next section.
2396 .P
2397 An alternative approach is to use named parentheses instead. The Perl syntax
2398 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2399 could rewrite the above example as follows:
2400 .sp
2401 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2402 .sp
2403 If there is more than one subpattern with the same name, the earliest one is
2404 used.
2405 .P
2406 This particular example pattern that we have been looking at contains nested
2407 unlimited repeats, and so the use of a possessive quantifier for matching
2408 strings of non-parentheses is important when applying the pattern to strings
2409 that do not match. For example, when this pattern is applied to
2410 .sp
2411 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2412 .sp
2413 it yields "no match" quickly. However, if a possessive quantifier is not used,
2414 the match runs for a very long time indeed because there are so many different
2415 ways the + and * repeats can carve up the subject, and all have to be tested
2416 before failure can be reported.
2417 .P
2418 At the end of a match, the values of capturing parentheses are those from
2419 the outermost level. If you want to obtain intermediate values, a callout
2420 function can be used (see below and the
2421 .\" HREF
2422 \fBpcrecallout\fP
2423 .\"
2424 documentation). If the pattern above is matched against
2425 .sp
2426 (ab(cd)ef)
2427 .sp
2428 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2429 the last value taken on at the top level. If a capturing subpattern is not
2430 matched at the top level, its final captured value is unset, even if it was
2431 (temporarily) set at a deeper level during the matching process.
2432 .P
2433 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2434 obtain extra memory to store data during a recursion, which it does by using
2435 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2436 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2437 .P
2438 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2439 Consider this pattern, which matches text in angle brackets, allowing for
2440 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2441 recursing), whereas any characters are permitted at the outer level.
2442 .sp
2443 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2444 .sp
2445 In this pattern, (?(R) is the start of a conditional subpattern, with two
2446 different alternatives for the recursive and non-recursive cases. The (?R) item
2447 is the actual recursive call.
2448 .
2449 .
2450 .\" HTML <a name="recursiondifference"></a>
2451 .SS "Differences in recursion processing between PCRE and Perl"
2452 .rs
2453 .sp
2454 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2455 (like Python, but unlike Perl), a recursive subpattern call is always treated
2456 as an atomic group. That is, once it has matched some of the subject string, it
2457 is never re-entered, even if it contains untried alternatives and there is a
2458 subsequent matching failure. This can be illustrated by the following pattern,
2459 which purports to match a palindromic string that contains an odd number of
2460 characters (for example, "a", "aba", "abcba", "abcdcba"):
2461 .sp
2462 ^(.|(.)(?1)\e2)$
2463 .sp
2464 The idea is that it either matches a single character, or two identical
2465 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2466 it does not if the pattern is longer than three characters. Consider the
2467 subject string "abcba":
2468 .P
2469 At the top level, the first character is matched, but as it is not at the end
2470 of the string, the first alternative fails; the second alternative is taken
2471 and the recursion kicks in. The recursive call to subpattern 1 successfully
2472 matches the next character ("b"). (Note that the beginning and end of line
2473 tests are not part of the recursion).
2474 .P
2475 Back at the top level, the next character ("c") is compared with what
2476 subpattern 2 matched, which was "a". This fails. Because the recursion is
2477 treated as an atomic group, there are now no backtracking points, and so the
2478 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2479 try the second alternative.) However, if the pattern is written with the
2480 alternatives in the other order, things are different:
2481 .sp
2482 ^((.)(?1)\e2|.)$
2483 .sp
2484 This time, the recursing alternative is tried first, and continues to recurse
2485 until it runs out of characters, at which point the recursion fails. But this
2486 time we do have another alternative to try at the higher level. That is the big
2487 difference: in the previous case the remaining alternative is at a deeper
2488 recursion level, which PCRE cannot use.
2489 .P
2490 To change the pattern so that it matches all palindromic strings, not just
2491 those with an odd number of characters, it is tempting to change the pattern to
2492 this:
2493 .sp
2494 ^((.)(?1)\e2|.?)$
2495 .sp
2496 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2497 deeper recursion has matched a single character, it cannot be entered again in
2498 order to match an empty string. The solution is to separate the two cases, and
2499 write out the odd and even cases as alternatives at the higher level:
2500 .sp
2501 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2502 .sp
2503 If you want to match typical palindromic phrases, the pattern has to ignore all
2504 non-word characters, which can be done like this:
2505 .sp
2506 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2507 .sp
2508 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2509 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2510 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2511 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2512 more) to match typical phrases, and Perl takes so long that you think it has
2513 gone into a loop.
2514 .P
2515 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2516 string does not start with a palindrome that is shorter than the entire string.
2517 For example, although "abcba" is correctly matched, if the subject is "ababa",
2518 PCRE finds the palindrome "aba" at the start, then fails at top level because
2519 the end of the string does not follow. Once again, it cannot jump back into the
2520 recursion to try other alternatives, so the entire match fails.
2521 .P
2522 The second way in which PCRE and Perl differ in their recursion processing is
2523 in the handling of captured values. In Perl, when a subpattern is called
2524 recursively or as a subpattern (see the next section), it has no access to any
2525 values that were captured outside the recursion, whereas in PCRE these values
2526 can be referenced. Consider this pattern:
2527 .sp
2528 ^(.)(\e1|a(?2))
2529 .sp
2530 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2531 then in the second group, when the back reference \e1 fails to match "b", the
2532 second alternative matches "a" and then recurses. In the recursion, \e1 does
2533 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2534 match because inside the recursive call \e1 cannot access the externally set
2535 value.
2536 .
2537 .
2538 .\" HTML <a name="subpatternsassubroutines"></a>
2540 .rs
2541 .sp
2542 If the syntax for a recursive subpattern call (either by number or by
2543 name) is used outside the parentheses to which it refers, it operates like a
2544 subroutine in a programming language. The called subpattern may be defined
2545 before or after the reference. A numbered reference can be absolute or
2546 relative, as in these examples:
2547 .sp
2548 (...(absolute)...)...(?2)...
2549 (...(relative)...)...(?-1)...
2550 (...(?+1)...(relative)...
2551 .sp
2552 An earlier example pointed out that the pattern
2553 .sp
2554 (sens|respons)e and \e1ibility
2555 .sp
2556 matches "sense and sensibility" and "response and responsibility", but not
2557 "sense and responsibility". If instead the pattern
2558 .sp
2559 (sens|respons)e and (?1)ibility
2560 .sp
2561 is used, it does match "sense and responsibility" as well as the other two
2562 strings. Another example is given in the discussion of DEFINE above.
2563 .P
2564 All subroutine calls, whether recursive or not, are always treated as atomic
2565 groups. That is, once a subroutine has matched some of the subject string, it
2566 is never re-entered, even if it contains untried alternatives and there is a
2567 subsequent matching failure. Any capturing parentheses that are set during the
2568 subroutine call revert to their previous values afterwards.
2569 .P
2570 Processing options such as case-independence are fixed when a subpattern is
2571 defined, so if it is used as a subroutine, such options cannot be changed for
2572 different calls. For example, consider this pattern:
2573 .sp
2574 (abc)(?i:(?-1))
2575 .sp
2576 It matches "abcabc". It does not match "abcABC" because the change of
2577 processing option does not affect the called subpattern.
2578 .
2579 .
2580 .\" HTML <a name="onigurumasubroutines"></a>
2582 .rs
2583 .sp
2584 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2585 a number enclosed either in angle brackets or single quotes, is an alternative
2586 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2587 are two of the examples used above, rewritten using this syntax:
2588 .sp
2589 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2590 (sens|respons)e and \eg'1'ibility
2591 .sp
2592 PCRE supports an extension to Oniguruma: if a number is preceded by a
2593 plus or a minus sign it is taken as a relative reference. For example:
2594 .sp
2595 (abc)(?i:\eg<-1>)
2596 .sp
2597 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2598 synonymous. The former is a back reference; the latter is a subroutine call.
2599 .
2600 .
2602 .rs
2603 .sp
2604 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2605 code to be obeyed in the middle of matching a regular expression. This makes it
2606 possible, amongst other things, to extract different substrings that match the
2607 same pair of parentheses when there is a repetition.
2608 .P
2609 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2610 code. The feature is called "callout". The caller of PCRE provides an external
2611 function by putting its entry point in the global variable \fIpcre_callout\fP
2612 (8-bit library) or \fIpcre[16|32]_callout\fP (16-bit or 32-bit library).
2613 By default, this variable contains NULL, which disables all calling out.
2614 .P
2615 Within a regular expression, (?C) indicates the points at which the external
2616 function is to be called. If you want to identify different callout points, you
2617 can put a number less than 256 after the letter C. The default value is zero.
2618 For example, this pattern has two callout points:
2619 .sp
2620 (?C1)abc(?C2)def
2621 .sp
2622 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2623 automatically installed before each item in the pattern. They are all numbered
2624 255.
2625 .P
2626 During matching, when PCRE reaches a callout point, the external function is
2627 called. It is provided with the number of the callout, the position in the
2628 pattern, and, optionally, one item of data originally supplied by the caller of
2629 the matching function. The callout function may cause matching to proceed, to
2630 backtrack, or to fail altogether. A complete description of the interface to
2631 the callout function is given in the
2632 .\" HREF
2633 \fBpcrecallout\fP
2634 .\"
2635 documentation.
2636 .
2637 .
2638 .\" HTML <a name="backtrackcontrol"></a>
2640 .rs
2641 .sp
2642 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2643 are described in the Perl documentation as "experimental and subject to change
2644 or removal in a future version of Perl". It goes on to say: "Their usage in
2645 production code should be noted to avoid problems during upgrades." The same
2646 remarks apply to the PCRE features described in this section.
2647 .P
2648 Since these verbs are specifically related to backtracking, most of them can be
2649 used only when the pattern is to be matched using one of the traditional
2650 matching functions, which use a backtracking algorithm. With the exception of
2651 (*FAIL), which behaves like a failing negative assertion, they cause an error
2652 if encountered by a DFA matching function.
2653 .P
2654 If any of these verbs are used in an assertion or in a subpattern that is
2655 called as a subroutine (whether or not recursively), their effect is confined
2656 to that subpattern; it does not extend to the surrounding pattern, with one
2657 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2658 a successful positive assertion \fIis\fP passed back when a match succeeds
2659 (compare capturing parentheses in assertions). Note that such subpatterns are
2660 processed as anchored at the point where they are tested. Note also that Perl's
2661 treatment of subroutines and assertions is different in some cases.
2662 .P
2663 The new verbs make use of what was previously invalid syntax: an opening
2664 parenthesis followed by an asterisk. They are generally of the form
2665 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2666 depending on whether or not an argument is present. A name is any sequence of
2667 characters that does not include a closing parenthesis. The maximum length of
2668 name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit library.
2669 If the name is empty, that is, if the closing parenthesis immediately follows
2670 the colon, the effect is as if the colon were not there. Any number of these
2671 verbs may occur in a pattern.
2672 .
2673 .
2674 .\" HTML <a name="nooptimize"></a>
2675 .SS "Optimizations that affect backtracking verbs"
2676 .rs
2677 .sp
2678 PCRE contains some optimizations that are used to speed up matching by running
2679 some checks at the start of each match attempt. For example, it may know the
2680 minimum length of matching subject, or that a particular character must be
2681 present. When one of these optimizations suppresses the running of a match, any
2682 included backtracking verbs will not, of course, be processed. You can suppress
2683 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2684 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2685 pattern with (*NO_START_OPT). There is more discussion of this option in the
2686 section entitled
2687 .\" HTML <a href="pcreapi.html#execoptions">
2688 .\" </a>
2689 "Option bits for \fBpcre_exec()\fP"
2690 .\"
2691 in the
2692 .\" HREF
2693 \fBpcreapi\fP
2694 .\"
2695 documentation.
2696 .P
2697 Experiments with Perl suggest that it too has similar optimizations, sometimes
2698 leading to anomalous results.
2699 .
2700 .
2701 .SS "Verbs that act immediately"
2702 .rs
2703 .sp
2704 The following verbs act as soon as they are encountered. They may not be
2705 followed by a name.
2706 .sp
2707 (*ACCEPT)
2708 .sp
2709 This verb causes the match to end successfully, skipping the remainder of the
2710 pattern. However, when it is inside a subpattern that is called as a
2711 subroutine, only that subpattern is ended successfully. Matching then continues
2712 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2713 far is captured. For example:
2714 .sp
2715 A((?:A|B(*ACCEPT)|C)D)
2716 .sp
2717 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2718 the outer parentheses.
2719 .sp
2720 (*FAIL) or (*F)
2721 .sp
2722 This verb causes a matching failure, forcing backtracking to occur. It is
2723 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2724 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2725 Perl features that are not present in PCRE. The nearest equivalent is the
2726 callout feature, as for example in this pattern:
2727 .sp
2728 a+(?C)(*FAIL)
2729 .sp
2730 A match with the string "aaaa" always fails, but the callout is taken before
2731 each backtrack happens (in this example, 10 times).
2732 .
2733 .
2734 .SS "Recording which path was taken"
2735 .rs
2736 .sp
2737 There is one verb whose main purpose is to track how a match was arrived at,
2738 though it also has a secondary use in conjunction with advancing the match
2739 starting point (see (*SKIP) below).
2740 .sp
2741 (*MARK:NAME) or (*:NAME)
2742 .sp
2743 A name is always required with this verb. There may be as many instances of
2744 (*MARK) as you like in a pattern, and their names do not have to be unique.
2745 .P
2746 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2747 path is passed back to the caller as described in the section entitled
2748 .\" HTML <a href="pcreapi.html#extradata">
2749 .\" </a>
2750 "Extra data for \fBpcre_exec()\fP"
2751 .\"
2752 in the
2753 .\" HREF
2754 \fBpcreapi\fP
2755 .\"
2756 documentation. Here is an example of \fBpcretest\fP output, where the /K
2757 modifier requests the retrieval and outputting of (*MARK) data:
2758 .sp
2759 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2760 data> XY
2761 0: XY
2762 MK: A
2763 XZ
2764 0: XZ
2765 MK: B
2766 .sp
2767 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2768 indicates which of the two alternatives matched. This is a more efficient way
2769 of obtaining this information than putting each alternative in its own
2770 capturing parentheses.
2771 .P
2772 If (*MARK) is encountered in a positive assertion, its name is recorded and
2773 passed back if it is the last-encountered. This does not happen for negative
2774 assertions.
2775 .P
2776 After a partial match or a failed match, the name of the last encountered
2777 (*MARK) in the entire match process is returned. For example:
2778 .sp
2779 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2780 data> XP
2781 No match, mark = B
2782 .sp
2783 Note that in this unanchored example the mark is retained from the match
2784 attempt that started at the letter "X" in the subject. Subsequent match
2785 attempts starting at "P" and then with an empty string do not get as far as the
2786 (*MARK) item, but nevertheless do not reset it.
2787 .P
2788 If you are interested in (*MARK) values after failed matches, you should
2789 probably set the PCRE_NO_START_OPTIMIZE option
2790 .\" HTML <a href="#nooptimize">
2791 .\" </a>
2792 (see above)
2793 .\"
2794 to ensure that the match is always attempted.
2795 .
2796 .
2797 .SS "Verbs that act after backtracking"
2798 .rs
2799 .sp
2800 The following verbs do nothing when they are encountered. Matching continues
2801 with what follows, but if there is no subsequent match, causing a backtrack to
2802 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2803 the verb. However, when one of these verbs appears inside an atomic group, its
2804 effect is confined to that group, because once the group has been matched,
2805 there is never any backtracking into it. In this situation, backtracking can
2806 "jump back" to the left of the entire atomic group. (Remember also, as stated
2807 above, that this localization also applies in subroutine calls and assertions.)
2808 .P
2809 These verbs differ in exactly what kind of failure occurs when backtracking
2810 reaches them.
2811 .sp
2812 (*COMMIT)
2813 .sp
2814 This verb, which may not be followed by a name, causes the whole match to fail
2815 outright if the rest of the pattern does not match. Even if the pattern is
2816 unanchored, no further attempts to find a match by advancing the starting point
2817 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2818 finding a match at the current starting point, or not at all. For example:
2819 .sp
2820 a+(*COMMIT)b
2821 .sp
2822 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2823 dynamic anchor, or "I've started, so I must finish." The name of the most
2824 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2825 match failure.
2826 .P
2827 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2828 unless PCRE's start-of-match optimizations are turned off, as shown in this
2829 \fBpcretest\fP example:
2830 .sp
2831 re> /(*COMMIT)abc/
2832 data> xyzabc
2833 0: abc
2834 xyzabc\eY
2835 No match
2836 .sp
2837 PCRE knows that any match must start with "a", so the optimization skips along
2838 the subject to "a" before running the first match attempt, which succeeds. When
2839 the optimization is disabled by the \eY escape in the second subject, the match
2840 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2841 starting points.
2842 .sp
2843 (*PRUNE) or (*PRUNE:NAME)
2844 .sp
2845 This verb causes the match to fail at the current starting position in the
2846 subject if the rest of the pattern does not match. If the pattern is
2847 unanchored, the normal "bumpalong" advance to the next starting character then
2848 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2849 reached, or when matching to the right of (*PRUNE), but if there is no match to
2850 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2851 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2852 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2853 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2854 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2855 .sp
2856 (*SKIP)
2857 .sp
2858 This verb, when given without a name, is like (*PRUNE), except that if the
2859 pattern is unanchored, the "bumpalong" advance is not to the next character,
2860 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2861 signifies that whatever text was matched leading up to it cannot be part of a
2862 successful match. Consider:
2863 .sp
2864 a+(*SKIP)b
2865 .sp
2866 If the subject is "aaaac...", after the first match attempt fails (starting at
2867 the first character in the string), the starting point skips on to start the
2868 next attempt at "c". Note that a possessive quantifer does not have the same
2869 effect as this example; although it would suppress backtracking during the
2870 first match attempt, the second attempt would start at the second character
2871 instead of skipping on to "c".
2872 .sp
2873 (*SKIP:NAME)
2874 .sp
2875 When (*SKIP) has an associated name, its behaviour is modified. If the
2876 following pattern fails to match, the previous path through the pattern is
2877 searched for the most recent (*MARK) that has the same name. If one is found,
2878 the "bumpalong" advance is to the subject position that corresponds to that
2879 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2880 matching name is found, the (*SKIP) is ignored.
2881 .sp
2882 (*THEN) or (*THEN:NAME)
2883 .sp
2884 This verb causes a skip to the next innermost alternative if the rest of the
2885 pattern does not match. That is, it cancels pending backtracking, but only
2886 within the current alternative. Its name comes from the observation that it can
2887 be used for a pattern-based if-then-else block:
2888 .sp
2889 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2890 .sp
2891 If the COND1 pattern matches, FOO is tried (and possibly further items after
2892 the end of the group if FOO succeeds); on failure, the matcher skips to the
2893 second alternative and tries COND2, without backtracking into COND1. The
2894 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2895 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2896 .P
2897 Note that a subpattern that does not contain a | character is just a part of
2898 the enclosing alternative; it is not a nested alternation with only one
2899 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2900 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2901 pattern fragments that do not contain any | characters at this level:
2902 .sp
2903 A (B(*THEN)C) | D
2904 .sp
2905 If A and B are matched, but there is a failure in C, matching does not
2906 backtrack into A; instead it moves to the next alternative, that is, D.
2907 However, if the subpattern containing (*THEN) is given an alternative, it
2908 behaves differently:
2909 .sp
2910 A (B(*THEN)C | (*FAIL)) | D
2911 .sp
2912 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2913 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2914 because there are no more alternatives to try. In this case, matching does now
2915 backtrack into A.
2916 .P
2917 Note also that a conditional subpattern is not considered as having two
2918 alternatives, because only one is ever used. In other words, the | character in
2919 a conditional subpattern has a different meaning. Ignoring white space,
2920 consider:
2921 .sp
2922 ^.*? (?(?=a) a | b(*THEN)c )
2923 .sp
2924 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2925 it initially matches zero characters. The condition (?=a) then fails, the
2926 character "b" is matched, but "c" is not. At this point, matching does not
2927 backtrack to .*? as might perhaps be expected from the presence of the |
2928 character. The conditional subpattern is part of the single alternative that
2929 comprises the whole pattern, and so the match fails. (If there was a backtrack
2930 into .*?, allowing it to match "b", the match would succeed.)
2931 .P
2932 The verbs just described provide four different "strengths" of control when
2933 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2934 next alternative. (*PRUNE) comes next, failing the match at the current
2935 starting position, but allowing an advance to the next character (for an
2936 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2937 than one character. (*COMMIT) is the strongest, causing the entire match to
2938 fail.
2939 .P
2940 If more than one such verb is present in a pattern, the "strongest" one wins.
2941 For example, consider this pattern, where A, B, etc. are complex pattern
2942 fragments:
2943 .sp
2944 (A(*COMMIT)B(*THEN)C|D)
2945 .sp
2946 Once A has matched, PCRE is committed to this match, at the current starting
2947 position. If subsequently B matches, but C does not, the normal (*THEN) action
2948 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2949 overrides.
2950 .
2951 .
2952 .SH "SEE ALSO"
2953 .rs
2954 .sp
2955 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2956 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP, \fBpcre32(3)\fP.
2957 .
2958 .
2960 .rs
2961 .sp
2962 .nf
2963 Philip Hazel
2964 University Computing Service
2965 Cambridge CB2 3QH, England.
2966 .fi
2967 .
2968 .
2970 .rs
2971 .sp
2972 .nf
2973 Last updated: 10 September 2012
2974 Copyright (c) 1997-2012 University of Cambridge.
2975 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5