ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1404 - (show annotations)
Tue Nov 19 15:36:57 2013 UTC (7 years, 2 months ago) by ph10
File MIME type: text/html
File size: 137808 byte(s)
Source tidies for 8.34-RC1.
1 <html>
2 <head>
3 <title>pcrepattern specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 <h1>pcrepattern man page</h1>
7 <p>
8 Return to the <a href="index.html">PCRE index page</a>.
9 </p>
10 <p>
11 This page is part of the PCRE HTML documentation. It was generated automatically
12 from the original man page. If there is any nonsense in it, please consult the
13 man page, in case the conversion went wrong.
14 <br>
15 <ul>
16 <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 <li><a name="TOC2" href="#SEC2">SPECIAL START-OF-PATTERN ITEMS</a>
18 <li><a name="TOC3" href="#SEC3">EBCDIC CHARACTER CODES</a>
19 <li><a name="TOC4" href="#SEC4">CHARACTERS AND METACHARACTERS</a>
20 <li><a name="TOC5" href="#SEC5">BACKSLASH</a>
21 <li><a name="TOC6" href="#SEC6">CIRCUMFLEX AND DOLLAR</a>
22 <li><a name="TOC7" href="#SEC7">FULL STOP (PERIOD, DOT) AND \N</a>
23 <li><a name="TOC8" href="#SEC8">MATCHING A SINGLE DATA UNIT</a>
25 <li><a name="TOC10" href="#SEC10">POSIX CHARACTER CLASSES</a>
26 <li><a name="TOC11" href="#SEC11">VERTICAL BAR</a>
27 <li><a name="TOC12" href="#SEC12">INTERNAL OPTION SETTING</a>
28 <li><a name="TOC13" href="#SEC13">SUBPATTERNS</a>
29 <li><a name="TOC14" href="#SEC14">DUPLICATE SUBPATTERN NUMBERS</a>
30 <li><a name="TOC15" href="#SEC15">NAMED SUBPATTERNS</a>
31 <li><a name="TOC16" href="#SEC16">REPETITION</a>
33 <li><a name="TOC18" href="#SEC18">BACK REFERENCES</a>
34 <li><a name="TOC19" href="#SEC19">ASSERTIONS</a>
35 <li><a name="TOC20" href="#SEC20">CONDITIONAL SUBPATTERNS</a>
36 <li><a name="TOC21" href="#SEC21">COMMENTS</a>
37 <li><a name="TOC22" href="#SEC22">RECURSIVE PATTERNS</a>
38 <li><a name="TOC23" href="#SEC23">SUBPATTERNS AS SUBROUTINES</a>
39 <li><a name="TOC24" href="#SEC24">ONIGURUMA SUBROUTINE SYNTAX</a>
40 <li><a name="TOC25" href="#SEC25">CALLOUTS</a>
41 <li><a name="TOC26" href="#SEC26">BACKTRACKING CONTROL</a>
42 <li><a name="TOC27" href="#SEC27">SEE ALSO</a>
43 <li><a name="TOC28" href="#SEC28">AUTHOR</a>
44 <li><a name="TOC29" href="#SEC29">REVISION</a>
45 </ul>
46 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
47 <P>
48 The syntax and semantics of the regular expressions that are supported by PCRE
49 are described in detail below. There is a quick-reference syntax summary in the
50 <a href="pcresyntax.html"><b>pcresyntax</b></a>
51 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
52 also supports some alternative regular expression syntax (which does not
53 conflict with the Perl syntax) in order to provide some compatibility with
54 regular expressions in Python, .NET, and Oniguruma.
55 </P>
56 <P>
57 Perl's regular expressions are described in its own documentation, and
58 regular expressions in general are covered in a number of books, some of which
59 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
60 published by O'Reilly, covers regular expressions in great detail. This
61 description of PCRE's regular expressions is intended as reference material.
62 </P>
63 <P>
64 This document discusses the patterns that are supported by PCRE when one its
65 main matching functions, <b>pcre_exec()</b> (8-bit) or <b>pcre[16|32]_exec()</b>
66 (16- or 32-bit), is used. PCRE also has alternative matching functions,
67 <b>pcre_dfa_exec()</b> and <b>pcre[16|32_dfa_exec()</b>, which match using a
68 different algorithm that is not Perl-compatible. Some of the features discussed
69 below are not available when DFA matching is used. The advantages and
70 disadvantages of the alternative functions, and how they differ from the normal
71 functions, are discussed in the
72 <a href="pcrematching.html"><b>pcrematching</b></a>
73 page.
74 </P>
75 <br><a name="SEC2" href="#TOC1">SPECIAL START-OF-PATTERN ITEMS</a><br>
76 <P>
77 A number of options that can be passed to <b>pcre_compile()</b> can also be set
78 by special items at the start of a pattern. These are not Perl-compatible, but
79 are provided to make these options accessible to pattern writers who are not
80 able to change the program that processes the pattern. Any number of these
81 items may appear, but they must all be together right at the start of the
82 pattern string, and the letters must be in upper case.
83 </P>
84 <br><b>
85 UTF support
86 </b><br>
87 <P>
88 The original operation of PCRE was on strings of one-byte characters. However,
89 there is now also support for UTF-8 strings in the original library, an
90 extra library that supports 16-bit and UTF-16 character strings, and a
91 third library that supports 32-bit and UTF-32 character strings. To use these
92 features, PCRE must be built to include appropriate support. When using UTF
93 strings you must either call the compiling function with the PCRE_UTF8,
94 PCRE_UTF16, or PCRE_UTF32 option, or the pattern must start with one of
95 these special sequences:
96 <pre>
97 (*UTF8)
98 (*UTF16)
99 (*UTF32)
100 (*UTF)
101 </pre>
102 (*UTF) is a generic sequence that can be used with any of the libraries.
103 Starting a pattern with such a sequence is equivalent to setting the relevant
104 option. How setting a UTF mode affects pattern matching is mentioned in several
105 places below. There is also a summary of features in the
106 <a href="pcreunicode.html"><b>pcreunicode</b></a>
107 page.
108 </P>
109 <P>
110 Some applications that allow their users to supply patterns may wish to
111 restrict them to non-UTF data for security reasons. If the PCRE_NEVER_UTF
112 option is set at compile time, (*UTF) etc. are not allowed, and their
113 appearance causes an error.
114 </P>
115 <br><b>
116 Unicode property support
117 </b><br>
118 <P>
119 Another special sequence that may appear at the start of a pattern is (*UCP).
120 This has the same effect as setting the PCRE_UCP option: it causes sequences
121 such as \d and \w to use Unicode properties to determine character types,
122 instead of recognizing only characters with codes less than 128 via a lookup
123 table.
124 </P>
125 <br><b>
126 Disabling auto-possessification
127 </b><br>
128 <P>
129 If a pattern starts with (*NO_AUTO_POSSESS), it has the same effect as setting
130 the PCRE_NO_AUTO_POSSESS option at compile time. This stops PCRE from making
131 quantifiers possessive when what follows cannot match the repeated item. For
132 example, by default a+b is treated as a++b. For more details, see the
133 <a href="pcreapi.html"><b>pcreapi</b></a>
134 documentation.
135 </P>
136 <br><b>
137 Disabling start-up optimizations
138 </b><br>
139 <P>
140 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
141 PCRE_NO_START_OPTIMIZE option either at compile or matching time. This disables
142 several optimizations for quickly reaching "no match" results. For more
143 details, see the
144 <a href="pcreapi.html"><b>pcreapi</b></a>
145 documentation.
146 <a name="newlines"></a></P>
147 <br><b>
148 Newline conventions
149 </b><br>
150 <P>
151 PCRE supports five different conventions for indicating line breaks in
152 strings: a single CR (carriage return) character, a single LF (linefeed)
153 character, the two-character sequence CRLF, any of the three preceding, or any
154 Unicode newline sequence. The
155 <a href="pcreapi.html"><b>pcreapi</b></a>
156 page has
157 <a href="pcreapi.html#newlines">further discussion</a>
158 about newlines, and shows how to set the newline convention in the
159 <i>options</i> arguments for the compiling and matching functions.
160 </P>
161 <P>
162 It is also possible to specify a newline convention by starting a pattern
163 string with one of the following five sequences:
164 <pre>
165 (*CR) carriage return
166 (*LF) linefeed
167 (*CRLF) carriage return, followed by linefeed
168 (*ANYCRLF) any of the three above
169 (*ANY) all Unicode newline sequences
170 </pre>
171 These override the default and the options given to the compiling function. For
172 example, on a Unix system where LF is the default newline sequence, the pattern
173 <pre>
174 (*CR)a.b
175 </pre>
176 changes the convention to CR. That pattern matches "a\nb" because LF is no
177 longer a newline. If more than one of these settings is present, the last one
178 is used.
179 </P>
180 <P>
181 The newline convention affects where the circumflex and dollar assertions are
182 true. It also affects the interpretation of the dot metacharacter when
183 PCRE_DOTALL is not set, and the behaviour of \N. However, it does not affect
184 what the \R escape sequence matches. By default, this is any Unicode newline
185 sequence, for Perl compatibility. However, this can be changed; see the
186 description of \R in the section entitled
187 <a href="#newlineseq">"Newline sequences"</a>
188 below. A change of \R setting can be combined with a change of newline
189 convention.
190 </P>
191 <br><b>
192 Setting match and recursion limits
193 </b><br>
194 <P>
195 The caller of <b>pcre_exec()</b> can set a limit on the number of times the
196 internal <b>match()</b> function is called and on the maximum depth of
197 recursive calls. These facilities are provided to catch runaway matches that
198 are provoked by patterns with huge matching trees (a typical example is a
199 pattern with nested unlimited repeats) and to avoid running out of system stack
200 by too much recursion. When one of these limits is reached, <b>pcre_exec()</b>
201 gives an error return. The limits can also be set by items at the start of the
202 pattern of the form
203 <pre>
204 (*LIMIT_MATCH=d)
206 </pre>
207 where d is any number of decimal digits. However, the value of the setting must
208 be less than the value set (or defaulted) by the caller of <b>pcre_exec()</b>
209 for it to have any effect. In other words, the pattern writer can lower the
210 limits set by the programmer, but not raise them. If there is more than one
211 setting of one of these limits, the lower value is used.
212 </P>
213 <br><a name="SEC3" href="#TOC1">EBCDIC CHARACTER CODES</a><br>
214 <P>
215 PCRE can be compiled to run in an environment that uses EBCDIC as its character
216 code rather than ASCII or Unicode (typically a mainframe system). In the
217 sections below, character code values are ASCII or Unicode; in an EBCDIC
218 environment these characters may have different code values, and there are no
219 code points greater than 255.
220 </P>
221 <br><a name="SEC4" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
222 <P>
223 A regular expression is a pattern that is matched against a subject string from
224 left to right. Most characters stand for themselves in a pattern, and match the
225 corresponding characters in the subject. As a trivial example, the pattern
226 <pre>
227 The quick brown fox
228 </pre>
229 matches a portion of a subject string that is identical to itself. When
230 caseless matching is specified (the PCRE_CASELESS option), letters are matched
231 independently of case. In a UTF mode, PCRE always understands the concept of
232 case for characters whose values are less than 128, so caseless matching is
233 always possible. For characters with higher values, the concept of case is
234 supported if PCRE is compiled with Unicode property support, but not otherwise.
235 If you want to use caseless matching for characters 128 and above, you must
236 ensure that PCRE is compiled with Unicode property support as well as with
237 UTF support.
238 </P>
239 <P>
240 The power of regular expressions comes from the ability to include alternatives
241 and repetitions in the pattern. These are encoded in the pattern by the use of
242 <i>metacharacters</i>, which do not stand for themselves but instead are
243 interpreted in some special way.
244 </P>
245 <P>
246 There are two different sets of metacharacters: those that are recognized
247 anywhere in the pattern except within square brackets, and those that are
248 recognized within square brackets. Outside square brackets, the metacharacters
249 are as follows:
250 <pre>
251 \ general escape character with several uses
252 ^ assert start of string (or line, in multiline mode)
253 $ assert end of string (or line, in multiline mode)
254 . match any character except newline (by default)
255 [ start character class definition
256 | start of alternative branch
257 ( start subpattern
258 ) end subpattern
259 ? extends the meaning of (
260 also 0 or 1 quantifier
261 also quantifier minimizer
262 * 0 or more quantifier
263 + 1 or more quantifier
264 also "possessive quantifier"
265 { start min/max quantifier
266 </pre>
267 Part of a pattern that is in square brackets is called a "character class". In
268 a character class the only metacharacters are:
269 <pre>
270 \ general escape character
271 ^ negate the class, but only if the first character
272 - indicates character range
273 [ POSIX character class (only if followed by POSIX syntax)
274 ] terminates the character class
275 </pre>
276 The following sections describe the use of each of the metacharacters.
277 </P>
278 <br><a name="SEC5" href="#TOC1">BACKSLASH</a><br>
279 <P>
280 The backslash character has several uses. Firstly, if it is followed by a
281 character that is not a number or a letter, it takes away any special meaning
282 that character may have. This use of backslash as an escape character applies
283 both inside and outside character classes.
284 </P>
285 <P>
286 For example, if you want to match a * character, you write \* in the pattern.
287 This escaping action applies whether or not the following character would
288 otherwise be interpreted as a metacharacter, so it is always safe to precede a
289 non-alphanumeric with backslash to specify that it stands for itself. In
290 particular, if you want to match a backslash, you write \\.
291 </P>
292 <P>
293 In a UTF mode, only ASCII numbers and letters have any special meaning after a
294 backslash. All other characters (in particular, those whose codepoints are
295 greater than 127) are treated as literals.
296 </P>
297 <P>
298 If a pattern is compiled with the PCRE_EXTENDED option, most white space in the
299 pattern (other than in a character class), and characters between a # outside a
300 character class and the next newline, inclusive, are ignored. An escaping
301 backslash can be used to include a white space or # character as part of the
302 pattern.
303 </P>
304 <P>
305 If you want to remove the special meaning from a sequence of characters, you
306 can do so by putting them between \Q and \E. This is different from Perl in
307 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
308 Perl, $ and @ cause variable interpolation. Note the following examples:
309 <pre>
310 Pattern PCRE matches Perl matches
312 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
313 \Qabc\$xyz\E abc\$xyz abc\$xyz
314 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
315 </pre>
316 The \Q...\E sequence is recognized both inside and outside character classes.
317 An isolated \E that is not preceded by \Q is ignored. If \Q is not followed
318 by \E later in the pattern, the literal interpretation continues to the end of
319 the pattern (that is, \E is assumed at the end). If the isolated \Q is inside
320 a character class, this causes an error, because the character class is not
321 terminated.
322 <a name="digitsafterbackslash"></a></P>
323 <br><b>
324 Non-printing characters
325 </b><br>
326 <P>
327 A second use of backslash provides a way of encoding non-printing characters
328 in patterns in a visible manner. There is no restriction on the appearance of
329 non-printing characters, apart from the binary zero that terminates a pattern,
330 but when a pattern is being prepared by text editing, it is often easier to use
331 one of the following escape sequences than the binary character it represents:
332 <pre>
333 \a alarm, that is, the BEL character (hex 07)
334 \cx "control-x", where x is any ASCII character
335 \e escape (hex 1B)
336 \f form feed (hex 0C)
337 \n linefeed (hex 0A)
338 \r carriage return (hex 0D)
339 \t tab (hex 09)
340 \0dd character with octal code 0dd
341 \ddd character with octal code ddd, or back reference
342 \o{ddd..} character with octal code ddd..
343 \xhh character with hex code hh
344 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
345 \uhhhh character with hex code hhhh (JavaScript mode only)
346 </pre>
347 The precise effect of \cx on ASCII characters is as follows: if x is a lower
348 case letter, it is converted to upper case. Then bit 6 of the character (hex
349 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A (A is 41, Z is 5A),
350 but \c{ becomes hex 3B ({ is 7B), and \c; becomes hex 7B (; is 3B). If the
351 data item (byte or 16-bit value) following \c has a value greater than 127, a
352 compile-time error occurs. This locks out non-ASCII characters in all modes.
353 </P>
354 <P>
355 The \c facility was designed for use with ASCII characters, but with the
356 extension to Unicode it is even less useful than it once was. It is, however,
357 recognized when PCRE is compiled in EBCDIC mode, where data items are always
358 bytes. In this mode, all values are valid after \c. If the next character is a
359 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
360 byte are inverted. Thus \cA becomes hex 01, as in ASCII (A is C1), but because
361 the EBCDIC letters are disjoint, \cZ becomes hex 29 (Z is E9), and other
362 characters also generate different values.
363 </P>
364 <P>
365 After \0 up to two further octal digits are read. If there are fewer than two
366 digits, just those that are present are used. Thus the sequence \0\x\07
367 specifies two binary zeros followed by a BEL character (code value 7). Make
368 sure you supply two digits after the initial zero if the pattern character that
369 follows is itself an octal digit.
370 </P>
371 <P>
372 The escape \o must be followed by a sequence of octal digits, enclosed in
373 braces. An error occurs if this is not the case. This escape is a recent
374 addition to Perl; it provides way of specifying character code points as octal
375 numbers greater than 0777, and it also allows octal numbers and back references
376 to be unambiguously specified.
377 </P>
378 <P>
379 For greater clarity and unambiguity, it is best to avoid following \ by a
380 digit greater than zero. Instead, use \o{} or \x{} to specify character
381 numbers, and \g{} to specify back references. The following paragraphs
382 describe the old, ambiguous syntax.
383 </P>
384 <P>
385 The handling of a backslash followed by a digit other than 0 is complicated,
386 and Perl has changed in recent releases, causing PCRE also to change. Outside a
387 character class, PCRE reads the digit and any following digits as a decimal
388 number. If the number is less than 8, or if there have been at least that many
389 previous capturing left parentheses in the expression, the entire sequence is
390 taken as a <i>back reference</i>. A description of how this works is given
391 <a href="#backreferences">later,</a>
392 following the discussion of
393 <a href="#subpattern">parenthesized subpatterns.</a>
394 </P>
395 <P>
396 Inside a character class, or if the decimal number following \ is greater than
397 7 and there have not been that many capturing subpatterns, PCRE handles \8 and
398 \9 as the literal characters "8" and "9", and otherwise re-reads up to three
399 octal digits following the backslash, using them to generate a data character.
400 Any subsequent digits stand for themselves. For example:
401 <pre>
402 \040 is another way of writing an ASCII space
403 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
404 \7 is always a back reference
405 \11 might be a back reference, or another way of writing a tab
406 \011 is always a tab
407 \0113 is a tab followed by the character "3"
408 \113 might be a back reference, otherwise the character with octal code 113
409 \377 might be a back reference, otherwise the value 255 (decimal)
410 \81 is either a back reference, or the two characters "8" and "1"
411 </pre>
412 Note that octal values of 100 or greater that are specified using this syntax
413 must not be introduced by a leading zero, because no more than three octal
414 digits are ever read.
415 </P>
416 <P>
417 By default, after \x that is not followed by {, from zero to two hexadecimal
418 digits are read (letters can be in upper or lower case). Any number of
419 hexadecimal digits may appear between \x{ and }. If a character other than
420 a hexadecimal digit appears between \x{ and }, or if there is no terminating
421 }, an error occurs.
422 </P>
423 <P>
424 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is
425 as just described only when it is followed by two hexadecimal digits.
426 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
427 code points greater than 256 is provided by \u, which must be followed by
428 four hexadecimal digits; otherwise it matches a literal "u" character.
429 </P>
430 <P>
431 Characters whose value is less than 256 can be defined by either of the two
432 syntaxes for \x (or by \u in JavaScript mode). There is no difference in the
433 way they are handled. For example, \xdc is exactly the same as \x{dc} (or
434 \u00dc in JavaScript mode).
435 </P>
436 <br><b>
437 Constraints on character values
438 </b><br>
439 <P>
440 Characters that are specified using octal or hexadecimal numbers are
441 limited to certain values, as follows:
442 <pre>
443 8-bit non-UTF mode less than 0x100
444 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
445 16-bit non-UTF mode less than 0x10000
446 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
447 32-bit non-UTF mode less than 0x100000000
448 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
449 </pre>
450 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
451 "surrogate" codepoints), and 0xffef.
452 </P>
453 <br><b>
454 Escape sequences in character classes
455 </b><br>
456 <P>
457 All the sequences that define a single character value can be used both inside
458 and outside character classes. In addition, inside a character class, \b is
459 interpreted as the backspace character (hex 08).
460 </P>
461 <P>
462 \N is not allowed in a character class. \B, \R, and \X are not special
463 inside a character class. Like other unrecognized escape sequences, they are
464 treated as the literal characters "B", "R", and "X" by default, but cause an
465 error if the PCRE_EXTRA option is set. Outside a character class, these
466 sequences have different meanings.
467 </P>
468 <br><b>
469 Unsupported escape sequences
470 </b><br>
471 <P>
472 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
473 handler and used to modify the case of following characters. By default, PCRE
474 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
475 option is set, \U matches a "U" character, and \u can be used to define a
476 character by code point, as described in the previous section.
477 </P>
478 <br><b>
479 Absolute and relative back references
480 </b><br>
481 <P>
482 The sequence \g followed by an unsigned or a negative number, optionally
483 enclosed in braces, is an absolute or relative back reference. A named back
484 reference can be coded as \g{name}. Back references are discussed
485 <a href="#backreferences">later,</a>
486 following the discussion of
487 <a href="#subpattern">parenthesized subpatterns.</a>
488 </P>
489 <br><b>
490 Absolute and relative subroutine calls
491 </b><br>
492 <P>
493 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
494 a number enclosed either in angle brackets or single quotes, is an alternative
495 syntax for referencing a subpattern as a "subroutine". Details are discussed
496 <a href="#onigurumasubroutines">later.</a>
497 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
498 synonymous. The former is a back reference; the latter is a
499 <a href="#subpatternsassubroutines">subroutine</a>
500 call.
501 <a name="genericchartypes"></a></P>
502 <br><b>
503 Generic character types
504 </b><br>
505 <P>
506 Another use of backslash is for specifying generic character types:
507 <pre>
508 \d any decimal digit
509 \D any character that is not a decimal digit
510 \h any horizontal white space character
511 \H any character that is not a horizontal white space character
512 \s any white space character
513 \S any character that is not a white space character
514 \v any vertical white space character
515 \V any character that is not a vertical white space character
516 \w any "word" character
517 \W any "non-word" character
518 </pre>
519 There is also the single sequence \N, which matches a non-newline character.
520 This is the same as
521 <a href="#fullstopdot">the "." metacharacter</a>
522 when PCRE_DOTALL is not set. Perl also uses \N to match characters by name;
523 PCRE does not support this.
524 </P>
525 <P>
526 Each pair of lower and upper case escape sequences partitions the complete set
527 of characters into two disjoint sets. Any given character matches one, and only
528 one, of each pair. The sequences can appear both inside and outside character
529 classes. They each match one character of the appropriate type. If the current
530 matching point is at the end of the subject string, all of them fail, because
531 there is no character to match.
532 </P>
533 <P>
534 For compatibility with Perl, \s did not used to match the VT character (code
535 11), which made it different from the the POSIX "space" class. However, Perl
536 added VT at release 5.18, and PCRE followed suit at release 8.34. The default
537 \s characters are now HT (9), LF (10), VT (11), FF (12), CR (13), and space
538 (32), which are defined as white space in the "C" locale. This list may vary if
539 locale-specific matching is taking place; in particular, in some locales the
540 "non-breaking space" character (\xA0) is recognized as white space.
541 </P>
542 <P>
543 A "word" character is an underscore or any character that is a letter or digit.
544 By default, the definition of letters and digits is controlled by PCRE's
545 low-valued character tables, and may vary if locale-specific matching is taking
546 place (see
547 <a href="pcreapi.html#localesupport">"Locale support"</a>
548 in the
549 <a href="pcreapi.html"><b>pcreapi</b></a>
550 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
551 or "french" in Windows, some character codes greater than 127 are used for
552 accented letters, and these are then matched by \w. The use of locales with
553 Unicode is discouraged.
554 </P>
555 <P>
556 By default, characters whose code points are greater than 127 never match \d,
557 \s, or \w, and always match \D, \S, and \W, although this may vary for
558 characters in the range 128-255 when locale-specific matching is happening.
559 These escape sequences retain their original meanings from before Unicode
560 support was available, mainly for efficiency reasons. If PCRE is compiled with
561 Unicode property support, and the PCRE_UCP option is set, the behaviour is
562 changed so that Unicode properties are used to determine character types, as
563 follows:
564 <pre>
565 \d any character that matches \p{Nd} (decimal digit)
566 \s any character that matches \p{Z} or \h or \v
567 \w any character that matches \p{L} or \p{N}, plus underscore
568 </pre>
569 The upper case escapes match the inverse sets of characters. Note that \d
570 matches only decimal digits, whereas \w matches any Unicode digit, as well as
571 any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and
572 \B because they are defined in terms of \w and \W. Matching these sequences
573 is noticeably slower when PCRE_UCP is set.
574 </P>
575 <P>
576 The sequences \h, \H, \v, and \V are features that were added to Perl at
577 release 5.10. In contrast to the other sequences, which match only ASCII
578 characters by default, these always match certain high-valued code points,
579 whether or not PCRE_UCP is set. The horizontal space characters are:
580 <pre>
581 U+0009 Horizontal tab (HT)
582 U+0020 Space
583 U+00A0 Non-break space
584 U+1680 Ogham space mark
585 U+180E Mongolian vowel separator
586 U+2000 En quad
587 U+2001 Em quad
588 U+2002 En space
589 U+2003 Em space
590 U+2004 Three-per-em space
591 U+2005 Four-per-em space
592 U+2006 Six-per-em space
593 U+2007 Figure space
594 U+2008 Punctuation space
595 U+2009 Thin space
596 U+200A Hair space
597 U+202F Narrow no-break space
598 U+205F Medium mathematical space
599 U+3000 Ideographic space
600 </pre>
601 The vertical space characters are:
602 <pre>
603 U+000A Linefeed (LF)
604 U+000B Vertical tab (VT)
605 U+000C Form feed (FF)
606 U+000D Carriage return (CR)
607 U+0085 Next line (NEL)
608 U+2028 Line separator
609 U+2029 Paragraph separator
610 </pre>
611 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
612 relevant.
613 <a name="newlineseq"></a></P>
614 <br><b>
615 Newline sequences
616 </b><br>
617 <P>
618 Outside a character class, by default, the escape sequence \R matches any
619 Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the
620 following:
621 <pre>
622 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
623 </pre>
624 This is an example of an "atomic group", details of which are given
625 <a href="#atomicgroup">below.</a>
626 This particular group matches either the two-character sequence CR followed by
627 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
628 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
629 line, U+0085). The two-character sequence is treated as a single unit that
630 cannot be split.
631 </P>
632 <P>
633 In other modes, two additional characters whose codepoints are greater than 255
634 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
635 Unicode character property support is not needed for these characters to be
636 recognized.
637 </P>
638 <P>
639 It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
640 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
641 either at compile time or when the pattern is matched. (BSR is an abbrevation
642 for "backslash R".) This can be made the default when PCRE is built; if this is
643 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
644 It is also possible to specify these settings by starting a pattern string with
645 one of the following sequences:
646 <pre>
647 (*BSR_ANYCRLF) CR, LF, or CRLF only
648 (*BSR_UNICODE) any Unicode newline sequence
649 </pre>
650 These override the default and the options given to the compiling function, but
651 they can themselves be overridden by options given to a matching function. Note
652 that these special settings, which are not Perl-compatible, are recognized only
653 at the very start of a pattern, and that they must be in upper case. If more
654 than one of them is present, the last one is used. They can be combined with a
655 change of newline convention; for example, a pattern can start with:
656 <pre>
658 </pre>
659 They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or
660 (*UCP) special sequences. Inside a character class, \R is treated as an
661 unrecognized escape sequence, and so matches the letter "R" by default, but
662 causes an error if PCRE_EXTRA is set.
663 <a name="uniextseq"></a></P>
664 <br><b>
665 Unicode character properties
666 </b><br>
667 <P>
668 When PCRE is built with Unicode character property support, three additional
669 escape sequences that match characters with specific properties are available.
670 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
671 characters whose codepoints are less than 256, but they do work in this mode.
672 The extra escape sequences are:
673 <pre>
674 \p{<i>xx</i>} a character with the <i>xx</i> property
675 \P{<i>xx</i>} a character without the <i>xx</i> property
676 \X a Unicode extended grapheme cluster
677 </pre>
678 The property names represented by <i>xx</i> above are limited to the Unicode
679 script names, the general category properties, "Any", which matches any
680 character (including newline), and some special PCRE properties (described
681 in the
682 <a href="#extraprops">next section).</a>
683 Other Perl properties such as "InMusicalSymbols" are not currently supported by
684 PCRE. Note that \P{Any} does not match any characters, so always causes a
685 match failure.
686 </P>
687 <P>
688 Sets of Unicode characters are defined as belonging to certain scripts. A
689 character from one of these sets can be matched using a script name. For
690 example:
691 <pre>
692 \p{Greek}
693 \P{Han}
694 </pre>
695 Those that are not part of an identified script are lumped together as
696 "Common". The current list of scripts is:
697 </P>
698 <P>
699 Arabic,
700 Armenian,
701 Avestan,
702 Balinese,
703 Bamum,
704 Batak,
705 Bengali,
706 Bopomofo,
707 Brahmi,
708 Braille,
709 Buginese,
710 Buhid,
711 Canadian_Aboriginal,
712 Carian,
713 Chakma,
714 Cham,
715 Cherokee,
716 Common,
717 Coptic,
718 Cuneiform,
719 Cypriot,
720 Cyrillic,
721 Deseret,
722 Devanagari,
723 Egyptian_Hieroglyphs,
724 Ethiopic,
725 Georgian,
726 Glagolitic,
727 Gothic,
728 Greek,
729 Gujarati,
730 Gurmukhi,
731 Han,
732 Hangul,
733 Hanunoo,
734 Hebrew,
735 Hiragana,
736 Imperial_Aramaic,
737 Inherited,
738 Inscriptional_Pahlavi,
739 Inscriptional_Parthian,
740 Javanese,
741 Kaithi,
742 Kannada,
743 Katakana,
744 Kayah_Li,
745 Kharoshthi,
746 Khmer,
747 Lao,
748 Latin,
749 Lepcha,
750 Limbu,
751 Linear_B,
752 Lisu,
753 Lycian,
754 Lydian,
755 Malayalam,
756 Mandaic,
757 Meetei_Mayek,
758 Meroitic_Cursive,
759 Meroitic_Hieroglyphs,
760 Miao,
761 Mongolian,
762 Myanmar,
763 New_Tai_Lue,
764 Nko,
765 Ogham,
766 Old_Italic,
767 Old_Persian,
768 Old_South_Arabian,
769 Old_Turkic,
770 Ol_Chiki,
771 Oriya,
772 Osmanya,
773 Phags_Pa,
774 Phoenician,
775 Rejang,
776 Runic,
777 Samaritan,
778 Saurashtra,
779 Sharada,
780 Shavian,
781 Sinhala,
782 Sora_Sompeng,
783 Sundanese,
784 Syloti_Nagri,
785 Syriac,
786 Tagalog,
787 Tagbanwa,
788 Tai_Le,
789 Tai_Tham,
790 Tai_Viet,
791 Takri,
792 Tamil,
793 Telugu,
794 Thaana,
795 Thai,
796 Tibetan,
797 Tifinagh,
798 Ugaritic,
799 Vai,
800 Yi.
801 </P>
802 <P>
803 Each character has exactly one Unicode general category property, specified by
804 a two-letter abbreviation. For compatibility with Perl, negation can be
805 specified by including a circumflex between the opening brace and the property
806 name. For example, \p{^Lu} is the same as \P{Lu}.
807 </P>
808 <P>
809 If only one letter is specified with \p or \P, it includes all the general
810 category properties that start with that letter. In this case, in the absence
811 of negation, the curly brackets in the escape sequence are optional; these two
812 examples have the same effect:
813 <pre>
814 \p{L}
815 \pL
816 </pre>
817 The following general category property codes are supported:
818 <pre>
819 C Other
820 Cc Control
821 Cf Format
822 Cn Unassigned
823 Co Private use
824 Cs Surrogate
826 L Letter
827 Ll Lower case letter
828 Lm Modifier letter
829 Lo Other letter
830 Lt Title case letter
831 Lu Upper case letter
833 M Mark
834 Mc Spacing mark
835 Me Enclosing mark
836 Mn Non-spacing mark
838 N Number
839 Nd Decimal number
840 Nl Letter number
841 No Other number
843 P Punctuation
844 Pc Connector punctuation
845 Pd Dash punctuation
846 Pe Close punctuation
847 Pf Final punctuation
848 Pi Initial punctuation
849 Po Other punctuation
850 Ps Open punctuation
852 S Symbol
853 Sc Currency symbol
854 Sk Modifier symbol
855 Sm Mathematical symbol
856 So Other symbol
858 Z Separator
859 Zl Line separator
860 Zp Paragraph separator
861 Zs Space separator
862 </pre>
863 The special property L& is also supported: it matches a character that has
864 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
865 a modifier or "other".
866 </P>
867 <P>
868 The Cs (Surrogate) property applies only to characters in the range U+D800 to
869 U+DFFF. Such characters are not valid in Unicode strings and so
870 cannot be tested by PCRE, unless UTF validity checking has been turned off
871 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
872 PCRE_NO_UTF32_CHECK in the
873 <a href="pcreapi.html"><b>pcreapi</b></a>
874 page). Perl does not support the Cs property.
875 </P>
876 <P>
877 The long synonyms for property names that Perl supports (such as \p{Letter})
878 are not supported by PCRE, nor is it permitted to prefix any of these
879 properties with "Is".
880 </P>
881 <P>
882 No character that is in the Unicode table has the Cn (unassigned) property.
883 Instead, this property is assumed for any code point that is not in the
884 Unicode table.
885 </P>
886 <P>
887 Specifying caseless matching does not affect these escape sequences. For
888 example, \p{Lu} always matches only upper case letters. This is different from
889 the behaviour of current versions of Perl.
890 </P>
891 <P>
892 Matching characters by Unicode property is not fast, because PCRE has to do a
893 multistage table lookup in order to find a character's property. That is why
894 the traditional escape sequences such as \d and \w do not use Unicode
895 properties in PCRE by default, though you can make them do so by setting the
896 PCRE_UCP option or by starting the pattern with (*UCP).
897 </P>
898 <br><b>
899 Extended grapheme clusters
900 </b><br>
901 <P>
902 The \X escape matches any number of Unicode characters that form an "extended
903 grapheme cluster", and treats the sequence as an atomic group
904 <a href="#atomicgroup">(see below).</a>
905 Up to and including release 8.31, PCRE matched an earlier, simpler definition
906 that was equivalent to
907 <pre>
908 (?&#62;\PM\pM*)
909 </pre>
910 That is, it matched a character without the "mark" property, followed by zero
911 or more characters with the "mark" property. Characters with the "mark"
912 property are typically non-spacing accents that affect the preceding character.
913 </P>
914 <P>
915 This simple definition was extended in Unicode to include more complicated
916 kinds of composite character by giving each character a grapheme breaking
917 property, and creating rules that use these properties to define the boundaries
918 of extended grapheme clusters. In releases of PCRE later than 8.31, \X matches
919 one of these clusters.
920 </P>
921 <P>
922 \X always matches at least one character. Then it decides whether to add
923 additional characters according to the following rules for ending a cluster:
924 </P>
925 <P>
926 1. End at the end of the subject string.
927 </P>
928 <P>
929 2. Do not end between CR and LF; otherwise end after any control character.
930 </P>
931 <P>
932 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
933 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
934 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
935 character; an LVT or T character may be follwed only by a T character.
936 </P>
937 <P>
938 4. Do not end before extending characters or spacing marks. Characters with
939 the "mark" property always have the "extend" grapheme breaking property.
940 </P>
941 <P>
942 5. Do not end after prepend characters.
943 </P>
944 <P>
945 6. Otherwise, end the cluster.
946 <a name="extraprops"></a></P>
947 <br><b>
948 PCRE's additional properties
949 </b><br>
950 <P>
951 As well as the standard Unicode properties described above, PCRE supports four
952 more that make it possible to convert traditional escape sequences such as \w
953 and \s to use Unicode properties. PCRE uses these non-standard, non-Perl
954 properties internally when PCRE_UCP is set. However, they may also be used
955 explicitly. These properties are:
956 <pre>
957 Xan Any alphanumeric character
958 Xps Any POSIX space character
959 Xsp Any Perl space character
960 Xwd Any Perl "word" character
961 </pre>
962 Xan matches characters that have either the L (letter) or the N (number)
963 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
964 carriage return, and any other character that has the Z (separator) property.
965 Xsp is the same as Xps; it used to exclude vertical tab, for Perl
966 compatibility, but Perl changed, and so PCRE followed at release 8.34. Xwd
967 matches the same characters as Xan, plus underscore.
968 </P>
969 <P>
970 There is another non-standard property, Xuc, which matches any character that
971 can be represented by a Universal Character Name in C++ and other programming
972 languages. These are the characters $, @, ` (grave accent), and all characters
973 with Unicode code points greater than or equal to U+00A0, except for the
974 surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are
975 excluded. (Universal Character Names are of the form \uHHHH or \UHHHHHHHH
976 where H is a hexadecimal digit. Note that the Xuc property does not match these
977 sequences but the characters that they represent.)
978 <a name="resetmatchstart"></a></P>
979 <br><b>
980 Resetting the match start
981 </b><br>
982 <P>
983 The escape sequence \K causes any previously matched characters not to be
984 included in the final matched sequence. For example, the pattern:
985 <pre>
986 foo\Kbar
987 </pre>
988 matches "foobar", but reports that it has matched "bar". This feature is
989 similar to a lookbehind assertion
990 <a href="#lookbehind">(described below).</a>
991 However, in this case, the part of the subject before the real match does not
992 have to be of fixed length, as lookbehind assertions do. The use of \K does
993 not interfere with the setting of
994 <a href="#subpattern">captured substrings.</a>
995 For example, when the pattern
996 <pre>
997 (foo)\Kbar
998 </pre>
999 matches "foobar", the first substring is still set to "foo".
1000 </P>
1001 <P>
1002 Perl documents that the use of \K within assertions is "not well defined". In
1003 PCRE, \K is acted upon when it occurs inside positive assertions, but is
1004 ignored in negative assertions.
1005 <a name="smallassertions"></a></P>
1006 <br><b>
1007 Simple assertions
1008 </b><br>
1009 <P>
1010 The final use of backslash is for certain simple assertions. An assertion
1011 specifies a condition that has to be met at a particular point in a match,
1012 without consuming any characters from the subject string. The use of
1013 subpatterns for more complicated assertions is described
1014 <a href="#bigassertions">below.</a>
1015 The backslashed assertions are:
1016 <pre>
1017 \b matches at a word boundary
1018 \B matches when not at a word boundary
1019 \A matches at the start of the subject
1020 \Z matches at the end of the subject
1021 also matches before a newline at the end of the subject
1022 \z matches only at the end of the subject
1023 \G matches at the first matching position in the subject
1024 </pre>
1025 Inside a character class, \b has a different meaning; it matches the backspace
1026 character. If any other of these assertions appears in a character class, by
1027 default it matches the corresponding literal character (for example, \B
1028 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
1029 escape sequence" error is generated instead.
1030 </P>
1031 <P>
1032 A word boundary is a position in the subject string where the current character
1033 and the previous character do not both match \w or \W (i.e. one matches
1034 \w and the other matches \W), or the start or end of the string if the
1035 first or last character matches \w, respectively. In a UTF mode, the meanings
1036 of \w and \W can be changed by setting the PCRE_UCP option. When this is
1037 done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start
1038 of word" or "end of word" metasequence. However, whatever follows \b normally
1039 determines which it is. For example, the fragment \ba matches "a" at the start
1040 of a word.
1041 </P>
1042 <P>
1043 The \A, \Z, and \z assertions differ from the traditional circumflex and
1044 dollar (described in the next section) in that they only ever match at the very
1045 start and end of the subject string, whatever options are set. Thus, they are
1046 independent of multiline mode. These three assertions are not affected by the
1047 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
1048 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
1049 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
1050 at a point other than the beginning of the subject, \A can never match. The
1051 difference between \Z and \z is that \Z matches before a newline at the end
1052 of the string as well as at the very end, whereas \z matches only at the end.
1053 </P>
1054 <P>
1055 The \G assertion is true only when the current matching position is at the
1056 start point of the match, as specified by the <i>startoffset</i> argument of
1057 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
1058 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
1059 arguments, you can mimic Perl's /g option, and it is in this kind of
1060 implementation where \G can be useful.
1061 </P>
1062 <P>
1063 Note, however, that PCRE's interpretation of \G, as the start of the current
1064 match, is subtly different from Perl's, which defines it as the end of the
1065 previous match. In Perl, these can be different when the previously matched
1066 string was empty. Because PCRE does just one match at a time, it cannot
1067 reproduce this behaviour.
1068 </P>
1069 <P>
1070 If all the alternatives of a pattern begin with \G, the expression is anchored
1071 to the starting match position, and the "anchored" flag is set in the compiled
1072 regular expression.
1073 </P>
1074 <br><a name="SEC6" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
1075 <P>
1076 The circumflex and dollar metacharacters are zero-width assertions. That is,
1077 they test for a particular condition being true without consuming any
1078 characters from the subject string.
1079 </P>
1080 <P>
1081 Outside a character class, in the default matching mode, the circumflex
1082 character is an assertion that is true only if the current matching point is at
1083 the start of the subject string. If the <i>startoffset</i> argument of
1084 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
1085 option is unset. Inside a character class, circumflex has an entirely different
1086 meaning
1087 <a href="#characterclass">(see below).</a>
1088 </P>
1089 <P>
1090 Circumflex need not be the first character of the pattern if a number of
1091 alternatives are involved, but it should be the first thing in each alternative
1092 in which it appears if the pattern is ever to match that branch. If all
1093 possible alternatives start with a circumflex, that is, if the pattern is
1094 constrained to match only at the start of the subject, it is said to be an
1095 "anchored" pattern. (There are also other constructs that can cause a pattern
1096 to be anchored.)
1097 </P>
1098 <P>
1099 The dollar character is an assertion that is true only if the current matching
1100 point is at the end of the subject string, or immediately before a newline at
1101 the end of the string (by default). Note, however, that it does not actually
1102 match the newline. Dollar need not be the last character of the pattern if a
1103 number of alternatives are involved, but it should be the last item in any
1104 branch in which it appears. Dollar has no special meaning in a character class.
1105 </P>
1106 <P>
1107 The meaning of dollar can be changed so that it matches only at the very end of
1108 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1109 does not affect the \Z assertion.
1110 </P>
1111 <P>
1112 The meanings of the circumflex and dollar characters are changed if the
1113 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1114 immediately after internal newlines as well as at the start of the subject
1115 string. It does not match after a newline that ends the string. A dollar
1116 matches before any newlines in the string, as well as at the very end, when
1117 PCRE_MULTILINE is set. When newline is specified as the two-character
1118 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1119 </P>
1120 <P>
1121 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
1122 \n represents a newline) in multiline mode, but not otherwise. Consequently,
1123 patterns that are anchored in single line mode because all branches start with
1124 ^ are not anchored in multiline mode, and a match for circumflex is possible
1125 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
1126 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1127 </P>
1128 <P>
1129 Note that the sequences \A, \Z, and \z can be used to match the start and
1130 end of the subject in both modes, and if all branches of a pattern start with
1131 \A it is always anchored, whether or not PCRE_MULTILINE is set.
1132 <a name="fullstopdot"></a></P>
1133 <br><a name="SEC7" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br>
1134 <P>
1135 Outside a character class, a dot in the pattern matches any one character in
1136 the subject string except (by default) a character that signifies the end of a
1137 line.
1138 </P>
1139 <P>
1140 When a line ending is defined as a single character, dot never matches that
1141 character; when the two-character sequence CRLF is used, dot does not match CR
1142 if it is immediately followed by LF, but otherwise it matches all characters
1143 (including isolated CRs and LFs). When any Unicode line endings are being
1144 recognized, dot does not match CR or LF or any of the other line ending
1145 characters.
1146 </P>
1147 <P>
1148 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1149 option is set, a dot matches any one character, without exception. If the
1150 two-character sequence CRLF is present in the subject string, it takes two dots
1151 to match it.
1152 </P>
1153 <P>
1154 The handling of dot is entirely independent of the handling of circumflex and
1155 dollar, the only relationship being that they both involve newlines. Dot has no
1156 special meaning in a character class.
1157 </P>
1158 <P>
1159 The escape sequence \N behaves like a dot, except that it is not affected by
1160 the PCRE_DOTALL option. In other words, it matches any character except one
1161 that signifies the end of a line. Perl also uses \N to match characters by
1162 name; PCRE does not support this.
1163 </P>
1164 <br><a name="SEC8" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br>
1165 <P>
1166 Outside a character class, the escape sequence \C matches any one data unit,
1167 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1168 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1169 a 32-bit unit. Unlike a dot, \C always
1170 matches line-ending characters. The feature is provided in Perl in order to
1171 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1172 used. Because \C breaks up characters into individual data units, matching one
1173 unit with \C in a UTF mode means that the rest of the string may start with a
1174 malformed UTF character. This has undefined results, because PCRE assumes that
1175 it is dealing with valid UTF strings (and by default it checks this at the
1176 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1177 PCRE_NO_UTF32_CHECK option is used).
1178 </P>
1179 <P>
1180 PCRE does not allow \C to appear in lookbehind assertions
1181 <a href="#lookbehind">(described below)</a>
1182 in a UTF mode, because this would make it impossible to calculate the length of
1183 the lookbehind.
1184 </P>
1185 <P>
1186 In general, the \C escape sequence is best avoided. However, one
1187 way of using it that avoids the problem of malformed UTF characters is to use a
1188 lookahead to check the length of the next character, as in this pattern, which
1189 could be used with a UTF-8 string (ignore white space and line breaks):
1190 <pre>
1191 (?| (?=[\x00-\x7f])(\C) |
1192 (?=[\x80-\x{7ff}])(\C)(\C) |
1193 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
1194 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
1195 </pre>
1196 A group that starts with (?| resets the capturing parentheses numbers in each
1197 alternative (see
1198 <a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a>
1199 below). The assertions at the start of each branch check the next UTF-8
1200 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1201 character's individual bytes are then captured by the appropriate number of
1202 groups.
1203 <a name="characterclass"></a></P>
1204 <br><a name="SEC9" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
1205 <P>
1206 An opening square bracket introduces a character class, terminated by a closing
1207 square bracket. A closing square bracket on its own is not special by default.
1208 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1209 bracket causes a compile-time error. If a closing square bracket is required as
1210 a member of the class, it should be the first data character in the class
1211 (after an initial circumflex, if present) or escaped with a backslash.
1212 </P>
1213 <P>
1214 A character class matches a single character in the subject. In a UTF mode, the
1215 character may be more than one data unit long. A matched character must be in
1216 the set of characters defined by the class, unless the first character in the
1217 class definition is a circumflex, in which case the subject character must not
1218 be in the set defined by the class. If a circumflex is actually required as a
1219 member of the class, ensure it is not the first character, or escape it with a
1220 backslash.
1221 </P>
1222 <P>
1223 For example, the character class [aeiou] matches any lower case vowel, while
1224 [^aeiou] matches any character that is not a lower case vowel. Note that a
1225 circumflex is just a convenient notation for specifying the characters that
1226 are in the class by enumerating those that are not. A class that starts with a
1227 circumflex is not an assertion; it still consumes a character from the subject
1228 string, and therefore it fails if the current pointer is at the end of the
1229 string.
1230 </P>
1231 <P>
1232 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1233 can be included in a class as a literal string of data units, or by using the
1234 \x{ escaping mechanism.
1235 </P>
1236 <P>
1237 When caseless matching is set, any letters in a class represent both their
1238 upper case and lower case versions, so for example, a caseless [aeiou] matches
1239 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1240 caseful version would. In a UTF mode, PCRE always understands the concept of
1241 case for characters whose values are less than 128, so caseless matching is
1242 always possible. For characters with higher values, the concept of case is
1243 supported if PCRE is compiled with Unicode property support, but not otherwise.
1244 If you want to use caseless matching in a UTF mode for characters 128 and
1245 above, you must ensure that PCRE is compiled with Unicode property support as
1246 well as with UTF support.
1247 </P>
1248 <P>
1249 Characters that might indicate line breaks are never treated in any special way
1250 when matching character classes, whatever line-ending sequence is in use, and
1251 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1252 such as [^a] always matches one of these characters.
1253 </P>
1254 <P>
1255 The minus (hyphen) character can be used to specify a range of characters in a
1256 character class. For example, [d-m] matches any letter between d and m,
1257 inclusive. If a minus character is required in a class, it must be escaped with
1258 a backslash or appear in a position where it cannot be interpreted as
1259 indicating a range, typically as the first or last character in the class, or
1260 immediately after a range. For example, [b-d-z] matches letters in the range b
1261 to d, a hyphen character, or z.
1262 </P>
1263 <P>
1264 It is not possible to have the literal character "]" as the end character of a
1265 range. A pattern such as [W-]46] is interpreted as a class of two characters
1266 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1267 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1268 the end of range, so [W-\]46] is interpreted as a class containing a range
1269 followed by two other characters. The octal or hexadecimal representation of
1270 "]" can also be used to end a range.
1271 </P>
1272 <P>
1273 An error is generated if a POSIX character class (see below) or an escape
1274 sequence other than one that defines a single character appears at a point
1275 where a range ending character is expected. For example, [z-\xff] is valid,
1276 but [A-\d] and [A-[:digit:]] are not.
1277 </P>
1278 <P>
1279 Ranges operate in the collating sequence of character values. They can also be
1280 used for characters specified numerically, for example [\000-\037]. Ranges
1281 can include any characters that are valid for the current mode.
1282 </P>
1283 <P>
1284 If a range that includes letters is used when caseless matching is set, it
1285 matches the letters in either case. For example, [W-c] is equivalent to
1286 [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1287 tables for a French locale are in use, [\xc8-\xcb] matches accented E
1288 characters in both cases. In UTF modes, PCRE supports the concept of case for
1289 characters with values greater than 128 only when it is compiled with Unicode
1290 property support.
1291 </P>
1292 <P>
1293 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v,
1294 \V, \w, and \W may appear in a character class, and add the characters that
1295 they match to the class. For example, [\dABCDEF] matches any hexadecimal
1296 digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w
1297 and their upper case partners, just as it does when they appear outside a
1298 character class, as described in the section entitled
1299 <a href="#genericchartypes">"Generic character types"</a>
1300 above. The escape sequence \b has a different meaning inside a character
1301 class; it matches the backspace character. The sequences \B, \N, \R, and \X
1302 are not special inside a character class. Like any other unrecognized escape
1303 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1304 default, but cause an error if the PCRE_EXTRA option is set.
1305 </P>
1306 <P>
1307 A circumflex can conveniently be used with the upper case character types to
1308 specify a more restricted set of characters than the matching lower case type.
1309 For example, the class [^\W_] matches any letter or digit, but not underscore,
1310 whereas [\w] includes underscore. A positive character class should be read as
1311 "something OR something OR ..." and a negative class as "NOT something AND NOT
1312 something AND NOT ...".
1313 </P>
1314 <P>
1315 The only metacharacters that are recognized in character classes are backslash,
1316 hyphen (only where it can be interpreted as specifying a range), circumflex
1317 (only at the start), opening square bracket (only when it can be interpreted as
1318 introducing a POSIX class name - see the next section), and the terminating
1319 closing square bracket. However, escaping other non-alphanumeric characters
1320 does no harm.
1321 </P>
1322 <br><a name="SEC10" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
1323 <P>
1324 Perl supports the POSIX notation for character classes. This uses names
1325 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1326 this notation. For example,
1327 <pre>
1328 [01[:alpha:]%]
1329 </pre>
1330 matches "0", "1", any alphabetic character, or "%". The supported class names
1331 are:
1332 <pre>
1333 alnum letters and digits
1334 alpha letters
1335 ascii character codes 0 - 127
1336 blank space or tab only
1337 cntrl control characters
1338 digit decimal digits (same as \d)
1339 graph printing characters, excluding space
1340 lower lower case letters
1341 print printing characters, including space
1342 punct printing characters, excluding letters and digits and space
1343 space white space (the same as \s from PCRE 8.34)
1344 upper upper case letters
1345 word "word" characters (same as \w)
1346 xdigit hexadecimal digits
1347 </pre>
1348 The default "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
1349 and space (32). If locale-specific matching is taking place, there may be
1350 additional space characters. "Space" used to be different to \s, which did not
1351 include VT, for Perl compatibility. However, Perl changed at release 5.18, and
1352 PCRE followed at release 8.34. "Space" and \s now match the same set of
1353 characters.
1354 </P>
1355 <P>
1356 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1357 5.8. Another Perl extension is negation, which is indicated by a ^ character
1358 after the colon. For example,
1359 <pre>
1360 [12[:^digit:]]
1361 </pre>
1362 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1363 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1364 supported, and an error is given if they are encountered.
1365 </P>
1366 <P>
1367 By default, characters with values greater than 128 do not match any of the
1368 POSIX character classes. However, if the PCRE_UCP option is passed to
1369 <b>pcre_compile()</b>, some of the classes are changed so that Unicode character
1370 properties are used. This is achieved by replacing certain POSIX classes by
1371 other sequences, as follows:
1372 <pre>
1373 [:alnum:] becomes \p{Xan}
1374 [:alpha:] becomes \p{L}
1375 [:blank:] becomes \h
1376 [:digit:] becomes \p{Nd}
1377 [:lower:] becomes \p{Ll}
1378 [:space:] becomes \p{Xps}
1379 [:upper:] becomes \p{Lu}
1380 [:word:] becomes \p{Xwd}
1381 </pre>
1382 Negated versions, such as [:^alpha:] use \P instead of \p. Three other POSIX
1383 classes are handled specially in UCP mode:
1384 </P>
1385 <P>
1386 [:graph:]
1387 This matches characters that have glyphs that mark the page when printed. In
1388 Unicode property terms, it matches all characters with the L, M, N, P, S, or Cf
1389 properties, except for:
1390 <pre>
1391 U+061C Arabic Letter Mark
1392 U+180E Mongolian Vowel Separator
1393 U+2066 - U+2069 Various "isolate"s
1395 </PRE>
1396 </P>
1397 <P>
1398 [:print:]
1399 This matches the same characters as [:graph:] plus space characters that are
1400 not controls, that is, characters with the Zs property.
1401 </P>
1402 <P>
1403 [:punct:]
1404 This matches all characters that have the Unicode P (punctuation) property,
1405 plus those characters whose code points are less than 128 that have the S
1406 (Symbol) property.
1407 </P>
1408 <P>
1409 The other POSIX classes are unchanged, and match only characters with code
1410 points less than 128.
1411 </P>
1412 <br><a name="SEC11" href="#TOC1">VERTICAL BAR</a><br>
1413 <P>
1414 Vertical bar characters are used to separate alternative patterns. For example,
1415 the pattern
1416 <pre>
1417 gilbert|sullivan
1418 </pre>
1419 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1420 and an empty alternative is permitted (matching the empty string). The matching
1421 process tries each alternative in turn, from left to right, and the first one
1422 that succeeds is used. If the alternatives are within a subpattern
1423 <a href="#subpattern">(defined below),</a>
1424 "succeeds" means matching the rest of the main pattern as well as the
1425 alternative in the subpattern.
1426 </P>
1427 <br><a name="SEC12" href="#TOC1">INTERNAL OPTION SETTING</a><br>
1428 <P>
1429 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1430 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1431 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1432 The option letters are
1433 <pre>
1434 i for PCRE_CASELESS
1436 s for PCRE_DOTALL
1437 x for PCRE_EXTENDED
1438 </pre>
1439 For example, (?im) sets caseless, multiline matching. It is also possible to
1440 unset these options by preceding the letter with a hyphen, and a combined
1441 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1442 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1443 permitted. If a letter appears both before and after the hyphen, the option is
1444 unset.
1445 </P>
1446 <P>
1447 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1448 changed in the same way as the Perl-compatible options by using the characters
1449 J, U and X respectively.
1450 </P>
1451 <P>
1452 When one of these option changes occurs at top level (that is, not inside
1453 subpattern parentheses), the change applies to the remainder of the pattern
1454 that follows. If the change is placed right at the start of a pattern, PCRE
1455 extracts it into the global options (and it will therefore show up in data
1456 extracted by the <b>pcre_fullinfo()</b> function).
1457 </P>
1458 <P>
1459 An option change within a subpattern (see below for a description of
1460 subpatterns) affects only that part of the subpattern that follows it, so
1461 <pre>
1462 (a(?i)b)c
1463 </pre>
1464 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1465 By this means, options can be made to have different settings in different
1466 parts of the pattern. Any changes made in one alternative do carry on
1467 into subsequent branches within the same subpattern. For example,
1468 <pre>
1469 (a(?i)b|c)
1470 </pre>
1471 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1472 branch is abandoned before the option setting. This is because the effects of
1473 option settings happen at compile time. There would be some very weird
1474 behaviour otherwise.
1475 </P>
1476 <P>
1477 <b>Note:</b> There are other PCRE-specific options that can be set by the
1478 application when the compiling or matching functions are called. In some cases
1479 the pattern can contain special leading sequences such as (*CRLF) to override
1480 what the application has set or what has been defaulted. Details are given in
1481 the section entitled
1482 <a href="#newlineseq">"Newline sequences"</a>
1483 above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading
1484 sequences that can be used to set UTF and Unicode property modes; they are
1485 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1486 options, respectively. The (*UTF) sequence is a generic version that can be
1487 used with any of the libraries. However, the application can set the
1488 PCRE_NEVER_UTF option, which locks out the use of the (*UTF) sequences.
1489 <a name="subpattern"></a></P>
1490 <br><a name="SEC13" href="#TOC1">SUBPATTERNS</a><br>
1491 <P>
1492 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1493 Turning part of a pattern into a subpattern does two things:
1494 <br>
1495 <br>
1496 1. It localizes a set of alternatives. For example, the pattern
1497 <pre>
1498 cat(aract|erpillar|)
1499 </pre>
1500 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1501 match "cataract", "erpillar" or an empty string.
1502 <br>
1503 <br>
1504 2. It sets up the subpattern as a capturing subpattern. This means that, when
1505 the whole pattern matches, that portion of the subject string that matched the
1506 subpattern is passed back to the caller via the <i>ovector</i> argument of the
1507 matching function. (This applies only to the traditional matching functions;
1508 the DFA matching functions do not support capturing.)
1509 </P>
1510 <P>
1511 Opening parentheses are counted from left to right (starting from 1) to obtain
1512 numbers for the capturing subpatterns. For example, if the string "the red
1513 king" is matched against the pattern
1514 <pre>
1515 the ((red|white) (king|queen))
1516 </pre>
1517 the captured substrings are "red king", "red", and "king", and are numbered 1,
1518 2, and 3, respectively.
1519 </P>
1520 <P>
1521 The fact that plain parentheses fulfil two functions is not always helpful.
1522 There are often times when a grouping subpattern is required without a
1523 capturing requirement. If an opening parenthesis is followed by a question mark
1524 and a colon, the subpattern does not do any capturing, and is not counted when
1525 computing the number of any subsequent capturing subpatterns. For example, if
1526 the string "the white queen" is matched against the pattern
1527 <pre>
1528 the ((?:red|white) (king|queen))
1529 </pre>
1530 the captured substrings are "white queen" and "queen", and are numbered 1 and
1531 2. The maximum number of capturing subpatterns is 65535.
1532 </P>
1533 <P>
1534 As a convenient shorthand, if any option settings are required at the start of
1535 a non-capturing subpattern, the option letters may appear between the "?" and
1536 the ":". Thus the two patterns
1537 <pre>
1538 (?i:saturday|sunday)
1539 (?:(?i)saturday|sunday)
1540 </pre>
1541 match exactly the same set of strings. Because alternative branches are tried
1542 from left to right, and options are not reset until the end of the subpattern
1543 is reached, an option setting in one branch does affect subsequent branches, so
1544 the above patterns match "SUNDAY" as well as "Saturday".
1545 <a name="dupsubpatternnumber"></a></P>
1546 <br><a name="SEC14" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1547 <P>
1548 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1549 the same numbers for its capturing parentheses. Such a subpattern starts with
1550 (?| and is itself a non-capturing subpattern. For example, consider this
1551 pattern:
1552 <pre>
1553 (?|(Sat)ur|(Sun))day
1554 </pre>
1555 Because the two alternatives are inside a (?| group, both sets of capturing
1556 parentheses are numbered one. Thus, when the pattern matches, you can look
1557 at captured substring number one, whichever alternative matched. This construct
1558 is useful when you want to capture part, but not all, of one of a number of
1559 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1560 number is reset at the start of each branch. The numbers of any capturing
1561 parentheses that follow the subpattern start after the highest number used in
1562 any branch. The following example is taken from the Perl documentation. The
1563 numbers underneath show in which buffer the captured content will be stored.
1564 <pre>
1565 # before ---------------branch-reset----------- after
1566 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1567 # 1 2 2 3 2 3 4
1568 </pre>
1569 A back reference to a numbered subpattern uses the most recent value that is
1570 set for that number by any subpattern. The following pattern matches "abcabc"
1571 or "defdef":
1572 <pre>
1573 /(?|(abc)|(def))\1/
1574 </pre>
1575 In contrast, a subroutine call to a numbered subpattern always refers to the
1576 first one in the pattern with the given number. The following pattern matches
1577 "abcabc" or "defabc":
1578 <pre>
1579 /(?|(abc)|(def))(?1)/
1580 </pre>
1581 If a
1582 <a href="#conditions">condition test</a>
1583 for a subpattern's having matched refers to a non-unique number, the test is
1584 true if any of the subpatterns of that number have matched.
1585 </P>
1586 <P>
1587 An alternative approach to using this "branch reset" feature is to use
1588 duplicate named subpatterns, as described in the next section.
1589 </P>
1590 <br><a name="SEC15" href="#TOC1">NAMED SUBPATTERNS</a><br>
1591 <P>
1592 Identifying capturing parentheses by number is simple, but it can be very hard
1593 to keep track of the numbers in complicated regular expressions. Furthermore,
1594 if an expression is modified, the numbers may change. To help with this
1595 difficulty, PCRE supports the naming of subpatterns. This feature was not
1596 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1597 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1598 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1599 have different names, but PCRE does not.
1600 </P>
1601 <P>
1602 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1603 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1604 parentheses from other parts of the pattern, such as
1605 <a href="#backreferences">back references,</a>
1606 <a href="#recursion">recursion,</a>
1607 and
1608 <a href="#conditions">conditions,</a>
1609 can be made by name as well as by number.
1610 </P>
1611 <P>
1612 Names consist of up to 32 alphanumeric characters and underscores, but must
1613 start with a non-digit. Named capturing parentheses are still allocated numbers
1614 as well as names, exactly as if the names were not present. The PCRE API
1615 provides function calls for extracting the name-to-number translation table
1616 from a compiled pattern. There is also a convenience function for extracting a
1617 captured substring by name.
1618 </P>
1619 <P>
1620 By default, a name must be unique within a pattern, but it is possible to relax
1621 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1622 names are also always permitted for subpatterns with the same number, set up as
1623 described in the previous section.) Duplicate names can be useful for patterns
1624 where only one instance of the named parentheses can match. Suppose you want to
1625 match the name of a weekday, either as a 3-letter abbreviation or as the full
1626 name, and in both cases you want to extract the abbreviation. This pattern
1627 (ignoring the line breaks) does the job:
1628 <pre>
1629 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1630 (?&#60;DN&#62;Tue)(?:sday)?|
1631 (?&#60;DN&#62;Wed)(?:nesday)?|
1632 (?&#60;DN&#62;Thu)(?:rsday)?|
1633 (?&#60;DN&#62;Sat)(?:urday)?
1634 </pre>
1635 There are five capturing substrings, but only one is ever set after a match.
1636 (An alternative way of solving this problem is to use a "branch reset"
1637 subpattern, as described in the previous section.)
1638 </P>
1639 <P>
1640 The convenience function for extracting the data by name returns the substring
1641 for the first (and in this example, the only) subpattern of that name that
1642 matched. This saves searching to find which numbered subpattern it was.
1643 </P>
1644 <P>
1645 If you make a back reference to a non-unique named subpattern from elsewhere in
1646 the pattern, the subpatterns to which the name refers are checked in the order
1647 in which they appear in the overall pattern. The first one that is set is used
1648 for the reference. For example, this pattern matches both "foofoo" and
1649 "barbar" but not "foobar" or "barfoo":
1650 <pre>
1651 (?:(?&#60;n&#62;foo)|(?&#60;n&#62;bar))\k&#60;n&#62;
1653 </PRE>
1654 </P>
1655 <P>
1656 If you make a subroutine call to a non-unique named subpattern, the one that
1657 corresponds to the first occurrence of the name is used. In the absence of
1658 duplicate numbers (see the previous section) this is the one with the lowest
1659 number.
1660 </P>
1661 <P>
1662 If you use a named reference in a condition
1663 test (see the
1664 <a href="#conditions">section about conditions</a>
1665 below), either to check whether a subpattern has matched, or to check for
1666 recursion, all subpatterns with the same name are tested. If the condition is
1667 true for any one of them, the overall condition is true. This is the same
1668 behaviour as testing by number. For further details of the interfaces for
1669 handling named subpatterns, see the
1670 <a href="pcreapi.html"><b>pcreapi</b></a>
1671 documentation.
1672 </P>
1673 <P>
1674 <b>Warning:</b> You cannot use different names to distinguish between two
1675 subpatterns with the same number because PCRE uses only the numbers when
1676 matching. For this reason, an error is given at compile time if different names
1677 are given to subpatterns with the same number. However, you can always give the
1678 same name to subpatterns with the same number, even when PCRE_DUPNAMES is not
1679 set.
1680 </P>
1681 <br><a name="SEC16" href="#TOC1">REPETITION</a><br>
1682 <P>
1683 Repetition is specified by quantifiers, which can follow any of the following
1684 items:
1685 <pre>
1686 a literal data character
1687 the dot metacharacter
1688 the \C escape sequence
1689 the \X escape sequence
1690 the \R escape sequence
1691 an escape such as \d or \pL that matches a single character
1692 a character class
1693 a back reference (see next section)
1694 a parenthesized subpattern (including assertions)
1695 a subroutine call to a subpattern (recursive or otherwise)
1696 </pre>
1697 The general repetition quantifier specifies a minimum and maximum number of
1698 permitted matches, by giving the two numbers in curly brackets (braces),
1699 separated by a comma. The numbers must be less than 65536, and the first must
1700 be less than or equal to the second. For example:
1701 <pre>
1702 z{2,4}
1703 </pre>
1704 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1705 character. If the second number is omitted, but the comma is present, there is
1706 no upper limit; if the second number and the comma are both omitted, the
1707 quantifier specifies an exact number of required matches. Thus
1708 <pre>
1709 [aeiou]{3,}
1710 </pre>
1711 matches at least 3 successive vowels, but may match many more, while
1712 <pre>
1713 \d{8}
1714 </pre>
1715 matches exactly 8 digits. An opening curly bracket that appears in a position
1716 where a quantifier is not allowed, or one that does not match the syntax of a
1717 quantifier, is taken as a literal character. For example, {,6} is not a
1718 quantifier, but a literal string of four characters.
1719 </P>
1720 <P>
1721 In UTF modes, quantifiers apply to characters rather than to individual data
1722 units. Thus, for example, \x{100}{2} matches two characters, each of
1723 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1724 \X{3} matches three Unicode extended grapheme clusters, each of which may be
1725 several data units long (and they may be of different lengths).
1726 </P>
1727 <P>
1728 The quantifier {0} is permitted, causing the expression to behave as if the
1729 previous item and the quantifier were not present. This may be useful for
1730 subpatterns that are referenced as
1731 <a href="#subpatternsassubroutines">subroutines</a>
1732 from elsewhere in the pattern (but see also the section entitled
1733 <a href="#subdefine">"Defining subpatterns for use by reference only"</a>
1734 below). Items other than subpatterns that have a {0} quantifier are omitted
1735 from the compiled pattern.
1736 </P>
1737 <P>
1738 For convenience, the three most common quantifiers have single-character
1739 abbreviations:
1740 <pre>
1741 * is equivalent to {0,}
1742 + is equivalent to {1,}
1743 ? is equivalent to {0,1}
1744 </pre>
1745 It is possible to construct infinite loops by following a subpattern that can
1746 match no characters with a quantifier that has no upper limit, for example:
1747 <pre>
1748 (a?)*
1749 </pre>
1750 Earlier versions of Perl and PCRE used to give an error at compile time for
1751 such patterns. However, because there are cases where this can be useful, such
1752 patterns are now accepted, but if any repetition of the subpattern does in fact
1753 match no characters, the loop is forcibly broken.
1754 </P>
1755 <P>
1756 By default, the quantifiers are "greedy", that is, they match as much as
1757 possible (up to the maximum number of permitted times), without causing the
1758 rest of the pattern to fail. The classic example of where this gives problems
1759 is in trying to match comments in C programs. These appear between /* and */
1760 and within the comment, individual * and / characters may appear. An attempt to
1761 match C comments by applying the pattern
1762 <pre>
1763 /\*.*\*/
1764 </pre>
1765 to the string
1766 <pre>
1767 /* first comment */ not comment /* second comment */
1768 </pre>
1769 fails, because it matches the entire string owing to the greediness of the .*
1770 item.
1771 </P>
1772 <P>
1773 However, if a quantifier is followed by a question mark, it ceases to be
1774 greedy, and instead matches the minimum number of times possible, so the
1775 pattern
1776 <pre>
1777 /\*.*?\*/
1778 </pre>
1779 does the right thing with the C comments. The meaning of the various
1780 quantifiers is not otherwise changed, just the preferred number of matches.
1781 Do not confuse this use of question mark with its use as a quantifier in its
1782 own right. Because it has two uses, it can sometimes appear doubled, as in
1783 <pre>
1784 \d??\d
1785 </pre>
1786 which matches one digit by preference, but can match two if that is the only
1787 way the rest of the pattern matches.
1788 </P>
1789 <P>
1790 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1791 the quantifiers are not greedy by default, but individual ones can be made
1792 greedy by following them with a question mark. In other words, it inverts the
1793 default behaviour.
1794 </P>
1795 <P>
1796 When a parenthesized subpattern is quantified with a minimum repeat count that
1797 is greater than 1 or with a limited maximum, more memory is required for the
1798 compiled pattern, in proportion to the size of the minimum or maximum.
1799 </P>
1800 <P>
1801 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1802 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1803 implicitly anchored, because whatever follows will be tried against every
1804 character position in the subject string, so there is no point in retrying the
1805 overall match at any position after the first. PCRE normally treats such a
1806 pattern as though it were preceded by \A.
1807 </P>
1808 <P>
1809 In cases where it is known that the subject string contains no newlines, it is
1810 worth setting PCRE_DOTALL in order to obtain this optimization, or
1811 alternatively using ^ to indicate anchoring explicitly.
1812 </P>
1813 <P>
1814 However, there are some cases where the optimization cannot be used. When .*
1815 is inside capturing parentheses that are the subject of a back reference
1816 elsewhere in the pattern, a match at the start may fail where a later one
1817 succeeds. Consider, for example:
1818 <pre>
1819 (.*)abc\1
1820 </pre>
1821 If the subject is "xyz123abc123" the match point is the fourth character. For
1822 this reason, such a pattern is not implicitly anchored.
1823 </P>
1824 <P>
1825 Another case where implicit anchoring is not applied is when the leading .* is
1826 inside an atomic group. Once again, a match at the start may fail where a later
1827 one succeeds. Consider this pattern:
1828 <pre>
1829 (?&#62;.*?a)b
1830 </pre>
1831 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1832 (*PRUNE) and (*SKIP) also disable this optimization.
1833 </P>
1834 <P>
1835 When a capturing subpattern is repeated, the value captured is the substring
1836 that matched the final iteration. For example, after
1837 <pre>
1838 (tweedle[dume]{3}\s*)+
1839 </pre>
1840 has matched "tweedledum tweedledee" the value of the captured substring is
1841 "tweedledee". However, if there are nested capturing subpatterns, the
1842 corresponding captured values may have been set in previous iterations. For
1843 example, after
1844 <pre>
1845 /(a|(b))+/
1846 </pre>
1847 matches "aba" the value of the second captured substring is "b".
1848 <a name="atomicgroup"></a></P>
1849 <br><a name="SEC17" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1850 <P>
1851 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1852 repetition, failure of what follows normally causes the repeated item to be
1853 re-evaluated to see if a different number of repeats allows the rest of the
1854 pattern to match. Sometimes it is useful to prevent this, either to change the
1855 nature of the match, or to cause it fail earlier than it otherwise might, when
1856 the author of the pattern knows there is no point in carrying on.
1857 </P>
1858 <P>
1859 Consider, for example, the pattern \d+foo when applied to the subject line
1860 <pre>
1861 123456bar
1862 </pre>
1863 After matching all 6 digits and then failing to match "foo", the normal
1864 action of the matcher is to try again with only 5 digits matching the \d+
1865 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1866 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1867 that once a subpattern has matched, it is not to be re-evaluated in this way.
1868 </P>
1869 <P>
1870 If we use atomic grouping for the previous example, the matcher gives up
1871 immediately on failing to match "foo" the first time. The notation is a kind of
1872 special parenthesis, starting with (?&#62; as in this example:
1873 <pre>
1874 (?&#62;\d+)foo
1875 </pre>
1876 This kind of parenthesis "locks up" the part of the pattern it contains once
1877 it has matched, and a failure further into the pattern is prevented from
1878 backtracking into it. Backtracking past it to previous items, however, works as
1879 normal.
1880 </P>
1881 <P>
1882 An alternative description is that a subpattern of this type matches the string
1883 of characters that an identical standalone pattern would match, if anchored at
1884 the current point in the subject string.
1885 </P>
1886 <P>
1887 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1888 the above example can be thought of as a maximizing repeat that must swallow
1889 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1890 number of digits they match in order to make the rest of the pattern match,
1891 (?&#62;\d+) can only match an entire sequence of digits.
1892 </P>
1893 <P>
1894 Atomic groups in general can of course contain arbitrarily complicated
1895 subpatterns, and can be nested. However, when the subpattern for an atomic
1896 group is just a single repeated item, as in the example above, a simpler
1897 notation, called a "possessive quantifier" can be used. This consists of an
1898 additional + character following a quantifier. Using this notation, the
1899 previous example can be rewritten as
1900 <pre>
1901 \d++foo
1902 </pre>
1903 Note that a possessive quantifier can be used with an entire group, for
1904 example:
1905 <pre>
1906 (abc|xyz){2,3}+
1907 </pre>
1908 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1909 option is ignored. They are a convenient notation for the simpler forms of
1910 atomic group. However, there is no difference in the meaning of a possessive
1911 quantifier and the equivalent atomic group, though there may be a performance
1912 difference; possessive quantifiers should be slightly faster.
1913 </P>
1914 <P>
1915 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1916 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1917 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1918 package, and PCRE copied it from there. It ultimately found its way into Perl
1919 at release 5.10.
1920 </P>
1921 <P>
1922 PCRE has an optimization that automatically "possessifies" certain simple
1923 pattern constructs. For example, the sequence A+B is treated as A++B because
1924 there is no point in backtracking into a sequence of A's when B must follow.
1925 </P>
1926 <P>
1927 When a pattern contains an unlimited repeat inside a subpattern that can itself
1928 be repeated an unlimited number of times, the use of an atomic group is the
1929 only way to avoid some failing matches taking a very long time indeed. The
1930 pattern
1931 <pre>
1932 (\D+|&#60;\d+&#62;)*[!?]
1933 </pre>
1934 matches an unlimited number of substrings that either consist of non-digits, or
1935 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1936 quickly. However, if it is applied to
1937 <pre>
1938 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1939 </pre>
1940 it takes a long time before reporting failure. This is because the string can
1941 be divided between the internal \D+ repeat and the external * repeat in a
1942 large number of ways, and all have to be tried. (The example uses [!?] rather
1943 than a single character at the end, because both PCRE and Perl have an
1944 optimization that allows for fast failure when a single character is used. They
1945 remember the last single character that is required for a match, and fail early
1946 if it is not present in the string.) If the pattern is changed so that it uses
1947 an atomic group, like this:
1948 <pre>
1949 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1950 </pre>
1951 sequences of non-digits cannot be broken, and failure happens quickly.
1952 <a name="backreferences"></a></P>
1953 <br><a name="SEC18" href="#TOC1">BACK REFERENCES</a><br>
1954 <P>
1955 Outside a character class, a backslash followed by a digit greater than 0 (and
1956 possibly further digits) is a back reference to a capturing subpattern earlier
1957 (that is, to its left) in the pattern, provided there have been that many
1958 previous capturing left parentheses.
1959 </P>
1960 <P>
1961 However, if the decimal number following the backslash is less than 10, it is
1962 always taken as a back reference, and causes an error only if there are not
1963 that many capturing left parentheses in the entire pattern. In other words, the
1964 parentheses that are referenced need not be to the left of the reference for
1965 numbers less than 10. A "forward back reference" of this type can make sense
1966 when a repetition is involved and the subpattern to the right has participated
1967 in an earlier iteration.
1968 </P>
1969 <P>
1970 It is not possible to have a numerical "forward back reference" to a subpattern
1971 whose number is 10 or more using this syntax because a sequence such as \50 is
1972 interpreted as a character defined in octal. See the subsection entitled
1973 "Non-printing characters"
1974 <a href="#digitsafterbackslash">above</a>
1975 for further details of the handling of digits following a backslash. There is
1976 no such problem when named parentheses are used. A back reference to any
1977 subpattern is possible using named parentheses (see below).
1978 </P>
1979 <P>
1980 Another way of avoiding the ambiguity inherent in the use of digits following a
1981 backslash is to use the \g escape sequence. This escape must be followed by an
1982 unsigned number or a negative number, optionally enclosed in braces. These
1983 examples are all identical:
1984 <pre>
1985 (ring), \1
1986 (ring), \g1
1987 (ring), \g{1}
1988 </pre>
1989 An unsigned number specifies an absolute reference without the ambiguity that
1990 is present in the older syntax. It is also useful when literal digits follow
1991 the reference. A negative number is a relative reference. Consider this
1992 example:
1993 <pre>
1994 (abc(def)ghi)\g{-1}
1995 </pre>
1996 The sequence \g{-1} is a reference to the most recently started capturing
1997 subpattern before \g, that is, is it equivalent to \2 in this example.
1998 Similarly, \g{-2} would be equivalent to \1. The use of relative references
1999 can be helpful in long patterns, and also in patterns that are created by
2000 joining together fragments that contain references within themselves.
2001 </P>
2002 <P>
2003 A back reference matches whatever actually matched the capturing subpattern in
2004 the current subject string, rather than anything matching the subpattern
2005 itself (see
2006 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
2007 below for a way of doing that). So the pattern
2008 <pre>
2009 (sens|respons)e and \1ibility
2010 </pre>
2011 matches "sense and sensibility" and "response and responsibility", but not
2012 "sense and responsibility". If caseful matching is in force at the time of the
2013 back reference, the case of letters is relevant. For example,
2014 <pre>
2015 ((?i)rah)\s+\1
2016 </pre>
2017 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
2018 capturing subpattern is matched caselessly.
2019 </P>
2020 <P>
2021 There are several different ways of writing back references to named
2022 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
2023 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
2024 back reference syntax, in which \g can be used for both numeric and named
2025 references, is also supported. We could rewrite the above example in any of
2026 the following ways:
2027 <pre>
2028 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
2029 (?'p1'(?i)rah)\s+\k{p1}
2030 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
2031 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
2032 </pre>
2033 A subpattern that is referenced by name may appear in the pattern before or
2034 after the reference.
2035 </P>
2036 <P>
2037 There may be more than one back reference to the same subpattern. If a
2038 subpattern has not actually been used in a particular match, any back
2039 references to it always fail by default. For example, the pattern
2040 <pre>
2041 (a|(bc))\2
2042 </pre>
2043 always fails if it starts to match "a" rather than "bc". However, if the
2044 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
2045 unset value matches an empty string.
2046 </P>
2047 <P>
2048 Because there may be many capturing parentheses in a pattern, all digits
2049 following a backslash are taken as part of a potential back reference number.
2050 If the pattern continues with a digit character, some delimiter must be used to
2051 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
2052 white space. Otherwise, the \g{ syntax or an empty comment (see
2053 <a href="#comments">"Comments"</a>
2054 below) can be used.
2055 </P>
2056 <br><b>
2057 Recursive back references
2058 </b><br>
2059 <P>
2060 A back reference that occurs inside the parentheses to which it refers fails
2061 when the subpattern is first used, so, for example, (a\1) never matches.
2062 However, such references can be useful inside repeated subpatterns. For
2063 example, the pattern
2064 <pre>
2065 (a|b\1)+
2066 </pre>
2067 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
2068 the subpattern, the back reference matches the character string corresponding
2069 to the previous iteration. In order for this to work, the pattern must be such
2070 that the first iteration does not need to match the back reference. This can be
2071 done using alternation, as in the example above, or by a quantifier with a
2072 minimum of zero.
2073 </P>
2074 <P>
2075 Back references of this type cause the group that they reference to be treated
2076 as an
2077 <a href="#atomicgroup">atomic group.</a>
2078 Once the whole group has been matched, a subsequent matching failure cannot
2079 cause backtracking into the middle of the group.
2080 <a name="bigassertions"></a></P>
2081 <br><a name="SEC19" href="#TOC1">ASSERTIONS</a><br>
2082 <P>
2083 An assertion is a test on the characters following or preceding the current
2084 matching point that does not actually consume any characters. The simple
2085 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
2086 <a href="#smallassertions">above.</a>
2087 </P>
2088 <P>
2089 More complicated assertions are coded as subpatterns. There are two kinds:
2090 those that look ahead of the current position in the subject string, and those
2091 that look behind it. An assertion subpattern is matched in the normal way,
2092 except that it does not cause the current matching position to be changed.
2093 </P>
2094 <P>
2095 Assertion subpatterns are not capturing subpatterns. If such an assertion
2096 contains capturing subpatterns within it, these are counted for the purposes of
2097 numbering the capturing subpatterns in the whole pattern. However, substring
2098 capturing is carried out only for positive assertions. (Perl sometimes, but not
2099 always, does do capturing in negative assertions.)
2100 </P>
2101 <P>
2102 For compatibility with Perl, assertion subpatterns may be repeated; though
2103 it makes no sense to assert the same thing several times, the side effect of
2104 capturing parentheses may occasionally be useful. In practice, there only three
2105 cases:
2106 <br>
2107 <br>
2108 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
2109 However, it may contain internal capturing parenthesized groups that are called
2110 from elsewhere via the
2111 <a href="#subpatternsassubroutines">subroutine mechanism.</a>
2112 <br>
2113 <br>
2114 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
2115 were {0,1}. At run time, the rest of the pattern match is tried with and
2116 without the assertion, the order depending on the greediness of the quantifier.
2117 <br>
2118 <br>
2119 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
2120 The assertion is obeyed just once when encountered during matching.
2121 </P>
2122 <br><b>
2123 Lookahead assertions
2124 </b><br>
2125 <P>
2126 Lookahead assertions start with (?= for positive assertions and (?! for
2127 negative assertions. For example,
2128 <pre>
2129 \w+(?=;)
2130 </pre>
2131 matches a word followed by a semicolon, but does not include the semicolon in
2132 the match, and
2133 <pre>
2134 foo(?!bar)
2135 </pre>
2136 matches any occurrence of "foo" that is not followed by "bar". Note that the
2137 apparently similar pattern
2138 <pre>
2139 (?!foo)bar
2140 </pre>
2141 does not find an occurrence of "bar" that is preceded by something other than
2142 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2143 (?!foo) is always true when the next three characters are "bar". A
2144 lookbehind assertion is needed to achieve the other effect.
2145 </P>
2146 <P>
2147 If you want to force a matching failure at some point in a pattern, the most
2148 convenient way to do it is with (?!) because an empty string always matches, so
2149 an assertion that requires there not to be an empty string must always fail.
2150 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2151 <a name="lookbehind"></a></P>
2152 <br><b>
2153 Lookbehind assertions
2154 </b><br>
2155 <P>
2156 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
2157 negative assertions. For example,
2158 <pre>
2159 (?&#60;!foo)bar
2160 </pre>
2161 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2162 a lookbehind assertion are restricted such that all the strings it matches must
2163 have a fixed length. However, if there are several top-level alternatives, they
2164 do not all have to have the same fixed length. Thus
2165 <pre>
2166 (?&#60;=bullock|donkey)
2167 </pre>
2168 is permitted, but
2169 <pre>
2170 (?&#60;!dogs?|cats?)
2171 </pre>
2172 causes an error at compile time. Branches that match different length strings
2173 are permitted only at the top level of a lookbehind assertion. This is an
2174 extension compared with Perl, which requires all branches to match the same
2175 length of string. An assertion such as
2176 <pre>
2177 (?&#60;=ab(c|de))
2178 </pre>
2179 is not permitted, because its single top-level branch can match two different
2180 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2181 branches:
2182 <pre>
2183 (?&#60;=abc|abde)
2184 </pre>
2185 In some cases, the escape sequence \K
2186 <a href="#resetmatchstart">(see above)</a>
2187 can be used instead of a lookbehind assertion to get round the fixed-length
2188 restriction.
2189 </P>
2190 <P>
2191 The implementation of lookbehind assertions is, for each alternative, to
2192 temporarily move the current position back by the fixed length and then try to
2193 match. If there are insufficient characters before the current position, the
2194 assertion fails.
2195 </P>
2196 <P>
2197 In a UTF mode, PCRE does not allow the \C escape (which matches a single data
2198 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2199 it impossible to calculate the length of the lookbehind. The \X and \R
2200 escapes, which can match different numbers of data units, are also not
2201 permitted.
2202 </P>
2203 <P>
2204 <a href="#subpatternsassubroutines">"Subroutine"</a>
2205 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2206 as the subpattern matches a fixed-length string.
2207 <a href="#recursion">Recursion,</a>
2208 however, is not supported.
2209 </P>
2210 <P>
2211 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2212 specify efficient matching of fixed-length strings at the end of subject
2213 strings. Consider a simple pattern such as
2214 <pre>
2215 abcd$
2216 </pre>
2217 when applied to a long string that does not match. Because matching proceeds
2218 from left to right, PCRE will look for each "a" in the subject and then see if
2219 what follows matches the rest of the pattern. If the pattern is specified as
2220 <pre>
2221 ^.*abcd$
2222 </pre>
2223 the initial .* matches the entire string at first, but when this fails (because
2224 there is no following "a"), it backtracks to match all but the last character,
2225 then all but the last two characters, and so on. Once again the search for "a"
2226 covers the entire string, from right to left, so we are no better off. However,
2227 if the pattern is written as
2228 <pre>
2229 ^.*+(?&#60;=abcd)
2230 </pre>
2231 there can be no backtracking for the .*+ item; it can match only the entire
2232 string. The subsequent lookbehind assertion does a single test on the last four
2233 characters. If it fails, the match fails immediately. For long strings, this
2234 approach makes a significant difference to the processing time.
2235 </P>
2236 <br><b>
2237 Using multiple assertions
2238 </b><br>
2239 <P>
2240 Several assertions (of any sort) may occur in succession. For example,
2241 <pre>
2242 (?&#60;=\d{3})(?&#60;!999)foo
2243 </pre>
2244 matches "foo" preceded by three digits that are not "999". Notice that each of
2245 the assertions is applied independently at the same point in the subject
2246 string. First there is a check that the previous three characters are all
2247 digits, and then there is a check that the same three characters are not "999".
2248 This pattern does <i>not</i> match "foo" preceded by six characters, the first
2249 of which are digits and the last three of which are not "999". For example, it
2250 doesn't match "123abcfoo". A pattern to do that is
2251 <pre>
2252 (?&#60;=\d{3}...)(?&#60;!999)foo
2253 </pre>
2254 This time the first assertion looks at the preceding six characters, checking
2255 that the first three are digits, and then the second assertion checks that the
2256 preceding three characters are not "999".
2257 </P>
2258 <P>
2259 Assertions can be nested in any combination. For example,
2260 <pre>
2261 (?&#60;=(?&#60;!foo)bar)baz
2262 </pre>
2263 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2264 preceded by "foo", while
2265 <pre>
2266 (?&#60;=\d{3}(?!999)...)foo
2267 </pre>
2268 is another pattern that matches "foo" preceded by three digits and any three
2269 characters that are not "999".
2270 <a name="conditions"></a></P>
2271 <br><a name="SEC20" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
2272 <P>
2273 It is possible to cause the matching process to obey a subpattern
2274 conditionally or to choose between two alternative subpatterns, depending on
2275 the result of an assertion, or whether a specific capturing subpattern has
2276 already been matched. The two possible forms of conditional subpattern are:
2277 <pre>
2278 (?(condition)yes-pattern)
2279 (?(condition)yes-pattern|no-pattern)
2280 </pre>
2281 If the condition is satisfied, the yes-pattern is used; otherwise the
2282 no-pattern (if present) is used. If there are more than two alternatives in the
2283 subpattern, a compile-time error occurs. Each of the two alternatives may
2284 itself contain nested subpatterns of any form, including conditional
2285 subpatterns; the restriction to two alternatives applies only at the level of
2286 the condition. This pattern fragment is an example where the alternatives are
2287 complex:
2288 <pre>
2289 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2291 </PRE>
2292 </P>
2293 <P>
2294 There are four kinds of condition: references to subpatterns, references to
2295 recursion, a pseudo-condition called DEFINE, and assertions.
2296 </P>
2297 <br><b>
2298 Checking for a used subpattern by number
2299 </b><br>
2300 <P>
2301 If the text between the parentheses consists of a sequence of digits, the
2302 condition is true if a capturing subpattern of that number has previously
2303 matched. If there is more than one capturing subpattern with the same number
2304 (see the earlier
2305 <a href="#recursion">section about duplicate subpattern numbers),</a>
2306 the condition is true if any of them have matched. An alternative notation is
2307 to precede the digits with a plus or minus sign. In this case, the subpattern
2308 number is relative rather than absolute. The most recently opened parentheses
2309 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2310 loops it can also make sense to refer to subsequent groups. The next
2311 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2312 zero in any of these forms is not used; it provokes a compile-time error.)
2313 </P>
2314 <P>
2315 Consider the following pattern, which contains non-significant white space to
2316 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2317 three parts for ease of discussion:
2318 <pre>
2319 ( \( )? [^()]+ (?(1) \) )
2320 </pre>
2321 The first part matches an optional opening parenthesis, and if that
2322 character is present, sets it as the first captured substring. The second part
2323 matches one or more characters that are not parentheses. The third part is a
2324 conditional subpattern that tests whether or not the first set of parentheses
2325 matched. If they did, that is, if subject started with an opening parenthesis,
2326 the condition is true, and so the yes-pattern is executed and a closing
2327 parenthesis is required. Otherwise, since no-pattern is not present, the
2328 subpattern matches nothing. In other words, this pattern matches a sequence of
2329 non-parentheses, optionally enclosed in parentheses.
2330 </P>
2331 <P>
2332 If you were embedding this pattern in a larger one, you could use a relative
2333 reference:
2334 <pre>
2335 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
2336 </pre>
2337 This makes the fragment independent of the parentheses in the larger pattern.
2338 </P>
2339 <br><b>
2340 Checking for a used subpattern by name
2341 </b><br>
2342 <P>
2343 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
2344 subpattern by name. For compatibility with earlier versions of PCRE, which had
2345 this facility before Perl, the syntax (?(name)...) is also recognized.
2346 </P>
2347 <P>
2348 Rewriting the above example to use a named subpattern gives this:
2349 <pre>
2350 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
2351 </pre>
2352 If the name used in a condition of this kind is a duplicate, the test is
2353 applied to all subpatterns of the same name, and is true if any one of them has
2354 matched.
2355 </P>
2356 <br><b>
2357 Checking for pattern recursion
2358 </b><br>
2359 <P>
2360 If the condition is the string (R), and there is no subpattern with the name R,
2361 the condition is true if a recursive call to the whole pattern or any
2362 subpattern has been made. If digits or a name preceded by ampersand follow the
2363 letter R, for example:
2364 <pre>
2365 (?(R3)...) or (?(R&name)...)
2366 </pre>
2367 the condition is true if the most recent recursion is into a subpattern whose
2368 number or name is given. This condition does not check the entire recursion
2369 stack. If the name used in a condition of this kind is a duplicate, the test is
2370 applied to all subpatterns of the same name, and is true if any one of them is
2371 the most recent recursion.
2372 </P>
2373 <P>
2374 At "top level", all these recursion test conditions are false.
2375 <a href="#recursion">The syntax for recursive patterns</a>
2376 is described below.
2377 <a name="subdefine"></a></P>
2378 <br><b>
2379 Defining subpatterns for use by reference only
2380 </b><br>
2381 <P>
2382 If the condition is the string (DEFINE), and there is no subpattern with the
2383 name DEFINE, the condition is always false. In this case, there may be only one
2384 alternative in the subpattern. It is always skipped if control reaches this
2385 point in the pattern; the idea of DEFINE is that it can be used to define
2386 subroutines that can be referenced from elsewhere. (The use of
2387 <a href="#subpatternsassubroutines">subroutines</a>
2388 is described below.) For example, a pattern to match an IPv4 address such as
2389 "" could be written like this (ignore white space and line
2390 breaks):
2391 <pre>
2392 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
2393 \b (?&byte) (\.(?&byte)){3} \b
2394 </pre>
2395 The first part of the pattern is a DEFINE group inside which a another group
2396 named "byte" is defined. This matches an individual component of an IPv4
2397 address (a number less than 256). When matching takes place, this part of the
2398 pattern is skipped because DEFINE acts like a false condition. The rest of the
2399 pattern uses references to the named group to match the four dot-separated
2400 components of an IPv4 address, insisting on a word boundary at each end.
2401 </P>
2402 <br><b>
2403 Assertion conditions
2404 </b><br>
2405 <P>
2406 If the condition is not in any of the above formats, it must be an assertion.
2407 This may be a positive or negative lookahead or lookbehind assertion. Consider
2408 this pattern, again containing non-significant white space, and with the two
2409 alternatives on the second line:
2410 <pre>
2411 (?(?=[^a-z]*[a-z])
2412 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
2413 </pre>
2414 The condition is a positive lookahead assertion that matches an optional
2415 sequence of non-letters followed by a letter. In other words, it tests for the
2416 presence of at least one letter in the subject. If a letter is found, the
2417 subject is matched against the first alternative; otherwise it is matched
2418 against the second. This pattern matches strings in one of the two forms
2419 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2420 <a name="comments"></a></P>
2421 <br><a name="SEC21" href="#TOC1">COMMENTS</a><br>
2422 <P>
2423 There are two ways of including comments in patterns that are processed by
2424 PCRE. In both cases, the start of the comment must not be in a character class,
2425 nor in the middle of any other sequence of related characters such as (?: or a
2426 subpattern name or number. The characters that make up a comment play no part
2427 in the pattern matching.
2428 </P>
2429 <P>
2430 The sequence (?# marks the start of a comment that continues up to the next
2431 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2432 option is set, an unescaped # character also introduces a comment, which in
2433 this case continues to immediately after the next newline character or
2434 character sequence in the pattern. Which characters are interpreted as newlines
2435 is controlled by the options passed to a compiling function or by a special
2436 sequence at the start of the pattern, as described in the section entitled
2437 <a href="#newlines">"Newline conventions"</a>
2438 above. Note that the end of this type of comment is a literal newline sequence
2439 in the pattern; escape sequences that happen to represent a newline do not
2440 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2441 default newline convention is in force:
2442 <pre>
2443 abc #comment \n still comment
2444 </pre>
2445 On encountering the # character, <b>pcre_compile()</b> skips along, looking for
2446 a newline in the pattern. The sequence \n is still literal at this stage, so
2447 it does not terminate the comment. Only an actual character with the code value
2448 0x0a (the default newline) does so.
2449 <a name="recursion"></a></P>
2450 <br><a name="SEC22" href="#TOC1">RECURSIVE PATTERNS</a><br>
2451 <P>
2452 Consider the problem of matching a string in parentheses, allowing for
2453 unlimited nested parentheses. Without the use of recursion, the best that can
2454 be done is to use a pattern that matches up to some fixed depth of nesting. It
2455 is not possible to handle an arbitrary nesting depth.
2456 </P>
2457 <P>
2458 For some time, Perl has provided a facility that allows regular expressions to
2459 recurse (amongst other things). It does this by interpolating Perl code in the
2460 expression at run time, and the code can refer to the expression itself. A Perl
2461 pattern using code interpolation to solve the parentheses problem can be
2462 created like this:
2463 <pre>
2464 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
2465 </pre>
2466 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2467 recursively to the pattern in which it appears.
2468 </P>
2469 <P>
2470 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2471 supports special syntax for recursion of the entire pattern, and also for
2472 individual subpattern recursion. After its introduction in PCRE and Python,
2473 this kind of recursion was subsequently introduced into Perl at release 5.10.
2474 </P>
2475 <P>
2476 A special item that consists of (? followed by a number greater than zero and a
2477 closing parenthesis is a recursive subroutine call of the subpattern of the
2478 given number, provided that it occurs inside that subpattern. (If not, it is a
2479 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2480 call, which is described in the next section.) The special item (?R) or (?0) is
2481 a recursive call of the entire regular expression.
2482 </P>
2483 <P>
2484 This PCRE pattern solves the nested parentheses problem (assume the
2485 PCRE_EXTENDED option is set so that white space is ignored):
2486 <pre>
2487 \( ( [^()]++ | (?R) )* \)
2488 </pre>
2489 First it matches an opening parenthesis. Then it matches any number of
2490 substrings which can either be a sequence of non-parentheses, or a recursive
2491 match of the pattern itself (that is, a correctly parenthesized substring).
2492 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2493 to avoid backtracking into sequences of non-parentheses.
2494 </P>
2495 <P>
2496 If this were part of a larger pattern, you would not want to recurse the entire
2497 pattern, so instead you could use this:
2498 <pre>
2499 ( \( ( [^()]++ | (?1) )* \) )
2500 </pre>
2501 We have put the pattern into parentheses, and caused the recursion to refer to
2502 them instead of the whole pattern.
2503 </P>
2504 <P>
2505 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2506 is made easier by the use of relative references. Instead of (?1) in the
2507 pattern above you can write (?-2) to refer to the second most recently opened
2508 parentheses preceding the recursion. In other words, a negative number counts
2509 capturing parentheses leftwards from the point at which it is encountered.
2510 </P>
2511 <P>
2512 It is also possible to refer to subsequently opened parentheses, by writing
2513 references such as (?+2). However, these cannot be recursive because the
2514 reference is not inside the parentheses that are referenced. They are always
2515 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2516 calls, as described in the next section.
2517 </P>
2518 <P>
2519 An alternative approach is to use named parentheses instead. The Perl syntax
2520 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
2521 could rewrite the above example as follows:
2522 <pre>
2523 (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
2524 </pre>
2525 If there is more than one subpattern with the same name, the earliest one is
2526 used.
2527 </P>
2528 <P>
2529 This particular example pattern that we have been looking at contains nested
2530 unlimited repeats, and so the use of a possessive quantifier for matching
2531 strings of non-parentheses is important when applying the pattern to strings
2532 that do not match. For example, when this pattern is applied to
2533 <pre>
2534 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2535 </pre>
2536 it yields "no match" quickly. However, if a possessive quantifier is not used,
2537 the match runs for a very long time indeed because there are so many different
2538 ways the + and * repeats can carve up the subject, and all have to be tested
2539 before failure can be reported.
2540 </P>
2541 <P>
2542 At the end of a match, the values of capturing parentheses are those from
2543 the outermost level. If you want to obtain intermediate values, a callout
2544 function can be used (see below and the
2545 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2546 documentation). If the pattern above is matched against
2547 <pre>
2548 (ab(cd)ef)
2549 </pre>
2550 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2551 the last value taken on at the top level. If a capturing subpattern is not
2552 matched at the top level, its final captured value is unset, even if it was
2553 (temporarily) set at a deeper level during the matching process.
2554 </P>
2555 <P>
2556 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2557 obtain extra memory to store data during a recursion, which it does by using
2558 <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
2559 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2560 </P>
2561 <P>
2562 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2563 Consider this pattern, which matches text in angle brackets, allowing for
2564 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2565 recursing), whereas any characters are permitted at the outer level.
2566 <pre>
2567 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
2568 </pre>
2569 In this pattern, (?(R) is the start of a conditional subpattern, with two
2570 different alternatives for the recursive and non-recursive cases. The (?R) item
2571 is the actual recursive call.
2572 <a name="recursiondifference"></a></P>
2573 <br><b>
2574 Differences in recursion processing between PCRE and Perl
2575 </b><br>
2576 <P>
2577 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2578 (like Python, but unlike Perl), a recursive subpattern call is always treated
2579 as an atomic group. That is, once it has matched some of the subject string, it
2580 is never re-entered, even if it contains untried alternatives and there is a
2581 subsequent matching failure. This can be illustrated by the following pattern,
2582 which purports to match a palindromic string that contains an odd number of
2583 characters (for example, "a", "aba", "abcba", "abcdcba"):
2584 <pre>
2585 ^(.|(.)(?1)\2)$
2586 </pre>
2587 The idea is that it either matches a single character, or two identical
2588 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2589 it does not if the pattern is longer than three characters. Consider the
2590 subject string "abcba":
2591 </P>
2592 <P>
2593 At the top level, the first character is matched, but as it is not at the end
2594 of the string, the first alternative fails; the second alternative is taken
2595 and the recursion kicks in. The recursive call to subpattern 1 successfully
2596 matches the next character ("b"). (Note that the beginning and end of line
2597 tests are not part of the recursion).
2598 </P>
2599 <P>
2600 Back at the top level, the next character ("c") is compared with what
2601 subpattern 2 matched, which was "a". This fails. Because the recursion is
2602 treated as an atomic group, there are now no backtracking points, and so the
2603 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2604 try the second alternative.) However, if the pattern is written with the
2605 alternatives in the other order, things are different:
2606 <pre>
2607 ^((.)(?1)\2|.)$
2608 </pre>
2609 This time, the recursing alternative is tried first, and continues to recurse
2610 until it runs out of characters, at which point the recursion fails. But this
2611 time we do have another alternative to try at the higher level. That is the big
2612 difference: in the previous case the remaining alternative is at a deeper
2613 recursion level, which PCRE cannot use.
2614 </P>
2615 <P>
2616 To change the pattern so that it matches all palindromic strings, not just
2617 those with an odd number of characters, it is tempting to change the pattern to
2618 this:
2619 <pre>
2620 ^((.)(?1)\2|.?)$
2621 </pre>
2622 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2623 deeper recursion has matched a single character, it cannot be entered again in
2624 order to match an empty string. The solution is to separate the two cases, and
2625 write out the odd and even cases as alternatives at the higher level:
2626 <pre>
2627 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
2628 </pre>
2629 If you want to match typical palindromic phrases, the pattern has to ignore all
2630 non-word characters, which can be done like this:
2631 <pre>
2632 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
2633 </pre>
2634 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2635 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2636 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2637 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2638 more) to match typical phrases, and Perl takes so long that you think it has
2639 gone into a loop.
2640 </P>
2641 <P>
2642 <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
2643 string does not start with a palindrome that is shorter than the entire string.
2644 For example, although "abcba" is correctly matched, if the subject is "ababa",
2645 PCRE finds the palindrome "aba" at the start, then fails at top level because
2646 the end of the string does not follow. Once again, it cannot jump back into the
2647 recursion to try other alternatives, so the entire match fails.
2648 </P>
2649 <P>
2650 The second way in which PCRE and Perl differ in their recursion processing is
2651 in the handling of captured values. In Perl, when a subpattern is called
2652 recursively or as a subpattern (see the next section), it has no access to any
2653 values that were captured outside the recursion, whereas in PCRE these values
2654 can be referenced. Consider this pattern:
2655 <pre>
2656 ^(.)(\1|a(?2))
2657 </pre>
2658 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2659 then in the second group, when the back reference \1 fails to match "b", the
2660 second alternative matches "a" and then recurses. In the recursion, \1 does
2661 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2662 match because inside the recursive call \1 cannot access the externally set
2663 value.
2664 <a name="subpatternsassubroutines"></a></P>
2665 <br><a name="SEC23" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
2666 <P>
2667 If the syntax for a recursive subpattern call (either by number or by
2668 name) is used outside the parentheses to which it refers, it operates like a
2669 subroutine in a programming language. The called subpattern may be defined
2670 before or after the reference. A numbered reference can be absolute or
2671 relative, as in these examples:
2672 <pre>
2673 (...(absolute)...)...(?2)...
2674 (...(relative)...)...(?-1)...
2675 (...(?+1)...(relative)...
2676 </pre>
2677 An earlier example pointed out that the pattern
2678 <pre>
2679 (sens|respons)e and \1ibility
2680 </pre>
2681 matches "sense and sensibility" and "response and responsibility", but not
2682 "sense and responsibility". If instead the pattern
2683 <pre>
2684 (sens|respons)e and (?1)ibility
2685 </pre>
2686 is used, it does match "sense and responsibility" as well as the other two
2687 strings. Another example is given in the discussion of DEFINE above.
2688 </P>
2689 <P>
2690 All subroutine calls, whether recursive or not, are always treated as atomic
2691 groups. That is, once a subroutine has matched some of the subject string, it
2692 is never re-entered, even if it contains untried alternatives and there is a
2693 subsequent matching failure. Any capturing parentheses that are set during the
2694 subroutine call revert to their previous values afterwards.
2695 </P>
2696 <P>
2697 Processing options such as case-independence are fixed when a subpattern is
2698 defined, so if it is used as a subroutine, such options cannot be changed for
2699 different calls. For example, consider this pattern:
2700 <pre>
2701 (abc)(?i:(?-1))
2702 </pre>
2703 It matches "abcabc". It does not match "abcABC" because the change of
2704 processing option does not affect the called subpattern.
2705 <a name="onigurumasubroutines"></a></P>
2706 <br><a name="SEC24" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
2707 <P>
2708 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
2709 a number enclosed either in angle brackets or single quotes, is an alternative
2710 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2711 are two of the examples used above, rewritten using this syntax:
2712 <pre>
2713 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
2714 (sens|respons)e and \g'1'ibility
2715 </pre>
2716 PCRE supports an extension to Oniguruma: if a number is preceded by a
2717 plus or a minus sign it is taken as a relative reference. For example:
2718 <pre>
2719 (abc)(?i:\g&#60;-1&#62;)
2720 </pre>
2721 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
2722 synonymous. The former is a back reference; the latter is a subroutine call.
2723 </P>
2724 <br><a name="SEC25" href="#TOC1">CALLOUTS</a><br>
2725 <P>
2726 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2727 code to be obeyed in the middle of matching a regular expression. This makes it
2728 possible, amongst other things, to extract different substrings that match the
2729 same pair of parentheses when there is a repetition.
2730 </P>
2731 <P>
2732 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2733 code. The feature is called "callout". The caller of PCRE provides an external
2734 function by putting its entry point in the global variable <i>pcre_callout</i>
2735 (8-bit library) or <i>pcre[16|32]_callout</i> (16-bit or 32-bit library).
2736 By default, this variable contains NULL, which disables all calling out.
2737 </P>
2738 <P>
2739 Within a regular expression, (?C) indicates the points at which the external
2740 function is to be called. If you want to identify different callout points, you
2741 can put a number less than 256 after the letter C. The default value is zero.
2742 For example, this pattern has two callout points:
2743 <pre>
2744 (?C1)abc(?C2)def
2745 </pre>
2746 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2747 automatically installed before each item in the pattern. They are all numbered
2748 255. If there is a conditional group in the pattern whose condition is an
2749 assertion, an additional callout is inserted just before the condition. An
2750 explicit callout may also be set at this position, as in this example:
2751 <pre>
2752 (?(?C9)(?=a)abc|def)
2753 </pre>
2754 Note that this applies only to assertion conditions, not to other types of
2755 condition.
2756 </P>
2757 <P>
2758 During matching, when PCRE reaches a callout point, the external function is
2759 called. It is provided with the number of the callout, the position in the
2760 pattern, and, optionally, one item of data originally supplied by the caller of
2761 the matching function. The callout function may cause matching to proceed, to
2762 backtrack, or to fail altogether.
2763 </P>
2764 <P>
2765 By default, PCRE implements a number of optimizations at compile time and
2766 matching time, and one side-effect is that sometimes callouts are skipped. If
2767 you need all possible callouts to happen, you need to set options that disable
2768 the relevant optimizations. More details, and a complete description of the
2769 interface to the callout function, are given in the
2770 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2771 documentation.
2772 <a name="backtrackcontrol"></a></P>
2773 <br><a name="SEC26" href="#TOC1">BACKTRACKING CONTROL</a><br>
2774 <P>
2775 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2776 are still described in the Perl documentation as "experimental and subject to
2777 change or removal in a future version of Perl". It goes on to say: "Their usage
2778 in production code should be noted to avoid problems during upgrades." The same
2779 remarks apply to the PCRE features described in this section.
2780 </P>
2781 <P>
2782 The new verbs make use of what was previously invalid syntax: an opening
2783 parenthesis followed by an asterisk. They are generally of the form
2784 (*VERB) or (*VERB:NAME). Some may take either form, possibly behaving
2785 differently depending on whether or not a name is present. A name is any
2786 sequence of characters that does not include a closing parenthesis. The maximum
2787 length of name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit
2788 libraries. If the name is empty, that is, if the closing parenthesis
2789 immediately follows the colon, the effect is as if the colon were not there.
2790 Any number of these verbs may occur in a pattern.
2791 </P>
2792 <P>
2793 Since these verbs are specifically related to backtracking, most of them can be
2794 used only when the pattern is to be matched using one of the traditional
2795 matching functions, because these use a backtracking algorithm. With the
2796 exception of (*FAIL), which behaves like a failing negative assertion, the
2797 backtracking control verbs cause an error if encountered by a DFA matching
2798 function.
2799 </P>
2800 <P>
2801 The behaviour of these verbs in
2802 <a href="#btrepeat">repeated groups,</a>
2803 <a href="#btassert">assertions,</a>
2804 and in
2805 <a href="#btsub">subpatterns called as subroutines</a>
2806 (whether or not recursively) is documented below.
2807 <a name="nooptimize"></a></P>
2808 <br><b>
2809 Optimizations that affect backtracking verbs
2810 </b><br>
2811 <P>
2812 PCRE contains some optimizations that are used to speed up matching by running
2813 some checks at the start of each match attempt. For example, it may know the
2814 minimum length of matching subject, or that a particular character must be
2815 present. When one of these optimizations bypasses the running of a match, any
2816 included backtracking verbs will not, of course, be processed. You can suppress
2817 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2818 when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the
2819 pattern with (*NO_START_OPT). There is more discussion of this option in the
2820 section entitled
2821 <a href="pcreapi.html#execoptions">"Option bits for <b>pcre_exec()</b>"</a>
2822 in the
2823 <a href="pcreapi.html"><b>pcreapi</b></a>
2824 documentation.
2825 </P>
2826 <P>
2827 Experiments with Perl suggest that it too has similar optimizations, sometimes
2828 leading to anomalous results.
2829 </P>
2830 <br><b>
2831 Verbs that act immediately
2832 </b><br>
2833 <P>
2834 The following verbs act as soon as they are encountered. They may not be
2835 followed by a name.
2836 <pre>
2837 (*ACCEPT)
2838 </pre>
2839 This verb causes the match to end successfully, skipping the remainder of the
2840 pattern. However, when it is inside a subpattern that is called as a
2841 subroutine, only that subpattern is ended successfully. Matching then continues
2842 at the outer level. If (*ACCEPT) in triggered in a positive assertion, the
2843 assertion succeeds; in a negative assertion, the assertion fails.
2844 </P>
2845 <P>
2846 If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For
2847 example:
2848 <pre>
2849 A((?:A|B(*ACCEPT)|C)D)
2850 </pre>
2851 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2852 the outer parentheses.
2853 <pre>
2854 (*FAIL) or (*F)
2855 </pre>
2856 This verb causes a matching failure, forcing backtracking to occur. It is
2857 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2858 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2859 Perl features that are not present in PCRE. The nearest equivalent is the
2860 callout feature, as for example in this pattern:
2861 <pre>
2862 a+(?C)(*FAIL)
2863 </pre>
2864 A match with the string "aaaa" always fails, but the callout is taken before
2865 each backtrack happens (in this example, 10 times).
2866 </P>
2867 <br><b>
2868 Recording which path was taken
2869 </b><br>
2870 <P>
2871 There is one verb whose main purpose is to track how a match was arrived at,
2872 though it also has a secondary use in conjunction with advancing the match
2873 starting point (see (*SKIP) below).
2874 <pre>
2875 (*MARK:NAME) or (*:NAME)
2876 </pre>
2877 A name is always required with this verb. There may be as many instances of
2878 (*MARK) as you like in a pattern, and their names do not have to be unique.
2879 </P>
2880 <P>
2881 When a match succeeds, the name of the last-encountered (*MARK:NAME),
2882 (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the
2883 caller as described in the section entitled
2884 <a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a>
2885 in the
2886 <a href="pcreapi.html"><b>pcreapi</b></a>
2887 documentation. Here is an example of <b>pcretest</b> output, where the /K
2888 modifier requests the retrieval and outputting of (*MARK) data:
2889 <pre>
2890 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2891 data&#62; XY
2892 0: XY
2893 MK: A
2894 XZ
2895 0: XZ
2896 MK: B
2897 </pre>
2898 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2899 indicates which of the two alternatives matched. This is a more efficient way
2900 of obtaining this information than putting each alternative in its own
2901 capturing parentheses.
2902 </P>
2903 <P>
2904 If a verb with a name is encountered in a positive assertion that is true, the
2905 name is recorded and passed back if it is the last-encountered. This does not
2906 happen for negative assertions or failing positive assertions.
2907 </P>
2908 <P>
2909 After a partial match or a failed match, the last encountered name in the
2910 entire match process is returned. For example:
2911 <pre>
2912 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2913 data&#62; XP
2914 No match, mark = B
2915 </pre>
2916 Note that in this unanchored example the mark is retained from the match
2917 attempt that started at the letter "X" in the subject. Subsequent match
2918 attempts starting at "P" and then with an empty string do not get as far as the
2919 (*MARK) item, but nevertheless do not reset it.
2920 </P>
2921 <P>
2922 If you are interested in (*MARK) values after failed matches, you should
2923 probably set the PCRE_NO_START_OPTIMIZE option
2924 <a href="#nooptimize">(see above)</a>
2925 to ensure that the match is always attempted.
2926 </P>
2927 <br><b>
2928 Verbs that act after backtracking
2929 </b><br>
2930 <P>
2931 The following verbs do nothing when they are encountered. Matching continues
2932 with what follows, but if there is no subsequent match, causing a backtrack to
2933 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2934 the verb. However, when one of these verbs appears inside an atomic group or an
2935 assertion that is true, its effect is confined to that group, because once the
2936 group has been matched, there is never any backtracking into it. In this
2937 situation, backtracking can "jump back" to the left of the entire atomic group
2938 or assertion. (Remember also, as stated above, that this localization also
2939 applies in subroutine calls.)
2940 </P>
2941 <P>
2942 These verbs differ in exactly what kind of failure occurs when backtracking
2943 reaches them. The behaviour described below is what happens when the verb is
2944 not in a subroutine or an assertion. Subsequent sections cover these special
2945 cases.
2946 <pre>
2947 (*COMMIT)
2948 </pre>
2949 This verb, which may not be followed by a name, causes the whole match to fail
2950 outright if there is a later matching failure that causes backtracking to reach
2951 it. Even if the pattern is unanchored, no further attempts to find a match by
2952 advancing the starting point take place. If (*COMMIT) is the only backtracking
2953 verb that is encountered, once it has been passed <b>pcre_exec()</b> is
2954 committed to finding a match at the current starting point, or not at all. For
2955 example:
2956 <pre>
2957 a+(*COMMIT)b
2958 </pre>
2959 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2960 dynamic anchor, or "I've started, so I must finish." The name of the most
2961 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2962 match failure.
2963 </P>
2964 <P>
2965 If there is more than one backtracking verb in a pattern, a different one that
2966 follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a
2967 match does not always guarantee that a match must be at this starting point.
2968 </P>
2969 <P>
2970 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2971 unless PCRE's start-of-match optimizations are turned off, as shown in this
2972 <b>pcretest</b> example:
2973 <pre>
2974 re&#62; /(*COMMIT)abc/
2975 data&#62; xyzabc
2976 0: abc
2977 xyzabc\Y
2978 No match
2979 </pre>
2980 PCRE knows that any match must start with "a", so the optimization skips along
2981 the subject to "a" before running the first match attempt, which succeeds. When
2982 the optimization is disabled by the \Y escape in the second subject, the match
2983 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2984 starting points.
2985 <pre>
2986 (*PRUNE) or (*PRUNE:NAME)
2987 </pre>
2988 This verb causes the match to fail at the current starting position in the
2989 subject if there is a later matching failure that causes backtracking to reach
2990 it. If the pattern is unanchored, the normal "bumpalong" advance to the next
2991 starting character then happens. Backtracking can occur as usual to the left of
2992 (*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but
2993 if there is no match to the right, backtracking cannot cross (*PRUNE). In
2994 simple cases, the use of (*PRUNE) is just an alternative to an atomic group or
2995 possessive quantifier, but there are some uses of (*PRUNE) that cannot be
2996 expressed in any other way. In an anchored pattern (*PRUNE) has the same effect
2997 as (*COMMIT).
2998 </P>
2999 <P>
3000 The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE).
3001 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3002 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3003 <pre>
3004 (*SKIP)
3005 </pre>
3006 This verb, when given without a name, is like (*PRUNE), except that if the
3007 pattern is unanchored, the "bumpalong" advance is not to the next character,
3008 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
3009 signifies that whatever text was matched leading up to it cannot be part of a
3010 successful match. Consider:
3011 <pre>
3012 a+(*SKIP)b
3013 </pre>
3014 If the subject is "aaaac...", after the first match attempt fails (starting at
3015 the first character in the string), the starting point skips on to start the
3016 next attempt at "c". Note that a possessive quantifer does not have the same
3017 effect as this example; although it would suppress backtracking during the
3018 first match attempt, the second attempt would start at the second character
3019 instead of skipping on to "c".
3020 <pre>
3021 (*SKIP:NAME)
3022 </pre>
3023 When (*SKIP) has an associated name, its behaviour is modified. When it is
3024 triggered, the previous path through the pattern is searched for the most
3025 recent (*MARK) that has the same name. If one is found, the "bumpalong" advance
3026 is to the subject position that corresponds to that (*MARK) instead of to where
3027 (*SKIP) was encountered. If no (*MARK) with a matching name is found, the
3028 (*SKIP) is ignored.
3029 </P>
3030 <P>
3031 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores
3032 names that are set by (*PRUNE:NAME) or (*THEN:NAME).
3033 <pre>
3034 (*THEN) or (*THEN:NAME)
3035 </pre>
3036 This verb causes a skip to the next innermost alternative when backtracking
3037 reaches it. That is, it cancels any further backtracking within the current
3038 alternative. Its name comes from the observation that it can be used for a
3039 pattern-based if-then-else block:
3040 <pre>
3041 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
3042 </pre>
3043 If the COND1 pattern matches, FOO is tried (and possibly further items after
3044 the end of the group if FOO succeeds); on failure, the matcher skips to the
3045 second alternative and tries COND2, without backtracking into COND1. If that
3046 succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no
3047 more alternatives, so there is a backtrack to whatever came before the entire
3048 group. If (*THEN) is not inside an alternation, it acts like (*PRUNE).
3049 </P>
3050 <P>
3051 The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN).
3052 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3053 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3054 </P>
3055 <P>
3056 A subpattern that does not contain a | character is just a part of the
3057 enclosing alternative; it is not a nested alternation with only one
3058 alternative. The effect of (*THEN) extends beyond such a subpattern to the
3059 enclosing alternative. Consider this pattern, where A, B, etc. are complex
3060 pattern fragments that do not contain any | characters at this level:
3061 <pre>
3062 A (B(*THEN)C) | D
3063 </pre>
3064 If A and B are matched, but there is a failure in C, matching does not
3065 backtrack into A; instead it moves to the next alternative, that is, D.
3066 However, if the subpattern containing (*THEN) is given an alternative, it
3067 behaves differently:
3068 <pre>
3069 A (B(*THEN)C | (*FAIL)) | D
3070 </pre>
3071 The effect of (*THEN) is now confined to the inner subpattern. After a failure
3072 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
3073 because there are no more alternatives to try. In this case, matching does now
3074 backtrack into A.
3075 </P>
3076 <P>
3077 Note that a conditional subpattern is not considered as having two
3078 alternatives, because only one is ever used. In other words, the | character in
3079 a conditional subpattern has a different meaning. Ignoring white space,
3080 consider:
3081 <pre>
3082 ^.*? (?(?=a) a | b(*THEN)c )
3083 </pre>
3084 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
3085 it initially matches zero characters. The condition (?=a) then fails, the
3086 character "b" is matched, but "c" is not. At this point, matching does not
3087 backtrack to .*? as might perhaps be expected from the presence of the |
3088 character. The conditional subpattern is part of the single alternative that
3089 comprises the whole pattern, and so the match fails. (If there was a backtrack
3090 into .*?, allowing it to match "b", the match would succeed.)
3091 </P>
3092 <P>
3093 The verbs just described provide four different "strengths" of control when
3094 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
3095 next alternative. (*PRUNE) comes next, failing the match at the current
3096 starting position, but allowing an advance to the next character (for an
3097 unanchored pattern). (*SKIP) is similar, except that the advance may be more
3098 than one character. (*COMMIT) is the strongest, causing the entire match to
3099 fail.
3100 </P>
3101 <br><b>
3102 More than one backtracking verb
3103 </b><br>
3104 <P>
3105 If more than one backtracking verb is present in a pattern, the one that is
3106 backtracked onto first acts. For example, consider this pattern, where A, B,
3107 etc. are complex pattern fragments:
3108 <pre>
3110 </pre>
3111 If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to
3112 fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes
3113 the next alternative (ABD) to be tried. This behaviour is consistent, but is
3114 not always the same as Perl's. It means that if two or more backtracking verbs
3115 appear in succession, all the the last of them has no effect. Consider this
3116 example:
3117 <pre>
3118 ...(*COMMIT)(*PRUNE)...
3119 </pre>
3120 If there is a matching failure to the right, backtracking onto (*PRUNE) causes
3121 it to be triggered, and its action is taken. There can never be a backtrack
3122 onto (*COMMIT).
3123 <a name="btrepeat"></a></P>
3124 <br><b>
3125 Backtracking verbs in repeated groups
3126 </b><br>
3127 <P>
3128 PCRE differs from Perl in its handling of backtracking verbs in repeated
3129 groups. For example, consider:
3130 <pre>
3131 /(a(*COMMIT)b)+ac/
3132 </pre>
3133 If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in
3134 the second repeat of the group acts.
3135 <a name="btassert"></a></P>
3136 <br><b>
3137 Backtracking verbs in assertions
3138 </b><br>
3139 <P>
3140 (*FAIL) in an assertion has its normal effect: it forces an immediate backtrack.
3141 </P>
3142 <P>
3143 (*ACCEPT) in a positive assertion causes the assertion to succeed without any
3144 further processing. In a negative assertion, (*ACCEPT) causes the assertion to
3145 fail without any further processing.
3146 </P>
3147 <P>
3148 The other backtracking verbs are not treated specially if they appear in a
3149 positive assertion. In particular, (*THEN) skips to the next alternative in the
3150 innermost enclosing group that has alternations, whether or not this is within
3151 the assertion.
3152 </P>
3153 <P>
3154 Negative assertions are, however, different, in order to ensure that changing a
3155 positive assertion into a negative assertion changes its result. Backtracking
3156 into (*COMMIT), (*SKIP), or (*PRUNE) causes a negative assertion to be true,
3157 without considering any further alternative branches in the assertion.
3158 Backtracking into (*THEN) causes it to skip to the next enclosing alternative
3159 within the assertion (the normal behaviour), but if the assertion does not have
3160 such an alternative, (*THEN) behaves like (*PRUNE).
3161 <a name="btsub"></a></P>
3162 <br><b>
3163 Backtracking verbs in subroutines
3164 </b><br>
3165 <P>
3166 These behaviours occur whether or not the subpattern is called recursively.
3167 Perl's treatment of subroutines is different in some cases.
3168 </P>
3169 <P>
3170 (*FAIL) in a subpattern called as a subroutine has its normal effect: it forces
3171 an immediate backtrack.
3172 </P>
3173 <P>
3174 (*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to
3175 succeed without any further processing. Matching then continues after the
3176 subroutine call.
3177 </P>
3178 <P>
3179 (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause
3180 the subroutine match to fail.
3181 </P>
3182 <P>
3183 (*THEN) skips to the next alternative in the innermost enclosing group within
3184 the subpattern that has alternatives. If there is no such group within the
3185 subpattern, (*THEN) causes the subroutine match to fail.
3186 </P>
3187 <br><a name="SEC27" href="#TOC1">SEE ALSO</a><br>
3188 <P>
3189 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
3190 <b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>, <b>pcre32(3)</b>.
3191 </P>
3192 <br><a name="SEC28" href="#TOC1">AUTHOR</a><br>
3193 <P>
3194 Philip Hazel
3195 <br>
3196 University Computing Service
3197 <br>
3198 Cambridge CB2 3QH, England.
3199 <br>
3200 </P>
3201 <br><a name="SEC29" href="#TOC1">REVISION</a><br>
3202 <P>
3203 Last updated: 12 November 2013
3204 <br>
3205 Copyright &copy; 1997-2013 University of Cambridge.
3206 <br>
3207 <p>
3208 Return to the <a href="index.html">PCRE index page</a>.
3209 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5