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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 507 - (hide annotations) (download) (as text)
Wed Mar 10 16:08:01 2010 UTC (4 years, 6 months ago) by ph10
File MIME type: text/html
File size: 103703 byte(s)
Tidies for 8.02-RC1 release.

1 nigel 63 <html>
2     <head>
3     <title>pcrepattern specification</title>
4     </head>
5     <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 nigel 75 <h1>pcrepattern man page</h1>
7     <p>
8     Return to the <a href="index.html">PCRE index page</a>.
9     </p>
10 ph10 111 <p>
11 nigel 75 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 ph10 111 <br>
15 nigel 63 <ul>
16     <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 ph10 227 <li><a name="TOC2" href="#SEC2">NEWLINE CONVENTIONS</a>
18     <li><a name="TOC3" href="#SEC3">CHARACTERS AND METACHARACTERS</a>
19     <li><a name="TOC4" href="#SEC4">BACKSLASH</a>
20     <li><a name="TOC5" href="#SEC5">CIRCUMFLEX AND DOLLAR</a>
21     <li><a name="TOC6" href="#SEC6">FULL STOP (PERIOD, DOT)</a>
22     <li><a name="TOC7" href="#SEC7">MATCHING A SINGLE BYTE</a>
23     <li><a name="TOC8" href="#SEC8">SQUARE BRACKETS AND CHARACTER CLASSES</a>
24     <li><a name="TOC9" href="#SEC9">POSIX CHARACTER CLASSES</a>
25     <li><a name="TOC10" href="#SEC10">VERTICAL BAR</a>
26     <li><a name="TOC11" href="#SEC11">INTERNAL OPTION SETTING</a>
27     <li><a name="TOC12" href="#SEC12">SUBPATTERNS</a>
28     <li><a name="TOC13" href="#SEC13">DUPLICATE SUBPATTERN NUMBERS</a>
29     <li><a name="TOC14" href="#SEC14">NAMED SUBPATTERNS</a>
30     <li><a name="TOC15" href="#SEC15">REPETITION</a>
31     <li><a name="TOC16" href="#SEC16">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a>
32     <li><a name="TOC17" href="#SEC17">BACK REFERENCES</a>
33     <li><a name="TOC18" href="#SEC18">ASSERTIONS</a>
34     <li><a name="TOC19" href="#SEC19">CONDITIONAL SUBPATTERNS</a>
35     <li><a name="TOC20" href="#SEC20">COMMENTS</a>
36     <li><a name="TOC21" href="#SEC21">RECURSIVE PATTERNS</a>
37     <li><a name="TOC22" href="#SEC22">SUBPATTERNS AS SUBROUTINES</a>
38 ph10 345 <li><a name="TOC23" href="#SEC23">ONIGURUMA SUBROUTINE SYNTAX</a>
39     <li><a name="TOC24" href="#SEC24">CALLOUTS</a>
40     <li><a name="TOC25" href="#SEC25">BACKTRACKING CONTROL</a>
41     <li><a name="TOC26" href="#SEC26">SEE ALSO</a>
42     <li><a name="TOC27" href="#SEC27">AUTHOR</a>
43     <li><a name="TOC28" href="#SEC28">REVISION</a>
44 nigel 63 </ul>
45     <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
46     <P>
47 ph10 208 The syntax and semantics of the regular expressions that are supported by PCRE
48     are described in detail below. There is a quick-reference syntax summary in the
49     <a href="pcresyntax.html"><b>pcresyntax</b></a>
50 ph10 345 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
51     also supports some alternative regular expression syntax (which does not
52     conflict with the Perl syntax) in order to provide some compatibility with
53     regular expressions in Python, .NET, and Oniguruma.
54     </P>
55     <P>
56     Perl's regular expressions are described in its own documentation, and
57 ph10 208 regular expressions in general are covered in a number of books, some of which
58     have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
59     published by O'Reilly, covers regular expressions in great detail. This
60     description of PCRE's regular expressions is intended as reference material.
61 nigel 63 </P>
62     <P>
63 nigel 75 The original operation of PCRE was on strings of one-byte characters. However,
64 ph10 461 there is now also support for UTF-8 character strings. To use this,
65     PCRE must be built to include UTF-8 support, and you must call
66     <b>pcre_compile()</b> or <b>pcre_compile2()</b> with the PCRE_UTF8 option. There
67     is also a special sequence that can be given at the start of a pattern:
68 ph10 416 <pre>
69     (*UTF8)
70     </pre>
71     Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
72     option. This feature is not Perl-compatible. How setting UTF-8 mode affects
73     pattern matching is mentioned in several places below. There is also a summary
74     of UTF-8 features in the
75 nigel 63 <a href="pcre.html#utf8support">section on UTF-8 support</a>
76     in the main
77     <a href="pcre.html"><b>pcre</b></a>
78     page.
79     </P>
80     <P>
81 nigel 77 The remainder of this document discusses the patterns that are supported by
82     PCRE when its main matching function, <b>pcre_exec()</b>, is used.
83     From release 6.0, PCRE offers a second matching function,
84     <b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not
85 ph10 172 Perl-compatible. Some of the features discussed below are not available when
86     <b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the
87     alternative function, and how it differs from the normal function, are
88     discussed in the
89 nigel 77 <a href="pcrematching.html"><b>pcrematching</b></a>
90     page.
91     </P>
92 ph10 227 <br><a name="SEC2" href="#TOC1">NEWLINE CONVENTIONS</a><br>
93 nigel 77 <P>
94 ph10 227 PCRE supports five different conventions for indicating line breaks in
95     strings: a single CR (carriage return) character, a single LF (linefeed)
96     character, the two-character sequence CRLF, any of the three preceding, or any
97     Unicode newline sequence. The
98     <a href="pcreapi.html"><b>pcreapi</b></a>
99     page has
100     <a href="pcreapi.html#newlines">further discussion</a>
101     about newlines, and shows how to set the newline convention in the
102     <i>options</i> arguments for the compiling and matching functions.
103     </P>
104     <P>
105     It is also possible to specify a newline convention by starting a pattern
106     string with one of the following five sequences:
107     <pre>
108     (*CR) carriage return
109     (*LF) linefeed
110     (*CRLF) carriage return, followed by linefeed
111     (*ANYCRLF) any of the three above
112     (*ANY) all Unicode newline sequences
113     </pre>
114 ph10 461 These override the default and the options given to <b>pcre_compile()</b> or
115     <b>pcre_compile2()</b>. For example, on a Unix system where LF is the default
116     newline sequence, the pattern
117 ph10 227 <pre>
118     (*CR)a.b
119     </pre>
120     changes the convention to CR. That pattern matches "a\nb" because LF is no
121     longer a newline. Note that these special settings, which are not
122     Perl-compatible, are recognized only at the very start of a pattern, and that
123 ph10 231 they must be in upper case. If more than one of them is present, the last one
124     is used.
125 ph10 227 </P>
126 ph10 231 <P>
127     The newline convention does not affect what the \R escape sequence matches. By
128     default, this is any Unicode newline sequence, for Perl compatibility. However,
129     this can be changed; see the description of \R in the section entitled
130     <a href="#newlineseq">"Newline sequences"</a>
131 ph10 247 below. A change of \R setting can be combined with a change of newline
132     convention.
133 ph10 231 </P>
134 ph10 227 <br><a name="SEC3" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
135     <P>
136 nigel 63 A regular expression is a pattern that is matched against a subject string from
137     left to right. Most characters stand for themselves in a pattern, and match the
138     corresponding characters in the subject. As a trivial example, the pattern
139     <pre>
140     The quick brown fox
141 nigel 75 </pre>
142 nigel 77 matches a portion of a subject string that is identical to itself. When
143     caseless matching is specified (the PCRE_CASELESS option), letters are matched
144     independently of case. In UTF-8 mode, PCRE always understands the concept of
145     case for characters whose values are less than 128, so caseless matching is
146     always possible. For characters with higher values, the concept of case is
147     supported if PCRE is compiled with Unicode property support, but not otherwise.
148     If you want to use caseless matching for characters 128 and above, you must
149     ensure that PCRE is compiled with Unicode property support as well as with
150     UTF-8 support.
151     </P>
152     <P>
153     The power of regular expressions comes from the ability to include alternatives
154     and repetitions in the pattern. These are encoded in the pattern by the use of
155 nigel 75 <i>metacharacters</i>, which do not stand for themselves but instead are
156 nigel 63 interpreted in some special way.
157     </P>
158     <P>
159 nigel 75 There are two different sets of metacharacters: those that are recognized
160 nigel 63 anywhere in the pattern except within square brackets, and those that are
161 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
162     are as follows:
163 nigel 63 <pre>
164     \ general escape character with several uses
165     ^ assert start of string (or line, in multiline mode)
166     $ assert end of string (or line, in multiline mode)
167     . match any character except newline (by default)
168     [ start character class definition
169     | start of alternative branch
170     ( start subpattern
171     ) end subpattern
172     ? extends the meaning of (
173     also 0 or 1 quantifier
174     also quantifier minimizer
175     * 0 or more quantifier
176     + 1 or more quantifier
177     also "possessive quantifier"
178     { start min/max quantifier
179 nigel 75 </pre>
180 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
181 nigel 75 a character class the only metacharacters are:
182 nigel 63 <pre>
183     \ general escape character
184     ^ negate the class, but only if the first character
185     - indicates character range
186 nigel 75 [ POSIX character class (only if followed by POSIX syntax)
187 nigel 63 ] terminates the character class
188 nigel 75 </pre>
189     The following sections describe the use of each of the metacharacters.
190 nigel 63 </P>
191 ph10 227 <br><a name="SEC4" href="#TOC1">BACKSLASH</a><br>
192 nigel 63 <P>
193     The backslash character has several uses. Firstly, if it is followed by a
194 nigel 91 non-alphanumeric character, it takes away any special meaning that character
195     may have. This use of backslash as an escape character applies both inside and
196 nigel 63 outside character classes.
197     </P>
198     <P>
199     For example, if you want to match a * character, you write \* in the pattern.
200     This escaping action applies whether or not the following character would
201 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
202     non-alphanumeric with backslash to specify that it stands for itself. In
203 nigel 63 particular, if you want to match a backslash, you write \\.
204     </P>
205     <P>
206     If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
207     pattern (other than in a character class) and characters between a # outside
208 nigel 91 a character class and the next newline are ignored. An escaping backslash can
209     be used to include a whitespace or # character as part of the pattern.
210 nigel 63 </P>
211     <P>
212     If you want to remove the special meaning from a sequence of characters, you
213     can do so by putting them between \Q and \E. This is different from Perl in
214     that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
215     Perl, $ and @ cause variable interpolation. Note the following examples:
216     <pre>
217     Pattern PCRE matches Perl matches
218 nigel 75
219     \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
220 nigel 63 \Qabc\$xyz\E abc\$xyz abc\$xyz
221     \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
222 nigel 75 </pre>
223 nigel 63 The \Q...\E sequence is recognized both inside and outside character classes.
224 nigel 75 <a name="digitsafterbackslash"></a></P>
225     <br><b>
226     Non-printing characters
227     </b><br>
228 nigel 63 <P>
229     A second use of backslash provides a way of encoding non-printing characters
230     in patterns in a visible manner. There is no restriction on the appearance of
231     non-printing characters, apart from the binary zero that terminates a pattern,
232 ph10 461 but when a pattern is being prepared by text editing, it is often easier to use
233     one of the following escape sequences than the binary character it represents:
234 nigel 63 <pre>
235     \a alarm, that is, the BEL character (hex 07)
236     \cx "control-x", where x is any character
237     \e escape (hex 1B)
238     \f formfeed (hex 0C)
239 ph10 227 \n linefeed (hex 0A)
240 nigel 63 \r carriage return (hex 0D)
241     \t tab (hex 09)
242 ph10 488 \ddd character with octal code ddd, or back reference
243 nigel 63 \xhh character with hex code hh
244 nigel 87 \x{hhh..} character with hex code hhh..
245 nigel 75 </pre>
246 nigel 63 The precise effect of \cx is as follows: if x is a lower case letter, it
247     is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
248     Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
249     7B.
250     </P>
251     <P>
252     After \x, from zero to two hexadecimal digits are read (letters can be in
253 nigel 87 upper or lower case). Any number of hexadecimal digits may appear between \x{
254     and }, but the value of the character code must be less than 256 in non-UTF-8
255 ph10 211 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
256     hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
257     point, which is 10FFFF.
258 nigel 63 </P>
259     <P>
260 ph10 211 If characters other than hexadecimal digits appear between \x{ and }, or if
261     there is no terminating }, this form of escape is not recognized. Instead, the
262     initial \x will be interpreted as a basic hexadecimal escape, with no
263     following digits, giving a character whose value is zero.
264     </P>
265     <P>
266 nigel 63 Characters whose value is less than 256 can be defined by either of the two
267 nigel 87 syntaxes for \x. There is no difference in the way they are handled. For
268     example, \xdc is exactly the same as \x{dc}.
269 nigel 63 </P>
270     <P>
271 nigel 91 After \0 up to two further octal digits are read. If there are fewer than two
272     digits, just those that are present are used. Thus the sequence \0\x\07
273     specifies two binary zeros followed by a BEL character (code value 7). Make
274     sure you supply two digits after the initial zero if the pattern character that
275     follows is itself an octal digit.
276 nigel 63 </P>
277     <P>
278     The handling of a backslash followed by a digit other than 0 is complicated.
279     Outside a character class, PCRE reads it and any following digits as a decimal
280     number. If the number is less than 10, or if there have been at least that many
281     previous capturing left parentheses in the expression, the entire sequence is
282     taken as a <i>back reference</i>. A description of how this works is given
283 nigel 75 <a href="#backreferences">later,</a>
284     following the discussion of
285     <a href="#subpattern">parenthesized subpatterns.</a>
286 nigel 63 </P>
287     <P>
288     Inside a character class, or if the decimal number is greater than 9 and there
289     have not been that many capturing subpatterns, PCRE re-reads up to three octal
290 nigel 93 digits following the backslash, and uses them to generate a data character. Any
291 nigel 91 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
292     character specified in octal must be less than \400. In UTF-8 mode, values up
293     to \777 are permitted. For example:
294 nigel 63 <pre>
295     \040 is another way of writing a space
296 nigel 75 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
297 nigel 63 \7 is always a back reference
298 nigel 75 \11 might be a back reference, or another way of writing a tab
299 nigel 63 \011 is always a tab
300     \0113 is a tab followed by the character "3"
301 nigel 75 \113 might be a back reference, otherwise the character with octal code 113
302     \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
303     \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
304     </pre>
305 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
306     zero, because no more than three octal digits are ever read.
307     </P>
308     <P>
309 nigel 91 All the sequences that define a single character value can be used both inside
310     and outside character classes. In addition, inside a character class, the
311     sequence \b is interpreted as the backspace character (hex 08), and the
312 nigel 93 sequences \R and \X are interpreted as the characters "R" and "X",
313     respectively. Outside a character class, these sequences have different
314     meanings
315 nigel 75 <a href="#uniextseq">(see below).</a>
316 nigel 63 </P>
317 nigel 75 <br><b>
318 nigel 93 Absolute and relative back references
319     </b><br>
320     <P>
321 ph10 208 The sequence \g followed by an unsigned or a negative number, optionally
322     enclosed in braces, is an absolute or relative back reference. A named back
323     reference can be coded as \g{name}. Back references are discussed
324 nigel 93 <a href="#backreferences">later,</a>
325     following the discussion of
326     <a href="#subpattern">parenthesized subpatterns.</a>
327     </P>
328     <br><b>
329 ph10 345 Absolute and relative subroutine calls
330     </b><br>
331     <P>
332     For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
333     a number enclosed either in angle brackets or single quotes, is an alternative
334     syntax for referencing a subpattern as a "subroutine". Details are discussed
335     <a href="#onigurumasubroutines">later.</a>
336     Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
337 ph10 461 synonymous. The former is a back reference; the latter is a
338 ph10 454 <a href="#subpatternsassubroutines">subroutine</a>
339     call.
340 ph10 345 </P>
341     <br><b>
342 nigel 75 Generic character types
343     </b><br>
344 nigel 63 <P>
345 nigel 93 Another use of backslash is for specifying generic character types. The
346 nigel 75 following are always recognized:
347 nigel 63 <pre>
348     \d any decimal digit
349     \D any character that is not a decimal digit
350 ph10 182 \h any horizontal whitespace character
351     \H any character that is not a horizontal whitespace character
352 nigel 63 \s any whitespace character
353     \S any character that is not a whitespace character
354 ph10 182 \v any vertical whitespace character
355     \V any character that is not a vertical whitespace character
356 nigel 63 \w any "word" character
357     \W any "non-word" character
358 nigel 75 </pre>
359 nigel 63 Each pair of escape sequences partitions the complete set of characters into
360     two disjoint sets. Any given character matches one, and only one, of each pair.
361     </P>
362     <P>
363 nigel 75 These character type sequences can appear both inside and outside character
364     classes. They each match one character of the appropriate type. If the current
365     matching point is at the end of the subject string, all of them fail, since
366     there is no character to match.
367 nigel 63 </P>
368     <P>
369     For compatibility with Perl, \s does not match the VT character (code 11).
370     This makes it different from the the POSIX "space" class. The \s characters
371 ph10 182 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
372 nigel 91 included in a Perl script, \s may match the VT character. In PCRE, it never
373 ph10 182 does.
374 nigel 63 </P>
375     <P>
376 ph10 182 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
377     \w, and always match \D, \S, and \W. This is true even when Unicode
378     character property support is available. These sequences retain their original
379     meanings from before UTF-8 support was available, mainly for efficiency
380 ph10 406 reasons. Note that this also affects \b, because it is defined in terms of \w
381     and \W.
382 ph10 182 </P>
383     <P>
384     The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to the
385     other sequences, these do match certain high-valued codepoints in UTF-8 mode.
386     The horizontal space characters are:
387     <pre>
388     U+0009 Horizontal tab
389     U+0020 Space
390     U+00A0 Non-break space
391     U+1680 Ogham space mark
392     U+180E Mongolian vowel separator
393     U+2000 En quad
394     U+2001 Em quad
395     U+2002 En space
396     U+2003 Em space
397     U+2004 Three-per-em space
398     U+2005 Four-per-em space
399     U+2006 Six-per-em space
400     U+2007 Figure space
401     U+2008 Punctuation space
402     U+2009 Thin space
403     U+200A Hair space
404     U+202F Narrow no-break space
405     U+205F Medium mathematical space
406     U+3000 Ideographic space
407     </pre>
408     The vertical space characters are:
409     <pre>
410     U+000A Linefeed
411     U+000B Vertical tab
412     U+000C Formfeed
413     U+000D Carriage return
414     U+0085 Next line
415     U+2028 Line separator
416     U+2029 Paragraph separator
417     </PRE>
418     </P>
419     <P>
420 nigel 75 A "word" character is an underscore or any character less than 256 that is a
421     letter or digit. The definition of letters and digits is controlled by PCRE's
422     low-valued character tables, and may vary if locale-specific matching is taking
423     place (see
424 nigel 63 <a href="pcreapi.html#localesupport">"Locale support"</a>
425     in the
426     <a href="pcreapi.html"><b>pcreapi</b></a>
427 ph10 142 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
428     or "french" in Windows, some character codes greater than 128 are used for
429 ph10 182 accented letters, and these are matched by \w. The use of locales with Unicode
430     is discouraged.
431 ph10 231 <a name="newlineseq"></a></P>
432 nigel 93 <br><b>
433     Newline sequences
434     </b><br>
435     <P>
436 ph10 231 Outside a character class, by default, the escape sequence \R matches any
437     Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is
438     equivalent to the following:
439 nigel 93 <pre>
440     (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
441     </pre>
442     This is an example of an "atomic group", details of which are given
443     <a href="#atomicgroup">below.</a>
444     This particular group matches either the two-character sequence CR followed by
445     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
446     U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
447     line, U+0085). The two-character sequence is treated as a single unit that
448     cannot be split.
449     </P>
450     <P>
451     In UTF-8 mode, two additional characters whose codepoints are greater than 255
452     are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
453     Unicode character property support is not needed for these characters to be
454     recognized.
455     </P>
456     <P>
457 ph10 231 It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
458     complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
459 ph10 247 either at compile time or when the pattern is matched. (BSR is an abbrevation
460     for "backslash R".) This can be made the default when PCRE is built; if this is
461     the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
462     It is also possible to specify these settings by starting a pattern string with
463     one of the following sequences:
464 ph10 231 <pre>
465     (*BSR_ANYCRLF) CR, LF, or CRLF only
466     (*BSR_UNICODE) any Unicode newline sequence
467     </pre>
468 ph10 461 These override the default and the options given to <b>pcre_compile()</b> or
469     <b>pcre_compile2()</b>, but they can be overridden by options given to
470     <b>pcre_exec()</b> or <b>pcre_dfa_exec()</b>. Note that these special settings,
471     which are not Perl-compatible, are recognized only at the very start of a
472     pattern, and that they must be in upper case. If more than one of them is
473     present, the last one is used. They can be combined with a change of newline
474     convention, for example, a pattern can start with:
475 ph10 247 <pre>
476     (*ANY)(*BSR_ANYCRLF)
477     </pre>
478 nigel 93 Inside a character class, \R matches the letter "R".
479 nigel 75 <a name="uniextseq"></a></P>
480     <br><b>
481     Unicode character properties
482     </b><br>
483     <P>
484     When PCRE is built with Unicode character property support, three additional
485 ph10 185 escape sequences that match characters with specific properties are available.
486     When not in UTF-8 mode, these sequences are of course limited to testing
487     characters whose codepoints are less than 256, but they do work in this mode.
488     The extra escape sequences are:
489 nigel 75 <pre>
490 nigel 87 \p{<i>xx</i>} a character with the <i>xx</i> property
491     \P{<i>xx</i>} a character without the <i>xx</i> property
492     \X an extended Unicode sequence
493 nigel 75 </pre>
494 nigel 87 The property names represented by <i>xx</i> above are limited to the Unicode
495     script names, the general category properties, and "Any", which matches any
496     character (including newline). Other properties such as "InMusicalSymbols" are
497     not currently supported by PCRE. Note that \P{Any} does not match any
498     characters, so always causes a match failure.
499 nigel 63 </P>
500     <P>
501 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
502     character from one of these sets can be matched using a script name. For
503     example:
504 nigel 75 <pre>
505 nigel 87 \p{Greek}
506     \P{Han}
507     </pre>
508     Those that are not part of an identified script are lumped together as
509     "Common". The current list of scripts is:
510     </P>
511     <P>
512     Arabic,
513     Armenian,
514 ph10 507 Avestan,
515 nigel 93 Balinese,
516 ph10 507 Bamum,
517 nigel 87 Bengali,
518     Bopomofo,
519     Braille,
520     Buginese,
521     Buhid,
522     Canadian_Aboriginal,
523 ph10 507 Carian,
524     Cham,
525 nigel 87 Cherokee,
526     Common,
527     Coptic,
528 nigel 93 Cuneiform,
529 nigel 87 Cypriot,
530     Cyrillic,
531     Deseret,
532     Devanagari,
533 ph10 507 Egyptian_Hieroglyphs,
534 nigel 87 Ethiopic,
535     Georgian,
536     Glagolitic,
537     Gothic,
538     Greek,
539     Gujarati,
540     Gurmukhi,
541     Han,
542     Hangul,
543     Hanunoo,
544     Hebrew,
545     Hiragana,
546 ph10 507 Imperial_Aramaic,
547 nigel 87 Inherited,
548 ph10 507 Inscriptional_Pahlavi,
549     Inscriptional_Parthian,
550     Javanese,
551     Kaithi,
552 nigel 87 Kannada,
553     Katakana,
554 ph10 507 Kayah_Li,
555 nigel 87 Kharoshthi,
556     Khmer,
557     Lao,
558     Latin,
559 ph10 507 Lepcha,
560 nigel 87 Limbu,
561     Linear_B,
562 ph10 507 Lisu,
563     Lycian,
564     Lydian,
565 nigel 87 Malayalam,
566 ph10 507 Meetei_Mayek,
567 nigel 87 Mongolian,
568     Myanmar,
569     New_Tai_Lue,
570 nigel 93 Nko,
571 nigel 87 Ogham,
572     Old_Italic,
573     Old_Persian,
574 ph10 507 Old_South_Arabian,
575     Old_Turkic,
576     Ol_Chiki,
577 nigel 87 Oriya,
578     Osmanya,
579 nigel 93 Phags_Pa,
580     Phoenician,
581 ph10 507 Rejang,
582 nigel 87 Runic,
583 ph10 507 Samaritan,
584     Saurashtra,
585 nigel 87 Shavian,
586     Sinhala,
587 ph10 507 Sundanese,
588 nigel 87 Syloti_Nagri,
589     Syriac,
590     Tagalog,
591     Tagbanwa,
592     Tai_Le,
593 ph10 507 Tai_Tham,
594     Tai_Viet,
595 nigel 87 Tamil,
596     Telugu,
597     Thaana,
598     Thai,
599     Tibetan,
600     Tifinagh,
601     Ugaritic,
602 ph10 507 Vai,
603 nigel 87 Yi.
604     </P>
605     <P>
606     Each character has exactly one general category property, specified by a
607     two-letter abbreviation. For compatibility with Perl, negation can be specified
608     by including a circumflex between the opening brace and the property name. For
609     example, \p{^Lu} is the same as \P{Lu}.
610     </P>
611     <P>
612     If only one letter is specified with \p or \P, it includes all the general
613     category properties that start with that letter. In this case, in the absence
614     of negation, the curly brackets in the escape sequence are optional; these two
615     examples have the same effect:
616     <pre>
617 nigel 75 \p{L}
618     \pL
619     </pre>
620 nigel 87 The following general category property codes are supported:
621 nigel 75 <pre>
622     C Other
623     Cc Control
624     Cf Format
625     Cn Unassigned
626     Co Private use
627     Cs Surrogate
628    
629     L Letter
630     Ll Lower case letter
631     Lm Modifier letter
632     Lo Other letter
633     Lt Title case letter
634     Lu Upper case letter
635    
636     M Mark
637     Mc Spacing mark
638     Me Enclosing mark
639     Mn Non-spacing mark
640    
641     N Number
642     Nd Decimal number
643     Nl Letter number
644     No Other number
645    
646     P Punctuation
647     Pc Connector punctuation
648     Pd Dash punctuation
649     Pe Close punctuation
650     Pf Final punctuation
651     Pi Initial punctuation
652     Po Other punctuation
653     Ps Open punctuation
654    
655     S Symbol
656     Sc Currency symbol
657     Sk Modifier symbol
658     Sm Mathematical symbol
659     So Other symbol
660    
661     Z Separator
662     Zl Line separator
663     Zp Paragraph separator
664     Zs Space separator
665     </pre>
666 nigel 87 The special property L& is also supported: it matches a character that has
667     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
668     a modifier or "other".
669 nigel 75 </P>
670     <P>
671 ph10 211 The Cs (Surrogate) property applies only to characters in the range U+D800 to
672     U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
673     cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
674     (see the discussion of PCRE_NO_UTF8_CHECK in the
675     <a href="pcreapi.html"><b>pcreapi</b></a>
676 ph10 453 page). Perl does not support the Cs property.
677 ph10 211 </P>
678     <P>
679 ph10 453 The long synonyms for property names that Perl supports (such as \p{Letter})
680 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
681 nigel 87 properties with "Is".
682     </P>
683     <P>
684     No character that is in the Unicode table has the Cn (unassigned) property.
685     Instead, this property is assumed for any code point that is not in the
686     Unicode table.
687     </P>
688     <P>
689 nigel 75 Specifying caseless matching does not affect these escape sequences. For
690     example, \p{Lu} always matches only upper case letters.
691     </P>
692     <P>
693     The \X escape matches any number of Unicode characters that form an extended
694     Unicode sequence. \X is equivalent to
695     <pre>
696     (?&#62;\PM\pM*)
697     </pre>
698     That is, it matches a character without the "mark" property, followed by zero
699     or more characters with the "mark" property, and treats the sequence as an
700     atomic group
701     <a href="#atomicgroup">(see below).</a>
702     Characters with the "mark" property are typically accents that affect the
703 ph10 185 preceding character. None of them have codepoints less than 256, so in
704     non-UTF-8 mode \X matches any one character.
705 nigel 75 </P>
706     <P>
707     Matching characters by Unicode property is not fast, because PCRE has to search
708     a structure that contains data for over fifteen thousand characters. That is
709     why the traditional escape sequences such as \d and \w do not use Unicode
710     properties in PCRE.
711 ph10 172 <a name="resetmatchstart"></a></P>
712     <br><b>
713     Resetting the match start
714     </b><br>
715     <P>
716     The escape sequence \K, which is a Perl 5.10 feature, causes any previously
717     matched characters not to be included in the final matched sequence. For
718     example, the pattern:
719     <pre>
720     foo\Kbar
721     </pre>
722     matches "foobar", but reports that it has matched "bar". This feature is
723     similar to a lookbehind assertion
724     <a href="#lookbehind">(described below).</a>
725     However, in this case, the part of the subject before the real match does not
726     have to be of fixed length, as lookbehind assertions do. The use of \K does
727     not interfere with the setting of
728     <a href="#subpattern">captured substrings.</a>
729     For example, when the pattern
730     <pre>
731     (foo)\Kbar
732     </pre>
733     matches "foobar", the first substring is still set to "foo".
734 ph10 507 </P>
735     <P>
736     Perl documents that the use of \K within assertions is "not well defined". In
737     PCRE, \K is acted upon when it occurs inside positive assertions, but is
738     ignored in negative assertions.
739 nigel 75 <a name="smallassertions"></a></P>
740     <br><b>
741     Simple assertions
742     </b><br>
743     <P>
744 nigel 93 The final use of backslash is for certain simple assertions. An assertion
745 nigel 63 specifies a condition that has to be met at a particular point in a match,
746     without consuming any characters from the subject string. The use of
747 nigel 75 subpatterns for more complicated assertions is described
748     <a href="#bigassertions">below.</a>
749 nigel 91 The backslashed assertions are:
750 nigel 63 <pre>
751     \b matches at a word boundary
752     \B matches when not at a word boundary
753 nigel 93 \A matches at the start of the subject
754     \Z matches at the end of the subject
755     also matches before a newline at the end of the subject
756     \z matches only at the end of the subject
757     \G matches at the first matching position in the subject
758 nigel 75 </pre>
759 nigel 63 These assertions may not appear in character classes (but note that \b has a
760     different meaning, namely the backspace character, inside a character class).
761     </P>
762     <P>
763     A word boundary is a position in the subject string where the current character
764     and the previous character do not both match \w or \W (i.e. one matches
765     \w and the other matches \W), or the start or end of the string if the
766 ph10 461 first or last character matches \w, respectively. Neither PCRE nor Perl has a
767     separte "start of word" or "end of word" metasequence. However, whatever
768     follows \b normally determines which it is. For example, the fragment
769     \ba matches "a" at the start of a word.
770 nigel 63 </P>
771     <P>
772     The \A, \Z, and \z assertions differ from the traditional circumflex and
773 nigel 75 dollar (described in the next section) in that they only ever match at the very
774     start and end of the subject string, whatever options are set. Thus, they are
775     independent of multiline mode. These three assertions are not affected by the
776     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
777     circumflex and dollar metacharacters. However, if the <i>startoffset</i>
778     argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
779     at a point other than the beginning of the subject, \A can never match. The
780 nigel 91 difference between \Z and \z is that \Z matches before a newline at the end
781     of the string as well as at the very end, whereas \z matches only at the end.
782 nigel 63 </P>
783     <P>
784     The \G assertion is true only when the current matching position is at the
785     start point of the match, as specified by the <i>startoffset</i> argument of
786     <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
787     non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
788     arguments, you can mimic Perl's /g option, and it is in this kind of
789     implementation where \G can be useful.
790     </P>
791     <P>
792     Note, however, that PCRE's interpretation of \G, as the start of the current
793     match, is subtly different from Perl's, which defines it as the end of the
794     previous match. In Perl, these can be different when the previously matched
795     string was empty. Because PCRE does just one match at a time, it cannot
796     reproduce this behaviour.
797     </P>
798     <P>
799     If all the alternatives of a pattern begin with \G, the expression is anchored
800     to the starting match position, and the "anchored" flag is set in the compiled
801     regular expression.
802     </P>
803 ph10 227 <br><a name="SEC5" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
804 nigel 63 <P>
805     Outside a character class, in the default matching mode, the circumflex
806 nigel 75 character is an assertion that is true only if the current matching point is
807 nigel 63 at the start of the subject string. If the <i>startoffset</i> argument of
808     <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
809     option is unset. Inside a character class, circumflex has an entirely different
810 nigel 75 meaning
811     <a href="#characterclass">(see below).</a>
812 nigel 63 </P>
813     <P>
814     Circumflex need not be the first character of the pattern if a number of
815     alternatives are involved, but it should be the first thing in each alternative
816     in which it appears if the pattern is ever to match that branch. If all
817     possible alternatives start with a circumflex, that is, if the pattern is
818     constrained to match only at the start of the subject, it is said to be an
819     "anchored" pattern. (There are also other constructs that can cause a pattern
820     to be anchored.)
821     </P>
822     <P>
823 nigel 75 A dollar character is an assertion that is true only if the current matching
824 nigel 63 point is at the end of the subject string, or immediately before a newline
825 nigel 91 at the end of the string (by default). Dollar need not be the last character of
826     the pattern if a number of alternatives are involved, but it should be the last
827     item in any branch in which it appears. Dollar has no special meaning in a
828     character class.
829 nigel 63 </P>
830     <P>
831     The meaning of dollar can be changed so that it matches only at the very end of
832     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
833     does not affect the \Z assertion.
834     </P>
835     <P>
836     The meanings of the circumflex and dollar characters are changed if the
837 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
838     immediately after internal newlines as well as at the start of the subject
839     string. It does not match after a newline that ends the string. A dollar
840     matches before any newlines in the string, as well as at the very end, when
841     PCRE_MULTILINE is set. When newline is specified as the two-character
842     sequence CRLF, isolated CR and LF characters do not indicate newlines.
843 nigel 63 </P>
844     <P>
845 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
846     \n represents a newline) in multiline mode, but not otherwise. Consequently,
847     patterns that are anchored in single line mode because all branches start with
848     ^ are not anchored in multiline mode, and a match for circumflex is possible
849     when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
850     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
851     </P>
852     <P>
853 nigel 63 Note that the sequences \A, \Z, and \z can be used to match the start and
854     end of the subject in both modes, and if all branches of a pattern start with
855 nigel 91 \A it is always anchored, whether or not PCRE_MULTILINE is set.
856 nigel 63 </P>
857 ph10 227 <br><a name="SEC6" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
858 nigel 63 <P>
859     Outside a character class, a dot in the pattern matches any one character in
860 nigel 91 the subject string except (by default) a character that signifies the end of a
861 nigel 93 line. In UTF-8 mode, the matched character may be more than one byte long.
862 nigel 63 </P>
863 nigel 91 <P>
864 nigel 93 When a line ending is defined as a single character, dot never matches that
865     character; when the two-character sequence CRLF is used, dot does not match CR
866     if it is immediately followed by LF, but otherwise it matches all characters
867     (including isolated CRs and LFs). When any Unicode line endings are being
868     recognized, dot does not match CR or LF or any of the other line ending
869     characters.
870     </P>
871     <P>
872 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
873 nigel 93 option is set, a dot matches any one character, without exception. If the
874     two-character sequence CRLF is present in the subject string, it takes two dots
875     to match it.
876 nigel 91 </P>
877     <P>
878     The handling of dot is entirely independent of the handling of circumflex and
879     dollar, the only relationship being that they both involve newlines. Dot has no
880     special meaning in a character class.
881     </P>
882 ph10 227 <br><a name="SEC7" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
883 nigel 63 <P>
884     Outside a character class, the escape sequence \C matches any one byte, both
885 nigel 93 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
886     characters. The feature is provided in Perl in order to match individual bytes
887     in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
888     what remains in the string may be a malformed UTF-8 string. For this reason,
889     the \C escape sequence is best avoided.
890 nigel 63 </P>
891     <P>
892 nigel 75 PCRE does not allow \C to appear in lookbehind assertions
893     <a href="#lookbehind">(described below),</a>
894     because in UTF-8 mode this would make it impossible to calculate the length of
895     the lookbehind.
896     <a name="characterclass"></a></P>
897 ph10 227 <br><a name="SEC8" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
898 nigel 63 <P>
899     An opening square bracket introduces a character class, terminated by a closing
900 ph10 461 square bracket. A closing square bracket on its own is not special by default.
901     However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
902     bracket causes a compile-time error. If a closing square bracket is required as
903     a member of the class, it should be the first data character in the class
904     (after an initial circumflex, if present) or escaped with a backslash.
905 nigel 63 </P>
906     <P>
907     A character class matches a single character in the subject. In UTF-8 mode, the
908 ph10 461 character may be more than one byte long. A matched character must be in the
909     set of characters defined by the class, unless the first character in the class
910 nigel 63 definition is a circumflex, in which case the subject character must not be in
911     the set defined by the class. If a circumflex is actually required as a member
912     of the class, ensure it is not the first character, or escape it with a
913     backslash.
914     </P>
915     <P>
916     For example, the character class [aeiou] matches any lower case vowel, while
917     [^aeiou] matches any character that is not a lower case vowel. Note that a
918 nigel 75 circumflex is just a convenient notation for specifying the characters that
919     are in the class by enumerating those that are not. A class that starts with a
920 ph10 461 circumflex is not an assertion; it still consumes a character from the subject
921 nigel 75 string, and therefore it fails if the current pointer is at the end of the
922     string.
923 nigel 63 </P>
924     <P>
925     In UTF-8 mode, characters with values greater than 255 can be included in a
926     class as a literal string of bytes, or by using the \x{ escaping mechanism.
927     </P>
928     <P>
929     When caseless matching is set, any letters in a class represent both their
930     upper case and lower case versions, so for example, a caseless [aeiou] matches
931     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
932 nigel 77 caseful version would. In UTF-8 mode, PCRE always understands the concept of
933     case for characters whose values are less than 128, so caseless matching is
934     always possible. For characters with higher values, the concept of case is
935     supported if PCRE is compiled with Unicode property support, but not otherwise.
936 ph10 461 If you want to use caseless matching in UTF8-mode for characters 128 and above,
937     you must ensure that PCRE is compiled with Unicode property support as well as
938     with UTF-8 support.
939 nigel 63 </P>
940     <P>
941 nigel 93 Characters that might indicate line breaks are never treated in any special way
942     when matching character classes, whatever line-ending sequence is in use, and
943     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
944     such as [^a] always matches one of these characters.
945 nigel 63 </P>
946     <P>
947     The minus (hyphen) character can be used to specify a range of characters in a
948     character class. For example, [d-m] matches any letter between d and m,
949     inclusive. If a minus character is required in a class, it must be escaped with
950     a backslash or appear in a position where it cannot be interpreted as
951     indicating a range, typically as the first or last character in the class.
952     </P>
953     <P>
954     It is not possible to have the literal character "]" as the end character of a
955     range. A pattern such as [W-]46] is interpreted as a class of two characters
956     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
957     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
958 nigel 75 the end of range, so [W-\]46] is interpreted as a class containing a range
959     followed by two other characters. The octal or hexadecimal representation of
960     "]" can also be used to end a range.
961 nigel 63 </P>
962     <P>
963     Ranges operate in the collating sequence of character values. They can also be
964     used for characters specified numerically, for example [\000-\037]. In UTF-8
965     mode, ranges can include characters whose values are greater than 255, for
966     example [\x{100}-\x{2ff}].
967     </P>
968     <P>
969     If a range that includes letters is used when caseless matching is set, it
970     matches the letters in either case. For example, [W-c] is equivalent to
971 nigel 75 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
972 ph10 142 tables for a French locale are in use, [\xc8-\xcb] matches accented E
973 nigel 75 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
974     characters with values greater than 128 only when it is compiled with Unicode
975     property support.
976 nigel 63 </P>
977     <P>
978 nigel 75 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
979     in a character class, and add the characters that they match to the class. For
980 nigel 63 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
981     conveniently be used with the upper case character types to specify a more
982     restricted set of characters than the matching lower case type. For example,
983     the class [^\W_] matches any letter or digit, but not underscore.
984     </P>
985     <P>
986 nigel 75 The only metacharacters that are recognized in character classes are backslash,
987     hyphen (only where it can be interpreted as specifying a range), circumflex
988     (only at the start), opening square bracket (only when it can be interpreted as
989     introducing a POSIX class name - see the next section), and the terminating
990     closing square bracket. However, escaping other non-alphanumeric characters
991     does no harm.
992 nigel 63 </P>
993 ph10 227 <br><a name="SEC9" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
994 nigel 63 <P>
995 nigel 75 Perl supports the POSIX notation for character classes. This uses names
996 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
997     this notation. For example,
998     <pre>
999     [01[:alpha:]%]
1000 nigel 75 </pre>
1001 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
1002     are
1003     <pre>
1004     alnum letters and digits
1005     alpha letters
1006     ascii character codes 0 - 127
1007     blank space or tab only
1008     cntrl control characters
1009     digit decimal digits (same as \d)
1010     graph printing characters, excluding space
1011     lower lower case letters
1012     print printing characters, including space
1013     punct printing characters, excluding letters and digits
1014     space white space (not quite the same as \s)
1015     upper upper case letters
1016     word "word" characters (same as \w)
1017     xdigit hexadecimal digits
1018 nigel 75 </pre>
1019 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1020     space (32). Notice that this list includes the VT character (code 11). This
1021     makes "space" different to \s, which does not include VT (for Perl
1022     compatibility).
1023     </P>
1024     <P>
1025     The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1026     5.8. Another Perl extension is negation, which is indicated by a ^ character
1027     after the colon. For example,
1028     <pre>
1029     [12[:^digit:]]
1030 nigel 75 </pre>
1031 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1032     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1033     supported, and an error is given if they are encountered.
1034     </P>
1035     <P>
1036 nigel 75 In UTF-8 mode, characters with values greater than 128 do not match any of
1037 nigel 63 the POSIX character classes.
1038     </P>
1039 ph10 227 <br><a name="SEC10" href="#TOC1">VERTICAL BAR</a><br>
1040 nigel 63 <P>
1041     Vertical bar characters are used to separate alternative patterns. For example,
1042     the pattern
1043     <pre>
1044     gilbert|sullivan
1045 nigel 75 </pre>
1046 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1047 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
1048     process tries each alternative in turn, from left to right, and the first one
1049     that succeeds is used. If the alternatives are within a subpattern
1050 nigel 75 <a href="#subpattern">(defined below),</a>
1051     "succeeds" means matching the rest of the main pattern as well as the
1052     alternative in the subpattern.
1053 nigel 63 </P>
1054 ph10 227 <br><a name="SEC11" href="#TOC1">INTERNAL OPTION SETTING</a><br>
1055 nigel 63 <P>
1056     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1057 ph10 243 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1058     the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1059     The option letters are
1060 nigel 63 <pre>
1061     i for PCRE_CASELESS
1062     m for PCRE_MULTILINE
1063     s for PCRE_DOTALL
1064     x for PCRE_EXTENDED
1065 nigel 75 </pre>
1066 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
1067     unset these options by preceding the letter with a hyphen, and a combined
1068     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1069     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1070     permitted. If a letter appears both before and after the hyphen, the option is
1071     unset.
1072     </P>
1073     <P>
1074 ph10 243 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1075     changed in the same way as the Perl-compatible options by using the characters
1076     J, U and X respectively.
1077     </P>
1078     <P>
1079 ph10 416 When one of these option changes occurs at top level (that is, not inside
1080     subpattern parentheses), the change applies to the remainder of the pattern
1081     that follows. If the change is placed right at the start of a pattern, PCRE
1082     extracts it into the global options (and it will therefore show up in data
1083     extracted by the <b>pcre_fullinfo()</b> function).
1084 nigel 63 </P>
1085     <P>
1086 nigel 93 An option change within a subpattern (see below for a description of
1087     subpatterns) affects only that part of the current pattern that follows it, so
1088 nigel 63 <pre>
1089     (a(?i)b)c
1090 nigel 75 </pre>
1091 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1092     By this means, options can be made to have different settings in different
1093     parts of the pattern. Any changes made in one alternative do carry on
1094     into subsequent branches within the same subpattern. For example,
1095     <pre>
1096     (a(?i)b|c)
1097 nigel 75 </pre>
1098 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1099     branch is abandoned before the option setting. This is because the effects of
1100     option settings happen at compile time. There would be some very weird
1101     behaviour otherwise.
1102 ph10 254 </P>
1103     <P>
1104     <b>Note:</b> There are other PCRE-specific options that can be set by the
1105     application when the compile or match functions are called. In some cases the
1106 ph10 416 pattern can contain special leading sequences such as (*CRLF) to override what
1107     the application has set or what has been defaulted. Details are given in the
1108     section entitled
1109 ph10 254 <a href="#newlineseq">"Newline sequences"</a>
1110 ph10 416 above. There is also the (*UTF8) leading sequence that can be used to set UTF-8
1111     mode; this is equivalent to setting the PCRE_UTF8 option.
1112 nigel 75 <a name="subpattern"></a></P>
1113 ph10 227 <br><a name="SEC12" href="#TOC1">SUBPATTERNS</a><br>
1114 nigel 63 <P>
1115     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1116 nigel 75 Turning part of a pattern into a subpattern does two things:
1117     <br>
1118     <br>
1119 nigel 63 1. It localizes a set of alternatives. For example, the pattern
1120     <pre>
1121     cat(aract|erpillar|)
1122 nigel 75 </pre>
1123 nigel 63 matches one of the words "cat", "cataract", or "caterpillar". Without the
1124 nigel 93 parentheses, it would match "cataract", "erpillar" or an empty string.
1125 nigel 75 <br>
1126     <br>
1127     2. It sets up the subpattern as a capturing subpattern. This means that, when
1128     the whole pattern matches, that portion of the subject string that matched the
1129     subpattern is passed back to the caller via the <i>ovector</i> argument of
1130 nigel 63 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
1131 nigel 75 from 1) to obtain numbers for the capturing subpatterns.
1132 nigel 63 </P>
1133     <P>
1134     For example, if the string "the red king" is matched against the pattern
1135     <pre>
1136     the ((red|white) (king|queen))
1137 nigel 75 </pre>
1138 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
1139     2, and 3, respectively.
1140     </P>
1141     <P>
1142     The fact that plain parentheses fulfil two functions is not always helpful.
1143     There are often times when a grouping subpattern is required without a
1144     capturing requirement. If an opening parenthesis is followed by a question mark
1145     and a colon, the subpattern does not do any capturing, and is not counted when
1146     computing the number of any subsequent capturing subpatterns. For example, if
1147     the string "the white queen" is matched against the pattern
1148     <pre>
1149     the ((?:red|white) (king|queen))
1150 nigel 75 </pre>
1151 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
1152 nigel 93 2. The maximum number of capturing subpatterns is 65535.
1153 nigel 63 </P>
1154     <P>
1155     As a convenient shorthand, if any option settings are required at the start of
1156     a non-capturing subpattern, the option letters may appear between the "?" and
1157     the ":". Thus the two patterns
1158     <pre>
1159     (?i:saturday|sunday)
1160     (?:(?i)saturday|sunday)
1161 nigel 75 </pre>
1162 nigel 63 match exactly the same set of strings. Because alternative branches are tried
1163     from left to right, and options are not reset until the end of the subpattern
1164     is reached, an option setting in one branch does affect subsequent branches, so
1165     the above patterns match "SUNDAY" as well as "Saturday".
1166 ph10 461 <a name="dupsubpatternnumber"></a></P>
1167 ph10 227 <br><a name="SEC13" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1168 nigel 63 <P>
1169 ph10 182 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1170     the same numbers for its capturing parentheses. Such a subpattern starts with
1171     (?| and is itself a non-capturing subpattern. For example, consider this
1172     pattern:
1173     <pre>
1174     (?|(Sat)ur|(Sun))day
1175     </pre>
1176     Because the two alternatives are inside a (?| group, both sets of capturing
1177     parentheses are numbered one. Thus, when the pattern matches, you can look
1178     at captured substring number one, whichever alternative matched. This construct
1179     is useful when you want to capture part, but not all, of one of a number of
1180     alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1181     number is reset at the start of each branch. The numbers of any capturing
1182     buffers that follow the subpattern start after the highest number used in any
1183     branch. The following example is taken from the Perl documentation.
1184     The numbers underneath show in which buffer the captured content will be
1185     stored.
1186     <pre>
1187     # before ---------------branch-reset----------- after
1188     / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1189     # 1 2 2 3 2 3 4
1190     </pre>
1191 ph10 488 A back reference to a numbered subpattern uses the most recent value that is
1192     set for that number by any subpattern. The following pattern matches "abcabc"
1193     or "defdef":
1194 ph10 461 <pre>
1195     /(?|(abc)|(def))\1/
1196     </pre>
1197     In contrast, a recursive or "subroutine" call to a numbered subpattern always
1198     refers to the first one in the pattern with the given number. The following
1199     pattern matches "abcabc" or "defabc":
1200     <pre>
1201     /(?|(abc)|(def))(?1)/
1202     </pre>
1203     If a
1204     <a href="#conditions">condition test</a>
1205     for a subpattern's having matched refers to a non-unique number, the test is
1206     true if any of the subpatterns of that number have matched.
1207 ph10 182 </P>
1208     <P>
1209     An alternative approach to using this "branch reset" feature is to use
1210     duplicate named subpatterns, as described in the next section.
1211     </P>
1212 ph10 227 <br><a name="SEC14" href="#TOC1">NAMED SUBPATTERNS</a><br>
1213 ph10 182 <P>
1214 nigel 63 Identifying capturing parentheses by number is simple, but it can be very hard
1215     to keep track of the numbers in complicated regular expressions. Furthermore,
1216 nigel 75 if an expression is modified, the numbers may change. To help with this
1217 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
1218     added to Perl until release 5.10. Python had the feature earlier, and PCRE
1219     introduced it at release 4.0, using the Python syntax. PCRE now supports both
1220 ph10 461 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1221     have different names, but PCRE does not.
1222 nigel 93 </P>
1223     <P>
1224     In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1225     (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1226 nigel 91 parentheses from other parts of the pattern, such as
1227 ph10 488 <a href="#backreferences">back references,</a>
1228 nigel 91 <a href="#recursion">recursion,</a>
1229     and
1230     <a href="#conditions">conditions,</a>
1231     can be made by name as well as by number.
1232 nigel 63 </P>
1233     <P>
1234 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
1235 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
1236     if the names were not present. The PCRE API provides function calls for
1237     extracting the name-to-number translation table from a compiled pattern. There
1238     is also a convenience function for extracting a captured substring by name.
1239 nigel 91 </P>
1240     <P>
1241     By default, a name must be unique within a pattern, but it is possible to relax
1242 ph10 461 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1243     names are also always permitted for subpatterns with the same number, set up as
1244     described in the previous section.) Duplicate names can be useful for patterns
1245     where only one instance of the named parentheses can match. Suppose you want to
1246     match the name of a weekday, either as a 3-letter abbreviation or as the full
1247     name, and in both cases you want to extract the abbreviation. This pattern
1248     (ignoring the line breaks) does the job:
1249 nigel 91 <pre>
1250 nigel 93 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1251     (?&#60;DN&#62;Tue)(?:sday)?|
1252     (?&#60;DN&#62;Wed)(?:nesday)?|
1253     (?&#60;DN&#62;Thu)(?:rsday)?|
1254     (?&#60;DN&#62;Sat)(?:urday)?
1255 nigel 91 </pre>
1256     There are five capturing substrings, but only one is ever set after a match.
1257 ph10 182 (An alternative way of solving this problem is to use a "branch reset"
1258     subpattern, as described in the previous section.)
1259     </P>
1260     <P>
1261 nigel 91 The convenience function for extracting the data by name returns the substring
1262 nigel 93 for the first (and in this example, the only) subpattern of that name that
1263 ph10 461 matched. This saves searching to find which numbered subpattern it was.
1264     </P>
1265     <P>
1266 ph10 488 If you make a back reference to a non-unique named subpattern from elsewhere in
1267 ph10 461 the pattern, the one that corresponds to the first occurrence of the name is
1268     used. In the absence of duplicate numbers (see the previous section) this is
1269     the one with the lowest number. If you use a named reference in a condition
1270     test (see the
1271     <a href="#conditions">section about conditions</a>
1272     below), either to check whether a subpattern has matched, or to check for
1273     recursion, all subpatterns with the same name are tested. If the condition is
1274     true for any one of them, the overall condition is true. This is the same
1275     behaviour as testing by number. For further details of the interfaces for
1276     handling named subpatterns, see the
1277 nigel 63 <a href="pcreapi.html"><b>pcreapi</b></a>
1278     documentation.
1279     </P>
1280 ph10 392 <P>
1281     <b>Warning:</b> You cannot use different names to distinguish between two
1282 ph10 461 subpatterns with the same number because PCRE uses only the numbers when
1283     matching. For this reason, an error is given at compile time if different names
1284     are given to subpatterns with the same number. However, you can give the same
1285     name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1286 ph10 392 </P>
1287 ph10 227 <br><a name="SEC15" href="#TOC1">REPETITION</a><br>
1288 nigel 63 <P>
1289     Repetition is specified by quantifiers, which can follow any of the following
1290     items:
1291     <pre>
1292     a literal data character
1293 nigel 93 the dot metacharacter
1294 nigel 63 the \C escape sequence
1295 nigel 75 the \X escape sequence (in UTF-8 mode with Unicode properties)
1296 nigel 93 the \R escape sequence
1297 nigel 75 an escape such as \d that matches a single character
1298 nigel 63 a character class
1299     a back reference (see next section)
1300     a parenthesized subpattern (unless it is an assertion)
1301 ph10 461 a recursive or "subroutine" call to a subpattern
1302 nigel 75 </pre>
1303 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1304     permitted matches, by giving the two numbers in curly brackets (braces),
1305     separated by a comma. The numbers must be less than 65536, and the first must
1306     be less than or equal to the second. For example:
1307     <pre>
1308     z{2,4}
1309 nigel 75 </pre>
1310 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1311     character. If the second number is omitted, but the comma is present, there is
1312     no upper limit; if the second number and the comma are both omitted, the
1313     quantifier specifies an exact number of required matches. Thus
1314     <pre>
1315     [aeiou]{3,}
1316 nigel 75 </pre>
1317 nigel 63 matches at least 3 successive vowels, but may match many more, while
1318     <pre>
1319     \d{8}
1320 nigel 75 </pre>
1321 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1322     where a quantifier is not allowed, or one that does not match the syntax of a
1323     quantifier, is taken as a literal character. For example, {,6} is not a
1324     quantifier, but a literal string of four characters.
1325     </P>
1326     <P>
1327     In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1328     bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
1329 nigel 75 which is represented by a two-byte sequence. Similarly, when Unicode property
1330     support is available, \X{3} matches three Unicode extended sequences, each of
1331     which may be several bytes long (and they may be of different lengths).
1332 nigel 63 </P>
1333     <P>
1334     The quantifier {0} is permitted, causing the expression to behave as if the
1335 ph10 345 previous item and the quantifier were not present. This may be useful for
1336     subpatterns that are referenced as
1337     <a href="#subpatternsassubroutines">subroutines</a>
1338     from elsewhere in the pattern. Items other than subpatterns that have a {0}
1339     quantifier are omitted from the compiled pattern.
1340 nigel 63 </P>
1341     <P>
1342 nigel 93 For convenience, the three most common quantifiers have single-character
1343     abbreviations:
1344 nigel 63 <pre>
1345     * is equivalent to {0,}
1346     + is equivalent to {1,}
1347     ? is equivalent to {0,1}
1348 nigel 75 </pre>
1349 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1350     match no characters with a quantifier that has no upper limit, for example:
1351     <pre>
1352     (a?)*
1353 nigel 75 </pre>
1354 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1355     such patterns. However, because there are cases where this can be useful, such
1356     patterns are now accepted, but if any repetition of the subpattern does in fact
1357     match no characters, the loop is forcibly broken.
1358     </P>
1359     <P>
1360     By default, the quantifiers are "greedy", that is, they match as much as
1361     possible (up to the maximum number of permitted times), without causing the
1362     rest of the pattern to fail. The classic example of where this gives problems
1363 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1364     and within the comment, individual * and / characters may appear. An attempt to
1365     match C comments by applying the pattern
1366 nigel 63 <pre>
1367     /\*.*\*/
1368 nigel 75 </pre>
1369 nigel 63 to the string
1370     <pre>
1371 nigel 75 /* first comment */ not comment /* second comment */
1372     </pre>
1373 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1374     item.
1375     </P>
1376     <P>
1377     However, if a quantifier is followed by a question mark, it ceases to be
1378     greedy, and instead matches the minimum number of times possible, so the
1379     pattern
1380     <pre>
1381     /\*.*?\*/
1382 nigel 75 </pre>
1383 nigel 63 does the right thing with the C comments. The meaning of the various
1384     quantifiers is not otherwise changed, just the preferred number of matches.
1385     Do not confuse this use of question mark with its use as a quantifier in its
1386     own right. Because it has two uses, it can sometimes appear doubled, as in
1387     <pre>
1388     \d??\d
1389 nigel 75 </pre>
1390 nigel 63 which matches one digit by preference, but can match two if that is the only
1391     way the rest of the pattern matches.
1392     </P>
1393     <P>
1394 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1395 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1396     greedy by following them with a question mark. In other words, it inverts the
1397     default behaviour.
1398     </P>
1399     <P>
1400     When a parenthesized subpattern is quantified with a minimum repeat count that
1401 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1402 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1403     </P>
1404     <P>
1405     If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1406 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1407 nigel 63 implicitly anchored, because whatever follows will be tried against every
1408     character position in the subject string, so there is no point in retrying the
1409     overall match at any position after the first. PCRE normally treats such a
1410     pattern as though it were preceded by \A.
1411     </P>
1412     <P>
1413     In cases where it is known that the subject string contains no newlines, it is
1414     worth setting PCRE_DOTALL in order to obtain this optimization, or
1415     alternatively using ^ to indicate anchoring explicitly.
1416     </P>
1417     <P>
1418     However, there is one situation where the optimization cannot be used. When .*
1419 ph10 488 is inside capturing parentheses that are the subject of a back reference
1420 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1421     succeeds. Consider, for example:
1422 nigel 63 <pre>
1423     (.*)abc\1
1424 nigel 75 </pre>
1425 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1426     this reason, such a pattern is not implicitly anchored.
1427     </P>
1428     <P>
1429     When a capturing subpattern is repeated, the value captured is the substring
1430     that matched the final iteration. For example, after
1431     <pre>
1432     (tweedle[dume]{3}\s*)+
1433 nigel 75 </pre>
1434 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1435     "tweedledee". However, if there are nested capturing subpatterns, the
1436     corresponding captured values may have been set in previous iterations. For
1437     example, after
1438     <pre>
1439     /(a|(b))+/
1440 nigel 75 </pre>
1441 nigel 63 matches "aba" the value of the second captured substring is "b".
1442 nigel 75 <a name="atomicgroup"></a></P>
1443 ph10 227 <br><a name="SEC16" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1444 nigel 63 <P>
1445 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1446     repetition, failure of what follows normally causes the repeated item to be
1447     re-evaluated to see if a different number of repeats allows the rest of the
1448     pattern to match. Sometimes it is useful to prevent this, either to change the
1449     nature of the match, or to cause it fail earlier than it otherwise might, when
1450     the author of the pattern knows there is no point in carrying on.
1451 nigel 63 </P>
1452     <P>
1453     Consider, for example, the pattern \d+foo when applied to the subject line
1454     <pre>
1455     123456bar
1456 nigel 75 </pre>
1457 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1458     action of the matcher is to try again with only 5 digits matching the \d+
1459     item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1460     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1461     that once a subpattern has matched, it is not to be re-evaluated in this way.
1462     </P>
1463     <P>
1464 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1465 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1466     special parenthesis, starting with (?&#62; as in this example:
1467     <pre>
1468 nigel 73 (?&#62;\d+)foo
1469 nigel 75 </pre>
1470 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1471     it has matched, and a failure further into the pattern is prevented from
1472     backtracking into it. Backtracking past it to previous items, however, works as
1473     normal.
1474     </P>
1475     <P>
1476     An alternative description is that a subpattern of this type matches the string
1477     of characters that an identical standalone pattern would match, if anchored at
1478     the current point in the subject string.
1479     </P>
1480     <P>
1481     Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1482     the above example can be thought of as a maximizing repeat that must swallow
1483     everything it can. So, while both \d+ and \d+? are prepared to adjust the
1484     number of digits they match in order to make the rest of the pattern match,
1485     (?&#62;\d+) can only match an entire sequence of digits.
1486     </P>
1487     <P>
1488     Atomic groups in general can of course contain arbitrarily complicated
1489     subpatterns, and can be nested. However, when the subpattern for an atomic
1490     group is just a single repeated item, as in the example above, a simpler
1491     notation, called a "possessive quantifier" can be used. This consists of an
1492     additional + character following a quantifier. Using this notation, the
1493     previous example can be rewritten as
1494     <pre>
1495 nigel 75 \d++foo
1496     </pre>
1497 ph10 208 Note that a possessive quantifier can be used with an entire group, for
1498     example:
1499     <pre>
1500     (abc|xyz){2,3}+
1501     </pre>
1502 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1503     option is ignored. They are a convenient notation for the simpler forms of
1504 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1505     quantifier and the equivalent atomic group, though there may be a performance
1506     difference; possessive quantifiers should be slightly faster.
1507 nigel 63 </P>
1508     <P>
1509 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1510     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1511     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1512     package, and PCRE copied it from there. It ultimately found its way into Perl
1513     at release 5.10.
1514 nigel 63 </P>
1515     <P>
1516 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1517     pattern constructs. For example, the sequence A+B is treated as A++B because
1518     there is no point in backtracking into a sequence of A's when B must follow.
1519     </P>
1520     <P>
1521 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1522     be repeated an unlimited number of times, the use of an atomic group is the
1523     only way to avoid some failing matches taking a very long time indeed. The
1524     pattern
1525     <pre>
1526     (\D+|&#60;\d+&#62;)*[!?]
1527 nigel 75 </pre>
1528 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1529     digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1530     quickly. However, if it is applied to
1531     <pre>
1532     aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1533 nigel 75 </pre>
1534 nigel 63 it takes a long time before reporting failure. This is because the string can
1535 nigel 75 be divided between the internal \D+ repeat and the external * repeat in a
1536     large number of ways, and all have to be tried. (The example uses [!?] rather
1537     than a single character at the end, because both PCRE and Perl have an
1538     optimization that allows for fast failure when a single character is used. They
1539     remember the last single character that is required for a match, and fail early
1540     if it is not present in the string.) If the pattern is changed so that it uses
1541     an atomic group, like this:
1542 nigel 63 <pre>
1543     ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1544 nigel 75 </pre>
1545 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1546 nigel 75 <a name="backreferences"></a></P>
1547 ph10 227 <br><a name="SEC17" href="#TOC1">BACK REFERENCES</a><br>
1548 nigel 63 <P>
1549     Outside a character class, a backslash followed by a digit greater than 0 (and
1550     possibly further digits) is a back reference to a capturing subpattern earlier
1551     (that is, to its left) in the pattern, provided there have been that many
1552     previous capturing left parentheses.
1553     </P>
1554     <P>
1555     However, if the decimal number following the backslash is less than 10, it is
1556     always taken as a back reference, and causes an error only if there are not
1557     that many capturing left parentheses in the entire pattern. In other words, the
1558     parentheses that are referenced need not be to the left of the reference for
1559 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1560     when a repetition is involved and the subpattern to the right has participated
1561     in an earlier iteration.
1562     </P>
1563     <P>
1564 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1565     whose number is 10 or more using this syntax because a sequence such as \50 is
1566     interpreted as a character defined in octal. See the subsection entitled
1567 nigel 91 "Non-printing characters"
1568 nigel 75 <a href="#digitsafterbackslash">above</a>
1569 nigel 93 for further details of the handling of digits following a backslash. There is
1570     no such problem when named parentheses are used. A back reference to any
1571     subpattern is possible using named parentheses (see below).
1572 nigel 63 </P>
1573     <P>
1574 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1575     backslash is to use the \g escape sequence, which is a feature introduced in
1576 ph10 208 Perl 5.10. This escape must be followed by an unsigned number or a negative
1577     number, optionally enclosed in braces. These examples are all identical:
1578 nigel 93 <pre>
1579     (ring), \1
1580     (ring), \g1
1581     (ring), \g{1}
1582     </pre>
1583 ph10 208 An unsigned number specifies an absolute reference without the ambiguity that
1584     is present in the older syntax. It is also useful when literal digits follow
1585     the reference. A negative number is a relative reference. Consider this
1586     example:
1587 nigel 93 <pre>
1588     (abc(def)ghi)\g{-1}
1589     </pre>
1590     The sequence \g{-1} is a reference to the most recently started capturing
1591     subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2}
1592     would be equivalent to \1. The use of relative references can be helpful in
1593     long patterns, and also in patterns that are created by joining together
1594     fragments that contain references within themselves.
1595     </P>
1596     <P>
1597 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1598     the current subject string, rather than anything matching the subpattern
1599     itself (see
1600     <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1601     below for a way of doing that). So the pattern
1602     <pre>
1603     (sens|respons)e and \1ibility
1604 nigel 75 </pre>
1605 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1606     "sense and responsibility". If caseful matching is in force at the time of the
1607     back reference, the case of letters is relevant. For example,
1608     <pre>
1609     ((?i)rah)\s+\1
1610 nigel 75 </pre>
1611 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1612     capturing subpattern is matched caselessly.
1613     </P>
1614     <P>
1615 ph10 172 There are several different ways of writing back references to named
1616     subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1617     \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1618     back reference syntax, in which \g can be used for both numeric and named
1619     references, is also supported. We could rewrite the above example in any of
1620 nigel 93 the following ways:
1621 nigel 63 <pre>
1622 nigel 93 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1623 ph10 172 (?'p1'(?i)rah)\s+\k{p1}
1624 nigel 91 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1625 ph10 172 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1626 nigel 75 </pre>
1627 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1628     after the reference.
1629     </P>
1630     <P>
1631 nigel 63 There may be more than one back reference to the same subpattern. If a
1632     subpattern has not actually been used in a particular match, any back
1633 ph10 461 references to it always fail by default. For example, the pattern
1634 nigel 63 <pre>
1635     (a|(bc))\2
1636 nigel 75 </pre>
1637 ph10 461 always fails if it starts to match "a" rather than "bc". However, if the
1638     PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1639     unset value matches an empty string.
1640     </P>
1641     <P>
1642     Because there may be many capturing parentheses in a pattern, all digits
1643     following a backslash are taken as part of a potential back reference number.
1644     If the pattern continues with a digit character, some delimiter must be used to
1645     terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1646     whitespace. Otherwise, the \g{ syntax or an empty comment (see
1647 nigel 75 <a href="#comments">"Comments"</a>
1648     below) can be used.
1649 nigel 63 </P>
1650 ph10 488 <br><b>
1651     Recursive back references
1652     </b><br>
1653 nigel 63 <P>
1654     A back reference that occurs inside the parentheses to which it refers fails
1655     when the subpattern is first used, so, for example, (a\1) never matches.
1656     However, such references can be useful inside repeated subpatterns. For
1657     example, the pattern
1658     <pre>
1659     (a|b\1)+
1660 nigel 75 </pre>
1661 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1662     the subpattern, the back reference matches the character string corresponding
1663     to the previous iteration. In order for this to work, the pattern must be such
1664     that the first iteration does not need to match the back reference. This can be
1665     done using alternation, as in the example above, or by a quantifier with a
1666     minimum of zero.
1667 ph10 488 </P>
1668     <P>
1669     Back references of this type cause the group that they reference to be treated
1670     as an
1671     <a href="#atomicgroup">atomic group.</a>
1672     Once the whole group has been matched, a subsequent matching failure cannot
1673     cause backtracking into the middle of the group.
1674 nigel 75 <a name="bigassertions"></a></P>
1675 ph10 227 <br><a name="SEC18" href="#TOC1">ASSERTIONS</a><br>
1676 nigel 63 <P>
1677     An assertion is a test on the characters following or preceding the current
1678     matching point that does not actually consume any characters. The simple
1679 nigel 75 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1680     <a href="#smallassertions">above.</a>
1681     </P>
1682     <P>
1683 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1684     those that look ahead of the current position in the subject string, and those
1685 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1686     except that it does not cause the current matching position to be changed.
1687 nigel 63 </P>
1688     <P>
1689 nigel 75 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1690     because it makes no sense to assert the same thing several times. If any kind
1691     of assertion contains capturing subpatterns within it, these are counted for
1692     the purposes of numbering the capturing subpatterns in the whole pattern.
1693     However, substring capturing is carried out only for positive assertions,
1694     because it does not make sense for negative assertions.
1695 nigel 63 </P>
1696 nigel 75 <br><b>
1697     Lookahead assertions
1698     </b><br>
1699 nigel 63 <P>
1700 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1701     negative assertions. For example,
1702 nigel 63 <pre>
1703     \w+(?=;)
1704 nigel 75 </pre>
1705 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1706     the match, and
1707     <pre>
1708     foo(?!bar)
1709 nigel 75 </pre>
1710 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1711     apparently similar pattern
1712     <pre>
1713     (?!foo)bar
1714 nigel 75 </pre>
1715 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1716     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1717     (?!foo) is always true when the next three characters are "bar". A
1718 nigel 75 lookbehind assertion is needed to achieve the other effect.
1719 nigel 63 </P>
1720     <P>
1721     If you want to force a matching failure at some point in a pattern, the most
1722     convenient way to do it is with (?!) because an empty string always matches, so
1723     an assertion that requires there not to be an empty string must always fail.
1724 ph10 461 The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a
1725     synonym for (?!).
1726 nigel 75 <a name="lookbehind"></a></P>
1727     <br><b>
1728     Lookbehind assertions
1729     </b><br>
1730 nigel 63 <P>
1731     Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1732     negative assertions. For example,
1733     <pre>
1734     (?&#60;!foo)bar
1735 nigel 75 </pre>
1736 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1737     a lookbehind assertion are restricted such that all the strings it matches must
1738 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1739     do not all have to have the same fixed length. Thus
1740 nigel 63 <pre>
1741     (?&#60;=bullock|donkey)
1742 nigel 75 </pre>
1743 nigel 63 is permitted, but
1744     <pre>
1745     (?&#60;!dogs?|cats?)
1746 nigel 75 </pre>
1747 nigel 63 causes an error at compile time. Branches that match different length strings
1748     are permitted only at the top level of a lookbehind assertion. This is an
1749 ph10 454 extension compared with Perl (5.8 and 5.10), which requires all branches to
1750 nigel 63 match the same length of string. An assertion such as
1751     <pre>
1752     (?&#60;=ab(c|de))
1753 nigel 75 </pre>
1754 nigel 63 is not permitted, because its single top-level branch can match two different
1755 ph10 454 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1756     branches:
1757 nigel 63 <pre>
1758     (?&#60;=abc|abde)
1759 nigel 75 </pre>
1760 ph10 172 In some cases, the Perl 5.10 escape sequence \K
1761     <a href="#resetmatchstart">(see above)</a>
1762 ph10 461 can be used instead of a lookbehind assertion to get round the fixed-length
1763 ph10 454 restriction.
1764 ph10 172 </P>
1765     <P>
1766 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
1767 nigel 93 temporarily move the current position back by the fixed length and then try to
1768 nigel 63 match. If there are insufficient characters before the current position, the
1769 nigel 93 assertion fails.
1770 nigel 63 </P>
1771     <P>
1772     PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1773     to appear in lookbehind assertions, because it makes it impossible to calculate
1774 nigel 93 the length of the lookbehind. The \X and \R escapes, which can match
1775     different numbers of bytes, are also not permitted.
1776 nigel 63 </P>
1777     <P>
1778 ph10 454 <a href="#subpatternsassubroutines">"Subroutine"</a>
1779     calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1780 ph10 461 as the subpattern matches a fixed-length string.
1781 ph10 454 <a href="#recursion">Recursion,</a>
1782     however, is not supported.
1783     </P>
1784     <P>
1785 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1786 ph10 461 specify efficient matching of fixed-length strings at the end of subject
1787     strings. Consider a simple pattern such as
1788 nigel 63 <pre>
1789     abcd$
1790 nigel 75 </pre>
1791 nigel 63 when applied to a long string that does not match. Because matching proceeds
1792     from left to right, PCRE will look for each "a" in the subject and then see if
1793     what follows matches the rest of the pattern. If the pattern is specified as
1794     <pre>
1795     ^.*abcd$
1796 nigel 75 </pre>
1797 nigel 63 the initial .* matches the entire string at first, but when this fails (because
1798     there is no following "a"), it backtracks to match all but the last character,
1799     then all but the last two characters, and so on. Once again the search for "a"
1800     covers the entire string, from right to left, so we are no better off. However,
1801     if the pattern is written as
1802     <pre>
1803     ^.*+(?&#60;=abcd)
1804 nigel 75 </pre>
1805 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
1806 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
1807     characters. If it fails, the match fails immediately. For long strings, this
1808     approach makes a significant difference to the processing time.
1809     </P>
1810 nigel 75 <br><b>
1811     Using multiple assertions
1812     </b><br>
1813 nigel 63 <P>
1814     Several assertions (of any sort) may occur in succession. For example,
1815     <pre>
1816     (?&#60;=\d{3})(?&#60;!999)foo
1817 nigel 75 </pre>
1818 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
1819     the assertions is applied independently at the same point in the subject
1820     string. First there is a check that the previous three characters are all
1821     digits, and then there is a check that the same three characters are not "999".
1822     This pattern does <i>not</i> match "foo" preceded by six characters, the first
1823     of which are digits and the last three of which are not "999". For example, it
1824     doesn't match "123abcfoo". A pattern to do that is
1825     <pre>
1826     (?&#60;=\d{3}...)(?&#60;!999)foo
1827 nigel 75 </pre>
1828 nigel 63 This time the first assertion looks at the preceding six characters, checking
1829     that the first three are digits, and then the second assertion checks that the
1830     preceding three characters are not "999".
1831     </P>
1832     <P>
1833     Assertions can be nested in any combination. For example,
1834     <pre>
1835     (?&#60;=(?&#60;!foo)bar)baz
1836 nigel 75 </pre>
1837 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1838     preceded by "foo", while
1839     <pre>
1840     (?&#60;=\d{3}(?!999)...)foo
1841 nigel 75 </pre>
1842     is another pattern that matches "foo" preceded by three digits and any three
1843 nigel 63 characters that are not "999".
1844 nigel 91 <a name="conditions"></a></P>
1845 ph10 227 <br><a name="SEC19" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1846 nigel 63 <P>
1847     It is possible to cause the matching process to obey a subpattern
1848     conditionally or to choose between two alternative subpatterns, depending on
1849 ph10 461 the result of an assertion, or whether a specific capturing subpattern has
1850     already been matched. The two possible forms of conditional subpattern are:
1851 nigel 63 <pre>
1852     (?(condition)yes-pattern)
1853     (?(condition)yes-pattern|no-pattern)
1854 nigel 75 </pre>
1855 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
1856     no-pattern (if present) is used. If there are more than two alternatives in the
1857     subpattern, a compile-time error occurs.
1858     </P>
1859     <P>
1860 nigel 93 There are four kinds of condition: references to subpatterns, references to
1861     recursion, a pseudo-condition called DEFINE, and assertions.
1862 nigel 91 </P>
1863 nigel 93 <br><b>
1864     Checking for a used subpattern by number
1865     </b><br>
1866 nigel 91 <P>
1867 nigel 93 If the text between the parentheses consists of a sequence of digits, the
1868 ph10 461 condition is true if a capturing subpattern of that number has previously
1869     matched. If there is more than one capturing subpattern with the same number
1870     (see the earlier
1871     <a href="#recursion">section about duplicate subpattern numbers),</a>
1872     the condition is true if any of them have been set. An alternative notation is
1873     to precede the digits with a plus or minus sign. In this case, the subpattern
1874     number is relative rather than absolute. The most recently opened parentheses
1875     can be referenced by (?(-1), the next most recent by (?(-2), and so on. In
1876     looping constructs it can also make sense to refer to subsequent groups with
1877     constructs such as (?(+2).
1878 nigel 93 </P>
1879     <P>
1880 nigel 91 Consider the following pattern, which contains non-significant white space to
1881     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1882     three parts for ease of discussion:
1883 nigel 63 <pre>
1884     ( \( )? [^()]+ (?(1) \) )
1885 nigel 75 </pre>
1886 nigel 63 The first part matches an optional opening parenthesis, and if that
1887     character is present, sets it as the first captured substring. The second part
1888     matches one or more characters that are not parentheses. The third part is a
1889     conditional subpattern that tests whether the first set of parentheses matched
1890     or not. If they did, that is, if subject started with an opening parenthesis,
1891     the condition is true, and so the yes-pattern is executed and a closing
1892     parenthesis is required. Otherwise, since no-pattern is not present, the
1893     subpattern matches nothing. In other words, this pattern matches a sequence of
1894 nigel 93 non-parentheses, optionally enclosed in parentheses.
1895     </P>
1896 ph10 172 <P>
1897     If you were embedding this pattern in a larger one, you could use a relative
1898     reference:
1899     <pre>
1900     ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
1901     </pre>
1902     This makes the fragment independent of the parentheses in the larger pattern.
1903     </P>
1904 nigel 93 <br><b>
1905     Checking for a used subpattern by name
1906     </b><br>
1907     <P>
1908     Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
1909     subpattern by name. For compatibility with earlier versions of PCRE, which had
1910     this facility before Perl, the syntax (?(name)...) is also recognized. However,
1911     there is a possible ambiguity with this syntax, because subpattern names may
1912     consist entirely of digits. PCRE looks first for a named subpattern; if it
1913     cannot find one and the name consists entirely of digits, PCRE looks for a
1914     subpattern of that number, which must be greater than zero. Using subpattern
1915     names that consist entirely of digits is not recommended.
1916     </P>
1917     <P>
1918     Rewriting the above example to use a named subpattern gives this:
1919 nigel 91 <pre>
1920 nigel 93 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
1921 ph10 461 </pre>
1922     If the name used in a condition of this kind is a duplicate, the test is
1923     applied to all subpatterns of the same name, and is true if any one of them has
1924     matched.
1925 nigel 93 </P>
1926     <br><b>
1927     Checking for pattern recursion
1928     </b><br>
1929     <P>
1930     If the condition is the string (R), and there is no subpattern with the name R,
1931     the condition is true if a recursive call to the whole pattern or any
1932     subpattern has been made. If digits or a name preceded by ampersand follow the
1933     letter R, for example:
1934     <pre>
1935     (?(R3)...) or (?(R&name)...)
1936 nigel 91 </pre>
1937 ph10 461 the condition is true if the most recent recursion is into a subpattern whose
1938 nigel 93 number or name is given. This condition does not check the entire recursion
1939 ph10 461 stack. If the name used in a condition of this kind is a duplicate, the test is
1940     applied to all subpatterns of the same name, and is true if any one of them is
1941     the most recent recursion.
1942 nigel 63 </P>
1943     <P>
1944 ph10 461 At "top level", all these recursion test conditions are false.
1945     <a href="#recursion">The syntax for recursive patterns</a>
1946     is described below.
1947 nigel 93 </P>
1948     <br><b>
1949     Defining subpatterns for use by reference only
1950     </b><br>
1951     <P>
1952     If the condition is the string (DEFINE), and there is no subpattern with the
1953     name DEFINE, the condition is always false. In this case, there may be only one
1954     alternative in the subpattern. It is always skipped if control reaches this
1955     point in the pattern; the idea of DEFINE is that it can be used to define
1956 ph10 461 "subroutines" that can be referenced from elsewhere. (The use of
1957 ph10 454 <a href="#subpatternsassubroutines">"subroutines"</a>
1958 nigel 93 is described below.) For example, a pattern to match an IPv4 address could be
1959     written like this (ignore whitespace and line breaks):
1960     <pre>
1961     (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
1962     \b (?&byte) (\.(?&byte)){3} \b
1963     </pre>
1964     The first part of the pattern is a DEFINE group inside which a another group
1965     named "byte" is defined. This matches an individual component of an IPv4
1966     address (a number less than 256). When matching takes place, this part of the
1967 ph10 461 pattern is skipped because DEFINE acts like a false condition. The rest of the
1968     pattern uses references to the named group to match the four dot-separated
1969     components of an IPv4 address, insisting on a word boundary at each end.
1970 nigel 93 </P>
1971     <br><b>
1972     Assertion conditions
1973     </b><br>
1974     <P>
1975     If the condition is not in any of the above formats, it must be an assertion.
1976 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
1977     this pattern, again containing non-significant white space, and with the two
1978     alternatives on the second line:
1979     <pre>
1980     (?(?=[^a-z]*[a-z])
1981     \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1982 nigel 75 </pre>
1983 nigel 63 The condition is a positive lookahead assertion that matches an optional
1984     sequence of non-letters followed by a letter. In other words, it tests for the
1985     presence of at least one letter in the subject. If a letter is found, the
1986     subject is matched against the first alternative; otherwise it is matched
1987     against the second. This pattern matches strings in one of the two forms
1988     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1989 nigel 75 <a name="comments"></a></P>
1990 ph10 227 <br><a name="SEC20" href="#TOC1">COMMENTS</a><br>
1991 nigel 63 <P>
1992 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
1993 nigel 63 closing parenthesis. Nested parentheses are not permitted. The characters
1994     that make up a comment play no part in the pattern matching at all.
1995     </P>
1996     <P>
1997     If the PCRE_EXTENDED option is set, an unescaped # character outside a
1998 nigel 91 character class introduces a comment that continues to immediately after the
1999     next newline in the pattern.
2000     <a name="recursion"></a></P>
2001 ph10 227 <br><a name="SEC21" href="#TOC1">RECURSIVE PATTERNS</a><br>
2002 nigel 63 <P>
2003     Consider the problem of matching a string in parentheses, allowing for
2004     unlimited nested parentheses. Without the use of recursion, the best that can
2005     be done is to use a pattern that matches up to some fixed depth of nesting. It
2006 nigel 93 is not possible to handle an arbitrary nesting depth.
2007     </P>
2008     <P>
2009     For some time, Perl has provided a facility that allows regular expressions to
2010     recurse (amongst other things). It does this by interpolating Perl code in the
2011     expression at run time, and the code can refer to the expression itself. A Perl
2012     pattern using code interpolation to solve the parentheses problem can be
2013     created like this:
2014 nigel 63 <pre>
2015     $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
2016 nigel 75 </pre>
2017 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2018 nigel 93 recursively to the pattern in which it appears.
2019 nigel 63 </P>
2020     <P>
2021 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2022     supports special syntax for recursion of the entire pattern, and also for
2023     individual subpattern recursion. After its introduction in PCRE and Python,
2024 ph10 453 this kind of recursion was subsequently introduced into Perl at release 5.10.
2025 nigel 63 </P>
2026     <P>
2027 nigel 93 A special item that consists of (? followed by a number greater than zero and a
2028     closing parenthesis is a recursive call of the subpattern of the given number,
2029 ph10 461 provided that it occurs inside that subpattern. (If not, it is a
2030 ph10 454 <a href="#subpatternsassubroutines">"subroutine"</a>
2031 nigel 93 call, which is described in the next section.) The special item (?R) or (?0) is
2032     a recursive call of the entire regular expression.
2033 nigel 87 </P>
2034     <P>
2035     This PCRE pattern solves the nested parentheses problem (assume the
2036     PCRE_EXTENDED option is set so that white space is ignored):
2037 nigel 63 <pre>
2038 ph10 461 \( ( [^()]++ | (?R) )* \)
2039 nigel 75 </pre>
2040 nigel 63 First it matches an opening parenthesis. Then it matches any number of
2041     substrings which can either be a sequence of non-parentheses, or a recursive
2042 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
2043 ph10 461 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2044     to avoid backtracking into sequences of non-parentheses.
2045 nigel 63 </P>
2046     <P>
2047     If this were part of a larger pattern, you would not want to recurse the entire
2048     pattern, so instead you could use this:
2049     <pre>
2050 ph10 461 ( \( ( [^()]++ | (?1) )* \) )
2051 nigel 75 </pre>
2052 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
2053 ph10 172 them instead of the whole pattern.
2054     </P>
2055     <P>
2056     In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2057 ph10 461 is made easier by the use of relative references (a Perl 5.10 feature).
2058 ph10 172 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
2059     most recently opened parentheses preceding the recursion. In other words, a
2060     negative number counts capturing parentheses leftwards from the point at which
2061     it is encountered.
2062     </P>
2063     <P>
2064     It is also possible to refer to subsequently opened parentheses, by writing
2065     references such as (?+2). However, these cannot be recursive because the
2066     reference is not inside the parentheses that are referenced. They are always
2067 ph10 454 <a href="#subpatternsassubroutines">"subroutine"</a>
2068     calls, as described in the next section.
2069 ph10 172 </P>
2070     <P>
2071     An alternative approach is to use named parentheses instead. The Perl syntax
2072     for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
2073     could rewrite the above example as follows:
2074 nigel 63 <pre>
2075 ph10 461 (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
2076 nigel 75 </pre>
2077 nigel 93 If there is more than one subpattern with the same name, the earliest one is
2078 ph10 172 used.
2079     </P>
2080     <P>
2081     This particular example pattern that we have been looking at contains nested
2082 ph10 461 unlimited repeats, and so the use of a possessive quantifier for matching
2083     strings of non-parentheses is important when applying the pattern to strings
2084     that do not match. For example, when this pattern is applied to
2085 nigel 63 <pre>
2086     (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2087 nigel 75 </pre>
2088 ph10 461 it yields "no match" quickly. However, if a possessive quantifier is not used,
2089 nigel 63 the match runs for a very long time indeed because there are so many different
2090     ways the + and * repeats can carve up the subject, and all have to be tested
2091     before failure can be reported.
2092     </P>
2093     <P>
2094 ph10 469 At the end of a match, the values of capturing parentheses are those from
2095     the outermost level. If you want to obtain intermediate values, a callout
2096     function can be used (see below and the
2097 nigel 63 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2098     documentation). If the pattern above is matched against
2099     <pre>
2100     (ab(cd)ef)
2101 nigel 75 </pre>
2102 ph10 469 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2103     the last value taken on at the top level. If a capturing subpattern is not
2104     matched at the top level, its final value is unset, even if it is (temporarily)
2105     set at a deeper level.
2106 nigel 63 </P>
2107     <P>
2108 ph10 469 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2109     obtain extra memory to store data during a recursion, which it does by using
2110     <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
2111     be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2112     </P>
2113     <P>
2114 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2115     Consider this pattern, which matches text in angle brackets, allowing for
2116     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2117     recursing), whereas any characters are permitted at the outer level.
2118     <pre>
2119     &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
2120 nigel 75 </pre>
2121 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
2122     different alternatives for the recursive and non-recursive cases. The (?R) item
2123     is the actual recursive call.
2124 ph10 453 <a name="recursiondifference"></a></P>
2125     <br><b>
2126     Recursion difference from Perl
2127     </b><br>
2128     <P>
2129     In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2130     treated as an atomic group. That is, once it has matched some of the subject
2131     string, it is never re-entered, even if it contains untried alternatives and
2132 ph10 461 there is a subsequent matching failure. This can be illustrated by the
2133     following pattern, which purports to match a palindromic string that contains
2134 ph10 453 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2135     <pre>
2136     ^(.|(.)(?1)\2)$
2137     </pre>
2138 ph10 461 The idea is that it either matches a single character, or two identical
2139     characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2140 ph10 453 it does not if the pattern is longer than three characters. Consider the
2141     subject string "abcba":
2142     </P>
2143     <P>
2144 ph10 461 At the top level, the first character is matched, but as it is not at the end
2145 ph10 453 of the string, the first alternative fails; the second alternative is taken
2146     and the recursion kicks in. The recursive call to subpattern 1 successfully
2147     matches the next character ("b"). (Note that the beginning and end of line
2148     tests are not part of the recursion).
2149     </P>
2150     <P>
2151     Back at the top level, the next character ("c") is compared with what
2152 ph10 461 subpattern 2 matched, which was "a". This fails. Because the recursion is
2153 ph10 453 treated as an atomic group, there are now no backtracking points, and so the
2154     entire match fails. (Perl is able, at this point, to re-enter the recursion and
2155     try the second alternative.) However, if the pattern is written with the
2156     alternatives in the other order, things are different:
2157     <pre>
2158     ^((.)(?1)\2|.)$
2159     </pre>
2160 ph10 461 This time, the recursing alternative is tried first, and continues to recurse
2161     until it runs out of characters, at which point the recursion fails. But this
2162     time we do have another alternative to try at the higher level. That is the big
2163 ph10 453 difference: in the previous case the remaining alternative is at a deeper
2164     recursion level, which PCRE cannot use.
2165     </P>
2166     <P>
2167 ph10 461 To change the pattern so that matches all palindromic strings, not just those
2168 ph10 453 with an odd number of characters, it is tempting to change the pattern to this:
2169     <pre>
2170     ^((.)(?1)\2|.?)$
2171     </pre>
2172 ph10 461 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2173     deeper recursion has matched a single character, it cannot be entered again in
2174     order to match an empty string. The solution is to separate the two cases, and
2175 ph10 453 write out the odd and even cases as alternatives at the higher level:
2176     <pre>
2177     ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
2178     </pre>
2179 ph10 461 If you want to match typical palindromic phrases, the pattern has to ignore all
2180 ph10 453 non-word characters, which can be done like this:
2181     <pre>
2182     ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
2183     </pre>
2184 ph10 461 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2185     man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2186     the use of the possessive quantifier *+ to avoid backtracking into sequences of
2187 ph10 453 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2188     more) to match typical phrases, and Perl takes so long that you think it has
2189     gone into a loop.
2190 ph10 461 </P>
2191     <P>
2192     <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
2193     string does not start with a palindrome that is shorter than the entire string.
2194     For example, although "abcba" is correctly matched, if the subject is "ababa",
2195     PCRE finds the palindrome "aba" at the start, then fails at top level because
2196     the end of the string does not follow. Once again, it cannot jump back into the
2197     recursion to try other alternatives, so the entire match fails.
2198 nigel 75 <a name="subpatternsassubroutines"></a></P>
2199 ph10 227 <br><a name="SEC22" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
2200 nigel 63 <P>
2201     If the syntax for a recursive subpattern reference (either by number or by
2202     name) is used outside the parentheses to which it refers, it operates like a
2203 nigel 93 subroutine in a programming language. The "called" subpattern may be defined
2204 ph10 172 before or after the reference. A numbered reference can be absolute or
2205     relative, as in these examples:
2206 nigel 63 <pre>
2207 ph10 172 (...(absolute)...)...(?2)...
2208     (...(relative)...)...(?-1)...
2209     (...(?+1)...(relative)...
2210     </pre>
2211     An earlier example pointed out that the pattern
2212     <pre>
2213 nigel 63 (sens|respons)e and \1ibility
2214 nigel 75 </pre>
2215 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
2216     "sense and responsibility". If instead the pattern
2217     <pre>
2218     (sens|respons)e and (?1)ibility
2219 nigel 75 </pre>
2220 nigel 63 is used, it does match "sense and responsibility" as well as the other two
2221 nigel 93 strings. Another example is given in the discussion of DEFINE above.
2222 nigel 63 </P>
2223 nigel 87 <P>
2224 ph10 469 Like recursive subpatterns, a subroutine call is always treated as an atomic
2225 nigel 87 group. That is, once it has matched some of the subject string, it is never
2226     re-entered, even if it contains untried alternatives and there is a subsequent
2227 ph10 469 matching failure. Any capturing parentheses that are set during the subroutine
2228     call revert to their previous values afterwards.
2229 nigel 87 </P>
2230 nigel 63 <P>
2231 nigel 93 When a subpattern is used as a subroutine, processing options such as
2232     case-independence are fixed when the subpattern is defined. They cannot be
2233     changed for different calls. For example, consider this pattern:
2234     <pre>
2235 ph10 172 (abc)(?i:(?-1))
2236 nigel 93 </pre>
2237     It matches "abcabc". It does not match "abcABC" because the change of
2238     processing option does not affect the called subpattern.
2239 ph10 345 <a name="onigurumasubroutines"></a></P>
2240     <br><a name="SEC23" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
2241     <P>
2242     For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
2243     a number enclosed either in angle brackets or single quotes, is an alternative
2244     syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2245     are two of the examples used above, rewritten using this syntax:
2246     <pre>
2247     (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
2248     (sens|respons)e and \g'1'ibility
2249     </pre>
2250     PCRE supports an extension to Oniguruma: if a number is preceded by a
2251     plus or a minus sign it is taken as a relative reference. For example:
2252     <pre>
2253     (abc)(?i:\g&#60;-1&#62;)
2254     </pre>
2255     Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
2256     synonymous. The former is a back reference; the latter is a subroutine call.
2257 nigel 93 </P>
2258 ph10 345 <br><a name="SEC24" href="#TOC1">CALLOUTS</a><br>
2259 nigel 93 <P>
2260 nigel 63 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2261     code to be obeyed in the middle of matching a regular expression. This makes it
2262     possible, amongst other things, to extract different substrings that match the
2263     same pair of parentheses when there is a repetition.
2264     </P>
2265     <P>
2266     PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2267     code. The feature is called "callout". The caller of PCRE provides an external
2268     function by putting its entry point in the global variable <i>pcre_callout</i>.
2269     By default, this variable contains NULL, which disables all calling out.
2270     </P>
2271     <P>
2272     Within a regular expression, (?C) indicates the points at which the external
2273     function is to be called. If you want to identify different callout points, you
2274     can put a number less than 256 after the letter C. The default value is zero.
2275     For example, this pattern has two callout points:
2276     <pre>
2277 ph10 155 (?C1)abc(?C2)def
2278 nigel 75 </pre>
2279     If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
2280     automatically installed before each item in the pattern. They are all numbered
2281     255.
2282 nigel 63 </P>
2283     <P>
2284     During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
2285     set), the external function is called. It is provided with the number of the
2286 nigel 75 callout, the position in the pattern, and, optionally, one item of data
2287     originally supplied by the caller of <b>pcre_exec()</b>. The callout function
2288     may cause matching to proceed, to backtrack, or to fail altogether. A complete
2289     description of the interface to the callout function is given in the
2290 nigel 63 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2291     documentation.
2292     </P>
2293 ph10 345 <br><a name="SEC25" href="#TOC1">BACKTRACKING CONTROL</a><br>
2294 nigel 63 <P>
2295 ph10 211 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2296     are described in the Perl documentation as "experimental and subject to change
2297     or removal in a future version of Perl". It goes on to say: "Their usage in
2298     production code should be noted to avoid problems during upgrades." The same
2299     remarks apply to the PCRE features described in this section.
2300     </P>
2301     <P>
2302 ph10 345 Since these verbs are specifically related to backtracking, most of them can be
2303     used only when the pattern is to be matched using <b>pcre_exec()</b>, which uses
2304     a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2305     failing negative assertion, they cause an error if encountered by
2306 ph10 211 <b>pcre_dfa_exec()</b>.
2307     </P>
2308     <P>
2309 ph10 469 If any of these verbs are used in an assertion or subroutine subpattern
2310     (including recursive subpatterns), their effect is confined to that subpattern;
2311     it does not extend to the surrounding pattern. Note that such subpatterns are
2312     processed as anchored at the point where they are tested.
2313 ph10 453 </P>
2314     <P>
2315 ph10 211 The new verbs make use of what was previously invalid syntax: an opening
2316     parenthesis followed by an asterisk. In Perl, they are generally of the form
2317     (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2318     form is just (*VERB). Any number of these verbs may occur in a pattern. There
2319     are two kinds:
2320     </P>
2321     <br><b>
2322     Verbs that act immediately
2323     </b><br>
2324     <P>
2325     The following verbs act as soon as they are encountered:
2326     <pre>
2327     (*ACCEPT)
2328     </pre>
2329     This verb causes the match to end successfully, skipping the remainder of the
2330     pattern. When inside a recursion, only the innermost pattern is ended
2331 ph10 461 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2332     captured. (This feature was added to PCRE at release 8.00.) For example:
2333 ph10 211 <pre>
2334 ph10 453 A((?:A|B(*ACCEPT)|C)D)
2335 ph10 211 </pre>
2336 ph10 461 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2337 ph10 453 the outer parentheses.
2338 ph10 211 <pre>
2339     (*FAIL) or (*F)
2340     </pre>
2341     This verb causes the match to fail, forcing backtracking to occur. It is
2342     equivalent to (?!) but easier to read. The Perl documentation notes that it is
2343     probably useful only when combined with (?{}) or (??{}). Those are, of course,
2344     Perl features that are not present in PCRE. The nearest equivalent is the
2345     callout feature, as for example in this pattern:
2346     <pre>
2347     a+(?C)(*FAIL)
2348     </pre>
2349     A match with the string "aaaa" always fails, but the callout is taken before
2350     each backtrack happens (in this example, 10 times).
2351     </P>
2352     <br><b>
2353     Verbs that act after backtracking
2354     </b><br>
2355     <P>
2356     The following verbs do nothing when they are encountered. Matching continues
2357     with what follows, but if there is no subsequent match, a failure is forced.
2358     The verbs differ in exactly what kind of failure occurs.
2359     <pre>
2360     (*COMMIT)
2361     </pre>
2362     This verb causes the whole match to fail outright if the rest of the pattern
2363     does not match. Even if the pattern is unanchored, no further attempts to find
2364 ph10 461 a match by advancing the starting point take place. Once (*COMMIT) has been
2365 ph10 211 passed, <b>pcre_exec()</b> is committed to finding a match at the current
2366     starting point, or not at all. For example:
2367     <pre>
2368     a+(*COMMIT)b
2369     </pre>
2370     This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2371     dynamic anchor, or "I've started, so I must finish."
2372     <pre>
2373     (*PRUNE)
2374     </pre>
2375     This verb causes the match to fail at the current position if the rest of the
2376     pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2377     advance to the next starting character then happens. Backtracking can occur as
2378     usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2379     if there is no match to the right, backtracking cannot cross (*PRUNE).
2380     In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2381     group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2382     be expressed in any other way.
2383     <pre>
2384     (*SKIP)
2385     </pre>
2386     This verb is like (*PRUNE), except that if the pattern is unanchored, the
2387     "bumpalong" advance is not to the next character, but to the position in the
2388     subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2389     was matched leading up to it cannot be part of a successful match. Consider:
2390     <pre>
2391     a+(*SKIP)b
2392     </pre>
2393     If the subject is "aaaac...", after the first match attempt fails (starting at
2394     the first character in the string), the starting point skips on to start the
2395     next attempt at "c". Note that a possessive quantifer does not have the same
2396 ph10 461 effect as this example; although it would suppress backtracking during the
2397 ph10 211 first match attempt, the second attempt would start at the second character
2398     instead of skipping on to "c".
2399     <pre>
2400     (*THEN)
2401     </pre>
2402     This verb causes a skip to the next alternation if the rest of the pattern does
2403     not match. That is, it cancels pending backtracking, but only within the
2404     current alternation. Its name comes from the observation that it can be used
2405     for a pattern-based if-then-else block:
2406     <pre>
2407     ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2408     </pre>
2409     If the COND1 pattern matches, FOO is tried (and possibly further items after
2410     the end of the group if FOO succeeds); on failure the matcher skips to the
2411     second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2412     is used outside of any alternation, it acts exactly like (*PRUNE).
2413     </P>
2414 ph10 345 <br><a name="SEC26" href="#TOC1">SEE ALSO</a><br>
2415 ph10 211 <P>
2416 ph10 461 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
2417     <b>pcresyntax</b>(3), <b>pcre</b>(3).
2418 nigel 93 </P>
2419 ph10 345 <br><a name="SEC27" href="#TOC1">AUTHOR</a><br>
2420 nigel 93 <P>
2421 ph10 99 Philip Hazel
2422 nigel 63 <br>
2423 ph10 99 University Computing Service
2424     <br>
2425     Cambridge CB2 3QH, England.
2426     <br>
2427     </P>
2428 ph10 345 <br><a name="SEC28" href="#TOC1">REVISION</a><br>
2429 ph10 99 <P>
2430 ph10 507 Last updated: 06 March 2010
2431 ph10 99 <br>
2432 ph10 488 Copyright &copy; 1997-2010 University of Cambridge.
2433 ph10 99 <br>
2434 nigel 75 <p>
2435     Return to the <a href="index.html">PCRE index page</a>.
2436     </p>

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12