ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 139 - (hide annotations) (download)
Fri Mar 30 13:41:47 2007 UTC (8 years ago) by ph10
File size: 74674 byte(s)
Documentation notes that fr_FR locale is "french" in Windows; add an 
out-of-tree built test to maint/ManyConfigTests.

1 nigel 79 .TH PCREPATTERN 3
2 nigel 63 .SH NAME
3     PCRE - Perl-compatible regular expressions
5 nigel 63 .rs
6     .sp
7     The syntax and semantics of the regular expressions supported by PCRE are
8     described below. Regular expressions are also described in the Perl
9 nigel 75 documentation and in a number of books, some of which have copious examples.
10     Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers
11     regular expressions in great detail. This description of PCRE's regular
12     expressions is intended as reference material.
13     .P
14     The original operation of PCRE was on strings of one-byte characters. However,
15     there is now also support for UTF-8 character strings. To use this, you must
16     build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
17     the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
18     places below. There is also a summary of UTF-8 features in the
19 nigel 63 .\" HTML <a href="pcre.html#utf8support">
20     .\" </a>
21     section on UTF-8 support
22     .\"
23     in the main
24     .\" HREF
25 nigel 75 \fBpcre\fP
26 nigel 63 .\"
27     page.
28 nigel 75 .P
29 nigel 77 The remainder of this document discusses the patterns that are supported by
30     PCRE when its main matching function, \fBpcre_exec()\fP, is used.
31     From release 6.0, PCRE offers a second matching function,
32     \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
33     Perl-compatible. The advantages and disadvantages of the alternative function,
34     and how it differs from the normal function, are discussed in the
35     .\" HREF
36     \fBpcrematching\fP
37     .\"
38     page.
39 nigel 93 .
40     .
42     .rs
43     .sp
44 nigel 63 A regular expression is a pattern that is matched against a subject string from
45     left to right. Most characters stand for themselves in a pattern, and match the
46     corresponding characters in the subject. As a trivial example, the pattern
47 nigel 75 .sp
48 nigel 63 The quick brown fox
49 nigel 75 .sp
50 nigel 77 matches a portion of a subject string that is identical to itself. When
51     caseless matching is specified (the PCRE_CASELESS option), letters are matched
52     independently of case. In UTF-8 mode, PCRE always understands the concept of
53     case for characters whose values are less than 128, so caseless matching is
54     always possible. For characters with higher values, the concept of case is
55     supported if PCRE is compiled with Unicode property support, but not otherwise.
56     If you want to use caseless matching for characters 128 and above, you must
57     ensure that PCRE is compiled with Unicode property support as well as with
58     UTF-8 support.
59     .P
60     The power of regular expressions comes from the ability to include alternatives
61     and repetitions in the pattern. These are encoded in the pattern by the use of
62 nigel 75 \fImetacharacters\fP, which do not stand for themselves but instead are
63 nigel 63 interpreted in some special way.
64 nigel 75 .P
65     There are two different sets of metacharacters: those that are recognized
66 nigel 63 anywhere in the pattern except within square brackets, and those that are
67 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
68     are as follows:
69 nigel 75 .sp
70     \e general escape character with several uses
71 nigel 63 ^ assert start of string (or line, in multiline mode)
72     $ assert end of string (or line, in multiline mode)
73     . match any character except newline (by default)
74     [ start character class definition
75     | start of alternative branch
76     ( start subpattern
77     ) end subpattern
78     ? extends the meaning of (
79     also 0 or 1 quantifier
80     also quantifier minimizer
81     * 0 or more quantifier
82     + 1 or more quantifier
83     also "possessive quantifier"
84     { start min/max quantifier
85 nigel 75 .sp
86 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
87 nigel 75 a character class the only metacharacters are:
88     .sp
89     \e general escape character
90 nigel 63 ^ negate the class, but only if the first character
91     - indicates character range
92 nigel 75 .\" JOIN
93 nigel 63 [ POSIX character class (only if followed by POSIX
94     syntax)
95     ] terminates the character class
96 nigel 75 .sp
97     The following sections describe the use of each of the metacharacters.
98     .
99 nigel 93 .
100 nigel 63 .SH BACKSLASH
101     .rs
102     .sp
103     The backslash character has several uses. Firstly, if it is followed by a
104 nigel 91 non-alphanumeric character, it takes away any special meaning that character
105     may have. This use of backslash as an escape character applies both inside and
106 nigel 63 outside character classes.
107 nigel 75 .P
108     For example, if you want to match a * character, you write \e* in the pattern.
109 nigel 63 This escaping action applies whether or not the following character would
110 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
111     non-alphanumeric with backslash to specify that it stands for itself. In
112     particular, if you want to match a backslash, you write \e\e.
113     .P
114 nigel 63 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
115     pattern (other than in a character class) and characters between a # outside
116 nigel 91 a character class and the next newline are ignored. An escaping backslash can
117     be used to include a whitespace or # character as part of the pattern.
118 nigel 75 .P
119 nigel 63 If you want to remove the special meaning from a sequence of characters, you
120 nigel 75 can do so by putting them between \eQ and \eE. This is different from Perl in
121     that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
122 nigel 63 Perl, $ and @ cause variable interpolation. Note the following examples:
123 nigel 75 .sp
124 nigel 63 Pattern PCRE matches Perl matches
125 nigel 75 .sp
126     .\" JOIN
127     \eQabc$xyz\eE abc$xyz abc followed by the
128 nigel 63 contents of $xyz
129 nigel 75 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
130     \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
131     .sp
132     The \eQ...\eE sequence is recognized both inside and outside character classes.
133     .
134     .
135     .\" HTML <a name="digitsafterbackslash"></a>
136     .SS "Non-printing characters"
137     .rs
138     .sp
139 nigel 63 A second use of backslash provides a way of encoding non-printing characters
140     in patterns in a visible manner. There is no restriction on the appearance of
141     non-printing characters, apart from the binary zero that terminates a pattern,
142     but when a pattern is being prepared by text editing, it is usually easier to
143     use one of the following escape sequences than the binary character it
144     represents:
145 nigel 75 .sp
146     \ea alarm, that is, the BEL character (hex 07)
147     \ecx "control-x", where x is any character
148     \ee escape (hex 1B)
149     \ef formfeed (hex 0C)
150     \en newline (hex 0A)
151     \er carriage return (hex 0D)
152     \et tab (hex 09)
153     \eddd character with octal code ddd, or backreference
154     \exhh character with hex code hh
155 nigel 87 \ex{hhh..} character with hex code hhh..
156 nigel 75 .sp
157     The precise effect of \ecx is as follows: if x is a lower case letter, it
158 nigel 63 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
159 nigel 75 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
160 nigel 63 7B.
161 nigel 75 .P
162     After \ex, from zero to two hexadecimal digits are read (letters can be in
163 nigel 87 upper or lower case). Any number of hexadecimal digits may appear between \ex{
164     and }, but the value of the character code must be less than 256 in non-UTF-8
165     mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
166     is 7FFFFFFF). If characters other than hexadecimal digits appear between \ex{
167     and }, or if there is no terminating }, this form of escape is not recognized.
168     Instead, the initial \ex will be interpreted as a basic hexadecimal escape,
169     with no following digits, giving a character whose value is zero.
170 nigel 75 .P
171 nigel 63 Characters whose value is less than 256 can be defined by either of the two
172 nigel 87 syntaxes for \ex. There is no difference in the way they are handled. For
173     example, \exdc is exactly the same as \ex{dc}.
174 nigel 75 .P
175 nigel 91 After \e0 up to two further octal digits are read. If there are fewer than two
176     digits, just those that are present are used. Thus the sequence \e0\ex\e07
177     specifies two binary zeros followed by a BEL character (code value 7). Make
178     sure you supply two digits after the initial zero if the pattern character that
179     follows is itself an octal digit.
180 nigel 75 .P
181 nigel 63 The handling of a backslash followed by a digit other than 0 is complicated.
182     Outside a character class, PCRE reads it and any following digits as a decimal
183     number. If the number is less than 10, or if there have been at least that many
184     previous capturing left parentheses in the expression, the entire sequence is
185 nigel 75 taken as a \fIback reference\fP. A description of how this works is given
186     .\" HTML <a href="#backreferences">
187     .\" </a>
188     later,
189     .\"
190     following the discussion of
191     .\" HTML <a href="#subpattern">
192     .\" </a>
193     parenthesized subpatterns.
194     .\"
195     .P
196 nigel 63 Inside a character class, or if the decimal number is greater than 9 and there
197     have not been that many capturing subpatterns, PCRE re-reads up to three octal
198 nigel 93 digits following the backslash, and uses them to generate a data character. Any
199 nigel 91 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
200     character specified in octal must be less than \e400. In UTF-8 mode, values up
201     to \e777 are permitted. For example:
202 nigel 75 .sp
203     \e040 is another way of writing a space
204     .\" JOIN
205     \e40 is the same, provided there are fewer than 40
206 nigel 63 previous capturing subpatterns
207 nigel 75 \e7 is always a back reference
208     .\" JOIN
209     \e11 might be a back reference, or another way of
210 nigel 63 writing a tab
211 nigel 75 \e011 is always a tab
212     \e0113 is a tab followed by the character "3"
213     .\" JOIN
214     \e113 might be a back reference, otherwise the
215 nigel 63 character with octal code 113
216 nigel 75 .\" JOIN
217     \e377 might be a back reference, otherwise
218 nigel 63 the byte consisting entirely of 1 bits
219 nigel 75 .\" JOIN
220     \e81 is either a back reference, or a binary zero
221 nigel 63 followed by the two characters "8" and "1"
222 nigel 75 .sp
223 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
224     zero, because no more than three octal digits are ever read.
225 nigel 75 .P
226 nigel 91 All the sequences that define a single character value can be used both inside
227     and outside character classes. In addition, inside a character class, the
228     sequence \eb is interpreted as the backspace character (hex 08), and the
229 nigel 93 sequences \eR and \eX are interpreted as the characters "R" and "X",
230     respectively. Outside a character class, these sequences have different
231     meanings
232 nigel 75 .\" HTML <a href="#uniextseq">
233     .\" </a>
234     (see below).
235     .\"
236     .
237     .
238 nigel 93 .SS "Absolute and relative back references"
239     .rs
240     .sp
241     The sequence \eg followed by a positive or negative number, optionally enclosed
242     in braces, is an absolute or relative back reference. Back references are
243     discussed
244     .\" HTML <a href="#backreferences">
245     .\" </a>
246     later,
247     .\"
248     following the discussion of
249     .\" HTML <a href="#subpattern">
250     .\" </a>
251     parenthesized subpatterns.
252     .\"
253     .
254     .
255 nigel 75 .SS "Generic character types"
256     .rs
257     .sp
258 nigel 93 Another use of backslash is for specifying generic character types. The
259 nigel 75 following are always recognized:
260     .sp
261     \ed any decimal digit
262     \eD any character that is not a decimal digit
263     \es any whitespace character
264     \eS any character that is not a whitespace character
265     \ew any "word" character
266     \eW any "non-word" character
267     .sp
268 nigel 63 Each pair of escape sequences partitions the complete set of characters into
269     two disjoint sets. Any given character matches one, and only one, of each pair.
270 nigel 75 .P
271     These character type sequences can appear both inside and outside character
272     classes. They each match one character of the appropriate type. If the current
273     matching point is at the end of the subject string, all of them fail, since
274     there is no character to match.
275     .P
276     For compatibility with Perl, \es does not match the VT character (code 11).
277     This makes it different from the the POSIX "space" class. The \es characters
278 nigel 91 are HT (9), LF (10), FF (12), CR (13), and space (32). (If "use locale;" is
279     included in a Perl script, \es may match the VT character. In PCRE, it never
280     does.)
281 nigel 75 .P
282     A "word" character is an underscore or any character less than 256 that is a
283     letter or digit. The definition of letters and digits is controlled by PCRE's
284     low-valued character tables, and may vary if locale-specific matching is taking
285     place (see
286 nigel 63 .\" HTML <a href="pcreapi.html#localesupport">
287     .\" </a>
288     "Locale support"
289     .\"
290     in the
291     .\" HREF
292 nigel 75 \fBpcreapi\fP
293 nigel 63 .\"
294 ph10 139 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
295     or "french" in Windows, some character codes greater than 128 are used for
296     accented letters, and these are matched by \ew.
297 nigel 75 .P
298     In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
299     \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
300 nigel 87 character property support is available. The use of locales with Unicode is
301     discouraged.
302 nigel 75 .
303     .
304 nigel 93 .SS "Newline sequences"
305     .rs
306     .sp
307     Outside a character class, the escape sequence \eR matches any Unicode newline
308     sequence. This is an extension to Perl. In non-UTF-8 mode \eR is equivalent to
309     the following:
310     .sp
311     (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
312     .sp
313     This is an example of an "atomic group", details of which are given
314     .\" HTML <a href="#atomicgroup">
315     .\" </a>
316     below.
317     .\"
318     This particular group matches either the two-character sequence CR followed by
319     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
320     U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
321     line, U+0085). The two-character sequence is treated as a single unit that
322     cannot be split.
323     .P
324     In UTF-8 mode, two additional characters whose codepoints are greater than 255
325     are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
326     Unicode character property support is not needed for these characters to be
327     recognized.
328     .P
329     Inside a character class, \eR matches the letter "R".
330     .
331     .
332 nigel 75 .\" HTML <a name="uniextseq"></a>
333     .SS Unicode character properties
334     .rs
335     .sp
336     When PCRE is built with Unicode character property support, three additional
337 nigel 87 escape sequences to match character properties are available when UTF-8 mode
338 nigel 75 is selected. They are:
339     .sp
340 nigel 87 \ep{\fIxx\fP} a character with the \fIxx\fP property
341     \eP{\fIxx\fP} a character without the \fIxx\fP property
342     \eX an extended Unicode sequence
343 nigel 75 .sp
344 nigel 87 The property names represented by \fIxx\fP above are limited to the Unicode
345     script names, the general category properties, and "Any", which matches any
346     character (including newline). Other properties such as "InMusicalSymbols" are
347     not currently supported by PCRE. Note that \eP{Any} does not match any
348     characters, so always causes a match failure.
349 nigel 75 .P
350 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
351     character from one of these sets can be matched using a script name. For
352     example:
353 nigel 75 .sp
354 nigel 87 \ep{Greek}
355     \eP{Han}
356     .sp
357     Those that are not part of an identified script are lumped together as
358     "Common". The current list of scripts is:
359     .P
360     Arabic,
361     Armenian,
362 nigel 93 Balinese,
363 nigel 87 Bengali,
364     Bopomofo,
365     Braille,
366     Buginese,
367     Buhid,
368     Canadian_Aboriginal,
369     Cherokee,
370     Common,
371     Coptic,
372 nigel 93 Cuneiform,
373 nigel 87 Cypriot,
374     Cyrillic,
375     Deseret,
376     Devanagari,
377     Ethiopic,
378     Georgian,
379     Glagolitic,
380     Gothic,
381     Greek,
382     Gujarati,
383     Gurmukhi,
384     Han,
385     Hangul,
386     Hanunoo,
387     Hebrew,
388     Hiragana,
389     Inherited,
390     Kannada,
391     Katakana,
392     Kharoshthi,
393     Khmer,
394     Lao,
395     Latin,
396     Limbu,
397     Linear_B,
398     Malayalam,
399     Mongolian,
400     Myanmar,
401     New_Tai_Lue,
402 nigel 93 Nko,
403 nigel 87 Ogham,
404     Old_Italic,
405     Old_Persian,
406     Oriya,
407     Osmanya,
408 nigel 93 Phags_Pa,
409     Phoenician,
410 nigel 87 Runic,
411     Shavian,
412     Sinhala,
413     Syloti_Nagri,
414     Syriac,
415     Tagalog,
416     Tagbanwa,
417     Tai_Le,
418     Tamil,
419     Telugu,
420     Thaana,
421     Thai,
422     Tibetan,
423     Tifinagh,
424     Ugaritic,
425     Yi.
426     .P
427     Each character has exactly one general category property, specified by a
428     two-letter abbreviation. For compatibility with Perl, negation can be specified
429     by including a circumflex between the opening brace and the property name. For
430     example, \ep{^Lu} is the same as \eP{Lu}.
431     .P
432     If only one letter is specified with \ep or \eP, it includes all the general
433     category properties that start with that letter. In this case, in the absence
434     of negation, the curly brackets in the escape sequence are optional; these two
435     examples have the same effect:
436     .sp
437 nigel 75 \ep{L}
438     \epL
439     .sp
440 nigel 87 The following general category property codes are supported:
441 nigel 75 .sp
442     C Other
443     Cc Control
444     Cf Format
445     Cn Unassigned
446     Co Private use
447     Cs Surrogate
448     .sp
449     L Letter
450     Ll Lower case letter
451     Lm Modifier letter
452     Lo Other letter
453     Lt Title case letter
454     Lu Upper case letter
455     .sp
456     M Mark
457     Mc Spacing mark
458     Me Enclosing mark
459     Mn Non-spacing mark
460     .sp
461     N Number
462     Nd Decimal number
463     Nl Letter number
464     No Other number
465     .sp
466     P Punctuation
467     Pc Connector punctuation
468     Pd Dash punctuation
469     Pe Close punctuation
470     Pf Final punctuation
471     Pi Initial punctuation
472     Po Other punctuation
473     Ps Open punctuation
474     .sp
475     S Symbol
476     Sc Currency symbol
477     Sk Modifier symbol
478     Sm Mathematical symbol
479     So Other symbol
480     .sp
481     Z Separator
482     Zl Line separator
483     Zp Paragraph separator
484     Zs Space separator
485     .sp
486 nigel 87 The special property L& is also supported: it matches a character that has
487     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
488     a modifier or "other".
489 nigel 75 .P
490 nigel 87 The long synonyms for these properties that Perl supports (such as \ep{Letter})
491 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
492 nigel 87 properties with "Is".
493     .P
494     No character that is in the Unicode table has the Cn (unassigned) property.
495     Instead, this property is assumed for any code point that is not in the
496     Unicode table.
497     .P
498 nigel 75 Specifying caseless matching does not affect these escape sequences. For
499     example, \ep{Lu} always matches only upper case letters.
500     .P
501     The \eX escape matches any number of Unicode characters that form an extended
502     Unicode sequence. \eX is equivalent to
503     .sp
504     (?>\ePM\epM*)
505     .sp
506     That is, it matches a character without the "mark" property, followed by zero
507     or more characters with the "mark" property, and treats the sequence as an
508     atomic group
509     .\" HTML <a href="#atomicgroup">
510     .\" </a>
511     (see below).
512     .\"
513     Characters with the "mark" property are typically accents that affect the
514     preceding character.
515     .P
516     Matching characters by Unicode property is not fast, because PCRE has to search
517     a structure that contains data for over fifteen thousand characters. That is
518     why the traditional escape sequences such as \ed and \ew do not use Unicode
519     properties in PCRE.
520     .
521     .
522     .\" HTML <a name="smallassertions"></a>
523     .SS "Simple assertions"
524     .rs
525     .sp
526 nigel 93 The final use of backslash is for certain simple assertions. An assertion
527 nigel 63 specifies a condition that has to be met at a particular point in a match,
528     without consuming any characters from the subject string. The use of
529 nigel 75 subpatterns for more complicated assertions is described
530     .\" HTML <a href="#bigassertions">
531     .\" </a>
532     below.
533     .\"
534 nigel 91 The backslashed assertions are:
535 nigel 75 .sp
536     \eb matches at a word boundary
537     \eB matches when not at a word boundary
538 nigel 93 \eA matches at the start of the subject
539     \eZ matches at the end of the subject
540     also matches before a newline at the end of the subject
541     \ez matches only at the end of the subject
542     \eG matches at the first matching position in the subject
543 nigel 75 .sp
544     These assertions may not appear in character classes (but note that \eb has a
545 nigel 63 different meaning, namely the backspace character, inside a character class).
546 nigel 75 .P
547 nigel 63 A word boundary is a position in the subject string where the current character
548 nigel 75 and the previous character do not both match \ew or \eW (i.e. one matches
549     \ew and the other matches \eW), or the start or end of the string if the
550     first or last character matches \ew, respectively.
551     .P
552     The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
553     dollar (described in the next section) in that they only ever match at the very
554     start and end of the subject string, whatever options are set. Thus, they are
555     independent of multiline mode. These three assertions are not affected by the
556     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
557     circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
558     argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
559     at a point other than the beginning of the subject, \eA can never match. The
560 nigel 91 difference between \eZ and \ez is that \eZ matches before a newline at the end
561     of the string as well as at the very end, whereas \ez matches only at the end.
562 nigel 75 .P
563     The \eG assertion is true only when the current matching position is at the
564     start point of the match, as specified by the \fIstartoffset\fP argument of
565     \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
566     non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
567 nigel 63 arguments, you can mimic Perl's /g option, and it is in this kind of
568 nigel 75 implementation where \eG can be useful.
569     .P
570     Note, however, that PCRE's interpretation of \eG, as the start of the current
571 nigel 63 match, is subtly different from Perl's, which defines it as the end of the
572     previous match. In Perl, these can be different when the previously matched
573     string was empty. Because PCRE does just one match at a time, it cannot
574     reproduce this behaviour.
575 nigel 75 .P
576     If all the alternatives of a pattern begin with \eG, the expression is anchored
577 nigel 63 to the starting match position, and the "anchored" flag is set in the compiled
578     regular expression.
579 nigel 75 .
580     .
582 nigel 63 .rs
583     .sp
584     Outside a character class, in the default matching mode, the circumflex
585 nigel 75 character is an assertion that is true only if the current matching point is
586     at the start of the subject string. If the \fIstartoffset\fP argument of
587     \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
588 nigel 63 option is unset. Inside a character class, circumflex has an entirely different
589 nigel 75 meaning
590     .\" HTML <a href="#characterclass">
591     .\" </a>
592     (see below).
593     .\"
594     .P
595 nigel 63 Circumflex need not be the first character of the pattern if a number of
596     alternatives are involved, but it should be the first thing in each alternative
597     in which it appears if the pattern is ever to match that branch. If all
598     possible alternatives start with a circumflex, that is, if the pattern is
599     constrained to match only at the start of the subject, it is said to be an
600     "anchored" pattern. (There are also other constructs that can cause a pattern
601     to be anchored.)
602 nigel 75 .P
603     A dollar character is an assertion that is true only if the current matching
604 nigel 63 point is at the end of the subject string, or immediately before a newline
605 nigel 91 at the end of the string (by default). Dollar need not be the last character of
606     the pattern if a number of alternatives are involved, but it should be the last
607     item in any branch in which it appears. Dollar has no special meaning in a
608     character class.
609 nigel 75 .P
610 nigel 63 The meaning of dollar can be changed so that it matches only at the very end of
611     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
612 nigel 75 does not affect the \eZ assertion.
613     .P
614 nigel 63 The meanings of the circumflex and dollar characters are changed if the
615 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
616     immediately after internal newlines as well as at the start of the subject
617     string. It does not match after a newline that ends the string. A dollar
618     matches before any newlines in the string, as well as at the very end, when
619     PCRE_MULTILINE is set. When newline is specified as the two-character
620     sequence CRLF, isolated CR and LF characters do not indicate newlines.
621 nigel 75 .P
622 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
623     \en represents a newline) in multiline mode, but not otherwise. Consequently,
624     patterns that are anchored in single line mode because all branches start with
625     ^ are not anchored in multiline mode, and a match for circumflex is possible
626     when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
627     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
628     .P
629 nigel 75 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
630 nigel 63 end of the subject in both modes, and if all branches of a pattern start with
631 nigel 91 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
632 nigel 75 .
633     .
635 nigel 63 .rs
636     .sp
637     Outside a character class, a dot in the pattern matches any one character in
638 nigel 91 the subject string except (by default) a character that signifies the end of a
639 nigel 93 line. In UTF-8 mode, the matched character may be more than one byte long.
640 nigel 91 .P
641 nigel 93 When a line ending is defined as a single character, dot never matches that
642     character; when the two-character sequence CRLF is used, dot does not match CR
643     if it is immediately followed by LF, but otherwise it matches all characters
644     (including isolated CRs and LFs). When any Unicode line endings are being
645     recognized, dot does not match CR or LF or any of the other line ending
646     characters.
647     .P
648 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
649 nigel 93 option is set, a dot matches any one character, without exception. If the
650     two-character sequence CRLF is present in the subject string, it takes two dots
651     to match it.
652 nigel 91 .P
653     The handling of dot is entirely independent of the handling of circumflex and
654     dollar, the only relationship being that they both involve newlines. Dot has no
655     special meaning in a character class.
656 nigel 75 .
657     .
659 nigel 63 .rs
660     .sp
661 nigel 75 Outside a character class, the escape sequence \eC matches any one byte, both
662 nigel 93 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
663     characters. The feature is provided in Perl in order to match individual bytes
664     in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
665     what remains in the string may be a malformed UTF-8 string. For this reason,
666     the \eC escape sequence is best avoided.
667 nigel 75 .P
668     PCRE does not allow \eC to appear in lookbehind assertions
669     .\" HTML <a href="#lookbehind">
670     .\" </a>
671     (described below),
672     .\"
673     because in UTF-8 mode this would make it impossible to calculate the length of
674     the lookbehind.
675     .
676     .
677     .\" HTML <a name="characterclass"></a>
679 nigel 63 .rs
680     .sp
681     An opening square bracket introduces a character class, terminated by a closing
682     square bracket. A closing square bracket on its own is not special. If a
683     closing square bracket is required as a member of the class, it should be the
684     first data character in the class (after an initial circumflex, if present) or
685     escaped with a backslash.
686 nigel 75 .P
687 nigel 63 A character class matches a single character in the subject. In UTF-8 mode, the
688     character may occupy more than one byte. A matched character must be in the set
689     of characters defined by the class, unless the first character in the class
690     definition is a circumflex, in which case the subject character must not be in
691     the set defined by the class. If a circumflex is actually required as a member
692     of the class, ensure it is not the first character, or escape it with a
693     backslash.
694 nigel 75 .P
695 nigel 63 For example, the character class [aeiou] matches any lower case vowel, while
696     [^aeiou] matches any character that is not a lower case vowel. Note that a
697 nigel 75 circumflex is just a convenient notation for specifying the characters that
698     are in the class by enumerating those that are not. A class that starts with a
699     circumflex is not an assertion: it still consumes a character from the subject
700     string, and therefore it fails if the current pointer is at the end of the
701     string.
702     .P
703 nigel 63 In UTF-8 mode, characters with values greater than 255 can be included in a
704 nigel 75 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
705     .P
706 nigel 63 When caseless matching is set, any letters in a class represent both their
707     upper case and lower case versions, so for example, a caseless [aeiou] matches
708     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
709 nigel 77 caseful version would. In UTF-8 mode, PCRE always understands the concept of
710     case for characters whose values are less than 128, so caseless matching is
711     always possible. For characters with higher values, the concept of case is
712     supported if PCRE is compiled with Unicode property support, but not otherwise.
713     If you want to use caseless matching for characters 128 and above, you must
714     ensure that PCRE is compiled with Unicode property support as well as with
715     UTF-8 support.
716 nigel 75 .P
717 nigel 93 Characters that might indicate line breaks are never treated in any special way
718     when matching character classes, whatever line-ending sequence is in use, and
719     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
720     such as [^a] always matches one of these characters.
721 nigel 75 .P
722 nigel 63 The minus (hyphen) character can be used to specify a range of characters in a
723     character class. For example, [d-m] matches any letter between d and m,
724     inclusive. If a minus character is required in a class, it must be escaped with
725     a backslash or appear in a position where it cannot be interpreted as
726     indicating a range, typically as the first or last character in the class.
727 nigel 75 .P
728 nigel 63 It is not possible to have the literal character "]" as the end character of a
729     range. A pattern such as [W-]46] is interpreted as a class of two characters
730     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
731     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
732 nigel 75 the end of range, so [W-\e]46] is interpreted as a class containing a range
733     followed by two other characters. The octal or hexadecimal representation of
734     "]" can also be used to end a range.
735     .P
736 nigel 63 Ranges operate in the collating sequence of character values. They can also be
737 nigel 75 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
738 nigel 63 mode, ranges can include characters whose values are greater than 255, for
739 nigel 75 example [\ex{100}-\ex{2ff}].
740     .P
741 nigel 63 If a range that includes letters is used when caseless matching is set, it
742     matches the letters in either case. For example, [W-c] is equivalent to
743 nigel 75 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
744 ph10 139 tables for a French locale are in use, [\exc8-\excb] matches accented E
745 nigel 75 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
746     characters with values greater than 128 only when it is compiled with Unicode
747     property support.
748     .P
749     The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
750     in a character class, and add the characters that they match to the class. For
751     example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
752 nigel 63 conveniently be used with the upper case character types to specify a more
753     restricted set of characters than the matching lower case type. For example,
754 nigel 75 the class [^\eW_] matches any letter or digit, but not underscore.
755     .P
756     The only metacharacters that are recognized in character classes are backslash,
757     hyphen (only where it can be interpreted as specifying a range), circumflex
758     (only at the start), opening square bracket (only when it can be interpreted as
759     introducing a POSIX class name - see the next section), and the terminating
760     closing square bracket. However, escaping other non-alphanumeric characters
761     does no harm.
762     .
763     .
765 nigel 63 .rs
766     .sp
767 nigel 75 Perl supports the POSIX notation for character classes. This uses names
768 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
769     this notation. For example,
770 nigel 75 .sp
771 nigel 63 [01[:alpha:]%]
772 nigel 75 .sp
773 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
774     are
775 nigel 75 .sp
776 nigel 63 alnum letters and digits
777     alpha letters
778     ascii character codes 0 - 127
779     blank space or tab only
780     cntrl control characters
781 nigel 75 digit decimal digits (same as \ed)
782 nigel 63 graph printing characters, excluding space
783     lower lower case letters
784     print printing characters, including space
785     punct printing characters, excluding letters and digits
786 nigel 75 space white space (not quite the same as \es)
787 nigel 63 upper upper case letters
788 nigel 75 word "word" characters (same as \ew)
789 nigel 63 xdigit hexadecimal digits
790 nigel 75 .sp
791 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
792     space (32). Notice that this list includes the VT character (code 11). This
793 nigel 75 makes "space" different to \es, which does not include VT (for Perl
794 nigel 63 compatibility).
795 nigel 75 .P
796 nigel 63 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
797     5.8. Another Perl extension is negation, which is indicated by a ^ character
798     after the colon. For example,
799 nigel 75 .sp
800 nigel 63 [12[:^digit:]]
801 nigel 75 .sp
802 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
803     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
804     supported, and an error is given if they are encountered.
805 nigel 75 .P
806     In UTF-8 mode, characters with values greater than 128 do not match any of
807 nigel 63 the POSIX character classes.
808 nigel 75 .
809     .
810     .SH "VERTICAL BAR"
811 nigel 63 .rs
812     .sp
813     Vertical bar characters are used to separate alternative patterns. For example,
814     the pattern
815 nigel 75 .sp
816 nigel 63 gilbert|sullivan
817 nigel 75 .sp
818 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
819 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
820     process tries each alternative in turn, from left to right, and the first one
821     that succeeds is used. If the alternatives are within a subpattern
822 nigel 75 .\" HTML <a href="#subpattern">
823     .\" </a>
824     (defined below),
825     .\"
826     "succeeds" means matching the rest of the main pattern as well as the
827     alternative in the subpattern.
828     .
829     .
831 nigel 63 .rs
832     .sp
833     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
834     PCRE_EXTENDED options can be changed from within the pattern by a sequence of
835     Perl option letters enclosed between "(?" and ")". The option letters are
836 nigel 75 .sp
837 nigel 63 i for PCRE_CASELESS
838     m for PCRE_MULTILINE
839     s for PCRE_DOTALL
840     x for PCRE_EXTENDED
841 nigel 75 .sp
842 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
843     unset these options by preceding the letter with a hyphen, and a combined
844     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
845     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
846     permitted. If a letter appears both before and after the hyphen, the option is
847     unset.
848 nigel 75 .P
849 nigel 63 When an option change occurs at top level (that is, not inside subpattern
850     parentheses), the change applies to the remainder of the pattern that follows.
851     If the change is placed right at the start of a pattern, PCRE extracts it into
852     the global options (and it will therefore show up in data extracted by the
853 nigel 75 \fBpcre_fullinfo()\fP function).
854     .P
855 nigel 93 An option change within a subpattern (see below for a description of
856     subpatterns) affects only that part of the current pattern that follows it, so
857 nigel 75 .sp
858 nigel 63 (a(?i)b)c
859 nigel 75 .sp
860 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
861     By this means, options can be made to have different settings in different
862     parts of the pattern. Any changes made in one alternative do carry on
863     into subsequent branches within the same subpattern. For example,
864 nigel 75 .sp
865 nigel 63 (a(?i)b|c)
866 nigel 75 .sp
867 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
868     branch is abandoned before the option setting. This is because the effects of
869     option settings happen at compile time. There would be some very weird
870     behaviour otherwise.
871 nigel 75 .P
872 nigel 91 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
873     changed in the same way as the Perl-compatible options by using the characters
874     J, U and X respectively.
875 nigel 75 .
876     .
877     .\" HTML <a name="subpattern"></a>
878 nigel 63 .SH SUBPATTERNS
879     .rs
880     .sp
881     Subpatterns are delimited by parentheses (round brackets), which can be nested.
882 nigel 75 Turning part of a pattern into a subpattern does two things:
883     .sp
884 nigel 63 1. It localizes a set of alternatives. For example, the pattern
885 nigel 75 .sp
886 nigel 63 cat(aract|erpillar|)
887 nigel 75 .sp
888 nigel 63 matches one of the words "cat", "cataract", or "caterpillar". Without the
889 nigel 93 parentheses, it would match "cataract", "erpillar" or an empty string.
890 nigel 75 .sp
891     2. It sets up the subpattern as a capturing subpattern. This means that, when
892     the whole pattern matches, that portion of the subject string that matched the
893     subpattern is passed back to the caller via the \fIovector\fP argument of
894     \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
895     from 1) to obtain numbers for the capturing subpatterns.
896     .P
897 nigel 63 For example, if the string "the red king" is matched against the pattern
898 nigel 75 .sp
899 nigel 63 the ((red|white) (king|queen))
900 nigel 75 .sp
901 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
902     2, and 3, respectively.
903 nigel 75 .P
904 nigel 63 The fact that plain parentheses fulfil two functions is not always helpful.
905     There are often times when a grouping subpattern is required without a
906     capturing requirement. If an opening parenthesis is followed by a question mark
907     and a colon, the subpattern does not do any capturing, and is not counted when
908     computing the number of any subsequent capturing subpatterns. For example, if
909     the string "the white queen" is matched against the pattern
910 nigel 75 .sp
911 nigel 63 the ((?:red|white) (king|queen))
912 nigel 75 .sp
913 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
914 nigel 93 2. The maximum number of capturing subpatterns is 65535.
915 nigel 75 .P
916 nigel 63 As a convenient shorthand, if any option settings are required at the start of
917     a non-capturing subpattern, the option letters may appear between the "?" and
918     the ":". Thus the two patterns
919 nigel 75 .sp
920 nigel 63 (?i:saturday|sunday)
921     (?:(?i)saturday|sunday)
922 nigel 75 .sp
923 nigel 63 match exactly the same set of strings. Because alternative branches are tried
924     from left to right, and options are not reset until the end of the subpattern
925     is reached, an option setting in one branch does affect subsequent branches, so
926     the above patterns match "SUNDAY" as well as "Saturday".
927 nigel 75 .
928     .
930 nigel 63 .rs
931     .sp
932     Identifying capturing parentheses by number is simple, but it can be very hard
933     to keep track of the numbers in complicated regular expressions. Furthermore,
934 nigel 75 if an expression is modified, the numbers may change. To help with this
935 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
936     added to Perl until release 5.10. Python had the feature earlier, and PCRE
937     introduced it at release 4.0, using the Python syntax. PCRE now supports both
938     the Perl and the Python syntax.
939     .P
940     In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
941     (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
942 nigel 91 parentheses from other parts of the pattern, such as
943     .\" HTML <a href="#backreferences">
944     .\" </a>
945     backreferences,
946     .\"
947     .\" HTML <a href="#recursion">
948     .\" </a>
949     recursion,
950     .\"
951     and
952     .\" HTML <a href="#conditions">
953     .\" </a>
954     conditions,
955     .\"
956     can be made by name as well as by number.
957 nigel 75 .P
958 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
959 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
960     if the names were not present. The PCRE API provides function calls for
961     extracting the name-to-number translation table from a compiled pattern. There
962     is also a convenience function for extracting a captured substring by name.
963 nigel 91 .P
964     By default, a name must be unique within a pattern, but it is possible to relax
965     this constraint by setting the PCRE_DUPNAMES option at compile time. This can
966     be useful for patterns where only one instance of the named parentheses can
967     match. Suppose you want to match the name of a weekday, either as a 3-letter
968     abbreviation or as the full name, and in both cases you want to extract the
969     abbreviation. This pattern (ignoring the line breaks) does the job:
970     .sp
971 nigel 93 (?<DN>Mon|Fri|Sun)(?:day)?|
972     (?<DN>Tue)(?:sday)?|
973     (?<DN>Wed)(?:nesday)?|
974     (?<DN>Thu)(?:rsday)?|
975     (?<DN>Sat)(?:urday)?
976 nigel 91 .sp
977     There are five capturing substrings, but only one is ever set after a match.
978     The convenience function for extracting the data by name returns the substring
979 nigel 93 for the first (and in this example, the only) subpattern of that name that
980 nigel 91 matched. This saves searching to find which numbered subpattern it was. If you
981     make a reference to a non-unique named subpattern from elsewhere in the
982     pattern, the one that corresponds to the lowest number is used. For further
983     details of the interfaces for handling named subpatterns, see the
984 nigel 63 .\" HREF
985 nigel 75 \fBpcreapi\fP
986 nigel 63 .\"
987     documentation.
988 nigel 75 .
989     .
990 nigel 63 .SH REPETITION
991     .rs
992     .sp
993     Repetition is specified by quantifiers, which can follow any of the following
994     items:
995 nigel 75 .sp
996 nigel 63 a literal data character
997 nigel 93 the dot metacharacter
998 nigel 75 the \eC escape sequence
999     the \eX escape sequence (in UTF-8 mode with Unicode properties)
1000 nigel 93 the \eR escape sequence
1001 nigel 75 an escape such as \ed that matches a single character
1002 nigel 63 a character class
1003     a back reference (see next section)
1004     a parenthesized subpattern (unless it is an assertion)
1005 nigel 75 .sp
1006 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1007     permitted matches, by giving the two numbers in curly brackets (braces),
1008     separated by a comma. The numbers must be less than 65536, and the first must
1009     be less than or equal to the second. For example:
1010 nigel 75 .sp
1011 nigel 63 z{2,4}
1012 nigel 75 .sp
1013 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1014     character. If the second number is omitted, but the comma is present, there is
1015     no upper limit; if the second number and the comma are both omitted, the
1016     quantifier specifies an exact number of required matches. Thus
1017 nigel 75 .sp
1018 nigel 63 [aeiou]{3,}
1019 nigel 75 .sp
1020 nigel 63 matches at least 3 successive vowels, but may match many more, while
1021 nigel 75 .sp
1022     \ed{8}
1023     .sp
1024 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1025     where a quantifier is not allowed, or one that does not match the syntax of a
1026     quantifier, is taken as a literal character. For example, {,6} is not a
1027     quantifier, but a literal string of four characters.
1028 nigel 75 .P
1029 nigel 63 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1030 nigel 75 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1031     which is represented by a two-byte sequence. Similarly, when Unicode property
1032     support is available, \eX{3} matches three Unicode extended sequences, each of
1033     which may be several bytes long (and they may be of different lengths).
1034     .P
1035 nigel 63 The quantifier {0} is permitted, causing the expression to behave as if the
1036     previous item and the quantifier were not present.
1037 nigel 75 .P
1038 nigel 93 For convenience, the three most common quantifiers have single-character
1039     abbreviations:
1040 nigel 75 .sp
1041 nigel 63 * is equivalent to {0,}
1042     + is equivalent to {1,}
1043     ? is equivalent to {0,1}
1044 nigel 75 .sp
1045 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1046     match no characters with a quantifier that has no upper limit, for example:
1047 nigel 75 .sp
1048 nigel 63 (a?)*
1049 nigel 75 .sp
1050 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1051     such patterns. However, because there are cases where this can be useful, such
1052     patterns are now accepted, but if any repetition of the subpattern does in fact
1053     match no characters, the loop is forcibly broken.
1054 nigel 75 .P
1055 nigel 63 By default, the quantifiers are "greedy", that is, they match as much as
1056     possible (up to the maximum number of permitted times), without causing the
1057     rest of the pattern to fail. The classic example of where this gives problems
1058 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1059     and within the comment, individual * and / characters may appear. An attempt to
1060     match C comments by applying the pattern
1061     .sp
1062     /\e*.*\e*/
1063     .sp
1064 nigel 63 to the string
1065 nigel 75 .sp
1066     /* first comment */ not comment /* second comment */
1067     .sp
1068 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1069     item.
1070 nigel 75 .P
1071 nigel 63 However, if a quantifier is followed by a question mark, it ceases to be
1072     greedy, and instead matches the minimum number of times possible, so the
1073     pattern
1074 nigel 75 .sp
1075     /\e*.*?\e*/
1076     .sp
1077 nigel 63 does the right thing with the C comments. The meaning of the various
1078     quantifiers is not otherwise changed, just the preferred number of matches.
1079     Do not confuse this use of question mark with its use as a quantifier in its
1080     own right. Because it has two uses, it can sometimes appear doubled, as in
1081 nigel 75 .sp
1082     \ed??\ed
1083     .sp
1084 nigel 63 which matches one digit by preference, but can match two if that is the only
1085     way the rest of the pattern matches.
1086 nigel 75 .P
1087 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1088 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1089     greedy by following them with a question mark. In other words, it inverts the
1090     default behaviour.
1091 nigel 75 .P
1092 nigel 63 When a parenthesized subpattern is quantified with a minimum repeat count that
1093 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1094 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1095 nigel 75 .P
1096 nigel 63 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1097 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1098 nigel 63 implicitly anchored, because whatever follows will be tried against every
1099     character position in the subject string, so there is no point in retrying the
1100     overall match at any position after the first. PCRE normally treats such a
1101 nigel 75 pattern as though it were preceded by \eA.
1102     .P
1103 nigel 63 In cases where it is known that the subject string contains no newlines, it is
1104     worth setting PCRE_DOTALL in order to obtain this optimization, or
1105     alternatively using ^ to indicate anchoring explicitly.
1106 nigel 75 .P
1107 nigel 63 However, there is one situation where the optimization cannot be used. When .*
1108     is inside capturing parentheses that are the subject of a backreference
1109 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1110     succeeds. Consider, for example:
1111 nigel 75 .sp
1112     (.*)abc\e1
1113     .sp
1114 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1115     this reason, such a pattern is not implicitly anchored.
1116 nigel 75 .P
1117 nigel 63 When a capturing subpattern is repeated, the value captured is the substring
1118     that matched the final iteration. For example, after
1119 nigel 75 .sp
1120     (tweedle[dume]{3}\es*)+
1121     .sp
1122 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1123     "tweedledee". However, if there are nested capturing subpatterns, the
1124     corresponding captured values may have been set in previous iterations. For
1125     example, after
1126 nigel 75 .sp
1127 nigel 63 /(a|(b))+/
1128 nigel 75 .sp
1129 nigel 63 matches "aba" the value of the second captured substring is "b".
1130 nigel 75 .
1131     .
1132     .\" HTML <a name="atomicgroup"></a>
1134 nigel 63 .rs
1135     .sp
1136 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1137     repetition, failure of what follows normally causes the repeated item to be
1138     re-evaluated to see if a different number of repeats allows the rest of the
1139     pattern to match. Sometimes it is useful to prevent this, either to change the
1140     nature of the match, or to cause it fail earlier than it otherwise might, when
1141     the author of the pattern knows there is no point in carrying on.
1142 nigel 75 .P
1143     Consider, for example, the pattern \ed+foo when applied to the subject line
1144     .sp
1145 nigel 63 123456bar
1146 nigel 75 .sp
1147 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1148 nigel 75 action of the matcher is to try again with only 5 digits matching the \ed+
1149 nigel 63 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1150     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1151     that once a subpattern has matched, it is not to be re-evaluated in this way.
1152 nigel 75 .P
1153 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1154 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1155     special parenthesis, starting with (?> as in this example:
1156 nigel 75 .sp
1157     (?>\ed+)foo
1158     .sp
1159 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1160     it has matched, and a failure further into the pattern is prevented from
1161     backtracking into it. Backtracking past it to previous items, however, works as
1162     normal.
1163 nigel 75 .P
1164 nigel 63 An alternative description is that a subpattern of this type matches the string
1165     of characters that an identical standalone pattern would match, if anchored at
1166     the current point in the subject string.
1167 nigel 75 .P
1168 nigel 63 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1169     the above example can be thought of as a maximizing repeat that must swallow
1170 nigel 75 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1171 nigel 63 number of digits they match in order to make the rest of the pattern match,
1172 nigel 75 (?>\ed+) can only match an entire sequence of digits.
1173     .P
1174 nigel 63 Atomic groups in general can of course contain arbitrarily complicated
1175     subpatterns, and can be nested. However, when the subpattern for an atomic
1176     group is just a single repeated item, as in the example above, a simpler
1177     notation, called a "possessive quantifier" can be used. This consists of an
1178     additional + character following a quantifier. Using this notation, the
1179     previous example can be rewritten as
1180 nigel 75 .sp
1181     \ed++foo
1182     .sp
1183 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1184     option is ignored. They are a convenient notation for the simpler forms of
1185 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1186     quantifier and the equivalent atomic group, though there may be a performance
1187     difference; possessive quantifiers should be slightly faster.
1188 nigel 75 .P
1189 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1190     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1191     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1192     package, and PCRE copied it from there. It ultimately found its way into Perl
1193     at release 5.10.
1194 nigel 75 .P
1195 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1196     pattern constructs. For example, the sequence A+B is treated as A++B because
1197     there is no point in backtracking into a sequence of A's when B must follow.
1198     .P
1199 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1200     be repeated an unlimited number of times, the use of an atomic group is the
1201     only way to avoid some failing matches taking a very long time indeed. The
1202     pattern
1203 nigel 75 .sp
1204     (\eD+|<\ed+>)*[!?]
1205     .sp
1206 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1207     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1208     quickly. However, if it is applied to
1209 nigel 75 .sp
1210 nigel 63 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1211 nigel 75 .sp
1212 nigel 63 it takes a long time before reporting failure. This is because the string can
1213 nigel 75 be divided between the internal \eD+ repeat and the external * repeat in a
1214     large number of ways, and all have to be tried. (The example uses [!?] rather
1215     than a single character at the end, because both PCRE and Perl have an
1216     optimization that allows for fast failure when a single character is used. They
1217     remember the last single character that is required for a match, and fail early
1218     if it is not present in the string.) If the pattern is changed so that it uses
1219     an atomic group, like this:
1220     .sp
1221     ((?>\eD+)|<\ed+>)*[!?]
1222     .sp
1223 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1224 nigel 75 .
1225     .
1226     .\" HTML <a name="backreferences"></a>
1228 nigel 63 .rs
1229     .sp
1230     Outside a character class, a backslash followed by a digit greater than 0 (and
1231     possibly further digits) is a back reference to a capturing subpattern earlier
1232     (that is, to its left) in the pattern, provided there have been that many
1233     previous capturing left parentheses.
1234 nigel 75 .P
1235 nigel 63 However, if the decimal number following the backslash is less than 10, it is
1236     always taken as a back reference, and causes an error only if there are not
1237     that many capturing left parentheses in the entire pattern. In other words, the
1238     parentheses that are referenced need not be to the left of the reference for
1239 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1240     when a repetition is involved and the subpattern to the right has participated
1241     in an earlier iteration.
1242     .P
1243 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1244     whose number is 10 or more using this syntax because a sequence such as \e50 is
1245     interpreted as a character defined in octal. See the subsection entitled
1246 nigel 91 "Non-printing characters"
1247 nigel 75 .\" HTML <a href="#digitsafterbackslash">
1248     .\" </a>
1249     above
1250     .\"
1251 nigel 93 for further details of the handling of digits following a backslash. There is
1252     no such problem when named parentheses are used. A back reference to any
1253     subpattern is possible using named parentheses (see below).
1254 nigel 75 .P
1255 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1256     backslash is to use the \eg escape sequence, which is a feature introduced in
1257     Perl 5.10. This escape must be followed by a positive or a negative number,
1258     optionally enclosed in braces. These examples are all identical:
1259     .sp
1260     (ring), \e1
1261     (ring), \eg1
1262     (ring), \eg{1}
1263     .sp
1264     A positive number specifies an absolute reference without the ambiguity that is
1265     present in the older syntax. It is also useful when literal digits follow the
1266     reference. A negative number is a relative reference. Consider this example:
1267     .sp
1268     (abc(def)ghi)\eg{-1}
1269     .sp
1270     The sequence \eg{-1} is a reference to the most recently started capturing
1271     subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1272     would be equivalent to \e1. The use of relative references can be helpful in
1273     long patterns, and also in patterns that are created by joining together
1274     fragments that contain references within themselves.
1275     .P
1276 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1277     the current subject string, rather than anything matching the subpattern
1278     itself (see
1279     .\" HTML <a href="#subpatternsassubroutines">
1280     .\" </a>
1281     "Subpatterns as subroutines"
1282     .\"
1283     below for a way of doing that). So the pattern
1284 nigel 75 .sp
1285     (sens|respons)e and \e1ibility
1286     .sp
1287 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1288     "sense and responsibility". If caseful matching is in force at the time of the
1289     back reference, the case of letters is relevant. For example,
1290 nigel 75 .sp
1291     ((?i)rah)\es+\e1
1292     .sp
1293 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1294     capturing subpattern is matched caselessly.
1295 nigel 75 .P
1296 nigel 93 Back references to named subpatterns use the Perl syntax \ek<name> or \ek'name'
1297     or the Python syntax (?P=name). We could rewrite the above example in either of
1298     the following ways:
1299 nigel 75 .sp
1300 nigel 93 (?<p1>(?i)rah)\es+\ek<p1>
1301 nigel 91 (?P<p1>(?i)rah)\es+(?P=p1)
1302 nigel 75 .sp
1303 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1304     after the reference.
1305     .P
1306 nigel 63 There may be more than one back reference to the same subpattern. If a
1307     subpattern has not actually been used in a particular match, any back
1308     references to it always fail. For example, the pattern
1309 nigel 75 .sp
1310     (a|(bc))\e2
1311     .sp
1312 nigel 63 always fails if it starts to match "a" rather than "bc". Because there may be
1313     many capturing parentheses in a pattern, all digits following the backslash are
1314     taken as part of a potential back reference number. If the pattern continues
1315     with a digit character, some delimiter must be used to terminate the back
1316     reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1317 nigel 75 Otherwise an empty comment (see
1318     .\" HTML <a href="#comments">
1319     .\" </a>
1320     "Comments"
1321     .\"
1322     below) can be used.
1323     .P
1324 nigel 63 A back reference that occurs inside the parentheses to which it refers fails
1325 nigel 75 when the subpattern is first used, so, for example, (a\e1) never matches.
1326 nigel 63 However, such references can be useful inside repeated subpatterns. For
1327     example, the pattern
1328 nigel 75 .sp
1329     (a|b\e1)+
1330     .sp
1331 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1332     the subpattern, the back reference matches the character string corresponding
1333     to the previous iteration. In order for this to work, the pattern must be such
1334     that the first iteration does not need to match the back reference. This can be
1335     done using alternation, as in the example above, or by a quantifier with a
1336     minimum of zero.
1337 nigel 75 .
1338     .
1339     .\" HTML <a name="bigassertions"></a>
1340 nigel 63 .SH ASSERTIONS
1341     .rs
1342     .sp
1343     An assertion is a test on the characters following or preceding the current
1344     matching point that does not actually consume any characters. The simple
1345 nigel 75 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1346     .\" HTML <a href="#smallassertions">
1347     .\" </a>
1348     above.
1349     .\"
1350     .P
1351 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1352     those that look ahead of the current position in the subject string, and those
1353 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1354     except that it does not cause the current matching position to be changed.
1355     .P
1356     Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1357     because it makes no sense to assert the same thing several times. If any kind
1358     of assertion contains capturing subpatterns within it, these are counted for
1359     the purposes of numbering the capturing subpatterns in the whole pattern.
1360     However, substring capturing is carried out only for positive assertions,
1361     because it does not make sense for negative assertions.
1362     .
1363     .
1364     .SS "Lookahead assertions"
1365     .rs
1366     .sp
1367 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1368     negative assertions. For example,
1369 nigel 75 .sp
1370     \ew+(?=;)
1371     .sp
1372 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1373     the match, and
1374 nigel 75 .sp
1375 nigel 63 foo(?!bar)
1376 nigel 75 .sp
1377 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1378     apparently similar pattern
1379 nigel 75 .sp
1380 nigel 63 (?!foo)bar
1381 nigel 75 .sp
1382 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1383     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1384     (?!foo) is always true when the next three characters are "bar". A
1385 nigel 75 lookbehind assertion is needed to achieve the other effect.
1386     .P
1387 nigel 63 If you want to force a matching failure at some point in a pattern, the most
1388     convenient way to do it is with (?!) because an empty string always matches, so
1389     an assertion that requires there not to be an empty string must always fail.
1390 nigel 75 .
1391     .
1392     .\" HTML <a name="lookbehind"></a>
1393     .SS "Lookbehind assertions"
1394     .rs
1395     .sp
1396 nigel 63 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1397     negative assertions. For example,
1398 nigel 75 .sp
1399 nigel 63 (?<!foo)bar
1400 nigel 75 .sp
1401 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1402     a lookbehind assertion are restricted such that all the strings it matches must
1403 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1404     do not all have to have the same fixed length. Thus
1405 nigel 75 .sp
1406 nigel 63 (?<=bullock|donkey)
1407 nigel 75 .sp
1408 nigel 63 is permitted, but
1409 nigel 75 .sp
1410 nigel 63 (?<!dogs?|cats?)
1411 nigel 75 .sp
1412 nigel 63 causes an error at compile time. Branches that match different length strings
1413     are permitted only at the top level of a lookbehind assertion. This is an
1414     extension compared with Perl (at least for 5.8), which requires all branches to
1415     match the same length of string. An assertion such as
1416 nigel 75 .sp
1417 nigel 63 (?<=ab(c|de))
1418 nigel 75 .sp
1419 nigel 63 is not permitted, because its single top-level branch can match two different
1420     lengths, but it is acceptable if rewritten to use two top-level branches:
1421 nigel 75 .sp
1422 nigel 63 (?<=abc|abde)
1423 nigel 75 .sp
1424 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
1425 nigel 93 temporarily move the current position back by the fixed length and then try to
1426 nigel 63 match. If there are insufficient characters before the current position, the
1427 nigel 93 assertion fails.
1428 nigel 75 .P
1429     PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1430 nigel 63 to appear in lookbehind assertions, because it makes it impossible to calculate
1431 nigel 93 the length of the lookbehind. The \eX and \eR escapes, which can match
1432     different numbers of bytes, are also not permitted.
1433 nigel 75 .P
1434 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1435     specify efficient matching at the end of the subject string. Consider a simple
1436     pattern such as
1437 nigel 75 .sp
1438 nigel 63 abcd$
1439 nigel 75 .sp
1440 nigel 63 when applied to a long string that does not match. Because matching proceeds
1441     from left to right, PCRE will look for each "a" in the subject and then see if
1442     what follows matches the rest of the pattern. If the pattern is specified as
1443 nigel 75 .sp
1444 nigel 63 ^.*abcd$
1445 nigel 75 .sp
1446 nigel 63 the initial .* matches the entire string at first, but when this fails (because
1447     there is no following "a"), it backtracks to match all but the last character,
1448     then all but the last two characters, and so on. Once again the search for "a"
1449     covers the entire string, from right to left, so we are no better off. However,
1450     if the pattern is written as
1451 nigel 75 .sp
1452 nigel 63 ^.*+(?<=abcd)
1453 nigel 75 .sp
1454 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
1455 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
1456     characters. If it fails, the match fails immediately. For long strings, this
1457     approach makes a significant difference to the processing time.
1458 nigel 75 .
1459     .
1460     .SS "Using multiple assertions"
1461     .rs
1462     .sp
1463 nigel 63 Several assertions (of any sort) may occur in succession. For example,
1464 nigel 75 .sp
1465     (?<=\ed{3})(?<!999)foo
1466     .sp
1467 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
1468     the assertions is applied independently at the same point in the subject
1469     string. First there is a check that the previous three characters are all
1470     digits, and then there is a check that the same three characters are not "999".
1471 nigel 75 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1472 nigel 63 of which are digits and the last three of which are not "999". For example, it
1473     doesn't match "123abcfoo". A pattern to do that is
1474 nigel 75 .sp
1475     (?<=\ed{3}...)(?<!999)foo
1476     .sp
1477 nigel 63 This time the first assertion looks at the preceding six characters, checking
1478     that the first three are digits, and then the second assertion checks that the
1479     preceding three characters are not "999".
1480 nigel 75 .P
1481 nigel 63 Assertions can be nested in any combination. For example,
1482 nigel 75 .sp
1483 nigel 63 (?<=(?<!foo)bar)baz
1484 nigel 75 .sp
1485 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1486     preceded by "foo", while
1487 nigel 75 .sp
1488     (?<=\ed{3}(?!999)...)foo
1489     .sp
1490     is another pattern that matches "foo" preceded by three digits and any three
1491 nigel 63 characters that are not "999".
1492 nigel 75 .
1493     .
1494 nigel 91 .\" HTML <a name="conditions"></a>
1496 nigel 63 .rs
1497     .sp
1498     It is possible to cause the matching process to obey a subpattern
1499     conditionally or to choose between two alternative subpatterns, depending on
1500     the result of an assertion, or whether a previous capturing subpattern matched
1501     or not. The two possible forms of conditional subpattern are
1502 nigel 75 .sp
1503 nigel 63 (?(condition)yes-pattern)
1504     (?(condition)yes-pattern|no-pattern)
1505 nigel 75 .sp
1506 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
1507     no-pattern (if present) is used. If there are more than two alternatives in the
1508     subpattern, a compile-time error occurs.
1509 nigel 75 .P
1510 nigel 93 There are four kinds of condition: references to subpatterns, references to
1511     recursion, a pseudo-condition called DEFINE, and assertions.
1512     .
1513     .SS "Checking for a used subpattern by number"
1514     .rs
1515     .sp
1516     If the text between the parentheses consists of a sequence of digits, the
1517     condition is true if the capturing subpattern of that number has previously
1518     matched.
1519 nigel 91 .P
1520     Consider the following pattern, which contains non-significant white space to
1521     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1522     three parts for ease of discussion:
1523 nigel 75 .sp
1524     ( \e( )? [^()]+ (?(1) \e) )
1525     .sp
1526 nigel 63 The first part matches an optional opening parenthesis, and if that
1527     character is present, sets it as the first captured substring. The second part
1528     matches one or more characters that are not parentheses. The third part is a
1529     conditional subpattern that tests whether the first set of parentheses matched
1530     or not. If they did, that is, if subject started with an opening parenthesis,
1531     the condition is true, and so the yes-pattern is executed and a closing
1532     parenthesis is required. Otherwise, since no-pattern is not present, the
1533     subpattern matches nothing. In other words, this pattern matches a sequence of
1534 nigel 93 non-parentheses, optionally enclosed in parentheses.
1535     .
1536     .SS "Checking for a used subpattern by name"
1537     .rs
1538 nigel 91 .sp
1539 nigel 93 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1540     subpattern by name. For compatibility with earlier versions of PCRE, which had
1541     this facility before Perl, the syntax (?(name)...) is also recognized. However,
1542     there is a possible ambiguity with this syntax, because subpattern names may
1543     consist entirely of digits. PCRE looks first for a named subpattern; if it
1544     cannot find one and the name consists entirely of digits, PCRE looks for a
1545     subpattern of that number, which must be greater than zero. Using subpattern
1546     names that consist entirely of digits is not recommended.
1547     .P
1548     Rewriting the above example to use a named subpattern gives this:
1549 nigel 91 .sp
1550 nigel 93 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1551     .sp
1552     .
1553     .SS "Checking for pattern recursion"
1554     .rs
1555     .sp
1556 nigel 91 If the condition is the string (R), and there is no subpattern with the name R,
1557 nigel 93 the condition is true if a recursive call to the whole pattern or any
1558     subpattern has been made. If digits or a name preceded by ampersand follow the
1559     letter R, for example:
1560     .sp
1561     (?(R3)...) or (?(R&name)...)
1562     .sp
1563     the condition is true if the most recent recursion is into the subpattern whose
1564     number or name is given. This condition does not check the entire recursion
1565     stack.
1566 nigel 75 .P
1567 nigel 93 At "top level", all these recursion test conditions are false. Recursive
1568     patterns are described below.
1569     .
1570     .SS "Defining subpatterns for use by reference only"
1571     .rs
1572     .sp
1573     If the condition is the string (DEFINE), and there is no subpattern with the
1574     name DEFINE, the condition is always false. In this case, there may be only one
1575     alternative in the subpattern. It is always skipped if control reaches this
1576     point in the pattern; the idea of DEFINE is that it can be used to define
1577     "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1578     is described below.) For example, a pattern to match an IPv4 address could be
1579     written like this (ignore whitespace and line breaks):
1580     .sp
1581     (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1582     \eb (?&byte) (\e.(?&byte)){3} \eb
1583     .sp
1584     The first part of the pattern is a DEFINE group inside which a another group
1585     named "byte" is defined. This matches an individual component of an IPv4
1586     address (a number less than 256). When matching takes place, this part of the
1587     pattern is skipped because DEFINE acts like a false condition.
1588     .P
1589     The rest of the pattern uses references to the named group to match the four
1590     dot-separated components of an IPv4 address, insisting on a word boundary at
1591     each end.
1592     .
1593     .SS "Assertion conditions"
1594     .rs
1595     .sp
1596     If the condition is not in any of the above formats, it must be an assertion.
1597 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
1598     this pattern, again containing non-significant white space, and with the two
1599     alternatives on the second line:
1600 nigel 75 .sp
1601 nigel 63 (?(?=[^a-z]*[a-z])
1602 nigel 75 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1603     .sp
1604 nigel 63 The condition is a positive lookahead assertion that matches an optional
1605     sequence of non-letters followed by a letter. In other words, it tests for the
1606     presence of at least one letter in the subject. If a letter is found, the
1607     subject is matched against the first alternative; otherwise it is matched
1608     against the second. This pattern matches strings in one of the two forms
1609     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1610 nigel 75 .
1611     .
1612     .\" HTML <a name="comments"></a>
1613 nigel 63 .SH COMMENTS
1614     .rs
1615     .sp
1616 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
1617 nigel 63 closing parenthesis. Nested parentheses are not permitted. The characters
1618     that make up a comment play no part in the pattern matching at all.
1619 nigel 75 .P
1620 nigel 63 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1621 nigel 91 character class introduces a comment that continues to immediately after the
1622     next newline in the pattern.
1623 nigel 75 .
1624     .
1625 nigel 91 .\" HTML <a name="recursion"></a>
1626 nigel 75 .SH "RECURSIVE PATTERNS"
1627 nigel 63 .rs
1628     .sp
1629     Consider the problem of matching a string in parentheses, allowing for
1630     unlimited nested parentheses. Without the use of recursion, the best that can
1631     be done is to use a pattern that matches up to some fixed depth of nesting. It
1632 nigel 93 is not possible to handle an arbitrary nesting depth.
1633     .P
1634     For some time, Perl has provided a facility that allows regular expressions to
1635     recurse (amongst other things). It does this by interpolating Perl code in the
1636     expression at run time, and the code can refer to the expression itself. A Perl
1637     pattern using code interpolation to solve the parentheses problem can be
1638     created like this:
1639 nigel 75 .sp
1640     $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1641     .sp
1642 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1643 nigel 93 recursively to the pattern in which it appears.
1644 nigel 75 .P
1645 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1646     supports special syntax for recursion of the entire pattern, and also for
1647     individual subpattern recursion. After its introduction in PCRE and Python,
1648     this kind of recursion was introduced into Perl at release 5.10.
1649 nigel 75 .P
1650 nigel 93 A special item that consists of (? followed by a number greater than zero and a
1651     closing parenthesis is a recursive call of the subpattern of the given number,
1652     provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1653     call, which is described in the next section.) The special item (?R) or (?0) is
1654     a recursive call of the entire regular expression.
1655 nigel 87 .P
1656 nigel 93 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1657     treated as an atomic group. That is, once it has matched some of the subject
1658     string, it is never re-entered, even if it contains untried alternatives and
1659     there is a subsequent matching failure.
1660     .P
1661 nigel 87 This PCRE pattern solves the nested parentheses problem (assume the
1662     PCRE_EXTENDED option is set so that white space is ignored):
1663 nigel 75 .sp
1664     \e( ( (?>[^()]+) | (?R) )* \e)
1665     .sp
1666 nigel 63 First it matches an opening parenthesis. Then it matches any number of
1667     substrings which can either be a sequence of non-parentheses, or a recursive
1668 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
1669 nigel 63 Finally there is a closing parenthesis.
1670 nigel 75 .P
1671 nigel 63 If this were part of a larger pattern, you would not want to recurse the entire
1672     pattern, so instead you could use this:
1673 nigel 75 .sp
1674     ( \e( ( (?>[^()]+) | (?1) )* \e) )
1675     .sp
1676 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
1677     them instead of the whole pattern. In a larger pattern, keeping track of
1678     parenthesis numbers can be tricky. It may be more convenient to use named
1679 nigel 93 parentheses instead. The Perl syntax for this is (?&name); PCRE's earlier
1680     syntax (?P>name) is also supported. We could rewrite the above example as
1681     follows:
1682 nigel 75 .sp
1683 nigel 93 (?<pn> \e( ( (?>[^()]+) | (?&pn) )* \e) )
1684 nigel 75 .sp
1685 nigel 93 If there is more than one subpattern with the same name, the earliest one is
1686     used. This particular example pattern contains nested unlimited repeats, and so
1687     the use of atomic grouping for matching strings of non-parentheses is important
1688 nigel 63 when applying the pattern to strings that do not match. For example, when this
1689     pattern is applied to
1690 nigel 75 .sp
1691 nigel 63 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1692 nigel 75 .sp
1693 nigel 63 it yields "no match" quickly. However, if atomic grouping is not used,
1694     the match runs for a very long time indeed because there are so many different
1695     ways the + and * repeats can carve up the subject, and all have to be tested
1696     before failure can be reported.
1697 nigel 75 .P
1698 nigel 63 At the end of a match, the values set for any capturing subpatterns are those
1699     from the outermost level of the recursion at which the subpattern value is set.
1700     If you want to obtain intermediate values, a callout function can be used (see
1701 nigel 93 below and the
1702 nigel 63 .\" HREF
1703 nigel 75 \fBpcrecallout\fP
1704 nigel 63 .\"
1705     documentation). If the pattern above is matched against
1706 nigel 75 .sp
1707 nigel 63 (ab(cd)ef)
1708 nigel 75 .sp
1709 nigel 63 the value for the capturing parentheses is "ef", which is the last value taken
1710     on at the top level. If additional parentheses are added, giving
1711 nigel 75 .sp
1712     \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1713 nigel 63 ^ ^
1714     ^ ^
1715 nigel 75 .sp
1716 nigel 63 the string they capture is "ab(cd)ef", the contents of the top level
1717     parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1718     has to obtain extra memory to store data during a recursion, which it does by
1719 nigel 75 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1720 nigel 63 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1721 nigel 75 .P
1722 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1723     Consider this pattern, which matches text in angle brackets, allowing for
1724     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1725     recursing), whereas any characters are permitted at the outer level.
1726 nigel 75 .sp
1727     < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1728     .sp
1729 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
1730     different alternatives for the recursive and non-recursive cases. The (?R) item
1731     is the actual recursive call.
1732 nigel 75 .
1733     .
1734 nigel 63 .\" HTML <a name="subpatternsassubroutines"></a>
1736 nigel 63 .rs
1737     .sp
1738     If the syntax for a recursive subpattern reference (either by number or by
1739     name) is used outside the parentheses to which it refers, it operates like a
1740 nigel 93 subroutine in a programming language. The "called" subpattern may be defined
1741     before or after the reference. An earlier example pointed out that the pattern
1742 nigel 75 .sp
1743     (sens|respons)e and \e1ibility
1744     .sp
1745 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1746     "sense and responsibility". If instead the pattern
1747 nigel 75 .sp
1748 nigel 63 (sens|respons)e and (?1)ibility
1749 nigel 75 .sp
1750 nigel 63 is used, it does match "sense and responsibility" as well as the other two
1751 nigel 93 strings. Another example is given in the discussion of DEFINE above.
1752 nigel 87 .P
1753     Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1754     group. That is, once it has matched some of the subject string, it is never
1755     re-entered, even if it contains untried alternatives and there is a subsequent
1756     matching failure.
1757 nigel 93 .P
1758     When a subpattern is used as a subroutine, processing options such as
1759     case-independence are fixed when the subpattern is defined. They cannot be
1760     changed for different calls. For example, consider this pattern:
1761     .sp
1762     (abc)(?i:(?1))
1763     .sp
1764     It matches "abcabc". It does not match "abcABC" because the change of
1765     processing option does not affect the called subpattern.
1766 nigel 75 .
1767     .
1768 nigel 63 .SH CALLOUTS
1769     .rs
1770     .sp
1771     Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1772     code to be obeyed in the middle of matching a regular expression. This makes it
1773     possible, amongst other things, to extract different substrings that match the
1774     same pair of parentheses when there is a repetition.
1775 nigel 75 .P
1776 nigel 63 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1777     code. The feature is called "callout". The caller of PCRE provides an external
1778 nigel 75 function by putting its entry point in the global variable \fIpcre_callout\fP.
1779 nigel 63 By default, this variable contains NULL, which disables all calling out.
1780 nigel 75 .P
1781 nigel 63 Within a regular expression, (?C) indicates the points at which the external
1782     function is to be called. If you want to identify different callout points, you
1783     can put a number less than 256 after the letter C. The default value is zero.
1784     For example, this pattern has two callout points:
1785 nigel 75 .sp
1786 nigel 63 (?C1)\dabc(?C2)def
1787 nigel 75 .sp
1788     If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
1789     automatically installed before each item in the pattern. They are all numbered
1790     255.
1791     .P
1792     During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
1793 nigel 63 set), the external function is called. It is provided with the number of the
1794 nigel 75 callout, the position in the pattern, and, optionally, one item of data
1795     originally supplied by the caller of \fBpcre_exec()\fP. The callout function
1796     may cause matching to proceed, to backtrack, or to fail altogether. A complete
1797     description of the interface to the callout function is given in the
1798 nigel 63 .\" HREF
1799 nigel 75 \fBpcrecallout\fP
1800 nigel 63 .\"
1801     documentation.
1802 nigel 93 .
1803     .
1804     .SH "SEE ALSO"
1805     .rs
1806     .sp
1807     \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3), \fBpcre\fP(3).
1808 ph10 99 .
1809     .
1810     .SH AUTHOR
1811     .rs
1812     .sp
1813     .nf
1814     Philip Hazel
1815     University Computing Service
1816     Cambridge CB2 3QH, England.
1817     .fi
1818     .
1819     .
1820     .SH REVISION
1821     .rs
1822     .sp
1823     .nf
1824     Last updated: 06 March 2007
1825     Copyright (c) 1997-2007 University of Cambridge.
1826     .fi


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

ViewVC Help
Powered by ViewVC 1.1.12