ViewVC logotype

Contents of /code/trunk/pcre.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 23 - (show annotations)
Sat Feb 24 21:38:41 2007 UTC (14 years, 8 months ago) by nigel
File size: 54947 byte(s)
Error occurred while calculating annotation data.
Load pcre-2.00 into code/trunk.
1 .TH PCRE 3
3 pcre - Perl-compatible regular expressions.
5 .B #include <pcre.h>
6 .PP
7 .SM
8 .br
9 .B pcre *pcre_compile(const char *\fIpattern\fR, int \fIoptions\fR,
10 .ti +5n
11 .B const char **\fIerrptr\fR, int *\fIerroffset\fR);
12 .PP
13 .br
14 .B pcre_extra *pcre_study(const pcre *\fIcode\fR, int \fIoptions\fR,
15 .ti +5n
16 .B const char **\fIerrptr\fR);
17 .PP
18 .br
19 .B int pcre_exec(const pcre *\fIcode\fR, "const pcre_extra *\fIextra\fR,"
20 .ti +5n
21 .B "const char *\fIsubject\fR," int \fIlength\fR, int \fIoptions\fR,
22 .ti +5n
23 .B int *\fIovector\fR, int \fIovecsize\fR);
24 .PP
25 .br
26 .B int pcre_info(const pcre *\fIcode\fR, int *\fIoptptr\fR, int
27 .B *\fIfirstcharptr\fR);
28 .PP
29 .br
30 .B char *pcre_version(void);
31 .PP
32 .br
33 .B void *(*pcre_malloc)(size_t);
34 .PP
35 .br
36 .B void (*pcre_free)(void *);
37 .PP
38 .br
39 .B unsigned char *pcre_cbits[128];
40 .PP
41 .br
42 .B unsigned char *pcre_ctypes[256];
43 .PP
44 .br
45 .B unsigned char *pcre_fcc[256];
46 .PP
47 .br
48 .B unsigned char *pcre_lcc[256];
53 The PCRE library is a set of functions that implement regular expression
54 pattern matching using the same syntax and semantics as Perl 5, with just a few
55 differences (see below). The current implementation corresponds to Perl 5.005.
57 PCRE has its own native API, which is described in this man page. There is also
58 a set of wrapper functions that correspond to the POSIX API. See
59 \fBpcreposix (3)\fR.
61 The three functions \fBpcre_compile()\fR, \fBpcre_study()\fR, and
62 \fBpcre_exec()\fR are used for compiling and matching regular expressions. The
63 function \fBpcre_info()\fR is used to find out information about a compiled
64 pattern, while the function \fBpcre_version()\fR returns a pointer to a string
65 containing the version of PCRE and its date of release.
67 The global variables \fBpcre_malloc\fR and \fBpcre_free\fR initially contain
68 the entry points of the standard \fBmalloc()\fR and \fBfree()\fR functions
69 respectively. PCRE calls the memory management functions via these variables,
70 so a calling program can replace them if it wishes to intercept the calls. This
71 should be done before calling any PCRE functions.
73 The other global variables are character tables. They are initialized when PCRE
74 is compiled, from source that is generated by reference to the C character type
75 functions, but which a user of PCRE is free to modify. In principle the tables
76 could also be modified at run time. See PCRE's README file for more details.
80 The PCRE functions can be used in multi-threading applications, with the
81 proviso that the character tables and the memory management functions pointed
82 to by \fBpcre_malloc\fR and \fBpcre_free\fR are shared by all threads.
84 The compiled form of a regular expression is not altered during matching, so
85 the same compiled pattern can safely be used by several threads at once.
89 The function \fBpcre_compile()\fR is called to compile a pattern into an
90 internal form. The pattern is a C string terminated by a binary zero, and
91 is passed in the argument \fIpattern\fR. A pointer to the compiled code block
92 is returned. The \fBpcre\fR type is defined for this for convenience, but in
93 fact \fBpcre\fR is just a typedef for \fBvoid\fR, since the contents of the
94 block are not defined.
95 .PP
96 The size of a compiled pattern is roughly proportional to the length of the
97 pattern string, except that each character class (other than those containing
98 just a single character, negated or not) requires 33 bytes, and repeat
99 quantifiers with a minimum greater than one or a bounded maximum cause the
100 relevant portions of the compiled pattern to be replicated.
101 .PP
102 The \fIoptions\fR argument contains independent bits that affect the
103 compilation. It should be zero if no options are required. Some of the options,
104 in particular, those that are compatible with Perl, can also be set and unset
105 from within the pattern (see the detailed description of regular expressions
106 below). For these options, the contents of the \fIoptions\fR argument specifies
107 their initial settings at the start of compilation and execution. The
108 PCRE_ANCHORED option can be set at the time of matching as well as at compile
109 time.
110 .PP
111 If \fIerrptr\fR is NULL, \fBpcre_compile()\fR returns NULL immediately.
112 Otherwise, if compilation of a pattern fails, \fBpcre_compile()\fR returns
113 NULL, and sets the variable pointed to by \fIerrptr\fR to point to a textual
114 error message.
116 The offset from the start of the pattern to the character where the error was
117 discovered is placed in the variable pointed to by \fIerroffset\fR, which must
118 not be NULL. If it is, an immediate error is given.
119 .PP
120 The following option bits are defined in the header file:
124 If this bit is set, the pattern is forced to be "anchored", that is, it is
125 constrained to match only at the start of the string which is being searched
126 (the "subject string"). This effect can also be achieved by appropriate
127 constructs in the pattern itself, which is the only way to do it in Perl.
131 If this bit is set, letters in the pattern match both upper and lower case
132 letters. It is equivalent to Perl's /i option.
136 If this bit is set, a dollar metacharacter in the pattern matches only at the
137 end of the subject string. Without this option, a dollar also matches
138 immediately before the final character if it is a newline (but not before any
139 other newlines). The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
140 set. There is no equivalent to this option in Perl.
144 If this bit is set, a dot metacharater in the pattern matches all characters,
145 including newlines. Without it, newlines are excluded. This option is
146 equivalent to Perl's /s option. A negative class such as [^a] always matches a
147 newline character, independent of the setting of this option.
151 If this bit is set, whitespace data characters in the pattern are totally
152 ignored except when escaped or inside a character class, and characters between
153 an unescaped # outside a character class and the next newline character,
154 inclusive, are also ignored. This is equivalent to Perl's /x option, and makes
155 it possible to include comments inside complicated patterns. Note, however,
156 that this applies only to data characters. Whitespace characters may never
157 appear within special character sequences in a pattern, for example within the
158 sequence (?( which introduces a conditional subpattern.
162 This option turns on additional functionality of PCRE that is incompatible with
163 Perl. Any backslash in a pattern that is followed by a letter that has no
164 special meaning causes an error, thus reserving these combinations for future
165 expansion. By default, as in Perl, a backslash followed by a letter with no
166 special meaning is treated as a literal. There are at present no other features
167 controlled by this option.
171 By default, PCRE treats the subject string as consisting of a single "line" of
172 characters (even if it actually contains several newlines). The "start of line"
173 metacharacter (^) matches only at the start of the string, while the "end of
174 line" metacharacter ($) matches only at the end of the string, or before a
175 terminating newline (unless PCRE_DOLLAR_ENDONLY is set). This is the same as
176 Perl.
178 When PCRE_MULTILINE it is set, the "start of line" and "end of line" constructs
179 match immediately following or immediately before any newline in the subject
180 string, respectively, as well as at the very start and end. This is equivalent
181 to Perl's /m option. If there are no "\\n" characters in a subject string, or
182 no occurrences of ^ or $ in a pattern, setting PCRE_MULTILINE has no
183 effect.
187 This option inverts the "greediness" of the quantifiers so that they are not
188 greedy by default, but become greedy if followed by "?". It is not compatible
189 with Perl. It can also be set by a (?U) option setting within the pattern.
193 When a pattern is going to be used several times, it is worth spending more
194 time analyzing it in order to speed up the time taken for matching. The
195 function \fBpcre_study()\fR takes a pointer to a compiled pattern as its first
196 argument, and returns a pointer to a \fBpcre_extra\fR block (another \fBvoid\fR
197 typedef) containing additional information about the pattern; this can be
198 passed to \fBpcre_exec()\fR. If no additional information is available, NULL
199 is returned.
201 The second argument contains option bits. At present, no options are defined
202 for \fBpcre_study()\fR, and this argument should always be zero.
204 The third argument for \fBpcre_study()\fR is a pointer to an error message. If
205 studying succeeds (even if no data is returned), the variable it points to is
206 set to NULL. Otherwise it points to a textual error message.
208 At present, studying a pattern is useful only for non-anchored patterns that do
209 not have a single fixed starting character. A bitmap of possible starting
210 characters is created.
214 The function \fBpcre_exec()\fR is called to match a subject string against a
215 pre-compiled pattern, which is passed in the \fIcode\fR argument. If the
216 pattern has been studied, the result of the study should be passed in the
217 \fIextra\fR argument. Otherwise this must be NULL.
219 The subject string is passed as a pointer in \fIsubject\fR and a length in
220 \fIlength\fR. Unlike the pattern string, it may contain binary zero characters.
222 The PCRE_ANCHORED option can be passed in the \fIoptions\fR argument, whose
223 unused bits must be zero. However, if a pattern was compiled with
224 PCRE_ANCHORED, or turned out to be anchored by virtue of its contents, it
225 cannot be made unachored at matching time.
227 There are also two further options that can be set only at matching time:
231 The first character of the string is not the beginning of a line, so the
232 circumflex metacharacter should not match before it. Setting this without
233 PCRE_MULTILINE (at compile time) causes circumflex never to match.
237 The end of the string is not the end of a line, so the dollar metacharacter
238 should not match it nor (except in multiline mode) a newline immediately before
239 it. Setting this without PCRE_MULTILINE (at compile time) causes dollar never
240 to match.
242 In general, a pattern matches a certain portion of the subject, and in
243 addition, further substrings from the subject may be picked out by parts of the
244 pattern. Following the usage in Jeffrey Friedl's book, this is called
245 "capturing" in what follows, and the phrase "capturing subpattern" is used for
246 a fragment of a pattern that picks out a substring. PCRE supports several other
247 kinds of parenthesized subpattern that do not cause substrings to be captured.
249 Captured substrings are returned to the caller via a vector of integer offsets
250 whose address is passed in \fIovector\fR. The number of elements in the vector
251 is passed in \fIovecsize\fR. The first two-thirds of the vector is used to pass
252 back captured substrings, each substring using a pair of integers. The
253 remaining third of the vector is used as workspace by \fBpcre_exec()\fR while
254 matching capturing subpatterns, and is not available for passing back
255 information. The length passed in \fIovecsize\fR should always be a multiple of
256 three. If it is not, it is rounded down.
258 When a match has been successful, information about captured substrings is
259 returned in pairs of integers, starting at the beginning of \fIovector\fR, and
260 continuing up to two-thirds of its length at the most. The first element of a
261 pair is set to the offset of the first character in a substring, and the second
262 is set to the offset of the first character after the end of a substring. The
263 first pair, \fIovector[0]\fR and \fIovector[1]\fR, identify the portion of the
264 subject string matched by the entire pattern. The next pair is used for the
265 first capturing subpattern, and so on. The value returned by \fBpcre_exec()\fR
266 is the number of pairs that have been set. If there are no capturing
267 subpatterns, the return value from a successful match is 1, indicating that
268 just the first pair of offsets has been set.
270 It is possible for an capturing subpattern number \fIn+1\fR to match some
271 part of the subject when subpattern \fIn\fR has not been used at all. For
272 example, if the string "abc" is matched against the pattern (a|(z))(bc)
273 subpatterns 1 and 3 are matched, but 2 is not. When this happens, both offset
274 values corresponding to the unused subpattern are set to -1.
276 If a capturing subpattern is matched repeatedly, it is the last portion of the
277 string that it matched that gets returned.
279 If the vector is too small to hold all the captured substrings, it is used as
280 far as possible (up to two-thirds of its length), and the function returns a
281 value of zero. In particular, if the substring offsets are not of interest,
282 \fBpcre_exec()\fR may be called with \fIovector\fR passed as NULL and
283 \fIovecsize\fR as zero. However, if the pattern contains back references and
284 the \fIovector\fR isn't big enough to remember the related substrings, PCRE has
285 to get additional memory for use during matching. Thus it is usually advisable
286 to supply an \fIovector\fR.
288 Note that \fBpcre_info()\fR can be used to find out how many capturing
289 subpatterns there are in a compiled pattern. The smallest size for
290 \fIovector\fR that will allow for \fIn\fR captured substrings in addition to
291 the offsets of the substring matched by the whole pattern is (\fIn\fR+1)*3.
293 If \fBpcre_exec()\fR fails, it returns a negative number. The following are
294 defined in the header file:
298 The subject string did not match the pattern.
302 Either \fIcode\fR or \fIsubject\fR was passed as NULL, or \fIovector\fR was
303 NULL and \fIovecsize\fR was not zero.
307 An unrecognized bit was set in the \fIoptions\fR argument.
311 PCRE stores a 4-byte "magic number" at the start of the compiled code, to catch
312 the case when it is passed a junk pointer. This is the error it gives when the
313 magic number isn't present.
317 While running the pattern match, an unknown item was encountered in the
318 compiled pattern. This error could be caused by a bug in PCRE or by overwriting
319 of the compiled pattern.
323 If a pattern contains back references, but the \fIovector\fR that is passed to
324 \fBpcre_exec()\fR is not big enough to remember the referenced substrings, PCRE
325 gets a block of memory at the start of matching to use for this purpose. If the
326 call via \fBpcre_malloc()\fR fails, this error is given. The memory is freed at
327 the end of matching.
331 The \fBpcre_info()\fR function returns information about a compiled pattern.
332 Its yield is the number of capturing subpatterns, or one of the following
333 negative numbers:
335 PCRE_ERROR_NULL the argument \fIcode\fR was NULL
336 PCRE_ERROR_BADMAGIC the "magic number" was not found
338 If the \fIoptptr\fR argument is not NULL, a copy of the options with which the
339 pattern was compiled is placed in the integer it points to.
341 If the \fIfirstcharptr\fR argument is not NULL, is is used to pass back
342 information about the first character of any matched string. If there is a
343 fixed first character, e.g. from a pattern such as (cat|cow|coyote), then it is
344 returned in the integer pointed to by \fIfirstcharptr\fR. Otherwise, if the
345 pattern was compiled with the PCRE_MULTILINE option, and every branch started
346 with "^", then -1 is returned, indicating that the pattern will match at the
347 start of a subject string or after any "\\n" within the string. Otherwise -2 is
348 returned.
352 There are some size limitations in PCRE but it is hoped that they will never in
353 practice be relevant.
354 The maximum length of a compiled pattern is 65539 (sic) bytes.
355 All values in repeating quantifiers must be less than 65536.
356 The maximum number of capturing subpatterns is 99.
357 The maximum number of all parenthesized subpatterns, including capturing
358 subpatterns, assertions, and other types of subpattern, is 200.
360 The maximum length of a subject string is the largest positive number that an
361 integer variable can hold. However, PCRE uses recursion to handle subpatterns
362 and indefinite repetition. This means that the available stack space may limit
363 the size of a subject string that can be processed by certain patterns.
367 The differences described here are with respect to Perl 5.005.
369 1. By default, a whitespace character is any character that the C library
370 function \fBisspace()\fR recognizes, though it is possible to compile PCRE with
371 alternative character type tables. Normally \fBisspace()\fR matches space,
372 formfeed, newline, carriage return, horizontal tab, and vertical tab. Perl 5
373 no longer includes vertical tab in its set of whitespace characters. The \\v
374 escape that was in the Perl documentation for a long time was never in fact
375 recognized. However, the character itself was treated as whitespace at least
376 up to 5.002. In 5.004 and 5.005 it does not match \\s.
378 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl permits
379 them, but they do not mean what you might think. For example, (?!a){3} does
380 not assert that the next three characters are not "a". It just asserts that the
381 next character is not "a" three times.
383 3. Capturing subpatterns that occur inside negative lookahead assertions are
384 counted, but their entries in the offsets vector are never set. Perl sets its
385 numerical variables from any such patterns that are matched before the
386 assertion fails to match something (thereby succeeding), but only if the
387 negative lookahead assertion contains just one branch.
389 4. Though binary zero characters are supported in the subject string, they are
390 not allowed in a pattern string because it is passed as a normal C string,
391 terminated by zero. The escape sequence "\\0" can be used in the pattern to
392 represent a binary zero.
394 5. The following Perl escape sequences are not supported: \\l, \\u, \\L, \\U,
395 \\E, \\Q. In fact these are implemented by Perl's general string-handling and
396 are not part of its pattern matching engine.
398 6. The Perl \\G assertion is not supported as it is not relevant to single
399 pattern matches.
401 7. Fairly obviously, PCRE does not support the (?{code}) construction.
403 8. There are at the time of writing some oddities in Perl 5.005_02 concerned
404 with the settings of captured strings when part of a pattern is repeated. For
405 example, matching "aba" against the pattern /^(a(b)?)+$/ sets $2 to the value
406 "b", but matching "aabbaa" against /^(aa(bb)?)+$/ leaves $2 unset. However, if
407 the pattern is changed to /^(aa(b(b))?)+$/ then $2 (and $3) get set.
409 In Perl 5.004 $2 is set in both cases, and that is also true of PCRE. If in the
410 future Perl changes to a consistent state that is different, PCRE may change to
411 follow.
413 9. Another as yet unresolved discrepancy is that in Perl 5.005_02 the pattern
414 /^(a)?(?(1)a|b)+$/ matches the string "a", whereas in PCRE it does not.
415 However, in both Perl and PCRE /^(a)?a/ matched against "a" leaves $1 unset.
417 10. PCRE provides some extensions to the Perl regular expression facilities:
419 (a) Although lookbehind assertions must match fixed length strings, each
420 alternative branch of a lookbehind assertion can match a different length of
421 string. Perl 5.005 requires them all to have the same length.
423 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $ meta-
424 character matches only at the very end of the string.
426 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no special
427 meaning is faulted.
429 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quantifiers is
430 inverted, that is, by default they are not greedy, but if followed by a
431 question mark they are.
435 The syntax and semantics of the regular expressions supported by PCRE are
436 described below. Regular expressions are also described in the Perl
437 documentation and in a number of other books, some of which have copious
438 examples. Jeffrey Friedl's "Mastering Regular Expressions", published by
439 O'Reilly (ISBN 1-56592-257-3), covers them in great detail. The description
440 here is intended as reference documentation.
442 A regular expression is a pattern that is matched against a subject string from
443 left to right. Most characters stand for themselves in a pattern, and match the
444 corresponding characters in the subject. As a trivial example, the pattern
446 The quick brown fox
448 matches a portion of a subject string that is identical to itself. The power of
449 regular expressions comes from the ability to include alternatives and
450 repetitions in the pattern. These are encoded in the pattern by the use of
451 \fImeta-characters\fR, which do not stand for themselves but instead are
452 interpreted in some special way.
454 There are two different sets of meta-characters: those that are recognized
455 anywhere in the pattern except within square brackets, and those that are
456 recognized in square brackets. Outside square brackets, the meta-characters are
457 as follows:
459 \\ general escape character with several uses
460 ^ assert start of subject (or line, in multiline mode)
461 $ assert end of subject (or line, in multiline mode)
462 . match any character except newline (by default)
463 [ start character class definition
464 | start of alternative branch
465 ( start subpattern
466 ) end subpattern
467 ? extends the meaning of (
468 also 0 or 1 quantifier
469 also quantifier minimizer
470 * 0 or more quantifier
471 + 1 or more quantifier
472 { start min/max quantifier
474 Part of a pattern that is in square brackets is called a "character class". In
475 a character class the only meta-characters are:
477 \\ general escape character
478 ^ negate the class, but only if the first character
479 - indicates character range
480 ] terminates the character class
482 The following sections describe the use of each of the meta-characters.
486 The backslash character has several uses. Firstly, if it is followed by a
487 non-alphameric character, it takes away any special meaning that character may
488 have. This use of backslash as an escape character applies both inside and
489 outside character classes.
491 For example, if you want to match a "*" character, you write "\\*" in the
492 pattern. This applies whether or not the following character would otherwise be
493 interpreted as a meta-character, so it is always safe to precede a
494 non-alphameric with "\\" to specify that it stands for itself. In particular,
495 if you want to match a backslash, you write "\\\\".
497 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
498 pattern (other than in a character class) and characters between a "#" outside
499 a character class and the next newline character are ignored. An escaping
500 backslash can be used to include a whitespace or "#" character as part of the
501 pattern.
503 A second use of backslash provides a way of encoding non-printing characters
504 in patterns in a visible manner. There is no restriction on the appearance of
505 non-printing characters, apart from the binary zero that terminates a pattern,
506 but when a pattern is being prepared by text editing, it is usually easier to
507 use one of the following escape sequences than the binary character it
508 represents:
510 \\a alarm, that is, the BEL character (hex 07)
511 \\cx "control-x", where x is any character
512 \\e escape (hex 1B)
513 \\f formfeed (hex 0C)
514 \\n newline (hex 0A)
515 \\r carriage return (hex 0D)
516 \\t tab (hex 09)
517 \\xhh character with hex code hh
518 \\ddd character with octal code ddd, or backreference
520 The precise effect of "\\cx" is as follows: if "x" is a lower case letter, it
521 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
522 Thus "\\cz" becomes hex 1A, but "\\c{" becomes hex 3B, while "\\c;" becomes hex
523 7B.
525 After "\\x", up to two hexadecimal digits are read (letters can be in upper or
526 lower case).
528 After "\\0" up to two further octal digits are read. In both cases, if there
529 are fewer than two digits, just those that are present are used. Thus the
530 sequence "\\0\\x\\07" specifies two binary zeros followed by a BEL character.
531 Make sure you supply two digits after the initial zero if the character that
532 follows is itself an octal digit.
534 The handling of a backslash followed by a digit other than 0 is complicated.
535 Outside a character class, PCRE reads it and any following digits as a decimal
536 number. If the number is less than 10, or if there have been at least that many
537 previous capturing left parentheses in the expression, the entire sequence is
538 taken as a \fIback reference\fR. A description of how this works is given
539 later, following the discussion of parenthesized subpatterns.
541 Inside a character class, or if the decimal number is greater than 9 and there
542 have not been that many capturing subpatterns, PCRE re-reads up to three octal
543 digits following the backslash, and generates a single byte from the least
544 significant 8 bits of the value. Any subsequent digits stand for themselves.
545 For example:
547 \\040 is another way of writing a space
548 \\40 is the same, provided there are fewer than 40
549 previous capturing subpatterns
550 \\7 is always a back reference
551 \\11 might be a back reference, or another way of
552 writing a tab
553 \\011 is always a tab
554 \\0113 is a tab followed by the character "3"
555 \\113 is the character with octal code 113 (since there
556 can be no more than 99 back references)
557 \\377 is a byte consisting entirely of 1 bits
558 \\81 is either a back reference, or a binary zero
559 followed by the two characters "8" and "1"
561 Note that octal values of 100 or greater must not be introduced by a leading
562 zero, because no more than three octal digits are ever read.
564 All the sequences that define a single byte value can be used both inside and
565 outside character classes. In addition, inside a character class, the sequence
566 "\\b" is interpreted as the backspace character (hex 08). Outside a character
567 class it has a different meaning (see below).
569 The third use of backslash is for specifying generic character types:
571 \\d any decimal digit
572 \\D any character that is not a decimal digit
573 \\s any whitespace character
574 \\S any character that is not a whitespace character
575 \\w any "word" character
576 \\W any "non-word" character
578 Each pair of escape sequences partitions the complete set of characters into
579 two disjoint sets. Any given character matches one, and only one, of each pair.
581 A "word" character is any letter or digit or the underscore character, that is,
582 any character which can be part of a Perl "word". These character type
583 sequences can appear both inside and outside character classes. They each match
584 one character of the appropriate type. If the current matching point is at the
585 end of the subject string, all of them fail, since there is no character to
586 match.
588 The fourth use of backslash is for certain simple assertions. An assertion
589 specifies a condition that has to be met at a particular point in a match,
590 without consuming any characters from the subject string. The use of
591 subpatterns for more complicated assertions is described below. The backslashed
592 assertions are
594 \\b word boundary
595 \\B not a word boundary
596 \\A start of subject (independent of multiline mode)
597 \\Z end of subject or newline at end (independent of multiline mode)
598 \\z end of subject (independent of multiline mode)
600 These assertions may not appear in character classes (but note that "\\b" has a
601 different meaning, namely the backspace character, inside a character class).
603 A word boundary is a position in the subject string where the current character
604 and the previous character do not both match \\w or \\W (i.e. one matches
605 \\w and the other matches \\W), or the start or end of the string if the
606 first or last character matches \\w, respectively.
608 The \\A, \\Z, and \\z assertions differ from the traditional circumflex and
609 dollar (described below) in that they only ever match at the very start and end
610 of the subject string, whatever options are set. They are not affected by the
611 PCRE_NOTBOL or PCRE_NOTEOL options. The difference between \\Z and \\z is that
612 \\Z matches before a newline that is the last character of the string as well
613 as at the end of the string, whereas \\z matches only at the end.
617 Outside a character class, in the default matching mode, the circumflex
618 character is an assertion which is true only if the current matching point is
619 at the start of the subject string. Inside a character class, circumflex has an
620 entirely different meaning (see below).
622 Circumflex need not be the first character of the pattern if a number of
623 alternatives are involved, but it should be the first thing in each alternative
624 in which it appears if the pattern is ever to match that branch. If all
625 possible alternatives start with a circumflex, that is, if the pattern is
626 constrained to match only at the start of the subject, it is said to be an
627 "anchored" pattern. (There are also other constructs that can cause a pattern
628 to be anchored.)
630 A dollar character is an assertion which is true only if the current matching
631 point is at the end of the subject string, or immediately before a newline
632 character that is the last character in the string (by default). Dollar need
633 not be the last character of the pattern if a number of alternatives are
634 involved, but it should be the last item in any branch in which it appears.
635 Dollar has no special meaning in a character class.
637 The meaning of dollar can be changed so that it matches only at the very end of
638 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile or matching
639 time. This does not affect the \\Z assertion.
641 The meanings of the circumflex and dollar characters are changed if the
642 PCRE_MULTILINE option is set. When this is the case, they match immediately
643 after and immediately before an internal "\\n" character, respectively, in
644 addition to matching at the start and end of the subject string. For example,
645 the pattern /^abc$/ matches the subject string "def\\nabc" in multiline mode,
646 but not otherwise. Consequently, patterns that are anchored in single line mode
647 because all branches start with "^" are not anchored in multiline mode. The
648 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
650 Note that the sequences \\A, \\Z, and \\z can be used to match the start and
651 end of the subject in both modes, and if all branches of a pattern start with
652 \\A is it always anchored, whether PCRE_MULTILINE is set or not.
656 Outside a character class, a dot in the pattern matches any one character in
657 the subject, including a non-printing character, but not (by default) newline.
658 If the PCRE_DOTALL option is set, then dots match newlines as well. The
659 handling of dot is entirely independent of the handling of circumflex and
660 dollar, the only relationship being that they both involve newline characters.
661 Dot has no special meaning in a character class.
665 An opening square bracket introduces a character class, terminated by a closing
666 square bracket. A closing square bracket on its own is not special. If a
667 closing square bracket is required as a member of the class, it should be the
668 first data character in the class (after an initial circumflex, if present) or
669 escaped with a backslash.
671 A character class matches a single character in the subject; the character must
672 be in the set of characters defined by the class, unless the first character in
673 the class is a circumflex, in which case the subject character must not be in
674 the set defined by the class. If a circumflex is actually required as a member
675 of the class, ensure it is not the first character, or escape it with a
676 backslash.
678 For example, the character class [aeiou] matches any lower case vowel, while
679 [^aeiou] matches any character that is not a lower case vowel. Note that a
680 circumflex is just a convenient notation for specifying the characters which
681 are in the class by enumerating those that are not. It is not an assertion: it
682 still consumes a character from the subject string, and fails if the current
683 pointer is at the end of the string.
685 When PCRE_CASELESS is set, any letters in a class represent both their upper
686 case and lower case versions, so for example, a caseless [aeiou] matches "A" as
687 well as "a", and a caseless [^aeiou] does not match "A", whereas a caseful
688 version would.
690 The newline character is never treated in any special way in character classes,
691 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
692 such as [^a] will always match a newline.
694 The minus (hyphen) character can be used to specify a range of characters in a
695 character class. For example, [d-m] matches any letter between d and m,
696 inclusive. If a minus character is required in a class, it must be escaped with
697 a backslash or appear in a position where it cannot be interpreted as
698 indicating a range, typically as the first or last character in the class. It
699 is not possible to have the character "]" as the end character of a range,
700 since a sequence such as [w-] is interpreted as a class of two characters. The
701 octal or hexadecimal representation of "]" can, however, be used to end a
702 range.
704 Ranges operate in ASCII collating sequence. They can also be used for
705 characters specified numerically, for example [\\000-\\037]. If a range such as
706 [W-c] is used when PCRE_CASELESS is set, it matches the letters involved in
707 either case, so is equivalent to [][\\^_`wxyzabc], matched caselessly.
709 The character types \\d, \\D, \\s, \\S, \\w, and \\W may also appear in a
710 character class, and add the characters that they match to the class. For
711 example, [\\dABCDEF] matches any hexadecimal digit. A circumflex can
712 conveniently be used with the upper case character types to specify a more
713 restricted set of characters than the matching lower case type. For example,
714 the class [^\\W_] matches any letter or digit, but not underscore.
716 All non-alphameric characters other than \\, -, ^ (at the start) and the
717 terminating ] are non-special in character classes, but it does no harm if they
718 are escaped.
722 Vertical bar characters are used to separate alternative patterns. For example,
723 the pattern
725 gilbert|sullivan
727 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
728 and an empty alternative is permitted (matching the empty string).
729 The matching process tries each alternative in turn, from left to right,
730 and the first one that succeeds is used. If the alternatives are within a
731 subpattern (defined below), "succeeds" means matching the rest of the main
732 pattern as well as the alternative in the subpattern.
737 can be changed from within the pattern by a sequence of Perl option letters
738 enclosed between "(?" and ")". The option letters are
742 s for PCRE_DOTALL
745 For example, (?im) sets caseless, multiline matching. It is also possible to
746 unset these options by preceding the letter with a hyphen, and a combined
747 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
748 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
749 permitted. If a letter appears both before and after the hyphen, the option is
750 unset.
752 The scope of these option changes depends on where in the pattern the setting
753 occurs. For settings that are outside any subpattern (defined below), the
754 effect is the same as if the options were set or unset at the start of
755 matching. The following patterns all behave in exactly the same way:
757 (?i)abc
758 a(?i)bc
759 ab(?i)c
760 abc(?i)
762 which in turn is the same as compiling the pattern abc with PCRE_CASELESS set.
763 In other words, such "top level" settings apply to the whole pattern (unless
764 there are other changes inside subpatterns). If there is more than one setting
765 of the same option at top level, the rightmost setting is used.
767 If an option change occurs inside a subpattern, the effect is different. This
768 is a change of behaviour in Perl 5.005. An option change inside a subpattern
769 affects only that part of the subpattern that follows it, so
771 (a(?i)b)c
773 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
774 By this means, options can be made to have different settings in different
775 parts of the pattern. Any changes made in one alternative do carry on
776 into subsequent branches within the same subpattern. For example,
778 (a(?i)b|c)
780 matches "ab", "aB", "c", and "C", even though when matching "C" the first
781 branch is abandoned before the option setting. This is because the effects of
782 option settings happen at compile time. There would be some very weird
783 behaviour otherwise.
785 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
786 same way as the Perl-compatible options by using the characters U and X
787 respectively. The (?X) flag setting is special in that it must always occur
788 earlier in the pattern than any of the additional features it turns on, even
789 when it is at top level. It is best put at the start.
793 Subpatterns are delimited by parentheses (round brackets), which can be nested.
794 Marking part of a pattern as a subpattern does two things:
796 1. It localizes a set of alternatives. For example, the pattern
798 cat(aract|erpillar|)
800 matches one of the words "cat", "cataract", or "caterpillar". Without the
801 parentheses, it would match "cataract", "erpillar" or the empty string.
803 2. It sets up the subpattern as a capturing subpattern (as defined above).
804 When the whole pattern matches, that portion of the subject string that matched
805 the subpattern is passed back to the caller via the \fIovector\fR argument of
806 \fBpcre_exec()\fR. Opening parentheses are counted from left to right (starting
807 from 1) to obtain the numbers of the capturing subpatterns.
809 For example, if the string "the red king" is matched against the pattern
811 the ((red|white) (king|queen))
813 the captured substrings are "red king", "red", and "king", and are numbered 1,
814 2, and 3.
816 The fact that plain parentheses fulfil two functions is not always helpful.
817 There are often times when a grouping subpattern is required without a
818 capturing requirement. If an opening parenthesis is followed by "?:", the
819 subpattern does not do any capturing, and is not counted when computing the
820 number of any subsequent capturing subpatterns. For example, if the string "the
821 white queen" is matched against the pattern
823 the ((?:red|white) (king|queen))
825 the captured substrings are "white queen" and "queen", and are numbered 1 and
826 2. The maximum number of captured substrings is 99, and the maximum number of
827 all subpatterns, both capturing and non-capturing, is 200.
829 As a convenient shorthand, if any option settings are required at the start of
830 a non-capturing subpattern, the option letters may appear between the "?" and
831 the ":". Thus the two patterns
833 (?i:saturday|sunday)
834 (?:(?i)saturday|sunday)
836 match exactly the same set of strings. Because alternative branches are tried
837 from left to right, and options are not reset until the end of the subpattern
838 is reached, an option setting in one branch does affect subsequent branches, so
839 the above patterns match "SUNDAY" as well as "Saturday".
843 Repetition is specified by quantifiers, which can follow any of the following
844 items:
846 a single character, possibly escaped
847 the . metacharacter
848 a character class
849 a back reference (see next section)
850 a parenthesized subpattern (unless it is an assertion - see below)
852 The general repetition quantifier specifies a minimum and maximum number of
853 permitted matches, by giving the two numbers in curly brackets (braces),
854 separated by a comma. The numbers must be less than 65536, and the first must
855 be less than or equal to the second. For example:
857 z{2,4}
859 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
860 character. If the second number is omitted, but the comma is present, there is
861 no upper limit; if the second number and the comma are both omitted, the
862 quantifier specifies an exact number of required matches. Thus
864 [aeiou]{3,}
866 matches at least 3 successive vowels, but may match many more, while
868 \\d{8}
870 matches exactly 8 digits. An opening curly bracket that appears in a position
871 where a quantifier is not allowed, or one that does not match the syntax of a
872 quantifier, is taken as a literal character. For example, {,6} is not a
873 quantifier, but a literal string of four characters.
875 The quantifier {0} is permitted, causing the expression to behave as if the
876 previous item and the quantifier were not present.
878 For convenience (and historical compatibility) the three most common
879 quantifiers have single-character abbreviations:
881 * is equivalent to {0,}
882 + is equivalent to {1,}
883 ? is equivalent to {0,1}
885 It is possible to construct infinite loops by following a subpattern that can
886 match no characters with a quantifier that has no upper limit, for example:
888 (a?)*
890 Earlier versions of Perl and PCRE used to give an error at compile time for
891 such patterns. However, because there are cases where this can be useful, such
892 patterns are now accepted, but if any repetition of the subpattern does in fact
893 match no characters, the loop is forcibly broken.
895 By default, the quantifiers are "greedy", that is, they match as much as
896 possible (up to the maximum number of permitted times), without causing the
897 rest of the pattern to fail. The classic example of where this gives problems
898 is in trying to match comments in C programs. These appear between the
899 sequences /* and */ and within the sequence, individual * and / characters may
900 appear. An attempt to match C comments by applying the pattern
902 /\\*.*\\*/
904 to the string
906 /* first command */ not comment /* second comment */
908 fails, because it matches the entire string due to the greediness of the .*
909 item.
911 However, if a quantifier is followed by a question mark, then it ceases to be
912 greedy, and instead matches the minimum number of times possible, so the
913 pattern
915 /\\*.*?\\*/
917 does the right thing with the C comments. The meaning of the various
918 quantifiers is not otherwise changed, just the preferred number of matches.
919 Do not confuse this use of question mark with its use as a quantifier in its
920 own right. Because it has two uses, it can sometimes appear doubled, as in
922 \\d??\\d
924 which matches one digit by preference, but can match two if that is the only
925 way the rest of the pattern matches.
927 If the PCRE_UNGREEDY option is set (an option which is not available in Perl)
928 then the quantifiers are not greedy by default, but individual ones can be made
929 greedy by following them with a question mark. In other words, it inverts the
930 default behaviour.
932 When a parenthesized subpattern is quantified with a minimum repeat count that
933 is greater than 1 or with a limited maximum, more store is required for the
934 compiled pattern, in proportion to the size of the minimum or maximum.
936 If a pattern starts with .* then it is implicitly anchored, since whatever
937 follows will be tried against every character position in the subject string.
938 PCRE treats this as though it were preceded by \\A.
940 When a capturing subpattern is repeated, the value captured is the substring
941 that matched the final iteration. For example, after
943 (tweedle[dume]{3}\\s*)+
945 has matched "tweedledum tweedledee" the value of the captured substring is
946 "tweedledee". However, if there are nested capturing subpatterns, the
947 corresponding captured values may have been set in previous iterations. For
948 example, after
950 /(a|(b))+/
952 matches "aba" the value of the second captured substring is "b".
956 Outside a character class, a backslash followed by a digit greater than 0 (and
957 possibly further digits) is a back reference to a capturing subpattern earlier
958 (i.e. to its left) in the pattern, provided there have been that many previous
959 capturing left parentheses.
961 However, if the decimal number following the backslash is less than 10, it is
962 always taken as a back reference, and causes an error only if there are not
963 that many capturing left parentheses in the entire pattern. In other words, the
964 parentheses that are referenced need not be to the left of the reference for
965 numbers less than 10. See the section entitled "Backslash" above for further
966 details of the handling of digits following a backslash.
968 A back reference matches whatever actually matched the capturing subpattern in
969 the current subject string, rather than anything matching the subpattern
970 itself. So the pattern
972 (sens|respons)e and \\1ibility
974 matches "sense and sensibility" and "response and responsibility", but not
975 "sense and responsibility". If caseful matching is in force at the time of the
976 back reference, then the case of letters is relevant. For example,
978 ((?i)rah)\\s+\\1
980 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
981 capturing subpattern is matched caselessly.
983 There may be more than one back reference to the same subpattern. If a
984 subpattern has not actually been used in a particular match, then any back
985 references to it always fail. For example, the pattern
987 (a|(bc))\\2
989 always fails if it starts to match "a" rather than "bc". Because there may be
990 up to 99 back references, all digits following the backslash are taken
991 as part of a potential back reference number. If the pattern continues with a
992 digit character, then some delimiter must be used to terminate the back
993 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
994 Otherwise an empty comment can be used.
996 A back reference that occurs inside the parentheses to which it refers fails
997 when the subpattern is first used, so, for example, (a\\1) never matches.
998 However, such references can be useful inside repeated subpatterns. For
999 example, the pattern
1001 (a|b\\1)+
1003 matches any number of "a"s and also "aba", "ababaa" etc. At each iteration of
1004 the subpattern, the back reference matches the character string corresponding
1005 to the previous iteration. In order for this to work, the pattern must be such
1006 that the first iteration does not need to match the back reference. This can be
1007 done using alternation, as in the example above, or by a quantifier with a
1008 minimum of zero.
1012 An assertion is a test on the characters following or preceding the current
1013 matching point that does not actually consume any characters. The simple
1014 assertions coded as \\b, \\B, \\A, \\Z, \\z, ^ and $ are described above. More
1015 complicated assertions are coded as subpatterns. There are two kinds: those
1016 that look ahead of the current position in the subject string, and those that
1017 look behind it.
1019 An assertion subpattern is matched in the normal way, except that it does not
1020 cause the current matching position to be changed. Lookahead assertions start
1021 with (?= for positive assertions and (?! for negative assertions. For example,
1023 \\w+(?=;)
1025 matches a word followed by a semicolon, but does not include the semicolon in
1026 the match, and
1028 foo(?!bar)
1030 matches any occurrence of "foo" that is not followed by "bar". Note that the
1031 apparently similar pattern
1033 (?!foo)bar
1035 does not find an occurrence of "bar" that is preceded by something other than
1036 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1037 (?!foo) is always true when the next three characters are "bar". A
1038 lookbehind assertion is needed to achieve this effect.
1040 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1041 negative assertions. For example,
1043 (?<!foo)bar
1045 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1046 a lookbehind assertion are restricted such that all the strings it matches must
1047 have a fixed length. However, if there are several alternatives, they do not
1048 all have to have the same fixed length. Thus
1050 (?<=bullock|donkey)
1052 is permitted, but
1054 (?<!dogs?|cats?)
1056 causes an error at compile time. Branches that match different length strings
1057 are permitted only at the top level of a lookbehind assertion. This is an
1058 extension compared with Perl 5.005, which requires all branches to match the
1059 same length of string. An assertion such as
1061 (?<=ab(c|de))
1063 is not permitted, because its single branch can match two different lengths,
1064 but it is acceptable if rewritten to use two branches:
1066 (?<=abc|abde)
1068 The implementation of lookbehind assertions is, for each alternative, to
1069 temporarily move the current position back by the fixed width and then try to
1070 match. If there are insufficient characters before the current position, the
1071 match is deemed to fail.
1073 Assertions can be nested in any combination. For example,
1075 (?<=(?<!foo)bar)baz
1077 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1078 preceded by "foo".
1080 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1081 because it makes no sense to assert the same thing several times. If an
1082 assertion contains capturing subpatterns within it, these are always counted
1083 for the purposes of numbering the capturing subpatterns in the whole pattern.
1084 Substring capturing is carried out for positive assertions, but it does not
1085 make sense for negative assertions.
1087 Assertions count towards the maximum of 200 parenthesized subpatterns.
1091 With both maximizing and minimizing repetition, failure of what follows
1092 normally causes the repeated item to be re-evaluated to see if a different
1093 number of repeats allows the rest of the pattern to match. Sometimes it is
1094 useful to prevent this, either to change the nature of the match, or to cause
1095 it fail earlier than it otherwise might, when the author of the pattern knows
1096 there is no point in carrying on.
1098 Consider, for example, the pattern \\d+foo when applied to the subject line
1100 123456bar
1102 After matching all 6 digits and then failing to match "foo", the normal
1103 action of the matcher is to try again with only 5 digits matching the \\d+
1104 item, and then with 4, and so on, before ultimately failing. Once-only
1105 subpatterns provide the means for specifying that once a portion of the pattern
1106 has matched, it is not to be re-evaluated in this way, so the matcher would
1107 give up immediately on failing to match "foo" the first time. The notation is
1108 another kind of special parenthesis, starting with (?> as in this example:
1110 (?>\\d+)bar
1112 This kind of parenthesis "locks up" the part of the pattern it contains once
1113 it has matched, and a failure further into the pattern is prevented from
1114 backtracking into it. Backtracking past it to previous items, however, works as
1115 normal.
1117 An alternative description is that a subpattern of this type matches the string
1118 of characters that an identical standalone pattern would match, if anchored at
1119 the current point in the subject string.
1121 Once-only subpatterns are not capturing subpatterns. Simple cases such as the
1122 above example can be though of as a maximizing repeat that must swallow
1123 everything it can. So, while both \\d+ and \\d+? are prepared to adjust the
1124 number of digits they match in order to make the rest of the pattern match,
1125 (?>\\d+) can only match an entire sequence of digits.
1127 This construction can of course contain arbitrarily complicated subpatterns,
1128 and it can be nested.
1132 It is possible to cause the matching process to obey a subpattern
1133 conditionally or to choose between two alternative subpatterns, depending on
1134 the result of an assertion, or whether a previous capturing subpattern matched
1135 or not. The two possible forms of conditional subpattern are
1137 (?(condition)yes-pattern)
1138 (?(condition)yes-pattern|no-pattern)
1140 If the condition is satisfied, the yes-pattern is used; otherwise the
1141 no-pattern (if present) is used. If there are more than two alternatives in the
1142 subpattern, a compile-time error occurs.
1144 There are two kinds of condition. If the text between the parentheses consists
1145 of a sequence of digits, then the condition is satisfied if the capturing
1146 subpattern of that number has previously matched. Consider the following
1147 pattern, which contains non-significant white space to make it more readable
1148 (assume the PCRE_EXTENDED option) and to divide it into three parts for ease
1149 of discussion:
1151 ( \\( )? [^()]+ (?(1) \\) )
1153 The first part matches an optional opening parenthesis, and if that
1154 character is present, sets it as the first captured substring. The second part
1155 matches one or more characters that are not parentheses. The third part is a
1156 conditional subpattern that tests whether the first set of parentheses matched
1157 or not. If they did, that is, if subject started with an opening parenthesis,
1158 the condition is true, and so the yes-pattern is executed and a closing
1159 parenthesis is required. Otherwise, since no-pattern is not present, the
1160 subpattern matches nothing. In other words, this pattern matches a sequence of
1161 non-parentheses, optionally enclosed in parentheses.
1163 If the condition is not a sequence of digits, it must be an assertion. This may
1164 be a positive or negative lookahead or lookbehind assertion. Consider this
1165 pattern, again containing non-significant white space, and with the two
1166 alternatives on the second line:
1168 (?(?=[^a-z]*[a-z])
1169 \\d{2}[a-z]{3}-\\d{2} | \\d{2}-\\d{2}-\\d{2} )
1171 The condition is a positive lookahead assertion that matches an optional
1172 sequence of non-letters followed by a letter. In other words, it tests for the
1173 presence of at least one letter in the subject. If a letter is found, the
1174 subject is matched against the first alternative; otherwise it is matched
1175 against the second. This pattern matches strings in one of the two forms
1176 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1180 The sequence (?# marks the start of a comment which continues up to the next
1181 closing parenthesis. Nested parentheses are not permitted. The characters
1182 that make up a comment play no part in the pattern matching at all.
1184 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1185 character class introduces a comment that continues up to the next newline
1186 character in the pattern.
1190 Certain items that may appear in patterns are more efficient than others. It is
1191 more efficient to use a character class like [aeiou] than a set of alternatives
1192 such as (a|e|i|o|u). In general, the simplest construction that provides the
1193 required behaviour is usually the most efficient. Jeffrey Friedl's book
1194 contains a lot of discussion about optimizing regular expressions for efficient
1195 performance.
1199 Philip Hazel <ph10@cam.ac.uk>
1200 .br
1201 University Computing Service,
1202 .br
1203 New Museums Site,
1204 .br
1205 Cambridge CB2 3QG, England.
1206 .br
1207 Phone: +44 1223 334714
1209 Copyright (c) 1998 University of Cambridge.

  ViewVC Help
Powered by ViewVC 1.1.5