ViewVC logotype

Contents of /code/tags/pcre-2.08/pcre.3.html

Parent Directory Parent Directory | Revision Log Revision Log

Revision 40 - (show annotations)
Sat Feb 24 21:39:15 2007 UTC (14 years, 1 month ago) by nigel
File MIME type: text/html
File size: 75992 byte(s)
Tag code/trunk as code/tags/pcre-2.08.
1 <HTML>
2 <HEAD>
3 <TITLE>pcre specification</TITLE>
4 </HEAD>
5 <body bgcolor="#FFFFFF" text="#00005A">
6 <H1>pcre specification</H1>
7 This HTML document has been generated automatically from the original man page.
8 If there is any nonsense in it, please consult the man page in case the
9 conversion went wrong.
10 <UL>
39 </UL>
41 <P>
42 pcre - Perl-compatible regular expressions.
43 </P>
45 <P>
46 <B>#include &#60;pcre.h&#62;</B>
47 </P>
48 <P>
49 <B>pcre *pcre_compile(const char *<I>pattern</I>, int <I>options</I>,</B>
50 <B>const char **<I>errptr</I>, int *<I>erroffset</I>,</B>
51 <B>const unsigned char *<I>tableptr</I>);</B>
52 </P>
53 <P>
54 <B>pcre_extra *pcre_study(const pcre *<I>code</I>, int <I>options</I>,</B>
55 <B>const char **<I>errptr</I>);</B>
56 </P>
57 <P>
58 <B>int pcre_exec(const pcre *<I>code</I>, const pcre_extra *<I>extra</I>,</B>
59 <B>const char *<I>subject</I>, int <I>length</I>, int <I>startoffset</I>,</B>
60 <B>int <I>options</I>, int *<I>ovector</I>, int <I>ovecsize</I>);</B>
61 </P>
62 <P>
63 <B>int pcre_copy_substring(const char *<I>subject</I>, int *<I>ovector</I>,</B>
64 <B>int <I>stringcount</I>, int <I>stringnumber</I>, char *<I>buffer</I>,</B>
65 <B>int <I>buffersize</I>);</B>
66 </P>
67 <P>
68 <B>int pcre_get_substring(const char *<I>subject</I>, int *<I>ovector</I>,</B>
69 <B>int <I>stringcount</I>, int <I>stringnumber</I>,</B>
70 <B>const char **<I>stringptr</I>);</B>
71 </P>
72 <P>
73 <B>int pcre_get_substring_list(const char *<I>subject</I>,</B>
74 <B>int *<I>ovector</I>, int <I>stringcount</I>, const char ***<I>listptr</I>);</B>
75 </P>
76 <P>
77 <B>const unsigned char *pcre_maketables(void);</B>
78 </P>
79 <P>
80 <B>int pcre_info(const pcre *<I>code</I>, int *<I>optptr</I>, int</B>
81 <B>*<I>firstcharptr</I>);</B>
82 </P>
83 <P>
84 <B>char *pcre_version(void);</B>
85 </P>
86 <P>
87 <B>void *(*pcre_malloc)(size_t);</B>
88 </P>
89 <P>
90 <B>void (*pcre_free)(void *);</B>
91 </P>
93 <P>
94 The PCRE library is a set of functions that implement regular expression
95 pattern matching using the same syntax and semantics as Perl 5, with just a few
96 differences (see below). The current implementation corresponds to Perl 5.005.
97 </P>
98 <P>
99 PCRE has its own native API, which is described in this document. There is also
100 a set of wrapper functions that correspond to the POSIX API. These are
101 described in the <B>pcreposix</B> documentation.
102 </P>
103 <P>
104 The native API function prototypes are defined in the header file <B>pcre.h</B>,
105 and on Unix systems the library itself is called <B>libpcre.a</B>, so can be
106 accessed by adding <B>-lpcre</B> to the command for linking an application which
107 calls it.
108 </P>
109 <P>
110 The functions <B>pcre_compile()</B>, <B>pcre_study()</B>, and <B>pcre_exec()</B>
111 are used for compiling and matching regular expressions, while
112 <B>pcre_copy_substring()</B>, <B>pcre_get_substring()</B>, and
113 <B>pcre_get_substring_list()</B> are convenience functions for extracting
114 captured substrings from a matched subject string. The function
115 <B>pcre_maketables()</B> is used (optionally) to build a set of character tables
116 in the current locale for passing to <B>pcre_compile()</B>.
117 </P>
118 <P>
119 The function <B>pcre_info()</B> is used to find out information about a compiled
120 pattern, while the function <B>pcre_version()</B> returns a pointer to a string
121 containing the version of PCRE and its date of release.
122 </P>
123 <P>
124 The global variables <B>pcre_malloc</B> and <B>pcre_free</B> initially contain
125 the entry points of the standard <B>malloc()</B> and <B>free()</B> functions
126 respectively. PCRE calls the memory management functions via these variables,
127 so a calling program can replace them if it wishes to intercept the calls. This
128 should be done before calling any PCRE functions.
129 </P>
131 <P>
132 The PCRE functions can be used in multi-threading applications, with the
133 proviso that the memory management functions pointed to by <B>pcre_malloc</B>
134 and <B>pcre_free</B> are shared by all threads.
135 </P>
136 <P>
137 The compiled form of a regular expression is not altered during matching, so
138 the same compiled pattern can safely be used by several threads at once.
139 </P>
141 <P>
142 The function <B>pcre_compile()</B> is called to compile a pattern into an
143 internal form. The pattern is a C string terminated by a binary zero, and
144 is passed in the argument <I>pattern</I>. A pointer to a single block of memory
145 that is obtained via <B>pcre_malloc</B> is returned. This contains the
146 compiled code and related data. The <B>pcre</B> type is defined for this for
147 convenience, but in fact <B>pcre</B> is just a typedef for <B>void</B>, since the
148 contents of the block are not externally defined. It is up to the caller to
149 free the memory when it is no longer required.
150 </P>
151 <P>
152 The size of a compiled pattern is roughly proportional to the length of the
153 pattern string, except that each character class (other than those containing
154 just a single character, negated or not) requires 33 bytes, and repeat
155 quantifiers with a minimum greater than one or a bounded maximum cause the
156 relevant portions of the compiled pattern to be replicated.
157 </P>
158 <P>
159 The <I>options</I> argument contains independent bits that affect the
160 compilation. It should be zero if no options are required. Some of the options,
161 in particular, those that are compatible with Perl, can also be set and unset
162 from within the pattern (see the detailed description of regular expressions
163 below). For these options, the contents of the <I>options</I> argument specifies
164 their initial settings at the start of compilation and execution. The
165 PCRE_ANCHORED option can be set at the time of matching as well as at compile
166 time.
167 </P>
168 <P>
169 If <I>errptr</I> is NULL, <B>pcre_compile()</B> returns NULL immediately.
170 Otherwise, if compilation of a pattern fails, <B>pcre_compile()</B> returns
171 NULL, and sets the variable pointed to by <I>errptr</I> to point to a textual
172 error message. The offset from the start of the pattern to the character where
173 the error was discovered is placed in the variable pointed to by
174 <I>erroffset</I>, which must not be NULL. If it is, an immediate error is given.
175 </P>
176 <P>
177 If the final argument, <I>tableptr</I>, is NULL, PCRE uses a default set of
178 character tables which are built when it is compiled, using the default C
179 locale. Otherwise, <I>tableptr</I> must be the result of a call to
180 <B>pcre_maketables()</B>. See the section on locale support below.
181 </P>
182 <P>
183 The following option bits are defined in the header file:
184 </P>
185 <P>
186 <PRE>
188 </PRE>
189 </P>
190 <P>
191 If this bit is set, the pattern is forced to be "anchored", that is, it is
192 constrained to match only at the start of the string which is being searched
193 (the "subject string"). This effect can also be achieved by appropriate
194 constructs in the pattern itself, which is the only way to do it in Perl.
195 </P>
196 <P>
197 <PRE>
199 </PRE>
200 </P>
201 <P>
202 If this bit is set, letters in the pattern match both upper and lower case
203 letters. It is equivalent to Perl's /i option.
204 </P>
205 <P>
206 <PRE>
208 </PRE>
209 </P>
210 <P>
211 If this bit is set, a dollar metacharacter in the pattern matches only at the
212 end of the subject string. Without this option, a dollar also matches
213 immediately before the final character if it is a newline (but not before any
214 other newlines). The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
215 set. There is no equivalent to this option in Perl.
216 </P>
217 <P>
218 <PRE>
220 </PRE>
221 </P>
222 <P>
223 If this bit is set, a dot metacharater in the pattern matches all characters,
224 including newlines. Without it, newlines are excluded. This option is
225 equivalent to Perl's /s option. A negative class such as [^a] always matches a
226 newline character, independent of the setting of this option.
227 </P>
228 <P>
229 <PRE>
231 </PRE>
232 </P>
233 <P>
234 If this bit is set, whitespace data characters in the pattern are totally
235 ignored except when escaped or inside a character class, and characters between
236 an unescaped # outside a character class and the next newline character,
237 inclusive, are also ignored. This is equivalent to Perl's /x option, and makes
238 it possible to include comments inside complicated patterns. Note, however,
239 that this applies only to data characters. Whitespace characters may never
240 appear within special character sequences in a pattern, for example within the
241 sequence (?( which introduces a conditional subpattern.
242 </P>
243 <P>
244 <PRE>
246 </PRE>
247 </P>
248 <P>
249 This option turns on additional functionality of PCRE that is incompatible with
250 Perl. Any backslash in a pattern that is followed by a letter that has no
251 special meaning causes an error, thus reserving these combinations for future
252 expansion. By default, as in Perl, a backslash followed by a letter with no
253 special meaning is treated as a literal. There are at present no other features
254 controlled by this option.
255 </P>
256 <P>
257 <PRE>
259 </PRE>
260 </P>
261 <P>
262 By default, PCRE treats the subject string as consisting of a single "line" of
263 characters (even if it actually contains several newlines). The "start of line"
264 metacharacter (^) matches only at the start of the string, while the "end of
265 line" metacharacter ($) matches only at the end of the string, or before a
266 terminating newline (unless PCRE_DOLLAR_ENDONLY is set). This is the same as
267 Perl.
268 </P>
269 <P>
270 When PCRE_MULTILINE it is set, the "start of line" and "end of line" constructs
271 match immediately following or immediately before any newline in the subject
272 string, respectively, as well as at the very start and end. This is equivalent
273 to Perl's /m option. If there are no "\n" characters in a subject string, or
274 no occurrences of ^ or $ in a pattern, setting PCRE_MULTILINE has no
275 effect.
276 </P>
277 <P>
278 <PRE>
280 </PRE>
281 </P>
282 <P>
283 This option inverts the "greediness" of the quantifiers so that they are not
284 greedy by default, but become greedy if followed by "?". It is not compatible
285 with Perl. It can also be set by a (?U) option setting within the pattern.
286 </P>
288 <P>
289 When a pattern is going to be used several times, it is worth spending more
290 time analyzing it in order to speed up the time taken for matching. The
291 function <B>pcre_study()</B> takes a pointer to a compiled pattern as its first
292 argument, and returns a pointer to a <B>pcre_extra</B> block (another <B>void</B>
293 typedef) containing additional information about the pattern; this can be
294 passed to <B>pcre_exec()</B>. If no additional information is available, NULL
295 is returned.
296 </P>
297 <P>
298 The second argument contains option bits. At present, no options are defined
299 for <B>pcre_study()</B>, and this argument should always be zero.
300 </P>
301 <P>
302 The third argument for <B>pcre_study()</B> is a pointer to an error message. If
303 studying succeeds (even if no data is returned), the variable it points to is
304 set to NULL. Otherwise it points to a textual error message.
305 </P>
306 <P>
307 At present, studying a pattern is useful only for non-anchored patterns that do
308 not have a single fixed starting character. A bitmap of possible starting
309 characters is created.
310 </P>
312 <P>
313 PCRE handles caseless matching, and determines whether characters are letters,
314 digits, or whatever, by reference to a set of tables. The library contains a
315 default set of tables which is created in the default C locale when PCRE is
316 compiled. This is used when the final argument of <B>pcre_compile()</B> is NULL,
317 and is sufficient for many applications.
318 </P>
319 <P>
320 An alternative set of tables can, however, be supplied. Such tables are built
321 by calling the <B>pcre_maketables()</B> function, which has no arguments, in the
322 relevant locale. The result can then be passed to <B>pcre_compile()</B> as often
323 as necessary. For example, to build and use tables that are appropriate for the
324 French locale (where accented characters with codes greater than 128 are
325 treated as letters), the following code could be used:
326 </P>
327 <P>
328 <PRE>
329 setlocale(LC_CTYPE, "fr");
330 tables = pcre_maketables();
331 re = pcre_compile(..., tables);
332 </PRE>
333 </P>
334 <P>
335 The tables are built in memory that is obtained via <B>pcre_malloc</B>. The
336 pointer that is passed to <B>pcre_compile</B> is saved with the compiled
337 pattern, and the same tables are used via this pointer by <B>pcre_study()</B>
338 and <B>pcre_exec()</B>. Thus for any single pattern, compilation, studying and
339 matching all happen in the same locale, but different patterns can be compiled
340 in different locales. It is the caller's responsibility to ensure that the
341 memory containing the tables remains available for as long as it is needed.
342 </P>
344 <P>
345 The <B>pcre_info()</B> function returns information about a compiled pattern.
346 Its yield is the number of capturing subpatterns, or one of the following
347 negative numbers:
348 </P>
349 <P>
350 <PRE>
351 PCRE_ERROR_NULL the argument <I>code</I> was NULL
352 PCRE_ERROR_BADMAGIC the "magic number" was not found
353 </PRE>
354 </P>
355 <P>
356 If the <I>optptr</I> argument is not NULL, a copy of the options with which the
357 pattern was compiled is placed in the integer it points to. These option bits
358 are those specified in the call to <B>pcre_compile()</B>, modified by any
359 top-level option settings within the pattern itself, and with the PCRE_ANCHORED
360 bit set if the form of the pattern implies that it can match only at the start
361 of a subject string.
362 </P>
363 <P>
364 If the pattern is not anchored and the <I>firstcharptr</I> argument is not NULL,
365 it is used to pass back information about the first character of any matched
366 string. If there is a fixed first character, e.g. from a pattern such as
367 (cat|cow|coyote), then it is returned in the integer pointed to by
368 <I>firstcharptr</I>. Otherwise, if either
369 </P>
370 <P>
371 (a) the pattern was compiled with the PCRE_MULTILINE option, and every branch
372 starts with "^", or
373 </P>
374 <P>
375 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not set
376 (if it were set, the pattern would be anchored),
377 </P>
378 <P>
379 then -1 is returned, indicating that the pattern matches only at the
380 start of a subject string or after any "\n" within the string. Otherwise -2 is
381 returned.
382 </P>
384 <P>
385 The function <B>pcre_exec()</B> is called to match a subject string against a
386 pre-compiled pattern, which is passed in the <I>code</I> argument. If the
387 pattern has been studied, the result of the study should be passed in the
388 <I>extra</I> argument. Otherwise this must be NULL.
389 </P>
390 <P>
391 The PCRE_ANCHORED option can be passed in the <I>options</I> argument, whose
392 unused bits must be zero. However, if a pattern was compiled with
393 PCRE_ANCHORED, or turned out to be anchored by virtue of its contents, it
394 cannot be made unachored at matching time.
395 </P>
396 <P>
397 There are also three further options that can be set only at matching time:
398 </P>
399 <P>
400 <PRE>
402 </PRE>
403 </P>
404 <P>
405 The first character of the string is not the beginning of a line, so the
406 circumflex metacharacter should not match before it. Setting this without
407 PCRE_MULTILINE (at compile time) causes circumflex never to match.
408 </P>
409 <P>
410 <PRE>
412 </PRE>
413 </P>
414 <P>
415 The end of the string is not the end of a line, so the dollar metacharacter
416 should not match it nor (except in multiline mode) a newline immediately before
417 it. Setting this without PCRE_MULTILINE (at compile time) causes dollar never
418 to match.
419 </P>
420 <P>
421 <PRE>
423 </PRE>
424 </P>
425 <P>
426 An empty string is not considered to be a valid match if this option is set. If
427 there are alternatives in the pattern, they are tried. If all the alternatives
428 match the empty string, the entire match fails. For example, if the pattern
429 </P>
430 <P>
431 <PRE>
432 a?b?
433 </PRE>
434 </P>
435 <P>
436 is applied to a string not beginning with "a" or "b", it matches the empty
437 string at the start of the subject. With PCRE_NOTEMPTY set, this match is not
438 valid, so PCRE searches further into the string for occurrences of "a" or "b".
439 Perl has no direct equivalent of this option, but it makes a special case of
440 a pattern match of the empty string within its <B>split()</B> function, or when
441 using the /g modifier. Using PCRE_NOTEMPTY it is possible to emulate this
442 behaviour.
443 </P>
444 <P>
445 The subject string is passed as a pointer in <I>subject</I>, a length in
446 <I>length</I>, and a starting offset in <I>startoffset</I>. Unlike the pattern
447 string, it may contain binary zero characters. When the starting offset is
448 zero, the search for a match starts at the beginning of the subject, and this
449 is by far the most common case.
450 </P>
451 <P>
452 A non-zero starting offset is useful when searching for another match in the
453 same subject by calling <B>pcre_exec()</B> again after a previous success.
454 Setting <I>startoffset</I> differs from just passing over a shortened string and
455 setting PCRE_NOTBOL in the case of a pattern that begins with any kind of
456 lookbehind. For example, consider the pattern
457 </P>
458 <P>
459 <PRE>
460 \Biss\B
461 </PRE>
462 </P>
463 <P>
464 which finds occurrences of "iss" in the middle of words. (\B matches only if
465 the current position in the subject is not a word boundary.) When applied to
466 the string "Mississipi" the first call to <B>pcre_exec()</B> finds the first
467 occurrence. If <B>pcre_exec()</B> is called again with just the remainder of the
468 subject, namely "issipi", it does not match, because \B is always false at the
469 start of the subject, which is deemed to be a word boundary. However, if
470 <B>pcre_exec()</B> is passed the entire string again, but with <I>startoffset</I>
471 set to 4, it finds the second occurrence of "iss" because it is able to look
472 behind the starting point to discover that it is preceded by a letter.
473 </P>
474 <P>
475 If a non-zero starting offset is passed when the pattern is anchored, one
476 attempt to match at the given offset is tried. This can only succeed if the
477 pattern does not require the match to be at the start of the subject.
478 </P>
479 <P>
480 In general, a pattern matches a certain portion of the subject, and in
481 addition, further substrings from the subject may be picked out by parts of the
482 pattern. Following the usage in Jeffrey Friedl's book, this is called
483 "capturing" in what follows, and the phrase "capturing subpattern" is used for
484 a fragment of a pattern that picks out a substring. PCRE supports several other
485 kinds of parenthesized subpattern that do not cause substrings to be captured.
486 </P>
487 <P>
488 Captured substrings are returned to the caller via a vector of integer offsets
489 whose address is passed in <I>ovector</I>. The number of elements in the vector
490 is passed in <I>ovecsize</I>. The first two-thirds of the vector is used to pass
491 back captured substrings, each substring using a pair of integers. The
492 remaining third of the vector is used as workspace by <B>pcre_exec()</B> while
493 matching capturing subpatterns, and is not available for passing back
494 information. The length passed in <I>ovecsize</I> should always be a multiple of
495 three. If it is not, it is rounded down.
496 </P>
497 <P>
498 When a match has been successful, information about captured substrings is
499 returned in pairs of integers, starting at the beginning of <I>ovector</I>, and
500 continuing up to two-thirds of its length at the most. The first element of a
501 pair is set to the offset of the first character in a substring, and the second
502 is set to the offset of the first character after the end of a substring. The
503 first pair, <I>ovector[0]</I> and <I>ovector[1]</I>, identify the portion of the
504 subject string matched by the entire pattern. The next pair is used for the
505 first capturing subpattern, and so on. The value returned by <B>pcre_exec()</B>
506 is the number of pairs that have been set. If there are no capturing
507 subpatterns, the return value from a successful match is 1, indicating that
508 just the first pair of offsets has been set.
509 </P>
510 <P>
511 Some convenience functions are provided for extracting the captured substrings
512 as separate strings. These are described in the following section.
513 </P>
514 <P>
515 It is possible for an capturing subpattern number <I>n+1</I> to match some
516 part of the subject when subpattern <I>n</I> has not been used at all. For
517 example, if the string "abc" is matched against the pattern (a|(z))(bc)
518 subpatterns 1 and 3 are matched, but 2 is not. When this happens, both offset
519 values corresponding to the unused subpattern are set to -1.
520 </P>
521 <P>
522 If a capturing subpattern is matched repeatedly, it is the last portion of the
523 string that it matched that gets returned.
524 </P>
525 <P>
526 If the vector is too small to hold all the captured substrings, it is used as
527 far as possible (up to two-thirds of its length), and the function returns a
528 value of zero. In particular, if the substring offsets are not of interest,
529 <B>pcre_exec()</B> may be called with <I>ovector</I> passed as NULL and
530 <I>ovecsize</I> as zero. However, if the pattern contains back references and
531 the <I>ovector</I> isn't big enough to remember the related substrings, PCRE has
532 to get additional memory for use during matching. Thus it is usually advisable
533 to supply an <I>ovector</I>.
534 </P>
535 <P>
536 Note that <B>pcre_info()</B> can be used to find out how many capturing
537 subpatterns there are in a compiled pattern. The smallest size for
538 <I>ovector</I> that will allow for <I>n</I> captured substrings in addition to
539 the offsets of the substring matched by the whole pattern is (<I>n</I>+1)*3.
540 </P>
541 <P>
542 If <B>pcre_exec()</B> fails, it returns a negative number. The following are
543 defined in the header file:
544 </P>
545 <P>
546 <PRE>
548 </PRE>
549 </P>
550 <P>
551 The subject string did not match the pattern.
552 </P>
553 <P>
554 <PRE>
556 </PRE>
557 </P>
558 <P>
559 Either <I>code</I> or <I>subject</I> was passed as NULL, or <I>ovector</I> was
560 NULL and <I>ovecsize</I> was not zero.
561 </P>
562 <P>
563 <PRE>
565 </PRE>
566 </P>
567 <P>
568 An unrecognized bit was set in the <I>options</I> argument.
569 </P>
570 <P>
571 <PRE>
573 </PRE>
574 </P>
575 <P>
576 PCRE stores a 4-byte "magic number" at the start of the compiled code, to catch
577 the case when it is passed a junk pointer. This is the error it gives when the
578 magic number isn't present.
579 </P>
580 <P>
581 <PRE>
583 </PRE>
584 </P>
585 <P>
586 While running the pattern match, an unknown item was encountered in the
587 compiled pattern. This error could be caused by a bug in PCRE or by overwriting
588 of the compiled pattern.
589 </P>
590 <P>
591 <PRE>
593 </PRE>
594 </P>
595 <P>
596 If a pattern contains back references, but the <I>ovector</I> that is passed to
597 <B>pcre_exec()</B> is not big enough to remember the referenced substrings, PCRE
598 gets a block of memory at the start of matching to use for this purpose. If the
599 call via <B>pcre_malloc()</B> fails, this error is given. The memory is freed at
600 the end of matching.
601 </P>
603 <P>
604 Captured substrings can be accessed directly by using the offsets returned by
605 <B>pcre_exec()</B> in <I>ovector</I>. For convenience, the functions
606 <B>pcre_copy_substring()</B>, <B>pcre_get_substring()</B>, and
607 <B>pcre_get_substring_list()</B> are provided for extracting captured substrings
608 as new, separate, zero-terminated strings. A substring that contains a binary
609 zero is correctly extracted and has a further zero added on the end, but the
610 result does not, of course, function as a C string.
611 </P>
612 <P>
613 The first three arguments are the same for all three functions: <I>subject</I>
614 is the subject string which has just been successfully matched, <I>ovector</I>
615 is a pointer to the vector of integer offsets that was passed to
616 <B>pcre_exec()</B>, and <I>stringcount</I> is the number of substrings that
617 were captured by the match, including the substring that matched the entire
618 regular expression. This is the value returned by <B>pcre_exec</B> if it
619 is greater than zero. If <B>pcre_exec()</B> returned zero, indicating that it
620 ran out of space in <I>ovector</I>, then the value passed as
621 <I>stringcount</I> should be the size of the vector divided by three.
622 </P>
623 <P>
624 The functions <B>pcre_copy_substring()</B> and <B>pcre_get_substring()</B>
625 extract a single substring, whose number is given as <I>stringnumber</I>. A
626 value of zero extracts the substring that matched the entire pattern, while
627 higher values extract the captured substrings. For <B>pcre_copy_substring()</B>,
628 the string is placed in <I>buffer</I>, whose length is given by
629 <I>buffersize</I>, while for <B>pcre_get_substring()</B> a new block of store is
630 obtained via <B>pcre_malloc</B>, and its address is returned via
631 <I>stringptr</I>. The yield of the function is the length of the string, not
632 including the terminating zero, or one of
633 </P>
634 <P>
635 <PRE>
637 </PRE>
638 </P>
639 <P>
640 The buffer was too small for <B>pcre_copy_substring()</B>, or the attempt to get
641 memory failed for <B>pcre_get_substring()</B>.
642 </P>
643 <P>
644 <PRE>
646 </PRE>
647 </P>
648 <P>
649 There is no substring whose number is <I>stringnumber</I>.
650 </P>
651 <P>
652 The <B>pcre_get_substring_list()</B> function extracts all available substrings
653 and builds a list of pointers to them. All this is done in a single block of
654 memory which is obtained via <B>pcre_malloc</B>. The address of the memory block
655 is returned via <I>listptr</I>, which is also the start of the list of string
656 pointers. The end of the list is marked by a NULL pointer. The yield of the
657 function is zero if all went well, or
658 </P>
659 <P>
660 <PRE>
662 </PRE>
663 </P>
664 <P>
665 if the attempt to get the memory block failed.
666 </P>
667 <P>
668 When any of these functions encounter a substring that is unset, which can
669 happen when capturing subpattern number <I>n+1</I> matches some part of the
670 subject, but subpattern <I>n</I> has not been used at all, they return an empty
671 string. This can be distinguished from a genuine zero-length substring by
672 inspecting the appropriate offset in <I>ovector</I>, which is negative for unset
673 substrings.
674 </P>
676 <P>
677 There are some size limitations in PCRE but it is hoped that they will never in
678 practice be relevant.
679 The maximum length of a compiled pattern is 65539 (sic) bytes.
680 All values in repeating quantifiers must be less than 65536.
681 The maximum number of capturing subpatterns is 99.
682 The maximum number of all parenthesized subpatterns, including capturing
683 subpatterns, assertions, and other types of subpattern, is 200.
684 </P>
685 <P>
686 The maximum length of a subject string is the largest positive number that an
687 integer variable can hold. However, PCRE uses recursion to handle subpatterns
688 and indefinite repetition. This means that the available stack space may limit
689 the size of a subject string that can be processed by certain patterns.
690 </P>
692 <P>
693 The differences described here are with respect to Perl 5.005.
694 </P>
695 <P>
696 1. By default, a whitespace character is any character that the C library
697 function <B>isspace()</B> recognizes, though it is possible to compile PCRE with
698 alternative character type tables. Normally <B>isspace()</B> matches space,
699 formfeed, newline, carriage return, horizontal tab, and vertical tab. Perl 5
700 no longer includes vertical tab in its set of whitespace characters. The \v
701 escape that was in the Perl documentation for a long time was never in fact
702 recognized. However, the character itself was treated as whitespace at least
703 up to 5.002. In 5.004 and 5.005 it does not match \s.
704 </P>
705 <P>
706 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl permits
707 them, but they do not mean what you might think. For example, (?!a){3} does
708 not assert that the next three characters are not "a". It just asserts that the
709 next character is not "a" three times.
710 </P>
711 <P>
712 3. Capturing subpatterns that occur inside negative lookahead assertions are
713 counted, but their entries in the offsets vector are never set. Perl sets its
714 numerical variables from any such patterns that are matched before the
715 assertion fails to match something (thereby succeeding), but only if the
716 negative lookahead assertion contains just one branch.
717 </P>
718 <P>
719 4. Though binary zero characters are supported in the subject string, they are
720 not allowed in a pattern string because it is passed as a normal C string,
721 terminated by zero. The escape sequence "\0" can be used in the pattern to
722 represent a binary zero.
723 </P>
724 <P>
725 5. The following Perl escape sequences are not supported: \l, \u, \L, \U,
726 \E, \Q. In fact these are implemented by Perl's general string-handling and
727 are not part of its pattern matching engine.
728 </P>
729 <P>
730 6. The Perl \G assertion is not supported as it is not relevant to single
731 pattern matches.
732 </P>
733 <P>
734 7. Fairly obviously, PCRE does not support the (?{code}) construction.
735 </P>
736 <P>
737 8. There are at the time of writing some oddities in Perl 5.005_02 concerned
738 with the settings of captured strings when part of a pattern is repeated. For
739 example, matching "aba" against the pattern /^(a(b)?)+$/ sets $2 to the value
740 "b", but matching "aabbaa" against /^(aa(bb)?)+$/ leaves $2 unset. However, if
741 the pattern is changed to /^(aa(b(b))?)+$/ then $2 (and $3) get set.
742 </P>
743 <P>
744 In Perl 5.004 $2 is set in both cases, and that is also true of PCRE. If in the
745 future Perl changes to a consistent state that is different, PCRE may change to
746 follow.
747 </P>
748 <P>
749 9. Another as yet unresolved discrepancy is that in Perl 5.005_02 the pattern
750 /^(a)?(?(1)a|b)+$/ matches the string "a", whereas in PCRE it does not.
751 However, in both Perl and PCRE /^(a)?a/ matched against "a" leaves $1 unset.
752 </P>
753 <P>
754 10. PCRE provides some extensions to the Perl regular expression facilities:
755 </P>
756 <P>
757 (a) Although lookbehind assertions must match fixed length strings, each
758 alternative branch of a lookbehind assertion can match a different length of
759 string. Perl 5.005 requires them all to have the same length.
760 </P>
761 <P>
762 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $ meta-
763 character matches only at the very end of the string.
764 </P>
765 <P>
766 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no special
767 meaning is faulted.
768 </P>
769 <P>
770 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quantifiers is
771 inverted, that is, by default they are not greedy, but if followed by a
772 question mark they are.
773 </P>
774 <P>
775 (e) PCRE_ANCHORED can be used to force a pattern to be tried only at the start
776 of the subject.
777 </P>
778 <P>
779 (f) The PCRE_NOTBOL, PCRE_NOTEOL, and PCRE_NOTEMPTY options for
780 <B>pcre_exec()</B> have no Perl equivalents.
781 </P>
783 <P>
784 The syntax and semantics of the regular expressions supported by PCRE are
785 described below. Regular expressions are also described in the Perl
786 documentation and in a number of other books, some of which have copious
787 examples. Jeffrey Friedl's "Mastering Regular Expressions", published by
788 O'Reilly (ISBN 1-56592-257-3), covers them in great detail. The description
789 here is intended as reference documentation.
790 </P>
791 <P>
792 A regular expression is a pattern that is matched against a subject string from
793 left to right. Most characters stand for themselves in a pattern, and match the
794 corresponding characters in the subject. As a trivial example, the pattern
795 </P>
796 <P>
797 <PRE>
798 The quick brown fox
799 </PRE>
800 </P>
801 <P>
802 matches a portion of a subject string that is identical to itself. The power of
803 regular expressions comes from the ability to include alternatives and
804 repetitions in the pattern. These are encoded in the pattern by the use of
805 <I>meta-characters</I>, which do not stand for themselves but instead are
806 interpreted in some special way.
807 </P>
808 <P>
809 There are two different sets of meta-characters: those that are recognized
810 anywhere in the pattern except within square brackets, and those that are
811 recognized in square brackets. Outside square brackets, the meta-characters are
812 as follows:
813 </P>
814 <P>
815 <PRE>
816 \ general escape character with several uses
817 ^ assert start of subject (or line, in multiline mode)
818 $ assert end of subject (or line, in multiline mode)
819 . match any character except newline (by default)
820 [ start character class definition
821 | start of alternative branch
822 ( start subpattern
823 ) end subpattern
824 ? extends the meaning of (
825 also 0 or 1 quantifier
826 also quantifier minimizer
827 * 0 or more quantifier
828 + 1 or more quantifier
829 { start min/max quantifier
830 </PRE>
831 </P>
832 <P>
833 Part of a pattern that is in square brackets is called a "character class". In
834 a character class the only meta-characters are:
835 </P>
836 <P>
837 <PRE>
838 \ general escape character
839 ^ negate the class, but only if the first character
840 - indicates character range
841 ] terminates the character class
842 </PRE>
843 </P>
844 <P>
845 The following sections describe the use of each of the meta-characters.
846 </P>
848 <P>
849 The backslash character has several uses. Firstly, if it is followed by a
850 non-alphameric character, it takes away any special meaning that character may
851 have. This use of backslash as an escape character applies both inside and
852 outside character classes.
853 </P>
854 <P>
855 For example, if you want to match a "*" character, you write "\*" in the
856 pattern. This applies whether or not the following character would otherwise be
857 interpreted as a meta-character, so it is always safe to precede a
858 non-alphameric with "\" to specify that it stands for itself. In particular,
859 if you want to match a backslash, you write "\\".
860 </P>
861 <P>
862 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
863 pattern (other than in a character class) and characters between a "#" outside
864 a character class and the next newline character are ignored. An escaping
865 backslash can be used to include a whitespace or "#" character as part of the
866 pattern.
867 </P>
868 <P>
869 A second use of backslash provides a way of encoding non-printing characters
870 in patterns in a visible manner. There is no restriction on the appearance of
871 non-printing characters, apart from the binary zero that terminates a pattern,
872 but when a pattern is being prepared by text editing, it is usually easier to
873 use one of the following escape sequences than the binary character it
874 represents:
875 </P>
876 <P>
877 <PRE>
878 \a alarm, that is, the BEL character (hex 07)
879 \cx "control-x", where x is any character
880 \e escape (hex 1B)
881 \f formfeed (hex 0C)
882 \n newline (hex 0A)
883 \r carriage return (hex 0D)
884 \t tab (hex 09)
885 \xhh character with hex code hh
886 \ddd character with octal code ddd, or backreference
887 </PRE>
888 </P>
889 <P>
890 The precise effect of "\cx" is as follows: if "x" is a lower case letter, it
891 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
892 Thus "\cz" becomes hex 1A, but "\c{" becomes hex 3B, while "\c;" becomes hex
893 7B.
894 </P>
895 <P>
896 After "\x", up to two hexadecimal digits are read (letters can be in upper or
897 lower case).
898 </P>
899 <P>
900 After "\0" up to two further octal digits are read. In both cases, if there
901 are fewer than two digits, just those that are present are used. Thus the
902 sequence "\0\x\07" specifies two binary zeros followed by a BEL character.
903 Make sure you supply two digits after the initial zero if the character that
904 follows is itself an octal digit.
905 </P>
906 <P>
907 The handling of a backslash followed by a digit other than 0 is complicated.
908 Outside a character class, PCRE reads it and any following digits as a decimal
909 number. If the number is less than 10, or if there have been at least that many
910 previous capturing left parentheses in the expression, the entire sequence is
911 taken as a <I>back reference</I>. A description of how this works is given
912 later, following the discussion of parenthesized subpatterns.
913 </P>
914 <P>
915 Inside a character class, or if the decimal number is greater than 9 and there
916 have not been that many capturing subpatterns, PCRE re-reads up to three octal
917 digits following the backslash, and generates a single byte from the least
918 significant 8 bits of the value. Any subsequent digits stand for themselves.
919 For example:
920 </P>
921 <P>
922 <PRE>
923 \040 is another way of writing a space
924 \40 is the same, provided there are fewer than 40
925 previous capturing subpatterns
926 \7 is always a back reference
927 \11 might be a back reference, or another way of
928 writing a tab
929 \011 is always a tab
930 \0113 is a tab followed by the character "3"
931 \113 is the character with octal code 113 (since there
932 can be no more than 99 back references)
933 \377 is a byte consisting entirely of 1 bits
934 \81 is either a back reference, or a binary zero
935 followed by the two characters "8" and "1"
936 </PRE>
937 </P>
938 <P>
939 Note that octal values of 100 or greater must not be introduced by a leading
940 zero, because no more than three octal digits are ever read.
941 </P>
942 <P>
943 All the sequences that define a single byte value can be used both inside and
944 outside character classes. In addition, inside a character class, the sequence
945 "\b" is interpreted as the backspace character (hex 08). Outside a character
946 class it has a different meaning (see below).
947 </P>
948 <P>
949 The third use of backslash is for specifying generic character types:
950 </P>
951 <P>
952 <PRE>
953 \d any decimal digit
954 \D any character that is not a decimal digit
955 \s any whitespace character
956 \S any character that is not a whitespace character
957 \w any "word" character
958 \W any "non-word" character
959 </PRE>
960 </P>
961 <P>
962 Each pair of escape sequences partitions the complete set of characters into
963 two disjoint sets. Any given character matches one, and only one, of each pair.
964 </P>
965 <P>
966 A "word" character is any letter or digit or the underscore character, that is,
967 any character which can be part of a Perl "word". The definition of letters and
968 digits is controlled by PCRE's character tables, and may vary if locale-
969 specific matching is taking place (see "Locale support" above). For example, in
970 the "fr" (French) locale, some character codes greater than 128 are used for
971 accented letters, and these are matched by \w.
972 </P>
973 <P>
974 These character type sequences can appear both inside and outside character
975 classes. They each match one character of the appropriate type. If the current
976 matching point is at the end of the subject string, all of them fail, since
977 there is no character to match.
978 </P>
979 <P>
980 The fourth use of backslash is for certain simple assertions. An assertion
981 specifies a condition that has to be met at a particular point in a match,
982 without consuming any characters from the subject string. The use of
983 subpatterns for more complicated assertions is described below. The backslashed
984 assertions are
985 </P>
986 <P>
987 <PRE>
988 \b word boundary
989 \B not a word boundary
990 \A start of subject (independent of multiline mode)
991 \Z end of subject or newline at end (independent of multiline mode)
992 \z end of subject (independent of multiline mode)
993 </PRE>
994 </P>
995 <P>
996 These assertions may not appear in character classes (but note that "\b" has a
997 different meaning, namely the backspace character, inside a character class).
998 </P>
999 <P>
1000 A word boundary is a position in the subject string where the current character
1001 and the previous character do not both match \w or \W (i.e. one matches
1002 \w and the other matches \W), or the start or end of the string if the
1003 first or last character matches \w, respectively.
1004 </P>
1005 <P>
1006 The \A, \Z, and \z assertions differ from the traditional circumflex and
1007 dollar (described below) in that they only ever match at the very start and end
1008 of the subject string, whatever options are set. They are not affected by the
1009 PCRE_NOTBOL or PCRE_NOTEOL options. If the <I>startoffset</I> argument of
1010 <B>pcre_exec()</B> is non-zero, \A can never match. The difference between \Z
1011 and \z is that \Z matches before a newline that is the last character of the
1012 string as well as at the end of the string, whereas \z matches only at the
1013 end.
1014 </P>
1016 <P>
1017 Outside a character class, in the default matching mode, the circumflex
1018 character is an assertion which is true only if the current matching point is
1019 at the start of the subject string. If the <I>startoffset</I> argument of
1020 <B>pcre_exec()</B> is non-zero, circumflex can never match. Inside a character
1021 class, circumflex has an entirely different meaning (see below).
1022 </P>
1023 <P>
1024 Circumflex need not be the first character of the pattern if a number of
1025 alternatives are involved, but it should be the first thing in each alternative
1026 in which it appears if the pattern is ever to match that branch. If all
1027 possible alternatives start with a circumflex, that is, if the pattern is
1028 constrained to match only at the start of the subject, it is said to be an
1029 "anchored" pattern. (There are also other constructs that can cause a pattern
1030 to be anchored.)
1031 </P>
1032 <P>
1033 A dollar character is an assertion which is true only if the current matching
1034 point is at the end of the subject string, or immediately before a newline
1035 character that is the last character in the string (by default). Dollar need
1036 not be the last character of the pattern if a number of alternatives are
1037 involved, but it should be the last item in any branch in which it appears.
1038 Dollar has no special meaning in a character class.
1039 </P>
1040 <P>
1041 The meaning of dollar can be changed so that it matches only at the very end of
1042 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile or matching
1043 time. This does not affect the \Z assertion.
1044 </P>
1045 <P>
1046 The meanings of the circumflex and dollar characters are changed if the
1047 PCRE_MULTILINE option is set. When this is the case, they match immediately
1048 after and immediately before an internal "\n" character, respectively, in
1049 addition to matching at the start and end of the subject string. For example,
1050 the pattern /^abc$/ matches the subject string "def\nabc" in multiline mode,
1051 but not otherwise. Consequently, patterns that are anchored in single line mode
1052 because all branches start with "^" are not anchored in multiline mode, and a
1053 match for circumflex is possible when the <I>startoffset</I> argument of
1054 <B>pcre_exec()</B> is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
1055 PCRE_MULTILINE is set.
1056 </P>
1057 <P>
1058 Note that the sequences \A, \Z, and \z can be used to match the start and
1059 end of the subject in both modes, and if all branches of a pattern start with
1060 \A is it always anchored, whether PCRE_MULTILINE is set or not.
1061 </P>
1063 <P>
1064 Outside a character class, a dot in the pattern matches any one character in
1065 the subject, including a non-printing character, but not (by default) newline.
1066 If the PCRE_DOTALL option is set, then dots match newlines as well. The
1067 handling of dot is entirely independent of the handling of circumflex and
1068 dollar, the only relationship being that they both involve newline characters.
1069 Dot has no special meaning in a character class.
1070 </P>
1072 <P>
1073 An opening square bracket introduces a character class, terminated by a closing
1074 square bracket. A closing square bracket on its own is not special. If a
1075 closing square bracket is required as a member of the class, it should be the
1076 first data character in the class (after an initial circumflex, if present) or
1077 escaped with a backslash.
1078 </P>
1079 <P>
1080 A character class matches a single character in the subject; the character must
1081 be in the set of characters defined by the class, unless the first character in
1082 the class is a circumflex, in which case the subject character must not be in
1083 the set defined by the class. If a circumflex is actually required as a member
1084 of the class, ensure it is not the first character, or escape it with a
1085 backslash.
1086 </P>
1087 <P>
1088 For example, the character class [aeiou] matches any lower case vowel, while
1089 [^aeiou] matches any character that is not a lower case vowel. Note that a
1090 circumflex is just a convenient notation for specifying the characters which
1091 are in the class by enumerating those that are not. It is not an assertion: it
1092 still consumes a character from the subject string, and fails if the current
1093 pointer is at the end of the string.
1094 </P>
1095 <P>
1096 When caseless matching is set, any letters in a class represent both their
1097 upper case and lower case versions, so for example, a caseless [aeiou] matches
1098 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1099 caseful version would.
1100 </P>
1101 <P>
1102 The newline character is never treated in any special way in character classes,
1103 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
1104 such as [^a] will always match a newline.
1105 </P>
1106 <P>
1107 The minus (hyphen) character can be used to specify a range of characters in a
1108 character class. For example, [d-m] matches any letter between d and m,
1109 inclusive. If a minus character is required in a class, it must be escaped with
1110 a backslash or appear in a position where it cannot be interpreted as
1111 indicating a range, typically as the first or last character in the class.
1112 </P>
1113 <P>
1114 It is not possible to have the literal character "]" as the end character of a
1115 range. A pattern such as [W-]46] is interpreted as a class of two characters
1116 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1117 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1118 the end of range, so [W-\]46] is interpreted as a single class containing a
1119 range followed by two separate characters. The octal or hexadecimal
1120 representation of "]" can also be used to end a range.
1121 </P>
1122 <P>
1123 Ranges operate in ASCII collating sequence. They can also be used for
1124 characters specified numerically, for example [\000-\037]. If a range that
1125 includes letters is used when caseless matching is set, it matches the letters
1126 in either case. For example, [W-c] is equivalent to [][\^_`wxyzabc], matched
1127 caselessly, and if character tables for the "fr" locale are in use,
1128 [\xc8-\xcb] matches accented E characters in both cases.
1129 </P>
1130 <P>
1131 The character types \d, \D, \s, \S, \w, and \W may also appear in a
1132 character class, and add the characters that they match to the class. For
1133 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
1134 conveniently be used with the upper case character types to specify a more
1135 restricted set of characters than the matching lower case type. For example,
1136 the class [^\W_] matches any letter or digit, but not underscore.
1137 </P>
1138 <P>
1139 All non-alphameric characters other than \, -, ^ (at the start) and the
1140 terminating ] are non-special in character classes, but it does no harm if they
1141 are escaped.
1142 </P>
1144 <P>
1145 Vertical bar characters are used to separate alternative patterns. For example,
1146 the pattern
1147 </P>
1148 <P>
1149 <PRE>
1150 gilbert|sullivan
1151 </PRE>
1152 </P>
1153 <P>
1154 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1155 and an empty alternative is permitted (matching the empty string).
1156 The matching process tries each alternative in turn, from left to right,
1157 and the first one that succeeds is used. If the alternatives are within a
1158 subpattern (defined below), "succeeds" means matching the rest of the main
1159 pattern as well as the alternative in the subpattern.
1160 </P>
1162 <P>
1164 can be changed from within the pattern by a sequence of Perl option letters
1165 enclosed between "(?" and ")". The option letters are
1166 </P>
1167 <P>
1168 <PRE>
1169 i for PCRE_CASELESS
1171 s for PCRE_DOTALL
1172 x for PCRE_EXTENDED
1173 </PRE>
1174 </P>
1175 <P>
1176 For example, (?im) sets caseless, multiline matching. It is also possible to
1177 unset these options by preceding the letter with a hyphen, and a combined
1178 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1179 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1180 permitted. If a letter appears both before and after the hyphen, the option is
1181 unset.
1182 </P>
1183 <P>
1184 The scope of these option changes depends on where in the pattern the setting
1185 occurs. For settings that are outside any subpattern (defined below), the
1186 effect is the same as if the options were set or unset at the start of
1187 matching. The following patterns all behave in exactly the same way:
1188 </P>
1189 <P>
1190 <PRE>
1191 (?i)abc
1192 a(?i)bc
1193 ab(?i)c
1194 abc(?i)
1195 </PRE>
1196 </P>
1197 <P>
1198 which in turn is the same as compiling the pattern abc with PCRE_CASELESS set.
1199 In other words, such "top level" settings apply to the whole pattern (unless
1200 there are other changes inside subpatterns). If there is more than one setting
1201 of the same option at top level, the rightmost setting is used.
1202 </P>
1203 <P>
1204 If an option change occurs inside a subpattern, the effect is different. This
1205 is a change of behaviour in Perl 5.005. An option change inside a subpattern
1206 affects only that part of the subpattern that follows it, so
1207 </P>
1208 <P>
1209 <PRE>
1210 (a(?i)b)c
1211 </PRE>
1212 </P>
1213 <P>
1214 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1215 By this means, options can be made to have different settings in different
1216 parts of the pattern. Any changes made in one alternative do carry on
1217 into subsequent branches within the same subpattern. For example,
1218 </P>
1219 <P>
1220 <PRE>
1221 (a(?i)b|c)
1222 </PRE>
1223 </P>
1224 <P>
1225 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1226 branch is abandoned before the option setting. This is because the effects of
1227 option settings happen at compile time. There would be some very weird
1228 behaviour otherwise.
1229 </P>
1230 <P>
1231 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
1232 same way as the Perl-compatible options by using the characters U and X
1233 respectively. The (?X) flag setting is special in that it must always occur
1234 earlier in the pattern than any of the additional features it turns on, even
1235 when it is at top level. It is best put at the start.
1236 </P>
1238 <P>
1239 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1240 Marking part of a pattern as a subpattern does two things:
1241 </P>
1242 <P>
1243 1. It localizes a set of alternatives. For example, the pattern
1244 </P>
1245 <P>
1246 <PRE>
1247 cat(aract|erpillar|)
1248 </PRE>
1249 </P>
1250 <P>
1251 matches one of the words "cat", "cataract", or "caterpillar". Without the
1252 parentheses, it would match "cataract", "erpillar" or the empty string.
1253 </P>
1254 <P>
1255 2. It sets up the subpattern as a capturing subpattern (as defined above).
1256 When the whole pattern matches, that portion of the subject string that matched
1257 the subpattern is passed back to the caller via the <I>ovector</I> argument of
1258 <B>pcre_exec()</B>. Opening parentheses are counted from left to right (starting
1259 from 1) to obtain the numbers of the capturing subpatterns.
1260 </P>
1261 <P>
1262 For example, if the string "the red king" is matched against the pattern
1263 </P>
1264 <P>
1265 <PRE>
1266 the ((red|white) (king|queen))
1267 </PRE>
1268 </P>
1269 <P>
1270 the captured substrings are "red king", "red", and "king", and are numbered 1,
1271 2, and 3.
1272 </P>
1273 <P>
1274 The fact that plain parentheses fulfil two functions is not always helpful.
1275 There are often times when a grouping subpattern is required without a
1276 capturing requirement. If an opening parenthesis is followed by "?:", the
1277 subpattern does not do any capturing, and is not counted when computing the
1278 number of any subsequent capturing subpatterns. For example, if the string "the
1279 white queen" is matched against the pattern
1280 </P>
1281 <P>
1282 <PRE>
1283 the ((?:red|white) (king|queen))
1284 </PRE>
1285 </P>
1286 <P>
1287 the captured substrings are "white queen" and "queen", and are numbered 1 and
1288 2. The maximum number of captured substrings is 99, and the maximum number of
1289 all subpatterns, both capturing and non-capturing, is 200.
1290 </P>
1291 <P>
1292 As a convenient shorthand, if any option settings are required at the start of
1293 a non-capturing subpattern, the option letters may appear between the "?" and
1294 the ":". Thus the two patterns
1295 </P>
1296 <P>
1297 <PRE>
1298 (?i:saturday|sunday)
1299 (?:(?i)saturday|sunday)
1300 </PRE>
1301 </P>
1302 <P>
1303 match exactly the same set of strings. Because alternative branches are tried
1304 from left to right, and options are not reset until the end of the subpattern
1305 is reached, an option setting in one branch does affect subsequent branches, so
1306 the above patterns match "SUNDAY" as well as "Saturday".
1307 </P>
1309 <P>
1310 Repetition is specified by quantifiers, which can follow any of the following
1311 items:
1312 </P>
1313 <P>
1314 <PRE>
1315 a single character, possibly escaped
1316 the . metacharacter
1317 a character class
1318 a back reference (see next section)
1319 a parenthesized subpattern (unless it is an assertion - see below)
1320 </PRE>
1321 </P>
1322 <P>
1323 The general repetition quantifier specifies a minimum and maximum number of
1324 permitted matches, by giving the two numbers in curly brackets (braces),
1325 separated by a comma. The numbers must be less than 65536, and the first must
1326 be less than or equal to the second. For example:
1327 </P>
1328 <P>
1329 <PRE>
1330 z{2,4}
1331 </PRE>
1332 </P>
1333 <P>
1334 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1335 character. If the second number is omitted, but the comma is present, there is
1336 no upper limit; if the second number and the comma are both omitted, the
1337 quantifier specifies an exact number of required matches. Thus
1338 </P>
1339 <P>
1340 <PRE>
1341 [aeiou]{3,}
1342 </PRE>
1343 </P>
1344 <P>
1345 matches at least 3 successive vowels, but may match many more, while
1346 </P>
1347 <P>
1348 <PRE>
1349 \d{8}
1350 </PRE>
1351 </P>
1352 <P>
1353 matches exactly 8 digits. An opening curly bracket that appears in a position
1354 where a quantifier is not allowed, or one that does not match the syntax of a
1355 quantifier, is taken as a literal character. For example, {,6} is not a
1356 quantifier, but a literal string of four characters.
1357 </P>
1358 <P>
1359 The quantifier {0} is permitted, causing the expression to behave as if the
1360 previous item and the quantifier were not present.
1361 </P>
1362 <P>
1363 For convenience (and historical compatibility) the three most common
1364 quantifiers have single-character abbreviations:
1365 </P>
1366 <P>
1367 <PRE>
1368 * is equivalent to {0,}
1369 + is equivalent to {1,}
1370 ? is equivalent to {0,1}
1371 </PRE>
1372 </P>
1373 <P>
1374 It is possible to construct infinite loops by following a subpattern that can
1375 match no characters with a quantifier that has no upper limit, for example:
1376 </P>
1377 <P>
1378 <PRE>
1379 (a?)*
1380 </PRE>
1381 </P>
1382 <P>
1383 Earlier versions of Perl and PCRE used to give an error at compile time for
1384 such patterns. However, because there are cases where this can be useful, such
1385 patterns are now accepted, but if any repetition of the subpattern does in fact
1386 match no characters, the loop is forcibly broken.
1387 </P>
1388 <P>
1389 By default, the quantifiers are "greedy", that is, they match as much as
1390 possible (up to the maximum number of permitted times), without causing the
1391 rest of the pattern to fail. The classic example of where this gives problems
1392 is in trying to match comments in C programs. These appear between the
1393 sequences /* and */ and within the sequence, individual * and / characters may
1394 appear. An attempt to match C comments by applying the pattern
1395 </P>
1396 <P>
1397 <PRE>
1398 /\*.*\*/
1399 </PRE>
1400 </P>
1401 <P>
1402 to the string
1403 </P>
1404 <P>
1405 <PRE>
1406 /* first command */ not comment /* second comment */
1407 </PRE>
1408 </P>
1409 <P>
1410 fails, because it matches the entire string due to the greediness of the .*
1411 item.
1412 </P>
1413 <P>
1414 However, if a quantifier is followed by a question mark, then it ceases to be
1415 greedy, and instead matches the minimum number of times possible, so the
1416 pattern
1417 </P>
1418 <P>
1419 <PRE>
1420 /\*.*?\*/
1421 </PRE>
1422 </P>
1423 <P>
1424 does the right thing with the C comments. The meaning of the various
1425 quantifiers is not otherwise changed, just the preferred number of matches.
1426 Do not confuse this use of question mark with its use as a quantifier in its
1427 own right. Because it has two uses, it can sometimes appear doubled, as in
1428 </P>
1429 <P>
1430 <PRE>
1431 \d??\d
1432 </PRE>
1433 </P>
1434 <P>
1435 which matches one digit by preference, but can match two if that is the only
1436 way the rest of the pattern matches.
1437 </P>
1438 <P>
1439 If the PCRE_UNGREEDY option is set (an option which is not available in Perl)
1440 then the quantifiers are not greedy by default, but individual ones can be made
1441 greedy by following them with a question mark. In other words, it inverts the
1442 default behaviour.
1443 </P>
1444 <P>
1445 When a parenthesized subpattern is quantified with a minimum repeat count that
1446 is greater than 1 or with a limited maximum, more store is required for the
1447 compiled pattern, in proportion to the size of the minimum or maximum.
1448 </P>
1449 <P>
1450 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1451 to Perl's /s) is set, thus allowing the . to match newlines, then the pattern
1452 is implicitly anchored, because whatever follows will be tried against every
1453 character position in the subject string, so there is no point in retrying the
1454 overall match at any position after the first. PCRE treats such a pattern as
1455 though it were preceded by \A. In cases where it is known that the subject
1456 string contains no newlines, it is worth setting PCRE_DOTALL when the pattern
1457 begins with .* in order to obtain this optimization, or alternatively using ^
1458 to indicate anchoring explicitly.
1459 </P>
1460 <P>
1461 When a capturing subpattern is repeated, the value captured is the substring
1462 that matched the final iteration. For example, after
1463 </P>
1464 <P>
1465 <PRE>
1466 (tweedle[dume]{3}\s*)+
1467 </PRE>
1468 </P>
1469 <P>
1470 has matched "tweedledum tweedledee" the value of the captured substring is
1471 "tweedledee". However, if there are nested capturing subpatterns, the
1472 corresponding captured values may have been set in previous iterations. For
1473 example, after
1474 </P>
1475 <P>
1476 <PRE>
1477 /(a|(b))+/
1478 </PRE>
1479 </P>
1480 <P>
1481 matches "aba" the value of the second captured substring is "b".
1482 </P>
1484 <P>
1485 Outside a character class, a backslash followed by a digit greater than 0 (and
1486 possibly further digits) is a back reference to a capturing subpattern earlier
1487 (i.e. to its left) in the pattern, provided there have been that many previous
1488 capturing left parentheses.
1489 </P>
1490 <P>
1491 However, if the decimal number following the backslash is less than 10, it is
1492 always taken as a back reference, and causes an error only if there are not
1493 that many capturing left parentheses in the entire pattern. In other words, the
1494 parentheses that are referenced need not be to the left of the reference for
1495 numbers less than 10. See the section entitled "Backslash" above for further
1496 details of the handling of digits following a backslash.
1497 </P>
1498 <P>
1499 A back reference matches whatever actually matched the capturing subpattern in
1500 the current subject string, rather than anything matching the subpattern
1501 itself. So the pattern
1502 </P>
1503 <P>
1504 <PRE>
1505 (sens|respons)e and \1ibility
1506 </PRE>
1507 </P>
1508 <P>
1509 matches "sense and sensibility" and "response and responsibility", but not
1510 "sense and responsibility". If caseful matching is in force at the time of the
1511 back reference, then the case of letters is relevant. For example,
1512 </P>
1513 <P>
1514 <PRE>
1515 ((?i)rah)\s+\1
1516 </PRE>
1517 </P>
1518 <P>
1519 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1520 capturing subpattern is matched caselessly.
1521 </P>
1522 <P>
1523 There may be more than one back reference to the same subpattern. If a
1524 subpattern has not actually been used in a particular match, then any back
1525 references to it always fail. For example, the pattern
1526 </P>
1527 <P>
1528 <PRE>
1529 (a|(bc))\2
1530 </PRE>
1531 </P>
1532 <P>
1533 always fails if it starts to match "a" rather than "bc". Because there may be
1534 up to 99 back references, all digits following the backslash are taken
1535 as part of a potential back reference number. If the pattern continues with a
1536 digit character, then some delimiter must be used to terminate the back
1537 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1538 Otherwise an empty comment can be used.
1539 </P>
1540 <P>
1541 A back reference that occurs inside the parentheses to which it refers fails
1542 when the subpattern is first used, so, for example, (a\1) never matches.
1543 However, such references can be useful inside repeated subpatterns. For
1544 example, the pattern
1545 </P>
1546 <P>
1547 <PRE>
1548 (a|b\1)+
1549 </PRE>
1550 </P>
1551 <P>
1552 matches any number of "a"s and also "aba", "ababaa" etc. At each iteration of
1553 the subpattern, the back reference matches the character string corresponding
1554 to the previous iteration. In order for this to work, the pattern must be such
1555 that the first iteration does not need to match the back reference. This can be
1556 done using alternation, as in the example above, or by a quantifier with a
1557 minimum of zero.
1558 </P>
1560 <P>
1561 An assertion is a test on the characters following or preceding the current
1562 matching point that does not actually consume any characters. The simple
1563 assertions coded as \b, \B, \A, \Z, \z, ^ and $ are described above. More
1564 complicated assertions are coded as subpatterns. There are two kinds: those
1565 that look ahead of the current position in the subject string, and those that
1566 look behind it.
1567 </P>
1568 <P>
1569 An assertion subpattern is matched in the normal way, except that it does not
1570 cause the current matching position to be changed. Lookahead assertions start
1571 with (?= for positive assertions and (?! for negative assertions. For example,
1572 </P>
1573 <P>
1574 <PRE>
1575 \w+(?=;)
1576 </PRE>
1577 </P>
1578 <P>
1579 matches a word followed by a semicolon, but does not include the semicolon in
1580 the match, and
1581 </P>
1582 <P>
1583 <PRE>
1584 foo(?!bar)
1585 </PRE>
1586 </P>
1587 <P>
1588 matches any occurrence of "foo" that is not followed by "bar". Note that the
1589 apparently similar pattern
1590 </P>
1591 <P>
1592 <PRE>
1593 (?!foo)bar
1594 </PRE>
1595 </P>
1596 <P>
1597 does not find an occurrence of "bar" that is preceded by something other than
1598 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1599 (?!foo) is always true when the next three characters are "bar". A
1600 lookbehind assertion is needed to achieve this effect.
1601 </P>
1602 <P>
1603 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1604 negative assertions. For example,
1605 </P>
1606 <P>
1607 <PRE>
1608 (?&#60;!foo)bar
1609 </PRE>
1610 </P>
1611 <P>
1612 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1613 a lookbehind assertion are restricted such that all the strings it matches must
1614 have a fixed length. However, if there are several alternatives, they do not
1615 all have to have the same fixed length. Thus
1616 </P>
1617 <P>
1618 <PRE>
1619 (?&#60;=bullock|donkey)
1620 </PRE>
1621 </P>
1622 <P>
1623 is permitted, but
1624 </P>
1625 <P>
1626 <PRE>
1627 (?&#60;!dogs?|cats?)
1628 </PRE>
1629 </P>
1630 <P>
1631 causes an error at compile time. Branches that match different length strings
1632 are permitted only at the top level of a lookbehind assertion. This is an
1633 extension compared with Perl 5.005, which requires all branches to match the
1634 same length of string. An assertion such as
1635 </P>
1636 <P>
1637 <PRE>
1638 (?&#60;=ab(c|de))
1639 </PRE>
1640 </P>
1641 <P>
1642 is not permitted, because its single top-level branch can match two different
1643 lengths, but it is acceptable if rewritten to use two top-level branches:
1644 </P>
1645 <P>
1646 <PRE>
1647 (?&#60;=abc|abde)
1648 </PRE>
1649 </P>
1650 <P>
1651 The implementation of lookbehind assertions is, for each alternative, to
1652 temporarily move the current position back by the fixed width and then try to
1653 match. If there are insufficient characters before the current position, the
1654 match is deemed to fail. Lookbehinds in conjunction with once-only subpatterns
1655 can be particularly useful for matching at the ends of strings; an example is
1656 given at the end of the section on once-only subpatterns.
1657 </P>
1658 <P>
1659 Several assertions (of any sort) may occur in succession. For example,
1660 </P>
1661 <P>
1662 <PRE>
1663 (?&#60;=\d{3})(?&#60;!999)foo
1664 </PRE>
1665 </P>
1666 <P>
1667 matches "foo" preceded by three digits that are not "999". Notice that each of
1668 the assertions is applied independently at the same point in the subject
1669 string. First there is a check that the previous three characters are all
1670 digits, then there is a check that the same three characters are not "999".
1671 This pattern does <I>not</I> match "foo" preceded by six characters, the first
1672 of which are digits and the last three of which are not "999". For example, it
1673 doesn't match "123abcfoo". A pattern to do that is
1674 </P>
1675 <P>
1676 <PRE>
1677 (?&#60;=\d{3}...)(?&#60;!999)foo
1678 </PRE>
1679 </P>
1680 <P>
1681 This time the first assertion looks at the preceding six characters, checking
1682 that the first three are digits, and then the second assertion checks that the
1683 preceding three characters are not "999".
1684 </P>
1685 <P>
1686 Assertions can be nested in any combination. For example,
1687 </P>
1688 <P>
1689 <PRE>
1690 (?&#60;=(?&#60;!foo)bar)baz
1691 </PRE>
1692 </P>
1693 <P>
1694 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1695 preceded by "foo", while
1696 </P>
1697 <P>
1698 <PRE>
1699 (?&#60;=\d{3}(?!999)...)foo
1700 </PRE>
1701 </P>
1702 <P>
1703 is another pattern which matches "foo" preceded by three digits and any three
1704 characters that are not "999".
1705 </P>
1706 <P>
1707 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1708 because it makes no sense to assert the same thing several times. If any kind
1709 of assertion contains capturing subpatterns within it, these are counted for
1710 the purposes of numbering the capturing subpatterns in the whole pattern.
1711 However, substring capturing is carried out only for positive assertions,
1712 because it does not make sense for negative assertions.
1713 </P>
1714 <P>
1715 Assertions count towards the maximum of 200 parenthesized subpatterns.
1716 </P>
1718 <P>
1719 With both maximizing and minimizing repetition, failure of what follows
1720 normally causes the repeated item to be re-evaluated to see if a different
1721 number of repeats allows the rest of the pattern to match. Sometimes it is
1722 useful to prevent this, either to change the nature of the match, or to cause
1723 it fail earlier than it otherwise might, when the author of the pattern knows
1724 there is no point in carrying on.
1725 </P>
1726 <P>
1727 Consider, for example, the pattern \d+foo when applied to the subject line
1728 </P>
1729 <P>
1730 <PRE>
1731 123456bar
1732 </PRE>
1733 </P>
1734 <P>
1735 After matching all 6 digits and then failing to match "foo", the normal
1736 action of the matcher is to try again with only 5 digits matching the \d+
1737 item, and then with 4, and so on, before ultimately failing. Once-only
1738 subpatterns provide the means for specifying that once a portion of the pattern
1739 has matched, it is not to be re-evaluated in this way, so the matcher would
1740 give up immediately on failing to match "foo" the first time. The notation is
1741 another kind of special parenthesis, starting with (?&#62; as in this example:
1742 </P>
1743 <P>
1744 <PRE>
1745 (?&#62;\d+)bar
1746 </PRE>
1747 </P>
1748 <P>
1749 This kind of parenthesis "locks up" the part of the pattern it contains once
1750 it has matched, and a failure further into the pattern is prevented from
1751 backtracking into it. Backtracking past it to previous items, however, works as
1752 normal.
1753 </P>
1754 <P>
1755 An alternative description is that a subpattern of this type matches the string
1756 of characters that an identical standalone pattern would match, if anchored at
1757 the current point in the subject string.
1758 </P>
1759 <P>
1760 Once-only subpatterns are not capturing subpatterns. Simple cases such as the
1761 above example can be thought of as a maximizing repeat that must swallow
1762 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1763 number of digits they match in order to make the rest of the pattern match,
1764 (?&#62;\d+) can only match an entire sequence of digits.
1765 </P>
1766 <P>
1767 This construction can of course contain arbitrarily complicated subpatterns,
1768 and it can be nested.
1769 </P>
1770 <P>
1771 Once-only subpatterns can be used in conjunction with lookbehind assertions to
1772 specify efficient matching at the end of the subject string. Consider a simple
1773 pattern such as
1774 </P>
1775 <P>
1776 <PRE>
1777 abcd$
1778 </PRE>
1779 </P>
1780 <P>
1781 when applied to a long string which does not match it. Because matching
1782 proceeds from left to right, PCRE will look for each "a" in the subject and
1783 then see if what follows matches the rest of the pattern. If the pattern is
1784 specified as
1785 </P>
1786 <P>
1787 <PRE>
1788 ^.*abcd$
1789 </PRE>
1790 </P>
1791 <P>
1792 then the initial .* matches the entire string at first, but when this fails, it
1793 backtracks to match all but the last character, then all but the last two
1794 characters, and so on. Once again the search for "a" covers the entire string,
1795 from right to left, so we are no better off. However, if the pattern is written
1796 as
1797 </P>
1798 <P>
1799 <PRE>
1800 ^(?&#62;.*)(?&#60;=abcd)
1801 </PRE>
1802 </P>
1803 <P>
1804 then there can be no backtracking for the .* item; it can match only the entire
1805 string. The subsequent lookbehind assertion does a single test on the last four
1806 characters. If it fails, the match fails immediately. For long strings, this
1807 approach makes a significant difference to the processing time.
1808 </P>
1810 <P>
1811 It is possible to cause the matching process to obey a subpattern
1812 conditionally or to choose between two alternative subpatterns, depending on
1813 the result of an assertion, or whether a previous capturing subpattern matched
1814 or not. The two possible forms of conditional subpattern are
1815 </P>
1816 <P>
1817 <PRE>
1818 (?(condition)yes-pattern)
1819 (?(condition)yes-pattern|no-pattern)
1820 </PRE>
1821 </P>
1822 <P>
1823 If the condition is satisfied, the yes-pattern is used; otherwise the
1824 no-pattern (if present) is used. If there are more than two alternatives in the
1825 subpattern, a compile-time error occurs.
1826 </P>
1827 <P>
1828 There are two kinds of condition. If the text between the parentheses consists
1829 of a sequence of digits, then the condition is satisfied if the capturing
1830 subpattern of that number has previously matched. Consider the following
1831 pattern, which contains non-significant white space to make it more readable
1832 (assume the PCRE_EXTENDED option) and to divide it into three parts for ease
1833 of discussion:
1834 </P>
1835 <P>
1836 <PRE>
1837 ( \( )? [^()]+ (?(1) \) )
1838 </PRE>
1839 </P>
1840 <P>
1841 The first part matches an optional opening parenthesis, and if that
1842 character is present, sets it as the first captured substring. The second part
1843 matches one or more characters that are not parentheses. The third part is a
1844 conditional subpattern that tests whether the first set of parentheses matched
1845 or not. If they did, that is, if subject started with an opening parenthesis,
1846 the condition is true, and so the yes-pattern is executed and a closing
1847 parenthesis is required. Otherwise, since no-pattern is not present, the
1848 subpattern matches nothing. In other words, this pattern matches a sequence of
1849 non-parentheses, optionally enclosed in parentheses.
1850 </P>
1851 <P>
1852 If the condition is not a sequence of digits, it must be an assertion. This may
1853 be a positive or negative lookahead or lookbehind assertion. Consider this
1854 pattern, again containing non-significant white space, and with the two
1855 alternatives on the second line:
1856 </P>
1857 <P>
1858 <PRE>
1859 (?(?=[^a-z]*[a-z])
1860 \d{2}[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1861 </PRE>
1862 </P>
1863 <P>
1864 The condition is a positive lookahead assertion that matches an optional
1865 sequence of non-letters followed by a letter. In other words, it tests for the
1866 presence of at least one letter in the subject. If a letter is found, the
1867 subject is matched against the first alternative; otherwise it is matched
1868 against the second. This pattern matches strings in one of the two forms
1869 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1870 </P>
1872 <P>
1873 The sequence (?# marks the start of a comment which continues up to the next
1874 closing parenthesis. Nested parentheses are not permitted. The characters
1875 that make up a comment play no part in the pattern matching at all.
1876 </P>
1877 <P>
1878 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1879 character class introduces a comment that continues up to the next newline
1880 character in the pattern.
1881 </P>
1883 <P>
1884 Certain items that may appear in patterns are more efficient than others. It is
1885 more efficient to use a character class like [aeiou] than a set of alternatives
1886 such as (a|e|i|o|u). In general, the simplest construction that provides the
1887 required behaviour is usually the most efficient. Jeffrey Friedl's book
1888 contains a lot of discussion about optimizing regular expressions for efficient
1889 performance.
1890 </P>
1891 <P>
1892 When a pattern begins with .* and the PCRE_DOTALL option is set, the pattern is
1893 implicitly anchored by PCRE, since it can match only at the start of a subject
1894 string. However, if PCRE_DOTALL is not set, PCRE cannot make this optimization,
1895 because the . metacharacter does not then match a newline, and if the subject
1896 string contains newlines, the pattern may match from the character immediately
1897 following one of them instead of from the very start. For example, the pattern
1898 </P>
1899 <P>
1900 <PRE>
1901 (.*) second
1902 </PRE>
1903 </P>
1904 <P>
1905 matches the subject "first\nand second" (where \n stands for a newline
1906 character) with the first captured substring being "and". In order to do this,
1907 PCRE has to retry the match starting after every newline in the subject.
1908 </P>
1909 <P>
1910 If you are using such a pattern with subject strings that do not contain
1911 newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
1912 the pattern with ^.* to indicate explicit anchoring. That saves PCRE from
1913 having to scan along the subject looking for a newline to restart at.
1914 </P>
1915 <P>
1916 Beware of patterns that contain nested indefinite repeats. These can take a
1917 long time to run when applied to a string that does not match. Consider the
1918 pattern fragment
1919 </P>
1920 <P>
1921 <PRE>
1922 (a+)*
1923 </PRE>
1924 </P>
1925 <P>
1926 This can match "aaaa" in 33 different ways, and this number increases very
1927 rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
1928 times, and for each of those cases other than 0, the + repeats can match
1929 different numbers of times.) When the remainder of the pattern is such that the
1930 entire match is going to fail, PCRE has in principle to try every possible
1931 variation, and this can take an extremely long time.
1932 </P>
1933 <P>
1934 An optimization catches some of the more simple cases such as
1935 </P>
1936 <P>
1937 <PRE>
1938 (a+)*b
1939 </PRE>
1940 </P>
1941 <P>
1942 where a literal character follows. Before embarking on the standard matching
1943 procedure, PCRE checks that there is a "b" later in the subject string, and if
1944 there is not, it fails the match immediately. However, when there is no
1945 following literal this optimization cannot be used. You can see the difference
1946 by comparing the behaviour of
1947 </P>
1948 <P>
1949 <PRE>
1950 (a+)*\d
1951 </PRE>
1952 </P>
1953 <P>
1954 with the pattern above. The former gives a failure almost instantly when
1955 applied to a whole line of "a" characters, whereas the latter takes an
1956 appreciable time with strings longer than about 20 characters.
1957 </P>
1958 <LI><A NAME="SEC28" HREF="#TOC1">AUTHOR</A>
1959 <P>
1960 Philip Hazel &#60;ph10@cam.ac.uk&#62;
1961 <BR>
1962 University Computing Service,
1963 <BR>
1964 New Museums Site,
1965 <BR>
1966 Cambridge CB2 3QG, England.
1967 <BR>
1968 Phone: +44 1223 334714
1969 </P>
1970 <P>
1971 Last updated: 29 July 1999
1972 <BR>
1973 Copyright (c) 1997-1999 University of Cambridge.

  ViewVC Help
Powered by ViewVC 1.1.5