Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | =head1 NAME |
2 | ||
3 | perlre - Perl regular expressions | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
cb1a09d0 | 7 | This page describes the syntax of regular expressions in Perl. For a |
5f05dabc | 8 | description of how to I<use> regular expressions in matching |
19799a22 | 9 | operations, plus various examples of the same, see discussions |
1e66bd83 | 10 | of C<m//>, C<s///>, C<qr//> and C<??> in L<perlop/"Regexp Quote-Like Operators">. |
cb1a09d0 | 11 | |
19799a22 | 12 | Matching operations can have various modifiers. Modifiers |
5a964f20 | 13 | that relate to the interpretation of the regular expression inside |
19799a22 GS |
14 | are listed below. Modifiers that alter the way a regular expression |
15 | is used by Perl are detailed in L<perlop/"Regexp Quote-Like Operators"> and | |
1e66bd83 | 16 | L<perlop/"Gory details of parsing quoted constructs">. |
a0d0e21e | 17 | |
55497cff | 18 | =over 4 |
19 | ||
20 | =item i | |
21 | ||
22 | Do case-insensitive pattern matching. | |
23 | ||
a034a98d DD |
24 | If C<use locale> is in effect, the case map is taken from the current |
25 | locale. See L<perllocale>. | |
26 | ||
54310121 | 27 | =item m |
55497cff | 28 | |
29 | Treat string as multiple lines. That is, change "^" and "$" from matching | |
14218588 | 30 | the start or end of the string to matching the start or end of any |
7f761169 | 31 | line anywhere within the string. |
55497cff | 32 | |
54310121 | 33 | =item s |
55497cff | 34 | |
35 | Treat string as single line. That is, change "." to match any character | |
19799a22 | 36 | whatsoever, even a newline, which normally it would not match. |
55497cff | 37 | |
19799a22 GS |
38 | The C</s> and C</m> modifiers both override the C<$*> setting. That |
39 | is, no matter what C<$*> contains, C</s> without C</m> will force | |
40 | "^" to match only at the beginning of the string and "$" to match | |
41 | only at the end (or just before a newline at the end) of the string. | |
42 | Together, as /ms, they let the "." match any character whatsoever, | |
43 | while yet allowing "^" and "$" to match, respectively, just after | |
44 | and just before newlines within the string. | |
7b8d334a | 45 | |
54310121 | 46 | =item x |
55497cff | 47 | |
48 | Extend your pattern's legibility by permitting whitespace and comments. | |
49 | ||
50 | =back | |
a0d0e21e LW |
51 | |
52 | These are usually written as "the C</x> modifier", even though the delimiter | |
14218588 | 53 | in question might not really be a slash. Any of these |
a0d0e21e | 54 | modifiers may also be embedded within the regular expression itself using |
14218588 | 55 | the C<(?...)> construct. See below. |
a0d0e21e | 56 | |
4633a7c4 | 57 | The C</x> modifier itself needs a little more explanation. It tells |
55497cff | 58 | the regular expression parser to ignore whitespace that is neither |
59 | backslashed nor within a character class. You can use this to break up | |
4633a7c4 | 60 | your regular expression into (slightly) more readable parts. The C<#> |
54310121 | 61 | character is also treated as a metacharacter introducing a comment, |
55497cff | 62 | just as in ordinary Perl code. This also means that if you want real |
14218588 | 63 | whitespace or C<#> characters in the pattern (outside a character |
5a964f20 | 64 | class, where they are unaffected by C</x>), that you'll either have to |
55497cff | 65 | escape them or encode them using octal or hex escapes. Taken together, |
66 | these features go a long way towards making Perl's regular expressions | |
0c815be9 HS |
67 | more readable. Note that you have to be careful not to include the |
68 | pattern delimiter in the comment--perl has no way of knowing you did | |
5a964f20 | 69 | not intend to close the pattern early. See the C-comment deletion code |
0c815be9 | 70 | in L<perlop>. |
a0d0e21e LW |
71 | |
72 | =head2 Regular Expressions | |
73 | ||
19799a22 | 74 | The patterns used in Perl pattern matching derive from supplied in |
14218588 | 75 | the Version 8 regex routines. (The routines are derived |
19799a22 GS |
76 | (distantly) from Henry Spencer's freely redistributable reimplementation |
77 | of the V8 routines.) See L<Version 8 Regular Expressions> for | |
78 | details. | |
a0d0e21e LW |
79 | |
80 | In particular the following metacharacters have their standard I<egrep>-ish | |
81 | meanings: | |
82 | ||
54310121 | 83 | \ Quote the next metacharacter |
a0d0e21e LW |
84 | ^ Match the beginning of the line |
85 | . Match any character (except newline) | |
c07a80fd | 86 | $ Match the end of the line (or before newline at the end) |
a0d0e21e LW |
87 | | Alternation |
88 | () Grouping | |
89 | [] Character class | |
90 | ||
14218588 GS |
91 | By default, the "^" character is guaranteed to match only the |
92 | beginning of the string, the "$" character only the end (or before the | |
93 | newline at the end), and Perl does certain optimizations with the | |
a0d0e21e LW |
94 | assumption that the string contains only one line. Embedded newlines |
95 | will not be matched by "^" or "$". You may, however, wish to treat a | |
4a6725af | 96 | string as a multi-line buffer, such that the "^" will match after any |
a0d0e21e LW |
97 | newline within the string, and "$" will match before any newline. At the |
98 | cost of a little more overhead, you can do this by using the /m modifier | |
99 | on the pattern match operator. (Older programs did this by setting C<$*>, | |
5f05dabc | 100 | but this practice is now deprecated.) |
a0d0e21e | 101 | |
14218588 | 102 | To simplify multi-line substitutions, the "." character never matches a |
55497cff | 103 | newline unless you use the C</s> modifier, which in effect tells Perl to pretend |
a0d0e21e LW |
104 | the string is a single line--even if it isn't. The C</s> modifier also |
105 | overrides the setting of C<$*>, in case you have some (badly behaved) older | |
106 | code that sets it in another module. | |
107 | ||
108 | The following standard quantifiers are recognized: | |
109 | ||
110 | * Match 0 or more times | |
111 | + Match 1 or more times | |
112 | ? Match 1 or 0 times | |
113 | {n} Match exactly n times | |
114 | {n,} Match at least n times | |
115 | {n,m} Match at least n but not more than m times | |
116 | ||
117 | (If a curly bracket occurs in any other context, it is treated | |
118 | as a regular character.) The "*" modifier is equivalent to C<{0,}>, the "+" | |
25f94b33 | 119 | modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited |
9c79236d GS |
120 | to integral values less than a preset limit defined when perl is built. |
121 | This is usually 32766 on the most common platforms. The actual limit can | |
122 | be seen in the error message generated by code such as this: | |
123 | ||
124 | $_ **= $_ , / {$_} / for 2 .. 42; | |
a0d0e21e | 125 | |
54310121 | 126 | By default, a quantified subpattern is "greedy", that is, it will match as |
127 | many times as possible (given a particular starting location) while still | |
128 | allowing the rest of the pattern to match. If you want it to match the | |
129 | minimum number of times possible, follow the quantifier with a "?". Note | |
130 | that the meanings don't change, just the "greediness": | |
a0d0e21e LW |
131 | |
132 | *? Match 0 or more times | |
133 | +? Match 1 or more times | |
134 | ?? Match 0 or 1 time | |
135 | {n}? Match exactly n times | |
136 | {n,}? Match at least n times | |
137 | {n,m}? Match at least n but not more than m times | |
138 | ||
5f05dabc | 139 | Because patterns are processed as double quoted strings, the following |
a0d0e21e LW |
140 | also work: |
141 | ||
0f36ee90 | 142 | \t tab (HT, TAB) |
143 | \n newline (LF, NL) | |
144 | \r return (CR) | |
145 | \f form feed (FF) | |
146 | \a alarm (bell) (BEL) | |
147 | \e escape (think troff) (ESC) | |
cb1a09d0 AD |
148 | \033 octal char (think of a PDP-11) |
149 | \x1B hex char | |
a0ed51b3 | 150 | \x{263a} wide hex char (Unicode SMILEY) |
a0d0e21e | 151 | \c[ control char |
cb1a09d0 AD |
152 | \l lowercase next char (think vi) |
153 | \u uppercase next char (think vi) | |
154 | \L lowercase till \E (think vi) | |
155 | \U uppercase till \E (think vi) | |
156 | \E end case modification (think vi) | |
5a964f20 | 157 | \Q quote (disable) pattern metacharacters till \E |
a0d0e21e | 158 | |
a034a98d | 159 | If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u> |
7b8d334a | 160 | and C<\U> is taken from the current locale. See L<perllocale>. |
a034a98d | 161 | |
1d2dff63 GS |
162 | You cannot include a literal C<$> or C<@> within a C<\Q> sequence. |
163 | An unescaped C<$> or C<@> interpolates the corresponding variable, | |
164 | while escaping will cause the literal string C<\$> to be matched. | |
165 | You'll need to write something like C<m/\Quser\E\@\Qhost/>. | |
166 | ||
a0d0e21e LW |
167 | In addition, Perl defines the following: |
168 | ||
169 | \w Match a "word" character (alphanumeric plus "_") | |
170 | \W Match a non-word character | |
171 | \s Match a whitespace character | |
172 | \S Match a non-whitespace character | |
173 | \d Match a digit character | |
174 | \D Match a non-digit character | |
a0ed51b3 LW |
175 | \pP Match P, named property. Use \p{Prop} for longer names. |
176 | \PP Match non-P | |
f244e06d GS |
177 | \X Match eXtended Unicode "combining character sequence", |
178 | equivalent to C<(?:\PM\pM*)> | |
a0ed51b3 | 179 | \C Match a single C char (octet) even under utf8. |
a0d0e21e | 180 | |
19799a22 | 181 | A C<\w> matches a single alphanumeric character, not a whole word. |
14218588 GS |
182 | Use C<\w+> to match a string of Perl-identifier characters (which isn't |
183 | the same as matching an English word). If C<use locale> is in effect, the | |
184 | list of alphabetic characters generated by C<\w> is taken from the | |
185 | current locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>, | |
186 | C<\d>, and C<\D> within character classes (though not as either end of | |
187 | a range). See L<utf8> for details about C<\pP>, C<\PP>, and C<\X>. | |
a0d0e21e | 188 | |
b8c5462f JH |
189 | The POSIX character class syntax |
190 | ||
191 | [:class:] | |
192 | ||
193 | is also available. The available classes and their \-equivalents | |
194 | (if any) are as follows: | |
195 | ||
196 | alpha | |
197 | alnum | |
198 | ascii | |
199 | cntrl | |
200 | digit \d | |
201 | graph | |
202 | lower | |
203 | ||
204 | punct | |
205 | space \s | |
206 | upper | |
207 | word \w | |
208 | xdigit | |
209 | ||
210 | Note that the [] are part of the [::] construct, not part of the whole | |
211 | character class. For example: | |
212 | ||
213 | [01[:alpha:]%] | |
214 | ||
215 | matches one, zero, any alphabetic character, and the percentage sign. | |
216 | ||
217 | The exact meanings of the above classes depend from many things: | |
218 | if the C<utf8> pragma is used, the following equivalenced to Unicode | |
219 | \p{} constructs hold: | |
220 | ||
221 | alpha IsAlpha | |
222 | alnum IsAlnum | |
223 | ascii IsASCII | |
224 | cntrl IsCntrl | |
225 | digit IsDigit | |
226 | graph IsGraph | |
227 | lower IsLower | |
228 | print IsPrint | |
229 | punct IsPunct | |
230 | space IsSpace | |
231 | upper IsUpper | |
232 | word IsWord | |
233 | xdigit IsXDigit | |
234 | ||
235 | For example, [:lower:] and \p{IsLower} are equivalent. | |
236 | ||
237 | If the C<utf8> pragma is not used but the C<locale> pragma is, the | |
238 | classes correlate with the isalpha(3) interface (except for `word', | |
239 | which is a Perl extension). | |
240 | ||
241 | The assumedly non-obviously named classes are: | |
242 | ||
243 | =over 4 | |
244 | ||
245 | =item cntrl | |
246 | ||
247 | Any control character. Usually characters that don't produce | |
248 | output as such but instead control the terminal somehow: | |
249 | for example newline and backspace are control characters. | |
250 | ||
251 | =item graph | |
252 | ||
253 | Any alphanumeric or punctuation character. | |
254 | ||
255 | =item print | |
256 | ||
257 | Any alphanumeric or punctuation character or space. | |
258 | ||
259 | =item punct | |
260 | ||
261 | Any punctuation character. | |
262 | ||
263 | =item xdigit | |
264 | ||
265 | Any hexadecimal digit. Though this may feel silly | |
266 | (/0-9a-f/i would work just fine) it is included | |
267 | for completeness. | |
268 | ||
269 | =item | |
270 | ||
271 | =back | |
272 | ||
273 | You can negate the [::] character classes by prefixing the class name | |
274 | with a '^'. This is a Perl extension. For example: | |
275 | ||
276 | ^digit \D \P{IsDigit} | |
277 | ^space \S \P{IsSpace} | |
278 | ^word \W \P{IsWord} | |
279 | ||
280 | The POSIX character classes [.cc.] and [=cc=] are B<not> supported | |
281 | and trying to use them will cause an error. | |
282 | ||
a0d0e21e LW |
283 | Perl defines the following zero-width assertions: |
284 | ||
285 | \b Match a word boundary | |
286 | \B Match a non-(word boundary) | |
b85d18e9 IZ |
287 | \A Match only at beginning of string |
288 | \Z Match only at end of string, or before newline at the end | |
289 | \z Match only at end of string | |
a99df21c | 290 | \G Match only where previous m//g left off (works only with /g) |
a0d0e21e | 291 | |
14218588 | 292 | A word boundary (C<\b>) is a spot between two characters |
19799a22 GS |
293 | that has a C<\w> on one side of it and a C<\W> on the other side |
294 | of it (in either order), counting the imaginary characters off the | |
295 | beginning and end of the string as matching a C<\W>. (Within | |
296 | character classes C<\b> represents backspace rather than a word | |
297 | boundary, just as it normally does in any double-quoted string.) | |
298 | The C<\A> and C<\Z> are just like "^" and "$", except that they | |
299 | won't match multiple times when the C</m> modifier is used, while | |
300 | "^" and "$" will match at every internal line boundary. To match | |
301 | the actual end of the string and not ignore an optional trailing | |
302 | newline, use C<\z>. | |
303 | ||
304 | The C<\G> assertion can be used to chain global matches (using | |
305 | C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">. | |
306 | It is also useful when writing C<lex>-like scanners, when you have | |
307 | several patterns that you want to match against consequent substrings | |
308 | of your string, see the previous reference. The actual location | |
309 | where C<\G> will match can also be influenced by using C<pos()> as | |
310 | an lvalue. See L<perlfunc/pos>. | |
14218588 GS |
311 | |
312 | The bracketing construct C<( ... )> creates capture buffers. To | |
313 | refer to the digit'th buffer use \E<lt>digitE<gt> within the | |
314 | match. Outside the match use "$" instead of "\". (The | |
315 | \E<lt>digitE<gt> notation works in certain circumstances outside | |
316 | the match. See the warning below about \1 vs $1 for details.) | |
317 | Referring back to another part of the match is called a | |
318 | I<backreference>. | |
319 | ||
320 | There is no limit to the number of captured substrings that you may | |
321 | use. However Perl also uses \10, \11, etc. as aliases for \010, | |
322 | \011, etc. (Recall that 0 means octal, so \011 is the 9'th ASCII | |
323 | character, a tab.) Perl resolves this ambiguity by interpreting | |
324 | \10 as a backreference only if at least 10 left parentheses have | |
325 | opened before it. Likewise \11 is a backreference only if at least | |
326 | 11 left parentheses have opened before it. And so on. \1 through | |
327 | \9 are always interpreted as backreferences." | |
328 | ||
329 | Examples: | |
a0d0e21e LW |
330 | |
331 | s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words | |
332 | ||
14218588 GS |
333 | if (/(.)\1/) { # find first doubled char |
334 | print "'$1' is the first doubled character\n"; | |
335 | } | |
336 | ||
337 | if (/Time: (..):(..):(..)/) { # parse out values | |
a0d0e21e LW |
338 | $hours = $1; |
339 | $minutes = $2; | |
340 | $seconds = $3; | |
341 | } | |
14218588 GS |
342 | |
343 | Several special variables also refer back to portions of the previous | |
344 | match. C<$+> returns whatever the last bracket match matched. | |
345 | C<$&> returns the entire matched string. (At one point C<$0> did | |
346 | also, but now it returns the name of the program.) C<$`> returns | |
347 | everything before the matched string. And C<$'> returns everything | |
348 | after the matched string. | |
349 | ||
350 | The numbered variables ($1, $2, $3, etc.) and the related punctuation | |
351 | set (C<<$+>, C<$&>, C<$`>, and C<$'>) are all dynamically scoped | |
352 | until the end of the enclosing block or until the next successful | |
353 | match, whichever comes first. (See L<perlsyn/"Compound Statements">.) | |
354 | ||
355 | B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or | |
356 | C<$'> anywhere in the program, it has to provide them for every | |
357 | pattern match. This may substantially slow your program. Perl | |
358 | uses the same mechanism to produce $1, $2, etc, so you also pay a | |
359 | price for each pattern that contains capturing parentheses. (To | |
360 | avoid this cost while retaining the grouping behaviour, use the | |
361 | extended regular expression C<(?: ... )> instead.) But if you never | |
362 | use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing | |
363 | parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`> | |
364 | if you can, but if you can't (and some algorithms really appreciate | |
365 | them), once you've used them once, use them at will, because you've | |
366 | already paid the price. As of 5.005, C<$&> is not so costly as the | |
367 | other two. | |
68dc0745 | 368 | |
19799a22 GS |
369 | Backslashed metacharacters in Perl are alphanumeric, such as C<\b>, |
370 | C<\w>, C<\n>. Unlike some other regular expression languages, there | |
371 | are no backslashed symbols that aren't alphanumeric. So anything | |
372 | that looks like \\, \(, \), \E<lt>, \E<gt>, \{, or \} is always | |
373 | interpreted as a literal character, not a metacharacter. This was | |
374 | once used in a common idiom to disable or quote the special meanings | |
375 | of regular expression metacharacters in a string that you want to | |
376 | use for a pattern. Simply quote all non-alphanumeric characters: | |
a0d0e21e LW |
377 | |
378 | $pattern =~ s/(\W)/\\$1/g; | |
379 | ||
14218588 GS |
380 | Today it is more common to use the quotemeta() function or the C<\Q> |
381 | metaquoting escape sequence to disable all metacharacters' special | |
382 | meanings like this: | |
a0d0e21e LW |
383 | |
384 | /$unquoted\Q$quoted\E$unquoted/ | |
385 | ||
19799a22 GS |
386 | =head2 Extended Patterns |
387 | ||
14218588 GS |
388 | Perl also defines a consistent extension syntax for features not |
389 | found in standard tools like B<awk> and B<lex>. The syntax is a | |
390 | pair of parentheses with a question mark as the first thing within | |
391 | the parentheses. The character after the question mark indicates | |
392 | the extension. | |
19799a22 | 393 | |
14218588 GS |
394 | The stability of these extensions varies widely. Some have been |
395 | part of the core language for many years. Others are experimental | |
396 | and may change without warning or be completely removed. Check | |
397 | the documentation on an individual feature to verify its current | |
398 | status. | |
19799a22 | 399 | |
14218588 GS |
400 | A question mark was chosen for this and for the minimal-matching |
401 | construct because 1) question marks are rare in older regular | |
402 | expressions, and 2) whenever you see one, you should stop and | |
403 | "question" exactly what is going on. That's psychology... | |
a0d0e21e LW |
404 | |
405 | =over 10 | |
406 | ||
cc6b7395 | 407 | =item C<(?#text)> |
a0d0e21e | 408 | |
14218588 | 409 | A comment. The text is ignored. If the C</x> modifier enables |
19799a22 | 410 | whitespace formatting, a simple C<#> will suffice. Note that Perl closes |
259138e3 GS |
411 | the comment as soon as it sees a C<)>, so there is no way to put a literal |
412 | C<)> in the comment. | |
a0d0e21e | 413 | |
19799a22 GS |
414 | =item C<(?imsx-imsx)> |
415 | ||
416 | One or more embedded pattern-match modifiers. This is particularly | |
417 | useful for dynamic patterns, such as those read in from a configuration | |
418 | file, read in as an argument, are specified in a table somewhere, | |
419 | etc. Consider the case that some of which want to be case sensitive | |
420 | and some do not. The case insensitive ones need to include merely | |
421 | C<(?i)> at the front of the pattern. For example: | |
422 | ||
423 | $pattern = "foobar"; | |
424 | if ( /$pattern/i ) { } | |
425 | ||
426 | # more flexible: | |
427 | ||
428 | $pattern = "(?i)foobar"; | |
429 | if ( /$pattern/ ) { } | |
430 | ||
431 | Letters after a C<-> turn those modifiers off. These modifiers are | |
432 | localized inside an enclosing group (if any). For example, | |
433 | ||
434 | ( (?i) blah ) \s+ \1 | |
435 | ||
436 | will match a repeated (I<including the case>!) word C<blah> in any | |
14218588 | 437 | case, assuming C<x> modifier, and no C<i> modifier outside this |
19799a22 GS |
438 | group. |
439 | ||
5a964f20 | 440 | =item C<(?:pattern)> |
a0d0e21e | 441 | |
ca9dfc88 IZ |
442 | =item C<(?imsx-imsx:pattern)> |
443 | ||
5a964f20 TC |
444 | This is for clustering, not capturing; it groups subexpressions like |
445 | "()", but doesn't make backreferences as "()" does. So | |
a0d0e21e | 446 | |
5a964f20 | 447 | @fields = split(/\b(?:a|b|c)\b/) |
a0d0e21e LW |
448 | |
449 | is like | |
450 | ||
5a964f20 | 451 | @fields = split(/\b(a|b|c)\b/) |
a0d0e21e | 452 | |
19799a22 GS |
453 | but doesn't spit out extra fields. It's also cheaper not to capture |
454 | characters if you don't need to. | |
a0d0e21e | 455 | |
19799a22 GS |
456 | Any letters between C<?> and C<:> act as flags modifiers as with |
457 | C<(?imsx-imsx)>. For example, | |
ca9dfc88 IZ |
458 | |
459 | /(?s-i:more.*than).*million/i | |
460 | ||
14218588 | 461 | is equivalent to the more verbose |
ca9dfc88 IZ |
462 | |
463 | /(?:(?s-i)more.*than).*million/i | |
464 | ||
5a964f20 | 465 | =item C<(?=pattern)> |
a0d0e21e | 466 | |
19799a22 | 467 | A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/> |
a0d0e21e LW |
468 | matches a word followed by a tab, without including the tab in C<$&>. |
469 | ||
5a964f20 | 470 | =item C<(?!pattern)> |
a0d0e21e | 471 | |
19799a22 | 472 | A zero-width negative look-ahead assertion. For example C</foo(?!bar)/> |
a0d0e21e | 473 | matches any occurrence of "foo" that isn't followed by "bar". Note |
19799a22 GS |
474 | however that look-ahead and look-behind are NOT the same thing. You cannot |
475 | use this for look-behind. | |
7b8d334a | 476 | |
5a964f20 | 477 | If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/> |
7b8d334a GS |
478 | will not do what you want. That's because the C<(?!foo)> is just saying that |
479 | the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will | |
480 | match. You would have to do something like C</(?!foo)...bar/> for that. We | |
481 | say "like" because there's the case of your "bar" not having three characters | |
482 | before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>. | |
483 | Sometimes it's still easier just to say: | |
a0d0e21e | 484 | |
a3cb178b | 485 | if (/bar/ && $` !~ /foo$/) |
a0d0e21e | 486 | |
19799a22 | 487 | For look-behind see below. |
c277df42 | 488 | |
5a964f20 | 489 | =item C<(?E<lt>=pattern)> |
c277df42 | 490 | |
19799a22 GS |
491 | A zero-width positive look-behind assertion. For example, C</(?E<lt>=\t)\w+/> |
492 | matches a word that follows a tab, without including the tab in C<$&>. | |
493 | Works only for fixed-width look-behind. | |
c277df42 | 494 | |
5a964f20 | 495 | =item C<(?<!pattern)> |
c277df42 | 496 | |
19799a22 GS |
497 | A zero-width negative look-behind assertion. For example C</(?<!bar)foo/> |
498 | matches any occurrence of "foo" that does not follow "bar". Works | |
499 | only for fixed-width look-behind. | |
c277df42 | 500 | |
cc6b7395 | 501 | =item C<(?{ code })> |
c277df42 | 502 | |
19799a22 GS |
503 | B<WARNING>: This extended regular expression feature is considered |
504 | highly experimental, and may be changed or deleted without notice. | |
c277df42 | 505 | |
19799a22 GS |
506 | This zero-width assertion evaluate any embedded Perl code. It |
507 | always succeeds, and its C<code> is not interpolated. Currently, | |
508 | the rules to determine where the C<code> ends are somewhat convoluted. | |
509 | ||
510 | The C<code> is properly scoped in the following sense: If the assertion | |
511 | is backtracked (compare L<"Backtracking">), all changes introduced after | |
512 | C<local>ization are undone, so that | |
b9ac3b5b GS |
513 | |
514 | $_ = 'a' x 8; | |
515 | m< | |
516 | (?{ $cnt = 0 }) # Initialize $cnt. | |
517 | ( | |
518 | a | |
519 | (?{ | |
520 | local $cnt = $cnt + 1; # Update $cnt, backtracking-safe. | |
521 | }) | |
522 | )* | |
523 | aaaa | |
524 | (?{ $res = $cnt }) # On success copy to non-localized | |
525 | # location. | |
526 | >x; | |
527 | ||
19799a22 | 528 | will set C<$res = 4>. Note that after the match, $cnt returns to the globally |
14218588 | 529 | introduced value, because the scopes that restrict C<local> operators |
b9ac3b5b GS |
530 | are unwound. |
531 | ||
19799a22 GS |
532 | This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)> |
533 | switch. If I<not> used in this way, the result of evaluation of | |
534 | C<code> is put into the special variable C<$^R>. This happens | |
535 | immediately, so C<$^R> can be used from other C<(?{ code })> assertions | |
536 | inside the same regular expression. | |
b9ac3b5b | 537 | |
19799a22 GS |
538 | The assignment to C<$^R> above is properly localized, so the old |
539 | value of C<$^R> is restored if the assertion is backtracked; compare | |
540 | L<"Backtracking">. | |
b9ac3b5b | 541 | |
19799a22 GS |
542 | For reasons of security, this construct is forbidden if the regular |
543 | expression involves run-time interpolation of variables, unless the | |
544 | perilous C<use re 'eval'> pragma has been used (see L<re>), or the | |
545 | variables contain results of C<qr//> operator (see | |
546 | L<perlop/"qr/STRING/imosx">). | |
871b0233 | 547 | |
14218588 | 548 | This restriction is because of the wide-spread and remarkably convenient |
19799a22 | 549 | custom of using run-time determined strings as patterns. For example: |
871b0233 IZ |
550 | |
551 | $re = <>; | |
552 | chomp $re; | |
553 | $string =~ /$re/; | |
554 | ||
14218588 GS |
555 | Before Perl knew how to execute interpolated code within a pattern, |
556 | this operation was completely safe from a security point of view, | |
557 | although it could raise an exception from an illegal pattern. If | |
558 | you turn on the C<use re 'eval'>, though, it is no longer secure, | |
559 | so you should only do so if you are also using taint checking. | |
560 | Better yet, use the carefully constrained evaluation within a Safe | |
561 | module. See L<perlsec> for details about both these mechanisms. | |
871b0233 | 562 | |
0f5d15d6 IZ |
563 | =item C<(?p{ code })> |
564 | ||
19799a22 GS |
565 | B<WARNING>: This extended regular expression feature is considered |
566 | highly experimental, and may be changed or deleted without notice. | |
0f5d15d6 | 567 | |
19799a22 GS |
568 | This is a "postponed" regular subexpression. The C<code> is evaluated |
569 | at run time, at the moment this subexpression may match. The result | |
570 | of evaluation is considered as a regular expression and matched as | |
571 | if it were inserted instead of this construct. | |
0f5d15d6 | 572 | |
428594d9 | 573 | The C<code> is not interpolated. As before, the rules to determine |
19799a22 GS |
574 | where the C<code> ends are currently somewhat convoluted. |
575 | ||
576 | The following pattern matches a parenthesized group: | |
0f5d15d6 IZ |
577 | |
578 | $re = qr{ | |
579 | \( | |
580 | (?: | |
581 | (?> [^()]+ ) # Non-parens without backtracking | |
582 | | | |
583 | (?p{ $re }) # Group with matching parens | |
584 | )* | |
585 | \) | |
586 | }x; | |
587 | ||
5a964f20 TC |
588 | =item C<(?E<gt>pattern)> |
589 | ||
19799a22 GS |
590 | B<WARNING>: This extended regular expression feature is considered |
591 | highly experimental, and may be changed or deleted without notice. | |
592 | ||
593 | An "independent" subexpression, one which matches the substring | |
594 | that a I<standalone> C<pattern> would match if anchored at the given | |
14218588 | 595 | position--but it matches no more than this substring. This |
19799a22 GS |
596 | construct is useful for optimizations of what would otherwise be |
597 | "eternal" matches, because it will not backtrack (see L<"Backtracking">). | |
598 | ||
599 | For example: C<^(?E<gt>a*)ab> will never match, since C<(?E<gt>a*)> | |
600 | (anchored at the beginning of string, as above) will match I<all> | |
601 | characters C<a> at the beginning of string, leaving no C<a> for | |
602 | C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>, | |
603 | since the match of the subgroup C<a*> is influenced by the following | |
604 | group C<ab> (see L<"Backtracking">). In particular, C<a*> inside | |
605 | C<a*ab> will match fewer characters than a standalone C<a*>, since | |
606 | this makes the tail match. | |
607 | ||
608 | An effect similar to C<(?E<gt>pattern)> may be achieved by writing | |
609 | C<(?=(pattern))\1>. This matches the same substring as a standalone | |
610 | C<a+>, and the following C<\1> eats the matched string; it therefore | |
611 | makes a zero-length assertion into an analogue of C<(?E<gt>...)>. | |
612 | (The difference between these two constructs is that the second one | |
613 | uses a capturing group, thus shifting ordinals of backreferences | |
614 | in the rest of a regular expression.) | |
615 | ||
616 | Consider this pattern: | |
c277df42 | 617 | |
871b0233 IZ |
618 | m{ \( |
619 | ( | |
620 | [^()]+ | |
621 | | | |
622 | \( [^()]* \) | |
623 | )+ | |
624 | \) | |
625 | }x | |
5a964f20 | 626 | |
19799a22 GS |
627 | That will efficiently match a nonempty group with matching parentheses |
628 | two levels deep or less. However, if there is no such group, it | |
629 | will take virtually forever on a long string. That's because there | |
630 | are so many different ways to split a long string into several | |
631 | substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar | |
632 | to a subpattern of the above pattern. Consider how the pattern | |
633 | above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several | |
634 | seconds, but that each extra letter doubles this time. This | |
635 | exponential performance will make it appear that your program has | |
14218588 | 636 | hung. However, a tiny change to this pattern |
5a964f20 | 637 | |
871b0233 IZ |
638 | m{ \( |
639 | ( | |
640 | (?> [^()]+ ) | |
641 | | | |
642 | \( [^()]* \) | |
643 | )+ | |
644 | \) | |
645 | }x | |
c277df42 | 646 | |
5a964f20 TC |
647 | which uses C<(?E<gt>...)> matches exactly when the one above does (verifying |
648 | this yourself would be a productive exercise), but finishes in a fourth | |
649 | the time when used on a similar string with 1000000 C<a>s. Be aware, | |
650 | however, that this pattern currently triggers a warning message under | |
651 | B<-w> saying it C<"matches the null string many times">): | |
c277df42 | 652 | |
8d300b32 | 653 | On simple groups, such as the pattern C<(?E<gt> [^()]+ )>, a comparable |
19799a22 | 654 | effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>. |
c277df42 IZ |
655 | This was only 4 times slower on a string with 1000000 C<a>s. |
656 | ||
5a964f20 | 657 | =item C<(?(condition)yes-pattern|no-pattern)> |
c277df42 | 658 | |
5a964f20 | 659 | =item C<(?(condition)yes-pattern)> |
c277df42 | 660 | |
19799a22 GS |
661 | B<WARNING>: This extended regular expression feature is considered |
662 | highly experimental, and may be changed or deleted without notice. | |
663 | ||
c277df42 IZ |
664 | Conditional expression. C<(condition)> should be either an integer in |
665 | parentheses (which is valid if the corresponding pair of parentheses | |
19799a22 | 666 | matched), or look-ahead/look-behind/evaluate zero-width assertion. |
c277df42 | 667 | |
19799a22 | 668 | For example: |
c277df42 | 669 | |
5a964f20 | 670 | m{ ( \( )? |
871b0233 | 671 | [^()]+ |
5a964f20 | 672 | (?(1) \) ) |
871b0233 | 673 | }x |
c277df42 IZ |
674 | |
675 | matches a chunk of non-parentheses, possibly included in parentheses | |
676 | themselves. | |
a0d0e21e | 677 | |
a0d0e21e LW |
678 | =back |
679 | ||
c07a80fd | 680 | =head2 Backtracking |
681 | ||
c277df42 | 682 | A fundamental feature of regular expression matching involves the |
5a964f20 | 683 | notion called I<backtracking>, which is currently used (when needed) |
c277df42 IZ |
684 | by all regular expression quantifiers, namely C<*>, C<*?>, C<+>, |
685 | C<+?>, C<{n,m}>, and C<{n,m}?>. | |
c07a80fd | 686 | |
687 | For a regular expression to match, the I<entire> regular expression must | |
688 | match, not just part of it. So if the beginning of a pattern containing a | |
689 | quantifier succeeds in a way that causes later parts in the pattern to | |
690 | fail, the matching engine backs up and recalculates the beginning | |
691 | part--that's why it's called backtracking. | |
692 | ||
693 | Here is an example of backtracking: Let's say you want to find the | |
694 | word following "foo" in the string "Food is on the foo table.": | |
695 | ||
696 | $_ = "Food is on the foo table."; | |
697 | if ( /\b(foo)\s+(\w+)/i ) { | |
698 | print "$2 follows $1.\n"; | |
699 | } | |
700 | ||
701 | When the match runs, the first part of the regular expression (C<\b(foo)>) | |
702 | finds a possible match right at the beginning of the string, and loads up | |
703 | $1 with "Foo". However, as soon as the matching engine sees that there's | |
704 | no whitespace following the "Foo" that it had saved in $1, it realizes its | |
68dc0745 | 705 | mistake and starts over again one character after where it had the |
c07a80fd | 706 | tentative match. This time it goes all the way until the next occurrence |
707 | of "foo". The complete regular expression matches this time, and you get | |
708 | the expected output of "table follows foo." | |
709 | ||
710 | Sometimes minimal matching can help a lot. Imagine you'd like to match | |
711 | everything between "foo" and "bar". Initially, you write something | |
712 | like this: | |
713 | ||
714 | $_ = "The food is under the bar in the barn."; | |
715 | if ( /foo(.*)bar/ ) { | |
716 | print "got <$1>\n"; | |
717 | } | |
718 | ||
719 | Which perhaps unexpectedly yields: | |
720 | ||
721 | got <d is under the bar in the > | |
722 | ||
723 | That's because C<.*> was greedy, so you get everything between the | |
14218588 | 724 | I<first> "foo" and the I<last> "bar". Here it's more effective |
c07a80fd | 725 | to use minimal matching to make sure you get the text between a "foo" |
726 | and the first "bar" thereafter. | |
727 | ||
728 | if ( /foo(.*?)bar/ ) { print "got <$1>\n" } | |
729 | got <d is under the > | |
730 | ||
731 | Here's another example: let's say you'd like to match a number at the end | |
732 | of a string, and you also want to keep the preceding part the match. | |
733 | So you write this: | |
734 | ||
735 | $_ = "I have 2 numbers: 53147"; | |
736 | if ( /(.*)(\d*)/ ) { # Wrong! | |
737 | print "Beginning is <$1>, number is <$2>.\n"; | |
738 | } | |
739 | ||
740 | That won't work at all, because C<.*> was greedy and gobbled up the | |
741 | whole string. As C<\d*> can match on an empty string the complete | |
742 | regular expression matched successfully. | |
743 | ||
8e1088bc | 744 | Beginning is <I have 2 numbers: 53147>, number is <>. |
c07a80fd | 745 | |
746 | Here are some variants, most of which don't work: | |
747 | ||
748 | $_ = "I have 2 numbers: 53147"; | |
749 | @pats = qw{ | |
750 | (.*)(\d*) | |
751 | (.*)(\d+) | |
752 | (.*?)(\d*) | |
753 | (.*?)(\d+) | |
754 | (.*)(\d+)$ | |
755 | (.*?)(\d+)$ | |
756 | (.*)\b(\d+)$ | |
757 | (.*\D)(\d+)$ | |
758 | }; | |
759 | ||
760 | for $pat (@pats) { | |
761 | printf "%-12s ", $pat; | |
762 | if ( /$pat/ ) { | |
763 | print "<$1> <$2>\n"; | |
764 | } else { | |
765 | print "FAIL\n"; | |
766 | } | |
767 | } | |
768 | ||
769 | That will print out: | |
770 | ||
771 | (.*)(\d*) <I have 2 numbers: 53147> <> | |
772 | (.*)(\d+) <I have 2 numbers: 5314> <7> | |
773 | (.*?)(\d*) <> <> | |
774 | (.*?)(\d+) <I have > <2> | |
775 | (.*)(\d+)$ <I have 2 numbers: 5314> <7> | |
776 | (.*?)(\d+)$ <I have 2 numbers: > <53147> | |
777 | (.*)\b(\d+)$ <I have 2 numbers: > <53147> | |
778 | (.*\D)(\d+)$ <I have 2 numbers: > <53147> | |
779 | ||
780 | As you see, this can be a bit tricky. It's important to realize that a | |
781 | regular expression is merely a set of assertions that gives a definition | |
782 | of success. There may be 0, 1, or several different ways that the | |
783 | definition might succeed against a particular string. And if there are | |
5a964f20 TC |
784 | multiple ways it might succeed, you need to understand backtracking to |
785 | know which variety of success you will achieve. | |
c07a80fd | 786 | |
19799a22 | 787 | When using look-ahead assertions and negations, this can all get even |
54310121 | 788 | tricker. Imagine you'd like to find a sequence of non-digits not |
c07a80fd | 789 | followed by "123". You might try to write that as |
790 | ||
871b0233 IZ |
791 | $_ = "ABC123"; |
792 | if ( /^\D*(?!123)/ ) { # Wrong! | |
793 | print "Yup, no 123 in $_\n"; | |
794 | } | |
c07a80fd | 795 | |
796 | But that isn't going to match; at least, not the way you're hoping. It | |
797 | claims that there is no 123 in the string. Here's a clearer picture of | |
798 | why it that pattern matches, contrary to popular expectations: | |
799 | ||
800 | $x = 'ABC123' ; | |
801 | $y = 'ABC445' ; | |
802 | ||
803 | print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ; | |
804 | print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ; | |
805 | ||
806 | print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ; | |
807 | print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ; | |
808 | ||
809 | This prints | |
810 | ||
811 | 2: got ABC | |
812 | 3: got AB | |
813 | 4: got ABC | |
814 | ||
5f05dabc | 815 | You might have expected test 3 to fail because it seems to a more |
c07a80fd | 816 | general purpose version of test 1. The important difference between |
817 | them is that test 3 contains a quantifier (C<\D*>) and so can use | |
818 | backtracking, whereas test 1 will not. What's happening is | |
819 | that you've asked "Is it true that at the start of $x, following 0 or more | |
5f05dabc | 820 | non-digits, you have something that's not 123?" If the pattern matcher had |
c07a80fd | 821 | let C<\D*> expand to "ABC", this would have caused the whole pattern to |
54310121 | 822 | fail. |
14218588 | 823 | |
c07a80fd | 824 | The search engine will initially match C<\D*> with "ABC". Then it will |
14218588 | 825 | try to match C<(?!123> with "123", which fails. But because |
c07a80fd | 826 | a quantifier (C<\D*>) has been used in the regular expression, the |
827 | search engine can backtrack and retry the match differently | |
54310121 | 828 | in the hope of matching the complete regular expression. |
c07a80fd | 829 | |
5a964f20 TC |
830 | The pattern really, I<really> wants to succeed, so it uses the |
831 | standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this | |
c07a80fd | 832 | time. Now there's indeed something following "AB" that is not |
14218588 | 833 | "123". It's "C123", which suffices. |
c07a80fd | 834 | |
14218588 GS |
835 | We can deal with this by using both an assertion and a negation. |
836 | We'll say that the first part in $1 must be followed both by a digit | |
837 | and by something that's not "123". Remember that the look-aheads | |
838 | are zero-width expressions--they only look, but don't consume any | |
839 | of the string in their match. So rewriting this way produces what | |
c07a80fd | 840 | you'd expect; that is, case 5 will fail, but case 6 succeeds: |
841 | ||
842 | print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ; | |
843 | print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ; | |
844 | ||
845 | 6: got ABC | |
846 | ||
5a964f20 | 847 | In other words, the two zero-width assertions next to each other work as though |
19799a22 | 848 | they're ANDed together, just as you'd use any built-in assertions: C</^$/> |
c07a80fd | 849 | matches only if you're at the beginning of the line AND the end of the |
850 | line simultaneously. The deeper underlying truth is that juxtaposition in | |
851 | regular expressions always means AND, except when you write an explicit OR | |
852 | using the vertical bar. C</ab/> means match "a" AND (then) match "b", | |
853 | although the attempted matches are made at different positions because "a" | |
854 | is not a zero-width assertion, but a one-width assertion. | |
855 | ||
19799a22 | 856 | B<WARNING>: particularly complicated regular expressions can take |
14218588 | 857 | exponential time to solve because of the immense number of possible |
19799a22 | 858 | ways they can use backtracking to try match. For example, this will |
14218588 | 859 | take a painfully long time to run |
c07a80fd | 860 | |
861 | /((a{0,5}){0,5}){0,5}/ | |
862 | ||
14218588 GS |
863 | And if you used C<*>'s instead of limiting it to 0 through 5 matches, |
864 | then it would take forever--or until you ran out of stack space. | |
c07a80fd | 865 | |
c277df42 | 866 | A powerful tool for optimizing such beasts is "independent" groups, |
5a964f20 | 867 | which do not backtrace (see L<C<(?E<gt>pattern)>>). Note also that |
19799a22 | 868 | zero-length look-ahead/look-behind assertions will not backtrace to make |
14218588 GS |
869 | the tail match, since they are in "logical" context: only |
870 | whether they match is considered relevant. For an example | |
19799a22 | 871 | where side-effects of a look-ahead I<might> have influenced the |
5a964f20 | 872 | following match, see L<C<(?E<gt>pattern)>>. |
c277df42 | 873 | |
a0d0e21e LW |
874 | =head2 Version 8 Regular Expressions |
875 | ||
5a964f20 | 876 | In case you're not familiar with the "regular" Version 8 regex |
a0d0e21e LW |
877 | routines, here are the pattern-matching rules not described above. |
878 | ||
54310121 | 879 | Any single character matches itself, unless it is a I<metacharacter> |
a0d0e21e | 880 | with a special meaning described here or above. You can cause |
5a964f20 | 881 | characters that normally function as metacharacters to be interpreted |
5f05dabc | 882 | literally by prefixing them with a "\" (e.g., "\." matches a ".", not any |
a0d0e21e LW |
883 | character; "\\" matches a "\"). A series of characters matches that |
884 | series of characters in the target string, so the pattern C<blurfl> | |
885 | would match "blurfl" in the target string. | |
886 | ||
887 | You can specify a character class, by enclosing a list of characters | |
5a964f20 | 888 | in C<[]>, which will match any one character from the list. If the |
a0d0e21e | 889 | first character after the "[" is "^", the class matches any character not |
14218588 | 890 | in the list. Within a list, the "-" character specifies a |
5a964f20 | 891 | range, so that C<a-z> represents all characters between "a" and "z", |
84850974 DD |
892 | inclusive. If you want "-" itself to be a member of a class, put it |
893 | at the start or end of the list, or escape it with a backslash. (The | |
894 | following all specify the same class of three characters: C<[-az]>, | |
895 | C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which | |
896 | specifies a class containing twenty-six characters.) | |
a0d0e21e | 897 | |
8ada0baa JH |
898 | Note also that the whole range idea is rather unportable between |
899 | character sets--and even within character sets they may cause results | |
900 | you probably didn't expect. A sound principle is to use only ranges | |
901 | that begin from and end at either alphabets of equal case ([a-e], | |
902 | [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt, | |
903 | spell out the character sets in full. | |
904 | ||
54310121 | 905 | Characters may be specified using a metacharacter syntax much like that |
a0d0e21e LW |
906 | used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return, |
907 | "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string | |
908 | of octal digits, matches the character whose ASCII value is I<nnn>. | |
0f36ee90 | 909 | Similarly, \xI<nn>, where I<nn> are hexadecimal digits, matches the |
a0d0e21e | 910 | character whose ASCII value is I<nn>. The expression \cI<x> matches the |
54310121 | 911 | ASCII character control-I<x>. Finally, the "." metacharacter matches any |
a0d0e21e LW |
912 | character except "\n" (unless you use C</s>). |
913 | ||
914 | You can specify a series of alternatives for a pattern using "|" to | |
915 | separate them, so that C<fee|fie|foe> will match any of "fee", "fie", | |
5a964f20 | 916 | or "foe" in the target string (as would C<f(e|i|o)e>). The |
a0d0e21e LW |
917 | first alternative includes everything from the last pattern delimiter |
918 | ("(", "[", or the beginning of the pattern) up to the first "|", and | |
919 | the last alternative contains everything from the last "|" to the next | |
14218588 GS |
920 | pattern delimiter. That's why it's common practice to include |
921 | alternatives in parentheses: to minimize confusion about where they | |
a3cb178b GS |
922 | start and end. |
923 | ||
5a964f20 | 924 | Alternatives are tried from left to right, so the first |
a3cb178b GS |
925 | alternative found for which the entire expression matches, is the one that |
926 | is chosen. This means that alternatives are not necessarily greedy. For | |
628afcb5 | 927 | example: when matching C<foo|foot> against "barefoot", only the "foo" |
a3cb178b GS |
928 | part will match, as that is the first alternative tried, and it successfully |
929 | matches the target string. (This might not seem important, but it is | |
930 | important when you are capturing matched text using parentheses.) | |
931 | ||
5a964f20 | 932 | Also remember that "|" is interpreted as a literal within square brackets, |
a3cb178b | 933 | so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>. |
a0d0e21e | 934 | |
14218588 GS |
935 | Within a pattern, you may designate subpatterns for later reference |
936 | by enclosing them in parentheses, and you may refer back to the | |
937 | I<n>th subpattern later in the pattern using the metacharacter | |
938 | \I<n>. Subpatterns are numbered based on the left to right order | |
939 | of their opening parenthesis. A backreference matches whatever | |
940 | actually matched the subpattern in the string being examined, not | |
941 | the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will | |
942 | match "0x1234 0x4321", but not "0x1234 01234", because subpattern | |
943 | 1 matched "0x", even though the rule C<0|0x> could potentially match | |
944 | the leading 0 in the second number. | |
cb1a09d0 | 945 | |
19799a22 | 946 | =head2 Warning on \1 vs $1 |
cb1a09d0 | 947 | |
5a964f20 | 948 | Some people get too used to writing things like: |
cb1a09d0 AD |
949 | |
950 | $pattern =~ s/(\W)/\\\1/g; | |
951 | ||
952 | This is grandfathered for the RHS of a substitute to avoid shocking the | |
953 | B<sed> addicts, but it's a dirty habit to get into. That's because in | |
5f05dabc | 954 | PerlThink, the righthand side of a C<s///> is a double-quoted string. C<\1> in |
cb1a09d0 AD |
955 | the usual double-quoted string means a control-A. The customary Unix |
956 | meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit | |
957 | of doing that, you get yourself into trouble if you then add an C</e> | |
958 | modifier. | |
959 | ||
5a964f20 | 960 | s/(\d+)/ \1 + 1 /eg; # causes warning under -w |
cb1a09d0 AD |
961 | |
962 | Or if you try to do | |
963 | ||
964 | s/(\d+)/\1000/; | |
965 | ||
966 | You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with | |
14218588 | 967 | C<${1}000>. The operation of interpolation should not be confused |
cb1a09d0 AD |
968 | with the operation of matching a backreference. Certainly they mean two |
969 | different things on the I<left> side of the C<s///>. | |
9fa51da4 | 970 | |
c84d73f1 IZ |
971 | =head2 Repeated patterns matching zero-length substring |
972 | ||
19799a22 | 973 | B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite. |
c84d73f1 IZ |
974 | |
975 | Regular expressions provide a terse and powerful programming language. As | |
976 | with most other power tools, power comes together with the ability | |
977 | to wreak havoc. | |
978 | ||
979 | A common abuse of this power stems from the ability to make infinite | |
628afcb5 | 980 | loops using regular expressions, with something as innocuous as: |
c84d73f1 IZ |
981 | |
982 | 'foo' =~ m{ ( o? )* }x; | |
983 | ||
984 | The C<o?> can match at the beginning of C<'foo'>, and since the position | |
985 | in the string is not moved by the match, C<o?> would match again and again | |
14218588 | 986 | because of the C<*> modifier. Another common way to create a similar cycle |
c84d73f1 IZ |
987 | is with the looping modifier C<//g>: |
988 | ||
989 | @matches = ( 'foo' =~ m{ o? }xg ); | |
990 | ||
991 | or | |
992 | ||
993 | print "match: <$&>\n" while 'foo' =~ m{ o? }xg; | |
994 | ||
995 | or the loop implied by split(). | |
996 | ||
997 | However, long experience has shown that many programming tasks may | |
14218588 GS |
998 | be significantly simplified by using repeated subexpressions that |
999 | may match zero-length substrings. Here's a simple example being: | |
c84d73f1 IZ |
1000 | |
1001 | @chars = split //, $string; # // is not magic in split | |
1002 | ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// / | |
1003 | ||
1004 | Thus Perl allows the C</()/> construct, which I<forcefully breaks | |
1005 | the infinite loop>. The rules for this are different for lower-level | |
1006 | loops given by the greedy modifiers C<*+{}>, and for higher-level | |
1007 | ones like the C</g> modifier or split() operator. | |
1008 | ||
19799a22 GS |
1009 | The lower-level loops are I<interrupted> (that is, the loop is |
1010 | broken) when Perl detects that a repeated expression matched a | |
1011 | zero-length substring. Thus | |
c84d73f1 IZ |
1012 | |
1013 | m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x; | |
1014 | ||
1015 | is made equivalent to | |
1016 | ||
1017 | m{ (?: NON_ZERO_LENGTH )* | |
1018 | | | |
1019 | (?: ZERO_LENGTH )? | |
1020 | }x; | |
1021 | ||
1022 | The higher level-loops preserve an additional state between iterations: | |
1023 | whether the last match was zero-length. To break the loop, the following | |
1024 | match after a zero-length match is prohibited to have a length of zero. | |
1025 | This prohibition interacts with backtracking (see L<"Backtracking">), | |
1026 | and so the I<second best> match is chosen if the I<best> match is of | |
1027 | zero length. | |
1028 | ||
19799a22 | 1029 | For example: |
c84d73f1 IZ |
1030 | |
1031 | $_ = 'bar'; | |
1032 | s/\w??/<$&>/g; | |
1033 | ||
1034 | results in C<"<><b><><a><><r><>">. At each position of the string the best | |
1035 | match given by non-greedy C<??> is the zero-length match, and the I<second | |
1036 | best> match is what is matched by C<\w>. Thus zero-length matches | |
1037 | alternate with one-character-long matches. | |
1038 | ||
1039 | Similarly, for repeated C<m/()/g> the second-best match is the match at the | |
1040 | position one notch further in the string. | |
1041 | ||
19799a22 | 1042 | The additional state of being I<matched with zero-length> is associated with |
c84d73f1 IZ |
1043 | the matched string, and is reset by each assignment to pos(). |
1044 | ||
1045 | =head2 Creating custom RE engines | |
1046 | ||
1047 | Overloaded constants (see L<overload>) provide a simple way to extend | |
1048 | the functionality of the RE engine. | |
1049 | ||
1050 | Suppose that we want to enable a new RE escape-sequence C<\Y|> which | |
1051 | matches at boundary between white-space characters and non-whitespace | |
1052 | characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly | |
1053 | at these positions, so we want to have each C<\Y|> in the place of the | |
1054 | more complicated version. We can create a module C<customre> to do | |
1055 | this: | |
1056 | ||
1057 | package customre; | |
1058 | use overload; | |
1059 | ||
1060 | sub import { | |
1061 | shift; | |
1062 | die "No argument to customre::import allowed" if @_; | |
1063 | overload::constant 'qr' => \&convert; | |
1064 | } | |
1065 | ||
1066 | sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"} | |
1067 | ||
1068 | my %rules = ( '\\' => '\\', | |
1069 | 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ ); | |
1070 | sub convert { | |
1071 | my $re = shift; | |
1072 | $re =~ s{ | |
1073 | \\ ( \\ | Y . ) | |
1074 | } | |
1075 | { $rules{$1} or invalid($re,$1) }sgex; | |
1076 | return $re; | |
1077 | } | |
1078 | ||
1079 | Now C<use customre> enables the new escape in constant regular | |
1080 | expressions, i.e., those without any runtime variable interpolations. | |
1081 | As documented in L<overload>, this conversion will work only over | |
1082 | literal parts of regular expressions. For C<\Y|$re\Y|> the variable | |
1083 | part of this regular expression needs to be converted explicitly | |
1084 | (but only if the special meaning of C<\Y|> should be enabled inside $re): | |
1085 | ||
1086 | use customre; | |
1087 | $re = <>; | |
1088 | chomp $re; | |
1089 | $re = customre::convert $re; | |
1090 | /\Y|$re\Y|/; | |
1091 | ||
19799a22 GS |
1092 | =head1 BUGS |
1093 | ||
1094 | This manpage is varies from difficult to understand to completely | |
1095 | and utterly opaque. | |
1096 | ||
1097 | =head1 SEE ALSO | |
9fa51da4 | 1098 | |
9b599b2a GS |
1099 | L<perlop/"Regexp Quote-Like Operators">. |
1100 | ||
1e66bd83 PP |
1101 | L<perlop/"Gory details of parsing quoted constructs">. |
1102 | ||
14218588 GS |
1103 | L<perlfaq6>. |
1104 | ||
9b599b2a GS |
1105 | L<perlfunc/pos>. |
1106 | ||
1107 | L<perllocale>. | |
1108 | ||
14218588 GS |
1109 | I<Mastering Regular Expressions> by Jeffrey Friedl, published |
1110 | by O'Reilly and Associates. |