Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | =head1 NAME |
2 | ||
3 | perlre - Perl regular expressions | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
91e0c79e MJD |
7 | This page describes the syntax of regular expressions in Perl. |
8 | ||
9 | if you haven't used regular expressions before, a quick-start | |
10 | introduction is available in L<perlrequick>, and a longer tutorial | |
11 | introduction is available in L<perlretut>. | |
12 | ||
13 | For reference on how regular expressions are used in matching | |
14 | operations, plus various examples of the same, see discussions of | |
15 | C<m//>, C<s///>, C<qr//> and C<??> in L<perlop/"Regexp Quote-Like | |
16 | Operators">. | |
cb1a09d0 | 17 | |
19799a22 | 18 | Matching operations can have various modifiers. Modifiers |
5a964f20 | 19 | that relate to the interpretation of the regular expression inside |
19799a22 GS |
20 | are listed below. Modifiers that alter the way a regular expression |
21 | is used by Perl are detailed in L<perlop/"Regexp Quote-Like Operators"> and | |
1e66bd83 | 22 | L<perlop/"Gory details of parsing quoted constructs">. |
a0d0e21e | 23 | |
55497cff | 24 | =over 4 |
25 | ||
26 | =item i | |
27 | ||
28 | Do case-insensitive pattern matching. | |
29 | ||
a034a98d DD |
30 | If C<use locale> is in effect, the case map is taken from the current |
31 | locale. See L<perllocale>. | |
32 | ||
54310121 | 33 | =item m |
55497cff | 34 | |
35 | Treat string as multiple lines. That is, change "^" and "$" from matching | |
14218588 | 36 | the start or end of the string to matching the start or end of any |
7f761169 | 37 | line anywhere within the string. |
55497cff | 38 | |
54310121 | 39 | =item s |
55497cff | 40 | |
41 | Treat string as single line. That is, change "." to match any character | |
19799a22 | 42 | whatsoever, even a newline, which normally it would not match. |
55497cff | 43 | |
19799a22 GS |
44 | The C</s> and C</m> modifiers both override the C<$*> setting. That |
45 | is, no matter what C<$*> contains, C</s> without C</m> will force | |
46 | "^" to match only at the beginning of the string and "$" to match | |
47 | only at the end (or just before a newline at the end) of the string. | |
48 | Together, as /ms, they let the "." match any character whatsoever, | |
fb55449c | 49 | while still allowing "^" and "$" to match, respectively, just after |
19799a22 | 50 | and just before newlines within the string. |
7b8d334a | 51 | |
54310121 | 52 | =item x |
55497cff | 53 | |
54 | Extend your pattern's legibility by permitting whitespace and comments. | |
55 | ||
56 | =back | |
a0d0e21e LW |
57 | |
58 | These are usually written as "the C</x> modifier", even though the delimiter | |
14218588 | 59 | in question might not really be a slash. Any of these |
a0d0e21e | 60 | modifiers may also be embedded within the regular expression itself using |
14218588 | 61 | the C<(?...)> construct. See below. |
a0d0e21e | 62 | |
4633a7c4 | 63 | The C</x> modifier itself needs a little more explanation. It tells |
55497cff | 64 | the regular expression parser to ignore whitespace that is neither |
65 | backslashed nor within a character class. You can use this to break up | |
4633a7c4 | 66 | your regular expression into (slightly) more readable parts. The C<#> |
54310121 | 67 | character is also treated as a metacharacter introducing a comment, |
55497cff | 68 | just as in ordinary Perl code. This also means that if you want real |
14218588 | 69 | whitespace or C<#> characters in the pattern (outside a character |
5a964f20 | 70 | class, where they are unaffected by C</x>), that you'll either have to |
55497cff | 71 | escape them or encode them using octal or hex escapes. Taken together, |
72 | these features go a long way towards making Perl's regular expressions | |
0c815be9 HS |
73 | more readable. Note that you have to be careful not to include the |
74 | pattern delimiter in the comment--perl has no way of knowing you did | |
5a964f20 | 75 | not intend to close the pattern early. See the C-comment deletion code |
0c815be9 | 76 | in L<perlop>. |
a0d0e21e LW |
77 | |
78 | =head2 Regular Expressions | |
79 | ||
19799a22 | 80 | The patterns used in Perl pattern matching derive from supplied in |
14218588 | 81 | the Version 8 regex routines. (The routines are derived |
19799a22 GS |
82 | (distantly) from Henry Spencer's freely redistributable reimplementation |
83 | of the V8 routines.) See L<Version 8 Regular Expressions> for | |
84 | details. | |
a0d0e21e LW |
85 | |
86 | In particular the following metacharacters have their standard I<egrep>-ish | |
87 | meanings: | |
88 | ||
54310121 | 89 | \ Quote the next metacharacter |
a0d0e21e LW |
90 | ^ Match the beginning of the line |
91 | . Match any character (except newline) | |
c07a80fd | 92 | $ Match the end of the line (or before newline at the end) |
a0d0e21e LW |
93 | | Alternation |
94 | () Grouping | |
95 | [] Character class | |
96 | ||
14218588 GS |
97 | By default, the "^" character is guaranteed to match only the |
98 | beginning of the string, the "$" character only the end (or before the | |
99 | newline at the end), and Perl does certain optimizations with the | |
a0d0e21e LW |
100 | assumption that the string contains only one line. Embedded newlines |
101 | will not be matched by "^" or "$". You may, however, wish to treat a | |
4a6725af | 102 | string as a multi-line buffer, such that the "^" will match after any |
a0d0e21e LW |
103 | newline within the string, and "$" will match before any newline. At the |
104 | cost of a little more overhead, you can do this by using the /m modifier | |
105 | on the pattern match operator. (Older programs did this by setting C<$*>, | |
5f05dabc | 106 | but this practice is now deprecated.) |
a0d0e21e | 107 | |
14218588 | 108 | To simplify multi-line substitutions, the "." character never matches a |
55497cff | 109 | newline unless you use the C</s> modifier, which in effect tells Perl to pretend |
a0d0e21e LW |
110 | the string is a single line--even if it isn't. The C</s> modifier also |
111 | overrides the setting of C<$*>, in case you have some (badly behaved) older | |
112 | code that sets it in another module. | |
113 | ||
114 | The following standard quantifiers are recognized: | |
115 | ||
116 | * Match 0 or more times | |
117 | + Match 1 or more times | |
118 | ? Match 1 or 0 times | |
119 | {n} Match exactly n times | |
120 | {n,} Match at least n times | |
121 | {n,m} Match at least n but not more than m times | |
122 | ||
123 | (If a curly bracket occurs in any other context, it is treated | |
124 | as a regular character.) The "*" modifier is equivalent to C<{0,}>, the "+" | |
25f94b33 | 125 | modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited |
9c79236d GS |
126 | to integral values less than a preset limit defined when perl is built. |
127 | This is usually 32766 on the most common platforms. The actual limit can | |
128 | be seen in the error message generated by code such as this: | |
129 | ||
820475bd | 130 | $_ **= $_ , / {$_} / for 2 .. 42; |
a0d0e21e | 131 | |
54310121 | 132 | By default, a quantified subpattern is "greedy", that is, it will match as |
133 | many times as possible (given a particular starting location) while still | |
134 | allowing the rest of the pattern to match. If you want it to match the | |
135 | minimum number of times possible, follow the quantifier with a "?". Note | |
136 | that the meanings don't change, just the "greediness": | |
a0d0e21e LW |
137 | |
138 | *? Match 0 or more times | |
139 | +? Match 1 or more times | |
140 | ?? Match 0 or 1 time | |
141 | {n}? Match exactly n times | |
142 | {n,}? Match at least n times | |
143 | {n,m}? Match at least n but not more than m times | |
144 | ||
5f05dabc | 145 | Because patterns are processed as double quoted strings, the following |
a0d0e21e LW |
146 | also work: |
147 | ||
0f36ee90 | 148 | \t tab (HT, TAB) |
149 | \n newline (LF, NL) | |
150 | \r return (CR) | |
151 | \f form feed (FF) | |
152 | \a alarm (bell) (BEL) | |
153 | \e escape (think troff) (ESC) | |
cb1a09d0 AD |
154 | \033 octal char (think of a PDP-11) |
155 | \x1B hex char | |
a0ed51b3 | 156 | \x{263a} wide hex char (Unicode SMILEY) |
a0d0e21e | 157 | \c[ control char |
4a2d328f | 158 | \N{name} named char |
cb1a09d0 AD |
159 | \l lowercase next char (think vi) |
160 | \u uppercase next char (think vi) | |
161 | \L lowercase till \E (think vi) | |
162 | \U uppercase till \E (think vi) | |
163 | \E end case modification (think vi) | |
5a964f20 | 164 | \Q quote (disable) pattern metacharacters till \E |
a0d0e21e | 165 | |
a034a98d | 166 | If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u> |
423cee85 | 167 | and C<\U> is taken from the current locale. See L<perllocale>. For |
4a2d328f | 168 | documentation of C<\N{name}>, see L<charnames>. |
a034a98d | 169 | |
1d2dff63 GS |
170 | You cannot include a literal C<$> or C<@> within a C<\Q> sequence. |
171 | An unescaped C<$> or C<@> interpolates the corresponding variable, | |
172 | while escaping will cause the literal string C<\$> to be matched. | |
173 | You'll need to write something like C<m/\Quser\E\@\Qhost/>. | |
174 | ||
a0d0e21e LW |
175 | In addition, Perl defines the following: |
176 | ||
177 | \w Match a "word" character (alphanumeric plus "_") | |
36bbe248 | 178 | \W Match a non-"word" character |
a0d0e21e LW |
179 | \s Match a whitespace character |
180 | \S Match a non-whitespace character | |
181 | \d Match a digit character | |
182 | \D Match a non-digit character | |
a0ed51b3 LW |
183 | \pP Match P, named property. Use \p{Prop} for longer names. |
184 | \PP Match non-P | |
f244e06d GS |
185 | \X Match eXtended Unicode "combining character sequence", |
186 | equivalent to C<(?:\PM\pM*)> | |
4a2d328f | 187 | \C Match a single C char (octet) even under utf8. |
a0d0e21e | 188 | |
36bbe248 | 189 | A C<\w> matches a single alphanumeric character or C<_>, not a whole word. |
14218588 GS |
190 | Use C<\w+> to match a string of Perl-identifier characters (which isn't |
191 | the same as matching an English word). If C<use locale> is in effect, the | |
192 | list of alphabetic characters generated by C<\w> is taken from the | |
193 | current locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>, | |
1209ba90 JH |
194 | C<\d>, and C<\D> within character classes, but if you try to use them |
195 | as endpoints of a range, that's not a range, the "-" is understood literally. | |
196 | See L<utf8> for details about C<\pP>, C<\PP>, and C<\X>. | |
a0d0e21e | 197 | |
b8c5462f JH |
198 | The POSIX character class syntax |
199 | ||
820475bd | 200 | [:class:] |
b8c5462f | 201 | |
26b44a0a JH |
202 | is also available. The available classes and their backslash |
203 | equivalents (if available) are as follows: | |
b8c5462f JH |
204 | |
205 | alpha | |
206 | alnum | |
207 | ascii | |
aaa51d5e | 208 | blank [1] |
b8c5462f JH |
209 | cntrl |
210 | digit \d | |
211 | graph | |
212 | lower | |
213 | ||
214 | punct | |
aaa51d5e | 215 | space \s [2] |
b8c5462f | 216 | upper |
aaa51d5e | 217 | word \w [3] |
b8c5462f JH |
218 | xdigit |
219 | ||
aaa51d5e JF |
220 | [1] A GNU extension equivalent to C<[ \t]>, `all horizontal whitespace'. |
221 | [2] Not I<exactly equivalent> to C<\s> since the C<[[:space:]]> includes | |
222 | also the (very rare) `vertical tabulator', "\ck", chr(11). | |
223 | [3] A Perl extension. | |
224 | ||
26b44a0a | 225 | For example use C<[:upper:]> to match all the uppercase characters. |
aaa51d5e JF |
226 | Note that the C<[]> are part of the C<[::]> construct, not part of the |
227 | whole character class. For example: | |
b8c5462f | 228 | |
820475bd | 229 | [01[:alpha:]%] |
b8c5462f | 230 | |
593df60c | 231 | matches zero, one, any alphabetic character, and the percentage sign. |
b8c5462f | 232 | |
26b44a0a | 233 | If the C<utf8> pragma is used, the following equivalences to Unicode |
3bec3564 JH |
234 | \p{} constructs and equivalent backslash character classes (if available), |
235 | will hold: | |
b8c5462f JH |
236 | |
237 | alpha IsAlpha | |
238 | alnum IsAlnum | |
239 | ascii IsASCII | |
aaa51d5e | 240 | blank IsSpace |
b8c5462f | 241 | cntrl IsCntrl |
3bec3564 | 242 | digit IsDigit \d |
b8c5462f JH |
243 | graph IsGraph |
244 | lower IsLower | |
245 | print IsPrint | |
246 | punct IsPunct | |
247 | space IsSpace | |
3bec3564 | 248 | IsSpacePerl \s |
b8c5462f JH |
249 | upper IsUpper |
250 | word IsWord | |
251 | xdigit IsXDigit | |
252 | ||
26b44a0a | 253 | For example C<[:lower:]> and C<\p{IsLower}> are equivalent. |
b8c5462f JH |
254 | |
255 | If the C<utf8> pragma is not used but the C<locale> pragma is, the | |
aaa51d5e JF |
256 | classes correlate with the usual isalpha(3) interface (except for |
257 | `word' and `blank'). | |
b8c5462f JH |
258 | |
259 | The assumedly non-obviously named classes are: | |
260 | ||
261 | =over 4 | |
262 | ||
263 | =item cntrl | |
264 | ||
820475bd GS |
265 | Any control character. Usually characters that don't produce output as |
266 | such but instead control the terminal somehow: for example newline and | |
267 | backspace are control characters. All characters with ord() less than | |
593df60c | 268 | 32 are most often classified as control characters (assuming ASCII, |
7be5a6cf JF |
269 | the ISO Latin character sets, and Unicode), as is the character with |
270 | the ord() value of 127 (C<DEL>). | |
b8c5462f JH |
271 | |
272 | =item graph | |
273 | ||
f1cbbd6e | 274 | Any alphanumeric or punctuation (special) character. |
b8c5462f JH |
275 | |
276 | =item print | |
277 | ||
f79b3095 | 278 | Any alphanumeric or punctuation (special) character or the space character. |
b8c5462f JH |
279 | |
280 | =item punct | |
281 | ||
f1cbbd6e | 282 | Any punctuation (special) character. |
b8c5462f JH |
283 | |
284 | =item xdigit | |
285 | ||
593df60c | 286 | Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would |
820475bd | 287 | work just fine) it is included for completeness. |
b8c5462f | 288 | |
b8c5462f JH |
289 | =back |
290 | ||
291 | You can negate the [::] character classes by prefixing the class name | |
292 | with a '^'. This is a Perl extension. For example: | |
293 | ||
93733859 JH |
294 | POSIX trad. Perl utf8 Perl |
295 | ||
296 | [:^digit:] \D \P{IsDigit} | |
297 | [:^space:] \S \P{IsSpace} | |
298 | [:^word:] \W \P{IsWord} | |
b8c5462f | 299 | |
26b44a0a JH |
300 | The POSIX character classes [.cc.] and [=cc=] are recognized but |
301 | B<not> supported and trying to use them will cause an error. | |
b8c5462f | 302 | |
a0d0e21e LW |
303 | Perl defines the following zero-width assertions: |
304 | ||
305 | \b Match a word boundary | |
306 | \B Match a non-(word boundary) | |
b85d18e9 IZ |
307 | \A Match only at beginning of string |
308 | \Z Match only at end of string, or before newline at the end | |
309 | \z Match only at end of string | |
9da458fc IZ |
310 | \G Match only at pos() (e.g. at the end-of-match position |
311 | of prior m//g) | |
a0d0e21e | 312 | |
14218588 | 313 | A word boundary (C<\b>) is a spot between two characters |
19799a22 GS |
314 | that has a C<\w> on one side of it and a C<\W> on the other side |
315 | of it (in either order), counting the imaginary characters off the | |
316 | beginning and end of the string as matching a C<\W>. (Within | |
317 | character classes C<\b> represents backspace rather than a word | |
318 | boundary, just as it normally does in any double-quoted string.) | |
319 | The C<\A> and C<\Z> are just like "^" and "$", except that they | |
320 | won't match multiple times when the C</m> modifier is used, while | |
321 | "^" and "$" will match at every internal line boundary. To match | |
322 | the actual end of the string and not ignore an optional trailing | |
323 | newline, use C<\z>. | |
324 | ||
325 | The C<\G> assertion can be used to chain global matches (using | |
326 | C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">. | |
327 | It is also useful when writing C<lex>-like scanners, when you have | |
328 | several patterns that you want to match against consequent substrings | |
329 | of your string, see the previous reference. The actual location | |
330 | where C<\G> will match can also be influenced by using C<pos()> as | |
331 | an lvalue. See L<perlfunc/pos>. | |
c47ff5f1 | 332 | |
14218588 | 333 | The bracketing construct C<( ... )> creates capture buffers. To |
c47ff5f1 | 334 | refer to the digit'th buffer use \<digit> within the |
14218588 | 335 | match. Outside the match use "$" instead of "\". (The |
c47ff5f1 | 336 | \<digit> notation works in certain circumstances outside |
14218588 GS |
337 | the match. See the warning below about \1 vs $1 for details.) |
338 | Referring back to another part of the match is called a | |
339 | I<backreference>. | |
340 | ||
341 | There is no limit to the number of captured substrings that you may | |
342 | use. However Perl also uses \10, \11, etc. as aliases for \010, | |
fb55449c JH |
343 | \011, etc. (Recall that 0 means octal, so \011 is the character at |
344 | number 9 in your coded character set; which would be the 10th character, | |
345 | a horizontal tab under ASCII.) Perl resolves this | |
346 | ambiguity by interpreting \10 as a backreference only if at least 10 | |
347 | left parentheses have opened before it. Likewise \11 is a | |
348 | backreference only if at least 11 left parentheses have opened | |
349 | before it. And so on. \1 through \9 are always interpreted as | |
350 | backreferences. | |
14218588 GS |
351 | |
352 | Examples: | |
a0d0e21e LW |
353 | |
354 | s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words | |
355 | ||
14218588 GS |
356 | if (/(.)\1/) { # find first doubled char |
357 | print "'$1' is the first doubled character\n"; | |
358 | } | |
c47ff5f1 | 359 | |
14218588 | 360 | if (/Time: (..):(..):(..)/) { # parse out values |
a0d0e21e LW |
361 | $hours = $1; |
362 | $minutes = $2; | |
363 | $seconds = $3; | |
364 | } | |
c47ff5f1 | 365 | |
14218588 GS |
366 | Several special variables also refer back to portions of the previous |
367 | match. C<$+> returns whatever the last bracket match matched. | |
368 | C<$&> returns the entire matched string. (At one point C<$0> did | |
369 | also, but now it returns the name of the program.) C<$`> returns | |
370 | everything before the matched string. And C<$'> returns everything | |
371 | after the matched string. | |
372 | ||
373 | The numbered variables ($1, $2, $3, etc.) and the related punctuation | |
262ffc8b | 374 | set (C<$+>, C<$&>, C<$`>, and C<$'>) are all dynamically scoped |
14218588 GS |
375 | until the end of the enclosing block or until the next successful |
376 | match, whichever comes first. (See L<perlsyn/"Compound Statements">.) | |
377 | ||
378 | B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or | |
379 | C<$'> anywhere in the program, it has to provide them for every | |
380 | pattern match. This may substantially slow your program. Perl | |
381 | uses the same mechanism to produce $1, $2, etc, so you also pay a | |
382 | price for each pattern that contains capturing parentheses. (To | |
383 | avoid this cost while retaining the grouping behaviour, use the | |
384 | extended regular expression C<(?: ... )> instead.) But if you never | |
385 | use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing | |
386 | parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`> | |
387 | if you can, but if you can't (and some algorithms really appreciate | |
388 | them), once you've used them once, use them at will, because you've | |
389 | already paid the price. As of 5.005, C<$&> is not so costly as the | |
390 | other two. | |
68dc0745 | 391 | |
19799a22 GS |
392 | Backslashed metacharacters in Perl are alphanumeric, such as C<\b>, |
393 | C<\w>, C<\n>. Unlike some other regular expression languages, there | |
394 | are no backslashed symbols that aren't alphanumeric. So anything | |
c47ff5f1 | 395 | that looks like \\, \(, \), \<, \>, \{, or \} is always |
19799a22 GS |
396 | interpreted as a literal character, not a metacharacter. This was |
397 | once used in a common idiom to disable or quote the special meanings | |
398 | of regular expression metacharacters in a string that you want to | |
36bbe248 | 399 | use for a pattern. Simply quote all non-"word" characters: |
a0d0e21e LW |
400 | |
401 | $pattern =~ s/(\W)/\\$1/g; | |
402 | ||
f1cbbd6e | 403 | (If C<use locale> is set, then this depends on the current locale.) |
14218588 GS |
404 | Today it is more common to use the quotemeta() function or the C<\Q> |
405 | metaquoting escape sequence to disable all metacharacters' special | |
406 | meanings like this: | |
a0d0e21e LW |
407 | |
408 | /$unquoted\Q$quoted\E$unquoted/ | |
409 | ||
9da458fc IZ |
410 | Beware that if you put literal backslashes (those not inside |
411 | interpolated variables) between C<\Q> and C<\E>, double-quotish | |
412 | backslash interpolation may lead to confusing results. If you | |
413 | I<need> to use literal backslashes within C<\Q...\E>, | |
414 | consult L<perlop/"Gory details of parsing quoted constructs">. | |
415 | ||
19799a22 GS |
416 | =head2 Extended Patterns |
417 | ||
14218588 GS |
418 | Perl also defines a consistent extension syntax for features not |
419 | found in standard tools like B<awk> and B<lex>. The syntax is a | |
420 | pair of parentheses with a question mark as the first thing within | |
421 | the parentheses. The character after the question mark indicates | |
422 | the extension. | |
19799a22 | 423 | |
14218588 GS |
424 | The stability of these extensions varies widely. Some have been |
425 | part of the core language for many years. Others are experimental | |
426 | and may change without warning or be completely removed. Check | |
427 | the documentation on an individual feature to verify its current | |
428 | status. | |
19799a22 | 429 | |
14218588 GS |
430 | A question mark was chosen for this and for the minimal-matching |
431 | construct because 1) question marks are rare in older regular | |
432 | expressions, and 2) whenever you see one, you should stop and | |
433 | "question" exactly what is going on. That's psychology... | |
a0d0e21e LW |
434 | |
435 | =over 10 | |
436 | ||
cc6b7395 | 437 | =item C<(?#text)> |
a0d0e21e | 438 | |
14218588 | 439 | A comment. The text is ignored. If the C</x> modifier enables |
19799a22 | 440 | whitespace formatting, a simple C<#> will suffice. Note that Perl closes |
259138e3 GS |
441 | the comment as soon as it sees a C<)>, so there is no way to put a literal |
442 | C<)> in the comment. | |
a0d0e21e | 443 | |
19799a22 GS |
444 | =item C<(?imsx-imsx)> |
445 | ||
446 | One or more embedded pattern-match modifiers. This is particularly | |
447 | useful for dynamic patterns, such as those read in from a configuration | |
448 | file, read in as an argument, are specified in a table somewhere, | |
449 | etc. Consider the case that some of which want to be case sensitive | |
450 | and some do not. The case insensitive ones need to include merely | |
451 | C<(?i)> at the front of the pattern. For example: | |
452 | ||
453 | $pattern = "foobar"; | |
454 | if ( /$pattern/i ) { } | |
455 | ||
456 | # more flexible: | |
457 | ||
458 | $pattern = "(?i)foobar"; | |
459 | if ( /$pattern/ ) { } | |
460 | ||
461 | Letters after a C<-> turn those modifiers off. These modifiers are | |
462 | localized inside an enclosing group (if any). For example, | |
463 | ||
464 | ( (?i) blah ) \s+ \1 | |
465 | ||
466 | will match a repeated (I<including the case>!) word C<blah> in any | |
14218588 | 467 | case, assuming C<x> modifier, and no C<i> modifier outside this |
19799a22 GS |
468 | group. |
469 | ||
5a964f20 | 470 | =item C<(?:pattern)> |
a0d0e21e | 471 | |
ca9dfc88 IZ |
472 | =item C<(?imsx-imsx:pattern)> |
473 | ||
5a964f20 TC |
474 | This is for clustering, not capturing; it groups subexpressions like |
475 | "()", but doesn't make backreferences as "()" does. So | |
a0d0e21e | 476 | |
5a964f20 | 477 | @fields = split(/\b(?:a|b|c)\b/) |
a0d0e21e LW |
478 | |
479 | is like | |
480 | ||
5a964f20 | 481 | @fields = split(/\b(a|b|c)\b/) |
a0d0e21e | 482 | |
19799a22 GS |
483 | but doesn't spit out extra fields. It's also cheaper not to capture |
484 | characters if you don't need to. | |
a0d0e21e | 485 | |
19799a22 GS |
486 | Any letters between C<?> and C<:> act as flags modifiers as with |
487 | C<(?imsx-imsx)>. For example, | |
ca9dfc88 IZ |
488 | |
489 | /(?s-i:more.*than).*million/i | |
490 | ||
14218588 | 491 | is equivalent to the more verbose |
ca9dfc88 IZ |
492 | |
493 | /(?:(?s-i)more.*than).*million/i | |
494 | ||
5a964f20 | 495 | =item C<(?=pattern)> |
a0d0e21e | 496 | |
19799a22 | 497 | A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/> |
a0d0e21e LW |
498 | matches a word followed by a tab, without including the tab in C<$&>. |
499 | ||
5a964f20 | 500 | =item C<(?!pattern)> |
a0d0e21e | 501 | |
19799a22 | 502 | A zero-width negative look-ahead assertion. For example C</foo(?!bar)/> |
a0d0e21e | 503 | matches any occurrence of "foo" that isn't followed by "bar". Note |
19799a22 GS |
504 | however that look-ahead and look-behind are NOT the same thing. You cannot |
505 | use this for look-behind. | |
7b8d334a | 506 | |
5a964f20 | 507 | If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/> |
7b8d334a GS |
508 | will not do what you want. That's because the C<(?!foo)> is just saying that |
509 | the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will | |
510 | match. You would have to do something like C</(?!foo)...bar/> for that. We | |
511 | say "like" because there's the case of your "bar" not having three characters | |
512 | before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>. | |
513 | Sometimes it's still easier just to say: | |
a0d0e21e | 514 | |
a3cb178b | 515 | if (/bar/ && $` !~ /foo$/) |
a0d0e21e | 516 | |
19799a22 | 517 | For look-behind see below. |
c277df42 | 518 | |
c47ff5f1 | 519 | =item C<(?<=pattern)> |
c277df42 | 520 | |
c47ff5f1 | 521 | A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/> |
19799a22 GS |
522 | matches a word that follows a tab, without including the tab in C<$&>. |
523 | Works only for fixed-width look-behind. | |
c277df42 | 524 | |
5a964f20 | 525 | =item C<(?<!pattern)> |
c277df42 | 526 | |
19799a22 GS |
527 | A zero-width negative look-behind assertion. For example C</(?<!bar)foo/> |
528 | matches any occurrence of "foo" that does not follow "bar". Works | |
529 | only for fixed-width look-behind. | |
c277df42 | 530 | |
cc6b7395 | 531 | =item C<(?{ code })> |
c277df42 | 532 | |
19799a22 GS |
533 | B<WARNING>: This extended regular expression feature is considered |
534 | highly experimental, and may be changed or deleted without notice. | |
c277df42 | 535 | |
19799a22 GS |
536 | This zero-width assertion evaluate any embedded Perl code. It |
537 | always succeeds, and its C<code> is not interpolated. Currently, | |
538 | the rules to determine where the C<code> ends are somewhat convoluted. | |
539 | ||
540 | The C<code> is properly scoped in the following sense: If the assertion | |
541 | is backtracked (compare L<"Backtracking">), all changes introduced after | |
542 | C<local>ization are undone, so that | |
b9ac3b5b GS |
543 | |
544 | $_ = 'a' x 8; | |
545 | m< | |
546 | (?{ $cnt = 0 }) # Initialize $cnt. | |
547 | ( | |
548 | a | |
549 | (?{ | |
550 | local $cnt = $cnt + 1; # Update $cnt, backtracking-safe. | |
551 | }) | |
552 | )* | |
553 | aaaa | |
554 | (?{ $res = $cnt }) # On success copy to non-localized | |
555 | # location. | |
556 | >x; | |
557 | ||
19799a22 | 558 | will set C<$res = 4>. Note that after the match, $cnt returns to the globally |
14218588 | 559 | introduced value, because the scopes that restrict C<local> operators |
b9ac3b5b GS |
560 | are unwound. |
561 | ||
19799a22 GS |
562 | This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)> |
563 | switch. If I<not> used in this way, the result of evaluation of | |
564 | C<code> is put into the special variable C<$^R>. This happens | |
565 | immediately, so C<$^R> can be used from other C<(?{ code })> assertions | |
566 | inside the same regular expression. | |
b9ac3b5b | 567 | |
19799a22 GS |
568 | The assignment to C<$^R> above is properly localized, so the old |
569 | value of C<$^R> is restored if the assertion is backtracked; compare | |
570 | L<"Backtracking">. | |
b9ac3b5b | 571 | |
19799a22 GS |
572 | For reasons of security, this construct is forbidden if the regular |
573 | expression involves run-time interpolation of variables, unless the | |
574 | perilous C<use re 'eval'> pragma has been used (see L<re>), or the | |
575 | variables contain results of C<qr//> operator (see | |
576 | L<perlop/"qr/STRING/imosx">). | |
871b0233 | 577 | |
14218588 | 578 | This restriction is because of the wide-spread and remarkably convenient |
19799a22 | 579 | custom of using run-time determined strings as patterns. For example: |
871b0233 IZ |
580 | |
581 | $re = <>; | |
582 | chomp $re; | |
583 | $string =~ /$re/; | |
584 | ||
14218588 GS |
585 | Before Perl knew how to execute interpolated code within a pattern, |
586 | this operation was completely safe from a security point of view, | |
587 | although it could raise an exception from an illegal pattern. If | |
588 | you turn on the C<use re 'eval'>, though, it is no longer secure, | |
589 | so you should only do so if you are also using taint checking. | |
590 | Better yet, use the carefully constrained evaluation within a Safe | |
591 | module. See L<perlsec> for details about both these mechanisms. | |
871b0233 | 592 | |
14455d6c | 593 | =item C<(??{ code })> |
0f5d15d6 | 594 | |
19799a22 GS |
595 | B<WARNING>: This extended regular expression feature is considered |
596 | highly experimental, and may be changed or deleted without notice. | |
9da458fc IZ |
597 | A simplified version of the syntax may be introduced for commonly |
598 | used idioms. | |
0f5d15d6 | 599 | |
19799a22 GS |
600 | This is a "postponed" regular subexpression. The C<code> is evaluated |
601 | at run time, at the moment this subexpression may match. The result | |
602 | of evaluation is considered as a regular expression and matched as | |
603 | if it were inserted instead of this construct. | |
0f5d15d6 | 604 | |
428594d9 | 605 | The C<code> is not interpolated. As before, the rules to determine |
19799a22 GS |
606 | where the C<code> ends are currently somewhat convoluted. |
607 | ||
608 | The following pattern matches a parenthesized group: | |
0f5d15d6 IZ |
609 | |
610 | $re = qr{ | |
611 | \( | |
612 | (?: | |
613 | (?> [^()]+ ) # Non-parens without backtracking | |
614 | | | |
14455d6c | 615 | (??{ $re }) # Group with matching parens |
0f5d15d6 IZ |
616 | )* |
617 | \) | |
618 | }x; | |
619 | ||
c47ff5f1 | 620 | =item C<< (?>pattern) >> |
5a964f20 | 621 | |
19799a22 GS |
622 | B<WARNING>: This extended regular expression feature is considered |
623 | highly experimental, and may be changed or deleted without notice. | |
624 | ||
625 | An "independent" subexpression, one which matches the substring | |
626 | that a I<standalone> C<pattern> would match if anchored at the given | |
9da458fc | 627 | position, and it matches I<nothing other than this substring>. This |
19799a22 GS |
628 | construct is useful for optimizations of what would otherwise be |
629 | "eternal" matches, because it will not backtrack (see L<"Backtracking">). | |
9da458fc IZ |
630 | It may also be useful in places where the "grab all you can, and do not |
631 | give anything back" semantic is desirable. | |
19799a22 | 632 | |
c47ff5f1 | 633 | For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >> |
19799a22 GS |
634 | (anchored at the beginning of string, as above) will match I<all> |
635 | characters C<a> at the beginning of string, leaving no C<a> for | |
636 | C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>, | |
637 | since the match of the subgroup C<a*> is influenced by the following | |
638 | group C<ab> (see L<"Backtracking">). In particular, C<a*> inside | |
639 | C<a*ab> will match fewer characters than a standalone C<a*>, since | |
640 | this makes the tail match. | |
641 | ||
c47ff5f1 | 642 | An effect similar to C<< (?>pattern) >> may be achieved by writing |
19799a22 GS |
643 | C<(?=(pattern))\1>. This matches the same substring as a standalone |
644 | C<a+>, and the following C<\1> eats the matched string; it therefore | |
c47ff5f1 | 645 | makes a zero-length assertion into an analogue of C<< (?>...) >>. |
19799a22 GS |
646 | (The difference between these two constructs is that the second one |
647 | uses a capturing group, thus shifting ordinals of backreferences | |
648 | in the rest of a regular expression.) | |
649 | ||
650 | Consider this pattern: | |
c277df42 | 651 | |
871b0233 IZ |
652 | m{ \( |
653 | ( | |
9da458fc | 654 | [^()]+ # x+ |
871b0233 IZ |
655 | | |
656 | \( [^()]* \) | |
657 | )+ | |
658 | \) | |
659 | }x | |
5a964f20 | 660 | |
19799a22 GS |
661 | That will efficiently match a nonempty group with matching parentheses |
662 | two levels deep or less. However, if there is no such group, it | |
663 | will take virtually forever on a long string. That's because there | |
664 | are so many different ways to split a long string into several | |
665 | substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar | |
666 | to a subpattern of the above pattern. Consider how the pattern | |
667 | above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several | |
668 | seconds, but that each extra letter doubles this time. This | |
669 | exponential performance will make it appear that your program has | |
14218588 | 670 | hung. However, a tiny change to this pattern |
5a964f20 | 671 | |
871b0233 IZ |
672 | m{ \( |
673 | ( | |
9da458fc | 674 | (?> [^()]+ ) # change x+ above to (?> x+ ) |
871b0233 IZ |
675 | | |
676 | \( [^()]* \) | |
677 | )+ | |
678 | \) | |
679 | }x | |
c277df42 | 680 | |
c47ff5f1 | 681 | which uses C<< (?>...) >> matches exactly when the one above does (verifying |
5a964f20 TC |
682 | this yourself would be a productive exercise), but finishes in a fourth |
683 | the time when used on a similar string with 1000000 C<a>s. Be aware, | |
684 | however, that this pattern currently triggers a warning message under | |
9f1b1f2d | 685 | the C<use warnings> pragma or B<-w> switch saying it |
6bab786b | 686 | C<"matches null string many times in regex">. |
c277df42 | 687 | |
c47ff5f1 | 688 | On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable |
19799a22 | 689 | effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>. |
c277df42 IZ |
690 | This was only 4 times slower on a string with 1000000 C<a>s. |
691 | ||
9da458fc IZ |
692 | The "grab all you can, and do not give anything back" semantic is desirable |
693 | in many situations where on the first sight a simple C<()*> looks like | |
694 | the correct solution. Suppose we parse text with comments being delimited | |
695 | by C<#> followed by some optional (horizontal) whitespace. Contrary to | |
4375e838 | 696 | its appearance, C<#[ \t]*> I<is not> the correct subexpression to match |
9da458fc IZ |
697 | the comment delimiter, because it may "give up" some whitespace if |
698 | the remainder of the pattern can be made to match that way. The correct | |
699 | answer is either one of these: | |
700 | ||
701 | (?>#[ \t]*) | |
702 | #[ \t]*(?![ \t]) | |
703 | ||
704 | For example, to grab non-empty comments into $1, one should use either | |
705 | one of these: | |
706 | ||
707 | / (?> \# [ \t]* ) ( .+ ) /x; | |
708 | / \# [ \t]* ( [^ \t] .* ) /x; | |
709 | ||
710 | Which one you pick depends on which of these expressions better reflects | |
711 | the above specification of comments. | |
712 | ||
5a964f20 | 713 | =item C<(?(condition)yes-pattern|no-pattern)> |
c277df42 | 714 | |
5a964f20 | 715 | =item C<(?(condition)yes-pattern)> |
c277df42 | 716 | |
19799a22 GS |
717 | B<WARNING>: This extended regular expression feature is considered |
718 | highly experimental, and may be changed or deleted without notice. | |
719 | ||
c277df42 IZ |
720 | Conditional expression. C<(condition)> should be either an integer in |
721 | parentheses (which is valid if the corresponding pair of parentheses | |
19799a22 | 722 | matched), or look-ahead/look-behind/evaluate zero-width assertion. |
c277df42 | 723 | |
19799a22 | 724 | For example: |
c277df42 | 725 | |
5a964f20 | 726 | m{ ( \( )? |
871b0233 | 727 | [^()]+ |
5a964f20 | 728 | (?(1) \) ) |
871b0233 | 729 | }x |
c277df42 IZ |
730 | |
731 | matches a chunk of non-parentheses, possibly included in parentheses | |
732 | themselves. | |
a0d0e21e | 733 | |
a0d0e21e LW |
734 | =back |
735 | ||
c07a80fd | 736 | =head2 Backtracking |
737 | ||
35a734be IZ |
738 | NOTE: This section presents an abstract approximation of regular |
739 | expression behavior. For a more rigorous (and complicated) view of | |
740 | the rules involved in selecting a match among possible alternatives, | |
741 | see L<Combining pieces together>. | |
742 | ||
c277df42 | 743 | A fundamental feature of regular expression matching involves the |
5a964f20 | 744 | notion called I<backtracking>, which is currently used (when needed) |
c277df42 | 745 | by all regular expression quantifiers, namely C<*>, C<*?>, C<+>, |
9da458fc IZ |
746 | C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized |
747 | internally, but the general principle outlined here is valid. | |
c07a80fd | 748 | |
749 | For a regular expression to match, the I<entire> regular expression must | |
750 | match, not just part of it. So if the beginning of a pattern containing a | |
751 | quantifier succeeds in a way that causes later parts in the pattern to | |
752 | fail, the matching engine backs up and recalculates the beginning | |
753 | part--that's why it's called backtracking. | |
754 | ||
755 | Here is an example of backtracking: Let's say you want to find the | |
756 | word following "foo" in the string "Food is on the foo table.": | |
757 | ||
758 | $_ = "Food is on the foo table."; | |
759 | if ( /\b(foo)\s+(\w+)/i ) { | |
760 | print "$2 follows $1.\n"; | |
761 | } | |
762 | ||
763 | When the match runs, the first part of the regular expression (C<\b(foo)>) | |
764 | finds a possible match right at the beginning of the string, and loads up | |
765 | $1 with "Foo". However, as soon as the matching engine sees that there's | |
766 | no whitespace following the "Foo" that it had saved in $1, it realizes its | |
68dc0745 | 767 | mistake and starts over again one character after where it had the |
c07a80fd | 768 | tentative match. This time it goes all the way until the next occurrence |
769 | of "foo". The complete regular expression matches this time, and you get | |
770 | the expected output of "table follows foo." | |
771 | ||
772 | Sometimes minimal matching can help a lot. Imagine you'd like to match | |
773 | everything between "foo" and "bar". Initially, you write something | |
774 | like this: | |
775 | ||
776 | $_ = "The food is under the bar in the barn."; | |
777 | if ( /foo(.*)bar/ ) { | |
778 | print "got <$1>\n"; | |
779 | } | |
780 | ||
781 | Which perhaps unexpectedly yields: | |
782 | ||
783 | got <d is under the bar in the > | |
784 | ||
785 | That's because C<.*> was greedy, so you get everything between the | |
14218588 | 786 | I<first> "foo" and the I<last> "bar". Here it's more effective |
c07a80fd | 787 | to use minimal matching to make sure you get the text between a "foo" |
788 | and the first "bar" thereafter. | |
789 | ||
790 | if ( /foo(.*?)bar/ ) { print "got <$1>\n" } | |
791 | got <d is under the > | |
792 | ||
793 | Here's another example: let's say you'd like to match a number at the end | |
b6e13d97 | 794 | of a string, and you also want to keep the preceding part of the match. |
c07a80fd | 795 | So you write this: |
796 | ||
797 | $_ = "I have 2 numbers: 53147"; | |
798 | if ( /(.*)(\d*)/ ) { # Wrong! | |
799 | print "Beginning is <$1>, number is <$2>.\n"; | |
800 | } | |
801 | ||
802 | That won't work at all, because C<.*> was greedy and gobbled up the | |
803 | whole string. As C<\d*> can match on an empty string the complete | |
804 | regular expression matched successfully. | |
805 | ||
8e1088bc | 806 | Beginning is <I have 2 numbers: 53147>, number is <>. |
c07a80fd | 807 | |
808 | Here are some variants, most of which don't work: | |
809 | ||
810 | $_ = "I have 2 numbers: 53147"; | |
811 | @pats = qw{ | |
812 | (.*)(\d*) | |
813 | (.*)(\d+) | |
814 | (.*?)(\d*) | |
815 | (.*?)(\d+) | |
816 | (.*)(\d+)$ | |
817 | (.*?)(\d+)$ | |
818 | (.*)\b(\d+)$ | |
819 | (.*\D)(\d+)$ | |
820 | }; | |
821 | ||
822 | for $pat (@pats) { | |
823 | printf "%-12s ", $pat; | |
824 | if ( /$pat/ ) { | |
825 | print "<$1> <$2>\n"; | |
826 | } else { | |
827 | print "FAIL\n"; | |
828 | } | |
829 | } | |
830 | ||
831 | That will print out: | |
832 | ||
833 | (.*)(\d*) <I have 2 numbers: 53147> <> | |
834 | (.*)(\d+) <I have 2 numbers: 5314> <7> | |
835 | (.*?)(\d*) <> <> | |
836 | (.*?)(\d+) <I have > <2> | |
837 | (.*)(\d+)$ <I have 2 numbers: 5314> <7> | |
838 | (.*?)(\d+)$ <I have 2 numbers: > <53147> | |
839 | (.*)\b(\d+)$ <I have 2 numbers: > <53147> | |
840 | (.*\D)(\d+)$ <I have 2 numbers: > <53147> | |
841 | ||
842 | As you see, this can be a bit tricky. It's important to realize that a | |
843 | regular expression is merely a set of assertions that gives a definition | |
844 | of success. There may be 0, 1, or several different ways that the | |
845 | definition might succeed against a particular string. And if there are | |
5a964f20 TC |
846 | multiple ways it might succeed, you need to understand backtracking to |
847 | know which variety of success you will achieve. | |
c07a80fd | 848 | |
19799a22 | 849 | When using look-ahead assertions and negations, this can all get even |
54310121 | 850 | tricker. Imagine you'd like to find a sequence of non-digits not |
c07a80fd | 851 | followed by "123". You might try to write that as |
852 | ||
871b0233 IZ |
853 | $_ = "ABC123"; |
854 | if ( /^\D*(?!123)/ ) { # Wrong! | |
855 | print "Yup, no 123 in $_\n"; | |
856 | } | |
c07a80fd | 857 | |
858 | But that isn't going to match; at least, not the way you're hoping. It | |
859 | claims that there is no 123 in the string. Here's a clearer picture of | |
9b9391b2 | 860 | why that pattern matches, contrary to popular expectations: |
c07a80fd | 861 | |
862 | $x = 'ABC123' ; | |
863 | $y = 'ABC445' ; | |
864 | ||
865 | print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ; | |
866 | print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ; | |
867 | ||
868 | print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ; | |
869 | print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ; | |
870 | ||
871 | This prints | |
872 | ||
873 | 2: got ABC | |
874 | 3: got AB | |
875 | 4: got ABC | |
876 | ||
5f05dabc | 877 | You might have expected test 3 to fail because it seems to a more |
c07a80fd | 878 | general purpose version of test 1. The important difference between |
879 | them is that test 3 contains a quantifier (C<\D*>) and so can use | |
880 | backtracking, whereas test 1 will not. What's happening is | |
881 | that you've asked "Is it true that at the start of $x, following 0 or more | |
5f05dabc | 882 | non-digits, you have something that's not 123?" If the pattern matcher had |
c07a80fd | 883 | let C<\D*> expand to "ABC", this would have caused the whole pattern to |
54310121 | 884 | fail. |
14218588 | 885 | |
c07a80fd | 886 | The search engine will initially match C<\D*> with "ABC". Then it will |
14218588 | 887 | try to match C<(?!123> with "123", which fails. But because |
c07a80fd | 888 | a quantifier (C<\D*>) has been used in the regular expression, the |
889 | search engine can backtrack and retry the match differently | |
54310121 | 890 | in the hope of matching the complete regular expression. |
c07a80fd | 891 | |
5a964f20 TC |
892 | The pattern really, I<really> wants to succeed, so it uses the |
893 | standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this | |
c07a80fd | 894 | time. Now there's indeed something following "AB" that is not |
14218588 | 895 | "123". It's "C123", which suffices. |
c07a80fd | 896 | |
14218588 GS |
897 | We can deal with this by using both an assertion and a negation. |
898 | We'll say that the first part in $1 must be followed both by a digit | |
899 | and by something that's not "123". Remember that the look-aheads | |
900 | are zero-width expressions--they only look, but don't consume any | |
901 | of the string in their match. So rewriting this way produces what | |
c07a80fd | 902 | you'd expect; that is, case 5 will fail, but case 6 succeeds: |
903 | ||
904 | print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ; | |
905 | print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ; | |
906 | ||
907 | 6: got ABC | |
908 | ||
5a964f20 | 909 | In other words, the two zero-width assertions next to each other work as though |
19799a22 | 910 | they're ANDed together, just as you'd use any built-in assertions: C</^$/> |
c07a80fd | 911 | matches only if you're at the beginning of the line AND the end of the |
912 | line simultaneously. The deeper underlying truth is that juxtaposition in | |
913 | regular expressions always means AND, except when you write an explicit OR | |
914 | using the vertical bar. C</ab/> means match "a" AND (then) match "b", | |
915 | although the attempted matches are made at different positions because "a" | |
916 | is not a zero-width assertion, but a one-width assertion. | |
917 | ||
19799a22 | 918 | B<WARNING>: particularly complicated regular expressions can take |
14218588 | 919 | exponential time to solve because of the immense number of possible |
9da458fc IZ |
920 | ways they can use backtracking to try match. For example, without |
921 | internal optimizations done by the regular expression engine, this will | |
922 | take a painfully long time to run: | |
c07a80fd | 923 | |
e1901655 IZ |
924 | 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/ |
925 | ||
926 | And if you used C<*>'s in the internal groups instead of limiting them | |
927 | to 0 through 5 matches, then it would take forever--or until you ran | |
928 | out of stack space. Moreover, these internal optimizations are not | |
929 | always applicable. For example, if you put C<{0,5}> instead of C<*> | |
930 | on the external group, no current optimization is applicable, and the | |
931 | match takes a long time to finish. | |
c07a80fd | 932 | |
9da458fc IZ |
933 | A powerful tool for optimizing such beasts is what is known as an |
934 | "independent group", | |
c47ff5f1 | 935 | which does not backtrack (see L<C<< (?>pattern) >>>). Note also that |
9da458fc | 936 | zero-length look-ahead/look-behind assertions will not backtrack to make |
14218588 GS |
937 | the tail match, since they are in "logical" context: only |
938 | whether they match is considered relevant. For an example | |
9da458fc | 939 | where side-effects of look-ahead I<might> have influenced the |
c47ff5f1 | 940 | following match, see L<C<< (?>pattern) >>>. |
c277df42 | 941 | |
a0d0e21e LW |
942 | =head2 Version 8 Regular Expressions |
943 | ||
5a964f20 | 944 | In case you're not familiar with the "regular" Version 8 regex |
a0d0e21e LW |
945 | routines, here are the pattern-matching rules not described above. |
946 | ||
54310121 | 947 | Any single character matches itself, unless it is a I<metacharacter> |
a0d0e21e | 948 | with a special meaning described here or above. You can cause |
5a964f20 | 949 | characters that normally function as metacharacters to be interpreted |
5f05dabc | 950 | literally by prefixing them with a "\" (e.g., "\." matches a ".", not any |
a0d0e21e LW |
951 | character; "\\" matches a "\"). A series of characters matches that |
952 | series of characters in the target string, so the pattern C<blurfl> | |
953 | would match "blurfl" in the target string. | |
954 | ||
955 | You can specify a character class, by enclosing a list of characters | |
5a964f20 | 956 | in C<[]>, which will match any one character from the list. If the |
a0d0e21e | 957 | first character after the "[" is "^", the class matches any character not |
14218588 | 958 | in the list. Within a list, the "-" character specifies a |
5a964f20 | 959 | range, so that C<a-z> represents all characters between "a" and "z", |
8a4f6ac2 GS |
960 | inclusive. If you want either "-" or "]" itself to be a member of a |
961 | class, put it at the start of the list (possibly after a "^"), or | |
962 | escape it with a backslash. "-" is also taken literally when it is | |
963 | at the end of the list, just before the closing "]". (The | |
84850974 DD |
964 | following all specify the same class of three characters: C<[-az]>, |
965 | C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which | |
fb55449c JH |
966 | specifies a class containing twenty-six characters, even on EBCDIC |
967 | based coded character sets.) Also, if you try to use the character | |
968 | classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of | |
969 | a range, that's not a range, the "-" is understood literally. | |
a0d0e21e | 970 | |
8ada0baa JH |
971 | Note also that the whole range idea is rather unportable between |
972 | character sets--and even within character sets they may cause results | |
973 | you probably didn't expect. A sound principle is to use only ranges | |
974 | that begin from and end at either alphabets of equal case ([a-e], | |
975 | [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt, | |
976 | spell out the character sets in full. | |
977 | ||
54310121 | 978 | Characters may be specified using a metacharacter syntax much like that |
a0d0e21e LW |
979 | used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return, |
980 | "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string | |
fb55449c JH |
981 | of octal digits, matches the character whose coded character set value |
982 | is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits, | |
983 | matches the character whose numeric value is I<nn>. The expression \cI<x> | |
984 | matches the character control-I<x>. Finally, the "." metacharacter | |
985 | matches any character except "\n" (unless you use C</s>). | |
a0d0e21e LW |
986 | |
987 | You can specify a series of alternatives for a pattern using "|" to | |
988 | separate them, so that C<fee|fie|foe> will match any of "fee", "fie", | |
5a964f20 | 989 | or "foe" in the target string (as would C<f(e|i|o)e>). The |
a0d0e21e LW |
990 | first alternative includes everything from the last pattern delimiter |
991 | ("(", "[", or the beginning of the pattern) up to the first "|", and | |
992 | the last alternative contains everything from the last "|" to the next | |
14218588 GS |
993 | pattern delimiter. That's why it's common practice to include |
994 | alternatives in parentheses: to minimize confusion about where they | |
a3cb178b GS |
995 | start and end. |
996 | ||
5a964f20 | 997 | Alternatives are tried from left to right, so the first |
a3cb178b GS |
998 | alternative found for which the entire expression matches, is the one that |
999 | is chosen. This means that alternatives are not necessarily greedy. For | |
628afcb5 | 1000 | example: when matching C<foo|foot> against "barefoot", only the "foo" |
a3cb178b GS |
1001 | part will match, as that is the first alternative tried, and it successfully |
1002 | matches the target string. (This might not seem important, but it is | |
1003 | important when you are capturing matched text using parentheses.) | |
1004 | ||
5a964f20 | 1005 | Also remember that "|" is interpreted as a literal within square brackets, |
a3cb178b | 1006 | so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>. |
a0d0e21e | 1007 | |
14218588 GS |
1008 | Within a pattern, you may designate subpatterns for later reference |
1009 | by enclosing them in parentheses, and you may refer back to the | |
1010 | I<n>th subpattern later in the pattern using the metacharacter | |
1011 | \I<n>. Subpatterns are numbered based on the left to right order | |
1012 | of their opening parenthesis. A backreference matches whatever | |
1013 | actually matched the subpattern in the string being examined, not | |
1014 | the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will | |
1015 | match "0x1234 0x4321", but not "0x1234 01234", because subpattern | |
1016 | 1 matched "0x", even though the rule C<0|0x> could potentially match | |
1017 | the leading 0 in the second number. | |
cb1a09d0 | 1018 | |
19799a22 | 1019 | =head2 Warning on \1 vs $1 |
cb1a09d0 | 1020 | |
5a964f20 | 1021 | Some people get too used to writing things like: |
cb1a09d0 AD |
1022 | |
1023 | $pattern =~ s/(\W)/\\\1/g; | |
1024 | ||
1025 | This is grandfathered for the RHS of a substitute to avoid shocking the | |
1026 | B<sed> addicts, but it's a dirty habit to get into. That's because in | |
5f05dabc | 1027 | PerlThink, the righthand side of a C<s///> is a double-quoted string. C<\1> in |
cb1a09d0 AD |
1028 | the usual double-quoted string means a control-A. The customary Unix |
1029 | meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit | |
1030 | of doing that, you get yourself into trouble if you then add an C</e> | |
1031 | modifier. | |
1032 | ||
5a964f20 | 1033 | s/(\d+)/ \1 + 1 /eg; # causes warning under -w |
cb1a09d0 AD |
1034 | |
1035 | Or if you try to do | |
1036 | ||
1037 | s/(\d+)/\1000/; | |
1038 | ||
1039 | You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with | |
14218588 | 1040 | C<${1}000>. The operation of interpolation should not be confused |
cb1a09d0 AD |
1041 | with the operation of matching a backreference. Certainly they mean two |
1042 | different things on the I<left> side of the C<s///>. | |
9fa51da4 | 1043 | |
c84d73f1 IZ |
1044 | =head2 Repeated patterns matching zero-length substring |
1045 | ||
19799a22 | 1046 | B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite. |
c84d73f1 IZ |
1047 | |
1048 | Regular expressions provide a terse and powerful programming language. As | |
1049 | with most other power tools, power comes together with the ability | |
1050 | to wreak havoc. | |
1051 | ||
1052 | A common abuse of this power stems from the ability to make infinite | |
628afcb5 | 1053 | loops using regular expressions, with something as innocuous as: |
c84d73f1 IZ |
1054 | |
1055 | 'foo' =~ m{ ( o? )* }x; | |
1056 | ||
1057 | The C<o?> can match at the beginning of C<'foo'>, and since the position | |
1058 | in the string is not moved by the match, C<o?> would match again and again | |
14218588 | 1059 | because of the C<*> modifier. Another common way to create a similar cycle |
c84d73f1 IZ |
1060 | is with the looping modifier C<//g>: |
1061 | ||
1062 | @matches = ( 'foo' =~ m{ o? }xg ); | |
1063 | ||
1064 | or | |
1065 | ||
1066 | print "match: <$&>\n" while 'foo' =~ m{ o? }xg; | |
1067 | ||
1068 | or the loop implied by split(). | |
1069 | ||
1070 | However, long experience has shown that many programming tasks may | |
14218588 GS |
1071 | be significantly simplified by using repeated subexpressions that |
1072 | may match zero-length substrings. Here's a simple example being: | |
c84d73f1 IZ |
1073 | |
1074 | @chars = split //, $string; # // is not magic in split | |
1075 | ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// / | |
1076 | ||
9da458fc | 1077 | Thus Perl allows such constructs, by I<forcefully breaking |
c84d73f1 IZ |
1078 | the infinite loop>. The rules for this are different for lower-level |
1079 | loops given by the greedy modifiers C<*+{}>, and for higher-level | |
1080 | ones like the C</g> modifier or split() operator. | |
1081 | ||
19799a22 GS |
1082 | The lower-level loops are I<interrupted> (that is, the loop is |
1083 | broken) when Perl detects that a repeated expression matched a | |
1084 | zero-length substring. Thus | |
c84d73f1 IZ |
1085 | |
1086 | m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x; | |
1087 | ||
1088 | is made equivalent to | |
1089 | ||
1090 | m{ (?: NON_ZERO_LENGTH )* | |
1091 | | | |
1092 | (?: ZERO_LENGTH )? | |
1093 | }x; | |
1094 | ||
1095 | The higher level-loops preserve an additional state between iterations: | |
1096 | whether the last match was zero-length. To break the loop, the following | |
1097 | match after a zero-length match is prohibited to have a length of zero. | |
1098 | This prohibition interacts with backtracking (see L<"Backtracking">), | |
1099 | and so the I<second best> match is chosen if the I<best> match is of | |
1100 | zero length. | |
1101 | ||
19799a22 | 1102 | For example: |
c84d73f1 IZ |
1103 | |
1104 | $_ = 'bar'; | |
1105 | s/\w??/<$&>/g; | |
1106 | ||
20fb949f | 1107 | results in C<< <><b><><a><><r><> >>. At each position of the string the best |
c84d73f1 IZ |
1108 | match given by non-greedy C<??> is the zero-length match, and the I<second |
1109 | best> match is what is matched by C<\w>. Thus zero-length matches | |
1110 | alternate with one-character-long matches. | |
1111 | ||
1112 | Similarly, for repeated C<m/()/g> the second-best match is the match at the | |
1113 | position one notch further in the string. | |
1114 | ||
19799a22 | 1115 | The additional state of being I<matched with zero-length> is associated with |
c84d73f1 | 1116 | the matched string, and is reset by each assignment to pos(). |
9da458fc IZ |
1117 | Zero-length matches at the end of the previous match are ignored |
1118 | during C<split>. | |
c84d73f1 | 1119 | |
35a734be IZ |
1120 | =head2 Combining pieces together |
1121 | ||
1122 | Each of the elementary pieces of regular expressions which were described | |
1123 | before (such as C<ab> or C<\Z>) could match at most one substring | |
1124 | at the given position of the input string. However, in a typical regular | |
1125 | expression these elementary pieces are combined into more complicated | |
1126 | patterns using combining operators C<ST>, C<S|T>, C<S*> etc | |
1127 | (in these examples C<S> and C<T> are regular subexpressions). | |
1128 | ||
1129 | Such combinations can include alternatives, leading to a problem of choice: | |
1130 | if we match a regular expression C<a|ab> against C<"abc">, will it match | |
1131 | substring C<"a"> or C<"ab">? One way to describe which substring is | |
1132 | actually matched is the concept of backtracking (see L<"Backtracking">). | |
1133 | However, this description is too low-level and makes you think | |
1134 | in terms of a particular implementation. | |
1135 | ||
1136 | Another description starts with notions of "better"/"worse". All the | |
1137 | substrings which may be matched by the given regular expression can be | |
1138 | sorted from the "best" match to the "worst" match, and it is the "best" | |
1139 | match which is chosen. This substitutes the question of "what is chosen?" | |
1140 | by the question of "which matches are better, and which are worse?". | |
1141 | ||
1142 | Again, for elementary pieces there is no such question, since at most | |
1143 | one match at a given position is possible. This section describes the | |
1144 | notion of better/worse for combining operators. In the description | |
1145 | below C<S> and C<T> are regular subexpressions. | |
1146 | ||
13a2d996 | 1147 | =over 4 |
35a734be IZ |
1148 | |
1149 | =item C<ST> | |
1150 | ||
1151 | Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are | |
1152 | substrings which can be matched by C<S>, C<B> and C<B'> are substrings | |
1153 | which can be matched by C<T>. | |
1154 | ||
1155 | If C<A> is better match for C<S> than C<A'>, C<AB> is a better | |
1156 | match than C<A'B'>. | |
1157 | ||
1158 | If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if | |
1159 | C<B> is better match for C<T> than C<B'>. | |
1160 | ||
1161 | =item C<S|T> | |
1162 | ||
1163 | When C<S> can match, it is a better match than when only C<T> can match. | |
1164 | ||
1165 | Ordering of two matches for C<S> is the same as for C<S>. Similar for | |
1166 | two matches for C<T>. | |
1167 | ||
1168 | =item C<S{REPEAT_COUNT}> | |
1169 | ||
1170 | Matches as C<SSS...S> (repeated as many times as necessary). | |
1171 | ||
1172 | =item C<S{min,max}> | |
1173 | ||
1174 | Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>. | |
1175 | ||
1176 | =item C<S{min,max}?> | |
1177 | ||
1178 | Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>. | |
1179 | ||
1180 | =item C<S?>, C<S*>, C<S+> | |
1181 | ||
1182 | Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively. | |
1183 | ||
1184 | =item C<S??>, C<S*?>, C<S+?> | |
1185 | ||
1186 | Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively. | |
1187 | ||
c47ff5f1 | 1188 | =item C<< (?>S) >> |
35a734be IZ |
1189 | |
1190 | Matches the best match for C<S> and only that. | |
1191 | ||
1192 | =item C<(?=S)>, C<(?<=S)> | |
1193 | ||
1194 | Only the best match for C<S> is considered. (This is important only if | |
1195 | C<S> has capturing parentheses, and backreferences are used somewhere | |
1196 | else in the whole regular expression.) | |
1197 | ||
1198 | =item C<(?!S)>, C<(?<!S)> | |
1199 | ||
1200 | For this grouping operator there is no need to describe the ordering, since | |
1201 | only whether or not C<S> can match is important. | |
1202 | ||
14455d6c | 1203 | =item C<(??{ EXPR })> |
35a734be IZ |
1204 | |
1205 | The ordering is the same as for the regular expression which is | |
1206 | the result of EXPR. | |
1207 | ||
1208 | =item C<(?(condition)yes-pattern|no-pattern)> | |
1209 | ||
1210 | Recall that which of C<yes-pattern> or C<no-pattern> actually matches is | |
1211 | already determined. The ordering of the matches is the same as for the | |
1212 | chosen subexpression. | |
1213 | ||
1214 | =back | |
1215 | ||
1216 | The above recipes describe the ordering of matches I<at a given position>. | |
1217 | One more rule is needed to understand how a match is determined for the | |
1218 | whole regular expression: a match at an earlier position is always better | |
1219 | than a match at a later position. | |
1220 | ||
c84d73f1 IZ |
1221 | =head2 Creating custom RE engines |
1222 | ||
1223 | Overloaded constants (see L<overload>) provide a simple way to extend | |
1224 | the functionality of the RE engine. | |
1225 | ||
1226 | Suppose that we want to enable a new RE escape-sequence C<\Y|> which | |
1227 | matches at boundary between white-space characters and non-whitespace | |
1228 | characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly | |
1229 | at these positions, so we want to have each C<\Y|> in the place of the | |
1230 | more complicated version. We can create a module C<customre> to do | |
1231 | this: | |
1232 | ||
1233 | package customre; | |
1234 | use overload; | |
1235 | ||
1236 | sub import { | |
1237 | shift; | |
1238 | die "No argument to customre::import allowed" if @_; | |
1239 | overload::constant 'qr' => \&convert; | |
1240 | } | |
1241 | ||
1242 | sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"} | |
1243 | ||
1244 | my %rules = ( '\\' => '\\', | |
1245 | 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ ); | |
1246 | sub convert { | |
1247 | my $re = shift; | |
1248 | $re =~ s{ | |
1249 | \\ ( \\ | Y . ) | |
1250 | } | |
1251 | { $rules{$1} or invalid($re,$1) }sgex; | |
1252 | return $re; | |
1253 | } | |
1254 | ||
1255 | Now C<use customre> enables the new escape in constant regular | |
1256 | expressions, i.e., those without any runtime variable interpolations. | |
1257 | As documented in L<overload>, this conversion will work only over | |
1258 | literal parts of regular expressions. For C<\Y|$re\Y|> the variable | |
1259 | part of this regular expression needs to be converted explicitly | |
1260 | (but only if the special meaning of C<\Y|> should be enabled inside $re): | |
1261 | ||
1262 | use customre; | |
1263 | $re = <>; | |
1264 | chomp $re; | |
1265 | $re = customre::convert $re; | |
1266 | /\Y|$re\Y|/; | |
1267 | ||
19799a22 GS |
1268 | =head1 BUGS |
1269 | ||
9da458fc IZ |
1270 | This document varies from difficult to understand to completely |
1271 | and utterly opaque. The wandering prose riddled with jargon is | |
1272 | hard to fathom in several places. | |
1273 | ||
1274 | This document needs a rewrite that separates the tutorial content | |
1275 | from the reference content. | |
19799a22 GS |
1276 | |
1277 | =head1 SEE ALSO | |
9fa51da4 | 1278 | |
91e0c79e MJD |
1279 | L<perlrequick>. |
1280 | ||
1281 | L<perlretut>. | |
1282 | ||
9b599b2a GS |
1283 | L<perlop/"Regexp Quote-Like Operators">. |
1284 | ||
1e66bd83 PP |
1285 | L<perlop/"Gory details of parsing quoted constructs">. |
1286 | ||
14218588 GS |
1287 | L<perlfaq6>. |
1288 | ||
9b599b2a GS |
1289 | L<perlfunc/pos>. |
1290 | ||
1291 | L<perllocale>. | |
1292 | ||
fb55449c JH |
1293 | L<perlebcdic>. |
1294 | ||
14218588 GS |
1295 | I<Mastering Regular Expressions> by Jeffrey Friedl, published |
1296 | by O'Reilly and Associates. |