This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Minor touches at the [:class:] description.
[perl5.git] / pod / perlre.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
3perlre - Perl regular expressions
4
5=head1 DESCRIPTION
6
cb1a09d0 7This page describes the syntax of regular expressions in Perl. For a
5f05dabc 8description of how to I<use> regular expressions in matching
19799a22 9operations, plus various examples of the same, see discussions
1e66bd83 10of C<m//>, C<s///>, C<qr//> and C<??> in L<perlop/"Regexp Quote-Like Operators">.
cb1a09d0 11
19799a22 12Matching operations can have various modifiers. Modifiers
5a964f20 13that relate to the interpretation of the regular expression inside
19799a22
GS
14are listed below. Modifiers that alter the way a regular expression
15is used by Perl are detailed in L<perlop/"Regexp Quote-Like Operators"> and
1e66bd83 16L<perlop/"Gory details of parsing quoted constructs">.
a0d0e21e 17
55497cff
PP
18=over 4
19
20=item i
21
22Do case-insensitive pattern matching.
23
a034a98d
DD
24If C<use locale> is in effect, the case map is taken from the current
25locale. See L<perllocale>.
26
54310121 27=item m
55497cff
PP
28
29Treat string as multiple lines. That is, change "^" and "$" from matching
14218588 30the start or end of the string to matching the start or end of any
7f761169 31line anywhere within the string.
55497cff 32
54310121 33=item s
55497cff
PP
34
35Treat string as single line. That is, change "." to match any character
19799a22 36whatsoever, even a newline, which normally it would not match.
55497cff 37
19799a22
GS
38The C</s> and C</m> modifiers both override the C<$*> setting. That
39is, 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
41only at the end (or just before a newline at the end) of the string.
42Together, as /ms, they let the "." match any character whatsoever,
43while yet allowing "^" and "$" to match, respectively, just after
44and just before newlines within the string.
7b8d334a 45
54310121 46=item x
55497cff
PP
47
48Extend your pattern's legibility by permitting whitespace and comments.
49
50=back
a0d0e21e
LW
51
52These are usually written as "the C</x> modifier", even though the delimiter
14218588 53in question might not really be a slash. Any of these
a0d0e21e 54modifiers may also be embedded within the regular expression itself using
14218588 55the C<(?...)> construct. See below.
a0d0e21e 56
4633a7c4 57The C</x> modifier itself needs a little more explanation. It tells
55497cff
PP
58the regular expression parser to ignore whitespace that is neither
59backslashed nor within a character class. You can use this to break up
4633a7c4 60your regular expression into (slightly) more readable parts. The C<#>
54310121 61character is also treated as a metacharacter introducing a comment,
55497cff 62just as in ordinary Perl code. This also means that if you want real
14218588 63whitespace or C<#> characters in the pattern (outside a character
5a964f20 64class, where they are unaffected by C</x>), that you'll either have to
55497cff
PP
65escape them or encode them using octal or hex escapes. Taken together,
66these features go a long way towards making Perl's regular expressions
0c815be9
HS
67more readable. Note that you have to be careful not to include the
68pattern delimiter in the comment--perl has no way of knowing you did
5a964f20 69not intend to close the pattern early. See the C-comment deletion code
0c815be9 70in L<perlop>.
a0d0e21e
LW
71
72=head2 Regular Expressions
73
19799a22 74The patterns used in Perl pattern matching derive from supplied in
14218588 75the Version 8 regex routines. (The routines are derived
19799a22
GS
76(distantly) from Henry Spencer's freely redistributable reimplementation
77of the V8 routines.) See L<Version 8 Regular Expressions> for
78details.
a0d0e21e
LW
79
80In particular the following metacharacters have their standard I<egrep>-ish
81meanings:
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
91By default, the "^" character is guaranteed to match only the
92beginning of the string, the "$" character only the end (or before the
93newline at the end), and Perl does certain optimizations with the
a0d0e21e
LW
94assumption that the string contains only one line. Embedded newlines
95will not be matched by "^" or "$". You may, however, wish to treat a
4a6725af 96string as a multi-line buffer, such that the "^" will match after any
a0d0e21e
LW
97newline within the string, and "$" will match before any newline. At the
98cost of a little more overhead, you can do this by using the /m modifier
99on the pattern match operator. (Older programs did this by setting C<$*>,
5f05dabc 100but this practice is now deprecated.)
a0d0e21e 101
14218588 102To simplify multi-line substitutions, the "." character never matches a
55497cff 103newline unless you use the C</s> modifier, which in effect tells Perl to pretend
a0d0e21e
LW
104the string is a single line--even if it isn't. The C</s> modifier also
105overrides the setting of C<$*>, in case you have some (badly behaved) older
106code that sets it in another module.
107
108The 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
118as a regular character.) The "*" modifier is equivalent to C<{0,}>, the "+"
25f94b33 119modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
9c79236d
GS
120to integral values less than a preset limit defined when perl is built.
121This is usually 32766 on the most common platforms. The actual limit can
122be seen in the error message generated by code such as this:
123
124 $_ **= $_ , / {$_} / for 2 .. 42;
a0d0e21e 125
54310121
PP
126By default, a quantified subpattern is "greedy", that is, it will match as
127many times as possible (given a particular starting location) while still
128allowing the rest of the pattern to match. If you want it to match the
129minimum number of times possible, follow the quantifier with a "?". Note
130that 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 139Because patterns are processed as double quoted strings, the following
a0d0e21e
LW
140also work:
141
0f36ee90
PP
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
4a2d328f 152 \N{name} named char
cb1a09d0
AD
153 \l lowercase next char (think vi)
154 \u uppercase next char (think vi)
155 \L lowercase till \E (think vi)
156 \U uppercase till \E (think vi)
157 \E end case modification (think vi)
5a964f20 158 \Q quote (disable) pattern metacharacters till \E
a0d0e21e 159
a034a98d 160If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
423cee85 161and C<\U> is taken from the current locale. See L<perllocale>. For
4a2d328f 162documentation of C<\N{name}>, see L<charnames>.
a034a98d 163
1d2dff63
GS
164You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
165An unescaped C<$> or C<@> interpolates the corresponding variable,
166while escaping will cause the literal string C<\$> to be matched.
167You'll need to write something like C<m/\Quser\E\@\Qhost/>.
168
a0d0e21e
LW
169In addition, Perl defines the following:
170
171 \w Match a "word" character (alphanumeric plus "_")
172 \W Match a non-word character
173 \s Match a whitespace character
174 \S Match a non-whitespace character
175 \d Match a digit character
176 \D Match a non-digit character
a0ed51b3
LW
177 \pP Match P, named property. Use \p{Prop} for longer names.
178 \PP Match non-P
f244e06d
GS
179 \X Match eXtended Unicode "combining character sequence",
180 equivalent to C<(?:\PM\pM*)>
4a2d328f 181 \C Match a single C char (octet) even under utf8.
a0d0e21e 182
19799a22 183A C<\w> matches a single alphanumeric character, not a whole word.
14218588
GS
184Use C<\w+> to match a string of Perl-identifier characters (which isn't
185the same as matching an English word). If C<use locale> is in effect, the
186list of alphabetic characters generated by C<\w> is taken from the
187current locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
188C<\d>, and C<\D> within character classes (though not as either end of
189a range). See L<utf8> for details about C<\pP>, C<\PP>, and C<\X>.
a0d0e21e 190
b8c5462f
JH
191The POSIX character class syntax
192
193 [:class:]
194
26b44a0a
JH
195is also available. The available classes and their backslash
196equivalents (if available) are as follows:
b8c5462f
JH
197
198 alpha
199 alnum
200 ascii
201 cntrl
202 digit \d
203 graph
204 lower
205 print
206 punct
207 space \s
208 upper
209 word \w
210 xdigit
211
26b44a0a
JH
212For example use C<[:upper:]> to match all the uppercase characters.
213Note that the C<[]> are part of the C<[::]> construct, not part of the whole
b8c5462f
JH
214character class. For example:
215
216 [01[:alpha:]%]
217
218matches one, zero, any alphabetic character, and the percentage sign.
219
26b44a0a 220If the C<utf8> pragma is used, the following equivalences to Unicode
b8c5462f
JH
221\p{} constructs hold:
222
223 alpha IsAlpha
224 alnum IsAlnum
225 ascii IsASCII
226 cntrl IsCntrl
227 digit IsDigit
228 graph IsGraph
229 lower IsLower
230 print IsPrint
231 punct IsPunct
232 space IsSpace
233 upper IsUpper
234 word IsWord
235 xdigit IsXDigit
236
26b44a0a 237For example C<[:lower:]> and C<\p{IsLower}> are equivalent.
b8c5462f
JH
238
239If the C<utf8> pragma is not used but the C<locale> pragma is, the
240classes correlate with the isalpha(3) interface (except for `word',
26b44a0a 241which is a Perl extension, mirroring C<\w>).
b8c5462f
JH
242
243The assumedly non-obviously named classes are:
244
245=over 4
246
247=item cntrl
248
249 Any control character. Usually characters that don't produce
250 output as such but instead control the terminal somehow:
251 for example newline and backspace are control characters.
93733859
JH
252 All characters with ord() less than 32 are most often control
253 classified as characters.
b8c5462f
JH
254
255=item graph
256
257 Any alphanumeric or punctuation character.
258
259=item print
260
261 Any alphanumeric or punctuation character or space.
262
263=item punct
264
265 Any punctuation character.
266
267=item xdigit
268
269 Any hexadecimal digit. Though this may feel silly
270 (/0-9a-f/i would work just fine) it is included
271 for completeness.
272
273=item
274
275=back
276
277You can negate the [::] character classes by prefixing the class name
278with a '^'. This is a Perl extension. For example:
279
93733859
JH
280 POSIX trad. Perl utf8 Perl
281
282 [:^digit:] \D \P{IsDigit}
283 [:^space:] \S \P{IsSpace}
284 [:^word:] \W \P{IsWord}
b8c5462f 285
26b44a0a
JH
286The POSIX character classes [.cc.] and [=cc=] are recognized but
287B<not> supported and trying to use them will cause an error.
b8c5462f 288
a0d0e21e
LW
289Perl defines the following zero-width assertions:
290
291 \b Match a word boundary
292 \B Match a non-(word boundary)
b85d18e9
IZ
293 \A Match only at beginning of string
294 \Z Match only at end of string, or before newline at the end
295 \z Match only at end of string
a99df21c 296 \G Match only where previous m//g left off (works only with /g)
a0d0e21e 297
14218588 298A word boundary (C<\b>) is a spot between two characters
19799a22
GS
299that has a C<\w> on one side of it and a C<\W> on the other side
300of it (in either order), counting the imaginary characters off the
301beginning and end of the string as matching a C<\W>. (Within
302character classes C<\b> represents backspace rather than a word
303boundary, just as it normally does in any double-quoted string.)
304The C<\A> and C<\Z> are just like "^" and "$", except that they
305won't match multiple times when the C</m> modifier is used, while
306"^" and "$" will match at every internal line boundary. To match
307the actual end of the string and not ignore an optional trailing
308newline, use C<\z>.
309
310The C<\G> assertion can be used to chain global matches (using
311C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
312It is also useful when writing C<lex>-like scanners, when you have
313several patterns that you want to match against consequent substrings
314of your string, see the previous reference. The actual location
315where C<\G> will match can also be influenced by using C<pos()> as
316an lvalue. See L<perlfunc/pos>.
14218588
GS
317
318The bracketing construct C<( ... )> creates capture buffers. To
319refer to the digit'th buffer use \E<lt>digitE<gt> within the
320match. Outside the match use "$" instead of "\". (The
321\E<lt>digitE<gt> notation works in certain circumstances outside
322the match. See the warning below about \1 vs $1 for details.)
323Referring back to another part of the match is called a
324I<backreference>.
325
326There is no limit to the number of captured substrings that you may
327use. However Perl also uses \10, \11, etc. as aliases for \010,
328\011, etc. (Recall that 0 means octal, so \011 is the 9'th ASCII
329character, a tab.) Perl resolves this ambiguity by interpreting
330\10 as a backreference only if at least 10 left parentheses have
331opened before it. Likewise \11 is a backreference only if at least
33211 left parentheses have opened before it. And so on. \1 through
333\9 are always interpreted as backreferences."
334
335Examples:
a0d0e21e
LW
336
337 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
338
14218588
GS
339 if (/(.)\1/) { # find first doubled char
340 print "'$1' is the first doubled character\n";
341 }
342
343 if (/Time: (..):(..):(..)/) { # parse out values
a0d0e21e
LW
344 $hours = $1;
345 $minutes = $2;
346 $seconds = $3;
347 }
14218588
GS
348
349Several special variables also refer back to portions of the previous
350match. C<$+> returns whatever the last bracket match matched.
351C<$&> returns the entire matched string. (At one point C<$0> did
352also, but now it returns the name of the program.) C<$`> returns
353everything before the matched string. And C<$'> returns everything
354after the matched string.
355
356The numbered variables ($1, $2, $3, etc.) and the related punctuation
357set (C<<$+>, C<$&>, C<$`>, and C<$'>) are all dynamically scoped
358until the end of the enclosing block or until the next successful
359match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
360
361B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
362C<$'> anywhere in the program, it has to provide them for every
363pattern match. This may substantially slow your program. Perl
364uses the same mechanism to produce $1, $2, etc, so you also pay a
365price for each pattern that contains capturing parentheses. (To
366avoid this cost while retaining the grouping behaviour, use the
367extended regular expression C<(?: ... )> instead.) But if you never
368use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
369parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
370if you can, but if you can't (and some algorithms really appreciate
371them), once you've used them once, use them at will, because you've
372already paid the price. As of 5.005, C<$&> is not so costly as the
373other two.
68dc0745 374
19799a22
GS
375Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
376C<\w>, C<\n>. Unlike some other regular expression languages, there
377are no backslashed symbols that aren't alphanumeric. So anything
378that looks like \\, \(, \), \E<lt>, \E<gt>, \{, or \} is always
379interpreted as a literal character, not a metacharacter. This was
380once used in a common idiom to disable or quote the special meanings
381of regular expression metacharacters in a string that you want to
382use for a pattern. Simply quote all non-alphanumeric characters:
a0d0e21e
LW
383
384 $pattern =~ s/(\W)/\\$1/g;
385
14218588
GS
386Today it is more common to use the quotemeta() function or the C<\Q>
387metaquoting escape sequence to disable all metacharacters' special
388meanings like this:
a0d0e21e
LW
389
390 /$unquoted\Q$quoted\E$unquoted/
391
19799a22
GS
392=head2 Extended Patterns
393
14218588
GS
394Perl also defines a consistent extension syntax for features not
395found in standard tools like B<awk> and B<lex>. The syntax is a
396pair of parentheses with a question mark as the first thing within
397the parentheses. The character after the question mark indicates
398the extension.
19799a22 399
14218588
GS
400The stability of these extensions varies widely. Some have been
401part of the core language for many years. Others are experimental
402and may change without warning or be completely removed. Check
403the documentation on an individual feature to verify its current
404status.
19799a22 405
14218588
GS
406A question mark was chosen for this and for the minimal-matching
407construct because 1) question marks are rare in older regular
408expressions, and 2) whenever you see one, you should stop and
409"question" exactly what is going on. That's psychology...
a0d0e21e
LW
410
411=over 10
412
cc6b7395 413=item C<(?#text)>
a0d0e21e 414
14218588 415A comment. The text is ignored. If the C</x> modifier enables
19799a22 416whitespace formatting, a simple C<#> will suffice. Note that Perl closes
259138e3
GS
417the comment as soon as it sees a C<)>, so there is no way to put a literal
418C<)> in the comment.
a0d0e21e 419
19799a22
GS
420=item C<(?imsx-imsx)>
421
422One or more embedded pattern-match modifiers. This is particularly
423useful for dynamic patterns, such as those read in from a configuration
424file, read in as an argument, are specified in a table somewhere,
425etc. Consider the case that some of which want to be case sensitive
426and some do not. The case insensitive ones need to include merely
427C<(?i)> at the front of the pattern. For example:
428
429 $pattern = "foobar";
430 if ( /$pattern/i ) { }
431
432 # more flexible:
433
434 $pattern = "(?i)foobar";
435 if ( /$pattern/ ) { }
436
437Letters after a C<-> turn those modifiers off. These modifiers are
438localized inside an enclosing group (if any). For example,
439
440 ( (?i) blah ) \s+ \1
441
442will match a repeated (I<including the case>!) word C<blah> in any
14218588 443case, assuming C<x> modifier, and no C<i> modifier outside this
19799a22
GS
444group.
445
5a964f20 446=item C<(?:pattern)>
a0d0e21e 447
ca9dfc88
IZ
448=item C<(?imsx-imsx:pattern)>
449
5a964f20
TC
450This is for clustering, not capturing; it groups subexpressions like
451"()", but doesn't make backreferences as "()" does. So
a0d0e21e 452
5a964f20 453 @fields = split(/\b(?:a|b|c)\b/)
a0d0e21e
LW
454
455is like
456
5a964f20 457 @fields = split(/\b(a|b|c)\b/)
a0d0e21e 458
19799a22
GS
459but doesn't spit out extra fields. It's also cheaper not to capture
460characters if you don't need to.
a0d0e21e 461
19799a22
GS
462Any letters between C<?> and C<:> act as flags modifiers as with
463C<(?imsx-imsx)>. For example,
ca9dfc88
IZ
464
465 /(?s-i:more.*than).*million/i
466
14218588 467is equivalent to the more verbose
ca9dfc88
IZ
468
469 /(?:(?s-i)more.*than).*million/i
470
5a964f20 471=item C<(?=pattern)>
a0d0e21e 472
19799a22 473A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
a0d0e21e
LW
474matches a word followed by a tab, without including the tab in C<$&>.
475
5a964f20 476=item C<(?!pattern)>
a0d0e21e 477
19799a22 478A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
a0d0e21e 479matches any occurrence of "foo" that isn't followed by "bar". Note
19799a22
GS
480however that look-ahead and look-behind are NOT the same thing. You cannot
481use this for look-behind.
7b8d334a 482
5a964f20 483If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
7b8d334a
GS
484will not do what you want. That's because the C<(?!foo)> is just saying that
485the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
486match. You would have to do something like C</(?!foo)...bar/> for that. We
487say "like" because there's the case of your "bar" not having three characters
488before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
489Sometimes it's still easier just to say:
a0d0e21e 490
a3cb178b 491 if (/bar/ && $` !~ /foo$/)
a0d0e21e 492
19799a22 493For look-behind see below.
c277df42 494
5a964f20 495=item C<(?E<lt>=pattern)>
c277df42 496
19799a22
GS
497A zero-width positive look-behind assertion. For example, C</(?E<lt>=\t)\w+/>
498matches a word that follows a tab, without including the tab in C<$&>.
499Works only for fixed-width look-behind.
c277df42 500
5a964f20 501=item C<(?<!pattern)>
c277df42 502
19799a22
GS
503A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
504matches any occurrence of "foo" that does not follow "bar". Works
505only for fixed-width look-behind.
c277df42 506
cc6b7395 507=item C<(?{ code })>
c277df42 508
19799a22
GS
509B<WARNING>: This extended regular expression feature is considered
510highly experimental, and may be changed or deleted without notice.
c277df42 511
19799a22
GS
512This zero-width assertion evaluate any embedded Perl code. It
513always succeeds, and its C<code> is not interpolated. Currently,
514the rules to determine where the C<code> ends are somewhat convoluted.
515
516The C<code> is properly scoped in the following sense: If the assertion
517is backtracked (compare L<"Backtracking">), all changes introduced after
518C<local>ization are undone, so that
b9ac3b5b
GS
519
520 $_ = 'a' x 8;
521 m<
522 (?{ $cnt = 0 }) # Initialize $cnt.
523 (
524 a
525 (?{
526 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
527 })
528 )*
529 aaaa
530 (?{ $res = $cnt }) # On success copy to non-localized
531 # location.
532 >x;
533
19799a22 534will set C<$res = 4>. Note that after the match, $cnt returns to the globally
14218588 535introduced value, because the scopes that restrict C<local> operators
b9ac3b5b
GS
536are unwound.
537
19799a22
GS
538This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
539switch. If I<not> used in this way, the result of evaluation of
540C<code> is put into the special variable C<$^R>. This happens
541immediately, so C<$^R> can be used from other C<(?{ code })> assertions
542inside the same regular expression.
b9ac3b5b 543
19799a22
GS
544The assignment to C<$^R> above is properly localized, so the old
545value of C<$^R> is restored if the assertion is backtracked; compare
546L<"Backtracking">.
b9ac3b5b 547
19799a22
GS
548For reasons of security, this construct is forbidden if the regular
549expression involves run-time interpolation of variables, unless the
550perilous C<use re 'eval'> pragma has been used (see L<re>), or the
551variables contain results of C<qr//> operator (see
552L<perlop/"qr/STRING/imosx">).
871b0233 553
14218588 554This restriction is because of the wide-spread and remarkably convenient
19799a22 555custom of using run-time determined strings as patterns. For example:
871b0233
IZ
556
557 $re = <>;
558 chomp $re;
559 $string =~ /$re/;
560
14218588
GS
561Before Perl knew how to execute interpolated code within a pattern,
562this operation was completely safe from a security point of view,
563although it could raise an exception from an illegal pattern. If
564you turn on the C<use re 'eval'>, though, it is no longer secure,
565so you should only do so if you are also using taint checking.
566Better yet, use the carefully constrained evaluation within a Safe
567module. See L<perlsec> for details about both these mechanisms.
871b0233 568
0f5d15d6
IZ
569=item C<(?p{ code })>
570
19799a22
GS
571B<WARNING>: This extended regular expression feature is considered
572highly experimental, and may be changed or deleted without notice.
0f5d15d6 573
19799a22
GS
574This is a "postponed" regular subexpression. The C<code> is evaluated
575at run time, at the moment this subexpression may match. The result
576of evaluation is considered as a regular expression and matched as
577if it were inserted instead of this construct.
0f5d15d6 578
428594d9 579The C<code> is not interpolated. As before, the rules to determine
19799a22
GS
580where the C<code> ends are currently somewhat convoluted.
581
582The following pattern matches a parenthesized group:
0f5d15d6
IZ
583
584 $re = qr{
585 \(
586 (?:
587 (?> [^()]+ ) # Non-parens without backtracking
588 |
589 (?p{ $re }) # Group with matching parens
590 )*
591 \)
592 }x;
593
5a964f20
TC
594=item C<(?E<gt>pattern)>
595
19799a22
GS
596B<WARNING>: This extended regular expression feature is considered
597highly experimental, and may be changed or deleted without notice.
598
599An "independent" subexpression, one which matches the substring
600that a I<standalone> C<pattern> would match if anchored at the given
14218588 601position--but it matches no more than this substring. This
19799a22
GS
602construct is useful for optimizations of what would otherwise be
603"eternal" matches, because it will not backtrack (see L<"Backtracking">).
604
605For example: C<^(?E<gt>a*)ab> will never match, since C<(?E<gt>a*)>
606(anchored at the beginning of string, as above) will match I<all>
607characters C<a> at the beginning of string, leaving no C<a> for
608C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
609since the match of the subgroup C<a*> is influenced by the following
610group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
611C<a*ab> will match fewer characters than a standalone C<a*>, since
612this makes the tail match.
613
614An effect similar to C<(?E<gt>pattern)> may be achieved by writing
615C<(?=(pattern))\1>. This matches the same substring as a standalone
616C<a+>, and the following C<\1> eats the matched string; it therefore
617makes a zero-length assertion into an analogue of C<(?E<gt>...)>.
618(The difference between these two constructs is that the second one
619uses a capturing group, thus shifting ordinals of backreferences
620in the rest of a regular expression.)
621
622Consider this pattern:
c277df42 623
871b0233
IZ
624 m{ \(
625 (
626 [^()]+
627 |
628 \( [^()]* \)
629 )+
630 \)
631 }x
5a964f20 632
19799a22
GS
633That will efficiently match a nonempty group with matching parentheses
634two levels deep or less. However, if there is no such group, it
635will take virtually forever on a long string. That's because there
636are so many different ways to split a long string into several
637substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
638to a subpattern of the above pattern. Consider how the pattern
639above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
640seconds, but that each extra letter doubles this time. This
641exponential performance will make it appear that your program has
14218588 642hung. However, a tiny change to this pattern
5a964f20 643
871b0233
IZ
644 m{ \(
645 (
646 (?> [^()]+ )
647 |
648 \( [^()]* \)
649 )+
650 \)
651 }x
c277df42 652
5a964f20
TC
653which uses C<(?E<gt>...)> matches exactly when the one above does (verifying
654this yourself would be a productive exercise), but finishes in a fourth
655the time when used on a similar string with 1000000 C<a>s. Be aware,
656however, that this pattern currently triggers a warning message under
657B<-w> saying it C<"matches the null string many times">):
c277df42 658
8d300b32 659On simple groups, such as the pattern C<(?E<gt> [^()]+ )>, a comparable
19799a22 660effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
c277df42
IZ
661This was only 4 times slower on a string with 1000000 C<a>s.
662
5a964f20 663=item C<(?(condition)yes-pattern|no-pattern)>
c277df42 664
5a964f20 665=item C<(?(condition)yes-pattern)>
c277df42 666
19799a22
GS
667B<WARNING>: This extended regular expression feature is considered
668highly experimental, and may be changed or deleted without notice.
669
c277df42
IZ
670Conditional expression. C<(condition)> should be either an integer in
671parentheses (which is valid if the corresponding pair of parentheses
19799a22 672matched), or look-ahead/look-behind/evaluate zero-width assertion.
c277df42 673
19799a22 674For example:
c277df42 675
5a964f20 676 m{ ( \( )?
871b0233 677 [^()]+
5a964f20 678 (?(1) \) )
871b0233 679 }x
c277df42
IZ
680
681matches a chunk of non-parentheses, possibly included in parentheses
682themselves.
a0d0e21e 683
a0d0e21e
LW
684=back
685
c07a80fd
PP
686=head2 Backtracking
687
c277df42 688A fundamental feature of regular expression matching involves the
5a964f20 689notion called I<backtracking>, which is currently used (when needed)
c277df42
IZ
690by all regular expression quantifiers, namely C<*>, C<*?>, C<+>,
691C<+?>, C<{n,m}>, and C<{n,m}?>.
c07a80fd
PP
692
693For a regular expression to match, the I<entire> regular expression must
694match, not just part of it. So if the beginning of a pattern containing a
695quantifier succeeds in a way that causes later parts in the pattern to
696fail, the matching engine backs up and recalculates the beginning
697part--that's why it's called backtracking.
698
699Here is an example of backtracking: Let's say you want to find the
700word following "foo" in the string "Food is on the foo table.":
701
702 $_ = "Food is on the foo table.";
703 if ( /\b(foo)\s+(\w+)/i ) {
704 print "$2 follows $1.\n";
705 }
706
707When the match runs, the first part of the regular expression (C<\b(foo)>)
708finds a possible match right at the beginning of the string, and loads up
709$1 with "Foo". However, as soon as the matching engine sees that there's
710no whitespace following the "Foo" that it had saved in $1, it realizes its
68dc0745 711mistake and starts over again one character after where it had the
c07a80fd
PP
712tentative match. This time it goes all the way until the next occurrence
713of "foo". The complete regular expression matches this time, and you get
714the expected output of "table follows foo."
715
716Sometimes minimal matching can help a lot. Imagine you'd like to match
717everything between "foo" and "bar". Initially, you write something
718like this:
719
720 $_ = "The food is under the bar in the barn.";
721 if ( /foo(.*)bar/ ) {
722 print "got <$1>\n";
723 }
724
725Which perhaps unexpectedly yields:
726
727 got <d is under the bar in the >
728
729That's because C<.*> was greedy, so you get everything between the
14218588 730I<first> "foo" and the I<last> "bar". Here it's more effective
c07a80fd
PP
731to use minimal matching to make sure you get the text between a "foo"
732and the first "bar" thereafter.
733
734 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
735 got <d is under the >
736
737Here's another example: let's say you'd like to match a number at the end
738of a string, and you also want to keep the preceding part the match.
739So you write this:
740
741 $_ = "I have 2 numbers: 53147";
742 if ( /(.*)(\d*)/ ) { # Wrong!
743 print "Beginning is <$1>, number is <$2>.\n";
744 }
745
746That won't work at all, because C<.*> was greedy and gobbled up the
747whole string. As C<\d*> can match on an empty string the complete
748regular expression matched successfully.
749
8e1088bc 750 Beginning is <I have 2 numbers: 53147>, number is <>.
c07a80fd
PP
751
752Here are some variants, most of which don't work:
753
754 $_ = "I have 2 numbers: 53147";
755 @pats = qw{
756 (.*)(\d*)
757 (.*)(\d+)
758 (.*?)(\d*)
759 (.*?)(\d+)
760 (.*)(\d+)$
761 (.*?)(\d+)$
762 (.*)\b(\d+)$
763 (.*\D)(\d+)$
764 };
765
766 for $pat (@pats) {
767 printf "%-12s ", $pat;
768 if ( /$pat/ ) {
769 print "<$1> <$2>\n";
770 } else {
771 print "FAIL\n";
772 }
773 }
774
775That will print out:
776
777 (.*)(\d*) <I have 2 numbers: 53147> <>
778 (.*)(\d+) <I have 2 numbers: 5314> <7>
779 (.*?)(\d*) <> <>
780 (.*?)(\d+) <I have > <2>
781 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
782 (.*?)(\d+)$ <I have 2 numbers: > <53147>
783 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
784 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
785
786As you see, this can be a bit tricky. It's important to realize that a
787regular expression is merely a set of assertions that gives a definition
788of success. There may be 0, 1, or several different ways that the
789definition might succeed against a particular string. And if there are
5a964f20
TC
790multiple ways it might succeed, you need to understand backtracking to
791know which variety of success you will achieve.
c07a80fd 792
19799a22 793When using look-ahead assertions and negations, this can all get even
54310121 794tricker. Imagine you'd like to find a sequence of non-digits not
c07a80fd
PP
795followed by "123". You might try to write that as
796
871b0233
IZ
797 $_ = "ABC123";
798 if ( /^\D*(?!123)/ ) { # Wrong!
799 print "Yup, no 123 in $_\n";
800 }
c07a80fd
PP
801
802But that isn't going to match; at least, not the way you're hoping. It
803claims that there is no 123 in the string. Here's a clearer picture of
804why it that pattern matches, contrary to popular expectations:
805
806 $x = 'ABC123' ;
807 $y = 'ABC445' ;
808
809 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ;
810 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ;
811
812 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ;
813 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ;
814
815This prints
816
817 2: got ABC
818 3: got AB
819 4: got ABC
820
5f05dabc 821You might have expected test 3 to fail because it seems to a more
c07a80fd
PP
822general purpose version of test 1. The important difference between
823them is that test 3 contains a quantifier (C<\D*>) and so can use
824backtracking, whereas test 1 will not. What's happening is
825that you've asked "Is it true that at the start of $x, following 0 or more
5f05dabc 826non-digits, you have something that's not 123?" If the pattern matcher had
c07a80fd 827let C<\D*> expand to "ABC", this would have caused the whole pattern to
54310121 828fail.
14218588 829
c07a80fd 830The search engine will initially match C<\D*> with "ABC". Then it will
14218588 831try to match C<(?!123> with "123", which fails. But because
c07a80fd
PP
832a quantifier (C<\D*>) has been used in the regular expression, the
833search engine can backtrack and retry the match differently
54310121 834in the hope of matching the complete regular expression.
c07a80fd 835
5a964f20
TC
836The pattern really, I<really> wants to succeed, so it uses the
837standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
c07a80fd 838time. Now there's indeed something following "AB" that is not
14218588 839"123". It's "C123", which suffices.
c07a80fd 840
14218588
GS
841We can deal with this by using both an assertion and a negation.
842We'll say that the first part in $1 must be followed both by a digit
843and by something that's not "123". Remember that the look-aheads
844are zero-width expressions--they only look, but don't consume any
845of the string in their match. So rewriting this way produces what
c07a80fd
PP
846you'd expect; that is, case 5 will fail, but case 6 succeeds:
847
848 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ;
849 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ;
850
851 6: got ABC
852
5a964f20 853In other words, the two zero-width assertions next to each other work as though
19799a22 854they're ANDed together, just as you'd use any built-in assertions: C</^$/>
c07a80fd
PP
855matches only if you're at the beginning of the line AND the end of the
856line simultaneously. The deeper underlying truth is that juxtaposition in
857regular expressions always means AND, except when you write an explicit OR
858using the vertical bar. C</ab/> means match "a" AND (then) match "b",
859although the attempted matches are made at different positions because "a"
860is not a zero-width assertion, but a one-width assertion.
861
19799a22 862B<WARNING>: particularly complicated regular expressions can take
14218588 863exponential time to solve because of the immense number of possible
19799a22 864ways they can use backtracking to try match. For example, this will
14218588 865take a painfully long time to run
c07a80fd
PP
866
867 /((a{0,5}){0,5}){0,5}/
868
14218588
GS
869And if you used C<*>'s instead of limiting it to 0 through 5 matches,
870then it would take forever--or until you ran out of stack space.
c07a80fd 871
c277df42 872A powerful tool for optimizing such beasts is "independent" groups,
5a964f20 873which do not backtrace (see L<C<(?E<gt>pattern)>>). Note also that
19799a22 874zero-length look-ahead/look-behind assertions will not backtrace to make
14218588
GS
875the tail match, since they are in "logical" context: only
876whether they match is considered relevant. For an example
19799a22 877where side-effects of a look-ahead I<might> have influenced the
5a964f20 878following match, see L<C<(?E<gt>pattern)>>.
c277df42 879
a0d0e21e
LW
880=head2 Version 8 Regular Expressions
881
5a964f20 882In case you're not familiar with the "regular" Version 8 regex
a0d0e21e
LW
883routines, here are the pattern-matching rules not described above.
884
54310121 885Any single character matches itself, unless it is a I<metacharacter>
a0d0e21e 886with a special meaning described here or above. You can cause
5a964f20 887characters that normally function as metacharacters to be interpreted
5f05dabc 888literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
a0d0e21e
LW
889character; "\\" matches a "\"). A series of characters matches that
890series of characters in the target string, so the pattern C<blurfl>
891would match "blurfl" in the target string.
892
893You can specify a character class, by enclosing a list of characters
5a964f20 894in C<[]>, which will match any one character from the list. If the
a0d0e21e 895first character after the "[" is "^", the class matches any character not
14218588 896in the list. Within a list, the "-" character specifies a
5a964f20 897range, so that C<a-z> represents all characters between "a" and "z",
84850974
DD
898inclusive. If you want "-" itself to be a member of a class, put it
899at the start or end of the list, or escape it with a backslash. (The
900following all specify the same class of three characters: C<[-az]>,
901C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
902specifies a class containing twenty-six characters.)
a0d0e21e 903
8ada0baa
JH
904Note also that the whole range idea is rather unportable between
905character sets--and even within character sets they may cause results
906you probably didn't expect. A sound principle is to use only ranges
907that begin from and end at either alphabets of equal case ([a-e],
908[A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
909spell out the character sets in full.
910
54310121 911Characters may be specified using a metacharacter syntax much like that
a0d0e21e
LW
912used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
913"\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
914of octal digits, matches the character whose ASCII value is I<nnn>.
0f36ee90 915Similarly, \xI<nn>, where I<nn> are hexadecimal digits, matches the
a0d0e21e 916character whose ASCII value is I<nn>. The expression \cI<x> matches the
54310121 917ASCII character control-I<x>. Finally, the "." metacharacter matches any
a0d0e21e
LW
918character except "\n" (unless you use C</s>).
919
920You can specify a series of alternatives for a pattern using "|" to
921separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
5a964f20 922or "foe" in the target string (as would C<f(e|i|o)e>). The
a0d0e21e
LW
923first alternative includes everything from the last pattern delimiter
924("(", "[", or the beginning of the pattern) up to the first "|", and
925the last alternative contains everything from the last "|" to the next
14218588
GS
926pattern delimiter. That's why it's common practice to include
927alternatives in parentheses: to minimize confusion about where they
a3cb178b
GS
928start and end.
929
5a964f20 930Alternatives are tried from left to right, so the first
a3cb178b
GS
931alternative found for which the entire expression matches, is the one that
932is chosen. This means that alternatives are not necessarily greedy. For
628afcb5 933example: when matching C<foo|foot> against "barefoot", only the "foo"
a3cb178b
GS
934part will match, as that is the first alternative tried, and it successfully
935matches the target string. (This might not seem important, but it is
936important when you are capturing matched text using parentheses.)
937
5a964f20 938Also remember that "|" is interpreted as a literal within square brackets,
a3cb178b 939so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
a0d0e21e 940
14218588
GS
941Within a pattern, you may designate subpatterns for later reference
942by enclosing them in parentheses, and you may refer back to the
943I<n>th subpattern later in the pattern using the metacharacter
944\I<n>. Subpatterns are numbered based on the left to right order
945of their opening parenthesis. A backreference matches whatever
946actually matched the subpattern in the string being examined, not
947the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
948match "0x1234 0x4321", but not "0x1234 01234", because subpattern
9491 matched "0x", even though the rule C<0|0x> could potentially match
950the leading 0 in the second number.
cb1a09d0 951
19799a22 952=head2 Warning on \1 vs $1
cb1a09d0 953
5a964f20 954Some people get too used to writing things like:
cb1a09d0
AD
955
956 $pattern =~ s/(\W)/\\\1/g;
957
958This is grandfathered for the RHS of a substitute to avoid shocking the
959B<sed> addicts, but it's a dirty habit to get into. That's because in
5f05dabc 960PerlThink, the righthand side of a C<s///> is a double-quoted string. C<\1> in
cb1a09d0
AD
961the usual double-quoted string means a control-A. The customary Unix
962meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
963of doing that, you get yourself into trouble if you then add an C</e>
964modifier.
965
5a964f20 966 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
cb1a09d0
AD
967
968Or if you try to do
969
970 s/(\d+)/\1000/;
971
972You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
14218588 973C<${1}000>. The operation of interpolation should not be confused
cb1a09d0
AD
974with the operation of matching a backreference. Certainly they mean two
975different things on the I<left> side of the C<s///>.
9fa51da4 976
c84d73f1
IZ
977=head2 Repeated patterns matching zero-length substring
978
19799a22 979B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
c84d73f1
IZ
980
981Regular expressions provide a terse and powerful programming language. As
982with most other power tools, power comes together with the ability
983to wreak havoc.
984
985A common abuse of this power stems from the ability to make infinite
628afcb5 986loops using regular expressions, with something as innocuous as:
c84d73f1
IZ
987
988 'foo' =~ m{ ( o? )* }x;
989
990The C<o?> can match at the beginning of C<'foo'>, and since the position
991in the string is not moved by the match, C<o?> would match again and again
14218588 992because of the C<*> modifier. Another common way to create a similar cycle
c84d73f1
IZ
993is with the looping modifier C<//g>:
994
995 @matches = ( 'foo' =~ m{ o? }xg );
996
997or
998
999 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1000
1001or the loop implied by split().
1002
1003However, long experience has shown that many programming tasks may
14218588
GS
1004be significantly simplified by using repeated subexpressions that
1005may match zero-length substrings. Here's a simple example being:
c84d73f1
IZ
1006
1007 @chars = split //, $string; # // is not magic in split
1008 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1009
1010Thus Perl allows the C</()/> construct, which I<forcefully breaks
1011the infinite loop>. The rules for this are different for lower-level
1012loops given by the greedy modifiers C<*+{}>, and for higher-level
1013ones like the C</g> modifier or split() operator.
1014
19799a22
GS
1015The lower-level loops are I<interrupted> (that is, the loop is
1016broken) when Perl detects that a repeated expression matched a
1017zero-length substring. Thus
c84d73f1
IZ
1018
1019 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1020
1021is made equivalent to
1022
1023 m{ (?: NON_ZERO_LENGTH )*
1024 |
1025 (?: ZERO_LENGTH )?
1026 }x;
1027
1028The higher level-loops preserve an additional state between iterations:
1029whether the last match was zero-length. To break the loop, the following
1030match after a zero-length match is prohibited to have a length of zero.
1031This prohibition interacts with backtracking (see L<"Backtracking">),
1032and so the I<second best> match is chosen if the I<best> match is of
1033zero length.
1034
19799a22 1035For example:
c84d73f1
IZ
1036
1037 $_ = 'bar';
1038 s/\w??/<$&>/g;
1039
1040results in C<"<><b><><a><><r><>">. At each position of the string the best
1041match given by non-greedy C<??> is the zero-length match, and the I<second
1042best> match is what is matched by C<\w>. Thus zero-length matches
1043alternate with one-character-long matches.
1044
1045Similarly, for repeated C<m/()/g> the second-best match is the match at the
1046position one notch further in the string.
1047
19799a22 1048The additional state of being I<matched with zero-length> is associated with
c84d73f1
IZ
1049the matched string, and is reset by each assignment to pos().
1050
1051=head2 Creating custom RE engines
1052
1053Overloaded constants (see L<overload>) provide a simple way to extend
1054the functionality of the RE engine.
1055
1056Suppose that we want to enable a new RE escape-sequence C<\Y|> which
1057matches at boundary between white-space characters and non-whitespace
1058characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
1059at these positions, so we want to have each C<\Y|> in the place of the
1060more complicated version. We can create a module C<customre> to do
1061this:
1062
1063 package customre;
1064 use overload;
1065
1066 sub import {
1067 shift;
1068 die "No argument to customre::import allowed" if @_;
1069 overload::constant 'qr' => \&convert;
1070 }
1071
1072 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
1073
1074 my %rules = ( '\\' => '\\',
1075 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
1076 sub convert {
1077 my $re = shift;
1078 $re =~ s{
1079 \\ ( \\ | Y . )
1080 }
1081 { $rules{$1} or invalid($re,$1) }sgex;
1082 return $re;
1083 }
1084
1085Now C<use customre> enables the new escape in constant regular
1086expressions, i.e., those without any runtime variable interpolations.
1087As documented in L<overload>, this conversion will work only over
1088literal parts of regular expressions. For C<\Y|$re\Y|> the variable
1089part of this regular expression needs to be converted explicitly
1090(but only if the special meaning of C<\Y|> should be enabled inside $re):
1091
1092 use customre;
1093 $re = <>;
1094 chomp $re;
1095 $re = customre::convert $re;
1096 /\Y|$re\Y|/;
1097
19799a22
GS
1098=head1 BUGS
1099
1100This manpage is varies from difficult to understand to completely
1101and utterly opaque.
1102
1103=head1 SEE ALSO
9fa51da4 1104
9b599b2a
GS
1105L<perlop/"Regexp Quote-Like Operators">.
1106
1e66bd83
PP
1107L<perlop/"Gory details of parsing quoted constructs">.
1108
14218588
GS
1109L<perlfaq6>.
1110
9b599b2a
GS
1111L<perlfunc/pos>.
1112
1113L<perllocale>.
1114
14218588
GS
1115I<Mastering Regular Expressions> by Jeffrey Friedl, published
1116by O'Reilly and Associates.