3 perlre - Perl regular expressions
7 This page describes the syntax of regular expressions in Perl.
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>.
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
18 Matching operations can have various modifiers. Modifiers
19 that relate to the interpretation of the regular expression inside
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
22 L<perlop/"Gory details of parsing quoted constructs">.
28 Do case-insensitive pattern matching.
30 If C<use locale> is in effect, the case map is taken from the current
31 locale. See L<perllocale>.
35 Treat string as multiple lines. That is, change "^" and "$" from matching
36 the start or end of the string to matching the start or end of any
37 line anywhere within the string.
41 Treat string as single line. That is, change "." to match any character
42 whatsoever, even a newline, which normally it would not match.
44 The C</s> and C</m> modifiers both override the C<$*> 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,
49 while still allowing "^" and "$" to match, respectively, just after
50 and just before newlines within the string.
54 Extend your pattern's legibility by permitting whitespace and comments.
58 These are usually written as "the C</x> modifier", even though the delimiter
59 in question might not really be a slash. Any of these
60 modifiers may also be embedded within the regular expression itself using
61 the C<(?...)> construct. See below.
63 The C</x> modifier itself needs a little more explanation. It tells
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
66 your regular expression into (slightly) more readable parts. The C<#>
67 character is also treated as a metacharacter introducing a comment,
68 just as in ordinary Perl code. This also means that if you want real
69 whitespace or C<#> characters in the pattern (outside a character
70 class, where they are unaffected by C</x>), that you'll either have to
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
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
75 not intend to close the pattern early. See the C-comment deletion code
78 =head2 Regular Expressions
80 The patterns used in Perl pattern matching derive from supplied in
81 the Version 8 regex routines. (The routines are derived
82 (distantly) from Henry Spencer's freely redistributable reimplementation
83 of the V8 routines.) See L<Version 8 Regular Expressions> for
86 In particular the following metacharacters have their standard I<egrep>-ish
89 \ Quote the next metacharacter
90 ^ Match the beginning of the line
91 . Match any character (except newline)
92 $ Match the end of the line (or before newline at the end)
97 By default, the "^" character is guaranteed to match only the
98 beginning of the string, the "$" character only the end (or before the
99 newline at the end), and Perl does certain optimizations with the
100 assumption that the string contains only one line. Embedded newlines
101 will not be matched by "^" or "$". You may, however, wish to treat a
102 string as a multi-line buffer, such that the "^" will match after any
103 newline within the string, and "$" will match before any newline. At the
104 cost of a little more overhead, you can do this by using the /m modifier
105 on the pattern match operator. (Older programs did this by setting C<$*>,
106 but this practice is now deprecated.)
108 To simplify multi-line substitutions, the "." character never matches a
109 newline unless you use the C</s> modifier, which in effect tells Perl to pretend
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.
114 The following standard quantifiers are recognized:
116 * Match 0 or more times
117 + Match 1 or more 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
123 (If a curly bracket occurs in any other context, it is treated
124 as a regular character. In particular, the lower bound
125 is not optional.) The "*" modifier is equivalent to C<{0,}>, the "+"
126 modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
127 to integral values less than a preset limit defined when perl is built.
128 This is usually 32766 on the most common platforms. The actual limit can
129 be seen in the error message generated by code such as this:
131 $_ **= $_ , / {$_} / for 2 .. 42;
133 By default, a quantified subpattern is "greedy", that is, it will match as
134 many times as possible (given a particular starting location) while still
135 allowing the rest of the pattern to match. If you want it to match the
136 minimum number of times possible, follow the quantifier with a "?". Note
137 that the meanings don't change, just the "greediness":
139 *? Match 0 or more times
140 +? Match 1 or more times
142 {n}? Match exactly n times
143 {n,}? Match at least n times
144 {n,m}? Match at least n but not more than m times
146 Because patterns are processed as double quoted strings, the following
153 \a alarm (bell) (BEL)
154 \e escape (think troff) (ESC)
155 \033 octal char (think of a PDP-11)
157 \x{263a} wide hex char (Unicode SMILEY)
160 \l lowercase next char (think vi)
161 \u uppercase next char (think vi)
162 \L lowercase till \E (think vi)
163 \U uppercase till \E (think vi)
164 \E end case modification (think vi)
165 \Q quote (disable) pattern metacharacters till \E
167 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
168 and C<\U> is taken from the current locale. See L<perllocale>. For
169 documentation of C<\N{name}>, see L<charnames>.
171 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
172 An unescaped C<$> or C<@> interpolates the corresponding variable,
173 while escaping will cause the literal string C<\$> to be matched.
174 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
176 In addition, Perl defines the following:
178 \w Match a "word" character (alphanumeric plus "_")
179 \W Match a non-"word" character
180 \s Match a whitespace character
181 \S Match a non-whitespace character
182 \d Match a digit character
183 \D Match a non-digit character
184 \pP Match P, named property. Use \p{Prop} for longer names.
186 \X Match eXtended Unicode "combining character sequence",
187 equivalent to (?:\PM\pM*)
188 \C Match a single C char (octet) even under Unicode.
189 NOTE: breaks up characters into their UTF-8 bytes,
190 so you may end up with malformed pieces of UTF-8.
192 A C<\w> matches a single alphanumeric character (an alphabetic
193 character, or a decimal digit) or C<_>, not a whole word. Use C<\w+>
194 to match a string of Perl-identifier characters (which isn't the same
195 as matching an English word). If C<use locale> is in effect, the list
196 of alphabetic characters generated by C<\w> is taken from the current
197 locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
198 C<\d>, and C<\D> within character classes, but if you try to use them
199 as endpoints of a range, that's not a range, the "-" is understood
200 literally. If Unicode is in effect, C<\s> matches also "\x{85}",
201 "\x{2028}, and "\x{2029}", see L<perlunicode> for more details about
202 C<\pP>, C<\PP>, and C<\X>, and L<perluniintro> about Unicode in general.
203 You can define your own C<\p> and C<\P> propreties, see L<perlunicode>.
205 The POSIX character class syntax
209 is also available. The available classes and their backslash
210 equivalents (if available) are as follows:
231 A GNU extension equivalent to C<[ \t]>, `all horizontal whitespace'.
235 Not exactly equivalent to C<\s> since the C<[[:space:]]> includes
236 also the (very rare) `vertical tabulator', "\ck", chr(11).
240 A Perl extension, see above.
244 For example use C<[:upper:]> to match all the uppercase characters.
245 Note that the C<[]> are part of the C<[::]> construct, not part of the
246 whole character class. For example:
250 matches zero, one, any alphabetic character, and the percentage sign.
252 The following equivalences to Unicode \p{} constructs and equivalent
253 backslash character classes (if available), will hold:
255 [:...:] \p{...} backslash
273 For example C<[:lower:]> and C<\p{IsLower}> are equivalent.
275 If the C<utf8> pragma is not used but the C<locale> pragma is, the
276 classes correlate with the usual isalpha(3) interface (except for
279 The assumedly non-obviously named classes are:
285 Any control character. Usually characters that don't produce output as
286 such but instead control the terminal somehow: for example newline and
287 backspace are control characters. All characters with ord() less than
288 32 are most often classified as control characters (assuming ASCII,
289 the ISO Latin character sets, and Unicode), as is the character with
290 the ord() value of 127 (C<DEL>).
294 Any alphanumeric or punctuation (special) character.
298 Any alphanumeric or punctuation (special) character or the space character.
302 Any punctuation (special) character.
306 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
307 work just fine) it is included for completeness.
311 You can negate the [::] character classes by prefixing the class name
312 with a '^'. This is a Perl extension. For example:
314 POSIX traditional Unicode
316 [:^digit:] \D \P{IsDigit}
317 [:^space:] \S \P{IsSpace}
318 [:^word:] \W \P{IsWord}
320 Perl respects the POSIX standard in that POSIX character classes are
321 only supported within a character class. The POSIX character classes
322 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
323 use them will cause an error.
325 Perl defines the following zero-width assertions:
327 \b Match a word boundary
328 \B Match a non-(word boundary)
329 \A Match only at beginning of string
330 \Z Match only at end of string, or before newline at the end
331 \z Match only at end of string
332 \G Match only at pos() (e.g. at the end-of-match position
335 A word boundary (C<\b>) is a spot between two characters
336 that has a C<\w> on one side of it and a C<\W> on the other side
337 of it (in either order), counting the imaginary characters off the
338 beginning and end of the string as matching a C<\W>. (Within
339 character classes C<\b> represents backspace rather than a word
340 boundary, just as it normally does in any double-quoted string.)
341 The C<\A> and C<\Z> are just like "^" and "$", except that they
342 won't match multiple times when the C</m> modifier is used, while
343 "^" and "$" will match at every internal line boundary. To match
344 the actual end of the string and not ignore an optional trailing
347 The C<\G> assertion can be used to chain global matches (using
348 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
349 It is also useful when writing C<lex>-like scanners, when you have
350 several patterns that you want to match against consequent substrings
351 of your string, see the previous reference. The actual location
352 where C<\G> will match can also be influenced by using C<pos()> as
353 an lvalue: see L<perlfunc/pos>. Currently C<\G> is only fully
354 supported when anchored to the start of the pattern; while it
355 is permitted to use it elsewhere, as in C</(?<=\G..)./g>, some
356 such uses (C</.\G/g>, for example) currently cause problems, and
357 it is recommended that you avoid such usage for now.
359 The bracketing construct C<( ... )> creates capture buffers. To
360 refer to the digit'th buffer use \<digit> within the
361 match. Outside the match use "$" instead of "\". (The
362 \<digit> notation works in certain circumstances outside
363 the match. See the warning below about \1 vs $1 for details.)
364 Referring back to another part of the match is called a
367 There is no limit to the number of captured substrings that you may
368 use. However Perl also uses \10, \11, etc. as aliases for \010,
369 \011, etc. (Recall that 0 means octal, so \011 is the character at
370 number 9 in your coded character set; which would be the 10th character,
371 a horizontal tab under ASCII.) Perl resolves this
372 ambiguity by interpreting \10 as a backreference only if at least 10
373 left parentheses have opened before it. Likewise \11 is a
374 backreference only if at least 11 left parentheses have opened
375 before it. And so on. \1 through \9 are always interpreted as
380 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
382 if (/(.)\1/) { # find first doubled char
383 print "'$1' is the first doubled character\n";
386 if (/Time: (..):(..):(..)/) { # parse out values
392 Several special variables also refer back to portions of the previous
393 match. C<$+> returns whatever the last bracket match matched.
394 C<$&> returns the entire matched string. (At one point C<$0> did
395 also, but now it returns the name of the program.) C<$`> returns
396 everything before the matched string. C<$'> returns everything
397 after the matched string. And C<$^N> contains whatever was matched by
398 the most-recently closed group (submatch). C<$^N> can be used in
399 extended patterns (see below), for example to assign a submatch to a
402 The numbered variables ($1, $2, $3, etc.) and the related punctuation
403 set (C<$+>, C<$&>, C<$`>, C<$'>, and C<$^N>) are all dynamically scoped
404 until the end of the enclosing block or until the next successful
405 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
407 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
408 C<$'> anywhere in the program, it has to provide them for every
409 pattern match. This may substantially slow your program. Perl
410 uses the same mechanism to produce $1, $2, etc, so you also pay a
411 price for each pattern that contains capturing parentheses. (To
412 avoid this cost while retaining the grouping behaviour, use the
413 extended regular expression C<(?: ... )> instead.) But if you never
414 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
415 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
416 if you can, but if you can't (and some algorithms really appreciate
417 them), once you've used them once, use them at will, because you've
418 already paid the price. As of 5.005, C<$&> is not so costly as the
421 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
422 C<\w>, C<\n>. Unlike some other regular expression languages, there
423 are no backslashed symbols that aren't alphanumeric. So anything
424 that looks like \\, \(, \), \<, \>, \{, or \} is always
425 interpreted as a literal character, not a metacharacter. This was
426 once used in a common idiom to disable or quote the special meanings
427 of regular expression metacharacters in a string that you want to
428 use for a pattern. Simply quote all non-"word" characters:
430 $pattern =~ s/(\W)/\\$1/g;
432 (If C<use locale> is set, then this depends on the current locale.)
433 Today it is more common to use the quotemeta() function or the C<\Q>
434 metaquoting escape sequence to disable all metacharacters' special
437 /$unquoted\Q$quoted\E$unquoted/
439 Beware that if you put literal backslashes (those not inside
440 interpolated variables) between C<\Q> and C<\E>, double-quotish
441 backslash interpolation may lead to confusing results. If you
442 I<need> to use literal backslashes within C<\Q...\E>,
443 consult L<perlop/"Gory details of parsing quoted constructs">.
445 =head2 Extended Patterns
447 Perl also defines a consistent extension syntax for features not
448 found in standard tools like B<awk> and B<lex>. The syntax is a
449 pair of parentheses with a question mark as the first thing within
450 the parentheses. The character after the question mark indicates
453 The stability of these extensions varies widely. Some have been
454 part of the core language for many years. Others are experimental
455 and may change without warning or be completely removed. Check
456 the documentation on an individual feature to verify its current
459 A question mark was chosen for this and for the minimal-matching
460 construct because 1) question marks are rare in older regular
461 expressions, and 2) whenever you see one, you should stop and
462 "question" exactly what is going on. That's psychology...
468 A comment. The text is ignored. If the C</x> modifier enables
469 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
470 the comment as soon as it sees a C<)>, so there is no way to put a literal
473 =item C<(?imsx-imsx)>
475 One or more embedded pattern-match modifiers, to be turned on (or
476 turned off, if preceded by C<->) for the remainder of the pattern or
477 the remainder of the enclosing pattern group (if any). This is
478 particularly useful for dynamic patterns, such as those read in from a
479 configuration file, read in as an argument, are specified in a table
480 somewhere, etc. Consider the case that some of which want to be case
481 sensitive and some do not. The case insensitive ones need to include
482 merely C<(?i)> at the front of the pattern. For example:
485 if ( /$pattern/i ) { }
489 $pattern = "(?i)foobar";
490 if ( /$pattern/ ) { }
492 These modifiers are restored at the end of the enclosing group. For example,
496 will match a repeated (I<including the case>!) word C<blah> in any
497 case, assuming C<x> modifier, and no C<i> modifier outside this
502 =item C<(?imsx-imsx:pattern)>
504 This is for clustering, not capturing; it groups subexpressions like
505 "()", but doesn't make backreferences as "()" does. So
507 @fields = split(/\b(?:a|b|c)\b/)
511 @fields = split(/\b(a|b|c)\b/)
513 but doesn't spit out extra fields. It's also cheaper not to capture
514 characters if you don't need to.
516 Any letters between C<?> and C<:> act as flags modifiers as with
517 C<(?imsx-imsx)>. For example,
519 /(?s-i:more.*than).*million/i
521 is equivalent to the more verbose
523 /(?:(?s-i)more.*than).*million/i
527 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
528 matches a word followed by a tab, without including the tab in C<$&>.
532 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
533 matches any occurrence of "foo" that isn't followed by "bar". Note
534 however that look-ahead and look-behind are NOT the same thing. You cannot
535 use this for look-behind.
537 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
538 will not do what you want. That's because the C<(?!foo)> is just saying that
539 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
540 match. You would have to do something like C</(?!foo)...bar/> for that. We
541 say "like" because there's the case of your "bar" not having three characters
542 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
543 Sometimes it's still easier just to say:
545 if (/bar/ && $` !~ /foo$/)
547 For look-behind see below.
549 =item C<(?<=pattern)>
551 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
552 matches a word that follows a tab, without including the tab in C<$&>.
553 Works only for fixed-width look-behind.
555 =item C<(?<!pattern)>
557 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
558 matches any occurrence of "foo" that does not follow "bar". Works
559 only for fixed-width look-behind.
563 B<WARNING>: This extended regular expression feature is considered
564 highly experimental, and may be changed or deleted without notice.
566 This zero-width assertion evaluate any embedded Perl code. It
567 always succeeds, and its C<code> is not interpolated. Currently,
568 the rules to determine where the C<code> ends are somewhat convoluted.
570 This feature can be used together with the special variable C<$^N> to
571 capture the results of submatches in variables without having to keep
572 track of the number of nested parentheses. For example:
574 $_ = "The brown fox jumps over the lazy dog";
575 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
576 print "color = $color, animal = $animal\n";
578 The C<code> is properly scoped in the following sense: If the assertion
579 is backtracked (compare L<"Backtracking">), all changes introduced after
580 C<local>ization are undone, so that
584 (?{ $cnt = 0 }) # Initialize $cnt.
588 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
592 (?{ $res = $cnt }) # On success copy to non-localized
596 will set C<$res = 4>. Note that after the match, $cnt returns to the globally
597 introduced value, because the scopes that restrict C<local> operators
600 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
601 switch. If I<not> used in this way, the result of evaluation of
602 C<code> is put into the special variable C<$^R>. This happens
603 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
604 inside the same regular expression.
606 The assignment to C<$^R> above is properly localized, so the old
607 value of C<$^R> is restored if the assertion is backtracked; compare
610 For reasons of security, this construct is forbidden if the regular
611 expression involves run-time interpolation of variables, unless the
612 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
613 variables contain results of C<qr//> operator (see
614 L<perlop/"qr/STRING/imosx">).
616 This restriction is because of the wide-spread and remarkably convenient
617 custom of using run-time determined strings as patterns. For example:
623 Before Perl knew how to execute interpolated code within a pattern,
624 this operation was completely safe from a security point of view,
625 although it could raise an exception from an illegal pattern. If
626 you turn on the C<use re 'eval'>, though, it is no longer secure,
627 so you should only do so if you are also using taint checking.
628 Better yet, use the carefully constrained evaluation within a Safe
629 module. See L<perlsec> for details about both these mechanisms.
631 =item C<(??{ code })>
633 B<WARNING>: This extended regular expression feature is considered
634 highly experimental, and may be changed or deleted without notice.
635 A simplified version of the syntax may be introduced for commonly
638 This is a "postponed" regular subexpression. The C<code> is evaluated
639 at run time, at the moment this subexpression may match. The result
640 of evaluation is considered as a regular expression and matched as
641 if it were inserted instead of this construct.
643 The C<code> is not interpolated. As before, the rules to determine
644 where the C<code> ends are currently somewhat convoluted.
646 The following pattern matches a parenthesized group:
651 (?> [^()]+ ) # Non-parens without backtracking
653 (??{ $re }) # Group with matching parens
658 =item C<< (?>pattern) >>
660 B<WARNING>: This extended regular expression feature is considered
661 highly experimental, and may be changed or deleted without notice.
663 An "independent" subexpression, one which matches the substring
664 that a I<standalone> C<pattern> would match if anchored at the given
665 position, and it matches I<nothing other than this substring>. This
666 construct is useful for optimizations of what would otherwise be
667 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
668 It may also be useful in places where the "grab all you can, and do not
669 give anything back" semantic is desirable.
671 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
672 (anchored at the beginning of string, as above) will match I<all>
673 characters C<a> at the beginning of string, leaving no C<a> for
674 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
675 since the match of the subgroup C<a*> is influenced by the following
676 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
677 C<a*ab> will match fewer characters than a standalone C<a*>, since
678 this makes the tail match.
680 An effect similar to C<< (?>pattern) >> may be achieved by writing
681 C<(?=(pattern))\1>. This matches the same substring as a standalone
682 C<a+>, and the following C<\1> eats the matched string; it therefore
683 makes a zero-length assertion into an analogue of C<< (?>...) >>.
684 (The difference between these two constructs is that the second one
685 uses a capturing group, thus shifting ordinals of backreferences
686 in the rest of a regular expression.)
688 Consider this pattern:
699 That will efficiently match a nonempty group with matching parentheses
700 two levels deep or less. However, if there is no such group, it
701 will take virtually forever on a long string. That's because there
702 are so many different ways to split a long string into several
703 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
704 to a subpattern of the above pattern. Consider how the pattern
705 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
706 seconds, but that each extra letter doubles this time. This
707 exponential performance will make it appear that your program has
708 hung. However, a tiny change to this pattern
712 (?> [^()]+ ) # change x+ above to (?> x+ )
719 which uses C<< (?>...) >> matches exactly when the one above does (verifying
720 this yourself would be a productive exercise), but finishes in a fourth
721 the time when used on a similar string with 1000000 C<a>s. Be aware,
722 however, that this pattern currently triggers a warning message under
723 the C<use warnings> pragma or B<-w> switch saying it
724 C<"matches null string many times in regex">.
726 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
727 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
728 This was only 4 times slower on a string with 1000000 C<a>s.
730 The "grab all you can, and do not give anything back" semantic is desirable
731 in many situations where on the first sight a simple C<()*> looks like
732 the correct solution. Suppose we parse text with comments being delimited
733 by C<#> followed by some optional (horizontal) whitespace. Contrary to
734 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
735 the comment delimiter, because it may "give up" some whitespace if
736 the remainder of the pattern can be made to match that way. The correct
737 answer is either one of these:
742 For example, to grab non-empty comments into $1, one should use either
745 / (?> \# [ \t]* ) ( .+ ) /x;
746 / \# [ \t]* ( [^ \t] .* ) /x;
748 Which one you pick depends on which of these expressions better reflects
749 the above specification of comments.
751 =item C<(?(condition)yes-pattern|no-pattern)>
753 =item C<(?(condition)yes-pattern)>
755 B<WARNING>: This extended regular expression feature is considered
756 highly experimental, and may be changed or deleted without notice.
758 Conditional expression. C<(condition)> should be either an integer in
759 parentheses (which is valid if the corresponding pair of parentheses
760 matched), or look-ahead/look-behind/evaluate zero-width assertion.
769 matches a chunk of non-parentheses, possibly included in parentheses
776 NOTE: This section presents an abstract approximation of regular
777 expression behavior. For a more rigorous (and complicated) view of
778 the rules involved in selecting a match among possible alternatives,
779 see L<Combining pieces together>.
781 A fundamental feature of regular expression matching involves the
782 notion called I<backtracking>, which is currently used (when needed)
783 by all regular expression quantifiers, namely C<*>, C<*?>, C<+>,
784 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
785 internally, but the general principle outlined here is valid.
787 For a regular expression to match, the I<entire> regular expression must
788 match, not just part of it. So if the beginning of a pattern containing a
789 quantifier succeeds in a way that causes later parts in the pattern to
790 fail, the matching engine backs up and recalculates the beginning
791 part--that's why it's called backtracking.
793 Here is an example of backtracking: Let's say you want to find the
794 word following "foo" in the string "Food is on the foo table.":
796 $_ = "Food is on the foo table.";
797 if ( /\b(foo)\s+(\w+)/i ) {
798 print "$2 follows $1.\n";
801 When the match runs, the first part of the regular expression (C<\b(foo)>)
802 finds a possible match right at the beginning of the string, and loads up
803 $1 with "Foo". However, as soon as the matching engine sees that there's
804 no whitespace following the "Foo" that it had saved in $1, it realizes its
805 mistake and starts over again one character after where it had the
806 tentative match. This time it goes all the way until the next occurrence
807 of "foo". The complete regular expression matches this time, and you get
808 the expected output of "table follows foo."
810 Sometimes minimal matching can help a lot. Imagine you'd like to match
811 everything between "foo" and "bar". Initially, you write something
814 $_ = "The food is under the bar in the barn.";
815 if ( /foo(.*)bar/ ) {
819 Which perhaps unexpectedly yields:
821 got <d is under the bar in the >
823 That's because C<.*> was greedy, so you get everything between the
824 I<first> "foo" and the I<last> "bar". Here it's more effective
825 to use minimal matching to make sure you get the text between a "foo"
826 and the first "bar" thereafter.
828 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
829 got <d is under the >
831 Here's another example: let's say you'd like to match a number at the end
832 of a string, and you also want to keep the preceding part of the match.
835 $_ = "I have 2 numbers: 53147";
836 if ( /(.*)(\d*)/ ) { # Wrong!
837 print "Beginning is <$1>, number is <$2>.\n";
840 That won't work at all, because C<.*> was greedy and gobbled up the
841 whole string. As C<\d*> can match on an empty string the complete
842 regular expression matched successfully.
844 Beginning is <I have 2 numbers: 53147>, number is <>.
846 Here are some variants, most of which don't work:
848 $_ = "I have 2 numbers: 53147";
861 printf "%-12s ", $pat;
871 (.*)(\d*) <I have 2 numbers: 53147> <>
872 (.*)(\d+) <I have 2 numbers: 5314> <7>
874 (.*?)(\d+) <I have > <2>
875 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
876 (.*?)(\d+)$ <I have 2 numbers: > <53147>
877 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
878 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
880 As you see, this can be a bit tricky. It's important to realize that a
881 regular expression is merely a set of assertions that gives a definition
882 of success. There may be 0, 1, or several different ways that the
883 definition might succeed against a particular string. And if there are
884 multiple ways it might succeed, you need to understand backtracking to
885 know which variety of success you will achieve.
887 When using look-ahead assertions and negations, this can all get even
888 trickier. Imagine you'd like to find a sequence of non-digits not
889 followed by "123". You might try to write that as
892 if ( /^\D*(?!123)/ ) { # Wrong!
893 print "Yup, no 123 in $_\n";
896 But that isn't going to match; at least, not the way you're hoping. It
897 claims that there is no 123 in the string. Here's a clearer picture of
898 why that pattern matches, contrary to popular expectations:
903 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ;
904 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ;
906 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ;
907 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ;
915 You might have expected test 3 to fail because it seems to a more
916 general purpose version of test 1. The important difference between
917 them is that test 3 contains a quantifier (C<\D*>) and so can use
918 backtracking, whereas test 1 will not. What's happening is
919 that you've asked "Is it true that at the start of $x, following 0 or more
920 non-digits, you have something that's not 123?" If the pattern matcher had
921 let C<\D*> expand to "ABC", this would have caused the whole pattern to
924 The search engine will initially match C<\D*> with "ABC". Then it will
925 try to match C<(?!123> with "123", which fails. But because
926 a quantifier (C<\D*>) has been used in the regular expression, the
927 search engine can backtrack and retry the match differently
928 in the hope of matching the complete regular expression.
930 The pattern really, I<really> wants to succeed, so it uses the
931 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
932 time. Now there's indeed something following "AB" that is not
933 "123". It's "C123", which suffices.
935 We can deal with this by using both an assertion and a negation.
936 We'll say that the first part in $1 must be followed both by a digit
937 and by something that's not "123". Remember that the look-aheads
938 are zero-width expressions--they only look, but don't consume any
939 of the string in their match. So rewriting this way produces what
940 you'd expect; that is, case 5 will fail, but case 6 succeeds:
942 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ;
943 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ;
947 In other words, the two zero-width assertions next to each other work as though
948 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
949 matches only if you're at the beginning of the line AND the end of the
950 line simultaneously. The deeper underlying truth is that juxtaposition in
951 regular expressions always means AND, except when you write an explicit OR
952 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
953 although the attempted matches are made at different positions because "a"
954 is not a zero-width assertion, but a one-width assertion.
956 B<WARNING>: particularly complicated regular expressions can take
957 exponential time to solve because of the immense number of possible
958 ways they can use backtracking to try match. For example, without
959 internal optimizations done by the regular expression engine, this will
960 take a painfully long time to run:
962 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
964 And if you used C<*>'s in the internal groups instead of limiting them
965 to 0 through 5 matches, then it would take forever--or until you ran
966 out of stack space. Moreover, these internal optimizations are not
967 always applicable. For example, if you put C<{0,5}> instead of C<*>
968 on the external group, no current optimization is applicable, and the
969 match takes a long time to finish.
971 A powerful tool for optimizing such beasts is what is known as an
973 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
974 zero-length look-ahead/look-behind assertions will not backtrack to make
975 the tail match, since they are in "logical" context: only
976 whether they match is considered relevant. For an example
977 where side-effects of look-ahead I<might> have influenced the
978 following match, see L<C<< (?>pattern) >>>.
980 =head2 Version 8 Regular Expressions
982 In case you're not familiar with the "regular" Version 8 regex
983 routines, here are the pattern-matching rules not described above.
985 Any single character matches itself, unless it is a I<metacharacter>
986 with a special meaning described here or above. You can cause
987 characters that normally function as metacharacters to be interpreted
988 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
989 character; "\\" matches a "\"). A series of characters matches that
990 series of characters in the target string, so the pattern C<blurfl>
991 would match "blurfl" in the target string.
993 You can specify a character class, by enclosing a list of characters
994 in C<[]>, which will match any one character from the list. If the
995 first character after the "[" is "^", the class matches any character not
996 in the list. Within a list, the "-" character specifies a
997 range, so that C<a-z> represents all characters between "a" and "z",
998 inclusive. If you want either "-" or "]" itself to be a member of a
999 class, put it at the start of the list (possibly after a "^"), or
1000 escape it with a backslash. "-" is also taken literally when it is
1001 at the end of the list, just before the closing "]". (The
1002 following all specify the same class of three characters: C<[-az]>,
1003 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1004 specifies a class containing twenty-six characters, even on EBCDIC
1005 based coded character sets.) Also, if you try to use the character
1006 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1007 a range, that's not a range, the "-" is understood literally.
1009 Note also that the whole range idea is rather unportable between
1010 character sets--and even within character sets they may cause results
1011 you probably didn't expect. A sound principle is to use only ranges
1012 that begin from and end at either alphabets of equal case ([a-e],
1013 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1014 spell out the character sets in full.
1016 Characters may be specified using a metacharacter syntax much like that
1017 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1018 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1019 of octal digits, matches the character whose coded character set value
1020 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1021 matches the character whose numeric value is I<nn>. The expression \cI<x>
1022 matches the character control-I<x>. Finally, the "." metacharacter
1023 matches any character except "\n" (unless you use C</s>).
1025 You can specify a series of alternatives for a pattern using "|" to
1026 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1027 or "foe" in the target string (as would C<f(e|i|o)e>). The
1028 first alternative includes everything from the last pattern delimiter
1029 ("(", "[", or the beginning of the pattern) up to the first "|", and
1030 the last alternative contains everything from the last "|" to the next
1031 pattern delimiter. That's why it's common practice to include
1032 alternatives in parentheses: to minimize confusion about where they
1035 Alternatives are tried from left to right, so the first
1036 alternative found for which the entire expression matches, is the one that
1037 is chosen. This means that alternatives are not necessarily greedy. For
1038 example: when matching C<foo|foot> against "barefoot", only the "foo"
1039 part will match, as that is the first alternative tried, and it successfully
1040 matches the target string. (This might not seem important, but it is
1041 important when you are capturing matched text using parentheses.)
1043 Also remember that "|" is interpreted as a literal within square brackets,
1044 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1046 Within a pattern, you may designate subpatterns for later reference
1047 by enclosing them in parentheses, and you may refer back to the
1048 I<n>th subpattern later in the pattern using the metacharacter
1049 \I<n>. Subpatterns are numbered based on the left to right order
1050 of their opening parenthesis. A backreference matches whatever
1051 actually matched the subpattern in the string being examined, not
1052 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1053 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1054 1 matched "0x", even though the rule C<0|0x> could potentially match
1055 the leading 0 in the second number.
1057 =head2 Warning on \1 vs $1
1059 Some people get too used to writing things like:
1061 $pattern =~ s/(\W)/\\\1/g;
1063 This is grandfathered for the RHS of a substitute to avoid shocking the
1064 B<sed> addicts, but it's a dirty habit to get into. That's because in
1065 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1066 the usual double-quoted string means a control-A. The customary Unix
1067 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1068 of doing that, you get yourself into trouble if you then add an C</e>
1071 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1077 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1078 C<${1}000>. The operation of interpolation should not be confused
1079 with the operation of matching a backreference. Certainly they mean two
1080 different things on the I<left> side of the C<s///>.
1082 =head2 Repeated patterns matching zero-length substring
1084 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1086 Regular expressions provide a terse and powerful programming language. As
1087 with most other power tools, power comes together with the ability
1090 A common abuse of this power stems from the ability to make infinite
1091 loops using regular expressions, with something as innocuous as:
1093 'foo' =~ m{ ( o? )* }x;
1095 The C<o?> can match at the beginning of C<'foo'>, and since the position
1096 in the string is not moved by the match, C<o?> would match again and again
1097 because of the C<*> modifier. Another common way to create a similar cycle
1098 is with the looping modifier C<//g>:
1100 @matches = ( 'foo' =~ m{ o? }xg );
1104 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1106 or the loop implied by split().
1108 However, long experience has shown that many programming tasks may
1109 be significantly simplified by using repeated subexpressions that
1110 may match zero-length substrings. Here's a simple example being:
1112 @chars = split //, $string; # // is not magic in split
1113 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1115 Thus Perl allows such constructs, by I<forcefully breaking
1116 the infinite loop>. The rules for this are different for lower-level
1117 loops given by the greedy modifiers C<*+{}>, and for higher-level
1118 ones like the C</g> modifier or split() operator.
1120 The lower-level loops are I<interrupted> (that is, the loop is
1121 broken) when Perl detects that a repeated expression matched a
1122 zero-length substring. Thus
1124 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1126 is made equivalent to
1128 m{ (?: NON_ZERO_LENGTH )*
1133 The higher level-loops preserve an additional state between iterations:
1134 whether the last match was zero-length. To break the loop, the following
1135 match after a zero-length match is prohibited to have a length of zero.
1136 This prohibition interacts with backtracking (see L<"Backtracking">),
1137 and so the I<second best> match is chosen if the I<best> match is of
1145 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1146 match given by non-greedy C<??> is the zero-length match, and the I<second
1147 best> match is what is matched by C<\w>. Thus zero-length matches
1148 alternate with one-character-long matches.
1150 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1151 position one notch further in the string.
1153 The additional state of being I<matched with zero-length> is associated with
1154 the matched string, and is reset by each assignment to pos().
1155 Zero-length matches at the end of the previous match are ignored
1158 =head2 Combining pieces together
1160 Each of the elementary pieces of regular expressions which were described
1161 before (such as C<ab> or C<\Z>) could match at most one substring
1162 at the given position of the input string. However, in a typical regular
1163 expression these elementary pieces are combined into more complicated
1164 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1165 (in these examples C<S> and C<T> are regular subexpressions).
1167 Such combinations can include alternatives, leading to a problem of choice:
1168 if we match a regular expression C<a|ab> against C<"abc">, will it match
1169 substring C<"a"> or C<"ab">? One way to describe which substring is
1170 actually matched is the concept of backtracking (see L<"Backtracking">).
1171 However, this description is too low-level and makes you think
1172 in terms of a particular implementation.
1174 Another description starts with notions of "better"/"worse". All the
1175 substrings which may be matched by the given regular expression can be
1176 sorted from the "best" match to the "worst" match, and it is the "best"
1177 match which is chosen. This substitutes the question of "what is chosen?"
1178 by the question of "which matches are better, and which are worse?".
1180 Again, for elementary pieces there is no such question, since at most
1181 one match at a given position is possible. This section describes the
1182 notion of better/worse for combining operators. In the description
1183 below C<S> and C<T> are regular subexpressions.
1189 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1190 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1191 which can be matched by C<T>.
1193 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
1196 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
1197 C<B> is better match for C<T> than C<B'>.
1201 When C<S> can match, it is a better match than when only C<T> can match.
1203 Ordering of two matches for C<S> is the same as for C<S>. Similar for
1204 two matches for C<T>.
1206 =item C<S{REPEAT_COUNT}>
1208 Matches as C<SSS...S> (repeated as many times as necessary).
1212 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
1214 =item C<S{min,max}?>
1216 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
1218 =item C<S?>, C<S*>, C<S+>
1220 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
1222 =item C<S??>, C<S*?>, C<S+?>
1224 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
1228 Matches the best match for C<S> and only that.
1230 =item C<(?=S)>, C<(?<=S)>
1232 Only the best match for C<S> is considered. (This is important only if
1233 C<S> has capturing parentheses, and backreferences are used somewhere
1234 else in the whole regular expression.)
1236 =item C<(?!S)>, C<(?<!S)>
1238 For this grouping operator there is no need to describe the ordering, since
1239 only whether or not C<S> can match is important.
1241 =item C<(??{ EXPR })>
1243 The ordering is the same as for the regular expression which is
1246 =item C<(?(condition)yes-pattern|no-pattern)>
1248 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
1249 already determined. The ordering of the matches is the same as for the
1250 chosen subexpression.
1254 The above recipes describe the ordering of matches I<at a given position>.
1255 One more rule is needed to understand how a match is determined for the
1256 whole regular expression: a match at an earlier position is always better
1257 than a match at a later position.
1259 =head2 Creating custom RE engines
1261 Overloaded constants (see L<overload>) provide a simple way to extend
1262 the functionality of the RE engine.
1264 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
1265 matches at boundary between white-space characters and non-whitespace
1266 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
1267 at these positions, so we want to have each C<\Y|> in the place of the
1268 more complicated version. We can create a module C<customre> to do
1276 die "No argument to customre::import allowed" if @_;
1277 overload::constant 'qr' => \&convert;
1280 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
1282 my %rules = ( '\\' => '\\',
1283 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
1289 { $rules{$1} or invalid($re,$1) }sgex;
1293 Now C<use customre> enables the new escape in constant regular
1294 expressions, i.e., those without any runtime variable interpolations.
1295 As documented in L<overload>, this conversion will work only over
1296 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
1297 part of this regular expression needs to be converted explicitly
1298 (but only if the special meaning of C<\Y|> should be enabled inside $re):
1303 $re = customre::convert $re;
1308 This document varies from difficult to understand to completely
1309 and utterly opaque. The wandering prose riddled with jargon is
1310 hard to fathom in several places.
1312 This document needs a rewrite that separates the tutorial content
1313 from the reference content.
1321 L<perlop/"Regexp Quote-Like Operators">.
1323 L<perlop/"Gory details of parsing quoted constructs">.
1333 I<Mastering Regular Expressions> by Jeffrey Friedl, published
1334 by O'Reilly and Associates.