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.) The "*" modifier is equivalent to C<{0,}>, the "+"
125 modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
126 to integral values less than a preset limit defined when perl is built.
127 This is usually 32766 on the most common platforms. The actual limit can
128 be seen in the error message generated by code such as this:
130 $_ **= $_ , / {$_} / for 2 .. 42;
132 By default, a quantified subpattern is "greedy", that is, it will match as
133 many times as possible (given a particular starting location) while still
134 allowing the rest of the pattern to match. If you want it to match the
135 minimum number of times possible, follow the quantifier with a "?". Note
136 that the meanings don't change, just the "greediness":
138 *? Match 0 or more times
139 +? Match 1 or more times
141 {n}? Match exactly n times
142 {n,}? Match at least n times
143 {n,m}? Match at least n but not more than m times
145 Because patterns are processed as double quoted strings, the following
152 \a alarm (bell) (BEL)
153 \e escape (think troff) (ESC)
154 \033 octal char (think of a PDP-11)
156 \x{263a} wide hex char (Unicode SMILEY)
159 \l lowercase next char (think vi)
160 \u uppercase next char (think vi)
161 \L lowercase till \E (think vi)
162 \U uppercase till \E (think vi)
163 \E end case modification (think vi)
164 \Q quote (disable) pattern metacharacters till \E
166 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
167 and C<\U> is taken from the current locale. See L<perllocale>. For
168 documentation of C<\N{name}>, see L<charnames>.
170 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
171 An unescaped C<$> or C<@> interpolates the corresponding variable,
172 while escaping will cause the literal string C<\$> to be matched.
173 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
175 In addition, Perl defines the following:
177 \w Match a "word" character (alphanumeric plus "_")
178 \W Match a non-"word" character
179 \s Match a whitespace character
180 \S Match a non-whitespace character
181 \d Match a digit character
182 \D Match a non-digit character
183 \pP Match P, named property. Use \p{Prop} for longer names.
185 \X Match eXtended Unicode "combining character sequence",
186 equivalent to (?:\PM\pM*)
187 \C Match a single C char (octet) even under Unicode.
188 NOTE: breaks up characters into their UTF-8 bytes,
189 so you may end up with malformed pieces of UTF-8.
191 A C<\w> matches a single alphanumeric character (an alphabetic
192 character, or a decimal digit) or C<_>, not a whole word. Use C<\w+>
193 to match a string of Perl-identifier characters (which isn't the same
194 as matching an English word). If C<use locale> is in effect, the list
195 of alphabetic characters generated by C<\w> is taken from the current
196 locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
197 C<\d>, and C<\D> within character classes, but if you try to use them
198 as endpoints of a range, that's not a range, the "-" is understood
199 literally. If Unicode is in effect, C<\s> matches also "\x{85}",
200 "\x{2028}, and "\x{2029}", see L<perlunicode> for more details about
201 C<\pP>, C<\PP>, and C<\X>, and L<perluniintro> about Unicode in general.
202 You can define your own C<\p> and C<\P> propreties, see L<perlunicode>.
204 The POSIX character class syntax
208 is also available. The available classes and their backslash
209 equivalents (if available) are as follows:
230 A GNU extension equivalent to C<[ \t]>, `all horizontal whitespace'.
234 Not exactly equivalent to C<\s> since the C<[[:space:]]> includes
235 also the (very rare) `vertical tabulator', "\ck", chr(11).
239 A Perl extension, see above.
243 For example use C<[:upper:]> to match all the uppercase characters.
244 Note that the C<[]> are part of the C<[::]> construct, not part of the
245 whole character class. For example:
249 matches zero, one, any alphabetic character, and the percentage sign.
251 The following equivalences to Unicode \p{} constructs and equivalent
252 backslash character classes (if available), will hold:
254 [:...:] \p{...} backslash
272 For example C<[:lower:]> and C<\p{IsLower}> are equivalent.
274 If the C<utf8> pragma is not used but the C<locale> pragma is, the
275 classes correlate with the usual isalpha(3) interface (except for
278 The assumedly non-obviously named classes are:
284 Any control character. Usually characters that don't produce output as
285 such but instead control the terminal somehow: for example newline and
286 backspace are control characters. All characters with ord() less than
287 32 are most often classified as control characters (assuming ASCII,
288 the ISO Latin character sets, and Unicode), as is the character with
289 the ord() value of 127 (C<DEL>).
293 Any alphanumeric or punctuation (special) character.
297 Any alphanumeric or punctuation (special) character or the space character.
301 Any punctuation (special) character.
305 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
306 work just fine) it is included for completeness.
310 You can negate the [::] character classes by prefixing the class name
311 with a '^'. This is a Perl extension. For example:
313 POSIX traditional Unicode
315 [:^digit:] \D \P{IsDigit}
316 [:^space:] \S \P{IsSpace}
317 [:^word:] \W \P{IsWord}
319 Perl respects the POSIX standard in that POSIX character classes are
320 only supported within a character class. The POSIX character classes
321 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
322 use them will cause an error.
324 Perl defines the following zero-width assertions:
326 \b Match a word boundary
327 \B Match a non-(word boundary)
328 \A Match only at beginning of string
329 \Z Match only at end of string, or before newline at the end
330 \z Match only at end of string
331 \G Match only at pos() (e.g. at the end-of-match position
334 A word boundary (C<\b>) is a spot between two characters
335 that has a C<\w> on one side of it and a C<\W> on the other side
336 of it (in either order), counting the imaginary characters off the
337 beginning and end of the string as matching a C<\W>. (Within
338 character classes C<\b> represents backspace rather than a word
339 boundary, just as it normally does in any double-quoted string.)
340 The C<\A> and C<\Z> are just like "^" and "$", except that they
341 won't match multiple times when the C</m> modifier is used, while
342 "^" and "$" will match at every internal line boundary. To match
343 the actual end of the string and not ignore an optional trailing
346 The C<\G> assertion can be used to chain global matches (using
347 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
348 It is also useful when writing C<lex>-like scanners, when you have
349 several patterns that you want to match against consequent substrings
350 of your string, see the previous reference. The actual location
351 where C<\G> will match can also be influenced by using C<pos()> as
352 an lvalue: see L<perlfunc/pos>. Currently C<\G> is only fully
353 supported when anchored to the start of the pattern; while it
354 is permitted to use it elsewhere, as in C</(?<=\G..)./g>, some
355 such uses (C</.\G/g>, for example) currently cause problems, and
356 it is recommended that you avoid such usage for now.
358 The bracketing construct C<( ... )> creates capture buffers. To
359 refer to the digit'th buffer use \<digit> within the
360 match. Outside the match use "$" instead of "\". (The
361 \<digit> notation works in certain circumstances outside
362 the match. See the warning below about \1 vs $1 for details.)
363 Referring back to another part of the match is called a
366 There is no limit to the number of captured substrings that you may
367 use. However Perl also uses \10, \11, etc. as aliases for \010,
368 \011, etc. (Recall that 0 means octal, so \011 is the character at
369 number 9 in your coded character set; which would be the 10th character,
370 a horizontal tab under ASCII.) Perl resolves this
371 ambiguity by interpreting \10 as a backreference only if at least 10
372 left parentheses have opened before it. Likewise \11 is a
373 backreference only if at least 11 left parentheses have opened
374 before it. And so on. \1 through \9 are always interpreted as
379 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
381 if (/(.)\1/) { # find first doubled char
382 print "'$1' is the first doubled character\n";
385 if (/Time: (..):(..):(..)/) { # parse out values
391 Several special variables also refer back to portions of the previous
392 match. C<$+> returns whatever the last bracket match matched.
393 C<$&> returns the entire matched string. (At one point C<$0> did
394 also, but now it returns the name of the program.) C<$`> returns
395 everything before the matched string. And C<$'> returns everything
396 after the matched string.
398 The numbered variables ($1, $2, $3, etc.) and the related punctuation
399 set (C<$+>, C<$&>, C<$`>, and C<$'>) are all dynamically scoped
400 until the end of the enclosing block or until the next successful
401 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
403 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
404 C<$'> anywhere in the program, it has to provide them for every
405 pattern match. This may substantially slow your program. Perl
406 uses the same mechanism to produce $1, $2, etc, so you also pay a
407 price for each pattern that contains capturing parentheses. (To
408 avoid this cost while retaining the grouping behaviour, use the
409 extended regular expression C<(?: ... )> instead.) But if you never
410 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
411 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
412 if you can, but if you can't (and some algorithms really appreciate
413 them), once you've used them once, use them at will, because you've
414 already paid the price. As of 5.005, C<$&> is not so costly as the
417 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
418 C<\w>, C<\n>. Unlike some other regular expression languages, there
419 are no backslashed symbols that aren't alphanumeric. So anything
420 that looks like \\, \(, \), \<, \>, \{, or \} is always
421 interpreted as a literal character, not a metacharacter. This was
422 once used in a common idiom to disable or quote the special meanings
423 of regular expression metacharacters in a string that you want to
424 use for a pattern. Simply quote all non-"word" characters:
426 $pattern =~ s/(\W)/\\$1/g;
428 (If C<use locale> is set, then this depends on the current locale.)
429 Today it is more common to use the quotemeta() function or the C<\Q>
430 metaquoting escape sequence to disable all metacharacters' special
433 /$unquoted\Q$quoted\E$unquoted/
435 Beware that if you put literal backslashes (those not inside
436 interpolated variables) between C<\Q> and C<\E>, double-quotish
437 backslash interpolation may lead to confusing results. If you
438 I<need> to use literal backslashes within C<\Q...\E>,
439 consult L<perlop/"Gory details of parsing quoted constructs">.
441 =head2 Extended Patterns
443 Perl also defines a consistent extension syntax for features not
444 found in standard tools like B<awk> and B<lex>. The syntax is a
445 pair of parentheses with a question mark as the first thing within
446 the parentheses. The character after the question mark indicates
449 The stability of these extensions varies widely. Some have been
450 part of the core language for many years. Others are experimental
451 and may change without warning or be completely removed. Check
452 the documentation on an individual feature to verify its current
455 A question mark was chosen for this and for the minimal-matching
456 construct because 1) question marks are rare in older regular
457 expressions, and 2) whenever you see one, you should stop and
458 "question" exactly what is going on. That's psychology...
464 A comment. The text is ignored. If the C</x> modifier enables
465 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
466 the comment as soon as it sees a C<)>, so there is no way to put a literal
469 =item C<(?imsx-imsx)>
471 One or more embedded pattern-match modifiers, to be turned on (or
472 turned off, if preceded by C<->) for the remainder of the pattern or
473 the remainder of the enclosing pattern group (if any). This is
474 particularly useful for dynamic patterns, such as those read in from a
475 configuration file, read in as an argument, are specified in a table
476 somewhere, etc. Consider the case that some of which want to be case
477 sensitive and some do not. The case insensitive ones need to include
478 merely C<(?i)> at the front of the pattern. For example:
481 if ( /$pattern/i ) { }
485 $pattern = "(?i)foobar";
486 if ( /$pattern/ ) { }
488 These modifiers are restored at the end of the enclosing group. For example,
492 will match a repeated (I<including the case>!) word C<blah> in any
493 case, assuming C<x> modifier, and no C<i> modifier outside this
498 =item C<(?imsx-imsx:pattern)>
500 This is for clustering, not capturing; it groups subexpressions like
501 "()", but doesn't make backreferences as "()" does. So
503 @fields = split(/\b(?:a|b|c)\b/)
507 @fields = split(/\b(a|b|c)\b/)
509 but doesn't spit out extra fields. It's also cheaper not to capture
510 characters if you don't need to.
512 Any letters between C<?> and C<:> act as flags modifiers as with
513 C<(?imsx-imsx)>. For example,
515 /(?s-i:more.*than).*million/i
517 is equivalent to the more verbose
519 /(?:(?s-i)more.*than).*million/i
523 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
524 matches a word followed by a tab, without including the tab in C<$&>.
528 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
529 matches any occurrence of "foo" that isn't followed by "bar". Note
530 however that look-ahead and look-behind are NOT the same thing. You cannot
531 use this for look-behind.
533 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
534 will not do what you want. That's because the C<(?!foo)> is just saying that
535 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
536 match. You would have to do something like C</(?!foo)...bar/> for that. We
537 say "like" because there's the case of your "bar" not having three characters
538 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
539 Sometimes it's still easier just to say:
541 if (/bar/ && $` !~ /foo$/)
543 For look-behind see below.
545 =item C<(?<=pattern)>
547 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
548 matches a word that follows a tab, without including the tab in C<$&>.
549 Works only for fixed-width look-behind.
551 =item C<(?<!pattern)>
553 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
554 matches any occurrence of "foo" that does not follow "bar". Works
555 only for fixed-width look-behind.
559 B<WARNING>: This extended regular expression feature is considered
560 highly experimental, and may be changed or deleted without notice.
562 This zero-width assertion evaluate any embedded Perl code. It
563 always succeeds, and its C<code> is not interpolated. Currently,
564 the rules to determine where the C<code> ends are somewhat convoluted.
566 The C<code> is properly scoped in the following sense: If the assertion
567 is backtracked (compare L<"Backtracking">), all changes introduced after
568 C<local>ization are undone, so that
572 (?{ $cnt = 0 }) # Initialize $cnt.
576 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
580 (?{ $res = $cnt }) # On success copy to non-localized
584 will set C<$res = 4>. Note that after the match, $cnt returns to the globally
585 introduced value, because the scopes that restrict C<local> operators
588 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
589 switch. If I<not> used in this way, the result of evaluation of
590 C<code> is put into the special variable C<$^R>. This happens
591 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
592 inside the same regular expression.
594 The assignment to C<$^R> above is properly localized, so the old
595 value of C<$^R> is restored if the assertion is backtracked; compare
598 For reasons of security, this construct is forbidden if the regular
599 expression involves run-time interpolation of variables, unless the
600 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
601 variables contain results of C<qr//> operator (see
602 L<perlop/"qr/STRING/imosx">).
604 This restriction is because of the wide-spread and remarkably convenient
605 custom of using run-time determined strings as patterns. For example:
611 Before Perl knew how to execute interpolated code within a pattern,
612 this operation was completely safe from a security point of view,
613 although it could raise an exception from an illegal pattern. If
614 you turn on the C<use re 'eval'>, though, it is no longer secure,
615 so you should only do so if you are also using taint checking.
616 Better yet, use the carefully constrained evaluation within a Safe
617 module. See L<perlsec> for details about both these mechanisms.
619 =item C<(??{ code })>
621 B<WARNING>: This extended regular expression feature is considered
622 highly experimental, and may be changed or deleted without notice.
623 A simplified version of the syntax may be introduced for commonly
626 This is a "postponed" regular subexpression. The C<code> is evaluated
627 at run time, at the moment this subexpression may match. The result
628 of evaluation is considered as a regular expression and matched as
629 if it were inserted instead of this construct.
631 The C<code> is not interpolated. As before, the rules to determine
632 where the C<code> ends are currently somewhat convoluted.
634 The following pattern matches a parenthesized group:
639 (?> [^()]+ ) # Non-parens without backtracking
641 (??{ $re }) # Group with matching parens
646 =item C<< (?>pattern) >>
648 B<WARNING>: This extended regular expression feature is considered
649 highly experimental, and may be changed or deleted without notice.
651 An "independent" subexpression, one which matches the substring
652 that a I<standalone> C<pattern> would match if anchored at the given
653 position, and it matches I<nothing other than this substring>. This
654 construct is useful for optimizations of what would otherwise be
655 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
656 It may also be useful in places where the "grab all you can, and do not
657 give anything back" semantic is desirable.
659 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
660 (anchored at the beginning of string, as above) will match I<all>
661 characters C<a> at the beginning of string, leaving no C<a> for
662 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
663 since the match of the subgroup C<a*> is influenced by the following
664 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
665 C<a*ab> will match fewer characters than a standalone C<a*>, since
666 this makes the tail match.
668 An effect similar to C<< (?>pattern) >> may be achieved by writing
669 C<(?=(pattern))\1>. This matches the same substring as a standalone
670 C<a+>, and the following C<\1> eats the matched string; it therefore
671 makes a zero-length assertion into an analogue of C<< (?>...) >>.
672 (The difference between these two constructs is that the second one
673 uses a capturing group, thus shifting ordinals of backreferences
674 in the rest of a regular expression.)
676 Consider this pattern:
687 That will efficiently match a nonempty group with matching parentheses
688 two levels deep or less. However, if there is no such group, it
689 will take virtually forever on a long string. That's because there
690 are so many different ways to split a long string into several
691 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
692 to a subpattern of the above pattern. Consider how the pattern
693 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
694 seconds, but that each extra letter doubles this time. This
695 exponential performance will make it appear that your program has
696 hung. However, a tiny change to this pattern
700 (?> [^()]+ ) # change x+ above to (?> x+ )
707 which uses C<< (?>...) >> matches exactly when the one above does (verifying
708 this yourself would be a productive exercise), but finishes in a fourth
709 the time when used on a similar string with 1000000 C<a>s. Be aware,
710 however, that this pattern currently triggers a warning message under
711 the C<use warnings> pragma or B<-w> switch saying it
712 C<"matches null string many times in regex">.
714 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
715 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
716 This was only 4 times slower on a string with 1000000 C<a>s.
718 The "grab all you can, and do not give anything back" semantic is desirable
719 in many situations where on the first sight a simple C<()*> looks like
720 the correct solution. Suppose we parse text with comments being delimited
721 by C<#> followed by some optional (horizontal) whitespace. Contrary to
722 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
723 the comment delimiter, because it may "give up" some whitespace if
724 the remainder of the pattern can be made to match that way. The correct
725 answer is either one of these:
730 For example, to grab non-empty comments into $1, one should use either
733 / (?> \# [ \t]* ) ( .+ ) /x;
734 / \# [ \t]* ( [^ \t] .* ) /x;
736 Which one you pick depends on which of these expressions better reflects
737 the above specification of comments.
739 =item C<(?(condition)yes-pattern|no-pattern)>
741 =item C<(?(condition)yes-pattern)>
743 B<WARNING>: This extended regular expression feature is considered
744 highly experimental, and may be changed or deleted without notice.
746 Conditional expression. C<(condition)> should be either an integer in
747 parentheses (which is valid if the corresponding pair of parentheses
748 matched), or look-ahead/look-behind/evaluate zero-width assertion.
757 matches a chunk of non-parentheses, possibly included in parentheses
764 NOTE: This section presents an abstract approximation of regular
765 expression behavior. For a more rigorous (and complicated) view of
766 the rules involved in selecting a match among possible alternatives,
767 see L<Combining pieces together>.
769 A fundamental feature of regular expression matching involves the
770 notion called I<backtracking>, which is currently used (when needed)
771 by all regular expression quantifiers, namely C<*>, C<*?>, C<+>,
772 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
773 internally, but the general principle outlined here is valid.
775 For a regular expression to match, the I<entire> regular expression must
776 match, not just part of it. So if the beginning of a pattern containing a
777 quantifier succeeds in a way that causes later parts in the pattern to
778 fail, the matching engine backs up and recalculates the beginning
779 part--that's why it's called backtracking.
781 Here is an example of backtracking: Let's say you want to find the
782 word following "foo" in the string "Food is on the foo table.":
784 $_ = "Food is on the foo table.";
785 if ( /\b(foo)\s+(\w+)/i ) {
786 print "$2 follows $1.\n";
789 When the match runs, the first part of the regular expression (C<\b(foo)>)
790 finds a possible match right at the beginning of the string, and loads up
791 $1 with "Foo". However, as soon as the matching engine sees that there's
792 no whitespace following the "Foo" that it had saved in $1, it realizes its
793 mistake and starts over again one character after where it had the
794 tentative match. This time it goes all the way until the next occurrence
795 of "foo". The complete regular expression matches this time, and you get
796 the expected output of "table follows foo."
798 Sometimes minimal matching can help a lot. Imagine you'd like to match
799 everything between "foo" and "bar". Initially, you write something
802 $_ = "The food is under the bar in the barn.";
803 if ( /foo(.*)bar/ ) {
807 Which perhaps unexpectedly yields:
809 got <d is under the bar in the >
811 That's because C<.*> was greedy, so you get everything between the
812 I<first> "foo" and the I<last> "bar". Here it's more effective
813 to use minimal matching to make sure you get the text between a "foo"
814 and the first "bar" thereafter.
816 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
817 got <d is under the >
819 Here's another example: let's say you'd like to match a number at the end
820 of a string, and you also want to keep the preceding part of the match.
823 $_ = "I have 2 numbers: 53147";
824 if ( /(.*)(\d*)/ ) { # Wrong!
825 print "Beginning is <$1>, number is <$2>.\n";
828 That won't work at all, because C<.*> was greedy and gobbled up the
829 whole string. As C<\d*> can match on an empty string the complete
830 regular expression matched successfully.
832 Beginning is <I have 2 numbers: 53147>, number is <>.
834 Here are some variants, most of which don't work:
836 $_ = "I have 2 numbers: 53147";
849 printf "%-12s ", $pat;
859 (.*)(\d*) <I have 2 numbers: 53147> <>
860 (.*)(\d+) <I have 2 numbers: 5314> <7>
862 (.*?)(\d+) <I have > <2>
863 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
864 (.*?)(\d+)$ <I have 2 numbers: > <53147>
865 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
866 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
868 As you see, this can be a bit tricky. It's important to realize that a
869 regular expression is merely a set of assertions that gives a definition
870 of success. There may be 0, 1, or several different ways that the
871 definition might succeed against a particular string. And if there are
872 multiple ways it might succeed, you need to understand backtracking to
873 know which variety of success you will achieve.
875 When using look-ahead assertions and negations, this can all get even
876 tricker. Imagine you'd like to find a sequence of non-digits not
877 followed by "123". You might try to write that as
880 if ( /^\D*(?!123)/ ) { # Wrong!
881 print "Yup, no 123 in $_\n";
884 But that isn't going to match; at least, not the way you're hoping. It
885 claims that there is no 123 in the string. Here's a clearer picture of
886 why that pattern matches, contrary to popular expectations:
891 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ;
892 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ;
894 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ;
895 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ;
903 You might have expected test 3 to fail because it seems to a more
904 general purpose version of test 1. The important difference between
905 them is that test 3 contains a quantifier (C<\D*>) and so can use
906 backtracking, whereas test 1 will not. What's happening is
907 that you've asked "Is it true that at the start of $x, following 0 or more
908 non-digits, you have something that's not 123?" If the pattern matcher had
909 let C<\D*> expand to "ABC", this would have caused the whole pattern to
912 The search engine will initially match C<\D*> with "ABC". Then it will
913 try to match C<(?!123> with "123", which fails. But because
914 a quantifier (C<\D*>) has been used in the regular expression, the
915 search engine can backtrack and retry the match differently
916 in the hope of matching the complete regular expression.
918 The pattern really, I<really> wants to succeed, so it uses the
919 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
920 time. Now there's indeed something following "AB" that is not
921 "123". It's "C123", which suffices.
923 We can deal with this by using both an assertion and a negation.
924 We'll say that the first part in $1 must be followed both by a digit
925 and by something that's not "123". Remember that the look-aheads
926 are zero-width expressions--they only look, but don't consume any
927 of the string in their match. So rewriting this way produces what
928 you'd expect; that is, case 5 will fail, but case 6 succeeds:
930 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ;
931 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ;
935 In other words, the two zero-width assertions next to each other work as though
936 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
937 matches only if you're at the beginning of the line AND the end of the
938 line simultaneously. The deeper underlying truth is that juxtaposition in
939 regular expressions always means AND, except when you write an explicit OR
940 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
941 although the attempted matches are made at different positions because "a"
942 is not a zero-width assertion, but a one-width assertion.
944 B<WARNING>: particularly complicated regular expressions can take
945 exponential time to solve because of the immense number of possible
946 ways they can use backtracking to try match. For example, without
947 internal optimizations done by the regular expression engine, this will
948 take a painfully long time to run:
950 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
952 And if you used C<*>'s in the internal groups instead of limiting them
953 to 0 through 5 matches, then it would take forever--or until you ran
954 out of stack space. Moreover, these internal optimizations are not
955 always applicable. For example, if you put C<{0,5}> instead of C<*>
956 on the external group, no current optimization is applicable, and the
957 match takes a long time to finish.
959 A powerful tool for optimizing such beasts is what is known as an
961 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
962 zero-length look-ahead/look-behind assertions will not backtrack to make
963 the tail match, since they are in "logical" context: only
964 whether they match is considered relevant. For an example
965 where side-effects of look-ahead I<might> have influenced the
966 following match, see L<C<< (?>pattern) >>>.
968 =head2 Version 8 Regular Expressions
970 In case you're not familiar with the "regular" Version 8 regex
971 routines, here are the pattern-matching rules not described above.
973 Any single character matches itself, unless it is a I<metacharacter>
974 with a special meaning described here or above. You can cause
975 characters that normally function as metacharacters to be interpreted
976 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
977 character; "\\" matches a "\"). A series of characters matches that
978 series of characters in the target string, so the pattern C<blurfl>
979 would match "blurfl" in the target string.
981 You can specify a character class, by enclosing a list of characters
982 in C<[]>, which will match any one character from the list. If the
983 first character after the "[" is "^", the class matches any character not
984 in the list. Within a list, the "-" character specifies a
985 range, so that C<a-z> represents all characters between "a" and "z",
986 inclusive. If you want either "-" or "]" itself to be a member of a
987 class, put it at the start of the list (possibly after a "^"), or
988 escape it with a backslash. "-" is also taken literally when it is
989 at the end of the list, just before the closing "]". (The
990 following all specify the same class of three characters: C<[-az]>,
991 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
992 specifies a class containing twenty-six characters, even on EBCDIC
993 based coded character sets.) Also, if you try to use the character
994 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
995 a range, that's not a range, the "-" is understood literally.
997 Note also that the whole range idea is rather unportable between
998 character sets--and even within character sets they may cause results
999 you probably didn't expect. A sound principle is to use only ranges
1000 that begin from and end at either alphabets of equal case ([a-e],
1001 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1002 spell out the character sets in full.
1004 Characters may be specified using a metacharacter syntax much like that
1005 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1006 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1007 of octal digits, matches the character whose coded character set value
1008 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1009 matches the character whose numeric value is I<nn>. The expression \cI<x>
1010 matches the character control-I<x>. Finally, the "." metacharacter
1011 matches any character except "\n" (unless you use C</s>).
1013 You can specify a series of alternatives for a pattern using "|" to
1014 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1015 or "foe" in the target string (as would C<f(e|i|o)e>). The
1016 first alternative includes everything from the last pattern delimiter
1017 ("(", "[", or the beginning of the pattern) up to the first "|", and
1018 the last alternative contains everything from the last "|" to the next
1019 pattern delimiter. That's why it's common practice to include
1020 alternatives in parentheses: to minimize confusion about where they
1023 Alternatives are tried from left to right, so the first
1024 alternative found for which the entire expression matches, is the one that
1025 is chosen. This means that alternatives are not necessarily greedy. For
1026 example: when matching C<foo|foot> against "barefoot", only the "foo"
1027 part will match, as that is the first alternative tried, and it successfully
1028 matches the target string. (This might not seem important, but it is
1029 important when you are capturing matched text using parentheses.)
1031 Also remember that "|" is interpreted as a literal within square brackets,
1032 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1034 Within a pattern, you may designate subpatterns for later reference
1035 by enclosing them in parentheses, and you may refer back to the
1036 I<n>th subpattern later in the pattern using the metacharacter
1037 \I<n>. Subpatterns are numbered based on the left to right order
1038 of their opening parenthesis. A backreference matches whatever
1039 actually matched the subpattern in the string being examined, not
1040 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1041 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1042 1 matched "0x", even though the rule C<0|0x> could potentially match
1043 the leading 0 in the second number.
1045 =head2 Warning on \1 vs $1
1047 Some people get too used to writing things like:
1049 $pattern =~ s/(\W)/\\\1/g;
1051 This is grandfathered for the RHS of a substitute to avoid shocking the
1052 B<sed> addicts, but it's a dirty habit to get into. That's because in
1053 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1054 the usual double-quoted string means a control-A. The customary Unix
1055 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1056 of doing that, you get yourself into trouble if you then add an C</e>
1059 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1065 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1066 C<${1}000>. The operation of interpolation should not be confused
1067 with the operation of matching a backreference. Certainly they mean two
1068 different things on the I<left> side of the C<s///>.
1070 =head2 Repeated patterns matching zero-length substring
1072 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1074 Regular expressions provide a terse and powerful programming language. As
1075 with most other power tools, power comes together with the ability
1078 A common abuse of this power stems from the ability to make infinite
1079 loops using regular expressions, with something as innocuous as:
1081 'foo' =~ m{ ( o? )* }x;
1083 The C<o?> can match at the beginning of C<'foo'>, and since the position
1084 in the string is not moved by the match, C<o?> would match again and again
1085 because of the C<*> modifier. Another common way to create a similar cycle
1086 is with the looping modifier C<//g>:
1088 @matches = ( 'foo' =~ m{ o? }xg );
1092 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1094 or the loop implied by split().
1096 However, long experience has shown that many programming tasks may
1097 be significantly simplified by using repeated subexpressions that
1098 may match zero-length substrings. Here's a simple example being:
1100 @chars = split //, $string; # // is not magic in split
1101 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1103 Thus Perl allows such constructs, by I<forcefully breaking
1104 the infinite loop>. The rules for this are different for lower-level
1105 loops given by the greedy modifiers C<*+{}>, and for higher-level
1106 ones like the C</g> modifier or split() operator.
1108 The lower-level loops are I<interrupted> (that is, the loop is
1109 broken) when Perl detects that a repeated expression matched a
1110 zero-length substring. Thus
1112 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1114 is made equivalent to
1116 m{ (?: NON_ZERO_LENGTH )*
1121 The higher level-loops preserve an additional state between iterations:
1122 whether the last match was zero-length. To break the loop, the following
1123 match after a zero-length match is prohibited to have a length of zero.
1124 This prohibition interacts with backtracking (see L<"Backtracking">),
1125 and so the I<second best> match is chosen if the I<best> match is of
1133 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1134 match given by non-greedy C<??> is the zero-length match, and the I<second
1135 best> match is what is matched by C<\w>. Thus zero-length matches
1136 alternate with one-character-long matches.
1138 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1139 position one notch further in the string.
1141 The additional state of being I<matched with zero-length> is associated with
1142 the matched string, and is reset by each assignment to pos().
1143 Zero-length matches at the end of the previous match are ignored
1146 =head2 Combining pieces together
1148 Each of the elementary pieces of regular expressions which were described
1149 before (such as C<ab> or C<\Z>) could match at most one substring
1150 at the given position of the input string. However, in a typical regular
1151 expression these elementary pieces are combined into more complicated
1152 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1153 (in these examples C<S> and C<T> are regular subexpressions).
1155 Such combinations can include alternatives, leading to a problem of choice:
1156 if we match a regular expression C<a|ab> against C<"abc">, will it match
1157 substring C<"a"> or C<"ab">? One way to describe which substring is
1158 actually matched is the concept of backtracking (see L<"Backtracking">).
1159 However, this description is too low-level and makes you think
1160 in terms of a particular implementation.
1162 Another description starts with notions of "better"/"worse". All the
1163 substrings which may be matched by the given regular expression can be
1164 sorted from the "best" match to the "worst" match, and it is the "best"
1165 match which is chosen. This substitutes the question of "what is chosen?"
1166 by the question of "which matches are better, and which are worse?".
1168 Again, for elementary pieces there is no such question, since at most
1169 one match at a given position is possible. This section describes the
1170 notion of better/worse for combining operators. In the description
1171 below C<S> and C<T> are regular subexpressions.
1177 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1178 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1179 which can be matched by C<T>.
1181 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
1184 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
1185 C<B> is better match for C<T> than C<B'>.
1189 When C<S> can match, it is a better match than when only C<T> can match.
1191 Ordering of two matches for C<S> is the same as for C<S>. Similar for
1192 two matches for C<T>.
1194 =item C<S{REPEAT_COUNT}>
1196 Matches as C<SSS...S> (repeated as many times as necessary).
1200 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
1202 =item C<S{min,max}?>
1204 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
1206 =item C<S?>, C<S*>, C<S+>
1208 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
1210 =item C<S??>, C<S*?>, C<S+?>
1212 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
1216 Matches the best match for C<S> and only that.
1218 =item C<(?=S)>, C<(?<=S)>
1220 Only the best match for C<S> is considered. (This is important only if
1221 C<S> has capturing parentheses, and backreferences are used somewhere
1222 else in the whole regular expression.)
1224 =item C<(?!S)>, C<(?<!S)>
1226 For this grouping operator there is no need to describe the ordering, since
1227 only whether or not C<S> can match is important.
1229 =item C<(??{ EXPR })>
1231 The ordering is the same as for the regular expression which is
1234 =item C<(?(condition)yes-pattern|no-pattern)>
1236 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
1237 already determined. The ordering of the matches is the same as for the
1238 chosen subexpression.
1242 The above recipes describe the ordering of matches I<at a given position>.
1243 One more rule is needed to understand how a match is determined for the
1244 whole regular expression: a match at an earlier position is always better
1245 than a match at a later position.
1247 =head2 Creating custom RE engines
1249 Overloaded constants (see L<overload>) provide a simple way to extend
1250 the functionality of the RE engine.
1252 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
1253 matches at boundary between white-space characters and non-whitespace
1254 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
1255 at these positions, so we want to have each C<\Y|> in the place of the
1256 more complicated version. We can create a module C<customre> to do
1264 die "No argument to customre::import allowed" if @_;
1265 overload::constant 'qr' => \&convert;
1268 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
1270 my %rules = ( '\\' => '\\',
1271 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
1277 { $rules{$1} or invalid($re,$1) }sgex;
1281 Now C<use customre> enables the new escape in constant regular
1282 expressions, i.e., those without any runtime variable interpolations.
1283 As documented in L<overload>, this conversion will work only over
1284 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
1285 part of this regular expression needs to be converted explicitly
1286 (but only if the special meaning of C<\Y|> should be enabled inside $re):
1291 $re = customre::convert $re;
1296 This document varies from difficult to understand to completely
1297 and utterly opaque. The wandering prose riddled with jargon is
1298 hard to fathom in several places.
1300 This document needs a rewrite that separates the tutorial content
1301 from the reference content.
1309 L<perlop/"Regexp Quote-Like Operators">.
1311 L<perlop/"Gory details of parsing quoted constructs">.
1321 I<Mastering Regular Expressions> by Jeffrey Friedl, published
1322 by O'Reilly and Associates.