3 perlre - Perl regular expressions
7 This page describes the syntax of regular expressions in Perl. For a
8 description of how to I<use> regular expressions in matching
9 operations, plus various examples of the same, see discussion
10 of C<m//>, C<s///>, C<qr//> and C<??> in L<perlop/"Regexp Quote-Like Operators">.
12 The matching operations can have various modifiers. The modifiers
13 that relate to the interpretation of the regular expression inside
14 are listed below. For the modifiers that alter the way a regular expression
15 is used by Perl, see L<perlop/"Regexp Quote-Like Operators"> and
16 L<perlop/"Gory details of parsing quoted constructs">.
22 Do case-insensitive pattern matching.
24 If C<use locale> is in effect, the case map is taken from the current
25 locale. See L<perllocale>.
29 Treat string as multiple lines. That is, change "^" and "$" from matching
30 at only the very start or end of the string to the start or end of any
31 line anywhere within the string,
35 Treat string as single line. That is, change "." to match any character
36 whatsoever, even a newline, which it normally would not match.
38 The C</s> and C</m> modifiers both override the C<$*> setting. That is, no matter
39 what C<$*> contains, C</s> without C</m> will force "^" to match only at the
40 beginning of the string and "$" to match only at the end (or just before a
41 newline at the end) of the string. Together, as /ms, they let the "." match
42 any character whatsoever, while yet allowing "^" and "$" to match,
43 respectively, just after and just before newlines within the string.
47 Extend your pattern's legibility by permitting whitespace and comments.
51 These are usually written as "the C</x> modifier", even though the delimiter
52 in question might not actually be a slash. In fact, any of these
53 modifiers may also be embedded within the regular expression itself using
54 the new C<(?...)> construct. See below.
56 The C</x> modifier itself needs a little more explanation. It tells
57 the regular expression parser to ignore whitespace that is neither
58 backslashed nor within a character class. You can use this to break up
59 your regular expression into (slightly) more readable parts. The C<#>
60 character is also treated as a metacharacter introducing a comment,
61 just as in ordinary Perl code. This also means that if you want real
62 whitespace or C<#> characters in the pattern (outside of a character
63 class, where they are unaffected by C</x>), that you'll either have to
64 escape them or encode them using octal or hex escapes. Taken together,
65 these features go a long way towards making Perl's regular expressions
66 more readable. Note that you have to be careful not to include the
67 pattern delimiter in the comment--perl has no way of knowing you did
68 not intend to close the pattern early. See the C-comment deletion code
71 =head2 Regular Expressions
73 The patterns used in pattern matching are regular expressions such as
74 those supplied in the Version 8 regex routines. (In fact, the
75 routines are derived (distantly) from Henry Spencer's freely
76 redistributable reimplementation of the V8 routines.)
77 See L<Version 8 Regular Expressions> for details.
79 In particular the following metacharacters have their standard I<egrep>-ish
82 \ Quote the next metacharacter
83 ^ Match the beginning of the line
84 . Match any character (except newline)
85 $ Match the end of the line (or before newline at the end)
90 By default, the "^" character is guaranteed to match at only the
91 beginning of the string, the "$" character at only the end (or before the
92 newline at the end) and Perl does certain optimizations with the
93 assumption that the string contains only one line. Embedded newlines
94 will not be matched by "^" or "$". You may, however, wish to treat a
95 string as a multi-line buffer, such that the "^" will match after any
96 newline within the string, and "$" will match before any newline. At the
97 cost of a little more overhead, you can do this by using the /m modifier
98 on the pattern match operator. (Older programs did this by setting C<$*>,
99 but this practice is now deprecated.)
101 To facilitate multi-line substitutions, the "." character never matches a
102 newline unless you use the C</s> modifier, which in effect tells Perl to pretend
103 the string is a single line--even if it isn't. The C</s> modifier also
104 overrides the setting of C<$*>, in case you have some (badly behaved) older
105 code that sets it in another module.
107 The following standard quantifiers are recognized:
109 * Match 0 or more times
110 + Match 1 or more times
112 {n} Match exactly n times
113 {n,} Match at least n times
114 {n,m} Match at least n but not more than m times
116 (If a curly bracket occurs in any other context, it is treated
117 as a regular character.) The "*" modifier is equivalent to C<{0,}>, the "+"
118 modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
119 to integral values less than a preset limit defined when perl is built.
120 This is usually 32766 on the most common platforms. The actual limit can
121 be seen in the error message generated by code such as this:
123 $_ **= $_ , / {$_} / for 2 .. 42;
125 By default, a quantified subpattern is "greedy", that is, it will match as
126 many times as possible (given a particular starting location) while still
127 allowing the rest of the pattern to match. If you want it to match the
128 minimum number of times possible, follow the quantifier with a "?". Note
129 that the meanings don't change, just the "greediness":
131 *? Match 0 or more times
132 +? Match 1 or more times
134 {n}? Match exactly n times
135 {n,}? Match at least n times
136 {n,m}? Match at least n but not more than m times
138 Because patterns are processed as double quoted strings, the following
145 \a alarm (bell) (BEL)
146 \e escape (think troff) (ESC)
147 \033 octal char (think of a PDP-11)
150 \l lowercase next char (think vi)
151 \u uppercase next char (think vi)
152 \L lowercase till \E (think vi)
153 \U uppercase till \E (think vi)
154 \E end case modification (think vi)
155 \Q quote (disable) pattern metacharacters till \E
157 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
158 and C<\U> is taken from the current locale. See L<perllocale>.
160 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
161 An unescaped C<$> or C<@> interpolates the corresponding variable,
162 while escaping will cause the literal string C<\$> to be matched.
163 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
165 In addition, Perl defines the following:
167 \w Match a "word" character (alphanumeric plus "_")
168 \W Match a non-word character
169 \s Match a whitespace character
170 \S Match a non-whitespace character
171 \d Match a digit character
172 \D Match a non-digit character
174 A C<\w> matches a single alphanumeric character, not a whole
175 word. To match a word you'd need to say C<\w+>. If C<use locale> is in
176 effect, the list of alphabetic characters generated by C<\w> is taken
177 from the current locale. See L<perllocale>. You may use C<\w>, C<\W>,
178 C<\s>, C<\S>, C<\d>, and C<\D> within character classes (though not as
179 either end of a range).
181 Perl defines the following zero-width assertions:
183 \b Match a word boundary
184 \B Match a non-(word boundary)
185 \A Match only at beginning of string
186 \Z Match only at end of string, or before newline at the end
187 \z Match only at end of string
188 \G Match only where previous m//g left off (works only with /g)
190 A word boundary (C<\b>) is defined as a spot between two characters that
191 has a C<\w> on one side of it and a C<\W> on the other side of it (in
192 either order), counting the imaginary characters off the beginning and
193 end of the string as matching a C<\W>. (Within character classes C<\b>
194 represents backspace rather than a word boundary.) The C<\A> and C<\Z> are
195 just like "^" and "$", except that they won't match multiple times when the
196 C</m> modifier is used, while "^" and "$" will match at every internal line
197 boundary. To match the actual end of the string, not ignoring newline,
198 you can use C<\z>. The C<\G> assertion can be used to chain global
199 matches (using C<m//g>), as described in
200 L<perlop/"Regexp Quote-Like Operators">.
202 It is also useful when writing C<lex>-like scanners, when you have several
203 patterns that you want to match against consequent substrings of your
204 string, see the previous reference.
205 The actual location where C<\G> will match can also be influenced
206 by using C<pos()> as an lvalue. See L<perlfunc/pos>.
208 When the bracketing construct C<( ... )> is used, \E<lt>digitE<gt> matches the
209 digit'th substring. Outside of the pattern, always use "$" instead of "\"
210 in front of the digit. (While the \E<lt>digitE<gt> notation can on rare occasion work
211 outside the current pattern, this should not be relied upon. See the
212 WARNING below.) The scope of $E<lt>digitE<gt> (and C<$`>, C<$&>, and C<$'>)
213 extends to the end of the enclosing BLOCK or eval string, or to the next
214 successful pattern match, whichever comes first. If you want to use
215 parentheses to delimit a subpattern (e.g., a set of alternatives) without
216 saving it as a subpattern, follow the ( with a ?:.
218 You may have as many parentheses as you wish. If you have more
219 than 9 substrings, the variables $10, $11, ... refer to the
220 corresponding substring. Within the pattern, \10, \11, etc. refer back
221 to substrings if there have been at least that many left parentheses before
222 the backreference. Otherwise (for backward compatibility) \10 is the
223 same as \010, a backspace, and \11 the same as \011, a tab. And so
224 on. (\1 through \9 are always backreferences.)
226 C<$+> returns whatever the last bracket match matched. C<$&> returns the
227 entire matched string. (C<$0> used to return the same thing, but not any
228 more.) C<$`> returns everything before the matched string. C<$'> returns
229 everything after the matched string. Examples:
231 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
233 if (/Time: (..):(..):(..)/) {
239 Once perl sees that you need one of C<$&>, C<$`> or C<$'> anywhere in
240 the program, it has to provide them on each and every pattern match.
241 This can slow your program down. The same mechanism that handles
242 these provides for the use of $1, $2, etc., so you pay the same price
243 for each pattern that contains capturing parentheses. But if you never
244 use $&, etc., in your script, then patterns I<without> capturing
245 parentheses won't be penalized. So avoid $&, $', and $` if you can,
246 but if you can't (and some algorithms really appreciate them), once
247 you've used them once, use them at will, because you've already paid
248 the price. As of 5.005, $& is not so costly as the other two.
250 Backslashed metacharacters in Perl are
251 alphanumeric, such as C<\b>, C<\w>, C<\n>. Unlike some other regular
252 expression languages, there are no backslashed symbols that aren't
253 alphanumeric. So anything that looks like \\, \(, \), \E<lt>, \E<gt>,
254 \{, or \} is always interpreted as a literal character, not a
255 metacharacter. This was once used in a common idiom to disable or
256 quote the special meanings of regular expression metacharacters in a
257 string that you want to use for a pattern. Simply quote all
258 non-alphanumeric characters:
260 $pattern =~ s/(\W)/\\$1/g;
262 Now it is much more common to see either the quotemeta() function or
263 the C<\Q> escape sequence used to disable all metacharacters' special
266 /$unquoted\Q$quoted\E$unquoted/
268 Perl defines a consistent extension syntax for regular expressions.
269 The syntax is a pair of parentheses with a question mark as the first
270 thing within the parentheses (this was a syntax error in older
271 versions of Perl). The character after the question mark gives the
272 function of the extension. Several extensions are already supported:
278 A comment. The text is ignored. If the C</x> switch is used to enable
279 whitespace formatting, a simple C<#> will suffice. Note that perl closes
280 the comment as soon as it sees a C<)>, so there is no way to put a literal
285 =item C<(?imsx-imsx:pattern)>
287 This is for clustering, not capturing; it groups subexpressions like
288 "()", but doesn't make backreferences as "()" does. So
290 @fields = split(/\b(?:a|b|c)\b/)
294 @fields = split(/\b(a|b|c)\b/)
296 but doesn't spit out extra fields.
298 The letters between C<?> and C<:> act as flags modifiers, see
299 L<C<(?imsx-imsx)>>. In particular,
301 /(?s-i:more.*than).*million/i
303 is equivalent to more verbose
305 /(?:(?s-i)more.*than).*million/i
309 A zero-width positive lookahead assertion. For example, C</\w+(?=\t)/>
310 matches a word followed by a tab, without including the tab in C<$&>.
314 A zero-width negative lookahead assertion. For example C</foo(?!bar)/>
315 matches any occurrence of "foo" that isn't followed by "bar". Note
316 however that lookahead and lookbehind are NOT the same thing. You cannot
317 use this for lookbehind.
319 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
320 will not do what you want. That's because the C<(?!foo)> is just saying that
321 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
322 match. You would have to do something like C</(?!foo)...bar/> for that. We
323 say "like" because there's the case of your "bar" not having three characters
324 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
325 Sometimes it's still easier just to say:
327 if (/bar/ && $` !~ /foo$/)
329 For lookbehind see below.
331 =item C<(?E<lt>=pattern)>
333 A zero-width positive lookbehind assertion. For example, C</(?E<lt>=\t)\w+/>
334 matches a word following a tab, without including the tab in C<$&>.
335 Works only for fixed-width lookbehind.
337 =item C<(?<!pattern)>
339 A zero-width negative lookbehind assertion. For example C</(?<!bar)foo/>
340 matches any occurrence of "foo" that isn't following "bar".
341 Works only for fixed-width lookbehind.
345 Experimental "evaluate any Perl code" zero-width assertion. Always
346 succeeds. C<code> is not interpolated. Currently the rules to
347 determine where the C<code> ends are somewhat convoluted.
349 The C<code> is properly scoped in the following sense: if the assertion
350 is backtracked (compare L<"Backtracking">), all the changes introduced after
351 C<local>isation are undone, so
355 (?{ $cnt = 0 }) # Initialize $cnt.
359 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
363 (?{ $res = $cnt }) # On success copy to non-localized
367 will set C<$res = 4>. Note that after the match $cnt returns to the globally
368 introduced value 0, since the scopes which restrict C<local> statements
371 This assertion may be used as L<C<(?(condition)yes-pattern|no-pattern)>>
372 switch. If I<not> used in this way, the result of evaluation of C<code>
373 is put into variable $^R. This happens immediately, so $^R can be used from
374 other C<(?{ code })> assertions inside the same regular expression.
376 The above assignment to $^R is properly localized, thus the old value of $^R
377 is restored if the assertion is backtracked (compare L<"Backtracking">).
379 Due to security concerns, this construction is not allowed if the regular
380 expression involves run-time interpolation of variables, unless
381 C<use re 'eval'> pragma is used (see L<re>), or the variables contain
382 results of qr() operator (see L<perlop/"qr/STRING/imosx">).
384 This restriction is due to the wide-spread (questionable) practice of
391 without tainting. While this code is frowned upon from security point
392 of view, when C<(?{})> was introduced, it was considered bad to add
393 I<new> security holes to existing scripts.
395 B<NOTE:> Use of the above insecure snippet without also enabling taint mode
396 is to be severely frowned upon. C<use re 'eval'> does not disable tainting
397 checks, thus to allow $re in the above snippet to contain C<(?{})>
398 I<with tainting enabled>, one needs both C<use re 'eval'> and untaint
401 =item C<(?E<gt>pattern)>
403 An "independent" subexpression. Matches the substring that a
404 I<standalone> C<pattern> would match if anchored at the given position,
405 B<and only this substring>.
407 Say, C<^(?E<gt>a*)ab> will never match, since C<(?E<gt>a*)> (anchored
408 at the beginning of string, as above) will match I<all> characters
409 C<a> at the beginning of string, leaving no C<a> for C<ab> to match.
410 In contrast, C<a*ab> will match the same as C<a+b>, since the match of
411 the subgroup C<a*> is influenced by the following group C<ab> (see
412 L<"Backtracking">). In particular, C<a*> inside C<a*ab> will match
413 fewer characters than a standalone C<a*>, since this makes the tail match.
415 An effect similar to C<(?E<gt>pattern)> may be achieved by
419 since the lookahead is in I<"logical"> context, thus matches the same
420 substring as a standalone C<a+>. The following C<\1> eats the matched
421 string, thus making a zero-length assertion into an analogue of
422 C<(?E<gt>...)>. (The difference between these two constructs is that the
423 second one uses a catching group, thus shifting ordinals of
424 backreferences in the rest of a regular expression.)
426 This construct is useful for optimizations of "eternal"
427 matches, because it will not backtrack (see L<"Backtracking">).
438 That will efficiently match a nonempty group with matching
439 two-or-less-level-deep parentheses. However, if there is no such group,
440 it will take virtually forever on a long string. That's because there are
441 so many different ways to split a long string into several substrings.
442 This is what C<(.+)+> is doing, and C<(.+)+> is similar to a subpattern
443 of the above pattern. Consider that the above pattern detects no-match
444 on C<((()aaaaaaaaaaaaaaaaaa> in several seconds, but that each extra
445 letter doubles this time. This exponential performance will make it
446 appear that your program has hung.
448 However, a tiny modification of this pattern
459 which uses C<(?E<gt>...)> matches exactly when the one above does (verifying
460 this yourself would be a productive exercise), but finishes in a fourth
461 the time when used on a similar string with 1000000 C<a>s. Be aware,
462 however, that this pattern currently triggers a warning message under
463 B<-w> saying it C<"matches the null string many times">):
465 On simple groups, such as the pattern C<(?E<gt> [^()]+ )>, a comparable
466 effect may be achieved by negative lookahead, as in C<[^()]+ (?! [^()] )>.
467 This was only 4 times slower on a string with 1000000 C<a>s.
469 =item C<(?(condition)yes-pattern|no-pattern)>
471 =item C<(?(condition)yes-pattern)>
473 Conditional expression. C<(condition)> should be either an integer in
474 parentheses (which is valid if the corresponding pair of parentheses
475 matched), or lookahead/lookbehind/evaluate zero-width assertion.
484 matches a chunk of non-parentheses, possibly included in parentheses
487 =item C<(?imsx-imsx)>
489 One or more embedded pattern-match modifiers. This is particularly
490 useful for patterns that are specified in a table somewhere, some of
491 which want to be case sensitive, and some of which don't. The case
492 insensitive ones need to include merely C<(?i)> at the front of the
493 pattern. For example:
496 if ( /$pattern/i ) { }
500 $pattern = "(?i)foobar";
501 if ( /$pattern/ ) { }
503 Letters after C<-> switch modifiers off.
505 These modifiers are localized inside an enclosing group (if any). Say,
509 (assuming C<x> modifier, and no C<i> modifier outside of this group)
510 will match a repeated (I<including the case>!) word C<blah> in any
515 A question mark was chosen for this and for the new minimal-matching
516 construct because 1) question mark is pretty rare in older regular
517 expressions, and 2) whenever you see one, you should stop and "question"
518 exactly what is going on. That's psychology...
522 A fundamental feature of regular expression matching involves the
523 notion called I<backtracking>, which is currently used (when needed)
524 by all regular expression quantifiers, namely C<*>, C<*?>, C<+>,
525 C<+?>, C<{n,m}>, and C<{n,m}?>.
527 For a regular expression to match, the I<entire> regular expression must
528 match, not just part of it. So if the beginning of a pattern containing a
529 quantifier succeeds in a way that causes later parts in the pattern to
530 fail, the matching engine backs up and recalculates the beginning
531 part--that's why it's called backtracking.
533 Here is an example of backtracking: Let's say you want to find the
534 word following "foo" in the string "Food is on the foo table.":
536 $_ = "Food is on the foo table.";
537 if ( /\b(foo)\s+(\w+)/i ) {
538 print "$2 follows $1.\n";
541 When the match runs, the first part of the regular expression (C<\b(foo)>)
542 finds a possible match right at the beginning of the string, and loads up
543 $1 with "Foo". However, as soon as the matching engine sees that there's
544 no whitespace following the "Foo" that it had saved in $1, it realizes its
545 mistake and starts over again one character after where it had the
546 tentative match. This time it goes all the way until the next occurrence
547 of "foo". The complete regular expression matches this time, and you get
548 the expected output of "table follows foo."
550 Sometimes minimal matching can help a lot. Imagine you'd like to match
551 everything between "foo" and "bar". Initially, you write something
554 $_ = "The food is under the bar in the barn.";
555 if ( /foo(.*)bar/ ) {
559 Which perhaps unexpectedly yields:
561 got <d is under the bar in the >
563 That's because C<.*> was greedy, so you get everything between the
564 I<first> "foo" and the I<last> "bar". In this case, it's more effective
565 to use minimal matching to make sure you get the text between a "foo"
566 and the first "bar" thereafter.
568 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
569 got <d is under the >
571 Here's another example: let's say you'd like to match a number at the end
572 of a string, and you also want to keep the preceding part the match.
575 $_ = "I have 2 numbers: 53147";
576 if ( /(.*)(\d*)/ ) { # Wrong!
577 print "Beginning is <$1>, number is <$2>.\n";
580 That won't work at all, because C<.*> was greedy and gobbled up the
581 whole string. As C<\d*> can match on an empty string the complete
582 regular expression matched successfully.
584 Beginning is <I have 2 numbers: 53147>, number is <>.
586 Here are some variants, most of which don't work:
588 $_ = "I have 2 numbers: 53147";
601 printf "%-12s ", $pat;
611 (.*)(\d*) <I have 2 numbers: 53147> <>
612 (.*)(\d+) <I have 2 numbers: 5314> <7>
614 (.*?)(\d+) <I have > <2>
615 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
616 (.*?)(\d+)$ <I have 2 numbers: > <53147>
617 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
618 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
620 As you see, this can be a bit tricky. It's important to realize that a
621 regular expression is merely a set of assertions that gives a definition
622 of success. There may be 0, 1, or several different ways that the
623 definition might succeed against a particular string. And if there are
624 multiple ways it might succeed, you need to understand backtracking to
625 know which variety of success you will achieve.
627 When using lookahead assertions and negations, this can all get even
628 tricker. Imagine you'd like to find a sequence of non-digits not
629 followed by "123". You might try to write that as
632 if ( /^\D*(?!123)/ ) { # Wrong!
633 print "Yup, no 123 in $_\n";
636 But that isn't going to match; at least, not the way you're hoping. It
637 claims that there is no 123 in the string. Here's a clearer picture of
638 why it that pattern matches, contrary to popular expectations:
643 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ;
644 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ;
646 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ;
647 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ;
655 You might have expected test 3 to fail because it seems to a more
656 general purpose version of test 1. The important difference between
657 them is that test 3 contains a quantifier (C<\D*>) and so can use
658 backtracking, whereas test 1 will not. What's happening is
659 that you've asked "Is it true that at the start of $x, following 0 or more
660 non-digits, you have something that's not 123?" If the pattern matcher had
661 let C<\D*> expand to "ABC", this would have caused the whole pattern to
663 The search engine will initially match C<\D*> with "ABC". Then it will
664 try to match C<(?!123> with "123", which of course fails. But because
665 a quantifier (C<\D*>) has been used in the regular expression, the
666 search engine can backtrack and retry the match differently
667 in the hope of matching the complete regular expression.
669 The pattern really, I<really> wants to succeed, so it uses the
670 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
671 time. Now there's indeed something following "AB" that is not
672 "123". It's in fact "C123", which suffices.
674 We can deal with this by using both an assertion and a negation. We'll
675 say that the first part in $1 must be followed by a digit, and in fact, it
676 must also be followed by something that's not "123". Remember that the
677 lookaheads are zero-width expressions--they only look, but don't consume
678 any of the string in their match. So rewriting this way produces what
679 you'd expect; that is, case 5 will fail, but case 6 succeeds:
681 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ;
682 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ;
686 In other words, the two zero-width assertions next to each other work as though
687 they're ANDed together, just as you'd use any builtin assertions: C</^$/>
688 matches only if you're at the beginning of the line AND the end of the
689 line simultaneously. The deeper underlying truth is that juxtaposition in
690 regular expressions always means AND, except when you write an explicit OR
691 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
692 although the attempted matches are made at different positions because "a"
693 is not a zero-width assertion, but a one-width assertion.
695 One warning: particularly complicated regular expressions can take
696 exponential time to solve due to the immense number of possible ways they
697 can use backtracking to try match. For example this will take a very long
700 /((a{0,5}){0,5}){0,5}/
702 And if you used C<*>'s instead of limiting it to 0 through 5 matches, then
703 it would take literally forever--or until you ran out of stack space.
705 A powerful tool for optimizing such beasts is "independent" groups,
706 which do not backtrace (see L<C<(?E<gt>pattern)>>). Note also that
707 zero-length lookahead/lookbehind assertions will not backtrace to make
708 the tail match, since they are in "logical" context: only the fact
709 whether they match or not is considered relevant. For an example
710 where side-effects of a lookahead I<might> have influenced the
711 following match, see L<C<(?E<gt>pattern)>>.
713 =head2 Version 8 Regular Expressions
715 In case you're not familiar with the "regular" Version 8 regex
716 routines, here are the pattern-matching rules not described above.
718 Any single character matches itself, unless it is a I<metacharacter>
719 with a special meaning described here or above. You can cause
720 characters that normally function as metacharacters to be interpreted
721 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
722 character; "\\" matches a "\"). A series of characters matches that
723 series of characters in the target string, so the pattern C<blurfl>
724 would match "blurfl" in the target string.
726 You can specify a character class, by enclosing a list of characters
727 in C<[]>, which will match any one character from the list. If the
728 first character after the "[" is "^", the class matches any character not
729 in the list. Within a list, the "-" character is used to specify a
730 range, so that C<a-z> represents all characters between "a" and "z",
731 inclusive. If you want "-" itself to be a member of a class, put it
732 at the start or end of the list, or escape it with a backslash. (The
733 following all specify the same class of three characters: C<[-az]>,
734 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
735 specifies a class containing twenty-six characters.)
737 Note also that the whole range idea is rather unportable between
738 character sets--and even within character sets they may cause results
739 you probably didn't expect. A sound principle is to use only ranges
740 that begin from and end at either alphabets of equal case ([a-e],
741 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
742 spell out the character sets in full.
744 Characters may be specified using a metacharacter syntax much like that
745 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
746 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
747 of octal digits, matches the character whose ASCII value is I<nnn>.
748 Similarly, \xI<nn>, where I<nn> are hexadecimal digits, matches the
749 character whose ASCII value is I<nn>. The expression \cI<x> matches the
750 ASCII character control-I<x>. Finally, the "." metacharacter matches any
751 character except "\n" (unless you use C</s>).
753 You can specify a series of alternatives for a pattern using "|" to
754 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
755 or "foe" in the target string (as would C<f(e|i|o)e>). The
756 first alternative includes everything from the last pattern delimiter
757 ("(", "[", or the beginning of the pattern) up to the first "|", and
758 the last alternative contains everything from the last "|" to the next
759 pattern delimiter. For this reason, it's common practice to include
760 alternatives in parentheses, to minimize confusion about where they
763 Alternatives are tried from left to right, so the first
764 alternative found for which the entire expression matches, is the one that
765 is chosen. This means that alternatives are not necessarily greedy. For
766 example: when matching C<foo|foot> against "barefoot", only the "foo"
767 part will match, as that is the first alternative tried, and it successfully
768 matches the target string. (This might not seem important, but it is
769 important when you are capturing matched text using parentheses.)
771 Also remember that "|" is interpreted as a literal within square brackets,
772 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
774 Within a pattern, you may designate subpatterns for later reference by
775 enclosing them in parentheses, and you may refer back to the I<n>th
776 subpattern later in the pattern using the metacharacter \I<n>.
777 Subpatterns are numbered based on the left to right order of their
778 opening parenthesis. A backreference matches whatever
779 actually matched the subpattern in the string being examined, not the
780 rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
781 match "0x1234 0x4321", but not "0x1234 01234", because subpattern 1
782 actually matched "0x", even though the rule C<0|0x> could
783 potentially match the leading 0 in the second number.
785 =head2 WARNING on \1 vs $1
787 Some people get too used to writing things like:
789 $pattern =~ s/(\W)/\\\1/g;
791 This is grandfathered for the RHS of a substitute to avoid shocking the
792 B<sed> addicts, but it's a dirty habit to get into. That's because in
793 PerlThink, the righthand side of a C<s///> is a double-quoted string. C<\1> in
794 the usual double-quoted string means a control-A. The customary Unix
795 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
796 of doing that, you get yourself into trouble if you then add an C</e>
799 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
805 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
806 C<${1}000>. Basically, the operation of interpolation should not be confused
807 with the operation of matching a backreference. Certainly they mean two
808 different things on the I<left> side of the C<s///>.
810 =head2 Repeated patterns matching zero-length substring
812 WARNING: Difficult material (and prose) ahead. This section needs a rewrite.
814 Regular expressions provide a terse and powerful programming language. As
815 with most other power tools, power comes together with the ability
818 A common abuse of this power stems from the ability to make infinite
819 loops using regular expressions, with something as innocuous as:
821 'foo' =~ m{ ( o? )* }x;
823 The C<o?> can match at the beginning of C<'foo'>, and since the position
824 in the string is not moved by the match, C<o?> would match again and again
825 due to the C<*> modifier. Another common way to create a similar cycle
826 is with the looping modifier C<//g>:
828 @matches = ( 'foo' =~ m{ o? }xg );
832 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
834 or the loop implied by split().
836 However, long experience has shown that many programming tasks may
837 be significantly simplified by using repeated subexpressions which
838 may match zero-length substrings, with a simple example being:
840 @chars = split //, $string; # // is not magic in split
841 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
843 Thus Perl allows the C</()/> construct, which I<forcefully breaks
844 the infinite loop>. The rules for this are different for lower-level
845 loops given by the greedy modifiers C<*+{}>, and for higher-level
846 ones like the C</g> modifier or split() operator.
848 The lower-level loops are I<interrupted> when it is detected that a
849 repeated expression did match a zero-length substring, thus
851 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
853 is made equivalent to
855 m{ (?: NON_ZERO_LENGTH )*
860 The higher level-loops preserve an additional state between iterations:
861 whether the last match was zero-length. To break the loop, the following
862 match after a zero-length match is prohibited to have a length of zero.
863 This prohibition interacts with backtracking (see L<"Backtracking">),
864 and so the I<second best> match is chosen if the I<best> match is of
872 results in C<"<><b><><a><><r><>">. At each position of the string the best
873 match given by non-greedy C<??> is the zero-length match, and the I<second
874 best> match is what is matched by C<\w>. Thus zero-length matches
875 alternate with one-character-long matches.
877 Similarly, for repeated C<m/()/g> the second-best match is the match at the
878 position one notch further in the string.
880 The additional state of being I<matched with zero-length> is associated to
881 the matched string, and is reset by each assignment to pos().
883 =head2 Creating custom RE engines
885 Overloaded constants (see L<overload>) provide a simple way to extend
886 the functionality of the RE engine.
888 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
889 matches at boundary between white-space characters and non-whitespace
890 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
891 at these positions, so we want to have each C<\Y|> in the place of the
892 more complicated version. We can create a module C<customre> to do
900 die "No argument to customre::import allowed" if @_;
901 overload::constant 'qr' => \&convert;
904 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
906 my %rules = ( '\\' => '\\',
907 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
913 { $rules{$1} or invalid($re,$1) }sgex;
917 Now C<use customre> enables the new escape in constant regular
918 expressions, i.e., those without any runtime variable interpolations.
919 As documented in L<overload>, this conversion will work only over
920 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
921 part of this regular expression needs to be converted explicitly
922 (but only if the special meaning of C<\Y|> should be enabled inside $re):
927 $re = customre::convert $re;
932 L<perlop/"Regexp Quote-Like Operators">.
934 L<perlop/"Gory details of parsing quoted constructs">.
940 I<Mastering Regular Expressions> (see L<perlbook>) by Jeffrey Friedl.