3 perlretut - Perl regular expressions tutorial
7 This page provides a basic tutorial on understanding, creating and
8 using regular expressions in Perl. It serves as a complement to the
9 reference page on regular expressions L<perlre>. Regular expressions
10 are an integral part of the C<m//>, C<s///>, C<qr//> and C<split>
11 operators and so this tutorial also overlaps with
12 L<perlop/"Regexp Quote-Like Operators"> and L<perlfunc/split>.
14 Perl is widely renowned for excellence in text processing, and regular
15 expressions are one of the big factors behind this fame. Perl regular
16 expressions display an efficiency and flexibility unknown in most
17 other computer languages. Mastering even the basics of regular
18 expressions will allow you to manipulate text with surprising ease.
20 What is a regular expression? At its most basic, a regular expression
21 is a template that is used to determine if a string has certain
22 characteristics. The string is most often some text, such as a line,
23 sentence, web page, or even a whole book, but it doesn't have to be. It
24 could be binary data, for example. Biologists often use Perl to look
25 for patterns in long DNA sequences.
27 Suppose we want to determine if the text in variable, C<$var> contains
28 the sequence of characters S<C<m u s h r o o m>>
29 (blanks added for legibility). We can write in Perl
33 The value of this expression will be TRUE if C<$var> contains that
34 sequence of characters anywhere within it, and FALSE otherwise. The
35 portion enclosed in C<'E<sol>'> characters denotes the characteristic we
37 We use the term I<pattern> for it. The process of looking to see if the
38 pattern occurs in the string is called I<matching>, and the C<"=~">
39 operator along with the C<m//> tell Perl to try to match the pattern
40 against the string. Note that the pattern is also a string, but a very
41 special kind of one, as we will see. Patterns are in common use these
43 examples are the patterns typed into a search engine to find web pages
44 and the patterns used to list files in a directory, I<e.g.>, "C<ls *.txt>"
45 or "C<dir *.*>". In Perl, the patterns described by regular expressions
46 are used not only to search strings, but to also extract desired parts
47 of strings, and to do search and replace operations.
49 Regular expressions have the undeserved reputation of being abstract
50 and difficult to understand. This really stems simply because the
51 notation used to express them tends to be terse and dense, and not
52 because of inherent complexity. We recommend using the C</x> regular
53 expression modifier (described below) along with plenty of white space
54 to make them less dense, and easier to read. Regular expressions are
56 simple concepts like conditionals and loops and are no more difficult
57 to understand than the corresponding C<if> conditionals and C<while>
58 loops in the Perl language itself.
60 This tutorial flattens the learning curve by discussing regular
61 expression concepts, along with their notation, one at a time and with
62 many examples. The first part of the tutorial will progress from the
63 simplest word searches to the basic regular expression concepts. If
64 you master the first part, you will have all the tools needed to solve
65 about 98% of your needs. The second part of the tutorial is for those
66 comfortable with the basics and hungry for more power tools. It
67 discusses the more advanced regular expression operators and
68 introduces the latest cutting-edge innovations.
70 A note: to save time, "regular expression" is often abbreviated as
71 regexp or regex. Regexp is a more natural abbreviation than regex, but
72 is harder to pronounce. The Perl pod documentation is evenly split on
73 regexp vs regex; in Perl, there is more than one way to abbreviate it.
74 We'll use regexp in this tutorial.
76 New in v5.22, L<C<use re 'strict'>|re/'strict' mode> applies stricter
77 rules than otherwise when compiling regular expression patterns. It can
78 find things that, while legal, may not be what you intended.
80 =head1 Part 1: The basics
82 =head2 Simple word matching
84 The simplest regexp is simply a word, or more generally, a string of
85 characters. A regexp consisting of just a word matches any string that
88 "Hello World" =~ /World/; # matches
90 What is this Perl statement all about? C<"Hello World"> is a simple
91 double-quoted string. C<World> is the regular expression and the
92 C<//> enclosing C</World/> tells Perl to search a string for a match.
93 The operator C<=~> associates the string with the regexp match and
94 produces a true value if the regexp matched, or false if the regexp
95 did not match. In our case, C<World> matches the second word in
96 C<"Hello World">, so the expression is true. Expressions like this
97 are useful in conditionals:
99 if ("Hello World" =~ /World/) {
100 print "It matches\n";
103 print "It doesn't match\n";
106 There are useful variations on this theme. The sense of the match can
107 be reversed by using the C<!~> operator:
109 if ("Hello World" !~ /World/) {
110 print "It doesn't match\n";
113 print "It matches\n";
116 The literal string in the regexp can be replaced by a variable:
118 my $greeting = "World";
119 if ("Hello World" =~ /$greeting/) {
120 print "It matches\n";
123 print "It doesn't match\n";
126 If you're matching against the special default variable C<$_>, the
127 C<$_ =~> part can be omitted:
131 print "It matches\n";
134 print "It doesn't match\n";
137 And finally, the C<//> default delimiters for a match can be changed
138 to arbitrary delimiters by putting an C<'m'> out front:
140 "Hello World" =~ m!World!; # matches, delimited by '!'
141 "Hello World" =~ m{World}; # matches, note the paired '{}'
142 "/usr/bin/perl" =~ m"/perl"; # matches after '/usr/bin',
143 # '/' becomes an ordinary char
145 C</World/>, C<m!World!>, and C<m{World}> all represent the
146 same thing. When, I<e.g.>, the quote (C<'"'>) is used as a delimiter, the forward
147 slash C<'/'> becomes an ordinary character and can be used in this regexp
150 Let's consider how different regexps would match C<"Hello World">:
152 "Hello World" =~ /world/; # doesn't match
153 "Hello World" =~ /o W/; # matches
154 "Hello World" =~ /oW/; # doesn't match
155 "Hello World" =~ /World /; # doesn't match
157 The first regexp C<world> doesn't match because regexps are by default
158 case-sensitive. The second regexp matches because the substring
159 S<C<'o W'>> occurs in the string S<C<"Hello World">>. The space
160 character C<' '> is treated like any other character in a regexp and is
161 needed to match in this case. The lack of a space character is the
162 reason the third regexp C<'oW'> doesn't match. The fourth regexp
163 "C<World >" doesn't match because there is a space at the end of the
164 regexp, but not at the end of the string. The lesson here is that
165 regexps must match a part of the string I<exactly> in order for the
166 statement to be true.
168 If a regexp matches in more than one place in the string, Perl will
169 always match at the earliest possible point in the string:
171 "Hello World" =~ /o/; # matches 'o' in 'Hello'
172 "That hat is red" =~ /hat/; # matches 'hat' in 'That'
174 With respect to character matching, there are a few more points you
175 need to know about. First of all, not all characters can be used
176 "as-is" in a match. Some characters, called I<metacharacters>, are
177 generally reserved for use in regexp notation. The metacharacters are
181 This list is not as definitive as it may appear (or be claimed to be in
182 other documentation). For example, C<"#"> is a metacharacter only when
183 the C</x> pattern modifier (described below) is used, and both C<"}">
184 and C<"]"> are metacharacters only when paired with opening C<"{"> or
185 C<"["> respectively; other gotchas apply.
187 The significance of each of these will be explained
188 in the rest of the tutorial, but for now, it is important only to know
189 that a metacharacter can be matched as-is by putting a backslash before
192 "2+2=4" =~ /2+2/; # doesn't match, + is a metacharacter
193 "2+2=4" =~ /2\+2/; # matches, \+ is treated like an ordinary +
194 "The interval is [0,1)." =~ /[0,1)./ # is a syntax error!
195 "The interval is [0,1)." =~ /\[0,1\)\./ # matches
196 "#!/usr/bin/perl" =~ /#!\/usr\/bin\/perl/; # matches
198 In the last regexp, the forward slash C<'/'> is also backslashed,
199 because it is used to delimit the regexp. This can lead to LTS
200 (leaning toothpick syndrome), however, and it is often more readable
201 to change delimiters.
203 "#!/usr/bin/perl" =~ m!#\!/usr/bin/perl!; # easier to read
205 The backslash character C<'\'> is a metacharacter itself and needs to
208 'C:\WIN32' =~ /C:\\WIN/; # matches
210 In situations where it doesn't make sense for a particular metacharacter
211 to mean what it normally does, it automatically loses its
212 metacharacter-ness and becomes an ordinary character that is to be
213 matched literally. For example, the C<'}'> is a metacharacter only when
214 it is the mate of a C<'{'> metacharacter. Otherwise it is treated as a
215 literal RIGHT CURLY BRACKET. This may lead to unexpected results.
216 L<C<use re 'strict'>|re/'strict' mode> can catch some of these.
218 In addition to the metacharacters, there are some ASCII characters
219 which don't have printable character equivalents and are instead
220 represented by I<escape sequences>. Common examples are C<\t> for a
221 tab, C<\n> for a newline, C<\r> for a carriage return and C<\a> for a
222 bell (or alert). If your string is better thought of as a sequence of arbitrary
223 bytes, the octal escape sequence, I<e.g.>, C<\033>, or hexadecimal escape
224 sequence, I<e.g.>, C<\x1B> may be a more natural representation for your
225 bytes. Here are some examples of escapes:
227 "1000\t2000" =~ m(0\t2) # matches
228 "1000\n2000" =~ /0\n20/ # matches
229 "1000\t2000" =~ /\000\t2/ # doesn't match, "0" ne "\000"
230 "cat" =~ /\o{143}\x61\x74/ # matches in ASCII, but a weird way
233 If you've been around Perl a while, all this talk of escape sequences
234 may seem familiar. Similar escape sequences are used in double-quoted
235 strings and in fact the regexps in Perl are mostly treated as
236 double-quoted strings. This means that variables can be used in
237 regexps as well. Just like double-quoted strings, the values of the
238 variables in the regexp will be substituted in before the regexp is
239 evaluated for matching purposes. So we have:
242 'housecat' =~ /$foo/; # matches
243 'cathouse' =~ /cat$foo/; # matches
244 'housecat' =~ /${foo}cat/; # matches
246 So far, so good. With the knowledge above you can already perform
247 searches with just about any literal string regexp you can dream up.
248 Here is a I<very simple> emulation of the Unix grep program:
258 % chmod +x simple_grep
260 % simple_grep abba /usr/dict/words
271 This program is easy to understand. C<#!/usr/bin/perl> is the standard
272 way to invoke a perl program from the shell.
273 S<C<$regexp = shift;>> saves the first command line argument as the
274 regexp to be used, leaving the rest of the command line arguments to
275 be treated as files. S<C<< while (<>) >>> loops over all the lines in
276 all the files. For each line, S<C<print if /$regexp/;>> prints the
277 line if the regexp matches the line. In this line, both C<print> and
278 C</$regexp/> use the default variable C<$_> implicitly.
280 With all of the regexps above, if the regexp matched anywhere in the
281 string, it was considered a match. Sometimes, however, we'd like to
282 specify I<where> in the string the regexp should try to match. To do
283 this, we would use the I<anchor> metacharacters C<'^'> and C<'$'>. The
284 anchor C<'^'> means match at the beginning of the string and the anchor
285 C<'$'> means match at the end of the string, or before a newline at the
286 end of the string. Here is how they are used:
288 "housekeeper" =~ /keeper/; # matches
289 "housekeeper" =~ /^keeper/; # doesn't match
290 "housekeeper" =~ /keeper$/; # matches
291 "housekeeper\n" =~ /keeper$/; # matches
293 The second regexp doesn't match because C<'^'> constrains C<keeper> to
294 match only at the beginning of the string, but C<"housekeeper"> has
295 keeper starting in the middle. The third regexp does match, since the
296 C<'$'> constrains C<keeper> to match only at the end of the string.
298 When both C<'^'> and C<'$'> are used at the same time, the regexp has to
299 match both the beginning and the end of the string, I<i.e.>, the regexp
300 matches the whole string. Consider
302 "keeper" =~ /^keep$/; # doesn't match
303 "keeper" =~ /^keeper$/; # matches
304 "" =~ /^$/; # ^$ matches an empty string
306 The first regexp doesn't match because the string has more to it than
307 C<keep>. Since the second regexp is exactly the string, it
308 matches. Using both C<'^'> and C<'$'> in a regexp forces the complete
309 string to match, so it gives you complete control over which strings
310 match and which don't. Suppose you are looking for a fellow named
311 bert, off in a string by himself:
313 "dogbert" =~ /bert/; # matches, but not what you want
315 "dilbert" =~ /^bert/; # doesn't match, but ..
316 "bertram" =~ /^bert/; # matches, so still not good enough
318 "bertram" =~ /^bert$/; # doesn't match, good
319 "dilbert" =~ /^bert$/; # doesn't match, good
320 "bert" =~ /^bert$/; # matches, perfect
322 Of course, in the case of a literal string, one could just as easily
323 use the string comparison S<C<$string eq 'bert'>> and it would be
324 more efficient. The C<^...$> regexp really becomes useful when we
325 add in the more powerful regexp tools below.
327 =head2 Using character classes
329 Although one can already do quite a lot with the literal string
330 regexps above, we've only scratched the surface of regular expression
331 technology. In this and subsequent sections we will introduce regexp
332 concepts (and associated metacharacter notations) that will allow a
333 regexp to represent not just a single character sequence, but a I<whole
336 One such concept is that of a I<character class>. A character class
337 allows a set of possible characters, rather than just a single
338 character, to match at a particular point in a regexp. You can define
339 your own custom character classes. These
340 are denoted by brackets C<[...]>, with the set of characters
341 to be possibly matched inside. Here are some examples:
343 /cat/; # matches 'cat'
344 /[bcr]at/; # matches 'bat, 'cat', or 'rat'
345 /item[0123456789]/; # matches 'item0' or ... or 'item9'
346 "abc" =~ /[cab]/; # matches 'a'
348 In the last statement, even though C<'c'> is the first character in
349 the class, C<'a'> matches because the first character position in the
350 string is the earliest point at which the regexp can match.
352 /[yY][eE][sS]/; # match 'yes' in a case-insensitive way
353 # 'yes', 'Yes', 'YES', etc.
355 This regexp displays a common task: perform a case-insensitive
356 match. Perl provides a way of avoiding all those brackets by simply
357 appending an C<'i'> to the end of the match. Then C</[yY][eE][sS]/;>
358 can be rewritten as C</yes/i;>. The C<'i'> stands for
359 case-insensitive and is an example of a I<modifier> of the matching
360 operation. We will meet other modifiers later in the tutorial.
362 We saw in the section above that there were ordinary characters, which
363 represented themselves, and special characters, which needed a
364 backslash C<'\'> to represent themselves. The same is true in a
365 character class, but the sets of ordinary and special characters
366 inside a character class are different than those outside a character
367 class. The special characters for a character class are C<-]\^$> (and
368 the pattern delimiter, whatever it is).
369 C<']'> is special because it denotes the end of a character class. C<'$'> is
370 special because it denotes a scalar variable. C<'\'> is special because
371 it is used in escape sequences, just like above. Here is how the
372 special characters C<]$\> are handled:
374 /[\]c]def/; # matches ']def' or 'cdef'
376 /[$x]at/; # matches 'bat', 'cat', or 'rat'
377 /[\$x]at/; # matches '$at' or 'xat'
378 /[\\$x]at/; # matches '\at', 'bat, 'cat', or 'rat'
380 The last two are a little tricky. In C<[\$x]>, the backslash protects
381 the dollar sign, so the character class has two members C<'$'> and C<'x'>.
382 In C<[\\$x]>, the backslash is protected, so C<$x> is treated as a
383 variable and substituted in double quote fashion.
385 The special character C<'-'> acts as a range operator within character
386 classes, so that a contiguous set of characters can be written as a
387 range. With ranges, the unwieldy C<[0123456789]> and C<[abc...xyz]>
388 become the svelte C<[0-9]> and C<[a-z]>. Some examples are
390 /item[0-9]/; # matches 'item0' or ... or 'item9'
391 /[0-9bx-z]aa/; # matches '0aa', ..., '9aa',
392 # 'baa', 'xaa', 'yaa', or 'zaa'
393 /[0-9a-fA-F]/; # matches a hexadecimal digit
394 /[0-9a-zA-Z_]/; # matches a "word" character,
395 # like those in a Perl variable name
397 If C<'-'> is the first or last character in a character class, it is
398 treated as an ordinary character; C<[-ab]>, C<[ab-]> and C<[a\-b]> are
401 The special character C<'^'> in the first position of a character class
402 denotes a I<negated character class>, which matches any character but
403 those in the brackets. Both C<[...]> and C<[^...]> must match a
404 character, or the match fails. Then
406 /[^a]at/; # doesn't match 'aat' or 'at', but matches
407 # all other 'bat', 'cat, '0at', '%at', etc.
408 /[^0-9]/; # matches a non-numeric character
409 /[a^]at/; # matches 'aat' or '^at'; here '^' is ordinary
411 Now, even C<[0-9]> can be a bother to write multiple times, so in the
412 interest of saving keystrokes and making regexps more readable, Perl
413 has several abbreviations for common character classes, as shown below.
414 Since the introduction of Unicode, unless the C</a> modifier is in
415 effect, these character classes match more than just a few characters in
422 C<\d> matches a digit, not just C<[0-9]> but also digits from non-roman scripts
426 C<\s> matches a whitespace character, the set C<[\ \t\r\n\f]> and others
430 C<\w> matches a word character (alphanumeric or C<'_'>), not just C<[0-9a-zA-Z_]>
431 but also digits and characters from non-roman scripts
435 C<\D> is a negated C<\d>; it represents any other character than a digit, or C<[^\d]>
439 C<\S> is a negated C<\s>; it represents any non-whitespace character C<[^\s]>
443 C<\W> is a negated C<\w>; it represents any non-word character C<[^\w]>
447 The period C<'.'> matches any character but C<"\n"> (unless the modifier C</s> is
448 in effect, as explained below).
452 C<\N>, like the period, matches any character but C<"\n">, but it does so
453 regardless of whether the modifier C</s> is in effect.
457 The C</a> modifier, available starting in Perl 5.14, is used to
458 restrict the matches of C<\d>, C<\s>, and C<\w> to just those in the ASCII range.
459 It is useful to keep your program from being needlessly exposed to full
460 Unicode (and its accompanying security considerations) when all you want
461 is to process English-like text. (The "a" may be doubled, C</aa>, to
462 provide even more restrictions, preventing case-insensitive matching of
463 ASCII with non-ASCII characters; otherwise a Unicode "Kelvin Sign"
464 would caselessly match a "k" or "K".)
466 The C<\d\s\w\D\S\W> abbreviations can be used both inside and outside
467 of bracketed character classes. Here are some in use:
469 /\d\d:\d\d:\d\d/; # matches a hh:mm:ss time format
470 /[\d\s]/; # matches any digit or whitespace character
471 /\w\W\w/; # matches a word char, followed by a
472 # non-word char, followed by a word char
473 /..rt/; # matches any two chars, followed by 'rt'
474 /end\./; # matches 'end.'
475 /end[.]/; # same thing, matches 'end.'
477 Because a period is a metacharacter, it needs to be escaped to match
478 as an ordinary period. Because, for example, C<\d> and C<\w> are sets
479 of characters, it is incorrect to think of C<[^\d\w]> as C<[\D\W]>; in
480 fact C<[^\d\w]> is the same as C<[^\w]>, which is the same as
481 C<[\W]>. Think De Morgan's laws.
483 In actuality, the period and C<\d\s\w\D\S\W> abbreviations are
484 themselves types of character classes, so the ones surrounded by
485 brackets are just one type of character class. When we need to make a
486 distinction, we refer to them as "bracketed character classes."
488 An anchor useful in basic regexps is the I<word anchor>
489 C<\b>. This matches a boundary between a word character and a non-word
490 character C<\w\W> or C<\W\w>:
492 $x = "Housecat catenates house and cat";
493 $x =~ /cat/; # matches cat in 'housecat'
494 $x =~ /\bcat/; # matches cat in 'catenates'
495 $x =~ /cat\b/; # matches cat in 'housecat'
496 $x =~ /\bcat\b/; # matches 'cat' at end of string
498 Note in the last example, the end of the string is considered a word
501 For natural language processing (so that, for example, apostrophes are
502 included in words), use instead C<\b{wb}>
504 "don't" =~ / .+? \b{wb} /x; # matches the whole string
506 You might wonder why C<'.'> matches everything but C<"\n"> - why not
507 every character? The reason is that often one is matching against
508 lines and would like to ignore the newline characters. For instance,
509 while the string C<"\n"> represents one line, we would like to think
512 "" =~ /^$/; # matches
513 "\n" =~ /^$/; # matches, $ anchors before "\n"
515 "" =~ /./; # doesn't match; it needs a char
516 "" =~ /^.$/; # doesn't match; it needs a char
517 "\n" =~ /^.$/; # doesn't match; it needs a char other than "\n"
518 "a" =~ /^.$/; # matches
519 "a\n" =~ /^.$/; # matches, $ anchors before "\n"
521 This behavior is convenient, because we usually want to ignore
522 newlines when we count and match characters in a line. Sometimes,
523 however, we want to keep track of newlines. We might even want C<'^'>
524 and C<'$'> to anchor at the beginning and end of lines within the
525 string, rather than just the beginning and end of the string. Perl
526 allows us to choose between ignoring and paying attention to newlines
527 by using the C</s> and C</m> modifiers. C</s> and C</m> stand for
528 single line and multi-line and they determine whether a string is to
529 be treated as one continuous string, or as a set of lines. The two
530 modifiers affect two aspects of how the regexp is interpreted: 1) how
531 the C<'.'> character class is defined, and 2) where the anchors C<'^'>
532 and C<'$'> are able to match. Here are the four possible combinations:
538 no modifiers: Default behavior. C<'.'> matches any character
539 except C<"\n">. C<'^'> matches only at the beginning of the string and
540 C<'$'> matches only at the end or before a newline at the end.
544 s modifier (C</s>): Treat string as a single long line. C<'.'> matches
545 any character, even C<"\n">. C<'^'> matches only at the beginning of
546 the string and C<'$'> matches only at the end or before a newline at the
551 m modifier (C</m>): Treat string as a set of multiple lines. C<'.'>
552 matches any character except C<"\n">. C<'^'> and C<'$'> are able to match
553 at the start or end of I<any> line within the string.
557 both s and m modifiers (C</sm>): Treat string as a single long line, but
558 detect multiple lines. C<'.'> matches any character, even
559 C<"\n">. C<'^'> and C<'$'>, however, are able to match at the start or end
560 of I<any> line within the string.
564 Here are examples of C</s> and C</m> in action:
566 $x = "There once was a girl\nWho programmed in Perl\n";
568 $x =~ /^Who/; # doesn't match, "Who" not at start of string
569 $x =~ /^Who/s; # doesn't match, "Who" not at start of string
570 $x =~ /^Who/m; # matches, "Who" at start of second line
571 $x =~ /^Who/sm; # matches, "Who" at start of second line
573 $x =~ /girl.Who/; # doesn't match, "." doesn't match "\n"
574 $x =~ /girl.Who/s; # matches, "." matches "\n"
575 $x =~ /girl.Who/m; # doesn't match, "." doesn't match "\n"
576 $x =~ /girl.Who/sm; # matches, "." matches "\n"
578 Most of the time, the default behavior is what is wanted, but C</s> and
579 C</m> are occasionally very useful. If C</m> is being used, the start
580 of the string can still be matched with C<\A> and the end of the string
581 can still be matched with the anchors C<\Z> (matches both the end and
582 the newline before, like C<'$'>), and C<\z> (matches only the end):
584 $x =~ /^Who/m; # matches, "Who" at start of second line
585 $x =~ /\AWho/m; # doesn't match, "Who" is not at start of string
587 $x =~ /girl$/m; # matches, "girl" at end of first line
588 $x =~ /girl\Z/m; # doesn't match, "girl" is not at end of string
590 $x =~ /Perl\Z/m; # matches, "Perl" is at newline before end
591 $x =~ /Perl\z/m; # doesn't match, "Perl" is not at end of string
593 We now know how to create choices among classes of characters in a
594 regexp. What about choices among words or character strings? Such
595 choices are described in the next section.
597 =head2 Matching this or that
599 Sometimes we would like our regexp to be able to match different
600 possible words or character strings. This is accomplished by using
601 the I<alternation> metacharacter C<'|'>. To match C<dog> or C<cat>, we
602 form the regexp C<dog|cat>. As before, Perl will try to match the
603 regexp at the earliest possible point in the string. At each
604 character position, Perl will first try to match the first
605 alternative, C<dog>. If C<dog> doesn't match, Perl will then try the
606 next alternative, C<cat>. If C<cat> doesn't match either, then the
607 match fails and Perl moves to the next position in the string. Some
610 "cats and dogs" =~ /cat|dog|bird/; # matches "cat"
611 "cats and dogs" =~ /dog|cat|bird/; # matches "cat"
613 Even though C<dog> is the first alternative in the second regexp,
614 C<cat> is able to match earlier in the string.
616 "cats" =~ /c|ca|cat|cats/; # matches "c"
617 "cats" =~ /cats|cat|ca|c/; # matches "cats"
619 Here, all the alternatives match at the first string position, so the
620 first alternative is the one that matches. If some of the
621 alternatives are truncations of the others, put the longest ones first
622 to give them a chance to match.
624 "cab" =~ /a|b|c/ # matches "c"
627 The last example points out that character classes are like
628 alternations of characters. At a given character position, the first
629 alternative that allows the regexp match to succeed will be the one
632 =head2 Grouping things and hierarchical matching
634 Alternation allows a regexp to choose among alternatives, but by
635 itself it is unsatisfying. The reason is that each alternative is a whole
636 regexp, but sometime we want alternatives for just part of a
637 regexp. For instance, suppose we want to search for housecats or
638 housekeepers. The regexp C<housecat|housekeeper> fits the bill, but is
639 inefficient because we had to type C<house> twice. It would be nice to
640 have parts of the regexp be constant, like C<house>, and some
641 parts have alternatives, like C<cat|keeper>.
643 The I<grouping> metacharacters C<()> solve this problem. Grouping
644 allows parts of a regexp to be treated as a single unit. Parts of a
645 regexp are grouped by enclosing them in parentheses. Thus we could solve
646 the C<housecat|housekeeper> by forming the regexp as
647 C<house(cat|keeper)>. The regexp C<house(cat|keeper)> means match
648 C<house> followed by either C<cat> or C<keeper>. Some more examples
651 /(a|b)b/; # matches 'ab' or 'bb'
652 /(ac|b)b/; # matches 'acb' or 'bb'
653 /(^a|b)c/; # matches 'ac' at start of string or 'bc' anywhere
654 /(a|[bc])d/; # matches 'ad', 'bd', or 'cd'
656 /house(cat|)/; # matches either 'housecat' or 'house'
657 /house(cat(s|)|)/; # matches either 'housecats' or 'housecat' or
658 # 'house'. Note groups can be nested.
660 /(19|20|)\d\d/; # match years 19xx, 20xx, or the Y2K problem, xx
661 "20" =~ /(19|20|)\d\d/; # matches the null alternative '()\d\d',
662 # because '20\d\d' can't match
664 Alternations behave the same way in groups as out of them: at a given
665 string position, the leftmost alternative that allows the regexp to
666 match is taken. So in the last example at the first string position,
667 C<"20"> matches the second alternative, but there is nothing left over
668 to match the next two digits C<\d\d>. So Perl moves on to the next
669 alternative, which is the null alternative and that works, since
670 C<"20"> is two digits.
672 The process of trying one alternative, seeing if it matches, and
673 moving on to the next alternative, while going back in the string
674 from where the previous alternative was tried, if it doesn't, is called
675 I<backtracking>. The term "backtracking" comes from the idea that
676 matching a regexp is like a walk in the woods. Successfully matching
677 a regexp is like arriving at a destination. There are many possible
678 trailheads, one for each string position, and each one is tried in
679 order, left to right. From each trailhead there may be many paths,
680 some of which get you there, and some which are dead ends. When you
681 walk along a trail and hit a dead end, you have to backtrack along the
682 trail to an earlier point to try another trail. If you hit your
683 destination, you stop immediately and forget about trying all the
684 other trails. You are persistent, and only if you have tried all the
685 trails from all the trailheads and not arrived at your destination, do
686 you declare failure. To be concrete, here is a step-by-step analysis
687 of what Perl does when it tries to match the regexp
689 "abcde" =~ /(abd|abc)(df|d|de)/;
693 =item Z<>0. Start with the first letter in the string C<'a'>.
697 =item Z<>1. Try the first alternative in the first group C<'abd'>.
701 =item Z<>2. Match C<'a'> followed by C<'b'>. So far so good.
705 =item Z<>3. C<'d'> in the regexp doesn't match C<'c'> in the string - a
706 dead end. So backtrack two characters and pick the second alternative
707 in the first group C<'abc'>.
711 =item Z<>4. Match C<'a'> followed by C<'b'> followed by C<'c'>. We are on a roll
712 and have satisfied the first group. Set C<$1> to C<'abc'>.
716 =item Z<>5 Move on to the second group and pick the first alternative C<'df'>.
720 =item Z<>6 Match the C<'d'>.
724 =item Z<>7. C<'f'> in the regexp doesn't match C<'e'> in the string, so a dead
725 end. Backtrack one character and pick the second alternative in the
730 =item Z<>8. C<'d'> matches. The second grouping is satisfied, so set
735 =item Z<>9. We are at the end of the regexp, so we are done! We have
736 matched C<'abcd'> out of the string C<"abcde">.
740 There are a couple of things to note about this analysis. First, the
741 third alternative in the second group C<'de'> also allows a match, but we
742 stopped before we got to it - at a given character position, leftmost
743 wins. Second, we were able to get a match at the first character
744 position of the string C<'a'>. If there were no matches at the first
745 position, Perl would move to the second character position C<'b'> and
746 attempt the match all over again. Only when all possible paths at all
747 possible character positions have been exhausted does Perl give
748 up and declare S<C<$string =~ /(abd|abc)(df|d|de)/;>> to be false.
750 Even with all this work, regexp matching happens remarkably fast. To
751 speed things up, Perl compiles the regexp into a compact sequence of
752 opcodes that can often fit inside a processor cache. When the code is
753 executed, these opcodes can then run at full throttle and search very
756 =head2 Extracting matches
758 The grouping metacharacters C<()> also serve another completely
759 different function: they allow the extraction of the parts of a string
760 that matched. This is very useful to find out what matched and for
761 text processing in general. For each grouping, the part that matched
762 inside goes into the special variables C<$1>, C<$2>, I<etc>. They can be
763 used just as ordinary variables:
765 # extract hours, minutes, seconds
766 if ($time =~ /(\d\d):(\d\d):(\d\d)/) { # match hh:mm:ss format
772 Now, we know that in scalar context,
773 S<C<$time =~ /(\d\d):(\d\d):(\d\d)/>> returns a true or false
774 value. In list context, however, it returns the list of matched values
775 C<($1,$2,$3)>. So we could write the code more compactly as
777 # extract hours, minutes, seconds
778 ($hours, $minutes, $second) = ($time =~ /(\d\d):(\d\d):(\d\d)/);
780 If the groupings in a regexp are nested, C<$1> gets the group with the
781 leftmost opening parenthesis, C<$2> the next opening parenthesis,
782 I<etc>. Here is a regexp with nested groups:
784 /(ab(cd|ef)((gi)|j))/;
787 If this regexp matches, C<$1> contains a string starting with
788 C<'ab'>, C<$2> is either set to C<'cd'> or C<'ef'>, C<$3> equals either
789 C<'gi'> or C<'j'>, and C<$4> is either set to C<'gi'>, just like C<$3>,
790 or it remains undefined.
792 For convenience, Perl sets C<$+> to the string held by the highest numbered
793 C<$1>, C<$2>,... that got assigned (and, somewhat related, C<$^N> to the
794 value of the C<$1>, C<$2>,... most-recently assigned; I<i.e.> the C<$1>,
795 C<$2>,... associated with the rightmost closing parenthesis used in the
799 =head2 Backreferences
801 Closely associated with the matching variables C<$1>, C<$2>, ... are
802 the I<backreferences> C<\g1>, C<\g2>,... Backreferences are simply
803 matching variables that can be used I<inside> a regexp. This is a
804 really nice feature; what matches later in a regexp is made to depend on
805 what matched earlier in the regexp. Suppose we wanted to look
806 for doubled words in a text, like "the the". The following regexp finds
807 all 3-letter doubles with a space in between:
811 The grouping assigns a value to C<\g1>, so that the same 3-letter sequence
812 is used for both parts.
814 A similar task is to find words consisting of two identical parts:
816 % simple_grep '^(\w\w\w\w|\w\w\w|\w\w|\w)\g1$' /usr/dict/words
824 The regexp has a single grouping which considers 4-letter
825 combinations, then 3-letter combinations, I<etc>., and uses C<\g1> to look for
826 a repeat. Although C<$1> and C<\g1> represent the same thing, care should be
827 taken to use matched variables C<$1>, C<$2>,... only I<outside> a regexp
828 and backreferences C<\g1>, C<\g2>,... only I<inside> a regexp; not doing
829 so may lead to surprising and unsatisfactory results.
832 =head2 Relative backreferences
834 Counting the opening parentheses to get the correct number for a
835 backreference is error-prone as soon as there is more than one
836 capturing group. A more convenient technique became available
837 with Perl 5.10: relative backreferences. To refer to the immediately
838 preceding capture group one now may write C<\g-1> or C<\g{-1}>, the next but
839 last is available via C<\g-2> or C<\g{-2}>, and so on.
841 Another good reason in addition to readability and maintainability
842 for using relative backreferences is illustrated by the following example,
843 where a simple pattern for matching peculiar strings is used:
845 $a99a = '([a-z])(\d)\g2\g1'; # matches a11a, g22g, x33x, etc.
847 Now that we have this pattern stored as a handy string, we might feel
848 tempted to use it as a part of some other pattern:
851 if ($line =~ /^(\w+)=$a99a$/){ # unexpected behavior!
852 print "$1 is valid\n";
854 print "bad line: '$line'\n";
857 But this doesn't match, at least not the way one might expect. Only
858 after inserting the interpolated C<$a99a> and looking at the resulting
859 full text of the regexp is it obvious that the backreferences have
860 backfired. The subexpression C<(\w+)> has snatched number 1 and
861 demoted the groups in C<$a99a> by one rank. This can be avoided by
862 using relative backreferences:
864 $a99a = '([a-z])(\d)\g{-1}\g{-2}'; # safe for being interpolated
867 =head2 Named backreferences
869 Perl 5.10 also introduced named capture groups and named backreferences.
870 To attach a name to a capturing group, you write either
871 C<< (?<name>...) >> or C<< (?'name'...) >>. The backreference may
872 then be written as C<\g{name}>. It is permissible to attach the
873 same name to more than one group, but then only the leftmost one of the
874 eponymous set can be referenced. Outside of the pattern a named
875 capture group is accessible through the C<%+> hash.
877 Assuming that we have to match calendar dates which may be given in one
878 of the three formats yyyy-mm-dd, mm/dd/yyyy or dd.mm.yyyy, we can write
879 three suitable patterns where we use C<'d'>, C<'m'> and C<'y'> respectively as the
880 names of the groups capturing the pertaining components of a date. The
881 matching operation combines the three patterns as alternatives:
883 $fmt1 = '(?<y>\d\d\d\d)-(?<m>\d\d)-(?<d>\d\d)';
884 $fmt2 = '(?<m>\d\d)/(?<d>\d\d)/(?<y>\d\d\d\d)';
885 $fmt3 = '(?<d>\d\d)\.(?<m>\d\d)\.(?<y>\d\d\d\d)';
886 for my $d (qw(2006-10-21 15.01.2007 10/31/2005)) {
887 if ( $d =~ m{$fmt1|$fmt2|$fmt3} ){
888 print "day=$+{d} month=$+{m} year=$+{y}\n";
892 If any of the alternatives matches, the hash C<%+> is bound to contain the
893 three key-value pairs.
896 =head2 Alternative capture group numbering
898 Yet another capturing group numbering technique (also as from Perl 5.10)
899 deals with the problem of referring to groups within a set of alternatives.
900 Consider a pattern for matching a time of the day, civil or military style:
902 if ( $time =~ /(\d\d|\d):(\d\d)|(\d\d)(\d\d)/ ){
903 # process hour and minute
906 Processing the results requires an additional if statement to determine
907 whether C<$1> and C<$2> or C<$3> and C<$4> contain the goodies. It would
908 be easier if we could use group numbers 1 and 2 in second alternative as
909 well, and this is exactly what the parenthesized construct C<(?|...)>,
910 set around an alternative achieves. Here is an extended version of the
913 if($time =~ /(?|(\d\d|\d):(\d\d)|(\d\d)(\d\d))\s+([A-Z][A-Z][A-Z])/){
914 print "hour=$1 minute=$2 zone=$3\n";
917 Within the alternative numbering group, group numbers start at the same
918 position for each alternative. After the group, numbering continues
919 with one higher than the maximum reached across all the alternatives.
921 =head2 Position information
923 In addition to what was matched, Perl also provides the
924 positions of what was matched as contents of the C<@-> and C<@+>
925 arrays. C<$-[0]> is the position of the start of the entire match and
926 C<$+[0]> is the position of the end. Similarly, C<$-[n]> is the
927 position of the start of the C<$n> match and C<$+[n]> is the position
928 of the end. If C<$n> is undefined, so are C<$-[n]> and C<$+[n]>. Then
931 $x = "Mmm...donut, thought Homer";
932 $x =~ /^(Mmm|Yech)\.\.\.(donut|peas)/; # matches
933 foreach $exp (1..$#-) {
935 print "Match $exp: '$$exp' at position ($-[$exp],$+[$exp])\n";
940 Match 1: 'Mmm' at position (0,3)
941 Match 2: 'donut' at position (6,11)
943 Even if there are no groupings in a regexp, it is still possible to
944 find out what exactly matched in a string. If you use them, Perl
945 will set C<$`> to the part of the string before the match, will set C<$&>
946 to the part of the string that matched, and will set C<$'> to the part
947 of the string after the match. An example:
949 $x = "the cat caught the mouse";
950 $x =~ /cat/; # $` = 'the ', $& = 'cat', $' = ' caught the mouse'
951 $x =~ /the/; # $` = '', $& = 'the', $' = ' cat caught the mouse'
953 In the second match, C<$`> equals C<''> because the regexp matched at the
954 first character position in the string and stopped; it never saw the
957 If your code is to run on Perl versions earlier than
958 5.20, it is worthwhile to note that using C<$`> and C<$'>
959 slows down regexp matching quite a bit, while C<$&> slows it down to a
960 lesser extent, because if they are used in one regexp in a program,
961 they are generated for I<all> regexps in the program. So if raw
962 performance is a goal of your application, they should be avoided.
963 If you need to extract the corresponding substrings, use C<@-> and
966 $` is the same as substr( $x, 0, $-[0] )
967 $& is the same as substr( $x, $-[0], $+[0]-$-[0] )
968 $' is the same as substr( $x, $+[0] )
970 As of Perl 5.10, the C<${^PREMATCH}>, C<${^MATCH}> and C<${^POSTMATCH}>
971 variables may be used. These are only set if the C</p> modifier is
972 present. Consequently they do not penalize the rest of the program. In
973 Perl 5.20, C<${^PREMATCH}>, C<${^MATCH}> and C<${^POSTMATCH}> are available
974 whether the C</p> has been used or not (the modifier is ignored), and
975 C<$`>, C<$'> and C<$&> do not cause any speed difference.
977 =head2 Non-capturing groupings
979 A group that is required to bundle a set of alternatives may or may not be
980 useful as a capturing group. If it isn't, it just creates a superfluous
981 addition to the set of available capture group values, inside as well as
982 outside the regexp. Non-capturing groupings, denoted by C<(?:regexp)>,
983 still allow the regexp to be treated as a single unit, but don't establish
984 a capturing group at the same time. Both capturing and non-capturing
985 groupings are allowed to co-exist in the same regexp. Because there is
986 no extraction, non-capturing groupings are faster than capturing
987 groupings. Non-capturing groupings are also handy for choosing exactly
988 which parts of a regexp are to be extracted to matching variables:
990 # match a number, $1-$4 are set, but we only want $1
991 /([+-]?\ *(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?)/;
993 # match a number faster , only $1 is set
994 /([+-]?\ *(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?)/;
996 # match a number, get $1 = whole number, $2 = exponent
997 /([+-]?\ *(?:\d+(?:\.\d*)?|\.\d+)(?:[eE]([+-]?\d+))?)/;
999 Non-capturing groupings are also useful for removing nuisance
1000 elements gathered from a split operation where parentheses are
1001 required for some reason:
1004 @num = split /(a|b)+/, $x; # @num = ('12','a','34','a','5')
1005 @num = split /(?:a|b)+/, $x; # @num = ('12','34','5')
1007 In Perl 5.22 and later, all groups within a regexp can be set to
1008 non-capturing by using the new C</n> flag:
1010 "hello" =~ /(hi|hello)/n; # $1 is not set!
1012 See L<perlre/"n"> for more information.
1014 =head2 Matching repetitions
1016 The examples in the previous section display an annoying weakness. We
1017 were only matching 3-letter words, or chunks of words of 4 letters or
1018 less. We'd like to be able to match words or, more generally, strings
1019 of any length, without writing out tedious alternatives like
1020 C<\w\w\w\w|\w\w\w|\w\w|\w>.
1022 This is exactly the problem the I<quantifier> metacharacters C<'?'>,
1023 C<'*'>, C<'+'>, and C<{}> were created for. They allow us to delimit the
1024 number of repeats for a portion of a regexp we consider to be a
1025 match. Quantifiers are put immediately after the character, character
1026 class, or grouping that we want to specify. They have the following
1033 C<a?> means: match C<'a'> 1 or 0 times
1037 C<a*> means: match C<'a'> 0 or more times, I<i.e.>, any number of times
1041 C<a+> means: match C<'a'> 1 or more times, I<i.e.>, at least once
1045 C<a{n,m}> means: match at least C<n> times, but not more than C<m>
1050 C<a{n,}> means: match at least C<n> or more times
1054 C<a{,n}> means: match at most C<n> times, or fewer
1058 C<a{n}> means: match exactly C<n> times
1062 If you like, you can add blanks (tab or space characters) within the
1063 braces, but adjacent to them, and/or next to the comma (if any).
1065 Here are some examples:
1067 /[a-z]+\s+\d*/; # match a lowercase word, at least one space, and
1068 # any number of digits
1069 /(\w+)\s+\g1/; # match doubled words of arbitrary length
1070 /y(es)?/i; # matches 'y', 'Y', or a case-insensitive 'yes'
1071 $year =~ /^\d{2,4}$/; # make sure year is at least 2 but not more
1073 $year =~ /^\d{ 2, 4 }$/; # Same; for those who like wide open
1075 $year =~ /^\d{2, 4}$/; # Same.
1076 $year =~ /^\d{4}$|^\d{2}$/; # better match; throw out 3-digit dates
1077 $year =~ /^\d{2}(\d{2})?$/; # same thing written differently.
1078 # However, this captures the last two
1079 # digits in $1 and the other does not.
1081 % simple_grep '^(\w+)\g1$' /usr/dict/words # isn't this easier?
1089 For all of these quantifiers, Perl will try to match as much of the
1090 string as possible, while still allowing the regexp to succeed. Thus
1091 with C</a?.../>, Perl will first try to match the regexp with the C<'a'>
1092 present; if that fails, Perl will try to match the regexp without the
1093 C<'a'> present. For the quantifier C<'*'>, we get the following:
1095 $x = "the cat in the hat";
1096 $x =~ /^(.*)(cat)(.*)$/; # matches,
1099 # $3 = ' in the hat'
1101 Which is what we might expect, the match finds the only C<cat> in the
1102 string and locks onto it. Consider, however, this regexp:
1104 $x =~ /^(.*)(at)(.*)$/; # matches,
1105 # $1 = 'the cat in the h'
1107 # $3 = '' (0 characters match)
1109 One might initially guess that Perl would find the C<at> in C<cat> and
1110 stop there, but that wouldn't give the longest possible string to the
1111 first quantifier C<.*>. Instead, the first quantifier C<.*> grabs as
1112 much of the string as possible while still having the regexp match. In
1113 this example, that means having the C<at> sequence with the final C<at>
1114 in the string. The other important principle illustrated here is that,
1115 when there are two or more elements in a regexp, the I<leftmost>
1116 quantifier, if there is one, gets to grab as much of the string as
1117 possible, leaving the rest of the regexp to fight over scraps. Thus in
1118 our example, the first quantifier C<.*> grabs most of the string, while
1119 the second quantifier C<.*> gets the empty string. Quantifiers that
1120 grab as much of the string as possible are called I<maximal match> or
1121 I<greedy> quantifiers.
1123 When a regexp can match a string in several different ways, we can use
1124 the principles above to predict which way the regexp will match:
1130 Principle 0: Taken as a whole, any regexp will be matched at the
1131 earliest possible position in the string.
1135 Principle 1: In an alternation C<a|b|c...>, the leftmost alternative
1136 that allows a match for the whole regexp will be the one used.
1140 Principle 2: The maximal matching quantifiers C<'?'>, C<'*'>, C<'+'> and
1141 C<{n,m}> will in general match as much of the string as possible while
1142 still allowing the whole regexp to match.
1146 Principle 3: If there are two or more elements in a regexp, the
1147 leftmost greedy quantifier, if any, will match as much of the string
1148 as possible while still allowing the whole regexp to match. The next
1149 leftmost greedy quantifier, if any, will try to match as much of the
1150 string remaining available to it as possible, while still allowing the
1151 whole regexp to match. And so on, until all the regexp elements are
1156 As we have seen above, Principle 0 overrides the others. The regexp
1157 will be matched as early as possible, with the other principles
1158 determining how the regexp matches at that earliest character
1161 Here is an example of these principles in action:
1163 $x = "The programming republic of Perl";
1164 $x =~ /^(.+)(e|r)(.*)$/; # matches,
1165 # $1 = 'The programming republic of Pe'
1169 This regexp matches at the earliest string position, C<'T'>. One
1170 might think that C<'e'>, being leftmost in the alternation, would be
1171 matched, but C<'r'> produces the longest string in the first quantifier.
1173 $x =~ /(m{1,2})(.*)$/; # matches,
1175 # $2 = 'ing republic of Perl'
1177 Here, The earliest possible match is at the first C<'m'> in
1178 C<programming>. C<m{1,2}> is the first quantifier, so it gets to match
1181 $x =~ /.*(m{1,2})(.*)$/; # matches,
1183 # $2 = 'ing republic of Perl'
1185 Here, the regexp matches at the start of the string. The first
1186 quantifier C<.*> grabs as much as possible, leaving just a single
1187 C<'m'> for the second quantifier C<m{1,2}>.
1189 $x =~ /(.?)(m{1,2})(.*)$/; # matches,
1192 # $3 = 'ing republic of Perl'
1194 Here, C<.?> eats its maximal one character at the earliest possible
1195 position in the string, C<'a'> in C<programming>, leaving C<m{1,2}>
1196 the opportunity to match both C<'m'>'s. Finally,
1198 "aXXXb" =~ /(X*)/; # matches with $1 = ''
1200 because it can match zero copies of C<'X'> at the beginning of the
1201 string. If you definitely want to match at least one C<'X'>, use
1204 Sometimes greed is not good. At times, we would like quantifiers to
1205 match a I<minimal> piece of string, rather than a maximal piece. For
1206 this purpose, Larry Wall created the I<minimal match> or
1207 I<non-greedy> quantifiers C<??>, C<*?>, C<+?>, and C<{}?>. These are
1208 the usual quantifiers with a C<'?'> appended to them. They have the
1215 C<a??> means: match C<'a'> 0 or 1 times. Try 0 first, then 1.
1219 C<a*?> means: match C<'a'> 0 or more times, I<i.e.>, any number of times,
1220 but as few times as possible
1224 C<a+?> means: match C<'a'> 1 or more times, I<i.e.>, at least once, but
1225 as few times as possible
1229 C<a{n,m}?> means: match at least C<n> times, not more than C<m>
1230 times, as few times as possible
1234 C<a{n,}?> means: match at least C<n> times, but as few times as
1239 C<a{,n}?> means: match at most C<n> times, but as few times as
1244 C<a{n}?> means: match exactly C<n> times. Because we match exactly
1245 C<n> times, C<a{n}?> is equivalent to C<a{n}> and is just there for
1246 notational consistency.
1250 Let's look at the example above, but with minimal quantifiers:
1252 $x = "The programming republic of Perl";
1253 $x =~ /^(.+?)(e|r)(.*)$/; # matches,
1256 # $3 = ' programming republic of Perl'
1258 The minimal string that will allow both the start of the string C<'^'>
1259 and the alternation to match is C<Th>, with the alternation C<e|r>
1260 matching C<'e'>. The second quantifier C<.*> is free to gobble up the
1263 $x =~ /(m{1,2}?)(.*?)$/; # matches,
1265 # $2 = 'ming republic of Perl'
1267 The first string position that this regexp can match is at the first
1268 C<'m'> in C<programming>. At this position, the minimal C<m{1,2}?>
1269 matches just one C<'m'>. Although the second quantifier C<.*?> would
1270 prefer to match no characters, it is constrained by the end-of-string
1271 anchor C<'$'> to match the rest of the string.
1273 $x =~ /(.*?)(m{1,2}?)(.*)$/; # matches,
1276 # $3 = 'ming republic of Perl'
1278 In this regexp, you might expect the first minimal quantifier C<.*?>
1279 to match the empty string, because it is not constrained by a C<'^'>
1280 anchor to match the beginning of the word. Principle 0 applies here,
1281 however. Because it is possible for the whole regexp to match at the
1282 start of the string, it I<will> match at the start of the string. Thus
1283 the first quantifier has to match everything up to the first C<'m'>. The
1284 second minimal quantifier matches just one C<'m'> and the third
1285 quantifier matches the rest of the string.
1287 $x =~ /(.??)(m{1,2})(.*)$/; # matches,
1290 # $3 = 'ing republic of Perl'
1292 Just as in the previous regexp, the first quantifier C<.??> can match
1293 earliest at position C<'a'>, so it does. The second quantifier is
1294 greedy, so it matches C<mm>, and the third matches the rest of the
1297 We can modify principle 3 above to take into account non-greedy
1304 Principle 3: If there are two or more elements in a regexp, the
1305 leftmost greedy (non-greedy) quantifier, if any, will match as much
1306 (little) of the string as possible while still allowing the whole
1307 regexp to match. The next leftmost greedy (non-greedy) quantifier, if
1308 any, will try to match as much (little) of the string remaining
1309 available to it as possible, while still allowing the whole regexp to
1310 match. And so on, until all the regexp elements are satisfied.
1314 Just like alternation, quantifiers are also susceptible to
1315 backtracking. Here is a step-by-step analysis of the example
1317 $x = "the cat in the hat";
1318 $x =~ /^(.*)(at)(.*)$/; # matches,
1319 # $1 = 'the cat in the h'
1321 # $3 = '' (0 matches)
1325 =item Z<>0. Start with the first letter in the string C<'t'>.
1329 =item Z<>1. The first quantifier C<'.*'> starts out by matching the whole
1330 string "C<the cat in the hat>".
1334 =item Z<>2. C<'a'> in the regexp element C<'at'> doesn't match the end
1335 of the string. Backtrack one character.
1339 =item Z<>3. C<'a'> in the regexp element C<'at'> still doesn't match
1340 the last letter of the string C<'t'>, so backtrack one more character.
1344 =item Z<>4. Now we can match the C<'a'> and the C<'t'>.
1348 =item Z<>5. Move on to the third element C<'.*'>. Since we are at the
1349 end of the string and C<'.*'> can match 0 times, assign it the empty
1354 =item Z<>6. We are done!
1358 Most of the time, all this moving forward and backtracking happens
1359 quickly and searching is fast. There are some pathological regexps,
1360 however, whose execution time exponentially grows with the size of the
1361 string. A typical structure that blows up in your face is of the form
1365 The problem is the nested indeterminate quantifiers. There are many
1366 different ways of partitioning a string of length n between the C<'+'>
1367 and C<'*'>: one repetition with C<b+> of length n, two repetitions with
1368 the first C<b+> length k and the second with length n-k, m repetitions
1369 whose bits add up to length n, I<etc>. In fact there are an exponential
1370 number of ways to partition a string as a function of its length. A
1371 regexp may get lucky and match early in the process, but if there is
1372 no match, Perl will try I<every> possibility before giving up. So be
1373 careful with nested C<'*'>'s, C<{n,m}>'s, and C<'+'>'s. The book
1374 I<Mastering Regular Expressions> by Jeffrey Friedl gives a wonderful
1375 discussion of this and other efficiency issues.
1378 =head2 Possessive quantifiers
1380 Backtracking during the relentless search for a match may be a waste
1381 of time, particularly when the match is bound to fail. Consider
1384 /^\w+\s+\w+$/; # a word, spaces, a word
1386 Whenever this is applied to a string which doesn't quite meet the
1387 pattern's expectations such as S<C<"abc ">> or S<C<"abc def ">>,
1388 the regexp engine will backtrack, approximately once for each character
1389 in the string. But we know that there is no way around taking I<all>
1390 of the initial word characters to match the first repetition, that I<all>
1391 spaces must be eaten by the middle part, and the same goes for the second
1394 With the introduction of the I<possessive quantifiers> in Perl 5.10, we
1395 have a way of instructing the regexp engine not to backtrack, with the
1396 usual quantifiers with a C<'+'> appended to them. This makes them greedy as
1397 well as stingy; once they succeed they won't give anything back to permit
1398 another solution. They have the following meanings:
1404 C<a{n,m}+> means: match at least C<n> times, not more than C<m> times,
1405 as many times as possible, and don't give anything up. C<a?+> is short
1410 C<a{n,}+> means: match at least C<n> times, but as many times as possible,
1411 and don't give anything up. C<a++> is short for C<a{1,}+>.
1415 C<a{,n}+> means: match as many times as possible up to at most C<n>
1416 times, and don't give anything up. C<a*+> is short for C<a{0,}+>.
1420 C<a{n}+> means: match exactly C<n> times. It is just there for
1421 notational consistency.
1425 These possessive quantifiers represent a special case of a more general
1426 concept, the I<independent subexpression>, see below.
1428 As an example where a possessive quantifier is suitable we consider
1429 matching a quoted string, as it appears in several programming languages.
1430 The backslash is used as an escape character that indicates that the
1431 next character is to be taken literally, as another character for the
1432 string. Therefore, after the opening quote, we expect a (possibly
1433 empty) sequence of alternatives: either some character except an
1434 unescaped quote or backslash or an escaped character.
1436 /"(?:[^"\\]++|\\.)*+"/;
1439 =head2 Building a regexp
1441 At this point, we have all the basic regexp concepts covered, so let's
1442 give a more involved example of a regular expression. We will build a
1443 regexp that matches numbers.
1445 The first task in building a regexp is to decide what we want to match
1446 and what we want to exclude. In our case, we want to match both
1447 integers and floating point numbers and we want to reject any string
1448 that isn't a number.
1450 The next task is to break the problem down into smaller problems that
1451 are easily converted into a regexp.
1453 The simplest case is integers. These consist of a sequence of digits,
1454 with an optional sign in front. The digits we can represent with
1455 C<\d+> and the sign can be matched with C<[+-]>. Thus the integer
1458 /[+-]?\d+/; # matches integers
1460 A floating point number potentially has a sign, an integral part, a
1461 decimal point, a fractional part, and an exponent. One or more of these
1462 parts is optional, so we need to check out the different
1463 possibilities. Floating point numbers which are in proper form include
1464 123., 0.345, .34, -1e6, and 25.4E-72. As with integers, the sign out
1465 front is completely optional and can be matched by C<[+-]?>. We can
1466 see that if there is no exponent, floating point numbers must have a
1467 decimal point, otherwise they are integers. We might be tempted to
1468 model these with C<\d*\.\d*>, but this would also match just a single
1469 decimal point, which is not a number. So the three cases of floating
1470 point number without exponent are
1472 /[+-]?\d+\./; # 1., 321., etc.
1473 /[+-]?\.\d+/; # .1, .234, etc.
1474 /[+-]?\d+\.\d+/; # 1.0, 30.56, etc.
1476 These can be combined into a single regexp with a three-way alternation:
1478 /[+-]?(\d+\.\d+|\d+\.|\.\d+)/; # floating point, no exponent
1480 In this alternation, it is important to put C<'\d+\.\d+'> before
1481 C<'\d+\.'>. If C<'\d+\.'> were first, the regexp would happily match that
1482 and ignore the fractional part of the number.
1484 Now consider floating point numbers with exponents. The key
1485 observation here is that I<both> integers and numbers with decimal
1486 points are allowed in front of an exponent. Then exponents, like the
1487 overall sign, are independent of whether we are matching numbers with
1488 or without decimal points, and can be "decoupled" from the
1489 mantissa. The overall form of the regexp now becomes clear:
1491 /^(optional sign)(integer | f.p. mantissa)(optional exponent)$/;
1493 The exponent is an C<'e'> or C<'E'>, followed by an integer. So the
1496 /[eE][+-]?\d+/; # exponent
1498 Putting all the parts together, we get a regexp that matches numbers:
1500 /^[+-]?(\d+\.\d+|\d+\.|\.\d+|\d+)([eE][+-]?\d+)?$/; # Ta da!
1502 Long regexps like this may impress your friends, but can be hard to
1503 decipher. In complex situations like this, the C</x> modifier for a
1504 match is invaluable. It allows one to put nearly arbitrary whitespace
1505 and comments into a regexp without affecting their meaning. Using it,
1506 we can rewrite our "extended" regexp in the more pleasing form
1509 [+-]? # first, match an optional sign
1510 ( # then match integers or f.p. mantissas:
1511 \d+\.\d+ # mantissa of the form a.b
1512 |\d+\. # mantissa of the form a.
1513 |\.\d+ # mantissa of the form .b
1514 |\d+ # integer of the form a
1516 ( [eE] [+-]? \d+ )? # finally, optionally match an exponent
1519 If whitespace is mostly irrelevant, how does one include space
1520 characters in an extended regexp? The answer is to backslash it
1521 S<C<'\ '>> or put it in a character class S<C<[ ]>>. The same thing
1522 goes for pound signs: use C<\#> or C<[#]>. For instance, Perl allows
1523 a space between the sign and the mantissa or integer, and we could add
1524 this to our regexp as follows:
1527 [+-]?\ * # first, match an optional sign *and space*
1528 ( # then match integers or f.p. mantissas:
1529 \d+\.\d+ # mantissa of the form a.b
1530 |\d+\. # mantissa of the form a.
1531 |\.\d+ # mantissa of the form .b
1532 |\d+ # integer of the form a
1534 ( [eE] [+-]? \d+ )? # finally, optionally match an exponent
1537 In this form, it is easier to see a way to simplify the
1538 alternation. Alternatives 1, 2, and 4 all start with C<\d+>, so it
1539 could be factored out:
1542 [+-]?\ * # first, match an optional sign
1543 ( # then match integers or f.p. mantissas:
1544 \d+ # start out with a ...
1546 \.\d* # mantissa of the form a.b or a.
1547 )? # ? takes care of integers of the form a
1548 |\.\d+ # mantissa of the form .b
1550 ( [eE] [+-]? \d+ )? # finally, optionally match an exponent
1553 Starting in Perl v5.26, specifying C</xx> changes the square-bracketed
1554 portions of a pattern to ignore tabs and space characters unless they
1555 are escaped by preceding them with a backslash. So, we could write
1558 [ + - ]?\ * # first, match an optional sign
1559 ( # then match integers or f.p. mantissas:
1560 \d+ # start out with a ...
1562 \.\d* # mantissa of the form a.b or a.
1563 )? # ? takes care of integers of the form a
1564 |\.\d+ # mantissa of the form .b
1566 ( [ e E ] [ + - ]? \d+ )? # finally, optionally match an exponent
1569 This doesn't really improve the legibility of this example, but it's
1570 available in case you want it. Squashing the pattern down to the
1571 compact form, we have
1573 /^[+-]?\ *(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?$/;
1575 This is our final regexp. To recap, we built a regexp by
1581 specifying the task in detail,
1585 breaking down the problem into smaller parts,
1589 translating the small parts into regexps,
1593 combining the regexps,
1597 and optimizing the final combined regexp.
1601 These are also the typical steps involved in writing a computer
1602 program. This makes perfect sense, because regular expressions are
1603 essentially programs written in a little computer language that specifies
1606 =head2 Using regular expressions in Perl
1608 The last topic of Part 1 briefly covers how regexps are used in Perl
1609 programs. Where do they fit into Perl syntax?
1611 We have already introduced the matching operator in its default
1612 C</regexp/> and arbitrary delimiter C<m!regexp!> forms. We have used
1613 the binding operator C<=~> and its negation C<!~> to test for string
1614 matches. Associated with the matching operator, we have discussed the
1615 single line C</s>, multi-line C</m>, case-insensitive C</i> and
1616 extended C</x> modifiers. There are a few more things you might
1617 want to know about matching operators.
1619 =head3 Prohibiting substitution
1621 If you change C<$pattern> after the first substitution happens, Perl
1622 will ignore it. If you don't want any substitutions at all, use the
1623 special delimiter C<m''>:
1625 @pattern = ('Seuss');
1627 print if m'@pattern'; # matches literal '@pattern', not 'Seuss'
1630 Similar to strings, C<m''> acts like apostrophes on a regexp; all other
1631 C<'m'> delimiters act like quotes. If the regexp evaluates to the empty string,
1632 the regexp in the I<last successful match> is used instead. So we have
1634 "dog" =~ /d/; # 'd' matches
1635 "dogbert" =~ //; # this matches the 'd' regexp used before
1638 =head3 Global matching
1640 The final two modifiers we will discuss here,
1641 C</g> and C</c>, concern multiple matches.
1642 The modifier C</g> stands for global matching and allows the
1643 matching operator to match within a string as many times as possible.
1644 In scalar context, successive invocations against a string will have
1645 C</g> jump from match to match, keeping track of position in the
1646 string as it goes along. You can get or set the position with the
1649 The use of C</g> is shown in the following example. Suppose we have
1650 a string that consists of words separated by spaces. If we know how
1651 many words there are in advance, we could extract the words using
1654 $x = "cat dog house"; # 3 words
1655 $x =~ /^\s*(\w+)\s+(\w+)\s+(\w+)\s*$/; # matches,
1660 But what if we had an indeterminate number of words? This is the sort
1661 of task C</g> was made for. To extract all words, form the simple
1662 regexp C<(\w+)> and loop over all matches with C</(\w+)/g>:
1664 while ($x =~ /(\w+)/g) {
1665 print "Word is $1, ends at position ", pos $x, "\n";
1670 Word is cat, ends at position 3
1671 Word is dog, ends at position 7
1672 Word is house, ends at position 13
1674 A failed match or changing the target string resets the position. If
1675 you don't want the position reset after failure to match, add the
1676 C</c>, as in C</regexp/gc>. The current position in the string is
1677 associated with the string, not the regexp. This means that different
1678 strings have different positions and their respective positions can be
1679 set or read independently.
1681 In list context, C</g> returns a list of matched groupings, or if
1682 there are no groupings, a list of matches to the whole regexp. So if
1683 we wanted just the words, we could use
1685 @words = ($x =~ /(\w+)/g); # matches,
1688 # $words[2] = 'house'
1690 Closely associated with the C</g> modifier is the C<\G> anchor. The
1691 C<\G> anchor matches at the point where the previous C</g> match left
1692 off. C<\G> allows us to easily do context-sensitive matching:
1694 $metric = 1; # use metric units
1696 $x = <FILE>; # read in measurement
1697 $x =~ /^([+-]?\d+)\s*/g; # get magnitude
1699 if ($metric) { # error checking
1700 print "Units error!" unless $x =~ /\Gkg\./g;
1703 print "Units error!" unless $x =~ /\Glbs\./g;
1705 $x =~ /\G\s+(widget|sprocket)/g; # continue processing
1707 The combination of C</g> and C<\G> allows us to process the string a
1708 bit at a time and use arbitrary Perl logic to decide what to do next.
1709 Currently, the C<\G> anchor is only fully supported when used to anchor
1710 to the start of the pattern.
1712 C<\G> is also invaluable in processing fixed-length records with
1713 regexps. Suppose we have a snippet of coding region DNA, encoded as
1714 base pair letters C<ATCGTTGAAT...> and we want to find all the stop
1715 codons C<TGA>. In a coding region, codons are 3-letter sequences, so
1716 we can think of the DNA snippet as a sequence of 3-letter records. The
1719 # expanded, this is "ATC GTT GAA TGC AAA TGA CAT GAC"
1720 $dna = "ATCGTTGAATGCAAATGACATGAC";
1723 doesn't work; it may match a C<TGA>, but there is no guarantee that
1724 the match is aligned with codon boundaries, I<e.g.>, the substring
1725 S<C<GTT GAA>> gives a match. A better solution is
1727 while ($dna =~ /(\w\w\w)*?TGA/g) { # note the minimal *?
1728 print "Got a TGA stop codon at position ", pos $dna, "\n";
1733 Got a TGA stop codon at position 18
1734 Got a TGA stop codon at position 23
1736 Position 18 is good, but position 23 is bogus. What happened?
1738 The answer is that our regexp works well until we get past the last
1739 real match. Then the regexp will fail to match a synchronized C<TGA>
1740 and start stepping ahead one character position at a time, not what we
1741 want. The solution is to use C<\G> to anchor the match to the codon
1744 while ($dna =~ /\G(\w\w\w)*?TGA/g) {
1745 print "Got a TGA stop codon at position ", pos $dna, "\n";
1750 Got a TGA stop codon at position 18
1752 which is the correct answer. This example illustrates that it is
1753 important not only to match what is desired, but to reject what is not
1756 (There are other regexp modifiers that are available, such as
1757 C</o>, but their specialized uses are beyond the
1758 scope of this introduction. )
1760 =head3 Search and replace
1762 Regular expressions also play a big role in I<search and replace>
1763 operations in Perl. Search and replace is accomplished with the
1764 C<s///> operator. The general form is
1765 C<s/regexp/replacement/modifiers>, with everything we know about
1766 regexps and modifiers applying in this case as well. The
1767 I<replacement> is a Perl double-quoted string that replaces in the
1768 string whatever is matched with the C<regexp>. The operator C<=~> is
1769 also used here to associate a string with C<s///>. If matching
1770 against C<$_>, the S<C<$_ =~>> can be dropped. If there is a match,
1771 C<s///> returns the number of substitutions made; otherwise it returns
1772 false. Here are a few examples:
1774 $x = "Time to feed the cat!";
1775 $x =~ s/cat/hacker/; # $x contains "Time to feed the hacker!"
1776 if ($x =~ s/^(Time.*hacker)!$/$1 now!/) {
1777 $more_insistent = 1;
1779 $y = "'quoted words'";
1780 $y =~ s/^'(.*)'$/$1/; # strip single quotes,
1781 # $y contains "quoted words"
1783 In the last example, the whole string was matched, but only the part
1784 inside the single quotes was grouped. With the C<s///> operator, the
1785 matched variables C<$1>, C<$2>, I<etc>. are immediately available for use
1786 in the replacement expression, so we use C<$1> to replace the quoted
1787 string with just what was quoted. With the global modifier, C<s///g>
1788 will search and replace all occurrences of the regexp in the string:
1790 $x = "I batted 4 for 4";
1791 $x =~ s/4/four/; # doesn't do it all:
1792 # $x contains "I batted four for 4"
1793 $x = "I batted 4 for 4";
1794 $x =~ s/4/four/g; # does it all:
1795 # $x contains "I batted four for four"
1797 If you prefer "regex" over "regexp" in this tutorial, you could use
1798 the following program to replace it:
1800 % cat > simple_replace
1803 $replacement = shift;
1805 s/$regexp/$replacement/g;
1810 % simple_replace regexp regex perlretut.pod
1812 In C<simple_replace> we used the C<s///g> modifier to replace all
1813 occurrences of the regexp on each line. (Even though the regular
1814 expression appears in a loop, Perl is smart enough to compile it
1815 only once.) As with C<simple_grep>, both the
1816 C<print> and the C<s/$regexp/$replacement/g> use C<$_> implicitly.
1818 If you don't want C<s///> to change your original variable you can use
1819 the non-destructive substitute modifier, C<s///r>. This changes the
1820 behavior so that C<s///r> returns the final substituted string
1821 (instead of the number of substitutions):
1823 $x = "I like dogs.";
1824 $y = $x =~ s/dogs/cats/r;
1827 That example will print "I like dogs. I like cats". Notice the original
1828 C<$x> variable has not been affected. The overall
1829 result of the substitution is instead stored in C<$y>. If the
1830 substitution doesn't affect anything then the original string is
1833 $x = "I like dogs.";
1834 $y = $x =~ s/elephants/cougars/r;
1835 print "$x $y\n"; # prints "I like dogs. I like dogs."
1837 One other interesting thing that the C<s///r> flag allows is chaining
1840 $x = "Cats are great.";
1841 print $x =~ s/Cats/Dogs/r =~ s/Dogs/Frogs/r =~
1842 s/Frogs/Hedgehogs/r, "\n";
1843 # prints "Hedgehogs are great."
1845 A modifier available specifically to search and replace is the
1846 C<s///e> evaluation modifier. C<s///e> treats the
1847 replacement text as Perl code, rather than a double-quoted
1848 string. The value that the code returns is substituted for the
1849 matched substring. C<s///e> is useful if you need to do a bit of
1850 computation in the process of replacing text. This example counts
1851 character frequencies in a line:
1853 $x = "Bill the cat";
1854 $x =~ s/(.)/$chars{$1}++;$1/eg; # final $1 replaces char with itself
1855 print "frequency of '$_' is $chars{$_}\n"
1856 foreach (sort {$chars{$b} <=> $chars{$a}} keys %chars);
1860 frequency of ' ' is 2
1861 frequency of 't' is 2
1862 frequency of 'l' is 2
1863 frequency of 'B' is 1
1864 frequency of 'c' is 1
1865 frequency of 'e' is 1
1866 frequency of 'h' is 1
1867 frequency of 'i' is 1
1868 frequency of 'a' is 1
1870 As with the match C<m//> operator, C<s///> can use other delimiters,
1871 such as C<s!!!> and C<s{}{}>, and even C<s{}//>. If single quotes are
1872 used C<s'''>, then the regexp and replacement are
1873 treated as single-quoted strings and there are no
1874 variable substitutions. C<s///> in list context
1875 returns the same thing as in scalar context, I<i.e.>, the number of
1878 =head3 The split function
1880 The C<split()> function is another place where a regexp is used.
1881 C<split /regexp/, string, limit> separates the C<string> operand into
1882 a list of substrings and returns that list. The regexp must be designed
1883 to match whatever constitutes the separators for the desired substrings.
1884 The C<limit>, if present, constrains splitting into no more than C<limit>
1885 number of strings. For example, to split a string into words, use
1887 $x = "Calvin and Hobbes";
1888 @words = split /\s+/, $x; # $word[0] = 'Calvin'
1890 # $word[2] = 'Hobbes'
1892 If the empty regexp C<//> is used, the regexp always matches and
1893 the string is split into individual characters. If the regexp has
1894 groupings, then the resulting list contains the matched substrings from the
1895 groupings as well. For instance,
1897 $x = "/usr/bin/perl";
1898 @dirs = split m!/!, $x; # $dirs[0] = ''
1902 @parts = split m!(/)!, $x; # $parts[0] = ''
1908 # $parts[6] = 'perl'
1910 Since the first character of C<$x> matched the regexp, C<split> prepended
1911 an empty initial element to the list.
1913 If you have read this far, congratulations! You now have all the basic
1914 tools needed to use regular expressions to solve a wide range of text
1915 processing problems. If this is your first time through the tutorial,
1916 why not stop here and play around with regexps a while.... S<Part 2>
1917 concerns the more esoteric aspects of regular expressions and those
1918 concepts certainly aren't needed right at the start.
1920 =head1 Part 2: Power tools
1922 OK, you know the basics of regexps and you want to know more. If
1923 matching regular expressions is analogous to a walk in the woods, then
1924 the tools discussed in Part 1 are analogous to topo maps and a
1925 compass, basic tools we use all the time. Most of the tools in part 2
1926 are analogous to flare guns and satellite phones. They aren't used
1927 too often on a hike, but when we are stuck, they can be invaluable.
1929 What follows are the more advanced, less used, or sometimes esoteric
1930 capabilities of Perl regexps. In Part 2, we will assume you are
1931 comfortable with the basics and concentrate on the advanced features.
1933 =head2 More on characters, strings, and character classes
1935 There are a number of escape sequences and character classes that we
1936 haven't covered yet.
1938 There are several escape sequences that convert characters or strings
1939 between upper and lower case, and they are also available within
1940 patterns. C<\l> and C<\u> convert the next character to lower or
1941 upper case, respectively:
1944 $string =~ /\u$x/; # matches 'Perl' in $string
1945 $x = "M(rs?|s)\\."; # note the double backslash
1946 $string =~ /\l$x/; # matches 'mr.', 'mrs.', and 'ms.',
1948 A C<\L> or C<\U> indicates a lasting conversion of case, until
1949 terminated by C<\E> or thrown over by another C<\U> or C<\L>:
1951 $x = "This word is in lower case:\L SHOUT\E";
1952 $x =~ /shout/; # matches
1953 $x = "I STILL KEYPUNCH CARDS FOR MY 360";
1954 $x =~ /\Ukeypunch/; # matches punch card string
1956 If there is no C<\E>, case is converted until the end of the
1957 string. The regexps C<\L\u$word> or C<\u\L$word> convert the first
1958 character of C<$word> to uppercase and the rest of the characters to
1961 Control characters can be escaped with C<\c>, so that a control-Z
1962 character would be matched with C<\cZ>. The escape sequence
1963 C<\Q>...C<\E> quotes, or protects most non-alphabetic characters. For
1966 $x = "\QThat !^*&%~& cat!";
1967 $x =~ /\Q!^*&%~&\E/; # check for rough language
1969 It does not protect C<'$'> or C<'@'>, so that variables can still be
1972 C<\Q>, C<\L>, C<\l>, C<\U>, C<\u> and C<\E> are actually part of
1973 double-quotish syntax, and not part of regexp syntax proper. They will
1974 work if they appear in a regular expression embedded directly in a
1975 program, but not when contained in a string that is interpolated in a
1978 Perl regexps can handle more than just the
1979 standard ASCII character set. Perl supports I<Unicode>, a standard
1980 for representing the alphabets from virtually all of the world's written
1981 languages, and a host of symbols. Perl's text strings are Unicode strings, so
1982 they can contain characters with a value (codepoint or character number) higher
1985 What does this mean for regexps? Well, regexp users don't need to know
1986 much about Perl's internal representation of strings. But they do need
1987 to know 1) how to represent Unicode characters in a regexp and 2) that
1988 a matching operation will treat the string to be searched as a sequence
1989 of characters, not bytes. The answer to 1) is that Unicode characters
1990 greater than C<chr(255)> are represented using the C<\x{hex}> notation, because
1991 C<\x>I<XY> (without curly braces and I<XY> are two hex digits) doesn't
1992 go further than 255. (Starting in Perl 5.14, if you're an octal fan,
1993 you can also use C<\o{oct}>.)
1995 /\x{263a}/; # match a Unicode smiley face :)
1996 /\x{ 263a }/; # Same
1998 B<NOTE>: In Perl 5.6.0 it used to be that one needed to say C<use
1999 utf8> to use any Unicode features. This is no longer the case: for
2000 almost all Unicode processing, the explicit C<utf8> pragma is not
2001 needed. (The only case where it matters is if your Perl script is in
2002 Unicode and encoded in UTF-8, then an explicit C<use utf8> is needed.)
2004 Figuring out the hexadecimal sequence of a Unicode character you want
2005 or deciphering someone else's hexadecimal Unicode regexp is about as
2006 much fun as programming in machine code. So another way to specify
2007 Unicode characters is to use the I<named character> escape
2008 sequence C<\N{I<name>}>. I<name> is a name for the Unicode character, as
2009 specified in the Unicode standard. For instance, if we wanted to
2010 represent or match the astrological sign for the planet Mercury, we
2013 $x = "abc\N{MERCURY}def";
2014 $x =~ /\N{MERCURY}/; # matches
2015 $x =~ /\N{ MERCURY }/; # Also matches
2017 One can also use "short" names:
2019 print "\N{GREEK SMALL LETTER SIGMA} is called sigma.\n";
2020 print "\N{greek:Sigma} is an upper-case sigma.\n";
2022 You can also restrict names to a certain alphabet by specifying the
2023 L<charnames> pragma:
2025 use charnames qw(greek);
2026 print "\N{sigma} is Greek sigma\n";
2028 An index of character names is available on-line from the Unicode
2029 Consortium, L<https://www.unicode.org/charts/charindex.html>; explanatory
2030 material with links to other resources at
2031 L<https://www.unicode.org/standard/where>.
2033 Starting in Perl v5.32, an alternative to C<\N{...}> for full names is
2034 available, and that is to say
2036 /\p{Name=greek small letter sigma}/
2038 The casing of the character name is irrelevant when used in C<\p{}>, as
2039 are most spaces, underscores and hyphens. (A few outlier characters
2040 cause problems with ignoring all of them always. The details (which you
2041 can look up when you get more proficient, and if ever needed) are in
2042 L<https://www.unicode.org/reports/tr44/tr44-24.html#UAX44-LM2>).
2044 The answer to requirement 2) is that a regexp (mostly)
2045 uses Unicode characters. The "mostly" is for messy backward
2046 compatibility reasons, but starting in Perl 5.14, any regexp compiled in
2047 the scope of a C<use feature 'unicode_strings'> (which is automatically
2048 turned on within the scope of a C<use 5.012> or higher) will turn that
2049 "mostly" into "always". If you want to handle Unicode properly, you
2050 should ensure that C<'unicode_strings'> is turned on.
2051 Internally, this is encoded to bytes using either UTF-8 or a native 8
2052 bit encoding, depending on the history of the string, but conceptually
2053 it is a sequence of characters, not bytes. See L<perlunitut> for a
2054 tutorial about that.
2056 Let us now discuss Unicode character classes, most usually called
2057 "character properties". These are represented by the C<\p{I<name>}>
2058 escape sequence. The negation of this is C<\P{I<name>}>. For example,
2059 to match lower and uppercase characters,
2062 $x =~ /^\p{IsUpper}/; # matches, uppercase char class
2063 $x =~ /^\P{IsUpper}/; # doesn't match, char class sans uppercase
2064 $x =~ /^\p{IsLower}/; # doesn't match, lowercase char class
2065 $x =~ /^\P{IsLower}/; # matches, char class sans lowercase
2067 (The "C<Is>" is optional.)
2069 There are many, many Unicode character properties. For the full list
2070 see L<perluniprops>. Most of them have synonyms with shorter names,
2071 also listed there. Some synonyms are a single character. For these,
2072 you can drop the braces. For instance, C<\pM> is the same thing as
2073 C<\p{Mark}>, meaning things like accent marks.
2075 The Unicode C<\p{Script}> and C<\p{Script_Extensions}> properties are
2076 used to categorize every Unicode character into the language script it
2077 is written in. For example,
2078 English, French, and a bunch of other European languages are written in
2079 the Latin script. But there is also the Greek script, the Thai script,
2080 the Katakana script, I<etc>. (C<Script> is an older, less advanced,
2081 form of C<Script_Extensions>, retained only for backwards
2082 compatibility.) You can test whether a character is in a particular
2083 script with, for example C<\p{Latin}>, C<\p{Greek}>, or
2084 C<\p{Katakana}>. To test if it isn't in the Balinese script, you would
2085 use C<\P{Balinese}>. (These all use C<Script_Extensions> under the
2086 hood, as that gives better results.)
2088 What we have described so far is the single form of the C<\p{...}> character
2089 classes. There is also a compound form which you may run into. These
2090 look like C<\p{I<name>=I<value>}> or C<\p{I<name>:I<value>}> (the equals sign and colon
2091 can be used interchangeably). These are more general than the single form,
2092 and in fact most of the single forms are just Perl-defined shortcuts for common
2093 compound forms. For example, the script examples in the previous paragraph
2094 could be written equivalently as C<\p{Script_Extensions=Latin}>, C<\p{Script_Extensions:Greek}>,
2095 C<\p{script_extensions=katakana}>, and C<\P{script_extensions=balinese}> (case is irrelevant
2096 between the C<{}> braces). You may
2097 never have to use the compound forms, but sometimes it is necessary, and their
2098 use can make your code easier to understand.
2100 C<\X> is an abbreviation for a character class that comprises
2101 a Unicode I<extended grapheme cluster>. This represents a "logical character":
2102 what appears to be a single character, but may be represented internally by more
2103 than one. As an example, using the Unicode full names, I<e.g.>, "S<A + COMBINING
2104 RING>" is a grapheme cluster with base character "A" and combining character
2105 "S<COMBINING RING>, which translates in Danish to "A" with the circle atop it,
2106 as in the word E<Aring>ngstrom.
2108 For the full and latest information about Unicode see the latest
2109 Unicode standard, or the Unicode Consortium's website L<https://www.unicode.org>
2111 As if all those classes weren't enough, Perl also defines POSIX-style
2112 character classes. These have the form C<[:I<name>:]>, with I<name> the
2113 name of the POSIX class. The POSIX classes are C<alpha>, C<alnum>,
2114 C<ascii>, C<cntrl>, C<digit>, C<graph>, C<lower>, C<print>, C<punct>,
2115 C<space>, C<upper>, and C<xdigit>, and two extensions, C<word> (a Perl
2116 extension to match C<\w>), and C<blank> (a GNU extension). The C</a>
2117 modifier restricts these to matching just in the ASCII range; otherwise
2118 they can match the same as their corresponding Perl Unicode classes:
2119 C<[:upper:]> is the same as C<\p{IsUpper}>, I<etc>. (There are some
2120 exceptions and gotchas with this; see L<perlrecharclass> for a full
2121 discussion.) The C<[:digit:]>, C<[:word:]>, and
2122 C<[:space:]> correspond to the familiar C<\d>, C<\w>, and C<\s>
2123 character classes. To negate a POSIX class, put a C<'^'> in front of
2124 the name, so that, I<e.g.>, C<[:^digit:]> corresponds to C<\D> and, under
2125 Unicode, C<\P{IsDigit}>. The Unicode and POSIX character classes can
2126 be used just like C<\d>, with the exception that POSIX character
2127 classes can only be used inside of a character class:
2129 /\s+[abc[:digit:]xyz]\s*/; # match a,b,c,x,y,z, or a digit
2130 /^=item\s[[:digit:]]/; # match '=item',
2131 # followed by a space and a digit
2132 /\s+[abc\p{IsDigit}xyz]\s+/; # match a,b,c,x,y,z, or a digit
2133 /^=item\s\p{IsDigit}/; # match '=item',
2134 # followed by a space and a digit
2136 Whew! That is all the rest of the characters and character classes.
2138 =head2 Compiling and saving regular expressions
2140 In Part 1 we mentioned that Perl compiles a regexp into a compact
2141 sequence of opcodes. Thus, a compiled regexp is a data structure
2142 that can be stored once and used again and again. The regexp quote
2143 C<qr//> does exactly that: C<qr/string/> compiles the C<string> as a
2144 regexp and transforms the result into a form that can be assigned to a
2147 $reg = qr/foo+bar?/; # reg contains a compiled regexp
2149 Then C<$reg> can be used as a regexp:
2152 $x =~ $reg; # matches, just like /foo+bar?/
2153 $x =~ /$reg/; # same thing, alternate form
2155 C<$reg> can also be interpolated into a larger regexp:
2157 $x =~ /(abc)?$reg/; # still matches
2159 As with the matching operator, the regexp quote can use different
2160 delimiters, I<e.g.>, C<qr!!>, C<qr{}> or C<qr~~>. Apostrophes
2161 as delimiters (C<qr''>) inhibit any interpolation.
2163 Pre-compiled regexps are useful for creating dynamic matches that
2164 don't need to be recompiled each time they are encountered. Using
2165 pre-compiled regexps, we write a C<grep_step> program which greps
2166 for a sequence of patterns, advancing to the next pattern as soon
2167 as one has been satisfied.
2171 # grep_step - match <number> regexps, one after the other
2172 # usage: multi_grep <number> regexp1 regexp2 ... file1 file2 ...
2175 $regexp[$_] = shift foreach (0..$number-1);
2176 @compiled = map qr/$_/, @regexp;
2177 while ($line = <>) {
2178 if ($line =~ /$compiled[0]/) {
2181 last unless @compiled;
2186 % grep_step 3 shift print last grep_step
2189 last unless @compiled;
2191 Storing pre-compiled regexps in an array C<@compiled> allows us to
2192 simply loop through the regexps without any recompilation, thus gaining
2193 flexibility without sacrificing speed.
2196 =head2 Composing regular expressions at runtime
2198 Backtracking is more efficient than repeated tries with different regular
2199 expressions. If there are several regular expressions and a match with
2200 any of them is acceptable, then it is possible to combine them into a set
2201 of alternatives. If the individual expressions are input data, this
2202 can be done by programming a join operation. We'll exploit this idea in
2203 an improved version of the C<simple_grep> program: a program that matches
2208 # multi_grep - match any of <number> regexps
2209 # usage: multi_grep <number> regexp1 regexp2 ... file1 file2 ...
2212 $regexp[$_] = shift foreach (0..$number-1);
2213 $pattern = join '|', @regexp;
2215 while ($line = <>) {
2216 print $line if $line =~ /$pattern/;
2220 % multi_grep 2 shift for multi_grep
2222 $regexp[$_] = shift foreach (0..$number-1);
2224 Sometimes it is advantageous to construct a pattern from the I<input>
2225 that is to be analyzed and use the permissible values on the left
2226 hand side of the matching operations. As an example for this somewhat
2227 paradoxical situation, let's assume that our input contains a command
2228 verb which should match one out of a set of available command verbs,
2229 with the additional twist that commands may be abbreviated as long as
2230 the given string is unique. The program below demonstrates the basic
2235 $kwds = 'copy compare list print';
2237 $cmd =~ s/^\s+|\s+$//g; # trim leading and trailing spaces
2238 if( ( @matches = $kwds =~ /\b$cmd\w*/g ) == 1 ){
2239 print "command: '@matches'\n";
2240 } elsif( @matches == 0 ){
2241 print "no such command: '$cmd'\n";
2243 print "not unique: '$cmd' (could be one of: @matches)\n";
2252 not unique: 'co' (could be one of: copy compare)
2254 no such command: 'printer'
2256 Rather than trying to match the input against the keywords, we match the
2257 combined set of keywords against the input. The pattern matching
2258 operation S<C<$kwds =~ /\b($cmd\w*)/g>> does several things at the
2259 same time. It makes sure that the given command begins where a keyword
2260 begins (C<\b>). It tolerates abbreviations due to the added C<\w*>. It
2261 tells us the number of matches (C<scalar @matches>) and all the keywords
2262 that were actually matched. You could hardly ask for more.
2264 =head2 Embedding comments and modifiers in a regular expression
2266 Starting with this section, we will be discussing Perl's set of
2267 I<extended patterns>. These are extensions to the traditional regular
2268 expression syntax that provide powerful new tools for pattern
2269 matching. We have already seen extensions in the form of the minimal
2270 matching constructs C<??>, C<*?>, C<+?>, C<{n,m}?>, C<{n,}?>, and
2271 C<{,n}?>. Most of the extensions below have the form C<(?char...)>,
2272 where the C<char> is a character that determines the type of extension.
2274 The first extension is an embedded comment C<(?#text)>. This embeds a
2275 comment into the regular expression without affecting its meaning. The
2276 comment should not have any closing parentheses in the text. An
2279 /(?# Match an integer:)[+-]?\d+/;
2281 This style of commenting has been largely superseded by the raw,
2282 freeform commenting that is allowed with the C</x> modifier.
2284 Most modifiers, such as C</i>, C</m>, C</s> and C</x> (or any
2285 combination thereof) can also be embedded in
2286 a regexp using C<(?i)>, C<(?m)>, C<(?s)>, and C<(?x)>. For instance,
2288 /(?i)yes/; # match 'yes' case insensitively
2289 /yes/i; # same thing
2290 /(?x)( # freeform version of an integer regexp
2291 [+-]? # match an optional sign
2292 \d+ # match a sequence of digits
2296 Embedded modifiers can have two important advantages over the usual
2297 modifiers. Embedded modifiers allow a custom set of modifiers for
2298 I<each> regexp pattern. This is great for matching an array of regexps
2299 that must have different modifiers:
2301 $pattern[0] = '(?i)doctor';
2302 $pattern[1] = 'Johnson';
2305 foreach $patt (@pattern) {
2310 The second advantage is that embedded modifiers (except C</p>, which
2311 modifies the entire regexp) only affect the regexp
2312 inside the group the embedded modifier is contained in. So grouping
2313 can be used to localize the modifier's effects:
2315 /Answer: ((?i)yes)/; # matches 'Answer: yes', 'Answer: YES', etc.
2317 Embedded modifiers can also turn off any modifiers already present
2318 by using, I<e.g.>, C<(?-i)>. Modifiers can also be combined into
2319 a single expression, I<e.g.>, C<(?s-i)> turns on single line mode and
2320 turns off case insensitivity.
2322 Embedded modifiers may also be added to a non-capturing grouping.
2323 C<(?i-m:regexp)> is a non-capturing grouping that matches C<regexp>
2324 case insensitively and turns off multi-line mode.
2327 =head2 Looking ahead and looking behind
2329 This section concerns the lookahead and lookbehind assertions. First,
2330 a little background.
2332 In Perl regular expressions, most regexp elements "eat up" a certain
2333 amount of string when they match. For instance, the regexp element
2334 C<[abc]> eats up one character of the string when it matches, in the
2335 sense that Perl moves to the next character position in the string
2336 after the match. There are some elements, however, that don't eat up
2337 characters (advance the character position) if they match. The examples
2338 we have seen so far are the anchors. The anchor C<'^'> matches the
2339 beginning of the line, but doesn't eat any characters. Similarly, the
2340 word boundary anchor C<\b> matches wherever a character matching C<\w>
2341 is next to a character that doesn't, but it doesn't eat up any
2342 characters itself. Anchors are examples of I<zero-width assertions>:
2343 zero-width, because they consume
2344 no characters, and assertions, because they test some property of the
2345 string. In the context of our walk in the woods analogy to regexp
2346 matching, most regexp elements move us along a trail, but anchors have
2347 us stop a moment and check our surroundings. If the local environment
2348 checks out, we can proceed forward. But if the local environment
2349 doesn't satisfy us, we must backtrack.
2351 Checking the environment entails either looking ahead on the trail,
2352 looking behind, or both. C<'^'> looks behind, to see that there are no
2353 characters before. C<'$'> looks ahead, to see that there are no
2354 characters after. C<\b> looks both ahead and behind, to see if the
2355 characters on either side differ in their "word-ness".
2357 The lookahead and lookbehind assertions are generalizations of the
2358 anchor concept. Lookahead and lookbehind are zero-width assertions
2359 that let us specify which characters we want to test for. The
2360 lookahead assertion is denoted by C<(?=regexp)> or (starting in 5.32,
2361 experimentally in 5.28) C<(*pla:regexp)> or
2362 C<(*positive_lookahead:regexp)>; and the lookbehind assertion is denoted
2363 by C<< (?<=fixed-regexp) >> or (starting in 5.32, experimentally in
2364 5.28) C<(*plb:fixed-regexp)> or C<(*positive_lookbehind:fixed-regexp)>.
2367 $x = "I catch the housecat 'Tom-cat' with catnip";
2368 $x =~ /cat(*pla:\s)/; # matches 'cat' in 'housecat'
2369 @catwords = ($x =~ /(?<=\s)cat\w+/g); # matches,
2370 # $catwords[0] = 'catch'
2371 # $catwords[1] = 'catnip'
2372 $x =~ /\bcat\b/; # matches 'cat' in 'Tom-cat'
2373 $x =~ /(?<=\s)cat(?=\s)/; # doesn't match; no isolated 'cat' in
2376 Note that the parentheses in these are
2377 non-capturing, since these are zero-width assertions. Thus in the
2378 second regexp, the substrings captured are those of the whole regexp
2379 itself. Lookahead can match arbitrary regexps, but
2380 lookbehind prior to 5.30 C<< (?<=fixed-regexp) >> only works for regexps
2381 of fixed width, I<i.e.>, a fixed number of characters long. Thus
2382 C<< (?<=(ab|bc)) >> is fine, but C<< (?<=(ab)*) >> prior to 5.30 is not.
2384 The negated versions of the lookahead and lookbehind assertions are
2385 denoted by C<(?!regexp)> and C<< (?<!fixed-regexp) >> respectively.
2386 Or, starting in 5.32 (experimentally in 5.28), C<(*nla:regexp)>,
2387 C<(*negative_lookahead:regexp)>, C<(*nlb:regexp)>, or
2388 C<(*negative_lookbehind:regexp)>.
2389 They evaluate true if the regexps do I<not> match:
2392 $x =~ /foo(?!bar)/; # doesn't match, 'bar' follows 'foo'
2393 $x =~ /foo(?!baz)/; # matches, 'baz' doesn't follow 'foo'
2394 $x =~ /(?<!\s)foo/; # matches, there is no \s before 'foo'
2396 Here is an example where a string containing blank-separated words,
2397 numbers and single dashes is to be split into its components.
2398 Using C</\s+/> alone won't work, because spaces are not required between
2399 dashes, or a word or a dash. Additional places for a split are established
2400 by looking ahead and behind:
2402 $str = "one two - --6-8";
2403 @toks = split / \s+ # a run of spaces
2404 | (?<=\S) (?=-) # any non-space followed by '-'
2405 | (?<=-) (?=\S) # a '-' followed by any non-space
2406 /x, $str; # @toks = qw(one two - - - 6 - 8)
2408 =head2 Using independent subexpressions to prevent backtracking
2410 I<Independent subexpressions> (or atomic subexpressions) are regular
2411 expressions, in the context of a larger regular expression, that
2412 function independently of the larger regular expression. That is, they
2413 consume as much or as little of the string as they wish without regard
2414 for the ability of the larger regexp to match. Independent
2415 subexpressions are represented by
2416 C<< (?>regexp) >> or (starting in 5.32, experimentally in 5.28)
2417 C<(*atomic:regexp)>. We can illustrate their behavior by first
2418 considering an ordinary regexp:
2421 $x =~ /a*ab/; # matches
2423 This obviously matches, but in the process of matching, the
2424 subexpression C<a*> first grabbed the C<'a'>. Doing so, however,
2425 wouldn't allow the whole regexp to match, so after backtracking, C<a*>
2426 eventually gave back the C<'a'> and matched the empty string. Here, what
2427 C<a*> matched was I<dependent> on what the rest of the regexp matched.
2429 Contrast that with an independent subexpression:
2431 $x =~ /(?>a*)ab/; # doesn't match!
2433 The independent subexpression C<< (?>a*) >> doesn't care about the rest
2434 of the regexp, so it sees an C<'a'> and grabs it. Then the rest of the
2435 regexp C<ab> cannot match. Because C<< (?>a*) >> is independent, there
2436 is no backtracking and the independent subexpression does not give
2437 up its C<'a'>. Thus the match of the regexp as a whole fails. A similar
2438 behavior occurs with completely independent regexps:
2441 $x =~ /a*/g; # matches, eats an 'a'
2442 $x =~ /\Gab/g; # doesn't match, no 'a' available
2444 Here C</g> and C<\G> create a "tag team" handoff of the string from
2445 one regexp to the other. Regexps with an independent subexpression are
2446 much like this, with a handoff of the string to the independent
2447 subexpression, and a handoff of the string back to the enclosing
2450 The ability of an independent subexpression to prevent backtracking
2451 can be quite useful. Suppose we want to match a non-empty string
2452 enclosed in parentheses up to two levels deep. Then the following
2455 $x = "abc(de(fg)h"; # unbalanced parentheses
2456 $x =~ /\( ( [ ^ () ]+ | \( [ ^ () ]* \) )+ \)/xx;
2458 The regexp matches an open parenthesis, one or more copies of an
2459 alternation, and a close parenthesis. The alternation is two-way, with
2460 the first alternative C<[^()]+> matching a substring with no
2461 parentheses and the second alternative C<\([^()]*\)> matching a
2462 substring delimited by parentheses. The problem with this regexp is
2463 that it is pathological: it has nested indeterminate quantifiers
2464 of the form C<(a+|b)+>. We discussed in Part 1 how nested quantifiers
2465 like this could take an exponentially long time to execute if no match
2466 were possible. To prevent the exponential blowup, we need to prevent
2467 useless backtracking at some point. This can be done by enclosing the
2468 inner quantifier as an independent subexpression:
2470 $x =~ /\( ( (?> [ ^ () ]+ ) | \([ ^ () ]* \) )+ \)/xx;
2472 Here, C<< (?>[^()]+) >> breaks the degeneracy of string partitioning
2473 by gobbling up as much of the string as possible and keeping it. Then
2474 match failures fail much more quickly.
2477 =head2 Conditional expressions
2479 A I<conditional expression> is a form of if-then-else statement
2480 that allows one to choose which patterns are to be matched, based on
2481 some condition. There are two types of conditional expression:
2482 C<(?(I<condition>)I<yes-regexp>)> and
2483 C<(?(condition)I<yes-regexp>|I<no-regexp>)>.
2484 C<(?(I<condition>)I<yes-regexp>)> is
2485 like an S<C<'if () {}'>> statement in Perl. If the I<condition> is true,
2486 the I<yes-regexp> will be matched. If the I<condition> is false, the
2487 I<yes-regexp> will be skipped and Perl will move onto the next regexp
2488 element. The second form is like an S<C<'if () {} else {}'>> statement
2489 in Perl. If the I<condition> is true, the I<yes-regexp> will be
2490 matched, otherwise the I<no-regexp> will be matched.
2492 The I<condition> can have several forms. The first form is simply an
2493 integer in parentheses C<(I<integer>)>. It is true if the corresponding
2494 backreference C<\I<integer>> matched earlier in the regexp. The same
2495 thing can be done with a name associated with a capture group, written
2496 as C<<< (E<lt>I<name>E<gt>) >>> or C<< ('I<name>') >>. The second form is a bare
2497 zero-width assertion C<(?...)>, either a lookahead, a lookbehind, or a
2498 code assertion (discussed in the next section). The third set of forms
2499 provides tests that return true if the expression is executed within
2500 a recursion (C<(R)>) or is being called from some capturing group,
2501 referenced either by number (C<(R1)>, C<(R2)>,...) or by name
2504 The integer or name form of the C<condition> allows us to choose,
2505 with more flexibility, what to match based on what matched earlier in the
2506 regexp. This searches for words of the form C<"$x$x"> or C<"$x$y$y$x">:
2508 % simple_grep '^(\w+)(\w+)?(?(2)\g2\g1|\g1)$' /usr/dict/words
2518 The lookbehind C<condition> allows, along with backreferences,
2519 an earlier part of the match to influence a later part of the
2520 match. For instance,
2522 /[ATGC]+(?(?<=AA)G|C)$/;
2524 matches a DNA sequence such that it either ends in C<AAG>, or some
2525 other base pair combination and C<'C'>. Note that the form is
2526 C<< (?(?<=AA)G|C) >> and not C<< (?((?<=AA))G|C) >>; for the
2527 lookahead, lookbehind or code assertions, the parentheses around the
2528 conditional are not needed.
2531 =head2 Defining named patterns
2533 Some regular expressions use identical subpatterns in several places.
2534 Starting with Perl 5.10, it is possible to define named subpatterns in
2535 a section of the pattern so that they can be called up by name
2536 anywhere in the pattern. This syntactic pattern for this definition
2537 group is C<< (?(DEFINE)(?<I<name>>I<pattern>)...) >>. An insertion
2538 of a named pattern is written as C<(?&I<name>)>.
2540 The example below illustrates this feature using the pattern for
2541 floating point numbers that was presented earlier on. The three
2542 subpatterns that are used more than once are the optional sign, the
2543 digit sequence for an integer and the decimal fraction. The C<DEFINE>
2544 group at the end of the pattern contains their definition. Notice
2545 that the decimal fraction pattern is the first place where we can
2546 reuse the integer pattern.
2548 /^ (?&osg)\ * ( (?&int)(?&dec)? | (?&dec) )
2549 (?: [eE](?&osg)(?&int) )?
2552 (?<osg>[-+]?) # optional sign
2553 (?<int>\d++) # integer
2554 (?<dec>\.(?&int)) # decimal fraction
2558 =head2 Recursive patterns
2560 This feature (introduced in Perl 5.10) significantly extends the
2561 power of Perl's pattern matching. By referring to some other
2562 capture group anywhere in the pattern with the construct
2563 C<(?I<group-ref>)>, the I<pattern> within the referenced group is used
2564 as an independent subpattern in place of the group reference itself.
2565 Because the group reference may be contained I<within> the group it
2566 refers to, it is now possible to apply pattern matching to tasks that
2567 hitherto required a recursive parser.
2569 To illustrate this feature, we'll design a pattern that matches if
2570 a string contains a palindrome. (This is a word or a sentence that,
2571 while ignoring spaces, interpunctuation and case, reads the same backwards
2572 as forwards. We begin by observing that the empty string or a string
2573 containing just one word character is a palindrome. Otherwise it must
2574 have a word character up front and the same at its end, with another
2575 palindrome in between.
2577 /(?: (\w) (?...Here be a palindrome...) \g{ -1 } | \w? )/x
2579 Adding C<\W*> at either end to eliminate what is to be ignored, we already
2580 have the full pattern:
2582 my $pp = qr/^(\W* (?: (\w) (?1) \g{-1} | \w? ) \W*)$/ix;
2583 for $s ( "saippuakauppias", "A man, a plan, a canal: Panama!" ){
2584 print "'$s' is a palindrome\n" if $s =~ /$pp/;
2587 In C<(?...)> both absolute and relative backreferences may be used.
2588 The entire pattern can be reinserted with C<(?R)> or C<(?0)>.
2589 If you prefer to name your groups, you can use C<(?&I<name>)> to
2590 recurse into that group.
2593 =head2 A bit of magic: executing Perl code in a regular expression
2595 Normally, regexps are a part of Perl expressions.
2596 I<Code evaluation> expressions turn that around by allowing
2597 arbitrary Perl code to be a part of a regexp. A code evaluation
2598 expression is denoted C<(?{I<code>})>, with I<code> a string of Perl
2601 Code expressions are zero-width assertions, and the value they return
2602 depends on their environment. There are two possibilities: either the
2603 code expression is used as a conditional in a conditional expression
2604 C<(?(I<condition>)...)>, or it is not. If the code expression is a
2605 conditional, the code is evaluated and the result (I<i.e.>, the result of
2606 the last statement) is used to determine truth or falsehood. If the
2607 code expression is not used as a conditional, the assertion always
2608 evaluates true and the result is put into the special variable
2609 C<$^R>. The variable C<$^R> can then be used in code expressions later
2610 in the regexp. Here are some silly examples:
2613 $x =~ /abc(?{print "Hi Mom!";})def/; # matches,
2615 $x =~ /aaa(?{print "Hi Mom!";})def/; # doesn't match,
2618 Pay careful attention to the next example:
2620 $x =~ /abc(?{print "Hi Mom!";})ddd/; # doesn't match,
2624 At first glance, you'd think that it shouldn't print, because obviously
2625 the C<ddd> isn't going to match the target string. But look at this
2628 $x =~ /abc(?{print "Hi Mom!";})[dD]dd/; # doesn't match,
2631 Hmm. What happened here? If you've been following along, you know that
2632 the above pattern should be effectively (almost) the same as the last one;
2633 enclosing the C<'d'> in a character class isn't going to change what it
2634 matches. So why does the first not print while the second one does?
2636 The answer lies in the optimizations the regexp engine makes. In the first
2637 case, all the engine sees are plain old characters (aside from the
2638 C<?{}> construct). It's smart enough to realize that the string C<'ddd'>
2639 doesn't occur in our target string before actually running the pattern
2640 through. But in the second case, we've tricked it into thinking that our
2641 pattern is more complicated. It takes a look, sees our
2642 character class, and decides that it will have to actually run the
2643 pattern to determine whether or not it matches, and in the process of
2644 running it hits the print statement before it discovers that we don't
2647 To take a closer look at how the engine does optimizations, see the
2648 section L</"Pragmas and debugging"> below.
2650 More fun with C<?{}>:
2652 $x =~ /(?{print "Hi Mom!";})/; # matches,
2654 $x =~ /(?{$c = 1;})(?{print "$c";})/; # matches,
2656 $x =~ /(?{$c = 1;})(?{print "$^R";})/; # matches,
2659 The bit of magic mentioned in the section title occurs when the regexp
2660 backtracks in the process of searching for a match. If the regexp
2661 backtracks over a code expression and if the variables used within are
2662 localized using C<local>, the changes in the variables produced by the
2663 code expression are undone! Thus, if we wanted to count how many times
2664 a character got matched inside a group, we could use, I<e.g.>,
2667 $count = 0; # initialize 'a' count
2668 $c = "bob"; # test if $c gets clobbered
2669 $x =~ /(?{local $c = 0;}) # initialize count
2671 (?{local $c = $c + 1;}) # increment count
2672 )* # do this any number of times,
2673 aa # but match 'aa' at the end
2674 (?{$count = $c;}) # copy local $c var into $count
2676 print "'a' count is $count, \$c variable is '$c'\n";
2680 'a' count is 2, $c variable is 'bob'
2682 If we replace the S<C< (?{local $c = $c + 1;})>> with
2683 S<C< (?{$c = $c + 1;})>>, the variable changes are I<not> undone
2684 during backtracking, and we get
2686 'a' count is 4, $c variable is 'bob'
2688 Note that only localized variable changes are undone. Other side
2689 effects of code expression execution are permanent. Thus
2692 $x =~ /(a(?{print "Yow\n";}))*aa/;
2701 The result C<$^R> is automatically localized, so that it will behave
2702 properly in the presence of backtracking.
2704 This example uses a code expression in a conditional to match a
2705 definite article, either C<'the'> in English or C<'der|die|das'> in
2708 $lang = 'DE'; # use German
2713 $lang eq 'EN'; # is the language English?
2715 the | # if so, then match 'the'
2716 (der|die|das) # else, match 'der|die|das'
2720 Note that the syntax here is C<(?(?{...})I<yes-regexp>|I<no-regexp>)>, not
2721 C<(?((?{...}))I<yes-regexp>|I<no-regexp>)>. In other words, in the case of a
2722 code expression, we don't need the extra parentheses around the
2725 If you try to use code expressions where the code text is contained within
2726 an interpolated variable, rather than appearing literally in the pattern,
2727 Perl may surprise you:
2731 /foo(?{ $bar })bar/; # compiles ok, $bar not interpolated
2732 /foo(?{ 1 })$bar/; # compiles ok, $bar interpolated
2733 /foo${pat}bar/; # compile error!
2735 $pat = qr/(?{ $foo = 1 })/; # precompile code regexp
2736 /foo${pat}bar/; # compiles ok
2738 If a regexp has a variable that interpolates a code expression, Perl
2739 treats the regexp as an error. If the code expression is precompiled into
2740 a variable, however, interpolating is ok. The question is, why is this an
2743 The reason is that variable interpolation and code expressions
2744 together pose a security risk. The combination is dangerous because
2745 many programmers who write search engines often take user input and
2746 plug it directly into a regexp:
2748 $regexp = <>; # read user-supplied regexp
2749 $chomp $regexp; # get rid of possible newline
2750 $text =~ /$regexp/; # search $text for the $regexp
2752 If the C<$regexp> variable contains a code expression, the user could
2753 then execute arbitrary Perl code. For instance, some joker could
2754 search for S<C<system('rm -rf *');>> to erase your files. In this
2755 sense, the combination of interpolation and code expressions I<taints>
2756 your regexp. So by default, using both interpolation and code
2757 expressions in the same regexp is not allowed. If you're not
2758 concerned about malicious users, it is possible to bypass this
2759 security check by invoking S<C<use re 'eval'>>:
2761 use re 'eval'; # throw caution out the door
2764 /foo${pat}bar/; # compiles ok
2766 Another form of code expression is the I<pattern code expression>.
2767 The pattern code expression is like a regular code expression, except
2768 that the result of the code evaluation is treated as a regular
2769 expression and matched immediately. A simple example is
2774 $x =~ /(??{$char x $length})/x; # matches, there are 5 of 'a'
2777 This final example contains both ordinary and pattern code
2778 expressions. It detects whether a binary string C<1101010010001...> has a
2779 Fibonacci spacing 0,1,1,2,3,5,... of the C<'1'>'s:
2781 $x = "1101010010001000001";
2782 $z0 = ''; $z1 = '0'; # initial conditions
2783 print "It is a Fibonacci sequence\n"
2784 if $x =~ /^1 # match an initial '1'
2786 ((??{ $z0 })) # match some '0'
2788 (?{ $z0 = $z1; $z1 .= $^N; })
2789 )+ # repeat as needed
2790 $ # that is all there is
2792 printf "Largest sequence matched was %d\n", length($z1)-length($z0);
2794 Remember that C<$^N> is set to whatever was matched by the last
2795 completed capture group. This prints
2797 It is a Fibonacci sequence
2798 Largest sequence matched was 5
2800 Ha! Try that with your garden variety regexp package...
2802 Note that the variables C<$z0> and C<$z1> are not substituted when the
2803 regexp is compiled, as happens for ordinary variables outside a code
2804 expression. Rather, the whole code block is parsed as perl code at the
2805 same time as perl is compiling the code containing the literal regexp
2808 This regexp without the C</x> modifier is
2810 /^1(?:((??{ $z0 }))1(?{ $z0 = $z1; $z1 .= $^N; }))+$/
2812 which shows that spaces are still possible in the code parts. Nevertheless,
2813 when working with code and conditional expressions, the extended form of
2814 regexps is almost necessary in creating and debugging regexps.
2817 =head2 Backtracking control verbs
2819 Perl 5.10 introduced a number of control verbs intended to provide
2820 detailed control over the backtracking process, by directly influencing
2821 the regexp engine and by providing monitoring techniques. See
2822 L<perlre/"Special Backtracking Control Verbs"> for a detailed
2825 Below is just one example, illustrating the control verb C<(*FAIL)>,
2826 which may be abbreviated as C<(*F)>. If this is inserted in a regexp
2827 it will cause it to fail, just as it would at some
2828 mismatch between the pattern and the string. Processing
2829 of the regexp continues as it would after any "normal"
2830 failure, so that, for instance, the next position in the string or another
2831 alternative will be tried. As failing to match doesn't preserve capture
2832 groups or produce results, it may be necessary to use this in
2833 combination with embedded code.
2836 "supercalifragilisticexpialidocious" =~
2837 /([aeiou])(?{ $count{$1}++; })(*FAIL)/i;
2838 printf "%3d '%s'\n", $count{$_}, $_ for (sort keys %count);
2840 The pattern begins with a class matching a subset of letters. Whenever
2841 this matches, a statement like C<$count{'a'}++;> is executed, incrementing
2842 the letter's counter. Then C<(*FAIL)> does what it says, and
2843 the regexp engine proceeds according to the book: as long as the end of
2844 the string hasn't been reached, the position is advanced before looking
2845 for another vowel. Thus, match or no match makes no difference, and the
2846 regexp engine proceeds until the entire string has been inspected.
2847 (It's remarkable that an alternative solution using something like
2849 $count{lc($_)}++ for split('', "supercalifragilisticexpialidocious");
2850 printf "%3d '%s'\n", $count2{$_}, $_ for ( qw{ a e i o u } );
2852 is considerably slower.)
2855 =head2 Pragmas and debugging
2857 Speaking of debugging, there are several pragmas available to control
2858 and debug regexps in Perl. We have already encountered one pragma in
2859 the previous section, S<C<use re 'eval';>>, that allows variable
2860 interpolation and code expressions to coexist in a regexp. The other
2865 @parts = ($tainted =~ /(\w+)\s+(\w+)/; # @parts is now tainted
2867 The C<taint> pragma causes any substrings from a match with a tainted
2868 variable to be tainted as well. This is not normally the case, as
2869 regexps are often used to extract the safe bits from a tainted
2870 variable. Use C<taint> when you are not extracting safe bits, but are
2871 performing some other processing. Both C<taint> and C<eval> pragmas
2872 are lexically scoped, which means they are in effect only until
2873 the end of the block enclosing the pragmas.
2875 use re '/m'; # or any other flags
2876 $multiline_string =~ /^foo/; # /m is implied
2878 The C<re '/flags'> pragma (introduced in Perl
2879 5.14) turns on the given regular expression flags
2880 until the end of the lexical scope. See
2881 L<re/"'E<sol>flags' mode"> for more
2885 /^(.*)$/s; # output debugging info
2887 use re 'debugcolor';
2888 /^(.*)$/s; # output debugging info in living color
2890 The global C<debug> and C<debugcolor> pragmas allow one to get
2891 detailed debugging info about regexp compilation and
2892 execution. C<debugcolor> is the same as debug, except the debugging
2893 information is displayed in color on terminals that can display
2894 termcap color sequences. Here is example output:
2896 % perl -e 'use re "debug"; "abc" =~ /a*b+c/;'
2897 Compiling REx 'a*b+c'
2905 floating 'bc' at 0..2147483647 (checking floating) minlen 2
2906 Guessing start of match, REx 'a*b+c' against 'abc'...
2907 Found floating substr 'bc' at offset 1...
2908 Guessed: match at offset 0
2909 Matching REx 'a*b+c' against 'abc'
2910 Setting an EVAL scope, savestack=3
2911 0 <> <abc> | 1: STAR
2912 EXACT <a> can match 1 times out of 32767...
2913 Setting an EVAL scope, savestack=3
2914 1 <a> <bc> | 4: PLUS
2915 EXACT <b> can match 1 times out of 32767...
2916 Setting an EVAL scope, savestack=3
2917 2 <ab> <c> | 7: EXACT <c>
2920 Freeing REx: 'a*b+c'
2922 If you have gotten this far into the tutorial, you can probably guess
2923 what the different parts of the debugging output tell you. The first
2926 Compiling REx 'a*b+c'
2935 describes the compilation stage. C<STAR(4)> means that there is a
2936 starred object, in this case C<'a'>, and if it matches, goto line 4,
2937 I<i.e.>, C<PLUS(7)>. The middle lines describe some heuristics and
2938 optimizations performed before a match:
2940 floating 'bc' at 0..2147483647 (checking floating) minlen 2
2941 Guessing start of match, REx 'a*b+c' against 'abc'...
2942 Found floating substr 'bc' at offset 1...
2943 Guessed: match at offset 0
2945 Then the match is executed and the remaining lines describe the
2948 Matching REx 'a*b+c' against 'abc'
2949 Setting an EVAL scope, savestack=3
2950 0 <> <abc> | 1: STAR
2951 EXACT <a> can match 1 times out of 32767...
2952 Setting an EVAL scope, savestack=3
2953 1 <a> <bc> | 4: PLUS
2954 EXACT <b> can match 1 times out of 32767...
2955 Setting an EVAL scope, savestack=3
2956 2 <ab> <c> | 7: EXACT <c>
2959 Freeing REx: 'a*b+c'
2961 Each step is of the form S<C<< n <x> <y> >>>, with C<< <x> >> the
2962 part of the string matched and C<< <y> >> the part not yet
2963 matched. The S<C<< | 1: STAR >>> says that Perl is at line number 1
2964 in the compilation list above. See
2965 L<perldebguts/"Debugging Regular Expressions"> for much more detail.
2967 An alternative method of debugging regexps is to embed C<print>
2968 statements within the regexp. This provides a blow-by-blow account of
2969 the backtracking in an alternation:
2971 "that this" =~ m@(?{print "Start at position ", pos, "\n";})
2981 (?{print "Done at position ", pos, "\n";})
2997 This is just a tutorial. For the full story on Perl regular
2998 expressions, see the L<perlre> regular expressions reference page.
3000 For more information on the matching C<m//> and substitution C<s///>
3001 operators, see L<perlop/"Regexp Quote-Like Operators">. For
3002 information on the C<split> operation, see L<perlfunc/split>.
3004 For an excellent all-around resource on the care and feeding of
3005 regular expressions, see the book I<Mastering Regular Expressions> by
3006 Jeffrey Friedl (published by O'Reilly, ISBN 1556592-257-3).
3008 =head1 AUTHOR AND COPYRIGHT
3010 Copyright (c) 2000 Mark Kvale.
3011 All rights reserved.
3012 Now maintained by Perl porters.
3014 This document may be distributed under the same terms as Perl itself.
3016 =head2 Acknowledgments
3018 The inspiration for the stop codon DNA example came from the ZIP
3019 code example in chapter 7 of I<Mastering Regular Expressions>.
3021 The author would like to thank Jeff Pinyan, Andrew Johnson, Peter
3022 Haworth, Ronald J Kimball, and Joe Smith for all their helpful