Commit | Line | Data |
a0d0e21e |
1 | =head1 NAME |
2 | |
3 | perlre - Perl regular expressions |
4 | |
5 | =head1 DESCRIPTION |
6 | |
cb1a09d0 |
7 | This page describes the syntax of regular expressions in Perl. For a |
5f05dabc |
8 | description of how to I<use> regular expressions in matching |
cb1a09d0 |
9 | operations, plus various examples of the same, see C<m//> and C<s///> in |
10 | L<perlop>. |
11 | |
68dc0745 |
12 | The matching operations can have various modifiers. The modifiers |
13 | which relate to the interpretation of the regular expression inside |
14 | are listed below. For the modifiers that alter the behaviour of the |
15 | operation, see L<perlop/"m//"> and L<perlop/"s//">. |
a0d0e21e |
16 | |
55497cff |
17 | =over 4 |
18 | |
19 | =item i |
20 | |
21 | Do case-insensitive pattern matching. |
22 | |
a034a98d |
23 | If C<use locale> is in effect, the case map is taken from the current |
24 | locale. See L<perllocale>. |
25 | |
54310121 |
26 | =item m |
55497cff |
27 | |
28 | Treat string as multiple lines. That is, change "^" and "$" from matching |
5f05dabc |
29 | at only the very start or end of the string to the start or end of any |
55497cff |
30 | line anywhere within the string, |
31 | |
54310121 |
32 | =item s |
55497cff |
33 | |
34 | Treat string as single line. That is, change "." to match any character |
35 | whatsoever, even a newline, which it normally would not match. |
36 | |
54310121 |
37 | =item x |
55497cff |
38 | |
39 | Extend your pattern's legibility by permitting whitespace and comments. |
40 | |
41 | =back |
a0d0e21e |
42 | |
43 | These are usually written as "the C</x> modifier", even though the delimiter |
44 | in question might not actually be a slash. In fact, any of these |
45 | modifiers may also be embedded within the regular expression itself using |
46 | the new C<(?...)> construct. See below. |
47 | |
4633a7c4 |
48 | The C</x> modifier itself needs a little more explanation. It tells |
55497cff |
49 | the regular expression parser to ignore whitespace that is neither |
50 | backslashed nor within a character class. You can use this to break up |
4633a7c4 |
51 | your regular expression into (slightly) more readable parts. The C<#> |
54310121 |
52 | character is also treated as a metacharacter introducing a comment, |
55497cff |
53 | just as in ordinary Perl code. This also means that if you want real |
54 | whitespace or C<#> characters in the pattern that you'll have to either |
55 | escape them or encode them using octal or hex escapes. Taken together, |
56 | these features go a long way towards making Perl's regular expressions |
57 | more readable. See the C comment deletion code in L<perlop>. |
a0d0e21e |
58 | |
59 | =head2 Regular Expressions |
60 | |
61 | The patterns used in pattern matching are regular expressions such as |
62 | those supplied in the Version 8 regexp routines. (In fact, the |
63 | routines are derived (distantly) from Henry Spencer's freely |
64 | redistributable reimplementation of the V8 routines.) |
65 | See L<Version 8 Regular Expressions> for details. |
66 | |
67 | In particular the following metacharacters have their standard I<egrep>-ish |
68 | meanings: |
69 | |
54310121 |
70 | \ Quote the next metacharacter |
a0d0e21e |
71 | ^ Match the beginning of the line |
72 | . Match any character (except newline) |
c07a80fd |
73 | $ Match the end of the line (or before newline at the end) |
a0d0e21e |
74 | | Alternation |
75 | () Grouping |
76 | [] Character class |
77 | |
5f05dabc |
78 | By default, the "^" character is guaranteed to match at only the |
79 | beginning of the string, the "$" character at only the end (or before the |
a0d0e21e |
80 | newline at the end) and Perl does certain optimizations with the |
81 | assumption that the string contains only one line. Embedded newlines |
82 | will not be matched by "^" or "$". You may, however, wish to treat a |
4a6725af |
83 | string as a multi-line buffer, such that the "^" will match after any |
a0d0e21e |
84 | newline within the string, and "$" will match before any newline. At the |
85 | cost of a little more overhead, you can do this by using the /m modifier |
86 | on the pattern match operator. (Older programs did this by setting C<$*>, |
5f05dabc |
87 | but this practice is now deprecated.) |
a0d0e21e |
88 | |
4a6725af |
89 | To facilitate multi-line substitutions, the "." character never matches a |
55497cff |
90 | newline unless you use the C</s> modifier, which in effect tells Perl to pretend |
a0d0e21e |
91 | the string is a single line--even if it isn't. The C</s> modifier also |
92 | overrides the setting of C<$*>, in case you have some (badly behaved) older |
93 | code that sets it in another module. |
94 | |
95 | The following standard quantifiers are recognized: |
96 | |
97 | * Match 0 or more times |
98 | + Match 1 or more times |
99 | ? Match 1 or 0 times |
100 | {n} Match exactly n times |
101 | {n,} Match at least n times |
102 | {n,m} Match at least n but not more than m times |
103 | |
104 | (If a curly bracket occurs in any other context, it is treated |
105 | as a regular character.) The "*" modifier is equivalent to C<{0,}>, the "+" |
25f94b33 |
106 | modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited |
c07a80fd |
107 | to integral values less than 65536. |
a0d0e21e |
108 | |
54310121 |
109 | By default, a quantified subpattern is "greedy", that is, it will match as |
110 | many times as possible (given a particular starting location) while still |
111 | allowing the rest of the pattern to match. If you want it to match the |
112 | minimum number of times possible, follow the quantifier with a "?". Note |
113 | that the meanings don't change, just the "greediness": |
a0d0e21e |
114 | |
115 | *? Match 0 or more times |
116 | +? Match 1 or more times |
117 | ?? Match 0 or 1 time |
118 | {n}? Match exactly n times |
119 | {n,}? Match at least n times |
120 | {n,m}? Match at least n but not more than m times |
121 | |
5f05dabc |
122 | Because patterns are processed as double quoted strings, the following |
a0d0e21e |
123 | also work: |
124 | |
0f36ee90 |
125 | \t tab (HT, TAB) |
126 | \n newline (LF, NL) |
127 | \r return (CR) |
128 | \f form feed (FF) |
129 | \a alarm (bell) (BEL) |
130 | \e escape (think troff) (ESC) |
cb1a09d0 |
131 | \033 octal char (think of a PDP-11) |
132 | \x1B hex char |
a0d0e21e |
133 | \c[ control char |
cb1a09d0 |
134 | \l lowercase next char (think vi) |
135 | \u uppercase next char (think vi) |
136 | \L lowercase till \E (think vi) |
137 | \U uppercase till \E (think vi) |
138 | \E end case modification (think vi) |
201ecf35 |
139 | \Q quote (disable) regexp metacharacters till \E |
a0d0e21e |
140 | |
a034a98d |
141 | If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u> |
142 | and <\U> is taken from the current locale. See L<perllocale>. |
143 | |
a0d0e21e |
144 | In addition, Perl defines the following: |
145 | |
146 | \w Match a "word" character (alphanumeric plus "_") |
147 | \W Match a non-word character |
148 | \s Match a whitespace character |
149 | \S Match a non-whitespace character |
150 | \d Match a digit character |
151 | \D Match a non-digit character |
152 | |
153 | Note that C<\w> matches a single alphanumeric character, not a whole |
a034a98d |
154 | word. To match a word you'd need to say C<\w+>. If C<use locale> is in |
155 | effect, the list of alphabetic characters generated by C<\w> is taken |
156 | from the current locale. See L<perllocale>. You may use C<\w>, C<\W>, |
157 | C<\s>, C<\S>, C<\d>, and C<\D> within character classes (though not as |
158 | either end of a range). |
a0d0e21e |
159 | |
160 | Perl defines the following zero-width assertions: |
161 | |
162 | \b Match a word boundary |
163 | \B Match a non-(word boundary) |
5f05dabc |
164 | \A Match at only beginning of string |
165 | \Z Match at only end of string (or before newline at the end) |
a99df21c |
166 | \G Match only where previous m//g left off (works only with /g) |
a0d0e21e |
167 | |
168 | A word boundary (C<\b>) is defined as a spot between two characters that |
68dc0745 |
169 | has a C<\w> on one side of it and a C<\W> on the other side of it (in |
a0d0e21e |
170 | either order), counting the imaginary characters off the beginning and |
171 | end of the string as matching a C<\W>. (Within character classes C<\b> |
172 | represents backspace rather than a word boundary.) The C<\A> and C<\Z> are |
173 | just like "^" and "$" except that they won't match multiple times when the |
174 | C</m> modifier is used, while "^" and "$" will match at every internal line |
c07a80fd |
175 | boundary. To match the actual end of the string, not ignoring newline, |
a99df21c |
176 | you can use C<\Z(?!\n)>. The C<\G> assertion can be used to chain global |
177 | matches (using C<m//g>), as described in |
e7ea3e70 |
178 | L<perlop/"Regexp Quote-Like Operators">. |
a99df21c |
179 | |
e7ea3e70 |
180 | It is also useful when writing C<lex>-like scanners, when you have several |
181 | regexps which you want to match against consequent substrings of your |
182 | string, see the previous reference. |
44a8e56a |
183 | The actual location where C<\G> will match can also be influenced |
184 | by using C<pos()> as an lvalue. See L<perlfunc/pos>. |
a0d0e21e |
185 | |
0f36ee90 |
186 | When the bracketing construct C<( ... )> is used, \E<lt>digitE<gt> matches the |
cb1a09d0 |
187 | digit'th substring. Outside of the pattern, always use "$" instead of "\" |
0f36ee90 |
188 | in front of the digit. (While the \E<lt>digitE<gt> notation can on rare occasion work |
cb1a09d0 |
189 | outside the current pattern, this should not be relied upon. See the |
0f36ee90 |
190 | WARNING below.) The scope of $E<lt>digitE<gt> (and C<$`>, C<$&>, and C<$'>) |
cb1a09d0 |
191 | extends to the end of the enclosing BLOCK or eval string, or to the next |
192 | successful pattern match, whichever comes first. If you want to use |
5f05dabc |
193 | parentheses to delimit a subpattern (e.g., a set of alternatives) without |
84dc3c4d |
194 | saving it as a subpattern, follow the ( with a ?:. |
cb1a09d0 |
195 | |
196 | You may have as many parentheses as you wish. If you have more |
a0d0e21e |
197 | than 9 substrings, the variables $10, $11, ... refer to the |
198 | corresponding substring. Within the pattern, \10, \11, etc. refer back |
5f05dabc |
199 | to substrings if there have been at least that many left parentheses before |
c07a80fd |
200 | the backreference. Otherwise (for backward compatibility) \10 is the |
a0d0e21e |
201 | same as \010, a backspace, and \11 the same as \011, a tab. And so |
202 | on. (\1 through \9 are always backreferences.) |
203 | |
204 | C<$+> returns whatever the last bracket match matched. C<$&> returns the |
0f36ee90 |
205 | entire matched string. (C<$0> used to return the same thing, but not any |
a0d0e21e |
206 | more.) C<$`> returns everything before the matched string. C<$'> returns |
207 | everything after the matched string. Examples: |
208 | |
209 | s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words |
210 | |
211 | if (/Time: (..):(..):(..)/) { |
212 | $hours = $1; |
213 | $minutes = $2; |
214 | $seconds = $3; |
215 | } |
216 | |
68dc0745 |
217 | Once perl sees that you need one of C<$&>, C<$`> or C<$'> anywhere in |
218 | the program, it has to provide them on each and every pattern match. |
219 | This can slow your program down. The same mechanism that handles |
220 | these provides for the use of $1, $2, etc., so you pay the same price |
221 | for each regexp that contains capturing parentheses. But if you never |
222 | use $&, etc., in your script, then regexps I<without> capturing |
223 | parentheses won't be penalized. So avoid $&, $', and $` if you can, |
224 | but if you can't (and some algorithms really appreciate them), once |
225 | you've used them once, use them at will, because you've already paid |
226 | the price. |
227 | |
a0d0e21e |
228 | You will note that all backslashed metacharacters in Perl are |
201ecf35 |
229 | alphanumeric, such as C<\b>, C<\w>, C<\n>. Unlike some other regular |
230 | expression languages, there are no backslashed symbols that aren't |
231 | alphanumeric. So anything that looks like \\, \(, \), \E<lt>, \E<gt>, |
232 | \{, or \} is always interpreted as a literal character, not a |
233 | metacharacter. This was once used in a common idiom to disable or |
234 | quote the special meanings of regular expression metacharacters in a |
235 | string that you want to use for a pattern. Simply quote all the |
a0d0e21e |
236 | non-alphanumeric characters: |
237 | |
238 | $pattern =~ s/(\W)/\\$1/g; |
239 | |
201ecf35 |
240 | Now it is much more common to see either the quotemeta() function or |
241 | the \Q escape sequence used to disable the metacharacters special |
242 | meanings like this: |
a0d0e21e |
243 | |
244 | /$unquoted\Q$quoted\E$unquoted/ |
245 | |
5f05dabc |
246 | Perl defines a consistent extension syntax for regular expressions. |
247 | The syntax is a pair of parentheses with a question mark as the first |
248 | thing within the parentheses (this was a syntax error in older |
249 | versions of Perl). The character after the question mark gives the |
250 | function of the extension. Several extensions are already supported: |
a0d0e21e |
251 | |
252 | =over 10 |
253 | |
254 | =item (?#text) |
255 | |
cb1a09d0 |
256 | A comment. The text is ignored. If the C</x> switch is used to enable |
257 | whitespace formatting, a simple C<#> will suffice. |
a0d0e21e |
258 | |
259 | =item (?:regexp) |
260 | |
0f36ee90 |
261 | This groups things like "()" but doesn't make backreferences like "()" does. So |
a0d0e21e |
262 | |
263 | split(/\b(?:a|b|c)\b/) |
264 | |
265 | is like |
266 | |
267 | split(/\b(a|b|c)\b/) |
268 | |
269 | but doesn't spit out extra fields. |
270 | |
271 | =item (?=regexp) |
272 | |
273 | A zero-width positive lookahead assertion. For example, C</\w+(?=\t)/> |
274 | matches a word followed by a tab, without including the tab in C<$&>. |
275 | |
276 | =item (?!regexp) |
277 | |
278 | A zero-width negative lookahead assertion. For example C</foo(?!bar)/> |
279 | matches any occurrence of "foo" that isn't followed by "bar". Note |
280 | however that lookahead and lookbehind are NOT the same thing. You cannot |
281 | use this for lookbehind: C</(?!foo)bar/> will not find an occurrence of |
282 | "bar" that is preceded by something which is not "foo". That's because |
283 | the C<(?!foo)> is just saying that the next thing cannot be "foo"--and |
284 | it's not, it's a "bar", so "foobar" will match. You would have to do |
0f36ee90 |
285 | something like C</(?!foo)...bar/> for that. We say "like" because there's |
a0d0e21e |
286 | the case of your "bar" not having three characters before it. You could |
c07a80fd |
287 | cover that this way: C</(?:(?!foo)...|^..?)bar/>. Sometimes it's still |
a0d0e21e |
288 | easier just to say: |
289 | |
c07a80fd |
290 | if (/foo/ && $` =~ /bar$/) |
a0d0e21e |
291 | |
c277df42 |
292 | For lookbehind see below. |
293 | |
294 | =item (?<=regexp) |
295 | |
296 | A zero-width positive lookbehind assertion. For example, C</(?=\t)\w+/> |
297 | matches a word following a tab, without including the tab in C<$&>. |
298 | Works only for fixed-width lookbehind. |
299 | |
300 | =item (?<!regexp) |
301 | |
302 | A zero-width negative lookbehind assertion. For example C</(?<!bar)foo/> |
303 | matches any occurrence of "foo" that isn't following "bar". |
304 | Works only for fixed-width lookbehind. |
305 | |
306 | =item (?{ code }) |
307 | |
308 | Experimental "evaluate any Perl code" zero-width assertion. Always |
309 | succeeds. Currently the quoting rules are somewhat convoluted, as is the |
310 | determination where the C<code> ends. |
311 | |
312 | |
313 | =item C<(?E<gt>regexp)> |
314 | |
315 | An "independend" subexpression. Matches the substring which a |
316 | I<standalone> C<regexp> would match if anchored at the given position, |
317 | B<and only this substring>. |
318 | |
319 | Say, C<^(?E<gt>a*)ab> will never match, since C<(?E<gt>a*)> (anchored |
320 | at the beginning of string, as above) will match I<all> the characters |
321 | C<a> at the beginning of string, leaving no C<a> for C<ab> to match. |
322 | In contrast, C<a*ab> will match the same as C<a+b>, since the match of |
323 | the subgroup C<a*> is influenced by the following group C<ab> (see |
324 | L<"Backtracking">). In particular, C<a*> inside C<a*ab> will match |
325 | less characters that a standalone C<a*>, since this makes the tail match. |
326 | |
327 | Note that a similar effect to C<(?E<gt>regexp)> may be achieved by |
328 | |
329 | (?=(regexp))\1 |
330 | |
331 | since the lookahead is in I<"logical"> context, thus matches the same |
332 | substring as a standalone C<a+>. The following C<\1> eats the matched |
333 | string, thus making a zero-length assertion into an analogue of |
334 | C<(?>...)>. (The difference of these two constructions is that the |
335 | second one uses a catching group, thus shifts ordinals of |
336 | backreferences in the rest of a regular expression.) |
337 | |
338 | This construction is very useful for optimizations of "eternal" |
339 | matches, since it will not backtrack (see L<"Backtracking">). Say, |
340 | |
341 | / \( ( |
342 | [^()]+ |
343 | | |
344 | \( [^()]* \) |
345 | )+ |
346 | \) /x |
347 | |
348 | will match a nonempty group with matching two-or-less-level-deep |
349 | parentheses. It is very efficient in finding such groups. However, |
350 | if there is no such group, it is going to take forever (on reasonably |
351 | long string), since there are so many different ways to split a long |
352 | string into several substrings (this is essentially what C<(.+)+> is |
353 | doing, and this is a subpattern of the above pattern). Say, on |
354 | C<((()aaaaaaaaaaaaaaaaaa> the above pattern detects no-match in 5sec |
355 | (on kitchentop'96 processor), and each extra letter doubles this time. |
356 | |
357 | However, a tiny modification of this |
358 | |
359 | / \( ( |
360 | (?> [^()]+ ) |
361 | | |
362 | \( [^()]* \) |
363 | )+ |
364 | \) /x |
365 | |
366 | which uses (?>...) matches exactly when the above one does (it is a |
367 | good excercise to check this), but finishes in a fourth of the above |
368 | time on a similar string with 1000000 C<a>s. |
369 | |
370 | Note that on simple groups like the above C<(?> [^()]+ )> a similar |
371 | effect may be achieved by negative lookahead, as in C<[^()]+ (?! [^()] )>. |
372 | This was only 4 times slower on a string with 1000000 C<a>s. |
373 | |
374 | =item (?(condition)yes-regexp|no-regexp) |
375 | |
376 | =item (?(condition)yes-regexp) |
377 | |
378 | Conditional expression. C<(condition)> should be either an integer in |
379 | parentheses (which is valid if the corresponding pair of parentheses |
380 | matched), or lookahead/lookbehind/evaluate zero-width assertion. |
381 | |
382 | Say, |
383 | |
384 | / ( \( )? |
385 | [^()]+ |
386 | (?(1) \) )/x |
387 | |
388 | matches a chunk of non-parentheses, possibly included in parentheses |
389 | themselves. |
a0d0e21e |
390 | |
391 | =item (?imsx) |
392 | |
393 | One or more embedded pattern-match modifiers. This is particularly |
394 | useful for patterns that are specified in a table somewhere, some of |
395 | which want to be case sensitive, and some of which don't. The case |
5f05dabc |
396 | insensitive ones need to include merely C<(?i)> at the front of the |
a0d0e21e |
397 | pattern. For example: |
398 | |
399 | $pattern = "foobar"; |
c07a80fd |
400 | if ( /$pattern/i ) |
a0d0e21e |
401 | |
402 | # more flexible: |
403 | |
404 | $pattern = "(?i)foobar"; |
c07a80fd |
405 | if ( /$pattern/ ) |
a0d0e21e |
406 | |
c277df42 |
407 | Note that these modifiers are localized inside an enclosing group (if |
408 | any). Say, |
409 | |
410 | ( (?i) blah ) \s+ \1 |
411 | |
412 | (assuming C<x> modifier, and no C<i> modifier outside of this group) |
413 | will match a repeated (I<including the case>!) word C<blah> in any |
414 | case. |
415 | |
a0d0e21e |
416 | =back |
417 | |
418 | The specific choice of question mark for this and the new minimal |
419 | matching construct was because 1) question mark is pretty rare in older |
420 | regular expressions, and 2) whenever you see one, you should stop |
421 | and "question" exactly what is going on. That's psychology... |
422 | |
c07a80fd |
423 | =head2 Backtracking |
424 | |
c277df42 |
425 | A fundamental feature of regular expression matching involves the |
426 | notion called I<backtracking>. which is currently used (when needed) |
427 | by all regular expression quantifiers, namely C<*>, C<*?>, C<+>, |
428 | C<+?>, C<{n,m}>, and C<{n,m}?>. |
c07a80fd |
429 | |
430 | For a regular expression to match, the I<entire> regular expression must |
431 | match, not just part of it. So if the beginning of a pattern containing a |
432 | quantifier succeeds in a way that causes later parts in the pattern to |
433 | fail, the matching engine backs up and recalculates the beginning |
434 | part--that's why it's called backtracking. |
435 | |
436 | Here is an example of backtracking: Let's say you want to find the |
437 | word following "foo" in the string "Food is on the foo table.": |
438 | |
439 | $_ = "Food is on the foo table."; |
440 | if ( /\b(foo)\s+(\w+)/i ) { |
441 | print "$2 follows $1.\n"; |
442 | } |
443 | |
444 | When the match runs, the first part of the regular expression (C<\b(foo)>) |
445 | finds a possible match right at the beginning of the string, and loads up |
446 | $1 with "Foo". However, as soon as the matching engine sees that there's |
447 | no whitespace following the "Foo" that it had saved in $1, it realizes its |
68dc0745 |
448 | mistake and starts over again one character after where it had the |
c07a80fd |
449 | tentative match. This time it goes all the way until the next occurrence |
450 | of "foo". The complete regular expression matches this time, and you get |
451 | the expected output of "table follows foo." |
452 | |
453 | Sometimes minimal matching can help a lot. Imagine you'd like to match |
454 | everything between "foo" and "bar". Initially, you write something |
455 | like this: |
456 | |
457 | $_ = "The food is under the bar in the barn."; |
458 | if ( /foo(.*)bar/ ) { |
459 | print "got <$1>\n"; |
460 | } |
461 | |
462 | Which perhaps unexpectedly yields: |
463 | |
464 | got <d is under the bar in the > |
465 | |
466 | That's because C<.*> was greedy, so you get everything between the |
467 | I<first> "foo" and the I<last> "bar". In this case, it's more effective |
468 | to use minimal matching to make sure you get the text between a "foo" |
469 | and the first "bar" thereafter. |
470 | |
471 | if ( /foo(.*?)bar/ ) { print "got <$1>\n" } |
472 | got <d is under the > |
473 | |
474 | Here's another example: let's say you'd like to match a number at the end |
475 | of a string, and you also want to keep the preceding part the match. |
476 | So you write this: |
477 | |
478 | $_ = "I have 2 numbers: 53147"; |
479 | if ( /(.*)(\d*)/ ) { # Wrong! |
480 | print "Beginning is <$1>, number is <$2>.\n"; |
481 | } |
482 | |
483 | That won't work at all, because C<.*> was greedy and gobbled up the |
484 | whole string. As C<\d*> can match on an empty string the complete |
485 | regular expression matched successfully. |
486 | |
8e1088bc |
487 | Beginning is <I have 2 numbers: 53147>, number is <>. |
c07a80fd |
488 | |
489 | Here are some variants, most of which don't work: |
490 | |
491 | $_ = "I have 2 numbers: 53147"; |
492 | @pats = qw{ |
493 | (.*)(\d*) |
494 | (.*)(\d+) |
495 | (.*?)(\d*) |
496 | (.*?)(\d+) |
497 | (.*)(\d+)$ |
498 | (.*?)(\d+)$ |
499 | (.*)\b(\d+)$ |
500 | (.*\D)(\d+)$ |
501 | }; |
502 | |
503 | for $pat (@pats) { |
504 | printf "%-12s ", $pat; |
505 | if ( /$pat/ ) { |
506 | print "<$1> <$2>\n"; |
507 | } else { |
508 | print "FAIL\n"; |
509 | } |
510 | } |
511 | |
512 | That will print out: |
513 | |
514 | (.*)(\d*) <I have 2 numbers: 53147> <> |
515 | (.*)(\d+) <I have 2 numbers: 5314> <7> |
516 | (.*?)(\d*) <> <> |
517 | (.*?)(\d+) <I have > <2> |
518 | (.*)(\d+)$ <I have 2 numbers: 5314> <7> |
519 | (.*?)(\d+)$ <I have 2 numbers: > <53147> |
520 | (.*)\b(\d+)$ <I have 2 numbers: > <53147> |
521 | (.*\D)(\d+)$ <I have 2 numbers: > <53147> |
522 | |
523 | As you see, this can be a bit tricky. It's important to realize that a |
524 | regular expression is merely a set of assertions that gives a definition |
525 | of success. There may be 0, 1, or several different ways that the |
526 | definition might succeed against a particular string. And if there are |
5f05dabc |
527 | multiple ways it might succeed, you need to understand backtracking to know which variety of success you will achieve. |
c07a80fd |
528 | |
529 | When using lookahead assertions and negations, this can all get even |
54310121 |
530 | tricker. Imagine you'd like to find a sequence of non-digits not |
c07a80fd |
531 | followed by "123". You might try to write that as |
532 | |
533 | $_ = "ABC123"; |
534 | if ( /^\D*(?!123)/ ) { # Wrong! |
535 | print "Yup, no 123 in $_\n"; |
536 | } |
537 | |
538 | But that isn't going to match; at least, not the way you're hoping. It |
539 | claims that there is no 123 in the string. Here's a clearer picture of |
540 | why it that pattern matches, contrary to popular expectations: |
541 | |
542 | $x = 'ABC123' ; |
543 | $y = 'ABC445' ; |
544 | |
545 | print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ; |
546 | print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ; |
547 | |
548 | print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ; |
549 | print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ; |
550 | |
551 | This prints |
552 | |
553 | 2: got ABC |
554 | 3: got AB |
555 | 4: got ABC |
556 | |
5f05dabc |
557 | You might have expected test 3 to fail because it seems to a more |
c07a80fd |
558 | general purpose version of test 1. The important difference between |
559 | them is that test 3 contains a quantifier (C<\D*>) and so can use |
560 | backtracking, whereas test 1 will not. What's happening is |
561 | that you've asked "Is it true that at the start of $x, following 0 or more |
5f05dabc |
562 | non-digits, you have something that's not 123?" If the pattern matcher had |
c07a80fd |
563 | let C<\D*> expand to "ABC", this would have caused the whole pattern to |
54310121 |
564 | fail. |
c07a80fd |
565 | The search engine will initially match C<\D*> with "ABC". Then it will |
566 | try to match C<(?!123> with "123" which, of course, fails. But because |
567 | a quantifier (C<\D*>) has been used in the regular expression, the |
568 | search engine can backtrack and retry the match differently |
54310121 |
569 | in the hope of matching the complete regular expression. |
c07a80fd |
570 | |
54310121 |
571 | Well now, |
c07a80fd |
572 | the pattern really, I<really> wants to succeed, so it uses the |
5f05dabc |
573 | standard regexp back-off-and-retry and lets C<\D*> expand to just "AB" this |
c07a80fd |
574 | time. Now there's indeed something following "AB" that is not |
575 | "123". It's in fact "C123", which suffices. |
576 | |
577 | We can deal with this by using both an assertion and a negation. We'll |
578 | say that the first part in $1 must be followed by a digit, and in fact, it |
579 | must also be followed by something that's not "123". Remember that the |
580 | lookaheads are zero-width expressions--they only look, but don't consume |
581 | any of the string in their match. So rewriting this way produces what |
582 | you'd expect; that is, case 5 will fail, but case 6 succeeds: |
583 | |
584 | print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ; |
585 | print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ; |
586 | |
587 | 6: got ABC |
588 | |
589 | In other words, the two zero-width assertions next to each other work like |
590 | they're ANDed together, just as you'd use any builtin assertions: C</^$/> |
591 | matches only if you're at the beginning of the line AND the end of the |
592 | line simultaneously. The deeper underlying truth is that juxtaposition in |
593 | regular expressions always means AND, except when you write an explicit OR |
594 | using the vertical bar. C</ab/> means match "a" AND (then) match "b", |
595 | although the attempted matches are made at different positions because "a" |
596 | is not a zero-width assertion, but a one-width assertion. |
597 | |
598 | One warning: particularly complicated regular expressions can take |
599 | exponential time to solve due to the immense number of possible ways they |
600 | can use backtracking to try match. For example this will take a very long |
601 | time to run |
602 | |
603 | /((a{0,5}){0,5}){0,5}/ |
604 | |
605 | And if you used C<*>'s instead of limiting it to 0 through 5 matches, then |
606 | it would take literally forever--or until you ran out of stack space. |
607 | |
c277df42 |
608 | A powerful tool for optimizing such beasts is "independent" groups, |
609 | which do not backtrace (see L<C<(?E<gt>regexp)>>). Note also that |
610 | zero-length lookahead/lookbehind assertions will not backtrace to make |
611 | the tail match, since they are in "logical" context: only the fact |
612 | whether they match or not is considered relevant. For an example |
613 | where side-effects of a lookahead I<might> have influenced the |
614 | following match, see L<C<(?E<gt>regexp)>>. |
615 | |
a0d0e21e |
616 | =head2 Version 8 Regular Expressions |
617 | |
618 | In case you're not familiar with the "regular" Version 8 regexp |
619 | routines, here are the pattern-matching rules not described above. |
620 | |
54310121 |
621 | Any single character matches itself, unless it is a I<metacharacter> |
a0d0e21e |
622 | with a special meaning described here or above. You can cause |
623 | characters which normally function as metacharacters to be interpreted |
5f05dabc |
624 | literally by prefixing them with a "\" (e.g., "\." matches a ".", not any |
a0d0e21e |
625 | character; "\\" matches a "\"). A series of characters matches that |
626 | series of characters in the target string, so the pattern C<blurfl> |
627 | would match "blurfl" in the target string. |
628 | |
629 | You can specify a character class, by enclosing a list of characters |
630 | in C<[]>, which will match any one of the characters in the list. If the |
631 | first character after the "[" is "^", the class matches any character not |
632 | in the list. Within a list, the "-" character is used to specify a |
633 | range, so that C<a-z> represents all the characters between "a" and "z", |
84850974 |
634 | inclusive. If you want "-" itself to be a member of a class, put it |
635 | at the start or end of the list, or escape it with a backslash. (The |
636 | following all specify the same class of three characters: C<[-az]>, |
637 | C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which |
638 | specifies a class containing twenty-six characters.) |
a0d0e21e |
639 | |
54310121 |
640 | Characters may be specified using a metacharacter syntax much like that |
a0d0e21e |
641 | used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return, |
642 | "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string |
643 | of octal digits, matches the character whose ASCII value is I<nnn>. |
0f36ee90 |
644 | Similarly, \xI<nn>, where I<nn> are hexadecimal digits, matches the |
a0d0e21e |
645 | character whose ASCII value is I<nn>. The expression \cI<x> matches the |
54310121 |
646 | ASCII character control-I<x>. Finally, the "." metacharacter matches any |
a0d0e21e |
647 | character except "\n" (unless you use C</s>). |
648 | |
649 | You can specify a series of alternatives for a pattern using "|" to |
650 | separate them, so that C<fee|fie|foe> will match any of "fee", "fie", |
651 | or "foe" in the target string (as would C<f(e|i|o)e>). Note that the |
652 | first alternative includes everything from the last pattern delimiter |
653 | ("(", "[", or the beginning of the pattern) up to the first "|", and |
654 | the last alternative contains everything from the last "|" to the next |
655 | pattern delimiter. For this reason, it's common practice to include |
656 | alternatives in parentheses, to minimize confusion about where they |
748a9306 |
657 | start and end. Note however that "|" is interpreted as a literal with |
658 | square brackets, so if you write C<[fee|fie|foe]> you're really only |
659 | matching C<[feio|]>. |
a0d0e21e |
660 | |
54310121 |
661 | Within a pattern, you may designate subpatterns for later reference by |
a0d0e21e |
662 | enclosing them in parentheses, and you may refer back to the I<n>th |
54310121 |
663 | subpattern later in the pattern using the metacharacter \I<n>. |
664 | Subpatterns are numbered based on the left to right order of their |
a0d0e21e |
665 | opening parenthesis. Note that a backreference matches whatever |
54310121 |
666 | actually matched the subpattern in the string being examined, not the |
667 | rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will |
668 | match "0x1234 0x4321",but not "0x1234 01234", because subpattern 1 |
748a9306 |
669 | actually matched "0x", even though the rule C<0|0x> could |
a0d0e21e |
670 | potentially match the leading 0 in the second number. |
cb1a09d0 |
671 | |
672 | =head2 WARNING on \1 vs $1 |
673 | |
674 | Some people get too used to writing things like |
675 | |
676 | $pattern =~ s/(\W)/\\\1/g; |
677 | |
678 | This is grandfathered for the RHS of a substitute to avoid shocking the |
679 | B<sed> addicts, but it's a dirty habit to get into. That's because in |
5f05dabc |
680 | PerlThink, the righthand side of a C<s///> is a double-quoted string. C<\1> in |
cb1a09d0 |
681 | the usual double-quoted string means a control-A. The customary Unix |
682 | meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit |
683 | of doing that, you get yourself into trouble if you then add an C</e> |
684 | modifier. |
685 | |
686 | s/(\d+)/ \1 + 1 /eg; |
687 | |
688 | Or if you try to do |
689 | |
690 | s/(\d+)/\1000/; |
691 | |
692 | You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with |
693 | C<${1}000>. Basically, the operation of interpolation should not be confused |
694 | with the operation of matching a backreference. Certainly they mean two |
695 | different things on the I<left> side of the C<s///>. |
9fa51da4 |
696 | |
697 | =head2 SEE ALSO |
698 | |
699 | "Mastering Regular Expressions" (see L<perlbook>) by Jeffrey Friedl. |