1=head1 NAME 2 3perlretut - Perl regular expressions tutorial 4 5=head1 DESCRIPTION 6 7This page provides a basic tutorial on understanding, creating and 8using regular expressions in Perl. It serves as a complement to the 9reference page on regular expressions L<perlre>. Regular expressions 10are an integral part of the C<m//>, C<s///>, C<qr//> and C<split> 11operators and so this tutorial also overlaps with 12L<perlop/"Regexp Quote-Like Operators"> and L<perlfunc/split>. 13 14Perl is widely renowned for excellence in text processing, and regular 15expressions are one of the big factors behind this fame. Perl regular 16expressions display an efficiency and flexibility unknown in most 17other computer languages. Mastering even the basics of regular 18expressions will allow you to manipulate text with surprising ease. 19 20What is a regular expression? A regular expression is simply a string 21that describes a pattern. Patterns are in common use these days; 22examples are the patterns typed into a search engine to find web pages 23and the patterns used to list files in a directory, e.g., C<ls *.txt> 24or C<dir *.*>. In Perl, the patterns described by regular expressions 25are used to search strings, extract desired parts of strings, and to 26do search and replace operations. 27 28Regular expressions have the undeserved reputation of being abstract 29and difficult to understand. Regular expressions are constructed using 30simple concepts like conditionals and loops and are no more difficult 31to understand than the corresponding C<if> conditionals and C<while> 32loops in the Perl language itself. In fact, the main challenge in 33learning regular expressions is just getting used to the terse 34notation used to express these concepts. 35 36This tutorial flattens the learning curve by discussing regular 37expression concepts, along with their notation, one at a time and with 38many examples. The first part of the tutorial will progress from the 39simplest word searches to the basic regular expression concepts. If 40you master the first part, you will have all the tools needed to solve 41about 98% of your needs. The second part of the tutorial is for those 42comfortable with the basics and hungry for more power tools. It 43discusses the more advanced regular expression operators and 44introduces the latest cutting edge innovations in 5.6.0. 45 46A note: to save time, 'regular expression' is often abbreviated as 47regexp or regex. Regexp is a more natural abbreviation than regex, but 48is harder to pronounce. The Perl pod documentation is evenly split on 49regexp vs regex; in Perl, there is more than one way to abbreviate it. 50We'll use regexp in this tutorial. 51 52=head1 Part 1: The basics 53 54=head2 Simple word matching 55 56The simplest regexp is simply a word, or more generally, a string of 57characters. A regexp consisting of a word matches any string that 58contains that word: 59 60 "Hello World" =~ /World/; # matches 61 62What is this Perl statement all about? C<"Hello World"> is a simple 63double quoted string. C<World> is the regular expression and the 64C<//> enclosing C</World/> tells Perl to search a string for a match. 65The operator C<=~> associates the string with the regexp match and 66produces a true value if the regexp matched, or false if the regexp 67did not match. In our case, C<World> matches the second word in 68C<"Hello World">, so the expression is true. Expressions like this 69are useful in conditionals: 70 71 if ("Hello World" =~ /World/) { 72 print "It matches\n"; 73 } 74 else { 75 print "It doesn't match\n"; 76 } 77 78There are useful variations on this theme. The sense of the match can 79be reversed by using the C<!~> operator: 80 81 if ("Hello World" !~ /World/) { 82 print "It doesn't match\n"; 83 } 84 else { 85 print "It matches\n"; 86 } 87 88The literal string in the regexp can be replaced by a variable: 89 90 $greeting = "World"; 91 if ("Hello World" =~ /$greeting/) { 92 print "It matches\n"; 93 } 94 else { 95 print "It doesn't match\n"; 96 } 97 98If you're matching against the special default variable C<$_>, the 99C<$_ =~> part can be omitted: 100 101 $_ = "Hello World"; 102 if (/World/) { 103 print "It matches\n"; 104 } 105 else { 106 print "It doesn't match\n"; 107 } 108 109And finally, the C<//> default delimiters for a match can be changed 110to arbitrary delimiters by putting an C<'m'> out front: 111 112 "Hello World" =~ m!World!; # matches, delimited by '!' 113 "Hello World" =~ m{World}; # matches, note the matching '{}' 114 "/usr/bin/perl" =~ m"/perl"; # matches after '/usr/bin', 115 # '/' becomes an ordinary char 116 117C</World/>, C<m!World!>, and C<m{World}> all represent the 118same thing. When, e.g., the quote (C<">) is used as a delimiter, the forward 119slash C<'/'> becomes an ordinary character and can be used in this regexp 120without trouble. 121 122Let's consider how different regexps would match C<"Hello World">: 123 124 "Hello World" =~ /world/; # doesn't match 125 "Hello World" =~ /o W/; # matches 126 "Hello World" =~ /oW/; # doesn't match 127 "Hello World" =~ /World /; # doesn't match 128 129The first regexp C<world> doesn't match because regexps are 130case-sensitive. The second regexp matches because the substring 131S<C<'o W'>> occurs in the string S<C<"Hello World">>. The space 132character ' ' is treated like any other character in a regexp and is 133needed to match in this case. The lack of a space character is the 134reason the third regexp C<'oW'> doesn't match. The fourth regexp 135C<'World '> doesn't match because there is a space at the end of the 136regexp, but not at the end of the string. The lesson here is that 137regexps must match a part of the string I<exactly> in order for the 138statement to be true. 139 140If a regexp matches in more than one place in the string, Perl will 141always match at the earliest possible point in the string: 142 143 "Hello World" =~ /o/; # matches 'o' in 'Hello' 144 "That hat is red" =~ /hat/; # matches 'hat' in 'That' 145 146With respect to character matching, there are a few more points you 147need to know about. First of all, not all characters can be used 'as 148is' in a match. Some characters, called I<metacharacters>, are reserved 149for use in regexp notation. The metacharacters are 150 151 {}[]()^$.|*+?\ 152 153The significance of each of these will be explained 154in the rest of the tutorial, but for now, it is important only to know 155that a metacharacter can be matched by putting a backslash before it: 156 157 "2+2=4" =~ /2+2/; # doesn't match, + is a metacharacter 158 "2+2=4" =~ /2\+2/; # matches, \+ is treated like an ordinary + 159 "The interval is [0,1)." =~ /[0,1)./ # is a syntax error! 160 "The interval is [0,1)." =~ /\[0,1\)\./ # matches 161 "#!/usr/bin/perl" =~ /#!\/usr\/bin\/perl/; # matches 162 163In the last regexp, the forward slash C<'/'> is also backslashed, 164because it is used to delimit the regexp. This can lead to LTS 165(leaning toothpick syndrome), however, and it is often more readable 166to change delimiters. 167 168 "#!/usr/bin/perl" =~ m!#\!/usr/bin/perl!; # easier to read 169 170The backslash character C<'\'> is a metacharacter itself and needs to 171be backslashed: 172 173 'C:\WIN32' =~ /C:\\WIN/; # matches 174 175In addition to the metacharacters, there are some ASCII characters 176which don't have printable character equivalents and are instead 177represented by I<escape sequences>. Common examples are C<\t> for a 178tab, C<\n> for a newline, C<\r> for a carriage return and C<\a> for a 179bell. If your string is better thought of as a sequence of arbitrary 180bytes, the octal escape sequence, e.g., C<\033>, or hexadecimal escape 181sequence, e.g., C<\x1B> may be a more natural representation for your 182bytes. Here are some examples of escapes: 183 184 "1000\t2000" =~ m(0\t2) # matches 185 "1000\n2000" =~ /0\n20/ # matches 186 "1000\t2000" =~ /\000\t2/ # doesn't match, "0" ne "\000" 187 "cat" =~ /\143\x61\x74/ # matches in ASCII, but a weird way to spell cat 188 189If you've been around Perl a while, all this talk of escape sequences 190may seem familiar. Similar escape sequences are used in double-quoted 191strings and in fact the regexps in Perl are mostly treated as 192double-quoted strings. This means that variables can be used in 193regexps as well. Just like double-quoted strings, the values of the 194variables in the regexp will be substituted in before the regexp is 195evaluated for matching purposes. So we have: 196 197 $foo = 'house'; 198 'housecat' =~ /$foo/; # matches 199 'cathouse' =~ /cat$foo/; # matches 200 'housecat' =~ /${foo}cat/; # matches 201 202So far, so good. With the knowledge above you can already perform 203searches with just about any literal string regexp you can dream up. 204Here is a I<very simple> emulation of the Unix grep program: 205 206 % cat > simple_grep 207 #!/usr/bin/perl 208 $regexp = shift; 209 while (<>) { 210 print if /$regexp/; 211 } 212 ^D 213 214 % chmod +x simple_grep 215 216 % simple_grep abba /usr/dict/words 217 Babbage 218 cabbage 219 cabbages 220 sabbath 221 Sabbathize 222 Sabbathizes 223 sabbatical 224 scabbard 225 scabbards 226 227This program is easy to understand. C<#!/usr/bin/perl> is the standard 228way to invoke a perl program from the shell. 229S<C<$regexp = shift;>> saves the first command line argument as the 230regexp to be used, leaving the rest of the command line arguments to 231be treated as files. S<C<< while (<>) >>> loops over all the lines in 232all the files. For each line, S<C<print if /$regexp/;>> prints the 233line if the regexp matches the line. In this line, both C<print> and 234C</$regexp/> use the default variable C<$_> implicitly. 235 236With all of the regexps above, if the regexp matched anywhere in the 237string, it was considered a match. Sometimes, however, we'd like to 238specify I<where> in the string the regexp should try to match. To do 239this, we would use the I<anchor> metacharacters C<^> and C<$>. The 240anchor C<^> means match at the beginning of the string and the anchor 241C<$> means match at the end of the string, or before a newline at the 242end of the string. Here is how they are used: 243 244 "housekeeper" =~ /keeper/; # matches 245 "housekeeper" =~ /^keeper/; # doesn't match 246 "housekeeper" =~ /keeper$/; # matches 247 "housekeeper\n" =~ /keeper$/; # matches 248 249The second regexp doesn't match because C<^> constrains C<keeper> to 250match only at the beginning of the string, but C<"housekeeper"> has 251keeper starting in the middle. The third regexp does match, since the 252C<$> constrains C<keeper> to match only at the end of the string. 253 254When both C<^> and C<$> are used at the same time, the regexp has to 255match both the beginning and the end of the string, i.e., the regexp 256matches the whole string. Consider 257 258 "keeper" =~ /^keep$/; # doesn't match 259 "keeper" =~ /^keeper$/; # matches 260 "" =~ /^$/; # ^$ matches an empty string 261 262The first regexp doesn't match because the string has more to it than 263C<keep>. Since the second regexp is exactly the string, it 264matches. Using both C<^> and C<$> in a regexp forces the complete 265string to match, so it gives you complete control over which strings 266match and which don't. Suppose you are looking for a fellow named 267bert, off in a string by himself: 268 269 "dogbert" =~ /bert/; # matches, but not what you want 270 271 "dilbert" =~ /^bert/; # doesn't match, but .. 272 "bertram" =~ /^bert/; # matches, so still not good enough 273 274 "bertram" =~ /^bert$/; # doesn't match, good 275 "dilbert" =~ /^bert$/; # doesn't match, good 276 "bert" =~ /^bert$/; # matches, perfect 277 278Of course, in the case of a literal string, one could just as easily 279use the string comparison S<C<$string eq 'bert'>> and it would be 280more efficient. The C<^...$> regexp really becomes useful when we 281add in the more powerful regexp tools below. 282 283=head2 Using character classes 284 285Although one can already do quite a lot with the literal string 286regexps above, we've only scratched the surface of regular expression 287technology. In this and subsequent sections we will introduce regexp 288concepts (and associated metacharacter notations) that will allow a 289regexp to not just represent a single character sequence, but a I<whole 290class> of them. 291 292One such concept is that of a I<character class>. A character class 293allows a set of possible characters, rather than just a single 294character, to match at a particular point in a regexp. Character 295classes are denoted by brackets C<[...]>, with the set of characters 296to be possibly matched inside. Here are some examples: 297 298 /cat/; # matches 'cat' 299 /[bcr]at/; # matches 'bat, 'cat', or 'rat' 300 /item[0123456789]/; # matches 'item0' or ... or 'item9' 301 "abc" =~ /[cab]/; # matches 'a' 302 303In the last statement, even though C<'c'> is the first character in 304the class, C<'a'> matches because the first character position in the 305string is the earliest point at which the regexp can match. 306 307 /[yY][eE][sS]/; # match 'yes' in a case-insensitive way 308 # 'yes', 'Yes', 'YES', etc. 309 310This regexp displays a common task: perform a case-insensitive 311match. Perl provides a way of avoiding all those brackets by simply 312appending an C<'i'> to the end of the match. Then C</[yY][eE][sS]/;> 313can be rewritten as C</yes/i;>. The C<'i'> stands for 314case-insensitive and is an example of a I<modifier> of the matching 315operation. We will meet other modifiers later in the tutorial. 316 317We saw in the section above that there were ordinary characters, which 318represented themselves, and special characters, which needed a 319backslash C<\> to represent themselves. The same is true in a 320character class, but the sets of ordinary and special characters 321inside a character class are different than those outside a character 322class. The special characters for a character class are C<-]\^$> (and 323the pattern delimiter, whatever it is). 324C<]> is special because it denotes the end of a character class. C<$> is 325special because it denotes a scalar variable. C<\> is special because 326it is used in escape sequences, just like above. Here is how the 327special characters C<]$\> are handled: 328 329 /[\]c]def/; # matches ']def' or 'cdef' 330 $x = 'bcr'; 331 /[$x]at/; # matches 'bat', 'cat', or 'rat' 332 /[\$x]at/; # matches '$at' or 'xat' 333 /[\\$x]at/; # matches '\at', 'bat, 'cat', or 'rat' 334 335The last two are a little tricky. In C<[\$x]>, the backslash protects 336the dollar sign, so the character class has two members C<$> and C<x>. 337In C<[\\$x]>, the backslash is protected, so C<$x> is treated as a 338variable and substituted in double quote fashion. 339 340The special character C<'-'> acts as a range operator within character 341classes, so that a contiguous set of characters can be written as a 342range. With ranges, the unwieldy C<[0123456789]> and C<[abc...xyz]> 343become the svelte C<[0-9]> and C<[a-z]>. Some examples are 344 345 /item[0-9]/; # matches 'item0' or ... or 'item9' 346 /[0-9bx-z]aa/; # matches '0aa', ..., '9aa', 347 # 'baa', 'xaa', 'yaa', or 'zaa' 348 /[0-9a-fA-F]/; # matches a hexadecimal digit 349 /[0-9a-zA-Z_]/; # matches a "word" character, 350 # like those in a Perl variable name 351 352If C<'-'> is the first or last character in a character class, it is 353treated as an ordinary character; C<[-ab]>, C<[ab-]> and C<[a\-b]> are 354all equivalent. 355 356The special character C<^> in the first position of a character class 357denotes a I<negated character class>, which matches any character but 358those in the brackets. Both C<[...]> and C<[^...]> must match a 359character, or the match fails. Then 360 361 /[^a]at/; # doesn't match 'aat' or 'at', but matches 362 # all other 'bat', 'cat, '0at', '%at', etc. 363 /[^0-9]/; # matches a non-numeric character 364 /[a^]at/; # matches 'aat' or '^at'; here '^' is ordinary 365 366Now, even C<[0-9]> can be a bother to write multiple times, so in the 367interest of saving keystrokes and making regexps more readable, Perl 368has several abbreviations for common character classes, as shown below. 369Since the introduction of Unicode, these character classes match more 370than just a few characters in the ISO 8859-1 range. 371 372=over 4 373 374=item * 375 376\d matches a digit, not just [0-9] but also digits from non-roman scripts 377 378=item * 379 380\s matches a whitespace character, the set [\ \t\r\n\f] and others 381 382=item * 383 384\w matches a word character (alphanumeric or _), not just [0-9a-zA-Z_] 385but also digits and characters from non-roman scripts 386 387=item * 388 389\D is a negated \d; it represents any other character than a digit, or [^\d] 390 391=item * 392 393\S is a negated \s; it represents any non-whitespace character [^\s] 394 395=item * 396 397\W is a negated \w; it represents any non-word character [^\w] 398 399=item * 400 401The period '.' matches any character but "\n" (unless the modifier C<//s> is 402in effect, as explained below). 403 404=back 405 406The C<\d\s\w\D\S\W> abbreviations can be used both inside and outside 407of character classes. Here are some in use: 408 409 /\d\d:\d\d:\d\d/; # matches a hh:mm:ss time format 410 /[\d\s]/; # matches any digit or whitespace character 411 /\w\W\w/; # matches a word char, followed by a 412 # non-word char, followed by a word char 413 /..rt/; # matches any two chars, followed by 'rt' 414 /end\./; # matches 'end.' 415 /end[.]/; # same thing, matches 'end.' 416 417Because a period is a metacharacter, it needs to be escaped to match 418as an ordinary period. Because, for example, C<\d> and C<\w> are sets 419of characters, it is incorrect to think of C<[^\d\w]> as C<[\D\W]>; in 420fact C<[^\d\w]> is the same as C<[^\w]>, which is the same as 421C<[\W]>. Think DeMorgan's laws. 422 423An anchor useful in basic regexps is the I<word anchor> 424C<\b>. This matches a boundary between a word character and a non-word 425character C<\w\W> or C<\W\w>: 426 427 $x = "Housecat catenates house and cat"; 428 $x =~ /cat/; # matches cat in 'housecat' 429 $x =~ /\bcat/; # matches cat in 'catenates' 430 $x =~ /cat\b/; # matches cat in 'housecat' 431 $x =~ /\bcat\b/; # matches 'cat' at end of string 432 433Note in the last example, the end of the string is considered a word 434boundary. 435 436You might wonder why C<'.'> matches everything but C<"\n"> - why not 437every character? The reason is that often one is matching against 438lines and would like to ignore the newline characters. For instance, 439while the string C<"\n"> represents one line, we would like to think 440of it as empty. Then 441 442 "" =~ /^$/; # matches 443 "\n" =~ /^$/; # matches, $ anchors before "\n" 444 445 "" =~ /./; # doesn't match; it needs a char 446 "" =~ /^.$/; # doesn't match; it needs a char 447 "\n" =~ /^.$/; # doesn't match; it needs a char other than "\n" 448 "a" =~ /^.$/; # matches 449 "a\n" =~ /^.$/; # matches, $ anchors before "\n" 450 451This behavior is convenient, because we usually want to ignore 452newlines when we count and match characters in a line. Sometimes, 453however, we want to keep track of newlines. We might even want C<^> 454and C<$> to anchor at the beginning and end of lines within the 455string, rather than just the beginning and end of the string. Perl 456allows us to choose between ignoring and paying attention to newlines 457by using the C<//s> and C<//m> modifiers. C<//s> and C<//m> stand for 458single line and multi-line and they determine whether a string is to 459be treated as one continuous string, or as a set of lines. The two 460modifiers affect two aspects of how the regexp is interpreted: 1) how 461the C<'.'> character class is defined, and 2) where the anchors C<^> 462and C<$> are able to match. Here are the four possible combinations: 463 464=over 4 465 466=item * 467 468no modifiers (//): Default behavior. C<'.'> matches any character 469except C<"\n">. C<^> matches only at the beginning of the string and 470C<$> matches only at the end or before a newline at the end. 471 472=item * 473 474s modifier (//s): Treat string as a single long line. C<'.'> matches 475any character, even C<"\n">. C<^> matches only at the beginning of 476the string and C<$> matches only at the end or before a newline at the 477end. 478 479=item * 480 481m modifier (//m): Treat string as a set of multiple lines. C<'.'> 482matches any character except C<"\n">. C<^> and C<$> are able to match 483at the start or end of I<any> line within the string. 484 485=item * 486 487both s and m modifiers (//sm): Treat string as a single long line, but 488detect multiple lines. C<'.'> matches any character, even 489C<"\n">. C<^> and C<$>, however, are able to match at the start or end 490of I<any> line within the string. 491 492=back 493 494Here are examples of C<//s> and C<//m> in action: 495 496 $x = "There once was a girl\nWho programmed in Perl\n"; 497 498 $x =~ /^Who/; # doesn't match, "Who" not at start of string 499 $x =~ /^Who/s; # doesn't match, "Who" not at start of string 500 $x =~ /^Who/m; # matches, "Who" at start of second line 501 $x =~ /^Who/sm; # matches, "Who" at start of second line 502 503 $x =~ /girl.Who/; # doesn't match, "." doesn't match "\n" 504 $x =~ /girl.Who/s; # matches, "." matches "\n" 505 $x =~ /girl.Who/m; # doesn't match, "." doesn't match "\n" 506 $x =~ /girl.Who/sm; # matches, "." matches "\n" 507 508Most of the time, the default behavior is what is wanted, but C<//s> and 509C<//m> are occasionally very useful. If C<//m> is being used, the start 510of the string can still be matched with C<\A> and the end of the string 511can still be matched with the anchors C<\Z> (matches both the end and 512the newline before, like C<$>), and C<\z> (matches only the end): 513 514 $x =~ /^Who/m; # matches, "Who" at start of second line 515 $x =~ /\AWho/m; # doesn't match, "Who" is not at start of string 516 517 $x =~ /girl$/m; # matches, "girl" at end of first line 518 $x =~ /girl\Z/m; # doesn't match, "girl" is not at end of string 519 520 $x =~ /Perl\Z/m; # matches, "Perl" is at newline before end 521 $x =~ /Perl\z/m; # doesn't match, "Perl" is not at end of string 522 523We now know how to create choices among classes of characters in a 524regexp. What about choices among words or character strings? Such 525choices are described in the next section. 526 527=head2 Matching this or that 528 529Sometimes we would like our regexp to be able to match different 530possible words or character strings. This is accomplished by using 531the I<alternation> metacharacter C<|>. To match C<dog> or C<cat>, we 532form the regexp C<dog|cat>. As before, Perl will try to match the 533regexp at the earliest possible point in the string. At each 534character position, Perl will first try to match the first 535alternative, C<dog>. If C<dog> doesn't match, Perl will then try the 536next alternative, C<cat>. If C<cat> doesn't match either, then the 537match fails and Perl moves to the next position in the string. Some 538examples: 539 540 "cats and dogs" =~ /cat|dog|bird/; # matches "cat" 541 "cats and dogs" =~ /dog|cat|bird/; # matches "cat" 542 543Even though C<dog> is the first alternative in the second regexp, 544C<cat> is able to match earlier in the string. 545 546 "cats" =~ /c|ca|cat|cats/; # matches "c" 547 "cats" =~ /cats|cat|ca|c/; # matches "cats" 548 549Here, all the alternatives match at the first string position, so the 550first alternative is the one that matches. If some of the 551alternatives are truncations of the others, put the longest ones first 552to give them a chance to match. 553 554 "cab" =~ /a|b|c/ # matches "c" 555 # /a|b|c/ == /[abc]/ 556 557The last example points out that character classes are like 558alternations of characters. At a given character position, the first 559alternative that allows the regexp match to succeed will be the one 560that matches. 561 562=head2 Grouping things and hierarchical matching 563 564Alternation allows a regexp to choose among alternatives, but by 565itself it is unsatisfying. The reason is that each alternative is a whole 566regexp, but sometime we want alternatives for just part of a 567regexp. For instance, suppose we want to search for housecats or 568housekeepers. The regexp C<housecat|housekeeper> fits the bill, but is 569inefficient because we had to type C<house> twice. It would be nice to 570have parts of the regexp be constant, like C<house>, and some 571parts have alternatives, like C<cat|keeper>. 572 573The I<grouping> metacharacters C<()> solve this problem. Grouping 574allows parts of a regexp to be treated as a single unit. Parts of a 575regexp are grouped by enclosing them in parentheses. Thus we could solve 576the C<housecat|housekeeper> by forming the regexp as 577C<house(cat|keeper)>. The regexp C<house(cat|keeper)> means match 578C<house> followed by either C<cat> or C<keeper>. Some more examples 579are 580 581 /(a|b)b/; # matches 'ab' or 'bb' 582 /(ac|b)b/; # matches 'acb' or 'bb' 583 /(^a|b)c/; # matches 'ac' at start of string or 'bc' anywhere 584 /(a|[bc])d/; # matches 'ad', 'bd', or 'cd' 585 586 /house(cat|)/; # matches either 'housecat' or 'house' 587 /house(cat(s|)|)/; # matches either 'housecats' or 'housecat' or 588 # 'house'. Note groups can be nested. 589 590 /(19|20|)\d\d/; # match years 19xx, 20xx, or the Y2K problem, xx 591 "20" =~ /(19|20|)\d\d/; # matches the null alternative '()\d\d', 592 # because '20\d\d' can't match 593 594Alternations behave the same way in groups as out of them: at a given 595string position, the leftmost alternative that allows the regexp to 596match is taken. So in the last example at the first string position, 597C<"20"> matches the second alternative, but there is nothing left over 598to match the next two digits C<\d\d>. So Perl moves on to the next 599alternative, which is the null alternative and that works, since 600C<"20"> is two digits. 601 602The process of trying one alternative, seeing if it matches, and 603moving on to the next alternative, while going back in the string 604from where the previous alternative was tried, if it doesn't, is called 605I<backtracking>. The term 'backtracking' comes from the idea that 606matching a regexp is like a walk in the woods. Successfully matching 607a regexp is like arriving at a destination. There are many possible 608trailheads, one for each string position, and each one is tried in 609order, left to right. From each trailhead there may be many paths, 610some of which get you there, and some which are dead ends. When you 611walk along a trail and hit a dead end, you have to backtrack along the 612trail to an earlier point to try another trail. If you hit your 613destination, you stop immediately and forget about trying all the 614other trails. You are persistent, and only if you have tried all the 615trails from all the trailheads and not arrived at your destination, do 616you declare failure. To be concrete, here is a step-by-step analysis 617of what Perl does when it tries to match the regexp 618 619 "abcde" =~ /(abd|abc)(df|d|de)/; 620 621=over 4 622 623=item 0 624 625Start with the first letter in the string 'a'. 626 627=item 1 628 629Try the first alternative in the first group 'abd'. 630 631=item 2 632 633Match 'a' followed by 'b'. So far so good. 634 635=item 3 636 637'd' in the regexp doesn't match 'c' in the string - a dead 638end. So backtrack two characters and pick the second alternative in 639the first group 'abc'. 640 641=item 4 642 643Match 'a' followed by 'b' followed by 'c'. We are on a roll 644and have satisfied the first group. Set $1 to 'abc'. 645 646=item 5 647 648Move on to the second group and pick the first alternative 649'df'. 650 651=item 6 652 653Match the 'd'. 654 655=item 7 656 657'f' in the regexp doesn't match 'e' in the string, so a dead 658end. Backtrack one character and pick the second alternative in the 659second group 'd'. 660 661=item 8 662 663'd' matches. The second grouping is satisfied, so set $2 to 664'd'. 665 666=item 9 667 668We are at the end of the regexp, so we are done! We have 669matched 'abcd' out of the string "abcde". 670 671=back 672 673There are a couple of things to note about this analysis. First, the 674third alternative in the second group 'de' also allows a match, but we 675stopped before we got to it - at a given character position, leftmost 676wins. Second, we were able to get a match at the first character 677position of the string 'a'. If there were no matches at the first 678position, Perl would move to the second character position 'b' and 679attempt the match all over again. Only when all possible paths at all 680possible character positions have been exhausted does Perl give 681up and declare S<C<$string =~ /(abd|abc)(df|d|de)/;>> to be false. 682 683Even with all this work, regexp matching happens remarkably fast. To 684speed things up, Perl compiles the regexp into a compact sequence of 685opcodes that can often fit inside a processor cache. When the code is 686executed, these opcodes can then run at full throttle and search very 687quickly. 688 689=head2 Extracting matches 690 691The grouping metacharacters C<()> also serve another completely 692different function: they allow the extraction of the parts of a string 693that matched. This is very useful to find out what matched and for 694text processing in general. For each grouping, the part that matched 695inside goes into the special variables C<$1>, C<$2>, etc. They can be 696used just as ordinary variables: 697 698 # extract hours, minutes, seconds 699 if ($time =~ /(\d\d):(\d\d):(\d\d)/) { # match hh:mm:ss format 700 $hours = $1; 701 $minutes = $2; 702 $seconds = $3; 703 } 704 705Now, we know that in scalar context, 706S<C<$time =~ /(\d\d):(\d\d):(\d\d)/>> returns a true or false 707value. In list context, however, it returns the list of matched values 708C<($1,$2,$3)>. So we could write the code more compactly as 709 710 # extract hours, minutes, seconds 711 ($hours, $minutes, $second) = ($time =~ /(\d\d):(\d\d):(\d\d)/); 712 713If the groupings in a regexp are nested, C<$1> gets the group with the 714leftmost opening parenthesis, C<$2> the next opening parenthesis, 715etc. Here is a regexp with nested groups: 716 717 /(ab(cd|ef)((gi)|j))/; 718 1 2 34 719 720If this regexp matches, C<$1> contains a string starting with 721C<'ab'>, C<$2> is either set to C<'cd'> or C<'ef'>, C<$3> equals either 722C<'gi'> or C<'j'>, and C<$4> is either set to C<'gi'>, just like C<$3>, 723or it remains undefined. 724 725For convenience, Perl sets C<$+> to the string held by the highest numbered 726C<$1>, C<$2>,... that got assigned (and, somewhat related, C<$^N> to the 727value of the C<$1>, C<$2>,... most-recently assigned; i.e. the C<$1>, 728C<$2>,... associated with the rightmost closing parenthesis used in the 729match). 730 731 732=head2 Backreferences 733 734Closely associated with the matching variables C<$1>, C<$2>, ... are 735the I<backreferences> C<\1>, C<\2>,... Backreferences are simply 736matching variables that can be used I<inside> a regexp. This is a 737really nice feature; what matches later in a regexp is made to depend on 738what matched earlier in the regexp. Suppose we wanted to look 739for doubled words in a text, like 'the the'. The following regexp finds 740all 3-letter doubles with a space in between: 741 742 /\b(\w\w\w)\s\1\b/; 743 744The grouping assigns a value to \1, so that the same 3 letter sequence 745is used for both parts. 746 747A similar task is to find words consisting of two identical parts: 748 749 % simple_grep '^(\w\w\w\w|\w\w\w|\w\w|\w)\1$' /usr/dict/words 750 beriberi 751 booboo 752 coco 753 mama 754 murmur 755 papa 756 757The regexp has a single grouping which considers 4-letter 758combinations, then 3-letter combinations, etc., and uses C<\1> to look for 759a repeat. Although C<$1> and C<\1> represent the same thing, care should be 760taken to use matched variables C<$1>, C<$2>,... only I<outside> a regexp 761and backreferences C<\1>, C<\2>,... only I<inside> a regexp; not doing 762so may lead to surprising and unsatisfactory results. 763 764 765=head2 Relative backreferences 766 767Counting the opening parentheses to get the correct number for a 768backreference is errorprone as soon as there is more than one 769capturing group. A more convenient technique became available 770with Perl 5.10: relative backreferences. To refer to the immediately 771preceding capture group one now may write C<\g{-1}>, the next but 772last is available via C<\g{-2}>, and so on. 773 774Another good reason in addition to readability and maintainability 775for using relative backreferences is illustrated by the following example, 776where a simple pattern for matching peculiar strings is used: 777 778 $a99a = '([a-z])(\d)\2\1'; # matches a11a, g22g, x33x, etc. 779 780Now that we have this pattern stored as a handy string, we might feel 781tempted to use it as a part of some other pattern: 782 783 $line = "code=e99e"; 784 if ($line =~ /^(\w+)=$a99a$/){ # unexpected behavior! 785 print "$1 is valid\n"; 786 } else { 787 print "bad line: '$line'\n"; 788 } 789 790But this doesn't match, at least not the way one might expect. Only 791after inserting the interpolated C<$a99a> and looking at the resulting 792full text of the regexp is it obvious that the backreferences have 793backfired. The subexpression C<(\w+)> has snatched number 1 and 794demoted the groups in C<$a99a> by one rank. This can be avoided by 795using relative backreferences: 796 797 $a99a = '([a-z])(\d)\g{-1}\g{-2}'; # safe for being interpolated 798 799 800=head2 Named backreferences 801 802Perl 5.10 also introduced named capture buffers and named backreferences. 803To attach a name to a capturing group, you write either 804C<< (?<name>...) >> or C<< (?'name'...) >>. The backreference may 805then be written as C<\g{name}>. It is permissible to attach the 806same name to more than one group, but then only the leftmost one of the 807eponymous set can be referenced. Outside of the pattern a named 808capture buffer is accessible through the C<%+> hash. 809 810Assuming that we have to match calendar dates which may be given in one 811of the three formats yyyy-mm-dd, mm/dd/yyyy or dd.mm.yyyy, we can write 812three suitable patterns where we use 'd', 'm' and 'y' respectively as the 813names of the buffers capturing the pertaining components of a date. The 814matching operation combines the three patterns as alternatives: 815 816 $fmt1 = '(?<y>\d\d\d\d)-(?<m>\d\d)-(?<d>\d\d)'; 817 $fmt2 = '(?<m>\d\d)/(?<d>\d\d)/(?<y>\d\d\d\d)'; 818 $fmt3 = '(?<d>\d\d)\.(?<m>\d\d)\.(?<y>\d\d\d\d)'; 819 for my $d qw( 2006-10-21 15.01.2007 10/31/2005 ){ 820 if ( $d =~ m{$fmt1|$fmt2|$fmt3} ){ 821 print "day=$+{d} month=$+{m} year=$+{y}\n"; 822 } 823 } 824 825If any of the alternatives matches, the hash C<%+> is bound to contain the 826three key-value pairs. 827 828 829=head2 Alternative capture group numbering 830 831Yet another capturing group numbering technique (also as from Perl 5.10) 832deals with the problem of referring to groups within a set of alternatives. 833Consider a pattern for matching a time of the day, civil or military style: 834 835 if ( $time =~ /(\d\d|\d):(\d\d)|(\d\d)(\d\d)/ ){ 836 # process hour and minute 837 } 838 839Processing the results requires an additional if statement to determine 840whether C<$1> and C<$2> or C<$3> and C<$4> contain the goodies. It would 841be easier if we could use buffer numbers 1 and 2 in second alternative as 842well, and this is exactly what the parenthesized construct C<(?|...)>, 843set around an alternative achieves. Here is an extended version of the 844previous pattern: 845 846 if ( $time =~ /(?|(\d\d|\d):(\d\d)|(\d\d)(\d\d))\s+([A-Z][A-Z][A-Z])/ ){ 847 print "hour=$1 minute=$2 zone=$3\n"; 848 } 849 850Within the alternative numbering group, buffer numbers start at the same 851position for each alternative. After the group, numbering continues 852with one higher than the maximum reached across all the alternatives. 853 854=head2 Position information 855 856In addition to what was matched, Perl (since 5.6.0) also provides the 857positions of what was matched as contents of the C<@-> and C<@+> 858arrays. C<$-[0]> is the position of the start of the entire match and 859C<$+[0]> is the position of the end. Similarly, C<$-[n]> is the 860position of the start of the C<$n> match and C<$+[n]> is the position 861of the end. If C<$n> is undefined, so are C<$-[n]> and C<$+[n]>. Then 862this code 863 864 $x = "Mmm...donut, thought Homer"; 865 $x =~ /^(Mmm|Yech)\.\.\.(donut|peas)/; # matches 866 foreach $expr (1..$#-) { 867 print "Match $expr: '${$expr}' at position ($-[$expr],$+[$expr])\n"; 868 } 869 870prints 871 872 Match 1: 'Mmm' at position (0,3) 873 Match 2: 'donut' at position (6,11) 874 875Even if there are no groupings in a regexp, it is still possible to 876find out what exactly matched in a string. If you use them, Perl 877will set C<$`> to the part of the string before the match, will set C<$&> 878to the part of the string that matched, and will set C<$'> to the part 879of the string after the match. An example: 880 881 $x = "the cat caught the mouse"; 882 $x =~ /cat/; # $` = 'the ', $& = 'cat', $' = ' caught the mouse' 883 $x =~ /the/; # $` = '', $& = 'the', $' = ' cat caught the mouse' 884 885In the second match, C<$`> equals C<''> because the regexp matched at the 886first character position in the string and stopped; it never saw the 887second 'the'. It is important to note that using C<$`> and C<$'> 888slows down regexp matching quite a bit, while C<$&> slows it down to a 889lesser extent, because if they are used in one regexp in a program, 890they are generated for I<all> regexps in the program. So if raw 891performance is a goal of your application, they should be avoided. 892If you need to extract the corresponding substrings, use C<@-> and 893C<@+> instead: 894 895 $` is the same as substr( $x, 0, $-[0] ) 896 $& is the same as substr( $x, $-[0], $+[0]-$-[0] ) 897 $' is the same as substr( $x, $+[0] ) 898 899 900=head2 Non-capturing groupings 901 902A group that is required to bundle a set of alternatives may or may not be 903useful as a capturing group. If it isn't, it just creates a superfluous 904addition to the set of available capture buffer values, inside as well as 905outside the regexp. Non-capturing groupings, denoted by C<(?:regexp)>, 906still allow the regexp to be treated as a single unit, but don't establish 907a capturing buffer at the same time. Both capturing and non-capturing 908groupings are allowed to co-exist in the same regexp. Because there is 909no extraction, non-capturing groupings are faster than capturing 910groupings. Non-capturing groupings are also handy for choosing exactly 911which parts of a regexp are to be extracted to matching variables: 912 913 # match a number, $1-$4 are set, but we only want $1 914 /([+-]?\ *(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?)/; 915 916 # match a number faster , only $1 is set 917 /([+-]?\ *(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?)/; 918 919 # match a number, get $1 = whole number, $2 = exponent 920 /([+-]?\ *(?:\d+(?:\.\d*)?|\.\d+)(?:[eE]([+-]?\d+))?)/; 921 922Non-capturing groupings are also useful for removing nuisance 923elements gathered from a split operation where parentheses are 924required for some reason: 925 926 $x = '12aba34ba5'; 927 @num = split /(a|b)+/, $x; # @num = ('12','a','34','b','5') 928 @num = split /(?:a|b)+/, $x; # @num = ('12','34','5') 929 930 931=head2 Matching repetitions 932 933The examples in the previous section display an annoying weakness. We 934were only matching 3-letter words, or chunks of words of 4 letters or 935less. We'd like to be able to match words or, more generally, strings 936of any length, without writing out tedious alternatives like 937C<\w\w\w\w|\w\w\w|\w\w|\w>. 938 939This is exactly the problem the I<quantifier> metacharacters C<?>, 940C<*>, C<+>, and C<{}> were created for. They allow us to delimit the 941number of repeats for a portion of a regexp we consider to be a 942match. Quantifiers are put immediately after the character, character 943class, or grouping that we want to specify. They have the following 944meanings: 945 946=over 4 947 948=item * 949 950C<a?> means: match 'a' 1 or 0 times 951 952=item * 953 954C<a*> means: match 'a' 0 or more times, i.e., any number of times 955 956=item * 957 958C<a+> means: match 'a' 1 or more times, i.e., at least once 959 960=item * 961 962C<a{n,m}> means: match at least C<n> times, but not more than C<m> 963times. 964 965=item * 966 967C<a{n,}> means: match at least C<n> or more times 968 969=item * 970 971C<a{n}> means: match exactly C<n> times 972 973=back 974 975Here are some examples: 976 977 /[a-z]+\s+\d*/; # match a lowercase word, at least one space, and 978 # any number of digits 979 /(\w+)\s+\1/; # match doubled words of arbitrary length 980 /y(es)?/i; # matches 'y', 'Y', or a case-insensitive 'yes' 981 $year =~ /\d{2,4}/; # make sure year is at least 2 but not more 982 # than 4 digits 983 $year =~ /\d{4}|\d{2}/; # better match; throw out 3 digit dates 984 $year =~ /\d{2}(\d{2})?/; # same thing written differently. However, 985 # this produces $1 and the other does not. 986 987 % simple_grep '^(\w+)\1$' /usr/dict/words # isn't this easier? 988 beriberi 989 booboo 990 coco 991 mama 992 murmur 993 papa 994 995For all of these quantifiers, Perl will try to match as much of the 996string as possible, while still allowing the regexp to succeed. Thus 997with C</a?.../>, Perl will first try to match the regexp with the C<a> 998present; if that fails, Perl will try to match the regexp without the 999C<a> present. For the quantifier C<*>, we get the following: 1000 1001 $x = "the cat in the hat"; 1002 $x =~ /^(.*)(cat)(.*)$/; # matches, 1003 # $1 = 'the ' 1004 # $2 = 'cat' 1005 # $3 = ' in the hat' 1006 1007Which is what we might expect, the match finds the only C<cat> in the 1008string and locks onto it. Consider, however, this regexp: 1009 1010 $x =~ /^(.*)(at)(.*)$/; # matches, 1011 # $1 = 'the cat in the h' 1012 # $2 = 'at' 1013 # $3 = '' (0 characters match) 1014 1015One might initially guess that Perl would find the C<at> in C<cat> and 1016stop there, but that wouldn't give the longest possible string to the 1017first quantifier C<.*>. Instead, the first quantifier C<.*> grabs as 1018much of the string as possible while still having the regexp match. In 1019this example, that means having the C<at> sequence with the final C<at> 1020in the string. The other important principle illustrated here is that 1021when there are two or more elements in a regexp, the I<leftmost> 1022quantifier, if there is one, gets to grab as much the string as 1023possible, leaving the rest of the regexp to fight over scraps. Thus in 1024our example, the first quantifier C<.*> grabs most of the string, while 1025the second quantifier C<.*> gets the empty string. Quantifiers that 1026grab as much of the string as possible are called I<maximal match> or 1027I<greedy> quantifiers. 1028 1029When a regexp can match a string in several different ways, we can use 1030the principles above to predict which way the regexp will match: 1031 1032=over 4 1033 1034=item * 1035 1036Principle 0: Taken as a whole, any regexp will be matched at the 1037earliest possible position in the string. 1038 1039=item * 1040 1041Principle 1: In an alternation C<a|b|c...>, the leftmost alternative 1042that allows a match for the whole regexp will be the one used. 1043 1044=item * 1045 1046Principle 2: The maximal matching quantifiers C<?>, C<*>, C<+> and 1047C<{n,m}> will in general match as much of the string as possible while 1048still allowing the whole regexp to match. 1049 1050=item * 1051 1052Principle 3: If there are two or more elements in a regexp, the 1053leftmost greedy quantifier, if any, will match as much of the string 1054as possible while still allowing the whole regexp to match. The next 1055leftmost greedy quantifier, if any, will try to match as much of the 1056string remaining available to it as possible, while still allowing the 1057whole regexp to match. And so on, until all the regexp elements are 1058satisfied. 1059 1060=back 1061 1062As we have seen above, Principle 0 overrides the others. The regexp 1063will be matched as early as possible, with the other principles 1064determining how the regexp matches at that earliest character 1065position. 1066 1067Here is an example of these principles in action: 1068 1069 $x = "The programming republic of Perl"; 1070 $x =~ /^(.+)(e|r)(.*)$/; # matches, 1071 # $1 = 'The programming republic of Pe' 1072 # $2 = 'r' 1073 # $3 = 'l' 1074 1075This regexp matches at the earliest string position, C<'T'>. One 1076might think that C<e>, being leftmost in the alternation, would be 1077matched, but C<r> produces the longest string in the first quantifier. 1078 1079 $x =~ /(m{1,2})(.*)$/; # matches, 1080 # $1 = 'mm' 1081 # $2 = 'ing republic of Perl' 1082 1083Here, The earliest possible match is at the first C<'m'> in 1084C<programming>. C<m{1,2}> is the first quantifier, so it gets to match 1085a maximal C<mm>. 1086 1087 $x =~ /.*(m{1,2})(.*)$/; # matches, 1088 # $1 = 'm' 1089 # $2 = 'ing republic of Perl' 1090 1091Here, the regexp matches at the start of the string. The first 1092quantifier C<.*> grabs as much as possible, leaving just a single 1093C<'m'> for the second quantifier C<m{1,2}>. 1094 1095 $x =~ /(.?)(m{1,2})(.*)$/; # matches, 1096 # $1 = 'a' 1097 # $2 = 'mm' 1098 # $3 = 'ing republic of Perl' 1099 1100Here, C<.?> eats its maximal one character at the earliest possible 1101position in the string, C<'a'> in C<programming>, leaving C<m{1,2}> 1102the opportunity to match both C<m>'s. Finally, 1103 1104 "aXXXb" =~ /(X*)/; # matches with $1 = '' 1105 1106because it can match zero copies of C<'X'> at the beginning of the 1107string. If you definitely want to match at least one C<'X'>, use 1108C<X+>, not C<X*>. 1109 1110Sometimes greed is not good. At times, we would like quantifiers to 1111match a I<minimal> piece of string, rather than a maximal piece. For 1112this purpose, Larry Wall created the I<minimal match> or 1113I<non-greedy> quantifiers C<??>, C<*?>, C<+?>, and C<{}?>. These are 1114the usual quantifiers with a C<?> appended to them. They have the 1115following meanings: 1116 1117=over 4 1118 1119=item * 1120 1121C<a??> means: match 'a' 0 or 1 times. Try 0 first, then 1. 1122 1123=item * 1124 1125C<a*?> means: match 'a' 0 or more times, i.e., any number of times, 1126but as few times as possible 1127 1128=item * 1129 1130C<a+?> means: match 'a' 1 or more times, i.e., at least once, but 1131as few times as possible 1132 1133=item * 1134 1135C<a{n,m}?> means: match at least C<n> times, not more than C<m> 1136times, as few times as possible 1137 1138=item * 1139 1140C<a{n,}?> means: match at least C<n> times, but as few times as 1141possible 1142 1143=item * 1144 1145C<a{n}?> means: match exactly C<n> times. Because we match exactly 1146C<n> times, C<a{n}?> is equivalent to C<a{n}> and is just there for 1147notational consistency. 1148 1149=back 1150 1151Let's look at the example above, but with minimal quantifiers: 1152 1153 $x = "The programming republic of Perl"; 1154 $x =~ /^(.+?)(e|r)(.*)$/; # matches, 1155 # $1 = 'Th' 1156 # $2 = 'e' 1157 # $3 = ' programming republic of Perl' 1158 1159The minimal string that will allow both the start of the string C<^> 1160and the alternation to match is C<Th>, with the alternation C<e|r> 1161matching C<e>. The second quantifier C<.*> is free to gobble up the 1162rest of the string. 1163 1164 $x =~ /(m{1,2}?)(.*?)$/; # matches, 1165 # $1 = 'm' 1166 # $2 = 'ming republic of Perl' 1167 1168The first string position that this regexp can match is at the first 1169C<'m'> in C<programming>. At this position, the minimal C<m{1,2}?> 1170matches just one C<'m'>. Although the second quantifier C<.*?> would 1171prefer to match no characters, it is constrained by the end-of-string 1172anchor C<$> to match the rest of the string. 1173 1174 $x =~ /(.*?)(m{1,2}?)(.*)$/; # matches, 1175 # $1 = 'The progra' 1176 # $2 = 'm' 1177 # $3 = 'ming republic of Perl' 1178 1179In this regexp, you might expect the first minimal quantifier C<.*?> 1180to match the empty string, because it is not constrained by a C<^> 1181anchor to match the beginning of the word. Principle 0 applies here, 1182however. Because it is possible for the whole regexp to match at the 1183start of the string, it I<will> match at the start of the string. Thus 1184the first quantifier has to match everything up to the first C<m>. The 1185second minimal quantifier matches just one C<m> and the third 1186quantifier matches the rest of the string. 1187 1188 $x =~ /(.??)(m{1,2})(.*)$/; # matches, 1189 # $1 = 'a' 1190 # $2 = 'mm' 1191 # $3 = 'ing republic of Perl' 1192 1193Just as in the previous regexp, the first quantifier C<.??> can match 1194earliest at position C<'a'>, so it does. The second quantifier is 1195greedy, so it matches C<mm>, and the third matches the rest of the 1196string. 1197 1198We can modify principle 3 above to take into account non-greedy 1199quantifiers: 1200 1201=over 4 1202 1203=item * 1204 1205Principle 3: If there are two or more elements in a regexp, the 1206leftmost greedy (non-greedy) quantifier, if any, will match as much 1207(little) of the string as possible while still allowing the whole 1208regexp to match. The next leftmost greedy (non-greedy) quantifier, if 1209any, will try to match as much (little) of the string remaining 1210available to it as possible, while still allowing the whole regexp to 1211match. And so on, until all the regexp elements are satisfied. 1212 1213=back 1214 1215Just like alternation, quantifiers are also susceptible to 1216backtracking. Here is a step-by-step analysis of the example 1217 1218 $x = "the cat in the hat"; 1219 $x =~ /^(.*)(at)(.*)$/; # matches, 1220 # $1 = 'the cat in the h' 1221 # $2 = 'at' 1222 # $3 = '' (0 matches) 1223 1224=over 4 1225 1226=item 0 1227 1228Start with the first letter in the string 't'. 1229 1230=item 1 1231 1232The first quantifier '.*' starts out by matching the whole 1233string 'the cat in the hat'. 1234 1235=item 2 1236 1237'a' in the regexp element 'at' doesn't match the end of the 1238string. Backtrack one character. 1239 1240=item 3 1241 1242'a' in the regexp element 'at' still doesn't match the last 1243letter of the string 't', so backtrack one more character. 1244 1245=item 4 1246 1247Now we can match the 'a' and the 't'. 1248 1249=item 5 1250 1251Move on to the third element '.*'. Since we are at the end of 1252the string and '.*' can match 0 times, assign it the empty string. 1253 1254=item 6 1255 1256We are done! 1257 1258=back 1259 1260Most of the time, all this moving forward and backtracking happens 1261quickly and searching is fast. There are some pathological regexps, 1262however, whose execution time exponentially grows with the size of the 1263string. A typical structure that blows up in your face is of the form 1264 1265 /(a|b+)*/; 1266 1267The problem is the nested indeterminate quantifiers. There are many 1268different ways of partitioning a string of length n between the C<+> 1269and C<*>: one repetition with C<b+> of length n, two repetitions with 1270the first C<b+> length k and the second with length n-k, m repetitions 1271whose bits add up to length n, etc. In fact there are an exponential 1272number of ways to partition a string as a function of its length. A 1273regexp may get lucky and match early in the process, but if there is 1274no match, Perl will try I<every> possibility before giving up. So be 1275careful with nested C<*>'s, C<{n,m}>'s, and C<+>'s. The book 1276I<Mastering Regular Expressions> by Jeffrey Friedl gives a wonderful 1277discussion of this and other efficiency issues. 1278 1279 1280=head2 Possessive quantifiers 1281 1282Backtracking during the relentless search for a match may be a waste 1283of time, particularly when the match is bound to fail. Consider 1284the simple pattern 1285 1286 /^\w+\s+\w+$/; # a word, spaces, a word 1287 1288Whenever this is applied to a string which doesn't quite meet the 1289pattern's expectations such as S<C<"abc ">> or S<C<"abc def ">>, 1290the regex engine will backtrack, approximately once for each character 1291in the string. But we know that there is no way around taking I<all> 1292of the initial word characters to match the first repetition, that I<all> 1293spaces must be eaten by the middle part, and the same goes for the second 1294word. 1295 1296With the introduction of the I<possessive quantifiers> in Perl 5.10, we 1297have a way of instructing the regex engine not to backtrack, with the 1298usual quantifiers with a C<+> appended to them. This makes them greedy as 1299well as stingy; once they succeed they won't give anything back to permit 1300another solution. They have the following meanings: 1301 1302=over 4 1303 1304=item * 1305 1306C<a{n,m}+> means: match at least C<n> times, not more than C<m> times, 1307as many times as possible, and don't give anything up. C<a?+> is short 1308for C<a{0,1}+> 1309 1310=item * 1311 1312C<a{n,}+> means: match at least C<n> times, but as many times as possible, 1313and don't give anything up. C<a*+> is short for C<a{0,}+> and C<a++> is 1314short for C<a{1,}+>. 1315 1316=item * 1317 1318C<a{n}+> means: match exactly C<n> times. It is just there for 1319notational consistency. 1320 1321=back 1322 1323These possessive quantifiers represent a special case of a more general 1324concept, the I<independent subexpression>, see below. 1325 1326As an example where a possessive quantifier is suitable we consider 1327matching a quoted string, as it appears in several programming languages. 1328The backslash is used as an escape character that indicates that the 1329next character is to be taken literally, as another character for the 1330string. Therefore, after the opening quote, we expect a (possibly 1331empty) sequence of alternatives: either some character except an 1332unescaped quote or backslash or an escaped character. 1333 1334 /"(?:[^"\\]++|\\.)*+"/; 1335 1336 1337=head2 Building a regexp 1338 1339At this point, we have all the basic regexp concepts covered, so let's 1340give a more involved example of a regular expression. We will build a 1341regexp that matches numbers. 1342 1343The first task in building a regexp is to decide what we want to match 1344and what we want to exclude. In our case, we want to match both 1345integers and floating point numbers and we want to reject any string 1346that isn't a number. 1347 1348The next task is to break the problem down into smaller problems that 1349are easily converted into a regexp. 1350 1351The simplest case is integers. These consist of a sequence of digits, 1352with an optional sign in front. The digits we can represent with 1353C<\d+> and the sign can be matched with C<[+-]>. Thus the integer 1354regexp is 1355 1356 /[+-]?\d+/; # matches integers 1357 1358A floating point number potentially has a sign, an integral part, a 1359decimal point, a fractional part, and an exponent. One or more of these 1360parts is optional, so we need to check out the different 1361possibilities. Floating point numbers which are in proper form include 1362123., 0.345, .34, -1e6, and 25.4E-72. As with integers, the sign out 1363front is completely optional and can be matched by C<[+-]?>. We can 1364see that if there is no exponent, floating point numbers must have a 1365decimal point, otherwise they are integers. We might be tempted to 1366model these with C<\d*\.\d*>, but this would also match just a single 1367decimal point, which is not a number. So the three cases of floating 1368point number without exponent are 1369 1370 /[+-]?\d+\./; # 1., 321., etc. 1371 /[+-]?\.\d+/; # .1, .234, etc. 1372 /[+-]?\d+\.\d+/; # 1.0, 30.56, etc. 1373 1374These can be combined into a single regexp with a three-way alternation: 1375 1376 /[+-]?(\d+\.\d+|\d+\.|\.\d+)/; # floating point, no exponent 1377 1378In this alternation, it is important to put C<'\d+\.\d+'> before 1379C<'\d+\.'>. If C<'\d+\.'> were first, the regexp would happily match that 1380and ignore the fractional part of the number. 1381 1382Now consider floating point numbers with exponents. The key 1383observation here is that I<both> integers and numbers with decimal 1384points are allowed in front of an exponent. Then exponents, like the 1385overall sign, are independent of whether we are matching numbers with 1386or without decimal points, and can be 'decoupled' from the 1387mantissa. The overall form of the regexp now becomes clear: 1388 1389 /^(optional sign)(integer | f.p. mantissa)(optional exponent)$/; 1390 1391The exponent is an C<e> or C<E>, followed by an integer. So the 1392exponent regexp is 1393 1394 /[eE][+-]?\d+/; # exponent 1395 1396Putting all the parts together, we get a regexp that matches numbers: 1397 1398 /^[+-]?(\d+\.\d+|\d+\.|\.\d+|\d+)([eE][+-]?\d+)?$/; # Ta da! 1399 1400Long regexps like this may impress your friends, but can be hard to 1401decipher. In complex situations like this, the C<//x> modifier for a 1402match is invaluable. It allows one to put nearly arbitrary whitespace 1403and comments into a regexp without affecting their meaning. Using it, 1404we can rewrite our 'extended' regexp in the more pleasing form 1405 1406 /^ 1407 [+-]? # first, match an optional sign 1408 ( # then match integers or f.p. mantissas: 1409 \d+\.\d+ # mantissa of the form a.b 1410 |\d+\. # mantissa of the form a. 1411 |\.\d+ # mantissa of the form .b 1412 |\d+ # integer of the form a 1413 ) 1414 ([eE][+-]?\d+)? # finally, optionally match an exponent 1415 $/x; 1416 1417If whitespace is mostly irrelevant, how does one include space 1418characters in an extended regexp? The answer is to backslash it 1419S<C<'\ '>> or put it in a character class S<C<[ ]>>. The same thing 1420goes for pound signs, use C<\#> or C<[#]>. For instance, Perl allows 1421a space between the sign and the mantissa or integer, and we could add 1422this to our regexp as follows: 1423 1424 /^ 1425 [+-]?\ * # first, match an optional sign *and space* 1426 ( # then match integers or f.p. mantissas: 1427 \d+\.\d+ # mantissa of the form a.b 1428 |\d+\. # mantissa of the form a. 1429 |\.\d+ # mantissa of the form .b 1430 |\d+ # integer of the form a 1431 ) 1432 ([eE][+-]?\d+)? # finally, optionally match an exponent 1433 $/x; 1434 1435In this form, it is easier to see a way to simplify the 1436alternation. Alternatives 1, 2, and 4 all start with C<\d+>, so it 1437could be factored out: 1438 1439 /^ 1440 [+-]?\ * # first, match an optional sign 1441 ( # then match integers or f.p. mantissas: 1442 \d+ # start out with a ... 1443 ( 1444 \.\d* # mantissa of the form a.b or a. 1445 )? # ? takes care of integers of the form a 1446 |\.\d+ # mantissa of the form .b 1447 ) 1448 ([eE][+-]?\d+)? # finally, optionally match an exponent 1449 $/x; 1450 1451or written in the compact form, 1452 1453 /^[+-]?\ *(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?$/; 1454 1455This is our final regexp. To recap, we built a regexp by 1456 1457=over 4 1458 1459=item * 1460 1461specifying the task in detail, 1462 1463=item * 1464 1465breaking down the problem into smaller parts, 1466 1467=item * 1468 1469translating the small parts into regexps, 1470 1471=item * 1472 1473combining the regexps, 1474 1475=item * 1476 1477and optimizing the final combined regexp. 1478 1479=back 1480 1481These are also the typical steps involved in writing a computer 1482program. This makes perfect sense, because regular expressions are 1483essentially programs written in a little computer language that specifies 1484patterns. 1485 1486=head2 Using regular expressions in Perl 1487 1488The last topic of Part 1 briefly covers how regexps are used in Perl 1489programs. Where do they fit into Perl syntax? 1490 1491We have already introduced the matching operator in its default 1492C</regexp/> and arbitrary delimiter C<m!regexp!> forms. We have used 1493the binding operator C<=~> and its negation C<!~> to test for string 1494matches. Associated with the matching operator, we have discussed the 1495single line C<//s>, multi-line C<//m>, case-insensitive C<//i> and 1496extended C<//x> modifiers. There are a few more things you might 1497want to know about matching operators. 1498 1499=head3 Optimizing pattern evaluation 1500 1501We pointed out earlier that variables in regexps are substituted 1502before the regexp is evaluated: 1503 1504 $pattern = 'Seuss'; 1505 while (<>) { 1506 print if /$pattern/; 1507 } 1508 1509This will print any lines containing the word C<Seuss>. It is not as 1510efficient as it could be, however, because Perl has to re-evaluate 1511(or compile) C<$pattern> each time through the loop. If C<$pattern> won't be 1512changing over the lifetime of the script, we can add the C<//o> 1513modifier, which directs Perl to only perform variable substitutions 1514once: 1515 1516 #!/usr/bin/perl 1517 # Improved simple_grep 1518 $regexp = shift; 1519 while (<>) { 1520 print if /$regexp/o; # a good deal faster 1521 } 1522 1523 1524=head3 Prohibiting substitution 1525 1526If you change C<$pattern> after the first substitution happens, Perl 1527will ignore it. If you don't want any substitutions at all, use the 1528special delimiter C<m''>: 1529 1530 @pattern = ('Seuss'); 1531 while (<>) { 1532 print if m'@pattern'; # matches literal '@pattern', not 'Seuss' 1533 } 1534 1535Similar to strings, C<m''> acts like apostrophes on a regexp; all other 1536C<m> delimiters act like quotes. If the regexp evaluates to the empty string, 1537the regexp in the I<last successful match> is used instead. So we have 1538 1539 "dog" =~ /d/; # 'd' matches 1540 "dogbert =~ //; # this matches the 'd' regexp used before 1541 1542 1543=head3 Global matching 1544 1545The final two modifiers C<//g> and C<//c> concern multiple matches. 1546The modifier C<//g> stands for global matching and allows the 1547matching operator to match within a string as many times as possible. 1548In scalar context, successive invocations against a string will have 1549`C<//g> jump from match to match, keeping track of position in the 1550string as it goes along. You can get or set the position with the 1551C<pos()> function. 1552 1553The use of C<//g> is shown in the following example. Suppose we have 1554a string that consists of words separated by spaces. If we know how 1555many words there are in advance, we could extract the words using 1556groupings: 1557 1558 $x = "cat dog house"; # 3 words 1559 $x =~ /^\s*(\w+)\s+(\w+)\s+(\w+)\s*$/; # matches, 1560 # $1 = 'cat' 1561 # $2 = 'dog' 1562 # $3 = 'house' 1563 1564But what if we had an indeterminate number of words? This is the sort 1565of task C<//g> was made for. To extract all words, form the simple 1566regexp C<(\w+)> and loop over all matches with C</(\w+)/g>: 1567 1568 while ($x =~ /(\w+)/g) { 1569 print "Word is $1, ends at position ", pos $x, "\n"; 1570 } 1571 1572prints 1573 1574 Word is cat, ends at position 3 1575 Word is dog, ends at position 7 1576 Word is house, ends at position 13 1577 1578A failed match or changing the target string resets the position. If 1579you don't want the position reset after failure to match, add the 1580C<//c>, as in C</regexp/gc>. The current position in the string is 1581associated with the string, not the regexp. This means that different 1582strings have different positions and their respective positions can be 1583set or read independently. 1584 1585In list context, C<//g> returns a list of matched groupings, or if 1586there are no groupings, a list of matches to the whole regexp. So if 1587we wanted just the words, we could use 1588 1589 @words = ($x =~ /(\w+)/g); # matches, 1590 # $word[0] = 'cat' 1591 # $word[1] = 'dog' 1592 # $word[2] = 'house' 1593 1594Closely associated with the C<//g> modifier is the C<\G> anchor. The 1595C<\G> anchor matches at the point where the previous C<//g> match left 1596off. C<\G> allows us to easily do context-sensitive matching: 1597 1598 $metric = 1; # use metric units 1599 ... 1600 $x = <FILE>; # read in measurement 1601 $x =~ /^([+-]?\d+)\s*/g; # get magnitude 1602 $weight = $1; 1603 if ($metric) { # error checking 1604 print "Units error!" unless $x =~ /\Gkg\./g; 1605 } 1606 else { 1607 print "Units error!" unless $x =~ /\Glbs\./g; 1608 } 1609 $x =~ /\G\s+(widget|sprocket)/g; # continue processing 1610 1611The combination of C<//g> and C<\G> allows us to process the string a 1612bit at a time and use arbitrary Perl logic to decide what to do next. 1613Currently, the C<\G> anchor is only fully supported when used to anchor 1614to the start of the pattern. 1615 1616C<\G> is also invaluable in processing fixed length records with 1617regexps. Suppose we have a snippet of coding region DNA, encoded as 1618base pair letters C<ATCGTTGAAT...> and we want to find all the stop 1619codons C<TGA>. In a coding region, codons are 3-letter sequences, so 1620we can think of the DNA snippet as a sequence of 3-letter records. The 1621naive regexp 1622 1623 # expanded, this is "ATC GTT GAA TGC AAA TGA CAT GAC" 1624 $dna = "ATCGTTGAATGCAAATGACATGAC"; 1625 $dna =~ /TGA/; 1626 1627doesn't work; it may match a C<TGA>, but there is no guarantee that 1628the match is aligned with codon boundaries, e.g., the substring 1629S<C<GTT GAA>> gives a match. A better solution is 1630 1631 while ($dna =~ /(\w\w\w)*?TGA/g) { # note the minimal *? 1632 print "Got a TGA stop codon at position ", pos $dna, "\n"; 1633 } 1634 1635which prints 1636 1637 Got a TGA stop codon at position 18 1638 Got a TGA stop codon at position 23 1639 1640Position 18 is good, but position 23 is bogus. What happened? 1641 1642The answer is that our regexp works well until we get past the last 1643real match. Then the regexp will fail to match a synchronized C<TGA> 1644and start stepping ahead one character position at a time, not what we 1645want. The solution is to use C<\G> to anchor the match to the codon 1646alignment: 1647 1648 while ($dna =~ /\G(\w\w\w)*?TGA/g) { 1649 print "Got a TGA stop codon at position ", pos $dna, "\n"; 1650 } 1651 1652This prints 1653 1654 Got a TGA stop codon at position 18 1655 1656which is the correct answer. This example illustrates that it is 1657important not only to match what is desired, but to reject what is not 1658desired. 1659 1660=head3 Search and replace 1661 1662Regular expressions also play a big role in I<search and replace> 1663operations in Perl. Search and replace is accomplished with the 1664C<s///> operator. The general form is 1665C<s/regexp/replacement/modifiers>, with everything we know about 1666regexps and modifiers applying in this case as well. The 1667C<replacement> is a Perl double quoted string that replaces in the 1668string whatever is matched with the C<regexp>. The operator C<=~> is 1669also used here to associate a string with C<s///>. If matching 1670against C<$_>, the S<C<$_ =~>> can be dropped. If there is a match, 1671C<s///> returns the number of substitutions made, otherwise it returns 1672false. Here are a few examples: 1673 1674 $x = "Time to feed the cat!"; 1675 $x =~ s/cat/hacker/; # $x contains "Time to feed the hacker!" 1676 if ($x =~ s/^(Time.*hacker)!$/$1 now!/) { 1677 $more_insistent = 1; 1678 } 1679 $y = "'quoted words'"; 1680 $y =~ s/^'(.*)'$/$1/; # strip single quotes, 1681 # $y contains "quoted words" 1682 1683In the last example, the whole string was matched, but only the part 1684inside the single quotes was grouped. With the C<s///> operator, the 1685matched variables C<$1>, C<$2>, etc. are immediately available for use 1686in the replacement expression, so we use C<$1> to replace the quoted 1687string with just what was quoted. With the global modifier, C<s///g> 1688will search and replace all occurrences of the regexp in the string: 1689 1690 $x = "I batted 4 for 4"; 1691 $x =~ s/4/four/; # doesn't do it all: 1692 # $x contains "I batted four for 4" 1693 $x = "I batted 4 for 4"; 1694 $x =~ s/4/four/g; # does it all: 1695 # $x contains "I batted four for four" 1696 1697If you prefer 'regex' over 'regexp' in this tutorial, you could use 1698the following program to replace it: 1699 1700 % cat > simple_replace 1701 #!/usr/bin/perl 1702 $regexp = shift; 1703 $replacement = shift; 1704 while (<>) { 1705 s/$regexp/$replacement/go; 1706 print; 1707 } 1708 ^D 1709 1710 % simple_replace regexp regex perlretut.pod 1711 1712In C<simple_replace> we used the C<s///g> modifier to replace all 1713occurrences of the regexp on each line and the C<s///o> modifier to 1714compile the regexp only once. As with C<simple_grep>, both the 1715C<print> and the C<s/$regexp/$replacement/go> use C<$_> implicitly. 1716 1717A modifier available specifically to search and replace is the 1718C<s///e> evaluation modifier. C<s///e> wraps an C<eval{...}> around 1719the replacement string and the evaluated result is substituted for the 1720matched substring. C<s///e> is useful if you need to do a bit of 1721computation in the process of replacing text. This example counts 1722character frequencies in a line: 1723 1724 $x = "Bill the cat"; 1725 $x =~ s/(.)/$chars{$1}++;$1/eg; # final $1 replaces char with itself 1726 print "frequency of '$_' is $chars{$_}\n" 1727 foreach (sort {$chars{$b} <=> $chars{$a}} keys %chars); 1728 1729This prints 1730 1731 frequency of ' ' is 2 1732 frequency of 't' is 2 1733 frequency of 'l' is 2 1734 frequency of 'B' is 1 1735 frequency of 'c' is 1 1736 frequency of 'e' is 1 1737 frequency of 'h' is 1 1738 frequency of 'i' is 1 1739 frequency of 'a' is 1 1740 1741As with the match C<m//> operator, C<s///> can use other delimiters, 1742such as C<s!!!> and C<s{}{}>, and even C<s{}//>. If single quotes are 1743used C<s'''>, then the regexp and replacement are treated as single 1744quoted strings and there are no substitutions. C<s///> in list context 1745returns the same thing as in scalar context, i.e., the number of 1746matches. 1747 1748=head3 The split function 1749 1750The C<split()> function is another place where a regexp is used. 1751C<split /regexp/, string, limit> separates the C<string> operand into 1752a list of substrings and returns that list. The regexp must be designed 1753to match whatever constitutes the separators for the desired substrings. 1754The C<limit>, if present, constrains splitting into no more than C<limit> 1755number of strings. For example, to split a string into words, use 1756 1757 $x = "Calvin and Hobbes"; 1758 @words = split /\s+/, $x; # $word[0] = 'Calvin' 1759 # $word[1] = 'and' 1760 # $word[2] = 'Hobbes' 1761 1762If the empty regexp C<//> is used, the regexp always matches and 1763the string is split into individual characters. If the regexp has 1764groupings, then the resulting list contains the matched substrings from the 1765groupings as well. For instance, 1766 1767 $x = "/usr/bin/perl"; 1768 @dirs = split m!/!, $x; # $dirs[0] = '' 1769 # $dirs[1] = 'usr' 1770 # $dirs[2] = 'bin' 1771 # $dirs[3] = 'perl' 1772 @parts = split m!(/)!, $x; # $parts[0] = '' 1773 # $parts[1] = '/' 1774 # $parts[2] = 'usr' 1775 # $parts[3] = '/' 1776 # $parts[4] = 'bin' 1777 # $parts[5] = '/' 1778 # $parts[6] = 'perl' 1779 1780Since the first character of $x matched the regexp, C<split> prepended 1781an empty initial element to the list. 1782 1783If you have read this far, congratulations! You now have all the basic 1784tools needed to use regular expressions to solve a wide range of text 1785processing problems. If this is your first time through the tutorial, 1786why not stop here and play around with regexps a while... S<Part 2> 1787concerns the more esoteric aspects of regular expressions and those 1788concepts certainly aren't needed right at the start. 1789 1790=head1 Part 2: Power tools 1791 1792OK, you know the basics of regexps and you want to know more. If 1793matching regular expressions is analogous to a walk in the woods, then 1794the tools discussed in Part 1 are analogous to topo maps and a 1795compass, basic tools we use all the time. Most of the tools in part 2 1796are analogous to flare guns and satellite phones. They aren't used 1797too often on a hike, but when we are stuck, they can be invaluable. 1798 1799What follows are the more advanced, less used, or sometimes esoteric 1800capabilities of Perl regexps. In Part 2, we will assume you are 1801comfortable with the basics and concentrate on the new features. 1802 1803=head2 More on characters, strings, and character classes 1804 1805There are a number of escape sequences and character classes that we 1806haven't covered yet. 1807 1808There are several escape sequences that convert characters or strings 1809between upper and lower case, and they are also available within 1810patterns. C<\l> and C<\u> convert the next character to lower or 1811upper case, respectively: 1812 1813 $x = "perl"; 1814 $string =~ /\u$x/; # matches 'Perl' in $string 1815 $x = "M(rs?|s)\\."; # note the double backslash 1816 $string =~ /\l$x/; # matches 'mr.', 'mrs.', and 'ms.', 1817 1818A C<\L> or C<\U> indicates a lasting conversion of case, until 1819terminated by C<\E> or thrown over by another C<\U> or C<\L>: 1820 1821 $x = "This word is in lower case:\L SHOUT\E"; 1822 $x =~ /shout/; # matches 1823 $x = "I STILL KEYPUNCH CARDS FOR MY 360" 1824 $x =~ /\Ukeypunch/; # matches punch card string 1825 1826If there is no C<\E>, case is converted until the end of the 1827string. The regexps C<\L\u$word> or C<\u\L$word> convert the first 1828character of C<$word> to uppercase and the rest of the characters to 1829lowercase. 1830 1831Control characters can be escaped with C<\c>, so that a control-Z 1832character would be matched with C<\cZ>. The escape sequence 1833C<\Q>...C<\E> quotes, or protects most non-alphabetic characters. For 1834instance, 1835 1836 $x = "\QThat !^*&%~& cat!"; 1837 $x =~ /\Q!^*&%~&\E/; # check for rough language 1838 1839It does not protect C<$> or C<@>, so that variables can still be 1840substituted. 1841 1842With the advent of 5.6.0, Perl regexps can handle more than just the 1843standard ASCII character set. Perl now supports I<Unicode>, a standard 1844for representing the alphabets from virtually all of the world's written 1845languages, and a host of symbols. Perl's text strings are Unicode strings, so 1846they can contain characters with a value (codepoint or character number) higher 1847than 255 1848 1849What does this mean for regexps? Well, regexp users don't need to know 1850much about Perl's internal representation of strings. But they do need 1851to know 1) how to represent Unicode characters in a regexp and 2) that 1852a matching operation will treat the string to be searched as a sequence 1853of characters, not bytes. The answer to 1) is that Unicode characters 1854greater than C<chr(255)> are represented using the C<\x{hex}> notation, 1855because the \0 octal and \x hex (without curly braces) don't go further 1856than 255. 1857 1858 /\x{263a}/; # match a Unicode smiley face :) 1859 1860B<NOTE>: In Perl 5.6.0 it used to be that one needed to say C<use 1861utf8> to use any Unicode features. This is no more the case: for 1862almost all Unicode processing, the explicit C<utf8> pragma is not 1863needed. (The only case where it matters is if your Perl script is in 1864Unicode and encoded in UTF-8, then an explicit C<use utf8> is needed.) 1865 1866Figuring out the hexadecimal sequence of a Unicode character you want 1867or deciphering someone else's hexadecimal Unicode regexp is about as 1868much fun as programming in machine code. So another way to specify 1869Unicode characters is to use the I<named character> escape 1870sequence C<\N{I<name>}>. I<name> is a name for the Unicode character, as 1871specified in the Unicode standard. For instance, if we wanted to 1872represent or match the astrological sign for the planet Mercury, we 1873could use 1874 1875 use charnames ":full"; # use named chars with Unicode full names 1876 $x = "abc\N{MERCURY}def"; 1877 $x =~ /\N{MERCURY}/; # matches 1878 1879One can also use short names or restrict names to a certain alphabet: 1880 1881 use charnames ':full'; 1882 print "\N{GREEK SMALL LETTER SIGMA} is called sigma.\n"; 1883 1884 use charnames ":short"; 1885 print "\N{greek:Sigma} is an upper-case sigma.\n"; 1886 1887 use charnames qw(greek); 1888 print "\N{sigma} is Greek sigma\n"; 1889 1890A list of full names is found in the file NamesList.txt in the 1891lib/perl5/X.X.X/unicore directory (where X.X.X is the perl 1892version number as it is installed on your system). 1893 1894The answer to requirement 2), as of 5.6.0, is that a regexp uses Unicode 1895characters. Internally, this is encoded to bytes using either UTF-8 or a 1896native 8 bit encoding, depending on the history of the string, but 1897conceptually it is a sequence of characters, not bytes. See 1898L<perlunitut> for a tutorial about that. 1899 1900Let us now discuss Unicode character classes. Just as with Unicode 1901characters, there are named Unicode character classes represented by the 1902C<\p{name}> escape sequence. Closely associated is the C<\P{name}> 1903character class, which is the negation of the C<\p{name}> class. For 1904example, to match lower and uppercase characters, 1905 1906 use charnames ":full"; # use named chars with Unicode full names 1907 $x = "BOB"; 1908 $x =~ /^\p{IsUpper}/; # matches, uppercase char class 1909 $x =~ /^\P{IsUpper}/; # doesn't match, char class sans uppercase 1910 $x =~ /^\p{IsLower}/; # doesn't match, lowercase char class 1911 $x =~ /^\P{IsLower}/; # matches, char class sans lowercase 1912 1913Here is the association between some Perl named classes and the 1914traditional Unicode classes: 1915 1916 Perl class name Unicode class name or regular expression 1917 1918 IsAlpha /^[LM]/ 1919 IsAlnum /^[LMN]/ 1920 IsASCII $code <= 127 1921 IsCntrl /^C/ 1922 IsBlank $code =~ /^(0020|0009)$/ || /^Z[^lp]/ 1923 IsDigit Nd 1924 IsGraph /^([LMNPS]|Co)/ 1925 IsLower Ll 1926 IsPrint /^([LMNPS]|Co|Zs)/ 1927 IsPunct /^P/ 1928 IsSpace /^Z/ || ($code =~ /^(0009|000A|000B|000C|000D)$/ 1929 IsSpacePerl /^Z/ || ($code =~ /^(0009|000A|000C|000D|0085|2028|2029)$/ 1930 IsUpper /^L[ut]/ 1931 IsWord /^[LMN]/ || $code eq "005F" 1932 IsXDigit $code =~ /^00(3[0-9]|[46][1-6])$/ 1933 1934You can also use the official Unicode class names with the C<\p> and 1935C<\P>, like C<\p{L}> for Unicode 'letters', or C<\p{Lu}> for uppercase 1936letters, or C<\P{Nd}> for non-digits. If a C<name> is just one 1937letter, the braces can be dropped. For instance, C<\pM> is the 1938character class of Unicode 'marks', for example accent marks. 1939For the full list see L<perlunicode>. 1940 1941The Unicode has also been separated into various sets of characters 1942which you can test with C<\p{...}> (in) and C<\P{...}> (not in). 1943To test whether a character is (or is not) an element of a script 1944you would use the script name, for example C<\p{Latin}>, C<\p{Greek}>, 1945or C<\P{Katakana}>. Other sets are the Unicode blocks, the names 1946of which begin with "In". One such block is dedicated to mathematical 1947operators, and its pattern formula is <C\p{InMathematicalOperators>}>. 1948For the full list see L<perluniprops>. 1949 1950What we have described so far is the single form of the C<\p{...}> character 1951classes. There is also a compound form which you may run into. These 1952look like C<\p{name=value}> or C<\p{name:value}> (the equals sign and colon 1953can be used interchangeably). These are more general than the single form, 1954and in fact most of the single forms are just Perl-defined shortcuts for common 1955compound forms. For example, the script examples in the previous paragraph 1956could be written equivalently as C<\p{Script=Latin}>, C<\p{Script:Greek}>, and 1957C<\P{script=katakana}> (case is irrelevant between the C<{}> braces). You may 1958never have to use the compound forms, but sometimes it is necessary, and their 1959use can make your code easier to understand. 1960 1961C<\X> is an abbreviation for a character class that comprises 1962a Unicode I<extended grapheme cluster>. This represents a "logical character", 1963what appears to be a single character, but may be represented internally by more 1964than one. As an example, using the Unicode full names, e.g., S<C<A + COMBINING 1965RING>> is a grapheme cluster with base character C<A> and combining character 1966S<C<COMBINING RING>>, which translates in Danish to A with the circle atop it, 1967as in the word Angstrom. 1968 1969For the full and latest information about Unicode see the latest 1970Unicode standard, or the Unicode Consortium's website L<http://www.unicode.org> 1971 1972As if all those classes weren't enough, Perl also defines POSIX style 1973character classes. These have the form C<[:name:]>, with C<name> the 1974name of the POSIX class. The POSIX classes are C<alpha>, C<alnum>, 1975C<ascii>, C<cntrl>, C<digit>, C<graph>, C<lower>, C<print>, C<punct>, 1976C<space>, C<upper>, and C<xdigit>, and two extensions, C<word> (a Perl 1977extension to match C<\w>), and C<blank> (a GNU extension). If C<utf8> 1978is being used, then these classes are defined the same as their 1979corresponding Perl Unicode classes: C<[:upper:]> is the same as 1980C<\p{IsUpper}>, etc. The POSIX character classes, however, don't 1981require using C<utf8>. The C<[:digit:]>, C<[:word:]>, and 1982C<[:space:]> correspond to the familiar C<\d>, C<\w>, and C<\s> 1983character classes. To negate a POSIX class, put a C<^> in front of 1984the name, so that, e.g., C<[:^digit:]> corresponds to C<\D> and under 1985C<utf8>, C<\P{IsDigit}>. The Unicode and POSIX character classes can 1986be used just like C<\d>, with the exception that POSIX character 1987classes can only be used inside of a character class: 1988 1989 /\s+[abc[:digit:]xyz]\s*/; # match a,b,c,x,y,z, or a digit 1990 /^=item\s[[:digit:]]/; # match '=item', 1991 # followed by a space and a digit 1992 use charnames ":full"; 1993 /\s+[abc\p{IsDigit}xyz]\s+/; # match a,b,c,x,y,z, or a digit 1994 /^=item\s\p{IsDigit}/; # match '=item', 1995 # followed by a space and a digit 1996 1997Whew! That is all the rest of the characters and character classes. 1998 1999=head2 Compiling and saving regular expressions 2000 2001In Part 1 we discussed the C<//o> modifier, which compiles a regexp 2002just once. This suggests that a compiled regexp is some data structure 2003that can be stored once and used again and again. The regexp quote 2004C<qr//> does exactly that: C<qr/string/> compiles the C<string> as a 2005regexp and transforms the result into a form that can be assigned to a 2006variable: 2007 2008 $reg = qr/foo+bar?/; # reg contains a compiled regexp 2009 2010Then C<$reg> can be used as a regexp: 2011 2012 $x = "fooooba"; 2013 $x =~ $reg; # matches, just like /foo+bar?/ 2014 $x =~ /$reg/; # same thing, alternate form 2015 2016C<$reg> can also be interpolated into a larger regexp: 2017 2018 $x =~ /(abc)?$reg/; # still matches 2019 2020As with the matching operator, the regexp quote can use different 2021delimiters, e.g., C<qr!!>, C<qr{}> or C<qr~~>. Apostrophes 2022as delimiters (C<qr''>) inhibit any interpolation. 2023 2024Pre-compiled regexps are useful for creating dynamic matches that 2025don't need to be recompiled each time they are encountered. Using 2026pre-compiled regexps, we write a C<grep_step> program which greps 2027for a sequence of patterns, advancing to the next pattern as soon 2028as one has been satisfied. 2029 2030 % cat > grep_step 2031 #!/usr/bin/perl 2032 # grep_step - match <number> regexps, one after the other 2033 # usage: multi_grep <number> regexp1 regexp2 ... file1 file2 ... 2034 2035 $number = shift; 2036 $regexp[$_] = shift foreach (0..$number-1); 2037 @compiled = map qr/$_/, @regexp; 2038 while ($line = <>) { 2039 if ($line =~ /$compiled[0]/) { 2040 print $line; 2041 shift @compiled; 2042 last unless @compiled; 2043 } 2044 } 2045 ^D 2046 2047 % grep_step 3 shift print last grep_step 2048 $number = shift; 2049 print $line; 2050 last unless @compiled; 2051 2052Storing pre-compiled regexps in an array C<@compiled> allows us to 2053simply loop through the regexps without any recompilation, thus gaining 2054flexibility without sacrificing speed. 2055 2056 2057=head2 Composing regular expressions at runtime 2058 2059Backtracking is more efficient than repeated tries with different regular 2060expressions. If there are several regular expressions and a match with 2061any of them is acceptable, then it is possible to combine them into a set 2062of alternatives. If the individual expressions are input data, this 2063can be done by programming a join operation. We'll exploit this idea in 2064an improved version of the C<simple_grep> program: a program that matches 2065multiple patterns: 2066 2067 % cat > multi_grep 2068 #!/usr/bin/perl 2069 # multi_grep - match any of <number> regexps 2070 # usage: multi_grep <number> regexp1 regexp2 ... file1 file2 ... 2071 2072 $number = shift; 2073 $regexp[$_] = shift foreach (0..$number-1); 2074 $pattern = join '|', @regexp; 2075 2076 while ($line = <>) { 2077 print $line if $line =~ /$pattern/o; 2078 } 2079 ^D 2080 2081 % multi_grep 2 shift for multi_grep 2082 $number = shift; 2083 $regexp[$_] = shift foreach (0..$number-1); 2084 2085Sometimes it is advantageous to construct a pattern from the I<input> 2086that is to be analyzed and use the permissible values on the left 2087hand side of the matching operations. As an example for this somewhat 2088paradoxical situation, let's assume that our input contains a command 2089verb which should match one out of a set of available command verbs, 2090with the additional twist that commands may be abbreviated as long as 2091the given string is unique. The program below demonstrates the basic 2092algorithm. 2093 2094 % cat > keymatch 2095 #!/usr/bin/perl 2096 $kwds = 'copy compare list print'; 2097 while( $command = <> ){ 2098 $command =~ s/^\s+|\s+$//g; # trim leading and trailing spaces 2099 if( ( @matches = $kwds =~ /\b$command\w*/g ) == 1 ){ 2100 print "command: '@matches'\n"; 2101 } elsif( @matches == 0 ){ 2102 print "no such command: '$command'\n"; 2103 } else { 2104 print "not unique: '$command' (could be one of: @matches)\n"; 2105 } 2106 } 2107 ^D 2108 2109 % keymatch 2110 li 2111 command: 'list' 2112 co 2113 not unique: 'co' (could be one of: copy compare) 2114 printer 2115 no such command: 'printer' 2116 2117Rather than trying to match the input against the keywords, we match the 2118combined set of keywords against the input. The pattern matching 2119operation S<C<$kwds =~ /\b($command\w*)/g>> does several things at the 2120same time. It makes sure that the given command begins where a keyword 2121begins (C<\b>). It tolerates abbreviations due to the added C<\w*>. It 2122tells us the number of matches (C<scalar @matches>) and all the keywords 2123that were actually matched. You could hardly ask for more. 2124 2125=head2 Embedding comments and modifiers in a regular expression 2126 2127Starting with this section, we will be discussing Perl's set of 2128I<extended patterns>. These are extensions to the traditional regular 2129expression syntax that provide powerful new tools for pattern 2130matching. We have already seen extensions in the form of the minimal 2131matching constructs C<??>, C<*?>, C<+?>, C<{n,m}?>, and C<{n,}?>. The 2132rest of the extensions below have the form C<(?char...)>, where the 2133C<char> is a character that determines the type of extension. 2134 2135The first extension is an embedded comment C<(?#text)>. This embeds a 2136comment into the regular expression without affecting its meaning. The 2137comment should not have any closing parentheses in the text. An 2138example is 2139 2140 /(?# Match an integer:)[+-]?\d+/; 2141 2142This style of commenting has been largely superseded by the raw, 2143freeform commenting that is allowed with the C<//x> modifier. 2144 2145The modifiers C<//i>, C<//m>, C<//s> and C<//x> (or any 2146combination thereof) can also be embedded in 2147a regexp using C<(?i)>, C<(?m)>, C<(?s)>, and C<(?x)>. For instance, 2148 2149 /(?i)yes/; # match 'yes' case insensitively 2150 /yes/i; # same thing 2151 /(?x)( # freeform version of an integer regexp 2152 [+-]? # match an optional sign 2153 \d+ # match a sequence of digits 2154 ) 2155 /x; 2156 2157Embedded modifiers can have two important advantages over the usual 2158modifiers. Embedded modifiers allow a custom set of modifiers to 2159I<each> regexp pattern. This is great for matching an array of regexps 2160that must have different modifiers: 2161 2162 $pattern[0] = '(?i)doctor'; 2163 $pattern[1] = 'Johnson'; 2164 ... 2165 while (<>) { 2166 foreach $patt (@pattern) { 2167 print if /$patt/; 2168 } 2169 } 2170 2171The second advantage is that embedded modifiers (except C<//p>, which 2172modifies the entire regexp) only affect the regexp 2173inside the group the embedded modifier is contained in. So grouping 2174can be used to localize the modifier's effects: 2175 2176 /Answer: ((?i)yes)/; # matches 'Answer: yes', 'Answer: YES', etc. 2177 2178Embedded modifiers can also turn off any modifiers already present 2179by using, e.g., C<(?-i)>. Modifiers can also be combined into 2180a single expression, e.g., C<(?s-i)> turns on single line mode and 2181turns off case insensitivity. 2182 2183Embedded modifiers may also be added to a non-capturing grouping. 2184C<(?i-m:regexp)> is a non-capturing grouping that matches C<regexp> 2185case insensitively and turns off multi-line mode. 2186 2187 2188=head2 Looking ahead and looking behind 2189 2190This section concerns the lookahead and lookbehind assertions. First, 2191a little background. 2192 2193In Perl regular expressions, most regexp elements 'eat up' a certain 2194amount of string when they match. For instance, the regexp element 2195C<[abc}]> eats up one character of the string when it matches, in the 2196sense that Perl moves to the next character position in the string 2197after the match. There are some elements, however, that don't eat up 2198characters (advance the character position) if they match. The examples 2199we have seen so far are the anchors. The anchor C<^> matches the 2200beginning of the line, but doesn't eat any characters. Similarly, the 2201word boundary anchor C<\b> matches wherever a character matching C<\w> 2202is next to a character that doesn't, but it doesn't eat up any 2203characters itself. Anchors are examples of I<zero-width assertions>. 2204Zero-width, because they consume 2205no characters, and assertions, because they test some property of the 2206string. In the context of our walk in the woods analogy to regexp 2207matching, most regexp elements move us along a trail, but anchors have 2208us stop a moment and check our surroundings. If the local environment 2209checks out, we can proceed forward. But if the local environment 2210doesn't satisfy us, we must backtrack. 2211 2212Checking the environment entails either looking ahead on the trail, 2213looking behind, or both. C<^> looks behind, to see that there are no 2214characters before. C<$> looks ahead, to see that there are no 2215characters after. C<\b> looks both ahead and behind, to see if the 2216characters on either side differ in their "word-ness". 2217 2218The lookahead and lookbehind assertions are generalizations of the 2219anchor concept. Lookahead and lookbehind are zero-width assertions 2220that let us specify which characters we want to test for. The 2221lookahead assertion is denoted by C<(?=regexp)> and the lookbehind 2222assertion is denoted by C<< (?<=fixed-regexp) >>. Some examples are 2223 2224 $x = "I catch the housecat 'Tom-cat' with catnip"; 2225 $x =~ /cat(?=\s)/; # matches 'cat' in 'housecat' 2226 @catwords = ($x =~ /(?<=\s)cat\w+/g); # matches, 2227 # $catwords[0] = 'catch' 2228 # $catwords[1] = 'catnip' 2229 $x =~ /\bcat\b/; # matches 'cat' in 'Tom-cat' 2230 $x =~ /(?<=\s)cat(?=\s)/; # doesn't match; no isolated 'cat' in 2231 # middle of $x 2232 2233Note that the parentheses in C<(?=regexp)> and C<< (?<=regexp) >> are 2234non-capturing, since these are zero-width assertions. Thus in the 2235second regexp, the substrings captured are those of the whole regexp 2236itself. Lookahead C<(?=regexp)> can match arbitrary regexps, but 2237lookbehind C<< (?<=fixed-regexp) >> only works for regexps of fixed 2238width, i.e., a fixed number of characters long. Thus 2239C<< (?<=(ab|bc)) >> is fine, but C<< (?<=(ab)*) >> is not. The 2240negated versions of the lookahead and lookbehind assertions are 2241denoted by C<(?!regexp)> and C<< (?<!fixed-regexp) >> respectively. 2242They evaluate true if the regexps do I<not> match: 2243 2244 $x = "foobar"; 2245 $x =~ /foo(?!bar)/; # doesn't match, 'bar' follows 'foo' 2246 $x =~ /foo(?!baz)/; # matches, 'baz' doesn't follow 'foo' 2247 $x =~ /(?<!\s)foo/; # matches, there is no \s before 'foo' 2248 2249The C<\C> is unsupported in lookbehind, because the already 2250treacherous definition of C<\C> would become even more so 2251when going backwards. 2252 2253Here is an example where a string containing blank-separated words, 2254numbers and single dashes is to be split into its components. 2255Using C</\s+/> alone won't work, because spaces are not required between 2256dashes, or a word or a dash. Additional places for a split are established 2257by looking ahead and behind: 2258 2259 $str = "one two - --6-8"; 2260 @toks = split / \s+ # a run of spaces 2261 | (?<=\S) (?=-) # any non-space followed by '-' 2262 | (?<=-) (?=\S) # a '-' followed by any non-space 2263 /x, $str; # @toks = qw(one two - - - 6 - 8) 2264 2265 2266=head2 Using independent subexpressions to prevent backtracking 2267 2268I<Independent subexpressions> are regular expressions, in the 2269context of a larger regular expression, that function independently of 2270the larger regular expression. That is, they consume as much or as 2271little of the string as they wish without regard for the ability of 2272the larger regexp to match. Independent subexpressions are represented 2273by C<< (?>regexp) >>. We can illustrate their behavior by first 2274considering an ordinary regexp: 2275 2276 $x = "ab"; 2277 $x =~ /a*ab/; # matches 2278 2279This obviously matches, but in the process of matching, the 2280subexpression C<a*> first grabbed the C<a>. Doing so, however, 2281wouldn't allow the whole regexp to match, so after backtracking, C<a*> 2282eventually gave back the C<a> and matched the empty string. Here, what 2283C<a*> matched was I<dependent> on what the rest of the regexp matched. 2284 2285Contrast that with an independent subexpression: 2286 2287 $x =~ /(?>a*)ab/; # doesn't match! 2288 2289The independent subexpression C<< (?>a*) >> doesn't care about the rest 2290of the regexp, so it sees an C<a> and grabs it. Then the rest of the 2291regexp C<ab> cannot match. Because C<< (?>a*) >> is independent, there 2292is no backtracking and the independent subexpression does not give 2293up its C<a>. Thus the match of the regexp as a whole fails. A similar 2294behavior occurs with completely independent regexps: 2295 2296 $x = "ab"; 2297 $x =~ /a*/g; # matches, eats an 'a' 2298 $x =~ /\Gab/g; # doesn't match, no 'a' available 2299 2300Here C<//g> and C<\G> create a 'tag team' handoff of the string from 2301one regexp to the other. Regexps with an independent subexpression are 2302much like this, with a handoff of the string to the independent 2303subexpression, and a handoff of the string back to the enclosing 2304regexp. 2305 2306The ability of an independent subexpression to prevent backtracking 2307can be quite useful. Suppose we want to match a non-empty string 2308enclosed in parentheses up to two levels deep. Then the following 2309regexp matches: 2310 2311 $x = "abc(de(fg)h"; # unbalanced parentheses 2312 $x =~ /\( ( [^()]+ | \([^()]*\) )+ \)/x; 2313 2314The regexp matches an open parenthesis, one or more copies of an 2315alternation, and a close parenthesis. The alternation is two-way, with 2316the first alternative C<[^()]+> matching a substring with no 2317parentheses and the second alternative C<\([^()]*\)> matching a 2318substring delimited by parentheses. The problem with this regexp is 2319that it is pathological: it has nested indeterminate quantifiers 2320of the form C<(a+|b)+>. We discussed in Part 1 how nested quantifiers 2321like this could take an exponentially long time to execute if there 2322was no match possible. To prevent the exponential blowup, we need to 2323prevent useless backtracking at some point. This can be done by 2324enclosing the inner quantifier as an independent subexpression: 2325 2326 $x =~ /\( ( (?>[^()]+) | \([^()]*\) )+ \)/x; 2327 2328Here, C<< (?>[^()]+) >> breaks the degeneracy of string partitioning 2329by gobbling up as much of the string as possible and keeping it. Then 2330match failures fail much more quickly. 2331 2332 2333=head2 Conditional expressions 2334 2335A I<conditional expression> is a form of if-then-else statement 2336that allows one to choose which patterns are to be matched, based on 2337some condition. There are two types of conditional expression: 2338C<(?(condition)yes-regexp)> and 2339C<(?(condition)yes-regexp|no-regexp)>. C<(?(condition)yes-regexp)> is 2340like an S<C<'if () {}'>> statement in Perl. If the C<condition> is true, 2341the C<yes-regexp> will be matched. If the C<condition> is false, the 2342C<yes-regexp> will be skipped and Perl will move onto the next regexp 2343element. The second form is like an S<C<'if () {} else {}'>> statement 2344in Perl. If the C<condition> is true, the C<yes-regexp> will be 2345matched, otherwise the C<no-regexp> will be matched. 2346 2347The C<condition> can have several forms. The first form is simply an 2348integer in parentheses C<(integer)>. It is true if the corresponding 2349backreference C<\integer> matched earlier in the regexp. The same 2350thing can be done with a name associated with a capture buffer, written 2351as C<< (<name>) >> or C<< ('name') >>. The second form is a bare 2352zero width assertion C<(?...)>, either a lookahead, a lookbehind, or a 2353code assertion (discussed in the next section). The third set of forms 2354provides tests that return true if the expression is executed within 2355a recursion (C<(R)>) or is being called from some capturing group, 2356referenced either by number (C<(R1)>, C<(R2)>,...) or by name 2357(C<(R&name)>). 2358 2359The integer or name form of the C<condition> allows us to choose, 2360with more flexibility, what to match based on what matched earlier in the 2361regexp. This searches for words of the form C<"$x$x"> or C<"$x$y$y$x">: 2362 2363 % simple_grep '^(\w+)(\w+)?(?(2)\2\1|\1)$' /usr/dict/words 2364 beriberi 2365 coco 2366 couscous 2367 deed 2368 ... 2369 toot 2370 toto 2371 tutu 2372 2373The lookbehind C<condition> allows, along with backreferences, 2374an earlier part of the match to influence a later part of the 2375match. For instance, 2376 2377 /[ATGC]+(?(?<=AA)G|C)$/; 2378 2379matches a DNA sequence such that it either ends in C<AAG>, or some 2380other base pair combination and C<C>. Note that the form is 2381C<< (?(?<=AA)G|C) >> and not C<< (?((?<=AA))G|C) >>; for the 2382lookahead, lookbehind or code assertions, the parentheses around the 2383conditional are not needed. 2384 2385 2386=head2 Defining named patterns 2387 2388Some regular expressions use identical subpatterns in several places. 2389Starting with Perl 5.10, it is possible to define named subpatterns in 2390a section of the pattern so that they can be called up by name 2391anywhere in the pattern. This syntactic pattern for this definition 2392group is C<< (?(DEFINE)(?<name>pattern)...) >>. An insertion 2393of a named pattern is written as C<(?&name)>. 2394 2395The example below illustrates this feature using the pattern for 2396floating point numbers that was presented earlier on. The three 2397subpatterns that are used more than once are the optional sign, the 2398digit sequence for an integer and the decimal fraction. The DEFINE 2399group at the end of the pattern contains their definition. Notice 2400that the decimal fraction pattern is the first place where we can 2401reuse the integer pattern. 2402 2403 /^ (?&osg)\ * ( (?&int)(?&dec)? | (?&dec) ) 2404 (?: [eE](?&osg)(?&int) )? 2405 $ 2406 (?(DEFINE) 2407 (?<osg>[-+]?) # optional sign 2408 (?<int>\d++) # integer 2409 (?<dec>\.(?&int)) # decimal fraction 2410 )/x 2411 2412 2413=head2 Recursive patterns 2414 2415This feature (introduced in Perl 5.10) significantly extends the 2416power of Perl's pattern matching. By referring to some other 2417capture group anywhere in the pattern with the construct 2418C<(?group-ref)>, the I<pattern> within the referenced group is used 2419as an independent subpattern in place of the group reference itself. 2420Because the group reference may be contained I<within> the group it 2421refers to, it is now possible to apply pattern matching to tasks that 2422hitherto required a recursive parser. 2423 2424To illustrate this feature, we'll design a pattern that matches if 2425a string contains a palindrome. (This is a word or a sentence that, 2426while ignoring spaces, interpunctuation and case, reads the same backwards 2427as forwards. We begin by observing that the empty string or a string 2428containing just one word character is a palindrome. Otherwise it must 2429have a word character up front and the same at its end, with another 2430palindrome in between. 2431 2432 /(?: (\w) (?...Here be a palindrome...) \g{-1} | \w? )/x 2433 2434Adding C<\W*> at either end to eliminate what is to be ignored, we already 2435have the full pattern: 2436 2437 my $pp = qr/^(\W* (?: (\w) (?1) \g{-1} | \w? ) \W*)$/ix; 2438 for $s ( "saippuakauppias", "A man, a plan, a canal: Panama!" ){ 2439 print "'$s' is a palindrome\n" if $s =~ /$pp/; 2440 } 2441 2442In C<(?...)> both absolute and relative backreferences may be used. 2443The entire pattern can be reinserted with C<(?R)> or C<(?0)>. 2444If you prefer to name your buffers, you can use C<(?&name)> to 2445recurse into that buffer. 2446 2447 2448=head2 A bit of magic: executing Perl code in a regular expression 2449 2450Normally, regexps are a part of Perl expressions. 2451I<Code evaluation> expressions turn that around by allowing 2452arbitrary Perl code to be a part of a regexp. A code evaluation 2453expression is denoted C<(?{code})>, with I<code> a string of Perl 2454statements. 2455 2456Be warned that this feature is considered experimental, and may be 2457changed without notice. 2458 2459Code expressions are zero-width assertions, and the value they return 2460depends on their environment. There are two possibilities: either the 2461code expression is used as a conditional in a conditional expression 2462C<(?(condition)...)>, or it is not. If the code expression is a 2463conditional, the code is evaluated and the result (i.e., the result of 2464the last statement) is used to determine truth or falsehood. If the 2465code expression is not used as a conditional, the assertion always 2466evaluates true and the result is put into the special variable 2467C<$^R>. The variable C<$^R> can then be used in code expressions later 2468in the regexp. Here are some silly examples: 2469 2470 $x = "abcdef"; 2471 $x =~ /abc(?{print "Hi Mom!";})def/; # matches, 2472 # prints 'Hi Mom!' 2473 $x =~ /aaa(?{print "Hi Mom!";})def/; # doesn't match, 2474 # no 'Hi Mom!' 2475 2476Pay careful attention to the next example: 2477 2478 $x =~ /abc(?{print "Hi Mom!";})ddd/; # doesn't match, 2479 # no 'Hi Mom!' 2480 # but why not? 2481 2482At first glance, you'd think that it shouldn't print, because obviously 2483the C<ddd> isn't going to match the target string. But look at this 2484example: 2485 2486 $x =~ /abc(?{print "Hi Mom!";})[dD]dd/; # doesn't match, 2487 # but _does_ print 2488 2489Hmm. What happened here? If you've been following along, you know that 2490the above pattern should be effectively (almost) the same as the last one; 2491enclosing the C<d> in a character class isn't going to change what it 2492matches. So why does the first not print while the second one does? 2493 2494The answer lies in the optimizations the regex engine makes. In the first 2495case, all the engine sees are plain old characters (aside from the 2496C<?{}> construct). It's smart enough to realize that the string 'ddd' 2497doesn't occur in our target string before actually running the pattern 2498through. But in the second case, we've tricked it into thinking that our 2499pattern is more complicated. It takes a look, sees our 2500character class, and decides that it will have to actually run the 2501pattern to determine whether or not it matches, and in the process of 2502running it hits the print statement before it discovers that we don't 2503have a match. 2504 2505To take a closer look at how the engine does optimizations, see the 2506section L<"Pragmas and debugging"> below. 2507 2508More fun with C<?{}>: 2509 2510 $x =~ /(?{print "Hi Mom!";})/; # matches, 2511 # prints 'Hi Mom!' 2512 $x =~ /(?{$c = 1;})(?{print "$c";})/; # matches, 2513 # prints '1' 2514 $x =~ /(?{$c = 1;})(?{print "$^R";})/; # matches, 2515 # prints '1' 2516 2517The bit of magic mentioned in the section title occurs when the regexp 2518backtracks in the process of searching for a match. If the regexp 2519backtracks over a code expression and if the variables used within are 2520localized using C<local>, the changes in the variables produced by the 2521code expression are undone! Thus, if we wanted to count how many times 2522a character got matched inside a group, we could use, e.g., 2523 2524 $x = "aaaa"; 2525 $count = 0; # initialize 'a' count 2526 $c = "bob"; # test if $c gets clobbered 2527 $x =~ /(?{local $c = 0;}) # initialize count 2528 ( a # match 'a' 2529 (?{local $c = $c + 1;}) # increment count 2530 )* # do this any number of times, 2531 aa # but match 'aa' at the end 2532 (?{$count = $c;}) # copy local $c var into $count 2533 /x; 2534 print "'a' count is $count, \$c variable is '$c'\n"; 2535 2536This prints 2537 2538 'a' count is 2, $c variable is 'bob' 2539 2540If we replace the S<C< (?{local $c = $c + 1;})>> with 2541S<C< (?{$c = $c + 1;})>>, the variable changes are I<not> undone 2542during backtracking, and we get 2543 2544 'a' count is 4, $c variable is 'bob' 2545 2546Note that only localized variable changes are undone. Other side 2547effects of code expression execution are permanent. Thus 2548 2549 $x = "aaaa"; 2550 $x =~ /(a(?{print "Yow\n";}))*aa/; 2551 2552produces 2553 2554 Yow 2555 Yow 2556 Yow 2557 Yow 2558 2559The result C<$^R> is automatically localized, so that it will behave 2560properly in the presence of backtracking. 2561 2562This example uses a code expression in a conditional to match a 2563definite article, either 'the' in English or 'der|die|das' in German: 2564 2565 $lang = 'DE'; # use German 2566 ... 2567 $text = "das"; 2568 print "matched\n" 2569 if $text =~ /(?(?{ 2570 $lang eq 'EN'; # is the language English? 2571 }) 2572 the | # if so, then match 'the' 2573 (der|die|das) # else, match 'der|die|das' 2574 ) 2575 /xi; 2576 2577Note that the syntax here is C<(?(?{...})yes-regexp|no-regexp)>, not 2578C<(?((?{...}))yes-regexp|no-regexp)>. In other words, in the case of a 2579code expression, we don't need the extra parentheses around the 2580conditional. 2581 2582If you try to use code expressions with interpolating variables, Perl 2583may surprise you: 2584 2585 $bar = 5; 2586 $pat = '(?{ 1 })'; 2587 /foo(?{ $bar })bar/; # compiles ok, $bar not interpolated 2588 /foo(?{ 1 })$bar/; # compile error! 2589 /foo${pat}bar/; # compile error! 2590 2591 $pat = qr/(?{ $foo = 1 })/; # precompile code regexp 2592 /foo${pat}bar/; # compiles ok 2593 2594If a regexp has (1) code expressions and interpolating variables, or 2595(2) a variable that interpolates a code expression, Perl treats the 2596regexp as an error. If the code expression is precompiled into a 2597variable, however, interpolating is ok. The question is, why is this 2598an error? 2599 2600The reason is that variable interpolation and code expressions 2601together pose a security risk. The combination is dangerous because 2602many programmers who write search engines often take user input and 2603plug it directly into a regexp: 2604 2605 $regexp = <>; # read user-supplied regexp 2606 $chomp $regexp; # get rid of possible newline 2607 $text =~ /$regexp/; # search $text for the $regexp 2608 2609If the C<$regexp> variable contains a code expression, the user could 2610then execute arbitrary Perl code. For instance, some joker could 2611search for S<C<system('rm -rf *');>> to erase your files. In this 2612sense, the combination of interpolation and code expressions I<taints> 2613your regexp. So by default, using both interpolation and code 2614expressions in the same regexp is not allowed. If you're not 2615concerned about malicious users, it is possible to bypass this 2616security check by invoking S<C<use re 'eval'>>: 2617 2618 use re 'eval'; # throw caution out the door 2619 $bar = 5; 2620 $pat = '(?{ 1 })'; 2621 /foo(?{ 1 })$bar/; # compiles ok 2622 /foo${pat}bar/; # compiles ok 2623 2624Another form of code expression is the I<pattern code expression>. 2625The pattern code expression is like a regular code expression, except 2626that the result of the code evaluation is treated as a regular 2627expression and matched immediately. A simple example is 2628 2629 $length = 5; 2630 $char = 'a'; 2631 $x = 'aaaaabb'; 2632 $x =~ /(??{$char x $length})/x; # matches, there are 5 of 'a' 2633 2634 2635This final example contains both ordinary and pattern code 2636expressions. It detects whether a binary string C<1101010010001...> has a 2637Fibonacci spacing 0,1,1,2,3,5,... of the C<1>'s: 2638 2639 $x = "1101010010001000001"; 2640 $z0 = ''; $z1 = '0'; # initial conditions 2641 print "It is a Fibonacci sequence\n" 2642 if $x =~ /^1 # match an initial '1' 2643 (?: 2644 ((??{ $z0 })) # match some '0' 2645 1 # and then a '1' 2646 (?{ $z0 = $z1; $z1 .= $^N; }) 2647 )+ # repeat as needed 2648 $ # that is all there is 2649 /x; 2650 printf "Largest sequence matched was %d\n", length($z1)-length($z0); 2651 2652Remember that C<$^N> is set to whatever was matched by the last 2653completed capture group. This prints 2654 2655 It is a Fibonacci sequence 2656 Largest sequence matched was 5 2657 2658Ha! Try that with your garden variety regexp package... 2659 2660Note that the variables C<$z0> and C<$z1> are not substituted when the 2661regexp is compiled, as happens for ordinary variables outside a code 2662expression. Rather, the code expressions are evaluated when Perl 2663encounters them during the search for a match. 2664 2665The regexp without the C<//x> modifier is 2666 2667 /^1(?:((??{ $z0 }))1(?{ $z0 = $z1; $z1 .= $^N; }))+$/ 2668 2669which shows that spaces are still possible in the code parts. Nevertheless, 2670when working with code and conditional expressions, the extended form of 2671regexps is almost necessary in creating and debugging regexps. 2672 2673 2674=head2 Backtracking control verbs 2675 2676Perl 5.10 introduced a number of control verbs intended to provide 2677detailed control over the backtracking process, by directly influencing 2678the regexp engine and by providing monitoring techniques. As all 2679the features in this group are experimental and subject to change or 2680removal in a future version of Perl, the interested reader is 2681referred to L<perlre/"Special Backtracking Control Verbs"> for a 2682detailed description. 2683 2684Below is just one example, illustrating the control verb C<(*FAIL)>, 2685which may be abbreviated as C<(*F)>. If this is inserted in a regexp 2686it will cause to fail, just like at some mismatch between the pattern 2687and the string. Processing of the regexp continues like after any "normal" 2688failure, so that, for instance, the next position in the string or another 2689alternative will be tried. As failing to match doesn't preserve capture 2690buffers or produce results, it may be necessary to use this in 2691combination with embedded code. 2692 2693 %count = (); 2694 "supercalifragilisticexpialidoceous" =~ 2695 /([aeiou])(?{ $count{$1}++; })(*FAIL)/oi; 2696 printf "%3d '%s'\n", $count{$_}, $_ for (sort keys %count); 2697 2698The pattern begins with a class matching a subset of letters. Whenever 2699this matches, a statement like C<$count{'a'}++;> is executed, incrementing 2700the letter's counter. Then C<(*FAIL)> does what it says, and 2701the regexp engine proceeds according to the book: as long as the end of 2702the string hasn't been reached, the position is advanced before looking 2703for another vowel. Thus, match or no match makes no difference, and the 2704regexp engine proceeds until the entire string has been inspected. 2705(It's remarkable that an alternative solution using something like 2706 2707 $count{lc($_)}++ for split('', "supercalifragilisticexpialidoceous"); 2708 printf "%3d '%s'\n", $count2{$_}, $_ for ( qw{ a e i o u } ); 2709 2710is considerably slower.) 2711 2712 2713=head2 Pragmas and debugging 2714 2715Speaking of debugging, there are several pragmas available to control 2716and debug regexps in Perl. We have already encountered one pragma in 2717the previous section, S<C<use re 'eval';>>, that allows variable 2718interpolation and code expressions to coexist in a regexp. The other 2719pragmas are 2720 2721 use re 'taint'; 2722 $tainted = <>; 2723 @parts = ($tainted =~ /(\w+)\s+(\w+)/; # @parts is now tainted 2724 2725The C<taint> pragma causes any substrings from a match with a tainted 2726variable to be tainted as well. This is not normally the case, as 2727regexps are often used to extract the safe bits from a tainted 2728variable. Use C<taint> when you are not extracting safe bits, but are 2729performing some other processing. Both C<taint> and C<eval> pragmas 2730are lexically scoped, which means they are in effect only until 2731the end of the block enclosing the pragmas. 2732 2733 use re 'debug'; 2734 /^(.*)$/s; # output debugging info 2735 2736 use re 'debugcolor'; 2737 /^(.*)$/s; # output debugging info in living color 2738 2739The global C<debug> and C<debugcolor> pragmas allow one to get 2740detailed debugging info about regexp compilation and 2741execution. C<debugcolor> is the same as debug, except the debugging 2742information is displayed in color on terminals that can display 2743termcap color sequences. Here is example output: 2744 2745 % perl -e 'use re "debug"; "abc" =~ /a*b+c/;' 2746 Compiling REx `a*b+c' 2747 size 9 first at 1 2748 1: STAR(4) 2749 2: EXACT <a>(0) 2750 4: PLUS(7) 2751 5: EXACT <b>(0) 2752 7: EXACT <c>(9) 2753 9: END(0) 2754 floating `bc' at 0..2147483647 (checking floating) minlen 2 2755 Guessing start of match, REx `a*b+c' against `abc'... 2756 Found floating substr `bc' at offset 1... 2757 Guessed: match at offset 0 2758 Matching REx `a*b+c' against `abc' 2759 Setting an EVAL scope, savestack=3 2760 0 <> <abc> | 1: STAR 2761 EXACT <a> can match 1 times out of 32767... 2762 Setting an EVAL scope, savestack=3 2763 1 <a> <bc> | 4: PLUS 2764 EXACT <b> can match 1 times out of 32767... 2765 Setting an EVAL scope, savestack=3 2766 2 <ab> <c> | 7: EXACT <c> 2767 3 <abc> <> | 9: END 2768 Match successful! 2769 Freeing REx: `a*b+c' 2770 2771If you have gotten this far into the tutorial, you can probably guess 2772what the different parts of the debugging output tell you. The first 2773part 2774 2775 Compiling REx `a*b+c' 2776 size 9 first at 1 2777 1: STAR(4) 2778 2: EXACT <a>(0) 2779 4: PLUS(7) 2780 5: EXACT <b>(0) 2781 7: EXACT <c>(9) 2782 9: END(0) 2783 2784describes the compilation stage. C<STAR(4)> means that there is a 2785starred object, in this case C<'a'>, and if it matches, goto line 4, 2786i.e., C<PLUS(7)>. The middle lines describe some heuristics and 2787optimizations performed before a match: 2788 2789 floating `bc' at 0..2147483647 (checking floating) minlen 2 2790 Guessing start of match, REx `a*b+c' against `abc'... 2791 Found floating substr `bc' at offset 1... 2792 Guessed: match at offset 0 2793 2794Then the match is executed and the remaining lines describe the 2795process: 2796 2797 Matching REx `a*b+c' against `abc' 2798 Setting an EVAL scope, savestack=3 2799 0 <> <abc> | 1: STAR 2800 EXACT <a> can match 1 times out of 32767... 2801 Setting an EVAL scope, savestack=3 2802 1 <a> <bc> | 4: PLUS 2803 EXACT <b> can match 1 times out of 32767... 2804 Setting an EVAL scope, savestack=3 2805 2 <ab> <c> | 7: EXACT <c> 2806 3 <abc> <> | 9: END 2807 Match successful! 2808 Freeing REx: `a*b+c' 2809 2810Each step is of the form S<C<< n <x> <y> >>>, with C<< <x> >> the 2811part of the string matched and C<< <y> >> the part not yet 2812matched. The S<C<< | 1: STAR >>> says that Perl is at line number 1 2813n the compilation list above. See 2814L<perldebguts/"Debugging regular expressions"> for much more detail. 2815 2816An alternative method of debugging regexps is to embed C<print> 2817statements within the regexp. This provides a blow-by-blow account of 2818the backtracking in an alternation: 2819 2820 "that this" =~ m@(?{print "Start at position ", pos, "\n";}) 2821 t(?{print "t1\n";}) 2822 h(?{print "h1\n";}) 2823 i(?{print "i1\n";}) 2824 s(?{print "s1\n";}) 2825 | 2826 t(?{print "t2\n";}) 2827 h(?{print "h2\n";}) 2828 a(?{print "a2\n";}) 2829 t(?{print "t2\n";}) 2830 (?{print "Done at position ", pos, "\n";}) 2831 @x; 2832 2833prints 2834 2835 Start at position 0 2836 t1 2837 h1 2838 t2 2839 h2 2840 a2 2841 t2 2842 Done at position 4 2843 2844=head1 BUGS 2845 2846Code expressions, conditional expressions, and independent expressions 2847are I<experimental>. Don't use them in production code. Yet. 2848 2849=head1 SEE ALSO 2850 2851This is just a tutorial. For the full story on Perl regular 2852expressions, see the L<perlre> regular expressions reference page. 2853 2854For more information on the matching C<m//> and substitution C<s///> 2855operators, see L<perlop/"Regexp Quote-Like Operators">. For 2856information on the C<split> operation, see L<perlfunc/split>. 2857 2858For an excellent all-around resource on the care and feeding of 2859regular expressions, see the book I<Mastering Regular Expressions> by 2860Jeffrey Friedl (published by O'Reilly, ISBN 1556592-257-3). 2861 2862=head1 AUTHOR AND COPYRIGHT 2863 2864Copyright (c) 2000 Mark Kvale 2865All rights reserved. 2866 2867This document may be distributed under the same terms as Perl itself. 2868 2869=head2 Acknowledgments 2870 2871The inspiration for the stop codon DNA example came from the ZIP 2872code example in chapter 7 of I<Mastering Regular Expressions>. 2873 2874The author would like to thank Jeff Pinyan, Andrew Johnson, Peter 2875Haworth, Ronald J Kimball, and Joe Smith for all their helpful 2876comments. 2877 2878=cut 2879 2880