Chapter 22. Regular Expressions
Each character matches itself, unless it is one of the special characters +?.*^$()[{|. The special meaning of these characters can be escaped using a .
The multiline and single-line modes are discussed in the section Chapter 23.
.Matches any character, but not a newline. In singleline mode, matches newlines as well.
( . . . )Groups a series of pattern elements to a single element. The text the group matches is captured for later use. It is also assigned immediately to
$^Nto be used during the match, e.g., in a(?{ ... }).^Matches the beginning of the target. In multiline mode, also matches after every newline character.
$Matches the end of the line, or before a final newline character. In multiline mode, also matches before every newline character.
[. . .]Denotes a class of characters to match.
[^. . .]negates the class.- ...
|...|... Matches the alternatives from left to right, until one succeeds.
(?#text)Comment.
(?[ modifier ] : pattern)Acts like
(pattern)but does not capture the text it matches. modifier can be one or more ofi,m,s, orx. Modifiers can be switched off by preceding the letter(s) with a minus sign, e.g.,si-xm. See page 37 for the meaning of the modifiers.(?=pattern)Zero-width positive look-ahead assertion.
(?!pattern)Zero-width negative look-ahead assertion.
(?<=pattern)Zero-width positive look-behind assertion.
(?<!pattern)Zero-width negative look-behind assertion.
(?{code})Executes Perl code while matching. Always succeeds with zero width. Can be used as the condition in a conditional pattern selection. If not, the result of executing code is stored in
$^R.(??{code})Executes Perl code while matching. Interprets the result as a pattern.
(?>pattern)Like
(?:pattern), but prevents backtracking inside.(?(cond)ptrue [|pfalse ])Selects a pattern depending on the condition. cond should be the number of a parenthesized subpattern, or one of the zero-width look-ahead, look-behind, and evaluate assertions.
(?modifier)Embedded pattern-match modifier. modifier can be one or more of
i,m,s, orx. Modifiers can be switched off by preceding the letter(s) with a minus sign, e.g.,(?si-xm).
Quantified subpatterns match as many times as possible. When followed with a ? they match the minimum number of times. These are the quantifiers:
| Matches the preceding pattern element one or more times. |
| Matches zero or one times. |
| Matches zero or more times. |
| Denotes the minimum n and maximum m match count. |
Patterns are processed as double-quoted strings, so standard string escapes have their usual meaning (see Chapter 6). An exception is , which matches word boundaries, except in a character class, where it denotes a backspace again.
A escapes any special meaning of nonalphanumeric characters, but it turns most alphanumeric characters into something special:
| Refer to matched subexpressions, grouped with |
| Matches alphanumeric plus |
| Matches whitespace. |
| Matches numeric. |
| Matches the beginning of the string. |
Matches the end of the string or before a newline at the end of the string. | |
| Matches the physical end of the string. |
Matches word boundaries. | |
| Matches where the previous search with a |
| Matches a named property. |
| Matches extended Unicode combining character sequence. |
| Matches a single 8-bit byte. |
| |
POSIX classes are used inside character classes, like [[:alpha:]]. These are the POSIX classes and their Unicode property names:
[:alpha:] p{IsAlpha}Matches one alphabetic character.
[:alnum:] p{IsAlnum}Matches one alphanumeric character.
[:ascii:] p{IsASCII}Matches one ASCII character.
[:blank:] p{IsSpace}Matches one whitespace character, almost like
s.[:cntrl:] p{IsCntrl}[:digit:] p{IsDigit}Matches one numeric character, like
d.[:graph:] p{IsGraph}Matches one alphanumeric or punctuation character.
[:lower:] p{IsLower}Matches one lowercase character.
[:print:] p{IsPrint}Matches one alphanumeric or punctuation character or space character.
[:punct:] p{IsPunct}Matches one punctuation character.
[:space:] p{IsSpace}Matches one whitespace character, almost like
s.[:upper:] p{IsUpper}Matches one uppercase character.
[:word:] p{IsWord}Matches one word character, like
w.[:xdigit:] p{IsXDigit}Matches one hexadecimal digit.
In general, the "Is" prefix may be omitted for property names.
The equivalent of s is p{IsSpacePerl}.
The POSIX classes can be negated with a ^, e.g., [:^print:], the named properties by using P, e.g., P{IsPrint}.
See also $1. . . $9, $+, $` , $&, $', $^R, and $^N, and @- and @+.
With modifier x, whitespace and comments can be embedded in the patterns.
Regular expression patterns can be compiled and used as values with the qr quoting operator: qr/string/modifiers compiles string as a pattern according to the (optional) modifiers, and returns the compiled pattern as a scalar value.
