perlman
gods
regex
<HR>
<P>
<H1><A NAME="NAME">NAME</A></H1>
<P>
perlre - Perl regular expressions
<P>
<HR>
<H1><A NAME="DESCRIPTION">DESCRIPTION</A></H1>
<P>
This page describes the syntax of regular expressions in Perl. For a
description of how to <EM>use</EM> regular expressions in matching operations, plus various examples of the
same, see discussion of [perlman:perlop], [perlman:perlop], [perlman:perlop] and <CODE>??</CODE> in [perlman:perlop].
<P>
The matching operations can have various modifiers. The modifiers that
relate to the interpretation of the regular expression inside are listed
below. For the modifiers that alter the way a regular expression is used by
Perl, see [perlman:perlop] and
[perlman:perlop].
<DL COMPACT>
<DT><STRONG><A NAME="item_i">i</A></STRONG>
<DD>
Do case-insensitive pattern matching.
<P>
If <CODE>use locale</CODE> is in effect, the case map is taken from the current locale. See [perlman:perllocale|the perllocale manpage].
<P><DT><STRONG><A NAME="item_m">m</A></STRONG>
<DD>
Treat string as multiple lines. That is, change ``^'' and ``$'' from
matching at only the very start or end of the string to the start or end of
any line anywhere within the string,
<P><DT><STRONG><A NAME="item_s">s</A></STRONG>
<DD>
Treat string as single line. That is, change ``.'' to match any character
whatsoever, even a newline, which it normally would not match.
<P>
The <A HREF="#item_s">/s</A> and <A HREF="#item_m">/m</A> modifiers both override the <CODE>$*</CODE> setting. That is, no matter what <CODE>$*</CODE> contains, <A HREF="#item_s">/s</A> without <A HREF="#item_m">/m</A> will force ``^'' to match only at the beginning of the string and ``$'' to
match only at the end (or just before a newline at the end) of the string.
Together, as <CODE>/ms</CODE>, they let the ``.'' match any character whatsoever, while
yet allowing ``^'' and ``$'' to match, respectively, just after and just
before newlines within the string.
<P><DT><STRONG><A NAME="item_x">x</A></STRONG>
<DD>
Extend your pattern's legibility by permitting whitespace and comments.
</DL>
<P>
These are usually written as ``the <CODE>/x</CODE> modifier'', even though the delimiter in question might not actually be a
slash. In fact, any of these modifiers may also be embedded within the
regular expression itself using the new <CODE>(?...)</CODE> construct. See below.
<P>
The <CODE>/x</CODE> modifier itself needs a little more explanation. It tells the regular
expression parser to ignore whitespace that is neither backslashed nor
within a character class. You can use this to break up your regular
expression into (slightly) more readable parts. The <CODE>#</CODE>
character is also treated as a metacharacter introducing a comment, just as
in ordinary Perl code. This also means that if you want real whitespace or <CODE>#</CODE> characters in the pattern (outside of a character class, where they are
unaffected by <CODE>/x</CODE>), that you'll either have to escape them or encode them using octal or hex
escapes. Taken together, these features go a long way towards making Perl's
regular expressions more readable. Note that you have to be careful not to
include the pattern delimiter in the comment--perl has no way of knowing
you did not intend to close the pattern early. See the C-comment deletion
code in [perlman:perlop|the perlop manpage].
<P>
<HR>
<H2><A NAME="Regular_Expressions">Regular Expressions</A></H2>
<P>
The patterns used in pattern matching are regular expressions such as those supplied in the Version 8 regex routines. (In fact, the routines are derived (distantly) from Henry Spencer's freely redistributable reimplementation of the
<FONT SIZE=-1>V8</FONT> routines.) See
<A HREF="#Version_8_Regular_Expressions">Version 8 Regular Expressions</A> for details.
<P>
In particular the following metacharacters have their standard <EM>egrep</EM>-ish meanings:
<P>
<PRE> \ Quote the next metacharacter
^ Match the beginning of the line
. Match any character (except newline)
$ Match the end of the line (or before newline at the end)
| Alternation
() Grouping
[] Character class
</PRE>
<P>
By default, the ``^'' character is guaranteed to match at only the
beginning of the string, the ``$'' character at only the end (or before the
newline at the end) and Perl does certain optimizations with the assumption
that the string contains only one line. Embedded newlines will not be
matched by ``^'' or ``$''. You may, however, wish to treat a string as a
multi-line buffer, such that the ``^'' will match after any newline within
the string, and ``$'' will match before any newline. At the cost of a
little more overhead, you can do this by using the <CODE>/m</CODE> modifier on the
pattern match operator. (Older programs did this by setting <CODE>$*</CODE>, but this practice is now deprecated.)
<P>
To facilitate multi-line substitutions, the ``.'' character never matches a
newline unless you use the <A HREF="#item_s">/s</A> modifier, which in effect tells Perl to pretend the string is a single
line--even if it isn't. The <A HREF="#item_s">/s</A> modifier also overrides the setting of <CODE>$*</CODE>, in case you have some (badly behaved) older code that sets it in another
module.
<P>
The following standard quantifiers are recognized:
<P>
<PRE> * Match 0 or more times
+ Match 1 or more times
? Match 1 or 0 times
{n} Match exactly n times
{n,} Match at least n times
{n,m} Match at least n but not more than m times
</PRE>
<P>
(If a curly bracket occurs in any other context, it is treated as a regular
character.) The ``*'' modifier is equivalent to <CODE>{0,}</CODE>, the ``+'' modifier to <CODE>{1,}</CODE>, and the ``?'' modifier to <CODE>{0,1}</CODE>. n and m are limited to integral values less than 65536.
<P>
By default, a quantified subpattern is ``greedy'', that is, it will match
as many times as possible (given a particular starting location) while
still allowing the rest of the pattern to match. If you want it to match
the minimum number of times possible, follow the quantifier with a ``?''.
Note that the meanings don't change, just the ``greediness'':
<P>
<PRE> *? Match 0 or more times
+? Match 1 or more times
?? Match 0 or 1 time
{n}? Match exactly n times
{n,}? Match at least n times
{n,m}? Match at least n but not more than m times
</PRE>
<P>
Because patterns are processed as double quoted strings, the following also
work:
<P>
<PRE> \t tab (HT, TAB)
\n newline (LF, NL)
\r return (CR)
\f form feed (FF)
\a alarm (bell) (BEL)
\e escape (think troff) (ESC)
\033 octal char (think of a PDP-11)
\x1B hex char
\c[ control char
\l lowercase next char (think vi)
\u uppercase next char (think vi)
\L lowercase till \E (think vi)
\U uppercase till \E (think vi)
\E end case modification (think vi)
\Q quote (disable) pattern metacharacters till \E
</PRE>
<P>
If <CODE>use locale</CODE> is in effect, the case map used by <CODE>\l</CODE>, <CODE>\L</CODE>, <CODE>\u</CODE>
and <CODE>\U</CODE> is taken from the current locale. See [perlman:perllocale|the perllocale manpage].
<P>
You cannot include a literal <CODE>$</CODE> or <CODE>@</CODE> within a <CODE>\Q</CODE> sequence. An unescaped <CODE>$</CODE> or <CODE>@</CODE> interpolates the corresponding variable, while escaping will cause the
literal string <CODE>\$</CODE> to be matched. You'll need to write something like <CODE>m/\Quser\E\@\Qhost/</CODE>.
<P>
In addition, Perl defines the following:
<P>
<PRE> \w Match a "word" character (alphanumeric plus "_")
\W Match a non-word character
\s Match a whitespace character
\S Match a non-whitespace character
\d Match a digit character
\D Match a non-digit character
</PRE>
<P>
<FONT SIZE=-1>A</FONT> <CODE>\w</CODE> matches a single alphanumeric character, not a whole word. To match a word
you'd need to say <CODE>\w+</CODE>. If <CODE>use locale</CODE> is in effect, the list of alphabetic characters generated by <CODE>\w</CODE> is taken from the current locale. See [perlman:perllocale|the perllocale manpage]. You may use <CODE>\w</CODE>, <CODE>\W</CODE>,
<CODE>\s</CODE>, <CODE>\S</CODE>, <CODE>\d</CODE>, and <CODE>\D</CODE> within character classes (though not as either end of a range).
<P>
Perl defines the following zero-width assertions:
<P>
<PRE> \b Match a word boundary
\B Match a non-(word boundary)
\A Match only at beginning of string
\Z Match only at end of string, or before newline at the end
\z Match only at end of string
\G Match only where previous m//g left off (works only with /g)
</PRE>
<P>
<FONT SIZE=-1>A</FONT> word boundary (<CODE>\b</CODE>) is defined as a spot between two characters that has a <CODE>\w</CODE> on one side of it and a <CODE>\W</CODE> on the other side of it (in either order), counting the imaginary
characters off the beginning and end of the string as matching a <CODE>\W</CODE>. (Within character classes <CODE>\b</CODE>
represents backspace rather than a word boundary.) The <CODE>\A</CODE> and <CODE>\Z</CODE> are just like ``^'' and ``$'', except that they won't match multiple times
when the
[perlman:perlop] modifier is used, while ``^'' and ``$'' will match at every internal line
boundary. To match the actual end of the string, not ignoring newline, you
can use <CODE>\z</CODE>. The <CODE>\G</CODE> assertion can be used to chain global matches (using <CODE>m//g</CODE>), as described in
[perlman:perlop].
<P>
It is also useful when writing <CODE>lex</CODE>-like scanners, when you have several patterns that you want to match
against consequent substrings of your string, see the previous reference.
The actual location where <CODE>\G</CODE> will match can also be influenced by using [perlfunc:pos|pos()] as an lvalue. See [perlfunc:pos|pos].
<P>
When the bracketing construct <CODE>( ... )</CODE> is used, \<digit> matches the digit'th substring. Outside of the pattern, always use ``$'' instead of ``\'' in front of the digit. (While the \<digit> notation can on rare occasion work outside the current pattern, this should not be relied upon. See the
<FONT SIZE=-1>WARNING</FONT> below.) The scope of $<digit> (and
<CODE>$`</CODE>, <CODE>$&</CODE>, and <CODE>$'</CODE>) extends to the end of the enclosing
<FONT SIZE=-1>BLOCK</FONT> or eval string, or to the next successful
pattern match, whichever comes first. If you want to use parentheses to
delimit a subpattern (e.g., a set of alternatives) without saving it as a
subpattern, follow the ( with a ?:.
<P>
You may have as many parentheses as you wish. If you have more than 9
substrings, the variables $10, $11, ... refer to the corresponding
substring. Within the pattern, \10, \11, etc. refer back to substrings if
there have been at least that many left parentheses before the
backreference. Otherwise (for backward compatibility) \10 is the same as
\010, a backspace, and \11 the same as \011, a tab. And so on. (\1 through
\9 are always backreferences.)
<P>
<CODE>$+</CODE> returns whatever the last bracket match matched. <CODE>$&</CODE> returns the entire matched string. (<CODE>$0</CODE> used to return the same thing, but not any more.) <CODE>$`</CODE> returns everything before the matched string. <CODE>$'</CODE> returns everything after the matched string. Examples:
<P>
<PRE> s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
</PRE>
<P>
<PRE> if (/Time: (..):(..):(..)/) {
$hours = $1;
$minutes = $2;
$seconds = $3;
}
</PRE>
<P>
Once perl sees that you need one of <CODE>$&</CODE>, <CODE>$`</CODE> or <CODE>$'</CODE> anywhere in the program, it has to provide them on each and every pattern
match. This can slow your program down. The same mechanism that handles
these provides for the use of $1, $2, etc., so you pay the same price for
each pattern that contains capturing parentheses. But if you never use
$&, etc., in your script, then patterns <EM>without</EM> capturing parentheses won't be penalized. So avoid $&, $', and $` if
you can, but if you can't (and some algorithms really appreciate them),
once you've used them once, use them at will, because you've already paid
the price. As of 5.005, $& is not so costly as the other two.
<P>
Backslashed metacharacters in Perl are alphanumeric, such as <CODE>\b</CODE>, <CODE>\w</CODE>, <CODE>\n</CODE>. Unlike some other regular expression languages, there are no backslashed
symbols that aren't alphanumeric. So anything that looks like \\, \(, \),
\<, \>, \{, or \} is always interpreted as a literal character, not
a metacharacter. This was once used in a common idiom to disable or quote
the special meanings of regular expression metacharacters in a string that
you want to use for a pattern. Simply quote all non-alphanumeric
characters:
<P>
<PRE> $pattern =~ s/(\W)/\\$1/g;
</PRE>
<P>
Now it is much more common to see either the
<CODE>quotemeta()</CODE> function or
the <CODE>\Q</CODE> escape sequence used to disable all metacharacters' special meanings like
this:
<P>
<PRE> /$unquoted\Q$quoted\E$unquoted/
</PRE>
<P>
Perl defines a consistent extension syntax for regular expressions. The
syntax is a pair of parentheses with a question mark as the first thing
within the parentheses (this was a syntax error in older versions of Perl).
The character after the question mark gives the function of the extension.
Several extensions are already supported:
<DL>
<DT><STRONG><A NAME="item__text_">(?#text)</A></STRONG><P>
<DD>
<FONT SIZE=-1>A</FONT> comment. The text is ignored. If the <CODE>/x</CODE> switch is used to enable whitespace formatting, a simple <CODE>#</CODE> will suffice. Note that perl closes the comment as soon as it sees a <CODE>)</CODE>, so there is no way to put a literal
<CODE>)</CODE> in the comment.
<P><DT><STRONG><A NAME="item__pattern_">(?:pattern)</A></STRONG><DD>
<DT><STRONG><A NAME="item__imsx_imsx_pattern_">(?imsx-imsx:pattern)</A></STRONG><P>
<DD>
This is for clustering, not capturing; it groups subexpressions like
``()'', but doesn't make backreferences as ``()'' does. So
<P>
<PRE> @fields = split(/\b(?:a|b|c)\b/)
</PRE>
<P>
is like
<P>
<PRE> @fields = split(/\b(a|b|c)\b/)
</PRE>
<P>
but doesn't spit out extra fields.
<P>
The letters between <CODE>?</CODE> and <CODE>:</CODE> act as flags modifiers, see
<EM><CODE>(?imsx-imsx)</CODE></EM>. In particular,
<P>
<PRE> /(?s-i:more.*than).*million/i
</PRE>
<P>
is equivalent to more verbose
<P>
<PRE> /(?:(?s-i)more.*than).*million/i
</PRE>
<DT><STRONG><A NAME="item__pattern_">(?=pattern)</A></STRONG><P>
<DD>
<FONT SIZE=-1>A</FONT> zero-width positive lookahead assertion. For
example, <CODE>/\w+(?=\t)/</CODE>
matches a word followed by a tab, without including the tab in <CODE>$&</CODE>.
<P><DT><STRONG><A NAME="item__pattern_">(?!pattern)</A></STRONG><P>
<DD>
<FONT SIZE=-1>A</FONT> zero-width negative lookahead assertion. For example <CODE>/foo(?!bar)/</CODE>
matches any occurrence of ``foo'' that isn't followed by ``bar''. Note however that lookahead and lookbehind are
<FONT SIZE=-1>NOT</FONT> the same thing. You cannot use this for lookbehind.
<P>
If you are looking for a ``bar'' that isn't preceded by a ``foo'', <CODE>/(?!foo)bar/</CODE>
will not do what you want. That's because the <CODE>(?!foo)</CODE> is just saying that the next thing cannot be ``foo''--and it's not, it's a
``bar'', so ``foobar'' will match. You would have to do something like <CODE>/(?!foo)...bar/</CODE> for that. We say ``like'' because there's the case of your ``bar'' not
having three characters before it. You could cover that this way: <CODE>/(?:(?!foo)...|^.{0,2})bar/</CODE>. Sometimes it's still easier just to say:
<P>
<PRE> if (/bar/ && $` !~ /foo$/)
</PRE>
<P>
For lookbehind see below.
<P><DT><STRONG><A NAME="item__lt_pattern_">(?<=pattern)</A></STRONG><P>
<DD>
<FONT SIZE=-1>A</FONT> zero-width positive lookbehind assertion. For
example, <CODE>/(?<=\t)\w+/</CODE>
matches a word following a tab, without including the tab in <CODE>$&</CODE>. Works only for fixed-width lookbehind.
<P><DT><STRONG><A NAME="item_C">(?<!pattern)</A></STRONG><P>
<DD>
<FONT SIZE=-1>A</FONT> zero-width negative lookbehind assertion. For
example <CODE>/(?<!bar)foo/</CODE>
matches any occurrence of ``foo'' that isn't following ``bar''. Works only
for fixed-width lookbehind.
<P><DT><STRONG>(?{ code })</STRONG><P>
<DD>
Experimental ``evaluate any Perl code'' zero-width assertion. Always
succeeds. <CODE>code</CODE> is not interpolated. Currently the rules to determine where the <CODE>code</CODE> ends are somewhat convoluted.
<P>
The <CODE>code</CODE> is properly scoped in the following sense: if the assertion is backtracked
(compare <A HREF="#Backtracking">Backtracking</A>), all the changes introduced after
[perlfunc:local|local]isation are undone, so
<P>
<PRE> $_ = 'a' x 8;
m<
(?{ $cnt = 0 }) # Initialize $cnt.
(
a
(?{
local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
})
)*
aaaa
(?{ $res = $cnt }) # On success copy to non-localized
# location.
>x;
</PRE>
<P>
will set <CODE>$res = 4</CODE>. Note that after the match <CODE>$cnt</CODE> returns to the globally
introduced value 0, since the scopes which restrict [perlfunc:local|local] statements are unwound.
<P>
This assertion may be used as <EM>(?(condition)yes-pattern)</EM>
switch. If <EM>not</EM> used in this way, the result of evaluation of <CODE>code</CODE>
is put into variable
<FONT SIZE=-1>$^R.</FONT> This happens immediately, so
<FONT SIZE=-1>$^R</FONT> can be used from other
<CODE>(?{ code })</CODE> assertions inside the same regular expression.
<P>
The above assignment to
<FONT SIZE=-1>$^R</FONT> is properly localized, thus the old value of
<FONT SIZE=-1>$^R</FONT> is restored if the assertion is backtracked (compare
<A HREF="#Backtracking">Backtracking</A>).
<P>
Due to security concerns, this construction is not allowed if the regular
expression involves run-time interpolation of variables, unless
<CODE>use re 'eval'</CODE> pragma is used (see <U>the re manpage</U><!--../lib/re.html-->), or the variables contain results of
<CODE>qr()</CODE> operator (see [perlman:perlop]).
<P>
This restriction is due to the wide-spread (questionable) practice of using
the construct
<P>
<PRE> $re = <>;
chomp $re;
$string =~ /$re/;
</PRE>
<P>
without tainting. While this code is frowned upon from security point of
view, when <CODE>(?{})</CODE> was introduced, it was considered bad to add
<EM>new</EM> security holes to existing scripts.
<P>
<STRONG>NOTE:</STRONG> Use of the above insecure snippet without also enabling taint mode is to be
severely frowned upon. <CODE>use re 'eval'</CODE> does not disable tainting checks, thus to allow <CODE>$re</CODE> in the
above snippet to contain <CODE>(?{})</CODE>
<EM>with tainting enabled</EM>, one needs both <CODE>use re 'eval'</CODE> and untaint the $re.
<P><DT><STRONG><A NAME="item__gtpattern_">(?>pattern)</A></STRONG><P>
<DD>
An ``independent'' subexpression. Matches the substring that a
<EM>standalone</EM> <CODE>pattern</CODE> would match if anchored at the given position,
<STRONG>and only this substring</STRONG>.
<P>
Say, <CODE>^(?>a*)ab</CODE> will never match, since <CODE>(?>a*)</CODE> (anchored at the beginning of string, as above) will match <EM>all</EM> characters
<CODE>a</CODE> at the beginning of string, leaving no <CODE>a</CODE> for <CODE>ab</CODE> to match. In contrast, <CODE>a*ab</CODE> will match the same as <CODE>a+b</CODE>, since the match of the subgroup <CODE>a*</CODE> is influenced by the following group <CODE>ab</CODE> (see
<A HREF="#Backtracking">Backtracking</A>). In particular, <CODE>a*</CODE> inside <CODE>a*ab</CODE> will match fewer characters than a standalone <CODE>a*</CODE>, since this makes the tail match.
<P>
An effect similar to <CODE>(?>pattern)</CODE> may be achieved by
<P>
<PRE> (?=(pattern))\1
</PRE>
<P>
since the lookahead is in <EM>"logical"</EM> context, thus matches the same substring as a standalone <CODE>a+</CODE>. The following <CODE>\1</CODE> eats the matched string, thus making a zero-length assertion into an
analogue of
<CODE>(?>...)</CODE>. (The difference between these two constructs is that the second one uses
a catching group, thus shifting ordinals of backreferences in the rest of a
regular expression.)
<P>
This construct is useful for optimizations of ``eternal'' matches, because
it will not backtrack (see <A HREF="#Backtracking">Backtracking</A>).
<P>
<PRE> m{ \(
(
[^()]+
|
\( [^()]* \)
)+
\)
}x
</PRE>
<P>
That will efficiently match a nonempty group with matching
two-or-less-level-deep parentheses. However, if there is no such group, it
will take virtually forever on a long string. That's because there are so
many different ways to split a long string into several substrings. This is
what <CODE>(.+)+</CODE> is doing, and <CODE>(.+)+</CODE> is similar to a subpattern of the above pattern. Consider that the above
pattern detects no-match on <CODE>((()aaaaaaaaaaaaaaaaaa</CODE> in several seconds, but that each extra letter doubles this time. This
exponential performance will make it appear that your program has hung.
<P>
However, a tiny modification of this pattern
<P>
<PRE> m{ \(
(
(?> [^()]+ )
|
\( [^()]* \)
)+
\)
}x
</PRE>
<P>
which uses <CODE>(?>...)</CODE> matches exactly when the one above does (verifying this yourself would be a
productive exercise), but finishes in a fourth the time when used on a
similar string with 1000000 <CODE>a</CODE>s. Be aware, however, that this pattern currently triggers a warning
message under
<STRONG>-w</STRONG> saying it <CODE>"matches the null string many times"</CODE>):
<P>
On simple groups, such as the pattern <CODE>(?</CODE> [^()]+ )>, a comparable effect may be achieved by negative lookahead, as
in <CODE>[^()]+ (?! [^()] )</CODE>. This was only 4 times slower on a string with 1000000 <CODE>a</CODE>s.
<P><DT><STRONG><A NAME="item__condition_yes_pattern_no_patt">(?(condition)yes-pattern|no-pattern)</A></STRONG><DD>
<DT><STRONG><A NAME="item__condition_yes_pattern_">(?(condition)yes-pattern)</A></STRONG><P>
<DD>
Conditional expression. <CODE>(condition)</CODE> should be either an integer in parentheses (which is valid if the
corresponding pair of parentheses matched), or
lookahead/lookbehind/evaluate zero-width assertion.
<P>
Say,
<P>
<PRE> m{ ( \( )?
[^()]+
(?(1) \) )
}x
</PRE>
<P>
matches a chunk of non-parentheses, possibly included in parentheses
themselves.
<P><DT><STRONG><A NAME="item__imsx_imsx_">(?imsx-imsx)</A></STRONG><P>
<DD>
One or more embedded pattern-match modifiers. This is particularly useful
for patterns that are specified in a table somewhere, some of which want to
be case sensitive, and some of which don't. The case insensitive ones need
to include merely <CODE>(?i)</CODE> at the front of the pattern. For example:
<P>
<PRE> $pattern = "foobar";
if ( /$pattern/i ) { }
</PRE>
<P>
<PRE> # more flexible:
</PRE>
<P>
<PRE> $pattern = "(?i)foobar";
if ( /$pattern/ ) { }
</PRE>
<P>
Letters after <CODE>-</CODE> switch modifiers off.
<P>
These modifiers are localized inside an enclosing group (if any). Say,
<P>
<PRE> ( (?i) blah ) \s+ \1
</PRE>
<P>
(assuming <CODE>x</CODE> modifier, and no <CODE>i</CODE> modifier outside of this group) will match a repeated (<EM>including the case</EM>!) word <CODE>blah</CODE> in any case.
</DL>
<P>
<FONT SIZE=-1>A</FONT> question mark was chosen for this and for the new
minimal-matching construct because 1) question mark is pretty rare in older
regular expressions, and 2) whenever you see one, you should stop and
``question'' exactly what is going on. That's psychology...
<P>
<HR>
<H2><A NAME="Backtracking">Backtracking</A></H2>
<P>
<FONT SIZE=-1>A</FONT> fundamental feature of regular expression matching
involves the notion called <EM>backtracking</EM>, which is currently used (when needed) by all regular expression
quantifiers, namely <CODE>*</CODE>, <CODE>*?</CODE>, <CODE>+</CODE>,
<CODE>+?</CODE>, <CODE>{n,m}</CODE>, and <CODE>{n,m}?</CODE>.
<P>
For a regular expression to match, the <EM>entire</EM> regular expression must match, not just part of it. So if the beginning of
a pattern containing a quantifier succeeds in a way that causes later parts
in the pattern to fail, the matching engine backs up and recalculates the
beginning part--that's why it's called backtracking.
<P>
Here is an example of backtracking: Let's say you want to find the word
following ``foo'' in the string ``Food is on the foo table.'':
<P>
<PRE> $_ = "Food is on the foo table.";
if ( /\b(foo)\s+(\w+)/i ) {
print "$2 follows $1.\n";
}
</PRE>
<P>
When the match runs, the first part of the regular expression (<CODE>\b(foo)</CODE>) finds a possible match right at the beginning of the string, and loads up
<CODE>$1</CODE> with ``Foo''. However, as soon as the matching engine sees
that there's no whitespace following the ``Foo'' that it had saved in $1,
it realizes its mistake and starts over again one character after where it
had the tentative match. This time it goes all the way until the next
occurrence of ``foo''. The complete regular expression matches this time,
and you get the expected output of ``table follows foo.''
<P>
Sometimes minimal matching can help a lot. Imagine you'd like to match
everything between ``foo'' and ``bar''. Initially, you write something like
this:
<P>
<PRE> $_ = "The food is under the bar in the barn.";
if ( /foo(.*)bar/ ) {
print "got <$1>\n";
}
</PRE>
<P>
Which perhaps unexpectedly yields:
<P>
<PRE> got <d is under the bar in the >
</PRE>
<P>
That's because <CODE>.*</CODE> was greedy, so you get everything between the
<EM>first</EM> ``foo'' and the <EM>last</EM> ``bar''. In this case, it's more effective to use minimal matching to make
sure you get the text between a ``foo'' and the first ``bar'' thereafter.
<P>
<PRE> if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
got <d is under the >
</PRE>
<P>
Here's another example: let's say you'd like to match a number at the end
of a string, and you also want to keep the preceding part the match. So you
write this:
<P>
<PRE> $_ = "I have 2 numbers: 53147";
if ( /(.*)(\d*)/ ) { # Wrong!
print "Beginning is <$1>, number is <$2>.\n";
}
</PRE>
<P>
That won't work at all, because <CODE>.*</CODE> was greedy and gobbled up the whole string. As <CODE>\d*</CODE> can match on an empty string the complete regular expression matched
successfully.
<P>
<PRE> Beginning is <I have 2 numbers: 53147>, number is <>.
</PRE>
<P>
Here are some variants, most of which don't work:
<P>
<PRE> $_ = "I have 2 numbers: 53147";
@pats = qw{
(.*)(\d*)
(.*)(\d+)
(.*?)(\d*)
(.*?)(\d+)
(.*)(\d+)$
(.*?)(\d+)$
(.*)\b(\d+)$
(.*\D)(\d+)$
};
</PRE>
<P>
<PRE> for $pat (@pats) {
printf "%-12s ", $pat;
if ( /$pat/ ) {
print "<$1> <$2>\n";
} else {
print "FAIL\n";
}
}
</PRE>
<P>
That will print out:
<P>
<PRE> (.*)(\d*) <I have 2 numbers: 53147> <>
(.*)(\d+) <I have 2 numbers: 5314> <7>
(.*?)(\d*) <> <>
(.*?)(\d+) <I have > <2>
(.*)(\d+)$ <I have 2 numbers: 5314> <7>
(.*?)(\d+)$ <I have 2 numbers: > <53147>
(.*)\b(\d+)$ <I have 2 numbers: > <53147>
(.*\D)(\d+)$ <I have 2 numbers: > <53147>
</PRE>
<P>
As you see, this can be a bit tricky. It's important to realize that a
regular expression is merely a set of assertions that gives a definition of
success. There may be 0, 1, or several different ways that the definition
might succeed against a particular string. And if there are multiple ways
it might succeed, you need to understand backtracking to know which variety
of success you will achieve.
<P>
When using lookahead assertions and negations, this can all get even
tricker. Imagine you'd like to find a sequence of non-digits not followed
by ``123''. You might try to write that as
<P>
<PRE> $_ = "ABC123";
if ( /^\D*(?!123)/ ) { # Wrong!
print "Yup, no 123 in $_\n";
}
</PRE>
<P>
But that isn't going to match; at least, not the way you're hoping. It
claims that there is no 123 in the string. Here's a clearer picture of why
it that pattern matches, contrary to popular expectations:
<P>
<PRE> $x = 'ABC123' ;
$y = 'ABC445' ;
</PRE>
<P>
<PRE> print "1: got $1\n" if $x =~ /^(ABC)(?!123)/ ;
print "2: got $1\n" if $y =~ /^(ABC)(?!123)/ ;
</PRE>
<P>
<PRE> print "3: got $1\n" if $x =~ /^(\D*)(?!123)/ ;
print "4: got $1\n" if $y =~ /^(\D*)(?!123)/ ;
</PRE>
<P>
This prints
<P>
<PRE> 2: got ABC
3: got AB
4: got ABC
</PRE>
<P>
You might have expected test 3 to fail because it seems to a more general
purpose version of test 1. The important difference between them is that
test 3 contains a quantifier (<CODE>\D*</CODE>) and so can use backtracking, whereas test 1 will not. What's happening is
that you've asked ``Is it true that at the start of $x, following 0 or more
non-digits, you have something that's not 123?'' If the pattern matcher had
let <CODE>\D*</CODE> expand to
<FONT SIZE=-1>``ABC'',</FONT> this would have caused the whole pattern to
fail. The search engine will initially match <CODE>\D*</CODE> with
<FONT SIZE=-1>``ABC''.</FONT> Then it will try to match <CODE>(?!123</CODE> with ``123'', which of course fails. But because a quantifier (<CODE>\D*</CODE>) has been used in the regular expression, the search engine can backtrack
and retry the match differently in the hope of matching the complete
regular expression.
<P>
The pattern really, <EM>really</EM> wants to succeed, so it uses the standard pattern back-off-and-retry and
lets <CODE>\D*</CODE> expand to just
<FONT SIZE=-1>``AB''</FONT> this time. Now there's indeed something following
<FONT SIZE=-1>``AB''</FONT> that is not ``123''. It's in fact
<FONT SIZE=-1>``C123'',</FONT> which suffices.
<P>
We can deal with this by using both an assertion and a negation. We'll say
that the first part in <CODE>$1</CODE> must be followed by a digit, and in
fact, it must also be followed by something that's not ``123''. Remember
that the lookaheads are zero-width expressions--they only look, but don't
consume any of the string in their match. So rewriting this way produces
what you'd expect; that is, case 5 will fail, but case 6 succeeds:
<P>
<PRE> print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/ ;
print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/ ;
</PRE>
<P>
<PRE> 6: got ABC
</PRE>
<P>
In other words, the two zero-width assertions next to each other work as
though they're ANDed together, just as you'd use any builtin assertions: <CODE>/^$/</CODE>
matches only if you're at the beginning of the line
<FONT SIZE=-1>AND</FONT> the end of the line simultaneously. The deeper underlying truth is that juxtaposition in regular expressions always means
<FONT SIZE=-1>AND,</FONT> except when you write an explicit
<FONT SIZE=-1>OR</FONT> using the vertical bar.
<CODE>/ab/</CODE> means match ``a''
<FONT SIZE=-1>AND</FONT> (then) match ``b'', although the attempted matches
are made at different positions because ``a'' is not a zero-width
assertion, but a one-width assertion.
<P>
One warning: particularly complicated regular expressions can take
exponential time to solve due to the immense number of possible ways they
can use backtracking to try match. For example this will take a very long
time to run
<P>
<PRE> /((a{0,5}){0,5}){0,5}/
</PRE>
<P>
And if you used <CODE>*</CODE>'s instead of limiting it to 0 through 5 matches, then it would take
literally forever--or until you ran out of stack space.
<P>
<FONT SIZE=-1>A</FONT> powerful tool for optimizing such beasts is
``independent'' groups, which do not backtrace (see <EM><CODE>(?>pattern)</CODE></EM>). Note also that zero-length lookahead/lookbehind assertions will not
backtrace to make the tail match, since they are in ``logical'' context:
only the fact whether they match or not is considered relevant. For an
example where side-effects of a lookahead <EM>might</EM> have influenced the following match, see <EM><CODE>(?>pattern)</CODE></EM>.
<P>
<HR>
<H2><A NAME="Version_8_Regular_Expressions">Version 8 Regular Expressions</A></H2>
<P>
In case you're not familiar with the ``regular'' Version 8 regex routines,
here are the pattern-matching rules not described above.
<P>
Any single character matches itself, unless it is a <EM>metacharacter</EM>
with a special meaning described here or above. You can cause characters that normally function as metacharacters to be interpreted literally by prefixing them with a ``\'' (e.g., ``\.'' matches a ``.'', not any character; ``\\'' matches a ``\'').
<FONT SIZE=-1>A</FONT> series of characters matches that series of characters in the target string, so the pattern
<CODE>blurfl</CODE>
would match ``blurfl'' in the target string.
<P>
You can specify a character class, by enclosing a list of characters in <CODE>[]</CODE>, which will match any one character from the list. If the first character
after the ``['' is ``^'', the class matches any character not in the list.
Within a list, the ``-'' character is used to specify a range, so that <CODE>a-z</CODE> represents all characters between ``a'' and ``z'', inclusive. If you want
``-'' itself to be a member of a class, put it at the start or end of the
list, or escape it with a backslash. (The following all specify the same
class of three characters: <CODE>[-az]</CODE>,
<CODE>[az-]</CODE>, and <CODE>[a\-z]</CODE>. All are different from <CODE>[a-z]</CODE>, which specifies a class containing twenty-six characters.)
<P>
Characters may be specified using a metacharacter syntax much like that used in
<FONT SIZE=-1>C:</FONT> ``\n'' matches a newline, ``\t'' a tab, ``\r'' a carriage return, ``\f'' a form feed, etc. More generally, \
<EM>nnn</EM>, where <EM>nnn</EM> is a string of octal digits, matches the character whose
<FONT SIZE=-1>ASCII</FONT> value is <EM>nnn</EM>. Similarly, \x<EM>nn</EM>, where <EM>nn</EM> are hexadecimal digits, matches the character whose
<FONT SIZE=-1>ASCII</FONT> value is <EM>nn</EM>. The expression \c<EM>x</EM> matches the
<FONT SIZE=-1>ASCII</FONT> character control-<EM>x</EM>. Finally, the ``.'' metacharacter matches any character except ``\n''
(unless you use <A HREF="#item_s">/s</A>).
<P>
You can specify a series of alternatives for a pattern using ``|'' to
separate them, so that <CODE>fee|fie|foe</CODE> will match any of ``fee'', ``fie'', or ``foe'' in the target string (as
would <CODE>f(e|i|o)e</CODE>). The first alternative includes everything from the last pattern
delimiter (``('', ``['', or the beginning of the pattern) up to the first
``|'', and the last alternative contains everything from the last ``|'' to
the next pattern delimiter. For this reason, it's common practice to
include alternatives in parentheses, to minimize confusion about where they
start and end.
<P>
Alternatives are tried from left to right, so the first alternative found
for which the entire expression matches, is the one that is chosen. This
means that alternatives are not necessarily greedy. For example: when
mathing <CODE>foo|foot</CODE> against ``barefoot'', only the ``foo'' part will match, as that is the
first alternative tried, and it successfully matches the target string.
(This might not seem important, but it is important when you are capturing
matched text using parentheses.)
<P>
Also remember that ``|'' is interpreted as a literal within square
brackets, so if you write <CODE>[fee|fie|foe]</CODE> you're really only matching <CODE>[feio|]</CODE>.
<P>
Within a pattern, you may designate subpatterns for later reference by
enclosing them in parentheses, and you may refer back to the <EM>n</EM>th subpattern later in the pattern using the metacharacter \<EM>n</EM>. Subpatterns are numbered based on the left to right order of their opening parenthesis.
<FONT SIZE=-1>A</FONT> backreference matches whatever actually matched the subpattern in the string being examined, not the rules for that subpattern. Therefore,
<CODE>(0|0x)\d*\s\1\d*</CODE> will match ``0x1234 0x4321'', but not ``0x1234 01234'', because subpattern
1 actually matched ``0x'', even though the rule <CODE>0|0x</CODE> could potentially match the leading 0 in the second number.
<P>
<HR>
<H2><A NAME="WARNING_on_1_vs_1">WARNING on \1 vs $1</A></H2>
<P>
Some people get too used to writing things like:
<P>
<PRE> $pattern =~ s/(\W)/\\\1/g;
</PRE>
<P>
This is grandfathered for the
<FONT SIZE=-1>RHS</FONT> of a substitute to avoid shocking the
<STRONG>sed</STRONG> addicts, but it's a dirty habit to get into. That's because in PerlThink,
the righthand side of a [perlman:perlop] is a double-quoted string. <CODE>\1</CODE> in the usual double-quoted string means a control-A. The customary Unix
meaning of <CODE>\1</CODE> is kludged in for [perlman:perlop]. However, if you get into the habit of doing that, you get yourself into
trouble if you then add an <CODE>/e</CODE>
modifier.
<P>
<PRE> s/(\d+)/ \1 + 1 /eg; # causes warning under -w
</PRE>
<P>
Or if you try to do
<P>
<PRE> s/(\d+)/\1000/;
</PRE>
<P>
You can't disambiguate that by saying <CODE>\{1}000</CODE>, whereas you can fix it with
<CODE>${1}000</CODE>. Basically, the operation of interpolation should not be confused with the
operation of matching a backreference. Certainly they mean two different
things on the <EM>left</EM> side of the [perlman:perlop].
<P>
<HR>
<H2><A NAME="Repeated_patterns_matching_zero_">Repeated patterns matching zero-length substring</A></H2>
<P>
<FONT SIZE=-1>WARNING:</FONT> Difficult material (and prose) ahead. This
section needs a rewrite.
<P>
Regular expressions provide a terse and powerful programming language. As
with most other power tools, power comes together with the ability to wreak
havoc.
<P>
<FONT SIZE=-1>A</FONT> common abuse of this power stems from the ability to
make infinite loops using regular expressions, with something as innocous
as:
<P>
<PRE> 'foo' =~ m{ ( o? )* }x;
</PRE>
<P>
The <CODE>o?</CODE> can match at the beginning of <CODE>'foo'</CODE>, and since the position in the string is not moved by the match, <CODE>o?</CODE> would match again and again due to the <CODE>*</CODE> modifier. Another common way to create a similar cycle is with the looping
modifier <CODE>//g</CODE>:
<P>
<PRE> @matches = ( 'foo' =~ m{ o? }xg );
</PRE>
<P>
or
<P>
<PRE> print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
</PRE>
<P>
or the loop implied by
<CODE>split().</CODE>
<P>
However, long experience has shown that many programming tasks may be
significantly simplified by using repeated subexpressions which may match
zero-length substrings, with a simple example being:
<P>
<PRE> @chars = split //, $string; # // is not magic in split
($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
</PRE>
<P>
Thus Perl allows the <CODE>/()/</CODE> construct, which <EM>forcefully breaks
the infinite loop</EM>. The rules for this are different for lower-level loops given by the
greedy modifiers <CODE>*+{}</CODE>, and for higher-level ones like the <CODE>/g</CODE> modifier or
<CODE>split()</CODE> operator.
<P>
The lower-level loops are <EM>interrupted</EM> when it is detected that a repeated expression did match a zero-length
substring, thus
<P>
<PRE> m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
</PRE>
<P>
is made equivalent to
<P>
<PRE> m{ (?: NON_ZERO_LENGTH )*
|
(?: ZERO_LENGTH )?
}x;
</PRE>
<P>
The higher level-loops preserve an additional state between iterations:
whether the last match was zero-length. To break the loop, the following
match after a zero-length match is prohibited to have a length of zero.
This prohibition interacts with backtracking (see <A HREF="#Backtracking">Backtracking</A>), and so the <EM>second best</EM> match is chosen if the <EM>best</EM> match is of zero length.
<P>
Say,
<P>
<PRE> $_ = 'bar';
s/\w??/<$&>/g;
</PRE>
<P>
results in <CODE>"<</CODE><b><><a><><r><>``>. At each position of the string the best match given by
non-greedy <CODE>??</CODE> is the zero-length match, and the <EM>second
best</EM> match is what is matched by <CODE>\w</CODE>. Thus zero-length matches alternate with one-character-long matches.
<P>
Similarly, for repeated <CODE>m/()/g</CODE> the second-best match is the match at the position one notch further in the
string.
<P>
The additional state of being <EM>matched with zero-length</EM> is associated to the matched string, and is reset by each assignment to
<CODE>pos().</CODE>
<P>
<HR>
<H2><A NAME="Creating_custom_RE_engines">Creating custom RE engines</A></H2>
<P>
Overloaded constants (see <U>the overload manpage</U><!--../lib/overload.html-->) provide a simple way to extend the functionality of the
<FONT SIZE=-1>RE</FONT> engine.
<P>
Suppose that we want to enable a new
<FONT SIZE=-1>RE</FONT> escape-sequence <CODE>\Y|</CODE> which matches at boundary between white-space characters and non-whitespace
characters. Note that <CODE>(?=\S)(?<!\S)|(?!\S)(?<=\S)</CODE> matches exactly at these positions, so we want to have each <CODE>\Y|</CODE> in the place of the more complicated version. We can create a module <CODE>customre</CODE> to do this:
<P>
<PRE> package customre;
use overload;
</PRE>
<P>
<PRE> sub import {
shift;
die "No argument to customre::import allowed" if @_;
overload::constant 'qr' => \&convert;
}
</PRE>
<P>
<PRE> sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
</PRE>
<P>
<PRE> my %rules = ( '\\' => '\\',
'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
sub convert {
my $re = shift;
$re =~ s{
\\ ( \\ | Y . )
}
{ $rules{$1} or invalid($re,$1) }sgex;
return $re;
}
</PRE>
<P>
Now <CODE>use customre</CODE> enables the new escape in constant regular expressions, i.e., those without
any runtime variable interpolations. As documented in <U>the overload manpage</U><!--../lib/overload.html-->, this conversion will work only over literal parts of regular expressions.
For <CODE>\Y|$re\Y|</CODE> the variable part of this regular expression needs to be converted
explicitly (but only if the special meaning of <CODE>\Y|</CODE> should be enabled inside $re):
<P>
<PRE> use customre;
$re = <>;
chomp $re;
$re = customre::convert $re;
/\Y|$re\Y|/;
</PRE>
<P>
<HR>
<H2><A NAME="SEE_ALSO">SEE ALSO</A></H2>
<P>
[perlman:perlop].
<P>
[perlman:perlop].
<P>
[perlfunc:pos|pos].
<P>
[perlman:perllocale|the perllocale manpage].
<P>
<EM>Mastering Regular Expressions</EM> (see [perlman:perlbook|the perlbook manpage]) by Jeffrey Friedl.
<HR>
<BR>Return to the [Library]<BR>