为什么 perl 正则表达式挂起？

Question

I have the following code (which uses a recursive regex). Things seem to work fine, except at one particular place, where I have an unmatched brace. I'm trying to understand what exactly is causing the hang. I have:

use strict;
use warnings;
use Carp;
use Data::Dumper;

my $matchBracePtrn =    qr/(?<brace>\((?:[^()]+|(?&brace))+\))/;
my $mkVarPtrn =         qr/\$(?:\w|${matchBracePtrn})/;

# my $testString='$(a) $(a $(c)';   # OK: reports $(a) and $(c)...
# my $testString='$(foreach i,$(LIST),$(eval $(call foo,$i)))'; # OK
my $testString='$(a) $(foreach i,$(LIST),$(eval $(call foo,$i))'; #not OK!

while ($testString =~ /($mkVarPtrn)/g) {
    print "$1\n"
}

This outputs:

$(a)

and then hangs. This is perl 5.22.0, in case that matters.

Answer 1

The (?<brace>\\((?:[^()]+|(?&brace))+\\)) pattern causes catastrophic backtracking .

In order to fix the recursive pattern, either use a ++ possessive quantifier with [^()] , [^()]++ , or use an atomic group instead of the non-capturing group, (?>[^()]+|(?&brace))+ .

So, use either of

my $matchBracePtrn =    qr/(?<brace>\((?:[^()]++|(?&brace))+\))/;
my $matchBracePtrn =    qr/(?<brace>\((?>[^()]+|(?&brace))+\))/;

See the online demo .