Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

Hi,

there's a funny regex in the source of TWiki we're trying to use here (and failing - that's why I'm going through the source) where it parses an RCS file: 

/^([0-9]+)\.([0-9]+)\s+date\s+(\d\d(\d\d)?(\.\d\d){5}?);$/o

[download]
note the last part: 
(\.\d\d){5}?;

[download]
I've never seen quantifiers strung together like that... I've checked the Camel Book - it doesn't specifically cover this case, but its talk about "quantified atoms" seems to suggest that the construct is illegal. OTOH since it does compile, so I would be tempted to interpret it as 
((\.\d\d){5})?;

[download]
- but that's not how it matches... :-( It does match 
.01.02.03.04.05;

[download]
but not a lone 
;

[download]
I get a feeling there's something wrong with this regex - what do you think?

Replies are listed 'Best First'.
Re: Regex quantifiers composed?
by blakem (Monsignor) on Oct 11, 2002 at 10:14 UTC
    The question mark after a quantifier turns off greeding matching. In your regex it doesn't make much sense though. You're trying to match exactly 5 times, so greedy and non-greedy are the same thing. 
    /a{1,5}/;  # match 1-5 'a's prefer the longest.
/a{1,5}?/; # match 1-5 'a's prefer the shortest.

    [download]
    Here is an example: 
    #!/usr/bin/perl -wT
use strict;

$_ = 'aaaab';

# greedily match upto 5 'a's followed by a word char 
print "Greedy:    ";
print /(a{1,5})\w/, "\n";

# non-greedily match upto 5 'a's followed by a word char
print "NonGreedy: ";
print /(a{1,5}?)\w/, "\n";

__END__
Greedy:    aaaa
NonGreedy: a

    [download]

    -Blake

[reply]
[d/l]
[select]
Re: Regex quantifiers composed?
by PodMaster (Abbot) on Oct 11, 2002 at 11:36 UTC
    I know blakem explained it, but when in doubt,

    YAPE::Regex::Explain to the rescue

    use YAPE::Regex::Explain;
my $regex = '/^([0-9]+)\.([0-9]+)\s+date\s+(\d\d(\d\d)?(\.\d\d){5}?);$
+/o';
#print YAPE::Regex::Explain->new($regex)->explain;
$regex = '(\.\d\d){5}?;';
print '#' x 60,"\n";
print YAPE::Regex::Explain->new($regex)->explain;
print '#' x 60,"\n";
print YAPE::Regex::Explain->new(qr/a{1,5}/)->explain;
print '#' x 60,"\n";
print YAPE::Regex::Explain->new(qr/a{1,5}?/)->explain;
__END__

    [download]
    Read more... (4 kB)
[reply]
[d/l]
[select]
Re: Regex quantifiers composed?
by Anonymous Monk on Oct 11, 2002 at 10:59 UTC

    Thanks!

    Another question: what does the /o modifier for the original regex actually do? I've looked it up, and Mark-Jason Dominus says it's an optimization for regexes that interpolate variables - but there aren't any variables in the TWiki regex...

[reply]
      heh heh, yep the /o is pointless as well.

      -Blake

[reply]
[d/l]

Back to Seekers of Perl Wisdom