Regexp Question

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

I have the following patterns that I would like all to match:

${ThisOne}
${This${One}}
${This${One}Too}
${This${One}and${also}this}
...
[download]

Basically the start pattern is '${' and the end pattern is '}'. I have tried everything I know, including the folding stuff (though I didn't understand that one so well). Here's what I'm working with now (which also doesn't work):

m/
  \$\{(
    [^\{\$\}]*
    (?:
       \$\{
          [^\$\{\}]+
       \}
       [^\$\{\}]*
    )*
  )\}
/gxs;
[download]

I am matching in an entire file and the results from strings like '${somekindof${PKG}}' get the string 'PKG}' in the $1 global. I just want 'somekindof${PKG}' in the $1. Where am I going wrong? Thanks for the help monks. ArmadilloMan

Comment on Regexp Question Select or Download Code

Replies are listed 'Best First'.
Re: Regexp Question by Zaxo (Archbishop) on Jun 11, 2002 at 11:17 UTC
Those nested brackets make me think of Text::Balanced. Maybe that would be more convenient. After Compline, Zaxo	[reply]
Re:^2 Regexp Question by flounder99 (Friar) on Jun 11, 2002 at 16:49 UTC
Or you could try Damian's Regexp::Common use Regexp::Common; use strict; my @examples = ( '${ThisOne}', 'do not match me %{}', '${This${One}}', '${This${One}Too}', '${This${One}and${also}this}', 'me either $}', 'dont match this line noise {} $ }{' ); foreach (@examples){ print "$_\n" if /(\$$RE{balanced}{-parens=>'{}'})/; } my $longstring = join "", @examples; print "\n", join ("\n", $longstring =~ /(\$$RE{balanced}{-parens=>'{}'})/g); __OUTPUT__ ${ThisOne} ${This${One}} ${This${One}Too} ${This${One}and${also}this} ${ThisOne} ${This${One}} ${This${One}Too} ${This${One}and${also}this} [download] flounder	[reply] [d/l]
Re: Regexp Question by I0 (Priest) on Jun 11, 2002 at 16:18 UTC
`$_ = '${ThisOne} ${This${One}} ${This${One}Too} ${This${One}and${also}this}'; ($re=$_)=~s/((\$\{)\|(\})\|.)/${[')','']}[!$3]\Q$1\E${['(','']}[!$2]/g; print join"\n",eval{/$re/},"",""; warn $@ if $@ =~ /unmatched/; $re = join '\|',map{quotemeta}eval{/$re/}; print join"\n",/\$\{($re)\}/g;` [download]	[reply] [d/l]
Re: Regexp Question by talexb (Chancellor) on Jun 11, 2002 at 12:44 UTC
I wonder if Parse::RecDescent would be suitable for this problem instead of writing a monster regexp. At least it would more easily handle the recursion that is suggested by your examples. --t. alex "Nyahhh (munch, munch) What's up, Doc?" --Bugs Bunny	[reply]
Re: Regexp Question by Anonymous Monk on Jun 11, 2002 at 11:27 UTC
`$test="\$\{This\$\{One\}and\$\{also\}this\}"; $test=~/\$\{(([^\{\$\}](?:\$\{[^\$\{\}]+\}[^\$\{\}])*))\}/gxs; print "test:$1";` [download] Try this.I haven't made any changes in your regexp i just added two brackets to get the required values($1)	[reply] [d/l]