in reply to Multiline Regex replacement in Multiline file

Howdy akamboj84,

In the replacement part of your regex, $dic{$matchkey} will always be undefined. $matchkey is the variable you defined above by joining all the keys in %dic, and is therefore not itself a key of %dic. What you actually meant was: 
$line =~ s%$matchkey%$dic{$&}%g;    # Use the matched string as a key 
+in $dic
$line =~ s%($matchkey)%$dic{$1}%g;  # Same as above without the perfor
+mance issues of $&

[download]
See the relevant section of perlvar.

Replies are listed 'Best First'.
Re^2: Multiline Regex replacement in Multiline file
by Athanasius (Archbishop) on Sep 15, 2014 at 09:37 UTC

    Hello Grimy,

    This is a nice idea, but unfortunately, as long as the regex contains metacharacters, it can’t be made to work, because $1 will never contain those metacharacters:

    #! perl
use strict;
use warnings;
use Data::Dump;

my $re  = 'aaa\s+bbb';
my %dic = ($re => 'new');
my $s   = 'aaa   bbb';

$s =~ s/($re)/$dic{$1}/;

print "\$1   = $1\n";
print "\$s   = $s\n";
print "\%dic = ";
dd     \%dic;

    [download]

    Output:

    19:34 >perl 1011_SoPW.pl
Use of uninitialized value within %dic in substitution iterator at 101
+1_SoPW.pl line 10.
$1   = aaa   bbb
$s   =
%dic = { "aaa\\s+bbb" => "new" }

19:34 >

    [download]

    Sorry, :-(

    Athanasius <°(((>< contra mundum Iustus alius egestas vitae, eros Piratica,

[reply]
[d/l]
[select]
      Hello Athanasius, Is there any workaround for this ? as my regex file is a long one and also i need to process a lot of files, so its kind of time/cpu consuming. I want to get rid of "for" loop as mentioned in reply to your previous post
[reply]

In Section Seekers of Perl Wisdom