#!/usr/bin/perl

use strict;
use warnings;

use POSIX qw(locale_h);
my $old_locale = setlocale(LC_CTYPE);
setlocale(LC_CTYPE, 'ca_ES.iso885915@euro');
use locale;

my @expressions;
my @contents;
my $content = qq{
    Abbott test1 test2 salud
};

push (@contents, $content);
$content = qq{
    salud test1 test2 Abbott
};
push (@contents, $content);

$content = qq{
    Abbott test1 test2 test2 test2 test2 test2 test2 test2 test2 salud
};
push (@contents, $content);

$content = qq{
    salud test1 test2 test2 test2 test2 test2 test2 test2 test2 Abbott
};
push (@contents, $content);

$content = qq{
    salud test1 test2 test2 test2 test2 test2 test2 test2 test2 test2 
+Abbott
};
push (@contents, $content);

$content = qq{
    Abbott test1 test2 test2 test2 test2 test2 test2 test2 test2 test2
+ salud
};
push (@contents, $content);

$content = qq{
    salud test1 test2 test2 test2 test2 test2 test2 test2 test2 test2 
+test2 test2 test2 test2 test2 test2 test2 test2 test2 test2 Abbott
};
push (@contents, $content);

$content = qq{
    salud test1 test2 test2 test2 test2 test2 test2 test2 test2 test2 
+test2 test2 test2 test2 test2 test2 test2 test2 test2 Abbott
};
push (@contents, $content);

$content = qq{
    salud test1 test2 test2 test2 test2 test2 test2 test2 test2 test2 
+test2 test2 test2 test2 test2 test2 test2 test2 test2 test2 Abbott
};
push (@contents, $content);

$content = qq{
    salud Abbott test1 test2 salud
};
push (@contents, $content);

my $par1 = '[a\xe0\xe1\xe4\xe2A\xc1\xc0\xc4\xc2]bb[o\xf2\xf3\xf6\xf4O\
+xd3\xd2\xd6\xd4]tt';
my $par2 = 's[a\xe0\xe1\xe4\xe2A\xc1\xc0\xc4\xc2]l[u\xf9\xfa\xfc\xfbU\
+xda\xd9\xdc\xdb]d';
my $expression = "$par1 $par2\:\:20";
push (@expressions, $expression);

warn "PART 1";

foreach my $cont (@contents){
    warn "CONTENT $cont";
    foreach my $exp (@expressions) {
        my $tag = 'span';
        my $class = "lighligth";
        next if ($exp !~ /::/);
        my ($exp, $distance) = split("::", $exp);
        my ($par1, $par2) = split(' ', $exp);
        # warn "Pars $par1 - $par2 - $distance";
        if ($cont =~ /$par1.*$par2/i) {
          if ($cont =~ /\b($par1)(\W+(?:\w+\W+){0,$distance})($par2)\b
+/i) {            
            # if ($cont =~ m/\b($par1)(\W+(?:\w*\W*){1,$distance})?($p
+ar2)\b/i){
             my ($par1, $par2, $par3, $par4) = ($1, $2, $3, $4);
              warn "FIND 1 Par1: $par1 Par2: $par2 Part3: $par3";
             $cont =~ s/$par1\Q$par2\E$par3/<$tag$class> $par1<\/$tag>
+$par2<$tag$class> $par3<\/$tag>/gi;
            }
        }
        warn "STEP";
        if ($cont =~ /$par2.*$par1/i) {
          if ($cont =~ /\b($par2)(\W+(?:\w+\W+){0,$distance})($par1)\b
+/i) {            
              my ($par1, $par2, $par3, $par4) = ($1, $2, $3, $4);
                 warn "FIND 2 Par1: $par1 Par2: $par2 Part3: $par3";
                $cont =~ s/$par1\Q$par2\E$par3/<$tag$class> $par1<\/$t
+ag>$par2<$tag$class> $par3<\/$tag>/gi;
            }
        }
    }
    warn "END";
    warn "\n\n";
}
[download]