Search for homophones

fp-phon-collisions.pl

#!/usr/bin/perl

use warnings;
use strict;
use utf8;

open(IN, "<", "$ARGV[0]");
binmode(IN, ":encoding(latin-1)");
#open(OUT, ">", "collisions.txt");
#binmode(OUT, ":utf8");
binmode(STDOUT, ":utf8");
binmode(STDERR, ":utf8");

my %p2g = ();

while(<IN>) {
	chomp;
	s/^\(//;
	my ($word, $pos, $pronraw) = split/\t/;
	next if ($pos eq 'pos');
	my $pron = $pronraw;
	$pron =~ s/[0-9]\)//g;
	$pron =~ s/[()]//g;
	$pron =~ s/^ //;
	$pron =~ s/\r//;
	$pron =~ s/ $//;
	my $value = $word . '__' . $pos;
	if(exists $p2g{$pron}) {
		if ($p2g{$pron} =~ /$value/) {
			#print STDERR "DUP: $p2g{$pron} $value\n";
		} else {
			$p2g{$pron} .= ", $value";
		}
	} else {
		$p2g{$pron} = $value;
	}
}

for my $pron (keys %p2g) {
	print $pron . "\t" . $p2g{$pron} . "\n";
}