#!/usr/bin/perl -s

our ($l1,$l2,$html,$debug,$v);
$v=1 if $debug;
die ("usage $0 -l1=en -l2=pt [file_which_listOfPathPaisr]\n") 
  unless $l1 && $l2;

while(<>){
  chomp;
  my @fs=split("\t",$_);
  if ($l1) {
    if($html){chomp ($res = `w3m -dump -T text/html '$fs[0]' |langident`);}
    else {    chomp ($res = `cat '$fs[0]' | langident`);}
    unless ($res eq $l1) {
        print STDERR "\t##unwanted languages $fs[0]($res not $l1)\n" if $v;
        next } }
  if ($l2) {
    if($html){chomp ($res = `w3m -dump -T text/html '$fs[1]' |langident`);}
    else {    chomp ($res = `cat '$fs[1]' | langident`);}
    unless ($res eq $l2) {
        print STDERR "\t##unwanted languages $fs[1]($res not $l2)\n" if $v;
        next } }
  print "$fs[0]\t$fs[1]\n";
}

__END__

=head1 NAME

langcheck - Perl script for language pairs checking

=head1 SYNOPSIS

langcheck -debug -l1=en -l2=pt [file]

=head1 DESCRIPTION

C<file> (or stdin) should have lines with 2 filenames separated with a TAB

Output is the list of filenamepairs with the proper languages.

=head1 AUTHOR

J.Joao Almeida, jj@di.uminho.pt

=head1 SEE ALSO

perl(1).

=cut