#!/usr/bin/perl -s

#undef $/;
#$/='';

my $f1=shift;
my $f2=shift;

our ( $l1, $l2,$o,$local);  ## options

$l1 ||= $f1;
$l2 ||= $f2;

$o ||= "_$l1-$l2.tmx";
$local = ($local ? "-local" : "");

## html2pml -latin1output -listofpairs -noimg  p$G.pairs  p$N.$O p$N.$T
## LANG=pt_PT xmlalign2cqp p$N.$O p$N.$T

 my $f1base= $f1;
 system("html2pml -latin1output -noimg '$f1' > _$l1.pml");

 my $f2base= $f2;
 system("html2pml -latin1output -noimg '$f2' > _$l2.pml");

 system("xmlalign2cqp $local _$l1.pml _$l2.pml");
 my $v=`cqpalign2tmx  $local _$l1.pml-_$l2.pml.align $o`;
 print $v;
 if ($v =~ /R=(\d+)/ && $1 > 80 ) {
    warn("problems: bad alignment($1)\n");
    rename($o,"$o.BADALIGNMENT") 
 }

 unlink ("_$l1.pml");
 unlink ("_$l2.pml");
 unlink ("_$l1.pml-_$l2.pml.align");
 unlink ("_$l2.pml-_$l1.pml.align");
 unlink (<c-*pml.corpus>);

__END__

=head1 NAME

htmlaligner - align html bitexts

=head1 SYNOPSIS

 htmlaligner -l1=en -l2=pt -o=en.pt.tmx f_en.html f_pt.html

=head1 DESCRIPTION


=head2 Pre-requesits

cwb-utils

cwb -- corpora workbench (Stutgard)

=head1 AUTHOR

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

=head1 SEE ALSO

perl(1).

=cut