#!/usr/bin/perl -s

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

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

our ( $l1, $l2,$o,$local);  ## options
$local = ($local ? "-local" : "");

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

$o ||= "_$l1-$l2.tmx";

## 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("pdftotext -enc UTF-8 $f1 _$l1.txt");
 system("html2pml -isutf8 -latin1output -txt _$l1.txt > _$l1.pml");

 my $f2base= $f2;
 system("pdftotext -enc UTF-8 $f2 _$l2.txt");
 system("html2pml -isutf8 -latin1output -txt _$l2.txt > _$l2.pml");

 system("xmlalign2cqp $local _$l1.pml _$l2.pml");
 system("cqpalign2tmx $local _$l1.pml-_$l2.pml.align $o");

 unlink ("_$l1.txt");
 unlink ("_$l2.txt");
 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

pdfaligner - align html bitexts

=head1 SYNOPSIS

 pdfaligner -l1=en -l2=pt -o=en.pt.tmx f_en.pdf f_pt.pdf

=head1 DESCRIPTION


=head2 Pre-requesits

cwb-utils (http://natura.di.uminho.pt)

cwb -- Corpora WorkBench (Stutgard)

pdftotext -- Xpdf

=head1 AUTHOR

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

=head1 SEE ALSO

perl(1).

=cut