#!/usr/bin/perl -w -s

use Lingua::StarDict::Gen;
use Biblio::Thesaurus;
use utf8;
use strict;

our ($name,$dir,$i,$langs);

my @extralang = split(/\W+/,$langs);
my %extralang = map{($_=>$_)} @extralang;

$dir ||= "D/";

my $t=shift or die("usage thesarus2stardict [-name=NNN] thesaurus_file\n");

$name ||= "Thesaurus_$t";
$name =~ s/\.the$//; 

my $s="/";
my $obj = thesaurusLoad($t);

#Lingua::StarDict::Gen::writeDict($obj->{$obj->{baselang}}, $name);
#mkdir($dir);
#Lingua::StarDict::Gen::writeDict(terms2xdxf($obj), $name, $dir);

if($i){ Lingua::StarDict::Gen::writeDict(terms2xdxf($obj),$name); }
else  { mkdir $dir;
        if ( $^O eq "MSWin32") {$s="\\"} 
        Lingua::StarDict::Gen::writeDict(terms2xdxf($obj),$name,"$dir$s"); }

sub terms2xdxf{
   my $t=shift;
   my $o={};
   my @l = sort $t->languages();
   my @okey = ();
  
   $t->downtr(
      {-default  => sub { 
          if(   $t->isLanguage($rel) and $extralang{$rel} ) 
               { "\n$rel: \t".join(",\n\t",map{push(@okey,$_);"<kref>$_</kref>"} @terms)}
          elsif(   $t->isLanguage($rel)) 
               { "\n$rel: \t".join(",\n\t",map{"<dtrn>$_</dtrn>"} @terms)}
          elsif($t->isExternal($rel))
               { "\n$rel: \t".join(",\n\t",map{"<abr>$_</abr>"} @terms)}
          else { "\n$rel→\t".join(",\n\t",map{"<kref>$_</kref>"} @terms)}
           },
       -eachTerm => sub { $o->{$term}="<ar>$_</ar>" ;
                          for my $k1(@okey){$o->{$k1}="<ar>$_</ar>";
                          @okey=(); }
           },
#      -end      => sub { "Thesaurus :\n $_ \nFIM\n"},
       -order    => ["DEF", "DOM", @l,"IOF","BT","NT","RT"],
   });
   $o;
}


__END__

=encoding utf8

=head1 NAME

thesaurus2stardict - generates a StarDict dictionary from a thesaurus

=head1 SYNOPSIS

 thesaurus2stardict [...] file.the

=head1 DESCRIPTION

By default, output is sent to directory D

 -dir=outputdir   (def: D)
 -name=NAME       (def: thesaurus_<file>)
 -i             build and install  (def: no install)
 -l=EN          use EN as the def. language
 -langs=RU,FR   also use the RU and FR languages

With C<-l=PT>, language will be "PT"

By default, language is the first language of the first concept.

With C<-i> (install) -- output directory is "/usr/share/stardict/dic/",
"c:\ProgramFiles\stardict\dic" or similar.

=head2 thesaurus format

Concepts separated by empty lines

Each line:

 relation  termlist (separeated by "," or ";")

Example:

 %encoding uft8
 %baselang PT
 %languages EN
 %languages FR 
 %desc[EN] NT Narrower Term
 %desc NT Termo Específico
 %desc BT Termo Genérico
 %desc INST Instâncias
 %desc IOF instância de 
 %inverse IOF INST

 _top_
 NT animal

 animal
 NT mamífero, animal doméstico, réptil
 EN animal

 mamífero
 INST gato, leão

 réptil
 INST crocodilo , cobra

 leão
 EN lion

 gato
 EN cat
 FR chat
 IOF animal doméstico
 SN Animal que tem quatro vidas e m(e)ia

 sapo
 IOF réptil
 EN frog

=head1 AUTHOR

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

=head1 SEE ALSO

perl(1).

Lingua::StarDict::Gen

term2stardict

Bibio::Thesaurus(3pm)

=cut