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

use CWB;
#use Lingua::PT::PLN;
use Lingua::PT::PLNbase;
use Data::Dumper;
use strict;

our ($local,$html, $quebra, $lema, $text, $filename);

$quebra = "-text" if $text;
my $corpname=shift or die("$0 ... CorpusName\n");

chomp(my $regis = `cwb-config -r`);
$regis="$ENV{HOME}/registry" if $local;

my $corpdir="/corpora";

if ($lema) { open(O,"| addradical > $corpname.corpus") 
             or die("Error in addradical"); }
else {       open(O,"> $corpname.corpus") 
             or die("cant create $corpname.corpus\n");}

my %t=( s => {}, v => {});

select(O);
my $tags=undef;

for my $n (@ARGV){
  if   ($html)   { $n = "html2pml $n |" }
  elsif($quebra) { $n = "quebraxmlsent $quebra < $n|" }
  print STDERR "$n\n";
  $tags = cqptokens({enc=>":utf8",outenc=>":utf8"},$n);
  %{$t{s}} = ( %{$t{s}}, %{$tags->{s}} );
  %{$t{v}} = ( %{$t{v}}, %{$tags->{v}} );
#  while(<F>){
#     print   $_ ; # xmlsentences($_);
#  }
}

close(O);
select (STDOUT);

  print STDERR "$corpname.corpus\n";
my ($r,$o)= tags2options($corpname,%t);

open(R,">$regis/$corpname") or die;
print R $r;
close R;

  mkdir("$corpdir/$corpname");
  mkdir("$corpdir/c1");
  mkdir($regis);

unlink(<$corpdir/$corpname/*>);

print STDERR "$CWB::Encode -d $corpdir/$corpname -0 corpus -s $o < $corpname.corpus";

!system ("$CWB::Encode -d $corpdir/$corpname -0 corpus -s $o < $corpname.corpus")||
    die ("Erro Encode -d $corpdir/$corpname -s $o < $corpname.corpus ($?)$@$!\n");
!system ("$CWB::Makeall $corpname") ||
    die ("Erro Makeall ($?)$@$!\n");

## REVER !system ("$CWB::CompressRdx ??? $corpname") || die ("Erro Makeall ($?)$@$!\n");
## REVER !system ("$CWB::Huffcode ??? $corpname") || die ("Erro Makeall ($?)$@$!\n");
## REVER  apagar alguns .cr?? .corpus corpuz.rdx corpus.rev e outros que ele diz que se poderao remover


sub tags2options{
 my ($name,%t)=@_;
 my $options="";
 
 my @v = grep !/corpus/, keys(%{$t{v}});
 my @s = grep !/corpus/, keys(%{$t{s}});
 my $registry= qq{
NAME "$name" 
ID $name
HOME $corpdir/$name
ATTRIBUTE word
};


 for (@v){ $registry .= "STRUCTURE $_\n"; }
 for my $s (@s){ $registry .= "STRUCTURE $s\n" unless grep(/^$s$/,@v); }

 $registry .= "ATTRIBUTE alema\n" if $lema;
 $registry .= "ATTRIBUTE apos\n" if $lema;
 $registry .= "ATTRIBUTE tit\n" if $filename;

 $options.= " -P alema -P apos " if $lema;
 $options.= " -P tit " if $filename;
 for (@v){ $options.=" -V $_";}
 for my $s (@s){ $options.=" -S $s" unless grep(/^$s$/,@v);}

 ($registry,$options)
}

__END__

=head1 NAME

txt2cqp - Perl script to make CWB-cqp corpora from pure text or XML

=head1 SYNOPSIS

 txt2cqp [options] corpusName file*
  options 
  -lema     to make lematization and POS atributes ("alema" and "apos")
  -text     to process plain text files
  -quebra   to break text in sentences
  -quebra=p for XML texts, to break p elements in sentences
  -filename to make a "tit" atribute  ### Not working

=head1 DESCRIPTION

=head1 Options

C<-text> - to proccess text with empty lines as paragraphs

C<-html> - to proccess html input

C<-lema> - it uses jspell morpulogical analyser to eval a lemma and a pos
and builds C<alema> (ambiguos lema) and C<apos> (ambiguos pos) attributes.

C<-filename> - to enabel the creation of an attribute with the filename called
C<tit>

C<-quebra> - to break sentences and insert a sentence split tag C<s>

C<-quebra=tagname> - to break XML elements names C<tagname> in sentences and 
insert a tag C<s>

=head1 Install

In order to install this script you need:

 . CWB  (to build and query the corpus)
 . jspell (in order do use C<-lema> option)
 . Lingua::PT::PLN (to make cqp tokenize)

=head1 Bugs

option C<-filename> is not working

=head1 AUTHOR

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

=head1 SEE ALSO

cwb -- corpus workbench (sttutgard)

jspell

perl(1).

Lingua::PT::PLN

=cut