#!/usr/bin/perl -s

use strict;
use warnings;

use File::Copy;

my $prefix = '__DIR__'; # This gets substituted by autoconf/automake

our (
     $p, # html like paragraphs
     $l, # each line is paragraph
     $s, # <s> sentence separator
     $n, # Natools like file
    );

my $f = shift;

my $t = "/tmp/_${$}_"; # temporary file

$/='';

if ($f) {
  if (-f $f) {
    open Fi, "sed -e 's/  *\$//' $f|" or die "Can't open pipe with sed!\n"
  }
  else {
    die "Can't find $f!\n"
  }
}
else {
  open Fi, "<&=STDIN "
}

# Maybe we must check if there are more than one of these options
$/ = '<p>' if $p;
$/ = "\n"  if $l;
$/ = '<s>' if $s;
$/ = "\n\$\n" if $n;

open Fo, ">$t.unt" or die "Error can't write to tmp file ($t.unt): $!\n";

while (<Fi>) {
  chomp;
  s/\s*([,.;!?()«»])\s*/ $1 /g;
  s/(\s|\n)+/ /g;
  s/\. \. \./.../g;
  s/ $/\n\n/;
  print Fo "$_\n\n"
}
close Fi;
close Fo;

system ("$prefix/bin/ems-tagger $prefix/share/ems/lexico $t.unt $prefix/share/ems/bigramas $prefix/share/ems/regras-lexicais $prefix/share/ems/regras-contextuais -i $t.int >  $t.out");
#if ($f) {
  #  move("$t.out", "$f.out")
  $\ = $/;
  $/ = "\n";
  open In, "$t.out" or die "Cant open $t.out for reading";
if ($f) {
  open Out, ">$f.out" or die "Cant open $f.out for writing";
  select Out;
}
  while (<In>) {
	chomp;
        print;
  }
if ($f) {
  close Out;
}
  close In;
  unlink "$t.out";
#}
#else {
  #open CAT, "$t.out" or die "Ooops. Someone deleted $t.out!\n";
  #while (<CAT>) { print }
  #close CAT
#}

unlink "$t.int" if -f "$t.int";
unlink "$t.unt" if -f "$t.unt";
unlink $t if -f $t;

1;
__END__

=head1 NAME

ems-tag - Top level interface to EMS

=head1 SYNOPSIS

 ems-tag [-p|-l|-s] [inputfile]

=head1 DESCRIPTION

This script is the top level interface to EMS. If you use a filename
in the command line, it is used as the input text. If not, the
standard input will be read.

The three switches can only be used independently:

=over 4

=item B<-p>

Input file uses HTML like paragraphs;

=item B<-l>

Input file uses each line as a different paragraph;

=item B<-s>

Input file uses <s> sentence separator;

=back

=head1 SEE ALSO

ems

=head1 COPYRIGHT

 Copyright (C)2002-2003 Jose Joao Almeida
 GNU GENERAL PUBLIC LICENSE (LGPL) Version 2 (June 1991)

=cut