#!/usr/bin/perl -s
  use Lingua::PT::pln;

  use vars qw{$comm $img $tag $latin1};

  @breakby=qw(table tr td p br h1 h2 h3 h4 h5 h6 li ul ol dl dt dd 
    div blockquote hr address center);
  @removtag=qw( body html font a b i u tt small);
  @remov=qw(frameset head meta);

  if(not $tag) { $tag="p" }
  unless($img)   { push (@removtag, "img"); }

  $patremovtag = '</?(?:'. join('|', @removtag) .')\b[^>]*>';
  $patremov = '<('. join('|', @remov) .')\b[^>]*>(.|\n)*?</\1>';
  $patsep = '\s*(?:</?(?:'. join('|', @breakby) .')\b[^>]*>\s*)+';

open(A,">_Aout") or die;
open(B,">_Bout") or die;

while(<>){
  my ($a,$b) = m!(.*)\t(.*)!;
  print STDERR "($a)($b)\n";
  $id ++;
  print A "<f id='$id' name='$a'>\n", html2p($a),"</f>\n";
  print B "<f id='$id' name='$b'>\n", html2p($b),"</f>\n";
}
 
sub html2p{
  my $f = shift;
  my $r = "";
  if($latin1){open(F,"recode -f html..latin1 < $f|") or die("cant open $f\n")}
  else       {open(F,$f) or die("cant open $f\n"); }
  local $/;
  undef $/;
  $_= <F>;
  close F;
  s#<!--.*?-->##sg;
  s#$patremovtag##ig;
  s#$patremov##ig;
  for(split(/$patsep/i,$_)){
     $r .=  "<!-- $& -->\n" if $comm;
     s/\s*\n\s*/ /g;
     $r .=  Lingua::PT::pln::xmlsentences({st=>$tag},$_)."\n";
  }
  $r
}

__END__

=head1 NAME

htmlpairlist2p - file with html pairs list to a single pair of C<P>s

=head1 SYNOPSIS

  htmlpairlist2p [-comm] [-img] [-latin1] file

=head1 DESCRIPTION

C<html2p> makes a html page with "<p>" with the independent text segments after
dividing it in sentences.

It was designed to help in the process of aligning texts.

The command C<recode> should be installed in order to be possible to use 
C<-latin1> option.

=head1 AUTHOR

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

=head1 SEE ALSO

perl(1).

=cut