# This is the translation module.  It is used by saying 
# "require 'translation.pl';" near the start of your program. o
#
# Then you can &translate("any string");
#
# By David Mankin

package translation;
use LWP::UserAgent;
use Carp;

$DEFAULT_TRANSLATION_DIRECTION = "fr_en";
$DEFAULT_CGI_URL="http://babelfish.altavista.com/cgi-bin/translate?";
$DEFAULT_COOKIE_URL="http://babelfish.altavista.com/";
$DEFAULT_CGI_METHOD="POST";
$DEFAULT_PROXY="http://cacheflow.cs.colorado.edu:3128";
$AGENT_NAME="FIT/0.1";

# http://babelfish.altavista.com/cgi-bin/translate?
# doit=done&urltext=Eat+my+lunch&lp=fr_en

my $cookie = undef;

sub translate {
  &translate_lang($DEFAULT_TRANSLATION_DIRECTION, @_);
}

## 
# translate_lang(string [, string, string, ...]);
##
sub translate_lang {
  my ($lang) = shift;
  &setup_cookie unless (defined $cookie);

  my (@strings) = @_;
  my (@results) = ();

  my $ua = new LWP::UserAgent;
  $ua->agent($AGENT_NAME . " " . $ua->agent);
  $ua->proxy('http', $DEFAULT_PROXY) if $DEFAULT_PROXY;

  my $req = new HTTP::Request $DEFAULT_CGI_METHOD => $DEFAULT_CGI_URL;

  $req->content_type('application/x-www-form-urlencoded');

  foreach (@strings) {
    my $content = &prepare_content($_, $lang);
  
    $req->content($content);

    my $res = $ua->request($req);

    if ($res->is_success) {
      my $content = $res->content;
      $content =~ s/(\n|\r|\r\n)/ /g;
      my $translation = &parse_results($content);
      push (@results, $translation);
    } else {
      warn "Couldn't retrieve results from server";
    }
  }

  if (wantarray) {
    return @results;
  } else {
    if (@results > 1) {
      carp "Warning: Using &translation() for multiple arguments in scalar context";
    }
    return $results[0];
  }

}

##
# private setup_cookie().
#
# Requests a page in order to get the cookies it returns so we can post 
# post to it later.  
#
# For now, does nothign
##
sub setup_cookie {

}

##
# private parse_results($HTML)
#
# Parses the results
#
# For now, does nothing.
##
sub parse_results{
  my ($html) = @_;
#return $html;

  $html =~ s/\n//g;

  $html =~ s/^.*<!-- Begin Content -->//s;

  #if ($html =~ /$PRE_HINT/) {
  #  print "Matches PRE HINT\n";
  #} else {
  #  print "Does not match PRE HINT\n";
  #}
  #if ($html =~ /$POST_HINT/) {
  #  print "Matches POST HINT\n";
  #} else {
  #  print "Does not match POST HINT\n";
  #}

  
  $html =~ m/$PRE_HINT((.|\n)*)$POST_HINT/s;
   

  return $1;
  return $html;
}

$PRE_HINT = '<font face="arial, helvetica">.*?:</b>\s*</font>\s*</td>\s*</tr>\s*<tr>\s*<td>\s*<img border=0 src="/gifs/clear.gif" width="15" height="0">\s*</td>\s*<td align="left">\s*<br>\s*<font face="arial, helvetica">\s*';

$POST_HINT = '<br><br>\s*</font>\s*</td>\s*</tr>';

sub prepare_content {
  # doit=done&urltext=Eat+my+lunch&lp=fr_en
  my ($string, $lang) = @_;

  #print ("String before encoding: $string\n");
  $string =~ s/%/%25/g;
  $string =~ s/\+/%2b/g;
  $string =~ s/ /+/g;

  $string =~ s/([^%+ A-Za-z0-9.-])/ sprintf("%%%02X", ord($1))/eg;
  #print ("String after encoding: $string\n");

  return "doit=done&urltext=$string&lp=$lang";
}



1;