#!/usr/bin/perl -w


## Usage: wsdOut.pl in_dir out_dir scoring_table_file
## Example: wsdOut.pl <path/topics_dir> <path_WSD/topics_dir> <scores_file>

#---------------------------------------------------------------------------------------#

##
##   INPUT ARGUMENTS:
##
##     - <in_dir> has to contain the directories which contains the files
##        to be desambiguate in "nam+id" format. 
##             OBLIGATORY TO FOLLOW THE ORIGINAL DIRECTORY STRUCTURE AND FILE FORMAT
##
##     - <out_dir>  folder for the out put of the desanbiguated files
##             THE OUTPUT WILL FOLLOW THE ORIGINAL DIRECTORY STRUCTURE
##
##     - <scoring_table_file> is a file where instances are desambiguated in Senseval Scorer Format:
##
##        word.pos  occurrenceID <synset_number-pos score>+
##       -------------------------------------------------
##
##       EXAMPLE:
##
##         village.n village.n.5-71-143.12 06107243-n 3.7281746031746
##         village.n village.n.5-71-279.21 06630722-n 1 06107243-n 2.7281746031746
##                     ...
##         leader.n leader.n.9-134-37.4 06950891-n 3.7281746031746
##         leader.n leader.n.9-135-78.4 06950891-n 3.7281746031746
##                     ...
##         lean.v lean.v.1-24-121.11 01106486-v 0.625 01392754-v 3.1031746031746
##         lean.v lean.v.15-91-235.20 01859256-v 0.714285714285714 00464659-v 0.555555555555556 01392754-v 2.45833333333333
##
##

#------------------------------------------------------------------------------------------------------------------------------#



use strict;


use vars qw ($in_dir $sc_file  @dirs $dir @files $file $in_file $out_file @main_dirs $out_dir);
use vars qw (%id2scores);



die "Usage: wsdOut.topics.pl <in_dir> <out_dir> <scoring_table_file>" if not defined $ARGV[2];


$in_dir = $ARGV[0];
$out_dir = $ARGV[1];
$sc_file = $ARGV[2];


print "\ngetting scoring table...\n";

%id2scores = &get_scoring_table($sc_file);

print " ... OK\n\n";



## Make directory for the out put path
if (not -e $out_dir){ mkdir ($out_dir,0770) || die $! }
if (not -e "$out_dir/cache_en"){mkdir ("$out_dir/cache_en",0770) || die $!;}


## Read all in put files. Only "nam+id" files
opendir(DIR,"$in_dir") || die "cannot open DIR(2): $in_dir\n" ;
@files = grep {/\.nam\+id$/} readdir(DIR);
close(DIR);


## print each file in the directory
print "Printing files in:\t$out_dir/cache_en ...\n";
foreach $file (@files){
  $file =~ s/\.nam\+id$//; ## remove extension (nam+id)
  
  ## Print out put file:
  &print_wsd_file("$in_dir", $file, "$out_dir/cache_en");
}

print " ... OK.\n";




#--------------------------------------------------------------------------------------#



#######
## IN: the file with all the disambiguated instances in Senseval Scorer Format.
## OUT: Normalized scores in a hash of hash structure:  $hash{instance_id} => $subhash{synset} => score
#
sub get_scoring_table{
  my ($table_f) = @_;

  my ($w, $ex_c,$c,$v,$id,$T,$i,$s,$lemma, $pos);
  my (@C,@score);
  my (%hscores, %hid);


  open (I, "$table_f") || die "cannot open FILE (scores file): $table_f\n";
  while (<I>){
    chomp;
    ($w,$ex_c,@score) = split;
    #senseval2/3 example code
    if (/(d\d+\.s\d+\.t\d+)/){
      $id = $1;
    }
    #clef example code
    else{
      ($lemma,$pos,$id,$s) = split (/\./,$ex_c);
    }
    ## NORMALIZE THE SCORES:
    # Compute the total sum of the scores
    $T = 0;
    for ($i=0; $i <= $#score-1; $i = $i+2){
      $T =  $T + $score[$i+1];
    }

    # Normalize the scores 
    for ($i=0; $i <= $#score-1; $i = $i+2){
      $c = $score[$i];
      $v = $score[$i+1] / $T;
      $hid{$c} = $v;
    }

    %{ $hscores{$id} } = %hid;
    undef %hid;
  }

  return %hscores;
}


#######
## IN: original "nam+id" file; out put path
## OUT: disambiguated "wsd" file
#
sub print_wsd_file{
  my ($target_dir, $file, $out_dir) = @_;
  my ($out_line, $chunk, $lemma, $pos, $sense, $id, $line,$new,$v);

  open(I,"$target_dir/$file.nam+id") || die "Cannot open file: $target_dir/$file.nam+id\n";
  open(O,">$out_dir/$file.wsd") || die "cannot open out_file: $out_dir/$file.wsd\n";

  ## print the first encoding line:
  ##  <?xml version='1.0' encoding="ISO-8859-1" ?>
  $line = <I>;
  if ($line =~ m/<\?xml.*\?>/){
    print O "$line";
  }else{
    die "ERROR: expecting xml encoding:\n\t<?xml version='1.0' encoding=\"ISO-8859-1\" ?>\n   File: $target_dir/$file.nam+id\n";
  }

  $out_line = "";
  while ($line = <I>){
    chomp $line;

    ### convert entity: &  => &amp ###
    $line =~ s/&/&amp\;/g if $line =~ m/&/ and not $line =~ m/&amp\;/;

    $out_line = "";
    while ($line =~ m/(^[^<]+|<WRD POS\=\"\d+\" ID\=\"[^\"]*\"><WF>[^<]*<\/WF><\/WRD>|<PHR>[^<]*<\/PHR>|<RP>[^<]*$|<[^>]+>)/g){
      $chunk = $1;

      if ($chunk =~ m/<WRD POS\=\"\d+\" ID\=\"([^\"]*)\"><WF>[^<]*<\/WF><\/WRD>/){
	$id = $1;
	if (not defined $id2scores{$id}){
	  $chunk =~ s/<\/WF><\/WRD>/<\/WF><SYNSET score\=\"1\">NO-SENSE<\/SYNSET><\/WRD>/;
	}
	else{
	  $new = "";
	  foreach $sense (keys %{ $id2scores{$id} }){
	    $v = $id2scores{$id}{$sense};
	    $new .= "<SYNSET score\=\"$v\">$sense<\/SYNSET>";
	  }
	  $chunk =~ s/<\/WF><\/WRD>/<\/WF>$new<\/WRD>/;
	}
      }
      $out_line .= $chunk;
    }
    print O "$out_line\n";
  }

  close(I);
  close(O);

}