#!/local/usr/bin/perl
# C&C NLP tools
# Copyright (c) Universities of Edinburgh, Oxford and Sydney
# Copyright (c) James R. Curran
#
# This software is covered by a non-commercial use licence.
# See LICENCE.txt for the full text of the licence.
#
# If LICENCE.txt is not included in this distribution
# please email candc@it.usyd.edu.au to obtain a copy.

# takes a file of dependencies with scores and filters those below a
# threshold; also compares against a gold standard

$deps = shift;
$gold = shift;
$thresh = shift;

open(DEPS, "$deps") || die("can't open deps file");
open(GOLD, "$gold") || die("can't open gold file");

print STDOUT "\# this file was generated by the following command(s):\n";
print STDOUT "\# perl partial_deps.pl $deps $gold $thresh\n\n";

$global_corr = 0;
$global_p = 0;
$global_r = 0;

$scorr = 0;
$stotal = 0;

sub eval_deps {
  $correct = 0;
  $sflag = 1;

  while(($k, $v) = each(%deps)){
    $correct++ if(defined($gold{$k}));

    if(!defined($gold{$k})){
      $sflag = 0;
      #print "incorrect: $k\n";
    }
  }

  if(keys(%deps) != 0 && keys(%gold) != 0){
    $global_corr += $correct;
    $global_p += scalar(keys(%deps));
    $global_r += scalar(keys(%gold));
  }

  if(keys(%gold) != 0 && keys(%gold) != 0){
    $stotal++;
    $scorr++ if($sflag && keys(%deps) == keys(%gold));
  }

  $gs = scalar(keys(%gold));
  $ps = scalar(keys(%deps));

   #print "correct: $correct\n";
   #print "gold size: $gs\n";
   #print "deps size: $ps\n";
   #print "\n";
}

sub read_gold {
  %gold = ();

  while(<GOLD>){
    # read comments at the beginning of gold file
    if(/^\# /){
      while(($g = <GOLD>) !~ /^$/){
	;
      }
      next;
    }

    return if(/^$/);

    chomp;
    $gold{$_} = 1;
  }
}

while(<DEPS>){
  if(/^\# /){
    while(($d = <DEPS>) !~ /^$/){
      ;
    }
    next;
  }

  if(/^$/){
    print STDOUT "\n";

    read_gold();
    eval_deps();

    %deps = ();
    next;
  }

  /^(\S+) (.*)$/;
  $nderivs = $1;
  $dep = $2;

  if($nderivs >= $thresh){
    print STDOUT "$dep\n";
    $deps{$dep} = 1;
  }
}

$p = ($global_corr * 100) / $global_p;
$r = ($global_corr * 100) / $global_r;
$sacc = ($scorr * 100) / $stotal;

print STDERR "precision: $global_corr out of $global_p = $p\n";
print STDERR "recall: $global_corr out of $global_r = $r\n";
print STDERR "sentences correct: $scorr out of $stotal = $sacc\n";