#!/usr/local/bin/perl5

# MI(word,speech_act) = ln_{2} ( P(word & speech_act) / P(speech_act)*P(word) )
# MI(word,speech_act) = ln_{2} ( P(speech act|word)/P(word)) 

$binDir = "/afs/cs.cmu.edu/project/cmt-8/enthusiast/mlb/CMU-Cam_Toolkit_v2/bin";
$homeDir = "/afs/cs.cmu.edu/project/cmt-8/enthusiast/mlb/MI";
$WORD  = 1;
$ACT   = 2;
$UNION = 3;

sub eat_parse_comments {
    for(my $i=0; $i<3; $i++) {
	$_ = <SOUP>;
    }
}

sub getSpeechAct {
    my $str = $_[0];
    $str =~ s/\[(.*)\]:.*/\1/;
    $str =~ s/(.*?)\+.*/\1/;
    $str;
}

sub loadActInventory {
    %speech_acts;
    my $inventory = "/afs/cs.cmu.edu/project/cmt-8/enthusiast/mlb/acts.txt";
    open(INV, "<$inventory") or die "Unable to open $inventory for input\n"; 
    while(<INV>) {
	chomp;
	$speech_acts{$_} = 1;
    }
    close(INV);
}

sub loadMI {
    %MI;
    my $data = "$homeDir/mi.data";
    open(MI, "<$data") or die "Unable to open $data for input\n";
    my $word;
    while(<MI>) {
	if(/.W/) {
	    $word = <MI>;
	    chomp($word);
	}
	else {
	    split(/ /, $_);
	    my $act = $_[0];
	    my $mi  = $_[1];
	    chomp($mi);
	    $MI{$word}{$act} = $mi;
	}
    }
    close(MI);
}

sub getPrediction {
    my $utt = $_[0];
    $max = 0;
    my $maxAct;
    my @words = split(/\s+/, $utt);
    for $act (keys %speech_acts) {
	my $totalMI = 0;
	for $word (@words) {
	    my $mi = $MI{$word}{$act};
	    $totalMI += $mi;
#	    print "MI($word,$act) = $mi\n";
	}
#	print "MI wrt $act = $totalMI\n";
	if($totalMI > $max) {
	    $max = $totalMI;
	    $maxAct = $act;
	}
    }
#    print "\n\tARGMAX = $maxAct, maxMI = $max\n\n";
    $maxAct;
}


sub Main {

    @ARGV == 2 or die "Usage:  MI.pl <infile> <outfile>";
    my $infile  = $ARGV[0];
    my $outfile = $ARGV[1];
    open(OUT, ">$outfile") or die "Unable to open file: $outfile for output\n";

    &loadActInventory;
    &loadMI;
    
    my $soupCall = "/afs/cs/project/cmt-46/trans/Soup/soup_s -grammar /afs/cs.cmu.edu/project/cmt-46/trans/Phoenix/Grammars/Travel/stable/all.1.gra";
    $position = 0;
    @sequence = ();
    my $line;
    open(SOUP, "$soupCall < $infile | ") or die "Could not run SOUP.";
    while($line = <SOUP>) {
	chomp($line);
	# at beginning of parse
	if($line =~ /\A\; [0-9]+:>/) {
	    $line =~ /\`{2}(.*)\'{2}/;
	    my $utt = $1;
	    if($utt) {
		print OUT "\"$utt\"\n";
		&eat_parse_comments;
		$done = 0;
		my @generated = ();
		while(!$done) {
		    my $interp = <SOUP>;
		    if($interp =~ /\A\n\Z/) {  $done = 1; }
		    else {
			chomp($interp);
			$interp =~ s/\A\s+//;
			push(@generated, $interp);
			print OUT "$interp\n";
		    }
		}
		my $Sn;
		
		# no ambiguity
		if(@generated == 1) {
		    $Sn = &getSpeechAct($generated[0]); 
		    print OUT "$Sn\n\n";
		}
		
		# get prediction from ngram component using previous two @seq
		else { 
		    # MI prediction
		    $Sn = &getPrediction($utt);
		    print OUT "$Sn\n$max\n\n";
		}
		push(@sequence, $Sn);

		$position++;
	    }
	}
    }
    close(SOUP);
}

Main;