#!/g/rcs/sw/bin/perl -w
#
# Create clusters of meeting data in a semi-unsupervised fashion
#
# parameters:
# 1) input directory
# 2) output data
####################################################################

use strict;

my $indir = $ARGV[0];
my $outfile = $ARGV[1];

opendir(INDIR, $indir) || die "directory open failed";
open(OUTPUT, ">$outfile");

my @files = readdir(INDIR); #get all the files in the directory
shift @files; shift @files; #shift off '.' and '..'

foreach my $file (@files) { #process files until none are left
    
    my $slash = '/';
    my $infile = $indir.$slash.$file;
    open(INPUT, $infile) || die "error on file open";

    my $within_spurt=0;
    my $previous_interupt = 0;
    my $first = 1;
    LINE: while (<INPUT>) { #inner loop to create initail clustering
	if( $first ) {
	    $first = 0; # skip first line which isn't data
	    next LINE;
	}
	my @line = split(' ');
	my $current_word = $line[2];
	my $word_in_spurt = $line[3];
	my $spurt_length =  $line[4];
	my $primary_speaker = $line[9];
	my $interupting_speakers = $line[11];
	my @current_spurt;
	#@debug = ($line[0], $line[1], $line[3], $line[4], $line[5]);

	# see if number of speakers increases and first_speaker == 0, 
	# this means the spurt is interupting, not the primary speaker
	if($primary_speaker == 0 && 
	       $interupting_speakers > $previous_interupt && 
               $word_in_spurt == 1) {
	    push @current_spurt, "<s> $current_word ";# start marker
	    $within_spurt = 1;
	    if($word_in_spurt == $spurt_length) {#only word in spurt
		push @current_spurt, "<\/s>\n";
		$within_spurt = 0;
		$interupting_speakers = 0; # the interupt is over
	    }
	}
	# if we are in a spurt and it is not the last word of that spurt
	elsif($within_spurt && $word_in_spurt != $spurt_length) {
	    push @current_spurt, "$current_word "; #add current word
	}
	# if this is the last word of a spurt
	elsif($word_in_spurt == $spurt_length && $within_spurt == 1) {
	    push @current_spurt, "$current_word <\/s>\n"; #end marker
	    $within_spurt = 0;
	    $interupting_speakers = 0; # the interupt is over
	}
        #make sure that at the end of spurts this flag is reduced
	if($word_in_spurt == $spurt_length) {
	    $interupting_speakers--;
	    # should not be less than 0
	    if($interupting_speakers < 0) { $interupting_speakers = 0; }
	}
	$previous_interupt = $interupting_speakers;

	my $yeah = 0;
	my $string = join('',@current_spurt);
	if( $string =~ /\bso\b/ ) {
	    $yeah = 1;
	}
	print STDOUT $string if $yeah;
	print $string;
    	print OUTPUT @current_spurt;
	undef @current_spurt;
	undef $string;
    }
    close(INPUT);
    last;
}