#!/usr/bin/perl -w
#
# Create clusters of meeting data in a semi-unsupervised fashion
#
# parameters:
# 1) input directory
# 2) output data
####################################################################
use strict;

my $indir = shift @ARGV or die "No indir supplied";
my $outfile = shift @ARGV or die "No outfile supplied";

die "$indir not a directory!" unless -d $indir;
open(OUTPUT, ">$outfile") || die "Could not open $outfile";
die "Could not chdir $indir" unless chdir($indir);
opendir(INDIR, ".") || die "directory open of $indir failed";

# I prefer the chdir instead of using the slash for reading the filenames

my @files = grep(!/^..?\z/, readdir(INDIR)); #get all the files in the directory
# I'm not sure if readdir will always give you . and .. first
# See the readdir function in the docs. That's my prefered method from the camel
closedir(INDIR); #just being neat

foreach my $infile (@files) { #process files until none are left
  next if -d $infile;
  open(INPUT, $infile) || die "error on file opening $infile";

  my $within_spurt=0;
  my $previous_interupt = 0;
  my @current_spurt;
  <INPUT>; # skip first line which isn't data
  while (<INPUT>) { #inner loop to create initail clustering
    my (undef, undef,
        $current_word,
        $word_in_spurt,
        $spurt_length,
        undef, undef, undef, undef,
        $primary_speaker,
        undef,
        $interupting_speakers) = split(' ');

# 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;
  }
  close(INPUT);

  if( grep { /\bso\b/ } @current_spurt) {;
    print STDOUT @current_spurt;
    print OUTPUT @current_spurt;
  }
}