#!/usr/bin/perl -w

$fileName = $ARGV[0];
%ctr = ();
@FMT=();
@cmp = ();

open(INPUT, "$fileName") || die "File not found\n";
@data = <INPUT>;
close(INPUT);

foreach $data(@data) {
	chomp($data);

	## *** The line below needs to be changed according to the dataset **
	($PYword, $KYword) = split("\\|",$data);
	$ctr{$PYword}++;
}

$count=0;
foreach $ctrData(sort keys %ctr) {
   #print "# $ctrData $ctr{$ctrData} $count\n";
   push (@FMT,"# $ctrData $ctr{$ctrData} $count\n");	  
   push (@cmp,"$ctrData#$count");
   $count++;
}

$tmp = "";
@wds_all = ();

foreach $data(@data) {
	@wds = ();
	   
	chomp($data);
#	print "Data : $data\n";

	($PYword, $KYword) = split("\\|",$data);

#	$KYword =~s/\*//g;    
#        $KYword =~s/\+//g;
#        $KYword =~s/^\s+//;
#        $KYword =~s/\s$//;
#	$KYword =~s/\"//g;
#	$KYword =~s/\s/-/gi;
#	print "$KYword\n";

	(@wds) = split(" ",$KYword);
	foreach $wds(@wds) {
		chomp ($wds);
	        $wds =~s/^\s+//;
        	$wds =~s/\s$//;
		$wds =~s/\*//g;    
	        $wds =~s/\+//g;
		$wds =~s/\"//g;
		$wds =~s/\s/_/gi;
		$wds =~s/://g;
#		print ">$KYword\n";

		if($wds ne '') {
	  		#print "$wds#$PYword\n";
			push (@wds_all,"$wds#$PYword");
		}
	}
}

%ctrWDS = ();
#print "printing word and year occurence\n";
foreach $wds_all(@wds_all) {
#	print ">>>:$wds_all\n";
	$ctrWDS{$wds_all}++;
}

$tmpStr = "";
$bin =0;
foreach $ctrWDSData(sort keys %ctrWDS) {

	#split the word-data array
	($wrd,$yer) = split("#",$ctrWDSData);
	   
	foreach $dat(@cmp) {
		#split the year-data array
		($w1,$w2) = split("#",$dat);

		#compare the years; if same we have found the bin number!
		if($yer eq $w1) {
			$bin = $w2;
		}
	}
	$tmpStr = "$bin $wrd";
	#construct the formatted output again
	push (@FMT,"$tmpStr $ctrWDS{$ctrWDSData}\n");
	#$count++;
} 

#print "---- printint output ----\n";
foreach $fmt(@FMT) {
	print "$fmt";
}