##
# FILE: mostFreqWord.pl
# AUTHOR: Daniel Jones
# CREATED: 12/18/2013
# MODIFIED: 21/18/2013
##

die "ERROR: Must enter one file name.\n" unless $#ARGV == 0;

open FILE, "<", $ARGV[0] or die "Could not open $ARGV[0] for reading.\n";

#the hash to contain the word-count pairs.
my %hash;
my @lines  = <FILE>;

#go through each line in the file
foreach my $line(@lines){
	#skip non-word characters
	my @words = split /\W+/, $line;
	
	#go through each word in the file
	foreach my $word(@words){
		chomp $word;
		#force all words to lowercase
		$word = lc $word;
		#remove beginning/trailing whitespace
		$word =~ s/^\s+|\s+$//g;
		my $key = $file.$word;
		
		#if the word exists, increment its value.
		#otherwise, set it to 1.
		if(exists $hash{$key}){
			$hash{$key}++;
		}
		else{
			$hash{$key} = 1;
		}
	}
}
close FILE;

my @values;
#get the values from the hash
foreach my $key(keys %hash){
	push @values, $hash{$key};
}

@values = sort @values;

my @keys = keys %hash;
my $idx = 0;

my $bestVal = @values[-1];
my $bestKey;

foreach my $key(@keys){
	if ($hash{$key} == $bestVal){
		$bestKey = $key;
		last;
	}
}

print "The most frequent word in $ARGV[0] is $bestKey, which was seen $bestVal times.\n";