#!/usr/bin/perl

# wordcloud.pl - given a hash of words and numbers, output a rudimentary word cloud

# Eric Lease Morgan <emorgan@nd.edu>
# August 25, 2008 - first cut; based on Infomotions Gallery word cloud

# require
use strict;

# initialize
my %database = {};
$database{ 'scout' } = 368;
$database{ 'heyward' } = 339;
$database{ 'though' } = 324;
$database{ 'duncan' } = 301;
$database{ 'uncas' } = 244;
$database{ 'little' } = 239;
$database{ 'without' } = 238;
$database{ 'own' } = 237;
$database{ 'eyes' } = 235;
$database{ 'before' } = 217;
$database{ 'indian' } = 205;
$database{ 'hawkeye' } = 205;
$database{ 'young' } = 204;
$database{ 'magua' } = 189;
$database{ 'much' } = 185;
$database{ 'place' } = 184;
$database{ 'long' } = 178;
$database{ 'time' } = 170;
$database{ 'moment' } = 164;
$database{ 'hand' } = 161;
$database{ 'cora' } = 161;
$database{ 'again' } = 154;
$database{ 'after' } = 153;
$database{ 'head' } = 152;
$database{ 'returned' } = 150;
$database{ 'among' } = 149;
$database{ 'most' } = 145;
$database{ 'huron' } = 144;
$database{ 'air' } = 144;
$database{ 'toward' } = 143;
$database{ 'well' } = 140;
$database{ 'few' } = 138;
$database{ 'seen' } = 132;
$database{ 'many' } = 130;
$database{ 'found' } = 127;
$database{ 'manner' } = 126;
$database{ 'alice' } = 126;
$database{ 'voice' } = 125;
$database{ 'hurons' } = 125;
$database{ 'david' } = 125;
$database{ 'see' } = 124;
$database{ 'chief' } = 124;
$database{ 'words' } = 123;
$database{ 'about' } = 122;
$database{ 'woods' } = 121;
$database{ 'never' } = 121;
$database{ 'know' } = 121;
$database{ 'great' } = 120;
$database{ 'rifle' } = 119;
$database{ 'here' } = 118;
$database{ 'until' } = 114;
$database{ 'soon' } = 113;
$database{ 'left' } = 113;
$database{ 'just' } = 113;
$database{ 'white' } = 112;
$database{ 'heard' } = 111;
$database{ 'look' } = 110;
$database{ 'father' } = 110;
$database{ 'yet' } = 109;
$database{ 'side' } = 109;
$database{ 'savage' } = 109;
$database{ 'eye' } = 109;
$database{ 'first' } = 108;
$database{ 'already' } = 108;
$database{ 'whole' } = 107;
$database{ 'party' } = 105;
$database{ 'delawares' } = 104;
$database{ 'light' } = 102;
$database{ 'enemy' } = 102;
$database{ 'within' } = 101;
$database{ 'water' } = 101;
$database{ 'warrior' } = 101;
$database{ 'continued' } = 101;
$database{ 'seemed' } = 100;
$database{ 'low' } = 100;
$database{ 'appeared' } = 100;
$database{ 'turned' } = 99;
$database{ 'once' } = 98;
$database{ 'same' } = 96;
$database{ 'must' } = 95;
$database{ 'dark' } = 95;
$database{ 'short' } = 93;
$database{ 'passed' } = 93;
$database{ 'friend' } = 92;
$database{ 'back' } = 91;
$database{ 'instant' } = 90;
$database{ 'project' } = 88;
$database{ 'people' } = 87;
$database{ 'around' } = 87;
$database{ 'way' } = 86;
$database{ 'enemies' } = 86;
$database{ 'between' } = 86;
$database{ 'against' } = 86;
$database{ 'form' } = 85;
$database{ 'munro' } = 84;
$database{ 'warriors' } = 83;
$database{ 'stood' } = 83;
$database{ 'nor' } = 83;
$database{ 'feet' } = 83;
$database{ 'far' } = 83;

# do the work, and sent it to stdout
print &build_cloud( 'rank' );

# the hard part
sub build_cloud {

	# get the input
	my $sortby = shift;
	if ( ! $sortby ) { $sortby = 'rank' }

	# initialize
	my %queries;
	my $pointer    = 99;
	my $html;
	my %alpha_list;
	my %seen;
	

	# invert the hash
	foreach ( sort keys %database ) { push( @{ $queries{ $database{ $_ } } }, $_ ) }
		
	# loop through the sorted values
	foreach my $value ( sort { $b <=> $a } values %database ) {
	
		# get the key for each value
		foreach my $key ( @{ $queries{ $value } } ) {
		
			# check to see if it exists
			if ( ! $seen{ $key } ) {
			
				# increment
				$seen{ $key }++;
				
				# branch according to sorty by
				if ( $sortby eq 'rank' ) { 
				
					# calculate the font size
					my $font_size = $pointer + 75;
					
					# build the return value
					my $query =  uc( $key );
					$query    =~ s/ /\+/g;
					$html    .=  "<span style='font-size: $font_size%'><a href='http://infomotions.com/alex/concordance/?cmd=wordSearch&phrase=zzzz&word=zzzz&bookcode=etext940&segsize=1000&usePre=1&word1=$query'>$key</a></span> &nbsp;";
					
				}
				
				# cache the query
				else { $alpha_list{ $key } = $pointer; }
				
				$pointer--;
				
			}
		
			last if ( $pointer <= 1 )
			
		}
	
	}
	
	if ( $sortby eq 'alpha' ) {
	
		foreach my $key ( sort keys %alpha_list ) {
		
			my $query = uc( $key );
			$query    =~ s/ /%20/g;
			my $font_size = $alpha_list{ $key } + 75 ;
			$key      =~ s/ /&nbsp;/g;
			$html     .= "<span style='font-size: $font_size%'><a href='http://infomotions.com/alex/concordance/?cmd=wordSearch&phrase=zzzz&word=zzzz&bookcode=etext940&segsize=1000&usePre=1&word1=$query'>$key</a></span> &nbsp;";
		
		}
		
	}
	
	# done
	return $html;

}