#ParseCompare.pm
#===============================================================#
#  Jenny Draper  (learithe@cats.ucsc.edu) 6/65/02                       
#  Parses a .compare file, and returns a hash of probailities   
#
#  USEAGE: perl parse_compare.pl compareFileName
#  INPUT:
#   ....4 lines of some stuff
#	             E       H       L    total
#	stride-mixed.alpha----------------------
#	  A   72963     572   40776  114311
#	  B   50309    5501  100326  156136
#	  C   15273    1783   57674   74730
#
#  OUPUT:
#	  $alphatype = "alpha"
#	  %output_hash = 
#		"A" -> [0.638285029437237, 0.00500389288869838, 0.356711077674065]
#		"B" -> [0.32221268637598, 0.035232105344059, 0.642555208279961]
#		"C" -> [0.204375752709755, 0.0238592265489094, 0.771765020741335]
#
#===============================================================#

package ParseCompare;
require Exporter;
@ISA = qw(Exporter);
@EXPORT = qw( ParseCompareFile );

sub ParseCompareFile( $ )
{	my $compFile = $_[0];
	%output_hash = ();
	my $alphaType = "";
	my $mutualInfo = "";
	my $goToEnd = 0;
	my @temp_ar = (); @temp2_ar = (); @splitCurrLine = (); 
	my $key = "";

	#open the .compare file
	open (COMPARE_FILE, $compFile) or die "ERROR: couldn't open $compFile\n\n";

	#get alphabet type
	@temp_ar = split(/\./, $compFile);
	@temp2_ar = split(/-/, $temp_ar[1]);
	$alphaType = $temp2_ar[0];

	#get hash of probabilities
	while(<COMPARE_FILE>)
	{	$currLine = $_;	
		if(($. < 7))	#skip the first 6 lines ($. counts input line#s)
		{	next;
		}	
		@splitCurrLine = split(/\s+/, $currLine); 	

		if( !(defined($splitCurrLine[1])) ) #skip blank lines
		{	next;
		}else
		{	if($splitCurrLine[1] eq "total")
			{	$goToEnd = 1;
				next;
			}elsif( ($goToEnd) and ($splitCurrLine[0] eq "Mutual"))
			{	$mutualInfo = $splitCurrLine[3];
				next;
			}elsif($goToEnd)	#skip all lines but the last line
			{	next;	
			}	
		}
		$output_hash{$splitCurrLine[1]} = [
			($splitCurrLine[2]/$splitCurrLine[5]),	#E-probability
			($splitCurrLine[3]/$splitCurrLine[5]),	#H-probability
        	($splitCurrLine[4]/$splitCurrLine[5])];	#L-probability
			
	}
	
    #code for debugging
	#foreach $key (keys %output_hash )
	#{   print $key, " ", @{$output_hash{$key}}, "\n";
	#}
	#print "\n";
	
close (COMPARE_FILE);
return($alphaType, $mutualInfo, \%output_hash);
}