#! usr/bin/perl

use strict;

# Check to make sure that one file was given on the command line, if not
# print a correct usage statement.
#
if (@ARGV != 1) {
    die "\nUsage: perl nuc_count.pl <fasta file>\n\n";
}

# Get the file name and open an instream to it.
#
my $file = shift;
open IN, $file or die "Couldn't open file: $file\n";

my $firstline = ""; #create an empty string

#First print sequence name
#When a line begins with "Query =" print the line
my $Queryname="";
while (my $line = <IN>) {
    if ($line =~ /^Query= /){
	my $Queryname=$line;
	print $line;
    }    
    
#The Plan
#Identify the region of the file with the list of hits
#Use an if loop to stay in that region
#Use a while loop to read through each line until it hits the end of the list
#In the while loop Pull out the Worm ID, Gene Annotation and E-value
#Concatenate those results all together
#At the end see if it is either the first line (b/c always print that)
#or if the E-value is less than 1e-10
#if so we will print all the values we collected
#otherwise we exit the while loop and move to the next Query 


#Pull out the Worm ID   
#Find the line that starts with "Sequences producing..."
    if ($line =~ /^Sequences producing High-scoring Segment Pairs/){
	$line = <IN>;
	$line = <IN>;  	# We skip to the third line
	$firstline = "true"; #define as true for first time through the loop
	while ($line ne "\n"){	
	    my @highhitarray = split /\s+/,$line; #turn it into an array split at spaces	
	    my $linetoprint = "";	   
	    $linetoprint .= "WormID = ".$highhitarray[0]; #print element 0, the WormID
	    $linetoprint .= "\n";
	    
	    	    
#Pull out the gene annotation
#In highhitarray, element 3 starts the Gene Annotation
#Concatenate each consecutive element starting with 3 until the string is all digits
	    my $count=3;
	    my $GeneAnnotation;
	    while (!($highhitarray[$count] =~ /^\d+$/)){
		$GeneAnnotation.=$highhitarray[$count]." ";
		$count++;	   
	    } #Print the concatenated string/Gene Annotation
	    $linetoprint .= "Gene Annotation = ".$GeneAnnotation;  
	    $linetoprint .= "\n";
	    
#Pull out the E-value
#Reverse the array so now the E-value is in Element 1 and print it
	    my @rev_array = reverse(@highhitarray);
	    $linetoprint .= "E-value = ".$rev_array[1];
	    $linetoprint .= "\n";
	    $linetoprint .= "\n";
	    if ($firstline eq "true" || $rev_array[1] <= 1e-10){ 
#if this was the first line or E-value <= 1e-10 we will print
		print $linetoprint;  #This has stored all the values we want
	      	$firstline = "false"; #Set firstline to false before going through loop again
	    }
	    $line = <IN>;   #Read the next line of while loop
	}
	print "---------------"; 
	print "\n";
	print "\n";
    }
}

close IN; 
exit;