#!/usr/bin/perl

use strict;
use ProbeSearch;
use constant SUCCESS => 0;
use constant ERROR => -1;

sub getParams {
    my %opts;
    my $errstr = "";
    use Getopt::Long;
    use Data::Dumper;

    GetOptions(\%opts, 'help');
    $errstr = " " if (defined($opts{help}));
    
   if ($errstr) {
   print <<"END_USAGE";
Description:

This file takes an html file and creates a probe_summary.txt file

Usage:
create_probe_summary.pl 

requires probes.html to run


END_USAGE
        print "\n$errstr" if ($errstr);
        die("\n");
    }

}


sub debug_print
{
   my $logfile = "./debug.log";
   my $msg = shift;
   open(LOGFILE, ">>".$logfile);  
   print LOGFILE $msg;
   close (LOGFILE);
}

sub reverse_compliment
{
   my $dna = shift;
   $dna = reverse($dna);
   $dna =~ tr/ATGCatgc/TACGtacg/;
   return $dna;
}


# Hard coding input file
my $location = "/home/bcarlson/data";
my @species = ("mouse","baboon","cat","chicken","chimp","cow","dog","fugu","pig","rat","tetra","zfish");
my $species_dna;
my $file = "probes.html";
my $line;
my $found_overgo = 0;
my $probe_name = "";
my $probe_results = "";
my $probe_sequence = "";
my $probe_idx;
my $reverse_compliment;
my $target;
my $target_start_idx;
my $target_file;
my $process_target;
my @data;
my $last_target;
my $last_line;
my $current_line;
my $input_file = "$location/$file";
my $temp;
my $alignment = new ProbeSearch;

open(INFILE, $input_file);

while (<INFILE>)
{
	if ($_ =~ /Overgo Characteristics/)
	{
		last;
	}

}

for (my $i=0; $i<26; $i++)
{ $_ = <INFILE>; }

PROBE_NAME: while (<INFILE>)
{
      if ($_ =~ /<tr valign=top>/)
      {
	      for (my $i=0; $i < 1; $i++)
	      { $_ = <INFILE>; }
	      ($probe_name) = ($_ =~ m/<td.*?>(.+?)&nbsp;<\/td>/);
              $probe_name =~ s/^\s+//;
              $probe_name =~ s/\s+$//;
	      ($target) = ($probe_name =~ m/(\d+)h.*/);
	      #print "Target: $target \tNAME: $probe_name\n"; 

	      for (my $i=0; $i < 7; $i++)
	      { $_ = <INFILE>; }
	      ($probe_results) = ($_ =~ m/<td.*?>(.+)&nbsp;<\/td>/);
	      #print "RESULT: $probe_results\n";
	      
	      for (my $i=0; $i<11; $i++)
	      { $_ = <INFILE>; }
	      ($probe_sequence) = ($_ =~ m/<td.*?>([ATGC]{36})&nbsp;<\/td>/);
	      #print "SEQUENCE: $probe_sequence\n\n";
              push @data, [$target, $probe_name, $probe_results, $probe_sequence, "", "", "", "", "", "", "", "", "", "", "", ""];
	     

      }

}
close (INFILE);
@data = sort {$a->[0] cmp $b->[0]} @data;

for (my $i=0; $i<=$#species ; $i++)
{
   $probe_idx = 0;
   $target_start_idx =0;
   $process_target = 0;
   $last_target = 0;
   debug_print "Processing species [$species[$i]]\n";
   while ($probe_idx <= $#data)
   {
      
      if ($data[$probe_idx][0] ne $last_target)
      {
        $target_start_idx = $probe_idx;
        $last_target = $data[$probe_idx][0];
        $target_file = "$location/Target$data[$probe_idx][0]/human_$species[$i]_T$data[$probe_idx][0].verbose";
	debug_print "checking $target_file\n";
        if (-e $target_file)
	{
          $last_target = $data[$probe_idx][0];
          # if file handle is open, close it.
          $alignment->close_file;

	  debug_print "Processing $target_file\n";
 	  debug_print "=================================================================\n\n";
	  # Open new target file
          $alignment->open_file($target_file);
          $process_target = 1;
        }
        else
        {
          $process_target = 0;
        }
      }
     
      if ($process_target == 1)
      {
        $species_dna = ""; 
	$alignment->set_probe($data[$probe_idx][3]);
             debug_print "Looking for regular                               [$data[$probe_idx][3]]\n";
	$species_dna = $alignment->find_probe_match;
        if ($species_dna eq "")
        {
	   $reverse_compliment = reverse_compliment($data[$probe_idx][3]);
	     debug_print "Looking for reverse compliment                    [$reverse_compliment]\n";
	   $alignment->set_probe($reverse_compliment);
	   $species_dna = $alignment->find_probe_match;
           if ($species_dna ne "")
           {
             debug_print "Species dna found, but in reverse compliment form [$species_dna]\n";
             $species_dna = reverse_compliment($species_dna);
           }
	     debug_print "Result of probe search                            [$species_dna]\n";
        }
             debug_print "Saving species dna                                [$species_dna]\n";
        $data[$probe_idx][$i+4] = $species_dna;
        debug_print "Target  : [$data[$probe_idx][0]]\n";
        debug_print "Target file : [$target_file]\n";
        debug_print "Probe   : [$data[$probe_idx][1]]\n";
        debug_print "Results : [$data[$probe_idx][2]]\n";
	debug_print "Human Probe   :     [$data[$probe_idx][3]]\n";
        debug_print "Found species DNA : [$species_dna]\n\n\n";
      }
      $probe_idx++;
   }
}
for (my $i=0; $i<=$#data; $i++)
{
   print "Probe $i\n";
   print "Target: $data[$i][0]\n";
   print "Name: $data[$i][1]\n";
   print "Results: $data[$i][2]\n";
   print "Human  : [$data[$i][3]]\n";
   print "Mouse  : [$data[$i][4]]\n";
   print "Baboon : [$data[$i][5]]\n";
   print "Cat    : [$data[$i][6]]\n";
   print "Chicken: [$data[$i][7]]\n";
   print "Chimp  : [$data[$i][8]]\n";
   print "Cow    : [$data[$i][9]]\n";
   print "Dog    : [$data[$i][10]]\n";
   print "Fugu   : [$data[$i][11]]\n";
   print "Pig    : [$data[$i][12]]\n";
   print "Rat    : [$data[$i][13]]\n";
   print "Tetra  : [$data[$i][14]]\n";
   print "Zfish  : [$data[$i][15]]\n";
   print "\n\n";
}