#!/usr/bin/perl

use strict;
use subs;
use POSIX;
use Data::Dumper;

use constant LINE_SIZE => 60;
use constant UNGAPPED_ALIGNMENT_INDEX_DENSITY => 10;  # Print an index every n pip lines


my %args = getParams();
my $alignment = 0;
my $alignment_offset=0;
my $human_start=0;
my $human_end=0;
my $species_start=0;
my $species_end=0;
my $fasta_name = "";
my $fasta_file = "";
my $offset = 0;
my $pip_location;
my $desc_location;
my $location = 0;
my $pip_count;
my $description;
my $pip_string="";
my @results;
my $temp_var = 0;

open (BLASTZ, $args{file});
while (<BLASTZ>)
{

  if (($_ =~ m/^#:lav/) && ($location != 0))
  {
    push @results, [$human_start, $human_end, $desc_location, $location,$pip_string];
    $human_end = 0;
    $human_start = 0;
    $desc_location = 0;
    $location = 0;
    $pip_string = ""; 
  }
   
  if ($_ =~ m/^h {/)
  {
    $_ = <BLASTZ>;
    $desc_location = tell(BLASTZ);
    next;
  }

  next if !($_ =~ m/^[h ] [{bel] \d/);
  # Only lines that need processing are here

  if ($_ =~ m/^ *b /)
  {
    
    if ($human_end ne 0)
    {
      push @results, [$human_start, $human_end, $desc_location, $location,$pip_string];
      $pip_string = "";
      $temp_var++;
    }
    ($human_start) = ($_ =~ m/^ *b *(\d+) /);

  }
  elsif ($_ =~ m/^ *e /)
  {

    ($human_end) = ($_ =~ m/^ *e *(\d+) /);
    $location = tell(BLASTZ);
    $pip_count=0;

  }
  else
  {
    $pip_count++;
    # Only create actual offset for every nth pip
    if ($pip_count >= UNGAPPED_ALIGNMENT_INDEX_DENSITY)
    {
      $pip_count = 0;
      (my $pip_start) = ($_ =~ m/^ *l (\d+) /);
      $pip_string .= "," if ($pip_string ne "");
      $pip_string .= "$pip_start,$pip_location";
    }
    $pip_location = tell(BLASTZ);
  }
}
push @results, [$human_start, $human_end, $desc_location, $location,$pip_string] if ($human_end != 0);


@results = sort { $a->[0] <=> $b->[0] } @results;

for (my $i=0; $i<@results; $i++)
{
  print "$results[$i][0]";
  print ",$results[$i][1]";
  print ",$results[$i][2]";
  print ",$results[$i][3]";
  print ",$results[$i][4]"  if ($results[$i][4] ne "");
  print "\n";
 
}


sub getParams {
    my %opts;
    my $errstr = "";
    use Getopt::Long;
    GetOptions(\%opts, 'file=s');

    if (!defined($opts{file}))
    {
      $errstr .= "ERROR : No blastz file to index\n";
    }

    if ($errstr) {
        print <<"END_USAGE";
Usage: index_blastz.pl -file <filename> 

Description:
This will create an index to the blsatz file for quick searching.
The index will be printed to stdout.

Usage: 
Probe Sources:
-file    <filename>  Blastz file to be index

END_USAGE
        print "\n$errstr" if ($errstr);
        die("\n");
    }
    return (file    => $opts{file});
}