#!/usr/bin/perl -w

# -----------------------------------------------------------------------
# repeatTable 1.0
# 
# This script extracts genoStart and the genoEnd value from a repeat 
# list file. The values are extracted and written into an output file.
# Every value pair is in a new line and surrounded by '()'.
# The values of the line from the repeat list are extracted if the 
# conditions of the passed parameters are true.
# For every value in the repeat list can be passed a condition.
#
# For example:
# You want to extract the start and end positions where the following
# conditions are true:
# - Repeat Family is L1
# - Strand is +
# - Repeat Class is LINE
# - Repeat Name is L1Md_A
# 
# You have to use the following command line:
# ./repeatTable input.txt output.txt -repFamily L1 -strand + -repClass LINE -repName L1Md_A
#
#  Copyright by Joern Hameister (2005)
# -----------------------------------------------------------------------

sub printHelp {
		print " repeatTable 1.0\n";
		print " \n";
		print " This script extracts genoStart and the genoEnd value from a repeat \n";
		print " list file. The values are extracted and written into an output file.\n";
		print " Every value pair is in a new line and surrounded by '()'.\n";
		print " The values of the line from the repeat list are extracted if the \n";
		print " conditions of the passed parameters are true.\n";
		print " For every value in the repeat list can be passed a condition.\n";
		print "\n";
		print " For example:\n";
		print " You want to extract the start and end positions where the following\n";
		print " conditions are true:\n";
		print " - Repeat Family is L1\n";
		print " - Strand is +\n";
		print " - Repeat Class is LINE\n";
		print " - Repeat Name is L1Md_A\n";
		print " \n";
		print " You have to use the following command line:\n";
		print " ./repeatTable input.txt output.txt -repFamily L1 -strand + -repClass LINE -repName L1Md_A	\n";
		print " \n";
		print " \n";
		print "./repeatTable INPUTFILE OUTPUTFILE -[OPTIONS]\n";
		print "\n";
	  print "-bin:\t\t 607 \n";
	  print "-swScore:\t 27368\n";
		print "-milliDiv:\t 25\n";
		print "-milliDel:\t 1\n";
		print "-milliIns:\t 0\n";
		print "-genoName:\t chr1\n";
		print "-genoStart:\t 3000000\n";
		print "-genoEnd:\t 3003389\n";
		print "-genoLeft:\t -192106223\n";
		print "-strand:\t +\n";
		print "-repName:\t L1Md_A\n";
		print "-repClass:\t LINE\n";
		print "-repFamily:\t L1\n";
		print "-repStart:\t 2193\n";
		print "-repEnd:\t 5586\n";
		print "-repLeft:\t -297\n";
		print "-id:\t\t 1\n";
}

if(@ARGV%2==1) {
	if($ARGV[0] ne '-h') { 
		print "Wrong arguments!\n";
	}
	printHelp();
	exit(-1);
}

	# load input file
	open(INPUT , "<", $ARGV[0]);
	open(OUTPUT, ">", $ARGV[1]);
	open(LOG, ">", "log.txt");
	
  my $bin = 0;
  my $swScore = 0;
	my $milliDiv = 0;
	my $milliDel = 0;
	my $milliIns = 0;
	my $genoName = 0;
	my $genoStart = 0;
	my $genoEnd = 0;
	my $genoLeft = 0;
	my $strand = 0;
	my $repName = 0;
	my $repClass = 0;
	my $repFamily = 0;
	my $repStart = 0;
	my $repEnd = 0;
	my $repLeft = 0;
	my $id = 0;

	my $binValue;
	my $swScoreValue;
	my $milliDivValue;
	my $milliDelValue;
	my $milliInsValue;
	my $genoNameValue;
	my $genoStartValue;
	my $genoEndValue;
	my $genoLeftValue;
	my $strandValue;
	my $repNameValue;
	my $repClassValue;
	my $repFamilyValue;
	my $repStartValue;
	my $repEndValue;
	my $repLeftValue;
	my $idValue;
	
	my $counter = 0;
	foreach $arg (@ARGV) {
		# print "$arg\n";
		
		if($arg eq '-h') {
			printHelp();
			
			exit(0);
		}
		if($arg eq '-bin') {
			$bin = 1;
			$binValue = $ARGV[$counter+1];
			print "Set value: $binValue\n";
		}
		elsif($arg eq '-swScore') {
			$swScore = 1;
			$swScoreValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-milliDivValue') {
			$milliDiv = 1;
			$milliDivValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-milliDel') {
			$milliDel = 1;
			$milliDelValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-milliIns') {
			$milliIns = 1;
			$milliInsValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-genoName') {
			$genoName = 1;
			$genoNameValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-genoStart') {
			$genoStart = 1;
			$genoStartValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-genoEnd') {
			$genoEnd = 1;
			$genoEndValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-genoLeft') {
			$genoLeft = 1;
			$genoLeftValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-strand') {
			$strand = 1;
			$strandValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repName') {
			$repName = 1;
			$repNameValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repClass') {
			$repClass = 1;
			$repClassValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repFamily') {
			$repFamily = 1;
			$repFamilyValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repStart') {
			$repStart = 1;
			$repStartValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repEnd') {
			$repEnd = 1;
			$repEndValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-repLeft') {
			$repLeft = 1;
			$repLeftValue = $ARGV[$counter+1];
		}
		elsif($arg eq '-id') {
			$id = 1;
			$idValue = $ARGV[$counter+1];
		}

		
		
		$counter++;
	}
	
	
	while(defined ($line = <INPUT>)) {
	  chomp($line);

		# Use tabs as separator	  
	  @fields = split /\t/,$line;
	  
	  my $miss = 0;
	  # Check fields and conditions
	  if($bin==1) {
	  	if($fields[0] ne $binValue) {
	  		$miss = 1;
	  	}
	  }
	  if($swScore==1) {
	  	if($fields[1] ne $swScoreValue) {
	  		$miss = 1;
	  	}
	  }
	  if($milliDiv==1) {
	  	if($fields[2] ne $milliDivValue) {
	  		$miss = 1;
	  	}
	  }
	  if($milliDel==1) {
	  	if($fields[3] ne $milliDelValue) {
	  		$miss = 1;
	  	}
	  }
	  if($milliIns==1) {
	  	if($fields[4] ne $milliInsValue) {
	  		$miss = 1;
	  	}
	  }
	  if($genoName==1) {
	  	if($fields[5] ne $genoNameValue) {
	  		$miss = 1;
	  	}
	  }
	  if($genoStart==1) {
	  	if($fields[6] ne $genoStartValue) {
	  		$miss = 1;
	  	}
	  }
	  if($genoEnd==1) {
	  	if($fields[7] ne $genoEndValue) {
	  		$miss = 1;
	  	}
	  }
	  if($genoLeft==1) {
	  	if($fields[8] ne $genoLeftValue) {
	  		$miss = 1;
	  	}
	  }
	  if($strand==1) {
	  	if($fields[9] ne $strandValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repName==1) {
	  	if($fields[10] ne $repNameValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repClass==1) {
	  	if($fields[11] ne $repClassValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repFamily==1) {
	  	if($fields[12] ne $repFamilyValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repStart==1) {
	  	if($fields[13] ne $repStartValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repEnd==1) {
	  	if($fields[14] ne $repEndValue) {
	  		$miss = 1;
	  	}
	  }
	  if($repLeft==1) {
	  	if($fields[15] ne $repLeftValue) {
	  		$miss = 1;
	  	}
	  }
	  if($id==1) {
	  	if($fields[16] ne $idValue) {
	  		$miss = 1;
	  	}
	  }

		if($miss!=1) {
			print LOG "$line";
			$startPos = $fields[6]+1;
			print OUTPUT "{$startPos, $fields[7]}\n";
			
		}
	}