#!/usr/bin/perl -w

use strict;

my $numArgs = $#ARGV + 1;  # calculate number of arguments passed to program from command line
our $inputfile;
our $filterfile;
our $outputfile;  # initialise filename variables
our $indexcolumn;

if ($numArgs >= 4) {
	$inputfile = "$ARGV[0]";
	print "Input filename received from command line.  Opening '$inputfile'...\n";
	open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	$filterfile = "$ARGV[1]";
	print "Filter filename received from command line.  Opening '$filterfile'...\n";
	open INPUTFILE, "<$filterfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	$outputfile = "$ARGV[2]";
	print "Output filename received from command line...\n";
	open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
	
	$indexcolumn = "$ARGV[3]"; 
	print "Index column ($indexcolumn) received from command line...\n";
}

elsif ($numArgs == 3) {
	$inputfile = "$ARGV[0]";
	print "Input filename received from command line.  Opening '$inputfile'...\n";
	open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	$filterfile = "$ARGV[1]";
	print "Filter filename received from command line.  Opening '$filterfile'...\n";
	open INPUTFILE, "<$filterfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	$outputfile = "$ARGV[2]";
	print "Output filename received from command line...\n";
	open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
	
	print "Please enter the index column of the input file: ";
	chomp($indexcolumn = <STDIN>);
}

elsif ($numArgs == 2) {
	$inputfile = "$ARGV[0]";
	print "Input filename received from command line.  Opening '$inputfile'...\n";
	open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	$filterfile = "$ARGV[1]";
	print "Filter filename received from command line.  Opening '$filterfile'...\n";
	open INPUTFILE, "<$filterfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	print "Please enter the path to your output file: ";
	chomp($outputfile = <STDIN>);
	open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
	
	print "Please enter the index column of the input file: ";
	chomp($indexcolumn = <STDIN>);
}

elsif ($numArgs == 1) {
	$inputfile = "$ARGV[0]";
	print "Input filename received from command line.  Opening '$inputfile'...\n";
	open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	print "Please enter the path to your filter file: ";
	chomp($filterfile = <STDIN>);
	open INPUTFILE, "<$filterfile" || die "An error occured whilst opening your filter file '$filterfile': $!";
	
	print "Please enter the path to your output file: ";
	chomp($outputfile = <STDIN>);
	open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
	
	print "Please enter the index column of the input file: ";
	chomp($indexcolumn = <STDIN>);
}

else {
	print "Please enter the path to your input file: ";
	chomp($inputfile = <STDIN>);
	open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";
	
	print "Please enter the path to your filter file: ";
	chomp($filterfile = <STDIN>);
	open INPUTFILE, "<$filterfile" || die "An error occured whilst opening your filter file '$filterfile': $!";
	
	print "Please enter the path to your output file: ";
	chomp($outputfile = <STDIN>);
	open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
	
	print "Please enter the index column of the input file: ";
	chomp($indexcolumn = <STDIN>);
}	

print "INPUT: $inputfile\n";
print "OUTPUT: $outputfile\n";  
print "FILTER: $filterfile\n"; #display input and output file names

my $lineindex = 0;
my @fields;
my $fields;
my $blanks = 0;
my $blocked = 0;
my $lineindex2 = 0;
my @blocklist;
my $blocklist;
my $isblocked = 0;
my $isblank = 0;
my $blocklistindex = 0;

while(defined(my $line = <INPUTFILE>)) {  # read through each line in turn while they still exist 

	$line =~ s/["]+//g;  # remove quotation marks from SMILES strings
	$line =~ s/[\n]+//g;  # remove new line characters

	$_ = $line;
	chomp $_;
		
	@fields = split;

	$blocklist[$blocklistindex] = $fields[1] - 1; # account for smiles index starting from 1 instead of 0.
	
	$blocklistindex++;

}

close INPUTFILE;
open INPUTFILE, "<$inputfile" || die "An error occured whilst opening your input file '$inputfile': $!";

while(defined(my $line = <INPUTFILE>)) {  # read through each line in turn while they still exist 
	
	$line =~ s/["]+//g;  # remove quotation marks from SMILES strings
	$line =~ s/[\n]+//g;  # remove new line characters

	@fields = split(/,\t*/, $line);  # split line into fields according to comma delimiter

	if (!$fields[0] && !$fields[1] && !$fields[2] && !$fields[3]) {
		print "Blank at line $lineindex\n";
		$blanks++;
		$isblank = 1;
	}
	
	for $lineindex2 (0 .. $blocklistindex - 1) {
	
		#print "lineindex2: $lineindex2\n";
		#print "fields[IC]: $fields[$indexcolumn]\n";
		#print "blocklist[LI2]: $blocklist[$lineindex2]\n";
	
		if ($fields[$indexcolumn] == $blocklist[$lineindex2]) {
			print "compound $fields[$indexcolumn] blocked...\n";
			$blocked++;
			$isblocked = 1;
		}	
	
	}
	
	if ($isblank == 0 && $isblocked == 0) {
		print OUTPUTFILE "$line\n"; 
	}
	
	$lineindex++;  # increment current line
	$isblank = 0;
	$isblocked = 0;
	$lineindex2 = 0;

}

print "Found $blanks blank lines.\n"; 
print "Blocked $blocked compounds.\n"; 

close OUTPUTFILE;
close INPUTFILE;


#print "molconvert -F '-3:[hydrogenize]' sdf $inputfile -o $outputfile";