#!/usr/bin/perl -w

use strict;

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

if ($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': $!";
	
	$outputdir = "$ARGV[1]";
	print "Output directory received from command line...\n";
	#open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
}

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 output directory: ";
	chomp($outputdir = <STDIN>);
	#open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
}

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 output directory: ";
	chomp($outputdir = <STDIN>);
	#open OUTPUTFILE, ">$outputfile" || die "An error occured whilst creating your output file '$outputfile': $!";
}	

print "INPUT: $inputfile\n";
print "OUTPUT DIRECTORY: $outputdir\n";  #display input and output file names

my $lineindex = 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

	my $molconvert_input = "./$outputdir/smiles/molconvert_input$lineindex.smi";	
	open OUTPUTFILE, "> $molconvert_input" || die "An error occured whilst creating your output file '$molconvert_input': $!";
	print OUTPUTFILE "$line\n";  # add new line character as all the fields in this row have been printed
	
	my $molconvert_output = "./$outputdir/sdf/molconvert_output$lineindex.sdf";
	
	system "molconvert -F \"-3:[hydrogenize]\" sdf $molconvert_input -o $molconvert_output";
	
	my $xedex_output = "./$outputdir/sdf_xedex/xedex_output$lineindex.sdf";
	my $xedex_input = $molconvert_output;
	
	system "xedex $xedex_input > $xedex_output";
	
	my $sdftoxed_output = "./$outputdir/xed/sdftoxed_output$lineindex.xed";
	my $sdftoxed_input = $xedex_output;	
	
	system "sdf_to_xed -F $xedex_output > $sdftoxed_output";
		
	$lineindex++;  # increment current line

}

close OUTPUTFILE;


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