#!/usr/bin/perl
#
# nuskaityti bindingdb sdf failus
# H-Se/DB26984_50278320.sdf kaip testas imtas
#
# VR
#

#use warnings;
use Data::Dumper;


#nuskaityti failus i @molekules
die "usage VR_bindingdb_parse.pl file.sdf" if (@ARGV != 1);
$FILE = $ARGV[0];
open FILE, "<", $FILE or die $!;
$x=0;
$key='str';
while(<FILE>) {
 chop($_);
 next if ($_ =~ /^$/);   #jei eilute tuscia praleidziam
 if ($_ =~ /\$\$\$\$/) { #jei eiluteje yra: $$$$
  $x++;
  $key = 'str';
  next;
 }
 #print $x, $_;
 if ($_ =~ /^\>/) {      #jei eilute prasideda: >
  chop();
  $_ =~ s/>  <//;
  $key = $_;
 } else {
  #$molekules[molekules nr.]{irasas}
  push @{ $molekules[$x]{$key} }, $_;
 }
}
close FILE;
#print STDERR Dumper \@molekules;
print $FILE, "\n";


#rasti nuo kur prasideda kitokia molekule
#pirmoji kita yra $ta_pati+1
kita();
sub kita {
$y=0;
print $y;
$ta_pati=0;
while($molekules[$y]{'str'}) {
 #print STDERR Dumper $molekules[$y]{'str'}, "\n";
 $z=0;
 while($molekules[$y]{'str'}[$z]) {
  #jei z eilute nesutampa y ir y+1 nors karta tai kita molekule
  if ($molekules[$y]{'str'}[$z] ne $molekules[$y+1]{'str'}[$z]) {
   $ta_pati=$y;
   #print STDERR $y, " :", $molekules[$y]{'str'}[$z], " ir ", $molekules[$y+1]{'str'}[$z], "\n";
   last;
  }
  $z++;
 }
 last if ($ta_pati);
 $y++;
}
}
print STDERR Dumper $molekules[$ta_pati+1]{'str'}, "\n";