#!/usr/bin/perl

use strict;
use subs;
use Data::Dumper;
require XML::Simple;

my %args = getParams();

my $probe_file = new XML::Simple();
my $db_file = new XML::Simple();

my $all_probes = $probe_file->XMLin($args{probefile}, keyattr => {});
my $db_probes = $db_file->XMLin($args{dbfile}, keyattr => {});


open (OUTFILE, ">".$args{outfile});
writeXMLheader( *OUTFILE );
for (my $i=0; $i<@{$all_probes->{probe}}; $i++)
{
  my $probe1 = $all_probes->{probe}->[$i]->{species}->[0]->{sequence};
  print "Validate Generic Probe [$probe1]\n" if ($args{verbose});
  for(my $j=0; $j<@{$db_probes->{parent_alignment}}; $j++)
  {
     my $probe2 = $db_probes->{parent_alignment}->[$j]->{probe}->{sequence};
     print "                   ... [$probe2]\n" if ($args{verbose});
     if ( $probe1 eq $probe2 )
     {
       print "Matching DB probe found, write generic probe to file\n\n" if ($args{verbose});
       writeProbe (*OUTFILE, $all_probes->{probe}->[$i]);
       last;
     }
  } 
}
writeXMLfooter( *OUTFILE );

sub writeProbe
{

  local *OUTFILE = shift;  
  my $probe   = shift;

  printf OUTFILE  "   <probe>\n";  
  printf OUTFILE  "      <time>%s</time>\n", $probe->{time};
  printf OUTFILE  "      <date>%s</date>\n",$probe->{date};
  printf OUTFILE  "      <name></name>\n";
  printf OUTFILE  "      <length>%d</length>\n",$probe->{length};
  printf OUTFILE  "      <alignment_identity>%d</alignment_identity>\n",$probe->{alignment_identity};
  printf OUTFILE  "      <probe_identity>%d</probe_identity>\n",$probe->{probe_identity};
  printf OUTFILE  "      <transversions>%d</transversions>\n", $probe->{transversions};
  printf OUTFILE  "      <transitions>%d</transitions>\n",$probe->{transitions};
  printf OUTFILE  "      <gaps>%d</gaps>\n", $probe->{gaps};
  printf OUTFILE  "      <probe_alignment>%s</probe_alignment>\n", $probe->{probe_alignment};
  printf OUTFILE  "      <species id='1'>\n";
  printf OUTFILE  "         <name></name>\n";
  printf OUTFILE  "         <file></file>\n";
  printf OUTFILE  "         <version></version>\n";
  printf OUTFILE  "         <alignment_start>%d</alignment_start>\n",$probe->{species}->[0]->{alignment_start};
  printf OUTFILE  "         <alignment_end>%d</alignment_end>\n",$probe->{species}->[0]->{alignment_end};
  printf OUTFILE  "         <chromosome></chromosome>\n";
  printf OUTFILE  "         <alignment_strand></alignment_strand>\n";
  printf OUTFILE  "         <probe_start>%d</probe_start>\n", $probe->{species}->[0]->{probe_start};
  printf OUTFILE  "         <probe_end>%d</probe_end>\n", $probe->{species}->[0]->{probe_end};
  printf OUTFILE  "         <sequence>%s</sequence>\n", $probe->{species}->[0]->{sequence};
  printf OUTFILE  "      </species>\n";
  printf OUTFILE  "      <species id='2'>\n";
  printf OUTFILE  "         <name></name>\n";
  printf OUTFILE  "         <file></file>\n";
  printf OUTFILE  "         <version></version>\n";
  printf OUTFILE  "         <alignment_start>%d</alignment_start>\n",$probe->{species}->[1]->{alignment_start};
  printf OUTFILE  "         <alignment_end>%d</alignment_end>\n", $probe->{species}->[1]->{alignment_end};
  printf OUTFILE  "         <alignment_strand></alignment_strand>\n";
  printf OUTFILE  "         <chromosome></chromosome>\n";
  printf OUTFILE  "         <probe_start>%d</probe_start>\n",$probe->{species}->[1]->{probe_start};
  printf OUTFILE  "         <probe_end>%d</probe_end>\n",$probe->{species}->[1]->{probe_end};
  printf OUTFILE  "         <sequence>%s</sequence>\n", $probe->{species}->[1]->{sequence};
  printf OUTFILE  "      </species>\n";
  printf OUTFILE  "   </probe>\n\n\n";  
}

sub writeXMLheader
{
  local *OUTFILE = shift;  
  print OUTFILE "<?xml version='1.0' encoding='ISO-8859-1'?>\n";
  print OUTFILE "<probe_results>\n\n";

}

sub writeXMLfooter
{
  local *OUTFILE = shift;  
  print OUTFILE "\n\n</probe_results>\n";
}


#################################################
#  getParams - This gets command line parameters
#################################################

sub getParams {
    my %opts;
    my $errstr = "";
    use Getopt::Long;
    GetOptions(\%opts, 'dbfile=s', 'probefile=s', 'verbose', 'outfile=s', 'help');

    if (defined($opts{help}))
    {
      $errstr = "\n";
    }
    else
    {
      if (!defined($opts{dbfile}))
      {
        $errstr .= "ERROR : File containing database probes not set\n";
      }
      elsif ( ! -f $opts{dbfile} )
      {
        $errstr .= "ERROR : Database file doesn't exist\n";
      }

      if (!defined($opts{probefile}))
      {
        $errstr .= "ERROR : Probe File not set\n";
      }
      elsif ( ! -f $opts{probefile} )
      {
        $errstr .= "ERROR : Probe file doesn't exist\n";
      }
    }

    if ($errstr) {
        print <<"END_USAGE";

Description: 

This script will take a database generated file (dbfile in xml v1 format) 
and a probe file (xml v2 format) and will create a xml file that contains 
only the probes from the database.  The xml file will be in v2 format, 
and will contain all the information from the probe file.

Usage: select_db_probes.pl -dbfile <filename> -probefile <filename> [-outfile <filename>] [-help] [-verbose]

Probe Sources:
-dbfile     Name of file that contains the probes that were generated from the DB (xml v1 format)
-probefile  Name of file that contains all probes (xml v2 format)
-outfile    Name of the output file, data will be in xml v1 format (default name : select_db_probes.xml)
-help       Display help for command
-verbose    Display all debug messages during execution.

END_USAGE
        print "\n$errstr" if ($errstr);
        die("\n");
    }


    return (dbfile    => $opts{dbfile},
            probefile => $opts{probefile}, 
            outfile   => $opts{outfile} || "select_db_probes.xml",
            verbose   => $opts{verbose} || 0)
}