#!/usr/local/bin/perl
# go2eugenes.pl - recode FBgo.acode database gene ids to eugenes ids
# add to Meow.pm ?


use Meow;
use Meow::Data;
use Meow::Mouse;
use Meow::Yeast;

# use Acodes;
use flybase::Datalib;

my $fbobs= 'fbobs'; 
my $acode= 'FBgo.acode';
my $meowpath= 'fbobs/';
my $newpath= 'fbobnew/';
my $idtag= 'FBgo';

main();

sub main {

	# $ENV{MO_WORK_PATH}=`pwd`; # for initEnv's chdir($workpath), def='/bio/work/meow/tmp/work/'
	Meow::initEnv();
	readArgs();

	$mouseob= Meow::getDataclass('mouse');
	my $data= $meowpath . $mouseob->targetdata;
	$mouseob->openIdDb($data, 'r');

	$yeastob= Meow::getDataclass('yeast');
	$data= $meowpath . $yeastob->targetdata;
	$yeastob->openIdDb($data, 'r');
	
	$oldname= $meowpath . $acode;
	$oldlib= new flybase::Datalib();
	$oldfile= $oldlib->open($oldname);
	$newname= $newpath . $acode;
	$newlib= new flybase::Datalib();
	$newlib->setIdtag( $idtag); # should get from fname
	$newfile= $newlib->create($newname,1);

	if ($dorun) {
		processrecs( $oldlib, $newlib);
		}
	else {
		print STDERR "opened $oldfile\n";
		print STDERR "created $newfile\n";
		print STDERR "yeast id S0003302=". $yeastob->hasDID('S0003302')."\n";  
		print STDERR "yeast id S0004363=". $yeastob->hasDID('S0004363')."\n";  
		print STDERR "mouse id MGI:1353510=". $mouseob->hasDID('MGI:1353510')."\n";  
		print STDERR "mouse id MGI:1858961=". $mouseob->hasDID('MGI:1858961')."\n";  
		print STDERR "mouse id MGI:95640=". $mouseob->hasDID('MGI:95640')."\n";  
		}
		
	$newlib->close();
	$oldlib->close();
	
	$mouseob->closeIdDb();
	$yeastob->closeIdDb();
}

# GNM|MGI:1353510
# |Arhgef1
# GNY|SGD:S0000039
# |CDC24
# GNY|SGD:S0003302
# |ROM1
# GNY|SGD:S0004363
# |ROM2

sub processRec {
	my $rdata= shift;
	my ($db, $did, $egid, $tag, $changed, $newdat);
	my @dat= split(/\n/, $$rdata);
	foreach (@dat) {
		if (/^(GN\w*\|)(\w+):(\w+)/) {
			$tag= $1; $db= $2; $did= $3; 
			if ($db eq 'SGD') { 
				$egid= $yeastob->hasDID($did);  
				$_= $tag.$egid if ($egid);
				}		
			elsif ($db eq 'MGI') { 
				$egid= $mouseob->hasDID('MGI:'.$did);  
				$_= $tag.$egid if ($egid);
				}		
			elsif ($db eq 'FB') { #? drop the FB: of FB:FBgn...
				$_=  $tag.$did;
				}		
			}
		$newdat .= "$_\n";
	}
		
	return $newdat;
}
			
## this is bad; 
# 	while ($$rdata =~ s/\nGN(M|Y)\|(\w+):(\w+)/GN$1|xxxx/g) {
# 		$db= $2; $did= $3; $tag= "GN$1";
# 		$egid= '';
# 		if ($db eq 'SGD') { 
# 			$egid= $yeastob->hasDID($did);  
# 			}
# 		elsif ($db eq 'MGI') { 
# 			$egid= $mouseob->hasDID('MGI:'.$did);  
# 			}
# 		unless($egid) {
# 			warn "Cannot find eugenes id for $db:$dbid\n";
# 			}
# 		else { $$rdata =~ s/$tag\|xxxx/$tag|$egid/; $changed= 1; }
# 		}	

sub processrecs {
	my( $oldlib, $newlib ) = @_;
	my ($data, $newdata, $i , $changed); 
	my $oldfh=  $oldlib->inlib;

  $/= "# EOR\n";
  while ( $data= <$oldfh> ) {	
  	my $id=''; 
  	if ($data =~ /\nLID\|(\d+)/) { $id= $idtag.$1; } 
  	else { warn "No id for $data\n"; next; }
		$newdata= processRec( \$data);
		$newlib->addRecord( $id, $newdata);
		print STDERR '.';
		if ((++$i % 100) == 0) { print STDERR " $id\n"; }
		if ($changed && $i>1 && $i<10) { # $debug
			 print STDERR "\nOld $id\n$data\n";
			 print STDERR "\nNew $id\n$newdata\n";
			}
  	}
  $/= "\n";
}

sub usage {
	my @dbs= Meow::datakeys();
	for my $i (0..$#dbs) { $dbs[$i] =~ s/^\w+:://; }
	print <<TEOF;

Convert FBgo.acode gene database ids to euGenes ids
Usage  
  go2eugenes -run
Arguments:
\t -help  \t== brief usage
\t -view  \t== show actions without performing
\t -debug \t== debug output 
\t -run   \t== perform commands
Valid data sections are 
   all  @dbs

TEOF
	exit(0);

}

sub readArgs {
	print STDERR "ARGV= ( @ARGV )\n";  

	require Getopt::Long;    
	## Getopt::Long::config( qw( autoabbrev ignorecase permute ) );
	$Getopt::Long::autoabbrev= 1;
	$Getopt::Long::ignorecase= 1;
	$Getopt::Long::permute= $Getopt::Long::PERMUTE;
	
	@optlist= (
		'debug!' => \$debug,
		'help!' => \$dohelp,
		'view!' => \$view,
		'run!' => \$dorun,
		);
		
	$optokay= Getopt::Long::GetOptions( @optlist);
	usage() if ($dohelp || !$optokay);
}