#!/usr/bin/perl
# gff2zff2.pl
## cant sort input gff by loc; loosing gene-exon ordering...

my($forxdef,$forhmm)=(0,0); ## forxdef is only one we need?
use Getopt::Long;
$ok= GetOptions("xdef!" => \$forxdef, "hmm!" => \$forhmm);
die "usage: gff2zff2.pl -hmm | -xdef < gff > zff\n -xdef is probably right.\n" 
unless($ok && ($forxdef||$forhmm));

if($forhmm) { zffForHmm(); }
else { zffForXdef(); }  # default

sub zffForXdef {
  my $zcmd="ADJ"; # or "SET" ?
  my $dt="."; # or "0" ?
  my ($pid, $lpid);
  while(<>){
    next unless(/^\w/);
    chomp;  
    my ($ref,$src,$type,$b,$e,$score,$strand,$shft,$attr)= split "\t";
    if($attr =~ m/(Parent|ID)=([^;\s]+)/) { $pid=$2; $pid =~ s/_[GSo]\d+$//; }
  
    if($ref ne $lref) {
      my($sb,$se,$tp)=($le,$le);
      $tp= ($lstrand eq '-') ? "Start" : "Stop";
      print join("\t", $tp,$sb,$se,$lstrand,'+50',$dt,$dt,$dt,$zcmd,$lref),"\n" if($le); $le=0;
      #print ">$ref\n";
      $lref= $ref; $lpid='';
    }
    $score =~ s/\s+//g;
    #$score = int($score); #? int($score/50); #?
    $score=100 if($score>100); $score= $score/100; $score=0.1 if ($score<0.1);
    ## for Coding, use values in 0.1 .. 1,2 range ??; higher for Start/Stop
    next if ($score <= 0);
    if($pid ne $lpid) {
      my($sb,$se,$tp)=($le,$le);
      $tp= ($lstrand eq '-') ? "Start" : "Stop";
      print join("\t", $tp,$sb,$se,$lstrand,'+50',$dt,$dt,$dt,$zcmd,$lref),"\n" if($le); $le=0;
      print ">$ref:$pid\n";
      ($sb,$se)=($b,$b);
      $tp= ($strand eq '-') ? "Stop" : "Start";
      print join("\t", $tp,$sb,$se,$strand,'+50',$dt,$dt,$dt,$zcmd,$ref),"\n";
      $lpid= $pid;
    }
    print join("\t", "Coding",$b,$e,$strand,'+'.$score,$dt,$dt,$dt,$zcmd,$ref),"\n";
    ($lb,$le,$lstrand,$lscore)=($b,$e,$strand,$score);
  }
}

=item hmm zff
>scaffold_4845
Einit   6571    6681    +       10.789  0       0       0       scaffold_4845-SNAP_dm.1
Eterm   6840    7199    +       11.685  0       0       2       scaffold_4845-SNAP_dm.1

=cut

sub zffForHmm {
  my ($pid, $lpid, @lv);
  my $etype="Einit";
  while(<>){
    next unless(/^\w/);
    chomp;  
    
    my @v= split "\t";
    my ($ref,$src,$type,$b,$e,$score,$strand,$shft,$attr)= @v;
    if($attr =~ m/(Parent|ID)=([^;\s]+)/) { $pid=$2; $pid =~ s/_[GSo]\d+$//; }
  
    $endgene= ($lpid ne $pid || $ref ne $lref);
    
    $score =~ s/\s+//g;
    $score = int($score); #? int($score/50); #?
    ## for Coding, use values in 0.1 .. 1,2 range ??; higher for Start/Stop
  
    $etype= "Eterm" if($etype eq "Exon" && $endgene);
    $etype= "Esngl" if($etype eq "Einit" && $endgene);
    print join("\t", $etype,$lb,$le,$lstrand,'+'.$lscore,0,0,0,$lpid,$lref),"\n" if($le); $le=0;

    if($ref ne $lref) {
      print ">$ref\n"; 
      $lref= $ref;  $lpid='';
    }
  
    next if ($score <= 0);
    if($pid ne $lpid) {
      $lpid= $pid;
      $etype="Einit";
    } else { 
      $etype="Exon"; 
    }
    ($lb,$le,$lstrand,$lscore)=($b,$e,$strand,$score);
    @lv= @v;
  }

  $etype= "Eterm" if($etype eq "Exon" && $endgene);
  $etype= "Esngl" if($etype eq "Einit" && $endgene);
  print join("\t", $etype,$lb,$le,$lstrand,'+'.$lscore,0,0,0,$lpid,$lref),"\n" if($le); $le=0;
}