pattern_mminput.pl

   1 #!/usr/bin/perl
   2 # pattern_mminput - convert patterns for Remi Coulom's MM tool
   3 #
   4 # First, run pattern_enumerate, then feed it to this one. The script will
   5 # also use the pattern.fdict file generated by pattern_enumerate, so direct
   6 # piping probably will not work.
   7 #
   8 # We use two-pass approach mainly basically to find out how many features
   9 # of certain kind there are.
  10 #
  11 # See mm/README for format of generated output.
  12
  13 my @fi, @fc, @fn; # start indices, counts, maps
  14
  15 my $total;
  16 my $df = "patterns.fdict";
  17 open D, $df or die "$df: $!";
  18 while (<D>) {
  19         chomp;
  20         my ($si, $fi, $fpi, $fn, $fp) = (/^(.*?) (.*?):(.*?) (.*?):(.*?)$/);
  21         $fn[$fi] = $fn;
  22         defined $fi[$fi] or $fi[$fi] = $si;
  23         $fc[$fi]++;
  24         $total = $si + 1;
  25 }
  26 close D;
  27
  28 print "! $total\n";
  29 print (@fi."\n");
  30 foreach my $i (0 .. $#fi) {
  31         print $fc[$i]." ".$fn[$i]."\n";
  32 }
  33 print "!\n";
  34
  35 while (<>) {
  36         chomp;
  37         s/^\(//; s/\)$//;
  38         my @patterns = split(/\) \(/, $_);
  39         my @teams = ();
  40         print "#\n";
  41         foreach my $pat (@patterns) {
  42                 my @feats = split(/ /, $pat);
  43                 my @tfeats = ();
  44                 foreach my $feat (@feats) {
  45                         my ($fi, $fp) = split(/:/, $feat);
  46                         push @tfeats, $fi[$fi] + $fp;
  47                 }
  48                 push @teams, join(' ', @tfeats);
  49         }
  50         print $teams[0]."\n";
  51         foreach my $i (0 .. $#teams) {
  52                 print $teams[$i]."\n";
  53         }
  54 }