tools/wn2rdf.pl

   1 #! /usr/bin/perl -w
   2
   3 # Wordnet dictionary database converter
   4 #
   5 # Converts the Wordnet prolog data to rockbox dictionary format.
   6 #
   7 # Written by Miika Pekkarinen <slasher@ihme.org>
   8 #
   9 # $Id$
  10
  11 use strict;
  12
  13 # Lookup tables
  14 my %words;
  15 my %descriptions;
  16
  17 sub getcatname {
  18         my ($id) = @_;
  19
  20         return 'N' if $id == 1;
  21         return 'V' if $id == 2;
  22         return 'A' if $id == 3;
  23         return 'A' if $id == 4;
  24         return '?';
  25 }
  26
  27 open IN_WORD, "wn_s.pl" or die "Open fail(#1): $!";
  28 open IN_DESC, "wn_g.pl" or die "Open fail(#2): $!";
  29 open OUTPUT, "> dict.preparsed" or die "Open fail(#3): $!";
  30
  31 print "Reading word file...\n";
  32
  33 # Read everything into memory
  34 while (<IN_WORD>) {
  35         chomp ;
  36
  37         # s(100001740,1,'entity',n,1,11). => 100001740,1,'entity',n,1,11
  38         s/(^s\()(.*)(\)\.$)/$2/;
  39
  40         my ($seqid, $n1, $word, $n2, $n3, $n4) = split /,/, $_, 6;
  41
  42         # 'entity' => entity
  43         $word =~ s/(^\')(.*)(\'$)/$2/;
  44         $word =~ s/\'\'/\'/s;
  45
  46         my $category = substr $seqid, 0, 1;
  47
  48         $words{lc $word}{$seqid} = $category;
  49 }
  50
  51 close IN_WORD;
  52
  53 print "Reading description file...\n";
  54 while (<IN_DESC>) {
  55         chomp ;
  56
  57         # g(100002056,'(a separate and self-contained entity)').
  58         # => 100002056,'(a separate and self-contained entity)'
  59         s/(^g\()(.*)(\)\.$)/$2/;
  60
  61         my ($seqid, $desc) = split /,/, $_, 2;
  62
  63         $desc =~ s/(^\'\()(.*)(\)\'$)/$2/;
  64         $desc =~ s/\'\'/\'/s;
  65
  66         $descriptions{$seqid} = $desc;
  67 }
  68
  69 close IN_DESC;
  70
  71 print "Sorting and writing output...\n";
  72
  73 # Now sort and find correct descriptions
  74 foreach my $word (sort keys %words) {
  75         my %categories;
  76
  77         # Find all definitions of the word
  78         foreach my $id (keys %{$words{$word}}) {
  79                 my $catid = $words{$word}{$id};
  80                 my $description = $descriptions{$id};
  81
  82                 if (!defined($description) or $description eq '') {
  83                         print "Error: Failed to link word: $word / ",
  84                           $words{$word}, "\n";
  85                         exit 1;
  86                 }
  87
  88                 push @{$categories{$catid}}, $description;
  89         }
  90
  91         my $finaldesc;
  92
  93         # 1 = noun
  94         # 2 = verb
  95         # 3 = adjective
  96         # 4 = adverb
  97         for my $catid (1 .. 4) {
  98                 my $n = 1;
  99                 my $catdesc;
 100
 101                 next unless $categories{$catid};
 102                 foreach my $desc ( @{$categories{$catid}} ) {
 103                         $catdesc .= " " if $catdesc;
 104                         $catdesc .= "$n. $desc";
 105                         $n++;
 106                 }
 107
 108                 next unless $catdesc;
 109                 $finaldesc .= "\t" if $finaldesc;
 110                 $finaldesc .= getcatname($catid) . ": $catdesc"
 111         }
 112
 113         die "Internal error" unless $finaldesc;
 114
 115         print OUTPUT "$word\t$finaldesc\n";
 116 }
 117
 118 close OUTPUT;
 119
 120 print "Done, output was successfully written!\n";
 121
 122