wiktionary/parsewords.pl

   1 #!/usr/bin/perl
   2
   3 if ($#ARGV != 0)
   4 {
   5     print "Usage: $0 <filename> \n";
   6     exit;
   7 }
   8
   9 $filename = $ARGV[0];
  10 print "File: $filename\n";
  11 open(FILE, $filename) or die("Could not open file $filename");
  12 @alllines = <FILE>;
  13 $collect = 'NONE';
  14
  15 foreach $line (@alllines)
  16 {
  17     if ($line =~ m/<title>(\w+)<\/title>/) {
  18         print "\n";
  19         print "Word: $1\n";
  20         $collect = 'NONE';
  21
  22     } elsif ($collect eq 'ETYMOLOGY') {
  23         print "Etymology: $line \n";
  24         $collect = 'NONE';
  25     } elsif ($line =~ m/===Etymology/) {
  26         $collect = 'ETYMOLOGY';
  27
  28     } elsif ($collect eq 'NOUN') {
  29         if ($line =~ m/^#:/) {
  30             print "Noun Usage: $line";
  31         } elsif ($line =~ m/^#/) {
  32             print "Noun Meaning: $line";
  33         } elsif ($line =~ m/^==/) {
  34             $collect = 'NONE';
  35         }
  36     } elsif ($line =~ m/==Noun/) {
  37         $collect = 'NOUN';
  38
  39     } elsif ($collect eq 'VERB') {
  40         if ($line =~ m/^#:/) {
  41             print "Verb Usage: $line";
  42         } elsif ($line =~ m/^#/) {
  43             print "Verb Meaning: $line";
  44         } elsif ($line =~ m/^==/) {
  45             $collect = 'NONE';
  46         }
  47     } elsif ($line =~ m/==Verb/) {
  48         $collect = 'VERB';
  49
  50     } elsif ($collect eq 'ADJECTIVE') {
  51         if ($line =~ m/^#:/) {
  52             print "Adjective Usage: $line";
  53         } elsif ($line =~ m/^#/) {
  54             print "Adjective Meaning: $line";
  55         } elsif ($line =~ m/^==/) {
  56             $collect = 'NONE';
  57         }
  58     } elsif ($line =~ m/==Adjective/) {
  59         $collect = 'ADJECTIVE';
  60     }
  61 }
  62 close (FILE);