#!/usr/local/bin/perl

# (C) Copyright 1996 by John J. Chew, III <jjchew@math.utoronto.ca>
# All rights reserved

# ddtag.pl - tag text according to the dictionaries in which words are to
#   be found, e.g. prior to submission to a crossword game mailing list.

# <title>ddtag.pl</title>

unshift(@INC, "$ENV{HOME}/lib/perl");

require 'dawg.pl';
require 'getopts.pl';

$gVersion = "2.11";

# edit the following two lines to point at default local DAWG files

$opt_C = "$ENV{'HOME'}/scrabble/tcd/words.dawg";
$opt_N = "$ENV{'HOME'}/scrabble/tcd/nonwords.dawg";
$opt_O = "$ENV{'HOME'}/scrabble/osw/words.dawg";
$opt_T = "$ENV{'HOME'}/scrabble/ospd/words.dawg";

sub Usage {
  die "Usage: $0 [-1] [-C chambers.dawg] [-N non-tcd.dawg] [-O osw.dawg] [-T tw
l.dawg] [file...]\n"
     ."  -1                tag only first occurrence of each word\n"
     ."  -i                alter file in place, for use as a Pine alternate edi
tor\n"
     ."  -C chambers.dawg  DAWG file containing partial Chambers lexicon\n"
     ."  -N non-tcd.dawg   DAWG file containing list of known non-Chambers word
s\n"
     ."  -O osw.dawg       DAWG file containing OSW lexicon\n"
     ."  -T twl.dawg       DAWG file containing TWL lexicon\n";
  }

&Getopts('1C:O:N:T:i') || &Usage;

$^I = 1 if $opt_i;

&dawg'open(*TWL, $opt_T) || die "Can't open $opt_T\n";
&dawg'open(*OSW, $opt_O) || die "Can't open $opt_O\n";
&dawg'open(*TCD, $opt_C) || die "Can't open $opt_C\n";
&dawg'open(*NTCD, $opt_N) || die "Can't open $opt_N\n";

while (<>) {
  while (length) {
    s/^\s+// && print $&;
    if (s/^\S+//) {
      $word = $&;
      $pre = $post = $tag = '';
      $word =~ s/^[([{'"`]+// && ($pre = $&);
      $word =~ s/[]}).,:;"'!?]+$// && ($post = $&);
      if (length($word) >= 2 && $word !~ /[^a-zA-Z]/) {
l1:     {
        if ($opt_1)
          { if (defined $seen{$word}) { last l1; } else { $seen{$word} = 1; } }
        $lc = "\L$word";
        if (&dawg'check(*TWL, $lc)) {
          unless (&dawg'check(*OSW, $lc) || &dawg'check(*TCD, $lc)) {
            $tag = (length($word) <= 9 || &dawg'check(*NTCD, $lc))
              ? '<$>' : '<?>';
            }
          }
        elsif (&dawg'check(*OSW, $lc) || &dawg'check(*TCD, $lc))
          { $tag = '<#>'; }
        elsif (length($word) <= 9 || &dawg'check(*NTCD, $lc)) { $tag = '<*>'; }
        else { $tag = '<?>'; }
        } # l1
        }
      print "$pre$word$tag$post";
      }
    }
  if (eof(ARGV)) {
    print "--\n";
    print "This message has been automatically tagged by ddtag $gVersion, a Per
l script\n";
    print "that tags words according to the lexica in which they are found:\n";
    print "  <\$> TWL only  <#> OSW only  <*> neither TWL nor OSW\n";
    print "ddtag is available at http://sushi.st.usm.edu/~jjchew/software/ddtag
.pl.\n";
    }
  }

&dawg'close(*TCD);
&dawg'close(*OSW);
&dawg'close(*TWL);

0;