5 #no warnings 'uninitialized';
7 # Usage: add-affixes [<inc_level>] [<flags>]
10 if ($ARGV[0] =~ /^[\d.]+$/) {
11 $inc_level = $ARGV[0];
15 #print STDERR "Include Level: $inc_level\n";
18 my $just_possessive = 0;
19 my $no_possessive = 0;
21 if ($_ eq 'use-all') {$use_all = 2}
22 elsif ($_ eq 'use-some') {$use_all = 1}
23 elsif ($_ eq 'just-possessive') {$just_possessive = 1}
24 elsif ($_ eq 'no-possessive') {$no_possessive = 1}
25 else {die "ERROR: Invalid flag $_\n";}
28 open F, "r/alt12dicts/2of12id.txt" or die;
36 # (flags, base word, part of speach, infl forms)
37 my ($d,$w,$p,$a) = /^([-@]?)(\w+) (.).*: ?(.*)$/ or die;
38 $possessive{$w} = "$w\'s\n" if $p eq 'N' && ($d eq '' || $use_all);
40 my @a = $a =~ /([-~@\w]+)/g;
41 @a = map {"$d$_"} @a if ($d);
43 foreach (@a) {if (s/^[~-]//) {push @a0, $_} else {push @a1, $_}}
44 $remove{"$w:$p"} = 1 unless @a1;
45 foreach (@a0) {$remove{"$w:$p:$_"} = 1}
48 # Maybe using AGID isn't a good idea here, many of the entries in AGID
49 # are unchecked, for example "associationism" may be an uncountable
50 # noun, but since the base entry is not in 2of12id it is not flagged
51 # that way, thus the plural "associationisms" gets included. However,
52 # AGID still needs to be used for uppercase words since they are not
53 # in 2of12id. For now I won't worry about it since it primary effects
56 open F, "r/infl/infl.txt" or die;
59 # (base word, part of speach, guess flag, infl forms)
60 my ($w,$p,$q,$a) = /(\S+) (.)(.*): (.+)/ or die;
62 # AGID things it is a noun and "use-some" or Uppercase
63 # (since 2of12id doesn't include uppercase)
64 # AGIG is guessing it is a noun and "use-all"
65 $possessive{$w} = "$w\'s\n" if $p eq 'N' && (($q eq '' && ($use_all || $w =~ /^[A-Z]/))
67 next if $remove{"$w:$p"};
68 next unless $q eq '' || $use_all >= 2;
69 my @a = split /, | \| /, $a;
70 @a = grep {my ($word,$tags,$level)
71 = /^([A-Za-z\']+)([~<!?]*)(| [\d.]+)(| {\S+})$/ or die $_;
73 $tags !~ /~|\?|!</ && $level <= $inc_level} @a;
74 @a = grep {not $remove{"$w:$p:$_"}} @a;
76 $lookup{$w} .= join("\n",@a)."\n";
79 unless ($no_possessive) {
81 open F, "r/special/not-possessive" or die;
85 delete $possessive{$_};
88 open F, "working/possessive-also.lst" or die;
92 $possessive{$_} = "$_\'s\n";
99 print $lookup{$_} unless $just_possessive;
100 print $possessive{$_} unless $no_possessive;