Commit | Line | Data |
5271a011 |
1 | package Text::Tradition::Language::Latin; |
2 | |
3 | use strict; |
4 | use warnings; |
5 | use Module::Load; |
0ce8c0cf |
6 | use parent qw/ Text::Tradition::Language::Perseus /; |
5271a011 |
7 | |
8 | =head1 NAME |
9 | |
10 | Text::Tradition::Language::Latin - language-specific module for Latin |
11 | |
12 | =head1 DESCRIPTION |
13 | |
0ce8c0cf |
14 | Implements morphology lookup for Latin words in context. This module |
f8862b58 |
15 | depends on the Lingua::Morph::Perseus module for access to PhiloLogic database data. |
5271a011 |
16 | |
17 | =head1 SUBROUTINES |
18 | |
19 | =head2 lemmatize( $text ) |
20 | |
0ce8c0cf |
21 | Evaluates the string using Treetagger and Perseus, and returns the results. |
5271a011 |
22 | |
23 | =begin testing |
24 | |
25 | use Text::Tradition; |
26 | use_ok( 'Text::Tradition::Language::Latin' ); |
27 | |
f8862b58 |
28 | eval "use Lingua::Morph::Perseus"; |
5271a011 |
29 | my $err = $@; |
30 | |
31 | SKIP: { |
f8862b58 |
32 | skip "Package Lingua::Morph::Perseus not found" if $err; |
5271a011 |
33 | |
34 | my $trad = Text::Tradition->new( |
35 | 'language' => 'Latin', |
36 | 'file' => 't/data/legendfrag.xml', |
37 | 'input' => 'Self' ); |
38 | $trad->lemmatize(); |
39 | my $ambig = 0; |
40 | foreach my $r ( $trad->collation->readings ) { |
41 | next if $r->is_meta; |
42 | ok( $r->has_lexemes, "Reading $r has one or more lexemes" ); |
43 | my @lex = $r->lexemes; |
44 | my $lexstr = join( '', map { $_->string } @lex ); |
45 | my $textstr = $r->text; |
46 | $textstr =~ s/\s+//g; |
47 | is( $textstr, $lexstr, "Lexemes for reading $r match the reading" ); |
48 | foreach my $l ( @lex ) { |
fe77efe0 |
49 | next unless $l->matches; |
5271a011 |
50 | next if $l->is_disambiguated; |
51 | printf( "Ambiguous lexeme %s for reading %s:\n\t%s\n", $l->string, $r->id, |
52 | join( "\n\t", map { $_->lemma . ': ' . $_->morphology->to_string } $l->matching_forms ) ); |
53 | $ambig++; |
54 | } |
55 | } |
f8862b58 |
56 | is( $ambig, 4, "Found 4 ambiguous forms as expected" ); |
5271a011 |
57 | } |
58 | |
59 | =end testing |
60 | |
61 | =cut |
62 | |
0ce8c0cf |
63 | our $dbhandle; |
64 | |
5271a011 |
65 | sub lemmatize { |
0ce8c0cf |
66 | return __PACKAGE__->perseus_lemmatize( @_ ); |
5271a011 |
67 | } |
68 | |
5271a011 |
69 | sub reading_lookup { |
0ce8c0cf |
70 | return __PACKAGE__->perseus_reading_lookup( @_ ); |
75ae2b25 |
71 | } |
72 | |
227d4a11 |
73 | 1; |
e92d4229 |
74 | |