Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition; |
2 | |
4a8828f0 |
3 | use Module::Load; |
dd3b58b0 |
4 | use Moose; |
8e1394aa |
5 | use Text::Tradition::Collation; |
6 | use Text::Tradition::Witness; |
dd3b58b0 |
7 | |
8 | has 'collation' => ( |
8e1394aa |
9 | is => 'ro', |
10 | isa => 'Text::Tradition::Collation', |
11 | writer => '_save_collation', |
12 | ); |
dd3b58b0 |
13 | |
14 | has 'witnesses' => ( |
8e1394aa |
15 | traits => ['Array'], |
16 | is => 'rw', |
17 | isa => 'ArrayRef[Text::Tradition::Witness]', |
18 | handles => { |
4a8828f0 |
19 | all => 'elements', |
20 | add => 'push', |
8e1394aa |
21 | }, |
4a8828f0 |
22 | default => sub { [] }, |
8e1394aa |
23 | ); |
c5104dc0 |
24 | |
8e1394aa |
25 | sub BUILD { |
26 | my( $self, $init_args ) = @_; |
c5104dc0 |
27 | |
8e1394aa |
28 | if( exists $init_args->{'witnesses'} ) { |
c5104dc0 |
29 | # We got passed an uncollated list of witnesses. Make a |
30 | # witness object for each witness, and then send them to the |
31 | # collator. |
32 | my $autosigil = 0; |
8e1394aa |
33 | foreach my $wit ( %{$init_args->{'witnesses'}} ) { |
c5104dc0 |
34 | # Each item in the list is either a string or an arrayref. |
35 | # If it's a string, it is a filename; if it's an arrayref, |
36 | # it is a tuple of 'sigil, file'. Handle either case. |
37 | my $args; |
38 | if( ref( $wit ) eq 'ARRAY' ) { |
39 | $args = { 'sigil' => $wit->[0], |
40 | 'file' => $wit->[1] }; |
41 | } else { |
42 | $args = { 'sigil' => chr( $autosigil+65 ), |
43 | 'file' => $wit }; |
44 | $autosigil++; |
45 | } |
8e1394aa |
46 | $self->witnesses->push( Text::Tradition::Witness->new( $args ) ); |
47 | # TODO Now how to collate these? |
c5104dc0 |
48 | } |
49 | } else { |
4a8828f0 |
50 | # Else we need to parse some collation data. Make a Collation object |
51 | my $collation = Text::Tradition::Collation->new( %$init_args, |
52 | 'tradition' => $self ); |
53 | $self->_save_collation( $collation ); |
54 | |
55 | # Call the appropriate parser on the given data |
56 | my @formats = grep { /^(GraphML|CSV|CTE|TEI)$/ } keys( %$init_args ); |
57 | my $format = shift( @formats ); |
58 | unless( $format ) { |
59 | warn "No data given to create a collation; will initialize an empty one"; |
60 | } |
61 | if( $format && $format =~ /^(CSV|CTE)$/ && |
62 | !exists $init_args->{'base'} ) { |
63 | warn "Cannot make a collation from $format without a base text"; |
64 | return; |
65 | } |
66 | |
67 | # Starting point for all texts |
68 | my $last_node = $collation->add_reading( '#START#' ); |
69 | |
70 | # Now do the parsing. |
71 | my @sigla; |
72 | if( $format ) { |
73 | my @parseargs; |
74 | if( $format =~ /^(CSV|CTE)$/ ) { |
75 | @parseargs = ( 'base' => $init_args->{'base'}, |
76 | 'data' => $init_args->{$format}, |
77 | 'format' => $format ); |
78 | $format = 'BaseText'; |
79 | } else { |
80 | @parseargs = ( $init_args->{ $format } ); |
81 | } |
82 | my $mod = "Text::Tradition::Parser::$format"; |
83 | load( $mod ); |
84 | $mod->can('parse')->( $self, @parseargs ); |
85 | } |
c5104dc0 |
86 | } |
8e1394aa |
87 | } |
c5104dc0 |
88 | |
de51424a |
89 | sub witness { |
90 | my( $self, $sigil ) = @_; |
91 | my $requested_wit; |
92 | foreach my $wit ( @{$self->witnesses} ) { |
93 | $requested_wit = $wit if $wit->sigil eq $sigil; |
94 | } |
e2902068 |
95 | # We depend on an undef return value for no such witness. |
96 | # warn "No such witness $sigil" unless $requested_wit; |
de51424a |
97 | return $requested_wit; |
98 | } |
99 | |
100 | |
4a8828f0 |
101 | sub add_witness { |
102 | my $self = shift; |
103 | my $new_wit = Text::Tradition::Witness->new( @_ ); |
104 | push( @{$self->witnesses}, $new_wit ); |
e2902068 |
105 | return $new_wit; |
4a8828f0 |
106 | } |
107 | |
dd3b58b0 |
108 | # The user will usually be instantiating a Tradition object, and |
109 | # examining its collation. The information about the tradition can |
110 | # come via several routes: |
111 | # - graphML from CollateX or elsewhere, standalone |
112 | # - TEI parallel segmentation |
113 | # - Leuven-style spreadsheet of variants, converted to CSV, plus base text |
114 | # - apparatus pulled from CTE, plus base text |
115 | # From this we should be able to get basic witness information. |
116 | # |
117 | # Alternatively the user can just give us the uncollated texts. Then |
118 | # instead of passing a collation, s/he is passing a set of witnesses |
119 | # from which we will generate a collation. Those witnesses can be in |
120 | # plaintext or in TEI with certain constraints adopted. |
121 | |
122 | # So the constructor for a tradition needs to take one of these infosets, |
123 | # and construct the collation and the witness objects. |
124 | |
125 | no Moose; |
126 | __PACKAGE__->meta->make_immutable; |