Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition; |
2 | |
65ed66b9 |
3 | use JSON qw / from_json /; |
4a8828f0 |
4 | use Module::Load; |
dd3b58b0 |
5 | use Moose; |
8e1394aa |
6 | use Text::Tradition::Collation; |
56cf65bd |
7 | use Text::Tradition::Stemma; |
8e1394aa |
8 | use Text::Tradition::Witness; |
dd3b58b0 |
9 | |
331c2dbf |
10 | use vars qw( $VERSION ); |
f242e74d |
11 | $VERSION = "0.5"; |
331c2dbf |
12 | |
dd3b58b0 |
13 | has 'collation' => ( |
8e1394aa |
14 | is => 'ro', |
15 | isa => 'Text::Tradition::Collation', |
16 | writer => '_save_collation', |
17 | ); |
dd3b58b0 |
18 | |
3b853983 |
19 | has 'witness_hash' => ( |
20 | traits => ['Hash'], |
21 | isa => 'HashRef[Text::Tradition::Witness]', |
8e1394aa |
22 | handles => { |
3b853983 |
23 | witness => 'get', |
24 | add_witness => 'set', |
25 | del_witness => 'delete', |
26 | has_witness => 'exists', |
27 | witnesses => 'values', |
8e1394aa |
28 | }, |
3b853983 |
29 | default => sub { {} }, |
8e1394aa |
30 | ); |
c5104dc0 |
31 | |
df6d9812 |
32 | has 'name' => ( |
33 | is => 'rw', |
34 | isa => 'Str', |
35 | default => 'Tradition', |
36 | ); |
56cf65bd |
37 | |
55bc8c78 |
38 | has 'language' => ( |
bbd064a9 |
39 | is => 'rw', |
55bc8c78 |
40 | isa => 'Str', |
fae52efd |
41 | predicate => 'has_language', |
55bc8c78 |
42 | ); |
43 | |
e0d617e6 |
44 | has 'stemmata' => ( |
45 | traits => ['Array'], |
46 | isa => 'ArrayRef[Text::Tradition::Stemma]', |
47 | handles => { |
7f52eac8 |
48 | stemmata => 'elements', |
e0d617e6 |
49 | _add_stemma => 'push', |
50 | stemma => 'get', |
51 | stemma_count => 'count', |
52 | clear_stemmata => 'clear', |
53 | }, |
c3c79612 |
54 | default => sub { [] }, |
56cf65bd |
55 | ); |
3b853983 |
56 | |
57 | # Create the witness before trying to add it |
910a0a6d |
58 | around 'add_witness' => sub { |
59 | my $orig = shift; |
60 | my $self = shift; |
331c2dbf |
61 | # TODO allow add of a Witness object? |
fae52efd |
62 | my %args = @_ == 1 ? %{$_[0]} : @_; |
63 | $args{'tradition'} = $self; |
64 | $args{'language'} = $self->language |
65 | if( $self->language && !exists $args{'language'} ); |
66 | my $new_wit = Text::Tradition::Witness->new( %args ); |
3b853983 |
67 | $self->$orig( $new_wit->sigil => $new_wit ); |
910a0a6d |
68 | return $new_wit; |
69 | }; |
331c2dbf |
70 | |
3b853983 |
71 | # Allow deletion of witness by object as well as by sigil |
72 | around 'del_witness' => sub { |
73 | my $orig = shift; |
74 | my $self = shift; |
75 | my @key_args; |
76 | foreach my $arg ( @_ ) { |
77 | push( @key_args, |
78 | ref( $arg ) eq 'Text::Tradition::Witness' ? $arg->sigil : $arg ); |
79 | } |
80 | return $self->$orig( @key_args ); |
81 | }; |
82 | |
83 | # Don't allow an empty hash value |
84 | around 'witness' => sub { |
85 | my( $orig, $self, $arg ) = @_; |
86 | return unless $self->has_witness( $arg ); |
87 | return $self->$orig( $arg ); |
88 | }; |
89 | |
331c2dbf |
90 | =head1 NAME |
91 | |
92 | Text::Tradition - a software model for a set of collated texts |
93 | |
94 | =head1 SYNOPSIS |
95 | |
96 | use Text::Tradition; |
97 | my $t = Text::Tradition->new( |
98 | 'name' => 'this is a text', |
99 | 'input' => 'TEI', |
100 | 'file' => '/path/to/tei_parallel_seg_file.xml' ); |
101 | |
102 | my @text_wits = $t->witnesses(); |
103 | my $manuscript_a = $t->witness( 'A' ); |
82fa4d57 |
104 | |
105 | $t = Text::Tradition->new(); |
106 | $t->add_witness( 'sourcetype' => 'xmldesc', |
107 | 'file' => '/path/to/teitranscription.xml' ); |
108 | $t->add_witness( 'sourcetype => 'plaintext', 'sigil' => 'Q', |
109 | 'string' => 'The quick brown fox jumped over the lazy dogs' ); |
110 | ## TODO |
111 | $t->collate_texts; |
331c2dbf |
112 | |
113 | my $text_path_svg = $t->collation->as_svg(); |
114 | ## See Text::Tradition::Collation for more on text collation itself |
115 | |
116 | =head1 DESCRIPTION |
117 | |
118 | Text::Tradition is a library for representation and analysis of collated |
119 | texts, particularly medieval ones. A 'tradition' refers to the aggregation |
120 | of surviving versions of a text, generally preserved in multiple |
121 | manuscripts (or 'witnesses'). A Tradition object thus has one more more |
122 | Witnesses, as well as a Collation that represents the unity of all versions |
123 | of the text. |
124 | |
125 | =head1 METHODS |
126 | |
127 | =head2 new |
128 | |
129 | Creates and returns a new text tradition object. The following options are |
130 | accepted. |
131 | |
132 | General options: |
133 | |
134 | =over 4 |
135 | |
136 | =item B<name> - The name of the text. |
137 | |
138 | =back |
139 | |
140 | Initialization based on a collation file: |
141 | |
142 | =over 4 |
143 | |
144 | =item B<input> - The input format of the collation file. Can be one of the |
145 | following: |
146 | |
147 | =over 4 |
148 | |
149 | =item * Self - a GraphML format produced by this module |
150 | |
151 | =item * CollateX - a GraphML format produced by CollateX |
152 | |
153 | =item * CTE - a TEI XML format produced by Classical Text Editor |
154 | |
a731e73a |
155 | =item * JSON - an alignment table in JSON format, as produced by CollateX and other tools |
156 | |
331c2dbf |
157 | =item * KUL - a specific CSV format for variants, not documented here |
158 | |
159 | =item * TEI - a TEI parallel segmentation format file |
160 | |
161 | =item * Tabular - a comma- or tab-separated collation. Takes an additional |
162 | option, 'sep_char', which defaults to the tab character. |
163 | |
164 | =back |
165 | |
166 | =item B<file> - The name of the file that contains the data. One of 'file' |
167 | or 'string' should be specified. |
168 | |
169 | =item B<string> - A text string that contains the data. One of 'file' or |
170 | 'string' should be specified. |
171 | |
172 | =item B<base> - The name of a text file that contains the base text, to be |
173 | used with input formats that require it (currently only KUL). |
174 | |
175 | =back |
176 | |
177 | Initialization based on a list of witnesses [NOT YET IMPLEMENTED]: |
178 | |
179 | =over 4 |
180 | |
181 | =item B<witnesses> - A reference to an array of Text::Tradition::Witness |
182 | objects that carry the text to be collated. |
183 | |
184 | =item B<collator> - A reference to a collation program that will accept |
185 | Witness objects. |
186 | |
187 | =back |
188 | |
189 | =head2 B<witnesses> |
190 | |
191 | Return the Text::Tradition::Witness objects associated with this tradition, |
192 | as an array. |
193 | |
044d1e45 |
194 | =head2 B<witness>( $sigil ) |
195 | |
196 | Returns the Text::Tradition::Witness object whose sigil is $sigil, or undef |
197 | if there is no such object within the tradition. |
198 | |
331c2dbf |
199 | =head2 B<add_witness>( %opts ) |
200 | |
201 | Instantiate a new witness with the given options (see documentation for |
202 | Text::Tradition::Witness) and add it to the tradition. |
203 | |
044d1e45 |
204 | =head2 B<del_witness>( $sigil ) |
205 | |
206 | Delete the witness with the given sigil from the tradition. Returns the |
207 | witness object for the deleted witness. |
208 | |
331c2dbf |
209 | =begin testing |
210 | |
211 | use_ok( 'Text::Tradition', "can use module" ); |
212 | |
213 | my $t = Text::Tradition->new( 'name' => 'empty' ); |
214 | is( ref( $t ), 'Text::Tradition', "initialized an empty Tradition object" ); |
215 | is( $t->name, 'empty', "object has the right name" ); |
216 | is( scalar $t->witnesses, 0, "object has no witnesses" ); |
217 | |
218 | my $simple = 't/data/simple.txt'; |
219 | my $s = Text::Tradition->new( |
220 | 'name' => 'inline', |
221 | 'input' => 'Tabular', |
222 | 'file' => $simple, |
223 | ); |
224 | is( ref( $s ), 'Text::Tradition', "initialized a Tradition object" ); |
225 | is( $s->name, 'inline', "object has the right name" ); |
226 | is( scalar $s->witnesses, 3, "object has three witnesses" ); |
227 | |
044d1e45 |
228 | my $wit_a = $s->witness('A'); |
229 | is( ref( $wit_a ), 'Text::Tradition::Witness', "Found a witness A" ); |
230 | if( $wit_a ) { |
231 | is( $wit_a->sigil, 'A', "Witness A has the right sigil" ); |
232 | } |
233 | is( $s->witness('X'), undef, "There is no witness X" ); |
234 | ok( !exists $s->{'witnesses'}->{'X'}, "Witness key X not created" ); |
235 | |
82fa4d57 |
236 | my $wit_d = $s->add_witness( 'sigil' => 'D', 'sourcetype' => 'collation' ); |
044d1e45 |
237 | is( ref( $wit_d ), 'Text::Tradition::Witness', "new witness created" ); |
238 | is( $wit_d->sigil, 'D', "witness has correct sigil" ); |
331c2dbf |
239 | is( scalar $s->witnesses, 4, "object now has four witnesses" ); |
240 | |
3b853983 |
241 | my $del = $s->del_witness( 'D' ); |
044d1e45 |
242 | is( $del, $wit_d, "Deleted correct witness" ); |
3b853983 |
243 | is( scalar $s->witnesses, 3, "object has three witnesses again" ); |
244 | |
331c2dbf |
245 | # TODO test initialization by witness list when we have it |
246 | |
247 | =end testing |
248 | |
249 | =cut |
910a0a6d |
250 | |
df6d9812 |
251 | |
8e1394aa |
252 | sub BUILD { |
253 | my( $self, $init_args ) = @_; |
fae52efd |
254 | |
255 | # First, make a collation object. This will use only those arguments in |
256 | # init_args that apply to the collation. |
257 | my $collation = Text::Tradition::Collation->new( %$init_args, |
258 | 'tradition' => $self ); |
259 | $self->_save_collation( $collation ); |
c5104dc0 |
260 | |
fae52efd |
261 | if( exists $init_args->{'input'} ) { |
910a0a6d |
262 | # Call the appropriate parser on the given data |
a731e73a |
263 | my @format_standalone = qw/ Self CollateText CollateX CTE JSON TEI Tabular /; |
dfc37e38 |
264 | my @format_basetext = qw/ KUL /; |
265 | my $use_base; |
266 | my $format = $init_args->{'input'}; |
dfc37e38 |
267 | if( $format && !( grep { $_ eq $format } @format_standalone ) |
268 | && !( grep { $_ eq $format } @format_basetext ) ) { |
269 | warn "Unrecognized input format $format; not parsing"; |
910a0a6d |
270 | return; |
271 | } |
dfc37e38 |
272 | if( $format && grep { $_ eq $format } @format_basetext ) { |
273 | $use_base = 1; |
274 | if( !exists $init_args->{'base'} ) { |
275 | warn "Cannot make a collation from $format without a base text"; |
276 | return; |
277 | } |
278 | } |
4a8828f0 |
279 | |
910a0a6d |
280 | # Now do the parsing. |
910a0a6d |
281 | if( $format ) { |
dfc37e38 |
282 | if( $use_base ) { |
283 | $format = 'BaseText'; # Use the BaseText module for parsing, |
284 | # but retain the original input arg. |
910a0a6d |
285 | } |
286 | my $mod = "Text::Tradition::Parser::$format"; |
287 | load( $mod ); |
dfc37e38 |
288 | $mod->can('parse')->( $self, $init_args ); |
910a0a6d |
289 | } |
c5104dc0 |
290 | } |
fae52efd |
291 | return $self; |
292 | } |
293 | |
294 | =head2 add_json_witnesses( $jsonstring, $options ) |
295 | |
296 | Adds a set of witnesses from a JSON array specification. This is a wrapper |
297 | to parse the JSON and call add_witness (with the specified $options) for |
298 | each element therein. |
299 | |
300 | =cut |
301 | |
302 | sub add_json_witnesses { |
303 | my( $self, $jsonstr, $extraopts ) = @_; |
65ed66b9 |
304 | my $witarray = from_json( $jsonstr ); |
305 | foreach my $witspec ( @{$witarray->{witnesses}} ) { |
fae52efd |
306 | my $opts = $extraopts || {}; |
307 | $opts->{'sourcetype'} = 'json'; |
308 | $opts->{'object'} = $witspec; |
309 | $self->add_witness( $opts ); |
310 | } |
8e1394aa |
311 | } |
c5104dc0 |
312 | |
56cf65bd |
313 | =head2 add_stemma( $dotfile ) |
314 | |
315 | Initializes a Text::Tradition::Stemma object from the given dotfile, |
316 | and associates it with the tradition. |
317 | |
318 | =begin testing |
319 | |
320 | use Text::Tradition; |
321 | |
322 | my $t = Text::Tradition->new( |
323 | 'name' => 'simple test', |
324 | 'input' => 'Tabular', |
325 | 'file' => 't/data/simple.txt', |
326 | ); |
327 | |
c3c79612 |
328 | is( $t->stemma_count, 0, "No stemmas added yet" ); |
56cf65bd |
329 | my $s; |
9ba651b9 |
330 | ok( $s = $t->add_stemma( dotfile => 't/data/simple.dot' ), "Added a simple stemma" ); |
56cf65bd |
331 | is( ref( $s ), 'Text::Tradition::Stemma', "Got a stemma object returned" ); |
e0d617e6 |
332 | is( $t->stemma_count, 1, "Tradition claims to have a stemma" ); |
333 | is( $t->stemma(0), $s, "Tradition hands back the right stemma" ); |
56cf65bd |
334 | |
335 | =end testing |
336 | |
337 | =cut |
338 | |
339 | sub add_stemma { |
bffafb73 |
340 | my $self = shift; |
341 | my %opts = @_; |
342 | my $stemma_fh; |
343 | if( $opts{'dotfile'} ) { |
344 | open $stemma_fh, '<', $opts{'dotfile'} |
345 | or warn "Could not open file " . $opts{'dotfile'}; |
346 | } elsif( $opts{'dot'} ) { |
347 | my $str = $opts{'dot'}; |
348 | open $stemma_fh, '<', \$str; |
349 | } |
350 | # Assume utf-8 |
351 | binmode $stemma_fh, ':utf8'; |
56cf65bd |
352 | my $stemma = Text::Tradition::Stemma->new( |
353 | 'collation' => $self->collation, |
354 | 'dot' => $stemma_fh ); |
355 | $self->_add_stemma( $stemma ) if $stemma; |
356 | return $stemma; |
357 | } |
358 | |
d3e7842a |
359 | sub lemmatize { |
360 | my $self = shift; |
361 | unless( $self->has_language ) { |
362 | warn "Please set a language to lemmatize a tradition"; |
363 | return; |
364 | } |
365 | my $mod = "Text::Tradition::Language::" . $self->language; |
366 | load( $mod ); |
367 | $mod->can( 'lemmatize' )->( $self ); |
368 | } |
369 | |
dd3b58b0 |
370 | no Moose; |
371 | __PACKAGE__->meta->make_immutable; |
331c2dbf |
372 | |
373 | |
374 | =head1 BUGS / TODO |
375 | |
376 | =over |
377 | |
378 | =item * Allow tradition to be initialized via passing to a collator. |
379 | |
380 | =back |
381 | |
382 | =head1 LICENSE |
383 | |
384 | This package is free software and is provided "as is" without express |
385 | or implied warranty. You can redistribute it and/or modify it under |
386 | the same terms as Perl itself. |
387 | |
388 | =head1 AUTHOR |
389 | |
390 | Tara L Andrews E<lt>aurum@cpan.orgE<gt> |