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; |
cf7e4e7b |
9 | use Text::Tradition::User; |
dd3b58b0 |
10 | |
331c2dbf |
11 | use vars qw( $VERSION ); |
f242e74d |
12 | $VERSION = "0.5"; |
331c2dbf |
13 | |
dd3b58b0 |
14 | has 'collation' => ( |
8e1394aa |
15 | is => 'ro', |
16 | isa => 'Text::Tradition::Collation', |
17 | writer => '_save_collation', |
18 | ); |
dd3b58b0 |
19 | |
3b853983 |
20 | has 'witness_hash' => ( |
21 | traits => ['Hash'], |
22 | isa => 'HashRef[Text::Tradition::Witness]', |
8e1394aa |
23 | handles => { |
3b853983 |
24 | witness => 'get', |
25 | add_witness => 'set', |
26 | del_witness => 'delete', |
27 | has_witness => 'exists', |
28 | witnesses => 'values', |
8e1394aa |
29 | }, |
3b853983 |
30 | default => sub { {} }, |
8e1394aa |
31 | ); |
c5104dc0 |
32 | |
df6d9812 |
33 | has 'name' => ( |
34 | is => 'rw', |
35 | isa => 'Str', |
36 | default => 'Tradition', |
37 | ); |
56cf65bd |
38 | |
55bc8c78 |
39 | has 'language' => ( |
bbd064a9 |
40 | is => 'rw', |
55bc8c78 |
41 | isa => 'Str', |
fae52efd |
42 | predicate => 'has_language', |
55bc8c78 |
43 | ); |
44 | |
e0d617e6 |
45 | has 'stemmata' => ( |
46 | traits => ['Array'], |
47 | isa => 'ArrayRef[Text::Tradition::Stemma]', |
48 | handles => { |
7f52eac8 |
49 | stemmata => 'elements', |
e0d617e6 |
50 | _add_stemma => 'push', |
51 | stemma => 'get', |
52 | stemma_count => 'count', |
53 | clear_stemmata => 'clear', |
54 | }, |
c3c79612 |
55 | default => sub { [] }, |
56cf65bd |
56 | ); |
f54b1ba7 |
57 | |
58 | has 'user' => ( |
59 | is => 'rw', |
60 | isa => 'Text::Tradition::User', |
61 | required => 0, |
62 | predicate => 'has_user', |
ec7ea4e6 |
63 | clearer => 'clear_user', |
5b3c1961 |
64 | weak_ref => 1 |
f54b1ba7 |
65 | ); |
3724dfa7 |
66 | |
67 | has 'public' => ( |
68 | is => 'rw', |
69 | isa => 'Bool', |
70 | required => 0, |
71 | default => sub { 0; }, |
72 | ); |
73 | |
3b853983 |
74 | # Create the witness before trying to add it |
910a0a6d |
75 | around 'add_witness' => sub { |
76 | my $orig = shift; |
77 | my $self = shift; |
331c2dbf |
78 | # TODO allow add of a Witness object? |
fae52efd |
79 | my %args = @_ == 1 ? %{$_[0]} : @_; |
80 | $args{'tradition'} = $self; |
81 | $args{'language'} = $self->language |
82 | if( $self->language && !exists $args{'language'} ); |
83 | my $new_wit = Text::Tradition::Witness->new( %args ); |
3b853983 |
84 | $self->$orig( $new_wit->sigil => $new_wit ); |
910a0a6d |
85 | return $new_wit; |
86 | }; |
331c2dbf |
87 | |
3b853983 |
88 | # Allow deletion of witness by object as well as by sigil |
89 | around 'del_witness' => sub { |
90 | my $orig = shift; |
91 | my $self = shift; |
92 | my @key_args; |
93 | foreach my $arg ( @_ ) { |
94 | push( @key_args, |
95 | ref( $arg ) eq 'Text::Tradition::Witness' ? $arg->sigil : $arg ); |
96 | } |
97 | return $self->$orig( @key_args ); |
98 | }; |
99 | |
100 | # Don't allow an empty hash value |
101 | around 'witness' => sub { |
102 | my( $orig, $self, $arg ) = @_; |
103 | return unless $self->has_witness( $arg ); |
104 | return $self->$orig( $arg ); |
105 | }; |
106 | |
331c2dbf |
107 | =head1 NAME |
108 | |
109 | Text::Tradition - a software model for a set of collated texts |
110 | |
111 | =head1 SYNOPSIS |
112 | |
113 | use Text::Tradition; |
114 | my $t = Text::Tradition->new( |
115 | 'name' => 'this is a text', |
116 | 'input' => 'TEI', |
117 | 'file' => '/path/to/tei_parallel_seg_file.xml' ); |
118 | |
119 | my @text_wits = $t->witnesses(); |
120 | my $manuscript_a = $t->witness( 'A' ); |
82fa4d57 |
121 | |
122 | $t = Text::Tradition->new(); |
123 | $t->add_witness( 'sourcetype' => 'xmldesc', |
124 | 'file' => '/path/to/teitranscription.xml' ); |
125 | $t->add_witness( 'sourcetype => 'plaintext', 'sigil' => 'Q', |
126 | 'string' => 'The quick brown fox jumped over the lazy dogs' ); |
127 | ## TODO |
128 | $t->collate_texts; |
331c2dbf |
129 | |
130 | my $text_path_svg = $t->collation->as_svg(); |
131 | ## See Text::Tradition::Collation for more on text collation itself |
132 | |
133 | =head1 DESCRIPTION |
134 | |
135 | Text::Tradition is a library for representation and analysis of collated |
136 | texts, particularly medieval ones. A 'tradition' refers to the aggregation |
137 | of surviving versions of a text, generally preserved in multiple |
138 | manuscripts (or 'witnesses'). A Tradition object thus has one more more |
139 | Witnesses, as well as a Collation that represents the unity of all versions |
140 | of the text. |
141 | |
142 | =head1 METHODS |
143 | |
144 | =head2 new |
145 | |
146 | Creates and returns a new text tradition object. The following options are |
147 | accepted. |
148 | |
149 | General options: |
150 | |
151 | =over 4 |
152 | |
153 | =item B<name> - The name of the text. |
154 | |
155 | =back |
156 | |
157 | Initialization based on a collation file: |
158 | |
159 | =over 4 |
160 | |
161 | =item B<input> - The input format of the collation file. Can be one of the |
162 | following: |
163 | |
164 | =over 4 |
165 | |
166 | =item * Self - a GraphML format produced by this module |
167 | |
168 | =item * CollateX - a GraphML format produced by CollateX |
169 | |
170 | =item * CTE - a TEI XML format produced by Classical Text Editor |
171 | |
a731e73a |
172 | =item * JSON - an alignment table in JSON format, as produced by CollateX and other tools |
173 | |
331c2dbf |
174 | =item * KUL - a specific CSV format for variants, not documented here |
175 | |
176 | =item * TEI - a TEI parallel segmentation format file |
177 | |
178 | =item * Tabular - a comma- or tab-separated collation. Takes an additional |
179 | option, 'sep_char', which defaults to the tab character. |
180 | |
181 | =back |
182 | |
183 | =item B<file> - The name of the file that contains the data. One of 'file' |
184 | or 'string' should be specified. |
185 | |
186 | =item B<string> - A text string that contains the data. One of 'file' or |
187 | 'string' should be specified. |
188 | |
189 | =item B<base> - The name of a text file that contains the base text, to be |
190 | used with input formats that require it (currently only KUL). |
191 | |
192 | =back |
193 | |
194 | Initialization based on a list of witnesses [NOT YET IMPLEMENTED]: |
195 | |
196 | =over 4 |
197 | |
198 | =item B<witnesses> - A reference to an array of Text::Tradition::Witness |
199 | objects that carry the text to be collated. |
200 | |
201 | =item B<collator> - A reference to a collation program that will accept |
202 | Witness objects. |
203 | |
204 | =back |
205 | |
206 | =head2 B<witnesses> |
207 | |
208 | Return the Text::Tradition::Witness objects associated with this tradition, |
209 | as an array. |
210 | |
044d1e45 |
211 | =head2 B<witness>( $sigil ) |
212 | |
213 | Returns the Text::Tradition::Witness object whose sigil is $sigil, or undef |
214 | if there is no such object within the tradition. |
215 | |
331c2dbf |
216 | =head2 B<add_witness>( %opts ) |
217 | |
218 | Instantiate a new witness with the given options (see documentation for |
219 | Text::Tradition::Witness) and add it to the tradition. |
220 | |
044d1e45 |
221 | =head2 B<del_witness>( $sigil ) |
222 | |
223 | Delete the witness with the given sigil from the tradition. Returns the |
224 | witness object for the deleted witness. |
225 | |
331c2dbf |
226 | =begin testing |
227 | |
228 | use_ok( 'Text::Tradition', "can use module" ); |
229 | |
230 | my $t = Text::Tradition->new( 'name' => 'empty' ); |
231 | is( ref( $t ), 'Text::Tradition', "initialized an empty Tradition object" ); |
232 | is( $t->name, 'empty', "object has the right name" ); |
233 | is( scalar $t->witnesses, 0, "object has no witnesses" ); |
234 | |
235 | my $simple = 't/data/simple.txt'; |
236 | my $s = Text::Tradition->new( |
237 | 'name' => 'inline', |
238 | 'input' => 'Tabular', |
239 | 'file' => $simple, |
240 | ); |
241 | is( ref( $s ), 'Text::Tradition', "initialized a Tradition object" ); |
242 | is( $s->name, 'inline', "object has the right name" ); |
243 | is( scalar $s->witnesses, 3, "object has three witnesses" ); |
244 | |
044d1e45 |
245 | my $wit_a = $s->witness('A'); |
246 | is( ref( $wit_a ), 'Text::Tradition::Witness', "Found a witness A" ); |
247 | if( $wit_a ) { |
248 | is( $wit_a->sigil, 'A', "Witness A has the right sigil" ); |
249 | } |
250 | is( $s->witness('X'), undef, "There is no witness X" ); |
251 | ok( !exists $s->{'witnesses'}->{'X'}, "Witness key X not created" ); |
252 | |
82fa4d57 |
253 | my $wit_d = $s->add_witness( 'sigil' => 'D', 'sourcetype' => 'collation' ); |
044d1e45 |
254 | is( ref( $wit_d ), 'Text::Tradition::Witness', "new witness created" ); |
255 | is( $wit_d->sigil, 'D', "witness has correct sigil" ); |
331c2dbf |
256 | is( scalar $s->witnesses, 4, "object now has four witnesses" ); |
257 | |
3b853983 |
258 | my $del = $s->del_witness( 'D' ); |
044d1e45 |
259 | is( $del, $wit_d, "Deleted correct witness" ); |
3b853983 |
260 | is( scalar $s->witnesses, 3, "object has three witnesses again" ); |
261 | |
331c2dbf |
262 | # TODO test initialization by witness list when we have it |
263 | |
264 | =end testing |
265 | |
266 | =cut |
910a0a6d |
267 | |
df6d9812 |
268 | |
8e1394aa |
269 | sub BUILD { |
270 | my( $self, $init_args ) = @_; |
fae52efd |
271 | |
272 | # First, make a collation object. This will use only those arguments in |
273 | # init_args that apply to the collation. |
274 | my $collation = Text::Tradition::Collation->new( %$init_args, |
275 | 'tradition' => $self ); |
276 | $self->_save_collation( $collation ); |
c5104dc0 |
277 | |
fae52efd |
278 | if( exists $init_args->{'input'} ) { |
910a0a6d |
279 | # Call the appropriate parser on the given data |
a731e73a |
280 | my @format_standalone = qw/ Self CollateText CollateX CTE JSON TEI Tabular /; |
dfc37e38 |
281 | my @format_basetext = qw/ KUL /; |
282 | my $use_base; |
283 | my $format = $init_args->{'input'}; |
dfc37e38 |
284 | if( $format && !( grep { $_ eq $format } @format_standalone ) |
285 | && !( grep { $_ eq $format } @format_basetext ) ) { |
286 | warn "Unrecognized input format $format; not parsing"; |
910a0a6d |
287 | return; |
288 | } |
dfc37e38 |
289 | if( $format && grep { $_ eq $format } @format_basetext ) { |
290 | $use_base = 1; |
291 | if( !exists $init_args->{'base'} ) { |
292 | warn "Cannot make a collation from $format without a base text"; |
293 | return; |
294 | } |
295 | } |
4a8828f0 |
296 | |
910a0a6d |
297 | # Now do the parsing. |
910a0a6d |
298 | if( $format ) { |
dfc37e38 |
299 | if( $use_base ) { |
300 | $format = 'BaseText'; # Use the BaseText module for parsing, |
301 | # but retain the original input arg. |
910a0a6d |
302 | } |
303 | my $mod = "Text::Tradition::Parser::$format"; |
304 | load( $mod ); |
dfc37e38 |
305 | $mod->can('parse')->( $self, $init_args ); |
910a0a6d |
306 | } |
c5104dc0 |
307 | } |
fae52efd |
308 | return $self; |
309 | } |
310 | |
311 | =head2 add_json_witnesses( $jsonstring, $options ) |
312 | |
313 | Adds a set of witnesses from a JSON array specification. This is a wrapper |
314 | to parse the JSON and call add_witness (with the specified $options) for |
315 | each element therein. |
316 | |
317 | =cut |
318 | |
319 | sub add_json_witnesses { |
320 | my( $self, $jsonstr, $extraopts ) = @_; |
65ed66b9 |
321 | my $witarray = from_json( $jsonstr ); |
322 | foreach my $witspec ( @{$witarray->{witnesses}} ) { |
fae52efd |
323 | my $opts = $extraopts || {}; |
324 | $opts->{'sourcetype'} = 'json'; |
325 | $opts->{'object'} = $witspec; |
326 | $self->add_witness( $opts ); |
327 | } |
8e1394aa |
328 | } |
c5104dc0 |
329 | |
56cf65bd |
330 | =head2 add_stemma( $dotfile ) |
331 | |
332 | Initializes a Text::Tradition::Stemma object from the given dotfile, |
333 | and associates it with the tradition. |
334 | |
335 | =begin testing |
336 | |
337 | use Text::Tradition; |
338 | |
339 | my $t = Text::Tradition->new( |
340 | 'name' => 'simple test', |
341 | 'input' => 'Tabular', |
342 | 'file' => 't/data/simple.txt', |
343 | ); |
344 | |
c3c79612 |
345 | is( $t->stemma_count, 0, "No stemmas added yet" ); |
56cf65bd |
346 | my $s; |
9ba651b9 |
347 | ok( $s = $t->add_stemma( dotfile => 't/data/simple.dot' ), "Added a simple stemma" ); |
56cf65bd |
348 | is( ref( $s ), 'Text::Tradition::Stemma', "Got a stemma object returned" ); |
e0d617e6 |
349 | is( $t->stemma_count, 1, "Tradition claims to have a stemma" ); |
350 | is( $t->stemma(0), $s, "Tradition hands back the right stemma" ); |
56cf65bd |
351 | |
352 | =end testing |
353 | |
354 | =cut |
355 | |
356 | sub add_stemma { |
bffafb73 |
357 | my $self = shift; |
358 | my %opts = @_; |
359 | my $stemma_fh; |
360 | if( $opts{'dotfile'} ) { |
361 | open $stemma_fh, '<', $opts{'dotfile'} |
362 | or warn "Could not open file " . $opts{'dotfile'}; |
363 | } elsif( $opts{'dot'} ) { |
364 | my $str = $opts{'dot'}; |
365 | open $stemma_fh, '<', \$str; |
366 | } |
367 | # Assume utf-8 |
368 | binmode $stemma_fh, ':utf8'; |
56cf65bd |
369 | my $stemma = Text::Tradition::Stemma->new( |
370 | 'collation' => $self->collation, |
371 | 'dot' => $stemma_fh ); |
372 | $self->_add_stemma( $stemma ) if $stemma; |
373 | return $stemma; |
374 | } |
375 | |
d3e7842a |
376 | sub lemmatize { |
377 | my $self = shift; |
378 | unless( $self->has_language ) { |
379 | warn "Please set a language to lemmatize a tradition"; |
380 | return; |
381 | } |
382 | my $mod = "Text::Tradition::Language::" . $self->language; |
383 | load( $mod ); |
384 | $mod->can( 'lemmatize' )->( $self ); |
385 | } |
386 | |
dd3b58b0 |
387 | no Moose; |
388 | __PACKAGE__->meta->make_immutable; |
331c2dbf |
389 | |
390 | |
391 | =head1 BUGS / TODO |
392 | |
393 | =over |
394 | |
395 | =item * Allow tradition to be initialized via passing to a collator. |
396 | |
397 | =back |
398 | |
399 | =head1 LICENSE |
400 | |
401 | This package is free software and is provided "as is" without express |
402 | or implied warranty. You can redistribute it and/or modify it under |
403 | the same terms as Perl itself. |
404 | |
405 | =head1 AUTHOR |
406 | |
407 | Tara L Andrews E<lt>aurum@cpan.orgE<gt> |