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; |
951ddfe8 |
6 | use Moose::Util qw/ does_role apply_all_roles /; |
8e1394aa |
7 | use Text::Tradition::Collation; |
951ddfe8 |
8 | use Text::Tradition::Error; |
8e1394aa |
9 | use Text::Tradition::Witness; |
cf7e4e7b |
10 | use Text::Tradition::User; |
951ddfe8 |
11 | use TryCatch; |
dd3b58b0 |
12 | |
331c2dbf |
13 | use vars qw( $VERSION ); |
37bf09f4 |
14 | $VERSION = "1.1"; |
15 | |
16 | # Enable plugin(s) if available |
17 | eval { with 'Text::Tradition::HasStemma'; }; |
ed5b9b70 |
18 | # Don't warn normally |
19 | # if( $@ ) { |
20 | # warn "Text::Tradition::Analysis not found. Disabling stemma analysis functionality"; |
21 | # }; |
e92d4229 |
22 | eval { with 'Text::Tradition::Language'; }; |
331c2dbf |
23 | |
dd3b58b0 |
24 | has 'collation' => ( |
8e1394aa |
25 | is => 'ro', |
26 | isa => 'Text::Tradition::Collation', |
27 | writer => '_save_collation', |
28 | ); |
dd3b58b0 |
29 | |
3b853983 |
30 | has 'witness_hash' => ( |
31 | traits => ['Hash'], |
32 | isa => 'HashRef[Text::Tradition::Witness]', |
8e1394aa |
33 | handles => { |
3b853983 |
34 | witness => 'get', |
35 | add_witness => 'set', |
36 | del_witness => 'delete', |
37 | has_witness => 'exists', |
38 | witnesses => 'values', |
8e1394aa |
39 | }, |
3b853983 |
40 | default => sub { {} }, |
8e1394aa |
41 | ); |
c5104dc0 |
42 | |
df6d9812 |
43 | has 'name' => ( |
44 | is => 'rw', |
45 | isa => 'Str', |
46 | default => 'Tradition', |
47 | ); |
56cf65bd |
48 | |
3579c22b |
49 | has '_initialized' => ( |
10943ab0 |
50 | is => 'ro', |
51 | isa => 'Bool', |
52 | default => undef, |
53 | writer => '_init_done', |
54 | ); |
55 | |
f54b1ba7 |
56 | has 'user' => ( |
57 | is => 'rw', |
58 | isa => 'Text::Tradition::User', |
59 | required => 0, |
60 | predicate => 'has_user', |
ec7ea4e6 |
61 | clearer => 'clear_user', |
d8a14401 |
62 | weak_ref => 1 |
f54b1ba7 |
63 | ); |
3724dfa7 |
64 | |
65 | has 'public' => ( |
66 | is => 'rw', |
67 | isa => 'Bool', |
68 | required => 0, |
69 | default => sub { 0; }, |
70 | ); |
71 | |
3b853983 |
72 | # Create the witness before trying to add it |
910a0a6d |
73 | around 'add_witness' => sub { |
74 | my $orig = shift; |
75 | my $self = shift; |
331c2dbf |
76 | # TODO allow add of a Witness object? |
fae52efd |
77 | my %args = @_ == 1 ? %{$_[0]} : @_; |
78 | $args{'tradition'} = $self; |
fae52efd |
79 | my $new_wit = Text::Tradition::Witness->new( %args ); |
3b853983 |
80 | $self->$orig( $new_wit->sigil => $new_wit ); |
910a0a6d |
81 | return $new_wit; |
82 | }; |
331c2dbf |
83 | |
3b853983 |
84 | # Allow deletion of witness by object as well as by sigil |
85 | around 'del_witness' => sub { |
86 | my $orig = shift; |
87 | my $self = shift; |
88 | my @key_args; |
89 | foreach my $arg ( @_ ) { |
90 | push( @key_args, |
91 | ref( $arg ) eq 'Text::Tradition::Witness' ? $arg->sigil : $arg ); |
92 | } |
93 | return $self->$orig( @key_args ); |
94 | }; |
95 | |
96 | # Don't allow an empty hash value |
97 | around 'witness' => sub { |
98 | my( $orig, $self, $arg ) = @_; |
99 | return unless $self->has_witness( $arg ); |
100 | return $self->$orig( $arg ); |
101 | }; |
102 | |
331c2dbf |
103 | =head1 NAME |
104 | |
105 | Text::Tradition - a software model for a set of collated texts |
106 | |
107 | =head1 SYNOPSIS |
108 | |
109 | use Text::Tradition; |
110 | my $t = Text::Tradition->new( |
111 | 'name' => 'this is a text', |
112 | 'input' => 'TEI', |
113 | 'file' => '/path/to/tei_parallel_seg_file.xml' ); |
114 | |
115 | my @text_wits = $t->witnesses(); |
116 | my $manuscript_a = $t->witness( 'A' ); |
82fa4d57 |
117 | |
118 | $t = Text::Tradition->new(); |
119 | $t->add_witness( 'sourcetype' => 'xmldesc', |
120 | 'file' => '/path/to/teitranscription.xml' ); |
121 | $t->add_witness( 'sourcetype => 'plaintext', 'sigil' => 'Q', |
122 | 'string' => 'The quick brown fox jumped over the lazy dogs' ); |
123 | ## TODO |
124 | $t->collate_texts; |
331c2dbf |
125 | |
126 | my $text_path_svg = $t->collation->as_svg(); |
127 | ## See Text::Tradition::Collation for more on text collation itself |
128 | |
129 | =head1 DESCRIPTION |
130 | |
131 | Text::Tradition is a library for representation and analysis of collated |
132 | texts, particularly medieval ones. A 'tradition' refers to the aggregation |
133 | of surviving versions of a text, generally preserved in multiple |
134 | manuscripts (or 'witnesses'). A Tradition object thus has one more more |
135 | Witnesses, as well as a Collation that represents the unity of all versions |
136 | of the text. |
137 | |
138 | =head1 METHODS |
139 | |
140 | =head2 new |
141 | |
142 | Creates and returns a new text tradition object. The following options are |
143 | accepted. |
144 | |
145 | General options: |
146 | |
147 | =over 4 |
148 | |
149 | =item B<name> - The name of the text. |
150 | |
151 | =back |
152 | |
153 | Initialization based on a collation file: |
154 | |
155 | =over 4 |
156 | |
157 | =item B<input> - The input format of the collation file. Can be one of the |
158 | following: |
159 | |
160 | =over 4 |
161 | |
162 | =item * Self - a GraphML format produced by this module |
163 | |
164 | =item * CollateX - a GraphML format produced by CollateX |
165 | |
166 | =item * CTE - a TEI XML format produced by Classical Text Editor |
167 | |
a445ce40 |
168 | =item * JSON - an alignment table in JSON format, as produced by CollateX and |
169 | other tools |
331c2dbf |
170 | |
171 | =item * TEI - a TEI parallel segmentation format file |
172 | |
a445ce40 |
173 | =item * Tabular - a spreadsheet collation. See the documentation for |
174 | L<Text::Tradition::Parser::Tabular> for an explanation of additional options. |
331c2dbf |
175 | |
176 | =back |
177 | |
178 | =item B<file> - The name of the file that contains the data. One of 'file' |
179 | or 'string' should be specified. |
180 | |
181 | =item B<string> - A text string that contains the data. One of 'file' or |
182 | 'string' should be specified. |
183 | |
331c2dbf |
184 | =back |
185 | |
186 | Initialization based on a list of witnesses [NOT YET IMPLEMENTED]: |
187 | |
188 | =over 4 |
189 | |
190 | =item B<witnesses> - A reference to an array of Text::Tradition::Witness |
191 | objects that carry the text to be collated. |
192 | |
193 | =item B<collator> - A reference to a collation program that will accept |
194 | Witness objects. |
195 | |
196 | =back |
197 | |
198 | =head2 B<witnesses> |
199 | |
200 | Return the Text::Tradition::Witness objects associated with this tradition, |
201 | as an array. |
202 | |
044d1e45 |
203 | =head2 B<witness>( $sigil ) |
204 | |
205 | Returns the Text::Tradition::Witness object whose sigil is $sigil, or undef |
206 | if there is no such object within the tradition. |
207 | |
331c2dbf |
208 | =head2 B<add_witness>( %opts ) |
209 | |
210 | Instantiate a new witness with the given options (see documentation for |
211 | Text::Tradition::Witness) and add it to the tradition. |
212 | |
044d1e45 |
213 | =head2 B<del_witness>( $sigil ) |
214 | |
215 | Delete the witness with the given sigil from the tradition. Returns the |
216 | witness object for the deleted witness. |
217 | |
331c2dbf |
218 | =begin testing |
219 | |
220 | use_ok( 'Text::Tradition', "can use module" ); |
221 | |
222 | my $t = Text::Tradition->new( 'name' => 'empty' ); |
223 | is( ref( $t ), 'Text::Tradition', "initialized an empty Tradition object" ); |
224 | is( $t->name, 'empty', "object has the right name" ); |
225 | is( scalar $t->witnesses, 0, "object has no witnesses" ); |
226 | |
227 | my $simple = 't/data/simple.txt'; |
228 | my $s = Text::Tradition->new( |
229 | 'name' => 'inline', |
230 | 'input' => 'Tabular', |
231 | 'file' => $simple, |
232 | ); |
233 | is( ref( $s ), 'Text::Tradition', "initialized a Tradition object" ); |
234 | is( $s->name, 'inline', "object has the right name" ); |
235 | is( scalar $s->witnesses, 3, "object has three witnesses" ); |
236 | |
044d1e45 |
237 | my $wit_a = $s->witness('A'); |
238 | is( ref( $wit_a ), 'Text::Tradition::Witness', "Found a witness A" ); |
239 | if( $wit_a ) { |
240 | is( $wit_a->sigil, 'A', "Witness A has the right sigil" ); |
241 | } |
242 | is( $s->witness('X'), undef, "There is no witness X" ); |
243 | ok( !exists $s->{'witnesses'}->{'X'}, "Witness key X not created" ); |
244 | |
82fa4d57 |
245 | my $wit_d = $s->add_witness( 'sigil' => 'D', 'sourcetype' => 'collation' ); |
044d1e45 |
246 | is( ref( $wit_d ), 'Text::Tradition::Witness', "new witness created" ); |
247 | is( $wit_d->sigil, 'D', "witness has correct sigil" ); |
331c2dbf |
248 | is( scalar $s->witnesses, 4, "object now has four witnesses" ); |
249 | |
3b853983 |
250 | my $del = $s->del_witness( 'D' ); |
044d1e45 |
251 | is( $del, $wit_d, "Deleted correct witness" ); |
3b853983 |
252 | is( scalar $s->witnesses, 3, "object has three witnesses again" ); |
253 | |
331c2dbf |
254 | # TODO test initialization by witness list when we have it |
255 | |
256 | =end testing |
257 | |
258 | =cut |
910a0a6d |
259 | |
df6d9812 |
260 | |
8e1394aa |
261 | sub BUILD { |
262 | my( $self, $init_args ) = @_; |
fae52efd |
263 | |
264 | # First, make a collation object. This will use only those arguments in |
265 | # init_args that apply to the collation. |
266 | my $collation = Text::Tradition::Collation->new( %$init_args, |
267 | 'tradition' => $self ); |
268 | $self->_save_collation( $collation ); |
c5104dc0 |
269 | |
fae52efd |
270 | if( exists $init_args->{'input'} ) { |
910a0a6d |
271 | # Call the appropriate parser on the given data |
a731e73a |
272 | my @format_standalone = qw/ Self CollateText CollateX CTE JSON TEI Tabular /; |
dfc37e38 |
273 | my @format_basetext = qw/ KUL /; |
274 | my $use_base; |
275 | my $format = $init_args->{'input'}; |
dfc37e38 |
276 | if( $format && !( grep { $_ eq $format } @format_standalone ) |
277 | && !( grep { $_ eq $format } @format_basetext ) ) { |
278 | warn "Unrecognized input format $format; not parsing"; |
910a0a6d |
279 | return; |
280 | } |
dfc37e38 |
281 | if( $format && grep { $_ eq $format } @format_basetext ) { |
282 | $use_base = 1; |
283 | if( !exists $init_args->{'base'} ) { |
284 | warn "Cannot make a collation from $format without a base text"; |
285 | return; |
286 | } |
287 | } |
4a8828f0 |
288 | |
910a0a6d |
289 | # Now do the parsing. |
910a0a6d |
290 | if( $format ) { |
dfc37e38 |
291 | if( $use_base ) { |
292 | $format = 'BaseText'; # Use the BaseText module for parsing, |
293 | # but retain the original input arg. |
910a0a6d |
294 | } |
295 | my $mod = "Text::Tradition::Parser::$format"; |
296 | load( $mod ); |
dfc37e38 |
297 | $mod->can('parse')->( $self, $init_args ); |
910a0a6d |
298 | } |
c5104dc0 |
299 | } |
10943ab0 |
300 | $self->_init_done( 1 ); |
fae52efd |
301 | return $self; |
302 | } |
303 | |
304 | =head2 add_json_witnesses( $jsonstring, $options ) |
305 | |
306 | Adds a set of witnesses from a JSON array specification. This is a wrapper |
307 | to parse the JSON and call add_witness (with the specified $options) for |
308 | each element therein. |
309 | |
310 | =cut |
311 | |
312 | sub add_json_witnesses { |
313 | my( $self, $jsonstr, $extraopts ) = @_; |
65ed66b9 |
314 | my $witarray = from_json( $jsonstr ); |
315 | foreach my $witspec ( @{$witarray->{witnesses}} ) { |
fae52efd |
316 | my $opts = $extraopts || {}; |
317 | $opts->{'sourcetype'} = 'json'; |
318 | $opts->{'object'} = $witspec; |
319 | $self->add_witness( $opts ); |
320 | } |
8e1394aa |
321 | } |
c5104dc0 |
322 | |
951ddfe8 |
323 | sub throw { |
324 | Text::Tradition::Error->throw( |
325 | 'ident' => 'Tradition error', |
326 | 'message' => $_[0], |
327 | ); |
328 | } |
329 | |
dd3b58b0 |
330 | no Moose; |
331 | __PACKAGE__->meta->make_immutable; |
331c2dbf |
332 | |
333 | |
334 | =head1 BUGS / TODO |
335 | |
336 | =over |
337 | |
338 | =item * Allow tradition to be initialized via passing to a collator. |
339 | |
340 | =back |
341 | |
342 | =head1 LICENSE |
343 | |
344 | This package is free software and is provided "as is" without express |
345 | or implied warranty. You can redistribute it and/or modify it under |
346 | the same terms as Perl itself. |
347 | |
348 | =head1 AUTHOR |
349 | |
350 | Tara L Andrews E<lt>aurum@cpan.orgE<gt> |