Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition::Collation; |
d047cd52 |
2 | |
6771a1b1 |
3 | use feature 'say'; |
910a0a6d |
4 | use Encode qw( decode_utf8 ); |
5 | use File::Temp; |
bfcbcecb |
6 | use File::Which; |
c9bf3dbf |
7 | use Graph; |
8e1394aa |
8 | use IPC::Run qw( run binary ); |
82fa4d57 |
9 | use Text::CSV; |
5c0072ef |
10 | use Text::Tradition::Collation::Data; |
b15511bf |
11 | use Text::Tradition::Collation::Reading; |
22222af9 |
12 | use Text::Tradition::Collation::RelationshipStore; |
63778331 |
13 | use Text::Tradition::Error; |
cc31ebaa |
14 | use XML::Easy::Syntax qw( $xml10_namestartchar_rx $xml10_namechar_rx ); |
5c0072ef |
15 | use XML::LibXML; |
16 | use XML::LibXML::XPathContext; |
dd3b58b0 |
17 | use Moose; |
18 | |
5c0072ef |
19 | has _data => ( |
20 | isa => 'Text::Tradition::Collation::Data', |
21 | is => 'ro', |
22 | required => 1, |
23 | handles => [ qw( |
24 | sequence |
25 | paths |
26 | _set_relations |
27 | relations |
28 | _set_start |
29 | _set_end |
30 | ac_label |
31 | has_cached_table |
32 | relationships |
33 | related_readings |
34 | get_relationship |
35 | del_relationship |
36 | equivalence |
37 | equivalence_graph |
38 | readings |
39 | reading |
40 | _add_reading |
41 | del_reading |
42 | has_reading |
43 | wit_list_separator |
44 | baselabel |
45 | linear |
46 | wordsep |
47 | start |
48 | end |
49 | cached_table |
50 | _graphcalc_done |
51 | has_cached_svg |
52 | wipe_table |
53 | )] |
54 | ); |
dd3b58b0 |
55 | |
3a2ebbf4 |
56 | has 'tradition' => ( |
57 | is => 'ro', |
d047cd52 |
58 | isa => 'Text::Tradition', |
8cfd99c4 |
59 | writer => '_set_tradition', |
8d9a1cd8 |
60 | weak_ref => 1, |
d047cd52 |
61 | ); |
dd3b58b0 |
62 | |
4e5a7b2c |
63 | =head1 NAME |
64 | |
65 | Text::Tradition::Collation - a software model for a text collation |
66 | |
67 | =head1 SYNOPSIS |
68 | |
69 | use Text::Tradition; |
70 | my $t = Text::Tradition->new( |
71 | 'name' => 'this is a text', |
72 | 'input' => 'TEI', |
73 | 'file' => '/path/to/tei_parallel_seg_file.xml' ); |
74 | |
75 | my $c = $t->collation; |
76 | my @readings = $c->readings; |
77 | my @paths = $c->paths; |
78 | my @relationships = $c->relationships; |
79 | |
80 | my $svg_variant_graph = $t->collation->as_svg(); |
81 | |
82 | =head1 DESCRIPTION |
83 | |
84 | Text::Tradition is a library for representation and analysis of collated |
85 | texts, particularly medieval ones. The Collation is the central feature of |
86 | a Tradition, where the text, its sequence of readings, and its relationships |
87 | between readings are actually kept. |
88 | |
89 | =head1 CONSTRUCTOR |
90 | |
91 | =head2 new |
92 | |
93 | The constructor. Takes a hash or hashref of the following arguments: |
94 | |
95 | =over |
96 | |
97 | =item * tradition - The Text::Tradition object to which the collation |
98 | belongs. Required. |
99 | |
100 | =item * linear - Whether the collation should be linear; that is, whether |
101 | transposed readings should be treated as two linked readings rather than one, |
102 | and therefore whether the collation graph is acyclic. Defaults to true. |
103 | |
4e5a7b2c |
104 | =item * baselabel - The default label for the path taken by a base text |
105 | (if any). Defaults to 'base text'. |
106 | |
107 | =item * wit_list_separator - The string to join a list of witnesses for |
108 | purposes of making labels in display graphs. Defaults to ', '. |
109 | |
110 | =item * ac_label - The extra label to tack onto a witness sigil when |
111 | representing another layer of path for the given witness - that is, when |
112 | a text has more than one possible reading due to scribal corrections or |
113 | the like. Defaults to ' (a.c.)'. |
114 | |
4e483aa5 |
115 | =item * wordsep - The string used to separate words in the original text. |
116 | Defaults to ' '. |
117 | |
4e5a7b2c |
118 | =back |
119 | |
120 | =head1 ACCESSORS |
121 | |
122 | =head2 tradition |
123 | |
124 | =head2 linear |
125 | |
4e5a7b2c |
126 | =head2 wit_list_separator |
127 | |
128 | =head2 baselabel |
129 | |
130 | =head2 ac_label |
131 | |
4e483aa5 |
132 | =head2 wordsep |
133 | |
4e5a7b2c |
134 | Simple accessors for collation attributes. |
135 | |
136 | =head2 start |
137 | |
138 | The meta-reading at the start of every witness path. |
139 | |
140 | =head2 end |
141 | |
142 | The meta-reading at the end of every witness path. |
143 | |
144 | =head2 readings |
145 | |
146 | Returns all Reading objects in the graph. |
147 | |
148 | =head2 reading( $id ) |
149 | |
150 | Returns the Reading object corresponding to the given ID. |
151 | |
152 | =head2 add_reading( $reading_args ) |
153 | |
154 | Adds a new reading object to the collation. |
155 | See L<Text::Tradition::Collation::Reading> for the available arguments. |
156 | |
157 | =head2 del_reading( $object_or_id ) |
158 | |
159 | Removes the given reading from the collation, implicitly removing its |
160 | paths and relationships. |
161 | |
4e5a7b2c |
162 | =head2 has_reading( $id ) |
163 | |
164 | Predicate to see whether a given reading ID is in the graph. |
165 | |
166 | =head2 reading_witnesses( $object_or_id ) |
167 | |
168 | Returns a list of sigils whose witnesses contain the reading. |
169 | |
170 | =head2 paths |
171 | |
172 | Returns all reading paths within the document - that is, all edges in the |
173 | collation graph. Each path is an arrayref of [ $source, $target ] reading IDs. |
174 | |
175 | =head2 add_path( $source, $target, $sigil ) |
176 | |
177 | Links the given readings in the collation in sequence, under the given witness |
178 | sigil. The readings may be specified by object or ID. |
179 | |
180 | =head2 del_path( $source, $target, $sigil ) |
181 | |
182 | Links the given readings in the collation in sequence, under the given witness |
183 | sigil. The readings may be specified by object or ID. |
184 | |
185 | =head2 has_path( $source, $target ); |
186 | |
187 | Returns true if the two readings are linked in sequence in any witness. |
188 | The readings may be specified by object or ID. |
189 | |
190 | =head2 relationships |
191 | |
192 | Returns all Relationship objects in the collation. |
193 | |
194 | =head2 add_relationship( $reading, $other_reading, $options ) |
195 | |
196 | Adds a new relationship of the type given in $options between the two readings, |
197 | which may be specified by object or ID. Returns a value of ( $status, @vectors) |
198 | where $status is true on success, and @vectors is a list of relationship edges |
199 | that were ultimately added. |
200 | See L<Text::Tradition::Collation::Relationship> for the available options. |
201 | |
202 | =cut |
dd3b58b0 |
203 | |
5c0072ef |
204 | sub BUILDARGS { |
205 | my ( $class, @args ) = @_; |
206 | my %args = @args == 1 ? %{ $args[0] } : @args; |
207 | # TODO determine these from the Moose::Meta object |
208 | my @delegate_attrs = qw(sequence relations readings wit_list_separator baselabel |
209 | linear wordsep start end cached_table _graphcalc_done); |
210 | my %data_args; |
211 | for my $attr (@delegate_attrs) { |
212 | $data_args{$attr} = delete $args{$attr} if exists $args{$attr}; |
213 | } |
214 | $args{_data} = Text::Tradition::Collation::Data->new(%data_args); |
215 | return \%args; |
216 | } |
217 | |
d047cd52 |
218 | sub BUILD { |
3a2ebbf4 |
219 | my $self = shift; |
22222af9 |
220 | $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) ); |
e4b73942 |
221 | $self->_set_start( $self->add_reading( |
222 | { 'collation' => $self, 'is_start' => 1, 'init' => 1 } ) ); |
223 | $self->_set_end( $self->add_reading( |
224 | { 'collation' => $self, 'is_end' => 1, 'init' => 1 } ) ); |
d047cd52 |
225 | } |
784877d9 |
226 | |
24efa55d |
227 | sub register_relationship_type { |
228 | my $self = shift; |
229 | my %args = @_ == 1 ? %{$_[0]} : @_; |
230 | if( $self->relations->has_type( $args{name} ) ) { |
231 | throw( 'Relationship type ' . $args{name} . ' already registered' ); |
232 | } |
233 | $self->relations->add_type( %args ); |
234 | } |
235 | |
3a2ebbf4 |
236 | ### Reading construct/destruct functions |
237 | |
238 | sub add_reading { |
239 | my( $self, $reading ) = @_; |
240 | unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) { |
241 | my %args = %$reading; |
e4b73942 |
242 | if( $args{'init'} ) { |
243 | # If we are initializing an empty collation, don't assume that we |
244 | # have set a tradition. |
245 | delete $args{'init'}; |
8943ff68 |
246 | } elsif( $self->tradition->can('language') && $self->tradition->has_language |
247 | && !exists $args{'language'} ) { |
6ad2ce78 |
248 | $args{'language'} = $self->tradition->language; |
249 | } |
3a2ebbf4 |
250 | $reading = Text::Tradition::Collation::Reading->new( |
251 | 'collation' => $self, |
252 | %args ); |
253 | } |
254 | # First check to see if a reading with this ID exists. |
255 | if( $self->reading( $reading->id ) ) { |
63778331 |
256 | throw( "Collation already has a reading with id " . $reading->id ); |
3a2ebbf4 |
257 | } |
c1915ab9 |
258 | $self->_graphcalc_done(0); |
3a2ebbf4 |
259 | $self->_add_reading( $reading->id => $reading ); |
260 | # Once the reading has been added, put it in both graphs. |
261 | $self->sequence->add_vertex( $reading->id ); |
cecbe56d |
262 | $self->relations->add_reading( $reading->id ); |
3a2ebbf4 |
263 | return $reading; |
eca16057 |
264 | }; |
265 | |
3a2ebbf4 |
266 | around del_reading => sub { |
267 | my $orig = shift; |
268 | my $self = shift; |
269 | my $arg = shift; |
270 | |
cecbe56d |
271 | if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) { |
272 | $arg = $arg->id; |
3a2ebbf4 |
273 | } |
3a2ebbf4 |
274 | # Remove the reading from the graphs. |
c1915ab9 |
275 | $self->_graphcalc_done(0); |
4e483aa5 |
276 | $self->_clear_cache; # Explicitly clear caches to GC the reading |
cecbe56d |
277 | $self->sequence->delete_vertex( $arg ); |
278 | $self->relations->delete_reading( $arg ); |
3a2ebbf4 |
279 | |
280 | # Carry on. |
cecbe56d |
281 | $self->$orig( $arg ); |
3a2ebbf4 |
282 | }; |
7854e12e |
283 | |
3c234eb6 |
284 | =head2 merge_readings( $main, $second, $concatenate, $with_str ) |
285 | |
286 | Merges the $second reading into the $main one. If $concatenate is true, then |
287 | the merged node will carry the text of both readings, concatenated with either |
288 | $with_str (if specified) or a sensible default (the empty string if the |
289 | appropriate 'join_*' flag is set on either reading, or else $self->wordsep.) |
290 | |
291 | The first two arguments may be either readings or reading IDs. |
292 | |
4e483aa5 |
293 | =begin testing |
294 | |
295 | use Text::Tradition; |
296 | |
297 | my $cxfile = 't/data/Collatex-16.xml'; |
298 | my $t = Text::Tradition->new( |
299 | 'name' => 'inline', |
300 | 'input' => 'CollateX', |
301 | 'file' => $cxfile, |
302 | ); |
303 | my $c = $t->collation; |
304 | |
305 | my $rno = scalar $c->readings; |
4ef65ab4 |
306 | # Split n21 ('unto') for testing purposes |
4e483aa5 |
307 | my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } ); |
308 | my $old_r = $c->reading( 'n21' ); |
309 | $old_r->alter_text( 'to' ); |
310 | $c->del_path( 'n20', 'n21', 'A' ); |
311 | $c->add_path( 'n20', 'n21p0', 'A' ); |
312 | $c->add_path( 'n21p0', 'n21', 'A' ); |
7a0956c1 |
313 | $c->add_relationship( 'n21', 'n22', { type => 'collated', scope => 'local' } ); |
4e483aa5 |
314 | $c->flatten_ranks(); |
315 | ok( $c->reading( 'n21p0' ), "New reading exists" ); |
316 | is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" ); |
317 | |
679f17e1 |
318 | # Combine n3 and n4 ( with his ) |
4e483aa5 |
319 | $c->merge_readings( 'n3', 'n4', 1 ); |
320 | ok( !$c->reading('n4'), "Reading n4 is gone" ); |
321 | is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" ); |
322 | |
679f17e1 |
323 | # Collapse n9 and n10 ( rood / root ) |
324 | $c->merge_readings( 'n9', 'n10' ); |
325 | ok( !$c->reading('n10'), "Reading n10 is gone" ); |
326 | is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" ); |
4e483aa5 |
327 | |
328 | # Combine n21 and n21p0 |
329 | my $remaining = $c->reading('n21'); |
330 | $remaining ||= $c->reading('n22'); # one of these should still exist |
331 | $c->merge_readings( 'n21p0', $remaining, 1 ); |
332 | ok( !$c->reading('n21'), "Reading $remaining is gone" ); |
333 | is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" ); |
334 | |
335 | =end testing |
336 | |
337 | =cut |
7854e12e |
338 | |
3a2ebbf4 |
339 | sub merge_readings { |
340 | my $self = shift; |
341 | |
56772e8c |
342 | # Sanity check |
343 | my( $kept_obj, $del_obj, $combine, $combine_char ) = $self->_objectify_args( @_ ); |
344 | my $mergemeta = $kept_obj->is_meta; |
345 | throw( "Cannot merge meta and non-meta reading" ) |
346 | unless ( $mergemeta && $del_obj->is_meta ) |
347 | || ( !$mergemeta && !$del_obj->is_meta ); |
348 | if( $mergemeta ) { |
349 | throw( "Cannot merge with start or end node" ) |
350 | if( $kept_obj eq $self->start || $kept_obj eq $self->end |
351 | || $del_obj eq $self->start || $del_obj eq $self->end ); |
a445ce40 |
352 | throw( "Cannot combine text of meta readings" ) if $combine; |
56772e8c |
353 | } |
3a2ebbf4 |
354 | # We only need the IDs for adding paths to the graph, not the reading |
355 | # objects themselves. |
56772e8c |
356 | my $kept = $kept_obj->id; |
357 | my $deleted = $del_obj->id; |
c1915ab9 |
358 | $self->_graphcalc_done(0); |
10943ab0 |
359 | |
3a2ebbf4 |
360 | # The kept reading should inherit the paths and the relationships |
361 | # of the deleted reading. |
362 | foreach my $path ( $self->sequence->edges_at( $deleted ) ) { |
363 | my @vector = ( $kept ); |
364 | push( @vector, $path->[1] ) if $path->[0] eq $deleted; |
365 | unshift( @vector, $path->[0] ) if $path->[1] eq $deleted; |
49d4f2ac |
366 | next if $vector[0] eq $vector[1]; # Don't add a self loop |
3a2ebbf4 |
367 | my %wits = %{$self->sequence->get_edge_attributes( @$path )}; |
368 | $self->sequence->add_edge( @vector ); |
369 | my $fwits = $self->sequence->get_edge_attributes( @vector ); |
370 | @wits{keys %$fwits} = values %$fwits; |
371 | $self->sequence->set_edge_attributes( @vector, \%wits ); |
372 | } |
cecbe56d |
373 | $self->relations->merge_readings( $kept, $deleted, $combine ); |
3a2ebbf4 |
374 | |
375 | # Do the deletion deed. |
4e483aa5 |
376 | if( $combine ) { |
869a1ada |
377 | # Combine the text of the readings |
4e483aa5 |
378 | my $joinstr = $combine_char; |
379 | unless( defined $joinstr ) { |
380 | $joinstr = '' if $kept_obj->join_next || $del_obj->join_prior; |
381 | $joinstr = $self->wordsep unless defined $joinstr; |
382 | } |
a445ce40 |
383 | $kept_obj->_combine( $del_obj, $joinstr ); |
49d4f2ac |
384 | } |
3a2ebbf4 |
385 | $self->del_reading( $deleted ); |
386 | } |
7854e12e |
387 | |
6771a1b1 |
388 | =head2 compress_readings |
389 | |
390 | Where possible in the graph, compresses plain sequences of readings into a |
391 | single reading. The sequences must consist of readings with no |
392 | relationships to other readings, with only a single witness path between |
393 | them and no other witness paths from either that would skip the other. The |
394 | readings must also not be marked as nonsense or bad grammar. |
395 | |
396 | WARNING: This operation cannot be undone. |
397 | |
398 | =cut |
399 | |
400 | sub compress_readings { |
401 | my $self = shift; |
402 | # Anywhere in the graph that there is a reading that joins only to a single |
403 | # successor, and neither of these have any relationships, just join the two |
404 | # readings. |
6771a1b1 |
405 | foreach my $rdg ( sort { $a->rank <=> $b->rank } $self->readings ) { |
428bcf0b |
406 | # Now look for readings that can be joined to their successors. |
a445ce40 |
407 | next unless $rdg->is_combinable; |
6771a1b1 |
408 | my %seen; |
409 | while( $self->sequence->successors( $rdg ) == 1 ) { |
410 | my( $next ) = $self->reading( $self->sequence->successors( $rdg ) ); |
411 | throw( "Infinite loop" ) if $seen{$next->id}; |
412 | $seen{$next->id} = 1; |
413 | last if $self->sequence->predecessors( $next ) > 1; |
a445ce40 |
414 | last unless $next->is_combinable; |
6771a1b1 |
415 | say "Joining readings $rdg and $next"; |
416 | $self->merge_readings( $rdg, $next, 1 ); |
417 | } |
418 | } |
419 | # Make sure we haven't screwed anything up |
420 | foreach my $wit ( $self->tradition->witnesses ) { |
421 | my $pathtext = $self->path_text( $wit->sigil ); |
422 | my $origtext = join( ' ', @{$wit->text} ); |
423 | throw( "Text differs for witness " . $wit->sigil ) |
424 | unless $pathtext eq $origtext; |
425 | if( $wit->is_layered ) { |
426 | $pathtext = $self->path_text( $wit->sigil.$self->ac_label ); |
427 | $origtext = join( ' ', @{$wit->layertext} ); |
428 | throw( "Ante-corr text differs for witness " . $wit->sigil ) |
429 | unless $pathtext eq $origtext; |
430 | } |
431 | } |
432 | |
433 | $self->relations->rebuild_equivalence(); |
434 | $self->calculate_ranks(); |
435 | } |
3265b0ce |
436 | |
3a2ebbf4 |
437 | # Helper function for manipulating the graph. |
438 | sub _stringify_args { |
4e483aa5 |
439 | my( $self, $first, $second, @args ) = @_; |
3a2ebbf4 |
440 | $first = $first->id |
441 | if ref( $first ) eq 'Text::Tradition::Collation::Reading'; |
442 | $second = $second->id |
443 | if ref( $second ) eq 'Text::Tradition::Collation::Reading'; |
4e483aa5 |
444 | return( $first, $second, @args ); |
3a2ebbf4 |
445 | } |
df6d9812 |
446 | |
4e5a7b2c |
447 | # Helper function for manipulating the graph. |
448 | sub _objectify_args { |
449 | my( $self, $first, $second, $arg ) = @_; |
450 | $first = $self->reading( $first ) |
451 | unless ref( $first ) eq 'Text::Tradition::Collation::Reading'; |
452 | $second = $self->reading( $second ) |
453 | unless ref( $second ) eq 'Text::Tradition::Collation::Reading'; |
454 | return( $first, $second, $arg ); |
455 | } |
f97ef19e |
456 | |
457 | =head2 duplicate_reading( $reading, @witlist ) |
458 | |
459 | Split the given reading into two, so that the new reading is in the path for |
ef73c20a |
460 | the witnesses given in @witlist. If the result is that certain non-colocated |
461 | relationships (e.g. transpositions) are no longer valid, these will be removed. |
462 | Returns the newly-created reading. |
f97ef19e |
463 | |
464 | =begin testing |
465 | |
466 | use Text::Tradition; |
467 | |
468 | my $st = Text::Tradition->new( 'input' => 'Self', 'file' => 't/data/collatecorr.xml' ); |
469 | is( ref( $st ), 'Text::Tradition', "Got a tradition from test file" ); |
470 | ok( $st->has_witness('Ba96'), "Tradition has the affected witness" ); |
471 | |
472 | my $sc = $st->collation; |
473 | my $numr = 17; |
474 | ok( $sc->reading('n131'), "Tradition has the affected reading" ); |
475 | is( scalar( $sc->readings ), $numr, "There are $numr readings in the graph" ); |
476 | is( $sc->end->rank, 14, "There are fourteen ranks in the graph" ); |
477 | |
478 | # Detach the erroneously collated reading |
2dcb5d11 |
479 | my( $newr, @del_rdgs ) = $sc->duplicate_reading( 'n131', 'Ba96' ); |
ef73c20a |
480 | ok( $newr, "New reading was created" ); |
f97ef19e |
481 | ok( $sc->reading('n131_0'), "Detached the bad collation with a new reading" ); |
482 | is( scalar( $sc->readings ), $numr + 1, "A reading was added to the graph" ); |
483 | is( $sc->end->rank, 10, "There are now only ten ranks in the graph" ); |
3c234eb6 |
484 | my $csucc = $sc->common_successor( 'n131', 'n131_0' ); |
485 | is( $csucc->id, 'n136', "Found correct common successor to duped reading" ); |
f97ef19e |
486 | |
487 | # Check that the bad transposition is gone |
2dcb5d11 |
488 | is( scalar @del_rdgs, 1, "Deleted reading was returned by API call" ); |
f97ef19e |
489 | is( $sc->get_relationship( 'n130', 'n135' ), undef, "Bad transposition relationship is gone" ); |
490 | |
e19635f8 |
491 | # The collation should not be fixed |
492 | my @pairs = $sc->identical_readings(); |
493 | is( scalar @pairs, 0, "Not re-collated yet" ); |
f97ef19e |
494 | # Fix the collation |
3c234eb6 |
495 | ok( $sc->merge_readings( 'n124', 'n131_0' ), "Collated the readings correctly" ); |
e19635f8 |
496 | @pairs = $sc->identical_readings( start => 'n124', end => $csucc->id ); |
3c234eb6 |
497 | is( scalar @pairs, 3, "Found three more identical readings" ); |
e19635f8 |
498 | is( $sc->end->rank, 11, "The ranks shifted appropriately" ); |
3c234eb6 |
499 | $sc->flatten_ranks(); |
f97ef19e |
500 | is( scalar( $sc->readings ), $numr - 3, "Now we are collated correctly" ); |
501 | |
502 | =end testing |
503 | |
504 | =cut |
505 | |
506 | sub duplicate_reading { |
507 | my( $self, $r, @wits ) = @_; |
508 | # Add the new reading, duplicating $r. |
509 | unless( ref( $r ) eq 'Text::Tradition::Collation::Reading' ) { |
510 | $r = $self->reading( $r ); |
511 | } |
512 | throw( "Cannot duplicate a meta-reading" ) |
513 | if $r->is_meta; |
514 | |
515 | # Get all the reading attributes and duplicate them. |
516 | my $rmeta = Text::Tradition::Collation::Reading->meta; |
517 | my %args; |
518 | foreach my $attr( $rmeta->get_all_attributes ) { |
519 | next if $attr->name =~ /^_/; |
520 | my $acc = $attr->get_read_method; |
521 | if( !$acc && $attr->has_applied_traits ) { |
522 | my $tr = $attr->applied_traits; |
523 | if( $tr->[0] =~ /::(Array|Hash)$/ ) { |
524 | my $which = $1; |
525 | my %methods = reverse %{$attr->handles}; |
526 | $acc = $methods{elements}; |
527 | $args{$attr->name} = $which eq 'Array' |
528 | ? [ $r->$acc ] : { $r->$acc }; |
529 | } |
530 | } else { |
531 | $args{$attr->name} = $r->$acc if $acc; |
532 | } |
533 | } |
534 | # By definition the new reading will no longer be common. |
535 | $args{is_common} = 0; |
536 | # The new reading also needs its own ID. |
537 | $args{id} = $self->_generate_dup_id( $r->id ); |
538 | |
539 | # Try to make the new reading. |
540 | my $newr = $self->add_reading( \%args ); |
541 | # The old reading is also no longer common. |
542 | $r->is_common( 0 ); |
543 | |
544 | # For each of the witnesses, dissociate from the old reading and |
545 | # associate with the new. |
546 | foreach my $wit ( @wits ) { |
547 | my $prior = $self->prior_reading( $r, $wit ); |
548 | my $next = $self->next_reading( $r, $wit ); |
549 | $self->del_path( $prior, $r, $wit ); |
550 | $self->add_path( $prior, $newr, $wit ); |
551 | $self->del_path( $r, $next, $wit ); |
552 | $self->add_path( $newr, $next, $wit ); |
553 | } |
554 | |
3c234eb6 |
555 | # If the graph is ranked, we need to look for relationships that are now |
556 | # invalid (i.e. 'non-colocation' types that might now be colocated) and |
557 | # remove them. If not, we can skip it. |
f97ef19e |
558 | my $succ; |
3c234eb6 |
559 | my %rrk; |
2dcb5d11 |
560 | my @deleted_relations; |
f97ef19e |
561 | if( $self->end->has_rank ) { |
3c234eb6 |
562 | # Find the point where we can stop checking |
f97ef19e |
563 | $succ = $self->common_successor( $r, $newr ); |
3c234eb6 |
564 | |
565 | # Hash the existing ranks |
f97ef19e |
566 | foreach my $rdg ( $self->readings ) { |
567 | $rrk{$rdg->id} = $rdg->rank; |
568 | } |
3c234eb6 |
569 | # Calculate the new ranks |
570 | $self->calculate_ranks(); |
f97ef19e |
571 | |
3c234eb6 |
572 | # Check for invalid non-colocated relationships among changed-rank readings |
573 | # from where the ranks start changing up to $succ |
f97ef19e |
574 | my $lastrank = $succ->rank; |
575 | foreach my $rdg ( $self->readings ) { |
576 | next if $rdg->rank > $lastrank; |
577 | next if $rdg->rank == $rrk{$rdg->id}; |
578 | my @noncolo = $rdg->related_readings( sub { !$_[0]->colocated } ); |
579 | next unless @noncolo; |
580 | foreach my $nc ( @noncolo ) { |
2dcb5d11 |
581 | unless( $self->relations->verify_or_delete( $rdg, $nc ) ) { |
582 | push( @deleted_relations, $nc ); |
583 | } |
f97ef19e |
584 | } |
585 | } |
ef73c20a |
586 | } |
2dcb5d11 |
587 | return ( $newr, @deleted_relations ); |
f97ef19e |
588 | } |
589 | |
590 | sub _generate_dup_id { |
591 | my( $self, $rid ) = @_; |
592 | my $newid; |
593 | my $i = 0; |
594 | while( !$newid ) { |
595 | $newid = $rid."_$i"; |
596 | if( $self->has_reading( $newid ) ) { |
597 | $newid = ''; |
598 | $i++; |
599 | } |
600 | } |
601 | return $newid; |
602 | } |
603 | |
3a2ebbf4 |
604 | ### Path logic |
605 | |
606 | sub add_path { |
607 | my $self = shift; |
608 | |
609 | # We only need the IDs for adding paths to the graph, not the reading |
610 | # objects themselves. |
cecbe56d |
611 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
3a2ebbf4 |
612 | |
c1915ab9 |
613 | $self->_graphcalc_done(0); |
3a2ebbf4 |
614 | # Connect the readings |
359944f7 |
615 | unless( $self->sequence->has_edge( $source, $target ) ) { |
616 | $self->sequence->add_edge( $source, $target ); |
617 | $self->relations->add_equivalence_edge( $source, $target ); |
618 | } |
3a2ebbf4 |
619 | # Note the witness in question |
620 | $self->sequence->set_edge_attribute( $source, $target, $wit, 1 ); |
359944f7 |
621 | } |
b15511bf |
622 | |
3a2ebbf4 |
623 | sub del_path { |
624 | my $self = shift; |
49d4f2ac |
625 | my @args; |
626 | if( ref( $_[0] ) eq 'ARRAY' ) { |
627 | my $e = shift @_; |
628 | @args = ( @$e, @_ ); |
629 | } else { |
630 | @args = @_; |
631 | } |
3a2ebbf4 |
632 | |
f97ef19e |
633 | # We only need the IDs for removing paths from the graph, not the reading |
3a2ebbf4 |
634 | # objects themselves. |
49d4f2ac |
635 | my( $source, $target, $wit ) = $self->_stringify_args( @args ); |
3a2ebbf4 |
636 | |
c1915ab9 |
637 | $self->_graphcalc_done(0); |
3a2ebbf4 |
638 | if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) { |
49d4f2ac |
639 | $self->sequence->delete_edge_attribute( $source, $target, $wit ); |
3a2ebbf4 |
640 | } |
30897024 |
641 | unless( $self->sequence->has_edge_attributes( $source, $target ) ) { |
3a2ebbf4 |
642 | $self->sequence->delete_edge( $source, $target ); |
359944f7 |
643 | $self->relations->delete_equivalence_edge( $source, $target ); |
3a2ebbf4 |
644 | } |
784877d9 |
645 | } |
646 | |
3a2ebbf4 |
647 | |
15d2d3df |
648 | # Extra graph-alike utility |
649 | sub has_path { |
3a2ebbf4 |
650 | my $self = shift; |
651 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
652 | return undef unless $self->sequence->has_edge( $source, $target ); |
653 | return $self->sequence->has_edge_attribute( $source, $target, $wit ); |
b15511bf |
654 | } |
655 | |
4e5a7b2c |
656 | =head2 clear_witness( @sigil_list ) |
3a2ebbf4 |
657 | |
4e5a7b2c |
658 | Clear the given witnesses out of the collation entirely, removing references |
659 | to them in paths, and removing readings that belong only to them. Should only |
660 | be called via $tradition->del_witness. |
3a2ebbf4 |
661 | |
662 | =cut |
663 | |
4e5a7b2c |
664 | sub clear_witness { |
665 | my( $self, @sigils ) = @_; |
666 | |
c1915ab9 |
667 | $self->_graphcalc_done(0); |
4e5a7b2c |
668 | # Clear the witness(es) out of the paths |
669 | foreach my $e ( $self->paths ) { |
670 | foreach my $sig ( @sigils ) { |
671 | $self->del_path( $e, $sig ); |
672 | } |
673 | } |
674 | |
675 | # Clear out the newly unused readings |
676 | foreach my $r ( $self->readings ) { |
677 | unless( $self->reading_witnesses( $r ) ) { |
678 | $self->del_reading( $r ); |
679 | } |
680 | } |
681 | } |
3a2ebbf4 |
682 | |
683 | sub add_relationship { |
684 | my $self = shift; |
22222af9 |
685 | my( $source, $target, $opts ) = $self->_stringify_args( @_ ); |
414cc046 |
686 | my( @vectors ) = $self->relations->add_relationship( $source, $target, $opts ); |
864ee4bf |
687 | foreach my $v ( @vectors ) { |
688 | next unless $self->get_relationship( $v )->colocated; |
689 | if( $self->reading( $v->[0] )->has_rank && $self->reading( $v->[1] )->has_rank |
690 | && $self->reading( $v->[0] )->rank ne $self->reading( $v->[1] )->rank ) { |
691 | $self->_graphcalc_done(0); |
692 | $self->_clear_cache; |
693 | last; |
694 | } |
695 | } |
63778331 |
696 | return @vectors; |
22222af9 |
697 | } |
ef9d481f |
698 | |
ca6e6095 |
699 | around qw/ get_relationship del_relationship / => sub { |
700 | my $orig = shift; |
701 | my $self = shift; |
702 | my @args = @_; |
703 | if( @args == 1 && ref( $args[0] ) eq 'ARRAY' ) { |
704 | @args = @{$_[0]}; |
705 | } |
e19635f8 |
706 | my @stringargs = $self->_stringify_args( @args ); |
707 | $self->$orig( @stringargs ); |
ca6e6095 |
708 | }; |
709 | |
22222af9 |
710 | =head2 reading_witnesses( $reading ) |
910a0a6d |
711 | |
22222af9 |
712 | Return a list of sigils corresponding to the witnesses in which the reading appears. |
3265b0ce |
713 | |
22222af9 |
714 | =cut |
1d310495 |
715 | |
1d310495 |
716 | sub reading_witnesses { |
717 | my( $self, $reading ) = @_; |
718 | # We need only check either the incoming or the outgoing edges; I have |
96dc90ec |
719 | # arbitrarily chosen "incoming". Thus, special-case the start node. |
720 | if( $reading eq $self->start ) { |
bed6ce83 |
721 | return map { $_->sigil } grep { $_->is_collated } $self->tradition->witnesses; |
96dc90ec |
722 | } |
1d310495 |
723 | my %all_witnesses; |
724 | foreach my $e ( $self->sequence->edges_to( $reading ) ) { |
725 | my $wits = $self->sequence->get_edge_attributes( @$e ); |
726 | @all_witnesses{ keys %$wits } = 1; |
727 | } |
c12bb878 |
728 | my $acstr = $self->ac_label; |
729 | foreach my $acwit ( grep { $_ =~ s/^(.*)\Q$acstr\E$/$1/ } keys %all_witnesses ) { |
730 | delete $all_witnesses{$acwit.$acstr} if exists $all_witnesses{$acwit}; |
731 | } |
1d310495 |
732 | return keys %all_witnesses; |
910a0a6d |
733 | } |
734 | |
4e5a7b2c |
735 | =head1 OUTPUT METHODS |
8e1394aa |
736 | |
0ecb975c |
737 | =head2 as_svg( \%options ) |
8e1394aa |
738 | |
0068967c |
739 | Returns an SVG string that represents the graph, via as_dot and graphviz. |
bfcbcecb |
740 | See as_dot for a list of options. Must have GraphViz (dot) installed to run. |
8e1394aa |
741 | |
742 | =cut |
743 | |
744 | sub as_svg { |
0ecb975c |
745 | my( $self, $opts ) = @_; |
bfcbcecb |
746 | throw( "Need GraphViz installed to output SVG" ) |
747 | unless File::Which::which( 'dot' ); |
e247aad1 |
748 | my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'}; |
1ff82d4f |
749 | $self->calculate_ranks() |
750 | unless( $self->_graphcalc_done || $opts->{'nocalc'} || !$self->linear ); |
be3af600 |
751 | my @cmd = qw/dot -Tsvg/; |
752 | my( $svg, $err ); |
753 | my $dotfile = File::Temp->new(); |
754 | ## USE FOR DEBUGGING |
755 | # $dotfile->unlink_on_destroy(0); |
756 | binmode $dotfile, ':utf8'; |
757 | print $dotfile $self->as_dot( $opts ); |
758 | push( @cmd, $dotfile->filename ); |
759 | run( \@cmd, ">", binary(), \$svg ); |
760 | $svg = decode_utf8( $svg ); |
761 | return $svg; |
8e1394aa |
762 | } |
763 | |
b22576c6 |
764 | |
0ecb975c |
765 | =head2 as_dot( \%options ) |
b22576c6 |
766 | |
0ecb975c |
767 | Returns a string that is the collation graph expressed in dot |
768 | (i.e. GraphViz) format. Options include: |
b22576c6 |
769 | |
0ecb975c |
770 | =over 4 |
b22576c6 |
771 | |
0ecb975c |
772 | =item * from |
b22576c6 |
773 | |
0ecb975c |
774 | =item * to |
df6d9812 |
775 | |
0ecb975c |
776 | =item * color_common |
777 | |
778 | =back |
df6d9812 |
779 | |
780 | =cut |
781 | |
782 | sub as_dot { |
0ecb975c |
783 | my( $self, $opts ) = @_; |
784 | my $startrank = $opts->{'from'} if $opts; |
785 | my $endrank = $opts->{'to'} if $opts; |
786 | my $color_common = $opts->{'color_common'} if $opts; |
b365fbae |
787 | my $STRAIGHTENHACK = !$startrank && !$endrank && $self->end->rank |
788 | && $self->end->rank > 100; |
6648ee3d |
789 | $STRAIGHTENHACK = 1 if $opts->{'straight'}; # even for subgraphs or small graphs |
b365fbae |
790 | |
b22576c6 |
791 | # Check the arguments |
792 | if( $startrank ) { |
793 | return if $endrank && $startrank > $endrank; |
794 | return if $startrank > $self->end->rank; |
795 | } |
796 | if( defined $endrank ) { |
797 | return if $endrank < 0; |
f1b3b33a |
798 | $endrank = undef if $endrank == $self->end->rank; |
b22576c6 |
799 | } |
800 | |
67da8d6c |
801 | my $graph_name = $self->tradition->name; |
802 | $graph_name =~ s/[^\w\s]//g; |
803 | $graph_name = join( '_', split( /\s+/, $graph_name ) ); |
f13b5582 |
804 | |
805 | my %graph_attrs = ( |
806 | 'rankdir' => 'LR', |
807 | 'bgcolor' => 'none', |
808 | ); |
809 | my %node_attrs = ( |
b8990398 |
810 | 'fontsize' => 14, |
f13b5582 |
811 | 'fillcolor' => 'white', |
812 | 'style' => 'filled', |
813 | 'shape' => 'ellipse' |
814 | ); |
815 | my %edge_attrs = ( |
816 | 'arrowhead' => 'open', |
817 | 'color' => '#000000', |
818 | 'fontcolor' => '#000000', |
819 | ); |
820 | |
67da8d6c |
821 | my $dot = sprintf( "digraph %s {\n", $graph_name ); |
f13b5582 |
822 | $dot .= "\tgraph " . _dot_attr_string( \%graph_attrs ) . ";\n"; |
823 | $dot .= "\tnode " . _dot_attr_string( \%node_attrs ) . ";\n"; |
df6d9812 |
824 | |
b22576c6 |
825 | # Output substitute start/end readings if necessary |
826 | if( $startrank ) { |
43c94341 |
827 | $dot .= "\t\"__SUBSTART__\" [ label=\"...\",id=\"__START__\" ];\n"; |
b22576c6 |
828 | } |
829 | if( $endrank ) { |
43c94341 |
830 | $dot .= "\t\"__SUBEND__\" [ label=\"...\",id=\"__END__\" ];\n"; |
b22576c6 |
831 | } |
b365fbae |
832 | if( $STRAIGHTENHACK ) { |
833 | ## HACK part 1 |
43c94341 |
834 | my $startlabel = $startrank ? '__SUBSTART__' : '__START__'; |
835 | $dot .= "\tsubgraph { rank=same \"$startlabel\" \"#SILENT#\" }\n"; |
b365fbae |
836 | $dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];" |
837 | } |
b22576c6 |
838 | my %used; # Keep track of the readings that actually appear in the graph |
30ddc24c |
839 | # Sort the readings by rank if we have ranks; this speeds layout. |
840 | my @all_readings = $self->end->has_rank |
841 | ? sort { $a->rank <=> $b->rank } $self->readings |
842 | : $self->readings; |
4633f9e4 |
843 | # TODO Refrain from outputting lacuna nodes - just grey out the edges. |
30ddc24c |
844 | foreach my $reading ( @all_readings ) { |
b22576c6 |
845 | # Only output readings within our rank range. |
846 | next if $startrank && $reading->rank < $startrank; |
847 | next if $endrank && $reading->rank > $endrank; |
848 | $used{$reading->id} = 1; |
910a0a6d |
849 | # Need not output nodes without separate labels |
3a2ebbf4 |
850 | next if $reading->id eq $reading->text; |
d4b75f44 |
851 | my $rattrs; |
30f0df34 |
852 | my $label = $reading->text; |
629e27b0 |
853 | $label .= '-' if $reading->join_next; |
854 | $label = "-$label" if $reading->join_prior; |
8f9cab7b |
855 | $label =~ s/\"/\\\"/g; |
d4b75f44 |
856 | $rattrs->{'label'} = $label; |
10e4b1ac |
857 | $rattrs->{'id'} = $reading->id; |
0ecb975c |
858 | $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common; |
d4b75f44 |
859 | $dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) ); |
df6d9812 |
860 | } |
3a2ebbf4 |
861 | |
30ddc24c |
862 | # Add the real edges. Need to weight one edge per rank jump, in a |
863 | # continuous line. |
b365fbae |
864 | # my $weighted = $self->_add_edge_weights; |
b22576c6 |
865 | my @edges = $self->paths; |
3bdec618 |
866 | my( %substart, %subend ); |
b22576c6 |
867 | foreach my $edge ( @edges ) { |
868 | # Do we need to output this edge? |
508fd430 |
869 | if( $used{$edge->[0]} && $used{$edge->[1]} ) { |
027d819c |
870 | my $label = $self->_path_display_label( $self->path_witnesses( $edge ) ); |
f13b5582 |
871 | my $variables = { %edge_attrs, 'label' => $label }; |
30ddc24c |
872 | |
b22576c6 |
873 | # Account for the rank gap if necessary |
30ddc24c |
874 | my $rank0 = $self->reading( $edge->[0] )->rank |
875 | if $self->reading( $edge->[0] )->has_rank; |
876 | my $rank1 = $self->reading( $edge->[1] )->rank |
877 | if $self->reading( $edge->[1] )->has_rank; |
878 | if( defined $rank0 && defined $rank1 && $rank1 - $rank0 > 1 ) { |
879 | $variables->{'minlen'} = $rank1 - $rank0; |
880 | } |
881 | |
882 | # Add the calculated edge weights |
b365fbae |
883 | # if( exists $weighted->{$edge->[0]} |
e247aad1 |
884 | # && $weighted->{$edge->[0]} eq $edge->[1] ) { |
885 | # # $variables->{'color'} = 'red'; |
886 | # $variables->{'weight'} = 3.0; |
887 | # } |
30ddc24c |
888 | |
508fd430 |
889 | # EXPERIMENTAL: make edge width reflect no. of witnesses |
890 | my $extrawidth = scalar( $self->path_witnesses( $edge ) ) * 0.2; |
891 | $variables->{'penwidth'} = $extrawidth + 0.8; # gives 1 for a single wit |
892 | |
f13b5582 |
893 | my $varopts = _dot_attr_string( $variables ); |
894 | $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n", |
895 | $edge->[0], $edge->[1], $varopts ); |
3bdec618 |
896 | } elsif( $used{$edge->[0]} ) { |
96ba0418 |
897 | $subend{$edge->[0]} = $edge->[1]; |
3bdec618 |
898 | } elsif( $used{$edge->[1]} ) { |
96ba0418 |
899 | $substart{$edge->[1]} = $edge->[0]; |
b22576c6 |
900 | } |
df6d9812 |
901 | } |
bed6ce83 |
902 | |
903 | # If we are asked to, add relationship links |
904 | if( exists $opts->{show_relations} ) { |
905 | my $filter = $opts->{show_relations}; # can be 'transposition' or 'all' |
906 | if( $filter eq 'transposition' ) { |
907 | $filter =~ qr/^transposition$/; |
908 | } |
909 | foreach my $redge ( $self->relationships ) { |
910 | if( $used{$redge->[0]} && $used{$redge->[1]} ) { |
911 | if( $filter ne 'all' ) { |
912 | my $rel = $self->get_relationship( $redge ); |
913 | next unless $rel->type =~ /$filter/; |
914 | my $variables = { |
915 | arrowhead => 'none', |
916 | color => '#FFA14F', |
917 | constraint => 'false', |
918 | label => uc( substr( $rel->type, 0, 4 ) ), |
919 | penwidth => '3', |
920 | }; |
921 | $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n", |
922 | $redge->[0], $redge->[1], _dot_attr_string( $variables ) ); |
923 | } |
924 | } |
925 | } |
926 | } |
927 | |
3bdec618 |
928 | # Add substitute start and end edges if necessary |
929 | foreach my $node ( keys %substart ) { |
96ba0418 |
930 | my $witstr = $self->_path_display_label ( $self->path_witnesses( $substart{$node}, $node ) ); |
f13b5582 |
931 | my $variables = { %edge_attrs, 'label' => $witstr }; |
96ba0418 |
932 | my $nrdg = $self->reading( $node ); |
933 | if( $nrdg->has_rank && $nrdg->rank > $startrank ) { |
934 | # Substart is actually one lower than $startrank |
935 | $variables->{'minlen'} = $nrdg->rank - ( $startrank - 1 ); |
936 | } |
f13b5582 |
937 | my $varopts = _dot_attr_string( $variables ); |
96ba0418 |
938 | $dot .= "\t\"__SUBSTART__\" -> \"$node\" $varopts;\n"; |
3bdec618 |
939 | } |
940 | foreach my $node ( keys %subend ) { |
96ba0418 |
941 | my $witstr = $self->_path_display_label ( $self->path_witnesses( $node, $subend{$node} ) ); |
f13b5582 |
942 | my $variables = { %edge_attrs, 'label' => $witstr }; |
943 | my $varopts = _dot_attr_string( $variables ); |
96ba0418 |
944 | $dot .= "\t\"$node\" -> \"__SUBEND__\" $varopts;\n"; |
3bdec618 |
945 | } |
b365fbae |
946 | # HACK part 2 |
947 | if( $STRAIGHTENHACK ) { |
43c94341 |
948 | my $endlabel = $endrank ? '__SUBEND__' : '__END__'; |
949 | $dot .= "\t\"$endlabel\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n"; |
b365fbae |
950 | } |
30ddc24c |
951 | |
df6d9812 |
952 | $dot .= "}\n"; |
953 | return $dot; |
954 | } |
955 | |
f13b5582 |
956 | sub _dot_attr_string { |
957 | my( $hash ) = @_; |
958 | my @attrs; |
959 | foreach my $k ( sort keys %$hash ) { |
960 | my $v = $hash->{$k}; |
961 | push( @attrs, $k.'="'.$v.'"' ); |
962 | } |
963 | return( '[ ' . join( ', ', @attrs ) . ' ]' ); |
964 | } |
965 | |
30ddc24c |
966 | sub _add_edge_weights { |
967 | my $self = shift; |
968 | # Walk the graph from START to END, choosing the successor node with |
969 | # the largest number of witness paths each time. |
970 | my $weighted = {}; |
971 | my $curr = $self->start->id; |
008fc8a6 |
972 | my $ranked = $self->end->has_rank; |
30ddc24c |
973 | while( $curr ne $self->end->id ) { |
008fc8a6 |
974 | my $rank = $ranked ? $self->reading( $curr )->rank : 0; |
30ddc24c |
975 | my @succ = sort { $self->path_witnesses( $curr, $a ) |
976 | <=> $self->path_witnesses( $curr, $b ) } |
977 | $self->sequence->successors( $curr ); |
978 | my $next = pop @succ; |
008fc8a6 |
979 | my $nextrank = $ranked ? $self->reading( $next )->rank : 0; |
30ddc24c |
980 | # Try to avoid lacunae in the weighted path. |
008fc8a6 |
981 | while( @succ && |
982 | ( $self->reading( $next )->is_lacuna || |
983 | $nextrank - $rank > 1 ) ){ |
30ddc24c |
984 | $next = pop @succ; |
985 | } |
986 | $weighted->{$curr} = $next; |
987 | $curr = $next; |
988 | } |
989 | return $weighted; |
990 | } |
991 | |
027d819c |
992 | =head2 path_witnesses( $edge ) |
993 | |
994 | Returns the list of sigils whose witnesses are associated with the given edge. |
995 | The edge can be passed as either an array or an arrayref of ( $source, $target ). |
996 | |
997 | =cut |
998 | |
3a2ebbf4 |
999 | sub path_witnesses { |
1000 | my( $self, @edge ) = @_; |
1001 | # If edge is an arrayref, cope. |
1002 | if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) { |
1003 | my $e = shift @edge; |
1004 | @edge = @$e; |
1005 | } |
1006 | my @wits = keys %{$self->sequence->get_edge_attributes( @edge )}; |
508fd430 |
1007 | return @wits; |
3a2ebbf4 |
1008 | } |
1009 | |
7f9f05e8 |
1010 | # Helper function. Make a display label for the given witnesses, showing a.c. |
1011 | # witnesses only where the main witness is not also in the list. |
027d819c |
1012 | sub _path_display_label { |
508fd430 |
1013 | my $self = shift; |
7f9f05e8 |
1014 | my %wits; |
1015 | map { $wits{$_} = 1 } @_; |
1016 | |
1017 | # If an a.c. wit is listed, remove it if the main wit is also listed. |
1018 | # Otherwise keep it for explicit listing. |
1019 | my $aclabel = $self->ac_label; |
1020 | my @disp_ac; |
1021 | foreach my $w ( sort keys %wits ) { |
1022 | if( $w =~ /^(.*)\Q$aclabel\E$/ ) { |
1023 | if( exists $wits{$1} ) { |
1024 | delete $wits{$w}; |
1025 | } else { |
1026 | push( @disp_ac, $w ); |
1027 | } |
1028 | } |
1029 | } |
1030 | |
1031 | # See if we are in a majority situation. |
8f9cab7b |
1032 | my $maj = scalar( $self->tradition->witnesses ) * 0.6; |
1ff82d4f |
1033 | $maj = $maj > 5 ? $maj : 5; |
7f9f05e8 |
1034 | if( scalar keys %wits > $maj ) { |
1035 | unshift( @disp_ac, 'majority' ); |
1036 | return join( ', ', @disp_ac ); |
8f9cab7b |
1037 | } else { |
7f9f05e8 |
1038 | return join( ', ', sort keys %wits ); |
8f9cab7b |
1039 | } |
1040 | } |
1dd07bda |
1041 | |
bf6e338d |
1042 | =head2 readings_at_rank( $rank ) |
1dd07bda |
1043 | |
bf6e338d |
1044 | Returns a list of readings at a given rank, taken from the alignment table. |
1dd07bda |
1045 | |
1046 | =cut |
1047 | |
bf6e338d |
1048 | sub readings_at_rank { |
1dd07bda |
1049 | my( $self, $rank ) = @_; |
bf6e338d |
1050 | my $table = $self->alignment_table; |
1051 | # Table rank is real rank - 1. |
1052 | my @elements = map { $_->{'tokens'}->[$rank-1] } @{$table->{'alignment'}}; |
1053 | my %readings; |
1054 | foreach my $e ( @elements ) { |
1055 | next unless ref( $e ) eq 'HASH'; |
1056 | next unless exists $e->{'t'}; |
1057 | $readings{$e->{'t'}->id} = $e->{'t'}; |
1058 | } |
1059 | return values %readings; |
1dd07bda |
1060 | } |
8f9cab7b |
1061 | |
4e5a7b2c |
1062 | =head2 as_graphml |
8e1394aa |
1063 | |
4e5a7b2c |
1064 | Returns a GraphML representation of the collation. The GraphML will contain |
1065 | two graphs. The first expresses the attributes of the readings and the witness |
1066 | paths that link them; the second expresses the relationships that link the |
1067 | readings. This is the native transfer format for a tradition. |
8e1394aa |
1068 | |
56eefa04 |
1069 | =begin testing |
1070 | |
1071 | use Text::Tradition; |
951ddfe8 |
1072 | use TryCatch; |
56eefa04 |
1073 | |
1074 | my $READINGS = 311; |
1075 | my $PATHS = 361; |
1076 | |
1077 | my $datafile = 't/data/florilegium_tei_ps.xml'; |
1078 | my $tradition = Text::Tradition->new( 'input' => 'TEI', |
1079 | 'name' => 'test0', |
1080 | 'file' => $datafile, |
1081 | 'linear' => 1 ); |
1082 | |
1083 | ok( $tradition, "Got a tradition object" ); |
1084 | is( scalar $tradition->witnesses, 13, "Found all witnesses" ); |
1085 | ok( $tradition->collation, "Tradition has a collation" ); |
1086 | |
1087 | my $c = $tradition->collation; |
1088 | is( scalar $c->readings, $READINGS, "Collation has all readings" ); |
1089 | is( scalar $c->paths, $PATHS, "Collation has all paths" ); |
1090 | is( scalar $c->relationships, 0, "Collation has all relationships" ); |
1091 | |
1092 | # Add a few relationships |
1093 | $c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } ); |
1094 | $c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } ); |
1095 | $c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } ); |
1096 | |
1097 | # Now write it to GraphML and parse it again. |
1098 | |
1099 | my $graphml = $c->as_graphml; |
1100 | my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml ); |
1101 | is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" ); |
1102 | is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" ); |
1103 | is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" ); |
1104 | |
9fef629b |
1105 | # Now add a stemma, write to GraphML, and look at the output. |
951ddfe8 |
1106 | SKIP: { |
37bf09f4 |
1107 | skip "Analysis module not present", 3 unless $tradition->can( 'add_stemma' ); |
951ddfe8 |
1108 | my $stemma = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' ); |
1109 | is( ref( $stemma ), 'Text::Tradition::Stemma', "Parsed dotfile into stemma" ); |
1110 | is( $tradition->stemmata, 1, "Tradition now has the stemma" ); |
1111 | $graphml = $c->as_graphml; |
1112 | like( $graphml, qr/digraph/, "Digraph declaration exists in GraphML" ); |
1113 | } |
2a812726 |
1114 | |
56eefa04 |
1115 | =end testing |
1116 | |
8e1394aa |
1117 | =cut |
1118 | |
a445ce40 |
1119 | ## TODO MOVE this to Tradition.pm and modularize it better |
8e1394aa |
1120 | sub as_graphml { |
a30ca502 |
1121 | my( $self, $options ) = @_; |
3d14b48e |
1122 | $self->calculate_ranks unless $self->_graphcalc_done; |
1123 | |
a30ca502 |
1124 | my $start = $options->{'from'} |
1125 | ? $self->reading( $options->{'from'} ) : $self->start; |
1126 | my $end = $options->{'to'} |
1127 | ? $self->reading( $options->{'to'} ) : $self->end; |
1128 | if( $start->has_rank && $end->has_rank && $end->rank < $start->rank ) { |
1129 | throw( 'Start node must be before end node' ); |
1130 | } |
1131 | # The readings need to be ranked for this to work. |
1132 | $start = $self->start unless $start->has_rank; |
1133 | $end = $self->end unless $end->has_rank; |
414cc046 |
1134 | my $rankoffset = 0; |
1135 | unless( $start eq $self->start ) { |
1136 | $rankoffset = $start->rank - 1; |
1137 | } |
a30ca502 |
1138 | my %use_readings; |
1139 | |
8e1394aa |
1140 | # Some namespaces |
1141 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
1142 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
1143 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
910a0a6d |
1144 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
8e1394aa |
1145 | |
1146 | # Create the document and root node |
428bcf0b |
1147 | require XML::LibXML; |
8e1394aa |
1148 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
1149 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
1150 | $graphml->setDocumentElement( $root ); |
1151 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
1152 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
bbd064a9 |
1153 | |
1154 | # List of attribute types to save on our objects and their corresponding |
1155 | # GraphML types |
1156 | my %save_types = ( |
1157 | 'Str' => 'string', |
1158 | 'Int' => 'int', |
1159 | 'Bool' => 'boolean', |
10e4b1ac |
1160 | 'ReadingID' => 'string', |
bbd064a9 |
1161 | 'RelationshipType' => 'string', |
1162 | 'RelationshipScope' => 'string', |
1163 | ); |
1164 | |
bbd064a9 |
1165 | # Add the data keys for the graph. Include an extra key 'version' for the |
1166 | # GraphML output version. |
e309421a |
1167 | my %graph_data_keys; |
1168 | my $gdi = 0; |
bbd064a9 |
1169 | my %graph_attributes = ( 'version' => 'string' ); |
1170 | # Graph attributes include those of Tradition and those of Collation. |
1171 | my %gattr_from; |
f97ef19e |
1172 | # TODO Use meta introspection method from duplicate_reading to do this |
1173 | # instead of naming custom keys. |
bbd064a9 |
1174 | my $tmeta = $self->tradition->meta; |
1175 | my $cmeta = $self->meta; |
1176 | map { $gattr_from{$_->name} = 'Tradition' } $tmeta->get_all_attributes; |
1177 | map { $gattr_from{$_->name} = 'Collation' } $cmeta->get_all_attributes; |
1178 | foreach my $attr ( ( $tmeta->get_all_attributes, $cmeta->get_all_attributes ) ) { |
1179 | next if $attr->name =~ /^_/; |
bbd064a9 |
1180 | next unless $save_types{$attr->type_constraint->name}; |
1181 | $graph_attributes{$attr->name} = $save_types{$attr->type_constraint->name}; |
1182 | } |
9fef629b |
1183 | # Extra custom keys for complex objects that should be saved in some form. |
1184 | # The subroutine should return a string, or undef/empty. |
951ddfe8 |
1185 | if( $tmeta->has_method('stemmata') ) { |
1186 | $graph_attributes{'stemmata'} = sub { |
1187 | my @stemstrs; |
1188 | map { push( @stemstrs, $_->editable( {linesep => ''} ) ) } |
1189 | $self->tradition->stemmata; |
1190 | join( "\n", @stemstrs ); |
1191 | }; |
1192 | } |
1193 | |
8943ff68 |
1194 | if( $tmeta->has_method('user') ) { |
1195 | $graph_attributes{'user'} = sub { |
1196 | $self->tradition->user ? $self->tradition->user->id : undef |
1197 | }; |
1198 | } |
bbd064a9 |
1199 | |
1200 | foreach my $datum ( sort keys %graph_attributes ) { |
e309421a |
1201 | $graph_data_keys{$datum} = 'dg'.$gdi++; |
1202 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
9fef629b |
1203 | my $dtype = ref( $graph_attributes{$datum} ) ? 'string' |
1204 | : $graph_attributes{$datum}; |
e309421a |
1205 | $key->setAttribute( 'attr.name', $datum ); |
9fef629b |
1206 | $key->setAttribute( 'attr.type', $dtype ); |
e309421a |
1207 | $key->setAttribute( 'for', 'graph' ); |
1208 | $key->setAttribute( 'id', $graph_data_keys{$datum} ); |
1209 | } |
f6066bac |
1210 | |
bbd064a9 |
1211 | # Add the data keys for reading nodes |
1212 | my %reading_attributes; |
1213 | my $rmeta = Text::Tradition::Collation::Reading->meta; |
1214 | foreach my $attr( $rmeta->get_all_attributes ) { |
1215 | next if $attr->name =~ /^_/; |
bbd064a9 |
1216 | next unless $save_types{$attr->type_constraint->name}; |
1217 | $reading_attributes{$attr->name} = $save_types{$attr->type_constraint->name}; |
1218 | } |
a445ce40 |
1219 | if( $self->start->does('Text::Tradition::Morphology' ) ) { |
1220 | # Extra custom key for the reading morphology |
1221 | $reading_attributes{'lexemes'} = 'string'; |
1222 | } |
7cd9f181 |
1223 | |
ef9d481f |
1224 | my %node_data_keys; |
1225 | my $ndi = 0; |
bbd064a9 |
1226 | foreach my $datum ( sort keys %reading_attributes ) { |
910a0a6d |
1227 | $node_data_keys{$datum} = 'dn'.$ndi++; |
1228 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
1229 | $key->setAttribute( 'attr.name', $datum ); |
bbd064a9 |
1230 | $key->setAttribute( 'attr.type', $reading_attributes{$datum} ); |
910a0a6d |
1231 | $key->setAttribute( 'for', 'node' ); |
1232 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
1233 | } |
1234 | |
bbd064a9 |
1235 | # Add the data keys for edges, that is, paths and relationships. Path |
1236 | # data does not come from a Moose class so is here manually. |
ef9d481f |
1237 | my $edi = 0; |
1238 | my %edge_data_keys; |
bbd064a9 |
1239 | my %edge_attributes = ( |
3a2ebbf4 |
1240 | witness => 'string', # ID/label for a path |
3a2ebbf4 |
1241 | extra => 'boolean', # Path key |
3a2ebbf4 |
1242 | ); |
bbd064a9 |
1243 | my @path_attributes = keys %edge_attributes; # track our manual additions |
1244 | my $pmeta = Text::Tradition::Collation::Relationship->meta; |
1245 | foreach my $attr( $pmeta->get_all_attributes ) { |
1246 | next if $attr->name =~ /^_/; |
bbd064a9 |
1247 | next unless $save_types{$attr->type_constraint->name}; |
1248 | $edge_attributes{$attr->name} = $save_types{$attr->type_constraint->name}; |
1249 | } |
1250 | foreach my $datum ( sort keys %edge_attributes ) { |
3a2ebbf4 |
1251 | $edge_data_keys{$datum} = 'de'.$edi++; |
910a0a6d |
1252 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
3a2ebbf4 |
1253 | $key->setAttribute( 'attr.name', $datum ); |
bbd064a9 |
1254 | $key->setAttribute( 'attr.type', $edge_attributes{$datum} ); |
910a0a6d |
1255 | $key->setAttribute( 'for', 'edge' ); |
3a2ebbf4 |
1256 | $key->setAttribute( 'id', $edge_data_keys{$datum} ); |
8e1394aa |
1257 | } |
3a2ebbf4 |
1258 | |
cc31ebaa |
1259 | # Add the collation graph itself. First, sanitize the name to a valid XML ID. |
1260 | my $xmlidname = $self->tradition->name; |
1261 | $xmlidname =~ s/(?!$xml10_namechar_rx)./_/g; |
1262 | if( $xmlidname !~ /^$xml10_namestartchar_rx/ ) { |
1263 | $xmlidname = '_'.$xmlidname; |
1264 | } |
2c669bca |
1265 | my $sgraph = $root->addNewChild( $graphml_ns, 'graph' ); |
1266 | $sgraph->setAttribute( 'edgedefault', 'directed' ); |
cc31ebaa |
1267 | $sgraph->setAttribute( 'id', $xmlidname ); |
2c669bca |
1268 | $sgraph->setAttribute( 'parse.edgeids', 'canonical' ); |
cc31ebaa |
1269 | $sgraph->setAttribute( 'parse.edges', 0 ); # fill in later |
2c669bca |
1270 | $sgraph->setAttribute( 'parse.nodeids', 'canonical' ); |
cc31ebaa |
1271 | $sgraph->setAttribute( 'parse.nodes', 0 ); # fill in later |
2c669bca |
1272 | $sgraph->setAttribute( 'parse.order', 'nodesfirst' ); |
22222af9 |
1273 | |
2a812726 |
1274 | # Tradition/collation attribute data |
bbd064a9 |
1275 | foreach my $datum ( keys %graph_attributes ) { |
1276 | my $value; |
1277 | if( $datum eq 'version' ) { |
2a812726 |
1278 | $value = '3.2'; |
9fef629b |
1279 | } elsif( ref( $graph_attributes{$datum} ) ) { |
1280 | my $sub = $graph_attributes{$datum}; |
1281 | $value = &$sub(); |
bbd064a9 |
1282 | } elsif( $gattr_from{$datum} eq 'Tradition' ) { |
1283 | $value = $self->tradition->$datum; |
1284 | } else { |
1285 | $value = $self->$datum; |
1286 | } |
2c669bca |
1287 | _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value ); |
e309421a |
1288 | } |
8e1394aa |
1289 | |
1290 | my $node_ctr = 0; |
1291 | my %node_hash; |
22222af9 |
1292 | # Add our readings to the graph |
3a2ebbf4 |
1293 | foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) { |
a30ca502 |
1294 | next if $n->has_rank && $n ne $self->start && $n ne $self->end && |
1295 | ( $n->rank < $start->rank || $n->rank > $end->rank ); |
1296 | $use_readings{$n->id} = 1; |
2c669bca |
1297 | # Add to the main graph |
1298 | my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' ); |
910a0a6d |
1299 | my $node_xmlid = 'n' . $node_ctr++; |
3a2ebbf4 |
1300 | $node_hash{ $n->id } = $node_xmlid; |
910a0a6d |
1301 | $node_el->setAttribute( 'id', $node_xmlid ); |
bbd064a9 |
1302 | foreach my $d ( keys %reading_attributes ) { |
255875b8 |
1303 | my $nval = $n->$d; |
7cd9f181 |
1304 | # Custom serialization |
1305 | if( $d eq 'lexemes' ) { |
1306 | # If nval is a true value, we have lexemes so we need to |
1307 | # serialize them. Otherwise set nval to undef so that the |
1308 | # key is excluded from this reading. |
1309 | $nval = $nval ? $n->_serialize_lexemes : undef; |
18c64d55 |
1310 | } elsif( $d eq 'normal_form' && $n->normal_form eq $n->text ) { |
1311 | $nval = undef; |
7cd9f181 |
1312 | } |
cc31ebaa |
1313 | if( $rankoffset && $d eq 'rank' && $n ne $self->start ) { |
414cc046 |
1314 | # Adjust the ranks within the subgraph. |
cc31ebaa |
1315 | $nval = $n eq $self->end ? $end->rank - $rankoffset + 1 |
1316 | : $nval - $rankoffset; |
414cc046 |
1317 | } |
255875b8 |
1318 | _add_graphml_data( $node_el, $node_data_keys{$d}, $nval ) |
1319 | if defined $nval; |
1320 | } |
b15511bf |
1321 | } |
1322 | |
2c669bca |
1323 | # Add the path edges to the sequence graph |
df6d9812 |
1324 | my $edge_ctr = 0; |
3a2ebbf4 |
1325 | foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) { |
1326 | # We add an edge in the graphml for every witness in $e. |
a30ca502 |
1327 | next unless( $use_readings{$e->[0]} || $use_readings{$e->[1]} ); |
1328 | my @edge_wits = sort $self->path_witnesses( $e ); |
cc31ebaa |
1329 | $e->[0] = $self->start->id unless $use_readings{$e->[0]}; |
1330 | $e->[1] = $self->end->id unless $use_readings{$e->[1]}; |
1331 | # Skip any path from start to end; that witness is not in the subgraph. |
1332 | next if ( $e->[0] eq $self->start->id && $e->[1] eq $self->end->id ); |
a30ca502 |
1333 | foreach my $wit ( @edge_wits ) { |
3a2ebbf4 |
1334 | my( $id, $from, $to ) = ( 'e'.$edge_ctr++, |
1335 | $node_hash{ $e->[0] }, |
1336 | $node_hash{ $e->[1] } ); |
2c669bca |
1337 | my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' ); |
3a2ebbf4 |
1338 | $edge_el->setAttribute( 'source', $from ); |
1339 | $edge_el->setAttribute( 'target', $to ); |
1340 | $edge_el->setAttribute( 'id', $id ); |
3a2ebbf4 |
1341 | |
1342 | # It's a witness path, so add the witness |
1343 | my $base = $wit; |
1344 | my $key = $edge_data_keys{'witness'}; |
1345 | # Is this an ante-corr witness? |
1346 | my $aclabel = $self->ac_label; |
1347 | if( $wit =~ /^(.*)\Q$aclabel\E$/ ) { |
1348 | # Keep the base witness |
1349 | $base = $1; |
1350 | # ...and record that this is an 'extra' reading path |
1351 | _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel ); |
1352 | } |
1353 | _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base ); |
1354 | } |
1355 | } |
1356 | |
cc31ebaa |
1357 | # Report the actual number of nodes and edges that went in |
1358 | $sgraph->setAttribute( 'parse.edges', $edge_ctr ); |
1359 | $sgraph->setAttribute( 'parse.nodes', $node_ctr ); |
1360 | |
22222af9 |
1361 | # Add the relationship graph to the XML |
bbd064a9 |
1362 | map { delete $edge_data_keys{$_} } @path_attributes; |
826d8773 |
1363 | $self->relations->_as_graphml( $graphml_ns, $root, \%node_hash, |
1364 | $node_data_keys{'id'}, \%edge_data_keys ); |
8e1394aa |
1365 | |
94c00c71 |
1366 | # Save and return the thing |
1367 | my $result = decode_utf8( $graphml->toString(1) ); |
94c00c71 |
1368 | return $result; |
df6d9812 |
1369 | } |
1370 | |
b15511bf |
1371 | sub _add_graphml_data { |
1372 | my( $el, $key, $value ) = @_; |
b15511bf |
1373 | return unless defined $value; |
c9bf3dbf |
1374 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
b15511bf |
1375 | $data_el->setAttribute( 'key', $key ); |
1376 | $data_el->appendText( $value ); |
8e1394aa |
1377 | } |
1378 | |
4e5a7b2c |
1379 | =head2 as_csv |
910a0a6d |
1380 | |
1381 | Returns a CSV alignment table representation of the collation graph, one |
2c669bca |
1382 | row per witness (or witness uncorrected.) |
910a0a6d |
1383 | |
1384 | =cut |
1385 | |
1386 | sub as_csv { |
3a2ebbf4 |
1387 | my( $self ) = @_; |
1dd07bda |
1388 | my $table = $self->alignment_table; |
82fa4d57 |
1389 | my $csv = Text::CSV->new( { binary => 1, quote_null => 0 } ); |
910a0a6d |
1390 | my @result; |
2c669bca |
1391 | # Make the header row |
1392 | $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} ); |
1393 | push( @result, decode_utf8( $csv->string ) ); |
1394 | # Make the rest of the rows |
1395 | foreach my $idx ( 0 .. $table->{'length'} - 1 ) { |
566f4595 |
1396 | my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}}; |
1dd07bda |
1397 | my @row = map { $_ ? $_->{'t'}->text : $_ } @rowobjs; |
2c669bca |
1398 | $csv->combine( @row ); |
910a0a6d |
1399 | push( @result, decode_utf8( $csv->string ) ); |
1400 | } |
3a2ebbf4 |
1401 | return join( "\n", @result ); |
910a0a6d |
1402 | } |
1403 | |
248276a2 |
1404 | =head2 alignment_table |
2c669bca |
1405 | |
566f4595 |
1406 | Return a reference to an alignment table, in a slightly enhanced CollateX |
1407 | format which looks like this: |
1408 | |
1409 | $table = { alignment => [ { witness => "SIGIL", |
4e5a7b2c |
1410 | tokens => [ { t => "TEXT" }, ... ] }, |
566f4595 |
1411 | { witness => "SIG2", |
4e5a7b2c |
1412 | tokens => [ { t => "TEXT" }, ... ] }, |
566f4595 |
1413 | ... ], |
1414 | length => TEXTLEN }; |
1415 | |
2c669bca |
1416 | =cut |
9f3ba6f7 |
1417 | |
1dd07bda |
1418 | sub alignment_table { |
1419 | my( $self ) = @_; |
1dd07bda |
1420 | return $self->cached_table if $self->has_cached_table; |
1421 | |
0ecb975c |
1422 | # Make sure we can do this |
1423 | throw( "Need a linear graph in order to make an alignment table" ) |
1424 | unless $self->linear; |
b39e7cb5 |
1425 | $self->calculate_ranks() |
1426 | unless $self->_graphcalc_done && $self->end->has_rank; |
1427 | |
2c669bca |
1428 | my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 }; |
3a2ebbf4 |
1429 | my @all_pos = ( 1 .. $self->end->rank - 1 ); |
68454b71 |
1430 | foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) { |
6771a1b1 |
1431 | # say STDERR "Making witness row(s) for " . $wit->sigil; |
1f7aa795 |
1432 | my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil ); |
1dd07bda |
1433 | my @row = _make_witness_row( \@wit_path, \@all_pos ); |
bed6ce83 |
1434 | my $witobj = { 'witness' => $wit->sigil, 'tokens' => \@row }; |
1435 | $witobj->{'identifier'} = $wit->identifier if $wit->identifier; |
1436 | push( @{$table->{'alignment'}}, $witobj ); |
1f7aa795 |
1437 | if( $wit->is_layered ) { |
1438 | my @wit_ac_path = $self->reading_sequence( $self->start, $self->end, |
861c3e27 |
1439 | $wit->sigil.$self->ac_label ); |
1dd07bda |
1440 | my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos ); |
bed6ce83 |
1441 | my $witacobj = { 'witness' => $wit->sigil.$self->ac_label, |
1442 | 'tokens' => \@ac_row }; |
1443 | $witacobj->{'identifier'} = $wit->identifier if $wit->identifier; |
1444 | push( @{$table->{'alignment'}}, $witacobj ); |
910a0a6d |
1445 | } |
1446 | } |
1dd07bda |
1447 | $self->cached_table( $table ); |
1448 | return $table; |
910a0a6d |
1449 | } |
1450 | |
1451 | sub _make_witness_row { |
1dd07bda |
1452 | my( $path, $positions ) = @_; |
910a0a6d |
1453 | my %char_hash; |
1454 | map { $char_hash{$_} = undef } @$positions; |
2c669bca |
1455 | my $debug = 0; |
910a0a6d |
1456 | foreach my $rdg ( @$path ) { |
6771a1b1 |
1457 | say STDERR "rank " . $rdg->rank if $debug; |
1458 | # say STDERR "No rank for " . $rdg->id unless defined $rdg->rank; |
1dd07bda |
1459 | $char_hash{$rdg->rank} = { 't' => $rdg }; |
910a0a6d |
1460 | } |
1461 | my @row = map { $char_hash{$_} } @$positions; |
eca16057 |
1462 | # Fill in lacuna markers for undef spots in the row |
1463 | my $last_el = shift @row; |
1464 | my @filled_row = ( $last_el ); |
1465 | foreach my $el ( @row ) { |
0e476982 |
1466 | # If we are using node reference, make the lacuna node appear many times |
1467 | # in the table. If not, use the lacuna tag. |
1dd07bda |
1468 | if( $last_el && $last_el->{'t'}->is_lacuna && !defined $el ) { |
1469 | $el = $last_el; |
eca16057 |
1470 | } |
1471 | push( @filled_row, $el ); |
1472 | $last_el = $el; |
1473 | } |
1474 | return @filled_row; |
910a0a6d |
1475 | } |
1476 | |
248276a2 |
1477 | |
4e5a7b2c |
1478 | =head1 NAVIGATION METHODS |
910a0a6d |
1479 | |
4e5a7b2c |
1480 | =head2 reading_sequence( $first, $last, $sigil, $backup ) |
e2902068 |
1481 | |
1482 | Returns the ordered list of readings, starting with $first and ending |
4e5a7b2c |
1483 | with $last, for the witness given in $sigil. If a $backup sigil is |
1484 | specified (e.g. when walking a layered witness), it will be used wherever |
1485 | no $sigil path exists. If there is a base text reading, that will be |
1486 | used wherever no path exists for $sigil or $backup. |
e2902068 |
1487 | |
1488 | =cut |
1489 | |
910a0a6d |
1490 | # TODO Think about returning some lazy-eval iterator. |
b0b4421a |
1491 | # TODO Get rid of backup; we should know from what witness is whether we need it. |
910a0a6d |
1492 | |
e2902068 |
1493 | sub reading_sequence { |
861c3e27 |
1494 | my( $self, $start, $end, $witness ) = @_; |
e2902068 |
1495 | |
930ff666 |
1496 | $witness = $self->baselabel unless $witness; |
e2902068 |
1497 | my @readings = ( $start ); |
1498 | my %seen; |
1499 | my $n = $start; |
3a2ebbf4 |
1500 | while( $n && $n->id ne $end->id ) { |
1501 | if( exists( $seen{$n->id} ) ) { |
63778331 |
1502 | throw( "Detected loop for $witness at " . $n->id ); |
910a0a6d |
1503 | } |
3a2ebbf4 |
1504 | $seen{$n->id} = 1; |
910a0a6d |
1505 | |
861c3e27 |
1506 | my $next = $self->next_reading( $n, $witness ); |
44771cf2 |
1507 | unless( $next ) { |
63778331 |
1508 | throw( "Did not find any path for $witness from reading " . $n->id ); |
44771cf2 |
1509 | } |
910a0a6d |
1510 | push( @readings, $next ); |
1511 | $n = $next; |
e2902068 |
1512 | } |
1513 | # Check that the last reading is our end reading. |
1514 | my $last = $readings[$#readings]; |
63778331 |
1515 | throw( "Last reading found from " . $start->text . |
1516 | " for witness $witness is not the end!" ) # TODO do we get this far? |
3a2ebbf4 |
1517 | unless $last->id eq $end->id; |
e2902068 |
1518 | |
1519 | return @readings; |
1520 | } |
1521 | |
4e5a7b2c |
1522 | =head2 next_reading( $reading, $sigil ); |
8e1394aa |
1523 | |
4a8828f0 |
1524 | Returns the reading that follows the given reading along the given witness |
930ff666 |
1525 | path. |
8e1394aa |
1526 | |
1527 | =cut |
1528 | |
4a8828f0 |
1529 | sub next_reading { |
e2902068 |
1530 | # Return the successor via the corresponding path. |
8e1394aa |
1531 | my $self = shift; |
3a2ebbf4 |
1532 | my $answer = $self->_find_linked_reading( 'next', @_ ); |
2c669bca |
1533 | return undef unless $answer; |
3a2ebbf4 |
1534 | return $self->reading( $answer ); |
8e1394aa |
1535 | } |
1536 | |
4e5a7b2c |
1537 | =head2 prior_reading( $reading, $sigil ) |
8e1394aa |
1538 | |
4a8828f0 |
1539 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
1540 | path. |
8e1394aa |
1541 | |
1542 | =cut |
1543 | |
4a8828f0 |
1544 | sub prior_reading { |
e2902068 |
1545 | # Return the predecessor via the corresponding path. |
8e1394aa |
1546 | my $self = shift; |
3a2ebbf4 |
1547 | my $answer = $self->_find_linked_reading( 'prior', @_ ); |
1548 | return $self->reading( $answer ); |
8e1394aa |
1549 | } |
1550 | |
4a8828f0 |
1551 | sub _find_linked_reading { |
861c3e27 |
1552 | my( $self, $direction, $node, $path ) = @_; |
1553 | |
1554 | # Get a backup if we are dealing with a layered witness |
1555 | my $alt_path; |
1556 | my $aclabel = $self->ac_label; |
1557 | if( $path && $path =~ /^(.*)\Q$aclabel\E$/ ) { |
1558 | $alt_path = $1; |
1559 | } |
1560 | |
e2902068 |
1561 | my @linked_paths = $direction eq 'next' |
3a2ebbf4 |
1562 | ? $self->sequence->edges_from( $node ) |
1563 | : $self->sequence->edges_to( $node ); |
e2902068 |
1564 | return undef unless scalar( @linked_paths ); |
8e1394aa |
1565 | |
e2902068 |
1566 | # We have to find the linked path that contains all of the |
1567 | # witnesses supplied in $path. |
1568 | my( @path_wits, @alt_path_wits ); |
4e5a7b2c |
1569 | @path_wits = sort( $self->_witnesses_of_label( $path ) ) if $path; |
1570 | @alt_path_wits = sort( $self->_witnesses_of_label( $alt_path ) ) if $alt_path; |
e2902068 |
1571 | my $base_le; |
1572 | my $alt_le; |
1573 | foreach my $le ( @linked_paths ) { |
3a2ebbf4 |
1574 | if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) { |
910a0a6d |
1575 | $base_le = $le; |
910a0a6d |
1576 | } |
508fd430 |
1577 | my @le_wits = sort $self->path_witnesses( $le ); |
3a2ebbf4 |
1578 | if( _is_within( \@path_wits, \@le_wits ) ) { |
1579 | # This is the right path. |
1580 | return $direction eq 'next' ? $le->[1] : $le->[0]; |
1581 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
1582 | $alt_le = $le; |
1583 | } |
8e1394aa |
1584 | } |
e2902068 |
1585 | # Got this far? Return the alternate path if it exists. |
3a2ebbf4 |
1586 | return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0] |
910a0a6d |
1587 | if $alt_le; |
e2902068 |
1588 | |
1589 | # Got this far? Return the base path if it exists. |
3a2ebbf4 |
1590 | return $direction eq 'next' ? $base_le->[1] : $base_le->[0] |
910a0a6d |
1591 | if $base_le; |
e2902068 |
1592 | |
1593 | # Got this far? We have no appropriate path. |
2c669bca |
1594 | warn "Could not find $direction node from " . $node->id |
910a0a6d |
1595 | . " along path $path"; |
8e1394aa |
1596 | return undef; |
1597 | } |
1598 | |
4a8828f0 |
1599 | # Some set logic. |
1600 | sub _is_within { |
1601 | my( $set1, $set2 ) = @_; |
7854e12e |
1602 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
1603 | foreach my $el ( @$set1 ) { |
910a0a6d |
1604 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
4a8828f0 |
1605 | } |
1606 | return $ret; |
1607 | } |
1608 | |
4e5a7b2c |
1609 | # Return the string that joins together a list of witnesses for |
1610 | # display on a single path. |
1611 | sub _witnesses_of_label { |
1612 | my( $self, $label ) = @_; |
1613 | my $regex = $self->wit_list_separator; |
1614 | my @answer = split( /\Q$regex\E/, $label ); |
1615 | return @answer; |
b0b4421a |
1616 | } |
1617 | |
d4b75f44 |
1618 | =head2 common_readings |
1619 | |
1620 | Returns the list of common readings in the graph (i.e. those readings that are |
1621 | shared by all non-lacunose witnesses.) |
1622 | |
1623 | =cut |
1624 | |
1625 | sub common_readings { |
1626 | my $self = shift; |
1627 | my @common = grep { $_->is_common } $self->readings; |
1628 | return @common; |
1629 | } |
1630 | |
fae52efd |
1631 | =head2 path_text( $sigil, [, $start, $end ] ) |
b0b4421a |
1632 | |
1633 | Returns the text of a witness (plus its backup, if we are using a layer) |
1634 | as stored in the collation. The text is returned as a string, where the |
1635 | individual readings are joined with spaces and the meta-readings (e.g. |
1636 | lacunae) are omitted. Optional specification of $start and $end allows |
1637 | the generation of a subset of the witness text. |
4e5a7b2c |
1638 | |
b0b4421a |
1639 | =cut |
1640 | |
1641 | sub path_text { |
861c3e27 |
1642 | my( $self, $wit, $start, $end ) = @_; |
b0b4421a |
1643 | $start = $self->start unless $start; |
1644 | $end = $self->end unless $end; |
861c3e27 |
1645 | my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit ); |
629e27b0 |
1646 | my $pathtext = ''; |
1647 | my $last; |
1648 | foreach my $r ( @path ) { |
6ad2ce78 |
1649 | unless ( $r->join_prior || !$last || $last->join_next ) { |
1650 | $pathtext .= ' '; |
1651 | } |
1652 | $pathtext .= $r->text; |
629e27b0 |
1653 | $last = $r; |
1654 | } |
1655 | return $pathtext; |
b0b4421a |
1656 | } |
4e5a7b2c |
1657 | |
1658 | =head1 INITIALIZATION METHODS |
1659 | |
1660 | These are mostly for use by parsers. |
1661 | |
1662 | =head2 make_witness_path( $witness ) |
1663 | |
1664 | Link the array of readings contained in $witness->path (and in |
1665 | $witness->uncorrected_path if it exists) into collation paths. |
1666 | Clear out the arrays when finished. |
de51424a |
1667 | |
4e5a7b2c |
1668 | =head2 make_witness_paths |
1669 | |
1670 | Call make_witness_path for all witnesses in the tradition. |
1671 | |
1672 | =cut |
930ff666 |
1673 | |
7e450e44 |
1674 | # For use when a collation is constructed from a base text and an apparatus. |
1675 | # We have the sequences of readings and just need to add path edges. |
1f7aa795 |
1676 | # When we are done, clear out the witness path attributes, as they are no |
1677 | # longer needed. |
1678 | # TODO Find a way to replace the witness path attributes with encapsulated functions? |
e2902068 |
1679 | |
6a222840 |
1680 | sub make_witness_paths { |
1681 | my( $self ) = @_; |
910a0a6d |
1682 | foreach my $wit ( $self->tradition->witnesses ) { |
6771a1b1 |
1683 | # say STDERR "Making path for " . $wit->sigil; |
910a0a6d |
1684 | $self->make_witness_path( $wit ); |
7854e12e |
1685 | } |
7854e12e |
1686 | } |
1687 | |
6a222840 |
1688 | sub make_witness_path { |
7854e12e |
1689 | my( $self, $wit ) = @_; |
1690 | my @chain = @{$wit->path}; |
15d2d3df |
1691 | my $sig = $wit->sigil; |
fae52efd |
1692 | # Add start and end if necessary |
1693 | unshift( @chain, $self->start ) unless $chain[0] eq $self->start; |
1694 | push( @chain, $self->end ) unless $chain[-1] eq $self->end; |
7854e12e |
1695 | foreach my $idx ( 0 .. $#chain-1 ) { |
910a0a6d |
1696 | $self->add_path( $chain[$idx], $chain[$idx+1], $sig ); |
7854e12e |
1697 | } |
1f7aa795 |
1698 | if( $wit->is_layered ) { |
d9e873d0 |
1699 | @chain = @{$wit->uncorrected_path}; |
fae52efd |
1700 | unshift( @chain, $self->start ) unless $chain[0] eq $self->start; |
1701 | push( @chain, $self->end ) unless $chain[-1] eq $self->end; |
d9e873d0 |
1702 | foreach my $idx( 0 .. $#chain-1 ) { |
1703 | my $source = $chain[$idx]; |
1704 | my $target = $chain[$idx+1]; |
1705 | $self->add_path( $source, $target, $sig.$self->ac_label ) |
1706 | unless $self->has_path( $source, $target, $sig ); |
1707 | } |
15d2d3df |
1708 | } |
1f7aa795 |
1709 | $wit->clear_path; |
1710 | $wit->clear_uncorrected_path; |
e2902068 |
1711 | } |
1712 | |
4e5a7b2c |
1713 | =head2 calculate_ranks |
1714 | |
1715 | Calculate the reading ranks (that is, their aligned positions relative |
1716 | to each other) for the graph. This can only be called on linear collations. |
1717 | |
b365fbae |
1718 | =begin testing |
1719 | |
1720 | use Text::Tradition; |
1721 | |
1722 | my $cxfile = 't/data/Collatex-16.xml'; |
1723 | my $t = Text::Tradition->new( |
1724 | 'name' => 'inline', |
1725 | 'input' => 'CollateX', |
1726 | 'file' => $cxfile, |
1727 | ); |
1728 | my $c = $t->collation; |
1729 | |
1730 | # Make an svg |
bfcbcecb |
1731 | my $table = $c->alignment_table; |
1732 | ok( $c->has_cached_table, "Alignment table was cached" ); |
1733 | is( $c->alignment_table, $table, "Cached table returned upon second call" ); |
b365fbae |
1734 | $c->calculate_ranks; |
bfcbcecb |
1735 | is( $c->alignment_table, $table, "Cached table retained with no rank change" ); |
864ee4bf |
1736 | $c->add_relationship( 'n13', 'n23', { type => 'repetition' } ); |
1737 | is( $c->alignment_table, $table, "Alignment table unchanged after non-colo relationship add" ); |
1738 | $c->add_relationship( 'n24', 'n23', { type => 'spelling' } ); |
1739 | isnt( $c->alignment_table, $table, "Alignment table changed after colo relationship add" ); |
b365fbae |
1740 | |
1741 | =end testing |
1742 | |
4e5a7b2c |
1743 | =cut |
1744 | |
910a0a6d |
1745 | sub calculate_ranks { |
1746 | my $self = shift; |
b365fbae |
1747 | # Save the existing ranks, in case we need to invalidate the cached SVG. |
1748 | my %existing_ranks; |
ac4d7ac2 |
1749 | map { $existing_ranks{$_} = $_->rank } $self->readings; |
359944f7 |
1750 | |
1751 | # Do the rankings based on the relationship equivalence graph, starting |
1752 | # with the start node. |
56772e8c |
1753 | my ( $node_ranks, $rank_nodes ) = $self->relations->equivalence_ranks(); |
1754 | |
910a0a6d |
1755 | # Transfer our rankings from the topological graph to the real one. |
1756 | foreach my $r ( $self->readings ) { |
cecbe56d |
1757 | if( defined $node_ranks->{$self->equivalence( $r->id )} ) { |
359944f7 |
1758 | $r->rank( $node_ranks->{$self->equivalence( $r->id )} ); |
67da8d6c |
1759 | } else { |
63778331 |
1760 | # Die. Find the last rank we calculated. |
359944f7 |
1761 | my @all_defined = sort { ( $node_ranks->{$self->equivalence( $a->id )}||-1 ) |
1762 | <=> ( $node_ranks->{$self->equivalence( $b->id )}||-1 ) } |
63778331 |
1763 | $self->readings; |
1764 | my $last = pop @all_defined; |
1765 | throw( "Ranks not calculated after $last - do you have a cycle in the graph?" ); |
67da8d6c |
1766 | } |
de51424a |
1767 | } |
bfcbcecb |
1768 | # Do we need to invalidate the cached data? |
be3af600 |
1769 | if( $self->has_cached_table ) { |
b365fbae |
1770 | foreach my $r ( $self->readings ) { |
7c293912 |
1771 | next if defined( $existing_ranks{$r} ) |
1772 | && $existing_ranks{$r} == $r->rank; |
c1915ab9 |
1773 | # Something has changed, so clear the cache |
bfcbcecb |
1774 | $self->_clear_cache; |
c1915ab9 |
1775 | # ...and recalculate the common readings. |
1776 | $self->calculate_common_readings(); |
b365fbae |
1777 | last; |
1778 | } |
1779 | } |
c1915ab9 |
1780 | # The graph calculation information is now up to date. |
1781 | $self->_graphcalc_done(1); |
8e1394aa |
1782 | } |
3a1f2523 |
1783 | |
c1915ab9 |
1784 | sub _clear_cache { |
1785 | my $self = shift; |
c1915ab9 |
1786 | $self->wipe_table if $self->has_cached_table; |
1787 | } |
1788 | |
1789 | |
4e5a7b2c |
1790 | =head2 flatten_ranks |
1791 | |
1792 | A convenience method for parsing collation data. Searches the graph for readings |
1793 | with the same text at the same rank, and merges any that are found. |
1794 | |
1795 | =cut |
1796 | |
0e476982 |
1797 | sub flatten_ranks { |
4ef65ab4 |
1798 | my ( $self, %args ) = shift; |
0e476982 |
1799 | my %unique_rank_rdg; |
bf6e338d |
1800 | my $changed; |
4ef65ab4 |
1801 | foreach my $p ( $self->identical_readings( %args ) ) { |
1802 | # say STDERR "Combining readings at same rank: @$p"; |
1803 | $changed = 1; |
1804 | $self->merge_readings( @$p ); |
1805 | # TODO see if this now makes a common point. |
7a0956c1 |
1806 | } |
1807 | # If we merged readings, the ranks are still fine but the alignment |
1808 | # table is wrong. Wipe it. |
1809 | $self->wipe_table() if $changed; |
1810 | } |
1811 | |
1812 | =head2 identical_readings |
1813 | =head2 identical_readings( start => $startnode, end => $endnode ) |
1814 | =head2 identical_readings( startrank => $startrank, endrank => $endrank ) |
1815 | |
1816 | Goes through the graph identifying all pairs of readings that appear to be |
1817 | identical, and therefore able to be merged into a single reading. Returns the |
1818 | relevant identical pairs. Can be restricted to run over only a part of the |
1819 | graph, specified either by node or by rank. |
1820 | |
1821 | =cut |
1822 | |
1823 | sub identical_readings { |
1824 | my ( $self, %args ) = @_; |
1825 | # Find where we should start and end. |
1826 | my $startrank = $args{startrank} || 0; |
1827 | if( $args{start} ) { |
1828 | throw( "Starting reading has no rank" ) unless $self->reading( $args{start} ) |
1829 | && $self->reading( $args{start} )->has_rank; |
1830 | $startrank = $self->reading( $args{start} )->rank; |
1831 | } |
1832 | my $endrank = $args{endrank} || $self->end->rank; |
1833 | if( $args{end} ) { |
1834 | throw( "Ending reading has no rank" ) unless $self->reading( $args{end} ) |
1835 | && $self->reading( $args{end} )->has_rank; |
3c234eb6 |
1836 | $endrank = $self->reading( $args{end} )->rank; |
7a0956c1 |
1837 | } |
1838 | |
1839 | # Make sure the ranks are correct. |
1840 | unless( $self->_graphcalc_done ) { |
1841 | $self->calculate_ranks; |
1842 | } |
1843 | # Go through the readings looking for duplicates. |
1844 | my %unique_rank_rdg; |
1845 | my @pairs; |
0e476982 |
1846 | foreach my $rdg ( $self->readings ) { |
1847 | next unless $rdg->has_rank; |
7a0956c1 |
1848 | my $rk = $rdg->rank; |
1849 | next if $rk > $endrank || $rk < $startrank; |
1850 | my $key = $rk . "||" . $rdg->text; |
0e476982 |
1851 | if( exists $unique_rank_rdg{$key} ) { |
07e6765f |
1852 | # Make sure they don't have different grammatical forms |
1853 | my $ur = $unique_rank_rdg{$key}; |
a445ce40 |
1854 | if( $rdg->is_identical( $ur ) ) { |
7a0956c1 |
1855 | push( @pairs, [ $ur, $rdg ] ); |
07e6765f |
1856 | } |
0e476982 |
1857 | } else { |
1858 | $unique_rank_rdg{$key} = $rdg; |
1859 | } |
7a0956c1 |
1860 | } |
1861 | |
1862 | return @pairs; |
0e476982 |
1863 | } |
4633f9e4 |
1864 | |
1865 | |
d4b75f44 |
1866 | =head2 calculate_common_readings |
1867 | |
1868 | Goes through the graph identifying the readings that appear in every witness |
1869 | (apart from those with lacunae at that spot.) Marks them as common and returns |
1870 | the list. |
1871 | |
1872 | =begin testing |
1873 | |
1874 | use Text::Tradition; |
1875 | |
1876 | my $cxfile = 't/data/Collatex-16.xml'; |
1877 | my $t = Text::Tradition->new( |
1878 | 'name' => 'inline', |
1879 | 'input' => 'CollateX', |
1880 | 'file' => $cxfile, |
1881 | ); |
1882 | my $c = $t->collation; |
1883 | |
1884 | my @common = $c->calculate_common_readings(); |
1885 | is( scalar @common, 8, "Found correct number of common readings" ); |
1886 | my @marked = sort $c->common_readings(); |
1887 | is( scalar @common, 8, "All common readings got marked as such" ); |
679f17e1 |
1888 | my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /; |
d4b75f44 |
1889 | is_deeply( \@marked, \@expected, "Found correct list of common readings" ); |
1890 | |
1891 | =end testing |
1892 | |
1893 | =cut |
1894 | |
1895 | sub calculate_common_readings { |
1896 | my $self = shift; |
1897 | my @common; |
c1915ab9 |
1898 | map { $_->is_common( 0 ) } $self->readings; |
1899 | # Implicitly calls calculate_ranks |
1dd07bda |
1900 | my $table = $self->alignment_table; |
d4b75f44 |
1901 | foreach my $idx ( 0 .. $table->{'length'} - 1 ) { |
7f52eac8 |
1902 | my @row = map { $_->{'tokens'}->[$idx] |
1903 | ? $_->{'tokens'}->[$idx]->{'t'} : '' } |
1904 | @{$table->{'alignment'}}; |
d4b75f44 |
1905 | my %hash; |
1906 | foreach my $r ( @row ) { |
1907 | if( $r ) { |
1908 | $hash{$r->id} = $r unless $r->is_meta; |
1909 | } else { |
1910 | $hash{'UNDEF'} = $r; |
1911 | } |
1912 | } |
1913 | if( keys %hash == 1 && !exists $hash{'UNDEF'} ) { |
1914 | my( $r ) = values %hash; |
1915 | $r->is_common( 1 ); |
1916 | push( @common, $r ); |
1917 | } |
1918 | } |
1919 | return @common; |
1920 | } |
1921 | |
861c3e27 |
1922 | =head2 text_from_paths |
1923 | |
1924 | Calculate the text array for all witnesses from the path, for later consistency |
1925 | checking. Only to be used if there is no non-graph-based way to know the |
1926 | original texts. |
1927 | |
1928 | =cut |
1929 | |
1930 | sub text_from_paths { |
1931 | my $self = shift; |
1932 | foreach my $wit ( $self->tradition->witnesses ) { |
5164a6f0 |
1933 | my @readings = $self->reading_sequence( $self->start, $self->end, $wit->sigil ); |
1934 | my @text; |
1935 | foreach my $r ( @readings ) { |
1936 | next if $r->is_meta; |
1937 | push( @text, $r->text ); |
1938 | } |
861c3e27 |
1939 | $wit->text( \@text ); |
1940 | if( $wit->is_layered ) { |
5164a6f0 |
1941 | my @ucrdgs = $self->reading_sequence( $self->start, $self->end, |
1942 | $wit->sigil.$self->ac_label ); |
1943 | my @uctext; |
1944 | foreach my $r ( @ucrdgs ) { |
1945 | next if $r->is_meta; |
1946 | push( @uctext, $r->text ); |
1947 | } |
1948 | $wit->layertext( \@uctext ); |
861c3e27 |
1949 | } |
1950 | } |
1951 | } |
0e476982 |
1952 | |
4e5a7b2c |
1953 | =head1 UTILITY FUNCTIONS |
1954 | |
1955 | =head2 common_predecessor( $reading_a, $reading_b ) |
8e1394aa |
1956 | |
4e5a7b2c |
1957 | Find the last reading that occurs in sequence before both the given readings. |
414cc046 |
1958 | At the very least this should be $self->start. |
4e5a7b2c |
1959 | |
1960 | =head2 common_successor( $reading_a, $reading_b ) |
1961 | |
1962 | Find the first reading that occurs in sequence after both the given readings. |
414cc046 |
1963 | At the very least this should be $self->end. |
4e5a7b2c |
1964 | |
22222af9 |
1965 | =begin testing |
1966 | |
1967 | use Text::Tradition; |
1968 | |
1969 | my $cxfile = 't/data/Collatex-16.xml'; |
1970 | my $t = Text::Tradition->new( |
1971 | 'name' => 'inline', |
1972 | 'input' => 'CollateX', |
1973 | 'file' => $cxfile, |
1974 | ); |
1975 | my $c = $t->collation; |
1976 | |
679f17e1 |
1977 | is( $c->common_predecessor( 'n24', 'n23' )->id, |
22222af9 |
1978 | 'n20', "Found correct common predecessor" ); |
679f17e1 |
1979 | is( $c->common_successor( 'n24', 'n23' )->id, |
10e4b1ac |
1980 | '__END__', "Found correct common successor" ); |
22222af9 |
1981 | |
4e5a7b2c |
1982 | is( $c->common_predecessor( 'n19', 'n17' )->id, |
22222af9 |
1983 | 'n16', "Found correct common predecessor for readings on same path" ); |
679f17e1 |
1984 | is( $c->common_successor( 'n21', 'n10' )->id, |
10e4b1ac |
1985 | '__END__', "Found correct common successor for readings on same path" ); |
22222af9 |
1986 | |
1987 | =end testing |
1988 | |
1989 | =cut |
1990 | |
1991 | ## Return the closest reading that is a predecessor of both the given readings. |
1992 | sub common_predecessor { |
1993 | my $self = shift; |
4e5a7b2c |
1994 | my( $r1, $r2 ) = $self->_objectify_args( @_ ); |
027d819c |
1995 | return $self->_common_in_path( $r1, $r2, 'predecessors' ); |
22222af9 |
1996 | } |
1997 | |
1998 | sub common_successor { |
1999 | my $self = shift; |
4e5a7b2c |
2000 | my( $r1, $r2 ) = $self->_objectify_args( @_ ); |
027d819c |
2001 | return $self->_common_in_path( $r1, $r2, 'successors' ); |
22222af9 |
2002 | } |
2003 | |
414cc046 |
2004 | |
2005 | # TODO think about how to do this without ranks... |
027d819c |
2006 | sub _common_in_path { |
22222af9 |
2007 | my( $self, $r1, $r2, $dir ) = @_; |
414cc046 |
2008 | my $iter = $self->end->rank; |
22222af9 |
2009 | my @candidates; |
414cc046 |
2010 | my @last_r1 = ( $r1 ); |
2011 | my @last_r2 = ( $r2 ); |
2012 | # my %all_seen = ( $r1 => 'r1', $r2 => 'r2' ); |
22222af9 |
2013 | my %all_seen; |
6771a1b1 |
2014 | # say STDERR "Finding common $dir for $r1, $r2"; |
22222af9 |
2015 | while( !@candidates ) { |
414cc046 |
2016 | last unless $iter--; # Avoid looping infinitely |
2017 | # Iterate separately down the graph from r1 and r2 |
2018 | my( @new_lc1, @new_lc2 ); |
2019 | foreach my $lc ( @last_r1 ) { |
2020 | foreach my $p ( $lc->$dir ) { |
2021 | if( $all_seen{$p->id} && $all_seen{$p->id} ne 'r1' ) { |
6771a1b1 |
2022 | # say STDERR "Path candidate $p from $lc"; |
414cc046 |
2023 | push( @candidates, $p ); |
002e3600 |
2024 | } elsif( !$all_seen{$p->id} ) { |
414cc046 |
2025 | $all_seen{$p->id} = 'r1'; |
2026 | push( @new_lc1, $p ); |
2027 | } |
2028 | } |
2029 | } |
2030 | foreach my $lc ( @last_r2 ) { |
22222af9 |
2031 | foreach my $p ( $lc->$dir ) { |
414cc046 |
2032 | if( $all_seen{$p->id} && $all_seen{$p->id} ne 'r2' ) { |
6771a1b1 |
2033 | # say STDERR "Path candidate $p from $lc"; |
22222af9 |
2034 | push( @candidates, $p ); |
002e3600 |
2035 | } elsif( !$all_seen{$p->id} ) { |
414cc046 |
2036 | $all_seen{$p->id} = 'r2'; |
2037 | push( @new_lc2, $p ); |
22222af9 |
2038 | } |
2039 | } |
2040 | } |
414cc046 |
2041 | @last_r1 = @new_lc1; |
2042 | @last_r2 = @new_lc2; |
22222af9 |
2043 | } |
2044 | my @answer = sort { $a->rank <=> $b->rank } @candidates; |
2045 | return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer ); |
2046 | } |
2047 | |
63778331 |
2048 | sub throw { |
2049 | Text::Tradition::Error->throw( |
2050 | 'ident' => 'Collation error', |
2051 | 'message' => $_[0], |
2052 | ); |
2053 | } |
2054 | |
dd3b58b0 |
2055 | no Moose; |
2056 | __PACKAGE__->meta->make_immutable; |
e867486f |
2057 | |
a445ce40 |
2058 | =head1 BUGS/TODO |
2059 | |
2060 | =over |
2061 | |
2062 | =item * Rework XML serialization in a more modular way |
2063 | |
2064 | =back |
2065 | |
027d819c |
2066 | =head1 LICENSE |
e867486f |
2067 | |
027d819c |
2068 | This package is free software and is provided "as is" without express |
2069 | or implied warranty. You can redistribute it and/or modify it under |
2070 | the same terms as Perl itself. |
e867486f |
2071 | |
027d819c |
2072 | =head1 AUTHOR |
e867486f |
2073 | |
027d819c |
2074 | Tara L Andrews E<lt>aurum@cpan.orgE<gt> |