Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition::Collation; |
d047cd52 |
2 | |
3 | use Graph::Easy; |
8e1394aa |
4 | use IPC::Run qw( run binary ); |
3265b0ce |
5 | use Text::Tradition::Collation::Path; |
4cdd82f1 |
6 | use Text::Tradition::Collation::Position; |
b15511bf |
7 | use Text::Tradition::Collation::Reading; |
8 | use Text::Tradition::Collation::Relationship; |
9 | use Text::Tradition::Collation::Segment; |
df6d9812 |
10 | use XML::LibXML; |
dd3b58b0 |
11 | use Moose; |
12 | |
13 | has 'graph' => ( |
d047cd52 |
14 | is => 'ro', |
15 | isa => 'Graph::Easy', |
16 | handles => { |
8e1394aa |
17 | add_reading => 'add_node', |
18 | del_reading => 'del_node', |
19 | add_path => 'add_edge', |
20 | del_path => 'del_edge', |
21 | reading => 'node', |
22 | path => 'edge', |
23 | readings => 'nodes', |
b15511bf |
24 | segments => 'nodes', |
8e1394aa |
25 | paths => 'edges', |
df6d9812 |
26 | relationships => 'edges', |
d047cd52 |
27 | }, |
28 | default => sub { Graph::Easy->new( undirected => 0 ) }, |
29 | ); |
784877d9 |
30 | |
dd3b58b0 |
31 | |
f6066bac |
32 | has 'tradition' => ( # TODO should this not be ro? |
8e1394aa |
33 | is => 'rw', |
d047cd52 |
34 | isa => 'Text::Tradition', |
35 | ); |
dd3b58b0 |
36 | |
8e1394aa |
37 | has 'svg' => ( |
38 | is => 'ro', |
39 | isa => 'Str', |
40 | writer => '_save_svg', |
41 | predicate => 'has_svg', |
42 | ); |
43 | |
8e1394aa |
44 | has 'graphml' => ( |
45 | is => 'ro', |
df6d9812 |
46 | isa => 'Str', |
8e1394aa |
47 | writer => '_save_graphml', |
48 | predicate => 'has_graphml', |
49 | ); |
50 | |
3a1f2523 |
51 | # Keeps track of the lemmas within the collation. At most one lemma |
52 | # per position in the graph. |
53 | has 'lemmata' => ( |
54 | is => 'ro', |
55 | isa => 'HashRef[Maybe[Str]]', |
56 | default => sub { {} }, |
57 | ); |
58 | |
4a8828f0 |
59 | has 'wit_list_separator' => ( |
7854e12e |
60 | is => 'rw', |
61 | isa => 'Str', |
62 | default => ', ', |
63 | ); |
64 | |
65 | has 'baselabel' => ( |
66 | is => 'rw', |
67 | isa => 'Str', |
68 | default => 'base text', |
69 | ); |
4a8828f0 |
70 | |
1f563ac3 |
71 | has 'collapsed' => ( |
15d2d3df |
72 | is => 'rw', |
73 | isa => 'Bool', |
74 | ); |
75 | |
76 | has 'linear' => ( |
77 | is => 'rw', |
78 | isa => 'Bool', |
79 | default => 1, |
80 | ); |
1f563ac3 |
81 | |
ef9d481f |
82 | has 'ac_label' => ( |
83 | is => 'rw', |
84 | isa => 'Str', |
85 | default => ' (a.c.)', |
86 | ); |
87 | |
1f563ac3 |
88 | |
dd3b58b0 |
89 | # The collation can be created two ways: |
90 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
91 | # the results as in 2. |
92 | # 2. Read a pre-prepared collation in one of a variety of formats, |
93 | # and make the graph from that. |
94 | |
95 | # The graph itself will (for now) be immutable, and the positions |
96 | # within the graph will also be immutable. We need to calculate those |
97 | # positions upon graph construction. The equivalences between graph |
98 | # nodes will be mutable, entirely determined by the user (or possibly |
99 | # by some semantic pre-processing provided by the user.) So the |
100 | # constructor should just make an empty equivalences object. The |
101 | # constructor will also need to make the witness objects, if we didn't |
102 | # come through option 1. |
103 | |
d047cd52 |
104 | sub BUILD { |
105 | my( $self, $args ) = @_; |
8e1394aa |
106 | $self->graph->use_class('node', 'Text::Tradition::Collation::Reading'); |
3265b0ce |
107 | $self->graph->use_class('edge', 'Text::Tradition::Collation::Path'); |
d047cd52 |
108 | |
4a8828f0 |
109 | # Pass through any graph-specific options. |
110 | my $shape = exists( $args->{'shape'} ) ? $args->{'shape'} : 'ellipse'; |
111 | $self->graph->set_attribute( 'node', 'shape', $shape ); |
d047cd52 |
112 | } |
784877d9 |
113 | |
7854e12e |
114 | # Wrapper around add_path |
115 | |
116 | around add_path => sub { |
117 | my $orig = shift; |
118 | my $self = shift; |
119 | |
120 | # Make sure there are three arguments |
121 | unless( @_ == 3 ) { |
122 | warn "Call add_path with args source, target, witness"; |
123 | return; |
124 | } |
125 | # Make sure the proposed path does not yet exist |
b15511bf |
126 | # NOTE 'reading' will currently return readings and segments |
7854e12e |
127 | my( $source, $target, $wit ) = @_; |
128 | $source = $self->reading( $source ) |
129 | unless ref( $source ) eq 'Text::Tradition::Collation::Reading'; |
130 | $target = $self->reading( $target ) |
131 | unless ref( $target ) eq 'Text::Tradition::Collation::Reading'; |
132 | foreach my $path ( $source->edges_to( $target ) ) { |
ef9d481f |
133 | if( $path->label eq $wit && $path->class eq 'edge.path' ) { |
7854e12e |
134 | return; |
135 | } |
136 | } |
137 | # Do the deed |
138 | $self->$orig( @_ ); |
139 | }; |
140 | |
3265b0ce |
141 | # Wrapper around paths |
142 | around paths => sub { |
143 | my $orig = shift; |
144 | my $self = shift; |
145 | |
b15511bf |
146 | my @result = grep { $_->sub_class eq 'path' } $self->$orig( @_ ); |
df6d9812 |
147 | return @result; |
148 | }; |
149 | |
150 | around relationships => sub { |
151 | my $orig = shift; |
152 | my $self = shift; |
b15511bf |
153 | my @result = grep { $_->sub_class eq 'relationship' } $self->$orig( @_ ); |
154 | return @result; |
155 | }; |
156 | |
157 | around readings => sub { |
158 | my $orig = shift; |
159 | my $self = shift; |
160 | my @result = grep { $_->sub_class ne 'segment' } $self->$orig( @_ ); |
161 | return @result; |
162 | }; |
163 | |
164 | around segments => sub { |
165 | my $orig = shift; |
166 | my $self = shift; |
167 | my @result = grep { $_->sub_class eq 'segment' } $self->$orig( @_ ); |
3265b0ce |
168 | return @result; |
169 | }; |
170 | |
7854e12e |
171 | # Wrapper around merge_nodes |
784877d9 |
172 | |
173 | sub merge_readings { |
174 | my $self = shift; |
175 | my $first_node = shift; |
176 | my $second_node = shift; |
177 | $first_node->merge_from( $second_node ); |
178 | unshift( @_, $first_node, $second_node ); |
179 | return $self->graph->merge_nodes( @_ ); |
180 | } |
181 | |
15d2d3df |
182 | # Extra graph-alike utility |
183 | sub has_path { |
184 | my( $self, $source, $target, $label ) = @_; |
185 | my @paths = $source->edges_to( $target ); |
186 | my @relevant = grep { $_->label eq $label } @paths; |
b15511bf |
187 | return scalar @relevant; |
188 | } |
189 | |
190 | ## Dealing with groups of readings, i.e. segments. |
191 | |
192 | sub add_segment { |
193 | my( $self, @items ) = @_; |
194 | my $segment = Text::Tradition::Collation::Segment->new( 'members' => \@items ); |
195 | return $segment; |
15d2d3df |
196 | } |
197 | |
3265b0ce |
198 | ## Dealing with relationships between readings. This is a different |
4cdd82f1 |
199 | ## sort of graph edge. Return a success/failure value and a list of |
200 | ## node pairs that have been linked. |
3265b0ce |
201 | |
202 | sub add_relationship { |
b15511bf |
203 | my( $self, $source, $target, $options ) = @_; |
ef9d481f |
204 | |
205 | # Make sure there is not another relationship between these two |
b15511bf |
206 | # readings or segments already |
ef9d481f |
207 | $source = $self->reading( $source ) |
b15511bf |
208 | unless ref( $source ) && $source->isa( 'Graph::Easy::Node' ); |
ef9d481f |
209 | $target = $self->reading( $target ) |
b15511bf |
210 | unless ref( $target ) && $target->isa( 'Graph::Easy::Node' ); |
4cdd82f1 |
211 | foreach my $rel ( $source->edges_to( $target ), $target->edges_to( $source ) ) { |
212 | if( $rel->class eq 'edge.relationship' ) { |
213 | return ( undef, "Relationship already exists between these readings" ); |
214 | } else { |
215 | return ( undef, "There is a witness path between these readings" ); |
216 | } |
217 | } |
218 | |
219 | if( $source->has_position && $target->has_position ) { |
220 | unless( grep { $_ eq $target } $self->same_position_as( $source ) ) { |
221 | return( undef, "Cannot set relationship at different positions" ); |
ef9d481f |
222 | } |
223 | } |
224 | |
4cdd82f1 |
225 | my @joined = ( [ $source->name, $target->name ] ); # Keep track of the nodes we join. |
226 | |
227 | $options->{'this_relation'} = [ $source, $target ]; |
f6066bac |
228 | my $rel; |
229 | eval { $rel = Text::Tradition::Collation::Relationship->new( %$options ) }; |
230 | if( $@ ) { |
231 | return ( undef, $@ ); |
232 | } |
3265b0ce |
233 | $self->graph->add_edge( $source, $target, $rel ); |
b15511bf |
234 | if( $options->{'global'} ) { |
3265b0ce |
235 | # Look for all readings with the source label, and if there are |
236 | # colocated readings with the target label, join them too. |
4cdd82f1 |
237 | foreach my $r ( grep { $_->label eq $source->label } $self->readings() ) { |
238 | next if $r->name eq $source->name; |
3265b0ce |
239 | my @colocated = grep { $_->label eq $target->label } |
240 | $self->same_position_as( $r ); |
241 | if( @colocated ) { |
242 | warn "Multiple readings with same label at same position!" |
243 | if @colocated > 1; |
4cdd82f1 |
244 | my $colo = $colocated[0]; |
245 | next if $colo->edges_to( $r ) || $r->edges_to( $colo ); |
246 | $options->{'primary_relation'} = $options->{'this_relation'}; |
247 | $options->{'this_relation'} = [ $r, $colocated[0] ]; |
b15511bf |
248 | my $dup_rel = Text::Tradition::Collation::Relationship->new( %$options ); |
3265b0ce |
249 | $self->graph->add_edge( $r, $colocated[0], $dup_rel ); |
4cdd82f1 |
250 | push( @joined, [ $r->name, $colocated[0]->name ] ); |
3265b0ce |
251 | } |
252 | } |
253 | } |
4cdd82f1 |
254 | return( 1, @joined ); |
3265b0ce |
255 | } |
256 | |
8e1394aa |
257 | =head2 Output method(s) |
258 | |
259 | =over |
260 | |
261 | =item B<as_svg> |
262 | |
263 | print $graph->as_svg( $recalculate ); |
264 | |
265 | Returns an SVG string that represents the graph. Uses GraphViz to do |
4a8828f0 |
266 | this, because Graph::Easy doesn\'t cope well with long graphs. Unless |
8e1394aa |
267 | $recalculate is passed (and is a true value), the method will return a |
268 | cached copy of the SVG after the first call to the method. |
269 | |
270 | =cut |
271 | |
272 | sub as_svg { |
273 | my( $self, $recalc ) = @_; |
274 | return $self->svg if $self->has_svg; |
275 | |
3265b0ce |
276 | $self->collapse_graph_paths(); |
8e1394aa |
277 | |
278 | my @cmd = qw/dot -Tsvg/; |
279 | my( $svg, $err ); |
df6d9812 |
280 | my $in = $self->as_dot(); |
8e1394aa |
281 | run( \@cmd, \$in, ">", binary(), \$svg ); |
df6d9812 |
282 | $self->_save_svg( $svg ); |
3265b0ce |
283 | $self->expand_graph_paths(); |
8e1394aa |
284 | return $svg; |
285 | } |
286 | |
df6d9812 |
287 | =item B<as_dot> |
288 | |
289 | print $graph->as_dot( $view, $recalculate ); |
290 | |
291 | Returns a string that is the collation graph expressed in dot |
292 | (i.e. GraphViz) format. The 'view' argument determines what kind of |
293 | graph is produced. |
294 | * 'path': a graph of witness paths through the collation (DEFAULT) |
295 | * 'relationship': a graph of how collation readings relate to |
296 | each other |
297 | |
298 | =cut |
299 | |
300 | sub as_dot { |
301 | my( $self, $view ) = @_; |
302 | $view = 'path' unless $view; |
303 | # TODO consider making some of these things configurable |
304 | my $dot = sprintf( "digraph %s {\n", $self->tradition->name ); |
305 | $dot .= "\tedge [ arrowhead=open ];\n"; |
306 | $dot .= "\tgraph [ rankdir=LR ];\n"; |
307 | $dot .= sprintf( "\tnode [ fontsize=%d, fillcolor=%s, style=%s, shape=%s ];\n", |
308 | 11, "white", "filled", $self->graph->get_attribute( 'node', 'shape' ) ); |
309 | |
310 | foreach my $reading ( $self->readings ) { |
b15511bf |
311 | # Need not output nodes without separate labels |
df6d9812 |
312 | next if $reading->name eq $reading->label; |
b15511bf |
313 | # TODO output readings or segments, but not both |
314 | next if $reading->class eq 'node.segment'; |
f2b9605f |
315 | $dot .= sprintf( "\t\"%s\" [ label=\"%s\" ];\n", $reading->name, $reading->label ); |
df6d9812 |
316 | } |
317 | |
318 | my @edges = $view eq 'relationship' ? $self->relationships : $self->paths; |
319 | foreach my $edge ( @edges ) { |
f6066bac |
320 | my %variables = ( 'color' => '#000000', |
321 | 'fontcolor' => '#000000', |
322 | 'label' => $edge->label, |
323 | ); |
324 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
f2b9605f |
325 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", |
f6066bac |
326 | $edge->from->name, $edge->to->name, $varopts ); |
df6d9812 |
327 | } |
df6d9812 |
328 | $dot .= "}\n"; |
329 | return $dot; |
330 | } |
331 | |
8e1394aa |
332 | =item B<as_graphml> |
333 | |
334 | print $graph->as_graphml( $recalculate ) |
335 | |
336 | Returns a GraphML representation of the collation graph, with |
337 | transposition information and position information. Unless |
338 | $recalculate is passed (and is a true value), the method will return a |
339 | cached copy of the SVG after the first call to the method. |
340 | |
341 | =cut |
342 | |
343 | sub as_graphml { |
344 | my( $self, $recalc ) = @_; |
345 | return $self->graphml if $self->has_graphml; |
346 | |
347 | # Some namespaces |
348 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
349 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
350 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
351 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
352 | |
353 | # Create the document and root node |
354 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
355 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
356 | $graphml->setDocumentElement( $root ); |
357 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
358 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
359 | |
f6066bac |
360 | # TODO Add some global graph data |
361 | |
8e1394aa |
362 | # Add the data keys for nodes |
ef9d481f |
363 | my %node_data_keys; |
364 | my $ndi = 0; |
b15511bf |
365 | foreach my $datum ( qw/ name reading identical position class / ) { |
ef9d481f |
366 | $node_data_keys{$datum} = 'dn'.$ndi++; |
8e1394aa |
367 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
ef9d481f |
368 | $key->setAttribute( 'attr.name', $datum ); |
8e1394aa |
369 | $key->setAttribute( 'attr.type', 'string' ); |
370 | $key->setAttribute( 'for', 'node' ); |
ef9d481f |
371 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
372 | } |
373 | |
df6d9812 |
374 | # Add the data keys for edges, i.e. witnesses |
ef9d481f |
375 | my $edi = 0; |
376 | my %edge_data_keys; |
b15511bf |
377 | foreach my $edge_key( qw/ witness_main witness_ante_corr relationship class / ) { |
ef9d481f |
378 | $edge_data_keys{$edge_key} = 'de'.$edi++; |
8e1394aa |
379 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
ef9d481f |
380 | $key->setAttribute( 'attr.name', $edge_key ); |
8e1394aa |
381 | $key->setAttribute( 'attr.type', 'string' ); |
382 | $key->setAttribute( 'for', 'edge' ); |
ef9d481f |
383 | $key->setAttribute( 'id', $edge_data_keys{$edge_key} ); |
8e1394aa |
384 | } |
ef9d481f |
385 | |
8e1394aa |
386 | # Add the graph, its nodes, and its edges |
387 | my $graph = $root->addNewChild( $graphml_ns, 'graph' ); |
388 | $graph->setAttribute( 'edgedefault', 'directed' ); |
389 | $graph->setAttribute( 'id', 'g0' ); # TODO make this meaningful |
390 | $graph->setAttribute( 'parse.edgeids', 'canonical' ); |
df6d9812 |
391 | $graph->setAttribute( 'parse.edges', scalar($self->paths) ); |
8e1394aa |
392 | $graph->setAttribute( 'parse.nodeids', 'canonical' ); |
df6d9812 |
393 | $graph->setAttribute( 'parse.nodes', scalar($self->readings) ); |
8e1394aa |
394 | $graph->setAttribute( 'parse.order', 'nodesfirst' ); |
395 | |
396 | my $node_ctr = 0; |
397 | my %node_hash; |
b15511bf |
398 | # Add our readings to the graph |
b5054ca9 |
399 | foreach my $n ( sort { $a->name cmp $b->name } $self->readings ) { |
8e1394aa |
400 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
401 | my $node_xmlid = 'n' . $node_ctr++; |
402 | $node_hash{ $n->name } = $node_xmlid; |
403 | $node_el->setAttribute( 'id', $node_xmlid ); |
b15511bf |
404 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
405 | _add_graphml_data( $node_el, $node_data_keys{'reading'}, $n->label ); |
f6066bac |
406 | _add_graphml_data( $node_el, $node_data_keys{'position'}, $n->position->reference ) |
407 | if $n->has_position; |
b15511bf |
408 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
409 | _add_graphml_data( $node_el, $node_data_keys{'identical'}, $n->primary->name ) |
410 | if $n->has_primary; |
8e1394aa |
411 | } |
412 | |
b15511bf |
413 | # Add any segments we have |
414 | foreach my $n ( sort { $a->name cmp $b->name } $self->segments ) { |
415 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
416 | my $node_xmlid = 'n' . $node_ctr++; |
417 | $node_hash{ $n->name } = $node_xmlid; |
418 | $node_el->setAttribute( 'id', $node_xmlid ); |
419 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
420 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
421 | } |
422 | |
423 | # Add the path, relationship, and segment edges |
df6d9812 |
424 | my $edge_ctr = 0; |
ef9d481f |
425 | foreach my $e ( sort { $a->from->name cmp $b->from->name } $self->graph->edges() ) { |
df6d9812 |
426 | my( $name, $from, $to ) = ( 'e'.$edge_ctr++, |
b5054ca9 |
427 | $node_hash{ $e->from->name() }, |
428 | $node_hash{ $e->to->name() } ); |
8e1394aa |
429 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
430 | $edge_el->setAttribute( 'source', $from ); |
431 | $edge_el->setAttribute( 'target', $to ); |
432 | $edge_el->setAttribute( 'id', $name ); |
b15511bf |
433 | # Add the edge class |
434 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, $e->sub_class ); |
435 | if( $e->sub_class eq 'path' ) { |
ef9d481f |
436 | # It's a witness path, so add the witness |
437 | my $base = $e->label; |
438 | my $key = $edge_data_keys{'witness_main'}; |
439 | # TODO kind of hacky |
440 | if( $e->label =~ /^(.*?)\s+(\(a\.c\.\))$/ ) { |
441 | $base = $1; |
442 | $key = $edge_data_keys{'witness_ante_corr'}; |
443 | } |
b15511bf |
444 | _add_graphml_data( $edge_el, $key, $base ); |
445 | } elsif( $e->sub_class eq 'relationship' ) { |
ef9d481f |
446 | # It's a relationship |
b15511bf |
447 | _add_graphml_data( $edge_el, $edge_data_keys{'relationship'}, $e->label ); |
448 | } # else a segment, nothing to record but source, target, class |
8e1394aa |
449 | } |
450 | |
451 | # Return the thing |
df6d9812 |
452 | $self->_save_graphml( $graphml->toString(1) ); |
453 | return $graphml->toString(1); |
454 | } |
455 | |
b15511bf |
456 | sub _add_graphml_data { |
457 | my( $el, $key, $value ) = @_; |
458 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
459 | return unless defined $value; |
460 | $data_el->setAttribute( 'key', $key ); |
461 | $data_el->appendText( $value ); |
8e1394aa |
462 | } |
463 | |
3265b0ce |
464 | sub collapse_graph_paths { |
1f563ac3 |
465 | my $self = shift; |
3265b0ce |
466 | # Our collation graph has an path per witness. This is great for |
1f563ac3 |
467 | # calculation purposes, but terrible for display. Thus we want to |
3265b0ce |
468 | # display only one path between any two nodes. |
1f563ac3 |
469 | |
470 | return if $self->collapsed; |
471 | |
3265b0ce |
472 | print STDERR "Collapsing witness paths in graph...\n"; |
1f563ac3 |
473 | |
474 | # Don't list out every witness if we have more than half to list. |
475 | my $majority = int( scalar( @{$self->tradition->witnesses} ) / 2 ) + 1; |
a0093bf2 |
476 | # But don't compress if there are only a few witnesses. |
477 | $majority = 4 if $majority < 4; |
b15511bf |
478 | foreach my $node ( $self->readings ) { |
1f563ac3 |
479 | my $newlabels = {}; |
480 | # We will visit each node, so we only look ahead. |
df6d9812 |
481 | foreach my $edge ( $node->outgoing() ) { |
482 | next unless $edge->class eq 'edge.path'; |
483 | add_hash_entry( $newlabels, $edge->to->name, $edge->name ); |
484 | $self->del_path( $edge ); |
1f563ac3 |
485 | } |
486 | |
487 | foreach my $newdest ( keys %$newlabels ) { |
488 | my $label; |
489 | my @compressed_wits = (); |
490 | if( @{$newlabels->{$newdest}} < $majority ) { |
a0093bf2 |
491 | $label = join( ', ', sort( @{$newlabels->{$newdest}} ) ); |
1f563ac3 |
492 | } else { |
493 | ## TODO FIX THIS HACK |
6a222840 |
494 | my @aclabels; |
1f563ac3 |
495 | foreach my $wit ( @{$newlabels->{$newdest}} ) { |
6a222840 |
496 | if( $wit =~ /^(.*?)(\s*\(?a\.\s*c\.\)?)$/ ) { |
497 | push( @aclabels, $wit ); |
1f563ac3 |
498 | } else { |
499 | push( @compressed_wits, $wit ); |
500 | } |
501 | } |
a0093bf2 |
502 | $label = join( ', ', 'majority', sort( @aclabels ) ); |
1f563ac3 |
503 | } |
504 | |
3265b0ce |
505 | my $newpath = |
1f563ac3 |
506 | $self->add_path( $node, $self->reading( $newdest ), $label ); |
507 | if( @compressed_wits ) { |
3265b0ce |
508 | $newpath->hidden_witnesses( \@compressed_wits ); |
1f563ac3 |
509 | } |
510 | } |
511 | } |
512 | |
513 | $self->collapsed( 1 ); |
514 | } |
515 | |
3265b0ce |
516 | sub expand_graph_paths { |
1f563ac3 |
517 | my $self = shift; |
3265b0ce |
518 | # Our collation graph has only one path between any two nodes. |
1f563ac3 |
519 | # This is great for display, but not so great for analysis. |
3265b0ce |
520 | # Expand this so that each witness has its own path between any |
1f563ac3 |
521 | # two reading nodes. |
522 | return unless $self->collapsed; |
523 | |
3265b0ce |
524 | print STDERR "Expanding witness paths in graph...\n"; |
3265b0ce |
525 | foreach my $path( $self->paths ) { |
526 | my $from = $path->from; |
527 | my $to = $path->to; |
528 | my @wits = split( /, /, $path->label ); |
529 | if( $path->has_hidden_witnesses ) { |
530 | push( @wits, @{$path->hidden_witnesses} ); |
1f563ac3 |
531 | } |
3265b0ce |
532 | $self->del_path( $path ); |
1f563ac3 |
533 | foreach ( @wits ) { |
534 | $self->add_path( $from, $to, $_ ); |
535 | } |
536 | } |
537 | $self->collapsed( 0 ); |
538 | } |
539 | |
8e1394aa |
540 | =back |
541 | |
de51424a |
542 | =head2 Navigation methods |
543 | |
544 | =over |
545 | |
8e1394aa |
546 | =item B<start> |
547 | |
548 | my $beginning = $collation->start(); |
549 | |
550 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
551 | |
552 | =cut |
553 | |
554 | sub start { |
4a8828f0 |
555 | # Return the beginning reading of the graph. |
8e1394aa |
556 | my $self = shift; |
557 | my( $new_start ) = @_; |
558 | if( $new_start ) { |
559 | $self->del_reading( '#START#' ); |
560 | $self->graph->rename_node( $new_start, '#START#' ); |
561 | } |
562 | return $self->reading('#START#'); |
563 | } |
564 | |
e2902068 |
565 | =item B<reading_sequence> |
566 | |
567 | my @readings = $graph->reading_sequence( $first, $last, $path[, $alt_path] ); |
568 | |
569 | Returns the ordered list of readings, starting with $first and ending |
570 | with $last, along the given witness path. If no path is specified, |
571 | assume that the path is that of the base text (if any.) |
572 | |
573 | =cut |
574 | |
575 | sub reading_sequence { |
576 | my( $self, $start, $end, $witness, $backup ) = @_; |
577 | |
930ff666 |
578 | $witness = $self->baselabel unless $witness; |
e2902068 |
579 | my @readings = ( $start ); |
580 | my %seen; |
581 | my $n = $start; |
930ff666 |
582 | while( $n && $n ne $end ) { |
e2902068 |
583 | if( exists( $seen{$n->name()} ) ) { |
584 | warn "Detected loop at " . $n->name(); |
585 | last; |
586 | } |
587 | $seen{$n->name()} = 1; |
588 | |
589 | my $next = $self->next_reading( $n, $witness, $backup ); |
590 | warn "Did not find any path for $witness from reading " . $n->name |
591 | unless $next; |
592 | push( @readings, $next ); |
593 | $n = $next; |
594 | } |
595 | # Check that the last reading is our end reading. |
596 | my $last = $readings[$#readings]; |
597 | warn "Last reading found from " . $start->label() . |
598 | " for witness $witness is not the end!" |
599 | unless $last eq $end; |
600 | |
601 | return @readings; |
602 | } |
603 | |
4a8828f0 |
604 | =item B<next_reading> |
8e1394aa |
605 | |
4a8828f0 |
606 | my $next_reading = $graph->next_reading( $reading, $witpath ); |
8e1394aa |
607 | |
4a8828f0 |
608 | Returns the reading that follows the given reading along the given witness |
930ff666 |
609 | path. |
8e1394aa |
610 | |
611 | =cut |
612 | |
4a8828f0 |
613 | sub next_reading { |
e2902068 |
614 | # Return the successor via the corresponding path. |
8e1394aa |
615 | my $self = shift; |
4a8828f0 |
616 | return $self->_find_linked_reading( 'next', @_ ); |
8e1394aa |
617 | } |
618 | |
4a8828f0 |
619 | =item B<prior_reading> |
8e1394aa |
620 | |
4a8828f0 |
621 | my $prior_reading = $graph->prior_reading( $reading, $witpath ); |
8e1394aa |
622 | |
4a8828f0 |
623 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
624 | path. |
8e1394aa |
625 | |
626 | =cut |
627 | |
4a8828f0 |
628 | sub prior_reading { |
e2902068 |
629 | # Return the predecessor via the corresponding path. |
8e1394aa |
630 | my $self = shift; |
4a8828f0 |
631 | return $self->_find_linked_reading( 'prior', @_ ); |
8e1394aa |
632 | } |
633 | |
4a8828f0 |
634 | sub _find_linked_reading { |
e2902068 |
635 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
636 | my @linked_paths = $direction eq 'next' |
8e1394aa |
637 | ? $node->outgoing() : $node->incoming(); |
e2902068 |
638 | return undef unless scalar( @linked_paths ); |
8e1394aa |
639 | |
e2902068 |
640 | # We have to find the linked path that contains all of the |
641 | # witnesses supplied in $path. |
642 | my( @path_wits, @alt_path_wits ); |
643 | @path_wits = $self->witnesses_of_label( $path ) if $path; |
644 | @alt_path_wits = $self->witnesses_of_label( $alt_path ) if $alt_path; |
645 | my $base_le; |
646 | my $alt_le; |
647 | foreach my $le ( @linked_paths ) { |
930ff666 |
648 | if( $le->name eq $self->baselabel ) { |
e2902068 |
649 | $base_le = $le; |
650 | } else { |
651 | my @le_wits = $self->witnesses_of_label( $le->name ); |
652 | if( _is_within( \@path_wits, \@le_wits ) ) { |
653 | # This is the right path. |
654 | return $direction eq 'next' ? $le->to() : $le->from(); |
655 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
656 | $alt_le = $le; |
657 | } |
8e1394aa |
658 | } |
659 | } |
e2902068 |
660 | # Got this far? Return the alternate path if it exists. |
661 | return $direction eq 'next' ? $alt_le->to() : $alt_le->from() |
662 | if $alt_le; |
663 | |
664 | # Got this far? Return the base path if it exists. |
665 | return $direction eq 'next' ? $base_le->to() : $base_le->from() |
666 | if $base_le; |
667 | |
668 | # Got this far? We have no appropriate path. |
8e1394aa |
669 | warn "Could not find $direction node from " . $node->label |
e2902068 |
670 | . " along path $path"; |
8e1394aa |
671 | return undef; |
672 | } |
673 | |
4a8828f0 |
674 | # Some set logic. |
675 | sub _is_within { |
676 | my( $set1, $set2 ) = @_; |
7854e12e |
677 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
678 | foreach my $el ( @$set1 ) { |
679 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
680 | } |
681 | return $ret; |
682 | } |
683 | |
de51424a |
684 | |
685 | ## INITIALIZATION METHODS - for use by parsers |
4a8828f0 |
686 | # Walk the paths for each witness in the graph, and return the nodes |
e2902068 |
687 | # that the graph has in common. If $using_base is true, some |
688 | # different logic is needed. |
4a8828f0 |
689 | |
690 | sub walk_witness_paths { |
691 | my( $self, $end ) = @_; |
692 | # For each witness, walk the path through the graph. |
693 | # Then we need to find the common nodes. |
694 | # TODO This method is going to fall down if we have a very gappy |
695 | # text in the collation. |
696 | my $paths = {}; |
3a1f2523 |
697 | my @common_readings; |
4a8828f0 |
698 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
699 | my $curr_reading = $self->start; |
e2902068 |
700 | my @wit_path = $self->reading_sequence( $self->start, $end, |
701 | $wit->sigil ); |
4a8828f0 |
702 | $wit->path( \@wit_path ); |
e2902068 |
703 | |
704 | # Detect the common readings. |
930ff666 |
705 | @common_readings = _find_common( \@common_readings, \@wit_path ); |
4a8828f0 |
706 | } |
707 | |
708 | # Mark all the nodes as either common or not. |
3a1f2523 |
709 | foreach my $cn ( @common_readings ) { |
e2902068 |
710 | print STDERR "Setting " . $cn->name . " / " . $cn->label |
711 | . " as common node\n"; |
4a8828f0 |
712 | $cn->make_common; |
713 | } |
714 | foreach my $n ( $self->readings() ) { |
715 | $n->make_variant unless $n->is_common; |
716 | } |
3a1f2523 |
717 | # Return an array of the common nodes in order. |
718 | return @common_readings; |
4a8828f0 |
719 | } |
720 | |
930ff666 |
721 | sub _find_common { |
722 | my( $common_readings, $new_path ) = @_; |
723 | my @cr; |
724 | if( @$common_readings ) { |
725 | foreach my $n ( @$new_path ) { |
726 | push( @cr, $n ) if grep { $_ eq $n } @$common_readings; |
727 | } |
728 | } else { |
729 | push( @cr, @$new_path ); |
730 | } |
731 | return @cr; |
732 | } |
733 | |
734 | sub _remove_common { |
735 | my( $common_readings, $divergence ) = @_; |
736 | my @cr; |
737 | my %diverged; |
738 | map { $diverged{$_->name} = 1 } @$divergence; |
739 | foreach( @$common_readings ) { |
740 | push( @cr, $_ ) unless $diverged{$_->name}; |
741 | } |
742 | return @cr; |
743 | } |
744 | |
745 | |
e2902068 |
746 | # An alternative to walk_witness_paths, for use when a collation is |
6a222840 |
747 | # constructed from a base text and an apparatus. We have the |
748 | # sequences of readings and just need to add path edges. |
e2902068 |
749 | |
6a222840 |
750 | sub make_witness_paths { |
751 | my( $self ) = @_; |
e2902068 |
752 | |
930ff666 |
753 | my @common_readings; |
e2902068 |
754 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
15d2d3df |
755 | print STDERR "Making path for " . $wit->sigil . "\n"; |
6a222840 |
756 | $self->make_witness_path( $wit ); |
757 | @common_readings = _find_common( \@common_readings, $wit->path ); |
15d2d3df |
758 | @common_readings = _find_common( \@common_readings, $wit->uncorrected_path ); |
7854e12e |
759 | } |
b15511bf |
760 | map { $_->make_common } @common_readings; |
6a222840 |
761 | return @common_readings; |
7854e12e |
762 | } |
763 | |
6a222840 |
764 | sub make_witness_path { |
7854e12e |
765 | my( $self, $wit ) = @_; |
766 | my @chain = @{$wit->path}; |
15d2d3df |
767 | my $sig = $wit->sigil; |
7854e12e |
768 | foreach my $idx ( 0 .. $#chain-1 ) { |
6a222840 |
769 | $self->add_path( $chain[$idx], $chain[$idx+1], $sig ); |
7854e12e |
770 | } |
15d2d3df |
771 | @chain = @{$wit->uncorrected_path}; |
772 | foreach my $idx( 0 .. $#chain-1 ) { |
773 | my $source = $chain[$idx]; |
774 | my $target = $chain[$idx+1]; |
b15511bf |
775 | $self->add_path( $source, $target, $sig.$self->ac_label ) |
15d2d3df |
776 | unless $self->has_path( $source, $target, $sig ); |
777 | } |
e2902068 |
778 | } |
779 | |
4a8828f0 |
780 | sub common_readings { |
781 | my $self = shift; |
782 | my @common = grep { $_->is_common } $self->readings(); |
4cdd82f1 |
783 | return sort { $a->position->cmp_with( $b->position ) } @common; |
4a8828f0 |
784 | } |
785 | |
786 | # Calculate the relative positions of nodes in the graph, if they |
787 | # were not given to us. |
788 | sub calculate_positions { |
3a1f2523 |
789 | my( $self, @ordered_common ) = @_; |
4a8828f0 |
790 | |
4cdd82f1 |
791 | # First assign positions to all the common nodes. |
792 | my $l = 1; |
793 | foreach my $oc ( @ordered_common ) { |
794 | $oc->position( $l++, 1 ); |
4a8828f0 |
795 | } |
3a1f2523 |
796 | |
4cdd82f1 |
797 | if( $self->linear ) { |
798 | # For the space between each common node, we have to find all the chains |
799 | # from all the witnesses. The longest chain gives us our max, and the |
800 | # others get min/max ranges to fit. |
801 | my $first = shift @ordered_common; |
802 | while( @ordered_common ) { |
803 | my %paths; |
804 | my $next = shift @ordered_common; |
805 | my $longest = 0; |
806 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
807 | # Key to the path is not important; we just have to get |
808 | # all unique paths. |
809 | my $length = $self->_track_paths( \%paths, $first, $next, $wit->sigil ); |
810 | $longest = $length unless $longest > $length; |
811 | if( $wit->has_ante_corr ) { |
812 | my $length = $self->_track_paths( \%paths, $first, $next, |
813 | $wit->sigil.$self->ac_label, $wit->sigil ); |
814 | $longest = $length unless $longest > $length; |
815 | } |
816 | } |
817 | |
818 | # Transform the path values from unique strings to arrays. |
f6066bac |
819 | my @all_paths; |
4cdd82f1 |
820 | foreach my $k ( keys %paths ) { |
f6066bac |
821 | my @v = split( /\s+/, $k ); |
822 | push( @all_paths, \@v ); |
4cdd82f1 |
823 | } |
f6066bac |
824 | @all_paths = sort { scalar @$b <=> scalar @$a } @all_paths; |
4cdd82f1 |
825 | |
826 | # Now %paths has all the unique paths, and we know how long the |
827 | # longest of these is. Assign positions, starting with the |
828 | # longest. All non-common positions start at 2. |
f6066bac |
829 | foreach my $path ( @all_paths ) { |
830 | # Initially each element has a minimum position of 2 |
831 | # plus its position in the array (1 is the common |
832 | # node), and a max position of the longest array |
833 | # length minus its position in the array. |
4cdd82f1 |
834 | my $range = $longest - scalar @$path; |
f6066bac |
835 | my $min = 2; |
4cdd82f1 |
836 | foreach my $i ( 0 .. $#{$path} ) { |
4cdd82f1 |
837 | my $rdg = $self->reading( $path->[$i] ); |
f6066bac |
838 | if( $rdg->has_position ) { |
839 | # This reading has already had a more specific |
840 | # position set, so we need to take that into |
841 | # account when calculating the min and max for |
842 | # the next reading. |
843 | my $rangeminus = $rdg->position->min - $min; |
844 | $min = $rdg->position->min + 1; |
845 | $range = $range - $rangeminus; |
846 | if( $range < 0 ) { |
847 | print STDERR "Negative range for position! " . $rdg->name . "\n"; # May remove this warning |
848 | $range = 0; |
849 | } |
850 | } else { |
4cdd82f1 |
851 | $rdg->position( $first->position->common, $min, $min+$range ); |
f6066bac |
852 | $min++; |
853 | $longest = $min+$range-2 unless $longest+2 > $min+$range; # min starts at 2 but longest assumes 0 start |
854 | } |
855 | } |
856 | } |
857 | # Now go through again and make sure the positions are |
858 | # monotonic. Do this until they are. |
859 | my $monotonic = 0; |
860 | my $counter = 0; |
861 | until( $monotonic ) { |
862 | $monotonic = 1; |
863 | $counter++; |
864 | foreach my $path ( @all_paths ) { |
865 | foreach my $i ( 0 .. $#{$path} ) { |
866 | my $rdg = $self->reading( $path->[$i] ); |
867 | my $prior = $self->reading( $path->[$i-1] ) if $i > 0; |
868 | my $next = $self->reading( $path->[$i+1] ) if $i < $#{$path}; |
869 | if( $prior && $rdg->position->min <= $prior->position->min ) { |
870 | $monotonic = 0; |
871 | $rdg->position->min( $prior->position->min + 1 ); |
872 | } |
873 | if( $next && $rdg->position->max >= $next->position->max ) { |
874 | $monotonic = 0; |
875 | if( $next->position->max - 1 >= $rdg->position->min ) { |
876 | # If moving rdg/max down would not send it below |
877 | # rdg/min, do that. |
878 | $rdg->position->max( $next->position->max - 1 ); |
879 | } else { |
880 | # Otherwise increase next/max. |
881 | $next->position->max( $rdg->position->max + 1 ); |
882 | # ...min will be fixed on the next pass. |
883 | } |
884 | } |
4cdd82f1 |
885 | } |
886 | } |
f6066bac |
887 | if( $counter > $#all_paths + 1 && !$monotonic ) { |
888 | # We risk an infinite loop. Get out of here. |
889 | warn "Still not monotonic after $counter passes at common point " |
890 | . $first->position->common; |
891 | last; |
892 | } |
4cdd82f1 |
893 | } |
f6066bac |
894 | print STDERR "Took $counter passes for monotonicity at " . $first->position->common. "\n" |
895 | if $counter > 1; |
4cdd82f1 |
896 | |
897 | $first = $next; |
930ff666 |
898 | } |
f6066bac |
899 | |
4cdd82f1 |
900 | } else { |
901 | |
902 | # Non-linear positions are pretty much impossible to pin down. |
903 | # Any reading might appear anywhere in the graph. I guess we |
904 | # can do positions where there aren't transpositions... |
930ff666 |
905 | |
de51424a |
906 | } |
4cdd82f1 |
907 | $self->init_lemmata(); |
8e1394aa |
908 | } |
3a1f2523 |
909 | |
4cdd82f1 |
910 | # Helper function for the guts of calculate_positions. |
911 | sub _track_paths { |
912 | my $self = shift; |
913 | my $track_hash = shift; |
914 | # Args are first, last, wit, backup |
915 | my @path = $self->reading_sequence( @_ ); |
916 | # Top and tail the array |
917 | shift @path; |
918 | pop @path; |
f6066bac |
919 | $track_hash->{join( ' ', map { $_->name } @path )} = $_[2] |
4cdd82f1 |
920 | if @path; |
921 | return @path; |
922 | } |
923 | |
924 | sub possible_positions { |
3a1f2523 |
925 | my $self = shift; |
4cdd82f1 |
926 | my @answer; |
3a1f2523 |
927 | my %positions = (); |
4cdd82f1 |
928 | foreach my $r ( $self->readings ) { |
929 | next unless $r->has_position; |
930 | $positions{$r->position->maxref} = 1; |
931 | } |
932 | @answer = keys %positions; |
de51424a |
933 | return @answer; |
3a1f2523 |
934 | } |
935 | |
4cdd82f1 |
936 | # TODO think about indexing this. |
3a1f2523 |
937 | sub readings_at_position { |
4cdd82f1 |
938 | my( $self, $position, $strict ) = @_; |
939 | unless( ref( $position ) eq 'Text::Tradition::Collation::Position' ) { |
940 | $position = Text::Tradition::Collation::Position->new( $position ); |
941 | } |
942 | my @answer; |
943 | foreach my $r ( $self->readings ) { |
944 | push( @answer, $r ) if $r->is_at_position( $position, $strict ); |
945 | } |
3a1f2523 |
946 | return @answer; |
947 | } |
948 | |
949 | ## Lemmatizer functions |
950 | |
951 | sub init_lemmata { |
952 | my $self = shift; |
4cdd82f1 |
953 | |
954 | foreach my $position ( $self->possible_positions ) { |
3a1f2523 |
955 | $self->lemmata->{$position} = undef; |
956 | } |
957 | |
958 | foreach my $cr ( $self->common_readings ) { |
4cdd82f1 |
959 | $self->lemmata->{$cr->position->maxref} = $cr->name; |
3a1f2523 |
960 | } |
961 | } |
962 | |
963 | =item B<lemma_readings> |
964 | |
965 | my @state = $graph->lemma_readings( @readings_delemmatized ); |
966 | |
967 | Takes a list of readings that have just been delemmatized, and returns |
968 | a set of tuples of the form ['reading', 'state'] that indicates what |
969 | changes need to be made to the graph. |
970 | |
971 | =over |
972 | |
973 | =item * |
974 | |
975 | A state of 1 means 'lemmatize this reading' |
976 | |
977 | =item * |
978 | |
979 | A state of 0 means 'delemmatize this reading' |
980 | |
981 | =item * |
982 | |
983 | A state of undef means 'an ellipsis belongs in the text here because |
984 | no decision has been made / an earlier decision was backed out' |
985 | |
986 | =back |
987 | |
988 | =cut |
989 | |
990 | sub lemma_readings { |
991 | my( $self, @toggled_off_nodes ) = @_; |
992 | |
993 | # First get the positions of those nodes which have been |
994 | # toggled off. |
995 | my $positions_off = {}; |
4cdd82f1 |
996 | map { $positions_off->{ $_->position->reference } = $_->name } |
997 | @toggled_off_nodes; |
de51424a |
998 | |
3a1f2523 |
999 | # Now for each position, we have to see if a node is on, and we |
4cdd82f1 |
1000 | # have to see if a node has been turned off. The lemmata hash |
1001 | # should contain fixed positions, range positions whose node was |
1002 | # just turned off, and range positions whose node is on. |
3a1f2523 |
1003 | my @answer; |
4cdd82f1 |
1004 | my %fixed_positions; |
1005 | # TODO One of these is probably redundant. |
1006 | map { $fixed_positions{$_} = 0 } keys %{$self->lemmata}; |
1007 | map { $fixed_positions{$_} = 0 } keys %{$positions_off}; |
1008 | map { $fixed_positions{$_} = 1 } $self->possible_positions; |
1009 | foreach my $pos ( sort { Text::Tradition::Collation::Position::str_cmp( $a, $b ) } keys %fixed_positions ) { |
3a1f2523 |
1010 | # Find the state of this position. If there is an active node, |
1011 | # its name will be the state; otherwise the state will be 0 |
1012 | # (nothing at this position) or undef (ellipsis at this position) |
4cdd82f1 |
1013 | my $active = undef; |
1014 | $active = $self->lemmata->{$pos} if exists $self->lemmata->{$pos}; |
3a1f2523 |
1015 | |
1016 | # Is there a formerly active node that was toggled off? |
1017 | if( exists( $positions_off->{$pos} ) ) { |
1018 | my $off_node = $positions_off->{$pos}; |
1019 | if( $active && $active ne $off_node) { |
1020 | push( @answer, [ $off_node, 0 ], [ $active, 1 ] ); |
1021 | } else { |
4cdd82f1 |
1022 | unless( $fixed_positions{$pos} ) { |
1023 | $active = 0; |
1024 | delete $self->lemmata->{$pos}; |
1025 | } |
3a1f2523 |
1026 | push( @answer, [ $off_node, $active ] ); |
1027 | } |
1028 | |
1029 | # No formerly active node, so we just see if there is a currently |
1030 | # active one. |
1031 | } elsif( $active ) { |
1032 | # Push the active node, whatever it is. |
1033 | push( @answer, [ $active, 1 ] ); |
1034 | } else { |
1035 | # Push the state that is there. Arbitrarily use the first node |
1036 | # at that position. |
1037 | my @pos_nodes = $self->readings_at_position( $pos ); |
de51424a |
1038 | push( @answer, [ $pos_nodes[0]->name, $self->lemmata->{$pos} ] ); |
4cdd82f1 |
1039 | delete $self->lemmata->{$pos} unless $fixed_positions{$pos}; |
3a1f2523 |
1040 | } |
1041 | } |
4cdd82f1 |
1042 | |
3a1f2523 |
1043 | return @answer; |
1044 | } |
1045 | |
de51424a |
1046 | =item B<toggle_reading> |
1047 | |
1048 | my @readings_delemmatized = $graph->toggle_reading( $reading_name ); |
1049 | |
1050 | Takes a reading node name, and either lemmatizes or de-lemmatizes |
1051 | it. Returns a list of all readings that are de-lemmatized as a result |
1052 | of the toggle. |
1053 | |
1054 | =cut |
1055 | |
1056 | sub toggle_reading { |
1057 | my( $self, $rname ) = @_; |
1058 | |
1059 | return unless $rname; |
1060 | my $reading = $self->reading( $rname ); |
1061 | if( !$reading || $reading->is_common() ) { |
1062 | # Do nothing, it's a common node. |
1063 | return; |
1064 | } |
1065 | |
1066 | my $pos = $reading->position; |
4cdd82f1 |
1067 | my $fixed = $reading->position->fixed; |
1068 | my $old_state = $self->lemmata->{$pos->reference}; |
1069 | |
de51424a |
1070 | my @readings_off; |
1071 | if( $old_state && $old_state eq $rname ) { |
1072 | # Turn off the node. We turn on no others by default. |
1073 | push( @readings_off, $reading ); |
1074 | } else { |
1075 | # Turn on the node. |
4cdd82f1 |
1076 | $self->lemmata->{$pos->reference} = $rname; |
1077 | # Any other 'on' readings in the same position should be off |
1078 | # if we have a fixed position. |
1079 | push( @readings_off, $self->same_position_as( $reading, 1 ) ) |
1080 | if $pos->fixed; |
de51424a |
1081 | # Any node that is an identical transposed one should be off. |
1082 | push( @readings_off, $reading->identical_readings ); |
1083 | } |
1084 | @readings_off = unique_list( @readings_off ); |
4cdd82f1 |
1085 | |
de51424a |
1086 | # Turn off the readings that need to be turned off. |
1087 | my @readings_delemmatized; |
1088 | foreach my $n ( @readings_off ) { |
4cdd82f1 |
1089 | my $npos = $n->position; |
1090 | my $state = undef; |
1091 | $state = $self->lemmata->{$npos->reference} |
1092 | if defined $self->lemmata->{$npos->reference}; |
de51424a |
1093 | if( $state && $state eq $n->name ) { |
1094 | # this reading is still on, so turn it off |
1095 | push( @readings_delemmatized, $n ); |
1096 | my $new_state = undef; |
4cdd82f1 |
1097 | if( $npos->fixed && $n eq $reading ) { |
de51424a |
1098 | # This is the reading that was clicked, so if there are no |
4cdd82f1 |
1099 | # other readings there and this is a fixed position, turn off |
1100 | # the position. In all other cases, restore the ellipsis. |
1101 | my @other_n = $self->same_position_as( $n ); # TODO do we need strict? |
de51424a |
1102 | $new_state = 0 unless @other_n; |
1103 | } |
4cdd82f1 |
1104 | $self->lemmata->{$npos->reference} = $new_state; |
de51424a |
1105 | } elsif( $old_state && $old_state eq $n->name ) { |
1106 | # another reading has already been turned on here |
1107 | push( @readings_delemmatized, $n ); |
1108 | } # else some other reading was on anyway, so pass. |
1109 | } |
1110 | return @readings_delemmatized; |
1111 | } |
1112 | |
1113 | sub same_position_as { |
4cdd82f1 |
1114 | my( $self, $reading, $strict ) = @_; |
de51424a |
1115 | my $pos = $reading->position; |
4cdd82f1 |
1116 | my %onpath = ( $reading->name => 1 ); |
1117 | # TODO This might not always be sufficient. We really want to |
1118 | # exclude all readings on this one's path between its two |
1119 | # common points. |
1120 | map { $onpath{$_->name} = 1 } $reading->neighbor_readings; |
1121 | my @same = grep { !$onpath{$_->name} } |
1122 | $self->readings_at_position( $reading->position, $strict ); |
de51424a |
1123 | return @same; |
1124 | } |
3a1f2523 |
1125 | |
4a8828f0 |
1126 | # Return the string that joins together a list of witnesses for |
1127 | # display on a single path. |
1128 | sub path_label { |
1129 | my $self = shift; |
1130 | return join( $self->wit_list_separator, @_ ); |
1131 | } |
1132 | |
1133 | sub witnesses_of_label { |
de51424a |
1134 | my( $self, $label ) = @_; |
4a8828f0 |
1135 | my $regex = $self->wit_list_separator; |
de51424a |
1136 | my @answer = split( /\Q$regex\E/, $label ); |
1137 | return @answer; |
4a8828f0 |
1138 | } |
8e1394aa |
1139 | |
de51424a |
1140 | sub unique_list { |
1141 | my( @list ) = @_; |
1142 | my %h; |
1143 | map { $h{$_->name} = $_ } @list; |
1144 | return values( %h ); |
1145 | } |
1146 | |
1f563ac3 |
1147 | sub add_hash_entry { |
1148 | my( $hash, $key, $entry ) = @_; |
1149 | if( exists $hash->{$key} ) { |
1150 | push( @{$hash->{$key}}, $entry ); |
1151 | } else { |
1152 | $hash->{$key} = [ $entry ]; |
1153 | } |
1154 | } |
1155 | |
dd3b58b0 |
1156 | no Moose; |
1157 | __PACKAGE__->meta->make_immutable; |