Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition::Collation; |
d047cd52 |
2 | |
910a0a6d |
3 | use Encode qw( decode_utf8 ); |
4 | use File::Temp; |
d047cd52 |
5 | use Graph::Easy; |
8e1394aa |
6 | use IPC::Run qw( run binary ); |
910a0a6d |
7 | use Text::CSV_XS; |
3265b0ce |
8 | use Text::Tradition::Collation::Path; |
b15511bf |
9 | use Text::Tradition::Collation::Reading; |
10 | use Text::Tradition::Collation::Relationship; |
11 | use Text::Tradition::Collation::Segment; |
df6d9812 |
12 | use XML::LibXML; |
dd3b58b0 |
13 | use Moose; |
14 | |
15 | has 'graph' => ( |
d047cd52 |
16 | is => 'ro', |
17 | isa => 'Graph::Easy', |
18 | handles => { |
910a0a6d |
19 | add_reading => 'add_node', |
20 | del_reading => 'del_node', |
21 | del_segment => 'del_node', |
22 | add_path => 'add_edge', |
23 | del_path => 'del_edge', |
24 | reading => 'node', |
25 | path => 'edge', |
26 | readings => 'nodes', |
27 | segments => 'nodes', |
28 | paths => 'edges', |
29 | relationships => 'edges', |
d047cd52 |
30 | }, |
31 | default => sub { Graph::Easy->new( undirected => 0 ) }, |
32 | ); |
910a0a6d |
33 | |
dd3b58b0 |
34 | |
f6066bac |
35 | has 'tradition' => ( # TODO should this not be ro? |
8e1394aa |
36 | is => 'rw', |
d047cd52 |
37 | isa => 'Text::Tradition', |
38 | ); |
dd3b58b0 |
39 | |
8e1394aa |
40 | has 'svg' => ( |
41 | is => 'ro', |
42 | isa => 'Str', |
43 | writer => '_save_svg', |
44 | predicate => 'has_svg', |
45 | ); |
46 | |
8e1394aa |
47 | has 'graphml' => ( |
48 | is => 'ro', |
df6d9812 |
49 | isa => 'Str', |
8e1394aa |
50 | writer => '_save_graphml', |
51 | predicate => 'has_graphml', |
52 | ); |
53 | |
910a0a6d |
54 | has 'csv' => ( |
55 | is => 'ro', |
56 | isa => 'Str', |
57 | writer => '_save_csv', |
58 | predicate => 'has_csv', |
59 | ); |
60 | |
3a1f2523 |
61 | # Keeps track of the lemmas within the collation. At most one lemma |
62 | # per position in the graph. |
63 | has 'lemmata' => ( |
64 | is => 'ro', |
65 | isa => 'HashRef[Maybe[Str]]', |
66 | default => sub { {} }, |
67 | ); |
68 | |
4a8828f0 |
69 | has 'wit_list_separator' => ( |
7854e12e |
70 | is => 'rw', |
71 | isa => 'Str', |
72 | default => ', ', |
73 | ); |
74 | |
75 | has 'baselabel' => ( |
76 | is => 'rw', |
77 | isa => 'Str', |
78 | default => 'base text', |
79 | ); |
4a8828f0 |
80 | |
1f563ac3 |
81 | has 'collapsed' => ( |
15d2d3df |
82 | is => 'rw', |
83 | isa => 'Bool', |
84 | ); |
85 | |
86 | has 'linear' => ( |
87 | is => 'rw', |
88 | isa => 'Bool', |
89 | default => 1, |
90 | ); |
1f563ac3 |
91 | |
ef9d481f |
92 | has 'ac_label' => ( |
93 | is => 'rw', |
94 | isa => 'Str', |
95 | default => ' (a.c.)', |
96 | ); |
97 | |
1f563ac3 |
98 | |
dd3b58b0 |
99 | # The collation can be created two ways: |
100 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
101 | # the results as in 2. |
102 | # 2. Read a pre-prepared collation in one of a variety of formats, |
103 | # and make the graph from that. |
104 | |
105 | # The graph itself will (for now) be immutable, and the positions |
106 | # within the graph will also be immutable. We need to calculate those |
107 | # positions upon graph construction. The equivalences between graph |
108 | # nodes will be mutable, entirely determined by the user (or possibly |
109 | # by some semantic pre-processing provided by the user.) So the |
110 | # constructor should just make an empty equivalences object. The |
111 | # constructor will also need to make the witness objects, if we didn't |
112 | # come through option 1. |
113 | |
d047cd52 |
114 | sub BUILD { |
115 | my( $self, $args ) = @_; |
8e1394aa |
116 | $self->graph->use_class('node', 'Text::Tradition::Collation::Reading'); |
3265b0ce |
117 | $self->graph->use_class('edge', 'Text::Tradition::Collation::Path'); |
d047cd52 |
118 | |
4a8828f0 |
119 | # Pass through any graph-specific options. |
120 | my $shape = exists( $args->{'shape'} ) ? $args->{'shape'} : 'ellipse'; |
121 | $self->graph->set_attribute( 'node', 'shape', $shape ); |
d047cd52 |
122 | } |
784877d9 |
123 | |
7854e12e |
124 | # Wrapper around add_path |
125 | |
126 | around add_path => sub { |
127 | my $orig = shift; |
128 | my $self = shift; |
129 | |
130 | # Make sure there are three arguments |
131 | unless( @_ == 3 ) { |
910a0a6d |
132 | warn "Call add_path with args source, target, witness"; |
133 | return; |
7854e12e |
134 | } |
135 | # Make sure the proposed path does not yet exist |
b15511bf |
136 | # NOTE 'reading' will currently return readings and segments |
7854e12e |
137 | my( $source, $target, $wit ) = @_; |
138 | $source = $self->reading( $source ) |
910a0a6d |
139 | unless ref( $source ) eq 'Text::Tradition::Collation::Reading'; |
7854e12e |
140 | $target = $self->reading( $target ) |
910a0a6d |
141 | unless ref( $target ) eq 'Text::Tradition::Collation::Reading'; |
7854e12e |
142 | foreach my $path ( $source->edges_to( $target ) ) { |
910a0a6d |
143 | if( $path->label eq $wit && $path->class eq 'edge.path' ) { |
144 | return; |
145 | } |
7854e12e |
146 | } |
147 | # Do the deed |
148 | $self->$orig( @_ ); |
149 | }; |
150 | |
3265b0ce |
151 | # Wrapper around paths |
152 | around paths => sub { |
153 | my $orig = shift; |
154 | my $self = shift; |
155 | |
b15511bf |
156 | my @result = grep { $_->sub_class eq 'path' } $self->$orig( @_ ); |
df6d9812 |
157 | return @result; |
158 | }; |
159 | |
160 | around relationships => sub { |
161 | my $orig = shift; |
162 | my $self = shift; |
b15511bf |
163 | my @result = grep { $_->sub_class eq 'relationship' } $self->$orig( @_ ); |
164 | return @result; |
165 | }; |
166 | |
167 | around readings => sub { |
168 | my $orig = shift; |
169 | my $self = shift; |
170 | my @result = grep { $_->sub_class ne 'segment' } $self->$orig( @_ ); |
171 | return @result; |
172 | }; |
173 | |
174 | around segments => sub { |
175 | my $orig = shift; |
176 | my $self = shift; |
177 | my @result = grep { $_->sub_class eq 'segment' } $self->$orig( @_ ); |
3265b0ce |
178 | return @result; |
179 | }; |
180 | |
7854e12e |
181 | # Wrapper around merge_nodes |
784877d9 |
182 | |
183 | sub merge_readings { |
184 | my $self = shift; |
185 | my $first_node = shift; |
186 | my $second_node = shift; |
187 | $first_node->merge_from( $second_node ); |
188 | unshift( @_, $first_node, $second_node ); |
189 | return $self->graph->merge_nodes( @_ ); |
190 | } |
191 | |
15d2d3df |
192 | # Extra graph-alike utility |
193 | sub has_path { |
194 | my( $self, $source, $target, $label ) = @_; |
195 | my @paths = $source->edges_to( $target ); |
196 | my @relevant = grep { $_->label eq $label } @paths; |
b15511bf |
197 | return scalar @relevant; |
198 | } |
199 | |
200 | ## Dealing with groups of readings, i.e. segments. |
201 | |
202 | sub add_segment { |
203 | my( $self, @items ) = @_; |
204 | my $segment = Text::Tradition::Collation::Segment->new( 'members' => \@items ); |
205 | return $segment; |
15d2d3df |
206 | } |
207 | |
3265b0ce |
208 | ## Dealing with relationships between readings. This is a different |
4cdd82f1 |
209 | ## sort of graph edge. Return a success/failure value and a list of |
210 | ## node pairs that have been linked. |
3265b0ce |
211 | |
212 | sub add_relationship { |
b15511bf |
213 | my( $self, $source, $target, $options ) = @_; |
ef9d481f |
214 | |
215 | # Make sure there is not another relationship between these two |
b15511bf |
216 | # readings or segments already |
ef9d481f |
217 | $source = $self->reading( $source ) |
910a0a6d |
218 | unless ref( $source ) && $source->isa( 'Graph::Easy::Node' ); |
ef9d481f |
219 | $target = $self->reading( $target ) |
910a0a6d |
220 | unless ref( $target ) && $target->isa( 'Graph::Easy::Node' ); |
4cdd82f1 |
221 | foreach my $rel ( $source->edges_to( $target ), $target->edges_to( $source ) ) { |
910a0a6d |
222 | if( $rel->class eq 'edge.relationship' ) { |
223 | return ( undef, "Relationship already exists between these readings" ); |
224 | } |
4cdd82f1 |
225 | } |
910a0a6d |
226 | if( $options->{'equal_rank'} && !relationship_valid( $source, $target ) ) { |
227 | return ( undef, 'Relationship creates witness loop' ); |
ef9d481f |
228 | } |
229 | |
910a0a6d |
230 | # TODO Think about positional hilarity if relationships are added after positions |
231 | # are assigned. |
232 | |
4cdd82f1 |
233 | my @joined = ( [ $source->name, $target->name ] ); # Keep track of the nodes we join. |
234 | |
235 | $options->{'this_relation'} = [ $source, $target ]; |
f6066bac |
236 | my $rel; |
237 | eval { $rel = Text::Tradition::Collation::Relationship->new( %$options ) }; |
238 | if( $@ ) { |
239 | return ( undef, $@ ); |
240 | } |
3265b0ce |
241 | $self->graph->add_edge( $source, $target, $rel ); |
910a0a6d |
242 | |
243 | # TODO Handle global relationship setting |
244 | |
4cdd82f1 |
245 | return( 1, @joined ); |
3265b0ce |
246 | } |
247 | |
910a0a6d |
248 | sub relationship_valid { |
249 | my( $source, $target ) = @_; |
250 | # Check that linking the source and target in a relationship won't lead |
251 | # to a path loop for any witness. |
252 | my @proposed_related = ( $source, $target ); |
253 | push( @proposed_related, $source->related_readings ); |
254 | push( @proposed_related, $target->related_readings ); |
255 | my %pr_ids; |
256 | map { $pr_ids{ $_->name } = 1 } @proposed_related; |
257 | # The lists of 'in' and 'out' should not have any element that appears |
258 | # in 'proposed_related'. |
259 | foreach my $pr ( @proposed_related ) { |
260 | foreach my $e ( $pr->incoming ) { |
261 | if( exists $pr_ids{ $e->from->name } ) { |
262 | return 0; |
263 | } |
264 | } |
265 | foreach my $e ( $pr->outgoing ) { |
266 | if( exists $pr_ids{ $e->to->name } ) { |
267 | return 0; |
268 | } |
269 | } |
270 | } |
271 | return 1; |
272 | } |
273 | |
8e1394aa |
274 | =head2 Output method(s) |
275 | |
276 | =over |
277 | |
278 | =item B<as_svg> |
279 | |
280 | print $graph->as_svg( $recalculate ); |
281 | |
282 | Returns an SVG string that represents the graph. Uses GraphViz to do |
4a8828f0 |
283 | this, because Graph::Easy doesn\'t cope well with long graphs. Unless |
8e1394aa |
284 | $recalculate is passed (and is a true value), the method will return a |
285 | cached copy of the SVG after the first call to the method. |
286 | |
287 | =cut |
288 | |
289 | sub as_svg { |
290 | my( $self, $recalc ) = @_; |
291 | return $self->svg if $self->has_svg; |
292 | |
3265b0ce |
293 | $self->collapse_graph_paths(); |
8e1394aa |
294 | |
295 | my @cmd = qw/dot -Tsvg/; |
296 | my( $svg, $err ); |
910a0a6d |
297 | my $dotfile = File::Temp->new(); |
d9e873d0 |
298 | ## TODO REMOVE |
299 | $dotfile->unlink_on_destroy(0); |
910a0a6d |
300 | binmode $dotfile, ':utf8'; |
301 | print $dotfile $self->as_dot(); |
302 | push( @cmd, $dotfile->filename ); |
303 | run( \@cmd, ">", binary(), \$svg ); |
304 | $svg = decode_utf8( $svg ); |
df6d9812 |
305 | $self->_save_svg( $svg ); |
3265b0ce |
306 | $self->expand_graph_paths(); |
8e1394aa |
307 | return $svg; |
308 | } |
309 | |
df6d9812 |
310 | =item B<as_dot> |
311 | |
312 | print $graph->as_dot( $view, $recalculate ); |
313 | |
314 | Returns a string that is the collation graph expressed in dot |
315 | (i.e. GraphViz) format. The 'view' argument determines what kind of |
316 | graph is produced. |
317 | * 'path': a graph of witness paths through the collation (DEFAULT) |
318 | * 'relationship': a graph of how collation readings relate to |
319 | each other |
320 | |
321 | =cut |
322 | |
323 | sub as_dot { |
324 | my( $self, $view ) = @_; |
325 | $view = 'path' unless $view; |
326 | # TODO consider making some of these things configurable |
327 | my $dot = sprintf( "digraph %s {\n", $self->tradition->name ); |
328 | $dot .= "\tedge [ arrowhead=open ];\n"; |
329 | $dot .= "\tgraph [ rankdir=LR ];\n"; |
330 | $dot .= sprintf( "\tnode [ fontsize=%d, fillcolor=%s, style=%s, shape=%s ];\n", |
910a0a6d |
331 | 11, "white", "filled", $self->graph->get_attribute( 'node', 'shape' ) ); |
df6d9812 |
332 | |
333 | foreach my $reading ( $self->readings ) { |
910a0a6d |
334 | # Need not output nodes without separate labels |
335 | next if $reading->name eq $reading->label; |
336 | # TODO output readings or segments, but not both |
337 | next if $reading->class eq 'node.segment'; |
338 | $dot .= sprintf( "\t\"%s\" [ label=\"%s\" ];\n", $reading->name, $reading->label ); |
df6d9812 |
339 | } |
340 | |
341 | my @edges = $view eq 'relationship' ? $self->relationships : $self->paths; |
342 | foreach my $edge ( @edges ) { |
910a0a6d |
343 | my %variables = ( 'color' => '#000000', |
344 | 'fontcolor' => '#000000', |
345 | 'label' => $edge->label, |
346 | ); |
347 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
348 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", |
349 | $edge->from->name, $edge->to->name, $varopts ); |
df6d9812 |
350 | } |
df6d9812 |
351 | $dot .= "}\n"; |
352 | return $dot; |
353 | } |
354 | |
8e1394aa |
355 | =item B<as_graphml> |
356 | |
357 | print $graph->as_graphml( $recalculate ) |
358 | |
359 | Returns a GraphML representation of the collation graph, with |
360 | transposition information and position information. Unless |
361 | $recalculate is passed (and is a true value), the method will return a |
362 | cached copy of the SVG after the first call to the method. |
363 | |
364 | =cut |
365 | |
366 | sub as_graphml { |
367 | my( $self, $recalc ) = @_; |
368 | return $self->graphml if $self->has_graphml; |
369 | |
370 | # Some namespaces |
371 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
372 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
373 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
910a0a6d |
374 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
8e1394aa |
375 | |
376 | # Create the document and root node |
377 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
378 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
379 | $graphml->setDocumentElement( $root ); |
380 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
381 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
382 | |
f6066bac |
383 | # TODO Add some global graph data |
384 | |
8e1394aa |
385 | # Add the data keys for nodes |
ef9d481f |
386 | my %node_data_keys; |
387 | my $ndi = 0; |
d9e873d0 |
388 | foreach my $datum ( qw/ name reading identical rank class / ) { |
910a0a6d |
389 | $node_data_keys{$datum} = 'dn'.$ndi++; |
390 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
391 | $key->setAttribute( 'attr.name', $datum ); |
392 | $key->setAttribute( 'attr.type', 'string' ); |
393 | $key->setAttribute( 'for', 'node' ); |
394 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
395 | } |
396 | |
df6d9812 |
397 | # Add the data keys for edges, i.e. witnesses |
ef9d481f |
398 | my $edi = 0; |
399 | my %edge_data_keys; |
b15511bf |
400 | foreach my $edge_key( qw/ witness_main witness_ante_corr relationship class / ) { |
910a0a6d |
401 | $edge_data_keys{$edge_key} = 'de'.$edi++; |
402 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
403 | $key->setAttribute( 'attr.name', $edge_key ); |
404 | $key->setAttribute( 'attr.type', 'string' ); |
405 | $key->setAttribute( 'for', 'edge' ); |
406 | $key->setAttribute( 'id', $edge_data_keys{$edge_key} ); |
8e1394aa |
407 | } |
ef9d481f |
408 | |
8e1394aa |
409 | # Add the graph, its nodes, and its edges |
410 | my $graph = $root->addNewChild( $graphml_ns, 'graph' ); |
411 | $graph->setAttribute( 'edgedefault', 'directed' ); |
412 | $graph->setAttribute( 'id', 'g0' ); # TODO make this meaningful |
413 | $graph->setAttribute( 'parse.edgeids', 'canonical' ); |
df6d9812 |
414 | $graph->setAttribute( 'parse.edges', scalar($self->paths) ); |
8e1394aa |
415 | $graph->setAttribute( 'parse.nodeids', 'canonical' ); |
df6d9812 |
416 | $graph->setAttribute( 'parse.nodes', scalar($self->readings) ); |
8e1394aa |
417 | $graph->setAttribute( 'parse.order', 'nodesfirst' ); |
418 | |
419 | my $node_ctr = 0; |
420 | my %node_hash; |
b15511bf |
421 | # Add our readings to the graph |
b5054ca9 |
422 | foreach my $n ( sort { $a->name cmp $b->name } $self->readings ) { |
910a0a6d |
423 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
424 | my $node_xmlid = 'n' . $node_ctr++; |
425 | $node_hash{ $n->name } = $node_xmlid; |
426 | $node_el->setAttribute( 'id', $node_xmlid ); |
427 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
428 | _add_graphml_data( $node_el, $node_data_keys{'reading'}, $n->label ); |
d9e873d0 |
429 | _add_graphml_data( $node_el, $node_data_keys{'rank'}, $n->rank ) |
430 | if $n->has_rank; |
910a0a6d |
431 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
432 | _add_graphml_data( $node_el, $node_data_keys{'identical'}, $n->primary->name ) |
433 | if $n->has_primary; |
8e1394aa |
434 | } |
435 | |
b15511bf |
436 | # Add any segments we have |
437 | foreach my $n ( sort { $a->name cmp $b->name } $self->segments ) { |
910a0a6d |
438 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
439 | my $node_xmlid = 'n' . $node_ctr++; |
440 | $node_hash{ $n->name } = $node_xmlid; |
441 | $node_el->setAttribute( 'id', $node_xmlid ); |
442 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
443 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
b15511bf |
444 | } |
445 | |
446 | # Add the path, relationship, and segment edges |
df6d9812 |
447 | my $edge_ctr = 0; |
ef9d481f |
448 | foreach my $e ( sort { $a->from->name cmp $b->from->name } $self->graph->edges() ) { |
910a0a6d |
449 | my( $name, $from, $to ) = ( 'e'.$edge_ctr++, |
450 | $node_hash{ $e->from->name() }, |
451 | $node_hash{ $e->to->name() } ); |
452 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
453 | $edge_el->setAttribute( 'source', $from ); |
454 | $edge_el->setAttribute( 'target', $to ); |
455 | $edge_el->setAttribute( 'id', $name ); |
456 | # Add the edge class |
457 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, $e->sub_class ); |
458 | if( $e->sub_class eq 'path' ) { |
459 | # It's a witness path, so add the witness |
460 | my $base = $e->label; |
461 | my $key = $edge_data_keys{'witness_main'}; |
462 | # TODO kind of hacky |
463 | if( $e->label =~ /^(.*?)\s+(\(a\.c\.\))$/ ) { |
464 | $base = $1; |
465 | $key = $edge_data_keys{'witness_ante_corr'}; |
466 | } |
467 | _add_graphml_data( $edge_el, $key, $base ); |
468 | } elsif( $e->sub_class eq 'relationship' ) { |
469 | # It's a relationship |
470 | _add_graphml_data( $edge_el, $edge_data_keys{'relationship'}, $e->label ); |
471 | } # else a segment, nothing to record but source, target, class |
8e1394aa |
472 | } |
473 | |
474 | # Return the thing |
df6d9812 |
475 | $self->_save_graphml( $graphml->toString(1) ); |
476 | return $graphml->toString(1); |
477 | } |
478 | |
b15511bf |
479 | sub _add_graphml_data { |
480 | my( $el, $key, $value ) = @_; |
481 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
482 | return unless defined $value; |
483 | $data_el->setAttribute( 'key', $key ); |
484 | $data_el->appendText( $value ); |
8e1394aa |
485 | } |
486 | |
910a0a6d |
487 | =item B<as_csv> |
488 | |
489 | print $graph->as_csv( $recalculate ) |
490 | |
491 | Returns a CSV alignment table representation of the collation graph, one |
492 | row per witness (or witness uncorrected.) Unless $recalculate is passed |
493 | (and is a true value), the method will return a cached copy of the CSV |
494 | after the first call to the method. |
495 | |
496 | =cut |
497 | |
498 | sub as_csv { |
499 | my( $self, $recalc ) = @_; |
500 | return $self->csv if $self->has_csv; |
501 | my $table = $self->make_alignment_table; |
502 | my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } ); |
503 | my @result; |
504 | foreach my $row ( @$table ) { |
505 | $csv->combine( @$row ); |
506 | push( @result, decode_utf8( $csv->string ) ); |
507 | } |
508 | $self->_save_csv( join( "\n", @result ) ); |
509 | return $self->csv; |
510 | } |
511 | |
512 | sub make_alignment_table { |
513 | my $self = shift; |
514 | unless( $self->linear ) { |
515 | warn "Need a linear graph in order to make an alignment table"; |
516 | return; |
517 | } |
518 | my $table; |
519 | my @all_pos = sort { $a <=> $b } $self->possible_positions; |
520 | foreach my $wit ( $self->tradition->witnesses ) { |
521 | my @row = _make_witness_row( $wit->path, \@all_pos ); |
522 | unshift( @row, $wit->sigil ); |
523 | push( @$table, \@row ); |
524 | if( $wit->has_ante_corr ) { |
525 | my @ac_row = _make_witness_row( $wit->uncorrected_path, \@all_pos ); |
526 | unshift( @ac_row, $wit->sigil . $self->ac_label ); |
527 | push( @$table, \@ac_row ); |
528 | } |
529 | } |
530 | # Return a table where the witnesses read in columns rather than rows. |
531 | my $turned = _turn_table( $table ); |
532 | return $turned; |
533 | } |
534 | |
535 | sub _make_witness_row { |
536 | my( $path, $positions ) = @_; |
537 | my %char_hash; |
538 | map { $char_hash{$_} = undef } @$positions; |
539 | foreach my $rdg ( @$path ) { |
540 | $char_hash{$rdg->rank} = $rdg->text; |
541 | } |
542 | my @row = map { $char_hash{$_} } @$positions; |
543 | return @row; |
544 | } |
545 | |
546 | # Helper to turn the witnesses along columns rather than rows. Assumes |
547 | # equal-sized rows. |
548 | sub _turn_table { |
549 | my( $table ) = @_; |
550 | my $result = []; |
551 | return $result unless scalar @$table; |
552 | my $nrows = scalar @{$table->[0]}; |
553 | foreach my $idx ( 0 .. $nrows - 1 ) { |
554 | foreach my $wit ( 0 .. $#{$table} ) { |
555 | $result->[$idx]->[$wit] = $table->[$wit]->[$idx]; |
556 | } |
557 | } |
558 | return $result; |
559 | } |
560 | |
561 | |
3265b0ce |
562 | sub collapse_graph_paths { |
1f563ac3 |
563 | my $self = shift; |
3265b0ce |
564 | # Our collation graph has an path per witness. This is great for |
1f563ac3 |
565 | # calculation purposes, but terrible for display. Thus we want to |
3265b0ce |
566 | # display only one path between any two nodes. |
1f563ac3 |
567 | |
568 | return if $self->collapsed; |
569 | |
3265b0ce |
570 | print STDERR "Collapsing witness paths in graph...\n"; |
1f563ac3 |
571 | |
572 | # Don't list out every witness if we have more than half to list. |
910a0a6d |
573 | my $majority = int( scalar( $self->tradition->witnesses ) / 2 ) + 1; |
a0093bf2 |
574 | # But don't compress if there are only a few witnesses. |
575 | $majority = 4 if $majority < 4; |
b15511bf |
576 | foreach my $node ( $self->readings ) { |
910a0a6d |
577 | my $newlabels = {}; |
578 | # We will visit each node, so we only look ahead. |
579 | foreach my $edge ( $node->outgoing() ) { |
580 | next unless $edge->class eq 'edge.path'; |
581 | add_hash_entry( $newlabels, $edge->to->name, $edge->name ); |
582 | $self->del_path( $edge ); |
583 | } |
584 | |
585 | foreach my $newdest ( keys %$newlabels ) { |
586 | my $label; |
587 | my @compressed_wits = @{$newlabels->{$newdest}}; |
588 | if( @compressed_wits < $majority ) { |
589 | $label = join( ', ', sort( @{$newlabels->{$newdest}} ) ); |
590 | } else { |
591 | ## TODO FIX THIS HACK |
592 | my @aclabels; |
593 | foreach my $wit ( @compressed_wits ) { |
594 | push( @aclabels, $wit ) if( $wit =~ /^(.*?)(\s*\(?a\.\s*c\.\)?)$/ ); |
595 | } |
596 | $label = join( ', ', 'majority', sort( @aclabels ) ); |
597 | } |
598 | |
599 | my $newpath = $self->add_path( $node, $self->reading( $newdest ), $label ); |
600 | $newpath->hidden_witnesses( \@compressed_wits ); |
601 | } |
1f563ac3 |
602 | } |
603 | |
604 | $self->collapsed( 1 ); |
605 | } |
606 | |
3265b0ce |
607 | sub expand_graph_paths { |
1f563ac3 |
608 | my $self = shift; |
3265b0ce |
609 | # Our collation graph has only one path between any two nodes. |
1f563ac3 |
610 | # This is great for display, but not so great for analysis. |
3265b0ce |
611 | # Expand this so that each witness has its own path between any |
1f563ac3 |
612 | # two reading nodes. |
613 | return unless $self->collapsed; |
614 | |
3265b0ce |
615 | print STDERR "Expanding witness paths in graph...\n"; |
3265b0ce |
616 | foreach my $path( $self->paths ) { |
910a0a6d |
617 | my $from = $path->from; |
618 | my $to = $path->to; |
619 | warn sprintf( "No hidden witnesses on %s -> %s ?", $from->name, $to->name ) |
620 | unless $path->has_hidden_witnesses; |
621 | my @wits = @{$path->hidden_witnesses}; |
622 | $self->del_path( $path ); |
623 | foreach ( @wits ) { |
624 | $self->add_path( $from, $to, $_ ); |
625 | } |
1f563ac3 |
626 | } |
627 | $self->collapsed( 0 ); |
628 | } |
629 | |
8e1394aa |
630 | =back |
631 | |
de51424a |
632 | =head2 Navigation methods |
633 | |
634 | =over |
635 | |
8e1394aa |
636 | =item B<start> |
637 | |
638 | my $beginning = $collation->start(); |
639 | |
640 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
641 | |
642 | =cut |
643 | |
644 | sub start { |
4a8828f0 |
645 | # Return the beginning reading of the graph. |
8e1394aa |
646 | my $self = shift; |
647 | my( $new_start ) = @_; |
648 | if( $new_start ) { |
910a0a6d |
649 | $self->del_reading( '#START#' ); |
650 | $self->graph->rename_node( $new_start, '#START#' ); |
651 | } |
652 | # Make sure the start node has a start position. |
d9e873d0 |
653 | unless( $self->reading( '#START#' )->has_rank ) { |
654 | $self->reading( '#START#' )->rank( '0' ); |
8e1394aa |
655 | } |
656 | return $self->reading('#START#'); |
657 | } |
658 | |
910a0a6d |
659 | =item B<end> |
660 | |
661 | my $end = $collation->end(); |
662 | |
663 | Returns the end of the collation, a meta-reading with label '#END#'. |
664 | |
665 | =cut |
666 | |
667 | sub end { |
668 | my $self = shift; |
669 | my( $new_end ) = @_; |
670 | if( $new_end ) { |
671 | $self->del_reading( '#END#' ); |
672 | $self->graph->rename_node( $new_end, '#END#' ); |
673 | } |
674 | return $self->reading('#END#'); |
675 | } |
676 | |
e2902068 |
677 | =item B<reading_sequence> |
678 | |
679 | my @readings = $graph->reading_sequence( $first, $last, $path[, $alt_path] ); |
680 | |
681 | Returns the ordered list of readings, starting with $first and ending |
682 | with $last, along the given witness path. If no path is specified, |
683 | assume that the path is that of the base text (if any.) |
684 | |
685 | =cut |
686 | |
910a0a6d |
687 | # TODO Think about returning some lazy-eval iterator. |
688 | |
e2902068 |
689 | sub reading_sequence { |
690 | my( $self, $start, $end, $witness, $backup ) = @_; |
691 | |
930ff666 |
692 | $witness = $self->baselabel unless $witness; |
e2902068 |
693 | my @readings = ( $start ); |
694 | my %seen; |
695 | my $n = $start; |
930ff666 |
696 | while( $n && $n ne $end ) { |
910a0a6d |
697 | if( exists( $seen{$n->name()} ) ) { |
698 | warn "Detected loop at " . $n->name(); |
699 | last; |
700 | } |
701 | $seen{$n->name()} = 1; |
702 | |
703 | my $next = $self->next_reading( $n, $witness, $backup ); |
704 | warn "Did not find any path for $witness from reading " . $n->name |
705 | unless $next; |
706 | push( @readings, $next ); |
707 | $n = $next; |
e2902068 |
708 | } |
709 | # Check that the last reading is our end reading. |
710 | my $last = $readings[$#readings]; |
711 | warn "Last reading found from " . $start->label() . |
910a0a6d |
712 | " for witness $witness is not the end!" |
713 | unless $last eq $end; |
e2902068 |
714 | |
715 | return @readings; |
716 | } |
717 | |
4a8828f0 |
718 | =item B<next_reading> |
8e1394aa |
719 | |
4a8828f0 |
720 | my $next_reading = $graph->next_reading( $reading, $witpath ); |
8e1394aa |
721 | |
4a8828f0 |
722 | Returns the reading that follows the given reading along the given witness |
930ff666 |
723 | path. |
8e1394aa |
724 | |
725 | =cut |
726 | |
4a8828f0 |
727 | sub next_reading { |
e2902068 |
728 | # Return the successor via the corresponding path. |
8e1394aa |
729 | my $self = shift; |
4a8828f0 |
730 | return $self->_find_linked_reading( 'next', @_ ); |
8e1394aa |
731 | } |
732 | |
4a8828f0 |
733 | =item B<prior_reading> |
8e1394aa |
734 | |
4a8828f0 |
735 | my $prior_reading = $graph->prior_reading( $reading, $witpath ); |
8e1394aa |
736 | |
4a8828f0 |
737 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
738 | path. |
8e1394aa |
739 | |
740 | =cut |
741 | |
4a8828f0 |
742 | sub prior_reading { |
e2902068 |
743 | # Return the predecessor via the corresponding path. |
8e1394aa |
744 | my $self = shift; |
4a8828f0 |
745 | return $self->_find_linked_reading( 'prior', @_ ); |
8e1394aa |
746 | } |
747 | |
4a8828f0 |
748 | sub _find_linked_reading { |
e2902068 |
749 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
750 | my @linked_paths = $direction eq 'next' |
910a0a6d |
751 | ? $node->outgoing() : $node->incoming(); |
e2902068 |
752 | return undef unless scalar( @linked_paths ); |
8e1394aa |
753 | |
e2902068 |
754 | # We have to find the linked path that contains all of the |
755 | # witnesses supplied in $path. |
756 | my( @path_wits, @alt_path_wits ); |
757 | @path_wits = $self->witnesses_of_label( $path ) if $path; |
758 | @alt_path_wits = $self->witnesses_of_label( $alt_path ) if $alt_path; |
759 | my $base_le; |
760 | my $alt_le; |
761 | foreach my $le ( @linked_paths ) { |
910a0a6d |
762 | if( $le->name eq $self->baselabel ) { |
763 | $base_le = $le; |
764 | } else { |
765 | my @le_wits = $self->witnesses_of_label( $le->name ); |
766 | if( _is_within( \@path_wits, \@le_wits ) ) { |
767 | # This is the right path. |
768 | return $direction eq 'next' ? $le->to() : $le->from(); |
769 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
770 | $alt_le = $le; |
771 | } |
772 | } |
8e1394aa |
773 | } |
e2902068 |
774 | # Got this far? Return the alternate path if it exists. |
775 | return $direction eq 'next' ? $alt_le->to() : $alt_le->from() |
910a0a6d |
776 | if $alt_le; |
e2902068 |
777 | |
778 | # Got this far? Return the base path if it exists. |
779 | return $direction eq 'next' ? $base_le->to() : $base_le->from() |
910a0a6d |
780 | if $base_le; |
e2902068 |
781 | |
782 | # Got this far? We have no appropriate path. |
8e1394aa |
783 | warn "Could not find $direction node from " . $node->label |
910a0a6d |
784 | . " along path $path"; |
8e1394aa |
785 | return undef; |
786 | } |
787 | |
4a8828f0 |
788 | # Some set logic. |
789 | sub _is_within { |
790 | my( $set1, $set2 ) = @_; |
7854e12e |
791 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
792 | foreach my $el ( @$set1 ) { |
910a0a6d |
793 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
4a8828f0 |
794 | } |
795 | return $ret; |
796 | } |
797 | |
de51424a |
798 | |
799 | ## INITIALIZATION METHODS - for use by parsers |
4a8828f0 |
800 | # Walk the paths for each witness in the graph, and return the nodes |
e2902068 |
801 | # that the graph has in common. If $using_base is true, some |
802 | # different logic is needed. |
7e450e44 |
803 | # NOTE This does not create paths; it merely finds common readings. |
4a8828f0 |
804 | |
805 | sub walk_witness_paths { |
7e450e44 |
806 | my( $self ) = @_; |
4a8828f0 |
807 | # For each witness, walk the path through the graph. |
808 | # Then we need to find the common nodes. |
809 | # TODO This method is going to fall down if we have a very gappy |
810 | # text in the collation. |
811 | my $paths = {}; |
3a1f2523 |
812 | my @common_readings; |
910a0a6d |
813 | foreach my $wit ( $self->tradition->witnesses ) { |
814 | my $curr_reading = $self->start; |
7e450e44 |
815 | my @wit_path = $self->reading_sequence( $self->start, $self->end, |
910a0a6d |
816 | $wit->sigil ); |
817 | $wit->path( \@wit_path ); |
818 | |
819 | # Detect the common readings. |
820 | @common_readings = _find_common( \@common_readings, \@wit_path ); |
4a8828f0 |
821 | } |
822 | |
823 | # Mark all the nodes as either common or not. |
3a1f2523 |
824 | foreach my $cn ( @common_readings ) { |
910a0a6d |
825 | print STDERR "Setting " . $cn->name . " / " . $cn->label |
826 | . " as common node\n"; |
827 | $cn->make_common; |
4a8828f0 |
828 | } |
829 | foreach my $n ( $self->readings() ) { |
910a0a6d |
830 | $n->make_variant unless $n->is_common; |
4a8828f0 |
831 | } |
3a1f2523 |
832 | # Return an array of the common nodes in order. |
833 | return @common_readings; |
4a8828f0 |
834 | } |
835 | |
930ff666 |
836 | sub _find_common { |
837 | my( $common_readings, $new_path ) = @_; |
838 | my @cr; |
839 | if( @$common_readings ) { |
910a0a6d |
840 | foreach my $n ( @$new_path ) { |
841 | push( @cr, $n ) if grep { $_ eq $n } @$common_readings; |
842 | } |
930ff666 |
843 | } else { |
910a0a6d |
844 | push( @cr, @$new_path ); |
930ff666 |
845 | } |
846 | return @cr; |
847 | } |
848 | |
849 | sub _remove_common { |
850 | my( $common_readings, $divergence ) = @_; |
851 | my @cr; |
852 | my %diverged; |
853 | map { $diverged{$_->name} = 1 } @$divergence; |
854 | foreach( @$common_readings ) { |
910a0a6d |
855 | push( @cr, $_ ) unless $diverged{$_->name}; |
930ff666 |
856 | } |
857 | return @cr; |
858 | } |
859 | |
860 | |
7e450e44 |
861 | # For use when a collation is constructed from a base text and an apparatus. |
862 | # We have the sequences of readings and just need to add path edges. |
e2902068 |
863 | |
6a222840 |
864 | sub make_witness_paths { |
865 | my( $self ) = @_; |
910a0a6d |
866 | foreach my $wit ( $self->tradition->witnesses ) { |
867 | print STDERR "Making path for " . $wit->sigil . "\n"; |
868 | $self->make_witness_path( $wit ); |
7854e12e |
869 | } |
7854e12e |
870 | } |
871 | |
6a222840 |
872 | sub make_witness_path { |
7854e12e |
873 | my( $self, $wit ) = @_; |
874 | my @chain = @{$wit->path}; |
15d2d3df |
875 | my $sig = $wit->sigil; |
7854e12e |
876 | foreach my $idx ( 0 .. $#chain-1 ) { |
910a0a6d |
877 | $self->add_path( $chain[$idx], $chain[$idx+1], $sig ); |
7854e12e |
878 | } |
d9e873d0 |
879 | if( $wit->has_ante_corr ) { |
880 | @chain = @{$wit->uncorrected_path}; |
881 | foreach my $idx( 0 .. $#chain-1 ) { |
882 | my $source = $chain[$idx]; |
883 | my $target = $chain[$idx+1]; |
884 | $self->add_path( $source, $target, $sig.$self->ac_label ) |
885 | unless $self->has_path( $source, $target, $sig ); |
886 | } |
15d2d3df |
887 | } |
e2902068 |
888 | } |
889 | |
910a0a6d |
890 | sub calculate_ranks { |
891 | my $self = shift; |
892 | # Walk a version of the graph where every node linked by a relationship |
893 | # edge is fundamentally the same node, and do a topological ranking on |
894 | # the nodes in this graph. |
895 | my $topo_graph = Graph::Easy->new(); |
896 | my %rel_containers; |
897 | my $rel_ctr = 0; |
898 | # Add the nodes |
899 | foreach my $r ( $self->readings ) { |
900 | next if exists $rel_containers{$r->name}; |
901 | my @rels = $r->related_readings( 'colocated' ); |
902 | if( @rels ) { |
903 | # Make a relationship container. |
904 | push( @rels, $r ); |
905 | my $rn = $topo_graph->add_node( 'rel_container_' . $rel_ctr++ ); |
906 | foreach( @rels ) { |
907 | $rel_containers{$_->name} = $rn; |
908 | } |
909 | } else { |
910 | # Add a new node to mirror the old node. |
911 | $rel_containers{$r->name} = $topo_graph->add_node( $r->name ); |
912 | } |
4a8828f0 |
913 | } |
3a1f2523 |
914 | |
910a0a6d |
915 | # Add the edges. Need only one edge between any pair of nodes. |
916 | foreach my $r ( $self->readings ) { |
917 | foreach my $n ( $r->neighbor_readings( 'forward' ) ) { |
918 | $topo_graph->add_edge_once( $rel_containers{$r->name}, |
919 | $rel_containers{$n->name} ); |
920 | } |
921 | } |
922 | |
923 | # Now do the rankings, starting with the start node. |
924 | my $topo_start = $rel_containers{$self->start->name}; |
925 | my $node_ranks = { $topo_start->name => 0 }; |
926 | my @curr_origin = ( $topo_start ); |
927 | # A little iterative function. |
928 | while( @curr_origin ) { |
929 | @curr_origin = _assign_rank( $node_ranks, @curr_origin ); |
930 | } |
931 | # Transfer our rankings from the topological graph to the real one. |
932 | foreach my $r ( $self->readings ) { |
933 | $r->rank( $node_ranks->{$rel_containers{$r->name}->name} ); |
de51424a |
934 | } |
8e1394aa |
935 | } |
3a1f2523 |
936 | |
910a0a6d |
937 | sub _assign_rank { |
938 | my( $node_ranks, @current_nodes ) = @_; |
939 | # Look at each of the children of @current_nodes. If all the child's |
940 | # parents have a rank, assign it the highest rank + 1 and add it to |
941 | # @next_nodes. Otherwise skip it. |
942 | my @next_nodes; |
943 | foreach my $c ( @current_nodes ) { |
944 | warn "Current reading " . $c->name . "has no rank!" |
945 | unless exists $node_ranks->{$c->name}; |
946 | # print STDERR "Looking at child of node " . $c->name . ", rank " |
947 | # . $node_ranks->{$c->name} . "\n"; |
948 | my @children = map { $_->to } $c->outgoing; |
949 | foreach my $child ( @children ) { |
950 | next if exists $node_ranks->{$child->name}; |
951 | my $highest_rank = -1; |
952 | my $skip = 0; |
953 | my @parents = map { $_->from } $child->incoming; |
954 | foreach my $parent ( @parents ) { |
955 | if( exists $node_ranks->{$parent->name} ) { |
956 | $highest_rank = $node_ranks->{$parent->name} |
957 | if $highest_rank <= $node_ranks->{$parent->name}; |
958 | } else { |
959 | $skip = 1; |
960 | last; |
961 | } |
962 | } |
963 | next if $skip; |
964 | # print STDERR "Assigning rank " . ( $highest_rank + 1 ) . " to node " . $child->name . "\n"; |
965 | $node_ranks->{$child->name} = $highest_rank + 1; |
966 | push( @next_nodes, $child ); |
967 | } |
968 | } |
969 | return @next_nodes; |
4cdd82f1 |
970 | } |
910a0a6d |
971 | |
4cdd82f1 |
972 | sub possible_positions { |
3a1f2523 |
973 | my $self = shift; |
910a0a6d |
974 | my %all_pos; |
975 | map { $all_pos{ $_->rank } = 1 } $self->readings; |
976 | return keys %all_pos; |
3a1f2523 |
977 | } |
978 | |
4cdd82f1 |
979 | # TODO think about indexing this. |
3a1f2523 |
980 | sub readings_at_position { |
4cdd82f1 |
981 | my( $self, $position, $strict ) = @_; |
4cdd82f1 |
982 | my @answer; |
983 | foreach my $r ( $self->readings ) { |
910a0a6d |
984 | push( @answer, $r ) if $r->is_at_position( $position, $strict ); |
4cdd82f1 |
985 | } |
3a1f2523 |
986 | return @answer; |
987 | } |
988 | |
989 | ## Lemmatizer functions |
990 | |
991 | sub init_lemmata { |
992 | my $self = shift; |
4cdd82f1 |
993 | |
994 | foreach my $position ( $self->possible_positions ) { |
910a0a6d |
995 | $self->lemmata->{$position} = undef; |
3a1f2523 |
996 | } |
997 | |
998 | foreach my $cr ( $self->common_readings ) { |
910a0a6d |
999 | $self->lemmata->{$cr->position->maxref} = $cr->name; |
3a1f2523 |
1000 | } |
1001 | } |
7e450e44 |
1002 | |
1003 | sub common_readings { |
1004 | my $self = shift; |
1005 | my @common = grep { $_->is_common } $self->readings(); |
d9e873d0 |
1006 | return sort { $a->rank <=> $b->rank } @common; |
7e450e44 |
1007 | } |
3a1f2523 |
1008 | |
1009 | =item B<lemma_readings> |
1010 | |
1011 | my @state = $graph->lemma_readings( @readings_delemmatized ); |
1012 | |
1013 | Takes a list of readings that have just been delemmatized, and returns |
1014 | a set of tuples of the form ['reading', 'state'] that indicates what |
1015 | changes need to be made to the graph. |
1016 | |
1017 | =over |
1018 | |
1019 | =item * |
1020 | |
1021 | A state of 1 means 'lemmatize this reading' |
1022 | |
1023 | =item * |
1024 | |
1025 | A state of 0 means 'delemmatize this reading' |
1026 | |
1027 | =item * |
1028 | |
1029 | A state of undef means 'an ellipsis belongs in the text here because |
1030 | no decision has been made / an earlier decision was backed out' |
1031 | |
1032 | =back |
1033 | |
1034 | =cut |
1035 | |
1036 | sub lemma_readings { |
1037 | my( $self, @toggled_off_nodes ) = @_; |
1038 | |
1039 | # First get the positions of those nodes which have been |
1040 | # toggled off. |
1041 | my $positions_off = {}; |
4cdd82f1 |
1042 | map { $positions_off->{ $_->position->reference } = $_->name } |
1043 | @toggled_off_nodes; |
de51424a |
1044 | |
3a1f2523 |
1045 | # Now for each position, we have to see if a node is on, and we |
4cdd82f1 |
1046 | # have to see if a node has been turned off. The lemmata hash |
1047 | # should contain fixed positions, range positions whose node was |
1048 | # just turned off, and range positions whose node is on. |
3a1f2523 |
1049 | my @answer; |
4cdd82f1 |
1050 | my %fixed_positions; |
1051 | # TODO One of these is probably redundant. |
1052 | map { $fixed_positions{$_} = 0 } keys %{$self->lemmata}; |
1053 | map { $fixed_positions{$_} = 0 } keys %{$positions_off}; |
1054 | map { $fixed_positions{$_} = 1 } $self->possible_positions; |
1055 | foreach my $pos ( sort { Text::Tradition::Collation::Position::str_cmp( $a, $b ) } keys %fixed_positions ) { |
910a0a6d |
1056 | # Find the state of this position. If there is an active node, |
1057 | # its name will be the state; otherwise the state will be 0 |
1058 | # (nothing at this position) or undef (ellipsis at this position) |
1059 | my $active = undef; |
1060 | $active = $self->lemmata->{$pos} if exists $self->lemmata->{$pos}; |
1061 | |
1062 | # Is there a formerly active node that was toggled off? |
1063 | if( exists( $positions_off->{$pos} ) ) { |
1064 | my $off_node = $positions_off->{$pos}; |
1065 | if( $active && $active ne $off_node) { |
1066 | push( @answer, [ $off_node, 0 ], [ $active, 1 ] ); |
1067 | } else { |
1068 | unless( $fixed_positions{$pos} ) { |
1069 | $active = 0; |
1070 | delete $self->lemmata->{$pos}; |
1071 | } |
1072 | push( @answer, [ $off_node, $active ] ); |
1073 | } |
1074 | |
1075 | # No formerly active node, so we just see if there is a currently |
1076 | # active one. |
1077 | } elsif( $active ) { |
1078 | # Push the active node, whatever it is. |
1079 | push( @answer, [ $active, 1 ] ); |
1080 | } else { |
1081 | # Push the state that is there. Arbitrarily use the first node |
1082 | # at that position. |
1083 | my @pos_nodes = $self->readings_at_position( $pos ); |
1084 | push( @answer, [ $pos_nodes[0]->name, $self->lemmata->{$pos} ] ); |
1085 | delete $self->lemmata->{$pos} unless $fixed_positions{$pos}; |
1086 | } |
3a1f2523 |
1087 | } |
4cdd82f1 |
1088 | |
3a1f2523 |
1089 | return @answer; |
1090 | } |
1091 | |
de51424a |
1092 | =item B<toggle_reading> |
1093 | |
1094 | my @readings_delemmatized = $graph->toggle_reading( $reading_name ); |
1095 | |
1096 | Takes a reading node name, and either lemmatizes or de-lemmatizes |
1097 | it. Returns a list of all readings that are de-lemmatized as a result |
1098 | of the toggle. |
1099 | |
1100 | =cut |
1101 | |
1102 | sub toggle_reading { |
1103 | my( $self, $rname ) = @_; |
1104 | |
1105 | return unless $rname; |
1106 | my $reading = $self->reading( $rname ); |
1107 | if( !$reading || $reading->is_common() ) { |
910a0a6d |
1108 | # Do nothing, it's a common node. |
1109 | return; |
de51424a |
1110 | } |
1111 | |
1112 | my $pos = $reading->position; |
4cdd82f1 |
1113 | my $fixed = $reading->position->fixed; |
1114 | my $old_state = $self->lemmata->{$pos->reference}; |
1115 | |
de51424a |
1116 | my @readings_off; |
1117 | if( $old_state && $old_state eq $rname ) { |
910a0a6d |
1118 | # Turn off the node. We turn on no others by default. |
1119 | push( @readings_off, $reading ); |
de51424a |
1120 | } else { |
910a0a6d |
1121 | # Turn on the node. |
1122 | $self->lemmata->{$pos->reference} = $rname; |
1123 | # Any other 'on' readings in the same position should be off |
1124 | # if we have a fixed position. |
1125 | push( @readings_off, $self->same_position_as( $reading, 1 ) ) |
1126 | if $pos->fixed; |
1127 | # Any node that is an identical transposed one should be off. |
1128 | push( @readings_off, $reading->identical_readings ); |
de51424a |
1129 | } |
1130 | @readings_off = unique_list( @readings_off ); |
910a0a6d |
1131 | |
de51424a |
1132 | # Turn off the readings that need to be turned off. |
1133 | my @readings_delemmatized; |
1134 | foreach my $n ( @readings_off ) { |
910a0a6d |
1135 | my $npos = $n->position; |
1136 | my $state = undef; |
1137 | $state = $self->lemmata->{$npos->reference} |
1138 | if defined $self->lemmata->{$npos->reference}; |
1139 | if( $state && $state eq $n->name ) { |
1140 | # this reading is still on, so turn it off |
1141 | push( @readings_delemmatized, $n ); |
1142 | my $new_state = undef; |
1143 | if( $npos->fixed && $n eq $reading ) { |
1144 | # This is the reading that was clicked, so if there are no |
1145 | # other readings there and this is a fixed position, turn off |
1146 | # the position. In all other cases, restore the ellipsis. |
1147 | my @other_n = $self->same_position_as( $n ); # TODO do we need strict? |
1148 | $new_state = 0 unless @other_n; |
1149 | } |
1150 | $self->lemmata->{$npos->reference} = $new_state; |
1151 | } elsif( $old_state && $old_state eq $n->name ) { |
1152 | # another reading has already been turned on here |
1153 | push( @readings_delemmatized, $n ); |
1154 | } # else some other reading was on anyway, so pass. |
de51424a |
1155 | } |
1156 | return @readings_delemmatized; |
1157 | } |
1158 | |
1159 | sub same_position_as { |
4cdd82f1 |
1160 | my( $self, $reading, $strict ) = @_; |
de51424a |
1161 | my $pos = $reading->position; |
4cdd82f1 |
1162 | my %onpath = ( $reading->name => 1 ); |
1163 | # TODO This might not always be sufficient. We really want to |
1164 | # exclude all readings on this one's path between its two |
1165 | # common points. |
1166 | map { $onpath{$_->name} = 1 } $reading->neighbor_readings; |
1167 | my @same = grep { !$onpath{$_->name} } |
1168 | $self->readings_at_position( $reading->position, $strict ); |
de51424a |
1169 | return @same; |
1170 | } |
3a1f2523 |
1171 | |
4a8828f0 |
1172 | # Return the string that joins together a list of witnesses for |
1173 | # display on a single path. |
1174 | sub path_label { |
1175 | my $self = shift; |
1176 | return join( $self->wit_list_separator, @_ ); |
1177 | } |
1178 | |
1179 | sub witnesses_of_label { |
de51424a |
1180 | my( $self, $label ) = @_; |
4a8828f0 |
1181 | my $regex = $self->wit_list_separator; |
de51424a |
1182 | my @answer = split( /\Q$regex\E/, $label ); |
1183 | return @answer; |
4a8828f0 |
1184 | } |
8e1394aa |
1185 | |
de51424a |
1186 | sub unique_list { |
1187 | my( @list ) = @_; |
1188 | my %h; |
1189 | map { $h{$_->name} = $_ } @list; |
1190 | return values( %h ); |
1191 | } |
1192 | |
1f563ac3 |
1193 | sub add_hash_entry { |
1194 | my( $hash, $key, $entry ) = @_; |
1195 | if( exists $hash->{$key} ) { |
910a0a6d |
1196 | push( @{$hash->{$key}}, $entry ); |
1f563ac3 |
1197 | } else { |
910a0a6d |
1198 | $hash->{$key} = [ $entry ]; |
1f563ac3 |
1199 | } |
1200 | } |
1201 | |
dd3b58b0 |
1202 | no Moose; |
1203 | __PACKAGE__->meta->make_immutable; |