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', |
eca16057 |
20 | add_lacuna => 'add_node', |
910a0a6d |
21 | del_reading => 'del_node', |
22 | del_segment => 'del_node', |
23 | add_path => 'add_edge', |
24 | del_path => 'del_edge', |
25 | reading => 'node', |
26 | path => 'edge', |
27 | readings => 'nodes', |
28 | segments => 'nodes', |
29 | paths => 'edges', |
30 | relationships => 'edges', |
d047cd52 |
31 | }, |
32 | default => sub { Graph::Easy->new( undirected => 0 ) }, |
33 | ); |
910a0a6d |
34 | |
dd3b58b0 |
35 | |
f6066bac |
36 | has 'tradition' => ( # TODO should this not be ro? |
8e1394aa |
37 | is => 'rw', |
d047cd52 |
38 | isa => 'Text::Tradition', |
39 | ); |
dd3b58b0 |
40 | |
8e1394aa |
41 | has 'svg' => ( |
42 | is => 'ro', |
43 | isa => 'Str', |
44 | writer => '_save_svg', |
45 | predicate => 'has_svg', |
46 | ); |
47 | |
8e1394aa |
48 | has 'graphml' => ( |
49 | is => 'ro', |
df6d9812 |
50 | isa => 'Str', |
8e1394aa |
51 | writer => '_save_graphml', |
52 | predicate => 'has_graphml', |
53 | ); |
54 | |
910a0a6d |
55 | has 'csv' => ( |
56 | is => 'ro', |
57 | isa => 'Str', |
58 | writer => '_save_csv', |
59 | predicate => 'has_csv', |
60 | ); |
61 | |
3a1f2523 |
62 | # Keeps track of the lemmas within the collation. At most one lemma |
63 | # per position in the graph. |
64 | has 'lemmata' => ( |
65 | is => 'ro', |
66 | isa => 'HashRef[Maybe[Str]]', |
67 | default => sub { {} }, |
68 | ); |
69 | |
4a8828f0 |
70 | has 'wit_list_separator' => ( |
7854e12e |
71 | is => 'rw', |
72 | isa => 'Str', |
73 | default => ', ', |
74 | ); |
75 | |
76 | has 'baselabel' => ( |
77 | is => 'rw', |
78 | isa => 'Str', |
79 | default => 'base text', |
80 | ); |
4a8828f0 |
81 | |
1f563ac3 |
82 | has 'collapsed' => ( |
15d2d3df |
83 | is => 'rw', |
84 | isa => 'Bool', |
85 | ); |
86 | |
87 | has 'linear' => ( |
88 | is => 'rw', |
89 | isa => 'Bool', |
90 | default => 1, |
91 | ); |
1f563ac3 |
92 | |
ef9d481f |
93 | has 'ac_label' => ( |
94 | is => 'rw', |
95 | isa => 'Str', |
96 | default => ' (a.c.)', |
97 | ); |
98 | |
1f563ac3 |
99 | |
dd3b58b0 |
100 | # The collation can be created two ways: |
101 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
102 | # the results as in 2. |
103 | # 2. Read a pre-prepared collation in one of a variety of formats, |
104 | # and make the graph from that. |
105 | |
106 | # The graph itself will (for now) be immutable, and the positions |
107 | # within the graph will also be immutable. We need to calculate those |
108 | # positions upon graph construction. The equivalences between graph |
109 | # nodes will be mutable, entirely determined by the user (or possibly |
110 | # by some semantic pre-processing provided by the user.) So the |
111 | # constructor should just make an empty equivalences object. The |
112 | # constructor will also need to make the witness objects, if we didn't |
113 | # come through option 1. |
114 | |
d047cd52 |
115 | sub BUILD { |
116 | my( $self, $args ) = @_; |
8e1394aa |
117 | $self->graph->use_class('node', 'Text::Tradition::Collation::Reading'); |
3265b0ce |
118 | $self->graph->use_class('edge', 'Text::Tradition::Collation::Path'); |
d047cd52 |
119 | |
4a8828f0 |
120 | # Pass through any graph-specific options. |
121 | my $shape = exists( $args->{'shape'} ) ? $args->{'shape'} : 'ellipse'; |
122 | $self->graph->set_attribute( 'node', 'shape', $shape ); |
d047cd52 |
123 | } |
784877d9 |
124 | |
eca16057 |
125 | around add_lacuna => sub { |
126 | my $orig = shift; |
127 | my $self = shift; |
128 | my $id = shift @_; |
129 | my $l = $self->$orig( '#LACUNA_' . $id . '#' ); |
130 | $l->is_lacuna( 1 ); |
131 | return $l; |
132 | }; |
133 | |
7854e12e |
134 | # Wrapper around add_path |
135 | |
136 | around add_path => sub { |
137 | my $orig = shift; |
138 | my $self = shift; |
139 | |
140 | # Make sure there are three arguments |
141 | unless( @_ == 3 ) { |
910a0a6d |
142 | warn "Call add_path with args source, target, witness"; |
143 | return; |
7854e12e |
144 | } |
145 | # Make sure the proposed path does not yet exist |
b15511bf |
146 | # NOTE 'reading' will currently return readings and segments |
7854e12e |
147 | my( $source, $target, $wit ) = @_; |
148 | $source = $self->reading( $source ) |
910a0a6d |
149 | unless ref( $source ) eq 'Text::Tradition::Collation::Reading'; |
7854e12e |
150 | $target = $self->reading( $target ) |
910a0a6d |
151 | unless ref( $target ) eq 'Text::Tradition::Collation::Reading'; |
7854e12e |
152 | foreach my $path ( $source->edges_to( $target ) ) { |
910a0a6d |
153 | if( $path->label eq $wit && $path->class eq 'edge.path' ) { |
154 | return; |
155 | } |
7854e12e |
156 | } |
157 | # Do the deed |
158 | $self->$orig( @_ ); |
159 | }; |
160 | |
3265b0ce |
161 | # Wrapper around paths |
162 | around paths => sub { |
163 | my $orig = shift; |
164 | my $self = shift; |
165 | |
b15511bf |
166 | my @result = grep { $_->sub_class eq 'path' } $self->$orig( @_ ); |
df6d9812 |
167 | return @result; |
168 | }; |
169 | |
170 | around relationships => sub { |
171 | my $orig = shift; |
172 | my $self = shift; |
b15511bf |
173 | my @result = grep { $_->sub_class eq 'relationship' } $self->$orig( @_ ); |
174 | return @result; |
175 | }; |
176 | |
177 | around readings => sub { |
178 | my $orig = shift; |
179 | my $self = shift; |
180 | my @result = grep { $_->sub_class ne 'segment' } $self->$orig( @_ ); |
181 | return @result; |
182 | }; |
183 | |
184 | around segments => sub { |
185 | my $orig = shift; |
186 | my $self = shift; |
187 | my @result = grep { $_->sub_class eq 'segment' } $self->$orig( @_ ); |
3265b0ce |
188 | return @result; |
189 | }; |
190 | |
7854e12e |
191 | # Wrapper around merge_nodes |
784877d9 |
192 | |
193 | sub merge_readings { |
194 | my $self = shift; |
195 | my $first_node = shift; |
196 | my $second_node = shift; |
197 | $first_node->merge_from( $second_node ); |
198 | unshift( @_, $first_node, $second_node ); |
199 | return $self->graph->merge_nodes( @_ ); |
200 | } |
201 | |
15d2d3df |
202 | # Extra graph-alike utility |
203 | sub has_path { |
204 | my( $self, $source, $target, $label ) = @_; |
205 | my @paths = $source->edges_to( $target ); |
206 | my @relevant = grep { $_->label eq $label } @paths; |
b15511bf |
207 | return scalar @relevant; |
208 | } |
209 | |
210 | ## Dealing with groups of readings, i.e. segments. |
211 | |
212 | sub add_segment { |
213 | my( $self, @items ) = @_; |
214 | my $segment = Text::Tradition::Collation::Segment->new( 'members' => \@items ); |
215 | return $segment; |
15d2d3df |
216 | } |
217 | |
3265b0ce |
218 | ## Dealing with relationships between readings. This is a different |
4cdd82f1 |
219 | ## sort of graph edge. Return a success/failure value and a list of |
220 | ## node pairs that have been linked. |
3265b0ce |
221 | |
222 | sub add_relationship { |
b15511bf |
223 | my( $self, $source, $target, $options ) = @_; |
ef9d481f |
224 | |
225 | # Make sure there is not another relationship between these two |
b15511bf |
226 | # readings or segments already |
ef9d481f |
227 | $source = $self->reading( $source ) |
910a0a6d |
228 | unless ref( $source ) && $source->isa( 'Graph::Easy::Node' ); |
ef9d481f |
229 | $target = $self->reading( $target ) |
910a0a6d |
230 | unless ref( $target ) && $target->isa( 'Graph::Easy::Node' ); |
4cdd82f1 |
231 | foreach my $rel ( $source->edges_to( $target ), $target->edges_to( $source ) ) { |
910a0a6d |
232 | if( $rel->class eq 'edge.relationship' ) { |
233 | return ( undef, "Relationship already exists between these readings" ); |
234 | } |
4cdd82f1 |
235 | } |
910a0a6d |
236 | if( $options->{'equal_rank'} && !relationship_valid( $source, $target ) ) { |
237 | return ( undef, 'Relationship creates witness loop' ); |
ef9d481f |
238 | } |
239 | |
910a0a6d |
240 | # TODO Think about positional hilarity if relationships are added after positions |
241 | # are assigned. |
242 | |
4cdd82f1 |
243 | my @joined = ( [ $source->name, $target->name ] ); # Keep track of the nodes we join. |
244 | |
245 | $options->{'this_relation'} = [ $source, $target ]; |
f6066bac |
246 | my $rel; |
247 | eval { $rel = Text::Tradition::Collation::Relationship->new( %$options ) }; |
248 | if( $@ ) { |
249 | return ( undef, $@ ); |
250 | } |
3265b0ce |
251 | $self->graph->add_edge( $source, $target, $rel ); |
910a0a6d |
252 | |
253 | # TODO Handle global relationship setting |
254 | |
4cdd82f1 |
255 | return( 1, @joined ); |
3265b0ce |
256 | } |
257 | |
910a0a6d |
258 | sub relationship_valid { |
259 | my( $source, $target ) = @_; |
260 | # Check that linking the source and target in a relationship won't lead |
261 | # to a path loop for any witness. |
262 | my @proposed_related = ( $source, $target ); |
263 | push( @proposed_related, $source->related_readings ); |
264 | push( @proposed_related, $target->related_readings ); |
265 | my %pr_ids; |
266 | map { $pr_ids{ $_->name } = 1 } @proposed_related; |
267 | # The lists of 'in' and 'out' should not have any element that appears |
268 | # in 'proposed_related'. |
269 | foreach my $pr ( @proposed_related ) { |
270 | foreach my $e ( $pr->incoming ) { |
271 | if( exists $pr_ids{ $e->from->name } ) { |
272 | return 0; |
273 | } |
274 | } |
275 | foreach my $e ( $pr->outgoing ) { |
276 | if( exists $pr_ids{ $e->to->name } ) { |
277 | return 0; |
278 | } |
279 | } |
280 | } |
281 | return 1; |
282 | } |
283 | |
8e1394aa |
284 | =head2 Output method(s) |
285 | |
286 | =over |
287 | |
288 | =item B<as_svg> |
289 | |
290 | print $graph->as_svg( $recalculate ); |
291 | |
292 | Returns an SVG string that represents the graph. Uses GraphViz to do |
4a8828f0 |
293 | this, because Graph::Easy doesn\'t cope well with long graphs. Unless |
8e1394aa |
294 | $recalculate is passed (and is a true value), the method will return a |
295 | cached copy of the SVG after the first call to the method. |
296 | |
297 | =cut |
298 | |
299 | sub as_svg { |
300 | my( $self, $recalc ) = @_; |
301 | return $self->svg if $self->has_svg; |
302 | |
3265b0ce |
303 | $self->collapse_graph_paths(); |
8e1394aa |
304 | |
305 | my @cmd = qw/dot -Tsvg/; |
306 | my( $svg, $err ); |
910a0a6d |
307 | my $dotfile = File::Temp->new(); |
d9e873d0 |
308 | ## TODO REMOVE |
eca16057 |
309 | # $dotfile->unlink_on_destroy(0); |
910a0a6d |
310 | binmode $dotfile, ':utf8'; |
311 | print $dotfile $self->as_dot(); |
312 | push( @cmd, $dotfile->filename ); |
313 | run( \@cmd, ">", binary(), \$svg ); |
314 | $svg = decode_utf8( $svg ); |
df6d9812 |
315 | $self->_save_svg( $svg ); |
3265b0ce |
316 | $self->expand_graph_paths(); |
8e1394aa |
317 | return $svg; |
318 | } |
319 | |
df6d9812 |
320 | =item B<as_dot> |
321 | |
322 | print $graph->as_dot( $view, $recalculate ); |
323 | |
324 | Returns a string that is the collation graph expressed in dot |
325 | (i.e. GraphViz) format. The 'view' argument determines what kind of |
326 | graph is produced. |
327 | * 'path': a graph of witness paths through the collation (DEFAULT) |
328 | * 'relationship': a graph of how collation readings relate to |
329 | each other |
330 | |
331 | =cut |
332 | |
333 | sub as_dot { |
334 | my( $self, $view ) = @_; |
335 | $view = 'path' unless $view; |
336 | # TODO consider making some of these things configurable |
337 | my $dot = sprintf( "digraph %s {\n", $self->tradition->name ); |
338 | $dot .= "\tedge [ arrowhead=open ];\n"; |
339 | $dot .= "\tgraph [ rankdir=LR ];\n"; |
340 | $dot .= sprintf( "\tnode [ fontsize=%d, fillcolor=%s, style=%s, shape=%s ];\n", |
910a0a6d |
341 | 11, "white", "filled", $self->graph->get_attribute( 'node', 'shape' ) ); |
df6d9812 |
342 | |
343 | foreach my $reading ( $self->readings ) { |
910a0a6d |
344 | # Need not output nodes without separate labels |
345 | next if $reading->name eq $reading->label; |
346 | # TODO output readings or segments, but not both |
347 | next if $reading->class eq 'node.segment'; |
348 | $dot .= sprintf( "\t\"%s\" [ label=\"%s\" ];\n", $reading->name, $reading->label ); |
df6d9812 |
349 | } |
350 | |
351 | my @edges = $view eq 'relationship' ? $self->relationships : $self->paths; |
352 | foreach my $edge ( @edges ) { |
910a0a6d |
353 | my %variables = ( 'color' => '#000000', |
354 | 'fontcolor' => '#000000', |
355 | 'label' => $edge->label, |
356 | ); |
357 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
358 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", |
359 | $edge->from->name, $edge->to->name, $varopts ); |
df6d9812 |
360 | } |
df6d9812 |
361 | $dot .= "}\n"; |
362 | return $dot; |
363 | } |
364 | |
8e1394aa |
365 | =item B<as_graphml> |
366 | |
367 | print $graph->as_graphml( $recalculate ) |
368 | |
369 | Returns a GraphML representation of the collation graph, with |
370 | transposition information and position information. Unless |
371 | $recalculate is passed (and is a true value), the method will return a |
372 | cached copy of the SVG after the first call to the method. |
373 | |
374 | =cut |
375 | |
376 | sub as_graphml { |
377 | my( $self, $recalc ) = @_; |
378 | return $self->graphml if $self->has_graphml; |
379 | |
380 | # Some namespaces |
381 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
382 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
383 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
910a0a6d |
384 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
8e1394aa |
385 | |
386 | # Create the document and root node |
387 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
388 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
389 | $graphml->setDocumentElement( $root ); |
390 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
391 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
392 | |
f6066bac |
393 | # TODO Add some global graph data |
394 | |
8e1394aa |
395 | # Add the data keys for nodes |
ef9d481f |
396 | my %node_data_keys; |
397 | my $ndi = 0; |
d9e873d0 |
398 | foreach my $datum ( qw/ name reading identical rank class / ) { |
910a0a6d |
399 | $node_data_keys{$datum} = 'dn'.$ndi++; |
400 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
401 | $key->setAttribute( 'attr.name', $datum ); |
402 | $key->setAttribute( 'attr.type', 'string' ); |
403 | $key->setAttribute( 'for', 'node' ); |
404 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
405 | } |
406 | |
df6d9812 |
407 | # Add the data keys for edges, i.e. witnesses |
ef9d481f |
408 | my $edi = 0; |
409 | my %edge_data_keys; |
b15511bf |
410 | foreach my $edge_key( qw/ witness_main witness_ante_corr relationship class / ) { |
910a0a6d |
411 | $edge_data_keys{$edge_key} = 'de'.$edi++; |
412 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
413 | $key->setAttribute( 'attr.name', $edge_key ); |
414 | $key->setAttribute( 'attr.type', 'string' ); |
415 | $key->setAttribute( 'for', 'edge' ); |
416 | $key->setAttribute( 'id', $edge_data_keys{$edge_key} ); |
8e1394aa |
417 | } |
ef9d481f |
418 | |
8e1394aa |
419 | # Add the graph, its nodes, and its edges |
420 | my $graph = $root->addNewChild( $graphml_ns, 'graph' ); |
421 | $graph->setAttribute( 'edgedefault', 'directed' ); |
422 | $graph->setAttribute( 'id', 'g0' ); # TODO make this meaningful |
423 | $graph->setAttribute( 'parse.edgeids', 'canonical' ); |
df6d9812 |
424 | $graph->setAttribute( 'parse.edges', scalar($self->paths) ); |
8e1394aa |
425 | $graph->setAttribute( 'parse.nodeids', 'canonical' ); |
df6d9812 |
426 | $graph->setAttribute( 'parse.nodes', scalar($self->readings) ); |
8e1394aa |
427 | $graph->setAttribute( 'parse.order', 'nodesfirst' ); |
428 | |
429 | my $node_ctr = 0; |
430 | my %node_hash; |
b15511bf |
431 | # Add our readings to the graph |
b5054ca9 |
432 | foreach my $n ( sort { $a->name cmp $b->name } $self->readings ) { |
910a0a6d |
433 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
434 | my $node_xmlid = 'n' . $node_ctr++; |
435 | $node_hash{ $n->name } = $node_xmlid; |
436 | $node_el->setAttribute( 'id', $node_xmlid ); |
437 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
438 | _add_graphml_data( $node_el, $node_data_keys{'reading'}, $n->label ); |
d9e873d0 |
439 | _add_graphml_data( $node_el, $node_data_keys{'rank'}, $n->rank ) |
440 | if $n->has_rank; |
910a0a6d |
441 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
442 | _add_graphml_data( $node_el, $node_data_keys{'identical'}, $n->primary->name ) |
443 | if $n->has_primary; |
8e1394aa |
444 | } |
445 | |
b15511bf |
446 | # Add any segments we have |
447 | foreach my $n ( sort { $a->name cmp $b->name } $self->segments ) { |
910a0a6d |
448 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
449 | my $node_xmlid = 'n' . $node_ctr++; |
450 | $node_hash{ $n->name } = $node_xmlid; |
451 | $node_el->setAttribute( 'id', $node_xmlid ); |
452 | _add_graphml_data( $node_el, $node_data_keys{'class'}, $n->sub_class ); |
453 | _add_graphml_data( $node_el, $node_data_keys{'name'}, $n->name ); |
b15511bf |
454 | } |
455 | |
456 | # Add the path, relationship, and segment edges |
df6d9812 |
457 | my $edge_ctr = 0; |
ef9d481f |
458 | foreach my $e ( sort { $a->from->name cmp $b->from->name } $self->graph->edges() ) { |
910a0a6d |
459 | my( $name, $from, $to ) = ( 'e'.$edge_ctr++, |
460 | $node_hash{ $e->from->name() }, |
461 | $node_hash{ $e->to->name() } ); |
462 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
463 | $edge_el->setAttribute( 'source', $from ); |
464 | $edge_el->setAttribute( 'target', $to ); |
465 | $edge_el->setAttribute( 'id', $name ); |
466 | # Add the edge class |
467 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, $e->sub_class ); |
468 | if( $e->sub_class eq 'path' ) { |
469 | # It's a witness path, so add the witness |
470 | my $base = $e->label; |
471 | my $key = $edge_data_keys{'witness_main'}; |
472 | # TODO kind of hacky |
473 | if( $e->label =~ /^(.*?)\s+(\(a\.c\.\))$/ ) { |
474 | $base = $1; |
475 | $key = $edge_data_keys{'witness_ante_corr'}; |
476 | } |
477 | _add_graphml_data( $edge_el, $key, $base ); |
478 | } elsif( $e->sub_class eq 'relationship' ) { |
479 | # It's a relationship |
480 | _add_graphml_data( $edge_el, $edge_data_keys{'relationship'}, $e->label ); |
481 | } # else a segment, nothing to record but source, target, class |
8e1394aa |
482 | } |
483 | |
484 | # Return the thing |
df6d9812 |
485 | $self->_save_graphml( $graphml->toString(1) ); |
486 | return $graphml->toString(1); |
487 | } |
488 | |
b15511bf |
489 | sub _add_graphml_data { |
490 | my( $el, $key, $value ) = @_; |
491 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
492 | return unless defined $value; |
493 | $data_el->setAttribute( 'key', $key ); |
494 | $data_el->appendText( $value ); |
8e1394aa |
495 | } |
496 | |
910a0a6d |
497 | =item B<as_csv> |
498 | |
499 | print $graph->as_csv( $recalculate ) |
500 | |
501 | Returns a CSV alignment table representation of the collation graph, one |
502 | row per witness (or witness uncorrected.) Unless $recalculate is passed |
503 | (and is a true value), the method will return a cached copy of the CSV |
504 | after the first call to the method. |
505 | |
506 | =cut |
507 | |
508 | sub as_csv { |
509 | my( $self, $recalc ) = @_; |
510 | return $self->csv if $self->has_csv; |
511 | my $table = $self->make_alignment_table; |
512 | my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } ); |
513 | my @result; |
514 | foreach my $row ( @$table ) { |
515 | $csv->combine( @$row ); |
516 | push( @result, decode_utf8( $csv->string ) ); |
517 | } |
518 | $self->_save_csv( join( "\n", @result ) ); |
519 | return $self->csv; |
520 | } |
521 | |
9f3ba6f7 |
522 | # TODO Make an alignment table at the end of initialization to check for |
523 | # duplicate nodes from mis-collation. |
524 | |
525 | |
526 | |
910a0a6d |
527 | sub make_alignment_table { |
9f3ba6f7 |
528 | my( $self, $in_rows ) = shift; |
910a0a6d |
529 | unless( $self->linear ) { |
530 | warn "Need a linear graph in order to make an alignment table"; |
531 | return; |
532 | } |
533 | my $table; |
534 | my @all_pos = sort { $a <=> $b } $self->possible_positions; |
535 | foreach my $wit ( $self->tradition->witnesses ) { |
eca16057 |
536 | # print STDERR "Making witness row(s) for " . $wit->sigil . "\n"; |
910a0a6d |
537 | my @row = _make_witness_row( $wit->path, \@all_pos ); |
538 | unshift( @row, $wit->sigil ); |
539 | push( @$table, \@row ); |
540 | if( $wit->has_ante_corr ) { |
541 | my @ac_row = _make_witness_row( $wit->uncorrected_path, \@all_pos ); |
542 | unshift( @ac_row, $wit->sigil . $self->ac_label ); |
543 | push( @$table, \@ac_row ); |
544 | } |
545 | } |
9f3ba6f7 |
546 | return $table if $in_rows; |
547 | |
910a0a6d |
548 | # Return a table where the witnesses read in columns rather than rows. |
549 | my $turned = _turn_table( $table ); |
550 | return $turned; |
551 | } |
552 | |
553 | sub _make_witness_row { |
554 | my( $path, $positions ) = @_; |
555 | my %char_hash; |
556 | map { $char_hash{$_} = undef } @$positions; |
557 | foreach my $rdg ( @$path ) { |
eca16057 |
558 | my $rtext = $rdg->text; |
559 | $rtext = '#LACUNA#' if $rdg->is_lacuna; |
560 | $char_hash{$rdg->rank} = $rtext; |
910a0a6d |
561 | } |
562 | my @row = map { $char_hash{$_} } @$positions; |
eca16057 |
563 | # Fill in lacuna markers for undef spots in the row |
564 | my $last_el = shift @row; |
565 | my @filled_row = ( $last_el ); |
566 | foreach my $el ( @row ) { |
567 | if( $last_el && $last_el eq '#LACUNA#' && !defined $el ) { |
568 | $el = '#LACUNA#'; |
569 | } |
570 | push( @filled_row, $el ); |
571 | $last_el = $el; |
572 | } |
573 | return @filled_row; |
910a0a6d |
574 | } |
575 | |
576 | # Helper to turn the witnesses along columns rather than rows. Assumes |
577 | # equal-sized rows. |
578 | sub _turn_table { |
579 | my( $table ) = @_; |
580 | my $result = []; |
581 | return $result unless scalar @$table; |
582 | my $nrows = scalar @{$table->[0]}; |
583 | foreach my $idx ( 0 .. $nrows - 1 ) { |
584 | foreach my $wit ( 0 .. $#{$table} ) { |
585 | $result->[$idx]->[$wit] = $table->[$wit]->[$idx]; |
586 | } |
587 | } |
588 | return $result; |
589 | } |
590 | |
591 | |
3265b0ce |
592 | sub collapse_graph_paths { |
1f563ac3 |
593 | my $self = shift; |
3265b0ce |
594 | # Our collation graph has an path per witness. This is great for |
1f563ac3 |
595 | # calculation purposes, but terrible for display. Thus we want to |
3265b0ce |
596 | # display only one path between any two nodes. |
1f563ac3 |
597 | |
598 | return if $self->collapsed; |
599 | |
3265b0ce |
600 | print STDERR "Collapsing witness paths in graph...\n"; |
1f563ac3 |
601 | |
602 | # Don't list out every witness if we have more than half to list. |
910a0a6d |
603 | my $majority = int( scalar( $self->tradition->witnesses ) / 2 ) + 1; |
a0093bf2 |
604 | # But don't compress if there are only a few witnesses. |
605 | $majority = 4 if $majority < 4; |
b15511bf |
606 | foreach my $node ( $self->readings ) { |
910a0a6d |
607 | my $newlabels = {}; |
608 | # We will visit each node, so we only look ahead. |
609 | foreach my $edge ( $node->outgoing() ) { |
610 | next unless $edge->class eq 'edge.path'; |
611 | add_hash_entry( $newlabels, $edge->to->name, $edge->name ); |
612 | $self->del_path( $edge ); |
613 | } |
614 | |
615 | foreach my $newdest ( keys %$newlabels ) { |
616 | my $label; |
617 | my @compressed_wits = @{$newlabels->{$newdest}}; |
618 | if( @compressed_wits < $majority ) { |
619 | $label = join( ', ', sort( @{$newlabels->{$newdest}} ) ); |
620 | } else { |
621 | ## TODO FIX THIS HACK |
622 | my @aclabels; |
623 | foreach my $wit ( @compressed_wits ) { |
624 | push( @aclabels, $wit ) if( $wit =~ /^(.*?)(\s*\(?a\.\s*c\.\)?)$/ ); |
625 | } |
626 | $label = join( ', ', 'majority', sort( @aclabels ) ); |
627 | } |
628 | |
629 | my $newpath = $self->add_path( $node, $self->reading( $newdest ), $label ); |
630 | $newpath->hidden_witnesses( \@compressed_wits ); |
631 | } |
1f563ac3 |
632 | } |
633 | |
634 | $self->collapsed( 1 ); |
635 | } |
636 | |
3265b0ce |
637 | sub expand_graph_paths { |
1f563ac3 |
638 | my $self = shift; |
3265b0ce |
639 | # Our collation graph has only one path between any two nodes. |
1f563ac3 |
640 | # This is great for display, but not so great for analysis. |
3265b0ce |
641 | # Expand this so that each witness has its own path between any |
1f563ac3 |
642 | # two reading nodes. |
643 | return unless $self->collapsed; |
644 | |
3265b0ce |
645 | print STDERR "Expanding witness paths in graph...\n"; |
3265b0ce |
646 | foreach my $path( $self->paths ) { |
910a0a6d |
647 | my $from = $path->from; |
648 | my $to = $path->to; |
649 | warn sprintf( "No hidden witnesses on %s -> %s ?", $from->name, $to->name ) |
650 | unless $path->has_hidden_witnesses; |
651 | my @wits = @{$path->hidden_witnesses}; |
652 | $self->del_path( $path ); |
653 | foreach ( @wits ) { |
654 | $self->add_path( $from, $to, $_ ); |
655 | } |
1f563ac3 |
656 | } |
657 | $self->collapsed( 0 ); |
658 | } |
659 | |
8e1394aa |
660 | =back |
661 | |
de51424a |
662 | =head2 Navigation methods |
663 | |
664 | =over |
665 | |
8e1394aa |
666 | =item B<start> |
667 | |
668 | my $beginning = $collation->start(); |
669 | |
670 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
671 | |
672 | =cut |
673 | |
674 | sub start { |
4a8828f0 |
675 | # Return the beginning reading of the graph. |
8e1394aa |
676 | my $self = shift; |
677 | my( $new_start ) = @_; |
678 | if( $new_start ) { |
910a0a6d |
679 | $self->del_reading( '#START#' ); |
680 | $self->graph->rename_node( $new_start, '#START#' ); |
681 | } |
682 | # Make sure the start node has a start position. |
d9e873d0 |
683 | unless( $self->reading( '#START#' )->has_rank ) { |
684 | $self->reading( '#START#' )->rank( '0' ); |
8e1394aa |
685 | } |
686 | return $self->reading('#START#'); |
687 | } |
688 | |
910a0a6d |
689 | =item B<end> |
690 | |
691 | my $end = $collation->end(); |
692 | |
693 | Returns the end of the collation, a meta-reading with label '#END#'. |
694 | |
695 | =cut |
696 | |
697 | sub end { |
698 | my $self = shift; |
699 | my( $new_end ) = @_; |
700 | if( $new_end ) { |
701 | $self->del_reading( '#END#' ); |
702 | $self->graph->rename_node( $new_end, '#END#' ); |
703 | } |
704 | return $self->reading('#END#'); |
705 | } |
706 | |
e2902068 |
707 | =item B<reading_sequence> |
708 | |
709 | my @readings = $graph->reading_sequence( $first, $last, $path[, $alt_path] ); |
710 | |
711 | Returns the ordered list of readings, starting with $first and ending |
712 | with $last, along the given witness path. If no path is specified, |
713 | assume that the path is that of the base text (if any.) |
714 | |
715 | =cut |
716 | |
910a0a6d |
717 | # TODO Think about returning some lazy-eval iterator. |
718 | |
e2902068 |
719 | sub reading_sequence { |
720 | my( $self, $start, $end, $witness, $backup ) = @_; |
721 | |
930ff666 |
722 | $witness = $self->baselabel unless $witness; |
e2902068 |
723 | my @readings = ( $start ); |
724 | my %seen; |
725 | my $n = $start; |
930ff666 |
726 | while( $n && $n ne $end ) { |
910a0a6d |
727 | if( exists( $seen{$n->name()} ) ) { |
728 | warn "Detected loop at " . $n->name(); |
729 | last; |
730 | } |
731 | $seen{$n->name()} = 1; |
732 | |
733 | my $next = $self->next_reading( $n, $witness, $backup ); |
734 | warn "Did not find any path for $witness from reading " . $n->name |
735 | unless $next; |
736 | push( @readings, $next ); |
737 | $n = $next; |
e2902068 |
738 | } |
739 | # Check that the last reading is our end reading. |
740 | my $last = $readings[$#readings]; |
741 | warn "Last reading found from " . $start->label() . |
910a0a6d |
742 | " for witness $witness is not the end!" |
743 | unless $last eq $end; |
e2902068 |
744 | |
745 | return @readings; |
746 | } |
747 | |
4a8828f0 |
748 | =item B<next_reading> |
8e1394aa |
749 | |
4a8828f0 |
750 | my $next_reading = $graph->next_reading( $reading, $witpath ); |
8e1394aa |
751 | |
4a8828f0 |
752 | Returns the reading that follows the given reading along the given witness |
930ff666 |
753 | path. |
8e1394aa |
754 | |
755 | =cut |
756 | |
4a8828f0 |
757 | sub next_reading { |
e2902068 |
758 | # Return the successor via the corresponding path. |
8e1394aa |
759 | my $self = shift; |
4a8828f0 |
760 | return $self->_find_linked_reading( 'next', @_ ); |
8e1394aa |
761 | } |
762 | |
4a8828f0 |
763 | =item B<prior_reading> |
8e1394aa |
764 | |
4a8828f0 |
765 | my $prior_reading = $graph->prior_reading( $reading, $witpath ); |
8e1394aa |
766 | |
4a8828f0 |
767 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
768 | path. |
8e1394aa |
769 | |
770 | =cut |
771 | |
4a8828f0 |
772 | sub prior_reading { |
e2902068 |
773 | # Return the predecessor via the corresponding path. |
8e1394aa |
774 | my $self = shift; |
4a8828f0 |
775 | return $self->_find_linked_reading( 'prior', @_ ); |
8e1394aa |
776 | } |
777 | |
4a8828f0 |
778 | sub _find_linked_reading { |
e2902068 |
779 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
780 | my @linked_paths = $direction eq 'next' |
910a0a6d |
781 | ? $node->outgoing() : $node->incoming(); |
e2902068 |
782 | return undef unless scalar( @linked_paths ); |
8e1394aa |
783 | |
e2902068 |
784 | # We have to find the linked path that contains all of the |
785 | # witnesses supplied in $path. |
786 | my( @path_wits, @alt_path_wits ); |
787 | @path_wits = $self->witnesses_of_label( $path ) if $path; |
788 | @alt_path_wits = $self->witnesses_of_label( $alt_path ) if $alt_path; |
789 | my $base_le; |
790 | my $alt_le; |
791 | foreach my $le ( @linked_paths ) { |
910a0a6d |
792 | if( $le->name eq $self->baselabel ) { |
793 | $base_le = $le; |
794 | } else { |
795 | my @le_wits = $self->witnesses_of_label( $le->name ); |
796 | if( _is_within( \@path_wits, \@le_wits ) ) { |
797 | # This is the right path. |
798 | return $direction eq 'next' ? $le->to() : $le->from(); |
799 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
800 | $alt_le = $le; |
801 | } |
802 | } |
8e1394aa |
803 | } |
e2902068 |
804 | # Got this far? Return the alternate path if it exists. |
805 | return $direction eq 'next' ? $alt_le->to() : $alt_le->from() |
910a0a6d |
806 | if $alt_le; |
e2902068 |
807 | |
808 | # Got this far? Return the base path if it exists. |
809 | return $direction eq 'next' ? $base_le->to() : $base_le->from() |
910a0a6d |
810 | if $base_le; |
e2902068 |
811 | |
812 | # Got this far? We have no appropriate path. |
8e1394aa |
813 | warn "Could not find $direction node from " . $node->label |
910a0a6d |
814 | . " along path $path"; |
8e1394aa |
815 | return undef; |
816 | } |
817 | |
4a8828f0 |
818 | # Some set logic. |
819 | sub _is_within { |
820 | my( $set1, $set2 ) = @_; |
7854e12e |
821 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
822 | foreach my $el ( @$set1 ) { |
910a0a6d |
823 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
4a8828f0 |
824 | } |
825 | return $ret; |
826 | } |
827 | |
de51424a |
828 | |
829 | ## INITIALIZATION METHODS - for use by parsers |
4a8828f0 |
830 | # Walk the paths for each witness in the graph, and return the nodes |
e2902068 |
831 | # that the graph has in common. If $using_base is true, some |
832 | # different logic is needed. |
7e450e44 |
833 | # NOTE This does not create paths; it merely finds common readings. |
4a8828f0 |
834 | |
835 | sub walk_witness_paths { |
7e450e44 |
836 | my( $self ) = @_; |
4a8828f0 |
837 | # For each witness, walk the path through the graph. |
838 | # Then we need to find the common nodes. |
839 | # TODO This method is going to fall down if we have a very gappy |
840 | # text in the collation. |
841 | my $paths = {}; |
3a1f2523 |
842 | my @common_readings; |
910a0a6d |
843 | foreach my $wit ( $self->tradition->witnesses ) { |
844 | my $curr_reading = $self->start; |
7e450e44 |
845 | my @wit_path = $self->reading_sequence( $self->start, $self->end, |
910a0a6d |
846 | $wit->sigil ); |
847 | $wit->path( \@wit_path ); |
848 | |
849 | # Detect the common readings. |
850 | @common_readings = _find_common( \@common_readings, \@wit_path ); |
4a8828f0 |
851 | } |
852 | |
853 | # Mark all the nodes as either common or not. |
3a1f2523 |
854 | foreach my $cn ( @common_readings ) { |
910a0a6d |
855 | print STDERR "Setting " . $cn->name . " / " . $cn->label |
856 | . " as common node\n"; |
857 | $cn->make_common; |
4a8828f0 |
858 | } |
859 | foreach my $n ( $self->readings() ) { |
910a0a6d |
860 | $n->make_variant unless $n->is_common; |
4a8828f0 |
861 | } |
3a1f2523 |
862 | # Return an array of the common nodes in order. |
863 | return @common_readings; |
4a8828f0 |
864 | } |
865 | |
930ff666 |
866 | sub _find_common { |
867 | my( $common_readings, $new_path ) = @_; |
868 | my @cr; |
869 | if( @$common_readings ) { |
910a0a6d |
870 | foreach my $n ( @$new_path ) { |
871 | push( @cr, $n ) if grep { $_ eq $n } @$common_readings; |
872 | } |
930ff666 |
873 | } else { |
910a0a6d |
874 | push( @cr, @$new_path ); |
930ff666 |
875 | } |
876 | return @cr; |
877 | } |
878 | |
879 | sub _remove_common { |
880 | my( $common_readings, $divergence ) = @_; |
881 | my @cr; |
882 | my %diverged; |
883 | map { $diverged{$_->name} = 1 } @$divergence; |
884 | foreach( @$common_readings ) { |
910a0a6d |
885 | push( @cr, $_ ) unless $diverged{$_->name}; |
930ff666 |
886 | } |
887 | return @cr; |
888 | } |
889 | |
890 | |
7e450e44 |
891 | # For use when a collation is constructed from a base text and an apparatus. |
892 | # We have the sequences of readings and just need to add path edges. |
e2902068 |
893 | |
6a222840 |
894 | sub make_witness_paths { |
895 | my( $self ) = @_; |
910a0a6d |
896 | foreach my $wit ( $self->tradition->witnesses ) { |
897 | print STDERR "Making path for " . $wit->sigil . "\n"; |
898 | $self->make_witness_path( $wit ); |
7854e12e |
899 | } |
7854e12e |
900 | } |
901 | |
6a222840 |
902 | sub make_witness_path { |
7854e12e |
903 | my( $self, $wit ) = @_; |
904 | my @chain = @{$wit->path}; |
15d2d3df |
905 | my $sig = $wit->sigil; |
7854e12e |
906 | foreach my $idx ( 0 .. $#chain-1 ) { |
910a0a6d |
907 | $self->add_path( $chain[$idx], $chain[$idx+1], $sig ); |
7854e12e |
908 | } |
d9e873d0 |
909 | if( $wit->has_ante_corr ) { |
910 | @chain = @{$wit->uncorrected_path}; |
911 | foreach my $idx( 0 .. $#chain-1 ) { |
912 | my $source = $chain[$idx]; |
913 | my $target = $chain[$idx+1]; |
914 | $self->add_path( $source, $target, $sig.$self->ac_label ) |
915 | unless $self->has_path( $source, $target, $sig ); |
916 | } |
15d2d3df |
917 | } |
e2902068 |
918 | } |
919 | |
910a0a6d |
920 | sub calculate_ranks { |
921 | my $self = shift; |
922 | # Walk a version of the graph where every node linked by a relationship |
923 | # edge is fundamentally the same node, and do a topological ranking on |
924 | # the nodes in this graph. |
925 | my $topo_graph = Graph::Easy->new(); |
926 | my %rel_containers; |
927 | my $rel_ctr = 0; |
928 | # Add the nodes |
929 | foreach my $r ( $self->readings ) { |
930 | next if exists $rel_containers{$r->name}; |
931 | my @rels = $r->related_readings( 'colocated' ); |
932 | if( @rels ) { |
933 | # Make a relationship container. |
934 | push( @rels, $r ); |
935 | my $rn = $topo_graph->add_node( 'rel_container_' . $rel_ctr++ ); |
936 | foreach( @rels ) { |
937 | $rel_containers{$_->name} = $rn; |
938 | } |
939 | } else { |
940 | # Add a new node to mirror the old node. |
941 | $rel_containers{$r->name} = $topo_graph->add_node( $r->name ); |
942 | } |
4a8828f0 |
943 | } |
3a1f2523 |
944 | |
910a0a6d |
945 | # Add the edges. Need only one edge between any pair of nodes. |
946 | foreach my $r ( $self->readings ) { |
947 | foreach my $n ( $r->neighbor_readings( 'forward' ) ) { |
948 | $topo_graph->add_edge_once( $rel_containers{$r->name}, |
949 | $rel_containers{$n->name} ); |
950 | } |
951 | } |
952 | |
953 | # Now do the rankings, starting with the start node. |
954 | my $topo_start = $rel_containers{$self->start->name}; |
955 | my $node_ranks = { $topo_start->name => 0 }; |
956 | my @curr_origin = ( $topo_start ); |
957 | # A little iterative function. |
958 | while( @curr_origin ) { |
959 | @curr_origin = _assign_rank( $node_ranks, @curr_origin ); |
960 | } |
961 | # Transfer our rankings from the topological graph to the real one. |
962 | foreach my $r ( $self->readings ) { |
963 | $r->rank( $node_ranks->{$rel_containers{$r->name}->name} ); |
de51424a |
964 | } |
8e1394aa |
965 | } |
3a1f2523 |
966 | |
910a0a6d |
967 | sub _assign_rank { |
968 | my( $node_ranks, @current_nodes ) = @_; |
969 | # Look at each of the children of @current_nodes. If all the child's |
970 | # parents have a rank, assign it the highest rank + 1 and add it to |
971 | # @next_nodes. Otherwise skip it. |
972 | my @next_nodes; |
973 | foreach my $c ( @current_nodes ) { |
974 | warn "Current reading " . $c->name . "has no rank!" |
975 | unless exists $node_ranks->{$c->name}; |
976 | # print STDERR "Looking at child of node " . $c->name . ", rank " |
977 | # . $node_ranks->{$c->name} . "\n"; |
978 | my @children = map { $_->to } $c->outgoing; |
979 | foreach my $child ( @children ) { |
980 | next if exists $node_ranks->{$child->name}; |
981 | my $highest_rank = -1; |
982 | my $skip = 0; |
983 | my @parents = map { $_->from } $child->incoming; |
984 | foreach my $parent ( @parents ) { |
985 | if( exists $node_ranks->{$parent->name} ) { |
986 | $highest_rank = $node_ranks->{$parent->name} |
987 | if $highest_rank <= $node_ranks->{$parent->name}; |
988 | } else { |
989 | $skip = 1; |
990 | last; |
991 | } |
992 | } |
993 | next if $skip; |
994 | # print STDERR "Assigning rank " . ( $highest_rank + 1 ) . " to node " . $child->name . "\n"; |
995 | $node_ranks->{$child->name} = $highest_rank + 1; |
996 | push( @next_nodes, $child ); |
997 | } |
998 | } |
999 | return @next_nodes; |
4cdd82f1 |
1000 | } |
910a0a6d |
1001 | |
4cdd82f1 |
1002 | sub possible_positions { |
3a1f2523 |
1003 | my $self = shift; |
910a0a6d |
1004 | my %all_pos; |
1005 | map { $all_pos{ $_->rank } = 1 } $self->readings; |
1006 | return keys %all_pos; |
3a1f2523 |
1007 | } |
1008 | |
4cdd82f1 |
1009 | # TODO think about indexing this. |
3a1f2523 |
1010 | sub readings_at_position { |
4cdd82f1 |
1011 | my( $self, $position, $strict ) = @_; |
4cdd82f1 |
1012 | my @answer; |
1013 | foreach my $r ( $self->readings ) { |
910a0a6d |
1014 | push( @answer, $r ) if $r->is_at_position( $position, $strict ); |
4cdd82f1 |
1015 | } |
3a1f2523 |
1016 | return @answer; |
1017 | } |
1018 | |
1019 | ## Lemmatizer functions |
1020 | |
1021 | sub init_lemmata { |
1022 | my $self = shift; |
4cdd82f1 |
1023 | |
1024 | foreach my $position ( $self->possible_positions ) { |
910a0a6d |
1025 | $self->lemmata->{$position} = undef; |
3a1f2523 |
1026 | } |
1027 | |
1028 | foreach my $cr ( $self->common_readings ) { |
910a0a6d |
1029 | $self->lemmata->{$cr->position->maxref} = $cr->name; |
3a1f2523 |
1030 | } |
1031 | } |
7e450e44 |
1032 | |
1033 | sub common_readings { |
1034 | my $self = shift; |
1035 | my @common = grep { $_->is_common } $self->readings(); |
d9e873d0 |
1036 | return sort { $a->rank <=> $b->rank } @common; |
7e450e44 |
1037 | } |
3a1f2523 |
1038 | |
1039 | =item B<lemma_readings> |
1040 | |
1041 | my @state = $graph->lemma_readings( @readings_delemmatized ); |
1042 | |
1043 | Takes a list of readings that have just been delemmatized, and returns |
1044 | a set of tuples of the form ['reading', 'state'] that indicates what |
1045 | changes need to be made to the graph. |
1046 | |
1047 | =over |
1048 | |
1049 | =item * |
1050 | |
1051 | A state of 1 means 'lemmatize this reading' |
1052 | |
1053 | =item * |
1054 | |
1055 | A state of 0 means 'delemmatize this reading' |
1056 | |
1057 | =item * |
1058 | |
1059 | A state of undef means 'an ellipsis belongs in the text here because |
1060 | no decision has been made / an earlier decision was backed out' |
1061 | |
1062 | =back |
1063 | |
1064 | =cut |
1065 | |
1066 | sub lemma_readings { |
1067 | my( $self, @toggled_off_nodes ) = @_; |
1068 | |
1069 | # First get the positions of those nodes which have been |
1070 | # toggled off. |
1071 | my $positions_off = {}; |
4cdd82f1 |
1072 | map { $positions_off->{ $_->position->reference } = $_->name } |
1073 | @toggled_off_nodes; |
de51424a |
1074 | |
3a1f2523 |
1075 | # Now for each position, we have to see if a node is on, and we |
4cdd82f1 |
1076 | # have to see if a node has been turned off. The lemmata hash |
1077 | # should contain fixed positions, range positions whose node was |
1078 | # just turned off, and range positions whose node is on. |
3a1f2523 |
1079 | my @answer; |
4cdd82f1 |
1080 | my %fixed_positions; |
1081 | # TODO One of these is probably redundant. |
1082 | map { $fixed_positions{$_} = 0 } keys %{$self->lemmata}; |
1083 | map { $fixed_positions{$_} = 0 } keys %{$positions_off}; |
1084 | map { $fixed_positions{$_} = 1 } $self->possible_positions; |
1085 | foreach my $pos ( sort { Text::Tradition::Collation::Position::str_cmp( $a, $b ) } keys %fixed_positions ) { |
910a0a6d |
1086 | # Find the state of this position. If there is an active node, |
1087 | # its name will be the state; otherwise the state will be 0 |
1088 | # (nothing at this position) or undef (ellipsis at this position) |
1089 | my $active = undef; |
1090 | $active = $self->lemmata->{$pos} if exists $self->lemmata->{$pos}; |
1091 | |
1092 | # Is there a formerly active node that was toggled off? |
1093 | if( exists( $positions_off->{$pos} ) ) { |
1094 | my $off_node = $positions_off->{$pos}; |
1095 | if( $active && $active ne $off_node) { |
1096 | push( @answer, [ $off_node, 0 ], [ $active, 1 ] ); |
1097 | } else { |
1098 | unless( $fixed_positions{$pos} ) { |
1099 | $active = 0; |
1100 | delete $self->lemmata->{$pos}; |
1101 | } |
1102 | push( @answer, [ $off_node, $active ] ); |
1103 | } |
1104 | |
1105 | # No formerly active node, so we just see if there is a currently |
1106 | # active one. |
1107 | } elsif( $active ) { |
1108 | # Push the active node, whatever it is. |
1109 | push( @answer, [ $active, 1 ] ); |
1110 | } else { |
1111 | # Push the state that is there. Arbitrarily use the first node |
1112 | # at that position. |
1113 | my @pos_nodes = $self->readings_at_position( $pos ); |
1114 | push( @answer, [ $pos_nodes[0]->name, $self->lemmata->{$pos} ] ); |
1115 | delete $self->lemmata->{$pos} unless $fixed_positions{$pos}; |
1116 | } |
3a1f2523 |
1117 | } |
4cdd82f1 |
1118 | |
3a1f2523 |
1119 | return @answer; |
1120 | } |
1121 | |
de51424a |
1122 | =item B<toggle_reading> |
1123 | |
1124 | my @readings_delemmatized = $graph->toggle_reading( $reading_name ); |
1125 | |
1126 | Takes a reading node name, and either lemmatizes or de-lemmatizes |
1127 | it. Returns a list of all readings that are de-lemmatized as a result |
1128 | of the toggle. |
1129 | |
1130 | =cut |
1131 | |
1132 | sub toggle_reading { |
1133 | my( $self, $rname ) = @_; |
1134 | |
1135 | return unless $rname; |
1136 | my $reading = $self->reading( $rname ); |
1137 | if( !$reading || $reading->is_common() ) { |
910a0a6d |
1138 | # Do nothing, it's a common node. |
1139 | return; |
de51424a |
1140 | } |
1141 | |
1142 | my $pos = $reading->position; |
4cdd82f1 |
1143 | my $fixed = $reading->position->fixed; |
1144 | my $old_state = $self->lemmata->{$pos->reference}; |
1145 | |
de51424a |
1146 | my @readings_off; |
1147 | if( $old_state && $old_state eq $rname ) { |
910a0a6d |
1148 | # Turn off the node. We turn on no others by default. |
1149 | push( @readings_off, $reading ); |
de51424a |
1150 | } else { |
910a0a6d |
1151 | # Turn on the node. |
1152 | $self->lemmata->{$pos->reference} = $rname; |
1153 | # Any other 'on' readings in the same position should be off |
1154 | # if we have a fixed position. |
1155 | push( @readings_off, $self->same_position_as( $reading, 1 ) ) |
1156 | if $pos->fixed; |
1157 | # Any node that is an identical transposed one should be off. |
1158 | push( @readings_off, $reading->identical_readings ); |
de51424a |
1159 | } |
1160 | @readings_off = unique_list( @readings_off ); |
910a0a6d |
1161 | |
de51424a |
1162 | # Turn off the readings that need to be turned off. |
1163 | my @readings_delemmatized; |
1164 | foreach my $n ( @readings_off ) { |
910a0a6d |
1165 | my $npos = $n->position; |
1166 | my $state = undef; |
1167 | $state = $self->lemmata->{$npos->reference} |
1168 | if defined $self->lemmata->{$npos->reference}; |
1169 | if( $state && $state eq $n->name ) { |
1170 | # this reading is still on, so turn it off |
1171 | push( @readings_delemmatized, $n ); |
1172 | my $new_state = undef; |
1173 | if( $npos->fixed && $n eq $reading ) { |
1174 | # This is the reading that was clicked, so if there are no |
1175 | # other readings there and this is a fixed position, turn off |
1176 | # the position. In all other cases, restore the ellipsis. |
1177 | my @other_n = $self->same_position_as( $n ); # TODO do we need strict? |
1178 | $new_state = 0 unless @other_n; |
1179 | } |
1180 | $self->lemmata->{$npos->reference} = $new_state; |
1181 | } elsif( $old_state && $old_state eq $n->name ) { |
1182 | # another reading has already been turned on here |
1183 | push( @readings_delemmatized, $n ); |
1184 | } # else some other reading was on anyway, so pass. |
de51424a |
1185 | } |
1186 | return @readings_delemmatized; |
1187 | } |
1188 | |
1189 | sub same_position_as { |
4cdd82f1 |
1190 | my( $self, $reading, $strict ) = @_; |
de51424a |
1191 | my $pos = $reading->position; |
4cdd82f1 |
1192 | my %onpath = ( $reading->name => 1 ); |
1193 | # TODO This might not always be sufficient. We really want to |
1194 | # exclude all readings on this one's path between its two |
1195 | # common points. |
1196 | map { $onpath{$_->name} = 1 } $reading->neighbor_readings; |
1197 | my @same = grep { !$onpath{$_->name} } |
1198 | $self->readings_at_position( $reading->position, $strict ); |
de51424a |
1199 | return @same; |
1200 | } |
3a1f2523 |
1201 | |
4a8828f0 |
1202 | # Return the string that joins together a list of witnesses for |
1203 | # display on a single path. |
1204 | sub path_label { |
1205 | my $self = shift; |
1206 | return join( $self->wit_list_separator, @_ ); |
1207 | } |
1208 | |
1209 | sub witnesses_of_label { |
de51424a |
1210 | my( $self, $label ) = @_; |
4a8828f0 |
1211 | my $regex = $self->wit_list_separator; |
de51424a |
1212 | my @answer = split( /\Q$regex\E/, $label ); |
1213 | return @answer; |
4a8828f0 |
1214 | } |
8e1394aa |
1215 | |
de51424a |
1216 | sub unique_list { |
1217 | my( @list ) = @_; |
1218 | my %h; |
1219 | map { $h{$_->name} = $_ } @list; |
1220 | return values( %h ); |
1221 | } |
1222 | |
1f563ac3 |
1223 | sub add_hash_entry { |
1224 | my( $hash, $key, $entry ) = @_; |
1225 | if( exists $hash->{$key} ) { |
910a0a6d |
1226 | push( @{$hash->{$key}}, $entry ); |
1f563ac3 |
1227 | } else { |
910a0a6d |
1228 | $hash->{$key} = [ $entry ]; |
1f563ac3 |
1229 | } |
1230 | } |
1231 | |
dd3b58b0 |
1232 | no Moose; |
1233 | __PACKAGE__->meta->make_immutable; |