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