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; |
dd3b58b0 |
6 | use Moose; |
7 | |
8 | has 'graph' => ( |
d047cd52 |
9 | is => 'ro', |
10 | isa => 'Graph::Easy', |
11 | handles => { |
8e1394aa |
12 | add_reading => 'add_node', |
13 | del_reading => 'del_node', |
14 | add_path => 'add_edge', |
15 | del_path => 'del_edge', |
16 | reading => 'node', |
17 | path => 'edge', |
18 | readings => 'nodes', |
19 | paths => 'edges', |
d047cd52 |
20 | }, |
21 | default => sub { Graph::Easy->new( undirected => 0 ) }, |
22 | ); |
784877d9 |
23 | |
dd3b58b0 |
24 | |
dd3b58b0 |
25 | has 'tradition' => ( |
8e1394aa |
26 | is => 'rw', |
d047cd52 |
27 | isa => 'Text::Tradition', |
28 | ); |
dd3b58b0 |
29 | |
8e1394aa |
30 | has 'svg' => ( |
31 | is => 'ro', |
32 | isa => 'Str', |
33 | writer => '_save_svg', |
34 | predicate => 'has_svg', |
35 | ); |
36 | |
37 | has 'graphviz' => ( |
38 | is => 'ro', |
39 | isa => 'Str', |
40 | writer => '_save_graphviz', |
41 | predicate => 'has_graphviz', |
42 | ); |
43 | |
44 | has 'graphml' => ( |
45 | is => 'ro', |
46 | isa => 'XML::LibXML::Document', |
47 | writer => '_save_graphml', |
48 | predicate => 'has_graphml', |
49 | ); |
50 | |
3a1f2523 |
51 | # Keeps track of the lemmas within the collation. At most one lemma |
52 | # per position in the graph. |
53 | has 'lemmata' => ( |
54 | is => 'ro', |
55 | isa => 'HashRef[Maybe[Str]]', |
56 | default => sub { {} }, |
57 | ); |
58 | |
4a8828f0 |
59 | has 'wit_list_separator' => ( |
7854e12e |
60 | is => 'rw', |
61 | isa => 'Str', |
62 | default => ', ', |
63 | ); |
64 | |
65 | has 'baselabel' => ( |
66 | is => 'rw', |
67 | isa => 'Str', |
68 | default => 'base text', |
69 | ); |
4a8828f0 |
70 | |
dd3b58b0 |
71 | # The collation can be created two ways: |
72 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
73 | # the results as in 2. |
74 | # 2. Read a pre-prepared collation in one of a variety of formats, |
75 | # and make the graph from that. |
76 | |
77 | # The graph itself will (for now) be immutable, and the positions |
78 | # within the graph will also be immutable. We need to calculate those |
79 | # positions upon graph construction. The equivalences between graph |
80 | # nodes will be mutable, entirely determined by the user (or possibly |
81 | # by some semantic pre-processing provided by the user.) So the |
82 | # constructor should just make an empty equivalences object. The |
83 | # constructor will also need to make the witness objects, if we didn't |
84 | # come through option 1. |
85 | |
d047cd52 |
86 | sub BUILD { |
87 | my( $self, $args ) = @_; |
8e1394aa |
88 | $self->graph->use_class('node', 'Text::Tradition::Collation::Reading'); |
d047cd52 |
89 | |
4a8828f0 |
90 | # Pass through any graph-specific options. |
91 | my $shape = exists( $args->{'shape'} ) ? $args->{'shape'} : 'ellipse'; |
92 | $self->graph->set_attribute( 'node', 'shape', $shape ); |
d047cd52 |
93 | } |
784877d9 |
94 | |
7854e12e |
95 | # Wrapper around add_path |
96 | |
97 | around add_path => sub { |
98 | my $orig = shift; |
99 | my $self = shift; |
100 | |
101 | # Make sure there are three arguments |
102 | unless( @_ == 3 ) { |
103 | warn "Call add_path with args source, target, witness"; |
104 | return; |
105 | } |
106 | # Make sure the proposed path does not yet exist |
107 | my( $source, $target, $wit ) = @_; |
108 | $source = $self->reading( $source ) |
109 | unless ref( $source ) eq 'Text::Tradition::Collation::Reading'; |
110 | $target = $self->reading( $target ) |
111 | unless ref( $target ) eq 'Text::Tradition::Collation::Reading'; |
112 | foreach my $path ( $source->edges_to( $target ) ) { |
113 | if( $path->label eq $wit ) { |
114 | return; |
115 | } |
116 | } |
117 | # Do the deed |
118 | $self->$orig( @_ ); |
119 | }; |
120 | |
121 | # Wrapper around merge_nodes |
784877d9 |
122 | |
123 | sub merge_readings { |
124 | my $self = shift; |
125 | my $first_node = shift; |
126 | my $second_node = shift; |
127 | $first_node->merge_from( $second_node ); |
128 | unshift( @_, $first_node, $second_node ); |
129 | return $self->graph->merge_nodes( @_ ); |
130 | } |
131 | |
8e1394aa |
132 | =head2 Output method(s) |
133 | |
134 | =over |
135 | |
136 | =item B<as_svg> |
137 | |
138 | print $graph->as_svg( $recalculate ); |
139 | |
140 | Returns an SVG string that represents the graph. Uses GraphViz to do |
4a8828f0 |
141 | this, because Graph::Easy doesn\'t cope well with long graphs. Unless |
8e1394aa |
142 | $recalculate is passed (and is a true value), the method will return a |
143 | cached copy of the SVG after the first call to the method. |
144 | |
145 | =cut |
146 | |
147 | sub as_svg { |
148 | my( $self, $recalc ) = @_; |
149 | return $self->svg if $self->has_svg; |
150 | |
151 | $self->_save_graphviz( $self->graph->as_graphviz() ) |
152 | unless( $self->has_graphviz && !$recalc ); |
153 | |
154 | my @cmd = qw/dot -Tsvg/; |
155 | my( $svg, $err ); |
156 | my $in = $self->graphviz; |
157 | run( \@cmd, \$in, ">", binary(), \$svg ); |
158 | $self->{'svg'} = $svg; |
159 | return $svg; |
160 | } |
161 | |
162 | =item B<as_graphml> |
163 | |
164 | print $graph->as_graphml( $recalculate ) |
165 | |
166 | Returns a GraphML representation of the collation graph, with |
167 | transposition information and position information. Unless |
168 | $recalculate is passed (and is a true value), the method will return a |
169 | cached copy of the SVG after the first call to the method. |
170 | |
171 | =cut |
172 | |
173 | sub as_graphml { |
174 | my( $self, $recalc ) = @_; |
175 | return $self->graphml if $self->has_graphml; |
176 | |
177 | # Some namespaces |
178 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
179 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
180 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
181 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
182 | |
183 | # Create the document and root node |
184 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
185 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
186 | $graphml->setDocumentElement( $root ); |
187 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
188 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
189 | |
190 | # Add the data keys for nodes |
191 | my @node_data = ( 'name', 'token', 'identical', 'position' ); |
192 | foreach my $ndi ( 0 .. $#node_data ) { |
193 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
194 | $key->setAttribute( 'attr.name', $node_data[$ndi] ); |
195 | $key->setAttribute( 'attr.type', 'string' ); |
196 | $key->setAttribute( 'for', 'node' ); |
197 | $key->setAttribute( 'id', 'd'.$ndi ); |
198 | } |
199 | |
200 | # Add the data keys for edges |
201 | my %wit_hash; |
202 | my $wit_ctr = 0; |
203 | foreach my $wit ( $self->getWitnessList ) { |
204 | my $wit_key = 'w' . $wit_ctr++; |
205 | $wit_hash{$wit} = $wit_key; |
206 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
207 | $key->setAttribute( 'attr.name', $wit ); |
208 | $key->setAttribute( 'attr.type', 'string' ); |
209 | $key->setAttribute( 'for', 'edge' ); |
210 | $key->setAttribute( 'id', $wit_key ); |
211 | } |
212 | |
213 | # Add the graph, its nodes, and its edges |
214 | my $graph = $root->addNewChild( $graphml_ns, 'graph' ); |
215 | $graph->setAttribute( 'edgedefault', 'directed' ); |
216 | $graph->setAttribute( 'id', 'g0' ); # TODO make this meaningful |
217 | $graph->setAttribute( 'parse.edgeids', 'canonical' ); |
218 | $graph->setAttribute( 'parse.edges', $self->edges() ); |
219 | $graph->setAttribute( 'parse.nodeids', 'canonical' ); |
220 | $graph->setAttribute( 'parse.nodes', $self->nodes() ); |
221 | $graph->setAttribute( 'parse.order', 'nodesfirst' ); |
222 | |
223 | my $node_ctr = 0; |
224 | my %node_hash; |
225 | foreach my $n ( $self->readings ) { |
226 | my %this_node_data = (); |
227 | foreach my $ndi ( 0 .. $#node_data ) { |
228 | my $value; |
229 | $this_node_data{'d'.$ndi} = $n->name if $node_data[$ndi] eq 'name'; |
230 | $this_node_data{'d'.$ndi} = $n->label |
231 | if $node_data[$ndi] eq 'token'; |
232 | $this_node_data{'d'.$ndi} = $n->primary->name if $n->has_primary; |
233 | $this_node_data{'d'.$ndi} = |
234 | $self->{'positions'}->node_position( $n ) |
235 | if $node_data[$ndi] eq 'position'; |
236 | } |
237 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
238 | my $node_xmlid = 'n' . $node_ctr++; |
239 | $node_hash{ $n->name } = $node_xmlid; |
240 | $node_el->setAttribute( 'id', $node_xmlid ); |
241 | |
242 | foreach my $dk ( keys %this_node_data ) { |
243 | my $d_el = $node_el->addNewChild( $graphml_ns, 'data' ); |
244 | $d_el->setAttribute( 'key', $dk ); |
245 | $d_el->appendTextChild( $this_node_data{$dk} ); |
246 | } |
247 | } |
248 | |
249 | foreach my $e ( $self->edges() ) { |
250 | my( $name, $from, $to ) = ( $e->name, |
251 | $node_hash{ $e->from()->name() }, |
252 | $node_hash{ $e->to()->name() } ); |
253 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
254 | $edge_el->setAttribute( 'source', $from ); |
255 | $edge_el->setAttribute( 'target', $to ); |
256 | $edge_el->setAttribute( 'id', $name ); |
257 | # TODO Got to add the witnesses |
258 | } |
259 | |
260 | # Return the thing |
261 | $self->_save_graphml( $graphml ); |
262 | return $graphml; |
263 | } |
264 | |
265 | =back |
266 | |
de51424a |
267 | =head2 Navigation methods |
268 | |
269 | =over |
270 | |
8e1394aa |
271 | =item B<start> |
272 | |
273 | my $beginning = $collation->start(); |
274 | |
275 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
276 | |
277 | =cut |
278 | |
279 | sub start { |
4a8828f0 |
280 | # Return the beginning reading of the graph. |
8e1394aa |
281 | my $self = shift; |
282 | my( $new_start ) = @_; |
283 | if( $new_start ) { |
284 | $self->del_reading( '#START#' ); |
285 | $self->graph->rename_node( $new_start, '#START#' ); |
286 | } |
287 | return $self->reading('#START#'); |
288 | } |
289 | |
e2902068 |
290 | =item B<reading_sequence> |
291 | |
292 | my @readings = $graph->reading_sequence( $first, $last, $path[, $alt_path] ); |
293 | |
294 | Returns the ordered list of readings, starting with $first and ending |
295 | with $last, along the given witness path. If no path is specified, |
296 | assume that the path is that of the base text (if any.) |
297 | |
298 | =cut |
299 | |
300 | sub reading_sequence { |
301 | my( $self, $start, $end, $witness, $backup ) = @_; |
302 | |
930ff666 |
303 | $witness = $self->baselabel unless $witness; |
e2902068 |
304 | my @readings = ( $start ); |
305 | my %seen; |
306 | my $n = $start; |
930ff666 |
307 | while( $n && $n ne $end ) { |
e2902068 |
308 | if( exists( $seen{$n->name()} ) ) { |
309 | warn "Detected loop at " . $n->name(); |
310 | last; |
311 | } |
312 | $seen{$n->name()} = 1; |
313 | |
314 | my $next = $self->next_reading( $n, $witness, $backup ); |
315 | warn "Did not find any path for $witness from reading " . $n->name |
316 | unless $next; |
317 | push( @readings, $next ); |
318 | $n = $next; |
319 | } |
320 | # Check that the last reading is our end reading. |
321 | my $last = $readings[$#readings]; |
322 | warn "Last reading found from " . $start->label() . |
323 | " for witness $witness is not the end!" |
324 | unless $last eq $end; |
325 | |
326 | return @readings; |
327 | } |
328 | |
4a8828f0 |
329 | =item B<next_reading> |
8e1394aa |
330 | |
4a8828f0 |
331 | my $next_reading = $graph->next_reading( $reading, $witpath ); |
8e1394aa |
332 | |
4a8828f0 |
333 | Returns the reading that follows the given reading along the given witness |
930ff666 |
334 | path. |
8e1394aa |
335 | |
336 | =cut |
337 | |
4a8828f0 |
338 | sub next_reading { |
e2902068 |
339 | # Return the successor via the corresponding path. |
8e1394aa |
340 | my $self = shift; |
4a8828f0 |
341 | return $self->_find_linked_reading( 'next', @_ ); |
8e1394aa |
342 | } |
343 | |
4a8828f0 |
344 | =item B<prior_reading> |
8e1394aa |
345 | |
4a8828f0 |
346 | my $prior_reading = $graph->prior_reading( $reading, $witpath ); |
8e1394aa |
347 | |
4a8828f0 |
348 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
349 | path. |
8e1394aa |
350 | |
351 | =cut |
352 | |
4a8828f0 |
353 | sub prior_reading { |
e2902068 |
354 | # Return the predecessor via the corresponding path. |
8e1394aa |
355 | my $self = shift; |
4a8828f0 |
356 | return $self->_find_linked_reading( 'prior', @_ ); |
8e1394aa |
357 | } |
358 | |
4a8828f0 |
359 | sub _find_linked_reading { |
e2902068 |
360 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
361 | my @linked_paths = $direction eq 'next' |
8e1394aa |
362 | ? $node->outgoing() : $node->incoming(); |
e2902068 |
363 | return undef unless scalar( @linked_paths ); |
8e1394aa |
364 | |
e2902068 |
365 | # We have to find the linked path that contains all of the |
366 | # witnesses supplied in $path. |
367 | my( @path_wits, @alt_path_wits ); |
368 | @path_wits = $self->witnesses_of_label( $path ) if $path; |
369 | @alt_path_wits = $self->witnesses_of_label( $alt_path ) if $alt_path; |
370 | my $base_le; |
371 | my $alt_le; |
372 | foreach my $le ( @linked_paths ) { |
930ff666 |
373 | if( $le->name eq $self->baselabel ) { |
e2902068 |
374 | $base_le = $le; |
375 | } else { |
376 | my @le_wits = $self->witnesses_of_label( $le->name ); |
377 | if( _is_within( \@path_wits, \@le_wits ) ) { |
378 | # This is the right path. |
379 | return $direction eq 'next' ? $le->to() : $le->from(); |
380 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
381 | $alt_le = $le; |
382 | } |
8e1394aa |
383 | } |
384 | } |
e2902068 |
385 | # Got this far? Return the alternate path if it exists. |
386 | return $direction eq 'next' ? $alt_le->to() : $alt_le->from() |
387 | if $alt_le; |
388 | |
389 | # Got this far? Return the base path if it exists. |
390 | return $direction eq 'next' ? $base_le->to() : $base_le->from() |
391 | if $base_le; |
392 | |
393 | # Got this far? We have no appropriate path. |
8e1394aa |
394 | warn "Could not find $direction node from " . $node->label |
e2902068 |
395 | . " along path $path"; |
8e1394aa |
396 | return undef; |
397 | } |
398 | |
4a8828f0 |
399 | # Some set logic. |
400 | sub _is_within { |
401 | my( $set1, $set2 ) = @_; |
7854e12e |
402 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
403 | foreach my $el ( @$set1 ) { |
404 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
405 | } |
406 | return $ret; |
407 | } |
408 | |
de51424a |
409 | |
410 | ## INITIALIZATION METHODS - for use by parsers |
4a8828f0 |
411 | # Walk the paths for each witness in the graph, and return the nodes |
e2902068 |
412 | # that the graph has in common. If $using_base is true, some |
413 | # different logic is needed. |
4a8828f0 |
414 | |
415 | sub walk_witness_paths { |
416 | my( $self, $end ) = @_; |
417 | # For each witness, walk the path through the graph. |
418 | # Then we need to find the common nodes. |
419 | # TODO This method is going to fall down if we have a very gappy |
420 | # text in the collation. |
421 | my $paths = {}; |
3a1f2523 |
422 | my @common_readings; |
4a8828f0 |
423 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
424 | my $curr_reading = $self->start; |
e2902068 |
425 | my @wit_path = $self->reading_sequence( $self->start, $end, |
426 | $wit->sigil ); |
4a8828f0 |
427 | $wit->path( \@wit_path ); |
e2902068 |
428 | |
429 | # Detect the common readings. |
930ff666 |
430 | @common_readings = _find_common( \@common_readings, \@wit_path ); |
4a8828f0 |
431 | } |
432 | |
433 | # Mark all the nodes as either common or not. |
3a1f2523 |
434 | foreach my $cn ( @common_readings ) { |
e2902068 |
435 | print STDERR "Setting " . $cn->name . " / " . $cn->label |
436 | . " as common node\n"; |
4a8828f0 |
437 | $cn->make_common; |
438 | } |
439 | foreach my $n ( $self->readings() ) { |
440 | $n->make_variant unless $n->is_common; |
441 | } |
3a1f2523 |
442 | # Return an array of the common nodes in order. |
443 | return @common_readings; |
4a8828f0 |
444 | } |
445 | |
930ff666 |
446 | sub _find_common { |
447 | my( $common_readings, $new_path ) = @_; |
448 | my @cr; |
449 | if( @$common_readings ) { |
450 | foreach my $n ( @$new_path ) { |
451 | push( @cr, $n ) if grep { $_ eq $n } @$common_readings; |
452 | } |
453 | } else { |
454 | push( @cr, @$new_path ); |
455 | } |
456 | return @cr; |
457 | } |
458 | |
459 | sub _remove_common { |
460 | my( $common_readings, $divergence ) = @_; |
461 | my @cr; |
462 | my %diverged; |
463 | map { $diverged{$_->name} = 1 } @$divergence; |
464 | foreach( @$common_readings ) { |
465 | push( @cr, $_ ) unless $diverged{$_->name}; |
466 | } |
467 | return @cr; |
468 | } |
469 | |
470 | |
e2902068 |
471 | # An alternative to walk_witness_paths, for use when a collation is |
472 | # constructed from a base text and an apparatus. Also modifies the |
473 | # collation graph to remove all 'base text' paths and replace them |
474 | # with real witness paths. |
475 | |
476 | sub walk_and_expand_base { |
477 | my( $self, $end ) = @_; |
478 | |
930ff666 |
479 | my @common_readings; |
e2902068 |
480 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
7854e12e |
481 | my $sig = $wit->sigil; |
e2902068 |
482 | my $post_sig; |
483 | $post_sig = $wit->post_correctione |
484 | if $wit->has_post_correctione; |
930ff666 |
485 | |
486 | # $DB::single = 1 if $wit->sigil eq 'Vb11'; |
7854e12e |
487 | my @wit_path = $self->reading_sequence( $self->start, $end, $sig ); |
488 | $wit->path( \@wit_path ); |
489 | $self->connect_readings_for_witness( $wit ); |
930ff666 |
490 | @common_readings = _find_common( \@common_readings, \@wit_path ); |
7854e12e |
491 | |
492 | # If there is a post-correctio, get its path and compare. |
493 | # Add a correction range for each divergence. |
494 | if( $post_sig ) { |
7854e12e |
495 | my @corr_wit_path = $self->reading_sequence( $self->start, $end, |
496 | "$sig$post_sig", $sig ); |
497 | |
498 | # Map ante-corr readings to their indices |
499 | my %in_orig; |
500 | my $i = 0; |
501 | map { $in_orig{$_->name} = $i++ } @wit_path; |
502 | |
503 | # Look for divergences |
504 | my $diverged = 0; |
505 | my $last_common; |
506 | my @correction; |
930ff666 |
507 | $DB::single = 1 if $sig eq 'Vb12'; |
7854e12e |
508 | foreach my $rdg ( @corr_wit_path ) { |
509 | if( exists( $in_orig{$rdg->name} ) && !$diverged ) { |
510 | # We are reading the same here |
511 | $last_common = $in_orig{$rdg->name}; |
7854e12e |
512 | } elsif ( exists( $in_orig{$rdg->name} ) ) { |
513 | # We have been diverging but are reading the same again. |
514 | # Add the correction to the witness. |
930ff666 |
515 | my $offset = $last_common + 1; |
516 | my $length = $in_orig{$rdg->name} - $offset; |
517 | $wit->add_correction( $offset, $length, @correction ); |
7854e12e |
518 | $diverged = 0; |
930ff666 |
519 | @common_readings = _remove_common( \@common_readings, \@correction ); |
7854e12e |
520 | @correction = (); |
521 | $last_common = $in_orig{$rdg->name}; |
522 | } elsif( $diverged ) { |
523 | # We are in the middle of a divergence. |
524 | push( @correction, $rdg ); |
525 | } else { |
526 | # We have started to diverge. Note it. |
527 | $diverged = 1; |
528 | push( @correction, $rdg ); |
529 | } |
e2902068 |
530 | } |
930ff666 |
531 | # Add any divergence that is at the end of the text |
7854e12e |
532 | if( $diverged ) { |
930ff666 |
533 | $wit->add_correction( $last_common+1, $#wit_path, \@correction ); |
e2902068 |
534 | } |
e2902068 |
535 | } |
7854e12e |
536 | } |
e2902068 |
537 | |
7854e12e |
538 | # Remove any 'base text' paths. |
539 | foreach my $path ( $self->paths ) { |
540 | $self->del_path( $path ) |
541 | if $path->label eq $self->baselabel; |
542 | } |
543 | } |
544 | |
545 | sub connect_readings_for_witness { |
546 | my( $self, $wit ) = @_; |
547 | my @chain = @{$wit->path}; |
548 | foreach my $idx ( 0 .. $#chain-1 ) { |
549 | $self->add_path( $chain[$idx], $chain[$idx+1], $wit->sigil ); |
550 | } |
e2902068 |
551 | } |
552 | |
4a8828f0 |
553 | sub common_readings { |
554 | my $self = shift; |
555 | my @common = grep { $_->is_common } $self->readings(); |
de51424a |
556 | return sort { _cmp_position( $a->position, $b->position ) } @common; |
4a8828f0 |
557 | } |
558 | |
559 | # Calculate the relative positions of nodes in the graph, if they |
560 | # were not given to us. |
561 | sub calculate_positions { |
3a1f2523 |
562 | my( $self, @ordered_common ) = @_; |
4a8828f0 |
563 | |
564 | # We have to calculate the position identifiers for each word, |
565 | # keyed on the common nodes. This will be 'fun'. The end result |
566 | # is a hash per witness, whose key is the word node and whose |
567 | # value is its position in the text. Common nodes are always N,1 |
568 | # so have identical positions in each text. |
4a8828f0 |
569 | |
570 | my $node_pos = {}; |
571 | foreach my $wit ( @{$self->tradition->witnesses} ) { |
930ff666 |
572 | print STDERR "Calculating positions in " . $wit->sigil . "\n"; |
573 | _update_positions_from_path( $wit->path, @ordered_common ); |
574 | _update_positions_from_path( $wit->corrected_path, @ordered_common ) |
575 | if $wit->has_post_correctione; |
576 | } |
577 | |
578 | # DEBUG |
579 | foreach my $r ( $self->readings() ) { |
580 | print STDERR "Reading " . $r->name . "/" . $r->label . " has no position\n" |
581 | unless( $r->has_position ); |
4a8828f0 |
582 | } |
3a1f2523 |
583 | |
584 | $self->init_lemmata(); |
4a8828f0 |
585 | } |
586 | |
930ff666 |
587 | sub _update_positions_from_path { |
588 | my( $path, @ordered_common ) = @_; |
589 | |
590 | # First we walk the given path, making a matrix for the witness |
591 | # that corresponds to its eventual position identifier. Common |
592 | # nodes always start a new row, and are thus always in the first |
593 | # column. |
594 | |
595 | my $wit_matrix = []; |
596 | my $cn = 0; # We should hit the common readings in order. |
597 | my $row = []; |
598 | foreach my $wn ( @{$path} ) { |
599 | if( $wn eq $ordered_common[$cn] ) { |
600 | # Set up to look for the next common node, and |
601 | # start a new row of words. |
602 | $cn++; |
603 | push( @$wit_matrix, $row ) if scalar( @$row ); |
604 | $row = []; |
605 | } |
606 | push( @$row, $wn ); |
607 | } |
608 | push( @$wit_matrix, $row ); # Push the last row onto the matrix |
609 | |
610 | # Now we have a matrix per witness, so that each row in the |
611 | # matrix begins with a common node, and continues with all the |
612 | # variant words that appear in the witness. We turn this into |
613 | # real positions in row,cell format. But we need some |
614 | # trickery in order to make sure that each node gets assigned |
615 | # to only one position. |
616 | |
617 | foreach my $li ( 1..scalar(@$wit_matrix) ) { |
618 | foreach my $di ( 1..scalar(@{$wit_matrix->[$li-1]}) ) { |
619 | my $reading = $wit_matrix->[$li-1]->[$di-1]; |
620 | my $position = "$li,$di"; |
621 | # If we have seen this node before, we need to compare |
622 | # its position with what went before. |
623 | unless( $reading->has_position && |
624 | _cmp_position( $position, $reading->position ) < 1 ) { |
625 | # The new position ID replaces the old one. |
626 | $reading->position( $position ); |
627 | } # otherwise, the old position needs to stay. |
628 | } |
629 | } |
630 | } |
631 | |
4a8828f0 |
632 | sub _cmp_position { |
633 | my( $a, $b ) = @_; |
de51424a |
634 | if ( $a && $b ) { |
635 | my @pos_a = split(/,/, $a ); |
636 | my @pos_b = split(/,/, $b ); |
637 | |
638 | my $big_cmp = $pos_a[0] <=> $pos_b[0]; |
639 | return $big_cmp if $big_cmp; |
640 | # else |
641 | return $pos_a[1] <=> $pos_b[1]; |
642 | } elsif ( $b ) { # a is undefined |
643 | return -1; |
644 | } elsif ( $a ) { # b is undefined |
645 | return 1; |
646 | } |
647 | return 0; # they are both undefined |
8e1394aa |
648 | } |
3a1f2523 |
649 | |
650 | sub all_positions { |
651 | my $self = shift; |
652 | my %positions = (); |
653 | map { $positions{$_->position} = 1 } $self->readings; |
de51424a |
654 | my @answer = sort { _cmp_position( $a, $b ) } keys( %positions ); |
655 | return @answer; |
3a1f2523 |
656 | } |
657 | |
658 | sub readings_at_position { |
659 | my( $self, $pos ) = @_; |
660 | my @answer = grep { $_->position eq $pos } $self->readings; |
661 | return @answer; |
662 | } |
663 | |
664 | ## Lemmatizer functions |
665 | |
666 | sub init_lemmata { |
667 | my $self = shift; |
668 | |
669 | foreach my $position ( $self->all_positions ) { |
670 | $self->lemmata->{$position} = undef; |
671 | } |
672 | |
673 | foreach my $cr ( $self->common_readings ) { |
674 | $self->lemmata->{$cr->position} = $cr->name; |
675 | } |
676 | } |
677 | |
678 | =item B<lemma_readings> |
679 | |
680 | my @state = $graph->lemma_readings( @readings_delemmatized ); |
681 | |
682 | Takes a list of readings that have just been delemmatized, and returns |
683 | a set of tuples of the form ['reading', 'state'] that indicates what |
684 | changes need to be made to the graph. |
685 | |
686 | =over |
687 | |
688 | =item * |
689 | |
690 | A state of 1 means 'lemmatize this reading' |
691 | |
692 | =item * |
693 | |
694 | A state of 0 means 'delemmatize this reading' |
695 | |
696 | =item * |
697 | |
698 | A state of undef means 'an ellipsis belongs in the text here because |
699 | no decision has been made / an earlier decision was backed out' |
700 | |
701 | =back |
702 | |
703 | =cut |
704 | |
705 | sub lemma_readings { |
706 | my( $self, @toggled_off_nodes ) = @_; |
707 | |
708 | # First get the positions of those nodes which have been |
709 | # toggled off. |
710 | my $positions_off = {}; |
711 | map { $positions_off->{ $_->position } = $_->name } @toggled_off_nodes; |
de51424a |
712 | |
3a1f2523 |
713 | # Now for each position, we have to see if a node is on, and we |
714 | # have to see if a node has been turned off. |
715 | my @answer; |
716 | foreach my $pos ( $self->all_positions() ) { |
717 | # Find the state of this position. If there is an active node, |
718 | # its name will be the state; otherwise the state will be 0 |
719 | # (nothing at this position) or undef (ellipsis at this position) |
720 | my $active = $self->lemmata->{$pos}; |
721 | |
722 | # Is there a formerly active node that was toggled off? |
723 | if( exists( $positions_off->{$pos} ) ) { |
724 | my $off_node = $positions_off->{$pos}; |
725 | if( $active && $active ne $off_node) { |
726 | push( @answer, [ $off_node, 0 ], [ $active, 1 ] ); |
727 | } else { |
728 | push( @answer, [ $off_node, $active ] ); |
729 | } |
730 | |
731 | # No formerly active node, so we just see if there is a currently |
732 | # active one. |
733 | } elsif( $active ) { |
734 | # Push the active node, whatever it is. |
735 | push( @answer, [ $active, 1 ] ); |
736 | } else { |
737 | # Push the state that is there. Arbitrarily use the first node |
738 | # at that position. |
739 | my @pos_nodes = $self->readings_at_position( $pos ); |
de51424a |
740 | push( @answer, [ $pos_nodes[0]->name, $self->lemmata->{$pos} ] ); |
3a1f2523 |
741 | } |
742 | } |
743 | |
744 | return @answer; |
745 | } |
746 | |
de51424a |
747 | =item B<toggle_reading> |
748 | |
749 | my @readings_delemmatized = $graph->toggle_reading( $reading_name ); |
750 | |
751 | Takes a reading node name, and either lemmatizes or de-lemmatizes |
752 | it. Returns a list of all readings that are de-lemmatized as a result |
753 | of the toggle. |
754 | |
755 | =cut |
756 | |
757 | sub toggle_reading { |
758 | my( $self, $rname ) = @_; |
759 | |
760 | return unless $rname; |
761 | my $reading = $self->reading( $rname ); |
762 | if( !$reading || $reading->is_common() ) { |
763 | # Do nothing, it's a common node. |
764 | return; |
765 | } |
766 | |
767 | my $pos = $reading->position; |
768 | my $old_state = $self->lemmata->{$pos}; |
769 | my @readings_off; |
770 | if( $old_state && $old_state eq $rname ) { |
771 | # Turn off the node. We turn on no others by default. |
772 | push( @readings_off, $reading ); |
773 | } else { |
774 | # Turn on the node. |
775 | $self->lemmata->{$pos} = $rname; |
776 | # Any other 'on' readings in the same position should be off. |
777 | push( @readings_off, $self->same_position_as( $reading ) ); |
778 | # Any node that is an identical transposed one should be off. |
779 | push( @readings_off, $reading->identical_readings ); |
780 | } |
781 | @readings_off = unique_list( @readings_off ); |
782 | |
783 | # Turn off the readings that need to be turned off. |
784 | my @readings_delemmatized; |
785 | foreach my $n ( @readings_off ) { |
786 | my $state = $self->lemmata->{$n->position}; |
787 | if( $state && $state eq $n->name ) { |
788 | # this reading is still on, so turn it off |
789 | push( @readings_delemmatized, $n ); |
790 | my $new_state = undef; |
791 | if( $n eq $reading ) { |
792 | # This is the reading that was clicked, so if there are no |
793 | # other readings there, turn off the position. In all other |
794 | # cases, restore the ellipsis. |
795 | my @other_n = $self->same_position_as( $n ); |
796 | $new_state = 0 unless @other_n; |
797 | } |
798 | $self->lemmata->{$n->position} = $new_state; |
799 | } elsif( $old_state && $old_state eq $n->name ) { |
800 | # another reading has already been turned on here |
801 | push( @readings_delemmatized, $n ); |
802 | } # else some other reading was on anyway, so pass. |
803 | } |
804 | return @readings_delemmatized; |
805 | } |
806 | |
807 | sub same_position_as { |
808 | my( $self, $reading ) = @_; |
809 | my $pos = $reading->position; |
810 | my @same = grep { $_ ne $reading } $self->readings_at_position( $reading->position ); |
811 | return @same; |
812 | } |
3a1f2523 |
813 | |
4a8828f0 |
814 | # Return the string that joins together a list of witnesses for |
815 | # display on a single path. |
816 | sub path_label { |
817 | my $self = shift; |
818 | return join( $self->wit_list_separator, @_ ); |
819 | } |
820 | |
821 | sub witnesses_of_label { |
de51424a |
822 | my( $self, $label ) = @_; |
4a8828f0 |
823 | my $regex = $self->wit_list_separator; |
de51424a |
824 | my @answer = split( /\Q$regex\E/, $label ); |
825 | return @answer; |
4a8828f0 |
826 | } |
8e1394aa |
827 | |
de51424a |
828 | sub unique_list { |
829 | my( @list ) = @_; |
830 | my %h; |
831 | map { $h{$_->name} = $_ } @list; |
832 | return values( %h ); |
833 | } |
834 | |
dd3b58b0 |
835 | no Moose; |
836 | __PACKAGE__->meta->make_immutable; |