Commit | Line | Data |
dd3b58b0 |
1 | package Text::Tradition::Collation; |
d047cd52 |
2 | |
910a0a6d |
3 | use Encode qw( decode_utf8 ); |
4 | use File::Temp; |
c9bf3dbf |
5 | use Graph; |
8e1394aa |
6 | use IPC::Run qw( run binary ); |
910a0a6d |
7 | use Text::CSV_XS; |
b15511bf |
8 | use Text::Tradition::Collation::Reading; |
df6d9812 |
9 | use XML::LibXML; |
dd3b58b0 |
10 | use Moose; |
11 | |
3a2ebbf4 |
12 | has 'sequence' => ( |
d047cd52 |
13 | is => 'ro', |
3a2ebbf4 |
14 | isa => 'Graph', |
15 | default => sub { Graph->new() }, |
d047cd52 |
16 | handles => { |
3a2ebbf4 |
17 | paths => 'edges', |
d047cd52 |
18 | }, |
d047cd52 |
19 | ); |
3a2ebbf4 |
20 | |
21 | has 'relations' => ( |
22 | is => 'ro', |
23 | isa => 'Graph', |
24 | default => sub { Graph->new( undirected => 1 ) }, |
25 | handles => { |
26 | relationships => 'edges', |
27 | }, |
28 | ); |
dd3b58b0 |
29 | |
3a2ebbf4 |
30 | has 'tradition' => ( |
31 | is => 'ro', |
d047cd52 |
32 | isa => 'Text::Tradition', |
8d9a1cd8 |
33 | weak_ref => 1, |
d047cd52 |
34 | ); |
dd3b58b0 |
35 | |
3a2ebbf4 |
36 | has 'readings' => ( |
37 | isa => 'HashRef[Text::Tradition::Collation::Reading]', |
38 | traits => ['Hash'], |
39 | handles => { |
40 | reading => 'get', |
41 | _add_reading => 'set', |
42 | del_reading => 'delete', |
43 | has_reading => 'exists', |
44 | readings => 'values', |
45 | }, |
46 | default => sub { {} }, |
47 | ); |
910a0a6d |
48 | |
4a8828f0 |
49 | has 'wit_list_separator' => ( |
7854e12e |
50 | is => 'rw', |
51 | isa => 'Str', |
52 | default => ', ', |
53 | ); |
54 | |
55 | has 'baselabel' => ( |
56 | is => 'rw', |
57 | isa => 'Str', |
58 | default => 'base text', |
59 | ); |
4a8828f0 |
60 | |
15d2d3df |
61 | has 'linear' => ( |
62 | is => 'rw', |
63 | isa => 'Bool', |
64 | default => 1, |
65 | ); |
1f563ac3 |
66 | |
ef9d481f |
67 | has 'ac_label' => ( |
68 | is => 'rw', |
69 | isa => 'Str', |
70 | default => ' (a.c.)', |
71 | ); |
3a2ebbf4 |
72 | |
73 | has 'start' => ( |
74 | is => 'ro', |
75 | isa => 'Text::Tradition::Collation::Reading', |
76 | writer => '_set_start', |
77 | weak_ref => 1, |
78 | ); |
79 | |
80 | has 'end' => ( |
81 | is => 'ro', |
82 | isa => 'Text::Tradition::Collation::Reading', |
83 | writer => '_set_end', |
84 | weak_ref => 1, |
85 | ); |
1f563ac3 |
86 | |
dd3b58b0 |
87 | # The collation can be created two ways: |
88 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
89 | # the results as in 2. |
90 | # 2. Read a pre-prepared collation in one of a variety of formats, |
91 | # and make the graph from that. |
92 | |
93 | # The graph itself will (for now) be immutable, and the positions |
94 | # within the graph will also be immutable. We need to calculate those |
95 | # positions upon graph construction. The equivalences between graph |
96 | # nodes will be mutable, entirely determined by the user (or possibly |
97 | # by some semantic pre-processing provided by the user.) So the |
98 | # constructor should just make an empty equivalences object. The |
99 | # constructor will also need to make the witness objects, if we didn't |
100 | # come through option 1. |
101 | |
d047cd52 |
102 | sub BUILD { |
3a2ebbf4 |
103 | my $self = shift; |
104 | $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) ); |
105 | $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) ); |
d047cd52 |
106 | } |
784877d9 |
107 | |
3a2ebbf4 |
108 | ### Reading construct/destruct functions |
109 | |
110 | sub add_reading { |
111 | my( $self, $reading ) = @_; |
112 | unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) { |
113 | my %args = %$reading; |
114 | $reading = Text::Tradition::Collation::Reading->new( |
115 | 'collation' => $self, |
116 | %args ); |
117 | } |
118 | # First check to see if a reading with this ID exists. |
119 | if( $self->reading( $reading->id ) ) { |
120 | warn "Collation already has a reading with id " . $reading->id; |
121 | return undef; |
122 | } |
123 | $self->_add_reading( $reading->id => $reading ); |
124 | # Once the reading has been added, put it in both graphs. |
125 | $self->sequence->add_vertex( $reading->id ); |
126 | $self->relations->add_vertex( $reading->id ); |
127 | return $reading; |
eca16057 |
128 | }; |
129 | |
3a2ebbf4 |
130 | around del_reading => sub { |
131 | my $orig = shift; |
132 | my $self = shift; |
133 | my $arg = shift; |
134 | |
135 | if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) { |
136 | $arg = $arg->id; |
137 | } |
138 | |
139 | # Remove the reading from the graphs. |
140 | $self->sequence->delete_vertex( $arg ); |
141 | $self->relations->delete_vertex( $arg ); |
142 | |
143 | # Carry on. |
144 | $self->$orig( $arg ); |
145 | }; |
7854e12e |
146 | |
3a2ebbf4 |
147 | # merge_readings( $main, $to_be_deleted ); |
7854e12e |
148 | |
3a2ebbf4 |
149 | sub merge_readings { |
150 | my $self = shift; |
151 | |
152 | # We only need the IDs for adding paths to the graph, not the reading |
153 | # objects themselves. |
154 | my( $kept, $deleted ) = $self->_stringify_args( @_ ); |
155 | |
156 | # The kept reading should inherit the paths and the relationships |
157 | # of the deleted reading. |
158 | foreach my $path ( $self->sequence->edges_at( $deleted ) ) { |
159 | my @vector = ( $kept ); |
160 | push( @vector, $path->[1] ) if $path->[0] eq $deleted; |
161 | unshift( @vector, $path->[0] ) if $path->[1] eq $deleted; |
162 | my %wits = %{$self->sequence->get_edge_attributes( @$path )}; |
163 | $self->sequence->add_edge( @vector ); |
164 | my $fwits = $self->sequence->get_edge_attributes( @vector ); |
165 | @wits{keys %$fwits} = values %$fwits; |
166 | $self->sequence->set_edge_attributes( @vector, \%wits ); |
167 | } |
168 | foreach my $rel ( $self->relations->edges_at( $deleted ) ) { |
169 | my @vector = ( $kept ); |
170 | push( @vector, $rel->[0] eq $deleted ? $rel->[1] : $rel->[0] ); |
171 | # Is there a relationship here already? If so, keep it. |
172 | # TODO Warn about conflicting relationships |
173 | next if $self->relations->has_edge( @vector ); |
174 | # If not, adopt the relationship that would be deleted. |
175 | $self->relations->add_edge( @vector ); |
176 | my $attr = $self->relations->get_edge_attributes( @$rel ); |
177 | $self->relations->set_edge_attributes( @vector, $attr ); |
178 | } |
179 | |
180 | # Do the deletion deed. |
181 | $self->del_reading( $deleted ); |
182 | } |
7854e12e |
183 | |
3265b0ce |
184 | |
3a2ebbf4 |
185 | # Helper function for manipulating the graph. |
186 | sub _stringify_args { |
187 | my( $self, $first, $second, $arg ) = @_; |
188 | $first = $first->id |
189 | if ref( $first ) eq 'Text::Tradition::Collation::Reading'; |
190 | $second = $second->id |
191 | if ref( $second ) eq 'Text::Tradition::Collation::Reading'; |
192 | return( $first, $second, $arg ); |
193 | } |
df6d9812 |
194 | |
3a2ebbf4 |
195 | ### Path logic |
196 | |
197 | sub add_path { |
198 | my $self = shift; |
199 | |
200 | # We only need the IDs for adding paths to the graph, not the reading |
201 | # objects themselves. |
202 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
203 | |
204 | # Connect the readings |
205 | $self->sequence->add_edge( $source, $target ); |
206 | # Note the witness in question |
207 | $self->sequence->set_edge_attribute( $source, $target, $wit, 1 ); |
b15511bf |
208 | }; |
209 | |
3a2ebbf4 |
210 | sub del_path { |
211 | my $self = shift; |
212 | |
213 | # We only need the IDs for adding paths to the graph, not the reading |
214 | # objects themselves. |
215 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
216 | |
217 | if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) { |
218 | $self->sequence->del_edge_attribute( $source, $target, $wit ); |
219 | } |
220 | unless( keys %{$self->sequence->get_edge_attributes( $source, $target )} ) { |
221 | $self->sequence->delete_edge( $source, $target ); |
222 | } |
784877d9 |
223 | } |
224 | |
3a2ebbf4 |
225 | |
15d2d3df |
226 | # Extra graph-alike utility |
227 | sub has_path { |
3a2ebbf4 |
228 | my $self = shift; |
229 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
230 | return undef unless $self->sequence->has_edge( $source, $target ); |
231 | return $self->sequence->has_edge_attribute( $source, $target, $wit ); |
b15511bf |
232 | } |
233 | |
3a2ebbf4 |
234 | ### Relationship logic |
3265b0ce |
235 | |
3a2ebbf4 |
236 | =head2 add_relationship( $reading1, $reading2, $definition ) |
237 | |
238 | Adds the specified relationship between the two readings. A relationship |
239 | is transitive (i.e. undirected), and must have the following attributes |
240 | specified in the hashref $definition: |
241 | |
242 | =over 4 |
243 | |
244 | =item * type - Can be one of spelling, orthographic, grammatical, meaning, repetition, transposition. The first three are only valid relationships between readings that occur at the same point in the text. |
245 | |
246 | =item * non_correctable - (Optional) True if the reading would not have been corrected independently. |
ef9d481f |
247 | |
3a2ebbf4 |
248 | =item * non_independent - (Optional) True if the variant is unlikely to have occurred independently in unrelated witnesses. |
249 | |
250 | =item * global - (Optional) A meta-attribute, to set the same relationship between readings with the same text whenever they occur in the same place. |
251 | |
252 | =back |
253 | |
254 | =cut |
255 | |
256 | # Wouldn't it be lovely if edges could be objects, and all this type checking |
257 | # and attribute management could be done via Moose? |
258 | |
259 | sub add_relationship { |
260 | my $self = shift; |
261 | my( $source, $target, $options ) = $self->_stringify_args( @_ ); |
262 | |
263 | # Check the options |
264 | if( !defined $options->{'type'} || |
265 | $options->{'type'} !~ /^(spelling|orthographic|grammatical|meaning|repetition|transposition)$/i ) { |
266 | my $t = $options->{'type'} ? $options->{'type'} : ''; |
267 | return( undef, "Invalid or missing type" . $options->{'type'} ); |
268 | } |
269 | if( $options->{'type'} =~ /^(spelling|orthographic|grammatical|meaning)$/ ) { |
270 | $options->{'colocated'} = 1; |
271 | } |
272 | |
ef9d481f |
273 | # Make sure there is not another relationship between these two |
94c00c71 |
274 | # readings already |
3a2ebbf4 |
275 | if( $self->relations->has_edge( $source, $target ) ) { |
276 | return ( undef, "Relationship already exists between these readings" ); |
4cdd82f1 |
277 | } |
1d310495 |
278 | if( !$self->relationship_valid( $source, $target, $options->{'type'} ) ) { |
910a0a6d |
279 | return ( undef, 'Relationship creates witness loop' ); |
ef9d481f |
280 | } |
281 | |
3a2ebbf4 |
282 | my @vector = ( $source, $target ); |
283 | $self->relations->add_edge( @vector ); |
284 | $self->relations->set_edge_attributes( @vector, $options ); |
910a0a6d |
285 | |
286 | # TODO Handle global relationship setting |
287 | |
3a2ebbf4 |
288 | return( 1, @vector ); |
3265b0ce |
289 | } |
290 | |
910a0a6d |
291 | sub relationship_valid { |
1d310495 |
292 | my( $self, $source, $target, $rel ) = @_; |
293 | if( $rel eq 'repetition' ) { |
294 | return 1; |
295 | } elsif ( $rel eq 'transposition' ) { |
296 | # Check that the two readings do not appear in the same witness. |
297 | my %seen_wits; |
298 | map { $seen_wits{$_} = 1 } $self->reading_witnesses( $source ); |
299 | foreach my $w ( $self->reading_witnesses( $target ) ) { |
300 | return 0 if $seen_wits{$w}; |
301 | } |
302 | return 1; |
303 | } else { |
304 | # Check that linking the source and target in a relationship won't lead |
305 | # to a path loop for any witness. First make a lookup table of all the |
306 | # readings related to either the source or the target. |
307 | my @proposed_related = ( $source, $target ); |
308 | push( @proposed_related, $source->related_readings( 'colocated' ) ); |
309 | push( @proposed_related, $target->related_readings( 'colocated' ) ); |
310 | my %pr_ids; |
311 | map { $pr_ids{ $_->id } = 1 } @proposed_related; |
312 | |
313 | # None of these proposed related readings should have a neighbor that |
314 | # is also in proposed_related. |
315 | foreach my $pr ( keys %pr_ids ) { |
316 | foreach my $neighbor( $self->sequence->neighbors( $pr ) ) { |
317 | return 0 if exists $pr_ids{$neighbor}; |
318 | } |
319 | } |
320 | return 1; |
321 | } |
322 | } |
323 | |
324 | # Return a list of the witnesses in which the reading appears. |
325 | sub reading_witnesses { |
326 | my( $self, $reading ) = @_; |
327 | # We need only check either the incoming or the outgoing edges; I have |
328 | # arbitrarily chosen "incoming". |
329 | my %all_witnesses; |
330 | foreach my $e ( $self->sequence->edges_to( $reading ) ) { |
331 | my $wits = $self->sequence->get_edge_attributes( @$e ); |
332 | @all_witnesses{ keys %$wits } = 1; |
333 | } |
334 | return keys %all_witnesses; |
910a0a6d |
335 | } |
336 | |
3a2ebbf4 |
337 | sub related_readings { |
338 | my( $self, $reading, $colocated ) = @_; |
a753cc84 |
339 | my $return_object; |
340 | if( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) { |
341 | $reading = $reading->id; |
342 | $return_object = 1; |
7035e3a6 |
343 | # print STDERR "Returning related objects\n"; |
344 | # } else { |
345 | # print STDERR "Returning related object names\n"; |
a753cc84 |
346 | } |
3a2ebbf4 |
347 | my @related = $self->relations->all_reachable( $reading ); |
348 | if( $colocated ) { |
349 | my @colo = grep { $self->relations->has_edge_attribute( $reading, $_, 'colocated' ) } @related; |
a753cc84 |
350 | @related = @colo; |
351 | } |
352 | return $return_object ? map { $self->reading( $_ ) } @related : @related; |
3a2ebbf4 |
353 | } |
354 | |
8e1394aa |
355 | =head2 Output method(s) |
356 | |
357 | =over |
358 | |
359 | =item B<as_svg> |
360 | |
361 | print $graph->as_svg( $recalculate ); |
362 | |
363 | Returns an SVG string that represents the graph. Uses GraphViz to do |
4a8828f0 |
364 | this, because Graph::Easy doesn\'t cope well with long graphs. Unless |
8e1394aa |
365 | $recalculate is passed (and is a true value), the method will return a |
366 | cached copy of the SVG after the first call to the method. |
367 | |
368 | =cut |
369 | |
370 | sub as_svg { |
3a2ebbf4 |
371 | my( $self ) = @_; |
372 | |
8e1394aa |
373 | my @cmd = qw/dot -Tsvg/; |
374 | my( $svg, $err ); |
910a0a6d |
375 | my $dotfile = File::Temp->new(); |
d9e873d0 |
376 | ## TODO REMOVE |
eca16057 |
377 | # $dotfile->unlink_on_destroy(0); |
910a0a6d |
378 | binmode $dotfile, ':utf8'; |
379 | print $dotfile $self->as_dot(); |
380 | push( @cmd, $dotfile->filename ); |
381 | run( \@cmd, ">", binary(), \$svg ); |
382 | $svg = decode_utf8( $svg ); |
8e1394aa |
383 | return $svg; |
384 | } |
385 | |
df6d9812 |
386 | =item B<as_dot> |
387 | |
388 | print $graph->as_dot( $view, $recalculate ); |
389 | |
390 | Returns a string that is the collation graph expressed in dot |
391 | (i.e. GraphViz) format. The 'view' argument determines what kind of |
392 | graph is produced. |
393 | * 'path': a graph of witness paths through the collation (DEFAULT) |
394 | * 'relationship': a graph of how collation readings relate to |
395 | each other |
396 | |
397 | =cut |
398 | |
399 | sub as_dot { |
400 | my( $self, $view ) = @_; |
3a2ebbf4 |
401 | $view = 'sequence' unless $view; |
df6d9812 |
402 | # TODO consider making some of these things configurable |
67da8d6c |
403 | my $graph_name = $self->tradition->name; |
404 | $graph_name =~ s/[^\w\s]//g; |
405 | $graph_name = join( '_', split( /\s+/, $graph_name ) ); |
406 | my $dot = sprintf( "digraph %s {\n", $graph_name ); |
df6d9812 |
407 | $dot .= "\tedge [ arrowhead=open ];\n"; |
408 | $dot .= "\tgraph [ rankdir=LR ];\n"; |
409 | $dot .= sprintf( "\tnode [ fontsize=%d, fillcolor=%s, style=%s, shape=%s ];\n", |
3a2ebbf4 |
410 | 11, "white", "filled", "ellipse" ); |
df6d9812 |
411 | |
412 | foreach my $reading ( $self->readings ) { |
910a0a6d |
413 | # Need not output nodes without separate labels |
3a2ebbf4 |
414 | next if $reading->id eq $reading->text; |
415 | $dot .= sprintf( "\t\"%s\" [ label=\"%s\" ];\n", $reading->id, $reading->text ); |
df6d9812 |
416 | } |
3a2ebbf4 |
417 | |
418 | # TODO do something sensible for relationships |
df6d9812 |
419 | |
3a2ebbf4 |
420 | my @edges = $self->paths; |
df6d9812 |
421 | foreach my $edge ( @edges ) { |
910a0a6d |
422 | my %variables = ( 'color' => '#000000', |
423 | 'fontcolor' => '#000000', |
3a2ebbf4 |
424 | 'label' => join( ', ', $self->path_witnesses( $edge ) ), |
910a0a6d |
425 | ); |
426 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
427 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", |
3a2ebbf4 |
428 | $edge->[0], $edge->[1], $varopts ); |
df6d9812 |
429 | } |
df6d9812 |
430 | $dot .= "}\n"; |
431 | return $dot; |
432 | } |
433 | |
3a2ebbf4 |
434 | sub path_witnesses { |
435 | my( $self, @edge ) = @_; |
436 | # If edge is an arrayref, cope. |
437 | if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) { |
438 | my $e = shift @edge; |
439 | @edge = @$e; |
440 | } |
441 | my @wits = keys %{$self->sequence->get_edge_attributes( @edge )}; |
442 | return sort @wits; |
443 | } |
444 | |
8e1394aa |
445 | =item B<as_graphml> |
446 | |
447 | print $graph->as_graphml( $recalculate ) |
448 | |
449 | Returns a GraphML representation of the collation graph, with |
450 | transposition information and position information. Unless |
451 | $recalculate is passed (and is a true value), the method will return a |
452 | cached copy of the SVG after the first call to the method. |
453 | |
454 | =cut |
455 | |
456 | sub as_graphml { |
3a2ebbf4 |
457 | my( $self ) = @_; |
8e1394aa |
458 | |
459 | # Some namespaces |
460 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
461 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
462 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
910a0a6d |
463 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
8e1394aa |
464 | |
465 | # Create the document and root node |
466 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
467 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
468 | $graphml->setDocumentElement( $root ); |
469 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
470 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
471 | |
e309421a |
472 | # Add the data keys for the graph |
473 | my %graph_data_keys; |
474 | my $gdi = 0; |
1d310495 |
475 | my @graph_attributes = qw/ version wit_list_separator baselabel linear ac_label /; |
e309421a |
476 | foreach my $datum ( @graph_attributes ) { |
477 | $graph_data_keys{$datum} = 'dg'.$gdi++; |
478 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
479 | $key->setAttribute( 'attr.name', $datum ); |
480 | $key->setAttribute( 'attr.type', $key eq 'linear' ? 'boolean' : 'string' ); |
481 | $key->setAttribute( 'for', 'graph' ); |
482 | $key->setAttribute( 'id', $graph_data_keys{$datum} ); |
483 | } |
f6066bac |
484 | |
8e1394aa |
485 | # Add the data keys for nodes |
ef9d481f |
486 | my %node_data_keys; |
487 | my $ndi = 0; |
3a2ebbf4 |
488 | my %node_data = ( |
489 | id => 'string', |
490 | reading => 'string', |
491 | rank => 'string', |
492 | is_start => 'boolean', |
493 | is_end => 'boolean', |
494 | is_lacuna => 'boolean', |
495 | ); |
496 | foreach my $datum ( keys %node_data ) { |
910a0a6d |
497 | $node_data_keys{$datum} = 'dn'.$ndi++; |
498 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
499 | $key->setAttribute( 'attr.name', $datum ); |
3a2ebbf4 |
500 | $key->setAttribute( 'attr.type', $node_data{$datum} ); |
910a0a6d |
501 | $key->setAttribute( 'for', 'node' ); |
502 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
503 | } |
504 | |
df6d9812 |
505 | # Add the data keys for edges, i.e. witnesses |
ef9d481f |
506 | my $edi = 0; |
507 | my %edge_data_keys; |
3a2ebbf4 |
508 | my %edge_data = ( |
509 | class => 'string', # Path or relationship? |
510 | witness => 'string', # ID/label for a path |
511 | relationship => 'string', # ID/label for a relationship |
512 | extra => 'boolean', # Path key |
513 | colocated => 'boolean', # Relationship key |
514 | non_correctable => 'boolean', # Relationship key |
515 | non_independent => 'boolean', # Relationship key |
516 | ); |
517 | foreach my $datum ( keys %edge_data ) { |
518 | $edge_data_keys{$datum} = 'de'.$edi++; |
910a0a6d |
519 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
3a2ebbf4 |
520 | $key->setAttribute( 'attr.name', $datum ); |
521 | $key->setAttribute( 'attr.type', $edge_data{$datum} ); |
910a0a6d |
522 | $key->setAttribute( 'for', 'edge' ); |
3a2ebbf4 |
523 | $key->setAttribute( 'id', $edge_data_keys{$datum} ); |
8e1394aa |
524 | } |
3a2ebbf4 |
525 | |
526 | # Add the collation graph itself |
8e1394aa |
527 | my $graph = $root->addNewChild( $graphml_ns, 'graph' ); |
528 | $graph->setAttribute( 'edgedefault', 'directed' ); |
94c00c71 |
529 | $graph->setAttribute( 'id', $self->tradition->name ); |
8e1394aa |
530 | $graph->setAttribute( 'parse.edgeids', 'canonical' ); |
df6d9812 |
531 | $graph->setAttribute( 'parse.edges', scalar($self->paths) ); |
8e1394aa |
532 | $graph->setAttribute( 'parse.nodeids', 'canonical' ); |
df6d9812 |
533 | $graph->setAttribute( 'parse.nodes', scalar($self->readings) ); |
8e1394aa |
534 | $graph->setAttribute( 'parse.order', 'nodesfirst' ); |
e309421a |
535 | |
536 | # Collation attribute data |
537 | foreach my $datum ( @graph_attributes ) { |
1d310495 |
538 | my $value = $datum eq 'version' ? '2.0' : $self->$datum; |
539 | _add_graphml_data( $graph, $graph_data_keys{$datum}, $value ); |
e309421a |
540 | } |
8e1394aa |
541 | |
542 | my $node_ctr = 0; |
543 | my %node_hash; |
b15511bf |
544 | # Add our readings to the graph |
3a2ebbf4 |
545 | foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) { |
910a0a6d |
546 | my $node_el = $graph->addNewChild( $graphml_ns, 'node' ); |
547 | my $node_xmlid = 'n' . $node_ctr++; |
3a2ebbf4 |
548 | $node_hash{ $n->id } = $node_xmlid; |
910a0a6d |
549 | $node_el->setAttribute( 'id', $node_xmlid ); |
3a2ebbf4 |
550 | _add_graphml_data( $node_el, $node_data_keys{'id'}, $n->id ); |
551 | _add_graphml_data( $node_el, $node_data_keys{'reading'}, $n->text ); |
d9e873d0 |
552 | _add_graphml_data( $node_el, $node_data_keys{'rank'}, $n->rank ) |
553 | if $n->has_rank; |
b15511bf |
554 | } |
555 | |
3a2ebbf4 |
556 | # Add the path edges |
df6d9812 |
557 | my $edge_ctr = 0; |
3a2ebbf4 |
558 | foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) { |
559 | # We add an edge in the graphml for every witness in $e. |
560 | foreach my $wit ( $self->path_witnesses( $e ) ) { |
561 | my( $id, $from, $to ) = ( 'e'.$edge_ctr++, |
562 | $node_hash{ $e->[0] }, |
563 | $node_hash{ $e->[1] } ); |
564 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
565 | $edge_el->setAttribute( 'source', $from ); |
566 | $edge_el->setAttribute( 'target', $to ); |
567 | $edge_el->setAttribute( 'id', $id ); |
568 | # Add the edge class |
569 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, 'path' ); |
570 | |
571 | # It's a witness path, so add the witness |
572 | my $base = $wit; |
573 | my $key = $edge_data_keys{'witness'}; |
574 | # Is this an ante-corr witness? |
575 | my $aclabel = $self->ac_label; |
576 | if( $wit =~ /^(.*)\Q$aclabel\E$/ ) { |
577 | # Keep the base witness |
578 | $base = $1; |
579 | # ...and record that this is an 'extra' reading path |
580 | _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel ); |
581 | } |
582 | _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base ); |
583 | } |
584 | } |
585 | |
586 | # Add the relationship edges |
587 | foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->relationships ) { |
588 | my( $id, $from, $to ) = ( 'e'.$edge_ctr++, |
589 | $node_hash{ $e->[0] }, |
590 | $node_hash{ $e->[1] } ); |
591 | my $edge_el = $graph->addNewChild( $graphml_ns, 'edge' ); |
592 | $edge_el->setAttribute( 'source', $from ); |
593 | $edge_el->setAttribute( 'target', $to ); |
594 | $edge_el->setAttribute( 'id', $id ); |
595 | # Add the edge class |
596 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, 'relationship' ); |
597 | |
598 | my $data = $self->relations->get_edge_attributes( @$e ); |
599 | # It's a relationship, so save the relationship data |
600 | _add_graphml_data( $edge_el, $edge_data_keys{'relationship'}, $data->{type} ); |
601 | _add_graphml_data( $edge_el, $edge_data_keys{'colocated'}, $data->{colocated} ); |
602 | if( exists $data->{non_correctable} ) { |
603 | _add_graphml_data( $edge_el, $edge_data_keys{'non_correctable'}, |
604 | $data->{non_correctable} ); |
605 | } |
606 | if( exists $data->{non_independent} ) { |
607 | _add_graphml_data( $edge_el, $edge_data_keys{'non_independent'}, |
608 | $data->{non_independent} ); |
609 | } |
8e1394aa |
610 | } |
611 | |
94c00c71 |
612 | # Save and return the thing |
613 | my $result = decode_utf8( $graphml->toString(1) ); |
94c00c71 |
614 | return $result; |
df6d9812 |
615 | } |
616 | |
b15511bf |
617 | sub _add_graphml_data { |
618 | my( $el, $key, $value ) = @_; |
b15511bf |
619 | return unless defined $value; |
c9bf3dbf |
620 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
b15511bf |
621 | $data_el->setAttribute( 'key', $key ); |
622 | $data_el->appendText( $value ); |
8e1394aa |
623 | } |
624 | |
910a0a6d |
625 | =item B<as_csv> |
626 | |
627 | print $graph->as_csv( $recalculate ) |
628 | |
629 | Returns a CSV alignment table representation of the collation graph, one |
630 | row per witness (or witness uncorrected.) Unless $recalculate is passed |
631 | (and is a true value), the method will return a cached copy of the CSV |
632 | after the first call to the method. |
633 | |
634 | =cut |
635 | |
636 | sub as_csv { |
3a2ebbf4 |
637 | my( $self ) = @_; |
910a0a6d |
638 | my $table = $self->make_alignment_table; |
639 | my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } ); |
640 | my @result; |
641 | foreach my $row ( @$table ) { |
642 | $csv->combine( @$row ); |
643 | push( @result, decode_utf8( $csv->string ) ); |
644 | } |
3a2ebbf4 |
645 | return join( "\n", @result ); |
910a0a6d |
646 | } |
647 | |
0e476982 |
648 | # Make an alignment table - $noderefs controls whether the objects |
649 | # in the table are the nodes or simply their readings. |
9f3ba6f7 |
650 | |
910a0a6d |
651 | sub make_alignment_table { |
08e0fb85 |
652 | my( $self, $noderefs, $include ) = @_; |
910a0a6d |
653 | unless( $self->linear ) { |
654 | warn "Need a linear graph in order to make an alignment table"; |
655 | return; |
656 | } |
657 | my $table; |
3a2ebbf4 |
658 | my @all_pos = ( 1 .. $self->end->rank - 1 ); |
910a0a6d |
659 | foreach my $wit ( $self->tradition->witnesses ) { |
eca16057 |
660 | # print STDERR "Making witness row(s) for " . $wit->sigil . "\n"; |
1f7aa795 |
661 | my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil ); |
662 | my @row = _make_witness_row( \@wit_path, \@all_pos, $noderefs ); |
910a0a6d |
663 | unshift( @row, $wit->sigil ); |
664 | push( @$table, \@row ); |
1f7aa795 |
665 | if( $wit->is_layered ) { |
666 | my @wit_ac_path = $self->reading_sequence( $self->start, $self->end, |
667 | $wit->sigil.$self->ac_label, $wit->sigil ); |
668 | my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos, $noderefs ); |
910a0a6d |
669 | unshift( @ac_row, $wit->sigil . $self->ac_label ); |
670 | push( @$table, \@ac_row ); |
671 | } |
672 | } |
0e476982 |
673 | |
08e0fb85 |
674 | if( $include ) { |
675 | my $winnowed = []; |
676 | # Winnow out the rows for any witness not included. |
677 | foreach my $row ( @$table ) { |
678 | next unless $include->{$row->[0]}; |
679 | push( @$winnowed, $row ); |
680 | } |
681 | $table = $winnowed; |
682 | } |
683 | |
910a0a6d |
684 | # Return a table where the witnesses read in columns rather than rows. |
685 | my $turned = _turn_table( $table ); |
08e0fb85 |
686 | # TODO We should really go through and delete empty rows. |
910a0a6d |
687 | return $turned; |
688 | } |
689 | |
690 | sub _make_witness_row { |
0e476982 |
691 | my( $path, $positions, $noderefs ) = @_; |
910a0a6d |
692 | my %char_hash; |
693 | map { $char_hash{$_} = undef } @$positions; |
694 | foreach my $rdg ( @$path ) { |
eca16057 |
695 | my $rtext = $rdg->text; |
696 | $rtext = '#LACUNA#' if $rdg->is_lacuna; |
3a2ebbf4 |
697 | # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank; |
0e476982 |
698 | $char_hash{$rdg->rank} = $noderefs ? $rdg : $rtext; |
910a0a6d |
699 | } |
700 | my @row = map { $char_hash{$_} } @$positions; |
eca16057 |
701 | # Fill in lacuna markers for undef spots in the row |
702 | my $last_el = shift @row; |
703 | my @filled_row = ( $last_el ); |
704 | foreach my $el ( @row ) { |
0e476982 |
705 | # If we are using node reference, make the lacuna node appear many times |
706 | # in the table. If not, use the lacuna tag. |
707 | if( $last_el && _el_is_lacuna( $last_el ) && !defined $el ) { |
708 | $el = $noderefs ? $last_el : '#LACUNA#'; |
eca16057 |
709 | } |
710 | push( @filled_row, $el ); |
711 | $last_el = $el; |
712 | } |
713 | return @filled_row; |
910a0a6d |
714 | } |
715 | |
0e476982 |
716 | # Tiny utility function to say if a table element is a lacuna |
717 | sub _el_is_lacuna { |
718 | my $el = shift; |
719 | return 1 if $el eq '#LACUNA#'; |
720 | return 1 if ref( $el ) eq 'Text::Tradition::Collation::Reading' |
721 | && $el->is_lacuna; |
722 | return 0; |
723 | } |
724 | |
910a0a6d |
725 | # Helper to turn the witnesses along columns rather than rows. Assumes |
726 | # equal-sized rows. |
727 | sub _turn_table { |
728 | my( $table ) = @_; |
729 | my $result = []; |
730 | return $result unless scalar @$table; |
731 | my $nrows = scalar @{$table->[0]}; |
732 | foreach my $idx ( 0 .. $nrows - 1 ) { |
733 | foreach my $wit ( 0 .. $#{$table} ) { |
734 | $result->[$idx]->[$wit] = $table->[$wit]->[$idx]; |
735 | } |
736 | } |
737 | return $result; |
738 | } |
739 | |
8e1394aa |
740 | =back |
741 | |
de51424a |
742 | =head2 Navigation methods |
743 | |
744 | =over |
745 | |
8e1394aa |
746 | =item B<start> |
747 | |
748 | my $beginning = $collation->start(); |
749 | |
750 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
751 | |
910a0a6d |
752 | =item B<end> |
753 | |
754 | my $end = $collation->end(); |
755 | |
756 | Returns the end of the collation, a meta-reading with label '#END#'. |
757 | |
910a0a6d |
758 | |
e2902068 |
759 | =item B<reading_sequence> |
760 | |
761 | my @readings = $graph->reading_sequence( $first, $last, $path[, $alt_path] ); |
762 | |
763 | Returns the ordered list of readings, starting with $first and ending |
764 | with $last, along the given witness path. If no path is specified, |
765 | assume that the path is that of the base text (if any.) |
766 | |
767 | =cut |
768 | |
910a0a6d |
769 | # TODO Think about returning some lazy-eval iterator. |
770 | |
e2902068 |
771 | sub reading_sequence { |
772 | my( $self, $start, $end, $witness, $backup ) = @_; |
773 | |
930ff666 |
774 | $witness = $self->baselabel unless $witness; |
e2902068 |
775 | my @readings = ( $start ); |
776 | my %seen; |
777 | my $n = $start; |
3a2ebbf4 |
778 | while( $n && $n->id ne $end->id ) { |
779 | if( exists( $seen{$n->id} ) ) { |
780 | warn "Detected loop at " . $n->id; |
910a0a6d |
781 | last; |
782 | } |
3a2ebbf4 |
783 | $seen{$n->id} = 1; |
910a0a6d |
784 | |
785 | my $next = $self->next_reading( $n, $witness, $backup ); |
3a2ebbf4 |
786 | $DB::single = 1 if $next->id eq $end->id; |
44771cf2 |
787 | unless( $next ) { |
3a2ebbf4 |
788 | warn "Did not find any path for $witness from reading " . $n->id; |
44771cf2 |
789 | last; |
790 | } |
910a0a6d |
791 | push( @readings, $next ); |
792 | $n = $next; |
e2902068 |
793 | } |
794 | # Check that the last reading is our end reading. |
795 | my $last = $readings[$#readings]; |
3a2ebbf4 |
796 | warn "Last reading found from " . $start->text . |
910a0a6d |
797 | " for witness $witness is not the end!" |
3a2ebbf4 |
798 | unless $last->id eq $end->id; |
e2902068 |
799 | |
800 | return @readings; |
801 | } |
802 | |
4a8828f0 |
803 | =item B<next_reading> |
8e1394aa |
804 | |
4a8828f0 |
805 | my $next_reading = $graph->next_reading( $reading, $witpath ); |
8e1394aa |
806 | |
4a8828f0 |
807 | Returns the reading that follows the given reading along the given witness |
930ff666 |
808 | path. |
8e1394aa |
809 | |
810 | =cut |
811 | |
4a8828f0 |
812 | sub next_reading { |
e2902068 |
813 | # Return the successor via the corresponding path. |
8e1394aa |
814 | my $self = shift; |
3a2ebbf4 |
815 | my $answer = $self->_find_linked_reading( 'next', @_ ); |
816 | return $self->reading( $answer ); |
8e1394aa |
817 | } |
818 | |
4a8828f0 |
819 | =item B<prior_reading> |
8e1394aa |
820 | |
4a8828f0 |
821 | my $prior_reading = $graph->prior_reading( $reading, $witpath ); |
8e1394aa |
822 | |
4a8828f0 |
823 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
824 | path. |
8e1394aa |
825 | |
826 | =cut |
827 | |
4a8828f0 |
828 | sub prior_reading { |
e2902068 |
829 | # Return the predecessor via the corresponding path. |
8e1394aa |
830 | my $self = shift; |
3a2ebbf4 |
831 | my $answer = $self->_find_linked_reading( 'prior', @_ ); |
832 | return $self->reading( $answer ); |
8e1394aa |
833 | } |
834 | |
4a8828f0 |
835 | sub _find_linked_reading { |
e2902068 |
836 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
837 | my @linked_paths = $direction eq 'next' |
3a2ebbf4 |
838 | ? $self->sequence->edges_from( $node ) |
839 | : $self->sequence->edges_to( $node ); |
e2902068 |
840 | return undef unless scalar( @linked_paths ); |
8e1394aa |
841 | |
e2902068 |
842 | # We have to find the linked path that contains all of the |
843 | # witnesses supplied in $path. |
844 | my( @path_wits, @alt_path_wits ); |
3a2ebbf4 |
845 | @path_wits = sort( $self->witnesses_of_label( $path ) ) if $path; |
846 | @alt_path_wits = sort( $self->witnesses_of_label( $alt_path ) ) if $alt_path; |
e2902068 |
847 | my $base_le; |
848 | my $alt_le; |
849 | foreach my $le ( @linked_paths ) { |
3a2ebbf4 |
850 | if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) { |
910a0a6d |
851 | $base_le = $le; |
910a0a6d |
852 | } |
3a2ebbf4 |
853 | my @le_wits = $self->path_witnesses( $le ); |
854 | if( _is_within( \@path_wits, \@le_wits ) ) { |
855 | # This is the right path. |
856 | return $direction eq 'next' ? $le->[1] : $le->[0]; |
857 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
858 | $alt_le = $le; |
859 | } |
8e1394aa |
860 | } |
e2902068 |
861 | # Got this far? Return the alternate path if it exists. |
3a2ebbf4 |
862 | return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0] |
910a0a6d |
863 | if $alt_le; |
e2902068 |
864 | |
865 | # Got this far? Return the base path if it exists. |
3a2ebbf4 |
866 | return $direction eq 'next' ? $base_le->[1] : $base_le->[0] |
910a0a6d |
867 | if $base_le; |
e2902068 |
868 | |
869 | # Got this far? We have no appropriate path. |
8e1394aa |
870 | warn "Could not find $direction node from " . $node->label |
910a0a6d |
871 | . " along path $path"; |
8e1394aa |
872 | return undef; |
873 | } |
874 | |
4a8828f0 |
875 | # Some set logic. |
876 | sub _is_within { |
877 | my( $set1, $set2 ) = @_; |
7854e12e |
878 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
879 | foreach my $el ( @$set1 ) { |
910a0a6d |
880 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
4a8828f0 |
881 | } |
882 | return $ret; |
883 | } |
884 | |
de51424a |
885 | |
886 | ## INITIALIZATION METHODS - for use by parsers |
930ff666 |
887 | |
7e450e44 |
888 | # For use when a collation is constructed from a base text and an apparatus. |
889 | # We have the sequences of readings and just need to add path edges. |
1f7aa795 |
890 | # When we are done, clear out the witness path attributes, as they are no |
891 | # longer needed. |
892 | # TODO Find a way to replace the witness path attributes with encapsulated functions? |
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 | } |
1f7aa795 |
909 | if( $wit->is_layered ) { |
d9e873d0 |
910 | @chain = @{$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 | } |
1f7aa795 |
918 | $wit->clear_path; |
919 | $wit->clear_uncorrected_path; |
e2902068 |
920 | } |
921 | |
910a0a6d |
922 | sub calculate_ranks { |
923 | my $self = shift; |
924 | # Walk a version of the graph where every node linked by a relationship |
925 | # edge is fundamentally the same node, and do a topological ranking on |
926 | # the nodes in this graph. |
c9bf3dbf |
927 | my $topo_graph = Graph->new(); |
910a0a6d |
928 | my %rel_containers; |
929 | my $rel_ctr = 0; |
930 | # Add the nodes |
931 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
932 | next if exists $rel_containers{$r->id}; |
910a0a6d |
933 | my @rels = $r->related_readings( 'colocated' ); |
934 | if( @rels ) { |
935 | # Make a relationship container. |
936 | push( @rels, $r ); |
c9bf3dbf |
937 | my $rn = 'rel_container_' . $rel_ctr++; |
938 | $topo_graph->add_vertex( $rn ); |
910a0a6d |
939 | foreach( @rels ) { |
3a2ebbf4 |
940 | $rel_containers{$_->id} = $rn; |
910a0a6d |
941 | } |
942 | } else { |
943 | # Add a new node to mirror the old node. |
3a2ebbf4 |
944 | $rel_containers{$r->id} = $r->id; |
945 | $topo_graph->add_vertex( $r->id ); |
910a0a6d |
946 | } |
4a8828f0 |
947 | } |
3a1f2523 |
948 | |
3a2ebbf4 |
949 | # Add the edges. |
910a0a6d |
950 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
951 | foreach my $n ( $self->sequence->successors( $r->id ) ) { |
952 | my( $tfrom, $tto ) = ( $rel_containers{$r->id}, |
953 | $rel_containers{$n} ); |
954 | $topo_graph->add_edge( $tfrom, $tto ); |
910a0a6d |
955 | } |
956 | } |
957 | |
958 | # Now do the rankings, starting with the start node. |
3a2ebbf4 |
959 | my $topo_start = $rel_containers{$self->start->id}; |
c9bf3dbf |
960 | my $node_ranks = { $topo_start => 0 }; |
910a0a6d |
961 | my @curr_origin = ( $topo_start ); |
962 | # A little iterative function. |
963 | while( @curr_origin ) { |
c9bf3dbf |
964 | @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin ); |
910a0a6d |
965 | } |
966 | # Transfer our rankings from the topological graph to the real one. |
967 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
968 | if( defined $node_ranks->{$rel_containers{$r->id}} ) { |
969 | $r->rank( $node_ranks->{$rel_containers{$r->id}} ); |
67da8d6c |
970 | } else { |
971 | $DB::single = 1; |
3a2ebbf4 |
972 | die "No rank calculated for node " . $r->id |
67da8d6c |
973 | . " - do you have a cycle in the graph?"; |
974 | } |
de51424a |
975 | } |
8e1394aa |
976 | } |
3a1f2523 |
977 | |
910a0a6d |
978 | sub _assign_rank { |
c9bf3dbf |
979 | my( $graph, $node_ranks, @current_nodes ) = @_; |
910a0a6d |
980 | # Look at each of the children of @current_nodes. If all the child's |
981 | # parents have a rank, assign it the highest rank + 1 and add it to |
c9bf3dbf |
982 | # @next_nodes. Otherwise skip it; we will return when the highest-ranked |
983 | # parent gets a rank. |
910a0a6d |
984 | my @next_nodes; |
985 | foreach my $c ( @current_nodes ) { |
c9bf3dbf |
986 | warn "Current reading $c has no rank!" |
987 | unless exists $node_ranks->{$c}; |
988 | # print STDERR "Looking at child of node $c, rank " |
989 | # . $node_ranks->{$c} . "\n"; |
990 | foreach my $child ( $graph->successors( $c ) ) { |
991 | next if exists $node_ranks->{$child}; |
910a0a6d |
992 | my $highest_rank = -1; |
993 | my $skip = 0; |
c9bf3dbf |
994 | foreach my $parent ( $graph->predecessors( $child ) ) { |
995 | if( exists $node_ranks->{$parent} ) { |
996 | $highest_rank = $node_ranks->{$parent} |
997 | if $highest_rank <= $node_ranks->{$parent}; |
910a0a6d |
998 | } else { |
999 | $skip = 1; |
1000 | last; |
1001 | } |
1002 | } |
1003 | next if $skip; |
c9bf3dbf |
1004 | my $c_rank = $highest_rank + 1; |
1005 | # print STDERR "Assigning rank $c_rank to node $child \n"; |
1006 | $node_ranks->{$child} = $c_rank; |
910a0a6d |
1007 | push( @next_nodes, $child ); |
1008 | } |
1009 | } |
1010 | return @next_nodes; |
4cdd82f1 |
1011 | } |
910a0a6d |
1012 | |
0e476982 |
1013 | # Another method to make up for rough collation methods. If the same reading |
1014 | # appears multiple times at the same rank, collapse the nodes. |
1015 | sub flatten_ranks { |
1016 | my $self = shift; |
1017 | my %unique_rank_rdg; |
1018 | foreach my $rdg ( $self->readings ) { |
1019 | next unless $rdg->has_rank; |
1020 | my $key = $rdg->rank . "||" . $rdg->text; |
1021 | if( exists $unique_rank_rdg{$key} ) { |
1022 | # Combine! |
1023 | print STDERR "Combining readings at same rank: $key\n"; |
1024 | $self->merge_readings( $unique_rank_rdg{$key}, $rdg ); |
1025 | } else { |
1026 | $unique_rank_rdg{$key} = $rdg; |
1027 | } |
1028 | } |
1029 | } |
1030 | |
1031 | |
fa954f4c |
1032 | ## Utility functions |
de51424a |
1033 | |
4a8828f0 |
1034 | # Return the string that joins together a list of witnesses for |
1035 | # display on a single path. |
4a8828f0 |
1036 | sub witnesses_of_label { |
de51424a |
1037 | my( $self, $label ) = @_; |
4a8828f0 |
1038 | my $regex = $self->wit_list_separator; |
de51424a |
1039 | my @answer = split( /\Q$regex\E/, $label ); |
1040 | return @answer; |
4a8828f0 |
1041 | } |
8e1394aa |
1042 | |
dd3b58b0 |
1043 | no Moose; |
1044 | __PACKAGE__->meta->make_immutable; |
e867486f |
1045 | |
1046 | =head1 BUGS / TODO |
1047 | |
1048 | =over |
1049 | |
1050 | =item * Rationalize edge classes |
1051 | |
1052 | =item * Port the internal graph from Graph::Easy to Graph |
1053 | |
1054 | =back |