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; |
22222af9 |
9 | use Text::Tradition::Collation::RelationshipStore; |
df6d9812 |
10 | use XML::LibXML; |
dd3b58b0 |
11 | use Moose; |
12 | |
3a2ebbf4 |
13 | has 'sequence' => ( |
d047cd52 |
14 | is => 'ro', |
3a2ebbf4 |
15 | isa => 'Graph', |
16 | default => sub { Graph->new() }, |
d047cd52 |
17 | handles => { |
3a2ebbf4 |
18 | paths => 'edges', |
d047cd52 |
19 | }, |
d047cd52 |
20 | ); |
3a2ebbf4 |
21 | |
22 | has 'relations' => ( |
23 | is => 'ro', |
22222af9 |
24 | isa => 'Text::Tradition::Collation::RelationshipStore', |
25 | handles => { |
26 | relationships => 'relationships', |
27 | related_readings => 'related_readings', |
28 | }, |
29 | writer => '_set_relations', |
3a2ebbf4 |
30 | ); |
dd3b58b0 |
31 | |
3a2ebbf4 |
32 | has 'tradition' => ( |
33 | is => 'ro', |
d047cd52 |
34 | isa => 'Text::Tradition', |
8d9a1cd8 |
35 | weak_ref => 1, |
d047cd52 |
36 | ); |
dd3b58b0 |
37 | |
3a2ebbf4 |
38 | has 'readings' => ( |
39 | isa => 'HashRef[Text::Tradition::Collation::Reading]', |
40 | traits => ['Hash'], |
41 | handles => { |
42 | reading => 'get', |
43 | _add_reading => 'set', |
44 | del_reading => 'delete', |
45 | has_reading => 'exists', |
46 | readings => 'values', |
47 | }, |
48 | default => sub { {} }, |
49 | ); |
910a0a6d |
50 | |
4a8828f0 |
51 | has 'wit_list_separator' => ( |
7854e12e |
52 | is => 'rw', |
53 | isa => 'Str', |
54 | default => ', ', |
55 | ); |
56 | |
57 | has 'baselabel' => ( |
58 | is => 'rw', |
59 | isa => 'Str', |
60 | default => 'base text', |
61 | ); |
4a8828f0 |
62 | |
15d2d3df |
63 | has 'linear' => ( |
64 | is => 'rw', |
65 | isa => 'Bool', |
66 | default => 1, |
67 | ); |
1f563ac3 |
68 | |
ef9d481f |
69 | has 'ac_label' => ( |
70 | is => 'rw', |
71 | isa => 'Str', |
72 | default => ' (a.c.)', |
73 | ); |
3a2ebbf4 |
74 | |
75 | has 'start' => ( |
76 | is => 'ro', |
77 | isa => 'Text::Tradition::Collation::Reading', |
78 | writer => '_set_start', |
79 | weak_ref => 1, |
80 | ); |
81 | |
82 | has 'end' => ( |
83 | is => 'ro', |
84 | isa => 'Text::Tradition::Collation::Reading', |
85 | writer => '_set_end', |
86 | weak_ref => 1, |
87 | ); |
1f563ac3 |
88 | |
dd3b58b0 |
89 | # The collation can be created two ways: |
90 | # 1. Collate a set of witnesses (with CollateX I guess) and process |
91 | # the results as in 2. |
92 | # 2. Read a pre-prepared collation in one of a variety of formats, |
93 | # and make the graph from that. |
94 | |
95 | # The graph itself will (for now) be immutable, and the positions |
96 | # within the graph will also be immutable. We need to calculate those |
97 | # positions upon graph construction. The equivalences between graph |
98 | # nodes will be mutable, entirely determined by the user (or possibly |
99 | # by some semantic pre-processing provided by the user.) So the |
100 | # constructor should just make an empty equivalences object. The |
101 | # constructor will also need to make the witness objects, if we didn't |
102 | # come through option 1. |
103 | |
d047cd52 |
104 | sub BUILD { |
3a2ebbf4 |
105 | my $self = shift; |
22222af9 |
106 | $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) ); |
3a2ebbf4 |
107 | $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) ); |
108 | $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) ); |
d047cd52 |
109 | } |
784877d9 |
110 | |
3a2ebbf4 |
111 | ### Reading construct/destruct functions |
112 | |
113 | sub add_reading { |
114 | my( $self, $reading ) = @_; |
115 | unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) { |
116 | my %args = %$reading; |
117 | $reading = Text::Tradition::Collation::Reading->new( |
118 | 'collation' => $self, |
119 | %args ); |
120 | } |
121 | # First check to see if a reading with this ID exists. |
122 | if( $self->reading( $reading->id ) ) { |
123 | warn "Collation already has a reading with id " . $reading->id; |
124 | return undef; |
125 | } |
126 | $self->_add_reading( $reading->id => $reading ); |
127 | # Once the reading has been added, put it in both graphs. |
128 | $self->sequence->add_vertex( $reading->id ); |
22222af9 |
129 | $self->relations->add_reading( $reading->id ); |
3a2ebbf4 |
130 | return $reading; |
eca16057 |
131 | }; |
132 | |
3a2ebbf4 |
133 | around del_reading => sub { |
134 | my $orig = shift; |
135 | my $self = shift; |
136 | my $arg = shift; |
137 | |
138 | if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) { |
139 | $arg = $arg->id; |
140 | } |
3a2ebbf4 |
141 | # Remove the reading from the graphs. |
142 | $self->sequence->delete_vertex( $arg ); |
22222af9 |
143 | $self->relations->delete_reading( $arg ); |
3a2ebbf4 |
144 | |
145 | # Carry on. |
146 | $self->$orig( $arg ); |
147 | }; |
7854e12e |
148 | |
3a2ebbf4 |
149 | # merge_readings( $main, $to_be_deleted ); |
7854e12e |
150 | |
3a2ebbf4 |
151 | sub merge_readings { |
152 | my $self = shift; |
153 | |
154 | # We only need the IDs for adding paths to the graph, not the reading |
155 | # objects themselves. |
49d4f2ac |
156 | my( $kept, $deleted, $combine_char ) = $self->_stringify_args( @_ ); |
3a2ebbf4 |
157 | |
158 | # The kept reading should inherit the paths and the relationships |
159 | # of the deleted reading. |
160 | foreach my $path ( $self->sequence->edges_at( $deleted ) ) { |
161 | my @vector = ( $kept ); |
162 | push( @vector, $path->[1] ) if $path->[0] eq $deleted; |
163 | unshift( @vector, $path->[0] ) if $path->[1] eq $deleted; |
49d4f2ac |
164 | next if $vector[0] eq $vector[1]; # Don't add a self loop |
3a2ebbf4 |
165 | my %wits = %{$self->sequence->get_edge_attributes( @$path )}; |
166 | $self->sequence->add_edge( @vector ); |
167 | my $fwits = $self->sequence->get_edge_attributes( @vector ); |
168 | @wits{keys %$fwits} = values %$fwits; |
169 | $self->sequence->set_edge_attributes( @vector, \%wits ); |
170 | } |
22222af9 |
171 | $self->relations->merge_readings( $kept, $deleted, $combine_char ); |
3a2ebbf4 |
172 | |
173 | # Do the deletion deed. |
49d4f2ac |
174 | if( $combine_char ) { |
175 | my $kept_obj = $self->reading( $kept ); |
176 | my $new_text = join( $combine_char, $kept_obj->text, |
177 | $self->reading( $deleted )->text ); |
178 | $kept_obj->alter_text( $new_text ); |
179 | } |
3a2ebbf4 |
180 | $self->del_reading( $deleted ); |
181 | } |
7854e12e |
182 | |
3265b0ce |
183 | |
3a2ebbf4 |
184 | # Helper function for manipulating the graph. |
185 | sub _stringify_args { |
186 | my( $self, $first, $second, $arg ) = @_; |
187 | $first = $first->id |
188 | if ref( $first ) eq 'Text::Tradition::Collation::Reading'; |
189 | $second = $second->id |
190 | if ref( $second ) eq 'Text::Tradition::Collation::Reading'; |
191 | return( $first, $second, $arg ); |
192 | } |
df6d9812 |
193 | |
3a2ebbf4 |
194 | ### Path logic |
195 | |
196 | sub add_path { |
197 | my $self = shift; |
198 | |
199 | # We only need the IDs for adding paths to the graph, not the reading |
200 | # objects themselves. |
201 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
202 | |
203 | # Connect the readings |
204 | $self->sequence->add_edge( $source, $target ); |
205 | # Note the witness in question |
206 | $self->sequence->set_edge_attribute( $source, $target, $wit, 1 ); |
b15511bf |
207 | }; |
208 | |
3a2ebbf4 |
209 | sub del_path { |
210 | my $self = shift; |
49d4f2ac |
211 | my @args; |
212 | if( ref( $_[0] ) eq 'ARRAY' ) { |
213 | my $e = shift @_; |
214 | @args = ( @$e, @_ ); |
215 | } else { |
216 | @args = @_; |
217 | } |
3a2ebbf4 |
218 | |
219 | # We only need the IDs for adding paths to the graph, not the reading |
220 | # objects themselves. |
49d4f2ac |
221 | my( $source, $target, $wit ) = $self->_stringify_args( @args ); |
3a2ebbf4 |
222 | |
223 | if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) { |
49d4f2ac |
224 | $self->sequence->delete_edge_attribute( $source, $target, $wit ); |
3a2ebbf4 |
225 | } |
226 | unless( keys %{$self->sequence->get_edge_attributes( $source, $target )} ) { |
227 | $self->sequence->delete_edge( $source, $target ); |
228 | } |
784877d9 |
229 | } |
230 | |
3a2ebbf4 |
231 | |
15d2d3df |
232 | # Extra graph-alike utility |
233 | sub has_path { |
3a2ebbf4 |
234 | my $self = shift; |
235 | my( $source, $target, $wit ) = $self->_stringify_args( @_ ); |
236 | return undef unless $self->sequence->has_edge( $source, $target ); |
237 | return $self->sequence->has_edge_attribute( $source, $target, $wit ); |
b15511bf |
238 | } |
239 | |
3a2ebbf4 |
240 | =head2 add_relationship( $reading1, $reading2, $definition ) |
241 | |
242 | Adds the specified relationship between the two readings. A relationship |
22222af9 |
243 | is transitive (i.e. undirected); the options for its definition may be found |
244 | in Text::Tradition::Collation::Relationship. |
3a2ebbf4 |
245 | |
246 | =cut |
247 | |
248 | # Wouldn't it be lovely if edges could be objects, and all this type checking |
249 | # and attribute management could be done via Moose? |
250 | |
251 | sub add_relationship { |
252 | my $self = shift; |
22222af9 |
253 | my( $source, $target, $opts ) = $self->_stringify_args( @_ ); |
64ae6270 |
254 | my( $ret, @vectors ) = $self->relations->add_relationship( $source, |
255 | $self->reading( $source ), $target, $self->reading( $target ), $opts ); |
256 | # Force a full rank recalculation every time. Yuck. |
257 | $self->calculate_ranks() if $ret && $self->end->has_rank; |
258 | return( $ret, @vectors ); |
22222af9 |
259 | } |
ef9d481f |
260 | |
22222af9 |
261 | =head2 reading_witnesses( $reading ) |
910a0a6d |
262 | |
22222af9 |
263 | Return a list of sigils corresponding to the witnesses in which the reading appears. |
3265b0ce |
264 | |
22222af9 |
265 | =cut |
1d310495 |
266 | |
1d310495 |
267 | sub reading_witnesses { |
268 | my( $self, $reading ) = @_; |
269 | # We need only check either the incoming or the outgoing edges; I have |
96dc90ec |
270 | # arbitrarily chosen "incoming". Thus, special-case the start node. |
271 | if( $reading eq $self->start ) { |
272 | return map { $_->sigil } $self->tradition->witnesses; |
273 | } |
1d310495 |
274 | my %all_witnesses; |
275 | foreach my $e ( $self->sequence->edges_to( $reading ) ) { |
276 | my $wits = $self->sequence->get_edge_attributes( @$e ); |
277 | @all_witnesses{ keys %$wits } = 1; |
278 | } |
279 | return keys %all_witnesses; |
910a0a6d |
280 | } |
281 | |
8e1394aa |
282 | =head2 Output method(s) |
283 | |
284 | =over |
285 | |
286 | =item B<as_svg> |
287 | |
b22576c6 |
288 | print $collation->as_svg(); |
8e1394aa |
289 | |
0068967c |
290 | Returns an SVG string that represents the graph, via as_dot and graphviz. |
8e1394aa |
291 | |
292 | =cut |
293 | |
294 | sub as_svg { |
3a2ebbf4 |
295 | my( $self ) = @_; |
296 | |
8e1394aa |
297 | my @cmd = qw/dot -Tsvg/; |
298 | my( $svg, $err ); |
910a0a6d |
299 | my $dotfile = File::Temp->new(); |
d9e873d0 |
300 | ## TODO REMOVE |
eca16057 |
301 | # $dotfile->unlink_on_destroy(0); |
910a0a6d |
302 | binmode $dotfile, ':utf8'; |
303 | print $dotfile $self->as_dot(); |
304 | push( @cmd, $dotfile->filename ); |
305 | run( \@cmd, ">", binary(), \$svg ); |
306 | $svg = decode_utf8( $svg ); |
8e1394aa |
307 | return $svg; |
308 | } |
309 | |
b22576c6 |
310 | =item B<svg_subgraph> |
311 | |
312 | print $collation->svg_subgraph( $from, $to ) |
313 | |
314 | Returns an SVG string that represents the portion of the graph given by the |
315 | specified range. The $from and $to variables refer to ranks within the graph. |
316 | |
317 | =cut |
318 | |
319 | sub svg_subgraph { |
320 | my( $self, $from, $to ) = @_; |
321 | |
322 | my $dot = $self->as_dot( $from, $to ); |
323 | unless( $dot ) { |
324 | warn "Could not output a graph with range $from - $to"; |
325 | return; |
326 | } |
327 | |
328 | my @cmd = qw/dot -Tsvg/; |
329 | my( $svg, $err ); |
330 | my $dotfile = File::Temp->new(); |
331 | ## TODO REMOVE |
332 | # $dotfile->unlink_on_destroy(0); |
333 | binmode $dotfile, ':utf8'; |
334 | print $dotfile $dot; |
335 | push( @cmd, $dotfile->filename ); |
336 | run( \@cmd, ">", binary(), \$svg ); |
337 | $svg = decode_utf8( $svg ); |
338 | return $svg; |
339 | } |
340 | |
341 | |
df6d9812 |
342 | =item B<as_dot> |
343 | |
b22576c6 |
344 | print $collation->as_dot(); |
df6d9812 |
345 | |
346 | Returns a string that is the collation graph expressed in dot |
347 | (i.e. GraphViz) format. The 'view' argument determines what kind of |
348 | graph is produced. |
349 | * 'path': a graph of witness paths through the collation (DEFAULT) |
350 | * 'relationship': a graph of how collation readings relate to |
351 | each other |
352 | |
353 | =cut |
354 | |
355 | sub as_dot { |
b22576c6 |
356 | my( $self, $startrank, $endrank ) = @_; |
357 | |
358 | # Check the arguments |
359 | if( $startrank ) { |
360 | return if $endrank && $startrank > $endrank; |
361 | return if $startrank > $self->end->rank; |
362 | } |
363 | if( defined $endrank ) { |
364 | return if $endrank < 0; |
365 | } |
366 | |
df6d9812 |
367 | # TODO consider making some of these things configurable |
67da8d6c |
368 | my $graph_name = $self->tradition->name; |
369 | $graph_name =~ s/[^\w\s]//g; |
370 | $graph_name = join( '_', split( /\s+/, $graph_name ) ); |
371 | my $dot = sprintf( "digraph %s {\n", $graph_name ); |
df6d9812 |
372 | $dot .= "\tedge [ arrowhead=open ];\n"; |
5f52ddca |
373 | $dot .= "\tgraph [ rankdir=LR,bgcolor=none ];\n"; |
df6d9812 |
374 | $dot .= sprintf( "\tnode [ fontsize=%d, fillcolor=%s, style=%s, shape=%s ];\n", |
3a2ebbf4 |
375 | 11, "white", "filled", "ellipse" ); |
df6d9812 |
376 | |
b22576c6 |
377 | # Output substitute start/end readings if necessary |
378 | if( $startrank ) { |
379 | $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n"; |
380 | } |
381 | if( $endrank ) { |
382 | $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n"; |
383 | } |
384 | my %used; # Keep track of the readings that actually appear in the graph |
385 | my %subedges; |
386 | my %subend; |
df6d9812 |
387 | foreach my $reading ( $self->readings ) { |
b22576c6 |
388 | # Only output readings within our rank range. |
389 | next if $startrank && $reading->rank < $startrank; |
390 | next if $endrank && $reading->rank > $endrank; |
391 | $used{$reading->id} = 1; |
392 | $subedges{$reading->id} = '#SUBSTART#' |
393 | if $startrank && $startrank == $reading->rank; |
394 | $subedges{$reading->id} = '#SUBEND#' |
395 | if $endrank && $endrank == $reading->rank; |
910a0a6d |
396 | # Need not output nodes without separate labels |
3a2ebbf4 |
397 | next if $reading->id eq $reading->text; |
8f9cab7b |
398 | my $label = $reading->text; |
399 | $label =~ s/\"/\\\"/g; |
400 | $dot .= sprintf( "\t\"%s\" [ label=\"%s\" ];\n", $reading->id, $label ); |
df6d9812 |
401 | } |
3a2ebbf4 |
402 | |
b22576c6 |
403 | # Add substitute start and end edges if necessary |
404 | foreach my $node ( keys %subedges ) { |
405 | my @vector = ( $subedges{$node}, $node ); |
406 | @vector = reverse( @vector ) if $vector[0] =~ /END/; |
407 | my $witstr = join( ', ', sort $self->reading_witnesses( $self->reading( $node ) ) ); |
408 | my %variables = ( 'color' => '#000000', |
910a0a6d |
409 | 'fontcolor' => '#000000', |
b22576c6 |
410 | 'label' => $witstr, |
910a0a6d |
411 | ); |
412 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
b22576c6 |
413 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", @vector, $varopts ); |
414 | } |
415 | |
416 | # Add the real edges |
417 | my @edges = $self->paths; |
418 | foreach my $edge ( @edges ) { |
419 | # Do we need to output this edge? |
420 | if( $used{$edge->[0]} && $used{$edge->[1]} ) {; |
421 | my %variables = ( 'color' => '#000000', |
422 | 'fontcolor' => '#000000', |
423 | 'label' => join( ', ', $self->path_display_label( $edge ) ), |
424 | ); |
425 | my $varopts = join( ', ', map { $_.'="'.$variables{$_}.'"' } sort keys %variables ); |
426 | # Account for the rank gap if necessary |
427 | my $rankgap = $self->reading( $edge->[1] )->rank |
428 | - $self->reading( $edge->[0] )->rank; |
429 | $varopts .= ", minlen=$rankgap" if $rankgap > 1; |
430 | $dot .= sprintf( "\t\"%s\" -> \"%s\" [ %s ];\n", |
431 | $edge->[0], $edge->[1], $varopts ); |
432 | } |
df6d9812 |
433 | } |
b22576c6 |
434 | |
df6d9812 |
435 | $dot .= "}\n"; |
436 | return $dot; |
437 | } |
438 | |
3a2ebbf4 |
439 | sub path_witnesses { |
440 | my( $self, @edge ) = @_; |
441 | # If edge is an arrayref, cope. |
442 | if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) { |
443 | my $e = shift @edge; |
444 | @edge = @$e; |
445 | } |
446 | my @wits = keys %{$self->sequence->get_edge_attributes( @edge )}; |
447 | return sort @wits; |
448 | } |
449 | |
8f9cab7b |
450 | sub path_display_label { |
451 | my( $self, $edge ) = @_; |
452 | my @wits = $self->path_witnesses( $edge ); |
453 | my $maj = scalar( $self->tradition->witnesses ) * 0.6; |
454 | if( scalar @wits > $maj ) { |
455 | return 'majority'; |
456 | } else { |
457 | return join( ', ', @wits ); |
458 | } |
459 | } |
460 | |
461 | |
8e1394aa |
462 | =item B<as_graphml> |
463 | |
b22576c6 |
464 | print $collation->as_graphml( $recalculate ) |
8e1394aa |
465 | |
466 | Returns a GraphML representation of the collation graph, with |
467 | transposition information and position information. Unless |
468 | $recalculate is passed (and is a true value), the method will return a |
469 | cached copy of the SVG after the first call to the method. |
470 | |
471 | =cut |
472 | |
473 | sub as_graphml { |
3a2ebbf4 |
474 | my( $self ) = @_; |
8e1394aa |
475 | |
476 | # Some namespaces |
477 | my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; |
478 | my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; |
479 | my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' . |
910a0a6d |
480 | 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd'; |
8e1394aa |
481 | |
482 | # Create the document and root node |
483 | my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" ); |
484 | my $root = $graphml->createElementNS( $graphml_ns, 'graphml' ); |
485 | $graphml->setDocumentElement( $root ); |
486 | $root->setNamespace( $xsi_ns, 'xsi', 0 ); |
487 | $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema ); |
488 | |
e309421a |
489 | # Add the data keys for the graph |
490 | my %graph_data_keys; |
491 | my $gdi = 0; |
1d310495 |
492 | my @graph_attributes = qw/ version wit_list_separator baselabel linear ac_label /; |
e309421a |
493 | foreach my $datum ( @graph_attributes ) { |
494 | $graph_data_keys{$datum} = 'dg'.$gdi++; |
495 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
496 | $key->setAttribute( 'attr.name', $datum ); |
497 | $key->setAttribute( 'attr.type', $key eq 'linear' ? 'boolean' : 'string' ); |
498 | $key->setAttribute( 'for', 'graph' ); |
499 | $key->setAttribute( 'id', $graph_data_keys{$datum} ); |
500 | } |
f6066bac |
501 | |
8e1394aa |
502 | # Add the data keys for nodes |
ef9d481f |
503 | my %node_data_keys; |
504 | my $ndi = 0; |
3a2ebbf4 |
505 | my %node_data = ( |
506 | id => 'string', |
255875b8 |
507 | text => 'string', |
3a2ebbf4 |
508 | rank => 'string', |
509 | is_start => 'boolean', |
510 | is_end => 'boolean', |
511 | is_lacuna => 'boolean', |
512 | ); |
513 | foreach my $datum ( keys %node_data ) { |
910a0a6d |
514 | $node_data_keys{$datum} = 'dn'.$ndi++; |
515 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
516 | $key->setAttribute( 'attr.name', $datum ); |
3a2ebbf4 |
517 | $key->setAttribute( 'attr.type', $node_data{$datum} ); |
910a0a6d |
518 | $key->setAttribute( 'for', 'node' ); |
519 | $key->setAttribute( 'id', $node_data_keys{$datum} ); |
8e1394aa |
520 | } |
521 | |
df6d9812 |
522 | # Add the data keys for edges, i.e. witnesses |
ef9d481f |
523 | my $edi = 0; |
524 | my %edge_data_keys; |
3a2ebbf4 |
525 | my %edge_data = ( |
f523c7a8 |
526 | class => 'string', # Class, deprecated soon |
3a2ebbf4 |
527 | witness => 'string', # ID/label for a path |
528 | relationship => 'string', # ID/label for a relationship |
529 | extra => 'boolean', # Path key |
530 | colocated => 'boolean', # Relationship key |
531 | non_correctable => 'boolean', # Relationship key |
532 | non_independent => 'boolean', # Relationship key |
533 | ); |
534 | foreach my $datum ( keys %edge_data ) { |
535 | $edge_data_keys{$datum} = 'de'.$edi++; |
910a0a6d |
536 | my $key = $root->addNewChild( $graphml_ns, 'key' ); |
3a2ebbf4 |
537 | $key->setAttribute( 'attr.name', $datum ); |
538 | $key->setAttribute( 'attr.type', $edge_data{$datum} ); |
910a0a6d |
539 | $key->setAttribute( 'for', 'edge' ); |
3a2ebbf4 |
540 | $key->setAttribute( 'id', $edge_data_keys{$datum} ); |
8e1394aa |
541 | } |
3a2ebbf4 |
542 | |
22222af9 |
543 | # Add the collation graph itself |
2c669bca |
544 | my $sgraph = $root->addNewChild( $graphml_ns, 'graph' ); |
545 | $sgraph->setAttribute( 'edgedefault', 'directed' ); |
546 | $sgraph->setAttribute( 'id', $self->tradition->name ); |
547 | $sgraph->setAttribute( 'parse.edgeids', 'canonical' ); |
548 | $sgraph->setAttribute( 'parse.edges', scalar($self->paths) ); |
549 | $sgraph->setAttribute( 'parse.nodeids', 'canonical' ); |
550 | $sgraph->setAttribute( 'parse.nodes', scalar($self->readings) ); |
551 | $sgraph->setAttribute( 'parse.order', 'nodesfirst' ); |
22222af9 |
552 | |
e309421a |
553 | # Collation attribute data |
554 | foreach my $datum ( @graph_attributes ) { |
2c669bca |
555 | my $value = $datum eq 'version' ? '3.0' : $self->$datum; |
556 | _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value ); |
e309421a |
557 | } |
8e1394aa |
558 | |
559 | my $node_ctr = 0; |
560 | my %node_hash; |
22222af9 |
561 | # Add our readings to the graph |
3a2ebbf4 |
562 | foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) { |
2c669bca |
563 | # Add to the main graph |
564 | my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' ); |
910a0a6d |
565 | my $node_xmlid = 'n' . $node_ctr++; |
3a2ebbf4 |
566 | $node_hash{ $n->id } = $node_xmlid; |
910a0a6d |
567 | $node_el->setAttribute( 'id', $node_xmlid ); |
255875b8 |
568 | foreach my $d ( keys %node_data ) { |
569 | my $nval = $n->$d; |
570 | _add_graphml_data( $node_el, $node_data_keys{$d}, $nval ) |
571 | if defined $nval; |
572 | } |
b15511bf |
573 | } |
574 | |
2c669bca |
575 | # Add the path edges to the sequence graph |
df6d9812 |
576 | my $edge_ctr = 0; |
3a2ebbf4 |
577 | foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) { |
578 | # We add an edge in the graphml for every witness in $e. |
579 | foreach my $wit ( $self->path_witnesses( $e ) ) { |
580 | my( $id, $from, $to ) = ( 'e'.$edge_ctr++, |
581 | $node_hash{ $e->[0] }, |
582 | $node_hash{ $e->[1] } ); |
2c669bca |
583 | my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' ); |
3a2ebbf4 |
584 | $edge_el->setAttribute( 'source', $from ); |
585 | $edge_el->setAttribute( 'target', $to ); |
586 | $edge_el->setAttribute( 'id', $id ); |
3a2ebbf4 |
587 | |
588 | # It's a witness path, so add the witness |
589 | my $base = $wit; |
590 | my $key = $edge_data_keys{'witness'}; |
591 | # Is this an ante-corr witness? |
592 | my $aclabel = $self->ac_label; |
593 | if( $wit =~ /^(.*)\Q$aclabel\E$/ ) { |
594 | # Keep the base witness |
595 | $base = $1; |
596 | # ...and record that this is an 'extra' reading path |
597 | _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel ); |
598 | } |
599 | _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base ); |
f523c7a8 |
600 | _add_graphml_data( $edge_el, $edge_data_keys{'class'}, 'path' ); |
3a2ebbf4 |
601 | } |
602 | } |
603 | |
22222af9 |
604 | # Add the relationship graph to the XML |
605 | $self->relations->as_graphml( $root ); |
8e1394aa |
606 | |
94c00c71 |
607 | # Save and return the thing |
608 | my $result = decode_utf8( $graphml->toString(1) ); |
94c00c71 |
609 | return $result; |
df6d9812 |
610 | } |
611 | |
b15511bf |
612 | sub _add_graphml_data { |
613 | my( $el, $key, $value ) = @_; |
b15511bf |
614 | return unless defined $value; |
c9bf3dbf |
615 | my $data_el = $el->addNewChild( $el->namespaceURI, 'data' ); |
b15511bf |
616 | $data_el->setAttribute( 'key', $key ); |
617 | $data_el->appendText( $value ); |
8e1394aa |
618 | } |
619 | |
910a0a6d |
620 | =item B<as_csv> |
621 | |
b22576c6 |
622 | print $collation->as_csv( $recalculate ) |
910a0a6d |
623 | |
624 | Returns a CSV alignment table representation of the collation graph, one |
2c669bca |
625 | row per witness (or witness uncorrected.) |
910a0a6d |
626 | |
627 | =cut |
628 | |
629 | sub as_csv { |
3a2ebbf4 |
630 | my( $self ) = @_; |
910a0a6d |
631 | my $table = $self->make_alignment_table; |
632 | my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } ); |
633 | my @result; |
2c669bca |
634 | # Make the header row |
635 | $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} ); |
636 | push( @result, decode_utf8( $csv->string ) ); |
637 | # Make the rest of the rows |
638 | foreach my $idx ( 0 .. $table->{'length'} - 1 ) { |
566f4595 |
639 | my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}}; |
2c669bca |
640 | my @row = map { $_ ? $_->{'t'} : $_ } @rowobjs; |
641 | $csv->combine( @row ); |
910a0a6d |
642 | push( @result, decode_utf8( $csv->string ) ); |
643 | } |
3a2ebbf4 |
644 | return join( "\n", @result ); |
910a0a6d |
645 | } |
646 | |
2c669bca |
647 | =item B<make_alignment_table> |
648 | |
b22576c6 |
649 | my $table = $collation->make_alignment_table( $use_refs, \@wits_to_include ) |
2c669bca |
650 | |
566f4595 |
651 | Return a reference to an alignment table, in a slightly enhanced CollateX |
652 | format which looks like this: |
653 | |
654 | $table = { alignment => [ { witness => "SIGIL", |
655 | tokens => [ { t => "READINGTEXT" }, ... ] }, |
656 | { witness => "SIG2", |
657 | tokens => [ { t => "READINGTEXT" }, ... ] }, |
658 | ... ], |
659 | length => TEXTLEN }; |
660 | |
661 | If $use_refs is set to 1, the reading object is returned in the table |
662 | instead of READINGTEXT; if not, the text of the reading is returned. |
663 | If $wits_to_include is set to a hashref, only the witnesses whose sigil |
664 | keys have a true hash value will be included. |
2c669bca |
665 | |
666 | =cut |
9f3ba6f7 |
667 | |
910a0a6d |
668 | sub make_alignment_table { |
08e0fb85 |
669 | my( $self, $noderefs, $include ) = @_; |
910a0a6d |
670 | unless( $self->linear ) { |
671 | warn "Need a linear graph in order to make an alignment table"; |
672 | return; |
673 | } |
2c669bca |
674 | my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 }; |
3a2ebbf4 |
675 | my @all_pos = ( 1 .. $self->end->rank - 1 ); |
68454b71 |
676 | foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) { |
2c669bca |
677 | if( $include ) { |
566f4595 |
678 | next unless $include->{$wit->sigil}; |
2c669bca |
679 | } |
eca16057 |
680 | # print STDERR "Making witness row(s) for " . $wit->sigil . "\n"; |
1f7aa795 |
681 | my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil ); |
682 | my @row = _make_witness_row( \@wit_path, \@all_pos, $noderefs ); |
2c669bca |
683 | push( @{$table->{'alignment'}}, |
684 | { 'witness' => $wit->sigil, 'tokens' => \@row } ); |
1f7aa795 |
685 | if( $wit->is_layered ) { |
686 | my @wit_ac_path = $self->reading_sequence( $self->start, $self->end, |
687 | $wit->sigil.$self->ac_label, $wit->sigil ); |
688 | my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos, $noderefs ); |
2c669bca |
689 | push( @{$table->{'alignment'}}, |
690 | { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } ); |
910a0a6d |
691 | } |
692 | } |
2c669bca |
693 | return $table; |
910a0a6d |
694 | } |
695 | |
696 | sub _make_witness_row { |
0e476982 |
697 | my( $path, $positions, $noderefs ) = @_; |
910a0a6d |
698 | my %char_hash; |
699 | map { $char_hash{$_} = undef } @$positions; |
2c669bca |
700 | my $debug = 0; |
910a0a6d |
701 | foreach my $rdg ( @$path ) { |
eca16057 |
702 | my $rtext = $rdg->text; |
703 | $rtext = '#LACUNA#' if $rdg->is_lacuna; |
2c669bca |
704 | print STDERR "rank " . $rdg->rank . "\n" if $debug; |
3a2ebbf4 |
705 | # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank; |
2c669bca |
706 | $char_hash{$rdg->rank} = $noderefs ? { 't' => $rdg } |
707 | : { 't' => $rtext }; |
910a0a6d |
708 | } |
709 | my @row = map { $char_hash{$_} } @$positions; |
eca16057 |
710 | # Fill in lacuna markers for undef spots in the row |
711 | my $last_el = shift @row; |
712 | my @filled_row = ( $last_el ); |
713 | foreach my $el ( @row ) { |
0e476982 |
714 | # If we are using node reference, make the lacuna node appear many times |
715 | # in the table. If not, use the lacuna tag. |
716 | if( $last_el && _el_is_lacuna( $last_el ) && !defined $el ) { |
566f4595 |
717 | $el = $noderefs ? $last_el : { 't' => '#LACUNA#' }; |
eca16057 |
718 | } |
719 | push( @filled_row, $el ); |
720 | $last_el = $el; |
721 | } |
722 | return @filled_row; |
910a0a6d |
723 | } |
724 | |
0e476982 |
725 | # Tiny utility function to say if a table element is a lacuna |
726 | sub _el_is_lacuna { |
727 | my $el = shift; |
2c669bca |
728 | return 1 if $el->{'t'} eq '#LACUNA#'; |
729 | return 1 if ref( $el->{'t'} ) eq 'Text::Tradition::Collation::Reading' |
730 | && $el->{'t'}->is_lacuna; |
0e476982 |
731 | return 0; |
732 | } |
733 | |
910a0a6d |
734 | # Helper to turn the witnesses along columns rather than rows. Assumes |
735 | # equal-sized rows. |
736 | sub _turn_table { |
737 | my( $table ) = @_; |
738 | my $result = []; |
739 | return $result unless scalar @$table; |
740 | my $nrows = scalar @{$table->[0]}; |
741 | foreach my $idx ( 0 .. $nrows - 1 ) { |
742 | foreach my $wit ( 0 .. $#{$table} ) { |
743 | $result->[$idx]->[$wit] = $table->[$wit]->[$idx]; |
744 | } |
745 | } |
746 | return $result; |
747 | } |
748 | |
8e1394aa |
749 | =back |
750 | |
de51424a |
751 | =head2 Navigation methods |
752 | |
753 | =over |
754 | |
8e1394aa |
755 | =item B<start> |
756 | |
757 | my $beginning = $collation->start(); |
758 | |
759 | Returns the beginning of the collation, a meta-reading with label '#START#'. |
760 | |
910a0a6d |
761 | =item B<end> |
762 | |
763 | my $end = $collation->end(); |
764 | |
765 | Returns the end of the collation, a meta-reading with label '#END#'. |
766 | |
910a0a6d |
767 | |
e2902068 |
768 | =item B<reading_sequence> |
769 | |
b22576c6 |
770 | my @readings = $collation->reading_sequence( $first, $last, $path[, $alt_path] ); |
e2902068 |
771 | |
772 | Returns the ordered list of readings, starting with $first and ending |
773 | with $last, along the given witness path. If no path is specified, |
774 | assume that the path is that of the base text (if any.) |
775 | |
776 | =cut |
777 | |
910a0a6d |
778 | # TODO Think about returning some lazy-eval iterator. |
779 | |
e2902068 |
780 | sub reading_sequence { |
781 | my( $self, $start, $end, $witness, $backup ) = @_; |
782 | |
930ff666 |
783 | $witness = $self->baselabel unless $witness; |
e2902068 |
784 | my @readings = ( $start ); |
785 | my %seen; |
786 | my $n = $start; |
3a2ebbf4 |
787 | while( $n && $n->id ne $end->id ) { |
788 | if( exists( $seen{$n->id} ) ) { |
789 | warn "Detected loop at " . $n->id; |
910a0a6d |
790 | last; |
791 | } |
3a2ebbf4 |
792 | $seen{$n->id} = 1; |
910a0a6d |
793 | |
794 | my $next = $self->next_reading( $n, $witness, $backup ); |
44771cf2 |
795 | unless( $next ) { |
3a2ebbf4 |
796 | warn "Did not find any path for $witness from reading " . $n->id; |
44771cf2 |
797 | last; |
798 | } |
910a0a6d |
799 | push( @readings, $next ); |
800 | $n = $next; |
e2902068 |
801 | } |
802 | # Check that the last reading is our end reading. |
803 | my $last = $readings[$#readings]; |
3a2ebbf4 |
804 | warn "Last reading found from " . $start->text . |
910a0a6d |
805 | " for witness $witness is not the end!" |
3a2ebbf4 |
806 | unless $last->id eq $end->id; |
e2902068 |
807 | |
808 | return @readings; |
809 | } |
810 | |
4a8828f0 |
811 | =item B<next_reading> |
8e1394aa |
812 | |
b22576c6 |
813 | my $next_reading = $collation->next_reading( $reading, $witpath ); |
8e1394aa |
814 | |
4a8828f0 |
815 | Returns the reading that follows the given reading along the given witness |
930ff666 |
816 | path. |
8e1394aa |
817 | |
818 | =cut |
819 | |
4a8828f0 |
820 | sub next_reading { |
e2902068 |
821 | # Return the successor via the corresponding path. |
8e1394aa |
822 | my $self = shift; |
3a2ebbf4 |
823 | my $answer = $self->_find_linked_reading( 'next', @_ ); |
2c669bca |
824 | return undef unless $answer; |
3a2ebbf4 |
825 | return $self->reading( $answer ); |
8e1394aa |
826 | } |
827 | |
4a8828f0 |
828 | =item B<prior_reading> |
8e1394aa |
829 | |
b22576c6 |
830 | my $prior_reading = $collation->prior_reading( $reading, $witpath ); |
8e1394aa |
831 | |
4a8828f0 |
832 | Returns the reading that precedes the given reading along the given witness |
930ff666 |
833 | path. |
8e1394aa |
834 | |
835 | =cut |
836 | |
4a8828f0 |
837 | sub prior_reading { |
e2902068 |
838 | # Return the predecessor via the corresponding path. |
8e1394aa |
839 | my $self = shift; |
3a2ebbf4 |
840 | my $answer = $self->_find_linked_reading( 'prior', @_ ); |
841 | return $self->reading( $answer ); |
8e1394aa |
842 | } |
843 | |
4a8828f0 |
844 | sub _find_linked_reading { |
e2902068 |
845 | my( $self, $direction, $node, $path, $alt_path ) = @_; |
846 | my @linked_paths = $direction eq 'next' |
3a2ebbf4 |
847 | ? $self->sequence->edges_from( $node ) |
848 | : $self->sequence->edges_to( $node ); |
e2902068 |
849 | return undef unless scalar( @linked_paths ); |
8e1394aa |
850 | |
e2902068 |
851 | # We have to find the linked path that contains all of the |
852 | # witnesses supplied in $path. |
853 | my( @path_wits, @alt_path_wits ); |
3a2ebbf4 |
854 | @path_wits = sort( $self->witnesses_of_label( $path ) ) if $path; |
855 | @alt_path_wits = sort( $self->witnesses_of_label( $alt_path ) ) if $alt_path; |
e2902068 |
856 | my $base_le; |
857 | my $alt_le; |
858 | foreach my $le ( @linked_paths ) { |
3a2ebbf4 |
859 | if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) { |
910a0a6d |
860 | $base_le = $le; |
910a0a6d |
861 | } |
3a2ebbf4 |
862 | my @le_wits = $self->path_witnesses( $le ); |
863 | if( _is_within( \@path_wits, \@le_wits ) ) { |
864 | # This is the right path. |
865 | return $direction eq 'next' ? $le->[1] : $le->[0]; |
866 | } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) { |
867 | $alt_le = $le; |
868 | } |
8e1394aa |
869 | } |
e2902068 |
870 | # Got this far? Return the alternate path if it exists. |
3a2ebbf4 |
871 | return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0] |
910a0a6d |
872 | if $alt_le; |
e2902068 |
873 | |
874 | # Got this far? Return the base path if it exists. |
3a2ebbf4 |
875 | return $direction eq 'next' ? $base_le->[1] : $base_le->[0] |
910a0a6d |
876 | if $base_le; |
e2902068 |
877 | |
878 | # Got this far? We have no appropriate path. |
2c669bca |
879 | warn "Could not find $direction node from " . $node->id |
910a0a6d |
880 | . " along path $path"; |
8e1394aa |
881 | return undef; |
882 | } |
883 | |
4a8828f0 |
884 | # Some set logic. |
885 | sub _is_within { |
886 | my( $set1, $set2 ) = @_; |
7854e12e |
887 | my $ret = @$set1; # will be 0, i.e. false, if set1 is empty |
4a8828f0 |
888 | foreach my $el ( @$set1 ) { |
910a0a6d |
889 | $ret = 0 unless grep { /^\Q$el\E$/ } @$set2; |
4a8828f0 |
890 | } |
891 | return $ret; |
892 | } |
893 | |
de51424a |
894 | |
895 | ## INITIALIZATION METHODS - for use by parsers |
930ff666 |
896 | |
7e450e44 |
897 | # For use when a collation is constructed from a base text and an apparatus. |
898 | # We have the sequences of readings and just need to add path edges. |
1f7aa795 |
899 | # When we are done, clear out the witness path attributes, as they are no |
900 | # longer needed. |
901 | # TODO Find a way to replace the witness path attributes with encapsulated functions? |
e2902068 |
902 | |
6a222840 |
903 | sub make_witness_paths { |
904 | my( $self ) = @_; |
910a0a6d |
905 | foreach my $wit ( $self->tradition->witnesses ) { |
0068967c |
906 | # print STDERR "Making path for " . $wit->sigil . "\n"; |
910a0a6d |
907 | $self->make_witness_path( $wit ); |
7854e12e |
908 | } |
7854e12e |
909 | } |
910 | |
6a222840 |
911 | sub make_witness_path { |
7854e12e |
912 | my( $self, $wit ) = @_; |
913 | my @chain = @{$wit->path}; |
15d2d3df |
914 | my $sig = $wit->sigil; |
7854e12e |
915 | foreach my $idx ( 0 .. $#chain-1 ) { |
910a0a6d |
916 | $self->add_path( $chain[$idx], $chain[$idx+1], $sig ); |
7854e12e |
917 | } |
1f7aa795 |
918 | if( $wit->is_layered ) { |
d9e873d0 |
919 | @chain = @{$wit->uncorrected_path}; |
920 | foreach my $idx( 0 .. $#chain-1 ) { |
921 | my $source = $chain[$idx]; |
922 | my $target = $chain[$idx+1]; |
923 | $self->add_path( $source, $target, $sig.$self->ac_label ) |
924 | unless $self->has_path( $source, $target, $sig ); |
925 | } |
15d2d3df |
926 | } |
1f7aa795 |
927 | $wit->clear_path; |
928 | $wit->clear_uncorrected_path; |
e2902068 |
929 | } |
930 | |
910a0a6d |
931 | sub calculate_ranks { |
932 | my $self = shift; |
933 | # Walk a version of the graph where every node linked by a relationship |
934 | # edge is fundamentally the same node, and do a topological ranking on |
935 | # the nodes in this graph. |
c9bf3dbf |
936 | my $topo_graph = Graph->new(); |
910a0a6d |
937 | my %rel_containers; |
938 | my $rel_ctr = 0; |
939 | # Add the nodes |
940 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
941 | next if exists $rel_containers{$r->id}; |
910a0a6d |
942 | my @rels = $r->related_readings( 'colocated' ); |
943 | if( @rels ) { |
944 | # Make a relationship container. |
945 | push( @rels, $r ); |
c9bf3dbf |
946 | my $rn = 'rel_container_' . $rel_ctr++; |
947 | $topo_graph->add_vertex( $rn ); |
910a0a6d |
948 | foreach( @rels ) { |
3a2ebbf4 |
949 | $rel_containers{$_->id} = $rn; |
910a0a6d |
950 | } |
951 | } else { |
952 | # Add a new node to mirror the old node. |
3a2ebbf4 |
953 | $rel_containers{$r->id} = $r->id; |
954 | $topo_graph->add_vertex( $r->id ); |
910a0a6d |
955 | } |
4a8828f0 |
956 | } |
3a1f2523 |
957 | |
3a2ebbf4 |
958 | # Add the edges. |
910a0a6d |
959 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
960 | foreach my $n ( $self->sequence->successors( $r->id ) ) { |
961 | my( $tfrom, $tto ) = ( $rel_containers{$r->id}, |
962 | $rel_containers{$n} ); |
49d4f2ac |
963 | $DB::single = 1 unless $tfrom && $tto; |
3a2ebbf4 |
964 | $topo_graph->add_edge( $tfrom, $tto ); |
910a0a6d |
965 | } |
966 | } |
967 | |
968 | # Now do the rankings, starting with the start node. |
3a2ebbf4 |
969 | my $topo_start = $rel_containers{$self->start->id}; |
c9bf3dbf |
970 | my $node_ranks = { $topo_start => 0 }; |
910a0a6d |
971 | my @curr_origin = ( $topo_start ); |
972 | # A little iterative function. |
973 | while( @curr_origin ) { |
c9bf3dbf |
974 | @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin ); |
910a0a6d |
975 | } |
976 | # Transfer our rankings from the topological graph to the real one. |
977 | foreach my $r ( $self->readings ) { |
3a2ebbf4 |
978 | if( defined $node_ranks->{$rel_containers{$r->id}} ) { |
979 | $r->rank( $node_ranks->{$rel_containers{$r->id}} ); |
67da8d6c |
980 | } else { |
981 | $DB::single = 1; |
3a2ebbf4 |
982 | die "No rank calculated for node " . $r->id |
67da8d6c |
983 | . " - do you have a cycle in the graph?"; |
984 | } |
de51424a |
985 | } |
8e1394aa |
986 | } |
3a1f2523 |
987 | |
910a0a6d |
988 | sub _assign_rank { |
c9bf3dbf |
989 | my( $graph, $node_ranks, @current_nodes ) = @_; |
910a0a6d |
990 | # Look at each of the children of @current_nodes. If all the child's |
991 | # parents have a rank, assign it the highest rank + 1 and add it to |
c9bf3dbf |
992 | # @next_nodes. Otherwise skip it; we will return when the highest-ranked |
993 | # parent gets a rank. |
910a0a6d |
994 | my @next_nodes; |
995 | foreach my $c ( @current_nodes ) { |
c9bf3dbf |
996 | warn "Current reading $c has no rank!" |
997 | unless exists $node_ranks->{$c}; |
998 | # print STDERR "Looking at child of node $c, rank " |
999 | # . $node_ranks->{$c} . "\n"; |
1000 | foreach my $child ( $graph->successors( $c ) ) { |
1001 | next if exists $node_ranks->{$child}; |
910a0a6d |
1002 | my $highest_rank = -1; |
1003 | my $skip = 0; |
c9bf3dbf |
1004 | foreach my $parent ( $graph->predecessors( $child ) ) { |
1005 | if( exists $node_ranks->{$parent} ) { |
1006 | $highest_rank = $node_ranks->{$parent} |
1007 | if $highest_rank <= $node_ranks->{$parent}; |
910a0a6d |
1008 | } else { |
1009 | $skip = 1; |
1010 | last; |
1011 | } |
1012 | } |
1013 | next if $skip; |
c9bf3dbf |
1014 | my $c_rank = $highest_rank + 1; |
1015 | # print STDERR "Assigning rank $c_rank to node $child \n"; |
1016 | $node_ranks->{$child} = $c_rank; |
910a0a6d |
1017 | push( @next_nodes, $child ); |
1018 | } |
1019 | } |
1020 | return @next_nodes; |
4cdd82f1 |
1021 | } |
910a0a6d |
1022 | |
0e476982 |
1023 | # Another method to make up for rough collation methods. If the same reading |
1024 | # appears multiple times at the same rank, collapse the nodes. |
1025 | sub flatten_ranks { |
1026 | my $self = shift; |
1027 | my %unique_rank_rdg; |
1028 | foreach my $rdg ( $self->readings ) { |
1029 | next unless $rdg->has_rank; |
1030 | my $key = $rdg->rank . "||" . $rdg->text; |
1031 | if( exists $unique_rank_rdg{$key} ) { |
1032 | # Combine! |
0068967c |
1033 | # print STDERR "Combining readings at same rank: $key\n"; |
0e476982 |
1034 | $self->merge_readings( $unique_rank_rdg{$key}, $rdg ); |
1035 | } else { |
1036 | $unique_rank_rdg{$key} = $rdg; |
1037 | } |
1038 | } |
1039 | } |
1040 | |
1041 | |
fa954f4c |
1042 | ## Utility functions |
de51424a |
1043 | |
4a8828f0 |
1044 | # Return the string that joins together a list of witnesses for |
1045 | # display on a single path. |
4a8828f0 |
1046 | sub witnesses_of_label { |
de51424a |
1047 | my( $self, $label ) = @_; |
4a8828f0 |
1048 | my $regex = $self->wit_list_separator; |
de51424a |
1049 | my @answer = split( /\Q$regex\E/, $label ); |
1050 | return @answer; |
4a8828f0 |
1051 | } |
8e1394aa |
1052 | |
22222af9 |
1053 | =begin testing |
1054 | |
1055 | use Text::Tradition; |
1056 | |
1057 | my $cxfile = 't/data/Collatex-16.xml'; |
1058 | my $t = Text::Tradition->new( |
1059 | 'name' => 'inline', |
1060 | 'input' => 'CollateX', |
1061 | 'file' => $cxfile, |
1062 | ); |
1063 | my $c = $t->collation; |
1064 | |
1065 | is( $c->common_predecessor( $c->reading('n9'), $c->reading('n23') )->id, |
1066 | 'n20', "Found correct common predecessor" ); |
1067 | is( $c->common_successor( $c->reading('n9'), $c->reading('n23') )->id, |
1068 | '#END#', "Found correct common successor" ); |
1069 | |
1070 | is( $c->common_predecessor( $c->reading('n19'), $c->reading('n17') )->id, |
1071 | 'n16', "Found correct common predecessor for readings on same path" ); |
1072 | is( $c->common_successor( $c->reading('n21'), $c->reading('n26') )->id, |
1073 | '#END#', "Found correct common successor for readings on same path" ); |
1074 | |
1075 | =end testing |
1076 | |
1077 | =cut |
1078 | |
1079 | ## Return the closest reading that is a predecessor of both the given readings. |
1080 | sub common_predecessor { |
1081 | my $self = shift; |
1082 | return $self->common_in_path( @_, 'predecessors' ); |
1083 | } |
1084 | |
1085 | sub common_successor { |
1086 | my $self = shift; |
1087 | return $self->common_in_path( @_, 'successors' ); |
1088 | } |
1089 | |
1090 | sub common_in_path { |
1091 | my( $self, $r1, $r2, $dir ) = @_; |
1092 | my $iter = $r1->rank > $r2->rank ? $r1->rank : $r2->rank; |
1093 | $iter = $self->end->rank - $iter if $dir eq 'successors'; |
1094 | my @candidates; |
1095 | my @last_checked = ( $r1, $r2 ); |
1096 | my %all_seen; |
1097 | while( !@candidates ) { |
1098 | my @new_lc; |
1099 | foreach my $lc ( @last_checked ) { |
1100 | foreach my $p ( $lc->$dir ) { |
1101 | if( $all_seen{$p->id} ) { |
1102 | push( @candidates, $p ); |
1103 | } else { |
1104 | $all_seen{$p->id} = 1; |
1105 | push( @new_lc, $p ); |
1106 | } |
1107 | } |
1108 | } |
1109 | @last_checked = @new_lc; |
1110 | } |
1111 | my @answer = sort { $a->rank <=> $b->rank } @candidates; |
1112 | return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer ); |
1113 | } |
1114 | |
dd3b58b0 |
1115 | no Moose; |
1116 | __PACKAGE__->meta->make_immutable; |
e867486f |
1117 | |
1118 | =head1 BUGS / TODO |
1119 | |
1120 | =over |
1121 | |
0068967c |
1122 | =item * Think about making Relationship objects again |
e867486f |
1123 | |
1124 | =back |