1 package Text::Tradition::Collation;
3 use Encode qw( decode_utf8 );
7 use IPC::Run qw( run binary );
9 use Text::Tradition::Collation::Reading;
10 use Text::Tradition::Collation::RelationshipStore;
11 use Text::Tradition::Error;
13 use XML::LibXML::XPathContext;
19 default => sub { Graph->new() },
27 isa => 'Text::Tradition::Collation::RelationshipStore',
29 relationships => 'relationships',
30 related_readings => 'related_readings',
31 get_relationship => 'get_relationship',
32 del_relationship => 'del_relationship',
34 writer => '_set_relations',
39 isa => 'Text::Tradition',
44 isa => 'HashRef[Text::Tradition::Collation::Reading]',
48 _add_reading => 'set',
49 del_reading => 'delete',
50 has_reading => 'exists',
53 default => sub { {} },
56 has 'wit_list_separator' => (
65 default => 'base text',
88 isa => 'Text::Tradition::Collation::Reading',
89 writer => '_set_start',
95 isa => 'Text::Tradition::Collation::Reading',
100 has 'cached_svg' => (
103 predicate => 'has_cached_svg',
104 clearer => 'wipe_svg',
107 has 'cached_table' => (
110 predicate => 'has_cached_table',
111 clearer => 'wipe_table',
114 has '_graphcalc_done' => (
122 Text::Tradition::Collation - a software model for a text collation
127 my $t = Text::Tradition->new(
128 'name' => 'this is a text',
130 'file' => '/path/to/tei_parallel_seg_file.xml' );
132 my $c = $t->collation;
133 my @readings = $c->readings;
134 my @paths = $c->paths;
135 my @relationships = $c->relationships;
137 my $svg_variant_graph = $t->collation->as_svg();
141 Text::Tradition is a library for representation and analysis of collated
142 texts, particularly medieval ones. The Collation is the central feature of
143 a Tradition, where the text, its sequence of readings, and its relationships
144 between readings are actually kept.
150 The constructor. Takes a hash or hashref of the following arguments:
154 =item * tradition - The Text::Tradition object to which the collation
157 =item * linear - Whether the collation should be linear; that is, whether
158 transposed readings should be treated as two linked readings rather than one,
159 and therefore whether the collation graph is acyclic. Defaults to true.
161 =item * baselabel - The default label for the path taken by a base text
162 (if any). Defaults to 'base text'.
164 =item * wit_list_separator - The string to join a list of witnesses for
165 purposes of making labels in display graphs. Defaults to ', '.
167 =item * ac_label - The extra label to tack onto a witness sigil when
168 representing another layer of path for the given witness - that is, when
169 a text has more than one possible reading due to scribal corrections or
170 the like. Defaults to ' (a.c.)'.
172 =item * wordsep - The string used to separate words in the original text.
183 =head2 wit_list_separator
191 Simple accessors for collation attributes.
195 The meta-reading at the start of every witness path.
199 The meta-reading at the end of every witness path.
203 Returns all Reading objects in the graph.
205 =head2 reading( $id )
207 Returns the Reading object corresponding to the given ID.
209 =head2 add_reading( $reading_args )
211 Adds a new reading object to the collation.
212 See L<Text::Tradition::Collation::Reading> for the available arguments.
214 =head2 del_reading( $object_or_id )
216 Removes the given reading from the collation, implicitly removing its
217 paths and relationships.
219 =head2 merge_readings( $main, $second, $concatenate, $with_str )
221 Merges the $second reading into the $main one. If $concatenate is true, then
222 the merged node will carry the text of both readings, concatenated with either
223 $with_str (if specified) or a sensible default (the empty string if the
224 appropriate 'join_*' flag is set on either reading, or else $self->wordsep.)
226 The first two arguments may be either readings or reading IDs.
228 =head2 has_reading( $id )
230 Predicate to see whether a given reading ID is in the graph.
232 =head2 reading_witnesses( $object_or_id )
234 Returns a list of sigils whose witnesses contain the reading.
238 Returns all reading paths within the document - that is, all edges in the
239 collation graph. Each path is an arrayref of [ $source, $target ] reading IDs.
241 =head2 add_path( $source, $target, $sigil )
243 Links the given readings in the collation in sequence, under the given witness
244 sigil. The readings may be specified by object or ID.
246 =head2 del_path( $source, $target, $sigil )
248 Links the given readings in the collation in sequence, under the given witness
249 sigil. The readings may be specified by object or ID.
251 =head2 has_path( $source, $target );
253 Returns true if the two readings are linked in sequence in any witness.
254 The readings may be specified by object or ID.
258 Returns all Relationship objects in the collation.
260 =head2 add_relationship( $reading, $other_reading, $options )
262 Adds a new relationship of the type given in $options between the two readings,
263 which may be specified by object or ID. Returns a value of ( $status, @vectors)
264 where $status is true on success, and @vectors is a list of relationship edges
265 that were ultimately added.
266 See L<Text::Tradition::Collation::Relationship> for the available options.
272 $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) );
273 $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) );
274 $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) );
277 ### Reading construct/destruct functions
280 my( $self, $reading ) = @_;
281 unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
282 my %args = %$reading;
283 $reading = Text::Tradition::Collation::Reading->new(
284 'collation' => $self,
287 # First check to see if a reading with this ID exists.
288 if( $self->reading( $reading->id ) ) {
289 throw( "Collation already has a reading with id " . $reading->id );
291 $self->_graphcalc_done(0);
292 $self->_add_reading( $reading->id => $reading );
293 # Once the reading has been added, put it in both graphs.
294 $self->sequence->add_vertex( $reading->id );
295 $self->relations->add_reading( $reading->id );
299 around del_reading => sub {
304 if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
307 # Remove the reading from the graphs.
308 $self->_graphcalc_done(0);
309 $self->_clear_cache; # Explicitly clear caches to GC the reading
310 $self->sequence->delete_vertex( $arg );
311 $self->relations->delete_reading( $arg );
314 $self->$orig( $arg );
321 my $cxfile = 't/data/Collatex-16.xml';
322 my $t = Text::Tradition->new(
324 'input' => 'CollateX',
327 my $c = $t->collation;
329 my $rno = scalar $c->readings;
330 # Split n21 for testing purposes
331 my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } );
332 my $old_r = $c->reading( 'n21' );
333 $old_r->alter_text( 'to' );
334 $c->del_path( 'n20', 'n21', 'A' );
335 $c->add_path( 'n20', 'n21p0', 'A' );
336 $c->add_path( 'n21p0', 'n21', 'A' );
338 ok( $c->reading( 'n21p0' ), "New reading exists" );
339 is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" );
341 # Combine n3 and n4 ( with his )
342 $c->merge_readings( 'n3', 'n4', 1 );
343 ok( !$c->reading('n4'), "Reading n4 is gone" );
344 is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" );
346 # Collapse n9 and n10 ( rood / root )
347 $c->merge_readings( 'n9', 'n10' );
348 ok( !$c->reading('n10'), "Reading n10 is gone" );
349 is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" );
351 # Combine n21 and n21p0
352 my $remaining = $c->reading('n21');
353 $remaining ||= $c->reading('n22'); # one of these should still exist
354 $c->merge_readings( 'n21p0', $remaining, 1 );
355 ok( !$c->reading('n21'), "Reading $remaining is gone" );
356 is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" );
365 # We only need the IDs for adding paths to the graph, not the reading
366 # objects themselves.
367 my( $kept, $deleted, $combine, $combine_char ) = $self->_stringify_args( @_ );
368 $self->_graphcalc_done(0);
370 # The kept reading should inherit the paths and the relationships
371 # of the deleted reading.
372 foreach my $path ( $self->sequence->edges_at( $deleted ) ) {
373 my @vector = ( $kept );
374 push( @vector, $path->[1] ) if $path->[0] eq $deleted;
375 unshift( @vector, $path->[0] ) if $path->[1] eq $deleted;
376 next if $vector[0] eq $vector[1]; # Don't add a self loop
377 my %wits = %{$self->sequence->get_edge_attributes( @$path )};
378 $self->sequence->add_edge( @vector );
379 my $fwits = $self->sequence->get_edge_attributes( @vector );
380 @wits{keys %$fwits} = values %$fwits;
381 $self->sequence->set_edge_attributes( @vector, \%wits );
383 $self->relations->merge_readings( $kept, $deleted, $combine_char );
385 # Do the deletion deed.
387 my $kept_obj = $self->reading( $kept );
388 my $del_obj = $self->reading( $deleted );
389 my $joinstr = $combine_char;
390 unless( defined $joinstr ) {
391 $joinstr = '' if $kept_obj->join_next || $del_obj->join_prior;
392 $joinstr = $self->wordsep unless defined $joinstr;
394 $kept_obj->alter_text( join( $joinstr, $kept_obj->text, $del_obj->text ) );
396 $self->del_reading( $deleted );
400 # Helper function for manipulating the graph.
401 sub _stringify_args {
402 my( $self, $first, $second, @args ) = @_;
404 if ref( $first ) eq 'Text::Tradition::Collation::Reading';
405 $second = $second->id
406 if ref( $second ) eq 'Text::Tradition::Collation::Reading';
407 return( $first, $second, @args );
410 # Helper function for manipulating the graph.
411 sub _objectify_args {
412 my( $self, $first, $second, $arg ) = @_;
413 $first = $self->reading( $first )
414 unless ref( $first ) eq 'Text::Tradition::Collation::Reading';
415 $second = $self->reading( $second )
416 unless ref( $second ) eq 'Text::Tradition::Collation::Reading';
417 return( $first, $second, $arg );
424 # We only need the IDs for adding paths to the graph, not the reading
425 # objects themselves.
426 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
428 $self->_graphcalc_done(0);
429 # Connect the readings
430 $self->sequence->add_edge( $source, $target );
431 # Note the witness in question
432 $self->sequence->set_edge_attribute( $source, $target, $wit, 1 );
438 if( ref( $_[0] ) eq 'ARRAY' ) {
445 # We only need the IDs for adding paths to the graph, not the reading
446 # objects themselves.
447 my( $source, $target, $wit ) = $self->_stringify_args( @args );
449 $self->_graphcalc_done(0);
450 if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) {
451 $self->sequence->delete_edge_attribute( $source, $target, $wit );
453 unless( keys %{$self->sequence->get_edge_attributes( $source, $target )} ) {
454 $self->sequence->delete_edge( $source, $target );
459 # Extra graph-alike utility
462 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
463 return undef unless $self->sequence->has_edge( $source, $target );
464 return $self->sequence->has_edge_attribute( $source, $target, $wit );
467 =head2 clear_witness( @sigil_list )
469 Clear the given witnesses out of the collation entirely, removing references
470 to them in paths, and removing readings that belong only to them. Should only
471 be called via $tradition->del_witness.
476 my( $self, @sigils ) = @_;
478 $self->_graphcalc_done(0);
479 # Clear the witness(es) out of the paths
480 foreach my $e ( $self->paths ) {
481 foreach my $sig ( @sigils ) {
482 $self->del_path( $e, $sig );
486 # Clear out the newly unused readings
487 foreach my $r ( $self->readings ) {
488 unless( $self->reading_witnesses( $r ) ) {
489 $self->del_reading( $r );
494 sub add_relationship {
496 my( $source, $target, $opts ) = $self->_stringify_args( @_ );
497 my( @vectors ) = $self->relations->add_relationship( $source,
498 $self->reading( $source ), $target, $self->reading( $target ), $opts );
499 $self->_graphcalc_done(0);
503 around qw/ get_relationship del_relationship / => sub {
507 if( @args == 1 && ref( $args[0] ) eq 'ARRAY' ) {
510 my( $source, $target ) = $self->_stringify_args( @args );
511 $self->$orig( $source, $target );
514 =head2 reading_witnesses( $reading )
516 Return a list of sigils corresponding to the witnesses in which the reading appears.
520 sub reading_witnesses {
521 my( $self, $reading ) = @_;
522 # We need only check either the incoming or the outgoing edges; I have
523 # arbitrarily chosen "incoming". Thus, special-case the start node.
524 if( $reading eq $self->start ) {
525 return map { $_->sigil } $self->tradition->witnesses;
528 foreach my $e ( $self->sequence->edges_to( $reading ) ) {
529 my $wits = $self->sequence->get_edge_attributes( @$e );
530 @all_witnesses{ keys %$wits } = 1;
532 my $acstr = $self->ac_label;
533 foreach my $acwit ( grep { $_ =~ s/^(.*)\Q$acstr\E$/$1/ } keys %all_witnesses ) {
534 delete $all_witnesses{$acwit.$acstr} if exists $all_witnesses{$acwit};
536 return keys %all_witnesses;
539 =head1 OUTPUT METHODS
541 =head2 as_svg( \%options )
543 Returns an SVG string that represents the graph, via as_dot and graphviz.
544 See as_dot for a list of options. Must have GraphViz (dot) installed to run.
549 my( $self, $opts ) = @_;
550 throw( "Need GraphViz installed to output SVG" )
551 unless File::Which::which( 'dot' );
552 my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'};
553 $self->calculate_ranks() unless $self->_graphcalc_done;
554 if( !$self->has_cached_svg || $opts->{'recalc'} || $want_subgraph ) {
555 my @cmd = qw/dot -Tsvg/;
557 my $dotfile = File::Temp->new();
559 # $dotfile->unlink_on_destroy(0);
560 binmode $dotfile, ':utf8';
561 print $dotfile $self->as_dot( $opts );
562 push( @cmd, $dotfile->filename );
563 run( \@cmd, ">", binary(), \$svg );
564 $svg = decode_utf8( $svg );
565 $self->cached_svg( $svg ) unless $want_subgraph;
568 return $self->cached_svg;
573 =head2 as_dot( \%options )
575 Returns a string that is the collation graph expressed in dot
576 (i.e. GraphViz) format. Options include:
591 my( $self, $opts ) = @_;
592 my $startrank = $opts->{'from'} if $opts;
593 my $endrank = $opts->{'to'} if $opts;
594 my $color_common = $opts->{'color_common'} if $opts;
595 my $STRAIGHTENHACK = !$startrank && !$endrank && $self->end->rank
596 && $self->end->rank > 100;
597 $STRAIGHTENHACK = 1 if $opts->{'straight'}; # even for subgraphs or small graphs
599 # Check the arguments
601 return if $endrank && $startrank > $endrank;
602 return if $startrank > $self->end->rank;
604 if( defined $endrank ) {
605 return if $endrank < 0;
606 $endrank = undef if $endrank == $self->end->rank;
609 my $graph_name = $self->tradition->name;
610 $graph_name =~ s/[^\w\s]//g;
611 $graph_name = join( '_', split( /\s+/, $graph_name ) );
619 'fillcolor' => 'white',
624 'arrowhead' => 'open',
625 'color' => '#000000',
626 'fontcolor' => '#000000',
629 my $dot = sprintf( "digraph %s {\n", $graph_name );
630 $dot .= "\tgraph " . _dot_attr_string( \%graph_attrs ) . ";\n";
631 $dot .= "\tnode " . _dot_attr_string( \%node_attrs ) . ";\n";
633 # Output substitute start/end readings if necessary
635 $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n";
638 $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n";
640 if( $STRAIGHTENHACK ) {
642 my $startlabel = $startrank ? 'SUBSTART' : 'START';
643 $dot .= "\tsubgraph { rank=same \"#$startlabel#\" \"#SILENT#\" }\n";
644 $dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];"
646 my %used; # Keep track of the readings that actually appear in the graph
647 # Sort the readings by rank if we have ranks; this speeds layout.
648 my @all_readings = $self->end->has_rank
649 ? sort { $a->rank <=> $b->rank } $self->readings
651 # TODO Refrain from outputting lacuna nodes - just grey out the edges.
652 foreach my $reading ( @all_readings ) {
653 # Only output readings within our rank range.
654 next if $startrank && $reading->rank < $startrank;
655 next if $endrank && $reading->rank > $endrank;
656 $used{$reading->id} = 1;
657 # Need not output nodes without separate labels
658 next if $reading->id eq $reading->text;
660 my $label = $reading->text;
661 $label .= '-' if $reading->join_next;
662 $label = "-$label" if $reading->join_prior;
663 $label =~ s/\"/\\\"/g;
664 $rattrs->{'label'} = $label;
665 $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common;
666 $dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) );
669 # Add the real edges. Need to weight one edge per rank jump, in a
671 # my $weighted = $self->_add_edge_weights;
672 my @edges = $self->paths;
673 my( %substart, %subend );
674 foreach my $edge ( @edges ) {
675 # Do we need to output this edge?
676 if( $used{$edge->[0]} && $used{$edge->[1]} ) {
677 my $label = $self->_path_display_label( $self->path_witnesses( $edge ) );
678 my $variables = { %edge_attrs, 'label' => $label };
680 # Account for the rank gap if necessary
681 my $rank0 = $self->reading( $edge->[0] )->rank
682 if $self->reading( $edge->[0] )->has_rank;
683 my $rank1 = $self->reading( $edge->[1] )->rank
684 if $self->reading( $edge->[1] )->has_rank;
685 if( defined $rank0 && defined $rank1 && $rank1 - $rank0 > 1 ) {
686 $variables->{'minlen'} = $rank1 - $rank0;
689 # Add the calculated edge weights
690 # if( exists $weighted->{$edge->[0]}
691 # && $weighted->{$edge->[0]} eq $edge->[1] ) {
692 # # $variables->{'color'} = 'red';
693 # $variables->{'weight'} = 3.0;
696 # EXPERIMENTAL: make edge width reflect no. of witnesses
697 my $extrawidth = scalar( $self->path_witnesses( $edge ) ) * 0.2;
698 $variables->{'penwidth'} = $extrawidth + 0.8; # gives 1 for a single wit
700 my $varopts = _dot_attr_string( $variables );
701 $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n",
702 $edge->[0], $edge->[1], $varopts );
703 } elsif( $used{$edge->[0]} ) {
704 $subend{$edge->[0]} = 1;
705 } elsif( $used{$edge->[1]} ) {
706 $substart{$edge->[1]} = 1;
709 # Add substitute start and end edges if necessary
710 foreach my $node ( keys %substart ) {
711 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
712 my $variables = { %edge_attrs, 'label' => $witstr };
713 my $varopts = _dot_attr_string( $variables );
714 $dot .= "\t\"#SUBSTART#\" -> \"$node\" $varopts;";
716 foreach my $node ( keys %subend ) {
717 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
718 my $variables = { %edge_attrs, 'label' => $witstr };
719 my $varopts = _dot_attr_string( $variables );
720 $dot .= "\t\"$node\" -> \"#SUBEND#\" $varopts;";
723 if( $STRAIGHTENHACK ) {
724 my $endlabel = $endrank ? 'SUBEND' : 'END';
725 $dot .= "\t\"#$endlabel#\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
732 sub _dot_attr_string {
735 foreach my $k ( sort keys %$hash ) {
737 push( @attrs, $k.'="'.$v.'"' );
739 return( '[ ' . join( ', ', @attrs ) . ' ]' );
742 sub _add_edge_weights {
744 # Walk the graph from START to END, choosing the successor node with
745 # the largest number of witness paths each time.
747 my $curr = $self->start->id;
748 my $ranked = $self->end->has_rank;
749 while( $curr ne $self->end->id ) {
750 my $rank = $ranked ? $self->reading( $curr )->rank : 0;
751 my @succ = sort { $self->path_witnesses( $curr, $a )
752 <=> $self->path_witnesses( $curr, $b ) }
753 $self->sequence->successors( $curr );
754 my $next = pop @succ;
755 my $nextrank = $ranked ? $self->reading( $next )->rank : 0;
756 # Try to avoid lacunae in the weighted path.
758 ( $self->reading( $next )->is_lacuna ||
759 $nextrank - $rank > 1 ) ){
762 $weighted->{$curr} = $next;
768 =head2 path_witnesses( $edge )
770 Returns the list of sigils whose witnesses are associated with the given edge.
771 The edge can be passed as either an array or an arrayref of ( $source, $target ).
776 my( $self, @edge ) = @_;
777 # If edge is an arrayref, cope.
778 if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) {
782 my @wits = keys %{$self->sequence->get_edge_attributes( @edge )};
786 sub _path_display_label {
789 my $maj = scalar( $self->tradition->witnesses ) * 0.6;
790 if( scalar @wits > $maj ) {
791 # TODO break out a.c. wits
794 return join( ', ', @wits );
798 =head2 readings_at_rank( $rank )
800 Returns a list of readings at a given rank, taken from the alignment table.
804 sub readings_at_rank {
805 my( $self, $rank ) = @_;
806 my $table = $self->alignment_table;
807 # Table rank is real rank - 1.
808 my @elements = map { $_->{'tokens'}->[$rank-1] } @{$table->{'alignment'}};
810 foreach my $e ( @elements ) {
811 next unless ref( $e ) eq 'HASH';
812 next unless exists $e->{'t'};
813 $readings{$e->{'t'}->id} = $e->{'t'};
815 return values %readings;
820 Returns a GraphML representation of the collation. The GraphML will contain
821 two graphs. The first expresses the attributes of the readings and the witness
822 paths that link them; the second expresses the relationships that link the
823 readings. This is the native transfer format for a tradition.
832 my $datafile = 't/data/florilegium_tei_ps.xml';
833 my $tradition = Text::Tradition->new( 'input' => 'TEI',
838 ok( $tradition, "Got a tradition object" );
839 is( scalar $tradition->witnesses, 13, "Found all witnesses" );
840 ok( $tradition->collation, "Tradition has a collation" );
842 my $c = $tradition->collation;
843 is( scalar $c->readings, $READINGS, "Collation has all readings" );
844 is( scalar $c->paths, $PATHS, "Collation has all paths" );
845 is( scalar $c->relationships, 0, "Collation has all relationships" );
847 # Add a few relationships
848 $c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
849 $c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
850 $c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );
852 # Now write it to GraphML and parse it again.
854 my $graphml = $c->as_graphml;
855 my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
856 is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
857 is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
858 is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
866 $self->calculate_ranks unless $self->_graphcalc_done;
869 my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns';
870 my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance';
871 my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' .
872 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd';
874 # Create the document and root node
875 my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" );
876 my $root = $graphml->createElementNS( $graphml_ns, 'graphml' );
877 $graphml->setDocumentElement( $root );
878 $root->setNamespace( $xsi_ns, 'xsi', 0 );
879 $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema );
881 # List of attribute types to save on our objects and their corresponding
887 'RelationshipType' => 'string',
888 'RelationshipScope' => 'string',
891 # List of attribute names *not* to save on our objects.
892 # We will also not save any attribute beginning with _.
894 map { $skipsave{$_} = 1 } qw/ cached_svg /;
896 # Add the data keys for the graph. Include an extra key 'version' for the
897 # GraphML output version.
900 my %graph_attributes = ( 'version' => 'string' );
901 # Graph attributes include those of Tradition and those of Collation.
903 my $tmeta = $self->tradition->meta;
904 my $cmeta = $self->meta;
905 map { $gattr_from{$_->name} = 'Tradition' } $tmeta->get_all_attributes;
906 map { $gattr_from{$_->name} = 'Collation' } $cmeta->get_all_attributes;
907 foreach my $attr ( ( $tmeta->get_all_attributes, $cmeta->get_all_attributes ) ) {
908 next if $attr->name =~ /^_/;
909 next if $skipsave{$attr->name};
910 next unless $save_types{$attr->type_constraint->name};
911 $graph_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
914 foreach my $datum ( sort keys %graph_attributes ) {
915 $graph_data_keys{$datum} = 'dg'.$gdi++;
916 my $key = $root->addNewChild( $graphml_ns, 'key' );
917 $key->setAttribute( 'attr.name', $datum );
918 $key->setAttribute( 'attr.type', $graph_attributes{$datum} );
919 $key->setAttribute( 'for', 'graph' );
920 $key->setAttribute( 'id', $graph_data_keys{$datum} );
923 # Add the data keys for reading nodes
924 my %reading_attributes;
925 my $rmeta = Text::Tradition::Collation::Reading->meta;
926 foreach my $attr( $rmeta->get_all_attributes ) {
927 next if $attr->name =~ /^_/;
928 next if $skipsave{$attr->name};
929 next unless $save_types{$attr->type_constraint->name};
930 $reading_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
934 foreach my $datum ( sort keys %reading_attributes ) {
935 $node_data_keys{$datum} = 'dn'.$ndi++;
936 my $key = $root->addNewChild( $graphml_ns, 'key' );
937 $key->setAttribute( 'attr.name', $datum );
938 $key->setAttribute( 'attr.type', $reading_attributes{$datum} );
939 $key->setAttribute( 'for', 'node' );
940 $key->setAttribute( 'id', $node_data_keys{$datum} );
943 # Add the data keys for edges, that is, paths and relationships. Path
944 # data does not come from a Moose class so is here manually.
947 my %edge_attributes = (
948 witness => 'string', # ID/label for a path
949 extra => 'boolean', # Path key
951 my @path_attributes = keys %edge_attributes; # track our manual additions
952 my $pmeta = Text::Tradition::Collation::Relationship->meta;
953 foreach my $attr( $pmeta->get_all_attributes ) {
954 next if $attr->name =~ /^_/;
955 next if $skipsave{$attr->name};
956 next unless $save_types{$attr->type_constraint->name};
957 $edge_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
959 foreach my $datum ( sort keys %edge_attributes ) {
960 $edge_data_keys{$datum} = 'de'.$edi++;
961 my $key = $root->addNewChild( $graphml_ns, 'key' );
962 $key->setAttribute( 'attr.name', $datum );
963 $key->setAttribute( 'attr.type', $edge_attributes{$datum} );
964 $key->setAttribute( 'for', 'edge' );
965 $key->setAttribute( 'id', $edge_data_keys{$datum} );
968 # Add the collation graph itself
969 my $sgraph = $root->addNewChild( $graphml_ns, 'graph' );
970 $sgraph->setAttribute( 'edgedefault', 'directed' );
971 $sgraph->setAttribute( 'id', $self->tradition->name );
972 $sgraph->setAttribute( 'parse.edgeids', 'canonical' );
973 $sgraph->setAttribute( 'parse.edges', scalar($self->paths) );
974 $sgraph->setAttribute( 'parse.nodeids', 'canonical' );
975 $sgraph->setAttribute( 'parse.nodes', scalar($self->readings) );
976 $sgraph->setAttribute( 'parse.order', 'nodesfirst' );
978 # Collation attribute data
979 foreach my $datum ( keys %graph_attributes ) {
981 if( $datum eq 'version' ) {
983 } elsif( $gattr_from{$datum} eq 'Tradition' ) {
984 $value = $self->tradition->$datum;
986 $value = $self->$datum;
988 _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value );
993 # Add our readings to the graph
994 foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) {
995 # Add to the main graph
996 my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' );
997 my $node_xmlid = 'n' . $node_ctr++;
998 $node_hash{ $n->id } = $node_xmlid;
999 $node_el->setAttribute( 'id', $node_xmlid );
1000 foreach my $d ( keys %reading_attributes ) {
1002 _add_graphml_data( $node_el, $node_data_keys{$d}, $nval )
1007 # Add the path edges to the sequence graph
1009 foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) {
1010 # We add an edge in the graphml for every witness in $e.
1011 foreach my $wit ( sort $self->path_witnesses( $e ) ) {
1012 my( $id, $from, $to ) = ( 'e'.$edge_ctr++,
1013 $node_hash{ $e->[0] },
1014 $node_hash{ $e->[1] } );
1015 my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' );
1016 $edge_el->setAttribute( 'source', $from );
1017 $edge_el->setAttribute( 'target', $to );
1018 $edge_el->setAttribute( 'id', $id );
1020 # It's a witness path, so add the witness
1022 my $key = $edge_data_keys{'witness'};
1023 # Is this an ante-corr witness?
1024 my $aclabel = $self->ac_label;
1025 if( $wit =~ /^(.*)\Q$aclabel\E$/ ) {
1026 # Keep the base witness
1028 # ...and record that this is an 'extra' reading path
1029 _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel );
1031 _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base );
1035 # Add the relationship graph to the XML
1036 map { delete $edge_data_keys{$_} } @path_attributes;
1037 $self->relations->_as_graphml( $graphml_ns, $root, \%node_hash,
1038 $node_data_keys{'id'}, \%edge_data_keys );
1040 # Save and return the thing
1041 my $result = decode_utf8( $graphml->toString(1) );
1045 sub _add_graphml_data {
1046 my( $el, $key, $value ) = @_;
1047 return unless defined $value;
1048 my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
1049 $data_el->setAttribute( 'key', $key );
1050 $data_el->appendText( $value );
1055 Returns a CSV alignment table representation of the collation graph, one
1056 row per witness (or witness uncorrected.)
1062 my $table = $self->alignment_table;
1063 my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } );
1065 # Make the header row
1066 $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} );
1067 push( @result, decode_utf8( $csv->string ) );
1068 # Make the rest of the rows
1069 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1070 my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}};
1071 my @row = map { $_ ? $_->{'t'}->text : $_ } @rowobjs;
1072 $csv->combine( @row );
1073 push( @result, decode_utf8( $csv->string ) );
1075 return join( "\n", @result );
1078 =head2 alignment_table( $use_refs, $include_witnesses )
1080 Return a reference to an alignment table, in a slightly enhanced CollateX
1081 format which looks like this:
1083 $table = { alignment => [ { witness => "SIGIL",
1084 tokens => [ { t => "TEXT" }, ... ] },
1085 { witness => "SIG2",
1086 tokens => [ { t => "TEXT" }, ... ] },
1088 length => TEXTLEN };
1090 If $use_refs is set to 1, the reading object is returned in the table
1091 instead of READINGTEXT; if not, the text of the reading is returned.
1093 If $include_witnesses is set to a hashref, only the witnesses whose sigil
1094 keys have a true hash value will be included.
1098 sub alignment_table {
1100 $self->calculate_ranks() unless $self->_graphcalc_done;
1101 return $self->cached_table if $self->has_cached_table;
1103 # Make sure we can do this
1104 throw( "Need a linear graph in order to make an alignment table" )
1105 unless $self->linear;
1106 $self->calculate_ranks unless $self->end->has_rank;
1108 my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 };
1109 my @all_pos = ( 1 .. $self->end->rank - 1 );
1110 foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) {
1111 # print STDERR "Making witness row(s) for " . $wit->sigil . "\n";
1112 my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
1113 my @row = _make_witness_row( \@wit_path, \@all_pos );
1114 push( @{$table->{'alignment'}},
1115 { 'witness' => $wit->sigil, 'tokens' => \@row } );
1116 if( $wit->is_layered ) {
1117 my @wit_ac_path = $self->reading_sequence( $self->start, $self->end,
1118 $wit->sigil.$self->ac_label );
1119 my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos );
1120 push( @{$table->{'alignment'}},
1121 { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } );
1124 $self->cached_table( $table );
1128 sub _make_witness_row {
1129 my( $path, $positions ) = @_;
1131 map { $char_hash{$_} = undef } @$positions;
1133 foreach my $rdg ( @$path ) {
1134 my $rtext = $rdg->text;
1135 $rtext = '#LACUNA#' if $rdg->is_lacuna;
1136 print STDERR "rank " . $rdg->rank . "\n" if $debug;
1137 # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank;
1138 $char_hash{$rdg->rank} = { 't' => $rdg };
1140 my @row = map { $char_hash{$_} } @$positions;
1141 # Fill in lacuna markers for undef spots in the row
1142 my $last_el = shift @row;
1143 my @filled_row = ( $last_el );
1144 foreach my $el ( @row ) {
1145 # If we are using node reference, make the lacuna node appear many times
1146 # in the table. If not, use the lacuna tag.
1147 if( $last_el && $last_el->{'t'}->is_lacuna && !defined $el ) {
1150 push( @filled_row, $el );
1156 =head1 NAVIGATION METHODS
1158 =head2 reading_sequence( $first, $last, $sigil, $backup )
1160 Returns the ordered list of readings, starting with $first and ending
1161 with $last, for the witness given in $sigil. If a $backup sigil is
1162 specified (e.g. when walking a layered witness), it will be used wherever
1163 no $sigil path exists. If there is a base text reading, that will be
1164 used wherever no path exists for $sigil or $backup.
1168 # TODO Think about returning some lazy-eval iterator.
1169 # TODO Get rid of backup; we should know from what witness is whether we need it.
1171 sub reading_sequence {
1172 my( $self, $start, $end, $witness ) = @_;
1174 $witness = $self->baselabel unless $witness;
1175 my @readings = ( $start );
1178 while( $n && $n->id ne $end->id ) {
1179 if( exists( $seen{$n->id} ) ) {
1180 throw( "Detected loop for $witness at " . $n->id );
1184 my $next = $self->next_reading( $n, $witness );
1186 throw( "Did not find any path for $witness from reading " . $n->id );
1188 push( @readings, $next );
1191 # Check that the last reading is our end reading.
1192 my $last = $readings[$#readings];
1193 throw( "Last reading found from " . $start->text .
1194 " for witness $witness is not the end!" ) # TODO do we get this far?
1195 unless $last->id eq $end->id;
1200 =head2 next_reading( $reading, $sigil );
1202 Returns the reading that follows the given reading along the given witness
1208 # Return the successor via the corresponding path.
1210 my $answer = $self->_find_linked_reading( 'next', @_ );
1211 return undef unless $answer;
1212 return $self->reading( $answer );
1215 =head2 prior_reading( $reading, $sigil )
1217 Returns the reading that precedes the given reading along the given witness
1223 # Return the predecessor via the corresponding path.
1225 my $answer = $self->_find_linked_reading( 'prior', @_ );
1226 return $self->reading( $answer );
1229 sub _find_linked_reading {
1230 my( $self, $direction, $node, $path ) = @_;
1232 # Get a backup if we are dealing with a layered witness
1234 my $aclabel = $self->ac_label;
1235 if( $path && $path =~ /^(.*)\Q$aclabel\E$/ ) {
1239 my @linked_paths = $direction eq 'next'
1240 ? $self->sequence->edges_from( $node )
1241 : $self->sequence->edges_to( $node );
1242 return undef unless scalar( @linked_paths );
1244 # We have to find the linked path that contains all of the
1245 # witnesses supplied in $path.
1246 my( @path_wits, @alt_path_wits );
1247 @path_wits = sort( $self->_witnesses_of_label( $path ) ) if $path;
1248 @alt_path_wits = sort( $self->_witnesses_of_label( $alt_path ) ) if $alt_path;
1251 foreach my $le ( @linked_paths ) {
1252 if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) {
1255 my @le_wits = sort $self->path_witnesses( $le );
1256 if( _is_within( \@path_wits, \@le_wits ) ) {
1257 # This is the right path.
1258 return $direction eq 'next' ? $le->[1] : $le->[0];
1259 } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) {
1263 # Got this far? Return the alternate path if it exists.
1264 return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0]
1267 # Got this far? Return the base path if it exists.
1268 return $direction eq 'next' ? $base_le->[1] : $base_le->[0]
1271 # Got this far? We have no appropriate path.
1272 warn "Could not find $direction node from " . $node->id
1273 . " along path $path";
1279 my( $set1, $set2 ) = @_;
1280 my $ret = @$set1; # will be 0, i.e. false, if set1 is empty
1281 foreach my $el ( @$set1 ) {
1282 $ret = 0 unless grep { /^\Q$el\E$/ } @$set2;
1287 # Return the string that joins together a list of witnesses for
1288 # display on a single path.
1289 sub _witnesses_of_label {
1290 my( $self, $label ) = @_;
1291 my $regex = $self->wit_list_separator;
1292 my @answer = split( /\Q$regex\E/, $label );
1296 =head2 common_readings
1298 Returns the list of common readings in the graph (i.e. those readings that are
1299 shared by all non-lacunose witnesses.)
1303 sub common_readings {
1305 my @common = grep { $_->is_common } $self->readings;
1309 =head2 path_text( $sigil, $mainsigil [, $start, $end ] )
1311 Returns the text of a witness (plus its backup, if we are using a layer)
1312 as stored in the collation. The text is returned as a string, where the
1313 individual readings are joined with spaces and the meta-readings (e.g.
1314 lacunae) are omitted. Optional specification of $start and $end allows
1315 the generation of a subset of the witness text.
1320 my( $self, $wit, $start, $end ) = @_;
1321 $start = $self->start unless $start;
1322 $end = $self->end unless $end;
1323 my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit );
1326 foreach my $r ( @path ) {
1327 if( $r->join_prior || !$last || $last->join_next ) {
1328 $pathtext .= $r->text;
1330 $pathtext .= ' ' . $r->text;
1337 =head1 INITIALIZATION METHODS
1339 These are mostly for use by parsers.
1341 =head2 make_witness_path( $witness )
1343 Link the array of readings contained in $witness->path (and in
1344 $witness->uncorrected_path if it exists) into collation paths.
1345 Clear out the arrays when finished.
1347 =head2 make_witness_paths
1349 Call make_witness_path for all witnesses in the tradition.
1353 # For use when a collation is constructed from a base text and an apparatus.
1354 # We have the sequences of readings and just need to add path edges.
1355 # When we are done, clear out the witness path attributes, as they are no
1357 # TODO Find a way to replace the witness path attributes with encapsulated functions?
1359 sub make_witness_paths {
1361 foreach my $wit ( $self->tradition->witnesses ) {
1362 # print STDERR "Making path for " . $wit->sigil . "\n";
1363 $self->make_witness_path( $wit );
1367 sub make_witness_path {
1368 my( $self, $wit ) = @_;
1369 my @chain = @{$wit->path};
1370 my $sig = $wit->sigil;
1371 foreach my $idx ( 0 .. $#chain-1 ) {
1372 $self->add_path( $chain[$idx], $chain[$idx+1], $sig );
1374 if( $wit->is_layered ) {
1375 @chain = @{$wit->uncorrected_path};
1376 foreach my $idx( 0 .. $#chain-1 ) {
1377 my $source = $chain[$idx];
1378 my $target = $chain[$idx+1];
1379 $self->add_path( $source, $target, $sig.$self->ac_label )
1380 unless $self->has_path( $source, $target, $sig );
1384 $wit->clear_uncorrected_path;
1387 =head2 calculate_ranks
1389 Calculate the reading ranks (that is, their aligned positions relative
1390 to each other) for the graph. This can only be called on linear collations.
1394 use Text::Tradition;
1396 my $cxfile = 't/data/Collatex-16.xml';
1397 my $t = Text::Tradition->new(
1399 'input' => 'CollateX',
1402 my $c = $t->collation;
1405 my $table = $c->alignment_table;
1406 ok( $c->has_cached_table, "Alignment table was cached" );
1407 is( $c->alignment_table, $table, "Cached table returned upon second call" );
1408 $c->calculate_ranks;
1409 is( $c->alignment_table, $table, "Cached table retained with no rank change" );
1410 $c->add_relationship( 'n24', 'n23', { 'type' => 'spelling' } );
1411 isnt( $c->alignment_table, $table, "Alignment table changed after relationship add" );
1417 sub calculate_ranks {
1419 # Save the existing ranks, in case we need to invalidate the cached SVG.
1421 # Walk a version of the graph where every node linked by a relationship
1422 # edge is fundamentally the same node, and do a topological ranking on
1423 # the nodes in this graph.
1424 my $topo_graph = Graph->new();
1428 foreach my $r ( $self->readings ) {
1429 next if exists $rel_containers{$r->id};
1430 my @rels = $r->related_readings( 'colocated' );
1432 # Make a relationship container.
1434 my $rn = 'rel_container_' . $rel_ctr++;
1435 $topo_graph->add_vertex( $rn );
1437 $rel_containers{$_->id} = $rn;
1440 # Add a new node to mirror the old node.
1441 $rel_containers{$r->id} = $r->id;
1442 $topo_graph->add_vertex( $r->id );
1447 foreach my $r ( $self->readings ) {
1448 $existing_ranks{$r} = $r->rank;
1449 foreach my $n ( $self->sequence->successors( $r->id ) ) {
1450 my( $tfrom, $tto ) = ( $rel_containers{$r->id},
1451 $rel_containers{$n} );
1452 # $DB::single = 1 unless $tfrom && $tto;
1453 $topo_graph->add_edge( $tfrom, $tto );
1457 # Now do the rankings, starting with the start node.
1458 my $topo_start = $rel_containers{$self->start->id};
1459 my $node_ranks = { $topo_start => 0 };
1460 my @curr_origin = ( $topo_start );
1461 # A little iterative function.
1462 while( @curr_origin ) {
1463 @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin );
1465 # Transfer our rankings from the topological graph to the real one.
1466 foreach my $r ( $self->readings ) {
1467 if( defined $node_ranks->{$rel_containers{$r->id}} ) {
1468 $r->rank( $node_ranks->{$rel_containers{$r->id}} );
1470 # Die. Find the last rank we calculated.
1471 my @all_defined = sort { $node_ranks->{$rel_containers{$a->id}}
1472 <=> $node_ranks->{$rel_containers{$b->id}} }
1474 my $last = pop @all_defined;
1475 throw( "Ranks not calculated after $last - do you have a cycle in the graph?" );
1478 # Do we need to invalidate the cached data?
1479 if( $self->has_cached_svg || $self->has_cached_table ) {
1480 foreach my $r ( $self->readings ) {
1481 next if defined( $existing_ranks{$r} )
1482 && $existing_ranks{$r} == $r->rank;
1483 # Something has changed, so clear the cache
1484 $self->_clear_cache;
1485 # ...and recalculate the common readings.
1486 $self->calculate_common_readings();
1490 # The graph calculation information is now up to date.
1491 $self->_graphcalc_done(1);
1495 my( $graph, $node_ranks, @current_nodes ) = @_;
1496 # Look at each of the children of @current_nodes. If all the child's
1497 # parents have a rank, assign it the highest rank + 1 and add it to
1498 # @next_nodes. Otherwise skip it; we will return when the highest-ranked
1499 # parent gets a rank.
1501 foreach my $c ( @current_nodes ) {
1502 warn "Current reading $c has no rank!"
1503 unless exists $node_ranks->{$c};
1504 # print STDERR "Looking at child of node $c, rank "
1505 # . $node_ranks->{$c} . "\n";
1506 foreach my $child ( $graph->successors( $c ) ) {
1507 next if exists $node_ranks->{$child};
1508 my $highest_rank = -1;
1510 foreach my $parent ( $graph->predecessors( $child ) ) {
1511 if( exists $node_ranks->{$parent} ) {
1512 $highest_rank = $node_ranks->{$parent}
1513 if $highest_rank <= $node_ranks->{$parent};
1520 my $c_rank = $highest_rank + 1;
1521 # print STDERR "Assigning rank $c_rank to node $child \n";
1522 $node_ranks->{$child} = $c_rank;
1523 push( @next_nodes, $child );
1531 $self->wipe_svg if $self->has_cached_svg;
1532 $self->wipe_table if $self->has_cached_table;
1536 =head2 flatten_ranks
1538 A convenience method for parsing collation data. Searches the graph for readings
1539 with the same text at the same rank, and merges any that are found.
1545 my %unique_rank_rdg;
1547 foreach my $rdg ( $self->readings ) {
1548 next unless $rdg->has_rank;
1549 my $key = $rdg->rank . "||" . $rdg->text;
1550 if( exists $unique_rank_rdg{$key} ) {
1552 # print STDERR "Combining readings at same rank: $key\n";
1554 $self->merge_readings( $unique_rank_rdg{$key}, $rdg );
1555 # TODO see if this now makes a common point.
1557 $unique_rank_rdg{$key} = $rdg;
1560 # If we merged readings, the ranks are still fine but the alignment
1561 # table is wrong. Wipe it.
1562 $self->wipe_table() if $changed;
1566 =head2 calculate_common_readings
1568 Goes through the graph identifying the readings that appear in every witness
1569 (apart from those with lacunae at that spot.) Marks them as common and returns
1574 use Text::Tradition;
1576 my $cxfile = 't/data/Collatex-16.xml';
1577 my $t = Text::Tradition->new(
1579 'input' => 'CollateX',
1582 my $c = $t->collation;
1584 my @common = $c->calculate_common_readings();
1585 is( scalar @common, 8, "Found correct number of common readings" );
1586 my @marked = sort $c->common_readings();
1587 is( scalar @common, 8, "All common readings got marked as such" );
1588 my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /;
1589 is_deeply( \@marked, \@expected, "Found correct list of common readings" );
1595 sub calculate_common_readings {
1598 map { $_->is_common( 0 ) } $self->readings;
1599 # Implicitly calls calculate_ranks
1600 my $table = $self->alignment_table;
1601 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1602 my @row = map { $_->{'tokens'}->[$idx]
1603 ? $_->{'tokens'}->[$idx]->{'t'} : '' }
1604 @{$table->{'alignment'}};
1606 foreach my $r ( @row ) {
1608 $hash{$r->id} = $r unless $r->is_meta;
1610 $hash{'UNDEF'} = $r;
1613 if( keys %hash == 1 && !exists $hash{'UNDEF'} ) {
1614 my( $r ) = values %hash;
1616 push( @common, $r );
1622 =head2 text_from_paths
1624 Calculate the text array for all witnesses from the path, for later consistency
1625 checking. Only to be used if there is no non-graph-based way to know the
1630 sub text_from_paths {
1632 foreach my $wit ( $self->tradition->witnesses ) {
1633 my @readings = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
1635 foreach my $r ( @readings ) {
1636 next if $r->is_meta;
1637 push( @text, $r->text );
1639 $wit->text( \@text );
1640 if( $wit->is_layered ) {
1641 my @ucrdgs = $self->reading_sequence( $self->start, $self->end,
1642 $wit->sigil.$self->ac_label );
1644 foreach my $r ( @ucrdgs ) {
1645 next if $r->is_meta;
1646 push( @uctext, $r->text );
1648 $wit->layertext( \@uctext );
1653 =head1 UTILITY FUNCTIONS
1655 =head2 common_predecessor( $reading_a, $reading_b )
1657 Find the last reading that occurs in sequence before both the given readings.
1659 =head2 common_successor( $reading_a, $reading_b )
1661 Find the first reading that occurs in sequence after both the given readings.
1665 use Text::Tradition;
1667 my $cxfile = 't/data/Collatex-16.xml';
1668 my $t = Text::Tradition->new(
1670 'input' => 'CollateX',
1673 my $c = $t->collation;
1675 is( $c->common_predecessor( 'n24', 'n23' )->id,
1676 'n20', "Found correct common predecessor" );
1677 is( $c->common_successor( 'n24', 'n23' )->id,
1678 '#END#', "Found correct common successor" );
1680 is( $c->common_predecessor( 'n19', 'n17' )->id,
1681 'n16', "Found correct common predecessor for readings on same path" );
1682 is( $c->common_successor( 'n21', 'n10' )->id,
1683 '#END#', "Found correct common successor for readings on same path" );
1689 ## Return the closest reading that is a predecessor of both the given readings.
1690 sub common_predecessor {
1692 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1693 return $self->_common_in_path( $r1, $r2, 'predecessors' );
1696 sub common_successor {
1698 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1699 return $self->_common_in_path( $r1, $r2, 'successors' );
1702 sub _common_in_path {
1703 my( $self, $r1, $r2, $dir ) = @_;
1704 my $iter = $r1->rank > $r2->rank ? $r1->rank : $r2->rank;
1705 $iter = $self->end->rank - $iter if $dir eq 'successors';
1707 my @last_checked = ( $r1, $r2 );
1709 while( !@candidates ) {
1711 foreach my $lc ( @last_checked ) {
1712 foreach my $p ( $lc->$dir ) {
1713 if( $all_seen{$p->id} ) {
1714 push( @candidates, $p );
1716 $all_seen{$p->id} = 1;
1717 push( @new_lc, $p );
1721 @last_checked = @new_lc;
1723 my @answer = sort { $a->rank <=> $b->rank } @candidates;
1724 return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer );
1728 Text::Tradition::Error->throw(
1729 'ident' => 'Collation error',
1735 __PACKAGE__->meta->make_immutable;
1739 This package is free software and is provided "as is" without express
1740 or implied warranty. You can redistribute it and/or modify it under
1741 the same terms as Perl itself.
1745 Tara L Andrews E<lt>aurum@cpan.orgE<gt>