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" );
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 n25 and n26
347 $c->merge_readings( 'n25', 'n26' );
348 ok( !$c->reading('n26'), "Reading n26 is gone" );
349 is( $c->reading('n25')->text, 'rood', "Reading n25 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;
598 # Check the arguments
600 return if $endrank && $startrank > $endrank;
601 return if $startrank > $self->end->rank;
603 if( defined $endrank ) {
604 return if $endrank < 0;
605 $endrank = undef if $endrank == $self->end->rank;
608 my $graph_name = $self->tradition->name;
609 $graph_name =~ s/[^\w\s]//g;
610 $graph_name = join( '_', split( /\s+/, $graph_name ) );
618 'fillcolor' => 'white',
623 'arrowhead' => 'open',
624 'color' => '#000000',
625 'fontcolor' => '#000000',
628 my $dot = sprintf( "digraph %s {\n", $graph_name );
629 $dot .= "\tgraph " . _dot_attr_string( \%graph_attrs ) . ";\n";
630 $dot .= "\tnode " . _dot_attr_string( \%node_attrs ) . ";\n";
632 # Output substitute start/end readings if necessary
634 $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n";
637 $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n";
639 if( $STRAIGHTENHACK ) {
641 $dot .= "\tsubgraph { rank=same \"#START#\" \"#SILENT#\" }\n";
642 $dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];"
644 my %used; # Keep track of the readings that actually appear in the graph
645 # Sort the readings by rank if we have ranks; this speeds layout.
646 my @all_readings = $self->end->has_rank
647 ? sort { $a->rank <=> $b->rank } $self->readings
649 # TODO Refrain from outputting lacuna nodes - just grey out the edges.
650 foreach my $reading ( @all_readings ) {
651 # Only output readings within our rank range.
652 next if $startrank && $reading->rank < $startrank;
653 next if $endrank && $reading->rank > $endrank;
654 $used{$reading->id} = 1;
655 # Need not output nodes without separate labels
656 next if $reading->id eq $reading->text;
658 my $label = $reading->text;
659 $label .= '-' if $reading->join_next;
660 $label = "-$label" if $reading->join_prior;
661 $label =~ s/\"/\\\"/g;
662 $rattrs->{'label'} = $label;
663 $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common;
664 $dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) );
667 # Add the real edges. Need to weight one edge per rank jump, in a
669 # my $weighted = $self->_add_edge_weights;
670 my @edges = $self->paths;
671 my( %substart, %subend );
672 foreach my $edge ( @edges ) {
673 # Do we need to output this edge?
674 if( $used{$edge->[0]} && $used{$edge->[1]} ) {
675 my $label = $self->_path_display_label( $self->path_witnesses( $edge ) );
676 my $variables = { %edge_attrs, 'label' => $label };
678 # Account for the rank gap if necessary
679 my $rank0 = $self->reading( $edge->[0] )->rank
680 if $self->reading( $edge->[0] )->has_rank;
681 my $rank1 = $self->reading( $edge->[1] )->rank
682 if $self->reading( $edge->[1] )->has_rank;
683 if( defined $rank0 && defined $rank1 && $rank1 - $rank0 > 1 ) {
684 $variables->{'minlen'} = $rank1 - $rank0;
687 # Add the calculated edge weights
688 # if( exists $weighted->{$edge->[0]}
689 # && $weighted->{$edge->[0]} eq $edge->[1] ) {
690 # # $variables->{'color'} = 'red';
691 # $variables->{'weight'} = 3.0;
694 # EXPERIMENTAL: make edge width reflect no. of witnesses
695 my $extrawidth = scalar( $self->path_witnesses( $edge ) ) * 0.2;
696 $variables->{'penwidth'} = $extrawidth + 0.8; # gives 1 for a single wit
698 my $varopts = _dot_attr_string( $variables );
699 $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n",
700 $edge->[0], $edge->[1], $varopts );
701 } elsif( $used{$edge->[0]} ) {
702 $subend{$edge->[0]} = 1;
703 } elsif( $used{$edge->[1]} ) {
704 $substart{$edge->[1]} = 1;
707 # Add substitute start and end edges if necessary
708 foreach my $node ( keys %substart ) {
709 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
710 my $variables = { %edge_attrs, 'label' => $witstr };
711 my $varopts = _dot_attr_string( $variables );
712 $dot .= "\t\"#SUBSTART#\" -> \"$node\" $varopts;";
714 foreach my $node ( keys %subend ) {
715 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
716 my $variables = { %edge_attrs, 'label' => $witstr };
717 my $varopts = _dot_attr_string( $variables );
718 $dot .= "\t\"$node\" -> \"#SUBEND#\" $varopts;";
721 if( $STRAIGHTENHACK ) {
722 $dot .= "\t\"#END#\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
729 sub _dot_attr_string {
732 foreach my $k ( sort keys %$hash ) {
734 push( @attrs, $k.'="'.$v.'"' );
736 return( '[ ' . join( ', ', @attrs ) . ' ]' );
739 sub _add_edge_weights {
741 # Walk the graph from START to END, choosing the successor node with
742 # the largest number of witness paths each time.
744 my $curr = $self->start->id;
745 my $ranked = $self->end->has_rank;
746 while( $curr ne $self->end->id ) {
747 my $rank = $ranked ? $self->reading( $curr )->rank : 0;
748 my @succ = sort { $self->path_witnesses( $curr, $a )
749 <=> $self->path_witnesses( $curr, $b ) }
750 $self->sequence->successors( $curr );
751 my $next = pop @succ;
752 my $nextrank = $ranked ? $self->reading( $next )->rank : 0;
753 # Try to avoid lacunae in the weighted path.
755 ( $self->reading( $next )->is_lacuna ||
756 $nextrank - $rank > 1 ) ){
759 $weighted->{$curr} = $next;
765 =head2 path_witnesses( $edge )
767 Returns the list of sigils whose witnesses are associated with the given edge.
768 The edge can be passed as either an array or an arrayref of ( $source, $target ).
773 my( $self, @edge ) = @_;
774 # If edge is an arrayref, cope.
775 if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) {
779 my @wits = keys %{$self->sequence->get_edge_attributes( @edge )};
783 sub _path_display_label {
786 my $maj = scalar( $self->tradition->witnesses ) * 0.6;
787 if( scalar @wits > $maj ) {
788 # TODO break out a.c. wits
791 return join( ', ', @wits );
795 =head2 witnesses_at_rank
797 Returns a list of witnesses that are not lacunose, for a given rank.
801 sub witnesses_at_rank {
802 my( $self, $rank ) = @_;
807 Returns a GraphML representation of the collation. The GraphML will contain
808 two graphs. The first expresses the attributes of the readings and the witness
809 paths that link them; the second expresses the relationships that link the
810 readings. This is the native transfer format for a tradition.
819 my $datafile = 't/data/florilegium_tei_ps.xml';
820 my $tradition = Text::Tradition->new( 'input' => 'TEI',
825 ok( $tradition, "Got a tradition object" );
826 is( scalar $tradition->witnesses, 13, "Found all witnesses" );
827 ok( $tradition->collation, "Tradition has a collation" );
829 my $c = $tradition->collation;
830 is( scalar $c->readings, $READINGS, "Collation has all readings" );
831 is( scalar $c->paths, $PATHS, "Collation has all paths" );
832 is( scalar $c->relationships, 0, "Collation has all relationships" );
834 # Add a few relationships
835 $c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
836 $c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
837 $c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );
839 # Now write it to GraphML and parse it again.
841 my $graphml = $c->as_graphml;
842 my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
843 is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
844 is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
845 is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
853 $self->calculate_ranks unless $self->_graphcalc_done;
856 my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns';
857 my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance';
858 my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' .
859 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd';
861 # Create the document and root node
862 my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" );
863 my $root = $graphml->createElementNS( $graphml_ns, 'graphml' );
864 $graphml->setDocumentElement( $root );
865 $root->setNamespace( $xsi_ns, 'xsi', 0 );
866 $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema );
868 # List of attribute types to save on our objects and their corresponding
874 'RelationshipType' => 'string',
875 'RelationshipScope' => 'string',
878 # List of attribute names *not* to save on our objects.
879 # We will also not save any attribute beginning with _.
881 map { $skipsave{$_} = 1 } qw/ cached_svg /;
883 # Add the data keys for the graph. Include an extra key 'version' for the
884 # GraphML output version.
887 my %graph_attributes = ( 'version' => 'string' );
888 # Graph attributes include those of Tradition and those of Collation.
890 my $tmeta = $self->tradition->meta;
891 my $cmeta = $self->meta;
892 map { $gattr_from{$_->name} = 'Tradition' } $tmeta->get_all_attributes;
893 map { $gattr_from{$_->name} = 'Collation' } $cmeta->get_all_attributes;
894 foreach my $attr ( ( $tmeta->get_all_attributes, $cmeta->get_all_attributes ) ) {
895 next if $attr->name =~ /^_/;
896 next if $skipsave{$attr->name};
897 next unless $save_types{$attr->type_constraint->name};
898 $graph_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
901 foreach my $datum ( sort keys %graph_attributes ) {
902 $graph_data_keys{$datum} = 'dg'.$gdi++;
903 my $key = $root->addNewChild( $graphml_ns, 'key' );
904 $key->setAttribute( 'attr.name', $datum );
905 $key->setAttribute( 'attr.type', $graph_attributes{$datum} );
906 $key->setAttribute( 'for', 'graph' );
907 $key->setAttribute( 'id', $graph_data_keys{$datum} );
910 # Add the data keys for reading nodes
911 my %reading_attributes;
912 my $rmeta = Text::Tradition::Collation::Reading->meta;
913 foreach my $attr( $rmeta->get_all_attributes ) {
914 next if $attr->name =~ /^_/;
915 next if $skipsave{$attr->name};
916 next unless $save_types{$attr->type_constraint->name};
917 $reading_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
921 foreach my $datum ( sort keys %reading_attributes ) {
922 $node_data_keys{$datum} = 'dn'.$ndi++;
923 my $key = $root->addNewChild( $graphml_ns, 'key' );
924 $key->setAttribute( 'attr.name', $datum );
925 $key->setAttribute( 'attr.type', $reading_attributes{$datum} );
926 $key->setAttribute( 'for', 'node' );
927 $key->setAttribute( 'id', $node_data_keys{$datum} );
930 # Add the data keys for edges, that is, paths and relationships. Path
931 # data does not come from a Moose class so is here manually.
934 my %edge_attributes = (
935 witness => 'string', # ID/label for a path
936 extra => 'boolean', # Path key
938 my @path_attributes = keys %edge_attributes; # track our manual additions
939 my $pmeta = Text::Tradition::Collation::Relationship->meta;
940 foreach my $attr( $pmeta->get_all_attributes ) {
941 next if $attr->name =~ /^_/;
942 next if $skipsave{$attr->name};
943 next unless $save_types{$attr->type_constraint->name};
944 $edge_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
946 foreach my $datum ( sort keys %edge_attributes ) {
947 $edge_data_keys{$datum} = 'de'.$edi++;
948 my $key = $root->addNewChild( $graphml_ns, 'key' );
949 $key->setAttribute( 'attr.name', $datum );
950 $key->setAttribute( 'attr.type', $edge_attributes{$datum} );
951 $key->setAttribute( 'for', 'edge' );
952 $key->setAttribute( 'id', $edge_data_keys{$datum} );
955 # Add the collation graph itself
956 my $sgraph = $root->addNewChild( $graphml_ns, 'graph' );
957 $sgraph->setAttribute( 'edgedefault', 'directed' );
958 $sgraph->setAttribute( 'id', $self->tradition->name );
959 $sgraph->setAttribute( 'parse.edgeids', 'canonical' );
960 $sgraph->setAttribute( 'parse.edges', scalar($self->paths) );
961 $sgraph->setAttribute( 'parse.nodeids', 'canonical' );
962 $sgraph->setAttribute( 'parse.nodes', scalar($self->readings) );
963 $sgraph->setAttribute( 'parse.order', 'nodesfirst' );
965 # Collation attribute data
966 foreach my $datum ( keys %graph_attributes ) {
968 if( $datum eq 'version' ) {
970 } elsif( $gattr_from{$datum} eq 'Tradition' ) {
971 $value = $self->tradition->$datum;
973 $value = $self->$datum;
975 _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value );
980 # Add our readings to the graph
981 foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) {
982 # Add to the main graph
983 my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' );
984 my $node_xmlid = 'n' . $node_ctr++;
985 $node_hash{ $n->id } = $node_xmlid;
986 $node_el->setAttribute( 'id', $node_xmlid );
987 foreach my $d ( keys %reading_attributes ) {
989 _add_graphml_data( $node_el, $node_data_keys{$d}, $nval )
994 # Add the path edges to the sequence graph
996 foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) {
997 # We add an edge in the graphml for every witness in $e.
998 foreach my $wit ( sort $self->path_witnesses( $e ) ) {
999 my( $id, $from, $to ) = ( 'e'.$edge_ctr++,
1000 $node_hash{ $e->[0] },
1001 $node_hash{ $e->[1] } );
1002 my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' );
1003 $edge_el->setAttribute( 'source', $from );
1004 $edge_el->setAttribute( 'target', $to );
1005 $edge_el->setAttribute( 'id', $id );
1007 # It's a witness path, so add the witness
1009 my $key = $edge_data_keys{'witness'};
1010 # Is this an ante-corr witness?
1011 my $aclabel = $self->ac_label;
1012 if( $wit =~ /^(.*)\Q$aclabel\E$/ ) {
1013 # Keep the base witness
1015 # ...and record that this is an 'extra' reading path
1016 _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel );
1018 _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base );
1022 # Add the relationship graph to the XML
1023 map { delete $edge_data_keys{$_} } @path_attributes;
1024 $self->relations->_as_graphml( $graphml_ns, $root, \%node_hash,
1025 $node_data_keys{'id'}, \%edge_data_keys );
1027 # Save and return the thing
1028 my $result = decode_utf8( $graphml->toString(1) );
1032 sub _add_graphml_data {
1033 my( $el, $key, $value ) = @_;
1034 return unless defined $value;
1035 my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
1036 $data_el->setAttribute( 'key', $key );
1037 $data_el->appendText( $value );
1042 Returns a CSV alignment table representation of the collation graph, one
1043 row per witness (or witness uncorrected.)
1049 my $table = $self->alignment_table;
1050 my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } );
1052 # Make the header row
1053 $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} );
1054 push( @result, decode_utf8( $csv->string ) );
1055 # Make the rest of the rows
1056 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1057 my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}};
1058 my @row = map { $_ ? $_->{'t'}->text : $_ } @rowobjs;
1059 $csv->combine( @row );
1060 push( @result, decode_utf8( $csv->string ) );
1062 return join( "\n", @result );
1065 =head2 alignment_table( $use_refs, $include_witnesses )
1067 Return a reference to an alignment table, in a slightly enhanced CollateX
1068 format which looks like this:
1070 $table = { alignment => [ { witness => "SIGIL",
1071 tokens => [ { t => "TEXT" }, ... ] },
1072 { witness => "SIG2",
1073 tokens => [ { t => "TEXT" }, ... ] },
1075 length => TEXTLEN };
1077 If $use_refs is set to 1, the reading object is returned in the table
1078 instead of READINGTEXT; if not, the text of the reading is returned.
1080 If $include_witnesses is set to a hashref, only the witnesses whose sigil
1081 keys have a true hash value will be included.
1085 sub alignment_table {
1087 $self->calculate_ranks() unless $self->_graphcalc_done;
1088 return $self->cached_table if $self->has_cached_table;
1090 # Make sure we can do this
1091 throw( "Need a linear graph in order to make an alignment table" )
1092 unless $self->linear;
1093 $self->calculate_ranks unless $self->end->has_rank;
1095 my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 };
1096 my @all_pos = ( 1 .. $self->end->rank - 1 );
1097 foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) {
1098 # print STDERR "Making witness row(s) for " . $wit->sigil . "\n";
1099 my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
1100 my @row = _make_witness_row( \@wit_path, \@all_pos );
1101 push( @{$table->{'alignment'}},
1102 { 'witness' => $wit->sigil, 'tokens' => \@row } );
1103 if( $wit->is_layered ) {
1104 my @wit_ac_path = $self->reading_sequence( $self->start, $self->end,
1105 $wit->sigil.$self->ac_label );
1106 my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos );
1107 push( @{$table->{'alignment'}},
1108 { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } );
1111 $self->cached_table( $table );
1115 sub _make_witness_row {
1116 my( $path, $positions ) = @_;
1118 map { $char_hash{$_} = undef } @$positions;
1120 foreach my $rdg ( @$path ) {
1121 my $rtext = $rdg->text;
1122 $rtext = '#LACUNA#' if $rdg->is_lacuna;
1123 print STDERR "rank " . $rdg->rank . "\n" if $debug;
1124 # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank;
1125 $char_hash{$rdg->rank} = { 't' => $rdg };
1127 my @row = map { $char_hash{$_} } @$positions;
1128 # Fill in lacuna markers for undef spots in the row
1129 my $last_el = shift @row;
1130 my @filled_row = ( $last_el );
1131 foreach my $el ( @row ) {
1132 # If we are using node reference, make the lacuna node appear many times
1133 # in the table. If not, use the lacuna tag.
1134 if( $last_el && $last_el->{'t'}->is_lacuna && !defined $el ) {
1137 push( @filled_row, $el );
1143 =head1 NAVIGATION METHODS
1145 =head2 reading_sequence( $first, $last, $sigil, $backup )
1147 Returns the ordered list of readings, starting with $first and ending
1148 with $last, for the witness given in $sigil. If a $backup sigil is
1149 specified (e.g. when walking a layered witness), it will be used wherever
1150 no $sigil path exists. If there is a base text reading, that will be
1151 used wherever no path exists for $sigil or $backup.
1155 # TODO Think about returning some lazy-eval iterator.
1156 # TODO Get rid of backup; we should know from what witness is whether we need it.
1158 sub reading_sequence {
1159 my( $self, $start, $end, $witness ) = @_;
1161 $witness = $self->baselabel unless $witness;
1162 my @readings = ( $start );
1165 while( $n && $n->id ne $end->id ) {
1166 if( exists( $seen{$n->id} ) ) {
1167 throw( "Detected loop for $witness at " . $n->id );
1171 my $next = $self->next_reading( $n, $witness );
1173 throw( "Did not find any path for $witness from reading " . $n->id );
1175 push( @readings, $next );
1178 # Check that the last reading is our end reading.
1179 my $last = $readings[$#readings];
1180 throw( "Last reading found from " . $start->text .
1181 " for witness $witness is not the end!" ) # TODO do we get this far?
1182 unless $last->id eq $end->id;
1187 =head2 next_reading( $reading, $sigil );
1189 Returns the reading that follows the given reading along the given witness
1195 # Return the successor via the corresponding path.
1197 my $answer = $self->_find_linked_reading( 'next', @_ );
1198 return undef unless $answer;
1199 return $self->reading( $answer );
1202 =head2 prior_reading( $reading, $sigil )
1204 Returns the reading that precedes the given reading along the given witness
1210 # Return the predecessor via the corresponding path.
1212 my $answer = $self->_find_linked_reading( 'prior', @_ );
1213 return $self->reading( $answer );
1216 sub _find_linked_reading {
1217 my( $self, $direction, $node, $path ) = @_;
1219 # Get a backup if we are dealing with a layered witness
1221 my $aclabel = $self->ac_label;
1222 if( $path && $path =~ /^(.*)\Q$aclabel\E$/ ) {
1226 my @linked_paths = $direction eq 'next'
1227 ? $self->sequence->edges_from( $node )
1228 : $self->sequence->edges_to( $node );
1229 return undef unless scalar( @linked_paths );
1231 # We have to find the linked path that contains all of the
1232 # witnesses supplied in $path.
1233 my( @path_wits, @alt_path_wits );
1234 @path_wits = sort( $self->_witnesses_of_label( $path ) ) if $path;
1235 @alt_path_wits = sort( $self->_witnesses_of_label( $alt_path ) ) if $alt_path;
1238 foreach my $le ( @linked_paths ) {
1239 if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) {
1242 my @le_wits = sort $self->path_witnesses( $le );
1243 if( _is_within( \@path_wits, \@le_wits ) ) {
1244 # This is the right path.
1245 return $direction eq 'next' ? $le->[1] : $le->[0];
1246 } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) {
1250 # Got this far? Return the alternate path if it exists.
1251 return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0]
1254 # Got this far? Return the base path if it exists.
1255 return $direction eq 'next' ? $base_le->[1] : $base_le->[0]
1258 # Got this far? We have no appropriate path.
1259 warn "Could not find $direction node from " . $node->id
1260 . " along path $path";
1266 my( $set1, $set2 ) = @_;
1267 my $ret = @$set1; # will be 0, i.e. false, if set1 is empty
1268 foreach my $el ( @$set1 ) {
1269 $ret = 0 unless grep { /^\Q$el\E$/ } @$set2;
1274 # Return the string that joins together a list of witnesses for
1275 # display on a single path.
1276 sub _witnesses_of_label {
1277 my( $self, $label ) = @_;
1278 my $regex = $self->wit_list_separator;
1279 my @answer = split( /\Q$regex\E/, $label );
1283 =head2 common_readings
1285 Returns the list of common readings in the graph (i.e. those readings that are
1286 shared by all non-lacunose witnesses.)
1290 sub common_readings {
1292 my @common = grep { $_->is_common } $self->readings;
1296 =head2 path_text( $sigil, $mainsigil [, $start, $end ] )
1298 Returns the text of a witness (plus its backup, if we are using a layer)
1299 as stored in the collation. The text is returned as a string, where the
1300 individual readings are joined with spaces and the meta-readings (e.g.
1301 lacunae) are omitted. Optional specification of $start and $end allows
1302 the generation of a subset of the witness text.
1307 my( $self, $wit, $start, $end ) = @_;
1308 $start = $self->start unless $start;
1309 $end = $self->end unless $end;
1310 my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit );
1313 foreach my $r ( @path ) {
1314 if( $r->join_prior || !$last || $last->join_next ) {
1315 $pathtext .= $r->text;
1317 $pathtext .= ' ' . $r->text;
1324 =head1 INITIALIZATION METHODS
1326 These are mostly for use by parsers.
1328 =head2 make_witness_path( $witness )
1330 Link the array of readings contained in $witness->path (and in
1331 $witness->uncorrected_path if it exists) into collation paths.
1332 Clear out the arrays when finished.
1334 =head2 make_witness_paths
1336 Call make_witness_path for all witnesses in the tradition.
1340 # For use when a collation is constructed from a base text and an apparatus.
1341 # We have the sequences of readings and just need to add path edges.
1342 # When we are done, clear out the witness path attributes, as they are no
1344 # TODO Find a way to replace the witness path attributes with encapsulated functions?
1346 sub make_witness_paths {
1348 foreach my $wit ( $self->tradition->witnesses ) {
1349 # print STDERR "Making path for " . $wit->sigil . "\n";
1350 $self->make_witness_path( $wit );
1354 sub make_witness_path {
1355 my( $self, $wit ) = @_;
1356 my @chain = @{$wit->path};
1357 my $sig = $wit->sigil;
1358 foreach my $idx ( 0 .. $#chain-1 ) {
1359 $self->add_path( $chain[$idx], $chain[$idx+1], $sig );
1361 if( $wit->is_layered ) {
1362 @chain = @{$wit->uncorrected_path};
1363 foreach my $idx( 0 .. $#chain-1 ) {
1364 my $source = $chain[$idx];
1365 my $target = $chain[$idx+1];
1366 $self->add_path( $source, $target, $sig.$self->ac_label )
1367 unless $self->has_path( $source, $target, $sig );
1371 $wit->clear_uncorrected_path;
1374 =head2 calculate_ranks
1376 Calculate the reading ranks (that is, their aligned positions relative
1377 to each other) for the graph. This can only be called on linear collations.
1381 use Text::Tradition;
1383 my $cxfile = 't/data/Collatex-16.xml';
1384 my $t = Text::Tradition->new(
1386 'input' => 'CollateX',
1389 my $c = $t->collation;
1392 my $table = $c->alignment_table;
1393 ok( $c->has_cached_table, "Alignment table was cached" );
1394 is( $c->alignment_table, $table, "Cached table returned upon second call" );
1395 $c->calculate_ranks;
1396 is( $c->alignment_table, $table, "Cached table retained with no rank change" );
1397 $c->add_relationship( 'n9', 'n23', { 'type' => 'spelling' } );
1398 isnt( $c->alignment_table, $table, "Alignment table changed after relationship add" );
1404 sub calculate_ranks {
1406 # Save the existing ranks, in case we need to invalidate the cached SVG.
1408 # Walk a version of the graph where every node linked by a relationship
1409 # edge is fundamentally the same node, and do a topological ranking on
1410 # the nodes in this graph.
1411 my $topo_graph = Graph->new();
1415 foreach my $r ( $self->readings ) {
1416 next if exists $rel_containers{$r->id};
1417 my @rels = $r->related_readings( 'colocated' );
1419 # Make a relationship container.
1421 my $rn = 'rel_container_' . $rel_ctr++;
1422 $topo_graph->add_vertex( $rn );
1424 $rel_containers{$_->id} = $rn;
1427 # Add a new node to mirror the old node.
1428 $rel_containers{$r->id} = $r->id;
1429 $topo_graph->add_vertex( $r->id );
1434 foreach my $r ( $self->readings ) {
1435 $existing_ranks{$r} = $r->rank;
1436 foreach my $n ( $self->sequence->successors( $r->id ) ) {
1437 my( $tfrom, $tto ) = ( $rel_containers{$r->id},
1438 $rel_containers{$n} );
1439 # $DB::single = 1 unless $tfrom && $tto;
1440 $topo_graph->add_edge( $tfrom, $tto );
1444 # Now do the rankings, starting with the start node.
1445 my $topo_start = $rel_containers{$self->start->id};
1446 my $node_ranks = { $topo_start => 0 };
1447 my @curr_origin = ( $topo_start );
1448 # A little iterative function.
1449 while( @curr_origin ) {
1450 @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin );
1452 # Transfer our rankings from the topological graph to the real one.
1453 foreach my $r ( $self->readings ) {
1454 if( defined $node_ranks->{$rel_containers{$r->id}} ) {
1455 $r->rank( $node_ranks->{$rel_containers{$r->id}} );
1457 # Die. Find the last rank we calculated.
1458 my @all_defined = sort { $node_ranks->{$rel_containers{$a->id}}
1459 <=> $node_ranks->{$rel_containers{$b->id}} }
1461 my $last = pop @all_defined;
1462 throw( "Ranks not calculated after $last - do you have a cycle in the graph?" );
1465 # Do we need to invalidate the cached data?
1466 if( $self->has_cached_svg || $self->has_cached_table ) {
1467 foreach my $r ( $self->readings ) {
1468 next if defined( $existing_ranks{$r} )
1469 && $existing_ranks{$r} == $r->rank;
1470 # Something has changed, so clear the cache
1471 $self->_clear_cache;
1472 # ...and recalculate the common readings.
1473 $self->calculate_common_readings();
1477 # The graph calculation information is now up to date.
1478 $self->_graphcalc_done(1);
1482 my( $graph, $node_ranks, @current_nodes ) = @_;
1483 # Look at each of the children of @current_nodes. If all the child's
1484 # parents have a rank, assign it the highest rank + 1 and add it to
1485 # @next_nodes. Otherwise skip it; we will return when the highest-ranked
1486 # parent gets a rank.
1488 foreach my $c ( @current_nodes ) {
1489 warn "Current reading $c has no rank!"
1490 unless exists $node_ranks->{$c};
1491 # print STDERR "Looking at child of node $c, rank "
1492 # . $node_ranks->{$c} . "\n";
1493 foreach my $child ( $graph->successors( $c ) ) {
1494 next if exists $node_ranks->{$child};
1495 my $highest_rank = -1;
1497 foreach my $parent ( $graph->predecessors( $child ) ) {
1498 if( exists $node_ranks->{$parent} ) {
1499 $highest_rank = $node_ranks->{$parent}
1500 if $highest_rank <= $node_ranks->{$parent};
1507 my $c_rank = $highest_rank + 1;
1508 # print STDERR "Assigning rank $c_rank to node $child \n";
1509 $node_ranks->{$child} = $c_rank;
1510 push( @next_nodes, $child );
1518 $self->wipe_svg if $self->has_cached_svg;
1519 $self->wipe_table if $self->has_cached_table;
1523 =head2 flatten_ranks
1525 A convenience method for parsing collation data. Searches the graph for readings
1526 with the same text at the same rank, and merges any that are found.
1532 my %unique_rank_rdg;
1533 foreach my $rdg ( $self->readings ) {
1534 next unless $rdg->has_rank;
1535 my $key = $rdg->rank . "||" . $rdg->text;
1536 if( exists $unique_rank_rdg{$key} ) {
1538 # print STDERR "Combining readings at same rank: $key\n";
1539 $self->merge_readings( $unique_rank_rdg{$key}, $rdg );
1540 # TODO see if this now makes a common point.
1542 $unique_rank_rdg{$key} = $rdg;
1548 =head2 calculate_common_readings
1550 Goes through the graph identifying the readings that appear in every witness
1551 (apart from those with lacunae at that spot.) Marks them as common and returns
1556 use Text::Tradition;
1558 my $cxfile = 't/data/Collatex-16.xml';
1559 my $t = Text::Tradition->new(
1561 'input' => 'CollateX',
1564 my $c = $t->collation;
1566 my @common = $c->calculate_common_readings();
1567 is( scalar @common, 8, "Found correct number of common readings" );
1568 my @marked = sort $c->common_readings();
1569 is( scalar @common, 8, "All common readings got marked as such" );
1570 my @expected = qw/ n1 n12 n16 n19 n20 n5 n6 n7 /;
1571 is_deeply( \@marked, \@expected, "Found correct list of common readings" );
1577 sub calculate_common_readings {
1580 map { $_->is_common( 0 ) } $self->readings;
1581 # Implicitly calls calculate_ranks
1582 my $table = $self->alignment_table;
1583 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1584 my @row = map { $_->{'tokens'}->[$idx]
1585 ? $_->{'tokens'}->[$idx]->{'t'} : '' }
1586 @{$table->{'alignment'}};
1588 foreach my $r ( @row ) {
1590 $hash{$r->id} = $r unless $r->is_meta;
1592 $hash{'UNDEF'} = $r;
1595 if( keys %hash == 1 && !exists $hash{'UNDEF'} ) {
1596 my( $r ) = values %hash;
1598 push( @common, $r );
1604 =head2 text_from_paths
1606 Calculate the text array for all witnesses from the path, for later consistency
1607 checking. Only to be used if there is no non-graph-based way to know the
1612 sub text_from_paths {
1614 foreach my $wit ( $self->tradition->witnesses ) {
1615 my @text = split( /\s+/,
1616 $self->reading_sequence( $self->start, $self->end, $wit->sigil ) );
1617 $wit->text( \@text );
1618 if( $wit->is_layered ) {
1619 my @uctext = split( /\s+/,
1620 $self->reading_sequence( $self->start, $self->end,
1621 $wit->sigil.$self->ac_label ) );
1622 $wit->text( \@uctext );
1627 =head1 UTILITY FUNCTIONS
1629 =head2 common_predecessor( $reading_a, $reading_b )
1631 Find the last reading that occurs in sequence before both the given readings.
1633 =head2 common_successor( $reading_a, $reading_b )
1635 Find the first reading that occurs in sequence after both the given readings.
1639 use Text::Tradition;
1641 my $cxfile = 't/data/Collatex-16.xml';
1642 my $t = Text::Tradition->new(
1644 'input' => 'CollateX',
1647 my $c = $t->collation;
1649 is( $c->common_predecessor( 'n9', 'n23' )->id,
1650 'n20', "Found correct common predecessor" );
1651 is( $c->common_successor( 'n9', 'n23' )->id,
1652 '#END#', "Found correct common successor" );
1654 is( $c->common_predecessor( 'n19', 'n17' )->id,
1655 'n16', "Found correct common predecessor for readings on same path" );
1656 is( $c->common_successor( 'n21', 'n26' )->id,
1657 '#END#', "Found correct common successor for readings on same path" );
1663 ## Return the closest reading that is a predecessor of both the given readings.
1664 sub common_predecessor {
1666 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1667 return $self->_common_in_path( $r1, $r2, 'predecessors' );
1670 sub common_successor {
1672 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1673 return $self->_common_in_path( $r1, $r2, 'successors' );
1676 sub _common_in_path {
1677 my( $self, $r1, $r2, $dir ) = @_;
1678 my $iter = $r1->rank > $r2->rank ? $r1->rank : $r2->rank;
1679 $iter = $self->end->rank - $iter if $dir eq 'successors';
1681 my @last_checked = ( $r1, $r2 );
1683 while( !@candidates ) {
1685 foreach my $lc ( @last_checked ) {
1686 foreach my $p ( $lc->$dir ) {
1687 if( $all_seen{$p->id} ) {
1688 push( @candidates, $p );
1690 $all_seen{$p->id} = 1;
1691 push( @new_lc, $p );
1695 @last_checked = @new_lc;
1697 my @answer = sort { $a->rank <=> $b->rank } @candidates;
1698 return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer );
1702 Text::Tradition::Error->throw(
1703 'ident' => 'Collation error',
1709 __PACKAGE__->meta->make_immutable;
1713 This package is free software and is provided "as is" without express
1714 or implied warranty. You can redistribute it and/or modify it under
1715 the same terms as Perl itself.
1719 Tara L Andrews E<lt>aurum@cpan.orgE<gt>