1 package Text::Tradition::Collation;
3 use Encode qw( decode_utf8 );
6 use IPC::Run qw( run binary );
8 use Text::Tradition::Collation::Reading;
9 use Text::Tradition::Collation::RelationshipStore;
10 use Text::Tradition::Error;
12 use XML::LibXML::XPathContext;
18 default => sub { Graph->new() },
26 isa => 'Text::Tradition::Collation::RelationshipStore',
28 relationships => 'relationships',
29 related_readings => 'related_readings',
31 writer => '_set_relations',
36 isa => 'Text::Tradition',
41 isa => 'HashRef[Text::Tradition::Collation::Reading]',
45 _add_reading => 'set',
46 del_reading => 'delete',
47 has_reading => 'exists',
50 default => sub { {} },
53 has 'wit_list_separator' => (
62 default => 'base text',
79 isa => 'Text::Tradition::Collation::Reading',
80 writer => '_set_start',
86 isa => 'Text::Tradition::Collation::Reading',
94 predicate => 'has_cached_svg',
95 clearer => 'wipe_svg',
100 Text::Tradition::Collation - a software model for a text collation
105 my $t = Text::Tradition->new(
106 'name' => 'this is a text',
108 'file' => '/path/to/tei_parallel_seg_file.xml' );
110 my $c = $t->collation;
111 my @readings = $c->readings;
112 my @paths = $c->paths;
113 my @relationships = $c->relationships;
115 my $svg_variant_graph = $t->collation->as_svg();
119 Text::Tradition is a library for representation and analysis of collated
120 texts, particularly medieval ones. The Collation is the central feature of
121 a Tradition, where the text, its sequence of readings, and its relationships
122 between readings are actually kept.
128 The constructor. Takes a hash or hashref of the following arguments:
132 =item * tradition - The Text::Tradition object to which the collation
135 =item * linear - Whether the collation should be linear; that is, whether
136 transposed readings should be treated as two linked readings rather than one,
137 and therefore whether the collation graph is acyclic. Defaults to true.
139 =item * baselabel - The default label for the path taken by a base text
140 (if any). Defaults to 'base text'.
142 =item * wit_list_separator - The string to join a list of witnesses for
143 purposes of making labels in display graphs. Defaults to ', '.
145 =item * ac_label - The extra label to tack onto a witness sigil when
146 representing another layer of path for the given witness - that is, when
147 a text has more than one possible reading due to scribal corrections or
148 the like. Defaults to ' (a.c.)'.
158 =head2 wit_list_separator
164 Simple accessors for collation attributes.
168 The meta-reading at the start of every witness path.
172 The meta-reading at the end of every witness path.
176 Returns all Reading objects in the graph.
178 =head2 reading( $id )
180 Returns the Reading object corresponding to the given ID.
182 =head2 add_reading( $reading_args )
184 Adds a new reading object to the collation.
185 See L<Text::Tradition::Collation::Reading> for the available arguments.
187 =head2 del_reading( $object_or_id )
189 Removes the given reading from the collation, implicitly removing its
190 paths and relationships.
192 =head2 merge_readings( $main, $second )
194 Merges the $second reading into the $main one.
195 The arguments may be either readings or reading IDs.
197 =head2 has_reading( $id )
199 Predicate to see whether a given reading ID is in the graph.
201 =head2 reading_witnesses( $object_or_id )
203 Returns a list of sigils whose witnesses contain the reading.
207 Returns all reading paths within the document - that is, all edges in the
208 collation graph. Each path is an arrayref of [ $source, $target ] reading IDs.
210 =head2 add_path( $source, $target, $sigil )
212 Links the given readings in the collation in sequence, under the given witness
213 sigil. The readings may be specified by object or ID.
215 =head2 del_path( $source, $target, $sigil )
217 Links the given readings in the collation in sequence, under the given witness
218 sigil. The readings may be specified by object or ID.
220 =head2 has_path( $source, $target );
222 Returns true if the two readings are linked in sequence in any witness.
223 The readings may be specified by object or ID.
227 Returns all Relationship objects in the collation.
229 =head2 add_relationship( $reading, $other_reading, $options )
231 Adds a new relationship of the type given in $options between the two readings,
232 which may be specified by object or ID. Returns a value of ( $status, @vectors)
233 where $status is true on success, and @vectors is a list of relationship edges
234 that were ultimately added.
235 See L<Text::Tradition::Collation::Relationship> for the available options.
241 $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) );
242 $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) );
243 $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) );
246 ### Reading construct/destruct functions
249 my( $self, $reading ) = @_;
250 unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
251 my %args = %$reading;
252 $reading = Text::Tradition::Collation::Reading->new(
253 'collation' => $self,
256 # First check to see if a reading with this ID exists.
257 if( $self->reading( $reading->id ) ) {
258 throw( "Collation already has a reading with id " . $reading->id );
260 $self->_add_reading( $reading->id => $reading );
261 # Once the reading has been added, put it in both graphs.
262 $self->sequence->add_vertex( $reading->id );
263 $self->relations->add_reading( $reading->id );
267 around del_reading => sub {
272 if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
275 # Remove the reading from the graphs.
276 $self->sequence->delete_vertex( $arg );
277 $self->relations->delete_reading( $arg );
280 $self->$orig( $arg );
283 # merge_readings( $main, $to_be_deleted );
288 # We only need the IDs for adding paths to the graph, not the reading
289 # objects themselves.
290 my( $kept, $deleted, $combine_char ) = $self->_stringify_args( @_ );
292 # The kept reading should inherit the paths and the relationships
293 # of the deleted reading.
294 foreach my $path ( $self->sequence->edges_at( $deleted ) ) {
295 my @vector = ( $kept );
296 push( @vector, $path->[1] ) if $path->[0] eq $deleted;
297 unshift( @vector, $path->[0] ) if $path->[1] eq $deleted;
298 next if $vector[0] eq $vector[1]; # Don't add a self loop
299 my %wits = %{$self->sequence->get_edge_attributes( @$path )};
300 $self->sequence->add_edge( @vector );
301 my $fwits = $self->sequence->get_edge_attributes( @vector );
302 @wits{keys %$fwits} = values %$fwits;
303 $self->sequence->set_edge_attributes( @vector, \%wits );
305 $self->relations->merge_readings( $kept, $deleted, $combine_char );
307 # Do the deletion deed.
308 if( $combine_char ) {
309 my $kept_obj = $self->reading( $kept );
310 my $new_text = join( $combine_char, $kept_obj->text,
311 $self->reading( $deleted )->text );
312 $kept_obj->alter_text( $new_text );
314 $self->del_reading( $deleted );
318 # Helper function for manipulating the graph.
319 sub _stringify_args {
320 my( $self, $first, $second, $arg ) = @_;
322 if ref( $first ) eq 'Text::Tradition::Collation::Reading';
323 $second = $second->id
324 if ref( $second ) eq 'Text::Tradition::Collation::Reading';
325 return( $first, $second, $arg );
328 # Helper function for manipulating the graph.
329 sub _objectify_args {
330 my( $self, $first, $second, $arg ) = @_;
331 $first = $self->reading( $first )
332 unless ref( $first ) eq 'Text::Tradition::Collation::Reading';
333 $second = $self->reading( $second )
334 unless ref( $second ) eq 'Text::Tradition::Collation::Reading';
335 return( $first, $second, $arg );
342 # We only need the IDs for adding paths to the graph, not the reading
343 # objects themselves.
344 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
346 # Connect the readings
347 $self->sequence->add_edge( $source, $target );
348 # Note the witness in question
349 $self->sequence->set_edge_attribute( $source, $target, $wit, 1 );
355 if( ref( $_[0] ) eq 'ARRAY' ) {
362 # We only need the IDs for adding paths to the graph, not the reading
363 # objects themselves.
364 my( $source, $target, $wit ) = $self->_stringify_args( @args );
366 if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) {
367 $self->sequence->delete_edge_attribute( $source, $target, $wit );
369 unless( keys %{$self->sequence->get_edge_attributes( $source, $target )} ) {
370 $self->sequence->delete_edge( $source, $target );
375 # Extra graph-alike utility
378 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
379 return undef unless $self->sequence->has_edge( $source, $target );
380 return $self->sequence->has_edge_attribute( $source, $target, $wit );
383 =head2 clear_witness( @sigil_list )
385 Clear the given witnesses out of the collation entirely, removing references
386 to them in paths, and removing readings that belong only to them. Should only
387 be called via $tradition->del_witness.
392 my( $self, @sigils ) = @_;
394 # Clear the witness(es) out of the paths
395 foreach my $e ( $self->paths ) {
396 foreach my $sig ( @sigils ) {
397 $self->del_path( $e, $sig );
401 # Clear out the newly unused readings
402 foreach my $r ( $self->readings ) {
403 unless( $self->reading_witnesses( $r ) ) {
404 $self->del_reading( $r );
409 sub add_relationship {
411 my( $source, $target, $opts ) = $self->_stringify_args( @_ );
412 my( @vectors ) = $self->relations->add_relationship( $source,
413 $self->reading( $source ), $target, $self->reading( $target ), $opts );
414 # Force a full rank recalculation every time. Yuck.
415 $self->calculate_ranks() if $self->end->has_rank;
419 =head2 reading_witnesses( $reading )
421 Return a list of sigils corresponding to the witnesses in which the reading appears.
425 sub reading_witnesses {
426 my( $self, $reading ) = @_;
427 # We need only check either the incoming or the outgoing edges; I have
428 # arbitrarily chosen "incoming". Thus, special-case the start node.
429 if( $reading eq $self->start ) {
430 return map { $_->sigil } $self->tradition->witnesses;
433 foreach my $e ( $self->sequence->edges_to( $reading ) ) {
434 my $wits = $self->sequence->get_edge_attributes( @$e );
435 @all_witnesses{ keys %$wits } = 1;
437 return keys %all_witnesses;
440 =head1 OUTPUT METHODS
442 =head2 as_svg( \%options )
444 Returns an SVG string that represents the graph, via as_dot and graphviz.
445 See as_dot for a list of options.
450 my( $self, $opts ) = @_;
451 if( !$self->has_cached_svg || $opts->{'recalc'} ) {
452 my @cmd = qw/dot -Tsvg/;
454 my $dotfile = File::Temp->new();
456 # $dotfile->unlink_on_destroy(0);
457 binmode $dotfile, ':utf8';
458 print $dotfile $self->as_dot( $opts );
459 push( @cmd, $dotfile->filename );
460 run( \@cmd, ">", binary(), \$svg );
461 $self->cached_svg( decode_utf8( $svg ) );
463 return $self->cached_svg;
467 =head2 as_dot( \%options )
469 Returns a string that is the collation graph expressed in dot
470 (i.e. GraphViz) format. Options include:
485 my( $self, $opts ) = @_;
486 my $startrank = $opts->{'from'} if $opts;
487 my $endrank = $opts->{'to'} if $opts;
488 my $color_common = $opts->{'color_common'} if $opts;
489 my $STRAIGHTENHACK = !$startrank && !$endrank && $self->end->rank
490 && $self->end->rank > 100;
492 # Check the arguments
494 return if $endrank && $startrank > $endrank;
495 return if $startrank > $self->end->rank;
497 if( defined $endrank ) {
498 return if $endrank < 0;
499 $endrank = undef if $endrank == $self->end->rank;
502 my $graph_name = $self->tradition->name;
503 $graph_name =~ s/[^\w\s]//g;
504 $graph_name = join( '_', split( /\s+/, $graph_name ) );
512 'fillcolor' => 'white',
517 'arrowhead' => 'open',
518 'color' => '#000000',
519 'fontcolor' => '#000000',
522 my $dot = sprintf( "digraph %s {\n", $graph_name );
523 $dot .= "\tgraph " . _dot_attr_string( \%graph_attrs ) . ";\n";
524 $dot .= "\tnode " . _dot_attr_string( \%node_attrs ) . ";\n";
526 # Output substitute start/end readings if necessary
528 $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n";
531 $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n";
533 if( $STRAIGHTENHACK ) {
535 $dot .= "\tsubgraph { rank=same \"#START#\" \"#SILENT#\" }\n";
536 $dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];"
538 my %used; # Keep track of the readings that actually appear in the graph
539 # Sort the readings by rank if we have ranks; this speeds layout.
540 my @all_readings = $self->end->has_rank
541 ? sort { $a->rank <=> $b->rank } $self->readings
543 # TODO Refrain from outputting lacuna nodes - just grey out the edges.
544 foreach my $reading ( @all_readings ) {
545 # Only output readings within our rank range.
546 next if $startrank && $reading->rank < $startrank;
547 next if $endrank && $reading->rank > $endrank;
548 $used{$reading->id} = 1;
549 # Need not output nodes without separate labels
550 next if $reading->id eq $reading->text;
552 my $label = $reading->text;
553 $label =~ s/\"/\\\"/g;
554 $rattrs->{'label'} = $label;
555 $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common;
556 $dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) );
559 # Add the real edges. Need to weight one edge per rank jump, in a
561 # my $weighted = $self->_add_edge_weights;
562 my @edges = $self->paths;
563 my( %substart, %subend );
564 foreach my $edge ( @edges ) {
565 # Do we need to output this edge?
566 if( $used{$edge->[0]} && $used{$edge->[1]} ) {
567 my $label = $self->_path_display_label( $self->path_witnesses( $edge ) );
568 my $variables = { %edge_attrs, 'label' => $label };
570 # Account for the rank gap if necessary
571 my $rank0 = $self->reading( $edge->[0] )->rank
572 if $self->reading( $edge->[0] )->has_rank;
573 my $rank1 = $self->reading( $edge->[1] )->rank
574 if $self->reading( $edge->[1] )->has_rank;
575 if( defined $rank0 && defined $rank1 && $rank1 - $rank0 > 1 ) {
576 $variables->{'minlen'} = $rank1 - $rank0;
579 # Add the calculated edge weights
580 # if( exists $weighted->{$edge->[0]}
581 # && $weighted->{$edge->[0]} eq $edge->[1] ) {
582 # # $variables->{'color'} = 'red';
583 # $variables->{'weight'} = 3.0;
586 # EXPERIMENTAL: make edge width reflect no. of witnesses
587 my $extrawidth = scalar( $self->path_witnesses( $edge ) ) * 0.2;
588 $variables->{'penwidth'} = $extrawidth + 0.8; # gives 1 for a single wit
590 my $varopts = _dot_attr_string( $variables );
591 $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n",
592 $edge->[0], $edge->[1], $varopts );
593 } elsif( $used{$edge->[0]} ) {
594 $subend{$edge->[0]} = 1;
595 } elsif( $used{$edge->[1]} ) {
596 $substart{$edge->[1]} = 1;
599 # Add substitute start and end edges if necessary
600 foreach my $node ( keys %substart ) {
601 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
602 my $variables = { %edge_attrs, 'label' => $witstr };
603 my $varopts = _dot_attr_string( $variables );
604 $dot .= "\t\"#SUBSTART#\" -> \"$node\" $varopts;";
606 foreach my $node ( keys %subend ) {
607 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
608 my $variables = { %edge_attrs, 'label' => $witstr };
609 my $varopts = _dot_attr_string( $variables );
610 $dot .= "\t\"$node\" -> \"#SUBEND#\" $varopts;";
613 if( $STRAIGHTENHACK ) {
614 $dot .= "\t\"#END#\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
621 sub _dot_attr_string {
624 foreach my $k ( sort keys %$hash ) {
626 push( @attrs, $k.'="'.$v.'"' );
628 return( '[ ' . join( ', ', @attrs ) . ' ]' );
631 sub _add_edge_weights {
633 # Walk the graph from START to END, choosing the successor node with
634 # the largest number of witness paths each time.
636 my $curr = $self->start->id;
637 my $ranked = $self->end->has_rank;
638 while( $curr ne $self->end->id ) {
639 my $rank = $ranked ? $self->reading( $curr )->rank : 0;
640 my @succ = sort { $self->path_witnesses( $curr, $a )
641 <=> $self->path_witnesses( $curr, $b ) }
642 $self->sequence->successors( $curr );
643 my $next = pop @succ;
644 my $nextrank = $ranked ? $self->reading( $next )->rank : 0;
645 # Try to avoid lacunae in the weighted path.
647 ( $self->reading( $next )->is_lacuna ||
648 $nextrank - $rank > 1 ) ){
651 $weighted->{$curr} = $next;
657 =head2 path_witnesses( $edge )
659 Returns the list of sigils whose witnesses are associated with the given edge.
660 The edge can be passed as either an array or an arrayref of ( $source, $target ).
665 my( $self, @edge ) = @_;
666 # If edge is an arrayref, cope.
667 if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) {
671 my @wits = keys %{$self->sequence->get_edge_attributes( @edge )};
675 sub _path_display_label {
678 my $maj = scalar( $self->tradition->witnesses ) * 0.6;
679 if( scalar @wits > $maj ) {
680 # TODO break out a.c. wits
683 return join( ', ', @wits );
690 Returns a GraphML representation of the collation. The GraphML will contain
691 two graphs. The first expresses the attributes of the readings and the witness
692 paths that link them; the second expresses the relationships that link the
693 readings. This is the native transfer format for a tradition.
702 my $datafile = 't/data/florilegium_tei_ps.xml';
703 my $tradition = Text::Tradition->new( 'input' => 'TEI',
708 ok( $tradition, "Got a tradition object" );
709 is( scalar $tradition->witnesses, 13, "Found all witnesses" );
710 ok( $tradition->collation, "Tradition has a collation" );
712 my $c = $tradition->collation;
713 is( scalar $c->readings, $READINGS, "Collation has all readings" );
714 is( scalar $c->paths, $PATHS, "Collation has all paths" );
715 is( scalar $c->relationships, 0, "Collation has all relationships" );
717 # Add a few relationships
718 $c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
719 $c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
720 $c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );
722 # Now write it to GraphML and parse it again.
724 my $graphml = $c->as_graphml;
725 my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
726 is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
727 is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
728 is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
738 my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns';
739 my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance';
740 my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' .
741 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd';
743 # Create the document and root node
744 my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" );
745 my $root = $graphml->createElementNS( $graphml_ns, 'graphml' );
746 $graphml->setDocumentElement( $root );
747 $root->setNamespace( $xsi_ns, 'xsi', 0 );
748 $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema );
750 # Add the data keys for the graph
753 my @graph_attributes = qw/ version wit_list_separator baselabel linear ac_label /;
754 foreach my $datum ( @graph_attributes ) {
755 $graph_data_keys{$datum} = 'dg'.$gdi++;
756 my $key = $root->addNewChild( $graphml_ns, 'key' );
757 $key->setAttribute( 'attr.name', $datum );
758 $key->setAttribute( 'attr.type', $key eq 'linear' ? 'boolean' : 'string' );
759 $key->setAttribute( 'for', 'graph' );
760 $key->setAttribute( 'id', $graph_data_keys{$datum} );
763 # Add the data keys for nodes
770 is_start => 'boolean',
772 is_lacuna => 'boolean',
774 foreach my $datum ( keys %node_data ) {
775 $node_data_keys{$datum} = 'dn'.$ndi++;
776 my $key = $root->addNewChild( $graphml_ns, 'key' );
777 $key->setAttribute( 'attr.name', $datum );
778 $key->setAttribute( 'attr.type', $node_data{$datum} );
779 $key->setAttribute( 'for', 'node' );
780 $key->setAttribute( 'id', $node_data_keys{$datum} );
783 # Add the data keys for edges, i.e. witnesses
787 class => 'string', # Class, deprecated soon
788 witness => 'string', # ID/label for a path
789 relationship => 'string', # ID/label for a relationship
790 extra => 'boolean', # Path key
791 scope => 'string', # Relationship key
792 non_correctable => 'boolean', # Relationship key
793 non_independent => 'boolean', # Relationship key
795 foreach my $datum ( keys %edge_data ) {
796 $edge_data_keys{$datum} = 'de'.$edi++;
797 my $key = $root->addNewChild( $graphml_ns, 'key' );
798 $key->setAttribute( 'attr.name', $datum );
799 $key->setAttribute( 'attr.type', $edge_data{$datum} );
800 $key->setAttribute( 'for', 'edge' );
801 $key->setAttribute( 'id', $edge_data_keys{$datum} );
804 # Add the collation graph itself
805 my $sgraph = $root->addNewChild( $graphml_ns, 'graph' );
806 $sgraph->setAttribute( 'edgedefault', 'directed' );
807 $sgraph->setAttribute( 'id', $self->tradition->name );
808 $sgraph->setAttribute( 'parse.edgeids', 'canonical' );
809 $sgraph->setAttribute( 'parse.edges', scalar($self->paths) );
810 $sgraph->setAttribute( 'parse.nodeids', 'canonical' );
811 $sgraph->setAttribute( 'parse.nodes', scalar($self->readings) );
812 $sgraph->setAttribute( 'parse.order', 'nodesfirst' );
814 # Collation attribute data
815 foreach my $datum ( @graph_attributes ) {
816 my $value = $datum eq 'version' ? '3.0' : $self->$datum;
817 _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value );
822 # Add our readings to the graph
823 foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) {
824 # Add to the main graph
825 my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' );
826 my $node_xmlid = 'n' . $node_ctr++;
827 $node_hash{ $n->id } = $node_xmlid;
828 $node_el->setAttribute( 'id', $node_xmlid );
829 foreach my $d ( keys %node_data ) {
831 _add_graphml_data( $node_el, $node_data_keys{$d}, $nval )
836 # Add the path edges to the sequence graph
838 foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) {
839 # We add an edge in the graphml for every witness in $e.
840 foreach my $wit ( sort $self->path_witnesses( $e ) ) {
841 my( $id, $from, $to ) = ( 'e'.$edge_ctr++,
842 $node_hash{ $e->[0] },
843 $node_hash{ $e->[1] } );
844 my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' );
845 $edge_el->setAttribute( 'source', $from );
846 $edge_el->setAttribute( 'target', $to );
847 $edge_el->setAttribute( 'id', $id );
849 # It's a witness path, so add the witness
851 my $key = $edge_data_keys{'witness'};
852 # Is this an ante-corr witness?
853 my $aclabel = $self->ac_label;
854 if( $wit =~ /^(.*)\Q$aclabel\E$/ ) {
855 # Keep the base witness
857 # ...and record that this is an 'extra' reading path
858 _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel );
860 _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base );
861 _add_graphml_data( $edge_el, $edge_data_keys{'class'}, 'path' );
865 # Add the relationship graph to the XML
866 $self->relations->_as_graphml( $graphml_ns, $root, \%node_hash,
867 $node_data_keys{'id'}, \%edge_data_keys );
869 # Save and return the thing
870 my $result = decode_utf8( $graphml->toString(1) );
874 sub _add_graphml_data {
875 my( $el, $key, $value ) = @_;
876 return unless defined $value;
877 my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
878 $data_el->setAttribute( 'key', $key );
879 $data_el->appendText( $value );
884 Returns a CSV alignment table representation of the collation graph, one
885 row per witness (or witness uncorrected.)
891 my $table = $self->make_alignment_table;
892 my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } );
894 # Make the header row
895 $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} );
896 push( @result, decode_utf8( $csv->string ) );
897 # Make the rest of the rows
898 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
899 my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}};
900 my @row = map { $_ ? $_->{'t'} : $_ } @rowobjs;
901 $csv->combine( @row );
902 push( @result, decode_utf8( $csv->string ) );
904 return join( "\n", @result );
907 =head2 make_alignment_table( $use_refs, $include_witnesses )
909 Return a reference to an alignment table, in a slightly enhanced CollateX
910 format which looks like this:
912 $table = { alignment => [ { witness => "SIGIL",
913 tokens => [ { t => "TEXT" }, ... ] },
915 tokens => [ { t => "TEXT" }, ... ] },
919 If $use_refs is set to 1, the reading object is returned in the table
920 instead of READINGTEXT; if not, the text of the reading is returned.
922 If $include_witnesses is set to a hashref, only the witnesses whose sigil
923 keys have a true hash value will be included.
927 sub make_alignment_table {
928 my( $self, $noderefs, $include ) = @_;
929 # Make sure we can do this
930 throw( "Need a linear graph in order to make an alignment table" )
931 unless $self->linear;
932 $self->calculate_ranks unless $self->end->has_rank;
934 my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 };
935 my @all_pos = ( 1 .. $self->end->rank - 1 );
936 foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) {
938 next unless $include->{$wit->sigil};
940 # print STDERR "Making witness row(s) for " . $wit->sigil . "\n";
941 my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
942 my @row = _make_witness_row( \@wit_path, \@all_pos, $noderefs );
943 push( @{$table->{'alignment'}},
944 { 'witness' => $wit->sigil, 'tokens' => \@row } );
945 if( $wit->is_layered ) {
946 my @wit_ac_path = $self->reading_sequence( $self->start, $self->end,
947 $wit->sigil.$self->ac_label );
948 my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos, $noderefs );
949 push( @{$table->{'alignment'}},
950 { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } );
956 sub _make_witness_row {
957 my( $path, $positions, $noderefs ) = @_;
959 map { $char_hash{$_} = undef } @$positions;
961 foreach my $rdg ( @$path ) {
962 my $rtext = $rdg->text;
963 $rtext = '#LACUNA#' if $rdg->is_lacuna;
964 print STDERR "rank " . $rdg->rank . "\n" if $debug;
965 # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank;
966 $char_hash{$rdg->rank} = $noderefs ? { 't' => $rdg }
969 my @row = map { $char_hash{$_} } @$positions;
970 # Fill in lacuna markers for undef spots in the row
971 my $last_el = shift @row;
972 my @filled_row = ( $last_el );
973 foreach my $el ( @row ) {
974 # If we are using node reference, make the lacuna node appear many times
975 # in the table. If not, use the lacuna tag.
976 if( $last_el && _el_is_lacuna( $last_el ) && !defined $el ) {
977 $el = $noderefs ? $last_el : { 't' => '#LACUNA#' };
979 push( @filled_row, $el );
985 # Tiny utility function to say if a table element is a lacuna
988 return 1 if $el->{'t'} eq '#LACUNA#';
989 return 1 if ref( $el->{'t'} ) eq 'Text::Tradition::Collation::Reading'
990 && $el->{'t'}->is_lacuna;
994 # Helper to turn the witnesses along columns rather than rows. Assumes
999 return $result unless scalar @$table;
1000 my $nrows = scalar @{$table->[0]};
1001 foreach my $idx ( 0 .. $nrows - 1 ) {
1002 foreach my $wit ( 0 .. $#{$table} ) {
1003 $result->[$idx]->[$wit] = $table->[$wit]->[$idx];
1009 =head1 NAVIGATION METHODS
1011 =head2 reading_sequence( $first, $last, $sigil, $backup )
1013 Returns the ordered list of readings, starting with $first and ending
1014 with $last, for the witness given in $sigil. If a $backup sigil is
1015 specified (e.g. when walking a layered witness), it will be used wherever
1016 no $sigil path exists. If there is a base text reading, that will be
1017 used wherever no path exists for $sigil or $backup.
1021 # TODO Think about returning some lazy-eval iterator.
1022 # TODO Get rid of backup; we should know from what witness is whether we need it.
1024 sub reading_sequence {
1025 my( $self, $start, $end, $witness ) = @_;
1027 $witness = $self->baselabel unless $witness;
1028 my @readings = ( $start );
1031 while( $n && $n->id ne $end->id ) {
1032 if( exists( $seen{$n->id} ) ) {
1033 throw( "Detected loop for $witness at " . $n->id );
1037 my $next = $self->next_reading( $n, $witness );
1039 throw( "Did not find any path for $witness from reading " . $n->id );
1041 push( @readings, $next );
1044 # Check that the last reading is our end reading.
1045 my $last = $readings[$#readings];
1046 throw( "Last reading found from " . $start->text .
1047 " for witness $witness is not the end!" ) # TODO do we get this far?
1048 unless $last->id eq $end->id;
1053 =head2 next_reading( $reading, $sigil );
1055 Returns the reading that follows the given reading along the given witness
1061 # Return the successor via the corresponding path.
1063 my $answer = $self->_find_linked_reading( 'next', @_ );
1064 return undef unless $answer;
1065 return $self->reading( $answer );
1068 =head2 prior_reading( $reading, $sigil )
1070 Returns the reading that precedes the given reading along the given witness
1076 # Return the predecessor via the corresponding path.
1078 my $answer = $self->_find_linked_reading( 'prior', @_ );
1079 return $self->reading( $answer );
1082 sub _find_linked_reading {
1083 my( $self, $direction, $node, $path ) = @_;
1085 # Get a backup if we are dealing with a layered witness
1087 my $aclabel = $self->ac_label;
1088 if( $path && $path =~ /^(.*)\Q$aclabel\E$/ ) {
1092 my @linked_paths = $direction eq 'next'
1093 ? $self->sequence->edges_from( $node )
1094 : $self->sequence->edges_to( $node );
1095 return undef unless scalar( @linked_paths );
1097 # We have to find the linked path that contains all of the
1098 # witnesses supplied in $path.
1099 my( @path_wits, @alt_path_wits );
1100 @path_wits = sort( $self->_witnesses_of_label( $path ) ) if $path;
1101 @alt_path_wits = sort( $self->_witnesses_of_label( $alt_path ) ) if $alt_path;
1104 foreach my $le ( @linked_paths ) {
1105 if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) {
1108 my @le_wits = sort $self->path_witnesses( $le );
1109 if( _is_within( \@path_wits, \@le_wits ) ) {
1110 # This is the right path.
1111 return $direction eq 'next' ? $le->[1] : $le->[0];
1112 } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) {
1116 # Got this far? Return the alternate path if it exists.
1117 return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0]
1120 # Got this far? Return the base path if it exists.
1121 return $direction eq 'next' ? $base_le->[1] : $base_le->[0]
1124 # Got this far? We have no appropriate path.
1125 warn "Could not find $direction node from " . $node->id
1126 . " along path $path";
1132 my( $set1, $set2 ) = @_;
1133 my $ret = @$set1; # will be 0, i.e. false, if set1 is empty
1134 foreach my $el ( @$set1 ) {
1135 $ret = 0 unless grep { /^\Q$el\E$/ } @$set2;
1140 # Return the string that joins together a list of witnesses for
1141 # display on a single path.
1142 sub _witnesses_of_label {
1143 my( $self, $label ) = @_;
1144 my $regex = $self->wit_list_separator;
1145 my @answer = split( /\Q$regex\E/, $label );
1149 =head2 common_readings
1151 Returns the list of common readings in the graph (i.e. those readings that are
1152 shared by all non-lacunose witnesses.)
1156 sub common_readings {
1158 my @common = grep { $_->is_common } $self->readings;
1162 =head2 path_text( $sigil, $mainsigil [, $start, $end ] )
1164 Returns the text of a witness (plus its backup, if we are using a layer)
1165 as stored in the collation. The text is returned as a string, where the
1166 individual readings are joined with spaces and the meta-readings (e.g.
1167 lacunae) are omitted. Optional specification of $start and $end allows
1168 the generation of a subset of the witness text.
1173 my( $self, $wit, $start, $end ) = @_;
1174 $start = $self->start unless $start;
1175 $end = $self->end unless $end;
1176 my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit );
1177 return join( ' ', map { $_->text } @path );
1180 =head1 INITIALIZATION METHODS
1182 These are mostly for use by parsers.
1184 =head2 make_witness_path( $witness )
1186 Link the array of readings contained in $witness->path (and in
1187 $witness->uncorrected_path if it exists) into collation paths.
1188 Clear out the arrays when finished.
1190 =head2 make_witness_paths
1192 Call make_witness_path for all witnesses in the tradition.
1196 # For use when a collation is constructed from a base text and an apparatus.
1197 # We have the sequences of readings and just need to add path edges.
1198 # When we are done, clear out the witness path attributes, as they are no
1200 # TODO Find a way to replace the witness path attributes with encapsulated functions?
1202 sub make_witness_paths {
1204 foreach my $wit ( $self->tradition->witnesses ) {
1205 # print STDERR "Making path for " . $wit->sigil . "\n";
1206 $self->make_witness_path( $wit );
1210 sub make_witness_path {
1211 my( $self, $wit ) = @_;
1212 my @chain = @{$wit->path};
1213 my $sig = $wit->sigil;
1214 foreach my $idx ( 0 .. $#chain-1 ) {
1215 $self->add_path( $chain[$idx], $chain[$idx+1], $sig );
1217 if( $wit->is_layered ) {
1218 @chain = @{$wit->uncorrected_path};
1219 foreach my $idx( 0 .. $#chain-1 ) {
1220 my $source = $chain[$idx];
1221 my $target = $chain[$idx+1];
1222 $self->add_path( $source, $target, $sig.$self->ac_label )
1223 unless $self->has_path( $source, $target, $sig );
1227 $wit->clear_uncorrected_path;
1230 =head2 calculate_ranks
1232 Calculate the reading ranks (that is, their aligned positions relative
1233 to each other) for the graph. This can only be called on linear collations.
1237 use Text::Tradition;
1239 my $cxfile = 't/data/Collatex-16.xml';
1240 my $t = Text::Tradition->new(
1242 'input' => 'CollateX',
1245 my $c = $t->collation;
1248 my $svg = $c->as_svg;
1249 is( substr( $svg, 0, 5 ), '<?xml', "Got XML doc for svg" );
1250 ok( $c->has_cached_svg, "SVG was cached" );
1251 is( $c->as_svg, $svg, "Cached SVG returned upon second call" );
1252 $c->calculate_ranks;
1253 is( $c->as_svg, $svg, "Cached SVG retained with no rank change" );
1254 $c->add_relationship( 'n9', 'n23', { 'type' => 'spelling' } );
1255 isnt( $c->as_svg, $svg, "SVG changed after relationship add" );
1261 sub calculate_ranks {
1263 # Save the existing ranks, in case we need to invalidate the cached SVG.
1265 # Walk a version of the graph where every node linked by a relationship
1266 # edge is fundamentally the same node, and do a topological ranking on
1267 # the nodes in this graph.
1268 my $topo_graph = Graph->new();
1272 foreach my $r ( $self->readings ) {
1273 next if exists $rel_containers{$r->id};
1274 my @rels = $r->related_readings( 'colocated' );
1276 # Make a relationship container.
1278 my $rn = 'rel_container_' . $rel_ctr++;
1279 $topo_graph->add_vertex( $rn );
1281 $rel_containers{$_->id} = $rn;
1284 # Add a new node to mirror the old node.
1285 $rel_containers{$r->id} = $r->id;
1286 $topo_graph->add_vertex( $r->id );
1291 foreach my $r ( $self->readings ) {
1292 $existing_ranks{$r} = $r->rank;
1293 foreach my $n ( $self->sequence->successors( $r->id ) ) {
1294 my( $tfrom, $tto ) = ( $rel_containers{$r->id},
1295 $rel_containers{$n} );
1296 # $DB::single = 1 unless $tfrom && $tto;
1297 $topo_graph->add_edge( $tfrom, $tto );
1301 # Now do the rankings, starting with the start node.
1302 my $topo_start = $rel_containers{$self->start->id};
1303 my $node_ranks = { $topo_start => 0 };
1304 my @curr_origin = ( $topo_start );
1305 # A little iterative function.
1306 while( @curr_origin ) {
1307 @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin );
1309 # Transfer our rankings from the topological graph to the real one.
1310 foreach my $r ( $self->readings ) {
1311 if( defined $node_ranks->{$rel_containers{$r->id}} ) {
1312 $r->rank( $node_ranks->{$rel_containers{$r->id}} );
1314 # Die. Find the last rank we calculated.
1315 my @all_defined = sort { $node_ranks->{$rel_containers{$a->id}}
1316 <=> $node_ranks->{$rel_containers{$b->id}} }
1318 my $last = pop @all_defined;
1319 throw( "Ranks not calculated after $last - do you have a cycle in the graph?" );
1322 # Do we need to invalidate the cached SVG?
1323 if( $self->has_cached_svg ) {
1324 foreach my $r ( $self->readings ) {
1325 next if $existing_ranks{$r} == $r->rank;
1333 my( $graph, $node_ranks, @current_nodes ) = @_;
1334 # Look at each of the children of @current_nodes. If all the child's
1335 # parents have a rank, assign it the highest rank + 1 and add it to
1336 # @next_nodes. Otherwise skip it; we will return when the highest-ranked
1337 # parent gets a rank.
1339 foreach my $c ( @current_nodes ) {
1340 warn "Current reading $c has no rank!"
1341 unless exists $node_ranks->{$c};
1342 # print STDERR "Looking at child of node $c, rank "
1343 # . $node_ranks->{$c} . "\n";
1344 foreach my $child ( $graph->successors( $c ) ) {
1345 next if exists $node_ranks->{$child};
1346 my $highest_rank = -1;
1348 foreach my $parent ( $graph->predecessors( $child ) ) {
1349 if( exists $node_ranks->{$parent} ) {
1350 $highest_rank = $node_ranks->{$parent}
1351 if $highest_rank <= $node_ranks->{$parent};
1358 my $c_rank = $highest_rank + 1;
1359 # print STDERR "Assigning rank $c_rank to node $child \n";
1360 $node_ranks->{$child} = $c_rank;
1361 push( @next_nodes, $child );
1367 =head2 flatten_ranks
1369 A convenience method for parsing collation data. Searches the graph for readings
1370 with the same text at the same rank, and merges any that are found.
1376 my %unique_rank_rdg;
1377 foreach my $rdg ( $self->readings ) {
1378 next unless $rdg->has_rank;
1379 my $key = $rdg->rank . "||" . $rdg->text;
1380 if( exists $unique_rank_rdg{$key} ) {
1382 # print STDERR "Combining readings at same rank: $key\n";
1383 $self->merge_readings( $unique_rank_rdg{$key}, $rdg );
1384 # TODO see if this now makes a common point.
1386 $unique_rank_rdg{$key} = $rdg;
1391 =head2 remove_collations
1393 Another convenience method for parsing. Removes all 'collation' relationships
1394 that were defined in order to get the reading ranks to be correct.
1398 use Text::Tradition;
1400 my $cxfile = 't/data/Collatex-16.xml';
1401 my $t = Text::Tradition->new(
1403 'input' => 'CollateX',
1406 my $c = $t->collation;
1408 isnt( $c->reading('n23')->rank, $c->reading('n9')->rank, "Rank skew exists" );
1409 $c->add_relationship( 'n23', 'n9', { 'type' => 'collated', 'scope' => 'local' } );
1410 is( scalar $c->relationships, 4, "Found all expected relationships" );
1411 $c->remove_collations;
1412 is( scalar $c->relationships, 3, "Collated relationships now gone" );
1413 is( $c->reading('n23')->rank, $c->reading('n9')->rank, "Aligned ranks were preserved" );
1419 sub remove_collations {
1421 foreach my $reledge ( $self->relationships ) {
1422 my $relobj = $self->relations->get_relationship( $reledge );
1423 if( $relobj && $relobj->type eq 'collated' ) {
1424 $self->relations->delete_relationship( $reledge );
1430 =head2 calculate_common_readings
1432 Goes through the graph identifying the readings that appear in every witness
1433 (apart from those with lacunae at that spot.) Marks them as common and returns
1438 use Text::Tradition;
1440 my $cxfile = 't/data/Collatex-16.xml';
1441 my $t = Text::Tradition->new(
1443 'input' => 'CollateX',
1446 my $c = $t->collation;
1448 my @common = $c->calculate_common_readings();
1449 is( scalar @common, 8, "Found correct number of common readings" );
1450 my @marked = sort $c->common_readings();
1451 is( scalar @common, 8, "All common readings got marked as such" );
1452 my @expected = qw/ n1 n12 n16 n19 n20 n5 n6 n7 /;
1453 is_deeply( \@marked, \@expected, "Found correct list of common readings" );
1459 sub calculate_common_readings {
1462 my $table = $self->make_alignment_table( 1 );
1463 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1464 my @row = map { $_->{'tokens'}->[$idx]->{'t'} } @{$table->{'alignment'}};
1466 foreach my $r ( @row ) {
1468 $hash{$r->id} = $r unless $r->is_meta;
1470 $hash{'UNDEF'} = $r;
1473 if( keys %hash == 1 && !exists $hash{'UNDEF'} ) {
1474 my( $r ) = values %hash;
1476 push( @common, $r );
1482 =head2 text_from_paths
1484 Calculate the text array for all witnesses from the path, for later consistency
1485 checking. Only to be used if there is no non-graph-based way to know the
1490 sub text_from_paths {
1492 foreach my $wit ( $self->tradition->witnesses ) {
1493 my @text = split( /\s+/,
1494 $self->reading_sequence( $self->start, $self->end, $wit->sigil ) );
1495 $wit->text( \@text );
1496 if( $wit->is_layered ) {
1497 my @uctext = split( /\s+/,
1498 $self->reading_sequence( $self->start, $self->end,
1499 $wit->sigil.$self->ac_label ) );
1500 $wit->text( \@uctext );
1505 =head1 UTILITY FUNCTIONS
1507 =head2 common_predecessor( $reading_a, $reading_b )
1509 Find the last reading that occurs in sequence before both the given readings.
1511 =head2 common_successor( $reading_a, $reading_b )
1513 Find the first reading that occurs in sequence after both the given readings.
1517 use Text::Tradition;
1519 my $cxfile = 't/data/Collatex-16.xml';
1520 my $t = Text::Tradition->new(
1522 'input' => 'CollateX',
1525 my $c = $t->collation;
1527 is( $c->common_predecessor( 'n9', 'n23' )->id,
1528 'n20', "Found correct common predecessor" );
1529 is( $c->common_successor( 'n9', 'n23' )->id,
1530 '#END#', "Found correct common successor" );
1532 is( $c->common_predecessor( 'n19', 'n17' )->id,
1533 'n16', "Found correct common predecessor for readings on same path" );
1534 is( $c->common_successor( 'n21', 'n26' )->id,
1535 '#END#', "Found correct common successor for readings on same path" );
1541 ## Return the closest reading that is a predecessor of both the given readings.
1542 sub common_predecessor {
1544 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1545 return $self->_common_in_path( $r1, $r2, 'predecessors' );
1548 sub common_successor {
1550 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1551 return $self->_common_in_path( $r1, $r2, 'successors' );
1554 sub _common_in_path {
1555 my( $self, $r1, $r2, $dir ) = @_;
1556 my $iter = $r1->rank > $r2->rank ? $r1->rank : $r2->rank;
1557 $iter = $self->end->rank - $iter if $dir eq 'successors';
1559 my @last_checked = ( $r1, $r2 );
1561 while( !@candidates ) {
1563 foreach my $lc ( @last_checked ) {
1564 foreach my $p ( $lc->$dir ) {
1565 if( $all_seen{$p->id} ) {
1566 push( @candidates, $p );
1568 $all_seen{$p->id} = 1;
1569 push( @new_lc, $p );
1573 @last_checked = @new_lc;
1575 my @answer = sort { $a->rank <=> $b->rank } @candidates;
1576 return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer );
1580 Text::Tradition::Error->throw(
1581 'ident' => 'Collation error',
1587 __PACKAGE__->meta->make_immutable;
1591 This package is free software and is provided "as is" without express
1592 or implied warranty. You can redistribute it and/or modify it under
1593 the same terms as Perl itself.
1597 Tara L Andrews E<lt>aurum@cpan.orgE<gt>