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',
40 writer => '_set_tradition',
45 isa => 'HashRef[Text::Tradition::Collation::Reading]',
49 _add_reading => 'set',
50 del_reading => 'delete',
51 has_reading => 'exists',
54 default => sub { {} },
57 has 'wit_list_separator' => (
66 default => 'base text',
89 isa => 'Text::Tradition::Collation::Reading',
90 writer => '_set_start',
96 isa => 'Text::Tradition::Collation::Reading',
101 has 'cached_svg' => (
104 predicate => 'has_cached_svg',
105 clearer => 'wipe_svg',
108 has 'cached_table' => (
111 predicate => 'has_cached_table',
112 clearer => 'wipe_table',
115 has '_graphcalc_done' => (
123 Text::Tradition::Collation - a software model for a text collation
128 my $t = Text::Tradition->new(
129 'name' => 'this is a text',
131 'file' => '/path/to/tei_parallel_seg_file.xml' );
133 my $c = $t->collation;
134 my @readings = $c->readings;
135 my @paths = $c->paths;
136 my @relationships = $c->relationships;
138 my $svg_variant_graph = $t->collation->as_svg();
142 Text::Tradition is a library for representation and analysis of collated
143 texts, particularly medieval ones. The Collation is the central feature of
144 a Tradition, where the text, its sequence of readings, and its relationships
145 between readings are actually kept.
151 The constructor. Takes a hash or hashref of the following arguments:
155 =item * tradition - The Text::Tradition object to which the collation
158 =item * linear - Whether the collation should be linear; that is, whether
159 transposed readings should be treated as two linked readings rather than one,
160 and therefore whether the collation graph is acyclic. Defaults to true.
162 =item * baselabel - The default label for the path taken by a base text
163 (if any). Defaults to 'base text'.
165 =item * wit_list_separator - The string to join a list of witnesses for
166 purposes of making labels in display graphs. Defaults to ', '.
168 =item * ac_label - The extra label to tack onto a witness sigil when
169 representing another layer of path for the given witness - that is, when
170 a text has more than one possible reading due to scribal corrections or
171 the like. Defaults to ' (a.c.)'.
173 =item * wordsep - The string used to separate words in the original text.
184 =head2 wit_list_separator
192 Simple accessors for collation attributes.
196 The meta-reading at the start of every witness path.
200 The meta-reading at the end of every witness path.
204 Returns all Reading objects in the graph.
206 =head2 reading( $id )
208 Returns the Reading object corresponding to the given ID.
210 =head2 add_reading( $reading_args )
212 Adds a new reading object to the collation.
213 See L<Text::Tradition::Collation::Reading> for the available arguments.
215 =head2 del_reading( $object_or_id )
217 Removes the given reading from the collation, implicitly removing its
218 paths and relationships.
220 =head2 merge_readings( $main, $second, $concatenate, $with_str )
222 Merges the $second reading into the $main one. If $concatenate is true, then
223 the merged node will carry the text of both readings, concatenated with either
224 $with_str (if specified) or a sensible default (the empty string if the
225 appropriate 'join_*' flag is set on either reading, or else $self->wordsep.)
227 The first two arguments may be either readings or reading IDs.
229 =head2 has_reading( $id )
231 Predicate to see whether a given reading ID is in the graph.
233 =head2 reading_witnesses( $object_or_id )
235 Returns a list of sigils whose witnesses contain the reading.
239 Returns all reading paths within the document - that is, all edges in the
240 collation graph. Each path is an arrayref of [ $source, $target ] reading IDs.
242 =head2 add_path( $source, $target, $sigil )
244 Links the given readings in the collation in sequence, under the given witness
245 sigil. The readings may be specified by object or ID.
247 =head2 del_path( $source, $target, $sigil )
249 Links the given readings in the collation in sequence, under the given witness
250 sigil. The readings may be specified by object or ID.
252 =head2 has_path( $source, $target );
254 Returns true if the two readings are linked in sequence in any witness.
255 The readings may be specified by object or ID.
259 Returns all Relationship objects in the collation.
261 =head2 add_relationship( $reading, $other_reading, $options )
263 Adds a new relationship of the type given in $options between the two readings,
264 which may be specified by object or ID. Returns a value of ( $status, @vectors)
265 where $status is true on success, and @vectors is a list of relationship edges
266 that were ultimately added.
267 See L<Text::Tradition::Collation::Relationship> for the available options.
273 $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) );
274 $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) );
275 $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) );
278 ### Reading construct/destruct functions
281 my( $self, $reading ) = @_;
282 unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
283 my %args = %$reading;
284 $reading = Text::Tradition::Collation::Reading->new(
285 'collation' => $self,
288 # First check to see if a reading with this ID exists.
289 if( $self->reading( $reading->id ) ) {
290 throw( "Collation already has a reading with id " . $reading->id );
292 $self->_graphcalc_done(0);
293 $self->_add_reading( $reading->id => $reading );
294 # Once the reading has been added, put it in both graphs.
295 $self->sequence->add_vertex( $reading->id );
296 $self->relations->add_reading( $reading->id );
300 around del_reading => sub {
305 if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
308 # Remove the reading from the graphs.
309 $self->_graphcalc_done(0);
310 $self->_clear_cache; # Explicitly clear caches to GC the reading
311 $self->sequence->delete_vertex( $arg );
312 $self->relations->delete_reading( $arg );
315 $self->$orig( $arg );
322 my $cxfile = 't/data/Collatex-16.xml';
323 my $t = Text::Tradition->new(
325 'input' => 'CollateX',
328 my $c = $t->collation;
330 my $rno = scalar $c->readings;
331 # Split n21 for testing purposes
332 my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } );
333 my $old_r = $c->reading( 'n21' );
334 $old_r->alter_text( 'to' );
335 $c->del_path( 'n20', 'n21', 'A' );
336 $c->add_path( 'n20', 'n21p0', 'A' );
337 $c->add_path( 'n21p0', 'n21', 'A' );
339 ok( $c->reading( 'n21p0' ), "New reading exists" );
340 is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" );
342 # Combine n3 and n4 ( with his )
343 $c->merge_readings( 'n3', 'n4', 1 );
344 ok( !$c->reading('n4'), "Reading n4 is gone" );
345 is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" );
347 # Collapse n9 and n10 ( rood / root )
348 $c->merge_readings( 'n9', 'n10' );
349 ok( !$c->reading('n10'), "Reading n10 is gone" );
350 is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" );
352 # Combine n21 and n21p0
353 my $remaining = $c->reading('n21');
354 $remaining ||= $c->reading('n22'); # one of these should still exist
355 $c->merge_readings( 'n21p0', $remaining, 1 );
356 ok( !$c->reading('n21'), "Reading $remaining is gone" );
357 is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" );
366 # We only need the IDs for adding paths to the graph, not the reading
367 # objects themselves.
368 my( $kept, $deleted, $combine, $combine_char ) = $self->_stringify_args( @_ );
369 $self->_graphcalc_done(0);
371 # The kept reading should inherit the paths and the relationships
372 # of the deleted reading.
373 foreach my $path ( $self->sequence->edges_at( $deleted ) ) {
374 my @vector = ( $kept );
375 push( @vector, $path->[1] ) if $path->[0] eq $deleted;
376 unshift( @vector, $path->[0] ) if $path->[1] eq $deleted;
377 next if $vector[0] eq $vector[1]; # Don't add a self loop
378 my %wits = %{$self->sequence->get_edge_attributes( @$path )};
379 $self->sequence->add_edge( @vector );
380 my $fwits = $self->sequence->get_edge_attributes( @vector );
381 @wits{keys %$fwits} = values %$fwits;
382 $self->sequence->set_edge_attributes( @vector, \%wits );
384 $self->relations->merge_readings( $kept, $deleted, $combine_char );
386 # Do the deletion deed.
388 my $kept_obj = $self->reading( $kept );
389 my $del_obj = $self->reading( $deleted );
390 my $joinstr = $combine_char;
391 unless( defined $joinstr ) {
392 $joinstr = '' if $kept_obj->join_next || $del_obj->join_prior;
393 $joinstr = $self->wordsep unless defined $joinstr;
395 $kept_obj->alter_text( join( $joinstr, $kept_obj->text, $del_obj->text ) );
397 $self->del_reading( $deleted );
401 # Helper function for manipulating the graph.
402 sub _stringify_args {
403 my( $self, $first, $second, @args ) = @_;
405 if ref( $first ) eq 'Text::Tradition::Collation::Reading';
406 $second = $second->id
407 if ref( $second ) eq 'Text::Tradition::Collation::Reading';
408 return( $first, $second, @args );
411 # Helper function for manipulating the graph.
412 sub _objectify_args {
413 my( $self, $first, $second, $arg ) = @_;
414 $first = $self->reading( $first )
415 unless ref( $first ) eq 'Text::Tradition::Collation::Reading';
416 $second = $self->reading( $second )
417 unless ref( $second ) eq 'Text::Tradition::Collation::Reading';
418 return( $first, $second, $arg );
425 # We only need the IDs for adding paths to the graph, not the reading
426 # objects themselves.
427 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
429 $self->_graphcalc_done(0);
430 # Connect the readings
431 $self->sequence->add_edge( $source, $target );
432 # Note the witness in question
433 $self->sequence->set_edge_attribute( $source, $target, $wit, 1 );
439 if( ref( $_[0] ) eq 'ARRAY' ) {
446 # We only need the IDs for adding paths to the graph, not the reading
447 # objects themselves.
448 my( $source, $target, $wit ) = $self->_stringify_args( @args );
450 $self->_graphcalc_done(0);
451 if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) {
452 $self->sequence->delete_edge_attribute( $source, $target, $wit );
454 unless( keys %{$self->sequence->get_edge_attributes( $source, $target )} ) {
455 $self->sequence->delete_edge( $source, $target );
460 # Extra graph-alike utility
463 my( $source, $target, $wit ) = $self->_stringify_args( @_ );
464 return undef unless $self->sequence->has_edge( $source, $target );
465 return $self->sequence->has_edge_attribute( $source, $target, $wit );
468 =head2 clear_witness( @sigil_list )
470 Clear the given witnesses out of the collation entirely, removing references
471 to them in paths, and removing readings that belong only to them. Should only
472 be called via $tradition->del_witness.
477 my( $self, @sigils ) = @_;
479 $self->_graphcalc_done(0);
480 # Clear the witness(es) out of the paths
481 foreach my $e ( $self->paths ) {
482 foreach my $sig ( @sigils ) {
483 $self->del_path( $e, $sig );
487 # Clear out the newly unused readings
488 foreach my $r ( $self->readings ) {
489 unless( $self->reading_witnesses( $r ) ) {
490 $self->del_reading( $r );
495 sub add_relationship {
497 my( $source, $target, $opts ) = $self->_stringify_args( @_ );
498 my( @vectors ) = $self->relations->add_relationship( $source,
499 $self->reading( $source ), $target, $self->reading( $target ), $opts );
500 $self->_graphcalc_done(0);
504 around qw/ get_relationship del_relationship / => sub {
508 if( @args == 1 && ref( $args[0] ) eq 'ARRAY' ) {
511 my( $source, $target ) = $self->_stringify_args( @args );
512 $self->$orig( $source, $target );
515 =head2 reading_witnesses( $reading )
517 Return a list of sigils corresponding to the witnesses in which the reading appears.
521 sub reading_witnesses {
522 my( $self, $reading ) = @_;
523 # We need only check either the incoming or the outgoing edges; I have
524 # arbitrarily chosen "incoming". Thus, special-case the start node.
525 if( $reading eq $self->start ) {
526 return map { $_->sigil } $self->tradition->witnesses;
529 foreach my $e ( $self->sequence->edges_to( $reading ) ) {
530 my $wits = $self->sequence->get_edge_attributes( @$e );
531 @all_witnesses{ keys %$wits } = 1;
533 my $acstr = $self->ac_label;
534 foreach my $acwit ( grep { $_ =~ s/^(.*)\Q$acstr\E$/$1/ } keys %all_witnesses ) {
535 delete $all_witnesses{$acwit.$acstr} if exists $all_witnesses{$acwit};
537 return keys %all_witnesses;
540 =head1 OUTPUT METHODS
542 =head2 as_svg( \%options )
544 Returns an SVG string that represents the graph, via as_dot and graphviz.
545 See as_dot for a list of options. Must have GraphViz (dot) installed to run.
550 my( $self, $opts ) = @_;
551 throw( "Need GraphViz installed to output SVG" )
552 unless File::Which::which( 'dot' );
553 my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'};
554 $self->calculate_ranks() unless( $self->_graphcalc_done || $opts->{'nocalc'} );
555 if( !$self->has_cached_svg || $opts->{'recalc'} || $want_subgraph ) {
556 my @cmd = qw/dot -Tsvg/;
558 my $dotfile = File::Temp->new();
560 # $dotfile->unlink_on_destroy(0);
561 binmode $dotfile, ':utf8';
562 print $dotfile $self->as_dot( $opts );
563 push( @cmd, $dotfile->filename );
564 run( \@cmd, ">", binary(), \$svg );
565 $svg = decode_utf8( $svg );
566 $self->cached_svg( $svg ) unless $want_subgraph;
569 return $self->cached_svg;
574 =head2 as_dot( \%options )
576 Returns a string that is the collation graph expressed in dot
577 (i.e. GraphViz) format. Options include:
592 my( $self, $opts ) = @_;
593 my $startrank = $opts->{'from'} if $opts;
594 my $endrank = $opts->{'to'} if $opts;
595 my $color_common = $opts->{'color_common'} if $opts;
596 my $STRAIGHTENHACK = !$startrank && !$endrank && $self->end->rank
597 && $self->end->rank > 100;
598 $STRAIGHTENHACK = 1 if $opts->{'straight'}; # even for subgraphs or small graphs
600 # Check the arguments
602 return if $endrank && $startrank > $endrank;
603 return if $startrank > $self->end->rank;
605 if( defined $endrank ) {
606 return if $endrank < 0;
607 $endrank = undef if $endrank == $self->end->rank;
610 my $graph_name = $self->tradition->name;
611 $graph_name =~ s/[^\w\s]//g;
612 $graph_name = join( '_', split( /\s+/, $graph_name ) );
620 'fillcolor' => 'white',
625 'arrowhead' => 'open',
626 'color' => '#000000',
627 'fontcolor' => '#000000',
630 my $dot = sprintf( "digraph %s {\n", $graph_name );
631 $dot .= "\tgraph " . _dot_attr_string( \%graph_attrs ) . ";\n";
632 $dot .= "\tnode " . _dot_attr_string( \%node_attrs ) . ";\n";
634 # Output substitute start/end readings if necessary
636 $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n";
639 $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n";
641 if( $STRAIGHTENHACK ) {
643 my $startlabel = $startrank ? 'SUBSTART' : 'START';
644 $dot .= "\tsubgraph { rank=same \"#$startlabel#\" \"#SILENT#\" }\n";
645 $dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];"
647 my %used; # Keep track of the readings that actually appear in the graph
648 # Sort the readings by rank if we have ranks; this speeds layout.
649 my @all_readings = $self->end->has_rank
650 ? sort { $a->rank <=> $b->rank } $self->readings
652 # TODO Refrain from outputting lacuna nodes - just grey out the edges.
653 foreach my $reading ( @all_readings ) {
654 # Only output readings within our rank range.
655 next if $startrank && $reading->rank < $startrank;
656 next if $endrank && $reading->rank > $endrank;
657 $used{$reading->id} = 1;
658 # Need not output nodes without separate labels
659 next if $reading->id eq $reading->text;
661 my $label = $reading->text;
662 $label .= '-' if $reading->join_next;
663 $label = "-$label" if $reading->join_prior;
664 $label =~ s/\"/\\\"/g;
665 $rattrs->{'label'} = $label;
666 $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common;
667 $dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) );
670 # Add the real edges. Need to weight one edge per rank jump, in a
672 # my $weighted = $self->_add_edge_weights;
673 my @edges = $self->paths;
674 my( %substart, %subend );
675 foreach my $edge ( @edges ) {
676 # Do we need to output this edge?
677 if( $used{$edge->[0]} && $used{$edge->[1]} ) {
678 my $label = $self->_path_display_label( $self->path_witnesses( $edge ) );
679 my $variables = { %edge_attrs, 'label' => $label };
681 # Account for the rank gap if necessary
682 my $rank0 = $self->reading( $edge->[0] )->rank
683 if $self->reading( $edge->[0] )->has_rank;
684 my $rank1 = $self->reading( $edge->[1] )->rank
685 if $self->reading( $edge->[1] )->has_rank;
686 if( defined $rank0 && defined $rank1 && $rank1 - $rank0 > 1 ) {
687 $variables->{'minlen'} = $rank1 - $rank0;
690 # Add the calculated edge weights
691 # if( exists $weighted->{$edge->[0]}
692 # && $weighted->{$edge->[0]} eq $edge->[1] ) {
693 # # $variables->{'color'} = 'red';
694 # $variables->{'weight'} = 3.0;
697 # EXPERIMENTAL: make edge width reflect no. of witnesses
698 my $extrawidth = scalar( $self->path_witnesses( $edge ) ) * 0.2;
699 $variables->{'penwidth'} = $extrawidth + 0.8; # gives 1 for a single wit
701 my $varopts = _dot_attr_string( $variables );
702 $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n",
703 $edge->[0], $edge->[1], $varopts );
704 } elsif( $used{$edge->[0]} ) {
705 $subend{$edge->[0]} = 1;
706 } elsif( $used{$edge->[1]} ) {
707 $substart{$edge->[1]} = 1;
710 # Add substitute start and end edges if necessary
711 foreach my $node ( keys %substart ) {
712 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
713 my $variables = { %edge_attrs, 'label' => $witstr };
714 my $varopts = _dot_attr_string( $variables );
715 $dot .= "\t\"#SUBSTART#\" -> \"$node\" $varopts;";
717 foreach my $node ( keys %subend ) {
718 my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
719 my $variables = { %edge_attrs, 'label' => $witstr };
720 my $varopts = _dot_attr_string( $variables );
721 $dot .= "\t\"$node\" -> \"#SUBEND#\" $varopts;";
724 if( $STRAIGHTENHACK ) {
725 my $endlabel = $endrank ? 'SUBEND' : 'END';
726 $dot .= "\t\"#$endlabel#\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
733 sub _dot_attr_string {
736 foreach my $k ( sort keys %$hash ) {
738 push( @attrs, $k.'="'.$v.'"' );
740 return( '[ ' . join( ', ', @attrs ) . ' ]' );
743 sub _add_edge_weights {
745 # Walk the graph from START to END, choosing the successor node with
746 # the largest number of witness paths each time.
748 my $curr = $self->start->id;
749 my $ranked = $self->end->has_rank;
750 while( $curr ne $self->end->id ) {
751 my $rank = $ranked ? $self->reading( $curr )->rank : 0;
752 my @succ = sort { $self->path_witnesses( $curr, $a )
753 <=> $self->path_witnesses( $curr, $b ) }
754 $self->sequence->successors( $curr );
755 my $next = pop @succ;
756 my $nextrank = $ranked ? $self->reading( $next )->rank : 0;
757 # Try to avoid lacunae in the weighted path.
759 ( $self->reading( $next )->is_lacuna ||
760 $nextrank - $rank > 1 ) ){
763 $weighted->{$curr} = $next;
769 =head2 path_witnesses( $edge )
771 Returns the list of sigils whose witnesses are associated with the given edge.
772 The edge can be passed as either an array or an arrayref of ( $source, $target ).
777 my( $self, @edge ) = @_;
778 # If edge is an arrayref, cope.
779 if( @edge == 1 && ref( $edge[0] ) eq 'ARRAY' ) {
783 my @wits = keys %{$self->sequence->get_edge_attributes( @edge )};
787 # Helper function. Make a display label for the given witnesses, showing a.c.
788 # witnesses only where the main witness is not also in the list.
789 sub _path_display_label {
792 map { $wits{$_} = 1 } @_;
794 # If an a.c. wit is listed, remove it if the main wit is also listed.
795 # Otherwise keep it for explicit listing.
796 my $aclabel = $self->ac_label;
798 foreach my $w ( sort keys %wits ) {
799 if( $w =~ /^(.*)\Q$aclabel\E$/ ) {
800 if( exists $wits{$1} ) {
803 push( @disp_ac, $w );
808 # See if we are in a majority situation.
809 my $maj = scalar( $self->tradition->witnesses ) * 0.6;
810 if( scalar keys %wits > $maj ) {
811 unshift( @disp_ac, 'majority' );
812 return join( ', ', @disp_ac );
814 return join( ', ', sort keys %wits );
818 =head2 readings_at_rank( $rank )
820 Returns a list of readings at a given rank, taken from the alignment table.
824 sub readings_at_rank {
825 my( $self, $rank ) = @_;
826 my $table = $self->alignment_table;
827 # Table rank is real rank - 1.
828 my @elements = map { $_->{'tokens'}->[$rank-1] } @{$table->{'alignment'}};
830 foreach my $e ( @elements ) {
831 next unless ref( $e ) eq 'HASH';
832 next unless exists $e->{'t'};
833 $readings{$e->{'t'}->id} = $e->{'t'};
835 return values %readings;
840 Returns a GraphML representation of the collation. The GraphML will contain
841 two graphs. The first expresses the attributes of the readings and the witness
842 paths that link them; the second expresses the relationships that link the
843 readings. This is the native transfer format for a tradition.
852 my $datafile = 't/data/florilegium_tei_ps.xml';
853 my $tradition = Text::Tradition->new( 'input' => 'TEI',
858 ok( $tradition, "Got a tradition object" );
859 is( scalar $tradition->witnesses, 13, "Found all witnesses" );
860 ok( $tradition->collation, "Tradition has a collation" );
862 my $c = $tradition->collation;
863 is( scalar $c->readings, $READINGS, "Collation has all readings" );
864 is( scalar $c->paths, $PATHS, "Collation has all paths" );
865 is( scalar $c->relationships, 0, "Collation has all relationships" );
867 # Add a few relationships
868 $c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
869 $c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
870 $c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );
872 # Now write it to GraphML and parse it again.
874 my $graphml = $c->as_graphml;
875 my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
876 is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
877 is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
878 is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
886 $self->calculate_ranks unless $self->_graphcalc_done;
889 my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns';
890 my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance';
891 my $graphml_schema = 'http://graphml.graphdrawing.org/xmlns ' .
892 'http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd';
894 # Create the document and root node
895 my $graphml = XML::LibXML->createDocument( "1.0", "UTF-8" );
896 my $root = $graphml->createElementNS( $graphml_ns, 'graphml' );
897 $graphml->setDocumentElement( $root );
898 $root->setNamespace( $xsi_ns, 'xsi', 0 );
899 $root->setAttributeNS( $xsi_ns, 'schemaLocation', $graphml_schema );
901 # List of attribute types to save on our objects and their corresponding
907 'RelationshipType' => 'string',
908 'RelationshipScope' => 'string',
911 # List of attribute names *not* to save on our objects.
912 # We will also not save any attribute beginning with _.
914 map { $skipsave{$_} = 1 } qw/ cached_svg /;
916 # Add the data keys for the graph. Include an extra key 'version' for the
917 # GraphML output version.
920 my %graph_attributes = ( 'version' => 'string' );
921 # Graph attributes include those of Tradition and those of Collation.
923 my $tmeta = $self->tradition->meta;
924 my $cmeta = $self->meta;
925 map { $gattr_from{$_->name} = 'Tradition' } $tmeta->get_all_attributes;
926 map { $gattr_from{$_->name} = 'Collation' } $cmeta->get_all_attributes;
927 foreach my $attr ( ( $tmeta->get_all_attributes, $cmeta->get_all_attributes ) ) {
928 next if $attr->name =~ /^_/;
929 next if $skipsave{$attr->name};
930 next unless $save_types{$attr->type_constraint->name};
931 $graph_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
934 foreach my $datum ( sort keys %graph_attributes ) {
935 $graph_data_keys{$datum} = 'dg'.$gdi++;
936 my $key = $root->addNewChild( $graphml_ns, 'key' );
937 $key->setAttribute( 'attr.name', $datum );
938 $key->setAttribute( 'attr.type', $graph_attributes{$datum} );
939 $key->setAttribute( 'for', 'graph' );
940 $key->setAttribute( 'id', $graph_data_keys{$datum} );
943 # Add the data keys for reading nodes
944 my %reading_attributes;
945 my $rmeta = Text::Tradition::Collation::Reading->meta;
946 foreach my $attr( $rmeta->get_all_attributes ) {
947 next if $attr->name =~ /^_/;
948 next if $skipsave{$attr->name};
949 next unless $save_types{$attr->type_constraint->name};
950 $reading_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
954 foreach my $datum ( sort keys %reading_attributes ) {
955 $node_data_keys{$datum} = 'dn'.$ndi++;
956 my $key = $root->addNewChild( $graphml_ns, 'key' );
957 $key->setAttribute( 'attr.name', $datum );
958 $key->setAttribute( 'attr.type', $reading_attributes{$datum} );
959 $key->setAttribute( 'for', 'node' );
960 $key->setAttribute( 'id', $node_data_keys{$datum} );
963 # Add the data keys for edges, that is, paths and relationships. Path
964 # data does not come from a Moose class so is here manually.
967 my %edge_attributes = (
968 witness => 'string', # ID/label for a path
969 extra => 'boolean', # Path key
971 my @path_attributes = keys %edge_attributes; # track our manual additions
972 my $pmeta = Text::Tradition::Collation::Relationship->meta;
973 foreach my $attr( $pmeta->get_all_attributes ) {
974 next if $attr->name =~ /^_/;
975 next if $skipsave{$attr->name};
976 next unless $save_types{$attr->type_constraint->name};
977 $edge_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
979 foreach my $datum ( sort keys %edge_attributes ) {
980 $edge_data_keys{$datum} = 'de'.$edi++;
981 my $key = $root->addNewChild( $graphml_ns, 'key' );
982 $key->setAttribute( 'attr.name', $datum );
983 $key->setAttribute( 'attr.type', $edge_attributes{$datum} );
984 $key->setAttribute( 'for', 'edge' );
985 $key->setAttribute( 'id', $edge_data_keys{$datum} );
988 # Add the collation graph itself
989 my $sgraph = $root->addNewChild( $graphml_ns, 'graph' );
990 $sgraph->setAttribute( 'edgedefault', 'directed' );
991 $sgraph->setAttribute( 'id', $self->tradition->name );
992 $sgraph->setAttribute( 'parse.edgeids', 'canonical' );
993 $sgraph->setAttribute( 'parse.edges', scalar($self->paths) );
994 $sgraph->setAttribute( 'parse.nodeids', 'canonical' );
995 $sgraph->setAttribute( 'parse.nodes', scalar($self->readings) );
996 $sgraph->setAttribute( 'parse.order', 'nodesfirst' );
998 # Collation attribute data
999 foreach my $datum ( keys %graph_attributes ) {
1001 if( $datum eq 'version' ) {
1003 } elsif( $gattr_from{$datum} eq 'Tradition' ) {
1004 $value = $self->tradition->$datum;
1006 $value = $self->$datum;
1008 _add_graphml_data( $sgraph, $graph_data_keys{$datum}, $value );
1013 # Add our readings to the graph
1014 foreach my $n ( sort { $a->id cmp $b->id } $self->readings ) {
1015 # Add to the main graph
1016 my $node_el = $sgraph->addNewChild( $graphml_ns, 'node' );
1017 my $node_xmlid = 'n' . $node_ctr++;
1018 $node_hash{ $n->id } = $node_xmlid;
1019 $node_el->setAttribute( 'id', $node_xmlid );
1020 foreach my $d ( keys %reading_attributes ) {
1022 _add_graphml_data( $node_el, $node_data_keys{$d}, $nval )
1027 # Add the path edges to the sequence graph
1029 foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->sequence->edges() ) {
1030 # We add an edge in the graphml for every witness in $e.
1031 foreach my $wit ( sort $self->path_witnesses( $e ) ) {
1032 my( $id, $from, $to ) = ( 'e'.$edge_ctr++,
1033 $node_hash{ $e->[0] },
1034 $node_hash{ $e->[1] } );
1035 my $edge_el = $sgraph->addNewChild( $graphml_ns, 'edge' );
1036 $edge_el->setAttribute( 'source', $from );
1037 $edge_el->setAttribute( 'target', $to );
1038 $edge_el->setAttribute( 'id', $id );
1040 # It's a witness path, so add the witness
1042 my $key = $edge_data_keys{'witness'};
1043 # Is this an ante-corr witness?
1044 my $aclabel = $self->ac_label;
1045 if( $wit =~ /^(.*)\Q$aclabel\E$/ ) {
1046 # Keep the base witness
1048 # ...and record that this is an 'extra' reading path
1049 _add_graphml_data( $edge_el, $edge_data_keys{'extra'}, $aclabel );
1051 _add_graphml_data( $edge_el, $edge_data_keys{'witness'}, $base );
1055 # Add the relationship graph to the XML
1056 map { delete $edge_data_keys{$_} } @path_attributes;
1057 $self->relations->_as_graphml( $graphml_ns, $root, \%node_hash,
1058 $node_data_keys{'id'}, \%edge_data_keys );
1060 # Save and return the thing
1061 my $result = decode_utf8( $graphml->toString(1) );
1065 sub _add_graphml_data {
1066 my( $el, $key, $value ) = @_;
1067 return unless defined $value;
1068 my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
1069 $data_el->setAttribute( 'key', $key );
1070 $data_el->appendText( $value );
1075 Returns a CSV alignment table representation of the collation graph, one
1076 row per witness (or witness uncorrected.)
1082 my $table = $self->alignment_table;
1083 my $csv = Text::CSV->new( { binary => 1, quote_null => 0 } );
1085 # Make the header row
1086 $csv->combine( map { $_->{'witness'} } @{$table->{'alignment'}} );
1087 push( @result, decode_utf8( $csv->string ) );
1088 # Make the rest of the rows
1089 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1090 my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}};
1091 my @row = map { $_ ? $_->{'t'}->text : $_ } @rowobjs;
1092 $csv->combine( @row );
1093 push( @result, decode_utf8( $csv->string ) );
1095 return join( "\n", @result );
1098 =head2 alignment_table( $use_refs, $include_witnesses )
1100 Return a reference to an alignment table, in a slightly enhanced CollateX
1101 format which looks like this:
1103 $table = { alignment => [ { witness => "SIGIL",
1104 tokens => [ { t => "TEXT" }, ... ] },
1105 { witness => "SIG2",
1106 tokens => [ { t => "TEXT" }, ... ] },
1108 length => TEXTLEN };
1110 If $use_refs is set to 1, the reading object is returned in the table
1111 instead of READINGTEXT; if not, the text of the reading is returned.
1113 If $include_witnesses is set to a hashref, only the witnesses whose sigil
1114 keys have a true hash value will be included.
1118 sub alignment_table {
1120 $self->calculate_ranks() unless $self->_graphcalc_done;
1121 return $self->cached_table if $self->has_cached_table;
1123 # Make sure we can do this
1124 throw( "Need a linear graph in order to make an alignment table" )
1125 unless $self->linear;
1126 $self->calculate_ranks unless $self->end->has_rank;
1128 my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 };
1129 my @all_pos = ( 1 .. $self->end->rank - 1 );
1130 foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) {
1131 # print STDERR "Making witness row(s) for " . $wit->sigil . "\n";
1132 my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
1133 my @row = _make_witness_row( \@wit_path, \@all_pos );
1134 push( @{$table->{'alignment'}},
1135 { 'witness' => $wit->sigil, 'tokens' => \@row } );
1136 if( $wit->is_layered ) {
1137 my @wit_ac_path = $self->reading_sequence( $self->start, $self->end,
1138 $wit->sigil.$self->ac_label );
1139 my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos );
1140 push( @{$table->{'alignment'}},
1141 { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } );
1144 $self->cached_table( $table );
1148 sub _make_witness_row {
1149 my( $path, $positions ) = @_;
1151 map { $char_hash{$_} = undef } @$positions;
1153 foreach my $rdg ( @$path ) {
1154 my $rtext = $rdg->text;
1155 $rtext = '#LACUNA#' if $rdg->is_lacuna;
1156 print STDERR "rank " . $rdg->rank . "\n" if $debug;
1157 # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank;
1158 $char_hash{$rdg->rank} = { 't' => $rdg };
1160 my @row = map { $char_hash{$_} } @$positions;
1161 # Fill in lacuna markers for undef spots in the row
1162 my $last_el = shift @row;
1163 my @filled_row = ( $last_el );
1164 foreach my $el ( @row ) {
1165 # If we are using node reference, make the lacuna node appear many times
1166 # in the table. If not, use the lacuna tag.
1167 if( $last_el && $last_el->{'t'}->is_lacuna && !defined $el ) {
1170 push( @filled_row, $el );
1176 =head1 NAVIGATION METHODS
1178 =head2 reading_sequence( $first, $last, $sigil, $backup )
1180 Returns the ordered list of readings, starting with $first and ending
1181 with $last, for the witness given in $sigil. If a $backup sigil is
1182 specified (e.g. when walking a layered witness), it will be used wherever
1183 no $sigil path exists. If there is a base text reading, that will be
1184 used wherever no path exists for $sigil or $backup.
1188 # TODO Think about returning some lazy-eval iterator.
1189 # TODO Get rid of backup; we should know from what witness is whether we need it.
1191 sub reading_sequence {
1192 my( $self, $start, $end, $witness ) = @_;
1194 $witness = $self->baselabel unless $witness;
1195 my @readings = ( $start );
1198 while( $n && $n->id ne $end->id ) {
1199 if( exists( $seen{$n->id} ) ) {
1200 throw( "Detected loop for $witness at " . $n->id );
1204 my $next = $self->next_reading( $n, $witness );
1206 throw( "Did not find any path for $witness from reading " . $n->id );
1208 push( @readings, $next );
1211 # Check that the last reading is our end reading.
1212 my $last = $readings[$#readings];
1213 throw( "Last reading found from " . $start->text .
1214 " for witness $witness is not the end!" ) # TODO do we get this far?
1215 unless $last->id eq $end->id;
1220 =head2 next_reading( $reading, $sigil );
1222 Returns the reading that follows the given reading along the given witness
1228 # Return the successor via the corresponding path.
1230 my $answer = $self->_find_linked_reading( 'next', @_ );
1231 return undef unless $answer;
1232 return $self->reading( $answer );
1235 =head2 prior_reading( $reading, $sigil )
1237 Returns the reading that precedes the given reading along the given witness
1243 # Return the predecessor via the corresponding path.
1245 my $answer = $self->_find_linked_reading( 'prior', @_ );
1246 return $self->reading( $answer );
1249 sub _find_linked_reading {
1250 my( $self, $direction, $node, $path ) = @_;
1252 # Get a backup if we are dealing with a layered witness
1254 my $aclabel = $self->ac_label;
1255 if( $path && $path =~ /^(.*)\Q$aclabel\E$/ ) {
1259 my @linked_paths = $direction eq 'next'
1260 ? $self->sequence->edges_from( $node )
1261 : $self->sequence->edges_to( $node );
1262 return undef unless scalar( @linked_paths );
1264 # We have to find the linked path that contains all of the
1265 # witnesses supplied in $path.
1266 my( @path_wits, @alt_path_wits );
1267 @path_wits = sort( $self->_witnesses_of_label( $path ) ) if $path;
1268 @alt_path_wits = sort( $self->_witnesses_of_label( $alt_path ) ) if $alt_path;
1271 foreach my $le ( @linked_paths ) {
1272 if( $self->sequence->has_edge_attribute( @$le, $self->baselabel ) ) {
1275 my @le_wits = sort $self->path_witnesses( $le );
1276 if( _is_within( \@path_wits, \@le_wits ) ) {
1277 # This is the right path.
1278 return $direction eq 'next' ? $le->[1] : $le->[0];
1279 } elsif( _is_within( \@alt_path_wits, \@le_wits ) ) {
1283 # Got this far? Return the alternate path if it exists.
1284 return $direction eq 'next' ? $alt_le->[1] : $alt_le->[0]
1287 # Got this far? Return the base path if it exists.
1288 return $direction eq 'next' ? $base_le->[1] : $base_le->[0]
1291 # Got this far? We have no appropriate path.
1292 warn "Could not find $direction node from " . $node->id
1293 . " along path $path";
1299 my( $set1, $set2 ) = @_;
1300 my $ret = @$set1; # will be 0, i.e. false, if set1 is empty
1301 foreach my $el ( @$set1 ) {
1302 $ret = 0 unless grep { /^\Q$el\E$/ } @$set2;
1307 # Return the string that joins together a list of witnesses for
1308 # display on a single path.
1309 sub _witnesses_of_label {
1310 my( $self, $label ) = @_;
1311 my $regex = $self->wit_list_separator;
1312 my @answer = split( /\Q$regex\E/, $label );
1316 =head2 common_readings
1318 Returns the list of common readings in the graph (i.e. those readings that are
1319 shared by all non-lacunose witnesses.)
1323 sub common_readings {
1325 my @common = grep { $_->is_common } $self->readings;
1329 =head2 path_text( $sigil, [, $start, $end ] )
1331 Returns the text of a witness (plus its backup, if we are using a layer)
1332 as stored in the collation. The text is returned as a string, where the
1333 individual readings are joined with spaces and the meta-readings (e.g.
1334 lacunae) are omitted. Optional specification of $start and $end allows
1335 the generation of a subset of the witness text.
1340 my( $self, $wit, $start, $end ) = @_;
1341 $start = $self->start unless $start;
1342 $end = $self->end unless $end;
1343 my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit );
1346 foreach my $r ( @path ) {
1347 if( $r->join_prior || !$last || $last->join_next ) {
1348 $pathtext .= $r->text;
1350 $pathtext .= ' ' . $r->text;
1357 =head1 INITIALIZATION METHODS
1359 These are mostly for use by parsers.
1361 =head2 make_witness_path( $witness )
1363 Link the array of readings contained in $witness->path (and in
1364 $witness->uncorrected_path if it exists) into collation paths.
1365 Clear out the arrays when finished.
1367 =head2 make_witness_paths
1369 Call make_witness_path for all witnesses in the tradition.
1373 # For use when a collation is constructed from a base text and an apparatus.
1374 # We have the sequences of readings and just need to add path edges.
1375 # When we are done, clear out the witness path attributes, as they are no
1377 # TODO Find a way to replace the witness path attributes with encapsulated functions?
1379 sub make_witness_paths {
1381 foreach my $wit ( $self->tradition->witnesses ) {
1382 # print STDERR "Making path for " . $wit->sigil . "\n";
1383 $self->make_witness_path( $wit );
1387 sub make_witness_path {
1388 my( $self, $wit ) = @_;
1389 my @chain = @{$wit->path};
1390 my $sig = $wit->sigil;
1391 # Add start and end if necessary
1392 unshift( @chain, $self->start ) unless $chain[0] eq $self->start;
1393 push( @chain, $self->end ) unless $chain[-1] eq $self->end;
1394 foreach my $idx ( 0 .. $#chain-1 ) {
1395 $self->add_path( $chain[$idx], $chain[$idx+1], $sig );
1397 if( $wit->is_layered ) {
1398 @chain = @{$wit->uncorrected_path};
1399 unshift( @chain, $self->start ) unless $chain[0] eq $self->start;
1400 push( @chain, $self->end ) unless $chain[-1] eq $self->end;
1401 foreach my $idx( 0 .. $#chain-1 ) {
1402 my $source = $chain[$idx];
1403 my $target = $chain[$idx+1];
1404 $self->add_path( $source, $target, $sig.$self->ac_label )
1405 unless $self->has_path( $source, $target, $sig );
1409 $wit->clear_uncorrected_path;
1412 =head2 calculate_ranks
1414 Calculate the reading ranks (that is, their aligned positions relative
1415 to each other) for the graph. This can only be called on linear collations.
1419 use Text::Tradition;
1421 my $cxfile = 't/data/Collatex-16.xml';
1422 my $t = Text::Tradition->new(
1424 'input' => 'CollateX',
1427 my $c = $t->collation;
1430 my $table = $c->alignment_table;
1431 ok( $c->has_cached_table, "Alignment table was cached" );
1432 is( $c->alignment_table, $table, "Cached table returned upon second call" );
1433 $c->calculate_ranks;
1434 is( $c->alignment_table, $table, "Cached table retained with no rank change" );
1435 $c->add_relationship( 'n24', 'n23', { 'type' => 'spelling' } );
1436 isnt( $c->alignment_table, $table, "Alignment table changed after relationship add" );
1442 sub calculate_ranks {
1444 # Save the existing ranks, in case we need to invalidate the cached SVG.
1446 # Walk a version of the graph where every node linked by a relationship
1447 # edge is fundamentally the same node, and do a topological ranking on
1448 # the nodes in this graph.
1449 my $topo_graph = Graph->new();
1453 foreach my $r ( $self->readings ) {
1454 next if exists $rel_containers{$r->id};
1455 my @rels = $r->related_readings( 'colocated' );
1457 # Make a relationship container.
1459 my $rn = 'rel_container_' . $rel_ctr++;
1460 $topo_graph->add_vertex( $rn );
1462 $rel_containers{$_->id} = $rn;
1465 # Add a new node to mirror the old node.
1466 $rel_containers{$r->id} = $r->id;
1467 $topo_graph->add_vertex( $r->id );
1472 foreach my $r ( $self->readings ) {
1473 $existing_ranks{$r} = $r->rank;
1474 foreach my $n ( $self->sequence->successors( $r->id ) ) {
1475 my( $tfrom, $tto ) = ( $rel_containers{$r->id},
1476 $rel_containers{$n} );
1477 # $DB::single = 1 unless $tfrom && $tto;
1478 $topo_graph->add_edge( $tfrom, $tto );
1482 # Now do the rankings, starting with the start node.
1483 my $topo_start = $rel_containers{$self->start->id};
1484 my $node_ranks = { $topo_start => 0 };
1485 my @curr_origin = ( $topo_start );
1486 # A little iterative function.
1487 while( @curr_origin ) {
1488 @curr_origin = _assign_rank( $topo_graph, $node_ranks, @curr_origin );
1490 # Transfer our rankings from the topological graph to the real one.
1491 foreach my $r ( $self->readings ) {
1492 if( defined $node_ranks->{$rel_containers{$r->id}} ) {
1493 $r->rank( $node_ranks->{$rel_containers{$r->id}} );
1495 # Die. Find the last rank we calculated.
1496 my @all_defined = sort { ( $node_ranks->{$rel_containers{$a->id}}||-1 )
1497 <=> ( $node_ranks->{$rel_containers{$b->id}}||-1 ) }
1499 my $last = pop @all_defined;
1500 throw( "Ranks not calculated after $last - do you have a cycle in the graph?" );
1503 # Do we need to invalidate the cached data?
1504 if( $self->has_cached_svg || $self->has_cached_table ) {
1505 foreach my $r ( $self->readings ) {
1506 next if defined( $existing_ranks{$r} )
1507 && $existing_ranks{$r} == $r->rank;
1508 # Something has changed, so clear the cache
1509 $self->_clear_cache;
1510 # ...and recalculate the common readings.
1511 $self->calculate_common_readings();
1515 # The graph calculation information is now up to date.
1516 $self->_graphcalc_done(1);
1520 my( $graph, $node_ranks, @current_nodes ) = @_;
1521 # Look at each of the children of @current_nodes. If all the child's
1522 # parents have a rank, assign it the highest rank + 1 and add it to
1523 # @next_nodes. Otherwise skip it; we will return when the highest-ranked
1524 # parent gets a rank.
1526 foreach my $c ( @current_nodes ) {
1527 warn "Current reading $c has no rank!"
1528 unless exists $node_ranks->{$c};
1529 # print STDERR "Looking at child of node $c, rank "
1530 # . $node_ranks->{$c} . "\n";
1531 foreach my $child ( $graph->successors( $c ) ) {
1532 next if exists $node_ranks->{$child};
1533 my $highest_rank = -1;
1535 foreach my $parent ( $graph->predecessors( $child ) ) {
1536 if( exists $node_ranks->{$parent} ) {
1537 $highest_rank = $node_ranks->{$parent}
1538 if $highest_rank <= $node_ranks->{$parent};
1545 my $c_rank = $highest_rank + 1;
1546 # print STDERR "Assigning rank $c_rank to node $child \n";
1547 $node_ranks->{$child} = $c_rank;
1548 push( @next_nodes, $child );
1556 $self->wipe_svg if $self->has_cached_svg;
1557 $self->wipe_table if $self->has_cached_table;
1561 =head2 flatten_ranks
1563 A convenience method for parsing collation data. Searches the graph for readings
1564 with the same text at the same rank, and merges any that are found.
1570 my %unique_rank_rdg;
1572 foreach my $rdg ( $self->readings ) {
1573 next unless $rdg->has_rank;
1574 my $key = $rdg->rank . "||" . $rdg->text;
1575 if( exists $unique_rank_rdg{$key} ) {
1577 # print STDERR "Combining readings at same rank: $key\n";
1579 $self->merge_readings( $unique_rank_rdg{$key}, $rdg );
1580 # TODO see if this now makes a common point.
1582 $unique_rank_rdg{$key} = $rdg;
1585 # If we merged readings, the ranks are still fine but the alignment
1586 # table is wrong. Wipe it.
1587 $self->wipe_table() if $changed;
1591 =head2 calculate_common_readings
1593 Goes through the graph identifying the readings that appear in every witness
1594 (apart from those with lacunae at that spot.) Marks them as common and returns
1599 use Text::Tradition;
1601 my $cxfile = 't/data/Collatex-16.xml';
1602 my $t = Text::Tradition->new(
1604 'input' => 'CollateX',
1607 my $c = $t->collation;
1609 my @common = $c->calculate_common_readings();
1610 is( scalar @common, 8, "Found correct number of common readings" );
1611 my @marked = sort $c->common_readings();
1612 is( scalar @common, 8, "All common readings got marked as such" );
1613 my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /;
1614 is_deeply( \@marked, \@expected, "Found correct list of common readings" );
1620 sub calculate_common_readings {
1623 map { $_->is_common( 0 ) } $self->readings;
1624 # Implicitly calls calculate_ranks
1625 my $table = $self->alignment_table;
1626 foreach my $idx ( 0 .. $table->{'length'} - 1 ) {
1627 my @row = map { $_->{'tokens'}->[$idx]
1628 ? $_->{'tokens'}->[$idx]->{'t'} : '' }
1629 @{$table->{'alignment'}};
1631 foreach my $r ( @row ) {
1633 $hash{$r->id} = $r unless $r->is_meta;
1635 $hash{'UNDEF'} = $r;
1638 if( keys %hash == 1 && !exists $hash{'UNDEF'} ) {
1639 my( $r ) = values %hash;
1641 push( @common, $r );
1647 =head2 text_from_paths
1649 Calculate the text array for all witnesses from the path, for later consistency
1650 checking. Only to be used if there is no non-graph-based way to know the
1655 sub text_from_paths {
1657 foreach my $wit ( $self->tradition->witnesses ) {
1658 my @readings = $self->reading_sequence( $self->start, $self->end, $wit->sigil );
1660 foreach my $r ( @readings ) {
1661 next if $r->is_meta;
1662 push( @text, $r->text );
1664 $wit->text( \@text );
1665 if( $wit->is_layered ) {
1666 my @ucrdgs = $self->reading_sequence( $self->start, $self->end,
1667 $wit->sigil.$self->ac_label );
1669 foreach my $r ( @ucrdgs ) {
1670 next if $r->is_meta;
1671 push( @uctext, $r->text );
1673 $wit->layertext( \@uctext );
1678 =head1 UTILITY FUNCTIONS
1680 =head2 common_predecessor( $reading_a, $reading_b )
1682 Find the last reading that occurs in sequence before both the given readings.
1684 =head2 common_successor( $reading_a, $reading_b )
1686 Find the first reading that occurs in sequence after both the given readings.
1690 use Text::Tradition;
1692 my $cxfile = 't/data/Collatex-16.xml';
1693 my $t = Text::Tradition->new(
1695 'input' => 'CollateX',
1698 my $c = $t->collation;
1700 is( $c->common_predecessor( 'n24', 'n23' )->id,
1701 'n20', "Found correct common predecessor" );
1702 is( $c->common_successor( 'n24', 'n23' )->id,
1703 '#END#', "Found correct common successor" );
1705 is( $c->common_predecessor( 'n19', 'n17' )->id,
1706 'n16', "Found correct common predecessor for readings on same path" );
1707 is( $c->common_successor( 'n21', 'n10' )->id,
1708 '#END#', "Found correct common successor for readings on same path" );
1714 ## Return the closest reading that is a predecessor of both the given readings.
1715 sub common_predecessor {
1717 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1718 return $self->_common_in_path( $r1, $r2, 'predecessors' );
1721 sub common_successor {
1723 my( $r1, $r2 ) = $self->_objectify_args( @_ );
1724 return $self->_common_in_path( $r1, $r2, 'successors' );
1727 sub _common_in_path {
1728 my( $self, $r1, $r2, $dir ) = @_;
1729 my $iter = $r1->rank > $r2->rank ? $r1->rank : $r2->rank;
1730 $iter = $self->end->rank - $iter if $dir eq 'successors';
1732 my @last_checked = ( $r1, $r2 );
1734 while( !@candidates ) {
1736 foreach my $lc ( @last_checked ) {
1737 foreach my $p ( $lc->$dir ) {
1738 if( $all_seen{$p->id} ) {
1739 push( @candidates, $p );
1741 $all_seen{$p->id} = 1;
1742 push( @new_lc, $p );
1746 @last_checked = @new_lc;
1748 my @answer = sort { $a->rank <=> $b->rank } @candidates;
1749 return $dir eq 'predecessors' ? pop( @answer ) : shift ( @answer );
1753 Text::Tradition::Error->throw(
1754 'ident' => 'Collation error',
1760 __PACKAGE__->meta->make_immutable;
1764 This package is free software and is provided "as is" without express
1765 or implied warranty. You can redistribute it and/or modify it under
1766 the same terms as Perl itself.
1770 Tara L Andrews E<lt>aurum@cpan.orgE<gt>