my( $self, $reading ) = @_;
unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
my %args = %$reading;
+ if( $self->tradition->has_language && !exists $args{'language'} ) {
+ $args{'language'} = $self->tradition->language;
+ }
$reading = Text::Tradition::Collation::Reading->new(
'collation' => $self,
%args );
$self->_add_reading( $reading->id => $reading );
# Once the reading has been added, put it in both graphs.
$self->sequence->add_vertex( $reading->id );
- $self->relations->add_reading( $reading->id );
+ # All meta readings save 'start' and 'end' get disregarded for relationships.
+ unless( $reading->is_nonrel ) {
+ $self->relations->add_reading( $reading->id );
+ }
return $reading;
};
my $self = shift;
my $arg = shift;
- if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
- $arg = $arg->id;
+ unless( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
+ $arg = $self->reading( $arg )
}
+ my $argid = $arg->id;
# Remove the reading from the graphs.
$self->_graphcalc_done(0);
$self->_clear_cache; # Explicitly clear caches to GC the reading
- $self->sequence->delete_vertex( $arg );
- $self->relations->delete_reading( $arg );
+ $self->sequence->delete_vertex( $argid );
+ $self->relations->delete_reading( $argid )
+ unless $arg->is_nonrel;
# Carry on.
- $self->$orig( $arg );
+ $self->$orig( $argid );
};
=begin testing
sub merge_readings {
my $self = shift;
+ # Sanity check
+ my( $kept_obj, $del_obj, $combine, $combine_char ) = $self->_objectify_args( @_ );
+ my $mergemeta = $kept_obj->is_meta;
+ throw( "Cannot merge meta and non-meta reading" )
+ unless ( $mergemeta && $del_obj->is_meta )
+ || ( !$mergemeta && !$del_obj->is_meta );
+ if( $mergemeta ) {
+ throw( "Cannot merge with start or end node" )
+ if( $kept_obj eq $self->start || $kept_obj eq $self->end
+ || $del_obj eq $self->start || $del_obj eq $self->end );
+ }
# We only need the IDs for adding paths to the graph, not the reading
# objects themselves.
- my( $kept, $deleted, $combine, $combine_char ) = $self->_stringify_args( @_ );
+ my $kept = $kept_obj->id;
+ my $deleted = $del_obj->id;
$self->_graphcalc_done(0);
-
+
# The kept reading should inherit the paths and the relationships
# of the deleted reading.
foreach my $path ( $self->sequence->edges_at( $deleted ) ) {
@wits{keys %$fwits} = values %$fwits;
$self->sequence->set_edge_attributes( @vector, \%wits );
}
- $self->relations->merge_readings( $kept, $deleted, $combine_char );
+ $self->relations->merge_readings( $kept, $deleted, $combine )
+ unless $mergemeta;
# Do the deletion deed.
if( $combine ) {
- my $kept_obj = $self->reading( $kept );
- my $del_obj = $self->reading( $deleted );
my $joinstr = $combine_char;
unless( defined $joinstr ) {
$joinstr = '' if $kept_obj->join_next || $del_obj->join_prior;
# We only need the IDs for adding paths to the graph, not the reading
# objects themselves.
- my( $source, $target, $wit ) = $self->_stringify_args( @_ );
+ my( $source, $target, $wit ) = $self->_objectify_args( @_ );
$self->_graphcalc_done(0);
# Connect the readings
throw( "Need GraphViz installed to output SVG" )
unless File::Which::which( 'dot' );
my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'};
- $self->calculate_ranks() unless( $self->_graphcalc_done || $opts->{'nocalc'} );
+ $self->calculate_ranks()
+ unless( $self->_graphcalc_done || $opts->{'nocalc'} || !$self->linear );
if( !$self->has_cached_svg || $opts->{'recalc'} || $want_subgraph ) {
my @cmd = qw/dot -Tsvg/;
my( $svg, $err );
# Output substitute start/end readings if necessary
if( $startrank ) {
- $dot .= "\t\"#SUBSTART#\" [ label=\"...\" ];\n";
+ $dot .= "\t\"__SUBSTART__\" [ label=\"...\",id=\"__START__\" ];\n";
}
if( $endrank ) {
- $dot .= "\t\"#SUBEND#\" [ label=\"...\" ];\n";
+ $dot .= "\t\"__SUBEND__\" [ label=\"...\",id=\"__END__\" ];\n";
}
if( $STRAIGHTENHACK ) {
## HACK part 1
- my $startlabel = $startrank ? 'SUBSTART' : 'START';
- $dot .= "\tsubgraph { rank=same \"#$startlabel#\" \"#SILENT#\" }\n";
+ my $startlabel = $startrank ? '__SUBSTART__' : '__START__';
+ $dot .= "\tsubgraph { rank=same \"$startlabel\" \"#SILENT#\" }\n";
$dot .= "\t\"#SILENT#\" [ shape=diamond,color=white,penwidth=0,label=\"\" ];"
}
my %used; # Keep track of the readings that actually appear in the graph
$label = "-$label" if $reading->join_prior;
$label =~ s/\"/\\\"/g;
$rattrs->{'label'} = $label;
+ $rattrs->{'id'} = $reading->id;
$rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common;
$dot .= sprintf( "\t\"%s\" %s;\n", $reading->id, _dot_attr_string( $rattrs ) );
}
my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
my $variables = { %edge_attrs, 'label' => $witstr };
my $varopts = _dot_attr_string( $variables );
- $dot .= "\t\"#SUBSTART#\" -> \"$node\" $varopts;";
+ $dot .= "\t\"__SUBSTART__\" -> \"$node\" $varopts;";
}
foreach my $node ( keys %subend ) {
my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) );
my $variables = { %edge_attrs, 'label' => $witstr };
my $varopts = _dot_attr_string( $variables );
- $dot .= "\t\"$node\" -> \"#SUBEND#\" $varopts;";
+ $dot .= "\t\"$node\" -> \"__SUBEND__\" $varopts;";
}
# HACK part 2
if( $STRAIGHTENHACK ) {
- my $endlabel = $endrank ? 'SUBEND' : 'END';
- $dot .= "\t\"#$endlabel#\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
+ my $endlabel = $endrank ? '__SUBEND__' : '__END__';
+ $dot .= "\t\"$endlabel\" -> \"#SILENT#\" [ color=white,penwidth=0 ];\n";
}
$dot .= "}\n";
# See if we are in a majority situation.
my $maj = scalar( $self->tradition->witnesses ) * 0.6;
+ $maj = $maj > 5 ? $maj : 5;
if( scalar keys %wits > $maj ) {
unshift( @disp_ac, 'majority' );
return join( ', ', @disp_ac );
is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
+# Now add a stemma, write to GraphML, and parse again.
+my $stemma = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' );
+is( ref( $stemma ), 'Text::Tradition::Stemma', "Parsed dotfile into stemma" );
+is( $tradition->stemmata, 1, "Tradition now has the stemma" );
+$graphml = $c->as_graphml;
+like( $graphml, qr/digraph/, "Digraph declaration exists in GraphML" );
+
=end testing
=cut
'Str' => 'string',
'Int' => 'int',
'Bool' => 'boolean',
+ 'ReadingID' => 'string',
'RelationshipType' => 'string',
'RelationshipScope' => 'string',
);
next unless $save_types{$attr->type_constraint->name};
$graph_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
}
+ # Extra custom key for the tradition stemma(ta)
+ $graph_attributes{'stemmata'} = 'string';
foreach my $datum ( sort keys %graph_attributes ) {
$graph_data_keys{$datum} = 'dg'.$gdi++;
next unless $save_types{$attr->type_constraint->name};
$reading_attributes{$attr->name} = $save_types{$attr->type_constraint->name};
}
+ # Extra custom key for the reading morphology
+ $reading_attributes{'lexemes'} = 'string';
+
my %node_data_keys;
my $ndi = 0;
foreach my $datum ( sort keys %reading_attributes ) {
$sgraph->setAttribute( 'parse.nodes', 0 ); # fill in later
$sgraph->setAttribute( 'parse.order', 'nodesfirst' );
- # Collation attribute data
+ # Tradition/collation attribute data
foreach my $datum ( keys %graph_attributes ) {
my $value;
if( $datum eq 'version' ) {
- $value = '3.1';
+ $value = '3.2';
+ } elsif( $datum eq 'stemmata' ) {
+ my @stemstrs;
+ map { push( @stemstrs, $_->editable( {linesep => ''} ) ) }
+ $self->tradition->stemmata;
+ $value = join( "\n", @stemstrs );
} elsif( $gattr_from{$datum} eq 'Tradition' ) {
$value = $self->tradition->$datum;
} else {
$node_el->setAttribute( 'id', $node_xmlid );
foreach my $d ( keys %reading_attributes ) {
my $nval = $n->$d;
+ # Custom serialization
+ if( $d eq 'lexemes' ) {
+ # If nval is a true value, we have lexemes so we need to
+ # serialize them. Otherwise set nval to undef so that the
+ # key is excluded from this reading.
+ $nval = $nval ? $n->_serialize_lexemes : undef;
+ } elsif( $d eq 'normal_form' && $n->normal_form eq $n->text ) {
+ $nval = undef;
+ }
if( $rankoffset && $d eq 'rank' && $n ne $self->start ) {
# Adjust the ranks within the subgraph.
$nval = $n eq $self->end ? $end->rank - $rankoffset + 1
$start = $self->start unless $start;
$end = $self->end unless $end;
my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit );
- return _text_from_path( @path );
-}
-
-# Utility function so that we can cheat and use it when we need both the path
-# and its text.
-sub _text_from_path {
- my( $self, @path ) = @_;
my $pathtext = '';
my $last;
foreach my $r ( @path ) {
- if( $r->join_prior || !$last || $last->join_next ) {
- $pathtext .= $r->text;
- } else {
- $pathtext .= ' ' . $r->text;
- }
+ unless ( $r->join_prior || !$last || $last->join_next ) {
+ $pathtext .= ' ';
+ }
+ $pathtext .= $r->text;
$last = $r;
}
return $pathtext;
# Do the rankings based on the relationship equivalence graph, starting
# with the start node.
- my $topo_start = $self->equivalence( $self->start->id );
- my $node_ranks = { $topo_start => 0 };
- my @curr_origin = ( $topo_start );
- # A little iterative function.
- while( @curr_origin ) {
- @curr_origin = _assign_rank( $self->equivalence_graph,
- $node_ranks, @curr_origin );
- }
+ my ( $node_ranks, $rank_nodes ) = $self->relations->equivalence_ranks();
+
# Transfer our rankings from the topological graph to the real one.
foreach my $r ( $self->readings ) {
- if( defined $node_ranks->{$self->equivalence( $r->id )} ) {
+ if( $r->is_nonrel ) {
+ # These are not in the equivalence graph. Grab the rank of the highest
+ # predecessor + 1.
+ my @preds = $self->sequence->predecessors( $r );
+ my $mrank = 0;
+ map { my $rk = $node_ranks->{$self->equivalence( $_ )} + 1;
+ $mrank = $rk > $mrank ? $rk : $mrank; }
+ $self->sequence->predecessors( $r );
+ throw( "All predecessors of $r unranked!" ) unless $mrank;
+ $r->rank( $mrank );
+ } elsif( defined $node_ranks->{$self->equivalence( $r->id )} ) {
$r->rank( $node_ranks->{$self->equivalence( $r->id )} );
} else {
# Die. Find the last rank we calculated.
$self->_graphcalc_done(1);
}
-sub _assign_rank {
- my( $graph, $node_ranks, @current_nodes ) = @_;
- # Look at each of the children of @current_nodes. If all the child's
- # parents have a rank, assign it the highest rank + 1 and add it to
- # @next_nodes. Otherwise skip it; we will return when the highest-ranked
- # parent gets a rank.
- my @next_nodes;
- foreach my $c ( @current_nodes ) {
- warn "Current reading $c has no rank!"
- unless exists $node_ranks->{$c};
- # print STDERR "Looking at child of node $c, rank "
- # . $node_ranks->{$c} . "\n";
- foreach my $child ( $graph->successors( $c ) ) {
- next if exists $node_ranks->{$child};
- my $highest_rank = -1;
- my $skip = 0;
- foreach my $parent ( $graph->predecessors( $child ) ) {
- if( exists $node_ranks->{$parent} ) {
- $highest_rank = $node_ranks->{$parent}
- if $highest_rank <= $node_ranks->{$parent};
- } else {
- $skip = 1;
- last;
- }
- }
- next if $skip;
- my $c_rank = $highest_rank + 1;
- # print STDERR "Assigning rank $c_rank to node $child \n";
- $node_ranks->{$child} = $c_rank;
- push( @next_nodes, $child );
- }
- }
- return @next_nodes;
-}
-
sub _clear_cache {
my $self = shift;
$self->wipe_svg if $self->has_cached_svg;
is( $c->common_predecessor( 'n24', 'n23' )->id,
'n20', "Found correct common predecessor" );
is( $c->common_successor( 'n24', 'n23' )->id,
- '#END#', "Found correct common successor" );
+ '__END__', "Found correct common successor" );
is( $c->common_predecessor( 'n19', 'n17' )->id,
'n16', "Found correct common predecessor for readings on same path" );
is( $c->common_successor( 'n21', 'n10' )->id,
- '#END#', "Found correct common successor for readings on same path" );
+ '__END__', "Found correct common successor for readings on same path" );
=end testing
projects / scpubgit/stemmatology.git / blobdiff