add exceptions to the rest of the Tradition library
[scpubgit/stemmatology.git] / lib / Text / Tradition / Collation / RelationshipStore.pm
index 39bbdc0..981fded 100644 (file)
@@ -2,13 +2,15 @@ package Text::Tradition::Collation::RelationshipStore;
 
 use strict;
 use warnings;
+use Text::Tradition::Error;
 use Text::Tradition::Collation::Relationship;
 
 use Moose;
 
 =head1 NAME
 
-Text::Tradition::Collation::Reading - represents a reading (usually a word) in a collation.
+Text::Tradition::Collation::RelationshipStore - Keeps track of the relationships
+between readings in a given collation
     
 =head1 DESCRIPTION
 
@@ -17,6 +19,14 @@ texts, particularly medieval ones.  The RelationshipStore is an internal object
 of the collation, to keep track of the defined relationships (both specific and
 general) between readings.
 
+=begin testing
+
+use Text::Tradition;
+
+use_ok( 'Text::Tradition::Collation::RelationshipStore' );
+
+=end testing
+
 =head1 METHODS
 
 =head2 new( collation => $collation );
@@ -49,6 +59,27 @@ has 'graph' => (
     },
        );
        
+=head2 get_relationship
+
+Return the relationship object, if any, that exists between two readings.
+
+=cut
+
+sub get_relationship {
+       my( $self, @vector ) = @_;
+       my $relationship;
+       if( $self->graph->has_edge_attribute( @vector, 'object' ) ) {
+               $relationship = $self->graph->get_edge_attribute( @vector, 'object' );
+       }
+       return $relationship;
+}
+
+sub _set_relationship {
+       my( $self, $relationship, @vector ) = @_;
+       $self->graph->add_edge( @vector );
+       $self->graph->set_edge_attribute( @vector, 'object', $relationship );
+}
+       
 =head2 create
 
 Create a new relationship with the given options and return it.
@@ -61,13 +92,11 @@ sub create {
        # Check to see if a relationship exists between the two given readings
        my $source = delete $options->{'orig_a'};
        my $target = delete $options->{'orig_b'};
-       my $rel;
-       if( $self->graph->has_edge( $source, $target ) ) {
-               $rel = $self->graph->get_edge_attribute( $source, $target, 'object' );
-               if( $rel->type ne $options->type ) {
-                       warn "Relationship of type " . $rel->type 
-                               . "already exists between $source and $target";
-                       return;
+       my $rel = $self->get_relationship( $source, $target );
+       if( $rel ) {
+               if( $rel->type ne $options->{'type'} ) {
+                       throw( "Another relationship of type " . $rel->type 
+                               . " already exists between $source and $target" );
                } else {
                        return $rel;
                }
@@ -79,8 +108,7 @@ sub create {
        if( $rel && $rel->type eq $options->{'type'} ) {
                return $rel;
        } elsif( $rel ) {
-               warn sprintf( "Relationship of type %s with scope %s already defined for readings %s and %s", $rel->type, $rel->scope, $options->{'reading_a'}, $options->{'reading_b'} );
-               return;
+               throw( sprintf( "Relationship of type %s with scope %s already defined for readings %s and %s", $rel->type, $rel->scope, $options->{'reading_a'}, $options->{'reading_b'} ) );
        } else {
                $rel = Text::Tradition::Collation::Relationship->new( $options );
                $self->add_scoped_relationship( $rel ) if $rel->nonlocal;
@@ -144,7 +172,7 @@ sub add_relationship {
        my( $is_valid, $reason ) = 
                $self->relationship_valid( $source, $target, $options->{'type'} );
     unless( $is_valid ) {
-        return ( undef, $reason );
+        throw( "Invalid relationship: $reason" );
     }
     
     # Try to create the relationship object.
@@ -152,8 +180,7 @@ sub add_relationship {
     $options->{'reading_b'} = $target_rdg->text;
     $options->{'orig_a'} = $source;
     $options->{'orig_b'} = $target;
-    my $relationship = $self->create( $options );
-       return( undef, "Relationship creation failed" ) unless $relationship;
+    my $relationship = $self->create( $options );  # Will throw on error
 
        # Find all the pairs for which we need to set the relationship.
        my @vectors = ( [ $source, $target ] ); 
@@ -181,23 +208,20 @@ sub add_relationship {
     # Now set the relationship(s).
     my @pairs_set;
     foreach my $v ( @vectors ) {
-       if( $self->graph->has_edge( @$v ) ) {
-               # Is it locally scoped?
-               my $rel = $self->graph->get_edge_attribute( @$v );
+               my $rel = $self->get_relationship( @$v );
+       if( $rel ) {
                if( $rel->nonlocal ) {
-                       # TODO I think we should not be able to get here.
-                       warn "Found conflicting relationship at @$v";
+                       throw( "Found conflicting relationship at @$v" );
                } else {
                        warn "Not overriding local relationship set at @$v";
-                       next;
                }
+               next;
        }
-       $self->graph->add_edge( @$v );
-       $self->graph->set_edge_attribute( @$v, 'object', $relationship );
+       $self->_set_relationship( $relationship, @$v );
        push( @pairs_set, $v );
     }
     
-    return( 1, @pairs_set );
+    return @pairs_set;
 }
 
 =head2 relationship_valid( $source, $target, $type )
@@ -262,20 +286,32 @@ sub related_readings {
                $reading = $reading->id;
                $return_object = 1;
        }
-       my @related = $self->graph->all_reachable( $reading );
+       my @answer;
        if( $colocated ) {
-               my @colo;
-               foreach my $r ( @related ) {
-                       my $obj = $self->graph->get_edge_attribute( $reading, $r, 'object' );
-                       push( @colo, $r ) if $obj->colocated;
+               my %found = ( $reading => 1 );
+               my $check = [ $reading ];
+               my $iter = 0;
+               while( @$check ) {
+                       my $more = [];
+                       foreach my $r ( @$check ) {
+                               foreach my $nr ( $self->graph->neighbors( $r ) ) {
+                                       if( $self->get_relationship( $r, $nr )->colocated ) {
+                                               push( @$more, $nr ) unless exists $found{$nr};
+                                               $found{$nr} = 1;
+                                       }
+                               }
+                       }
+                       $check = $more;
                }
-               @related = @colo;
+               @answer = keys %found;
+       } else {
+               @answer = $self->graph->all_reachable( $reading );
        }
        if( $return_object ) {
                my $c = $self->collation;
-               return map { $c->reading( $_ ) } @related;
+               return map { $c->reading( $_ ) } @answer;
        } else {
-               return @related;
+               return @answer;
        }
 }
 
@@ -298,19 +334,81 @@ sub merge_readings {
                next if $combined;
                        
                # If kept / rel already has a relationship, warn and keep the old
-               if( $self->graph->has_edge( @vector ) ) {
+               my $rel = $self->get_relationship( @vector );
+               if( $rel ) {
                        warn sprintf( "Readings %s and %s have existing relationship; dropping link with %s", @vector, $deleted );
                        next;
                }
                
                # Otherwise, adopt the relationship that would be deleted.
-               my $rel = $self->graph->get_edge_attribute( @$edge, 'object' );
-               $self->graph->add_edge( @vector );
-               $self->graph->set_edge_attribute( @vector, 'object', $rel );
+               $rel = $self->get_relationship( @$edge );
+               $self->_set_relationship( $rel, @vector );
        }
        $self->delete_reading( $deleted );
 }
 
+sub as_graphml { 
+       my( $self, $graphml_ns, $xmlroot, $node_hash, $nodeid_key, $edge_keys ) = @_;
+       
+    my $rgraph = $xmlroot->addNewChild( $graphml_ns, 'graph' );
+       $rgraph->setAttribute( 'edgedefault', 'directed' );
+    $rgraph->setAttribute( 'id', 'relationships', );
+    $rgraph->setAttribute( 'parse.edgeids', 'canonical' );
+    $rgraph->setAttribute( 'parse.edges', scalar($self->graph->edges) );
+    $rgraph->setAttribute( 'parse.nodeids', 'canonical' );
+    $rgraph->setAttribute( 'parse.nodes', scalar($self->graph->vertices) );
+    $rgraph->setAttribute( 'parse.order', 'nodesfirst' );
+    
+    # Add the vertices according to their XML IDs
+    my %rdg_lookup = ( reverse %$node_hash );
+    foreach my $n ( sort _by_xmlid keys( %rdg_lookup ) ) {
+       my $n_el = $rgraph->addNewChild( $graphml_ns, 'node' );
+       $n_el->setAttribute( 'id', $n );
+       _add_graphml_data( $n_el, $nodeid_key, $rdg_lookup{$n} );
+    }
+    
+    # Add the relationship edges, with their object information
+    my $edge_ctr = 0;
+    foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->graph->edges ) {
+       # Add an edge and fill in its relationship info.
+               my $edge_el = $rgraph->addNewChild( $graphml_ns, 'edge' );
+               $edge_el->setAttribute( 'source', $node_hash->{$e->[0]} );
+               $edge_el->setAttribute( 'target', $node_hash->{$e->[1]} );
+               $edge_el->setAttribute( 'id', 'e'.$edge_ctr++ );
+
+               my $rel_obj = $self->get_relationship( @$e );
+               _add_graphml_data( $edge_el, $edge_keys->{'relationship'}, $rel_obj->type );
+               _add_graphml_data( $edge_el, $edge_keys->{'scope'}, $rel_obj->scope );
+               _add_graphml_data( $edge_el, $edge_keys->{'non_correctable'}, 
+                       $rel_obj->non_correctable ) if $rel_obj->noncorr_set;
+               _add_graphml_data( $edge_el, $edge_keys->{'non_independent'}, 
+                       $rel_obj->non_independent ) if $rel_obj->nonind_set;
+       }
+}
+
+sub _by_xmlid {
+       my $tmp_a = $a;
+       my $tmp_b = $b;
+       $tmp_a =~ s/\D//g;
+       $tmp_b =~ s/\D//g;
+       return $tmp_a <=> $tmp_b;
+}
+
+sub _add_graphml_data {
+    my( $el, $key, $value ) = @_;
+    return unless defined $value;
+    my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
+    $data_el->setAttribute( 'key', $key );
+    $data_el->appendText( $value );
+}
+
+sub throw {
+       Text::Tradition::Error->throw( 
+               'ident' => 'Relationship error',
+               'message' => $_[0],
+               );
+}
+
 no Moose;
 __PACKAGE__->meta->make_immutable;