[scpubgit/stemmatology.git] / lib / Text / Tradition / Collation / RelationshipStore.pm

package Text::Tradition::Collation::RelationshipStore;

use strict;
use warnings;
use Text::Tradition::Error;
use Text::Tradition::Collation::Relationship;
use TryCatch;

use Moose;

=head1 NAME

Text::Tradition::Collation::RelationshipStore - Keeps track of the relationships
between readings in a given collation
    
=head1 DESCRIPTION

Text::Tradition is a library for representation and analysis of collated
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 TryCatch;

use_ok( 'Text::Tradition::Collation::RelationshipStore' );

# Add some relationships, and delete them

my $cxfile = 't/data/Collatex-16.xml';
my $t = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'CollateX',
    'file'  => $cxfile,
    );
my $c = $t->collation;

my @v1 = $c->add_relationship( 'n21', 'n22', { 'type' => 'meaning' } );
is( scalar @v1, 1, "Added a single relationship" );
is( $v1[0]->[0], 'n21', "Got correct node 1" );
is( $v1[0]->[1], 'n22', "Got correct node 2" );
my @v2 = $c->add_relationship( 'n9', 'n23', 
	{ 'type' => 'spelling', 'scope' => 'global' } );
is( scalar @v2, 2, "Added a global relationship with two instances" );
@v1 = $c->del_relationship( 'n22', 'n21' );
is( scalar @v1, 1, "Deleted first relationship" );
@v2 = $c->del_relationship( 'n8', 'n13' );
is( scalar @v2, 2, "Deleted second global relationship" );
try {
	my @v3 = $c->del_relationship( 'n1', 'n2' );
	ok( 0, "Should have errored on non-existent relationship" );
} catch( Text::Tradition::Error $e ) {
	like( $e->message, qr/No relationship defined/, "Attempt to delete non-existent relationship errored" );
}

=end testing

=head1 METHODS

=head2 new( collation => $collation );

Creates a new relationship store for the given collation.

=cut

has 'collation' => (
	is => 'ro',
	isa => 'Text::Tradition::Collation',
	required => 1,
	weak_ref => 1,
	);

has 'scopedrels' => (
	is => 'ro',
	isa => 'HashRef[HashRef[Text::Tradition::Collation::Relationship]]',
	default => sub { {} },
	);

has 'graph' => (
	is => 'ro',
	isa => 'Graph',
	default => sub { Graph->new( undirected => 1 ) },
    handles => {
    	relationships => 'edges',
    	add_reading => 'add_vertex',
    	delete_reading => 'delete_vertex',
    },
	);
	
=head2 get_relationship

Return the relationship object, if any, that exists between two readings.

=cut

sub get_relationship {
	my $self = shift;
	my @vector;
	if( @_ == 1 && ref( $_[0] ) eq 'ARRAY' ) {
		# Dereference the edge arrayref that was passed.
		my $edge = shift;
		@vector = @$edge;
	} else {
		@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.
Warn and return undef if the relationship cannot be created.

=cut

sub create {
	my( $self, $options ) = @_;
	# 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 = $self->get_relationship( $source, $target );
	if( $rel ) {
		if( $rel->type eq 'collated' ) {
			# Always replace a 'collated' relationship with a more descriptive
			# one, if asked.
			$self->del_relationship( $source, $target );
		} elsif( $rel->type ne $options->{'type'} ) {
			throw( "Another relationship of type " . $rel->type 
				. " already exists between $source and $target" );
		} else {
			return $rel;
		}
	}
	
	# Check to see if a nonlocal relationship is defined for the two readings
	$rel = $self->scoped_relationship( $options->{'reading_a'}, 
		$options->{'reading_b'} );
	if( $rel && $rel->type eq $options->{'type'} ) {
		return $rel;
	} elsif( $rel ) {
		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;
		return $rel;
	}
}

=head2 add_scoped_relationship( $rel )

Keep track of relationships defined between specific readings that are scoped
non-locally.  Key on whichever reading occurs first alphabetically.

=cut

sub add_scoped_relationship {
	my( $self, $rel ) = @_;
	my $r = $self->scoped_relationship( $rel->reading_a, $rel->reading_b );
	if( $r ) {
		warn sprintf( "Scoped relationship of type %s already exists between %s and %s",
			$r->type, $rel->reading_a, $rel->reading_b );
		return;
	}
	$self->scopedrels->{$rel->reading_a}->{$rel->reading_b} = $rel;
}

=head2 scoped_relationship( $reading_a, $reading_b )

Returns the general (document-level or global) relationship that has been defined 
between the two reading strings. Returns undef if there is no general relationship.

=cut

sub scoped_relationship {
	my( $self, $rdga, $rdgb ) = @_;
	my( $first, $second ) = sort( $rdga, $rdgb );
	if( exists $self->scopedrels->{$first}->{$second} ) {
		return $self->scopedrels->{$first}->{$second};
	} else {
		return undef;
	}
}

=head2 add_relationship( $self, $source, $sourcetext, $target, $targettext, $opts )

Adds the relationship specified in $opts (see Text::Tradition::Collation::Relationship 
for the possible options) between the readings given in $source and $target.  Sets
up a scoped relationship between $sourcetext and $targettext if the relationship is
scoped non-locally.

Returns a status boolean and a list of all reading pairs connected by the call to
add_relationship.

=cut

sub add_relationship {
	my( $self, $source, $source_rdg, $target, $target_rdg, $options ) = @_;

	my $relationship;
	my $thispaironly;
	if( ref( $options ) eq 'Text::Tradition::Collation::Relationship' ) {
		$relationship = $options;
		$thispaironly = 1;  # If existing rel, set only where asked.
	} else {
		# Check the options
		$options->{'scope'} = 'local' unless $options->{'scope'};
		
		my( $is_valid, $reason ) = 
			$self->relationship_valid( $source, $target, $options->{'type'} );
		unless( $is_valid ) {
			throw( "Invalid relationship: $reason" );
		}
		
		# Try to create the relationship object.
		$options->{'reading_a'} = $source_rdg->text;
		$options->{'reading_b'} = $target_rdg->text;
		$options->{'orig_a'} = $source;
		$options->{'orig_b'} = $target;
    	if( $options->{'scope'} ne 'local' ) {
			# Is there a relationship with this a & b already?
			my $otherrel = $self->scoped_relationship( $options->{reading_a}, 
				$options->{reading_b} );
			if( $otherrel && $otherrel->type eq $options->{type}
				&& $otherrel->scope eq $options->{scope} ) {
				warn "Applying existing scoped relationship";
				$relationship = $otherrel;
			}
    	}
		$relationship = $self->create( $options ) unless $relationship;  # Will throw on error
    }


	# Find all the pairs for which we need to set the relationship.
	my @vectors = ( [ $source, $target ] );	
    if( $relationship->colocated && $relationship->nonlocal && !$thispaironly ) {
    	my $c = $self->collation;
    	# Set the same relationship everywhere we can, throughout the graph.
    	my @identical_readings = grep { $_->text eq $relationship->reading_a }
    		$c->readings;
    	foreach my $ir ( @identical_readings ) {
    		next if $ir->id eq $source;
    		# Check to see if there is a target reading with the same text at
    		# the same rank.
    		my @itarget = grep 
    			{ $_->rank == $ir->rank && $_->text eq $relationship->reading_b }
    			$c->readings;
    		if( @itarget ) {
    			# We found a hit.
    			warn "More than one reading with text " . $target_rdg->text
    				. " at rank " . $ir->rank . "!" if @itarget > 1;
    			push( @vectors, [ $ir->id, $itarget[0]->id ] );
    		}
    	}	
    }
    
    # Now set the relationship(s).
    my @pairs_set;
    foreach my $v ( @vectors ) {
		my $rel = $self->get_relationship( @$v );
    	if( $rel && $rel ne $relationship ) {
    		if( $rel->nonlocal ) {
    			$DB::single = 1;
    			throw( "Found conflicting relationship at @$v" );
    		} else {
    			warn "Not overriding local relationship set at @$v";
    		}
    		next;
    	}
    	$self->_set_relationship( $relationship, @$v );
    	push( @pairs_set, $v );
    }
    
    return @pairs_set;
}

=head2 del_relationship( $source, $target )

Removes the relationship between the given readings. If the relationship is
non-local, removes the relationship everywhere in the graph.

=cut

sub del_relationship {
	my( $self, $source, $target ) = @_;
	my $rel = $self->get_relationship( $source, $target );
	throw( "No relationship defined between $source and $target" ) unless $rel;
	my @vectors = ( [ $source, $target ] );
	$self->_remove_relationship( $source, $target );
	if( $rel->nonlocal ) {
		# Remove the relationship wherever it occurs.
		my @rel_edges = grep { $self->get_relationship( @$_ ) == $rel }
			$self->relationships;
		foreach my $re ( @rel_edges ) {
			$self->_remove_relationship( @$re );
			push( @vectors, $re );
		}
	}
	return @vectors;
}

sub _remove_relationship {
	my( $self, @vector ) = @_;
	$self->graph->delete_edge( @vector );
}
	
=head2 relationship_valid( $source, $target, $type )

Checks whether a relationship of type $type may exist between the readings given
in $source and $target.  Returns a tuple of ( status, message ) where status is
a yes/no boolean and, if the answer is no, message gives the reason why.

=cut

sub relationship_valid {
    my( $self, $source, $target, $rel ) = @_;
    my $c = $self->collation;
    if ( $rel eq 'transposition' || $rel eq 'repetition' ) {
		# Check that the two readings do (for a repetition) or do not (for
		# a transposition) appear in the same witness.
		# TODO this might be called before witness paths are set...
		my %seen_wits;
		map { $seen_wits{$_} = 1 } $c->reading_witnesses( $source );
		foreach my $w ( $c->reading_witnesses( $target ) ) {
			if( $seen_wits{$w} ) {
				return ( 0, "Readings both occur in witness $w" ) 
					if $rel eq 'transposition';
				return ( 1, "ok" ) if $rel eq 'repetition';
		}
		return $rel eq 'transposition' ? ( 1, "ok" )
			: ( 0, "Readings occur only in distinct witnesses" );
		}
	} else {
		# Check that linking the source and target in a relationship won't lead
		# to a path loop for any witness.  If they have the same rank then fine.
		return( 1, "ok" ) 
			if $c->reading( $source )->has_rank
				&& $c->reading( $target )->has_rank
				&& $c->reading( $source )->rank == $c->reading( $target )->rank;
		
		# Otherwise, first make a lookup table of all the
		# readings related to either the source or the target.
		my @proposed_related = ( $source, $target );
		push( @proposed_related, $self->related_readings( $source, 'colocated' ) );
		push( @proposed_related, $self->related_readings( $target, 'colocated' ) );
		my %pr_ids;
		map { $pr_ids{ $_ } = 1 } @proposed_related;
	
		# The cumulative predecessors and successors of the proposed-related readings
		# should not overlap.
		my %all_pred;
		my %all_succ;
		foreach my $pr ( keys %pr_ids ) {
			map { $all_pred{$_} = 1 } $c->sequence->all_predecessors( $pr );
			map { $all_succ{$_} = 1 } $c->sequence->all_successors( $pr );
		}
		foreach my $k ( keys %all_pred ) {
			return( 0, "Relationship would create witness loop" )
				if exists $all_succ{$k};
		}
		foreach my $k ( keys %pr_ids ) {
			return( 0, "Relationship would create witness loop" )
				if exists $all_pred{$k} || exists $all_succ{$k};
		}
		return ( 1, "ok" );
	}
}

=head2 related_readings( $reading, $filter )

Returns a list of readings that are connected via relationship links to $reading.
If $filter is set to a subroutine ref, returns only those related readings where
$filter( $relationship ) returns a true value.

=cut

sub related_readings {
	my( $self, $reading, $filter ) = @_;
	my $return_object;
	if( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
		$reading = $reading->id;
		$return_object = 1;
	}
	my @answer;
	if( $filter ) {
		# Backwards compat
		if( $filter eq 'colocated' ) {
			$filter = sub { $_[0]->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( $r ) ) {
					if( &$filter( $self->get_relationship( $r, $nr ) ) ) {
						push( @$more, $nr ) unless exists $found{$nr};
						$found{$nr} = 1;
					}
				}
			}
			$check = $more;
		}
		delete $found{$reading};
		@answer = keys %found;
	} else {
		@answer = $self->graph->all_reachable( $reading );
	}
	if( $return_object ) {
		my $c = $self->collation;
		return map { $c->reading( $_ ) } @answer;
	} else {
		return @answer;
	}
}

=head2 merge_readings( $kept, $deleted );

Makes a best-effort merge of the relationship links between the given readings, and
stops tracking the to-be-deleted reading.

=cut

sub merge_readings {
	my( $self, $kept, $deleted, $combined ) = @_;
	foreach my $edge ( $self->graph->edges_at( $deleted ) ) {
		# Get the pair of kept / rel
		my @vector = ( $kept );
		push( @vector, $edge->[0] eq $deleted ? $edge->[1] : $edge->[0] );
		next if $vector[0] eq $vector[1]; # Don't add a self loop
		
		# If kept changes its text, drop the relationship.
		next if $combined;
			
		# If kept / rel already has a relationship, warn and keep the old
		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.
		$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->{'annotation'}, $rel_obj->annotation );
		_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;

1;
Commit	Line	Data
22222af9	1	package Text::Tradition::Collation::RelationshipStore;
	2
	3	use strict;
	4	use warnings;
63778331	5	use Text::Tradition::Error;
22222af9	6	use Text::Tradition::Collation::Relationship;
a1615ee4	7	use TryCatch;
22222af9	8
	9	use Moose;
	10
	11	=head1 NAME
	12
2626f709	13	Text::Tradition::Collation::RelationshipStore - Keeps track of the relationships
2626f709	14	between readings in a given collation
22222af9	15
	16	=head1 DESCRIPTION
	17
	18	Text::Tradition is a library for representation and analysis of collated
	19	texts, particularly medieval ones. The RelationshipStore is an internal object
	20	of the collation, to keep track of the defined relationships (both specific and
	21	general) between readings.
	22
3ae5e2ad	23	=begin testing
	24
	25	use Text::Tradition;
ee801e17	26	use TryCatch;
3ae5e2ad	27
	28	use_ok( 'Text::Tradition::Collation::RelationshipStore' );
	29
ee801e17	30	# Add some relationships, and delete them
	31
	32	my $cxfile = 't/data/Collatex-16.xml';
	33	my $t = Text::Tradition->new(
	34	'name' => 'inline',
	35	'input' => 'CollateX',
	36	'file' => $cxfile,
	37	);
	38	my $c = $t->collation;
	39
	40	my @v1 = $c->add_relationship( 'n21', 'n22', { 'type' => 'meaning' } );
	41	is( scalar @v1, 1, "Added a single relationship" );
	42	is( $v1[0]->[0], 'n21', "Got correct node 1" );
	43	is( $v1[0]->[1], 'n22', "Got correct node 2" );
	44	my @v2 = $c->add_relationship( 'n9', 'n23',
	45	{ 'type' => 'spelling', 'scope' => 'global' } );
	46	is( scalar @v2, 2, "Added a global relationship with two instances" );
	47	@v1 = $c->del_relationship( 'n22', 'n21' );
	48	is( scalar @v1, 1, "Deleted first relationship" );
	49	@v2 = $c->del_relationship( 'n8', 'n13' );
	50	is( scalar @v2, 2, "Deleted second global relationship" );
	51	try {
	52	my @v3 = $c->del_relationship( 'n1', 'n2' );
	53	ok( 0, "Should have errored on non-existent relationship" );
	54	} catch( Text::Tradition::Error $e ) {
	55	like( $e->message, qr/No relationship defined/, "Attempt to delete non-existent relationship errored" );
	56	}
	57
3ae5e2ad	58	=end testing
3ae5e2ad	59
22222af9	60	=head1 METHODS
	61
	62	=head2 new( collation => $collation );
	63
	64	Creates a new relationship store for the given collation.
	65
	66	=cut
	67
	68	has 'collation' => (
	69	is => 'ro',
	70	isa => 'Text::Tradition::Collation',
	71	required => 1,
	72	weak_ref => 1,
	73	);
	74
	75	has 'scopedrels' => (
	76	is => 'ro',
	77	isa => 'HashRef[HashRef[Text::Tradition::Collation::Relationship]]',
	78	default => sub { {} },
	79	);
	80
	81	has 'graph' => (
	82	is => 'ro',
	83	isa => 'Graph',
	84	default => sub { Graph->new( undirected => 1 ) },
	85	handles => {
	86	relationships => 'edges',
	87	add_reading => 'add_vertex',
	88	delete_reading => 'delete_vertex',
	89	},
	90	);
	91
3ae5e2ad	92	=head2 get_relationship
	93
	94	Return the relationship object, if any, that exists between two readings.
	95
	96	=cut
	97
	98	sub get_relationship {
4633f9e4	99	my $self = shift;
	100	my @vector;
	101	if( @_ == 1 && ref( $_[0] ) eq 'ARRAY' ) {
	102	# Dereference the edge arrayref that was passed.
	103	my $edge = shift;
	104	@vector = @$edge;
	105	} else {
	106	@vector = @_;
	107	}
3ae5e2ad	108	my $relationship;
	109	if( $self->graph->has_edge_attribute( @vector, 'object' ) ) {
	110	$relationship = $self->graph->get_edge_attribute( @vector, 'object' );
ca6e6095	111	}
3ae5e2ad	112	return $relationship;
	113	}
	114
	115	sub _set_relationship {
	116	my( $self, $relationship, @vector ) = @_;
	117	$self->graph->add_edge( @vector );
	118	$self->graph->set_edge_attribute( @vector, 'object', $relationship );
	119	}
a1615ee4	120
22222af9	121	=head2 create
	122
	123	Create a new relationship with the given options and return it.
	124	Warn and return undef if the relationship cannot be created.
	125
	126	=cut
	127
	128	sub create {
	129	my( $self, $options ) = @_;
	130	# Check to see if a relationship exists between the two given readings
	131	my $source = delete $options->{'orig_a'};
	132	my $target = delete $options->{'orig_b'};
3ae5e2ad	133	my $rel = $self->get_relationship( $source, $target );
3ae5e2ad	134	if( $rel ) {
3d14b48e	135	if( $rel->type eq 'collated' ) {
	136	# Always replace a 'collated' relationship with a more descriptive
	137	# one, if asked.
	138	$self->del_relationship( $source, $target );
	139	} elsif( $rel->type ne $options->{'type'} ) {
63778331	140	throw( "Another relationship of type " . $rel->type
63778331	141	. " already exists between $source and $target" );
22222af9	142	} else {
	143	return $rel;
	144	}
	145	}
	146
	147	# Check to see if a nonlocal relationship is defined for the two readings
	148	$rel = $self->scoped_relationship( $options->{'reading_a'},
	149	$options->{'reading_b'} );
	150	if( $rel && $rel->type eq $options->{'type'} ) {
	151	return $rel;
	152	} elsif( $rel ) {
63778331	153	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'} ) );
22222af9	154	} else {
	155	$rel = Text::Tradition::Collation::Relationship->new( $options );
	156	$self->add_scoped_relationship( $rel ) if $rel->nonlocal;
	157	return $rel;
	158	}
	159	}
	160
	161	=head2 add_scoped_relationship( $rel )
	162
	163	Keep track of relationships defined between specific readings that are scoped
	164	non-locally. Key on whichever reading occurs first alphabetically.
	165
	166	=cut
	167
	168	sub add_scoped_relationship {
	169	my( $self, $rel ) = @_;
	170	my $r = $self->scoped_relationship( $rel->reading_a, $rel->reading_b );
	171	if( $r ) {
	172	warn sprintf( "Scoped relationship of type %s already exists between %s and %s",
	173	$r->type, $rel->reading_a, $rel->reading_b );
	174	return;
	175	}
	176	$self->scopedrels->{$rel->reading_a}->{$rel->reading_b} = $rel;
	177	}
	178
	179	=head2 scoped_relationship( $reading_a, $reading_b )
	180
	181	Returns the general (document-level or global) relationship that has been defined
	182	between the two reading strings. Returns undef if there is no general relationship.
	183
	184	=cut
	185
	186	sub scoped_relationship {
	187	my( $self, $rdga, $rdgb ) = @_;
	188	my( $first, $second ) = sort( $rdga, $rdgb );
	189	if( exists $self->scopedrels->{$first}->{$second} ) {
	190	return $self->scopedrels->{$first}->{$second};
	191	} else {
	192	return undef;
	193	}
	194	}
	195
	196	=head2 add_relationship( $self, $source, $sourcetext, $target, $targettext, $opts )
	197
	198	Adds the relationship specified in $opts (see Text::Tradition::Collation::Relationship
	199	for the possible options) between the readings given in $source and $target. Sets
	200	up a scoped relationship between $sourcetext and $targettext if the relationship is
	201	scoped non-locally.
	202
	203	Returns a status boolean and a list of all reading pairs connected by the call to
	204	add_relationship.
	205
	206	=cut
	207
	208	sub add_relationship {
	209	my( $self, $source, $source_rdg, $target, $target_rdg, $options ) = @_;
	210
ca6e6095	211	my $relationship;
	212	my $thispaironly;
	213	if( ref( $options ) eq 'Text::Tradition::Collation::Relationship' ) {
	214	$relationship = $options;
	215	$thispaironly = 1; # If existing rel, set only where asked.
	216	} else {
	217	# Check the options
	218	$options->{'scope'} = 'local' unless $options->{'scope'};
	219
	220	my( $is_valid, $reason ) =
	221	$self->relationship_valid( $source, $target, $options->{'type'} );
	222	unless( $is_valid ) {
	223	throw( "Invalid relationship: $reason" );
	224	}
	225
	226	# Try to create the relationship object.
	227	$options->{'reading_a'} = $source_rdg->text;
	228	$options->{'reading_b'} = $target_rdg->text;
	229	$options->{'orig_a'} = $source;
	230	$options->{'orig_b'} = $target;
0ac5e750	231	if( $options->{'scope'} ne 'local' ) {
	232	# Is there a relationship with this a & b already?
	233	my $otherrel = $self->scoped_relationship( $options->{reading_a},
	234	$options->{reading_b} );
	235	if( $otherrel && $otherrel->type eq $options->{type}
	236	&& $otherrel->scope eq $options->{scope} ) {
	237	warn "Applying existing scoped relationship";
	238	$relationship = $otherrel;
	239	}
	240	}
	241	$relationship = $self->create( $options ) unless $relationship; # Will throw on error
22222af9	242	}
ca6e6095	243
22222af9	244
	245	# Find all the pairs for which we need to set the relationship.
	246	my @vectors = ( [ $source, $target ] );
ca6e6095	247	if( $relationship->colocated && $relationship->nonlocal && !$thispaironly ) {
22222af9	248	my $c = $self->collation;
	249	# Set the same relationship everywhere we can, throughout the graph.
	250	my @identical_readings = grep { $_->text eq $relationship->reading_a }
	251	$c->readings;
	252	foreach my $ir ( @identical_readings ) {
cf6c01be	253	next if $ir->id eq $source;
22222af9	254	# Check to see if there is a target reading with the same text at
	255	# the same rank.
	256	my @itarget = grep
	257	{ $_->rank == $ir->rank && $_->text eq $relationship->reading_b }
	258	$c->readings;
	259	if( @itarget ) {
	260	# We found a hit.
	261	warn "More than one reading with text " . $target_rdg->text
	262	. " at rank " . $ir->rank . "!" if @itarget > 1;
cf6c01be	263	push( @vectors, [ $ir->id, $itarget[0]->id ] );
22222af9	264	}
	265	}
	266	}
	267
	268	# Now set the relationship(s).
	269	my @pairs_set;
	270	foreach my $v ( @vectors ) {
3ae5e2ad	271	my $rel = $self->get_relationship( @$v );
ca6e6095	272	if( $rel && $rel ne $relationship ) {
63778331	273	if( $rel->nonlocal ) {
ca6e6095	274	$DB::single = 1;
63778331	275	throw( "Found conflicting relationship at @$v" );
	276	} else {
	277	warn "Not overriding local relationship set at @$v";
	278	}
3ae5e2ad	279	next;
22222af9	280	}
3ae5e2ad	281	$self->_set_relationship( $relationship, @$v );
22222af9	282	push( @pairs_set, $v );
	283	}
	284
63778331	285	return @pairs_set;
22222af9	286	}
22222af9	287
ee801e17	288	=head2 del_relationship( $source, $target )
	289
	290	Removes the relationship between the given readings. If the relationship is
	291	non-local, removes the relationship everywhere in the graph.
	292
	293	=cut
	294
	295	sub del_relationship {
	296	my( $self, $source, $target ) = @_;
	297	my $rel = $self->get_relationship( $source, $target );
	298	throw( "No relationship defined between $source and $target" ) unless $rel;
	299	my @vectors = ( [ $source, $target ] );
	300	$self->_remove_relationship( $source, $target );
	301	if( $rel->nonlocal ) {
	302	# Remove the relationship wherever it occurs.
	303	my @rel_edges = grep { $self->get_relationship( @$_ ) == $rel }
	304	$self->relationships;
	305	foreach my $re ( @rel_edges ) {
	306	$self->_remove_relationship( @$re );
	307	push( @vectors, $re );
	308	}
	309	}
	310	return @vectors;
	311	}
	312
ca6e6095	313	sub _remove_relationship {
	314	my( $self, @vector ) = @_;
	315	$self->graph->delete_edge( @vector );
	316	}
	317
22222af9	318	=head2 relationship_valid( $source, $target, $type )
	319
	320	Checks whether a relationship of type $type may exist between the readings given
	321	in $source and $target. Returns a tuple of ( status, message ) where status is
	322	a yes/no boolean and, if the answer is no, message gives the reason why.
	323
	324	=cut
	325
	326	sub relationship_valid {
	327	my( $self, $source, $target, $rel ) = @_;
	328	my $c = $self->collation;
	329	if ( $rel eq 'transposition' \|\| $rel eq 'repetition' ) {
	330	# Check that the two readings do (for a repetition) or do not (for
	331	# a transposition) appear in the same witness.
32e95735	332	# TODO this might be called before witness paths are set...
22222af9	333	my %seen_wits;
	334	map { $seen_wits{$_} = 1 } $c->reading_witnesses( $source );
	335	foreach my $w ( $c->reading_witnesses( $target ) ) {
	336	if( $seen_wits{$w} ) {
	337	return ( 0, "Readings both occur in witness $w" )
	338	if $rel eq 'transposition';
	339	return ( 1, "ok" ) if $rel eq 'repetition';
	340	}
	341	return $rel eq 'transposition' ? ( 1, "ok" )
	342	: ( 0, "Readings occur only in distinct witnesses" );
	343	}
	344	} else {
	345	# Check that linking the source and target in a relationship won't lead
a1615ee4	346	# to a path loop for any witness. If they have the same rank then fine.
a1615ee4	347	return( 1, "ok" )
84d4ca78	348	if $c->reading( $source )->has_rank
	349	&& $c->reading( $target )->has_rank
	350	&& $c->reading( $source )->rank == $c->reading( $target )->rank;
a1615ee4	351
a1615ee4	352	# Otherwise, first make a lookup table of all the
22222af9	353	# readings related to either the source or the target.
	354	my @proposed_related = ( $source, $target );
	355	push( @proposed_related, $self->related_readings( $source, 'colocated' ) );
	356	push( @proposed_related, $self->related_readings( $target, 'colocated' ) );
	357	my %pr_ids;
	358	map { $pr_ids{ $_ } = 1 } @proposed_related;
	359
a1615ee4	360	# The cumulative predecessors and successors of the proposed-related readings
	361	# should not overlap.
	362	my %all_pred;
	363	my %all_succ;
22222af9	364	foreach my $pr ( keys %pr_ids ) {
a1615ee4	365	map { $all_pred{$_} = 1 } $c->sequence->all_predecessors( $pr );
	366	map { $all_succ{$_} = 1 } $c->sequence->all_successors( $pr );
	367	}
	368	foreach my $k ( keys %all_pred ) {
	369	return( 0, "Relationship would create witness loop" )
	370	if exists $all_succ{$k};
	371	}
	372	foreach my $k ( keys %pr_ids ) {
	373	return( 0, "Relationship would create witness loop" )
	374	if exists $all_pred{$k} \|\| exists $all_succ{$k};
	375	}
22222af9	376	return ( 1, "ok" );
	377	}
	378	}
	379
7f52eac8	380	=head2 related_readings( $reading, $filter )
22222af9	381
22222af9	382	Returns a list of readings that are connected via relationship links to $reading.
7f52eac8	383	If $filter is set to a subroutine ref, returns only those related readings where
7f52eac8	384	$filter( $relationship ) returns a true value.
22222af9	385
	386	=cut
	387
	388	sub related_readings {
7f52eac8	389	my( $self, $reading, $filter ) = @_;
22222af9	390	my $return_object;
	391	if( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) {
	392	$reading = $reading->id;
	393	$return_object = 1;
	394	}
c84275ff	395	my @answer;
7f52eac8	396	if( $filter ) {
	397	# Backwards compat
	398	if( $filter eq 'colocated' ) {
	399	$filter = sub { $_[0]->colocated };
	400	}
c84275ff	401	my %found = ( $reading => 1 );
	402	my $check = [ $reading ];
	403	my $iter = 0;
	404	while( @$check ) {
c84275ff	405	my $more = [];
	406	foreach my $r ( @$check ) {
	407	foreach my $nr ( $self->graph->neighbors( $r ) ) {
7f52eac8	408	if( &$filter( $self->get_relationship( $r, $nr ) ) ) {
c84275ff	409	push( @$more, $nr ) unless exists $found{$nr};
	410	$found{$nr} = 1;
	411	}
	412	}
	413	}
	414	$check = $more;
22222af9	415	}
7f52eac8	416	delete $found{$reading};
c84275ff	417	@answer = keys %found;
	418	} else {
	419	@answer = $self->graph->all_reachable( $reading );
22222af9	420	}
	421	if( $return_object ) {
	422	my $c = $self->collation;
c84275ff	423	return map { $c->reading( $_ ) } @answer;
22222af9	424	} else {
c84275ff	425	return @answer;
22222af9	426	}
	427	}
	428
	429	=head2 merge_readings( $kept, $deleted );
	430
	431	Makes a best-effort merge of the relationship links between the given readings, and
	432	stops tracking the to-be-deleted reading.
	433
	434	=cut
	435
	436	sub merge_readings {
	437	my( $self, $kept, $deleted, $combined ) = @_;
	438	foreach my $edge ( $self->graph->edges_at( $deleted ) ) {
	439	# Get the pair of kept / rel
	440	my @vector = ( $kept );
	441	push( @vector, $edge->[0] eq $deleted ? $edge->[1] : $edge->[0] );
	442	next if $vector[0] eq $vector[1]; # Don't add a self loop
	443
	444	# If kept changes its text, drop the relationship.
	445	next if $combined;
	446
	447	# If kept / rel already has a relationship, warn and keep the old
3ae5e2ad	448	my $rel = $self->get_relationship( @vector );
3ae5e2ad	449	if( $rel ) {
22222af9	450	warn sprintf( "Readings %s and %s have existing relationship; dropping link with %s", @vector, $deleted );
	451	next;
	452	}
	453
	454	# Otherwise, adopt the relationship that would be deleted.
3ae5e2ad	455	$rel = $self->get_relationship( @$edge );
3ae5e2ad	456	$self->_set_relationship( $rel, @vector );
22222af9	457	}
	458	$self->delete_reading( $deleted );
	459	}
	460
027d819c	461	sub _as_graphml {
2626f709	462	my( $self, $graphml_ns, $xmlroot, $node_hash, $nodeid_key, $edge_keys ) = @_;
c84275ff	463
	464	my $rgraph = $xmlroot->addNewChild( $graphml_ns, 'graph' );
	465	$rgraph->setAttribute( 'edgedefault', 'directed' );
	466	$rgraph->setAttribute( 'id', 'relationships', );
	467	$rgraph->setAttribute( 'parse.edgeids', 'canonical' );
	468	$rgraph->setAttribute( 'parse.edges', scalar($self->graph->edges) );
	469	$rgraph->setAttribute( 'parse.nodeids', 'canonical' );
	470	$rgraph->setAttribute( 'parse.nodes', scalar($self->graph->vertices) );
	471	$rgraph->setAttribute( 'parse.order', 'nodesfirst' );
	472
	473	# Add the vertices according to their XML IDs
2626f709	474	my %rdg_lookup = ( reverse %$node_hash );
2626f709	475	foreach my $n ( sort _by_xmlid keys( %rdg_lookup ) ) {
c84275ff	476	my $n_el = $rgraph->addNewChild( $graphml_ns, 'node' );
c84275ff	477	$n_el->setAttribute( 'id', $n );
2626f709	478	_add_graphml_data( $n_el, $nodeid_key, $rdg_lookup{$n} );
c84275ff	479	}
	480
	481	# Add the relationship edges, with their object information
	482	my $edge_ctr = 0;
	483	foreach my $e ( sort { $a->[0] cmp $b->[0] } $self->graph->edges ) {
	484	# Add an edge and fill in its relationship info.
	485	my $edge_el = $rgraph->addNewChild( $graphml_ns, 'edge' );
	486	$edge_el->setAttribute( 'source', $node_hash->{$e->[0]} );
	487	$edge_el->setAttribute( 'target', $node_hash->{$e->[1]} );
	488	$edge_el->setAttribute( 'id', 'e'.$edge_ctr++ );
	489
3ae5e2ad	490	my $rel_obj = $self->get_relationship( @$e );
c84275ff	491	_add_graphml_data( $edge_el, $edge_keys->{'relationship'}, $rel_obj->type );
c84275ff	492	_add_graphml_data( $edge_el, $edge_keys->{'scope'}, $rel_obj->scope );
3d14b48e	493	_add_graphml_data( $edge_el, $edge_keys->{'annotation'}, $rel_obj->annotation );
c84275ff	494	_add_graphml_data( $edge_el, $edge_keys->{'non_correctable'},
	495	$rel_obj->non_correctable ) if $rel_obj->noncorr_set;
	496	_add_graphml_data( $edge_el, $edge_keys->{'non_independent'},
	497	$rel_obj->non_independent ) if $rel_obj->nonind_set;
	498	}
	499	}
	500
	501	sub _by_xmlid {
2626f709	502	my $tmp_a = $a;
	503	my $tmp_b = $b;
	504	$tmp_a =~ s/\D//g;
	505	$tmp_b =~ s/\D//g;
	506	return $tmp_a <=> $tmp_b;
c84275ff	507	}
	508
	509	sub _add_graphml_data {
	510	my( $el, $key, $value ) = @_;
	511	return unless defined $value;
	512	my $data_el = $el->addNewChild( $el->namespaceURI, 'data' );
	513	$data_el->setAttribute( 'key', $key );
	514	$data_el->appendText( $value );
83d5ac3a	515	}
83d5ac3a	516
63778331	517	sub throw {
	518	Text::Tradition::Error->throw(
	519	'ident' => 'Relationship error',
	520	'message' => $_[0],
	521	);
	522	}
	523
22222af9	524	no Moose;
	525	__PACKAGE__->meta->make_immutable;
	526
	527	1;