package Text::Tradition::Stemma;
use Bio::Phylo::IO;
+use Encode qw( decode_utf8 );
use File::chdir;
use File::Temp;
use Graph;
use Graph::Reader::Dot;
-use IPC::Run qw/ run /;
+use IPC::Run qw/ run binary /;
use Moose;
-use Text::Balanced qw/ extract_bracketed /;
has collation => (
is => 'ro',
isa => 'Text::Tradition::Collation',
required => 1,
+ weak_ref => 1,
);
-has character_matrix => (
- is => 'ro',
- isa => 'ArrayRef[ArrayRef[Str]]',
- writer => '_save_character_matrix',
- predicate => 'has_character_matrix',
- );
-
has graph => (
is => 'rw',
isa => 'Graph',
predicate => 'has_graph',
);
-has apsp => (
- is => 'rw',
- isa => 'Graph',
- );
-
has distance_trees => (
is => 'ro',
isa => 'ArrayRef[Graph]',
writer => '_save_distance_trees',
predicate => 'has_distance_trees',
);
-
+
sub BUILD {
my( $self, $args ) = @_;
# If we have been handed a dotfile, initialize it into a graph.
if( exists $args->{'dot'} ) {
- my $reader = Graph::Reader::Dot->new();
- my $graph = $reader->read_graph( $args->{'dot'} );
- $graph
- ? $self->graph( $graph )
- : warn "Failed to parse dot file " . $args->{'dot'};
+ $self->graph_from_dot( $args->{'dot'} );
}
+}
+
+sub graph_from_dot {
+ my( $self, $dotfh ) = @_;
+ # Assume utf-8
+ binmode( $dotfh, ':utf8' );
+ my $reader = Graph::Reader::Dot->new();
+ my $graph = $reader->read_graph( $dotfh );
+ if( $graph ) {
+ $self->graph( $graph );
+ # Go through the nodes and set any non-hypothetical node to extant.
+ foreach my $v ( $self->graph->vertices ) {
+ $self->graph->set_vertex_attribute( $v, 'class', 'extant' )
+ unless $self->graph->has_vertex_attribute( $v, 'class' );
+ }
+ } else {
+ warn "Failed to parse dot in $dotfh";
+ }
+}
+
+sub as_dot {
+ my( $self, $opts ) = @_;
- # If we have a graph, calculate all the shortest paths between nodes,
- # disregarding direction.
- if( $self->has_graph ) {
- my $undirected;
- if( $self->graph->is_directed ) {
- # Make an undirected version.
- $undirected = Graph->new( 'undirected' => 1 );
- foreach my $v ( $self->graph->vertices ) {
- $undirected->add_vertex( $v );
- }
- foreach my $e ( $self->graph->edges ) {
- $undirected->add_edge( @$e );
- }
+ # Get default and specified options
+ my %graphopts = ();
+ my %nodeopts = (
+ 'fontsize' => 11,
+ 'hshape' => 'plaintext', # Shape for the hypothetical nodes
+ 'htext' => '*',
+ 'style' => 'filled',
+ 'fillcolor' => 'white',
+ 'shape' => 'ellipse', # Shape for the extant nodes
+ );
+ my %edgeopts = (
+ 'arrowhead' => 'open',
+ );
+ @graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}}
+ if $opts->{'graph'};
+ @nodeopts{ keys %{$opts->{'node'}} } = values %{$opts->{'node'}}
+ if $opts->{'node'};
+ @edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}}
+ if $opts->{'edge'};
+
+ my @dotlines;
+ push( @dotlines, 'digraph stemma {' );
+ ## Print out the global attributes
+ push( @dotlines, _make_dotline( 'graph', %graphopts ) ) if keys %graphopts;
+ push( @dotlines, _make_dotline( 'edge', %edgeopts ) ) if keys %edgeopts;
+ ## Delete our special attributes from the node set before continuing
+ my $hshape = delete $nodeopts{'hshape'};
+ my $htext = delete $nodeopts{'htext'};
+ push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts;
+
+ # Add each of the nodes.
+ foreach my $n ( $self->graph->vertices ) {
+ if( $self->graph->get_vertex_attribute( $n, 'class' ) eq 'hypothetical' ) {
+ # Apply our display settings for hypothetical nodes.
+ push( @dotlines, _make_dotline( $n, 'shape' => $hshape, 'label' => $htext ) );
} else {
- $undirected = $self->graph;
+ # Use the default display settings.
+ push( @dotlines, " $n;" );
}
- $self->apsp( $undirected->APSP_Floyd_Warshall() );
}
+ # Add each of our edges.
+ foreach my $e ( $self->graph->edges ) {
+ my( $from, $to ) = @$e;
+ push( @dotlines, " $from -> $to;" );
+ }
+ push( @dotlines, '}' );
+
+ return join( "\n", @dotlines );
+}
+
+
+# Another version of dot output meant for graph editing, thus
+# much simpler.
+sub editable {
+ my $self = shift;
+ my @dotlines;
+ push( @dotlines, 'digraph stemma {' );
+ my @real; # A cheap sort
+ foreach my $n ( sort $self->graph->vertices ) {
+ my $c = $self->graph->get_vertex_attribute( $n, 'class' );
+ $c = 'extant' unless $c;
+ if( $c eq 'extant' ) {
+ push( @real, $n );
+ } else {
+ push( @dotlines, _make_dotline( $n, 'class' => $c ) );
+ }
+ }
+ # Now do the real ones
+ foreach my $n ( @real ) {
+ push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) );
+ }
+ foreach my $e ( sort _by_vertex $self->graph->edges ) {
+ my( $from, $to ) = @$e;
+ push( @dotlines, " $from -> $to;" );
+ }
+ push( @dotlines, '}' );
+ return join( "\n", @dotlines );
+}
+
+sub _make_dotline {
+ my( $obj, %attr ) = @_;
+ my @pairs;
+ foreach my $k ( keys %attr ) {
+ my $v = $attr{$k};
+ $v =~ s/\"/\\\"/g;
+ push( @pairs, "$k=\"$v\"" );
+ }
+ return sprintf( " %s [ %s ];", $obj, join( ', ', @pairs ) );
+}
+
+sub _by_vertex {
+ return $a->[0].$a->[1] cmp $b->[0].$b->[1];
}
+# Render the stemma as SVG.
+sub as_svg {
+ my( $self, $opts ) = @_;
+ my $dot = $self->as_dot( $opts );
+ my @cmd = qw/dot -Tsvg/;
+ my( $svg, $err );
+ my $dotfile = File::Temp->new();
+ ## TODO REMOVE
+ # $dotfile->unlink_on_destroy(0);
+ binmode $dotfile, ':utf8';
+ print $dotfile $dot;
+ push( @cmd, $dotfile->filename );
+ run( \@cmd, ">", binary(), \$svg );
+ $svg = decode_utf8( $svg );
+ return $svg;
+}
+
+sub witnesses {
+ my $self = shift;
+ my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
+ $self->graph->vertices;
+ return @wits;
+}
+
+#### Methods for calculating phylogenetic trees ####
+
before 'distance_trees' => sub {
my $self = shift;
my %args = @_;
# We need to make a tree before we can return it.
my( $ok, $result ) = $self->run_phylip_pars();
if( $ok ) {
- $self->_save_distance_trees( _parse_newick( $result ) );
+ # Save the resulting trees
+ my $trees = _parse_newick( $result );
+ $self->_save_distance_trees( $trees );
} else {
- warn "Failed to calculate distance tree: $result";
+ warn "Failed to calculate distance trees: $result";
}
}
};
my $table = $self->collation->make_alignment_table;
# Push the names of the witnesses to initialize the rows of the matrix.
my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]};
- $DB::single = 1;
foreach my $token_index ( 1 .. $#{$table} ) {
# First implementation: make dumb alignment table, caring about
# nothing except which reading is in which position.
push( @{$matrix[$idx]}, $chars[$idx] );
}
}
- $self->_save_character_matrix( \@matrix );
+ return \@matrix;
}
sub _normalize_ac {
my %unique = ( '__UNDEF__' => 'X',
'#LACUNA#' => '?',
);
+ my %count;
my $ctr = 0;
foreach my $word ( @$row ) {
if( $word && !exists $unique{$word} ) {
$unique{$word} = chr( 65 + $ctr );
$ctr++;
}
+ $count{$word}++ if $word;
}
+ # Try to keep variants under 8 by lacunizing any singletons.
if( scalar( keys %unique ) > 8 ) {
+ foreach my $word ( keys %count ) {
+ if( $count{$word} == 1 ) {
+ $unique{$word} = '?';
+ }
+ }
+ }
+ my %u = reverse %unique;
+ if( scalar( keys %u ) > 8 ) {
warn "Have more than 8 variants on this location; phylip will break";
}
my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row;
sub phylip_pars_input {
my $self = shift;
- $self->make_character_matrix unless $self->has_character_matrix;
- my $matrix = '';
- my $rows = scalar @{$self->character_matrix};
- my $columns = scalar @{$self->character_matrix->[0]} - 1;
- $matrix .= "\t$rows\t$columns\n";
- foreach my $row ( @{$self->character_matrix} ) {
- $matrix .= join( '', @$row ) . "\n";
+ my $character_matrix = $self->make_character_matrix;
+ my $input = '';
+ my $rows = scalar @{$character_matrix};
+ my $columns = scalar @{$character_matrix->[0]} - 1;
+ $input .= "\t$rows\t$columns\n";
+ foreach my $row ( @{$character_matrix} ) {
+ $input .= join( '', @$row ) . "\n";
}
- return $matrix;
+ return $input;
}
sub run_phylip_pars {