provide for stemma graphic display

[scpubgit/stemmatology.git] / lib / Text / Tradition / Stemma.pm

diff --git a/lib/Text/Tradition/Stemma.pm b/lib/Text/Tradition/Stemma.pm
index df8cf4a..e04811c 100644 (file)
--- a/lib/Text/Tradition/Stemma.pm
+++ b/lib/Text/Tradition/Stemma.pm
@@ -1,12 +1,15 @@
 package Text::Tradition::Stemma;
 
+use Bio::Phylo::IO;
+use Encode qw( decode_utf8 );
 use File::chdir;
 use File::Temp;
-use IPC::Run qw/ run /;
-use Moose;
-use Text::Tradition::Collation::Position;
 use Graph;
+use Graph::Convert;
 use Graph::Reader::Dot;
+use IPC::Run qw/ run binary /;
+use Moose;
+use Text::Balanced qw/ extract_bracketed /;
 
 has collation => (
     is => 'ro',
@@ -31,13 +34,24 @@ 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'} ) {
+        # Open the file, assume UTF-8
+        open( my $dot, $args->{'dot'} ) or warn "Failed to read dot file";
+        # TODO don't bother if we haven't opened
+        binmode $dot, ":utf8";
         my $reader = Graph::Reader::Dot->new();
-        my $graph = $reader->read_graph( $args->{'dot'} );
+        my $graph = $reader->read_graph( $dot );
         $graph 
             ? $self->graph( $graph ) 
             : warn "Failed to parse dot file " . $args->{'dot'};
@@ -62,7 +76,53 @@ sub BUILD {
         $self->apsp( $undirected->APSP_Floyd_Warshall() );
     }
 }
-        
+
+# Render the stemma as SVG.
+sub as_svg {
+    my $self = shift;
+    # TODO add options for display, someday
+    my $dgraph = Graph::Convert->as_graph_easy( $self->graph );
+    # Set some class display attributes for 'hypothetical' and 'extant' nodes
+    $dgraph->set_attribute( 'flow', 'south' );
+    foreach my $n ( $dgraph->nodes ) {
+        if( $n->attribute( 'class' ) eq 'hypothetical' ) {
+            $n->set_attribute( 'shape', 'point' );
+            $n->set_attribute( 'pointshape', 'diamond' );
+        } else {
+            $n->set_attribute( 'shape', 'ellipse' );
+        }
+    }
+    
+    # Render to svg via graphviz
+    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 $dgraph->as_graphviz();
+    push( @cmd, $dotfile->filename );
+    run( \@cmd, ">", binary(), \$svg );
+    $svg = decode_utf8( $svg );
+    return $svg;
+}
+
+#### Methods for calculating phylogenetic trees ####
+
+before 'distance_trees' => sub {
+    my $self = shift;
+    my %args = @_;
+    # TODO allow specification of method for calculating distance tree
+    if( $args{'recalc'} || !$self->has_distance_trees ) {
+        # 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 ) );
+        } else {
+            warn "Failed to calculate distance tree: $result";
+        }
+    }
+};
         
 sub make_character_matrix {
     my $self = shift;
@@ -110,13 +170,13 @@ sub convert_characters {
         }
     }
     if( scalar( keys %unique ) > 8 ) {
-        warn "Have more than 8 variants on this location; pars will break";
+        warn "Have more than 8 variants on this location; phylip will break";
     }
     my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row;
     return @chars;
 }
 
-sub pars_input {
+sub phylip_pars_input {
     my $self = shift;
     $self->make_character_matrix unless $self->has_character_matrix;
     my $matrix = '';
@@ -129,16 +189,15 @@ sub pars_input {
     return $matrix;
 }
 
-sub run_pars {
+sub run_phylip_pars {
     my $self = shift;
 
     # Set up a temporary directory for all the default Phylip files.
     my $phylip_dir = File::Temp->newdir();
-    print STDERR $phylip_dir . "\n";
     # $phylip_dir->unlink_on_destroy(0);
     # We need an infile, and we need a command input file.
     open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
-    print MATRIX $self->pars_input();
+    print MATRIX $self->phylip_pars_input();
     close MATRIX;
 
     open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
@@ -201,6 +260,46 @@ sub run_pars {
     return( undef, join( '', @error ) );
 }
 
+sub _parse_newick {
+    my $newick = shift;
+    my @trees;
+    # Parse the result into a tree
+    my $forest = Bio::Phylo::IO->parse( 
+        -format => 'newick',
+        -string => $newick,
+        );
+    # Turn the tree into a graph, starting with the root node
+    foreach my $tree ( @{$forest->get_entities} ) {
+        push( @trees, _graph_from_bio( $tree ) );
+    }
+    return \@trees;
+}
+
+sub _graph_from_bio {
+    my $tree = shift;
+    my $graph = Graph->new( 'undirected' => 1 );
+    # Give all the intermediate anonymous nodes a name.
+    my $i = 0;
+    foreach my $n ( @{$tree->get_entities} ) {
+        next if $n->get_name;
+        $n->set_name( $i++ );
+    }
+    my $root = $tree->get_root->get_name;
+    $graph->add_vertex( $root );
+    _add_tree_children( $graph, $root, $tree->get_root->get_children() );
+    return $graph;
+}
+
+sub _add_tree_children {
+    my( $graph, $parent, $tree_children ) = @_;
+    foreach my $c ( @$tree_children ) {
+        my $child = $c->get_name;
+        $graph->add_vertex( $child );
+        $graph->add_path( $parent, $child );
+        _add_tree_children( $graph, $child, $c->get_children() );
+    }
+}
+
 no Moose;
 __PACKAGE__->meta->make_immutable;