use Encode qw( decode_utf8 );
use File::chdir;
use File::Temp;
+use File::Which;
use Graph;
-use Graph::Convert;
use Graph::Reader::Dot;
use IPC::Run qw/ run binary /;
+use Text::Tradition::StemmaUtil qw/ phylip_pars_input /;
use Moose;
-use Text::Balanced qw/ extract_bracketed /;
has collation => (
is => 'ro',
predicate => 'has_distance_trees',
);
+has distance_program => (
+ is => 'rw',
+ isa => 'Str',
+ default => '',
+ );
+
sub BUILD {
my( $self, $args ) = @_;
# If we have been handed a dotfile, initialize it into a graph.
binmode( $dotfh, ':utf8' );
my $reader = Graph::Reader::Dot->new();
my $graph = $reader->read_graph( $dotfh );
- $graph
- ? $self->graph( $graph )
- : warn "Failed to parse dot in $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 ) = @_;
- # TODO add options for display, someday
- # TODO see what happens with Graph::Writer::Dot 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' );
- $n->set_attribute( 'pointshape', 'diamond' );
+
+ # Get default and specified options
+ my %graphopts = ();
+ my %nodeopts = (
+ 'fontsize' => 11,
+ 'hshape' => 'plaintext', # Shape for the hypothetical nodes
+ 'htext' => '*',
+ 'style' => 'filled',
+ 'fillcolor' => 'white',
+ 'shape' => 'ellipse', # Shape for the extant nodes
+ );
+ my %edgeopts = (
+ 'arrowhead' => 'open',
+ );
+ @graphopts{ keys %{$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 {
- $n->set_attribute( 'shape', 'ellipse' );
+ # Use the default display settings.
+ push( @dotlines, " $n;" );
}
}
+ # Add each of our edges.
+ foreach my $e ( $self->graph->edges ) {
+ my( $from, $to ) = @$e;
+ push( @dotlines, " $from -> $to;" );
+ }
+ push( @dotlines, '}' );
- # Render to svg via graphviz
- my @lines = split( /\n/, $dgraph->as_graphviz() );
- # Add the size attribute
- if( $opts->{'size'} ) {
- my $sizeline = " graph [ size=\"" . $opts->{'size'} . "\" ]";
- splice( @lines, 1, 0, $sizeline );
+ 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 ) );
+ }
}
- return join( "\n", @lines );
+ # 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 {
before 'distance_trees' => sub {
my $self = shift;
- my %args = @_;
+ my %args = (
+ 'program' => 'phylip_pars',
+ @_ );
# TODO allow specification of method for calculating distance tree
- if( $args{'recalc'} || !$self->has_distance_trees ) {
+ if( !$self->has_distance_trees
+ || $args{'program'} ne $self->distance_program ) {
# We need to make a tree before we can return it.
- my( $ok, $result ) = $self->run_phylip_pars();
+ my $dsub = 'run_' . $args{'program'};
+ my( $ok, $result ) = $self->$dsub();
if( $ok ) {
# Save the resulting trees
my $trees = _parse_newick( $result );
$self->_save_distance_trees( $trees );
+ $self->distance_program( $args{'program'} );
} else {
warn "Failed to calculate distance trees: $result";
}
}
};
-
-sub make_character_matrix {
- my $self = shift;
- unless( $self->collation->linear ) {
- warn "Need a linear graph in order to make an alignment table";
- return;
- }
- 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]};
- foreach my $token_index ( 1 .. $#{$table} ) {
- # First implementation: make dumb alignment table, caring about
- # nothing except which reading is in which position.
- my @chars = convert_characters( $table->[$token_index] );
- foreach my $idx ( 0 .. $#matrix ) {
- push( @{$matrix[$idx]}, $chars[$idx] );
- }
- }
- return \@matrix;
-}
-
-sub _normalize_ac {
- my( $self, $witname ) = @_;
- my $ac = $self->collation->ac_label;
- if( $witname =~ /(.*)\Q$ac\E$/ ) {
- $witname = $1 . '_ac';
- }
- return sprintf( "%-10s", $witname );
-}
-
-sub convert_characters {
- my $row = shift;
- # This is a simple algorithm that treats every reading as different.
- # Eventually we will want to be able to specify how relationships
- # affect the character matrix.
- 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;
- return @chars;
-}
-
-sub phylip_pars_input {
- my $self = shift;
- 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 $input;
-}
sub run_phylip_pars {
my $self = shift;
# $phylip_dir->unlink_on_destroy(0);
# We need an infile, and we need a command input file.
open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
- print MATRIX $self->phylip_pars_input();
+ print MATRIX phylip_pars_input( $self->collation->make_alignment_table() );
close MATRIX;
open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
close CMD;
# And then we run the program.
- ### HACKY HACKY
- my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe';
- my $program = "pars";
- if( $^O eq 'darwin' ) {
- $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program";
- } else {
- $program = "$PHYLIP_PATH/$program";
+ my $program = File::Which::which( 'pars' );
+ unless( -x $program ) {
+ return( undef, "Phylip pars not found in path" );
}
{
projects / scpubgit/stemmatology.git / blobdiff