use Bio::Phylo::IO;
use Encode qw( decode_utf8 );
-use File::chdir;
use File::Temp;
-use File::Which;
use Graph;
use Graph::Reader::Dot;
use IPC::Run qw/ run binary /;
-use Text::Tradition::StemmaUtil qw/ phylip_pars_input /;
+use Text::Tradition::Error;
+use Text::Tradition::StemmaUtil qw/ character_input phylip_pars parse_newick /;
+use XML::LibXML;
use Moose;
+=head1 NAME
+
+Text::Tradition::Stemma - a representation of a I<stemma codicum> for a Text::Tradition
+
+=head1 SYNOPSIS
+
+ use Text::Tradition;
+ my $t = Text::Tradition->new(
+ 'name' => 'this is a text',
+ 'input' => 'TEI',
+ 'file' => '/path/to/tei_parallel_seg_file.xml' );
+
+ my $s = $tradition->add_stemma( dotfile => '/path/to/stemma.dot' );
+
+=head1 DESCRIPTION
+
+Text::Tradition is a library for representation and analysis of collated
+texts, particularly medieval ones. The Stemma is a representation of the
+copying relationships between the witnesses in a Tradition, modelled with
+a connected rooted directed acyclic graph (CRDAG).
+
+=head1 DOT SYNTAX
+
+The easiest way to define a stemma is to use a special form of the 'dot'
+syntax of GraphViz.
+
+Each stemma opens with the line
+
+ digraph Stemma {
+
+and continues with a list of all manuscript witnesses in the stemma, whether
+extant witnesses or missing archetypes or hyparchetypes. Each of these is
+listed by its sigil on its own line, e.g.:
+
+ alpha [ class=hypothetical ]
+ 1 [ class=hypothetical,label=* ]
+ Ms4 [ class=extant ]
+
+Extant witnesses are listed with class=extant; missing or postulated witnesses
+are listed with class=hypothetical. Anonymous hyparchetypes must be given a
+unique name or number, but can be represented as anonymous with the addition
+of 'label=*' to their lines. Greek letters or other special characters may be
+used as names, but they must always be wrapped in double quotes.
+
+Links between manuscripts are then listed with arrow notation, as below. These
+lines show the direction of copying, one step at a time, for the entire stemma.
+
+ alpha -> 1
+ 1 -> Ms4
+
+The final line in the definition should be the closing brace:
+
+ }
+
+Thus for a set of extant manuscripts A, B, and C, where A and B were copied
+from the archetype O and C was copied from B, the definition would be:
+
+ digraph Stemma {
+ O [ class=hypothetical]
+ A [ class=extant ]
+ B [ class=extant ]
+ C [ class=extant ]
+ O -> A
+ O -> B
+ B -> C
+ }
+
+=head1 CONSTRUCTOR
+
+=head2 new
+
+The constructor. This should generally be called from Text::Tradition, but
+if called directly it takes the following options:
+
+=over
+
+=item * collation - The collation with which the stemma is associated.
+
+=item * dot - A filehandle open to a DOT representation of the stemma graph.
+
+=back
+
+=cut
+
has collation => (
is => 'ro',
isa => 'Text::Tradition::Collation',
isa => 'Graph',
predicate => 'has_graph',
);
-
-has distance_trees => (
- is => 'ro',
- isa => 'ArrayRef[Graph]',
- writer => '_save_distance_trees',
- 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.
if( exists $args->{'dot'} ) {
- $self->graph_from_dot( $args->{'dot'} );
+ $self->_graph_from_dot( $args->{'dot'} );
}
}
-sub graph_from_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 ) {
unless $self->graph->has_vertex_attribute( $v, 'class' );
}
} else {
- warn "Failed to parse dot in $dotfh";
+ throw( "Failed to parse dot in $dotfh" );
}
}
+=head1 METHODS
+
+=head2 as_dot( \%options )
+
+Returns a normal dot representation of the stemma layout, suitable for rendering
+with GraphViz. Options include:
+
+=over
+
+=item * graph - A hashref of global graph options.
+
+=item * node - A hashref of global node options.
+
+=item * edge - A hashref of global edge options.
+
+=back
+
+See the GraphViz documentation for the list of available options.
+
+=cut
+
sub as_dot {
my( $self, $opts ) = @_;
+ ## See if we are including any a.c. witnesses in this graph.
+ my $graph = $self->graph;
+ if( exists $opts->{'layerwits'} ) {
+ $graph = $self->extend_graph( $opts->{'layerwits'} );
+ }
+
# Get default and specified options
- my %graphopts = ();
+ my %graphopts = (
+ # 'ratio' => 1,
+ );
my %nodeopts = (
'fontsize' => 11,
- 'hshape' => 'plaintext', # Shape for the hypothetical nodes
- 'htext' => '*',
'style' => 'filled',
'fillcolor' => 'white',
+ 'color' => 'white',
'shape' => 'ellipse', # Shape for the extant nodes
);
my %edgeopts = (
- 'arrowhead' => 'open',
+ 'arrowhead' => 'none',
);
@graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}}
if $opts->{'graph'};
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 ) );
+ foreach my $n ( $graph->vertices ) {
+ if( $graph->has_vertex_attribute( $n, 'label' ) ) {
+ my $ltext = $graph->get_vertex_attribute( $n, 'label' );
+ push( @dotlines, _make_dotline( $n, 'label' => $ltext ) );
} else {
# Use the default display settings.
+ $n = _dotquote( $n );
push( @dotlines, " $n;" );
}
}
# Add each of our edges.
- foreach my $e ( $self->graph->edges ) {
- my( $from, $to ) = @$e;
+ foreach my $e ( $graph->edges ) {
+ my( $from, $to ) = map { _dotquote( $_ ) } @$e;
push( @dotlines, " $from -> $to;" );
}
push( @dotlines, '}' );
return join( "\n", @dotlines );
}
+=head2 editable( $opts )
+
+Returns a version of the graph rendered in our definition format. The
+output separates statements with a newline; set $opts->{'linesep'} to the
+empty string or to a space if the result is to be sent via JSON.
+
+Any layer witnesses to be included should be passed via $opts->{'layerwits'}.
+
+=cut
-# Another version of dot output meant for graph editing, thus
-# much simpler.
sub editable {
- my $self = shift;
+ my( $self, $opts ) = @_;
+
+ ## See if we are including any a.c. witnesses in this graph.
+ my $graph = $self->graph;
+ if( exists $opts->{'layerwits'} ) {
+ $graph = $self->extend_graph( $opts->{'layerwits'} );
+ }
+
+ # Create the graph
+ my $join = ( $opts && exists $opts->{'linesep'} ) ? $opts->{'linesep'} : "\n";
my @dotlines;
push( @dotlines, 'digraph stemma {' );
my @real; # A cheap sort
push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) );
}
foreach my $e ( sort _by_vertex $self->graph->edges ) {
- my( $from, $to ) = @$e;
+ my( $from, $to ) = map { _dotquote( $_ ) } @$e;
push( @dotlines, " $from -> $to;" );
}
push( @dotlines, '}' );
- return join( "\n", @dotlines );
+ return join( $join, @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\"" );
+ my $v = _dotquote( $attr{$k} );
+ push( @pairs, "$k=$v" );
}
- return sprintf( " %s [ %s ];", $obj, join( ', ', @pairs ) );
+ return sprintf( " %s [ %s ];", _dotquote( $obj ), join( ', ', @pairs ) );
}
+sub _dotquote {
+ my( $str ) = @_;
+ return $str if $str =~ /^[A-Za-z0-9]+$/;
+ $str =~ s/\"/\\\"/g;
+ $str = '"' . $str . '"';
+ return $str;
+}
+
sub _by_vertex {
return $a->[0].$a->[1] cmp $b->[0].$b->[1];
}
-# Render the stemma as SVG.
+=head2 extend_graph( $layered_witnesses )
+
+Returns a graph which is the original stemma with witness layers added for the
+list in @$layered_witnesses. A layered (a.c.) witness is added as a parent
+of its main version, and additionally shares all other parents and children with
+that version.
+
+=cut
+
+sub extend_graph {
+ my( $self, $layerwits ) = @_;
+ # For each 'layered' witness in the layerwits array, add it to the graph
+ # as an ancestor of the 'main' witness, and otherwise with the same parent/
+ # child links as its main analogue.
+ # TOOD Handle case where B is copied from A but corrected from C
+
+ # Iterate through, adding a.c. witnesses
+ my $actag = $self->collation->ac_label;
+ my $graph = $self->graph->deep_copy;
+ foreach my $lw ( @$layerwits ) {
+ # Add the layered witness and set it with the same attributes as
+ # its 'main' analogue
+ my $lwac = $lw . $self->collation->ac_label;
+ $graph->add_vertex( $lwac );
+ $graph->set_vertex_attributes( $lwac,
+ $graph->get_vertex_attributes( $lw ) );
+
+ # Set it as ancestor to the main witness
+ $graph->add_edge( $lwac, $lw );
+
+ # Give it the same ancestors and descendants as the main witness has,
+ # bearing in mind that those ancestors and descendants might also just
+ # have had a layered witness defined.
+ foreach my $v ( $graph->predecessors( $lw ) ) {
+ next if $v eq $lwac; # Don't add a loop
+ $graph->add_edge( $v, $lwac );
+ $graph->add_edge( $v.$self->collation->ac_label, $lwac )
+ if $graph->has_vertex( $v.$self->collation->ac_label );
+ }
+ foreach my $v ( $graph->successors( $lw ) ) {
+ next if $v eq $lwac; # but this shouldn't occur
+ $graph->add_edge( $lwac, $v );
+ $graph->add_edge( $lwac, $v.$self->collation->ac_label )
+ if $graph->has_vertex( $v.$self->collation->ac_label );
+ }
+ }
+ return $graph;
+}
+
+=head2 as_svg
+
+Returns an SVG representation of the graph, calling as_dot first.
+
+=cut
+
sub as_svg {
my( $self, $opts ) = @_;
my $dot = $self->as_dot( $opts );
my @cmd = qw/dot -Tsvg/;
- my( $svg, $err );
+ my $svg;
my $dotfile = File::Temp->new();
## TODO REMOVE
# $dotfile->unlink_on_destroy(0);
print $dotfile $dot;
push( @cmd, $dotfile->filename );
run( \@cmd, ">", binary(), \$svg );
- $svg = decode_utf8( $svg );
- return $svg;
+ # HACK: Parse the SVG and change the dimensions.
+ # Get rid of width and height attributes to allow scaling.
+ my $parser = XML::LibXML->new();
+ my $svgdoc = $parser->parse_string( decode_utf8( $svg ) );
+ $svgdoc->documentElement->removeAttribute('width');
+ $svgdoc->documentElement->removeAttribute('height');
+ # Return the result
+ return decode_utf8( $svgdoc->toString );
}
+=head2 witnesses
+
+Returns a list of the extant witnesses represented in the stemma.
+
+=cut
+
sub witnesses {
my $self = shift;
my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
return @wits;
}
-#### Methods for calculating phylogenetic trees ####
+=head2 hypotheticals
-before 'distance_trees' => sub {
- my $self = shift;
- my %args = (
- 'program' => 'phylip_pars',
- @_ );
- # TODO allow specification of method for calculating distance tree
- if( !$self->has_distance_trees
- || $args{'program'} ne $self->distance_program ) {
- # We need to make a tree before we can return it.
- 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";
- }
- }
-};
+Returns a list of the hypothetical witnesses represented in the stemma.
-sub run_phylip_pars {
- my $self = shift;
-
- # Set up a temporary directory for all the default Phylip files.
- my $phylip_dir = File::Temp->newdir();
- # $phylip_dir->unlink_on_destroy(0);
- # We need an infile, and we need a command input file.
- open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
- print MATRIX phylip_pars_input( $self->collation->make_alignment_table() );
- close MATRIX;
-
- open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
- ## TODO any configuration parameters we want to set here
-# U Search for best tree? Yes
-# S Search option? More thorough search
-# V Number of trees to save? 100
-# J Randomize input order of species? No. Use input order
-# O Outgroup root? No, use as outgroup species 1
-# T Use Threshold parsimony? No, use ordinary parsimony
-# W Sites weighted? No
-# M Analyze multiple data sets? No
-# I Input species interleaved? Yes
-# 0 Terminal type (IBM PC, ANSI, none)? ANSI
-# 1 Print out the data at start of run No
-# 2 Print indications of progress of run Yes
-# 3 Print out tree Yes
-# 4 Print out steps in each site No
-# 5 Print character at all nodes of tree No
-# 6 Write out trees onto tree file? Yes
- print CMD "Y\n";
- close CMD;
-
- # And then we run the program.
- my $program = File::Which::which( 'pars' );
- unless( -x $program ) {
- return( undef, "Phylip pars not found in path" );
- }
-
- {
- # We need to run it in our temporary directory where we have created
- # all the expected files.
- local $CWD = $phylip_dir;
- my @cmd = ( $program );
- run \@cmd, '<', 'cmdfile', '>', '/dev/null';
- }
- # Now our output should be in 'outfile' and our tree in 'outtree',
- # both in the temp directory.
-
- my @outtree;
- if( -f "$phylip_dir/outtree" ) {
- open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read";
- @outtree = <TREE>;
- close TREE;
- }
- return( 1, join( '', @outtree ) ) if @outtree;
-
- my @error;
- if( -f "$phylip_dir/outfile" ) {
- open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read";
- @error = <OUTPUT>;
- close OUTPUT;
- } else {
- push( @error, "Neither outtree nor output file was produced!" );
- }
- return( undef, join( '', @error ) );
-}
+=cut
-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 hypotheticals {
+ my $self = shift;
+ my @wits = grep
+ { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'hypothetical' }
+ $self->graph->vertices;
+ return @wits;
}
-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 throw {
+ Text::Tradition::Error->throw(
+ 'ident' => 'Stemma error',
+ 'message' => $_[0],
+ );
}
-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;
1;
+
+=head1 LICENSE
+
+This package is free software and is provided "as is" without express
+or implied warranty. You can redistribute it and/or modify it under
+the same terms as Perl itself.
+
+=head1 AUTHOR
+
+Tara L Andrews E<lt>aurum@cpan.orgE<gt>