X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FStemma.pm;h=ec1714520261157338e0f6a98e0a282b3d6cb02e;hb=335a62ef9e58093170ab3e2da0945fc545e9f982;hp=b875e4e958a52d2c55c7326d77831ad1f798616d;hpb=027d819cfec7c990f32bf810203481c9f7dc1f60;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Stemma.pm b/lib/Text/Tradition/Stemma.pm index b875e4e..ec17145 100644 --- a/lib/Text/Tradition/Stemma.pm +++ b/lib/Text/Tradition/Stemma.pm @@ -8,6 +8,7 @@ use Graph::Reader::Dot; use IPC::Run qw/ run binary /; use Text::Tradition::Error; use Text::Tradition::StemmaUtil qw/ character_input phylip_pars parse_newick /; +use XML::LibXML; use Moose; =head1 NAME @@ -27,15 +28,14 @@ Text::Tradition::Stemma - a representation of a I for a Text::Tr =head1 DESCRIPTION Text::Tradition is a library for representation and analysis of collated -texts, particularly medieval ones. The Collation is the central feature of -a Tradition, where the text, its sequence of readings, and its relationships -between readings are actually kept. +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 (which is a directed acyclic graph, denoting -the scholar's hypothesis concerning which text(s) were copied from which other(s)) -is to use a special form of the 'dot' syntax of GraphViz. +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 @@ -107,20 +107,7 @@ has graph => ( 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. @@ -169,6 +156,12 @@ See the GraphViz documentation for the list of available options. 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 = ( # 'ratio' => 1, @@ -189,7 +182,7 @@ sub as_dot { if $opts->{'node'}; @edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}} if $opts->{'edge'}; - + my @dotlines; push( @dotlines, 'digraph stemma {' ); ## Print out the global attributes @@ -198,18 +191,19 @@ sub as_dot { push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts; # Add each of the nodes. - foreach my $n ( $self->graph->vertices ) { - if( $self->graph->has_vertex_attribute( $n, 'label' ) ) { - my $ltext = $self->graph->get_vertex_attribute( $n, 'label' ); + 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, '}' ); @@ -217,14 +211,27 @@ sub as_dot { return join( "\n", @dotlines ); } -=head2 editable +=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. -Returns a version of the graph rendered in our definition format. +Any layer witnesses to be included should be passed via $opts->{'layerwits'}. =cut 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 @@ -242,28 +249,84 @@ sub editable { 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]; } +=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->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. @@ -274,7 +337,7 @@ 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); @@ -282,8 +345,14 @@ sub as_svg { 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 @@ -299,45 +368,12 @@ sub witnesses { return @wits; } -=head2 distance_trees( program => $program ) - -Returns a set of undirected graphs, which are the result of running a distance -tree calculation program on the collation. Currently the only supported -program is phylip_pars. - -=cut - -#### Methods for calculating phylogenetic trees #### - -before 'distance_trees' => sub { +sub hypotheticals { 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 $result = $self->$dsub(); # this might throw an error - catch it? - # Save the resulting trees - my $trees = parse_newick( $result ); - $self->_save_distance_trees( $trees ); - $self->distance_program( $args{'program'} ); - } -}; - -=head2 run_phylip_pars - -Runs Phylip Pars on the collation, returning the results in Newick format. -Used for the distance_trees calculation. - -=cut - -sub run_phylip_pars { - my $self = shift; - my $cdata = character_input( $self->collation->make_alignment_table() ); - return phylip_pars( $cdata ); + my @wits = grep + { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'hypothetical' } + $self->graph->vertices; + return @wits; } sub throw {