X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FStemma.pm;h=c7114303e0ca1a7551194a367f19035f002e67aa;hb=933df8d301764db30b5362b5ee296aa6ef11ccf5;hp=fbf9ff7a448440c3e29fe879cea0f0598b68e119;hpb=63778331994b16f1890a1d6aa1a1e0dbfad73a21;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Stemma.pm b/lib/Text/Tradition/Stemma.pm index fbf9ff7..c711430 100644 --- a/lib/Text/Tradition/Stemma.pm +++ b/lib/Text/Tradition/Stemma.pm @@ -8,8 +8,93 @@ 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 + +Text::Tradition::Stemma - a representation of a I 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', @@ -22,32 +107,17 @@ 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. 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 ) { @@ -62,21 +132,49 @@ sub graph_from_dot { } } +=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'}; @@ -84,30 +182,28 @@ 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 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, '}' ); @@ -115,11 +211,27 @@ sub as_dot { 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 @@ -137,34 +249,95 @@ 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]; } -# 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->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); @@ -172,10 +345,22 @@ 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 + +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' } @@ -183,34 +368,18 @@ sub witnesses { 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 { - throw( "Failed to calculate distance trees: $result" ); - } - } -}; +Returns a list of the hypothetical witnesses represented in the stemma. + +=cut -sub run_phylip_pars { - my $self = shift; - my $cdata = character_input( $self->collation->make_alignment_table() ); - return phylip_pars( $cdata ); +sub hypotheticals { + my $self = shift; + my @wits = grep + { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'hypothetical' } + $self->graph->vertices; + return @wits; } sub throw { @@ -225,3 +394,13 @@ 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 Eaurum@cpan.orgE