Commit | Line | Data |
9463b0bf |
1 | package Text::Tradition::Stemma; |
2 | |
40f19742 |
3 | use Bio::Phylo::IO; |
e79c23c7 |
4 | use Encode qw( decode_utf8 ); |
9463b0bf |
5 | use File::Temp; |
e05997e2 |
6 | use Graph; |
7 | use Graph::Reader::Dot; |
e79c23c7 |
8 | use IPC::Run qw/ run binary /; |
63778331 |
9 | use Text::Tradition::Error; |
b02332ca |
10 | use Text::Tradition::StemmaUtil qw/ character_input phylip_pars parse_newick /; |
40f19742 |
11 | use Moose; |
9463b0bf |
12 | |
13 | has collation => ( |
14 | is => 'ro', |
15 | isa => 'Text::Tradition::Collation', |
16 | required => 1, |
8d9a1cd8 |
17 | weak_ref => 1, |
9463b0bf |
18 | ); |
19 | |
e05997e2 |
20 | has graph => ( |
21 | is => 'rw', |
22 | isa => 'Graph', |
23 | predicate => 'has_graph', |
24 | ); |
25 | |
40f19742 |
26 | has distance_trees => ( |
27 | is => 'ro', |
28 | isa => 'ArrayRef[Graph]', |
29 | writer => '_save_distance_trees', |
30 | predicate => 'has_distance_trees', |
31 | ); |
c0ccdb62 |
32 | |
0f5d05c6 |
33 | has distance_program => ( |
34 | is => 'rw', |
35 | isa => 'Str', |
36 | default => '', |
37 | ); |
38 | |
e05997e2 |
39 | sub BUILD { |
40 | my( $self, $args ) = @_; |
41 | # If we have been handed a dotfile, initialize it into a graph. |
42 | if( exists $args->{'dot'} ) { |
8d9a1cd8 |
43 | $self->graph_from_dot( $args->{'dot'} ); |
e05997e2 |
44 | } |
c0ccdb62 |
45 | } |
46 | |
8d9a1cd8 |
47 | sub graph_from_dot { |
48 | my( $self, $dotfh ) = @_; |
49 | # Assume utf-8 |
50 | binmode( $dotfh, ':utf8' ); |
51 | my $reader = Graph::Reader::Dot->new(); |
52 | my $graph = $reader->read_graph( $dotfh ); |
7a7c249c |
53 | if( $graph ) { |
54 | $self->graph( $graph ); |
55 | # Go through the nodes and set any non-hypothetical node to extant. |
56 | foreach my $v ( $self->graph->vertices ) { |
57 | $self->graph->set_vertex_attribute( $v, 'class', 'extant' ) |
58 | unless $self->graph->has_vertex_attribute( $v, 'class' ); |
59 | } |
60 | } else { |
63778331 |
61 | throw( "Failed to parse dot in $dotfh" ); |
7a7c249c |
62 | } |
8d9a1cd8 |
63 | } |
64 | |
65 | sub as_dot { |
e367f5c0 |
66 | my( $self, $opts ) = @_; |
7a7c249c |
67 | |
68 | # Get default and specified options |
69 | my %graphopts = (); |
70 | my %nodeopts = ( |
71 | 'fontsize' => 11, |
72 | 'hshape' => 'plaintext', # Shape for the hypothetical nodes |
73 | 'htext' => '*', |
74 | 'style' => 'filled', |
75 | 'fillcolor' => 'white', |
76 | 'shape' => 'ellipse', # Shape for the extant nodes |
77 | ); |
78 | my %edgeopts = ( |
79 | 'arrowhead' => 'open', |
80 | ); |
81 | @graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}} |
82 | if $opts->{'graph'}; |
83 | @nodeopts{ keys %{$opts->{'node'}} } = values %{$opts->{'node'}} |
84 | if $opts->{'node'}; |
85 | @edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}} |
86 | if $opts->{'edge'}; |
87 | |
88 | my @dotlines; |
89 | push( @dotlines, 'digraph stemma {' ); |
90 | ## Print out the global attributes |
91 | push( @dotlines, _make_dotline( 'graph', %graphopts ) ) if keys %graphopts; |
92 | push( @dotlines, _make_dotline( 'edge', %edgeopts ) ) if keys %edgeopts; |
93 | ## Delete our special attributes from the node set before continuing |
94 | my $hshape = delete $nodeopts{'hshape'}; |
95 | my $htext = delete $nodeopts{'htext'}; |
96 | push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts; |
97 | |
98 | # Add each of the nodes. |
99 | foreach my $n ( $self->graph->vertices ) { |
100 | if( $self->graph->get_vertex_attribute( $n, 'class' ) eq 'hypothetical' ) { |
101 | # Apply our display settings for hypothetical nodes. |
102 | push( @dotlines, _make_dotline( $n, 'shape' => $hshape, 'label' => $htext ) ); |
e79c23c7 |
103 | } else { |
7a7c249c |
104 | # Use the default display settings. |
105 | push( @dotlines, " $n;" ); |
e79c23c7 |
106 | } |
107 | } |
7a7c249c |
108 | # Add each of our edges. |
109 | foreach my $e ( $self->graph->edges ) { |
110 | my( $from, $to ) = @$e; |
111 | push( @dotlines, " $from -> $to;" ); |
112 | } |
113 | push( @dotlines, '}' ); |
e79c23c7 |
114 | |
7a7c249c |
115 | return join( "\n", @dotlines ); |
116 | } |
117 | |
118 | |
119 | # Another version of dot output meant for graph editing, thus |
120 | # much simpler. |
121 | sub editable { |
122 | my $self = shift; |
123 | my @dotlines; |
124 | push( @dotlines, 'digraph stemma {' ); |
125 | my @real; # A cheap sort |
126 | foreach my $n ( sort $self->graph->vertices ) { |
127 | my $c = $self->graph->get_vertex_attribute( $n, 'class' ); |
128 | $c = 'extant' unless $c; |
129 | if( $c eq 'extant' ) { |
130 | push( @real, $n ); |
131 | } else { |
132 | push( @dotlines, _make_dotline( $n, 'class' => $c ) ); |
133 | } |
e367f5c0 |
134 | } |
7a7c249c |
135 | # Now do the real ones |
136 | foreach my $n ( @real ) { |
137 | push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) ); |
138 | } |
139 | foreach my $e ( sort _by_vertex $self->graph->edges ) { |
140 | my( $from, $to ) = @$e; |
141 | push( @dotlines, " $from -> $to;" ); |
142 | } |
143 | push( @dotlines, '}' ); |
144 | return join( "\n", @dotlines ); |
145 | } |
146 | |
147 | sub _make_dotline { |
148 | my( $obj, %attr ) = @_; |
149 | my @pairs; |
150 | foreach my $k ( keys %attr ) { |
151 | my $v = $attr{$k}; |
152 | $v =~ s/\"/\\\"/g; |
153 | push( @pairs, "$k=\"$v\"" ); |
154 | } |
155 | return sprintf( " %s [ %s ];", $obj, join( ', ', @pairs ) ); |
8d9a1cd8 |
156 | } |
157 | |
7a7c249c |
158 | sub _by_vertex { |
159 | return $a->[0].$a->[1] cmp $b->[0].$b->[1]; |
160 | } |
8d9a1cd8 |
161 | |
162 | # Render the stemma as SVG. |
163 | sub as_svg { |
164 | my( $self, $opts ) = @_; |
165 | my $dot = $self->as_dot( $opts ); |
e79c23c7 |
166 | my @cmd = qw/dot -Tsvg/; |
167 | my( $svg, $err ); |
168 | my $dotfile = File::Temp->new(); |
169 | ## TODO REMOVE |
170 | # $dotfile->unlink_on_destroy(0); |
171 | binmode $dotfile, ':utf8'; |
8d9a1cd8 |
172 | print $dotfile $dot; |
e79c23c7 |
173 | push( @cmd, $dotfile->filename ); |
174 | run( \@cmd, ">", binary(), \$svg ); |
175 | $svg = decode_utf8( $svg ); |
176 | return $svg; |
177 | } |
178 | |
08e0fb85 |
179 | sub witnesses { |
180 | my $self = shift; |
181 | my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' } |
182 | $self->graph->vertices; |
183 | return @wits; |
184 | } |
185 | |
e79c23c7 |
186 | #### Methods for calculating phylogenetic trees #### |
187 | |
40f19742 |
188 | before 'distance_trees' => sub { |
189 | my $self = shift; |
0f5d05c6 |
190 | my %args = ( |
191 | 'program' => 'phylip_pars', |
192 | @_ ); |
40f19742 |
193 | # TODO allow specification of method for calculating distance tree |
0f5d05c6 |
194 | if( !$self->has_distance_trees |
195 | || $args{'program'} ne $self->distance_program ) { |
40f19742 |
196 | # We need to make a tree before we can return it. |
0f5d05c6 |
197 | my $dsub = 'run_' . $args{'program'}; |
198 | my( $ok, $result ) = $self->$dsub(); |
40f19742 |
199 | if( $ok ) { |
c0ccdb62 |
200 | # Save the resulting trees |
b02332ca |
201 | my $trees = parse_newick( $result ); |
c0ccdb62 |
202 | $self->_save_distance_trees( $trees ); |
0f5d05c6 |
203 | $self->distance_program( $args{'program'} ); |
40f19742 |
204 | } else { |
63778331 |
205 | throw( "Failed to calculate distance trees: $result" ); |
40f19742 |
206 | } |
207 | } |
208 | }; |
f6066bac |
209 | |
40f19742 |
210 | sub run_phylip_pars { |
b02332ca |
211 | my $self = shift; |
212 | my $cdata = character_input( $self->collation->make_alignment_table() ); |
213 | return phylip_pars( $cdata ); |
40f19742 |
214 | } |
215 | |
63778331 |
216 | sub throw { |
217 | Text::Tradition::Error->throw( |
218 | 'ident' => 'Stemma error', |
219 | 'message' => $_[0], |
220 | ); |
221 | } |
222 | |
223 | |
9463b0bf |
224 | no Moose; |
225 | __PACKAGE__->meta->make_immutable; |
226 | |
227 | 1; |