1 package Text::Tradition::Parser::GraphML;
6 use vars qw/ @EXPORT_OK $xpc /;
7 use Text::Tradition::Error;
9 use XML::LibXML::XPathContext;
11 @EXPORT_OK = qw/ graphml_parse /;
15 Text::Tradition::Parser::GraphML
19 Parser module for Text::Tradition, given a GraphML file that describes
20 a collation graph. Returns the information about the graph that has
21 been parsed out from the GraphML. This module is meant to be used
22 with a module (e.g. CollateX or Self) that interprets the specific
23 GraphML conventions of the source program.
27 =head2 B<graphml_parse>( $init_opts )
31 Takes a set of Tradition initialization options, among which should be either
32 'file' or 'string'; parses that file or string and returns a list of nodes, edges,
33 and their associated data.
37 # Return graph -> nodeid -> { key1/val1, key2/val2, key3/val3 ... }
38 # -> edgeid -> { source, target, wit1/val1, wit2/val2 ...}
43 my $parser = XML::LibXML->new();
45 if( exists $opts->{'string'} ) {
46 $doc = $parser->parse_string( $opts->{'string'} );
47 } elsif ( exists $opts->{'file'} ) {
48 $doc = $parser->parse_file( $opts->{'file'} );
49 } elsif ( exists $opts->{'xmlobj'} ) {
50 $doc = $opts->{'xmlobj'};
52 warn "Could not find string or file option to parse";
56 my( $graphattr, $nodedata, $edgedata ) = ( {}, {}, {} );
57 my $graphml = $doc->documentElement();
58 $xpc = XML::LibXML::XPathContext->new( $graphml );
59 $xpc->registerNs( 'g', 'http://graphml.graphdrawing.org/xmlns' );
61 # First get the ID keys, for node/edge data and for collation data
62 foreach my $k ( $xpc->findnodes( '//g:key' ) ) {
63 # Each key has a 'for' attribute to say whether it is for graph,
65 my $keyid = $k->getAttribute( 'id' );
66 my $keyname = $k->getAttribute( 'attr.name' );
68 # Keep track of the XML identifiers for the data carried
69 # in each node element.
70 my $dtype = $k->getAttribute( 'for' );
71 if( $dtype eq 'graph' ) {
72 $graphattr->{$keyid} = $keyname;
73 } elsif( $dtype eq 'node' ) {
74 $nodedata->{$keyid} = $keyname;
76 $edgedata->{$keyid} = $keyname;
80 my @graph_elements = $xpc->findnodes( '/g:graphml/g:graph' );
81 unless( @graph_elements ) {
82 throw( "No graph elements found in graph XML - is this really GraphML?" );
86 foreach my $graph_el ( @graph_elements ) {
87 my $graph_hash = { 'nodes' => [],
89 'name' => $graph_el->getAttribute( 'id' ) };
93 # Read in graph globals (if any).
94 # print STDERR "Reading graphml global data\n";
95 foreach my $dkey ( keys %$graphattr ) {
96 my $keyname = $graphattr->{$dkey};
97 my $keyvalue = _lookup_node_data( $graph_el, $dkey );
98 $graph_hash->{'global'}->{$keyname} = $keyvalue if defined $keyvalue;
101 # Add the nodes to the graph hash.
102 # print STDERR "Reading graphml nodes\n";
103 my @nodes = $xpc->findnodes( './/g:node', $graph_el );
104 foreach my $n ( @nodes ) {
105 # Could use a better way of registering these
107 foreach my $dkey ( keys %$nodedata ) {
108 my $keyname = $nodedata->{$dkey};
109 my $keyvalue = _lookup_node_data( $n, $dkey );
110 $node_hash->{$keyname} = $keyvalue if defined $keyvalue;
112 $node_reg->{$n->getAttribute( 'id' )} = $node_hash;
113 push( @{$graph_hash->{'nodes'}}, $node_hash );
116 # Now add the edges, and cross-ref with the node objects.
117 # print STDERR "Reading graphml edges\n";
118 my @edges = $xpc->findnodes( './/g:edge', $graph_el );
119 foreach my $e ( @edges ) {
120 my $from = $e->getAttribute('source');
121 my $to = $e->getAttribute('target');
123 # We don't know whether the edge data is one per witness
124 # or one per witness type, or something else. So we just
125 # save it and let our calling parser decide.
127 'source' => $node_reg->{$from},
128 'target' => $node_reg->{$to},
130 foreach my $wkey( keys %$edgedata ) {
131 my $wname = $edgedata->{$wkey};
132 my $wlabel = _lookup_node_data( $e, $wkey );
133 $edge_hash->{$wname} = $wlabel if $wlabel;
135 push( @{$graph_hash->{'edges'}}, $edge_hash );
137 push( @returned_graphs, $graph_hash );
139 return @returned_graphs;
143 sub _lookup_node_data {
144 my( $xmlnode, $key ) = @_;
145 my $lookup_xpath = './g:data[@key="%s"]/child::text()';
146 my $data = $xpc->find( sprintf( $lookup_xpath, $key ), $xmlnode );
147 # If we get back an empty nodelist, we return undef.
149 return undef unless $data->size;
150 return $data->to_literal->value;
152 # Otherwise we got back a value. Return it.
157 Text::Tradition::Error->throw(
158 'ident' => 'Parser::GraphML error',
165 This package is free software and is provided "as is" without express
166 or implied warranty. You can redistribute it and/or modify it under
167 the same terms as Perl itself.
171 Tara L Andrews, aurum@cpan.org