1 package Text::Tradition::Parser::GraphML;
6 use vars qw/ @EXPORT_OK $xpc /;
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 $graph_hash = { 'nodes' => [],
46 my $parser = XML::LibXML->new();
48 if( exists $opts->{'string'} ) {
49 $doc = $parser->parse_string( $opts->{'string'} );
50 } elsif ( exists $opts->{'file'} ) {
51 $doc = $parser->parse_file( $opts->{'file'} );
53 warn "Could not find string or file option to parse";
57 my( $graphattr, $nodedata, $witnesses ) = ( {}, {}, {} );
58 my $graphml = $doc->documentElement();
59 $xpc = XML::LibXML::XPathContext->new( $graphml );
60 $xpc->registerNs( 'g', 'http://graphml.graphdrawing.org/xmlns' );
62 # First get the ID keys, for witnesses and for collation data
63 foreach my $k ( $xpc->findnodes( '//g:key' ) ) {
64 # Each key has a 'for' attribute; the edge keys are witnesses, and
65 # the node keys contain an ID and string for each node.
66 my $keyid = $k->getAttribute( 'id' );
67 my $keyname = $k->getAttribute( 'attr.name' );
69 # Keep track of the XML identifiers for the data carried
70 # in each node element.
71 my $dtype = $k->getAttribute( 'for' );
72 if( $dtype eq 'graph' ) {
73 $graphattr->{$keyid} = $keyname;
74 } elsif( $dtype eq 'node' ) {
75 $nodedata->{$keyid} = $keyname;
77 $witnesses->{$keyid} = $keyname;
81 my $graph_el = $xpc->find( '/g:graphml/g:graph' )->[0];
82 $graph_hash->{'name'} = $graph_el->getAttribute( 'id' );
86 # Read in graph globals (if any).
87 print STDERR "Reading graphml global data\n";
88 foreach my $dkey ( keys %$graphattr ) {
89 my $keyname = $graphattr->{$dkey};
90 my $keyvalue = _lookup_node_data( $graph_el, $dkey );
91 $graph_hash->{'global'}->{$keyname} = $keyvalue;
94 # Add the nodes to the graph hash.
95 print STDERR "Reading graphml nodes\n";
96 my @nodes = $xpc->findnodes( '//g:node' );
97 foreach my $n ( @nodes ) {
98 # Could use a better way of registering these
100 foreach my $dkey ( keys %$nodedata ) {
101 my $keyname = $nodedata->{$dkey};
102 my $keyvalue = _lookup_node_data( $n, $dkey );
103 $node_hash->{$keyname} = $keyvalue if defined $keyvalue;
105 $node_reg->{$n->getAttribute( 'id' )} = $node_hash;
106 push( @{$graph_hash->{'nodes'}}, $node_hash );
109 # Now add the edges, and cross-ref with the node objects.
110 print STDERR "Reading graphml edges\n";
111 my @edges = $xpc->findnodes( '//g:edge' );
112 foreach my $e ( @edges ) {
113 my $from = $e->getAttribute('source');
114 my $to = $e->getAttribute('target');
116 # We don't know whether the edge data is one per witness
117 # or one per witness type, or something else. So we just
118 # save it and let our calling parser decide.
120 'source' => $node_reg->{$from},
121 'target' => $node_reg->{$to},
123 foreach my $wkey( keys %$witnesses ) {
124 my $wname = $witnesses->{$wkey};
125 my $wlabel = _lookup_node_data( $e, $wkey );
126 $edge_hash->{$wname} = $wlabel if $wlabel;
128 push( @{$graph_hash->{'edges'}}, $edge_hash );
134 sub _lookup_node_data {
135 my( $xmlnode, $key ) = @_;
136 my $lookup_xpath = './g:data[@key="%s"]/child::text()';
137 my $data = $xpc->find( sprintf( $lookup_xpath, $key ), $xmlnode );
138 # If we get back an empty nodelist, we return undef.
140 return undef unless $data->size;
141 return $data->to_literal->value;
143 # Otherwise we got back a value. Return it.
151 This package is free software and is provided "as is" without express
152 or implied warranty. You can redistribute it and/or modify it under
153 the same terms as Perl itself.
157 Tara L Andrews, aurum@cpan.org