X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FParser%2FCollateX.pm;h=68948d58698c7ad5f98f37861bb63e1c58210d13;hb=679f17e1a60a81370df8cbb49b94a2b5d19e3a98;hp=5526d3a1407d72c85e3198e18fee96d3f9ca5cbf;hpb=7e450e44cb188a835f5ca2b9f007b9782fcc753c;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Parser/CollateX.pm b/lib/Text/Tradition/Parser/CollateX.pm index 5526d3a..68948d5 100644 --- a/lib/Text/Tradition/Parser/CollateX.pm +++ b/lib/Text/Tradition/Parser/CollateX.pm @@ -2,12 +2,28 @@ package Text::Tradition::Parser::CollateX; use strict; use warnings; -use Text::Tradition::Parser::GraphML; +use Text::Tradition::Parser::GraphML qw/ graphml_parse /; =head1 NAME Text::Tradition::Parser::CollateX +=head1 SYNOPSIS + + use Text::Tradition; + + my $t_from_file = Text::Tradition->new( + 'name' => 'my text', + 'input' => 'CollateX', + 'file' => '/path/to/collation.xml' + ); + + my $t_from_string = Text::Tradition->new( + 'name' => 'my text', + 'input' => 'CollateX', + 'string' => $collation_xml, + ); + =head1 DESCRIPTION Parser module for Text::Tradition, given a GraphML file from the @@ -17,115 +33,125 @@ http://gregor.middell.net/collatex/ =head1 METHODS -=over +=head2 B + +parse( $tradition, $init_options ); + +Takes an initialized Text::Tradition object and a set of options; creates +the appropriate nodes and edges on the graph. The options hash should +include either a 'file' argument or a 'string' argument, depending on the +source of the XML to be parsed. -=item B +=begin testing -parse( $graph, $graphml_string ); +use Text::Tradition; +binmode STDOUT, ":utf8"; +binmode STDERR, ":utf8"; +eval { no warnings; binmode $DB::OUT, ":utf8"; }; -Takes an initialized Text::Tradition::Graph object and a string -containing the GraphML; creates the appropriate nodes and edges on the -graph. +my $cxfile = 't/data/Collatex-16.xml'; +my $t = Text::Tradition->new( + 'name' => 'inline', + 'input' => 'CollateX', + 'file' => $cxfile, + ); + +is( ref( $t ), 'Text::Tradition', "Parsed a CollateX input" ); +if( $t ) { + is( scalar $t->collation->readings, 26, "Collation has all readings" ); + is( scalar $t->collation->paths, 32, "Collation has all paths" ); + is( scalar $t->witnesses, 3, "Collation has all witnesses" ); + + # Check an 'identical' node + my $transposed = $t->collation->reading( 'n15' ); + my @related = $transposed->related_readings; + is( scalar @related, 1, "Reading links to transposed version" ); + is( $related[0]->id, 'n18', "Correct transposition link" ); +} + +=end testing =cut my $IDKEY = 'number'; -my $CONTENTKEY = 'token'; -my $TRANSKEY = 'identical'; +my $CONTENTKEY = 'tokens'; +my $EDGETYPEKEY = 'type'; +my $WITKEY = 'witnesses'; sub parse { - my( $tradition, $graphml_str ) = @_; - my $graph_data = Text::Tradition::Parser::GraphML::parse( $graphml_str ); + my( $tradition, $opts ) = @_; + my( $graph_data ) = graphml_parse( $opts ); my $collation = $tradition->collation; - my %witnesses; # Keep track of the witnesses we encounter as we - # run through the graph data. - - # Add the nodes to the graph. First delete the start node, because - # GraphML graphs will have their own start nodes. - $collation->del_reading( $collation->start() ); - $collation->del_reading( $collation->end() ); - my $extra_data = {}; # Keep track of info to be processed after all - # nodes have been created + # First add the readings to the graph. + ## Assume the start node has no text and id 0, and the end node has + ## no text and ID [number of nodes] - 1. + my $endnode = scalar @{$graph_data->{'nodes'}} - 1; foreach my $n ( @{$graph_data->{'nodes'}} ) { - my %node_data = %$n; - my $nodeid = delete $node_data{$IDKEY}; - my $token = delete $node_data{$CONTENTKEY}; - unless( defined $nodeid && defined $token ) { - warn "Did not find an ID or token for graph node, can't add it"; + unless( defined $n->{$IDKEY} && defined $n->{$CONTENTKEY} ) { + if( defined $n->{$IDKEY} && $n->{$IDKEY} == 0 ) { + # It's the start node. + $n->{$IDKEY} = $collation->start->id; + } elsif ( defined $n->{$IDKEY} && $n->{$IDKEY} == $endnode ) { + # It's the end node. + $n->{$IDKEY} = $collation->end->id; + } else { + # Something is probably wrong. + warn "Did not find an ID or token for graph node, can't add it"; + } next; } - my $gnode = $collation->add_reading( $nodeid ); - $gnode->text( $token ); - - # Whatever is left is extra info to be processed later. - if( keys %node_data ) { - $extra_data->{$nodeid} = \%node_data; - } + # Node ID should be an XML name, so prepend an 'n' if necessary. + if( $n->{$IDKEY} =~ /^\d/ ) { + $n->{$IDKEY} = 'n' . $n->{$IDKEY}; + } + # Create the reading. + my $gnode_args = { + 'id' => $n->{$IDKEY}, + 'text' => $n->{$CONTENTKEY}, + }; + my $gnode = $collation->add_reading( $gnode_args ); } - # Now add the edges. + # Now add the path edges. foreach my $e ( @{$graph_data->{'edges'}} ) { - my %edge_data = %$e; - my $from = delete $edge_data{'source'}; - my $to = delete $edge_data{'target'}; - - # In CollateX, we have a distinct witness data ID per witness, - # so that we can have multiple witnesses per edge. We want to - # translate this to one witness per edge in our own - # representation. - foreach my $ekey ( keys %edge_data ) { - my $wit = $edge_data{$ekey}; - # Create the witness object if it does not yet exist. - unless( $witnesses{$wit} ) { - $tradition->add_witness( 'sigil' => $wit ); - $witnesses{$wit} = 1; - } - $collation->add_path( $from->{$IDKEY}, $to->{$IDKEY}, $wit ); + my $from = $e->{'source'}; + my $to = $e->{'target'}; + + ## Edge data keys are ID (which we don't need), witnesses, and type. + ## Type can be 'path' or 'relationship'; + ## witnesses is a comma-separated list. + if( $e->{$EDGETYPEKEY} eq 'path' ) { + ## Add the path for each witness listesd. + # Create the witness objects if they does not yet exist. + foreach my $wit ( split( /, /, $e->{$WITKEY} ) ) { + unless( $tradition->witness( $wit ) ) { + $tradition->add_witness( 'sigil' => $wit ); + } + $collation->add_path( $from->{$IDKEY}, $to->{$IDKEY}, $wit ); + } + } else { # type 'relationship' + $collation->add_relationship( $from->{$IDKEY}, $to->{$IDKEY}, + { 'type' => 'transposition' } ); } } - # Process the extra node data if it exists. - foreach my $nodeid ( keys %$extra_data ) { - my $ed = $extra_data->{$nodeid}; - if( exists $ed->{$TRANSKEY} ) { - - my $tn_reading = $collation->reading( $nodeid ); - my $main_reading = $collation->reading( $ed->{$TRANSKEY} ); - if( $collation->linear ) { - $tn_reading->set_identical( $main_reading ); - } else { - $collation->merge_readings( $main_reading, $tn_reading ); - } - } # else we don't have any other tags to process yet. - } - - # Find the beginning and end nodes of the graph. The beginning node - # has no incoming edges; the end node has no outgoing edges. - my( $begin_node, $end_node ); - foreach my $gnode ( $collation->readings() ) { - # print STDERR "Checking node " . $gnode->name . "\n"; - my @outgoing = $gnode->outgoing(); - my @incoming = $gnode->incoming(); - - unless( scalar @incoming ) { - warn "Already have a beginning node" if $begin_node; - $begin_node = $gnode; - $collation->start( $gnode ); - } - unless( scalar @outgoing ) { - warn "Already have an ending node" if $end_node; - $end_node = $gnode; - $collation->end( $gnode ); - } - } + # Rank the readings. + $collation->calculate_common_readings(); # will implicitly rank - # Now we have added the witnesses and their paths, so we can - # calculate their explicit positions. - # TODO CollateX does this, and we should just have it exported there. - $collation->calculate_positions(); + # Save the text for each witness so that we can ensure consistency + # later on + $tradition->collation->text_from_paths(); } +=head1 BUGS / TODO + +=over + +=item * Make this into a stream parser with GraphML + +=item * Use CollateX-calculated ranks instead of recalculating our own + =back =head1 LICENSE