X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FParser%2FSelf.pm;h=895ca4d2d8a3ffe018f1648fcd3e71bcc9d4fa86;hb=e3196b2a34d403f922af145e2bb1b21356d5c94b;hp=5311660a7369b02ae864e09c0ccc2204cd3b66b0;hpb=d9e873d0f90f1b7aec072c16a0eed37878f7f47f;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Parser/Self.pm b/lib/Text/Tradition/Parser/Self.pm index 5311660..895ca4d 100644 --- a/lib/Text/Tradition/Parser/Self.pm +++ b/lib/Text/Tradition/Parser/Self.pm @@ -2,129 +2,240 @@ package Text::Tradition::Parser::Self; use strict; use warnings; -use Text::Tradition::Parser::GraphML; +use Text::Tradition::Parser::GraphML qw/ graphml_parse populate_witness_path /; =head1 NAME Text::Tradition::Parser::GraphML +=head1 SYNOPSIS + + use Text::Tradition; + + my $t_from_file = Text::Tradition->new( + 'name' => 'my text', + 'input' => 'Self', + 'file' => '/path/to/tradition.xml' + ); + + my $t_from_string = Text::Tradition->new( + 'name' => 'my text', + 'input' => 'Self', + 'string' => $tradition_xml, + ); + =head1 DESCRIPTION Parser module for Text::Tradition to read in its own GraphML output format. -TODO document what this format is. +GraphML is a relatively simple graph description language; a 'graph' element +can have 'node' and 'edge' elements, and each of these can have simple 'data' +elements for attributes to be saved. -=head1 METHODS +The graph itself has attributes as in the Collation object: + +=over + +=item * linear + +=item * ac_label + +=item * baselabel + +=item * wit_list_separator + +=back + +The node objects have the following attributes: + +=over + +=item * name + +=item * reading + +=item * identical + +=item * rank + +=item * class + +=back + +The edge objects have the following attributes: =over -=item B +=item * class + +=item * witness (for 'path' class edges) + +=item * extra (for 'path' class edges) + +=item * relationship (for 'relationship' class edges) + +=item * equal_rank (for 'relationship' class edges) + +=item * non_correctable (for 'relationship' class edges) + +=item * non_independent (for 'relationship' class edges) + +=back + +=head1 METHODS + +=head2 B + +parse( $graph, $opts ); + +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. + +=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 $tradition = 't/data/florilegium_graphml.xml'; +my $t = Text::Tradition->new( + 'name' => 'inline', + 'input' => 'Self', + 'file' => $tradition, + ); + +is( ref( $t ), 'Text::Tradition', "Parsed our own GraphML" ); +if( $t ) { + is( scalar $t->collation->readings, 319, "Collation has all readings" ); + is( scalar $t->collation->paths, 2854, "Collation has all paths" ); + is( scalar $t->witnesses, 13, "Collation has all witnesses" ); +} + +=end testing =cut -my( $IDKEY, $TOKENKEY, $TRANSPOS_KEY, $RANK_KEY, $CLASS_KEY ) - = qw/ name reading identical rank class /; +my( $IDKEY, $TOKENKEY, $TRANSPOS_KEY, $RANK_KEY, $CLASS_KEY, + $SOURCE_KEY, $TARGET_KEY, $WITNESS_KEY, $EXTRA_KEY, $RELATIONSHIP_KEY ) + = qw/ name reading identical rank class + source target witness extra relationship/; 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; - + + # Set up the graph-global attributes. They will appear in the + # hash under their accessor names. + print STDERR "Setting graph globals\n"; + $tradition->name( $graph_data->{'name'} ); + foreach my $gkey ( keys %{$graph_data->{'attr'}} ) { + my $val = $graph_data->{'attr'}->{$gkey}; + $collation->$gkey( $val ); + } + # Add the nodes to the graph. - # TODO Are we adding extra start/end nodes? my $extra_data = {}; # Keep track of data that needs to be processed # after the nodes & edges are created. print STDERR "Adding graph nodes\n"; foreach my $n ( @{$graph_data->{'nodes'}} ) { - # Each node is either a segment or a reading, depending on - # its class. Readings have text, segments don't. - my %node_data = %$n; + # First extract the data that we can use without reference to + # anything else. + my %node_data = %$n; # Need $n itself untouched for edge processing my $nodeid = delete $node_data{$IDKEY}; my $reading = delete $node_data{$TOKENKEY}; - my $class = $node_data{$CLASS_KEY} || ''; - # TODO this is a hack, fix it? - $class = 'reading' unless $class eq 'segment'; - my $method = $class eq 'segment' ? "add_$class" : "add_reading"; - my $gnode = $collation->$method( $nodeid ); - $gnode->label( $reading ); - $gnode->set_common if $class eq 'common'; - - # Now save the rest of the data, i.e. not the ID or label, + my $class = delete $node_data{$CLASS_KEY} || ''; + my $rank = delete $node_data{$RANK_KEY}; + + # Create the node. Current valid classes are common and meta. + # Everything else is a normal reading. + my $gnode = $collation->add_reading( $nodeid ); + $gnode->text( $reading ); + $gnode->make_common if $class eq 'common'; + $gnode->is_meta( 1 ) if $class eq 'meta'; + # This is a horrible hack. + $gnode->is_lacuna( $reading =~ /^\#LACUNA/ ); + $gnode->rank( $rank ) if defined $rank; + + # Now save the data that we need for post-processing, # if it exists. if ( keys %node_data ) { - $extra_data->{$nodeid} = \%node_data; + $extra_data->{$nodeid} = \%node_data } } # Now add the edges. print STDERR "Adding graph edges\n"; + my $has_ante_corr = {}; foreach my $e ( @{$graph_data->{'edges'}} ) { - my %edge_data = %$e; - my $from = delete $edge_data{'source'}; - my $to = delete $edge_data{'target'}; - my $class = delete $edge_data{'class'}; - - # Whatever is left tells us what kind of edge it is. - foreach my $wkey ( keys %edge_data ) { - if( $wkey =~ /^witness/ ) { - unless( $class eq 'path' ) { - warn "Cannot add witness label to a $class edge"; - next; - } - my $wit = $edge_data{$wkey}; - unless( $witnesses{$wit} ) { - $tradition->add_witness( sigil => $wit ); - $witnesses{$wit} = 1; - } - my $label = $wkey eq 'witness_ante_corr' - ? $wit . $collation->ac_label : $wit; - $collation->add_path( $from->{$IDKEY}, $to->{$IDKEY}, $label ); - } elsif( $wkey eq 'relationship' ) { - unless( $class eq 'relationship' ) { - warn "Cannot add relationship label to a $class edge"; - next; - } - my $rel = $edge_data{$wkey}; - # TODO handle global relationships - $collation->add_relationship( $rel, $from->{$IDKEY}, $to->{$IDKEY} ); - } else { - my $seg_edge = $collation->graph->add_edge( $from->{$IDKEY}, $to->{$IDKEY} ); - $seg_edge->set_attribute( 'class', 'segment' ); - } - } + my $from = $e->{$SOURCE_KEY}; + my $to = $e->{$TARGET_KEY}; + my $class = $e->{$CLASS_KEY}; + + # We may have more information depending on the class. + if( $class eq 'path' ) { + # We need the witness, and whether it is an 'extra' reading path. + my $wit = $e->{$WITNESS_KEY}; + warn "No witness label on path edge!" unless $wit; + my $extra = $e->{$EXTRA_KEY}; + my $label = $wit . ( $extra ? $collation->ac_label : '' ); + $collation->add_path( $from->{$IDKEY}, $to->{$IDKEY}, $label ); + # Add the witness if we don't have it already. + unless( $witnesses{$wit} ) { + $tradition->add_witness( sigil => $wit ); + $witnesses{$wit} = 1; + } + $has_ante_corr->{$wit} = 1 if $extra; + } elsif( $class eq 'relationship' ) { + # We need the metadata about the relationship. + my $opts = { 'type' => $e->{$RELATIONSHIP_KEY} }; + $opts->{'equal_rank'} = $e->{'equal_rank'} + if exists $e->{'equal_rank'}; + $opts->{'non_correctable'} = $e->{'non_correctable'} + if exists $e->{'non_correctable'}; + $opts->{'non_independent'} = $e->{'non_independent'} + if exists $e->{'non_independent'}; + warn "No relationship type for relationship edge!" unless $opts->{'type'}; + $collation->add_relationship( $from->{$IDKEY}, $to->{$IDKEY}, $opts ); + } } ## Deal with node information (transposition, relationships, etc.) that ## needs to be processed after all the nodes are created. - print STDERR "Adding second-pass data\n"; - my $linear = undef; + print STDERR "Adding second-pass node data\n"; foreach my $nkey ( keys %$extra_data ) { foreach my $edkey ( keys %{$extra_data->{$nkey}} ) { my $this_reading = $collation->reading( $nkey ); if( $edkey eq $TRANSPOS_KEY ) { my $other_reading = $collation->reading( $extra_data->{$nkey}->{$edkey} ); - # We evidently have a linear graph. - $linear = 1; $this_reading->set_identical( $other_reading ); - } elsif ( $edkey eq $RANK_KEY ) { - $this_reading->rank( $extra_data->{$nkey}->{$edkey} ); } else { warn "Unfamiliar reading node data $edkey for $nkey"; } } } - $collation->linear( $linear ); - # TODO We probably need to set the $witness->path arrays for each wit. + + # Set the $witness->path arrays for each wit. + populate_witness_path( $tradition, $has_ante_corr ); } +1; + +=head1 BUGS / TODO + +=over + +=item * Make this into a stream parser with GraphML + +=item * Simply field -> attribute correspondence for nodes and edges + +=item * Share key name constants with Collation.pm + =back =head1 LICENSE @@ -135,8 +246,4 @@ the same terms as Perl itself. =head1 AUTHOR -Tara L Andrews, aurum@cpan.org - -=cut - -1; +Tara L Andrews Eaurum@cpan.orgE