X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FParser%2FSelf.pm;h=5499fc711138bc923c8fe7ff66a2bcb4668e1fc5;hb=3bf5d6f1070741054a890165852039422030d963;hp=a1725d6f3601b0bd2946f454ce028caad9453231;hpb=49d4f2accb29737fdbaa42f75062d2055c4bc2ef;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Parser/Self.pm b/lib/Text/Tradition/Parser/Self.pm index a1725d6..5499fc7 100644 --- a/lib/Text/Tradition/Parser/Self.pm +++ b/lib/Text/Tradition/Parser/Self.pm @@ -106,147 +106,144 @@ my $t = Text::Tradition->new( 'file' => $tradition, ); -is( ref( $t ), 'Text::Tradition', "Parsed our own GraphML" ); +is( ref( $t ), 'Text::Tradition', "Parsed GraphML version 2" ); if( $t ) { is( scalar $t->collation->readings, 319, "Collation has all readings" ); is( scalar $t->collation->paths, 376, "Collation has all paths" ); is( scalar $t->witnesses, 13, "Collation has all witnesses" ); } +# TODO add a relationship, write graphml, reparse it, check that the rel +# is still there +$t->language('Greek'); +$t->collation->add_relationship( 'w12', 'w13', + { 'type' => 'grammatical', 'scope' => 'global', + 'annotation' => 'This is some note' } ); +ok( $t->collation->get_relationship( 'w12', 'w13' ), "Relationship set" ); +my $graphml_str = $t->collation->as_graphml; + +my $newt = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml_str ); +is( ref( $newt ), 'Text::Tradition', "Parsed current GraphML version" ); +if( $newt ) { + is( scalar $newt->collation->readings, 319, "Collation has all readings" ); + is( scalar $newt->collation->paths, 376, "Collation has all paths" ); + is( scalar $newt->witnesses, 13, "Collation has all witnesses" ); + is( scalar $newt->collation->relationships, 1, "Collation has added relationship" ); + is( $newt->language, 'Greek', "Tradition has correct language setting" ); + my $rel = $newt->collation->get_relationship( 'w12', 'w13' ); + ok( $rel, "Found set relationship" ); + is( $rel->annotation, 'This is some note', "Relationship has its properties" ); +} + + =end testing =cut -my( $IDKEY, $TOKENKEY, $TRANSPOS_KEY, $RANK_KEY, $CLASS_KEY, - $START_KEY, $END_KEY, $LACUNA_KEY, - $SOURCE_KEY, $TARGET_KEY, $WITNESS_KEY, $EXTRA_KEY, $RELATIONSHIP_KEY, - $COLO_KEY, $CORRECT_KEY, $INDEP_KEY ) - = qw/ name reading identical rank class - is_start is_end is_lacuna - source target witness extra relationship - equal_rank non_correctable non_independent /; - sub parse { my( $tradition, $opts ) = @_; - my $graph_data = graphml_parse( $opts ); + + # Collation data is in the first graph; relationship-specific stuff + # is in the second. + my( $graph_data, $rel_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. - my $use_version; - print STDERR "Setting graph globals\n"; + # print STDERR "Setting graph globals\n"; $tradition->name( $graph_data->{'name'} ); + my $use_version; + my $tmeta = $tradition->meta; + my $cmeta = $collation->meta; foreach my $gkey ( keys %{$graph_data->{'global'}} ) { my $val = $graph_data->{'global'}->{$gkey}; if( $gkey eq 'version' ) { $use_version = $val; + } elsif( $tmeta->has_attribute( $gkey ) ) { + $tradition->$gkey( $val ); } else { $collation->$gkey( $val ); } } - if( $use_version ) { - # Many of our tags changed. - $IDKEY = 'id'; - $TOKENKEY = 'text'; - $COLO_KEY = 'colocated'; - } # Add the nodes to the graph. - 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'}} ) { - # If it is the start or end node, we already have one, so skip it. - next if defined $n->{$START_KEY} || defined $n->{$END_KEY}; - - # First extract the data that we can use without reference to - # anything else. - my %node_data = %$n; # Need $n itself untouched for edge processing - - # Create the node. - my $reading_options = { - 'id' => delete $node_data{$IDKEY}, - 'is_lacuna' => delete $node_data{$LACUNA_KEY}, - }; - my $rank = delete $node_data{$RANK_KEY}; - $reading_options->{'rank'} = $rank if $rank; - my $text = delete $node_data{$TOKENKEY}; - $reading_options->{'text'} = $text if $text; - - # This is a horrible hack for backwards compatibility. - unless( $use_version ) { - $reading_options->{'is_lacuna'} = 1 - if $reading_options->{'text'} =~ /^\#LACUNA/; - } - - delete $node_data{$CLASS_KEY}; # Not actually used - my $gnode = $collation->add_reading( $reading_options ); - - # Now save the data that we need for post-processing, - # if it exists. TODO this is unneeded after conversion - if ( keys %node_data ) { - $extra_data->{$gnode->id} = \%node_data - } + # print STDERR "Adding collation readings\n"; + foreach my $n ( @{$graph_data->{'nodes'}} ) { + # If it is the start or end node, we already have one, so + # grab the rank and go. + next if( defined $n->{'is_start'} ); + if( defined $n->{'is_end'} ) { + $collation->end->rank( $n->{'rank'} ); + next; + } + my $gnode = $collation->add_reading( $n ); } # Now add the edges. - print STDERR "Adding graph edges\n"; + # print STDERR "Adding collation path edges\n"; foreach my $e ( @{$graph_data->{'edges'}} ) { - 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; - } - $tradition->witness( $wit )->is_layered( 1 ) if $extra; - } elsif( $class eq 'relationship' ) { - # We need the metadata about the relationship. - my $opts = { 'type' => $e->{$RELATIONSHIP_KEY} }; - $opts->{$COLO_KEY} = $e->{$COLO_KEY} - if exists $e->{$COLO_KEY}; - $opts->{$CORRECT_KEY} = $e->{$CORRECT_KEY} - if exists $e->{$CORRECT_KEY}; - $opts->{$INDEP_KEY} = $e->{$INDEP_KEY} - if exists $e->{$INDEP_KEY}; - warn "No relationship type for relationship edge!" unless $opts->{'type'}; - my( $ok, @result ) = $collation->add_relationship( $from->{$IDKEY}, $to->{$IDKEY}, $opts ); - unless( $ok ) { - warn "Did not add relationship: @result"; - } - } - } + my $from = $collation->reading( $e->{'source'}->{'id'} ); + my $to = $collation->reading( $e->{'target'}->{'id'} ); - ## Deal with node information (transposition, relationships, etc.) that - ## needs to be processed after all the nodes are created. - ## TODO unneeded after conversion - unless( $use_version ) { - 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} ); - $this_reading->set_identical( $other_reading ); - } else { - warn "Unfamiliar reading node data $edkey for $nkey"; - } - } + warn "No witness label on path edge!" unless $e->{'witness'}; + my $label = $e->{'witness'} . ( $e->{'extra'} ? $collation->ac_label : '' ); + $collation->add_path( $from, $to, $label ); + + # Add the witness if we don't have it already. + unless( $witnesses{$e->{'witness'}} ) { + $tradition->add_witness( sigil => $e->{'witness'} ); + $witnesses{$e->{'witness'}} = 1; } + $tradition->witness( $e->{'witness'} )->is_layered( 1 ) if $e->{'extra'}; } + + ## Done with the main graph, now look at the relationships. + # Nodes are added via the call to add_reading above. We only need + # add the relationships themselves. + # TODO check that scoping does trt + $rel_data->{'edges'} ||= []; # so that the next line doesn't break on no rels + foreach my $e ( sort { _layersort_rel( $a, $b ) } @{$rel_data->{'edges'}} ) { + my $from = $collation->reading( $e->{'source'}->{'id'} ); + my $to = $collation->reading( $e->{'target'}->{'id'} ); + delete $e->{'source'}; + delete $e->{'target'}; + # The remaining keys are relationship attributes. + # Backward compatibility... + if( $use_version eq '2.0' || $use_version eq '3.0' ) { + delete $e->{'class'}; + $e->{'type'} = delete $e->{'relationship'} if exists $e->{'relationship'}; + } + # Add the specified relationship unless we already have done. + my $rel_exists; + if( $e->{'scope'} ne 'local' ) { + my $relobj = $collation->get_relationship( $from, $to ); + if( $relobj && $relobj->scope eq $e->{'scope'} + && $relobj->type eq $e->{'type'} ) { + $rel_exists = 1; + } + } + $collation->add_relationship( $from, $to, $e ) unless $rel_exists; + } + + # Save the text for each witness so that we can ensure consistency + # later on + $collation->text_from_paths(); +} + +## Return the relationship that comes first in priority. +my %LAYERS = ( + 'collated' => 1, + 'orthographic' => 2, + 'spelling' => 3, + ); + +sub _layersort_rel { + my( $a, $b ) = @_; + my $key = exists $a->{'type'} ? 'type' : 'relationship'; + my $at = $LAYERS{$a->{$key}} || 99; + my $bt = $LAYERS{$b->{$key}} || 99; + return $at <=> $bt; } 1;