X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?p=scpubgit%2Fstemmatology.git;a=blobdiff_plain;f=lib%2FText%2FTradition%2FCollation.pm;h=d3a8567cdcd80a3ebcae936b3807bcce26bd6db8;hp=3c65b4ce3f372acc2e57a2cb6447b642712de93c;hb=96ba0418c65f3450b419aea78db41bf697612b63;hpb=10943ab0b79fbd489f6beb3b81a13ed8cbcfafcf diff --git a/lib/Text/Tradition/Collation.pm b/lib/Text/Tradition/Collation.pm index 3c65b4c..d3a8567 100644 --- a/lib/Text/Tradition/Collation.pm +++ b/lib/Text/Tradition/Collation.pm @@ -274,8 +274,10 @@ See L for the available options. sub BUILD { my $self = shift; $self->_set_relations( Text::Tradition::Collation::RelationshipStore->new( 'collation' => $self ) ); - $self->_set_start( $self->add_reading( { 'collation' => $self, 'is_start' => 1 } ) ); - $self->_set_end( $self->add_reading( { 'collation' => $self, 'is_end' => 1 } ) ); + $self->_set_start( $self->add_reading( + { 'collation' => $self, 'is_start' => 1, 'init' => 1 } ) ); + $self->_set_end( $self->add_reading( + { 'collation' => $self, 'is_end' => 1, 'init' => 1 } ) ); } ### Reading construct/destruct functions @@ -284,7 +286,11 @@ sub add_reading { my( $self, $reading ) = @_; unless( ref( $reading ) eq 'Text::Tradition::Collation::Reading' ) { my %args = %$reading; - if( $self->tradition->has_language && !exists $args{'language'} ) { + if( $args{'init'} ) { + # If we are initializing an empty collation, don't assume that we + # have set a tradition. + delete $args{'init'}; + } elsif( $self->tradition->has_language && !exists $args{'language'} ) { $args{'language'} = $self->tradition->language; } $reading = Text::Tradition::Collation::Reading->new( @@ -369,9 +375,21 @@ is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" ); sub merge_readings { my $self = shift; + # Sanity check + my( $kept_obj, $del_obj, $combine, $combine_char ) = $self->_objectify_args( @_ ); + my $mergemeta = $kept_obj->is_meta; + throw( "Cannot merge meta and non-meta reading" ) + unless ( $mergemeta && $del_obj->is_meta ) + || ( !$mergemeta && !$del_obj->is_meta ); + if( $mergemeta ) { + throw( "Cannot merge with start or end node" ) + if( $kept_obj eq $self->start || $kept_obj eq $self->end + || $del_obj eq $self->start || $del_obj eq $self->end ); + } # We only need the IDs for adding paths to the graph, not the reading # objects themselves. - my( $kept, $deleted, $combine, $combine_char ) = $self->_stringify_args( @_ ); + my $kept = $kept_obj->id; + my $deleted = $del_obj->id; $self->_graphcalc_done(0); # The kept reading should inherit the paths and the relationships @@ -391,8 +409,6 @@ sub merge_readings { # Do the deletion deed. if( $combine ) { - my $kept_obj = $self->reading( $kept ); - my $del_obj = $self->reading( $deleted ); my $joinstr = $combine_char; unless( defined $joinstr ) { $joinstr = '' if $kept_obj->join_next || $del_obj->join_prior; @@ -560,7 +576,8 @@ sub as_svg { throw( "Need GraphViz installed to output SVG" ) unless File::Which::which( 'dot' ); my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'}; - $self->calculate_ranks() unless( $self->_graphcalc_done || $opts->{'nocalc'} ); + $self->calculate_ranks() + unless( $self->_graphcalc_done || $opts->{'nocalc'} || !$self->linear ); if( !$self->has_cached_svg || $opts->{'recalc'} || $want_subgraph ) { my @cmd = qw/dot -Tsvg/; my( $svg, $err ); @@ -712,23 +729,28 @@ sub as_dot { $dot .= sprintf( "\t\"%s\" -> \"%s\" %s;\n", $edge->[0], $edge->[1], $varopts ); } elsif( $used{$edge->[0]} ) { - $subend{$edge->[0]} = 1; + $subend{$edge->[0]} = $edge->[1]; } elsif( $used{$edge->[1]} ) { - $substart{$edge->[1]} = 1; + $substart{$edge->[1]} = $edge->[0]; } } # Add substitute start and end edges if necessary foreach my $node ( keys %substart ) { - my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) ); + my $witstr = $self->_path_display_label ( $self->path_witnesses( $substart{$node}, $node ) ); my $variables = { %edge_attrs, 'label' => $witstr }; + my $nrdg = $self->reading( $node ); + if( $nrdg->has_rank && $nrdg->rank > $startrank ) { + # Substart is actually one lower than $startrank + $variables->{'minlen'} = $nrdg->rank - ( $startrank - 1 ); + } my $varopts = _dot_attr_string( $variables ); - $dot .= "\t\"__SUBSTART__\" -> \"$node\" $varopts;"; + $dot .= "\t\"__SUBSTART__\" -> \"$node\" $varopts;\n"; } foreach my $node ( keys %subend ) { - my $witstr = $self->_path_display_label ( $self->reading_witnesses( $self->reading( $node ) ) ); + my $witstr = $self->_path_display_label ( $self->path_witnesses( $node, $subend{$node} ) ); my $variables = { %edge_attrs, 'label' => $witstr }; my $varopts = _dot_attr_string( $variables ); - $dot .= "\t\"$node\" -> \"__SUBEND__\" $varopts;"; + $dot .= "\t\"$node\" -> \"__SUBEND__\" $varopts;\n"; } # HACK part 2 if( $STRAIGHTENHACK ) { @@ -817,6 +839,7 @@ sub _path_display_label { # See if we are in a majority situation. my $maj = scalar( $self->tradition->witnesses ) * 0.6; + $maj = $maj > 5 ? $maj : 5; if( scalar keys %wits > $maj ) { unshift( @disp_ac, 'majority' ); return join( ', ', @disp_ac ); @@ -1077,6 +1100,8 @@ sub as_graphml { # serialize them. Otherwise set nval to undef so that the # key is excluded from this reading. $nval = $nval ? $n->_serialize_lexemes : undef; + } elsif( $d eq 'normal_form' && $n->normal_form eq $n->text ) { + $nval = undef; } if( $rankoffset && $d eq 'rank' && $n ne $self->start ) { # Adjust the ranks within the subgraph. @@ -1520,14 +1545,8 @@ sub calculate_ranks { # Do the rankings based on the relationship equivalence graph, starting # with the start node. - my $topo_start = $self->equivalence( $self->start->id ); - my $node_ranks = { $topo_start => 0 }; - my @curr_origin = ( $topo_start ); - # A little iterative function. - while( @curr_origin ) { - @curr_origin = _assign_rank( $self->equivalence_graph, - $node_ranks, @curr_origin ); - } + my ( $node_ranks, $rank_nodes ) = $self->relations->equivalence_ranks(); + # Transfer our rankings from the topological graph to the real one. foreach my $r ( $self->readings ) { if( defined $node_ranks->{$self->equivalence( $r->id )} ) { @@ -1557,41 +1576,6 @@ sub calculate_ranks { $self->_graphcalc_done(1); } -sub _assign_rank { - my( $graph, $node_ranks, @current_nodes ) = @_; - # Look at each of the children of @current_nodes. If all the child's - # parents have a rank, assign it the highest rank + 1 and add it to - # @next_nodes. Otherwise skip it; we will return when the highest-ranked - # parent gets a rank. - my @next_nodes; - foreach my $c ( @current_nodes ) { - warn "Current reading $c has no rank!" - unless exists $node_ranks->{$c}; - # print STDERR "Looking at child of node $c, rank " - # . $node_ranks->{$c} . "\n"; - foreach my $child ( $graph->successors( $c ) ) { - next if exists $node_ranks->{$child}; - my $highest_rank = -1; - my $skip = 0; - foreach my $parent ( $graph->predecessors( $child ) ) { - if( exists $node_ranks->{$parent} ) { - $highest_rank = $node_ranks->{$parent} - if $highest_rank <= $node_ranks->{$parent}; - } else { - $skip = 1; - last; - } - } - next if $skip; - my $c_rank = $highest_rank + 1; - # print STDERR "Assigning rank $c_rank to node $child \n"; - $node_ranks->{$child} = $c_rank; - push( @next_nodes, $child ); - } - } - return @next_nodes; -} - sub _clear_cache { my $self = shift; $self->wipe_svg if $self->has_cached_svg; @@ -1614,8 +1598,17 @@ sub flatten_ranks { next unless $rdg->has_rank; my $key = $rdg->rank . "||" . $rdg->text; if( exists $unique_rank_rdg{$key} ) { + # Make sure they don't have different grammatical forms + my $ur = $unique_rank_rdg{$key}; + if( $rdg->disambiguated && $ur->disambiguated ) { + my $rform = join( '//', map { $_->form->to_string } $rdg->lexemes ); + my $uform = join( '//', map { $_->form->to_string } $ur->lexemes ); + next unless $rform eq $uform; + } elsif( $rdg->disambiguated xor $ur->disambiguated ) { + next; + } # Combine! - # print STDERR "Combining readings at same rank: $key\n"; + #print STDERR "Combining readings at same rank: $key\n"; $changed = 1; $self->merge_readings( $unique_rank_rdg{$key}, $rdg ); # TODO see if this now makes a common point.