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
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(
$self->_add_reading( $reading->id => $reading );
# Once the reading has been added, put it in both graphs.
$self->sequence->add_vertex( $reading->id );
- # All meta readings save 'start' and 'end' get disregarded for relationships.
- unless( $reading->is_nonrel ) {
- $self->relations->add_reading( $reading->id );
- }
+ $self->relations->add_reading( $reading->id );
return $reading;
};
my $self = shift;
my $arg = shift;
- unless( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
- $arg = $self->reading( $arg )
+ if( ref( $arg ) eq 'Text::Tradition::Collation::Reading' ) {
+ $arg = $arg->id;
}
- my $argid = $arg->id;
# Remove the reading from the graphs.
$self->_graphcalc_done(0);
$self->_clear_cache; # Explicitly clear caches to GC the reading
- $self->sequence->delete_vertex( $argid );
- $self->relations->delete_reading( $argid )
- unless $arg->is_nonrel;
+ $self->sequence->delete_vertex( $arg );
+ $self->relations->delete_reading( $arg );
# Carry on.
- $self->$orig( $argid );
+ $self->$orig( $arg );
};
=begin testing
@wits{keys %$fwits} = values %$fwits;
$self->sequence->set_edge_attributes( @vector, \%wits );
}
- $self->relations->merge_readings( $kept, $deleted, $combine )
- unless $mergemeta;
+ $self->relations->merge_readings( $kept, $deleted, $combine );
# Do the deletion deed.
if( $combine ) {
+ # Combine the text of the readings
my $joinstr = $combine_char;
unless( defined $joinstr ) {
$joinstr = '' if $kept_obj->join_next || $del_obj->join_prior;
$joinstr = $self->wordsep unless defined $joinstr;
}
$kept_obj->alter_text( join( $joinstr, $kept_obj->text, $del_obj->text ) );
+ $kept_obj->normal_form(
+ join( $joinstr, $kept_obj->normal_form, $del_obj->normal_form ) );
+ # Combine the lexemes present in the readings
+ if( $kept_obj->has_lexemes && $del_obj->has_lexemes ) {
+ $kept_obj->add_lexeme( $del_obj->lexemes );
+ }
}
$self->del_reading( $deleted );
}
# We only need the IDs for adding paths to the graph, not the reading
# objects themselves.
- my( $source, $target, $wit ) = $self->_objectify_args( @_ );
+ my( $source, $target, $wit ) = $self->_stringify_args( @_ );
$self->_graphcalc_done(0);
# Connect the readings
$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 ) {
# 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.
# Transfer our rankings from the topological graph to the real one.
foreach my $r ( $self->readings ) {
- if( $r->is_nonrel ) {
- # These are not in the equivalence graph. Grab the rank of the highest
- # predecessor + 1.
- my @preds = $self->sequence->predecessors( $r );
- my $mrank = 0;
- map { my $rk = $node_ranks->{$self->equivalence( $_ )} + 1;
- $mrank = $rk > $mrank ? $rk : $mrank; }
- $self->sequence->predecessors( $r );
- throw( "All predecessors of $r unranked!" ) unless $mrank;
- $r->rank( $mrank );
- } elsif( defined $node_ranks->{$self->equivalence( $r->id )} ) {
+ if( defined $node_ranks->{$self->equivalence( $r->id )} ) {
$r->rank( $node_ranks->{$self->equivalence( $r->id )} );
} else {
# Die. Find the last rank we calculated.
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.
projects / scpubgit/stemmatology.git / blobdiff