is( ref( $t ), 'Text::Tradition', "Parsed parallel-segmentation TEI" );
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->collation->readings, 311, "Collation has all readings" );
+ is( scalar $t->collation->paths, 361, "Collation has all paths" );
}
=end testing
foreach my $wit_el ( $xpc->findnodes( "//$LISTWIT/$WITNESS" ) ) {
my $sig = $wit_el->getAttribute( 'xml:id' );
my $source = $wit_el->toString();
- $tradition->add_witness( sigil => $sig, source => $source );
+ $tradition->add_witness( sigil => $sig, sourcetype => 'collation' );
}
map { $text->{$_->sigil} = [] } $tradition->witnesses;
my $sequence = $text->{$sig};
my @real_sequence = ( $c->start );
push( @$sequence, $c->end );
- my $source = $c->start;
- foreach( _clean_sequence( $sig, $sequence ) ) {
- my $rdg = _return_rdg( $_ );
- push( @real_sequence, $rdg );
- $c->add_path( $source, $rdg, $sig );
- $source = $rdg;
+ foreach( _clean_sequence( $sig, $sequence, 1 ) ) {
+ push( @real_sequence, _return_rdg( $_ ) );
}
# See if we need to make an a.c. version of the witness.
if( exists $app_ac->{$sig} ) {
my @uncorrected;
push( @uncorrected, @real_sequence );
+ # Get rid of any remaining placeholders.
+ @real_sequence = _clean_sequence( $sig, \@uncorrected );
+ # Do the uncorrections
foreach my $app ( keys %{$app_ac->{$sig}} ) {
my $start = _return_rdg( $app_anchors->{$app}->{$sig}->{'start'} );
my $end = _return_rdg( $app_anchors->{$app}->{$sig}->{'end'} );
my @new = map { _return_rdg( $_ ) } @{$app_ac->{$sig}->{$app}};
_replace_sequence( \@uncorrected, $start, $end, @new );
}
- my $source = $c->start;
- foreach my $rdg ( @uncorrected ) {
- my $has_base = grep { $_->label eq $sig } $source->edges_to( $rdg );
- if( $rdg ne $c->start && !$has_base ) {
- print STDERR sprintf( "Adding path %s from %s -> %s\n",
- $sig.$c->ac_label, $source->name, $rdg->name );
- $c->add_path( $source, $rdg, $sig.$c->ac_label );
- }
- $source = $rdg;
- }
- print STDERR "Adding a.c. version for witness $sig\n";
+ # and record the results.
+ $tradition->witness( $sig )->uncorrected_path( \@uncorrected );
$tradition->witness( $sig )->is_layered( 1 );
}
+ $tradition->witness( $sig )->path( \@real_sequence );
}
+ # Now make our witness paths.
+ $tradition->collation->make_witness_paths();
# Calculate the ranks for the nodes.
$tradition->collation->calculate_ranks();
# Now that we have ranks, see if we have distinct nodes with identical
# text and identical rank that can be merged.
$tradition->collation->flatten_ranks();
+
+ # And now that we've done that, calculate the common nodes.
+ $tradition->collation->calculate_common_readings();
+
+ # Save the text for each witness so that we can ensure consistency
+ # later on
+ $tradition->collation->text_from_paths();
}
sub _clean_sequence {
- my( $wit, $sequence ) = @_;
+ my( $wit, $sequence, $keep_ac ) = @_;
my @clean_sequence;
foreach my $rdg ( @$sequence ) {
if( $rdg =~ /^PH-(.*)$/ ) {
- # It is a placeholder. Keep it only if we need it.
+ # It is a placeholder. Keep it only if we need it for a later
+ # a.c. run.
my $app_id = $1;
- if( exists $app_ac->{$wit} &&
+ if( $keep_ac && exists $app_ac->{$wit} &&
exists $app_ac->{$wit}->{$app_id} ) {
- print STDERR "Retaining empty placeholder for $app_id\n";
- push( @clean_sequence, $rdg );
+ # print STDERR "Retaining empty placeholder for $app_id\n";
+ push( @clean_sequence, $rdg );
}
} else {
push( @clean_sequence, $rdg );
my( $arr, $start, $end, @new ) = @_;
my( $start_idx, $end_idx );
foreach my $i ( 0 .. $#{$arr} ) {
- $start_idx = $i if( $arr->[$i]->name eq $start );
- if( $arr->[$i]->name eq $end ) {
+ # If $arr->[$i] is a placeholder, cope.
+ my $iid = ref( $arr->[$i] ) ? $arr->[$i]->id : $arr->[$i];
+ $start_idx = $i if( $iid eq $start );
+ if( $iid eq $end ) {
$end_idx = $i;
last;
}
# passed a reading object, return the object.
my $wantobj = ref( $rdg ) eq 'Text::Tradition::Collation::Reading';
my $real = $rdg;
- if( exists $substitutions->{ $wantobj ? $rdg->name : $rdg } ) {
- $real = $substitutions->{ $wantobj ? $rdg->name : $rdg };
- $real = $real->name unless $wantobj;
+ if( exists $substitutions->{ $wantobj ? $rdg->id : $rdg } ) {
+ $real = $substitutions->{ $wantobj ? $rdg->id : $rdg };
+ $real = $real->id unless $wantobj;
}
return $real;
}
next if $w !~ /[[:alnum:]]/;
my $rdg = _make_reading( $tradition->collation, $w );
push( @new_readings, $rdg );
- unless( $in_var ) {
- $rdg->make_common;
- }
foreach ( @cur_wits ) {
warn "Empty wit!" unless $_;
warn "Empty reading!" unless $rdg;
my $xml_id = $xn->getAttribute( 'xml:id' );
my $rdg = _make_reading( $tradition->collation, $xn->textContent, $xml_id );
push( @new_readings, $rdg );
- unless( $in_var ) {
- $rdg->make_common;
- }
foreach( @cur_wits ) {
warn "Empty wit!" unless $_;
warn "Empty reading!" unless $rdg;
# Now collate these sets if we have more than one.
my $subs = collate_variants( $tradition->collation, @sets ) if @sets > 1;
map { $substitutions->{$_} = $subs->{$_} } keys %$subs;
- # TODO Look through substitutions to see if we can make anything common now.
# Return the entire set of unique readings.
my %unique;
foreach my $s ( @sets ) {
- map { $unique{$_->name} = $_ } @$s;
+ map { $unique{$_->id} = $_ } @$s;
}
push( @new_readings, values( %unique ) );
# Exit the current app.
# Add the reading set to the app anchors for each witness
# or put in placeholders for empty p.c. readings
foreach ( @rdg_wits ) {
- my $start = @words ? $words[0]->name : "PH-$current_app";
- my $end = @words ? $words[-1]->name : "PH-$current_app";
+ my $start = @words ? $words[0]->id : "PH-$current_app";
+ my $end = @words ? $words[-1]->id : "PH-$current_app";
$app_anchors->{$current_app}->{$_}->{'start'} = $start;
$app_anchors->{$current_app}->{$_}->{'end'} = $end;
push( @{$text->{$_}}, $start ) unless @words;
my $i = 0;
foreach my $sig ( keys %$text ) {
next if $active_wits{$sig};
- my $l = $tradition->collation->add_lacuna( $current_app . "_$i" );
+ my $l = $tradition->collation->add_reading( {
+ 'id' => $current_app . "_$i",
+ 'is_lacuna' => 1 } );
$i++;
push( @{$text->{$sig}}, $l );
}
unless( $seen_apps == $app_count ) {
foreach my $i ( 0 .. $#cur_wits ) {
my $w = $cur_wits[$i];
- my $l = $tradition->collation->add_lacuna( $current_app . "_$i" );
+ my $l = $tradition->collation->add_reading( {
+ 'id' => $current_app . "_$i",
+ 'is_lacuna' => 1 } );
push( @{$text->{$w}}, $l );
}
}
$xml_id = $try_id;
}
}
- my $rdg = $graph->add_reading( $xml_id );
- $rdg->text( $word );
+ my $rdg = $graph->add_reading(
+ { 'id' => $xml_id,
+ 'text' => $word }
+ );
$used_nodeids{$xml_id} = $rdg;
return $rdg;
}
projects / scpubgit/stemmatology.git / blobdiff