some parser bugfixes / workarounds

diff --git a/base/lib/Text/Tradition/Parser/CTE.pm b/base/lib/Text/Tradition/Parser/CTE.pm
index d7dc584..f161068 100644 (file)
--- a/base/lib/Text/Tradition/Parser/CTE.pm
+++ b/base/lib/Text/Tradition/Parser/CTE.pm
@@ -138,10 +138,14 @@ sub _remove_formatting {
                my $parent = $n->parentNode();
                my @children = $n->childNodes();
                my $first = shift @children;
-               $parent->replaceChild( $first, $n );
-               foreach my $c ( @children ) {
-                       $parent->insertAfter( $c, $first );
-                       $first = $c;
+               if( $first ) {
+                       $parent->replaceChild( $first, $n );
+                       foreach my $c ( @children ) {
+                               $parent->insertAfter( $c, $first );
+                               $first = $c;
+                       }
+               } else {
+                       $parent->removeChild( $n );
                }
        }
        
@@ -176,7 +180,7 @@ sub _get_base {
                # Anchor to mark the end of some apparatus; save its ID.
                push( @readings, { 'type' => 'anchor', 
                    'content' => $xn->getAttribute( 'xml:id' ) } );
-       } elsif ( $xn->nodeName ne 'note' ) {  # Any tag we don't know to disregard
+       } elsif ( $xn->nodeName !~ /^(note|seg)$/ ) {  # Any tag we don't know to disregard
            print STDERR "Unrecognized tag " . $xn->nodeName . "\n";
        }
        return @readings;
@@ -414,6 +418,13 @@ sub _expand_all_paths {
     
     # Make the path edges
     $c->make_witness_paths();
+    
+    # Now remove any orphan nodes, and warn that we are doing so.
+    foreach my $v ( $c->sequence->isolated_vertices ) {
+       my $r = $c->reading( $v );
+       print STDERR "Deleting orphan reading $r / " . $r->text;
+       $c->del_reading( $r );
+    }
 }
 
 sub _add_wit_path {

diff --git a/base/lib/Text/Tradition/Parser/JSON.pm b/base/lib/Text/Tradition/Parser/JSON.pm
index b32c904..b8e09c7 100644 (file)
--- a/base/lib/Text/Tradition/Parser/JSON.pm
+++ b/base/lib/Text/Tradition/Parser/JSON.pm
@@ -109,7 +109,7 @@ sub parse {
     my %ac_wits;  # Track these for later removal
     foreach my $sigil ( map { $_->{'witness'} } @{$table->{'alignment'}} ) {
        my $wit;
-       if( $tradition->has_witness( $sigil ) {
+       if( $tradition->has_witness( $sigil ) ) {
                $wit = $tradition->witness( $sigil );
                $wit->is_collated( 1 );
        } else {