no quoting needed for tabsep files

[scpubgit/stemmatology.git] / lib / Text / Tradition / Parser / Tabular.pm

diff --git a/lib/Text/Tradition/Parser/Tabular.pm b/lib/Text/Tradition/Parser/Tabular.pm
index 4b53e3c..cdd1804 100644 (file)
--- a/lib/Text/Tradition/Parser/Tabular.pm
+++ b/lib/Text/Tradition/Parser/Tabular.pm
@@ -100,7 +100,7 @@ foreach my $k ( keys %seen_wits ) {
                ok( $wit->has_layertext, "Witness $k has an a.c. version" );
                my $origtext = join( ' ', @{$wit->layertext} );
                my $acsig = $wit->sigil . $t->collation->ac_label;
-               my $graphtext = $t->collation->path_text( $acsig, $wit->sigil );
+               my $graphtext = $t->collation->path_text( $acsig );
                is( $graphtext, $origtext, "Collation matches original a.c. for witness $k" );
        } else {
                ok( !$wit->is_layered, "Witness $k not marked as layered" );
@@ -115,10 +115,13 @@ foreach my $k ( keys %seen_wits ) {
 sub parse {
     my( $tradition, $opts ) = @_;
     my $c = $tradition->collation; # shorthand
-    my $csv = Text::CSV_XS->new( { 
-        binary => 1, # binary for UTF-8
-        sep_char => exists $opts->{'sep_char'} ? $opts->{'sep_char'} : "\t" } 
-        );
+    my $csv_options = { 'binary' => 1 };
+    $csv_options->{'sep_char'} = $opts->{'sep_char'} || "\t";
+    if( $csv_options->{'sep_char'} eq "\t" ) {
+       # If it is really tab separated, nothing is an escape char.
+       $csv_options->{'quote_char'} = undef;
+    }
+    my $csv = Text::CSV_XS->new( $csv_options );
     
     my $alignment_table;
     if( exists $opts->{'string' } ) {
@@ -209,6 +212,7 @@ sub parse {
        my $ac_wit = $tradition->witness( $a );
         my $main_wit = $tradition->witness( $ac_wits{$a} );
         next unless $main_wit;
+        $main_wit->is_layered(1);
         $main_wit->uncorrected_path( $ac_wit->path );
         $tradition->del_witness( $ac_wit );
     }
@@ -219,6 +223,23 @@ sub parse {
        foreach my $rdg ( grep { $_->is_lacuna } $c->readings ) {
                $c->del_reading( $rdg ) unless $c->reading_witnesses( $rdg );
        }
+       
+       # Do a consistency check.
+       foreach my $wit ( $tradition->witnesses ) {
+               my $pathtext = $c->path_text( $wit->sigil );
+               my $origtext = join( ' ', @{$wit->text} );
+               warn "Text differs for witness " . $wit->sigil 
+                       unless $pathtext eq $origtext;
+               if( $wit->is_layered ) {
+                       $pathtext = $c->path_text( $wit->sigil.$c->ac_label );
+                       $origtext = join( ' ', @{$wit->layertext} );
+                       warn "Ante-corr text differs for witness " . $wit->sigil
+                               unless $pathtext eq $origtext;
+               } else {
+                       warn "Text " . $wit->sigil . " has a layered text but is not marked as layered"
+                               if $wit->has_layertext;
+               }
+       }
 }
 
 sub make_nodes {