--- trunk2/all2all.pl 2004/10/10 11:25:10 501 +++ trunk2/all2all.pl 2004/10/17 17:35:32 511 @@ -35,13 +35,13 @@ my $filter = { 'CROVOC' => sub { my $tmp = shift || return; - return undef unless ($tmp =~ s/\s+CROVOC.*$/ */); + return undef unless ($tmp =~ s/\s*CROVOC.*$/ */); return $tmp; }, 'CROVOC_tree' => sub { my $tmp = shift || return; - $tmp =~ s/\s+CROVOC.*$/ */; - $tmp =~ s/\s+EUROVOC.*//; + $tmp =~ s/\s*CROVOC.*$/ */; + $tmp =~ s/\s*EUROVOC.*//; return $tmp; }, }; @@ -138,11 +138,18 @@ $words =~ s/\W*\s+\W*/ /g; $words =~ s/\W+$//; + # first try to generate headline for this entry from index + my $h = shift @{$ds->{'index'}}; + # then, from display + $h ||= shift @{$ds->{'display'}}; + # and as last resport, fallback to headline + $h ||= $headline; + $index->insert( index_name => $ds->{'tag'}, #path => $f, path => $webpac->mfn, - headline => $headline, + headline => $h, words => $words, ); } @@ -151,7 +158,7 @@ foreach my $ds (@ds) { next if (! $ds->{'index'}); - $thes->{$ds->{'tag'}} ||= new WebPAC::Index; + $thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); foreach my $h (@{$ds->{'index'}}) { $thes->{$ds->{'tag'}}->insert(