--- trunk2/all2all.pl 2004/10/10 17:36:43 504 +++ trunk2/all2all.pl 2004/10/26 20:53:03 549 @@ -11,6 +11,7 @@ use YAML; use Carp; use Getopt::Long; +use Text::Unaccent 1.02; use lib './lib'; use WebPAC; @@ -46,6 +47,29 @@ }, }; +## remove accented characters +# +sub unac { + my $string = shift || return; + $string = unac_string($code_page,$string); + $string =~ tr/ðÐ/dD/; + return $string; +} +sub unac_2 { + my $string = shift || return; + if (length($string) > 2) { + my $pr = substr($string,0,2); + $string = unac_string($code_page,substr($string,2)); + $string =~ tr/ðÐ/dD/; + $string = $pr . $string; + } + $string =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; + $string =~ s/ð/d/g; + $string =~ s/&E[tT][hH];/D/g; + $string =~ s/&(\w\w)lig;/$1/gi; + return lc($string); +} + # create WebPAC object # my $webpac = new WebPAC( @@ -139,9 +163,9 @@ $words =~ s/\W+$//; # first try to generate headline for this entry from index - my $h = shift @{$ds->{'index'}}; + my $h = $ds->{'index'}->[0]; # then, from display - $h ||= shift @{$ds->{'display'}}; + $h ||= $ds->{'display'}->[0]; # and as last resport, fallback to headline $h ||= $headline; @@ -150,7 +174,7 @@ #path => $f, path => $webpac->mfn, headline => $h, - words => $words, + words => unac($words), ); } @@ -158,12 +182,13 @@ 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( - headline => $h, + sort_by => unac_2($h), mfn => $webpac->mfn, + headline => $h, ); } } @@ -201,7 +226,7 @@ } # XXX if(0) -$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); +#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); $log->info("creating tree"); @@ -310,13 +335,24 @@ }); my $tree = new WebPAC::Tree( - dir => './out', - html => 'browse.html', + tree => \@tree, +); + +$tree->output( + dir => './out', + html => 'browse.html', template => './output_template/tree.tt', js => 'tree-ids.js', - tree => \@tree, ); +$tree->output( + dir => './eurovoc', + html => 'hijerarhija.html', + template => './output_template/hijerarhija.tt', + js => 'tree-ids.js', +); + + $log->info("closing index"); $index->close;