X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-debtags-voc;h=be251781a89616dafd66b880f84473a15675ca88;hb=49a00e7a46d7a32787c5deb6bc348fa7527c7d87;hp=b6391564402db3bde8a04d621cfe8273ccff9f7c;hpb=2ad60a1122641f5c264f36a690847b76d38120b3;p=deb%2Fpackages.git diff --git a/bin/parse-debtags-voc b/bin/parse-debtags-voc index b639156..be25178 100755 --- a/bin/parse-debtags-voc +++ b/bin/parse-debtags-voc @@ -22,6 +22,7 @@ use lib './lib'; $| = 1; +use POSIX; use DB_File; use File::Path; use Data::Dumper; @@ -30,9 +31,10 @@ use URI::Escape; use Deb::Versions; use Packages::Template; -use Packages::Config qw( $TOPDIR ); -use Packages::CGI; +use Packages::Config qw( $TOPDIR @LANGUAGES ); +use Packages::I18N::Locale; &Packages::Config::init( './' ); +&Packages::I18N::Locale::load( "$TOPDIR/po" ); my $debtagsdir = "$TOPDIR/files/debtags"; my $wwwdir = "$TOPDIR/www/about"; my $voc_file = "$debtagsdir/vocabulary"; @@ -40,10 +42,20 @@ my (%voc, %voc_db); $/ = ""; -delete $ENV{'LANGUAGE'}; -delete $ENV{'LANG'}; -delete $ENV{'LC_ALL'}; -delete $ENV{'LC_MESSAGES'}; +sub process_desc { + my ($desc) = @_; + + if ($desc) { + $desc = encode_entities($desc); + + $desc =~ s,((ftp|http|https)://[\S~-]+?/?)((\>\;)?[)]?[']?[:.\,]?(\s|$)),$1$3,go; # syntax highlighting -> ']; + $desc =~ s/\A //o; + $desc =~ s/\n /\n/sgo; + $desc =~ s/\n.\n/\n
\n/go; + $desc =~ s/(((\n|\A) [^\n]*)+)/\n
$1\n<\/pre>/sgo; + } + return $desc; +} print "Parsing Vocabulary...\n"; tie %voc_db, "DB_File", "$debtagsdir/vocabulary.db.new", @@ -72,18 +84,22 @@ while () { warn "Duplicated key found: $voc_key\n"; next; } - my ($sdesc,$ldesc) = split /\n/, encode_entities($data{description}), 2; - - if ($ldesc) { - $ldesc =~ s,((ftp|http|https)://[\S~-]+?/?)((\>\;)?[)]?[']?[:.\,]?(\s|$)),$1$3,go; # syntax highlighting -> ']; - $ldesc =~ s/\A //o; - $ldesc =~ s/\n /\n/sgo; - $ldesc =~ s/\n.\n/\n \n/go; - $ldesc =~ s/(((\n|\A) [^\n]*)+)/\n
$1\n<\/pre>/sgo; - } - $data{html_description} = [ $sdesc, $ldesc||"" ]; + my ($sdesc,$ldesc) = split /\n/, $data{description}, 2; + $data{html_description} = [ encode_entities($sdesc), process_desc($ldesc)||"" ]; $voc_db{$voc_key} = $sdesc || ""; + + foreach my $lang (@LANGUAGES) { + next if $lang eq 'en'; + + my $cat = Packages::I18N::Locale->get_handle($lang) + or die "get_handle failed for $lang"; + + my $sdesc_trans = $cat->maketext($sdesc); + $voc_db{"$voc_key-$lang"} = $sdesc_trans + if $sdesc_trans and $sdesc_trans ne $sdesc; + } + $voc{$voc_key} = \%data; } @@ -102,7 +118,7 @@ my @facets = sort( grep { exists $voc{$_}{facet} } keys %voc ); my @tags = sort( grep { exists $voc{$_}{tag} } keys %voc ); my %tags_by_facet; foreach (@tags) { - my ($facet, $tag) = split /::/, $_, 2; + my ($facet, $tag) = split m/::/, $_, 2; warn "No facet data available for $facet\n" unless exists $voc{$facet}; $tags_by_facet{$facet} ||= [];