* Add back some parts of the cron job (100syncarchive_*)

[deb/packages.git] / bin / parse-packages

diff --git a/bin/parse-packages b/bin/parse-packages
index 40c4356eed0e140360366b4cb0eb29c360a0b0cd..7a67e691a9dd0a9e95abc61dcbd8a7ed2a034e0e 100755 (executable)
--- a/bin/parse-packages
+++ b/bin/parse-packages
@@ -35,18 +35,20 @@ my @descriptions = ("we count lines one-based\000");
 my %packages_descriptions = ();
 my %descriptions_packages = ();
 
+my @archives = ( 'us', 'non-US', 'security', 'volatile', 'backports' );
 my @suites = ('oldstable', 'stable', 'testing', 'unstable', 'experimental');
 
 $/ = "";
 
-for my $suite (@suites) {
+for my $archive (@archives) {
+    for my $suite (@suites) {
 
-       print "Reading $suite...\n";
+       print "Reading $archive/$suite...\n";
        my %packages_all_db;
        tie %packages_all_db, "DB_File", "packages_all_$suite.db.new",
                O_RDWR|O_CREAT, 0666, $DB_BTREE
                or die "Error creating DB: $!";
-       open PKG, "zcat /org/ftp.debian.org/ftp/dists/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
+       open PKG, "zcat /org/packages.debian.org/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
        while (<PKG>) {
                next if /^\s*$/;
                my $data = "";
@@ -86,19 +88,22 @@ for my $suite (@suites) {
                my $sdescr = $descr;
                $sdescr =~ s/\n.*//s;
                my $section = 'main';
-               my $subsection = $data{section};
+               my $subsection = $data{section} || '-';
                if ($data{section} && ($data{section} =~ m=/=o)) {
                    ($section, $subsection) = split m=/=o, $data{section}, 2;
                }
                $data{'section'} = $section;
                $data{'subsection'} = $subsection;
-               $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ".
+               $data{'priority'} ||= '-';
+               $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
                        "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
                $sources_packages{$src} .=
-                       "$suite $data{'package'} $data{'version'} $data{'architecture'}\000";
+                       "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
+               $data{archive} = $archive;
                while (my ($key, $value) = each (%data)) {
                        next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
                                or $key eq 'maintainer';
+                       print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
                        $data .= "$key: $value\n";
                }
                $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
@@ -106,6 +111,7 @@ for my $suite (@suites) {
        }
 
        untie %packages_all_db;
+    }
 }
 
 print "Writing databases...\n";
@@ -154,9 +160,10 @@ tie %descriptions_db, "DB_File", "descriptions.db.new",
        or die "Error creating DB: $!";
 open DESCR, "> descriptions.txt" or die "Error creating descriptions textfile";
 for (my $i=1; $i<= $#descriptions; $i++) {
-       my $null_d = $descriptions[$i];
-       $null_d =~ s/\n/\000/g;
-       print DESCR "$null_d\n";
+       my $plain_description = $descriptions[$i];
+       $plain_description =~ s/\n .\n/ /og;
+       $plain_description =~ s/[\n \t]+/ /og;
+       print DESCR "$plain_description\n";
        $descriptions_db{$i} = $descriptions[$i];
 }
 close DESCR;
@@ -167,6 +174,7 @@ for my $pkg (keys %package_names) {
        for (my $i=0;$i<length($pkg)-1;$i++) {
                my $before = substr($pkg, 0, $i);
                my $after = substr($pkg, $i);
+               $before = "^" if $before eq ""; # otherwise split doesn't work properly
                $package_postfixes{$after} .= "$before\0";
        }
 }