X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-packages;h=7a67e691a9dd0a9e95abc61dcbd8a7ed2a034e0e;hb=f7dae3f81707c8cab707df20ba2742a0087baf33;hp=b5c2cebac67eddae65c50ceafc1ca4a3d743736c;hpb=c049f68b67a8ffdce5280f39f96b38abf005641d;p=deb%2Fpackages.git diff --git a/bin/parse-packages b/bin/parse-packages index b5c2ceb..7a67e69 100755 --- a/bin/parse-packages +++ b/bin/parse-packages @@ -21,7 +21,7 @@ use strict; -my $what = $1 ? "non-free" : "*"; +my $what = $ARGV[0] ? "non-free" : "*"; # max. distinct results for a given package postfix my $MAX_PACKAGE_POSTFIXES = 100; @@ -35,47 +35,42 @@ 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 () { next if /^\s*$/; my $data = ""; my %data = (); chomp; s/\n /\377/g; - #s/\376\377\s*\376\377/\376\377/og; while (/^(\S+):\s*(.*)\s*$/mg) { my ($key, $value) = ($1, $2); $value =~ s/\377/\n /g; - $data .= "$key: $value\n"; $key =~ tr [A-Z] [a-z]; $data{$key} = $value; } # Skip double package next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}); - $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"} - = $data; $package_names{$data{'package'}} = 1; my $src = $data{'package'}; - my $srcversion = $data{'version'}; if ($data{'source'}) { $src = $data{'source'}; - if ($src =~ /(\S+) \((\S+)\)/) { - $src = $1; - $srcversion = $2; - } + $src =~ s/ .*//; # strip version info } + $data{'source'} = $src; my $descr = $data{'description'}; my $did = undef; if (exists($descriptions{$descr})) { @@ -85,6 +80,7 @@ for my $suite (@suites) { $descriptions[$did] = $descr; $descriptions{$descr} = $did; } + $data{'description'} = $did; $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did; $descriptions_packages{$did} .= "$data{'package'} $data{'version'} $data{'architecture'}\000"; @@ -92,17 +88,30 @@ 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; } - $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ". + $data{'section'} = $section; + $data{'subsection'} = $subsection; + $data{'priority'} ||= '-'; + $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ". "$section $subsection $data{'priority'} $data{'version'} $sdescr\000"; $sources_packages{$src} .= - "$data{'package'} $data{'architecture'} $data{'version'}\000"; + "$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'}"} + = $data; } untie %packages_all_db; + } } print "Writing databases...\n"; @@ -151,9 +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; @@ -164,6 +174,7 @@ for my $pkg (keys %package_names) { for (my $i=0;$i