X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-packages;h=e6a30829cd5ad30bb5f853d4ff3c8e328d624aec;hb=bbe90f9136f133811370f0d3c58a96c7ee2dc795;hp=4fd7e346113955634e25bde914faeed5dea12fc9;hpb=4180f4a4fedc35765e9f9063cbfa305d5d55d55b;p=deb%2Fpackages.git diff --git a/bin/parse-packages b/bin/parse-packages index 4fd7e34..e6a3082 100755 --- a/bin/parse-packages +++ b/bin/parse-packages @@ -20,12 +20,16 @@ # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA use strict; +use lib './lib'; my $what = $ARGV[0] ? "non-free" : "*"; # max. distinct results for a given package postfix my $MAX_PACKAGE_POSTFIXES = 100; use DB_File; +use Storable; +use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES ); +&Packages::Config::init( './' ); my %packages_small = (); my %package_names = (); my %package_postfixes = (); @@ -35,18 +39,22 @@ my @descriptions = ("we count lines one-based\000"); my %packages_descriptions = (); my %descriptions_packages = (); -my @suites = ('oldstable', 'stable', 'testing', 'unstable', 'experimental'); +my %sections = (); +my %subsections = (); +my %priorities = (); $/ = ""; -for my $suite (@suites) { +for my $archive (@ARCHIVES) { + for my $suite (@SUITES) { - print "Reading $suite...\n"; + my %virtual_packages = (); + print "Reading $archive/$suite...\n"; my %packages_all_db; - tie %packages_all_db, "DB_File", "packages_all_$suite.db.new", + tie %packages_all_db, "DB_File", "$DBDIR/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 $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|"; while () { next if /^\s*$/; my $data = ""; @@ -62,11 +70,17 @@ for my $suite (@suites) { # Skip double package next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}); + if ($data{'provides'}) { + foreach (split /\s*,\s*/, $data{'provides'}) { + $virtual_packages{$_}{$data{'package'}}++; + } + } $package_names{$data{'package'}} = 1; my $src = $data{'package'}; + my $src_version = ''; if ($data{'source'}) { $src = $data{'source'}; - $src =~ s/ .*//; # strip version info + $src =~ s/\s+.*//; # strip version info } $data{'source'} = $src; my $descr = $data{'description'}; @@ -86,31 +100,46 @@ 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, $data{section}, 2; + ($subsection, $section) = split m=/=o, $data{section}, 2 + if $section eq 'non-US'; } $data{'section'} = $section; $data{'subsection'} = $subsection; - $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ". + $data{'priority'} ||= '-'; + $sections{$suite}{$archive}{$section}++; + $subsections{$suite}{$archive}{$subsection}++; + $priorities{$suite}{$archive}{$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'; - $data .= "$key: $value\n"; + print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value; + $data .= "$key\00$value\00"; } + $data =~ s/.$//s; $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"} = $data; } + foreach (keys %virtual_packages) { + my @provided_by = keys %{$virtual_packages{$_}}; + $packages_small{$_} .= "$archive $suite virtual v v v v @provided_by\000"; + } + untie %packages_all_db; + } } print "Writing databases...\n"; my %packages_small_db; -tie %packages_small_db, "DB_File", "packages_small.db.new", +tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; while (my ($k, $v) = each(%packages_small)) { @@ -120,7 +149,7 @@ while (my ($k, $v) = each(%packages_small)) { untie %packages_small_db; my %sources_packages_db; -tie %sources_packages_db, "DB_File", "sources_packages.db.new", +tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; while (my ($k, $v) = each(%sources_packages)) { @@ -130,7 +159,7 @@ while (my ($k, $v) = each(%sources_packages)) { untie %sources_packages_db; my %packages_descriptions_db; -tie %packages_descriptions_db, "DB_File", "packages_descriptions.db.new", +tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; while (my ($k, $v) = each(%packages_descriptions)) { @@ -139,7 +168,7 @@ while (my ($k, $v) = each(%packages_descriptions)) { untie %packages_descriptions_db; my %descriptions_packages_db; -tie %descriptions_packages_db, "DB_File", "descriptions_packages.db.new", +tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; while (my ($k, $v) = each(%descriptions_packages)) { @@ -149,14 +178,19 @@ while (my ($k, $v) = each(%descriptions_packages)) { untie %descriptions_packages_db; my %descriptions_db; -tie %descriptions_db, "DB_File", "descriptions.db.new", +tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; -open DESCR, "> descriptions.txt" or die "Error creating descriptions textfile"; +open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile"; for (my $i=1; $i<= $#descriptions; $i++) { my $plain_description = $descriptions[$i]; - $plain_description =~ s/\n .\n/ /og; - $plain_description =~ s/[\n \t]+/ /og; +# WARNING: This needs to correspond with what happens in +# Packages/Search.pm:do_fulltext_search + $plain_description =~ tr [A-Z] [a-z]; + # ensure one space on both ends + $plain_description = " $plain_description "; + $plain_description =~ s/[(),.-]+//og; + $plain_description =~ s#[^a-z0-9_/+]+# #og; print DESCR "$plain_description\n"; $descriptions_db{$i} = $descriptions[$i]; } @@ -168,11 +202,12 @@ for my $pkg (keys %package_names) { for (my $i=0;$i