X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-packages;h=399a62ca060656a82314f4f80000a7462899b938;hb=fcd2a8e03040589f3048e91737d4b2863c8521fd;hp=7a67e691a9dd0a9e95abc61dcbd8a7ed2a034e0e;hpb=f7dae3f81707c8cab707df20ba2742a0087baf33;p=deb%2Fpackages.git diff --git a/bin/parse-packages b/bin/parse-packages index 7a67e69..399a62c 100755 --- a/bin/parse-packages +++ b/bin/parse-packages @@ -20,12 +20,18 @@ # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA use strict; +use lib './lib'; + +$| = 1; 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 = (); @@ -35,20 +41,22 @@ 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'); +my %sections = (); +my %subsections = (); +my %priorities = (); $/ = ""; -for my $archive (@archives) { - for my $suite (@suites) { +for my $archive (@ARCHIVES) { + for my $suite (@SUITES) { + 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: $!"; - open PKG, "zcat /org/packages.debian.org/archive/$archive/$suite/$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 = ""; @@ -64,11 +72,17 @@ for my $archive (@archives) { # Skip double package next if exists($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'}; @@ -91,10 +105,15 @@ for my $archive (@archives) { 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; $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} .= @@ -104,19 +123,25 @@ for my $archive (@archives) { 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"; + $data .= "$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, $v) = each(%packages_small)) { @@ -126,7 +151,7 @@ while (my ($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, $v) = each(%sources_packages)) { @@ -136,7 +161,7 @@ while (my ($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, $v) = each(%packages_descriptions)) { @@ -145,7 +170,7 @@ while (my ($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, $v) = each(%descriptions_packages)) { @@ -155,14 +180,19 @@ while (my ($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]; } @@ -179,7 +209,7 @@ for my $pkg (keys %package_names) { } } my %package_postfixes_db; -tie %package_postfixes_db, "DB_File", "package_postfixes.db.new", +tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new", O_RDWR|O_CREAT, 0666, $DB_BTREE or die "Error creating DB: $!"; while (my ($k, $v) = each(%package_postfixes)) { @@ -194,13 +224,20 @@ while (my ($k, $v) = each(%package_postfixes)) { } untie %package_postfixes_db; -rename("packages_small.db.new", "packages_small.db"); -rename("sources_packages.db.new", "sources_packages.db"); -for my $suite (@suites) { - rename("packages_all_$suite.db.new", "packages_all_$suite.db"); +store \%sections, "$DBDIR/sections.info"; +store \%subsections, "$DBDIR/subsections.info"; +store \%priorities, "$DBDIR/priorities.info"; + +rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db"); +rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db"); +for my $suite (@SUITES) { + rename("$DBDIR/packages_all_$suite.db.new", + "$DBDIR/packages_all_$suite.db"); } -rename("packages_descriptions.db.new", "packages_descriptions.db"); -rename("descriptions_packages.db.new", "descriptions_packages.db"); -rename("descriptions.txt.new", "descriptions.txt"); -rename("descriptions.db.new", "descriptions.db"); -rename("package_postfixes.db.new", "package_postfixes.db"); +rename("$DBDIR/packages_descriptions.db.new", + "$DBDIR/packages_descriptions.db"); +rename("$DBDIR/descriptions_packages.db.new", + "$DBDIR/descriptions_packages.db"); +rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt"); +rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db"); +rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");