]> git.deb.at Git - deb/packages.git/blobdiff - bin/parse-packages
Change _small.db layout a bit, to include virtual packages consisely at the
[deb/packages.git] / bin / parse-packages
index aad66889f510bb97bb5825ad80b9a5bd697969a5..b3aab25fff2ee63b7e779edb86a2928023cba511 100755 (executable)
 # 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 Deb::Versions;
+use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES );
+&Packages::Config::init( './' );
 my %packages_small = ();
+my %virtual_packages = ();
 my %package_names = ();
 my %package_postfixes = ();
 my %sources_packages = ();
@@ -35,43 +43,49 @@ 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";
+       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 (<PKG>) {
                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;
 
+               if ($data{'provides'}) {
+                   foreach (split /\s*,\s*/, $data{'provides'}) {
+                       $virtual_packages{$_}{$suite}{$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'};
                my $did = undef;
                if (exists($descriptions{$descr})) {
@@ -81,6 +95,7 @@ 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";
@@ -88,32 +103,78 @@ 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';
                }
-               $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ".
-                       "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
+               $data{'section'} = $section;
+               $data{'subsection'} = $subsection;
+               $data{'priority'} ||= '-';
+               $sections{$suite}{$archive}{$section}++;
+               $subsections{$suite}{$archive}{$subsection}++;
+               $priorities{$suite}{$archive}{$data{priority}}++;
+               my $pkgitem = "$archive $suite $data{'architecture'} ".
+                       "$section $subsection $data{'priority'} $data{'version'} $sdescr\0";
+               my $previtem = $packages_small{$data{'package'}}{$suite}{$data{'architecture'}}
+                   || $pkgitem;
+               $packages_small{$data{'package'}}{$suite}{$data{'architecture'}} = $pkgitem
+                   if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) >= 0;
+               $previtem = $packages_small{$data{'package'}}{$suite}{'any'}
+                   || $pkgitem;
+               $packages_small{$data{'package'}}{$suite}{'any'} = $pkgitem
+                   if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) >= 0;
                $sources_packages{$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\00$value\00";
+               }
+               $data =~ s/.$//s;
+               $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
+                       = $data;
        }
 
        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)) {
-       $v =~ s/.$//s;
-       $packages_small_db{$k} = $v;
+while (my ($pkg, $v) = each(%packages_small)) {
+       my ($res1, $res2, $res3) = ("", "", "");
+       while (my ($suite, $v2) = each %$v) {
+           $res2 .= $v2->{'any'};
+           while (my ($arch, $v3) = each %$v2) {
+               next if $arch eq 'any' or $v3 eq $v2->{'any'};
+               $res3 .= $v3;
+           }
+       }
+       
+       if (exists $virtual_packages{$pkg}) {
+               while (my ($suite, $v2) = each %{$virtual_packages{$pkg}}) {
+                       $res1 .= "$suite\01".(join ' ', keys %$v2)."\01";
+               }
+       } else {
+               $res1 .= "-\01-\01";
+       }
+       $res1 =~ s/.$/\0/s;
+
+       my $res = "$res1$res2$res3";
+       $res =~ s/.$//s;
+       $packages_small_db{$pkg} = $res;
 }
 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, $v) = each(%sources_packages)) {
@@ -123,7 +184,7 @@ while (my ($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, $v) = each(%packages_descriptions)) {
@@ -132,7 +193,7 @@ while (my ($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, $v) = each(%descriptions_packages)) {
@@ -142,14 +203,20 @@ while (my ($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 $null_d = $descriptions[$i];
-       $null_d =~ s/\n/\000/g;
-       print DESCR "$null_d\n";
+       my $plain_description = $descriptions[$i];
+# 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];
 }
 close DESCR;
@@ -160,11 +227,12 @@ 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} .= "$before\0";
        }
 }
 my %package_postfixes_db;
-tie %package_postfixes_db, "DB_File", "package_postfixes.db.new",
+tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
        O_RDWR|O_CREAT, 0666, $DB_BTREE
        or die "Error creating DB: $!";
 while (my ($k, $v) = each(%package_postfixes)) {
@@ -179,13 +247,20 @@ while (my ($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");