Change _small.db layout a bit, to include virtual packages consisely at the

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

diff --git a/bin/parse-packages b/bin/parse-packages
index ab5fbcef5b12e357fbfcce47cea2093e14356e8c..b3aab25fff2ee63b7e779edb86a2928023cba511 100755 (executable)
--- a/bin/parse-packages
+++ b/bin/parse-packages
@@ -22,14 +22,19 @@
 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 = ();
@@ -38,6 +43,10 @@ my @descriptions = ("we count lines one-based\000");
 my %packages_descriptions = ();
 my %descriptions_packages = ();
 
+my %sections = ();
+my %subsections = ();
+my %priorities = ();
+
 $/ = "";
 
 for my $archive (@ARCHIVES) {
@@ -64,16 +73,19 @@ 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'}) {
+                       $virtual_packages{$_}{$suite}{$data{'package'}}++;
+                   }
+               }
                $package_names{$data{'package'}} = 1;
                my $src = $data{'package'};
                my $src_version = '';
                if ($data{'source'}) {
                        $src = $data{'source'};
-                       $src_version = $1
-                           if $src =~ s/\s+\(\s*=\s*(.*)\).*//; # strip version info
+                       $src =~ s/\s+.*//; # strip version info
                }
                $data{'source'} = $src;
-               $data{'source-version'} = $src_version;
                my $descr = $data{'description'};
                my $did = undef;
                if (exists($descriptions{$descr})) {
@@ -94,12 +106,25 @@ 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'} ||= '-';
-               $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
-                       "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
+               $sections{$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} .=
                        "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
                $data{archive} = $archive;
@@ -107,8 +132,9 @@ 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'}"}
                        = $data;
        }
@@ -122,9 +148,28 @@ my %packages_small_db;
 tie %packages_small_db, "DB_File", "$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;
 
@@ -164,8 +209,13 @@ tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
 open DESCR, ">", "$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];
 }
@@ -197,6 +247,10 @@ while (my ($k, $v) = each(%package_postfixes)) {
 }
 untie %package_postfixes_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) {