parse-packages

   1 #!/usr/bin/perl -w
   2 # Convert Packages.gz files into Sleepycat db files for efficient usage of
   3 # data
   4 #
   5 # $Id$
   6 #
   7 # Copyright (C) 2006  Jeroen van Wolffelaar <jeroen@wolffelaar.nl>
   8 # This program is free software; you can redistribute it and/or modify
   9 # it under the terms of the GNU General Public License as published by
  10 # the Free Software Foundation; either version 2 of the License, or
  11 # (at your option) any later version.
  12
  13 # This program is distributed in the hope that it will be useful,
  14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 # GNU General Public License for more details.
  17
  18 # You should have received a copy of the GNU General Public License
  19 # along with this program; if not, write to the Free Software
  20 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21
  22 use strict;
  23 use lib './lib';
  24
  25 $| = 1;
  26
  27 my $what = $ARGV[0] ? "non-free" : "*";
  28 # max. distinct results for a given package postfix
  29 my $MAX_PACKAGE_POSTFIXES = 100;
  30
  31 use DB_File;
  32 use Storable;
  33 use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES );
  34 &Packages::Config::init( './' );
  35 my %packages_small = ();
  36 my %package_names = ();
  37 my %package_postfixes = ();
  38 my %sources_packages = ();
  39 my %descriptions = ();
  40 my @descriptions = ("we count lines one-based\000");
  41 my %packages_descriptions = ();
  42 my %descriptions_packages = ();
  43
  44 my %sections = ();
  45 my %subsections = ();
  46 my %priorities = ();
  47
  48 $/ = "";
  49
  50 for my $archive (@ARCHIVES) {
  51     for my $suite (@SUITES) {
  52
  53         my %virtual_packages = ();
  54         print "Reading $archive/$suite...\n";
  55         my %packages_all_db;
  56         tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
  57                 O_RDWR|O_CREAT, 0666, $DB_BTREE
  58                 or die "Error creating DB: $!";
  59         open PKG, "zcat $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
  60         while (<PKG>) {
  61                 next if /^\s*$/;
  62                 my $data = "";
  63                 my %data = ();
  64                 chomp;
  65                 s/\n /\377/g;
  66                 while (/^(\S+):\s*(.*)\s*$/mg) {
  67                         my ($key, $value) = ($1, $2);
  68                         $value =~ s/\377/\n /g;
  69                         $key =~ tr [A-Z] [a-z];
  70                         $data{$key} = $value;
  71                 }
  72                 # Skip double package
  73                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
  74
  75                 if ($data{'provides'}) {
  76                     foreach (split /\s*,\s*/, $data{'provides'}) {
  77                         $virtual_packages{$_}{$data{'package'}}++;
  78                     }
  79                 }
  80                 $package_names{$data{'package'}} = 1;
  81                 my $src = $data{'package'};
  82                 my $src_version = '';
  83                 if ($data{'source'}) {
  84                         $src = $data{'source'};
  85                         $src =~ s/\s+.*//; # strip version info
  86                 }
  87                 $data{'source'} = $src;
  88                 my $descr = $data{'description'};
  89                 my $did = undef;
  90                 if (exists($descriptions{$descr})) {
  91                         $did  = $descriptions{$descr};
  92                 } else {
  93                         $did = 1 + $#descriptions;
  94                         $descriptions[$did] = $descr;
  95                         $descriptions{$descr} = $did;
  96                 }
  97                 $data{'description'} = $did;
  98                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
  99                 $descriptions_packages{$did} .=
 100                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
 101
 102                 my $sdescr = $descr;
 103                 $sdescr =~ s/\n.*//s;
 104                 my $section = 'main';
 105                 my $subsection = $data{section} || '-';
 106                 if ($data{section} && ($data{section} =~ m=/=o)) {
 107                     ($section, $subsection) = split m=/=o, $data{section}, 2;
 108                     ($subsection, $section) = split m=/=o, $data{section}, 2
 109                         if $section eq 'non-US';
 110                 }
 111                 $data{'section'} = $section;
 112                 $data{'subsection'} = $subsection;
 113                 $data{'priority'} ||= '-';
 114                 $sections{$suite}{$archive}{$section}++;
 115                 $subsections{$suite}{$archive}{$subsection}++;
 116                 $priorities{$suite}{$archive}{$data{priority}}++;
 117                 $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
 118                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
 119                 $sources_packages{$src} .=
 120                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
 121                 $data{archive} = $archive;
 122                 while (my ($key, $value) = each (%data)) {
 123                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
 124                                 or $key eq 'maintainer';
 125                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
 126                         $data .= "$key\00$value\00";
 127                 }
 128                 $data =~ s/.$//s;
 129                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
 130                         = $data;
 131         }
 132
 133         foreach (keys %virtual_packages) {
 134             my @provided_by = keys %{$virtual_packages{$_}};
 135             $packages_small{$_} .= "$archive $suite virtual v v v v @provided_by\000";
 136         }
 137
 138         untie %packages_all_db;
 139     }
 140 }
 141
 142 print "Writing databases...\n";
 143 my %packages_small_db;
 144 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
 145         O_RDWR|O_CREAT, 0666, $DB_BTREE
 146         or die "Error creating DB: $!";
 147 while (my ($k, $v) = each(%packages_small)) {
 148         $v =~ s/.$//s;
 149         $packages_small_db{$k} = $v;
 150 }
 151 untie %packages_small_db;
 152
 153 my %sources_packages_db;
 154 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
 155         O_RDWR|O_CREAT, 0666, $DB_BTREE
 156         or die "Error creating DB: $!";
 157 while (my ($k, $v) = each(%sources_packages)) {
 158         $v =~ s/.$//s;
 159         $sources_packages_db{$k} = $v;
 160 }
 161 untie %sources_packages_db;
 162
 163 my %packages_descriptions_db;
 164 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
 165         O_RDWR|O_CREAT, 0666, $DB_BTREE
 166         or die "Error creating DB: $!";
 167 while (my ($k, $v) = each(%packages_descriptions)) {
 168         $packages_descriptions_db{$k} = $v;
 169 }
 170 untie %packages_descriptions_db;
 171
 172 my %descriptions_packages_db;
 173 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
 174         O_RDWR|O_CREAT, 0666, $DB_BTREE
 175         or die "Error creating DB: $!";
 176 while (my ($k, $v) = each(%descriptions_packages)) {
 177         $v =~ s/.$//s;
 178         $descriptions_packages_db{$k} = $v;
 179 }
 180 untie %descriptions_packages_db;
 181
 182 my %descriptions_db;
 183 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
 184         O_RDWR|O_CREAT, 0666, $DB_BTREE
 185         or die "Error creating DB: $!";
 186 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
 187 for (my $i=1; $i<= $#descriptions; $i++) {
 188         my $plain_description = $descriptions[$i];
 189 # WARNING: This needs to correspond with what happens in
 190 # Packages/Search.pm:do_fulltext_search
 191         $plain_description =~ tr [A-Z] [a-z];
 192         # ensure one space on both ends
 193         $plain_description = " $plain_description ";
 194         $plain_description =~ s/[(),.-]+//og;
 195         $plain_description =~ s#[^a-z0-9_/+]+# #og;
 196         print DESCR "$plain_description\n";
 197         $descriptions_db{$i} = $descriptions[$i];
 198 }
 199 close DESCR;
 200 untie %descriptions_db;
 201
 202 # package names stuff:
 203 for my $pkg (keys %package_names) {
 204         for (my $i=0;$i<length($pkg)-1;$i++) {
 205                 my $before = substr($pkg, 0, $i);
 206                 my $after = substr($pkg, $i);
 207                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
 208                 $package_postfixes{$after} .= "$before\0";
 209         }
 210 }
 211 my %package_postfixes_db;
 212 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
 213         O_RDWR|O_CREAT, 0666, $DB_BTREE
 214         or die "Error creating DB: $!";
 215 while (my ($k, $v) = each(%package_postfixes)) {
 216         $v =~ s/.$//s;
 217         my $nr = $v;
 218         $nr =~ s/[^\000]//g;
 219         $nr = length($nr) + 1; # < number of hits
 220         if ($nr > $MAX_PACKAGE_POSTFIXES) {
 221                 $v = "\001" . $nr;
 222         }
 223         $package_postfixes_db{$k} = $v;
 224 }
 225 untie %package_postfixes_db;
 226
 227 store \%sections, "$DBDIR/sections.info";
 228 store \%subsections, "$DBDIR/subsections.info";
 229 store \%priorities, "$DBDIR/priorities.info";
 230
 231 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
 232 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
 233 for my $suite (@SUITES) {
 234         rename("$DBDIR/packages_all_$suite.db.new",
 235                "$DBDIR/packages_all_$suite.db");
 236 }
 237 rename("$DBDIR/packages_descriptions.db.new",
 238        "$DBDIR/packages_descriptions.db");
 239 rename("$DBDIR/descriptions_packages.db.new",
 240        "$DBDIR/descriptions_packages.db");
 241 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
 242 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
 243 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");