bin/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 warnings;
  24 use lib './lib';
  25
  26 $| = 1;
  27
  28 my $what = $ARGV[0] ? "non-free" : "*";
  29 # max. distinct results for a given package postfix
  30 my $MAX_PACKAGE_POSTFIXES = 100;
  31
  32 use DB_File;
  33 use Storable;
  34 use File::Path;
  35 use Deb::Versions;
  36 use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES );
  37 &Packages::Config::init( './' );
  38 my %packages_small = ();
  39 my %virtual_packages = ();
  40 my %package_names = ();
  41 my %package_postfixes = ();
  42 my %sources_packages = ();
  43 my %descriptions = ();
  44 my @descriptions = ("we count lines one-based\000");
  45 my %packages_descriptions = ();
  46 my %descriptions_packages = ();
  47
  48 my %sections = ();
  49 my %subsections = ();
  50 my %priorities = ();
  51
  52 $/ = "";
  53
  54 -d $DBDIR || mkpath( $DBDIR );
  55
  56 for my $suite (@SUITES) {
  57     my %package_names_suite = ();
  58     my %packages_all_db;
  59     tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
  60             O_RDWR|O_CREAT, 0666, $DB_BTREE
  61             or die "Error creating DB: $!";
  62
  63     for my $archive (@ARCHIVES) {
  64         print "Reading $archive/$suite...\n";
  65         if (!-d "$TOPDIR/archive/$archive/$suite/") {
  66                 print "\tseems not to exist, skipping...\n";
  67                 next;
  68         }
  69         open PKG, "zcat $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
  70         while (<PKG>) {
  71                 next if /^\s*$/;
  72                 my $data = "";
  73                 my %data = ();
  74                 chomp;
  75                 s/\n /\377/g;
  76                 while (/^(\S+):\s*(.*)\s*$/mg) {
  77                         my ($key, $value) = ($1, $2);
  78                         $value =~ s/\377/\n /g;
  79                         $key =~ tr [A-Z] [a-z];
  80                         $data{$key} = $value;
  81                 }
  82                 # Skip double package
  83                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
  84                 # Skip arch:all for amd64&kfreebsd, too often broken
  85                 next if ($archive eq 'amd64' or $archive eq 'kfreebsd')
  86                     and $data{architecture} eq 'all';
  87
  88                 if ($data{'provides'}) {
  89                     foreach (split /\s*,\s*/, $data{'provides'}) {
  90                         $virtual_packages{$_}{$suite}{$data{'package'}}++;
  91                         $packages_small{$_} ||= {};
  92                     }
  93                 }
  94                 $package_names{$data{'package'}} = 1;
  95                 $package_names_suite{$data{'package'}} = 1;
  96                 my $src = $data{'package'};
  97                 my $src_version = '';
  98                 if ($data{'source'}) {
  99                         $src = $data{'source'};
 100                         $src =~ s/\s+.*//; # strip version info
 101                 }
 102                 $data{'source'} = $src;
 103                 my $descr = $data{'description'};
 104                 my $did = undef;
 105                 if (exists($descriptions{$descr})) {
 106                         $did  = $descriptions{$descr};
 107                 } else {
 108                         $did = 1 + $#descriptions;
 109                         $descriptions[$did] = $descr;
 110                         $descriptions{$descr} = $did;
 111                 }
 112                 $data{'description'} = $did;
 113                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
 114                 $descriptions_packages{$did} .=
 115                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
 116
 117                 my $sdescr = $descr;
 118                 $sdescr =~ s/\n.*//s;
 119                 my $section = 'main';
 120                 my $subsection = $data{section} || '-';
 121                 if ($data{section} && ($data{section} =~ m=/=o)) {
 122                     ($section, $subsection) = split m=/=o, $data{section}, 2;
 123                     ($subsection, $section) = split m=/=o, $data{section}, 2
 124                         if $section eq 'non-US';
 125                 }
 126                 $data{'section'} = $section;
 127                 $data{'subsection'} = $subsection;
 128                 $data{'priority'} ||= '-';
 129                 $sections{$suite}{$section}++;
 130                 $subsections{$suite}{$subsection}++;
 131                 $priorities{$suite}{$data{priority}}++;
 132                 my $pkgitem = "$archive $suite $data{'architecture'} ".
 133                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\0";
 134                 my $previtem = ($packages_small{$data{'package'}}{$suite}{$data{'architecture'}}
 135                     ||= $pkgitem);
 136                 $packages_small{$data{'package'}}{$suite}{$data{'architecture'}} = $pkgitem
 137                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
 138                 $previtem = ($packages_small{$data{'package'}}{$suite}{'any'}
 139                     ||= $pkgitem);
 140                 $packages_small{$data{'package'}}{$suite}{'any'} = $pkgitem
 141                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
 142                 $sources_packages{$src} .=
 143                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
 144                 $data{archive} = $archive;
 145                 while (my ($key, $value) = each (%data)) {
 146                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
 147                                 or $key eq 'maintainer';
 148                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
 149                         $data .= "$key\00$value\00";
 150                 }
 151                 $data =~ s/.$//s;
 152                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
 153                         = $data;
 154         }
 155     }
 156
 157     open NAMES, '>', "$DBDIR/package_names_$suite.txt.new"
 158         or die "Error creating package names list: $!";
 159     foreach (sort keys %package_names_suite) {
 160         print NAMES "$_\n";
 161     }
 162     close NAMES;
 163
 164     untie %packages_all_db;
 165 }
 166
 167 print "Writing databases...\n";
 168 my %packages_small_db;
 169 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
 170         O_RDWR|O_CREAT, 0666, $DB_BTREE
 171         or die "Error creating DB: $!";
 172 while (my ($pkg, $v) = each(%packages_small)) {
 173         my ($res1, $res2, $res3) = ("", "", "");
 174         while (my ($suite, $v2) = each %$v) {
 175             $res2 .= $v2->{'any'};
 176             while (my ($arch, $v3) = each %$v2) {
 177                 next if $arch eq 'any' or $v3 eq $v2->{'any'};
 178                 $res3 .= $v3;
 179             }
 180         }
 181
 182         if (exists $virtual_packages{$pkg}) {
 183                 while (my ($suite, $v2) = each %{$virtual_packages{$pkg}}) {
 184                         $res1 .= "$suite\01".(join ' ', keys %$v2)."\01";
 185                 }
 186         } else {
 187                 $res1 .= "-\01-\01";
 188         }
 189         $res1 =~ s/.$/\0/s;
 190
 191         my $res = "$res1$res2$res3";
 192         $res =~ s/.$//s;
 193         $packages_small_db{$pkg} = $res;
 194 }
 195 untie %packages_small_db;
 196
 197 my %sources_packages_db;
 198 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
 199         O_RDWR|O_CREAT, 0666, $DB_BTREE
 200         or die "Error creating DB: $!";
 201 while (my ($k, $v) = each(%sources_packages)) {
 202         $v =~ s/.$//s;
 203         $sources_packages_db{$k} = $v;
 204 }
 205 untie %sources_packages_db;
 206
 207 my %packages_descriptions_db;
 208 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
 209         O_RDWR|O_CREAT, 0666, $DB_BTREE
 210         or die "Error creating DB: $!";
 211 while (my ($k, $v) = each(%packages_descriptions)) {
 212         $packages_descriptions_db{$k} = $v;
 213 }
 214 untie %packages_descriptions_db;
 215
 216 my %descriptions_packages_db;
 217 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
 218         O_RDWR|O_CREAT, 0666, $DB_BTREE
 219         or die "Error creating DB: $!";
 220 while (my ($k, $v) = each(%descriptions_packages)) {
 221         $v =~ s/.$//s;
 222         $descriptions_packages_db{$k} = $v;
 223 }
 224 untie %descriptions_packages_db;
 225
 226 my %descriptions_db;
 227 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
 228         O_RDWR|O_CREAT, 0666, $DB_BTREE
 229         or die "Error creating DB: $!";
 230 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
 231 for (my $i=1; $i<= $#descriptions; $i++) {
 232         my $plain_description = $descriptions[$i];
 233 # WARNING: This needs to correspond with what happens in
 234 # Packages/Search.pm:do_fulltext_search
 235         $plain_description =~ tr [A-Z] [a-z];
 236         # ensure one space on both ends
 237         $plain_description = " $plain_description ";
 238         $plain_description =~ s/[(),.-]+//og;
 239         $plain_description =~ s#[^a-z0-9_/+]+# #og;
 240         print DESCR "$plain_description\n";
 241         $descriptions_db{$i} = $descriptions[$i];
 242 }
 243 close DESCR;
 244 untie %descriptions_db;
 245
 246 # package names stuff:
 247 for my $pkg (keys %package_names) {
 248         for (my $i=0;$i<length($pkg)-1;$i++) {
 249                 my $before = substr($pkg, 0, $i);
 250                 my $after = substr($pkg, $i);
 251                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
 252                 $package_postfixes{$after} .= "$before\0";
 253         }
 254 }
 255 my %package_postfixes_db;
 256 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
 257         O_RDWR|O_CREAT, 0666, $DB_BTREE
 258         or die "Error creating DB: $!";
 259 while (my ($k, $v) = each(%package_postfixes)) {
 260         $v =~ s/.$//s;
 261         my $nr = $v;
 262         $nr =~ s/[^\000]//g;
 263         $nr = length($nr) + 1; # < number of hits
 264         if ($nr > $MAX_PACKAGE_POSTFIXES) {
 265                 $v = "\001" . $nr;
 266         }
 267         $package_postfixes_db{$k} = $v;
 268 }
 269 untie %package_postfixes_db;
 270
 271 store \%sections, "$DBDIR/sections.info";
 272 store \%subsections, "$DBDIR/subsections.info";
 273 store \%priorities, "$DBDIR/priorities.info";
 274
 275 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
 276 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
 277 for my $suite (@SUITES) {
 278         rename("$DBDIR/packages_all_$suite.db.new",
 279                "$DBDIR/packages_all_$suite.db");
 280         rename("$DBDIR/package_names_$suite.txt.new",
 281                "$DBDIR/package_names_$suite.txt");
 282 }
 283 rename("$DBDIR/packages_descriptions.db.new",
 284        "$DBDIR/packages_descriptions.db");
 285 rename("$DBDIR/descriptions_packages.db.new",
 286        "$DBDIR/descriptions_packages.db");
 287 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
 288 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
 289 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");