]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
Show filelist from suite/arch specific db
[deb/packages.git] / 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 lib './lib';
24
25 my $what = $ARGV[0] ? "non-free" : "*";
26 # max. distinct results for a given package postfix
27 my $MAX_PACKAGE_POSTFIXES = 100;
28
29 use DB_File;
30 use Storable;
31 use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES );
32 &Packages::Config::init( './' );
33 my %packages_small = ();
34 my %package_names = ();
35 my %package_postfixes = ();
36 my %sources_packages = ();
37 my %descriptions = ();
38 my @descriptions = ("we count lines one-based\000");
39 my %packages_descriptions = ();
40 my %descriptions_packages = ();
41
42 my %sections = ();
43 my %subsections = ();
44 my %priorities = ();
45
46 $/ = "";
47
48 for my $archive (@ARCHIVES) {
49     for my $suite (@SUITES) {
50
51         print "Reading $archive/$suite...\n";
52         my %packages_all_db;
53         tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
54                 O_RDWR|O_CREAT, 0666, $DB_BTREE
55                 or die "Error creating DB: $!";
56         open PKG, "zcat $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
57         while (<PKG>) {
58                 next if /^\s*$/;
59                 my $data = "";
60                 my %data = ();
61                 chomp;
62                 s/\n /\377/g;
63                 while (/^(\S+):\s*(.*)\s*$/mg) {
64                         my ($key, $value) = ($1, $2);
65                         $value =~ s/\377/\n /g;
66                         $key =~ tr [A-Z] [a-z];
67                         $data{$key} = $value;
68                 }
69                 # Skip double package
70                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
71
72                 $package_names{$data{'package'}} = 1;
73                 my $src = $data{'package'};
74                 my $src_version = '';
75                 if ($data{'source'}) {
76                         $src = $data{'source'};
77                         $src =~ s/\s+.*//; # strip version info
78                 }
79                 $data{'source'} = $src;
80                 my $descr = $data{'description'};
81                 my $did = undef;
82                 if (exists($descriptions{$descr})) {
83                         $did  = $descriptions{$descr};
84                 } else {
85                         $did = 1 + $#descriptions;
86                         $descriptions[$did] = $descr;
87                         $descriptions{$descr} = $did;
88                 }
89                 $data{'description'} = $did;
90                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
91                 $descriptions_packages{$did} .=
92                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
93
94                 my $sdescr = $descr;
95                 $sdescr =~ s/\n.*//s;
96                 my $section = 'main';
97                 my $subsection = $data{section} || '-';
98                 if ($data{section} && ($data{section} =~ m=/=o)) {
99                     ($section, $subsection) = split m=/=o, $data{section}, 2;
100                     ($subsection, $section) = split m=/=o, $data{section}, 2
101                         if $section eq 'non-US';
102                 }
103                 $data{'section'} = $section;
104                 $data{'subsection'} = $subsection;
105                 $data{'priority'} ||= '-';
106                 $sections{$suite}{$archive}{$section}++;
107                 $subsections{$suite}{$archive}{$subsection}++;
108                 $priorities{$suite}{$archive}{$data{priority}}++;
109                 $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
110                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
111                 $sources_packages{$src} .=
112                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
113                 $data{archive} = $archive;
114                 while (my ($key, $value) = each (%data)) {
115                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
116                                 or $key eq 'maintainer';
117                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
118                         $data .= "$key\00$value\00";
119                 }
120                 $data =~ s/.$//s;
121                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
122                         = $data;
123         }
124
125         untie %packages_all_db;
126     }
127 }
128
129 print "Writing databases...\n";
130 my %packages_small_db;
131 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
132         O_RDWR|O_CREAT, 0666, $DB_BTREE
133         or die "Error creating DB: $!";
134 while (my ($k, $v) = each(%packages_small)) {
135         $v =~ s/.$//s;
136         $packages_small_db{$k} = $v;
137 }
138 untie %packages_small_db;
139
140 my %sources_packages_db;
141 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
142         O_RDWR|O_CREAT, 0666, $DB_BTREE
143         or die "Error creating DB: $!";
144 while (my ($k, $v) = each(%sources_packages)) {
145         $v =~ s/.$//s;
146         $sources_packages_db{$k} = $v;
147 }
148 untie %sources_packages_db;
149
150 my %packages_descriptions_db;
151 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
152         O_RDWR|O_CREAT, 0666, $DB_BTREE
153         or die "Error creating DB: $!";
154 while (my ($k, $v) = each(%packages_descriptions)) {
155         $packages_descriptions_db{$k} = $v;
156 }
157 untie %packages_descriptions_db;
158
159 my %descriptions_packages_db;
160 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
161         O_RDWR|O_CREAT, 0666, $DB_BTREE
162         or die "Error creating DB: $!";
163 while (my ($k, $v) = each(%descriptions_packages)) {
164         $v =~ s/.$//s;
165         $descriptions_packages_db{$k} = $v;
166 }
167 untie %descriptions_packages_db;
168
169 my %descriptions_db;
170 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
171         O_RDWR|O_CREAT, 0666, $DB_BTREE
172         or die "Error creating DB: $!";
173 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
174 for (my $i=1; $i<= $#descriptions; $i++) {
175         my $plain_description = $descriptions[$i];
176         $plain_description =~ s/\n .\n/ /og;
177         $plain_description =~ s/[\n \t]+/ /og;
178         print DESCR "$plain_description\n";
179         $descriptions_db{$i} = $descriptions[$i];
180 }
181 close DESCR;
182 untie %descriptions_db;
183
184 # package names stuff:
185 for my $pkg (keys %package_names) {
186         for (my $i=0;$i<length($pkg)-1;$i++) {
187                 my $before = substr($pkg, 0, $i);
188                 my $after = substr($pkg, $i);
189                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
190                 $package_postfixes{$after} .= "$before\0";
191         }
192 }
193 my %package_postfixes_db;
194 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
195         O_RDWR|O_CREAT, 0666, $DB_BTREE
196         or die "Error creating DB: $!";
197 while (my ($k, $v) = each(%package_postfixes)) {
198         $v =~ s/.$//s;
199         my $nr = $v;
200         $nr =~ s/[^\000]//g;
201         $nr = length($nr) + 1; # < number of hits
202         if ($nr > $MAX_PACKAGE_POSTFIXES) {
203                 $v = "\001" . $nr;
204         }
205         $package_postfixes_db{$k} = $v;
206 }
207 untie %package_postfixes_db;
208
209 store \%sections, "$DBDIR/sections.info";
210 store \%subsections, "$DBDIR/subsections.info";
211 store \%priorities, "$DBDIR/priorities.info";
212
213 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
214 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
215 for my $suite (@SUITES) {
216         rename("$DBDIR/packages_all_$suite.db.new",
217                "$DBDIR/packages_all_$suite.db");
218 }
219 rename("$DBDIR/packages_descriptions.db.new",
220        "$DBDIR/packages_descriptions.db");
221 rename("$DBDIR/descriptions_packages.db.new",
222        "$DBDIR/descriptions_packages.db");
223 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
224 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
225 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");