]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
In packages_all_$suite, store only description Id, not full description, nor
[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
24 my $what = $ARGV[0] ? "non-free" : "*";
25 # max. distinct results for a given package postfix
26 my $MAX_PACKAGE_POSTFIXES = 100;
27
28 use DB_File;
29 my %packages_small = ();
30 my %package_names = ();
31 my %package_postfixes = ();
32 my %sources_packages = ();
33 my %descriptions = ();
34 my @descriptions = ("we count lines one-based\000");
35 my %packages_descriptions = ();
36 my %descriptions_packages = ();
37
38 my @suites = ('oldstable', 'stable', 'testing', 'unstable', 'experimental');
39
40 $/ = "";
41
42 for my $suite (@suites) {
43
44         print "Reading $suite...\n";
45         my %packages_all_db;
46         tie %packages_all_db, "DB_File", "packages_all_$suite.db.new",
47                 O_RDWR|O_CREAT, 0666, $DB_BTREE
48                 or die "Error creating DB: $!";
49         open PKG, "zcat /org/ftp.debian.org/ftp/dists/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
50         while (<PKG>) {
51                 next if /^\s*$/;
52                 my $data = "";
53                 my %data = ();
54                 chomp;
55                 s/\n /\377/g;
56                 while (/^(\S+):\s*(.*)\s*$/mg) {
57                         my ($key, $value) = ($1, $2);
58                         $value =~ s/\377/\n /g;
59                         $key =~ tr [A-Z] [a-z];
60                         $data{$key} = $value;
61                 }
62                 # Skip double package
63                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
64
65                 $package_names{$data{'package'}} = 1;
66                 my $src = $data{'package'};
67                 if ($data{'source'}) {
68                         $src = $data{'source'};
69                         $src =~ s/ .*//; # strip version info
70                 }
71                 $data{'source'} = $src;
72                 my $descr = $data{'description'};
73                 my $did = undef;
74                 if (exists($descriptions{$descr})) {
75                         $did  = $descriptions{$descr};
76                 } else {
77                         $did = 1 + $#descriptions;
78                         $descriptions[$did] = $descr;
79                         $descriptions{$descr} = $did;
80                 }
81                 $data{'description'} = $did;
82                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
83                 $descriptions_packages{$did} .=
84                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
85
86                 my $sdescr = $descr;
87                 $sdescr =~ s/\n.*//s;
88                 my $section = 'main';
89                 my $subsection = $data{section};
90                 if ($data{section} && ($data{section} =~ m=/=o)) {
91                     ($section, $subsection) = split m=/=o, $data{section}, 2;
92                 }
93                 $data{'section'} = $section;
94                 $data{'subsection'} = $subsection;
95                 $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ".
96                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
97                 $sources_packages{$src} .=
98                         "$suite $data{'package'} $data{'version'} $data{'architecture'}\000";
99                 while (my ($key, $value) = each (%data)) {
100                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
101                                 or $key eq 'maintainer';
102                         $data .= "$key: $value\n";
103                 }
104                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
105                         = $data;
106         }
107
108         untie %packages_all_db;
109 }
110
111 print "Writing databases...\n";
112 my %packages_small_db;
113 tie %packages_small_db, "DB_File", "packages_small.db.new",
114         O_RDWR|O_CREAT, 0666, $DB_BTREE
115         or die "Error creating DB: $!";
116 while (my ($k, $v) = each(%packages_small)) {
117         $v =~ s/.$//s;
118         $packages_small_db{$k} = $v;
119 }
120 untie %packages_small_db;
121
122 my %sources_packages_db;
123 tie %sources_packages_db, "DB_File", "sources_packages.db.new",
124         O_RDWR|O_CREAT, 0666, $DB_BTREE
125         or die "Error creating DB: $!";
126 while (my ($k, $v) = each(%sources_packages)) {
127         $v =~ s/.$//s;
128         $sources_packages_db{$k} = $v;
129 }
130 untie %sources_packages_db;
131
132 my %packages_descriptions_db;
133 tie %packages_descriptions_db, "DB_File", "packages_descriptions.db.new",
134         O_RDWR|O_CREAT, 0666, $DB_BTREE
135         or die "Error creating DB: $!";
136 while (my ($k, $v) = each(%packages_descriptions)) {
137         $packages_descriptions_db{$k} = $v;
138 }
139 untie %packages_descriptions_db;
140
141 my %descriptions_packages_db;
142 tie %descriptions_packages_db, "DB_File", "descriptions_packages.db.new",
143         O_RDWR|O_CREAT, 0666, $DB_BTREE
144         or die "Error creating DB: $!";
145 while (my ($k, $v) = each(%descriptions_packages)) {
146         $v =~ s/.$//s;
147         $descriptions_packages_db{$k} = $v;
148 }
149 untie %descriptions_packages_db;
150
151 my %descriptions_db;
152 tie %descriptions_db, "DB_File", "descriptions.db.new",
153         O_RDWR|O_CREAT, 0666, $DB_BTREE
154         or die "Error creating DB: $!";
155 open DESCR, "> descriptions.txt" or die "Error creating descriptions textfile";
156 for (my $i=1; $i<= $#descriptions; $i++) {
157         my $null_d = $descriptions[$i];
158         $null_d =~ s/\n/\000/g;
159         print DESCR "$null_d\n";
160         $descriptions_db{$i} = $descriptions[$i];
161 }
162 close DESCR;
163 untie %descriptions_db;
164
165 # package names stuff:
166 for my $pkg (keys %package_names) {
167         for (my $i=0;$i<length($pkg)-1;$i++) {
168                 my $before = substr($pkg, 0, $i);
169                 my $after = substr($pkg, $i);
170                 $package_postfixes{$after} .= "$before\0";
171         }
172 }
173 my %package_postfixes_db;
174 tie %package_postfixes_db, "DB_File", "package_postfixes.db.new",
175         O_RDWR|O_CREAT, 0666, $DB_BTREE
176         or die "Error creating DB: $!";
177 while (my ($k, $v) = each(%package_postfixes)) {
178         $v =~ s/.$//s;
179         my $nr = $v;
180         $nr =~ s/[^\000]//g;
181         $nr = length($nr) + 1; # < number of hits
182         if ($nr > $MAX_PACKAGE_POSTFIXES) {
183                 $v = "\001" . $nr;
184         }
185         $package_postfixes_db{$k} = $v;
186 }
187 untie %package_postfixes_db;
188
189 rename("packages_small.db.new", "packages_small.db");
190 rename("sources_packages.db.new", "sources_packages.db");
191 for my $suite (@suites) {
192         rename("packages_all_$suite.db.new", "packages_all_$suite.db");
193 }
194 rename("packages_descriptions.db.new", "packages_descriptions.db");
195 rename("descriptions_packages.db.new", "descriptions_packages.db");
196 rename("descriptions.txt.new", "descriptions.txt");
197 rename("descriptions.db.new", "descriptions.db");
198 rename("package_postfixes.db.new", "package_postfixes.db");