]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
aad66889f510bb97bb5825ad80b9a5bd697969a5
[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 $data = "";
53                 my %data = ();
54                 chomp;
55                 s/\n /\377/g;
56                 #s/\376\377\s*\376\377/\376\377/og;
57                 while (/^(\S+):\s*(.*)\s*$/mg) {
58                         my ($key, $value) = ($1, $2);
59                         $value =~ s/\377/\n /g;
60                         $data .= "$key: $value\n";
61                         $key =~ tr [A-Z] [a-z];
62                         $data{$key} = $value;
63                 }
64                 # Skip double package
65                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
66                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
67                         = $data;
68
69                 $package_names{$data{'package'}} = 1;
70                 my $src = $data{'package'};
71                 if ($data{'source'}) {
72                         $src = $data{'source'};
73                         $src =~ s/ .*//; # strip version info
74                 }
75                 my $descr = $data{'description'};
76                 my $did = undef;
77                 if (exists($descriptions{$descr})) {
78                         $did  = $descriptions{$descr};
79                 } else {
80                         $did = 1 + $#descriptions;
81                         $descriptions[$did] = $descr;
82                         $descriptions{$descr} = $did;
83                 }
84                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
85                 $descriptions_packages{$did} .=
86                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
87
88                 my $sdescr = $descr;
89                 $sdescr =~ s/\n.*//s;
90                 my $section = 'main';
91                 my $subsection = $data{section};
92                 if ($data{section} && ($data{section} =~ m=/=o)) {
93                     ($section, $subsection) = split m=/=o, $data{section}, 2;
94                 }
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         }
100
101         untie %packages_all_db;
102 }
103
104 print "Writing databases...\n";
105 my %packages_small_db;
106 tie %packages_small_db, "DB_File", "packages_small.db.new",
107         O_RDWR|O_CREAT, 0666, $DB_BTREE
108         or die "Error creating DB: $!";
109 while (my ($k, $v) = each(%packages_small)) {
110         $v =~ s/.$//s;
111         $packages_small_db{$k} = $v;
112 }
113 untie %packages_small_db;
114
115 my %sources_packages_db;
116 tie %sources_packages_db, "DB_File", "sources_packages.db.new",
117         O_RDWR|O_CREAT, 0666, $DB_BTREE
118         or die "Error creating DB: $!";
119 while (my ($k, $v) = each(%sources_packages)) {
120         $v =~ s/.$//s;
121         $sources_packages_db{$k} = $v;
122 }
123 untie %sources_packages_db;
124
125 my %packages_descriptions_db;
126 tie %packages_descriptions_db, "DB_File", "packages_descriptions.db.new",
127         O_RDWR|O_CREAT, 0666, $DB_BTREE
128         or die "Error creating DB: $!";
129 while (my ($k, $v) = each(%packages_descriptions)) {
130         $packages_descriptions_db{$k} = $v;
131 }
132 untie %packages_descriptions_db;
133
134 my %descriptions_packages_db;
135 tie %descriptions_packages_db, "DB_File", "descriptions_packages.db.new",
136         O_RDWR|O_CREAT, 0666, $DB_BTREE
137         or die "Error creating DB: $!";
138 while (my ($k, $v) = each(%descriptions_packages)) {
139         $v =~ s/.$//s;
140         $descriptions_packages_db{$k} = $v;
141 }
142 untie %descriptions_packages_db;
143
144 my %descriptions_db;
145 tie %descriptions_db, "DB_File", "descriptions.db.new",
146         O_RDWR|O_CREAT, 0666, $DB_BTREE
147         or die "Error creating DB: $!";
148 open DESCR, "> descriptions.txt" or die "Error creating descriptions textfile";
149 for (my $i=1; $i<= $#descriptions; $i++) {
150         my $null_d = $descriptions[$i];
151         $null_d =~ s/\n/\000/g;
152         print DESCR "$null_d\n";
153         $descriptions_db{$i} = $descriptions[$i];
154 }
155 close DESCR;
156 untie %descriptions_db;
157
158 # package names stuff:
159 for my $pkg (keys %package_names) {
160         for (my $i=0;$i<length($pkg)-1;$i++) {
161                 my $before = substr($pkg, 0, $i);
162                 my $after = substr($pkg, $i);
163                 $package_postfixes{$after} .= "$before\0";
164         }
165 }
166 my %package_postfixes_db;
167 tie %package_postfixes_db, "DB_File", "package_postfixes.db.new",
168         O_RDWR|O_CREAT, 0666, $DB_BTREE
169         or die "Error creating DB: $!";
170 while (my ($k, $v) = each(%package_postfixes)) {
171         $v =~ s/.$//s;
172         my $nr = $v;
173         $nr =~ s/[^\000]//g;
174         $nr = length($nr) + 1; # < number of hits
175         if ($nr > $MAX_PACKAGE_POSTFIXES) {
176                 $v = "\001" . $nr;
177         }
178         $package_postfixes_db{$k} = $v;
179 }
180 untie %package_postfixes_db;
181
182 rename("packages_small.db.new", "packages_small.db");
183 rename("sources_packages.db.new", "sources_packages.db");
184 for my $suite (@suites) {
185         rename("packages_all_$suite.db.new", "packages_all_$suite.db");
186 }
187 rename("packages_descriptions.db.new", "packages_descriptions.db");
188 rename("descriptions_packages.db.new", "descriptions_packages.db");
189 rename("descriptions.txt.new", "descriptions.txt");
190 rename("descriptions.db.new", "descriptions.db");
191 rename("package_postfixes.db.new", "package_postfixes.db");