]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
238a47b9ec9f5ffe50c751afe05e81c2e8d392fc
[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_version = $1
78                             if $src =~ s/\s+\((.*)\).*//; # strip version info
79                 }
80                 $data{'source'} = $src;
81                 $data{'source-version'} = $src_version if $src_version;
82                 my $descr = $data{'description'};
83                 my $did = undef;
84                 if (exists($descriptions{$descr})) {
85                         $did  = $descriptions{$descr};
86                 } else {
87                         $did = 1 + $#descriptions;
88                         $descriptions[$did] = $descr;
89                         $descriptions{$descr} = $did;
90                 }
91                 $data{'description'} = $did;
92                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
93                 $descriptions_packages{$did} .=
94                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
95
96                 my $sdescr = $descr;
97                 $sdescr =~ s/\n.*//s;
98                 my $section = 'main';
99                 my $subsection = $data{section} || '-';
100                 if ($data{section} && ($data{section} =~ m=/=o)) {
101                     ($section, $subsection) = split m=/=o, $data{section}, 2;
102                     ($subsection, $section) = split m=/=o, $data{section}, 2
103                         if $section eq 'non-US';
104                 }
105                 $data{'section'} = $section;
106                 $data{'subsection'} = $subsection;
107                 $data{'priority'} ||= '-';
108                 $sections{$suite}{$archive}{$section}++;
109                 $subsections{$suite}{$archive}{$subsection}++;
110                 $priorities{$suite}{$archive}{$data{priority}}++;
111                 $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
112                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
113                 $sources_packages{$src} .=
114                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
115                 $data{archive} = $archive;
116                 while (my ($key, $value) = each (%data)) {
117                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
118                                 or $key eq 'maintainer';
119                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
120                         $data .= "$key: $value\n";
121                 }
122                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
123                         = $data;
124         }
125
126         untie %packages_all_db;
127     }
128 }
129
130 print "Writing databases...\n";
131 my %packages_small_db;
132 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
133         O_RDWR|O_CREAT, 0666, $DB_BTREE
134         or die "Error creating DB: $!";
135 while (my ($k, $v) = each(%packages_small)) {
136         $v =~ s/.$//s;
137         $packages_small_db{$k} = $v;
138 }
139 untie %packages_small_db;
140
141 my %sources_packages_db;
142 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
143         O_RDWR|O_CREAT, 0666, $DB_BTREE
144         or die "Error creating DB: $!";
145 while (my ($k, $v) = each(%sources_packages)) {
146         $v =~ s/.$//s;
147         $sources_packages_db{$k} = $v;
148 }
149 untie %sources_packages_db;
150
151 my %packages_descriptions_db;
152 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
153         O_RDWR|O_CREAT, 0666, $DB_BTREE
154         or die "Error creating DB: $!";
155 while (my ($k, $v) = each(%packages_descriptions)) {
156         $packages_descriptions_db{$k} = $v;
157 }
158 untie %packages_descriptions_db;
159
160 my %descriptions_packages_db;
161 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
162         O_RDWR|O_CREAT, 0666, $DB_BTREE
163         or die "Error creating DB: $!";
164 while (my ($k, $v) = each(%descriptions_packages)) {
165         $v =~ s/.$//s;
166         $descriptions_packages_db{$k} = $v;
167 }
168 untie %descriptions_packages_db;
169
170 my %descriptions_db;
171 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
172         O_RDWR|O_CREAT, 0666, $DB_BTREE
173         or die "Error creating DB: $!";
174 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
175 for (my $i=1; $i<= $#descriptions; $i++) {
176         my $plain_description = $descriptions[$i];
177         $plain_description =~ s/\n .\n/ /og;
178         $plain_description =~ s/[\n \t]+/ /og;
179         print DESCR "$plain_description\n";
180         $descriptions_db{$i} = $descriptions[$i];
181 }
182 close DESCR;
183 untie %descriptions_db;
184
185 # package names stuff:
186 for my $pkg (keys %package_names) {
187         for (my $i=0;$i<length($pkg)-1;$i++) {
188                 my $before = substr($pkg, 0, $i);
189                 my $after = substr($pkg, $i);
190                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
191                 $package_postfixes{$after} .= "$before\0";
192         }
193 }
194 my %package_postfixes_db;
195 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
196         O_RDWR|O_CREAT, 0666, $DB_BTREE
197         or die "Error creating DB: $!";
198 while (my ($k, $v) = each(%package_postfixes)) {
199         $v =~ s/.$//s;
200         my $nr = $v;
201         $nr =~ s/[^\000]//g;
202         $nr = length($nr) + 1; # < number of hits
203         if ($nr > $MAX_PACKAGE_POSTFIXES) {
204                 $v = "\001" . $nr;
205         }
206         $package_postfixes_db{$k} = $v;
207 }
208 untie %package_postfixes_db;
209
210 store \%sections, "$DBDIR/sections.info";
211 store \%subsections, "$DBDIR/subsections.info";
212 store \%priorities, "$DBDIR/priorities.info";
213
214 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
215 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
216 for my $suite (@SUITES) {
217         rename("$DBDIR/packages_all_$suite.db.new",
218                "$DBDIR/packages_all_$suite.db");
219 }
220 rename("$DBDIR/packages_descriptions.db.new",
221        "$DBDIR/packages_descriptions.db");
222 rename("$DBDIR/descriptions_packages.db.new",
223        "$DBDIR/descriptions_packages.db");
224 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
225 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
226 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");