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