]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
Some fixes to archive handling
[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 $suite (@SUITES) {
54     my %package_names_suite = ();
55     my %packages_all_db;
56     tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
57             O_RDWR|O_CREAT, 0666, $DB_BTREE
58             or die "Error creating DB: $!";
59
60     for my $archive (@ARCHIVES) {
61         print "Reading $archive/$suite...\n";
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                         $packages_small{$_} ||= {};
82                     }
83                 }
84                 $package_names{$data{'package'}} = 1;
85                 $package_names_suite{$data{'package'}} = 1;
86                 my $src = $data{'package'};
87                 my $src_version = '';
88                 if ($data{'source'}) {
89                         $src = $data{'source'};
90                         $src =~ s/\s+.*//; # strip version info
91                 }
92                 $data{'source'} = $src;
93                 my $descr = $data{'description'};
94                 my $did = undef;
95                 if (exists($descriptions{$descr})) {
96                         $did  = $descriptions{$descr};
97                 } else {
98                         $did = 1 + $#descriptions;
99                         $descriptions[$did] = $descr;
100                         $descriptions{$descr} = $did;
101                 }
102                 $data{'description'} = $did;
103                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
104                 $descriptions_packages{$did} .=
105                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
106
107                 my $sdescr = $descr;
108                 $sdescr =~ s/\n.*//s;
109                 my $section = 'main';
110                 my $subsection = $data{section} || '-';
111                 if ($data{section} && ($data{section} =~ m=/=o)) {
112                     ($section, $subsection) = split m=/=o, $data{section}, 2;
113                     ($subsection, $section) = split m=/=o, $data{section}, 2
114                         if $section eq 'non-US';
115                 }
116                 $data{'section'} = $section;
117                 $data{'subsection'} = $subsection;
118                 $data{'priority'} ||= '-';
119                 $sections{$suite}{$section}++;
120                 $subsections{$suite}{$subsection}++;
121                 $priorities{$suite}{$data{priority}}++;
122                 my $pkgitem = "$archive $suite $data{'architecture'} ".
123                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\0";
124                 my $previtem = ($packages_small{$data{'package'}}{$suite}{$data{'architecture'}}
125                     ||= $pkgitem);
126                 $packages_small{$data{'package'}}{$suite}{$data{'architecture'}} = $pkgitem
127                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
128                 $previtem = ($packages_small{$data{'package'}}{$suite}{'any'}
129                     ||= $pkgitem);
130                 $packages_small{$data{'package'}}{$suite}{'any'} = $pkgitem
131                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
132                 $sources_packages{$src} .=
133                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
134                 $data{archive} = $archive;
135                 while (my ($key, $value) = each (%data)) {
136                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
137                                 or $key eq 'maintainer';
138                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
139                         $data .= "$key\00$value\00";
140                 }
141                 $data =~ s/.$//s;
142                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
143                         = $data;
144         }
145     }
146
147     open NAMES, '>', "$DBDIR/package_names_$suite.txt.new"
148         or die "Error creating package names list: $!";
149     foreach (sort keys %package_names_suite) {
150         print NAMES "$_\n";
151     }
152     close NAMES;
153
154     untie %packages_all_db;
155 }
156
157 print "Writing databases...\n";
158 my %packages_small_db;
159 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
160         O_RDWR|O_CREAT, 0666, $DB_BTREE
161         or die "Error creating DB: $!";
162 while (my ($pkg, $v) = each(%packages_small)) {
163         my ($res1, $res2, $res3) = ("", "", "");
164         while (my ($suite, $v2) = each %$v) {
165             $res2 .= $v2->{'any'};
166             while (my ($arch, $v3) = each %$v2) {
167                 next if $arch eq 'any' or $v3 eq $v2->{'any'};
168                 $res3 .= $v3;
169             }
170         }
171         
172         if (exists $virtual_packages{$pkg}) {
173                 while (my ($suite, $v2) = each %{$virtual_packages{$pkg}}) {
174                         $res1 .= "$suite\01".(join ' ', keys %$v2)."\01";
175                 }
176         } else {
177                 $res1 .= "-\01-\01";
178         }
179         $res1 =~ s/.$/\0/s;
180
181         my $res = "$res1$res2$res3";
182         $res =~ s/.$//s;
183         $packages_small_db{$pkg} = $res;
184 }
185 untie %packages_small_db;
186
187 my %sources_packages_db;
188 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
189         O_RDWR|O_CREAT, 0666, $DB_BTREE
190         or die "Error creating DB: $!";
191 while (my ($k, $v) = each(%sources_packages)) {
192         $v =~ s/.$//s;
193         $sources_packages_db{$k} = $v;
194 }
195 untie %sources_packages_db;
196
197 my %packages_descriptions_db;
198 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
199         O_RDWR|O_CREAT, 0666, $DB_BTREE
200         or die "Error creating DB: $!";
201 while (my ($k, $v) = each(%packages_descriptions)) {
202         $packages_descriptions_db{$k} = $v;
203 }
204 untie %packages_descriptions_db;
205
206 my %descriptions_packages_db;
207 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
208         O_RDWR|O_CREAT, 0666, $DB_BTREE
209         or die "Error creating DB: $!";
210 while (my ($k, $v) = each(%descriptions_packages)) {
211         $v =~ s/.$//s;
212         $descriptions_packages_db{$k} = $v;
213 }
214 untie %descriptions_packages_db;
215
216 my %descriptions_db;
217 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
218         O_RDWR|O_CREAT, 0666, $DB_BTREE
219         or die "Error creating DB: $!";
220 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
221 for (my $i=1; $i<= $#descriptions; $i++) {
222         my $plain_description = $descriptions[$i];
223 # WARNING: This needs to correspond with what happens in
224 # Packages/Search.pm:do_fulltext_search
225         $plain_description =~ tr [A-Z] [a-z];
226         # ensure one space on both ends
227         $plain_description = " $plain_description ";
228         $plain_description =~ s/[(),.-]+//og;
229         $plain_description =~ s#[^a-z0-9_/+]+# #og;
230         print DESCR "$plain_description\n";
231         $descriptions_db{$i} = $descriptions[$i];
232 }
233 close DESCR;
234 untie %descriptions_db;
235
236 # package names stuff:
237 for my $pkg (keys %package_names) {
238         for (my $i=0;$i<length($pkg)-1;$i++) {
239                 my $before = substr($pkg, 0, $i);
240                 my $after = substr($pkg, $i);
241                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
242                 $package_postfixes{$after} .= "$before\0";
243         }
244 }
245 my %package_postfixes_db;
246 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
247         O_RDWR|O_CREAT, 0666, $DB_BTREE
248         or die "Error creating DB: $!";
249 while (my ($k, $v) = each(%package_postfixes)) {
250         $v =~ s/.$//s;
251         my $nr = $v;
252         $nr =~ s/[^\000]//g;
253         $nr = length($nr) + 1; # < number of hits
254         if ($nr > $MAX_PACKAGE_POSTFIXES) {
255                 $v = "\001" . $nr;
256         }
257         $package_postfixes_db{$k} = $v;
258 }
259 untie %package_postfixes_db;
260
261 store \%sections, "$DBDIR/sections.info";
262 store \%subsections, "$DBDIR/subsections.info";
263 store \%priorities, "$DBDIR/priorities.info";
264
265 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
266 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
267 for my $suite (@SUITES) {
268         rename("$DBDIR/packages_all_$suite.db.new",
269                "$DBDIR/packages_all_$suite.db");
270         rename("$DBDIR/package_names_$suite.txt.new",
271                "$DBDIR/package_names_$suite.txt");
272 }
273 rename("$DBDIR/packages_descriptions.db.new",
274        "$DBDIR/packages_descriptions.db");
275 rename("$DBDIR/descriptions_packages.db.new",
276        "$DBDIR/descriptions_packages.db");
277 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
278 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
279 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");