]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
fe61ca2269afc5b9f5e29104d7eb42f6ab713b3c
[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 File::Path;
35 use Deb::Versions;
36 use Lingua::Stem v0.82;
37 use Search::Xapian;
38 use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES );
39 &Packages::Config::init( './' );
40 my %packages_small = ();
41 my %virtual_packages = ();
42 my %package_names = ();
43 my %package_postfixes = ();
44 my %sources_packages = ();
45 my %descriptions = ();
46 my @descriptions = ("we count lines one-based\000");
47 my %packages_descriptions = ();
48 my %descriptions_packages = ();
49
50 my %sections = ();
51 my %subsections = ();
52 my %priorities = ();
53
54 $/ = "";
55
56 -d $DBDIR || mkpath( $DBDIR );
57 -d "$DBDIR/xapian.new" && rmtree("$DBDIR/xapian.new");
58 -d "$DBDIR/xapian.old" && rmtree("$DBDIR/xapian.old");
59 mkpath( "$DBDIR/xapian.new" );
60
61 for my $suite (@SUITES) {
62     my %package_names_suite = ();
63     my %packages_all_db;
64     tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
65             O_RDWR|O_CREAT, 0666, $DB_BTREE
66             or die "Error creating DB: $!";
67
68     for my $archive (@ARCHIVES) {
69         print "Reading $archive/$suite...\n";
70         if (!-d "$TOPDIR/archive/$archive/$suite/") {
71                 print "\tseems not to exist, skipping...\n";
72                 next;
73         }
74         open PKG, "zcat $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
75         while (<PKG>) {
76                 next if /^\s*$/;
77                 my $data = "";
78                 my %data = ();
79                 chomp;
80                 s/\n /\377/g;
81                 while (/^(\S+):\s*(.*)\s*$/mg) {
82                         my ($key, $value) = ($1, $2);
83                         $value =~ s/\377/\n /g;
84                         $key =~ tr [A-Z] [a-z];
85                         $data{$key} = $value;
86                 }
87                 # Skip double package
88                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
89                 # Skip arch:all for amd64 & gnuab, any non-redundancy is
90                 # usually a bug anyway
91                 next if ($archive eq 'amd64' or $archive eq 'gnuab')
92                     and $data{architecture} eq 'all';
93
94                 if ($data{'provides'}) {
95                     foreach (split /\s*,\s*/, $data{'provides'}) {
96                         $virtual_packages{$_}{$suite}{$data{'package'}}++;
97                         $packages_small{$_} ||= {};
98                     }
99                 }
100                 $package_names{$data{'package'}} = 1;
101                 $package_names_suite{$data{'package'}} = 1;
102                 my $src = $data{'package'};
103                 my $src_version = '';
104                 if ($data{'source'}) {
105                         $src = $data{'source'};
106                         $src =~ s/\s+.*//o; # strip version info
107                 }
108                 $data{'source'} = $src;
109                 # we add some additional data here
110                 my $descr = "$data{'description'}\000$data{'package'}\000"
111                     .($data{'tag'}||'');
112                 my $sdescr = $data{'description'};
113                 $sdescr =~ s/\n.*//os;
114                 my $did = undef;
115                 if (exists($descriptions{$descr})) {
116                         $did  = $descriptions{$descr};
117                 } else {
118                         $did = 1 + $#descriptions;
119                         $descriptions[$did] = $descr;
120                         $descriptions{$descr} = $did;
121                 }
122                 $data{'description'} = $did;
123                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
124                 $descriptions_packages{$did} .=
125                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
126
127                 my $section = 'main';
128                 my $subsection = $data{section} || '-';
129                 if ($data{section} && ($data{section} =~ m=/=o)) {
130                     ($section, $subsection) = split m=/=o, $data{section}, 2;
131                     ($subsection, $section) = split m=/=o, $data{section}, 2
132                         if $section eq 'non-US';
133                 }
134                 $data{'section'} = $section;
135                 $data{'subsection'} = $subsection;
136                 $data{'priority'} ||= '-';
137                 $sections{$suite}{$section}++;
138                 $subsections{$suite}{$subsection}++;
139                 $priorities{$suite}{$data{priority}}++;
140                 my $pkgitem = "$archive $suite $data{'architecture'} ".
141                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\0";
142                 my $previtem = ($packages_small{$data{'package'}}{$suite}{$data{'architecture'}}
143                     ||= $pkgitem);
144                 $packages_small{$data{'package'}}{$suite}{$data{'architecture'}} = $pkgitem
145                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
146                 $previtem = ($packages_small{$data{'package'}}{$suite}{'any'}
147                     ||= $pkgitem);
148                 $packages_small{$data{'package'}}{$suite}{'any'} = $pkgitem
149                     if version_cmp($data{'version'}, (split /\s/o, $previtem)[6]) > 0;
150                 $sources_packages{$src} .=
151                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
152                 $data{archive} = $archive;
153                 while (my ($key, $value) = each (%data)) {
154                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
155                                 or $key eq 'maintainer';
156                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
157                         $data .= "$key\00$value\00";
158                 }
159                 $data =~ s/.$//s;
160                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
161                         = $data;
162         }
163     }
164
165     open NAMES, '>', "$DBDIR/package_names_$suite.txt.new"
166         or die "Error creating package names list: $!";
167     foreach (sort keys %package_names_suite) {
168         print NAMES "$_\n";
169     }
170     close NAMES;
171
172     untie %packages_all_db;
173 }
174
175 print "Writing databases...\n";
176 my %packages_small_db;
177 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
178         O_RDWR|O_CREAT, 0666, $DB_BTREE
179         or die "Error creating DB: $!";
180 while (my ($pkg, $v) = each(%packages_small)) {
181         my ($res1, $res2, $res3) = ("", "", "");
182         while (my ($suite, $v2) = each %$v) {
183             $res2 .= $v2->{'any'};
184             while (my ($arch, $v3) = each %$v2) {
185                 next if $arch eq 'any' or $v3 eq $v2->{'any'};
186                 $res3 .= $v3;
187             }
188         }
189
190         if (exists $virtual_packages{$pkg}) {
191                 while (my ($suite, $v2) = each %{$virtual_packages{$pkg}}) {
192                         $res1 .= "$suite\01".(join ' ', keys %$v2)."\01";
193                 }
194         } else {
195                 $res1 .= "-\01-\01";
196         }
197         $res1 =~ s/.$/\0/s;
198
199         my $res = "$res1$res2$res3";
200         $res =~ s/.$//s;
201         $packages_small_db{$pkg} = $res;
202 }
203 untie %packages_small_db;
204
205 my %sources_packages_db;
206 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
207         O_RDWR|O_CREAT, 0666, $DB_BTREE
208         or die "Error creating DB: $!";
209 while (my ($k, $v) = each(%sources_packages)) {
210         $v =~ s/.$//s;
211         $sources_packages_db{$k} = $v;
212 }
213 untie %sources_packages_db;
214
215 my %packages_descriptions_db;
216 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
217         O_RDWR|O_CREAT, 0666, $DB_BTREE
218         or die "Error creating DB: $!";
219 while (my ($k, $v) = each(%packages_descriptions)) {
220         $packages_descriptions_db{$k} = $v;
221 }
222 untie %packages_descriptions_db;
223
224 my %descriptions_packages_db;
225 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
226         O_RDWR|O_CREAT, 0666, $DB_BTREE
227         or die "Error creating DB: $!";
228 while (my ($k, $v) = each(%descriptions_packages)) {
229         $v =~ s/.$//s;
230         $descriptions_packages_db{$k} = $v;
231 }
232 untie %descriptions_packages_db;
233
234 my $stemmer = Lingua::Stem->new();
235 $stemmer->stem_caching({ -level => 2 });
236 my $xapian_db;
237 eval {
238     $xapian_db = Search::Xapian::WritableDatabase->new("$DBDIR/xapian.new",
239                                                        Search::Xapian::DB_CREATE_OR_OPEN)
240         or die "can't create write-able db object: $!\n";
241 };
242 die $@ if $@;
243 my %descriptions_db;
244 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
245         O_RDWR|O_CREAT, 0666, $DB_BTREE
246         or die "Error creating DB: $!";
247 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
248 print "Index $#descriptions descriptions\n";
249 for (my $i=1; $i<= $#descriptions; $i++) {
250         my $plain_description = $descriptions[$i];
251         # strip away additional data
252         my ($only_desc) = split /\000/o, $plain_description, 2;
253 # WARNING: This needs to correspond with what happens in
254 # Packages/Search.pm:do_fulltext_search
255         $plain_description =~ tr [A-Z] [a-z];
256         # ensure one space on both ends
257         $plain_description = " $plain_description ";
258         $plain_description =~ s/[(),.-]+//og;
259         $plain_description =~ s#[^a-z0-9_/+]+# #og;
260         print DESCR "$plain_description\n";
261
262         #XAPIAN
263         eval {
264             my @words = split /\s+/, $plain_description;
265             my $stem_words = $stemmer->stem( \@words );
266             my $doc = Search::Xapian::Document->new()
267                 or die "can't create doc object for $i: $!\n";
268             if ($doc->set_data($i)){
269                 warn "can't set_data in doc object for $i: $!\n";
270             }
271             for my $j (0 .. (@$stem_words-1)) {
272                 next if $stem_words->[$j] =~ /^\s*$/o;
273                 if ($doc->add_posting($stem_words->[$j], $j)) {
274                     warn "can't add word $stem_words->[$j] $j: $!\n";
275                 }
276             }
277             $xapian_db->add_document($doc)
278                 or warn "failed to add document: $i\n";
279         };
280         die $@ if $@;
281
282         $descriptions_db{$i} = $only_desc;
283 }
284 close DESCR;
285 untie %descriptions_db;
286 $xapian_db->flush;
287 undef $xapian_db;
288
289 # package names stuff:
290 for my $pkg (keys %package_names) {
291         for (my $i=0;$i<length($pkg)-1;$i++) {
292                 my $before = substr($pkg, 0, $i);
293                 my $after = substr($pkg, $i);
294                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
295                 $package_postfixes{$after} .= "$before\0";
296         }
297 }
298 my %package_postfixes_db;
299 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
300         O_RDWR|O_CREAT, 0666, $DB_BTREE
301         or die "Error creating DB: $!";
302 while (my ($k, $v) = each(%package_postfixes)) {
303         $v =~ s/.$//s;
304         my $nr = $v;
305         $nr =~ s/[^\000]//g;
306         $nr = length($nr) + 1; # < number of hits
307         if ($nr > $MAX_PACKAGE_POSTFIXES) {
308                 $v = "\001" . $nr;
309         }
310         $package_postfixes_db{$k} = $v;
311 }
312 untie %package_postfixes_db;
313
314 store \%sections, "$DBDIR/sections.info";
315 store \%subsections, "$DBDIR/subsections.info";
316 store \%priorities, "$DBDIR/priorities.info";
317
318 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
319 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
320 for my $suite (@SUITES) {
321         rename("$DBDIR/packages_all_$suite.db.new",
322                "$DBDIR/packages_all_$suite.db");
323         rename("$DBDIR/package_names_$suite.txt.new",
324                "$DBDIR/package_names_$suite.txt");
325 }
326 rename("$DBDIR/xapian", "$DBDIR/xapian.old");
327 rename("$DBDIR/xapian.new","$DBDIR/xapian");
328 rmtree("$DBDIR/xapian.old");
329 rename("$DBDIR/packages_descriptions.db.new",
330        "$DBDIR/packages_descriptions.db");
331 rename("$DBDIR/descriptions_packages.db.new",
332        "$DBDIR/descriptions_packages.db");
333 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
334 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
335 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");