]> git.deb.at Git - deb/packages.git/blob - bin/parse-packages
Optimize and improve full description search by stripping away insignificant
[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         my %virtual_packages = ();
52         print "Reading $archive/$suite...\n";
53         my %packages_all_db;
54         tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new",
55                 O_RDWR|O_CREAT, 0666, $DB_BTREE
56                 or die "Error creating DB: $!";
57         open PKG, "zcat $TOPDIR/archive/$archive/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
58         while (<PKG>) {
59                 next if /^\s*$/;
60                 my $data = "";
61                 my %data = ();
62                 chomp;
63                 s/\n /\377/g;
64                 while (/^(\S+):\s*(.*)\s*$/mg) {
65                         my ($key, $value) = ($1, $2);
66                         $value =~ s/\377/\n /g;
67                         $key =~ tr [A-Z] [a-z];
68                         $data{$key} = $value;
69                 }
70                 # Skip double package
71                 next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
72
73                 if ($data{'provides'}) {
74                     foreach (split /\s*,\s*/, $data{'provides'}) {
75                         $virtual_packages{$_}{$data{'package'}}++;
76                     }
77                 }
78                 $package_names{$data{'package'}} = 1;
79                 my $src = $data{'package'};
80                 my $src_version = '';
81                 if ($data{'source'}) {
82                         $src = $data{'source'};
83                         $src =~ s/\s+.*//; # strip version info
84                 }
85                 $data{'source'} = $src;
86                 my $descr = $data{'description'};
87                 my $did = undef;
88                 if (exists($descriptions{$descr})) {
89                         $did  = $descriptions{$descr};
90                 } else {
91                         $did = 1 + $#descriptions;
92                         $descriptions[$did] = $descr;
93                         $descriptions{$descr} = $did;
94                 }
95                 $data{'description'} = $did;
96                 $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
97                 $descriptions_packages{$did} .=
98                         "$data{'package'} $data{'version'} $data{'architecture'}\000";
99
100                 my $sdescr = $descr;
101                 $sdescr =~ s/\n.*//s;
102                 my $section = 'main';
103                 my $subsection = $data{section} || '-';
104                 if ($data{section} && ($data{section} =~ m=/=o)) {
105                     ($section, $subsection) = split m=/=o, $data{section}, 2;
106                     ($subsection, $section) = split m=/=o, $data{section}, 2
107                         if $section eq 'non-US';
108                 }
109                 $data{'section'} = $section;
110                 $data{'subsection'} = $subsection;
111                 $data{'priority'} ||= '-';
112                 $sections{$suite}{$archive}{$section}++;
113                 $subsections{$suite}{$archive}{$subsection}++;
114                 $priorities{$suite}{$archive}{$data{priority}}++;
115                 $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ".
116                         "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
117                 $sources_packages{$src} .=
118                         "$archive $suite $data{'package'} $data{'version'} $data{'architecture'}\000";
119                 $data{archive} = $archive;
120                 while (my ($key, $value) = each (%data)) {
121                         next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
122                                 or $key eq 'maintainer';
123                         print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value;
124                         $data .= "$key\00$value\00";
125                 }
126                 $data =~ s/.$//s;
127                 $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
128                         = $data;
129         }
130
131         foreach (keys %virtual_packages) {
132             my @provided_by = keys %{$virtual_packages{$_}};
133             $packages_small{$_} .= "$archive $suite virtual v v v v @provided_by\000";
134         }
135
136         untie %packages_all_db;
137     }
138 }
139
140 print "Writing databases...\n";
141 my %packages_small_db;
142 tie %packages_small_db, "DB_File", "$DBDIR/packages_small.db.new",
143         O_RDWR|O_CREAT, 0666, $DB_BTREE
144         or die "Error creating DB: $!";
145 while (my ($k, $v) = each(%packages_small)) {
146         $v =~ s/.$//s;
147         $packages_small_db{$k} = $v;
148 }
149 untie %packages_small_db;
150
151 my %sources_packages_db;
152 tie %sources_packages_db, "DB_File", "$DBDIR/sources_packages.db.new",
153         O_RDWR|O_CREAT, 0666, $DB_BTREE
154         or die "Error creating DB: $!";
155 while (my ($k, $v) = each(%sources_packages)) {
156         $v =~ s/.$//s;
157         $sources_packages_db{$k} = $v;
158 }
159 untie %sources_packages_db;
160
161 my %packages_descriptions_db;
162 tie %packages_descriptions_db, "DB_File", "$DBDIR/packages_descriptions.db.new",
163         O_RDWR|O_CREAT, 0666, $DB_BTREE
164         or die "Error creating DB: $!";
165 while (my ($k, $v) = each(%packages_descriptions)) {
166         $packages_descriptions_db{$k} = $v;
167 }
168 untie %packages_descriptions_db;
169
170 my %descriptions_packages_db;
171 tie %descriptions_packages_db, "DB_File", "$DBDIR/descriptions_packages.db.new",
172         O_RDWR|O_CREAT, 0666, $DB_BTREE
173         or die "Error creating DB: $!";
174 while (my ($k, $v) = each(%descriptions_packages)) {
175         $v =~ s/.$//s;
176         $descriptions_packages_db{$k} = $v;
177 }
178 untie %descriptions_packages_db;
179
180 my %descriptions_db;
181 tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
182         O_RDWR|O_CREAT, 0666, $DB_BTREE
183         or die "Error creating DB: $!";
184 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
185 for (my $i=1; $i<= $#descriptions; $i++) {
186         my $plain_description = $descriptions[$i];
187 # WARNING: This needs to correspond with what happens in
188 # Packages/Search.pm:do_fulltext_search
189         $plain_description =~ tr [A-Z] [a-z];
190         # ensure one space on both ends
191         $plain_description = " $plain_description ";
192         $plain_description =~ s/[(),.-]+//og;
193         $plain_description =~ s#[^a-z0-9_/+]+# #og;
194         print DESCR "$plain_description\n";
195         $descriptions_db{$i} = $descriptions[$i];
196 }
197 close DESCR;
198 untie %descriptions_db;
199
200 # package names stuff:
201 for my $pkg (keys %package_names) {
202         for (my $i=0;$i<length($pkg)-1;$i++) {
203                 my $before = substr($pkg, 0, $i);
204                 my $after = substr($pkg, $i);
205                 $before = "^" if $before eq ""; # otherwise split doesn't work properly
206                 $package_postfixes{$after} .= "$before\0";
207         }
208 }
209 my %package_postfixes_db;
210 tie %package_postfixes_db, "DB_File", "$DBDIR/package_postfixes.db.new",
211         O_RDWR|O_CREAT, 0666, $DB_BTREE
212         or die "Error creating DB: $!";
213 while (my ($k, $v) = each(%package_postfixes)) {
214         $v =~ s/.$//s;
215         my $nr = $v;
216         $nr =~ s/[^\000]//g;
217         $nr = length($nr) + 1; # < number of hits
218         if ($nr > $MAX_PACKAGE_POSTFIXES) {
219                 $v = "\001" . $nr;
220         }
221         $package_postfixes_db{$k} = $v;
222 }
223 untie %package_postfixes_db;
224
225 store \%sections, "$DBDIR/sections.info";
226 store \%subsections, "$DBDIR/subsections.info";
227 store \%priorities, "$DBDIR/priorities.info";
228
229 rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db");
230 rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db");
231 for my $suite (@SUITES) {
232         rename("$DBDIR/packages_all_$suite.db.new",
233                "$DBDIR/packages_all_$suite.db");
234 }
235 rename("$DBDIR/packages_descriptions.db.new",
236        "$DBDIR/packages_descriptions.db");
237 rename("$DBDIR/descriptions_packages.db.new",
238        "$DBDIR/descriptions_packages.db");
239 rename("$DBDIR/descriptions.txt.new", "$DBDIR/descriptions.txt");
240 rename("$DBDIR/descriptions.db.new", "$DBDIR/descriptions.db");
241 rename("$DBDIR/package_postfixes.db.new", "$DBDIR/package_postfixes.db");