]> git.deb.at Git - deb/packages.git/blob - lib/Packages/DoSearch.pm
9fa8d3daad5db7423de408e7d99ca0df98e8ee76
[deb/packages.git] / lib / Packages / DoSearch.pm
1 package Packages::DoSearch;
2
3 use strict;
4 use warnings;
5
6 use Benchmark ':hireswallclock';
7 use DB_File;
8 use Exporter;
9 our @ISA = qw( Exporter );
10 our @EXPORT = qw( do_search );
11
12 use Deb::Versions;
13 use Packages::Search qw( :all );
14 use Packages::CGI qw( :DEFAULT );
15 use Packages::DB;
16 use Packages::Config qw( $DBDIR @SUITES @ARCHIVES $ROOT );
17
18 sub do_search {
19     my ($params, $opts, $page_content) = @_;
20     my $cat = $opts->{cat};
21
22     $Params::Search::too_many_hits = 0;
23
24     if ($params->{errors}{keywords}) {
25         fatal_error( $cat->g( "keyword not valid or missing" ) );
26         $opts->{keywords} = [];
27     } elsif (grep { length($_) < 2 } @{$opts->{keywords}}) {
28         fatal_error( $cat->g( "keyword too short (keywords need to have at least two characters)" ) );
29     }
30
31     my @keywords = @{$opts->{keywords}};
32     my $searchon = $opts->{searchon};
33     $page_content->{search_keywords} = \@keywords;
34
35     my $st0 = new Benchmark;
36     my (@results, @non_results);
37
38     unless (@Packages::CGI::fatal_errors) {
39
40         if ($searchon eq 'names') {
41             if ($opts->{source}) {
42                 do_names_search( [ @keywords ], \%sources, $sp_obj,
43                                  \&read_src_entry_all, $opts,
44                                  \@results, \@non_results );
45             } else {
46                 do_names_search( [ @keywords ], \%packages, $p_obj,
47                                  \&read_entry_all, $opts,
48                                  \@results, \@non_results );
49             }
50         } else {
51             do_names_search( [ @keywords ], \%packages, $p_obj,
52                              \&read_entry_all, $opts,
53                              \@results, \@non_results );
54             my $fts1 = new Benchmark;
55             do_xapian_search( [ @keywords ], "$DBDIR/xapian/",
56                                 \%did2pkg, \%packages,
57                                 \&read_entry_all, $opts,
58                                 \@results, \@non_results );
59             my $fts2 = new Benchmark;
60             my $fts_xapian = timediff($fts2,$fts1);
61             debug( "Fulltext search took ".timestr($fts_xapian) )
62                 if DEBUG;
63         }
64     }
65
66 #    use Data::Dumper;
67 #    debug( join( "", Dumper( \@results, \@non_results )) ) if DEBUG;
68     my $st1 = new Benchmark;
69     my $std = timediff($st1, $st0);
70     debug( "Search took ".timestr($std) ) if DEBUG;
71
72     $page_content->{too_many_hits} = $Packages::Search::too_many_hits;
73     #FIXME: non_results can't be compared to results since it is
74     # not normalized to unique packages
75     $page_content->{non_results} = scalar @non_results;
76
77     if (@results) {
78         my (%pkgs, %subsect, %sect, %archives, %desc, %binaries, %provided_by);
79
80         my %sort_by_relevance;
81         for (1 ... scalar @results) {
82 #           debug("$results[$_][0] => $_", 4) if DEBUG;
83             $sort_by_relevance{$results[$_-1][0]} = $_;
84         }
85 #       use Data::Dumper;
86 #       debug( "sort_by_relevance=".Dumper(\%sort_by_relevance), 4);
87
88         unless ($opts->{source}) {
89             foreach (@results) {
90                 my ($pkg_t, $archive, $suite, $arch, $section, $subsection,
91                     $priority, $version, $desc_md5, $desc) = @$_;
92
93                 my ($pkg) = $pkg_t =~ m/^(.+)/; # untaint
94                 if ($arch ne 'virtual') {
95                     $pkgs{$pkg}{$suite}{$version}{$arch} = 1;
96                     $subsect{$pkg}{$suite}{$version} = $subsection;
97                     $sect{$pkg}{$suite}{$version} = $section;
98                     $archives{$pkg}{$suite}{$version} ||= $archive;
99
100                     $desc{$pkg}{$suite}{$version} = [ $desc_md5, $desc ];
101                 } else {
102                     $provided_by{$pkg}{$suite} = [ split /\s+/, $desc ];
103                 }
104             }
105
106             my %uniq_pkgs = map { $_ => 1 } (keys %pkgs, keys %provided_by);
107             my @pkgs;
108             if ($searchon eq 'names') {
109                 @pkgs = sort keys %uniq_pkgs;
110             } else {
111                 @pkgs = sort { $sort_by_relevance{$a} <=> $sort_by_relevance{$b} } keys %uniq_pkgs;
112             }
113             process_packages( $page_content, 'packages', \%pkgs, \@pkgs,
114                               $opts, \@keywords,
115                               \&process_package, \%provided_by,
116                               \%archives, \%sect, \%subsect,
117                               \%desc );
118
119         } else { # unless $opts->{source}
120             foreach (@results) {
121                 my ($pkg, $archive, $suite, $section, $subsection, $priority,
122                     $version) = @$_;
123
124                 my $real_archive = '';
125                 if ($archive eq 'security') {
126                     $real_archive = $archive;
127                     $archive = 'us';
128                 }
129                 if (($real_archive eq $archive) &&
130                     $pkgs{$pkg}{$suite}{$archive} &&
131                     (version_cmp( $pkgs{$pkg}{$suite}{$archive}, $version ) >= 0)) {
132                     next;
133                 }
134                 $pkgs{$pkg}{$suite}{$archive} = $version;
135                 $subsect{$pkg}{$suite}{$archive}{source} = $subsection;
136                 $sect{$pkg}{$suite}{$archive}{source} = $section
137                     unless $section eq 'main';
138                 $archives{$pkg}{$suite}{$archive}{source} = $real_archive
139                     if $real_archive;
140
141                 $binaries{$pkg}{$suite}{$archive} = find_binaries( $pkg, $archive, $suite, \%src2bin );
142             }
143
144             my @pkgs = sort keys %pkgs;
145             process_packages( $page_content, 'src_packages', \%pkgs, \@pkgs,
146                               $opts, \@keywords,
147                               \&process_src_package, \%archives,
148                               \%sect, \%subsect, \%binaries );
149         } # else unless $opts->{source}
150     } # if @results
151 } # sub do_search
152
153 sub process_packages {
154     my ($content, $target, $pkgs, $pkgs_list, $opts, $keywords, $print_func, @func_args) = @_;
155
156     my @categories;
157     $content->{results} = scalar @$pkgs_list;
158
159     my $keyword;
160     $keyword = $keywords->[0] if @$keywords == 1;
161             
162     my $have_exact;
163     if ($keyword && grep { $_ eq $keyword } @$pkgs_list) {
164         $have_exact = 1;
165         $categories[0]{name} = $opts->{cat}->g( "Exact hits" );
166
167         $categories[0]{$target} = [ &$print_func( $opts, $keyword,
168                                                   $pkgs->{$keyword}||{},
169                                                   map { $_->{$keyword}||{} } @func_args ) ];
170         @$pkgs_list = grep { $_ ne $keyword } @$pkgs_list;
171     }
172             
173     if (@$pkgs_list && (($opts->{searchon} ne 'names') || !$opts->{exact})) {
174         my %cat;
175         $cat{name} = $opts->{cat}->g( 'Other hits' ) if $have_exact;
176         
177         $cat{packages} = [];
178         foreach my $pkg (@$pkgs_list) {
179             push @{$cat{$target}}, &$print_func( $opts, $pkg, $pkgs->{$pkg}||{},
180                                                  map { $_->{$pkg}||{} } @func_args );
181         }
182         push @categories, \%cat;
183     } elsif (@$pkgs_list) {
184         $content->{skipped} = scalar @$pkgs_list;
185     }
186
187     $content->{categories} = \@categories;
188 }
189
190 sub process_package {
191     my ($opts, $pkg, $pkgs, $provided_by,
192         $archives, $sect, $subsect, $desc) = @_;
193
194     my %pkg = ( pkg => $pkg,
195                 suites => [] );
196
197     foreach my $suite (@SUITES) {
198         my %suite = ( suite => $suite );
199         if (exists $pkgs->{$suite}) {
200             my %archs_printed;
201             my @versions = version_sort keys %{$pkgs->{$suite}};
202             $suite{section} = $sect->{$suite}{$versions[0]};
203             $suite{subsection} = $subsect->{$suite}{$versions[0]};
204             my $desc_md5 = $desc->{$suite}{$versions[0]}[0];
205             $suite{desc} = $desc->{$suite}{$versions[0]}[1];
206             $suite{versions} = [];
207
208             my $trans_desc = $desctrans{$desc_md5};
209             my %sdescs;
210             if ($trans_desc) {
211                 my %trans_desc = split /\000|\001/, $trans_desc;
212                 while (my ($l, $d) = each %trans_desc) {
213                     $d =~ s/\n.*//os;
214
215                     $sdescs{$l} = $d;
216                 }
217                 $suite{trans_desc} = \%sdescs;
218             }
219
220             foreach my $v (@versions) {
221                 my %version;
222                 $version{version} = $v;
223                 $version{archive} = $archives->{$suite}{$v};
224                     
225                 $version{architectures} = [ grep { !$archs_printed{$_} } sort keys %{$pkgs->{$suite}{$v}} ];
226                 push @{$suite{versions}}, \%version if @{$version{architectures}};
227
228                 $archs_printed{$_}++ foreach @{$version{architectures}};
229             }
230             if (my $p =  $provided_by->{$suite}) {
231                 $suite{providers} = $p;
232             }
233         } elsif (my $p =  $provided_by->{$suite}) {
234             $suite{desc} = $opts->{cat}->g('Virtual package');
235             $suite{providers} = $p;
236         }
237         push @{$pkg{suites}}, \%suite if $suite{versions} || $suite{providers};
238     }
239
240     return \%pkg;
241 }
242
243 sub process_src_package {
244     my ($opts, $pkg, $pkgs, $archives, $sect, $subsect, $binaries) = @_;
245
246     my %pkg = ( pkg => $pkg,
247                 origins => [] );
248
249     foreach my $suite (@SUITES) {
250         foreach my $archive (@ARCHIVES) {
251             if (exists $pkgs->{$suite}{$archive}) {
252                 my %origin;
253                 $origin{version} = $pkgs->{$suite}{$archive};
254                 $origin{suite} = $suite;
255                 $origin{archive} = $archive; 
256                 $origin{section} = $sect->{$suite}{$archive}{source};
257                 $origin{subsection} = $subsect->{$suite}{$archive}{source};
258                 $origin{real_archive} = $archives->{$suite}{$archive}{source};
259
260                 $origin{binaries} = $binaries->{$suite}{$archive};
261                 push @{$pkg{origins}}, \%origin;
262             }
263         }
264     }
265
266     return \%pkg;
267 }
268
269 1;