1 package Packages::DoSearchContents;
6 use Benchmark ':hireswallclock';
11 our @ISA = qw( Exporter );
12 our @EXPORT = qw( do_search_contents );
15 use Packages::I18N::Locale;
16 use Packages::Search qw( :all );
19 use Packages::Config qw( $DBDIR $SEARCH_URL $SEARCH_PAGE
20 @SUITES @ARCHIVES @ARCHITECTURES $ROOT );
22 sub do_search_contents {
23 my ($params, $opts, $html_header, $menu, $page_content) = @_;
25 if ($params->{errors}{keywords}) {
26 fatal_error( _g( "keyword not valid or missing" ) );
27 } elsif (grep { length($_) < 2 } @{$opts->{keywords}}) {
28 fatal_error( _g( "keyword too short (keywords need to have at least two characters)" ) );
30 if ($params->{errors}{suite}) {
31 fatal_error( _g( "suite not valid or not specified" ) );
34 #FIXME: that's extremely hacky atm
35 if ($params->{values}{suite}{no_replace}[0] eq 'default') {
36 $params->{values}{suite}{no_replace} =
37 $params->{values}{suite}{final} = $opts->{suite} = [ 'stable' ];
40 if (@{$opts->{suite}} > 1) {
41 fatal_error( sprintf( _g( "more than one suite specified for contents search (%s)" ), "@{$opts->{suite}}" ) );
46 my @keywords = @{$opts->{keywords}};
47 my $mode = $opts->{mode} || '';
48 my $suite = $opts->{suite}[0];
49 my $archive = $opts->{archive}[0] ||'';
50 $Packages::Search::too_many_hits = 0;
52 # for URL construction
53 my $keyword_esc = uri_escape( "@keywords" );
54 my $suites_param = join ',', @{$params->{values}{suite}{no_replace}};
55 my $sections_param = join ',', @{$params->{values}{section}{no_replace}};
56 my $archs_param = join ',', @{$params->{values}{arch}{no_replace}};
59 my $keyword_enc = encode_entities "@keywords" || '';
60 my $suites_enc = encode_entities( join( ', ', @{$params->{values}{suite}{no_replace}} ), '&<>"' );
61 my $sections_enc = encode_entities( join( ', ', @{$params->{values}{section}{no_replace}} ), '&<>"' );
62 my $archs_enc = encode_entities( join( ', ', @{$params->{values}{arch}{no_replace}} ), '&<>"' );
64 my $st0 = new Benchmark;
67 unless (@Packages::CGI::fatal_errors) {
71 my $first_kw = lc shift @keywords;
72 # full filename search is tricky
73 my $ffn = $mode eq 'filename';
75 my $reverses = tie my %reverses, 'DB_File', "$DBDIR/contents/reverse_$suite.db",
76 O_RDONLY, 0666, $DB_BTREE
77 or die "Failed opening reverse DB: $!";
80 open FILENAMES, '-|', 'fgrep', '--', $first_kw, "$DBDIR/contents/filenames_$suite.txt"
81 or die "Failed opening filename table: $!";
86 foreach my $kw (@keywords) {
87 next FILENAME unless /\Q$kw\E/;
89 &searchfile(\@results, reverse($_)."/", \$nres, $reverses);
90 last if $Packages::Search::too_many_hits;
92 close FILENAMES or warn "fgrep error: $!\n";
95 error(_g("The search mode you selected doesn't support more than one keyword."))
98 my $kw = reverse $first_kw;
100 # exact filename searching follows trivially:
101 $kw = "$kw/" if $mode eq 'exactfilename';
103 &searchfile(\@results, $kw, \$nres, $reverses);
109 my $st1 = new Benchmark;
110 my $std = timediff($st1, $st0);
111 debug( "Search took ".timestr($std) ) if DEBUG;
114 my $suite_wording = sprintf(_g("suite <em>%s</em>"), $suites_enc );
115 my $section_wording = $sections_enc eq 'all' ? _g("all sections")
116 : sprintf(_g("section(s) <em>%s</em>"), $sections_enc );
117 my $arch_wording = $archs_enc eq 'any' ? _g("all architectures")
118 : sprintf(_g("architecture(s) <em>%s</em>"), $archs_enc );
119 my $wording = _g("paths that end with");
120 if ($mode eq 'filename') {
121 $wording = _g("files named");
122 } elsif ($mode eq 'exactfilename') {
123 $wording = _g("filenames that contain");
125 msg( sprintf( _g("You have searched for %s <em>%s</em> in %s, %s, and %s." ),
126 $wording, $keyword_enc,
127 $suite_wording, $section_wording, $arch_wording ) );
129 if ($mode ne 'filename') {
130 msg( '<a href="'.make_search_url('',"keywords=$keyword_esc",{mode=>'filename'}).
131 "\">"._g("Search within filenames")."</a>");
133 if ($mode ne 'exactfilename') {
134 msg( '<a href="'.make_search_url('',"keywords=$keyword_esc",{mode=>'exactfilename'}).
135 "\">"._g("Search exact filename")."</a>");
137 if ($mode eq 'exactfilename' || $mode eq 'filename') {
138 msg( '<a href="'.make_search_url('',"keywords=$keyword_esc",{mode=>undef}).
139 "\">"._g("Search for paths ending with")."</a>");
142 msg( _g("Search in other suite:")." ".
143 join( ' ', map { '[<a href="'.make_search_url('',"keywords=$keyword_esc",{suite=>$_}).
144 "\">$_</a>]" } @SUITES ) );
146 if ($Packages::Search::too_many_hits) {
147 error( _g( "Your search was too wide so we will only display only the first about 100 matches. Please consider using a longer keyword or more keywords." ) );
150 %$html_header = ( title => _g( 'Package Contents Search Results' ),
151 lang => $opts->{lang},
152 title_tag => _g( 'Debian Package Contents Search Results' ),
154 print_search_field => 'packages',
155 search_field_values => {
156 keywords => $keyword_enc,
157 searchon => 'contents',
159 suite => $suites_enc,
160 section => $sections_enc,
161 exact => $opts->{exact},
162 debug => $opts->{debug},
167 my (%results,%archs);
168 foreach my $result (sort { $a->[0] cmp $b->[0] } @results) {
169 my $file = shift @$result;
172 my ($pkg, $arch) = split /:/, $_;
173 next unless $opts->{h_archs}{$arch};
174 $pkgs{$pkg}{$arch}++;
175 $archs{$arch}++ unless $arch eq 'all';
177 next unless keys %pkgs;
178 $results{$file} = \%pkgs;
180 my @all_archs = keys %archs;
181 @all_archs = @ARCHITECTURES unless @all_archs;
182 debug( "all_archs = @all_archs", 1 ) if DEBUG;
183 msg(_g("Limit search to a specific architecture:")." ".
184 join( ' ', map { '[<a href="'.make_search_url('',"keywords=$keyword_esc",{arch=>$_}).
185 "\">$_</a>]" } @all_archs ) )
186 unless (@{$opts->{arch}} == 1) || (@all_archs == 1);
187 msg(sprintf(_g('Search in <a href="%s">all architectures</a>'),
188 make_search_url('',"keywords=$keyword_esc",{arch=>undef})))
189 if @{$opts->{arch}} == 1;
191 if (!@Packages::CGI::fatal_errors && !keys(%results)) {
192 error( _g( "Nothing found" ) );
196 $$page_content .= "<p>".sprintf( _g( 'Found %s results' ),
197 scalar keys %results )."</p>";
198 $$page_content .= '<div
199 id="pcontentsres"><table><colgroup><col><col></colgroup><tr><th>'._g('File').'</th><th>'._g('Packages')
201 foreach my $file (sort keys %results) {
202 my $file_enc = encode_entities($file);
203 foreach my $kw (@{$opts->{keywords}}) {
204 my $kw_enc = encode_entities($kw);
205 $file_enc =~ s#(\Q$kw_enc\E)#<span class="keyword">$1</span>#g;
207 $$page_content .= "<tr><td class=\"file\">/$file_enc</td><td>";
209 foreach my $pkg (sort keys %{$results{$file}}) {
211 my @archs = keys %{$results{$file}{$pkg}};
212 unless ($results{$file}{$pkg}{all} ||
213 (@archs == @all_archs)) {
214 if (@archs < @all_archs/2) {
215 $arch_str = ' ['.join(' ',sort @archs).']';
217 $arch_str = ' ['._g('not').' '.
218 join(' ', grep { !$results{$file}{$pkg}{$_} } @all_archs).']';
221 push @pkgs, "<a href=\"".make_url($pkg,'',{suite=>$suite})."\">$pkg</a>$arch_str";
223 $$page_content .= join( ", ", @pkgs);
224 $$page_content .= "</td></tr>\n";
226 $$page_content .= '<tr><th>'._g('File').'</th><th>'._g('Packages')."</th></tr>\n" if @results > 20;
227 $$page_content .= '</table></div>';
229 } # sub do_search_contents
233 my ($results, $kw, $nres, $reverses) = @_;
235 my ($key, $value) = ($kw, "");
236 debug( "searchfile: kw=$kw", 1 ) if DEBUG;
237 for (my $status = $reverses->seq($key, $value, R_CURSOR);
239 $status = $reverses->seq( $key, $value, R_NEXT)) {
241 # FIXME: what's the most efficient "is prefix of" thingy? We only want to know
242 # whether $kw is or is not a prefix of $key
243 last unless index($key, $kw) == 0;
244 debug( "found $key", 2 ) if DEBUG;
246 my @hits = split /\0/o, $value;
247 push @$results, [ scalar reverse($key), @hits ];
248 last if ($$nres)++ > 100;
251 $Packages::Search::too_many_hits += $$nres - 100 if $$nres > 100;