3 # search_packages.pl -- CGI interface to the Packages files on packages.debian.org
5 # Copyright (C) 1998 James Treacy
6 # Copyright (C) 2000, 2001 Josip Rodin
7 # Copyright (C) 2001 Adam Heath
8 # Copyright (C) 2004 Martin Schulze
9 # Copyright (C) 2004-2006 Frank Lichtenheld
11 # use is allowed under the terms of the GNU Public License (GPL)
12 # see http://www.fsf.org/copyleft/gpl.html for a copy of the license
16 use CGI qw( -oldstyle_urls );
17 use CGI::Carp qw( fatalsToBrowser );
25 use Packages::Config qw( $DBDIR $ROOT $SEARCH_CGI $SEARCH_PAGE
26 @SUITES @SECTIONS @ARCHIVES @ARCHITECTURES );
28 use Packages::Search qw( :all );
29 use Packages::HTML ();
31 &Packages::CGI::reset;
33 $ENV{PATH} = "/bin:/usr/bin";
35 # Read in all the variables set by the form
37 if ($ARGV[0] && ($ARGV[0] eq 'php')) {
38 $input = new CGI(\*STDIN);
43 my $pet0 = new Benchmark;
44 my $tet0 = new Benchmark;
45 # use this to disable debugging in production mode completly
46 my $debug_allowed = 1;
47 my $debug = $debug_allowed && $input->param("debug");
48 $debug = 0 if !defined($debug) || $debug !~ /^\d+$/o;
49 $Packages::CGI::debug = $debug;
51 # read the configuration
52 our $db_read_time ||= 0;
54 &Packages::Config::init( '../' );
56 if (my $path = $input->param('path')) {
57 my @components = map { lc $_ } split /\//, $path;
59 my %SUITES = map { $_ => 1 } @SUITES;
60 my %SECTIONS = map { $_ => 1 } @SECTIONS;
61 my %ARCHIVES = map { $_ => 1 } @ARCHIVES;
62 my %ARCHITECTURES = map { $_ => 1 } @ARCHITECTURES;
64 foreach (@components) {
66 $input->param('suite', $_);
67 } elsif ($SECTIONS{$_}) {
68 $input->param('section', $_);
69 } elsif ($ARCHIVES{$_}) {
70 $input->param('archive', $_);
71 } elsif ($ARCHITECTURES{$_}) {
72 $input->param('arch', $_);
73 } elsif ($_ eq 'source') {
74 $input->param('searchon','sourcenames');
79 my ( $format, $keyword, $case, $subword, $exact, $searchon,
80 @suites, @sections, @archives, @archs );
82 my %params_def = ( keywords => { default => undef,
83 match => '^\s*([-+\@\w\/.:]+)\s*$',
85 suite => { default => 'stable', match => '^([\w-]+)$',
86 alias => 'version', array => ',',
88 replace => { all => \@SUITES } },
89 archive => { default => 'all', match => '^([\w-]+)$',
90 array => ',', var => \@archives,
91 replace => { all => \@ARCHIVES } },
92 case => { default => 'insensitive', match => '^(\w+)$',
94 official => { default => 0, match => '^(\w+)$' },
95 subword => { default => 0, match => '^(\w+)$',
97 exact => { default => undef, match => '^(\w+)$',
99 searchon => { default => 'all', match => '^(\w+)$',
101 section => { default => 'all', match => '^([\w-]+)$',
102 alias => 'release', array => ',',
104 replace => { all => \@SECTIONS } },
105 arch => { default => 'any', match => '^(\w+)$',
106 array => ',', var => \@archs, replace =>
107 { any => \@ARCHITECTURES } },
108 format => { default => 'html', match => '^(\w+)$',
112 my %params = Packages::Search::parse_params( $input, \%params_def, \%opts );
114 #XXX: Don't use alternative output formats yet
116 if ($format eq 'html') {
117 print $input->header;
120 if ($params{errors}{keywords}) {
121 fatal_error( "keyword not valid or missing" );
122 } elsif (length($keyword) < 2) {
123 fatal_error( "keyword too short (keywords need to have at least two characters)" );
126 my $case_bool = ( $case !~ /insensitive/ );
127 $exact = !$subword unless defined $exact;
128 $opts{h_suites} = { map { $_ => 1 } @suites };
129 $opts{h_sections} = { map { $_ => 1 } @sections };
130 $opts{h_archives} = { map { $_ => 1 } @archives };
131 $opts{h_archs} = { map { $_ => 1 } @archs };
133 # for URL construction
134 my $suites_param = join ',', @{$params{values}{suite}{no_replace}};
135 my $sections_param = join ',', @{$params{values}{section}{no_replace}};
136 my $archs_param = join ',', @{$params{values}{arch}{no_replace}};
139 my $keyword_enc = encode_entities $keyword || '';
140 my $searchon_enc = encode_entities $searchon;
141 my $suites_enc = encode_entities join ', ', @{$params{values}{suite}{no_replace}};
142 my $sections_enc = encode_entities join ', ', @{$params{values}{section}{no_replace}};
143 my $archs_enc = encode_entities join ', ', @{$params{values}{arch}{no_replace}};
144 my $pet1 = new Benchmark;
145 my $petd = timediff($pet1, $pet0);
146 debug( "Parameter evaluation took ".timestr($petd) );
148 my $st0 = new Benchmark;
151 our ($obj, $s_obj, $p_obj, $sp_obj,
152 %packages, %sources, %postf, %spostf, %src2bin, %did2pkg );
154 unless (@Packages::CGI::fatal_errors) {
156 my $dbmodtime = (stat("$DBDIR/packages_small.db"))[9];
157 if ($dbmodtime > $db_read_time) {
158 $obj = tie %packages, 'DB_File', "$DBDIR/packages_small.db",
159 O_RDONLY, 0666, $DB_BTREE
160 or die "couldn't tie DB $DBDIR/packages_small.db: $!";
161 $s_obj = tie %sources, 'DB_File', "$DBDIR/sources_small.db",
162 O_RDONLY, 0666, $DB_BTREE
163 or die "couldn't tie DB $DBDIR/sources_small.db: $!";
164 $p_obj = tie %postf, 'DB_File', "$DBDIR/package_postfixes.db",
165 O_RDONLY, 0666, $DB_BTREE
166 or die "couldn't tie postfix db $DBDIR/package_postfixes.db: $!";
167 $sp_obj = tie %spostf, 'DB_File', "$DBDIR/source_postfixes.db",
168 O_RDONLY, 0666, $DB_BTREE
169 or die "couldn't tie postfix db $DBDIR/source_postfixes.db: $!";
170 tie %src2bin, 'DB_File', "$DBDIR/sources_packages.db",
171 O_RDONLY, 0666, $DB_BTREE
172 or die "couldn't open $DBDIR/sources_packages.db: $!";
173 tie %did2pkg, 'DB_File', "$DBDIR/descriptions_packages.db",
174 O_RDONLY, 0666, $DB_BTREE
175 or die "couldn't tie DB $DBDIR/descriptions_packages.db: $!";
177 debug( "tied databases ($dbmodtime > $db_read_time)" );
178 $db_read_time = $dbmodtime;
181 if ($searchon eq 'names') {
182 push @results, @{ do_names_search( $keyword, \%packages,
184 \&read_entry, \%opts ) };
185 } elsif ($searchon eq 'sourcenames') {
186 push @results, @{ do_names_search( $keyword, \%sources,
188 \&read_src_entry, \%opts ) };
190 push @results, @{ do_names_search( $keyword, \%packages,
192 \&read_entry, \%opts ) };
193 push @results, @{ do_fulltext_search( $keyword, "$DBDIR/descriptions.txt",
196 \&read_entry, \%opts ) };
200 my $st1 = new Benchmark;
201 my $std = timediff($st1, $st0);
202 debug( "Search took ".timestr($std) );
204 if ($format eq 'html') {
205 my $suite_wording = $suites_enc eq "all" ? "all suites"
206 : "suite(s) <em>$suites_enc</em>";
207 my $section_wording = $sections_enc eq 'all' ? "all sections"
208 : "section(s) <em>$sections_enc</em>";
209 my $arch_wording = $archs_enc eq 'any' ? "all architectures"
210 : "architecture(s) <em>$archs_enc</em>";
211 if (($searchon eq "names") || ($searchon eq 'sourcenames')) {
212 my $source_wording = ( $searchon eq 'sourcenames' ) ? "source " : "";
213 my $exact_wording = $exact ? "named" : "that names contain";
214 msg( "You have searched for ${source_wording}packages $exact_wording <em>$keyword_enc</em> in $suite_wording, $section_wording, and $arch_wording." );
216 my $exact_wording = $exact ? "" : " (including subword matching)";
217 msg( "You have searched for <em>$keyword_enc</em> in packages names and descriptions in $suite_wording, $section_wording, and $arch_wording$exact_wording." );
221 if ($Packages::Search::too_many_hits) {
222 error( "Your search was too wide so we will only display exact matches. At least <em>$Packages::Search::too_many_hits</em> results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords." );
225 if (!@Packages::CGI::fatal_errors && !@results) {
226 if ($format eq 'html') {
227 my $keyword_esc = uri_escape( $keyword );
229 if (($searchon eq "names") || ($searchon eq 'sourcenames')) {
230 if (($suites_enc eq 'all')
231 && ($archs_enc eq 'any')
232 && ($sections_enc eq 'all')) {
233 error( "Can't find that package." );
235 error( "Can't find that package, at least not in that suite ".
236 ( ( $searchon eq 'sourcenames' ) ? "" : " and on that architecture" ) )
241 hint( "You have searched only for exact matches of the package name. You can try to search for <a href=\"$SEARCH_CGI?exact=0&searchon=$searchon&suite=$suites_param&case=$case&section=$sections_param&keywords=$keyword_esc&arch=$archs_param\">package names that contain your search string</a>." );
244 if (($suites_enc eq 'all')
245 && ($archs_enc eq 'any')
246 && ($sections_enc eq 'all')) {
247 error( "Can't find that string." );
249 error( "Can't find that string, at least not in that suite ($suites_enc, section $sections_enc) and on that architecture ($archs_enc)." );
254 hint( "You have searched only for words exactly matching your keywords. You can try to search <a href=\"$SEARCH_CGI?subword=1&searchon=$searchon&suite=$suites_param&case=$case&section=$sections_param&keywords=$keyword_esc&arch=$archs_param\">allowing subword matching</a>." );
257 hint( ( $printed ? "Or you" : "You" )." can try a different search on the <a href=\"$SEARCH_PAGE#search_packages\">Packages search page</a>." );
262 print Packages::HTML::header( title => 'Package Search Results' ,
264 title_tag => 'Debian Package Search Results',
265 print_title_above => 1,
266 print_search_field => 'packages',
267 search_field_values => {
268 keywords => $keyword_enc,
269 searchon => $searchon,
271 suite => $suites_enc,
272 section => $sections_enc,
284 my (%pkgs, %subsect, %sect, %desc, %binaries);
286 unless ($opts{searchon} eq 'sourcenames') {
288 my ($pkg_t, $archive, $suite, $arch, $section, $subsection,
289 $priority, $version, $desc) = @$_;
291 my ($pkg) = $pkg_t =~ m/^(.+)/; # untaint
292 $pkgs{$pkg}{$suite}{$archive}{$version}{$arch} = 1;
293 $subsect{$pkg}{$suite}{$archive}{$version} = $subsection;
294 $sect{$pkg}{$suite}{$archive}{$version} = $section
295 unless $section eq 'main';
297 $desc{$pkg}{$suite}{$archive}{$version} = $desc;
300 if ($opts{format} eq 'html') {
301 my ($start, $end) = multipageheader( $input, scalar keys %pkgs, \%opts );
304 foreach my $pkg (sort keys %pkgs) {
306 next if $count < $start or $count > $end;
307 printf "<h3>Package %s</h3>\n", $pkg;
309 foreach my $suite (@SUITES) {
310 foreach my $archive (@ARCHIVES) {
311 if (exists $pkgs{$pkg}{$suite}{$archive}) {
312 my @versions = version_sort keys %{$pkgs{$pkg}{$suite}{$archive}};
314 if ($sect{$pkg}{$suite}{$archive}{$versions[0]}) {
315 $origin_str .= " [<span style=\"color:red\">$sect{$pkg}{$suite}{$versions[0]}</span>]";
317 printf "<li><a href=\"$ROOT/%s/%s\">%s</a> (%s): %s %s\n",
318 $suite.(($archive ne 'us')?"/$archive":''), $pkg, $suite.(($archive ne 'us')?"/$archive":''), $subsect{$pkg}{$suite}{$archive}{$versions[0]},
319 $desc{$pkg}{$suite}{$archive}{$versions[0]}, $origin_str;
321 foreach my $v (@versions) {
322 printf "<br>%s: %s\n",
323 $v, join (" ", (sort keys %{$pkgs{$pkg}{$suite}{$archive}{$v}}) );
334 my ($pkg, $archive, $suite, $section, $subsection, $priority,
337 $pkgs{$pkg}{$suite}{$archive} = $version;
338 $subsect{$pkg}{$suite}{$archive}{source} = $subsection;
339 $sect{$pkg}{$suite}{$archive}{source} = $section
340 unless $section eq 'main';
342 $binaries{$pkg}{$suite}{$archive} = find_binaries( $pkg, $archive, $suite, \%src2bin );
345 if ($opts{format} eq 'html') {
346 my ($start, $end) = multipageheader( $input, scalar keys %pkgs, \%opts );
349 foreach my $pkg (sort keys %pkgs) {
351 next if ($count < $start) or ($count > $end);
352 printf "<h3>Source package %s</h3>\n", $pkg;
354 foreach my $suite (@SUITES) {
355 foreach my $archive (@ARCHIVES) {
356 if (exists $pkgs{$pkg}{$suite}{$archive}) {
358 if ($sect{$pkg}{$suite}{$archive}{source}) {
359 $origin_str .= " [<span style=\"color:red\">$sect{$pkg}{$suite}{$archive}{source}</span>]";
361 printf( "<li><a href=\"$ROOT/%s/source/%s\">%s</a> (%s): %s %s",
362 $suite.(($archive ne 'us')?"/$archive":''), $pkg, $suite.(($archive ne 'us')?"/$archive":''), $subsect{$pkg}{$suite}{$archive}{source},
363 $pkgs{$pkg}{$suite}{$archive}, $origin_str );
365 print "<br>Binary packages: ";
367 foreach my $bp (@{$binaries{$pkg}{$suite}{$archive}}) {
368 my $bp_link = sprintf( "<a href=\"$ROOT/%s/%s\">%s</a>",
369 $suite.(($archive ne 'us')?"/$archive":''), uri_escape( $bp ), $bp );
370 push @bp_links, $bp_link;
372 print join( ", ", @bp_links );
381 printindexline( $input, scalar keys %pkgs, \%opts );
383 #print_results(\@results, \%opts) if @results;;
384 my $tet1 = new Benchmark;
385 my $tetd = timediff($tet1, $tet0);
386 print "Total page evaluation took ".timestr($tetd)."<br>"
389 my $trailer = Packages::HTML::trailer( $ROOT );
390 $trailer =~ s/LAST_MODIFIED_DATE/gmtime()/e; #FIXME