use Packages::Search qw( :all );
use Packages::HTML ();
-my $thisscript = "search_packages.pl";
+my $thisscript = $Packages::HTML::SEARCH_CGI;
my $HOME = "http://www.debian.org";
my $ROOT = "";
my $SEARCHPAGE = "http://packages.debian.org/";
my @SUITES = qw( oldstable stable testing unstable experimental );
-my @DISTS = @SUITES;
my @SECTIONS = qw( main contrib non-free );
my @ARCHIVES = qw( us security installer );
my @ARCHITECTURES = qw( alpha amd64 arm hppa hurd-i386 i386 ia64
$ENV{PATH} = "/bin:/usr/bin";
# Read in all the variables set by the form
-my $input = new CGI;
+my $input;
+if ($ARGV[0] && ($ARGV[0] eq 'php')) {
+ $input = new CGI(\*STDIN);
+} else {
+ $input = new CGI;
+}
my $pet0 = new Benchmark;
# use this to disable debugging in production mode completly
my $debug_allowed = 1;
my $debug = $debug_allowed && $input->param("debug");
+$debug = 0 if not defined($debug);
$Search::Param::debug = 1 if $debug > 1;
# If you want, just print out a list of all of the variables and exit.
my @archs = @{$params{values}{arch}{final}};
my $page = $params{values}{page}{final};
my $results_per_page = $params{values}{number}{final};
+my %opts = ( case_bool => $case_bool, exact => $exact );
# for URL construction
my $suites_param = join ',', @{$params{values}{suite}{no_replace}};
my $st0 = new Benchmark;
my @results;
+my $too_many_hits;
if ($searchon eq 'sourcenames') {
$search_on_sources = 1;
}
my %sections = map { $_ => 1 } @sections;
my %archs = map { $_ => 1 } @archs;
-print "DEBUG: suites=@suites, sections=@sections, archs=@archs<br>" if $debug > 2;
-
-if ($searchon eq 'names') {
+print "DEBUG: suites=@suites, sections=@sections, archs=@archs<br>"
+ if $debug > 2;
+
+sub read_entry {
+ my ($hash, $key, $results) = @_;
+ my $result = $hash->{$key};
+ foreach (split /\000/, $result) {
+ my @data = split ( /\s/, $_, 7 );
+ print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
+ if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
+ && $sections{$data[2]}) {
+ print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
+ push @$results, [ $key, @data ];
+ }
+ }
+}
+sub read_src_entry {
+ my ($hash, $key, $results) = @_;
+ my $result = $hash->{$key};
+
+ foreach (split /\000/, $result) {
+ my @data = split ( /\s/, $_, 5 );
+ print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
+ if ($suites{$data[0]} && $sections{$data[1]}) {
+ print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
+ push @$results, [ $key, @data ];
+ }
+ }
+}
+sub do_names_search {
+ my ($keyword, $file, $postfix_file, $read_entry, $opts) = @_;
+ my @results;
- $keyword = lc $keyword unless $case_bool;
+ $keyword = lc $keyword unless $opts->{case_bool};
- my $obj = tie my %packages, 'DB_File', "$DBDIR/packages_small.db", O_RDONLY, 0666, $DB_BTREE
- or die "couldn't tie DB $DBDIR/packages_small.db: $!";
+ my $obj = tie my %packages, 'DB_File', "$DBDIR/$file", O_RDONLY, 0666, $DB_BTREE
+ or die "couldn't tie DB $DBDIR/$file: $!";
- if ($exact) {
- my $result = $packages{$keyword};
- foreach (split /\000/, $result) {
- my @data = split ( /\s/, $_, 7 );
- print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
- if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
- && $sections{$data[2]}) {
- print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
- push @results, [ $keyword, @data ];
- }
- }
+ if ($opts->{exact}) {
+ &$read_entry( \%packages, $keyword, \@results );
} else {
- while (my ($pkg, $result) = each %packages) {
- #what's faster? I can't really see a difference
- (index($pkg, $keyword) >= 0) or next;
- #$pkg =~ /\Q$keyword\E/ or next;
- foreach (split /\000/, $packages{$pkg}) {
- my @data = split ( /\s/, $_, 7 );
- print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
- if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
- && $sections{$data[2]}) {
- print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
- push @results, [ $pkg , @data ];
+ my ($key, $prefixes) = ($keyword, '');
+ my %pkgs;
+ my $p_obj = tie my %pref, 'DB_File', "$DBDIR/$postfix_file", O_RDONLY, 0666, $DB_BTREE
+ or die "couldn't tie postfix db $DBDIR/$postfix_file: $!";
+ $p_obj->seq( $key, $prefixes, R_CURSOR );
+ while (index($key, $keyword) >= 0) {
+ if ($prefixes =~ /^\001(\d+)/o) {
+ $too_many_hits += $1;
+ } else {
+ foreach (split /\000/o, $prefixes) {
+ $_ = '' if $_ eq '^';
+ print "DEBUG: add word $_$key<br>" if $debug > 2;
+ $pkgs{$_.$key}++;
}
}
+ last if $p_obj->seq( $key, $prefixes, R_NEXT ) != 0;
+ last if $too_many_hits or keys %pkgs >= 100;
+ }
+
+ my $no_results = keys %pkgs;
+ if ($too_many_hits || ($no_results >= 100)) {
+ $too_many_hits += $no_results;
+ %pkgs = ( $keyword => 1 );
+ }
+ foreach my $pkg (sort keys %pkgs) {
+ &$read_entry( \%packages, $pkg, \@results );
}
}
-} elsif ($searchon eq 'sourcenames') {
-
- $keyword = lc $keyword unless $case_bool;
-
- my $obj = tie my %packages, 'DB_File', "$DBDIR/sources_small.db", O_RDONLY, 0666, $DB_BTREE
- or die "couldn't tie DB $DBDIR/sources_small.db: $!";
-
- if ($exact) {
- my $result = $packages{$keyword};
- foreach (split /\000/, $result) {
- my @data = split ( /\s/, $_, 5 );
- print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
- if ($suites{$data[0]} && $sections{$data[1]}) {
- print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
- push @results, [ $keyword, @data ];
- }
+ return \@results;
+}
+sub do_fulltext_search {
+ my ($keword, $file, $mapping, $lookup, $read_entry, $opts) = @_;
+ my @results;
+
+ my @lines;
+ my $regex;
+ if ($opts->{case_bool}) {
+ if ($opts->{exact}) {
+ $regex = qr/\b\Q$keyword\E\b/o;
+ } else {
+ $regex = qr/\Q$keyword\E/o;
}
} else {
- while (my ($pkg, $result) = each %packages) {
- #what's faster? I can't really see a difference
- (index($pkg, $keyword) >= 0) or next;
- #$pkg =~ /\Q$keyword\E/ or next;
- foreach (split /\000/, $packages{$pkg}) {
- my @data = split ( /\s/, $_, 5 );
- print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
- if ($suites{$data[0]} && $sections{$data[1]}) {
- print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
- push @results, [ $pkg , @data ];
- }
- }
+ if ($exact) {
+ $regex = qr/\b\Q$keyword\E\b/io;
+ } else {
+ $regex = qr/\Q$keyword\E/io;
}
}
+
+ open DESC, '<', "$DBDIR/$file"
+ or die "couldn't open $DBDIR/$file: $!";
+ while (<DESC>) {
+ $_ =~ $regex or next;
+ print "DEBUG: Matched line $.<br>" if $debug > 2;
+ push @lines, $.;
+ }
+ close DESC;
+
+ tie my %packages, 'DB_File', "$DBDIR/$lookup", O_RDONLY, 0666, $DB_BTREE
+ or die "couldn't tie DB $DBDIR/$lookup: $!";
+ tie my %did2pkg, 'DB_File', "$DBDIR/$mapping", O_RDONLY, 0666, $DB_BTREE
+ or die "couldn't tie DB $DBDIR/$mapping: $!";
+
+ my %tmp_results;
+ foreach my $l (@lines) {
+ my $result = $did2pkg{$l};
+ foreach (split /\000/o, $result) {
+ my @data = split /\s/, $_, 3;
+ next unless $archs{$data[2]};
+ $tmp_results{$data[0]}++;
+ }
+ }
+ foreach my $pkg (keys %tmp_results) {
+ &$read_entry( \%packages, $pkg, \@results );
+ }
+ return \@results;
+}
+
+if ($searchon eq 'names') {
+ push @results, @{ do_names_search( $keyword, 'packages_small.db',
+ 'package_postfixes.db',
+ \&read_entry, \%opts ) };
+} elsif ($searchon eq 'sourcenames') {
+ push @results, @{ do_names_search( $keyword, 'sources_small.db',
+ 'source_postfixes.db',
+ \&read_src_entry, \%opts ) };
+} else {
+ push @results, @{ do_names_search( $keyword, 'packages_small.db',
+ 'package_postfixes.db',
+ \&read_entry, \%opts ) };
+ push @results, @{ do_fulltext_search( $keyword, 'descriptions.txt',
+ 'descriptions_packages.db',
+ 'packages_small.db',
+ \&read_entry, \%opts ) };
}
my $st1 = new Benchmark;
}
}
+if ($too_many_hits) {
+print "<p><strong>Your search was too wide so we will only display exact matches. At least <em>$too_many_hits</em> results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords.</strong></p>";
+}
+
if (!@results) {
if ($format eq 'html') {
my $keyword_esc = uri_escape( $keyword );
$rdf->addns( debpkg => 'http://packages.debian.org/xml/01-debian-packages-rdf' );
my @triples;
foreach my $pkg (sort keys %pkgs) {
- foreach my $ver (@DISTS) {
+ foreach my $ver (@SUITES) {
if (exists $pkgs{$pkg}{$ver}) {
my @versions = version_sort keys %{$pkgs{$pkg}{$ver}};
foreach my $version (@versions) {
</div>
END
+my $pete = new Benchmark;
+my $petd = timediff($pete, $pet0);
+print "Total page evaluation took ".timestr($petd)."<br>"
+ if $debug_allowed;
print $input->end_html;
}
+
+# vim: ts=8 sw=4