use POSIX;
use HTML::Entities;
use DB_File;
+use Lingua::Stem v0.82;
+use Search::Xapian qw(:ops);
use Deb::Versions;
use Packages::CGI;
our @EXPORT_OK = qw( read_entry read_entry_all read_entry_simple
read_src_entry read_src_entry_all find_binaries
- do_names_search do_fulltext_search
+ do_names_search do_fulltext_search do_xapian_search
);
our %EXPORT_TAGS = ( all => [ @EXPORT_OK ] );
#FIXME: make configurable
my %fallback_suites = (
+ 'oldstable-backports' => 'oldstable',
+ 'oldstable-volatile' => 'oldstable',
'stable-backports' => 'stable',
'stable-volatile' => 'stable',
experimental => 'unstable' );
&$read_entry( $packages, $pkg, $results, $non_results, $opts );
}
}
-sub do_fulltext_search {
- my ($keywords, $file, $did2pkg, $packages, $read_entry, $opts,
+
+sub do_xapian_search {
+ my ($keywords, $db, $did2pkg, $packages, $read_entry, $opts,
$results, $non_results) = @_;
# NOTE: this needs to correspond with parse-packages!
$keyword =~ s;[^a-z0-9_/+]+; ;og;
push @tmp, $keyword;
}
- my $first_keyword = shift @tmp;
- @$keywords = @tmp;
+ my $stemmer = Lingua::Stem->new();
+ $keywords = $stemmer->stem( @tmp );
+
+ my $db = Search::Xapian::Database->new( $db );
+ my $enq = $db->enquire( OP_AND, @$keywords );
+ debug( "Xapian Query was: ".$enq->get_query()->get_description(), 1) if DEBUG;
+ my @matches = $enq->matches(0, 100);
my $numres = 0;
my %tmp_results;
- # fgrep is seriously faster than using perl
- open DESC, '-|', 'fgrep', '-n', '--', $first_keyword, $file
- or die "couldn't open $file: $!";
- LINE:
- while (<DESC>) {
- foreach my $k (@$keywords) {
- next LINE unless /\Q$k\E/;
- }
- /^(\d+)/;
- my $nr = $1;
- debug( "Matched line $_", 2) if DEBUG;
- my $result = $did2pkg->{$nr};
+ foreach my $match ( @matches ) {
+ my $id = $match->get_docid();
+ my $result = $did2pkg->{$id};
+
foreach (split /\000/o, $result) {
my @data = split /\s/, $_, 3;
# debug ("Considering $data[0], arch = $data[2]", 3) if DEBUG;
}
last if $numres > 100;
}
- close DESC;
+ undef $db;
$too_many_hits++ if $numres > 100;
- my @results;
foreach my $pkg (keys %tmp_results) {
&$read_entry( $packages, $pkg, $results, $non_results, $opts );
}
- }
+}
sub find_binaries {
my ($pkg, $archive, $suite, $src2bin) = @_;