X-Git-Url: https://git.deb.at/?a=blobdiff_plain;ds=sidebyside;f=lib%2FPackages%2FSearch.pm;fp=lib%2FPackages%2FSearch.pm;h=021f0fd46196fad9847ce57616e3ebd47da5f802;hb=9f0c86ad48e5ccdd89c2ae10fc35baee286f3adf;hp=66b09441ba1ea72d84de6578e200a488971bf099;hpb=bf2a8d3268fcfd6ee90d65f4c55cd3b64b9d3461;p=deb%2Fpackages.git diff --git a/lib/Packages/Search.pm b/lib/Packages/Search.pm index 66b0944..021f0fd 100644 --- a/lib/Packages/Search.pm +++ b/lib/Packages/Search.pm @@ -58,6 +58,7 @@ our @ISA = qw( Exporter ); 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_xapian_search + find_similar ); our %EXPORT_TAGS = ( all => [ @EXPORT_OK ] ); @@ -240,6 +241,50 @@ sub do_xapian_search { } } +sub find_similar { + my ($pkg, $db, $did2pkg) = @_; + + my $db = Search::Xapian::Database->new( $db ); + my $enq = $db->enquire( "P$pkg" ); + debug( "Xapian Query was: ".$enq->get_query()->get_description(), 1) if DEBUG; + my $first_match = ($enq->matches(0,1))[0]->get_document(); + + my @terms; + my $term_it = $first_match->termlist_begin(); + my $term_end = $first_match->termlist_end(); + + for ($term_it; $term_it ne $term_end; $term_it++) { + debug( "TERM: ".$term_it->get_termname(), 3); + push @terms, $term_it->get_termname(); + } + + my $rel_enq = $db->enquire( OP_OR, @terms ); + debug( "Xapian Query was: ".$rel_enq->get_query()->get_description(), 1) if DEBUG; + my @rel_pkg = $rel_enq->matches(2,20); + + use Data::Dumper; + debug(Dumper(\@rel_pkg),1); + + my (@order, %tmp_results); + foreach my $match ( @rel_pkg ) { + my $id = $match->get_docid(); + my $result = $did2pkg->{$id}; + + foreach (split /\000/o, $result) { + my @data = split /\s/, $_, 3; + debug ("Considering $data[0], arch = $data[2], relevance=".$match->get_percent(), 3) if DEBUG; + next if $data[0] eq $pkg; + unless ($tmp_results{$data[0]}++) { + push @order, $data[0]; + } + } + } + undef $db; + + debug ("ORDER: @order", 2) if DEBUG; + return @order[0..10]; +} + sub find_binaries { my ($pkg, $archive, $suite, $src2bin) = @_;