X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=lib%2FPackages%2FDoSearch.pm;h=1aa67f1d327bf3f10a5e7da9730a628320862f8b;hb=cf56c281cb5054022e986de58ca288c71f939474;hp=2603ff9a10c848333643f27e98a6f450e4c36611;hpb=32f5a5ba7f66aedf7afc095dce485073fe030b88;p=deb%2Fpackages.git diff --git a/lib/Packages/DoSearch.pm b/lib/Packages/DoSearch.pm index 2603ff9..1aa67f1 100644 --- a/lib/Packages/DoSearch.pm +++ b/lib/Packages/DoSearch.pm @@ -5,8 +5,6 @@ use warnings; use Benchmark ':hireswallclock'; use DB_File; -use URI::Escape; -use HTML::Entities; use Exporter; our @ISA = qw( Exporter ); our @EXPORT = qw( do_search ); @@ -19,15 +17,16 @@ use Packages::DB; use Packages::Config qw( $DBDIR @SUITES @ARCHIVES $ROOT ); sub do_search { - my ($params, $opts, $html_header, $page_content) = @_; + my ($params, $opts, $page_content) = @_; + my $cat = $opts->{cat}; $Params::Search::too_many_hits = 0; if ($params->{errors}{keywords}) { - fatal_error( _g( "keyword not valid or missing" ) ); + fatal_error( $cat->g( "keyword not valid or missing" ) ); $opts->{keywords} = []; } elsif (grep { length($_) < 2 } @{$opts->{keywords}}) { - fatal_error( _g( "keyword too short (keywords need to have at least two characters)" ) ); + fatal_error( $cat->g( "keyword too short (keywords need to have at least two characters)" ) ); } my @keywords = @{$opts->{keywords}}; @@ -53,10 +52,15 @@ sub do_search { do_names_search( [ @keywords ], \%packages, $p_obj, \&read_entry_all, $opts, \@results, \@non_results ); - do_fulltext_search( [ @keywords ], "$DBDIR/descriptions.txt", + my $fts1 = new Benchmark; + do_xapian_search( [ @keywords ], "$DBDIR/xapian/", \%did2pkg, \%packages, \&read_entry_all, $opts, \@results, \@non_results ); + my $fts2 = new Benchmark; + my $fts_xapian = timediff($fts2,$fts1); + debug( "Fulltext search took ".timestr($fts_xapian) ) + if DEBUG; } } @@ -74,10 +78,18 @@ sub do_search { if (@results) { my (%pkgs, %subsect, %sect, %archives, %desc, %binaries, %provided_by); + my %sort_by_relevance; + for (1 ... scalar @results) { +# debug("$results[$_][0] => $_", 4) if DEBUG; + $sort_by_relevance{$results[$_-1][0]} = $_; + } +# use Data::Dumper; +# debug( "sort_by_relevance=".Dumper(\%sort_by_relevance), 4); + unless ($opts->{source}) { foreach (@results) { my ($pkg_t, $archive, $suite, $arch, $section, $subsection, - $priority, $version, $desc) = @$_; + $priority, $version, $desc_md5, $desc) = @$_; my ($pkg) = $pkg_t =~ m/^(.+)/; # untaint if ($arch ne 'virtual') { @@ -86,15 +98,21 @@ sub do_search { $sect{$pkg}{$suite}{$version} = $section; $archives{$pkg}{$suite}{$version} ||= $archive; - $desc{$pkg}{$suite}{$version} = $desc; + $desc{$pkg}{$suite}{$version} = [ $desc_md5, $desc ]; } else { $provided_by{$pkg}{$suite} = [ split /\s+/, $desc ]; } } my %uniq_pkgs = map { $_ => 1 } (keys %pkgs, keys %provided_by); - my @pkgs = sort keys %uniq_pkgs; - process_packages( $page_content, 'packages', \%pkgs, \@pkgs, $opts, \@keywords, + my @pkgs; + if ($searchon eq 'names') { + @pkgs = sort keys %uniq_pkgs; + } else { + @pkgs = sort { $sort_by_relevance{$a} <=> $sort_by_relevance{$b} } keys %uniq_pkgs; + } + process_packages( $page_content, 'packages', \%pkgs, \@pkgs, + $opts, \@keywords, \&process_package, \%provided_by, \%archives, \%sect, \%subsect, \%desc ); @@ -105,7 +123,7 @@ sub do_search { $version) = @$_; my $real_archive = ''; - if ($archive =~ /^(security|non-US)$/) { + if ($archive eq 'security') { $real_archive = $archive; $archive = 'us'; } @@ -125,7 +143,8 @@ sub do_search { } my @pkgs = sort keys %pkgs; - process_packages( $page_content, 'src_packages', \%pkgs, \@pkgs, $opts, \@keywords, + process_packages( $page_content, 'src_packages', \%pkgs, \@pkgs, + $opts, \@keywords, \&process_src_package, \%archives, \%sect, \%subsect, \%binaries ); } # else unless $opts->{source} @@ -144,20 +163,21 @@ sub process_packages { my $have_exact; if ($keyword && grep { $_ eq $keyword } @$pkgs_list) { $have_exact = 1; - $categories[0]{name} = _g( "Exact hits" ); + $categories[0]{name} = $opts->{cat}->g( "Exact hits" ); - $categories[0]{$target} = [ &$print_func( $keyword, $pkgs->{$keyword}||{}, - map { $_->{$keyword}||{} } @func_args ) ]; + $categories[0]{$target} = [ &$print_func( $opts, $keyword, + $pkgs->{$keyword}||{}, + map { $_->{$keyword}||{} } @func_args ) ]; @$pkgs_list = grep { $_ ne $keyword } @$pkgs_list; } if (@$pkgs_list && (($opts->{searchon} ne 'names') || !$opts->{exact})) { my %cat; - $cat{name} = _g( 'Other hits' ) if $have_exact; + $cat{name} = $opts->{cat}->g( 'Other hits' ) if $have_exact; $cat{packages} = []; foreach my $pkg (@$pkgs_list) { - push @{$cat{$target}}, &$print_func( $pkg, $pkgs->{$pkg}||{}, + push @{$cat{$target}}, &$print_func( $opts, $pkg, $pkgs->{$pkg}||{}, map { $_->{$pkg}||{} } @func_args ); } push @categories, \%cat; @@ -169,7 +189,8 @@ sub process_packages { } sub process_package { - my ($pkg, $pkgs, $provided_by, $archives, $sect, $subsect, $desc) = @_; + my ($opts, $pkg, $pkgs, $provided_by, + $archives, $sect, $subsect, $desc) = @_; my %pkg = ( pkg => $pkg, suites => [] ); @@ -181,9 +202,22 @@ sub process_package { my @versions = version_sort keys %{$pkgs->{$suite}}; $suite{section} = $sect->{$suite}{$versions[0]}; $suite{subsection} = $subsect->{$suite}{$versions[0]}; - $suite{desc} = $desc->{$suite}{$versions[0]}; + my $desc_md5 = $desc->{$suite}{$versions[0]}[0]; + $suite{desc} = $desc->{$suite}{$versions[0]}[1]; $suite{versions} = []; - + + my $trans_desc = $desctrans{$desc_md5}; + my %sdescs; + if ($trans_desc) { + my %trans_desc = split /\000|\001/, $trans_desc; + while (my ($l, $d) = each %trans_desc) { + $d =~ s/\n.*//os; + + $sdescs{$l} = $d; + } + $suite{trans_desc} = \%sdescs; + } + foreach my $v (@versions) { my %version; $version{version} = $v; @@ -198,7 +232,7 @@ sub process_package { $suite{providers} = $p; } } elsif (my $p = $provided_by->{$suite}) { - $suite{desc} = _g('Virtual package'); + $suite{desc} = $opts->{cat}->g('Virtual package'); $suite{providers} = $p; } push @{$pkg{suites}}, \%suite if $suite{versions} || $suite{providers}; @@ -208,7 +242,7 @@ sub process_package { } sub process_src_package { - my ($pkg, $pkgs, $archives, $sect, $subsect, $binaries) = @_; + my ($opts, $pkg, $pkgs, $archives, $sect, $subsect, $binaries) = @_; my %pkg = ( pkg => $pkg, origins => [] );