X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=lib%2FPackages%2FDoSearchContents.pm;h=b1057f83bd51a084baed256bb4eac4b2630e96e9;hb=a38d98d27880c8fc0f4d98646e318ecc1d39884f;hp=f66ba875a98c7fa54f2ddbb6930c87c6766b4c5a;hpb=dd9f98daf94d68c800bb08d831627a4641c8de4c;p=deb%2Fpackages.git diff --git a/lib/Packages/DoSearchContents.pm b/lib/Packages/DoSearchContents.pm index f66ba87..b1057f8 100644 --- a/lib/Packages/DoSearchContents.pm +++ b/lib/Packages/DoSearchContents.pm @@ -3,7 +3,7 @@ package Packages::DoSearchContents; use strict; use warnings; -use Benchmark; +use Benchmark ':hireswallclock'; use DB_File; use URI::Escape; use HTML::Entities; @@ -12,26 +12,42 @@ our @ISA = qw( Exporter ); our @EXPORT = qw( do_search_contents ); use Deb::Versions; +use Packages::I18N::Locale; use Packages::Search qw( :all ); use Packages::CGI; use Packages::DB; -use Packages::Config qw( $DBDIR $SEARCH_URL $SEARCH_CGI $SEARCH_PAGE - @SUITES @ARCHIVES $ROOT ); +use Packages::Config qw( $DBDIR $SEARCH_URL $SEARCH_PAGE + @SUITES @ARCHIVES @ARCHITECTURES $ROOT ); sub do_search_contents { my ($params, $opts, $html_header, $menu, $page_content) = @_; if ($params->{errors}{keywords}) { - fatal_error( "keyword not valid or missing" ); + fatal_error( _g( "keyword not valid or missing" ) ); } elsif (length($opts->{keywords}) < 2) { - fatal_error( "keyword too short (keywords need to have at least two characters)" ); + fatal_error( _g( "keyword too short (keywords need to have at least two characters)" ) ); + } + if ($params->{errors}{suite}) { + fatal_error( _g( "suite not valid or not specified" ) ); + } + + #FIXME: that's extremely hacky atm + if ($params->{values}{suite}{no_replace}[0] eq 'default') { + $params->{values}{suite}{no_replace} = + $params->{values}{suite}{final} = $opts->{suite} = [ 'stable' ]; + } + + if (@{$opts->{suite}} > 1) { + fatal_error( sprintf( _g( "more than one suite specified for contents search (%s)" ), "@{$opts->{suite}}" ) ); } $$menu = ""; my $keyword = $opts->{keywords}; - my $searchon = $opts->{searchon}; - my $exact = $opts->{exact}; + my $mode = $opts->{mode} || ''; + my $suite = $opts->{suite}[0]; + my $archive = $opts->{archive}[0] ||''; + $Packages::Search::too_many_hits = 0; # for URL construction my $keyword_esc = uri_escape( $keyword ); @@ -41,13 +57,12 @@ sub do_search_contents { # for output my $keyword_enc = encode_entities $keyword || ''; - my $searchon_enc = encode_entities $searchon; - my $suites_enc = encode_entities( join( ', ', @{$params->{values}{suite}{no_replace}} ) ); - my $sections_enc = encode_entities( join( ', ', @{$params->{values}{section}{no_replace}} ) ); - my $archs_enc = encode_entities( join( ', ', @{$params->{values}{arch}{no_replace}} ) ); + my $suites_enc = encode_entities( join( ', ', @{$params->{values}{suite}{no_replace}} ), '&<>"' ); + my $sections_enc = encode_entities( join( ', ', @{$params->{values}{section}{no_replace}} ), '&<>"' ); + my $archs_enc = encode_entities( join( ', ', @{$params->{values}{arch}{no_replace}} ), '&<>"' ); my $st0 = new Benchmark; - my (@results, @non_results); + my (@results); unless (@Packages::CGI::fatal_errors) { @@ -55,30 +70,28 @@ sub do_search_contents { my $kw = lc $keyword; # full filename search is tricky - my $ffn = $searchon eq 'filenames'; - - my $suite = 'stable'; #fixme + my $ffn = $mode eq 'filename'; my $reverses = tie my %reverses, 'DB_File', "$DBDIR/contents/reverse_$suite.db", O_RDONLY, 0666, $DB_BTREE or die "Failed opening reverse DB: $!"; if ($ffn) { - open FILENAMES, '-|', 'fgrep', '--', "$kw", "$DBDIR/contents/filenames_$suite.txt" + open FILENAMES, '-|', 'fgrep', '--', $kw, "$DBDIR/contents/filenames_$suite.txt" or die "Failed opening filename table: $!"; + while () { chomp; - last unless &searchfile(\@results, reverse($_)."/", \$nres, $reverses); + &searchfile(\@results, reverse($_)."/", \$nres, $reverses); + last if $Packages::Search::too_many_hits; } - close FILENAMES; + close FILENAMES or warn "fgrep error: $!\n"; } else { $kw = reverse $kw; # exact filename searching follows trivially: - $kw = "$kw/" if $exact; - - print "ERROR: Exact and fullfilenamesearch don't go along" if $ffn and $exact; + $kw = "$kw/" if $mode eq 'exactfilename'; &searchfile(\@results, $kw, \$nres, $reverses); } @@ -88,31 +101,48 @@ sub do_search_contents { my $st1 = new Benchmark; my $std = timediff($st1, $st0); - debug( "Search took ".timestr($std) ); + debug( "Search took ".timestr($std) ) if DEBUG; } - my $suite_wording = $suites_enc eq "all" ? "all suites" - : "suite(s) $suites_enc"; - my $section_wording = $sections_enc eq 'all' ? "all sections" - : "section(s) $sections_enc"; - my $arch_wording = $archs_enc eq 'any' ? "all architectures" - : "architecture(s) $archs_enc"; - my $wording = $opts->{exact} ? "exact filenames" : "filenames that contain"; - $wording = "paths that end with" if $searchon eq "contents"; - msg( "You have searched for ${wording} $keyword_enc in $suite_wording, $section_wording, and $arch_wording." ); + my $suite_wording = sprintf(_g("suite %s"), $suites_enc ); + my $section_wording = $sections_enc eq 'all' ? _g("all sections") + : sprintf(_g("section(s) %s"), $sections_enc ); + my $arch_wording = $archs_enc eq 'any' ? _g("all architectures") + : sprintf(_g("architecture(s) %s"), $archs_enc ); + my $wording = _g("paths that end with"); + if ($mode eq 'filename') { + $wording = _g("files named"); + } elsif ($mode eq 'exactfilename') { + $wording = _g("filenames that contain"); + } + msg( sprintf( _g("You have searched for %s %s in %s, %s, and %s." ), + $wording, $keyword_enc, + $suite_wording, $section_wording, $arch_wording ) ); - if ($Packages::Search::too_many_hits) { - error( "Your search was too wide so we will only display exact matches. At least $Packages::Search::too_many_hits results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords." ); + if ($mode ne 'filename') { + msg( ''filename'}). + "\">"._g("Search within filenames").""); } - - $$page_content = ''; - if (!@Packages::CGI::fatal_errors && !@results) { - $$page_content .= "No results"; + if ($mode ne 'exactfilename') { + msg( ''exactfilename'}). + "\">"._g("Search exact filename").""); } + if ($mode eq 'exactfilename' || $mode eq 'filename') { + msg( 'undef}). + "\">"._g("Search for paths ending with").""); + } + + msg( _g("Search in other suite:")." ". + join( ' ', map { '[$_}). + "\">$_]" } @SUITES ) ); - %$html_header = ( title => 'Package Contents Search Results' , - lang => 'en', - title_tag => 'Debian Package Contents Search Results', + if ($Packages::Search::too_many_hits) { + error( _g( "Your search was too wide so we will only display only the first about 100 matches. Please consider using a longer keyword or more keywords." ) ); + } + + %$html_header = ( title => _g( 'Package Contents Search Results' ), + lang => $opts->{lang}, + title_tag => _g( 'Debian Package Contents Search Results' ), print_title => 1, print_search_field => 'packages', search_field_values => { @@ -126,11 +156,65 @@ sub do_search_contents { }, ); - if (@results) { - $$page_content .= scalar @results . " results displayed:
"; - foreach (@results) { - $$page_content .= "$_
\n"; + $$page_content = ''; + my (%results,%archs); + foreach my $result (sort { $a->[0] cmp $b->[0] } @results) { + my $file = shift @$result; + my %pkgs; + foreach (@$result) { + my ($pkg, $arch) = split /:/, $_; + next unless $opts->{h_archs}{$arch}; + $pkgs{$pkg}{$arch}++; + $archs{$arch}++ unless $arch eq 'all'; + } + next unless keys %pkgs; + $results{$file} = \%pkgs; + } + my @all_archs = keys %archs; + @all_archs = @ARCHITECTURES unless @all_archs; + debug( "all_archs = @all_archs", 1 ) if DEBUG; + msg(_g("Limit search to a specific architecture:")." ". + join( ' ', map { '[$_}). + "\">$_]" } @all_archs ) ) + unless (@{$opts->{arch}} == 1) || (@all_archs == 1); + msg(sprintf(_g('Search in all architectures'), + make_search_url('',"keywords=$keyword_esc",{arch=>undef}))) + if @{$opts->{arch}} == 1; + + if (!@Packages::CGI::fatal_errors && !keys(%results)) { + error( _g( "Nothing found" ) ); + } + + if (keys %results) { + $$page_content .= "

".sprintf( _g( 'Found %s results' ), + scalar keys %results )."

"; + $$page_content .= '
\n"; + foreach my $file (sort keys %results) { + my $file_enc = encode_entities($file); + $file_enc =~ s#(\Q$keyword_enc\E)#$1#g; + $$page_content .= "\n"; } + $$page_content .= '\n" if @results > 20; + $$page_content .= '
'._g('File').''._g('Packages') + ."
/$file_enc"; + my @pkgs; + foreach my $pkg (sort keys %{$results{$file}}) { + my $arch_str = ''; + my @archs = keys %{$results{$file}{$pkg}}; + unless ($results{$file}{$pkg}{all} || + (@archs == @all_archs)) { + if (@archs < @all_archs/2) { + $arch_str = ' ['.join(' ',sort @archs).']'; + } else { + $arch_str = ' ['._g('not').' '. + join(' ', grep { !$results{$file}{$pkg}{$_} } @all_archs).']'; + } + } + push @pkgs, "$suite})."\">$pkg$arch_str"; + } + $$page_content .= join( ", ", @pkgs); + $$page_content .= "
'._g('File').''._g('Packages')."
'; } } # sub do_search_contents @@ -139,6 +223,7 @@ sub searchfile my ($results, $kw, $nres, $reverses) = @_; my ($key, $value) = ($kw, ""); + debug( "searchfile: kw=$kw", 1 ) if DEBUG; for (my $status = $reverses->seq($key, $value, R_CURSOR); $status == 0; $status = $reverses->seq( $key, $value, R_NEXT)) { @@ -146,14 +231,14 @@ sub searchfile # FIXME: what's the most efficient "is prefix of" thingy? We only want to know # whether $kw is or is not a prefix of $key last unless index($key, $kw) == 0; + debug( "found $key", 2 ) if DEBUG; my @hits = split /\0/o, $value; - push @$results, reverse($key)." is found in @hits"; + push @$results, [ scalar reverse($key), @hits ]; last if ($$nres)++ > 100; } -# FIXME: use too_many_hits - return $$nres<100; + $Packages::Search::too_many_hits += $$nres - 100 if $$nres > 100; }