X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=cgi-bin%2Fsearch_packages.pl;h=2e141bb5e6f3374daedc6ba192a9850a759e0717;hb=186b663bf3f69e02137d247021400e5a3f90d666;hp=f6df10c95635a882bc87a825eaf698266ae0777c;hpb=7837241d743061df9f58680a9b786e48d3365a48;p=deb%2Fpackages.git diff --git a/cgi-bin/search_packages.pl b/cgi-bin/search_packages.pl index f6df10c..2e141bb 100755 --- a/cgi-bin/search_packages.pl +++ b/cgi-bin/search_packages.pl @@ -26,12 +26,11 @@ use Deb::Versions; 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 @@ -44,12 +43,18 @@ my %ARCHITECTURES = map { $_ => 1 } @ARCHITECTURES; $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 $debug > 1; # If you want, just print out a list of all of the variables and exit. @@ -125,6 +130,7 @@ my @sections = @{$params{values}{section}{final}}; 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}}; @@ -177,7 +183,7 @@ my $search_on_sources = 0; my $st0 = new Benchmark; my @results; -my $too_much_hits; +my $too_many_hits; if ($searchon eq 'sourcenames') { $search_on_sources = 1; } @@ -186,7 +192,8 @@ my %suites = map { $_ => 1 } @suites; my %sections = map { $_ => 1 } @sections; my %archs = map { $_ => 1 } @archs; -print "DEBUG: suites=@suites, sections=@sections, archs=@archs
" if $debug > 2; +print "DEBUG: suites=@suites, sections=@sections, archs=@archs
" + if $debug > 2; sub read_entry { my ($hash, $key, $results) = @_; @@ -214,78 +221,56 @@ sub read_src_entry { } } } +sub do_names_search { + my ($keyword, $file, $postfix_file, $read_entry, $opts) = @_; + my @results; - -if ($searchon eq 'names') { - - $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) { - read_entry( \%packages, $keyword, \@results ); + if ($opts->{exact}) { + &$read_entry( \%packages, $keyword, \@results ); } else { my ($key, $prefixes) = ($keyword, ''); my %pkgs; - my $p_obj = tie my %pref, 'DB_File', "$DBDIR/package_postfixes.db", O_RDONLY, 0666, $DB_BTREE - or die "couldn't tie postfix db $DBDIR/package_postfixes.db: $!"; + 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, $prefixes, R_CURSOR ); - do { + while (index($key, $keyword) >= 0) { if ($prefixes =~ /^\001(\d+)/o) { - $too_much_hits += $1; + $too_many_hits += $1; } else { - print "DEBUG: add word $key
" if $debug > 2; - $pkgs{$key}++; foreach (split /\000/o, $prefixes) { + $_ = '' if $_ eq '^'; print "DEBUG: add word $_$key
" if $debug > 2; $pkgs{$_.$key}++; } } - } while (($p_obj->seq( $key, $prefixes, R_NEXT ) == 0) - && (index($key, $keyword) >= 0) - && !$too_much_hits - && (keys %pkgs < 100)); + 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_much_hits || ($no_results >= 100)) { - $too_much_hits += $no_results; + if ($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) { - read_src_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/, $result) { - my @data = split ( /\s/, $_, 5 ); - print "DEBUG: Considering entry ".join( ':', @data)."
" if $debug > 2; - if ($suites{$data[0]} && $sections{$data[1]}) { - print "DEBUG: Using entry ".join( ':', @data)."
" if $debug > 2; - push @results, [ $pkg , @data ]; - } - } + &$read_entry( \%packages, $pkg, \@results ); } } -} else { + return \@results; +} +sub do_fulltext_search { + my ($keword, $file, $mapping, $lookup, $read_entry, $opts) = @_; + my @results; my @lines; my $regex; - if ($case_bool) { - if ($exact) { + if ($opts->{case_bool}) { + if ($opts->{exact}) { $regex = qr/\b\Q$keyword\E\b/o; } else { $regex = qr/\Q$keyword\E/o; @@ -298,7 +283,8 @@ if ($searchon eq 'names') { } } - open DESC, '<', "$DBDIR/descriptions.txt" or die "couldn't open $DBDIR/descriptions.txt: $!"; + open DESC, '<', "$DBDIR/$file" + or die "couldn't open $DBDIR/$file: $!"; while () { $_ =~ $regex or next; print "DEBUG: Matched line $.
" if $debug > 2; @@ -306,10 +292,10 @@ if ($searchon eq 'names') { } close DESC; - 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 $obj = tie my %did2pkg, 'DB_File', "$DBDIR/descriptions_packages.db", O_RDONLY, 0666, $DB_BTREE - or die "couldn't tie DB $DBDIR/descriptions_packages.db: $!"; + tie my %packages, 'DB_File', "$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 $l (@lines) { @@ -321,8 +307,27 @@ if ($searchon eq 'names') { } } foreach my $pkg (keys %tmp_results) { - read_entry( \%packages, $pkg, \@results ); + &$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 $st1 = new Benchmark; @@ -346,8 +351,8 @@ if ($format eq 'html') { } } -if ($too_much_hits) { -print "

Your search was too wide so we will only display exact matches. At least $too_much_hits results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords.

"; +if ($too_many_hits) { +print "

Your search was too wide so we will only display exact matches. At least $too_many_hits results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords.

"; } if (!@results) { @@ -440,7 +445,7 @@ unless ($search_on_sources) { $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) { @@ -610,5 +615,11 @@ print < END +my $pete = new Benchmark; +my $petd = timediff($pete, $pet0); +print "Total page evaluation took ".timestr($petd)."
" + if $debug_allowed; print $input->end_html; } + +# vim: ts=8 sw=4