Implement prefix search for package name subword matches. Should be way

faster now

diff --git a/cgi-bin/search_packages.pl b/cgi-bin/search_packages.pl
index 4e2225ca25114b8f650ca68e666143078b79b84b..7aa7ffdcfa27122b7a678b81033afb89b1c02ba8 100755 (executable)
--- a/cgi-bin/search_packages.pl
+++ b/cgi-bin/search_packages.pl
@@ -177,6 +177,7 @@ my $search_on_sources = 0;
 
 my $st0 = new Benchmark;
 my @results;
+my $too_much_hits;
 if ($searchon eq 'sourcenames') {
     $search_on_sources = 1;
 }
@@ -187,6 +188,34 @@ my %archs = map { $_ => 1 } @archs;
 
 print "DEBUG: suites=@suites, sections=@sections, archs=@archs<br>" if $debug > 2;
 
+sub read_entry {
+    my ($hash, $key, $results) = @_;
+    my $result = $hash->{$key};
+    foreach (split /\000/, $result) {
+       my @data = split ( /\s/, $_, 7 );
+       print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
+       if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
+           && $sections{$data[2]}) {
+           print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
+           push @$results, [ $key, @data ];
+       }
+    }
+}
+sub read_src_entry {
+    my ($hash, $key, $results) = @_;
+    my $result = $hash->{$key};
+
+    foreach (split /\000/, $result) {
+       my @data = split ( /\s/, $_, 5 );
+       print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
+       if ($suites{$data[0]} && $sections{$data[1]}) {
+           print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
+           push @$results, [ $key, @data ];
+       }
+    }
+}
+
+
 if ($searchon eq 'names') {
 
     $keyword = lc $keyword unless $case_bool;
@@ -195,30 +224,36 @@ if ($searchon eq 'names') {
        or die "couldn't tie DB $DBDIR/packages_small.db: $!";
     
     if ($exact) {
-       my $result = $packages{$keyword};
-       foreach (split /\000/, $result) {
-           my @data = split ( /\s/, $_, 7 );
-           print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
-           if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
-               && $sections{$data[2]}) {
-               print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
-               push @results, [ $keyword, @data ];
-           }
-       }
+       read_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/, $packages{$pkg}) {
-               my @data = split ( /\s/, $_, 7 );
-               print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
-               if ($suites{$data[0]} && ($archs{$data[1]} || $data[1] eq 'all')
-                   && $sections{$data[2]}) {
-                   print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
-                   push @results, [ $pkg , @data ];
+       my ($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: $!";
+       $p_obj->seq( $key, $prefixes, R_CURSOR );
+       do {
+            if ($prefixes =~ /^\001(\d+)/o) {
+                $too_much_hits += $1;
+            } else {
+               print "DEBUG: add word $key<br>" if $debug > 2;
+               $pkgs{$key}++;
+               foreach (split /\000/o, $prefixes) {
+                   print "DEBUG: add word $_$key<br>" if $debug > 2;
+                   $pkgs{$_.$key}++;
                }
            }
+       } while (($p_obj->seq( $key, $prefixes, R_NEXT ) == 0)
+                && (index($key, $keyword) >= 0)
+                && !$too_much_hits
+                && (keys %pkgs < 100));
+        
+        my $no_results = keys %pkgs;
+        if ($too_much_hits || ($no_results >= 100)) {
+           $too_much_hits += $no_results;
+           %pkgs = ( $keyword => 1 );
+       }
+       foreach my $pkg (sort keys %pkgs) {
+           read_entry( \%packages, $pkg, \@results );
        }
     }
 } elsif ($searchon eq 'sourcenames') {
@@ -229,21 +264,13 @@ if ($searchon eq 'names') {
        or die "couldn't tie DB $DBDIR/sources_small.db: $!";
     
     if ($exact) {
-       my $result = $packages{$keyword};
-       foreach (split /\000/, $result) {
-           my @data = split ( /\s/, $_, 5 );
-           print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
-           if ($suites{$data[0]} && $sections{$data[1]}) {
-               print "DEBUG: Using entry ".join( ':', @data)."<br>" if $debug > 2;
-               push @results, [ $keyword, @data ];
-           }
-       }
+       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/, $packages{$pkg}) {
+           foreach (split /\000/, $result) {
                my @data = split ( /\s/, $_, 5 );
                print "DEBUG: Considering entry ".join( ':', @data)."<br>" if $debug > 2;
                if ($suites{$data[0]} && $sections{$data[1]}) {
@@ -276,6 +303,10 @@ if ($format eq 'html') {
     }
 }
 
+if ($too_much_hits) {
+print "<p><strong>Your search was too wide so we will only display exact matches. At least <em>$too_much_hits</em> results have been omitted and will not be displayed. Please consider using a longer keyword or more keywords.</strong></p>";
+}
+
 if (!@results) {
     if ($format eq 'html') {
        my $keyword_esc = uri_escape( $keyword );