really fix the caching in all cases

[spider.git] / perl / Prefix.pm

diff --git a/perl/Prefix.pm b/perl/Prefix.pm
index 005d3309020ec41cf61093fc8ec33aa0e91c0e60..0e67801d9e8f10ce494db023877ab5abcf67a439 100644 (file)
--- a/perl/Prefix.pm
+++ b/perl/Prefix.pm
@@ -24,28 +24,38 @@ $BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/  || (0,0))
 $main::build += $VERSION;
 $main::branch += $BRANCH;
 
-use vars qw($db  %prefix_loc %pre %cache $lasttime $hits $matchtotal);
+use vars qw($db  %prefix_loc %pre %cache $misses $hits $matchtotal $lasttime);
 
 $db = undef;                                   # the DB_File handle
 %prefix_loc = ();                              # the meat of the info
 %pre = ();                                             # the prefix list
 %cache = ();                                   # a runtime cache of matched prefixes
 $lasttime = 0;                                 # last time this cache was cleared
-$hits = $matchtotal = 1;               # cache stats
+$hits = $misses = $matchtotal = 1;             # cache stats
+
+#my $cachefn = "$main::data/prefix_cache";
 
 sub load
 {
+       # untie every thing
+#      unlink $cachefn;
+       
        if ($db) {
                undef $db;
                untie %pre;
                %pre = ();
                %prefix_loc = ();
+               untie %cache;
        }
-       $db = tie(%pre, "DB_File", undef, O_RDWR|O_CREAT, 0666, $DB_BTREE) or confess "can't tie \%pre ($!)";  
+
+       # tie the main prefix database
+       $db = tie(%pre, "DB_File", undef, O_RDWR|O_CREAT, 0664, $DB_BTREE) or confess "can't tie \%pre ($!)";  
        my $out = $@ if $@;
        do "$main::data/prefix_data.pl" if !$out;
        $out = $@ if $@;
-       #  print Data::Dumper->Dump([\%pre, \%prefix_loc], [qw(pre prefix_loc)]);
+
+       # tie the prefix cache
+#      tie (%cache, "DB_File", $cachefn, O_RDWR|O_CREAT, 0664, $DB_HASH) or confess "can't tie prefix cache to $cachefn $!";
        return $out;
 }
 
@@ -136,16 +146,34 @@ sub next
 sub matchprefix
 {
        my $pref = shift;
+       my @partials;
 
        for (my $i = length $pref; $i; $i--) {
+               $matchtotal++;
                my $s = substr($pref, 0, $i);
-               my @out = get($s);
-               if (isdbg('prefix')) {
-                       my $part = $out[0] || "*";
-                       $part .= '*' unless $part eq '*' || $part eq $s;
-                       dbg("Partial prefix: $pref $s $part" );
-               } 
-               return @out if @out && $out[0] eq $s;
+               push @partials, $s;
+               my $p = $cache{$s};
+               if ($p) {
+                       $hits++;
+                       if (isdbg('prefix')) {
+                               my $percent = sprintf "%.1f", $hits * 100 / $misses;
+                               dbg("Partial Prefix Cache Hit: $s Hits: $hits/$misses of $matchtotal = $percent\%");
+                       }
+                       $cache{$_} = $p for @partials;
+                       return @$p;
+               } else {
+                       $misses++;
+                       my @out = get($s);
+                       if (isdbg('prefix')) {
+                               my $part = $out[0] || "*";
+                               $part .= '*' unless $part eq '*' || $part eq $s;
+                               dbg("Partial prefix: $pref $s $part" );
+                       } 
+                       if (@out && $out[0] eq $s) {
+                               $cache{$_} =  \@out for @partials;
+                               return @out;
+                       } 
+               }
        }
        return ();
 }
@@ -165,19 +193,17 @@ sub extract
        my $p;
        my @parts;
        my ($call, $sp, $i);
-  
+
        # clear out the cache periodically to stop it growing for ever.
-       if ($main::systime - $lasttime >= 15*60) {
+       if ($main::systime - $lasttime >= 20*60) {
                if (isdbg('prefix')) {
-                       my $percent = $hits * 100 / $matchtotal;
-                       dbg("Prefix Cache Cleared, Hits: $hits of $matchtotal = $percent\%") 
+                       my $percent = sprintf "%.1f", $hits * 100 / $misses;
+                       dbg("Prefix Cache Cleared, Hits: $hits/$misses of $matchtotal = $percent\%") ;
                }
-               my $percent = $hits * 100 / $matchtotal;
-               dbg("Prefix Cache Cleared, $percent\% hits") if isdbg('prefix');
                %cache =();
                $lasttime = $main::systime;
                $hits = $matchtotal = 0;
-       } 
+       }
 
 LM:    foreach $call (split /,/, $calls) {
 
@@ -189,14 +215,15 @@ LM:       foreach $call (split /,/, $calls) {
                if ($p) {
                        $hits++;
                        if (isdbg('prefix')) {
-                               my $percent = $hits * 100 / $matchtotal;
-                               dbg("Prefix Cache Hit: $call Hits: $hits of $matchtotal = $percent\%") 
+                               my $percent = sprintf "%.1f", $hits * 100 / $misses;
+                               dbg("Prefix Cache Hit: $call Hits: $hits/$misses of $matchtotal = $percent\%");
                        }
                        push @out, @$p;
                        next;
                } else {
                        @nout =  get($call);
                        if (@nout && $nout[0] eq $call) {
+                               $misses++;
                                $cache{$call} = \@nout;
                                dbg("got exact prefix: $nout[0]") if isdbg('prefix');
                                push @out, @nout;
@@ -217,6 +244,7 @@ LM: foreach $call (split /,/, $calls) {
                        @nout = get($s);
                        if (@nout && $nout[0] eq $s) {
                                dbg("got exact multipart prefix: $call $s") if isdbg('prefix');
+                               $misses++;
                                $cache{$call} = \@nout;
                                push @out, @nout;
                                next;
@@ -236,6 +264,7 @@ LM: foreach $call (split /,/, $calls) {
                                my @try = get($s);
                                if (@try && $try[0] eq $s) {
                                        dbg("got 3 part prefix: $call $s") if isdbg('prefix');
+                                       $misses++;
                                        $cache{$call} = \@try;
                                        push @out, @try;
                                        next;
@@ -258,6 +287,7 @@ LM: foreach $call (split /,/, $calls) {
                                my @try = get($s);
                                if (@try && $try[0] eq $s) {
                                        dbg("got 2 part prefix: $call $s") if isdbg('prefix');
+                                       $misses++;
                                        $cache{$call} = \@try;
                                        push @out, @try;
                                        next;
@@ -272,7 +302,8 @@ LM: foreach $call (split /,/, $calls) {
                if (@parts == 1) {
                        @nout = matchprefix($parts[0]);
                        if (@nout) {
-                               dbg("got prefix: $call ]") if isdbg('prefix');
+                               dbg("got prefix: $call = $nout[0]") if isdbg('prefix');
+                               $misses++;
                                $cache{$call} = \@nout;
                                push @out, @nout;
                                next;
@@ -317,13 +348,16 @@ L1:               for ($n = 0; $n < @parts; $n++) {
                                                dbg("Compound prefix: $try $part" );
                                        }
                                        if (@try && $try eq $try[0]) {
+                                               $misses++;
                                                $cache{$call} = \@try;
                                                push @out, @try;
                                        } else {
+                                               $misses++;
                                                $cache{$call} = \@nout;
                                                push @out, @nout;
                                        }
                                } else {
+                                       $misses++;
                                        $cache{$call} = \@nout;
                                        push @out, @nout;
                                }
@@ -333,11 +367,12 @@ L1:               for ($n = 0; $n < @parts; $n++) {
 
                # we are a pirate!
                @nout = matchprefix('Q');
+               $misses++;
                $cache{$call} = \@nout;
                push @out, @nout;
        }
        
-       if (isdbg('prefix')) {
+       if (isdbg('prefixdata')) {
                my $dd = new Data::Dumper([ \@out ], [qw(@out)]);
                dbg($dd->Dumpxs);
        }