Remove a duplicate

[privoxy.git] / tools / privoxy-log-parser.pl

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 0fdc1d2..812da0b 100755 (executable)
--- a/tools/privoxy-log-parser.pl
+++ b/tools/privoxy-log-parser.pl
@@ -8,7 +8,7 @@
 #
 # http://www.fabiankeil.de/sourcecode/privoxy-log-parser/
 #
-# $Id: privoxy-log-parser.pl,v 1.89 2010/08/28 13:20:23 fabiankeil Exp $
+# $Id: privoxy-log-parser.pl,v 1.95 2010/10/23 08:01:36 fabiankeil Exp $
 #
 # TODO:
 #       - LOG_LEVEL_CGI, LOG_LEVEL_ERROR, LOG_LEVEL_WRITE content highlighting
@@ -61,6 +61,7 @@ use constant {
     CLI_OPTION_SHOW_INEFFECTIVE_FILTERS => 0,
     CLI_OPTION_ACCEPT_UNKNOWN_MESSAGES => 0,
     CLI_OPTION_STATISTICS => 0,
+    CLI_OPTION_URL_STATISTICS_THRESHOLD => 0,
 
     SUPPRESS_SUCCEEDED_FILTER_ADDITIONS => 1,
     SHOW_SCAN_INTRO => 0,
@@ -1860,6 +1861,15 @@ sub gather_loglevel_crunch_stats ($$) {
 
     $stats{requests}++;
     $stats{crunches}++;
+
+    if ($c =~ m/^Redirected:/) {
+        # Redirected: http://www.example.org/http://p.p/
+        $stats{'fast-redirections'}++;
+
+    } elsif ($c =~ m/^Blocked:/) {
+        # Blocked: blogger.googleusercontent.com:443
+        $stats{'blocked'}++;
+    }
 }
 
 
@@ -1959,6 +1969,7 @@ sub get_percentage ($$) {
 sub print_stats () {
 
     our %stats;
+    our %cli_options;
     my $new_connections = $stats{requests} - $stats{crunches} - $stats{'reused-connections'};
     my $outgoing_requests = $stats{requests} - $stats{crunches};
 
@@ -1970,6 +1981,10 @@ sub print_stats () {
     print "Client requests total: " . $stats{requests} . "\n";
     print "Crunches: " . $stats{crunches} . " (" .
         get_percentage($stats{requests}, $stats{crunches}) . ")\n";
+    print "Blocks: " . $stats{'blocked'} . " (" .
+        get_percentage($stats{requests}, $stats{'blocked'}) . ")\n";
+    print "Fast redirections: " . $stats{'fast-redirections'} . " (" .
+        get_percentage($stats{requests}, $stats{'fast-redirections'}) . ")\n";
     print "Outgoing requests: " . $outgoing_requests . " (" .
         get_percentage($stats{requests}, $outgoing_requests) . ")\n";
     print "Server keep-alive offers: " . $stats{'server-keep-alive'} . " (" .
@@ -1977,7 +1992,9 @@ sub print_stats () {
     print "New outgoing connections: " . $new_connections . " (" .
         get_percentage($stats{requests}, $new_connections) . ")\n";
     print "Reused connections: " . $stats{'reused-connections'} . " (" .
-        get_percentage($stats{requests}, $stats{'reused-connections'}) . ")\n";
+        get_percentage($stats{requests}, $stats{'reused-connections'}) .
+        "; server offers accepted: " .
+        get_percentage($stats{'server-keep-alive'}, $stats{'reused-connections'}) . ")\n";
     print "Empty responses: " . $stats{'empty-responses'} . " (" .
         get_percentage($stats{requests}, $stats{'empty-responses'}) . ")\n";
     print "Empty responses on new connections: "
@@ -2001,9 +2018,18 @@ sub print_stats () {
     foreach my $http_version (sort {$stats{'http-version'}{$b} <=> $stats{'http-version'}{$a}} keys %{$stats{'http-version'}}) {
         printf "%d : %s\n",  $stats{'http-version'}{$http_version}, $http_version;
     }
-    print "Requested ressources:\n";
-    foreach my $ressource (sort {$stats{'ressource'}{$b} <=> $stats{'ressource'}{$a}} keys %{$stats{'ressource'}}) {
-        printf "%d : %s\n", $stats{'ressource'}{$ressource}, $ressource;
+
+    if ($cli_options{'url-statistics-threshold'} == 0) {
+        print "URL statistics are disabled. Increase --url-statistics-threshold to enable them.\n";
+    } else {
+        print "Requested URLs:\n";
+        foreach my $ressource (sort {$stats{'ressource'}{$b} <=> $stats{'ressource'}{$a}} keys %{$stats{'ressource'}}) {
+            if ($stats{'ressource'}{$ressource} < $cli_options{'url-statistics-threshold'}) {
+                print "Skipped statistics for URLs below the treshold.\n";
+                last;
+            }
+            printf "%d : %s\n", $stats{'ressource'}{$ressource}, $ressource;
+        }
     }
 }
 
@@ -2239,6 +2265,7 @@ sub get_cli_options () {
         'show-ineffective-filters' => CLI_OPTION_SHOW_INEFFECTIVE_FILTERS,
         'accept-unknown-messages'  => CLI_OPTION_ACCEPT_UNKNOWN_MESSAGES,
         'statistics'               => CLI_OPTION_STATISTICS,
+        'url-statistics-threshold'  => CLI_OPTION_URL_STATISTICS_THRESHOLD,
     );
 
     GetOptions (
@@ -2251,6 +2278,7 @@ sub get_cli_options () {
         'show-ineffective-filters' => \$cli_options{'show-ineffective-filters'},
         'accept-unknown-messages'  => \$cli_options{'accept-unknown-messages'},
         'statistics'               => \$cli_options{'statistics'},
+        'url-statistics-threshold=s'=> \$cli_options{'url-statistics-threshold'},
         'version'                  => sub { VersionMessage && exit(0) },
         'help'                     => \&help,
    ) or exit(1);
@@ -2278,6 +2306,7 @@ Options and their default values if they have any:
     [--shorten-thread-ids]
     [--show-ineffective-filters]
     [--statistics]
+    [--url-statistics-threshold $cli_options{'url-statistics-threshold'}]
     [--title $cli_options{'title'}]
     [--version]
 see "perldoc $0" for more information
@@ -2316,7 +2345,8 @@ B<privoxy-log-parser> - A parser and syntax-highlighter for Privoxy log messages
 
 B<privoxy-log-parser> [B<--accept-unknown-messages>] [B<--html-output>]
 [B<--no-msecs>] [B<--no-syntax-higlighting>] [B<--statistics>]
-[B<--shorten-thread-ids>] [B<--show-ineffective-filters>] [B<--version>]
+[B<--shorten-thread-ids>] [B<--show-ineffective-filters>]
+[B<--url-statistics-threshold>] [B<--version>]
 
 =head1 DESCRIPTION
 
@@ -2367,6 +2397,10 @@ log messages. This is an experimental feature, if the results look wrong
 they very well might be. Also note that the results are pretty much guaranteed
 to be incorrect if Privoxy and Privoxy-Log-Parser aren't in sync.
 
+[B<--url-statistics-threshold>] Only show the request count for a ressource
+if it's above or equal to the given threshold. If the threshold is 0, URL
+statistics are disabled.
+
 [B<--version>] Print version and exit.
 
 =head1 EXAMPLES