From: Fabian Keil <fk@fabiankeil.de>
Date: Sat, 30 May 2020 09:34:31 +0000 (+0200)
Subject: privoxy-log-parser.pl: Prefer the number of CLF lines to get the total number of... 
X-Git-Tag: v_3_0_29~330
X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=commitdiff_plain;h=c402a37c55bb679f8b5b4a1b6eb3624f43ef5260

privoxy-log-parser.pl: Prefer the number of CLF lines to get the total number of requests

As it works with older Privoxy versions as well.
---

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 90966a07..df28db95 100755
--- a/tools/privoxy-log-parser.pl
+++ b/tools/privoxy-log-parser.pl
@@ -2178,56 +2178,62 @@ sub print_stats() {
     our %cli_options;
     my $new_connections = $stats{requests} - $stats{crunches} - $stats{'reused-connections'};
     my $client_requests_checksum = 0;
+    my $requests_total;
 
     if ($stats{requests_clf} && $stats{requests}
         && $stats{requests_clf} != $stats{requests}) {
         print "Inconsistent request counts: " . $stats{requests} . "/" . $stats{requests_clf} . "\n";
     }
-    if ($stats{requests_clf} && $stats{requests} eq 0) {
-        $stats{requests} = $stats{requests_clf};
-    }
 
-    if ($stats{requests} eq 0) {
+    # To get the total number of requests we can use either the number
+    # of Common-Log-Format lines or the number of "Request:" messages.
+    # We prefer the number of CLF lines if available because using
+    # it works when analysing old log files from Privoxy versions before 3.0.29.
+    # In Privoxy 3.0.28 and earlier "Request:" messages excluded
+    # crunched messages.
+    $requests_total = $stats{requests_clf} ? $stats{requests_clf} : $stats{requests};
+
+    if ($requests_total eq 0) {
         print "No requests yet.\n";
         return;
     }
 
-    print "Client requests total: " . $stats{requests} . "\n";
+    print "Client requests total: " . $requests_total . "\n";
     if ($stats{crunches}) {
-        my $outgoing_requests = $stats{requests} - $stats{crunches};
+        my $outgoing_requests = $requests_total - $stats{crunches};
         print "Crunches: " . $stats{crunches} . " (" .
-            get_percentage($stats{requests}, $stats{crunches}) . ")\n";
+            get_percentage($requests_total, $stats{crunches}) . ")\n";
         print "Blocks: " . $stats{'blocked'} . " (" .
-            get_percentage($stats{requests}, $stats{'blocked'}) . ")\n";
+            get_percentage($requests_total, $stats{'blocked'}) . ")\n";
         print "Fast redirections: " . $stats{'fast-redirections'} . " (" .
-            get_percentage($stats{requests}, $stats{'fast-redirections'}) . ")\n";
+            get_percentage($requests_total, $stats{'fast-redirections'}) . ")\n";
         print "Connection timeouts: " . $stats{'connection-timeout'} . " (" .
-            get_percentage($stats{requests}, $stats{'connection-timeout'}) . ")\n";
+            get_percentage($requests_total, $stats{'connection-timeout'}) . ")\n";
         print "Connection failures: " . $stats{'connection-failure'} . " (" .
-            get_percentage($stats{requests}, $stats{'connection-failure'}) . ")\n";
+            get_percentage($requests_total, $stats{'connection-failure'}) . ")\n";
         print "Outgoing requests: " . $outgoing_requests . " (" .
-            get_percentage($stats{requests}, $outgoing_requests) . ")\n";
+            get_percentage($requests_total, $outgoing_requests) . ")\n";
     } else {
         print "No crunches detected. Is 'debug 1024' enabled?\n";
     }
 
     print "Server keep-alive offers: " . $stats{'server-keep-alive'} . " (" .
-        get_percentage($stats{requests}, $stats{'server-keep-alive'}) . ")\n";
+        get_percentage($requests_total, $stats{'server-keep-alive'}) . ")\n";
     print "New outgoing connections: " . $new_connections . " (" .
-        get_percentage($stats{requests}, $new_connections) . ")\n";
+        get_percentage($requests_total, $new_connections) . ")\n";
     print "Reused connections: " . $stats{'reused-connections'} . " (" .
-        get_percentage($stats{requests}, $stats{'reused-connections'}) .
+        get_percentage($requests_total, $stats{'reused-connections'}) .
         "; server offers accepted: " .
         get_percentage($stats{'server-keep-alive'}, $stats{'reused-connections'}) . ")\n";
     print "Empty responses: " . $stats{'empty-responses'} . " (" .
-        get_percentage($stats{requests}, $stats{'empty-responses'}) . ")\n";
+        get_percentage($requests_total, $stats{'empty-responses'}) . ")\n";
     print "Empty responses on new connections: "
          . $stats{'empty-responses-on-new-connections'} . " (" .
-        get_percentage($stats{requests}, $stats{'empty-responses-on-new-connections'})
+        get_percentage($requests_total, $stats{'empty-responses-on-new-connections'})
         . ")\n";
     print "Empty responses on reused connections: " .
         $stats{'empty-responses-on-reused-connections'} . " (" .
-        get_percentage($stats{requests}, $stats{'empty-responses-on-reused-connections'}) .
+        get_percentage($requests_total, $stats{'empty-responses-on-reused-connections'}) .
         ")\n";
     print "Client connections: " .  $stats{'closed-client-connections'} . "\n";
     if ($stats{'content-size-total'}) {
@@ -2251,7 +2257,7 @@ sub print_stats() {
         printf "Enable --show-complete-request-distribution to get less common numbers as well.\n";
     }
     # Due to log rotation we may not have a complete picture for all the requests
-    printf "Improperly accounted requests: ~%d\n", abs($stats{requests} - $client_requests_checksum);
+    printf "Improperly accounted requests: ~%d\n", abs($requests_total - $client_requests_checksum);
 
     if (exists $stats{method}) {
         print "Method distribution:\n";