privoxy-log-parser: Highlight 'The client socket 16 has become unusable while the...

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

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 6749543..dce3b31 100755 (executable)
--- a/tools/privoxy-log-parser.pl
+++ b/tools/privoxy-log-parser.pl
@@ -1061,7 +1061,7 @@ sub handle_loglevel_re_filter($) {
                 return '';
         }
 
-        $c =~ s@(?<=\(size )(\d+)\)(?= with)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=\(size )(\d+)@$h{'Number'}$1$h{'Standard'}@;
         $c =~ s@(?<=\(new size )(\d+)@$h{'Number'}$1$h{'Standard'}@;
         $c =~ s@(?<=produced )(\d+)(?= hits)@$h{'Number'}$1$h{'Standard'}@;
 
@@ -1210,6 +1210,12 @@ sub handle_loglevel_re_filter($) {
 
         return '' unless SHOW_FILTER_READIN_IN;
 
+    } elsif ($c =~ m/^Decompression didn't result/) {
+
+        # Decompression didn't result in any content.
+
+        # Nothing to highlight.
+
     } else {
 
         found_unknown_content($content);
@@ -1627,6 +1633,7 @@ sub handle_loglevel_connect($) {
         # Reusing server socket 7 connected to www.privoxy.org. Total requests: 2.
         # Closing server socket 6 connected to d.asset.soup.io. Keep-alive: 0.\
         #  Tainted: 1. Socket alive: 1. Timeout: 60. Configuration file change detected: 0.
+        # Reusing server socket 35 connected to nl.wikipedia.org. Requests already sent: 5.
 
         $c =~ s@(?<= socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
         $c = highlight_matched_host($c, '(?<=for )[^\s]+(?=\.)');
@@ -1634,6 +1641,7 @@ sub handle_loglevel_connect($) {
         for my $number_pattern ('requests', 'Keep-alive', 'Tainted', ' alive', 'Timeout', 'detected') {
             $c = highlight_matched_pattern($c, 'Number', '(?<='. $number_pattern . ': )\d+');
         }
+        $c =~ s@(?<=already sent: )(\d+)@$h{'Number'}$1$h{'Standard'}@;
 
     } elsif ($c =~ m/^Connected to /) {
 
@@ -1754,6 +1762,33 @@ sub handle_loglevel_connect($) {
         # Forwarding 1954 bytes of encrypted POST data
         $c =~ s@(?<=Forwarding )(\d+)@$h{'Number'}$1$h{'Standard'}@;
 
+    } elsif ($c =~ m/^Forwarded the last \d+ bytes/) {
+
+        # Forwarded the last 1954 bytes
+        $c =~ s@(?<=the last )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
+    } elsif ($c =~ m/^Waiting for the next client connection. Currently active threads:/) {
+
+        # Waiting for the next client connection. Currently active threads: 30
+        $c =~ s@(?<=threads: )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
+    } elsif ($c =~ m/^Data arrived in time on client socket/) {
+
+        # Data arrived in time on client socket 6. Requests so far: 3
+        $c =~ s@(?<=client socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=Requests so far: )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
+    } elsif ($c =~ m/^Dropping the client connection on socket/) {
+
+        # Dropping the client connection on socket 71. The server connection has not been established yet.
+        $c =~ s@(?<=on socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
+    } elsif ($c =~ m/^The client socket \d+ has become unusable while the server/) {
+
+        # The client socket 16 has become unusable while the server socket 24 is still open.
+        $c =~ s@(?<=client socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=server socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
     } elsif ($c =~ m/^Looks like we / or
              $c =~ m/^Unsetting keep-alive flag/ or
              $c =~ m/^No connections to wait/ or
@@ -2007,14 +2042,19 @@ sub gather_loglevel_clf_stats($) {
     our %cli_options;
 
     # +0200] "GET https://www.youtube.com/watch?v=JmcA9LIIXWw HTTP/1.1" 200 68004
-    $content =~ m/^[+-]\d{4}\] "(\w+) (.+) (HTTP\/\d\.\d)" (\d+) (\d+)/;
+    # +0200] "VERSION-CONTROL http://p.p/ HTTP/1.1" 200 2787
+    $content =~ m/^[+-]\d{4}\] "([^ ]+) (.+) (HTTP\/\d\.\d)" (\d+) (\d+)/;
     $method       = $1;
     $resource     = $2;
     $http_version = $3;
     $status_code  = $4;
     $size         = $5;
 
+    $stats{requests_clf}++;
+
     unless (defined $method) {
+        # +0200] "Invalid request" 400 0
+        return if ($content =~ m/^[+-]\d{4}\] "Invalid request"/);
         print("Failed to parse: $content\n");
         return;
     }
@@ -2025,12 +2065,11 @@ sub gather_loglevel_clf_stats($) {
     $stats{'http-version'}{$http_version}++;
 
     if ($cli_options{'host-statistics-threshold'} != 0) {
-        $resource =~ m@(?:http[s]://)([^/]+)/?@;
+        $resource =~ m@(?:https?://)?([^/]+)/?@;
         $stats{'hosts'}{$1}++;
     }
     $stats{'content-size-total'} += $size;
     $stats{'status-code'}{$status_code}++;
-    $stats{requests_clf}++;
 }
 
 sub gather_loglevel_request_stats($$) {