privoxy-log-parser.pl: Let gather_loglevel_clf_stats() tolerate another 'invalid...

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

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 68bafcd..ed574c7 100755 (executable)
--- a/tools/privoxy-log-parser.pl
+++ b/tools/privoxy-log-parser.pl
@@ -1078,6 +1078,17 @@ sub handle_loglevel_re_filter($) {
         $c .= $req{$t}{'content_size_change'} . $h{'Standard'} . ")";
         $content = $c;
 
+    } elsif ($c =~ m/^filtering request body from client /) {
+
+        # filtering request body from client 127.0.0.1 (size 958) with 'null-filter' produced 0 hits (new size 958).
+
+        $c =~ s@(?<=from client )([^\s]+)@$h{'ip-address'}$1$h{'Standard'}@;
+        $c =~ s@(?<=\(size )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@([^\s]+?)(\'? produced)@$h{'filter'}$1$h{'Standard'}$2@;
+        $c =~ s@(?<=\(new size )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=produced )(\d+)(?= hits)@$h{'Number'}$1$h{'Standard'}@;
+        $content = $c;
+
   } elsif ($c =~ /\.{3}$/
         and $c =~ m/^(?:re_)?filtering \'?(.*?)\'? \(size (\d*)\) with (?:filter )?\'?([^\s]*?)\'? ?\.{3}$/) {
 
@@ -1245,7 +1256,22 @@ sub handle_loglevel_tagging($) {
         $c =~ s@(?<=^Tagger \')([^\']*)@$h{'tagger'}$1$h{'Standard'}@;
         $c =~ s@(?<=added tag \')([^\']*)@$h{'tag'}$1$h{'Standard'}@;
         $c =~ s@(?<=Action bits )(updated)@$h{'action-bits-update'}$1$h{'Standard'}@;
+
+    } elsif ($c =~ /^Enlisting tag/) {
+
+        # Enlisting tag 'forward-directly' for client 127.0.0.1.
+
+        $c =~ s@(?<=tag \')([^\']*)@$h{'tag'}$1$h{'Standard'}@;
+        $c = highlight_matched_host($c, '[^\s]+(?=\.$)');
+
+    } elsif ($c =~ /^Client tag/) {
+
+        # Client tag 'forward-directly' matches
+
+        $c =~ s@(?<=tag \')([^\']*)@$h{'tag'}$1$h{'Standard'}@;
+
     }
+
     return $c;
 }
 
@@ -1839,6 +1865,11 @@ sub handle_loglevel_connect($) {
         # Flushed 3153 bytes of request body
         $c =~ s@(?<=Flushed )(\d+)@$h{'Number'}$1$h{'Standard'}@;
 
+    } elsif ($c =~ m/^Complete client request followed by/) {
+
+        # Complete client request followed by 59 bytes of pipelined data received.
+        $c =~ s@(?<=followed by )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
     } elsif ($c =~ m/^Looks like we / or
              $c =~ m/^Unsetting keep-alive flag/ or
              $c =~ m/^No connections to wait/ or
@@ -2078,6 +2109,13 @@ sub handle_loglevel_error($) {
         # Sending data on socket 33 over TLS/SSL failed: no TLS/SSL errors detected
         $c =~ s@(?<=on socket )(\d+)@$h{'Number'}$1$h{'Standard'}@;
 
+    } elsif ($c =~ m/^Chunk size \d+ exceeds buffered data left/) {
+
+        # Chunk size 291 exceeds buffered data left. Already digested 69894 of 69957 buffered bytes.
+        $c =~ s@(?<=size )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=digested )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+        $c =~ s@(?<=of )(\d+)@$h{'Number'}$1$h{'Standard'}@;
+
     }
 
     # XXX: There are probably more messages that deserve highlighting.
@@ -2111,6 +2149,8 @@ sub gather_loglevel_clf_stats($) {
     unless (defined $method) {
         # +0200] "Invalid request" 400 0
         return if ($content =~ m/^[+-]\d{4}\] "Invalid request"/);
+        # +0100] "Failed reading chunked client body" 400 0
+        return if ($content =~ m/^[+-]\d{4}\] "Failed reading chunked client body"/);
         # +0100] "GET https://securepubads.g.doubleclick.net/gampad/ads?gd[...]... [too long, truncated]
         if ($content =~ m/\[too long, truncated\]$/) {
             print("Skipped LOG_LEVEL_CLF message that got truncated by Privoxy. Statistics will be inprecise.\n");