Accept and ignore: Looks like we got the last chunk together with the server headers...

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

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 4f02230..0acec31 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 1.70 2009/12/31 11:55:27 fabiankeil Exp $
+# $Id: privoxy-log-parser.pl,v 1.80 2010/07/21 14:40:09 fabiankeil Exp $
 #
 # TODO:
 #       - LOG_LEVEL_CGI, LOG_LEVEL_ERROR, LOG_LEVEL_WRITE content highlighting
@@ -44,7 +44,7 @@ use warnings;
 use Getopt::Long;
 
 use constant {
-    PRIVOXY_LOG_PARSER_VERSION => '0.5',
+    PRIVOXY_LOG_PARSER_VERSION => '0.6',
     # Feel free to mess with these ...
     DEFAULT_BACKGROUND => 'black',  # Choose registered colour (like 'black')
     DEFAULT_TEXT_COLOUR => 'white', # Choose registered colour (like 'black')
@@ -56,7 +56,6 @@ use constant {
     CLI_OPTION_NO_EMBEDDED_CSS => 0,
     CLI_OPTION_NO_MSECS => 0,
     CLI_OPTION_NO_SYNTAX_HIGHLIGHTING => 0,
-    CLI_OPTION_ERROR_LOG_FILE => '/var/log/privoxy-log.log',
     CLI_OPTION_SHOW_INEFFECTIVE_FILTERS => 0,
     CLI_OPTION_ACCEPT_UNKNOWN_MESSAGES => 0,
     CLI_OPTION_STATISTICS => 0,
@@ -80,6 +79,7 @@ use constant {
     PUNISH_MISSING_HIGHLIGHT_KNOWLEDGE_WITH_DEATH => 1,
 
     LOG_UNPARSED_LINES_TO_EXTRA_FILE => 0,
+    ERROR_LOG_FILE => '/var/log/privoxy-log-parser',
 
     # You better leave these alone unless you know what you're doing.
     COLOUR_RESET      => "\033[0;0m",
@@ -828,9 +828,9 @@ sub handle_loglevel_header ($) {
 
     } elsif ($c =~ m/^New host is: ([^\s]*)\./) {
 
-        # New host is: trac.vidalia-project.net. Crunching Referer: http://www.vidalia-project.net/
-        $c = highlight_matched_host($c, '(?<=New host is: )[^\s]+');
-        $c = highlight_matched_url($c, '(?<=Crunching Referer: )[^\s]+');
+        # New host is: trac.vidalia-project.net. Crunching Referer: http://www.vidalia-project.net/!
+        $c = highlight_matched_host($c, '(?<=New host is: )[^\s]+(?=\.)');
+        $c = highlight_matched_url($c, '(?<=Crunching Referer: )[^\s!]+');
 
     } elsif ($c =~ m/^Text mode enabled by force. (Take cover)!/) {
 
@@ -908,6 +908,7 @@ sub handle_loglevel_header ($) {
           or $c =~ m/^Appended client IP address to/
           or $c =~ m/^Removing 'Connection: close' to imply keep-alive./
           or $c =~ m/^keep-alive support is disabled/
+          or $c =~ m/^Continue hack in da house/
             )
     {
         # XXX: Some of these may need highlighting
@@ -953,6 +954,7 @@ sub handle_loglevel_header ($) {
         # Appended client IP address to X-Forwarded-For: 10.0.0.2, 10.0.0.1
         # Removing 'Connection: close' to imply keep-alive.
         # keep-alive support is disabled. Crunching: Keep-Alive: 300.
+        # Continue hack in da house.
 
     } elsif ($c =~ m/^scanning headers for:/) {
 
@@ -1622,7 +1624,7 @@ sub handle_loglevel_connect ($) {
         # Waiting for up to 4999 bytes from the client.
         $c =~ s@(?<=up to )(\d+)@$h{'Number'}$1$h{'Standard'}@;
 
-    } elsif ($c =~ m/^Looks like we rea/ or
+    } elsif ($c =~ m/^Looks like we / or
              $c =~ m/^Unsetting keep-alive flag/ or
              $c =~ m/^No connections to wait/ or
              $c =~ m/^Complete client request received/ or
@@ -1637,6 +1639,8 @@ sub handle_loglevel_connect ($) {
         # Looks like we reached the end of the last chunk. We better stop reading.
         # Looks like we read the end of the last chunk together with the server \
         #  headers. We better stop reading.
+        # Looks like we got the last chunk together with the server headers. \
+        #  We better stop reading.
         # Unsetting keep-alive flag.
         # No connections to wait for left.
         # Client request arrived in time or the client closed the connection.
@@ -1687,6 +1691,12 @@ sub handle_loglevel_info ($) {
         # Reloading configuration file '/usr/local/etc/privoxy/config'
         $c =~ s@(?<=loading configuration file \')([^\']*)@$h{'file'}$1$h{'Standard'}@;
 
+    } elsif ($c =~ m/^Loading (actions|filter) file: /) {
+
+        # Loading actions file: /usr/local/etc/privoxy/default.action
+        # Loading filter file: /usr/local/etc/privoxy/default.filter
+        $c =~ s@(?<= file: )(.*)$@$h{'file'}$1$h{'Standard'}@;
+
     } elsif ($c =~ m/^exiting by signal/) {
 
         # exiting by signal 15 .. bye
@@ -1744,7 +1754,8 @@ sub handle_loglevel_info ($) {
              $c =~ m/^Malformerd HTTP headers detected and MS IIS5 hack enabled/ or
              $c =~ m/^Invalid \"chunked\" transfer/ or
              $c =~ m/^Support for/ or
-             $c =~ m/^Flushing header and buffers/
+             $c =~ m/^Flushing header and buffers/ or
+             $c =~ m/^Can not resolve/
              ) {
 
         # No logfile configured. Please enable it before reporting any problems.
@@ -1755,6 +1766,7 @@ sub handle_loglevel_info ($) {
         # Support for 'Connection: keep-alive' is experimental, incomplete and\
         #  known not to work properly in some situations.
         # Flushing header and buffers. Stepping back from filtering.
+        # Can not resolve doesnotexist: hostname nor servname provided, or not known
 
     } else {
 
@@ -1872,8 +1884,7 @@ sub gather_loglevel_error_stats ($$) {
 
 sub gather_loglevel_connect_stats ($$) {
 
-    my $c = shift;
-    my $thread = shift;
+    my ($c, $thread) = @_;
     our %thread_data;
     our %stats;
 
@@ -1905,10 +1916,9 @@ sub gather_loglevel_connect_stats ($$) {
     }
 }
 
-sub gather_loglevel_header_stats ($) {
+sub gather_loglevel_header_stats ($$) {
 
-    my $c = shift;
-    my $thread = shift;
+    my ($c, $thread) = @_;
     our %stats;
 
     if ($c =~ m/^A HTTP\/1\.1 response without/ or
@@ -2125,7 +2135,7 @@ sub parse_loop () {
 
 sub stats_loop () {
 
-    my ($day, $time_stamp, $thread, $log_level, $content, $c, $msecs);
+    my ($day, $time_stamp, $msecs, $thread, $log_level, $content);
     my %log_level_handlers = (
          'Re-Filter'         => \&handle_loglevel_ignore,
          'Header'            => \&gather_loglevel_header_stats,
@@ -2148,9 +2158,9 @@ sub stats_loop () {
             $day = $1;
             $time_stamp = $2;
             $msecs = $3 ? $3 : 0;
-            $log_level = $5;
-            $content = $c = $6;
             $thread = $4;
+            $log_level = $5;
+            $content = $6;
 
             if (defined($log_level_handlers{$log_level})) {
 
@@ -2307,7 +2317,7 @@ that didn't modify the content.
 
 [B<--statistics>] Gather various statistics instead of syntax highlighting
 log messages. This is an experimental feature, if the results look wrong
-they very well might be. Also note that the results a pretty much guaranteed
+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<--version>] Print version and exit.