Consistently recommending socks5 in the Tor entry and mention its advantage compared...

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

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 98e3d01..12f8af8 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 1.69 2009/12/31 11:54:55 fabiankeil Exp $
+# $Id: privoxy-log-parser.pl,v 1.228 2010/02/13 15:25:10 fk 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",
@@ -784,25 +784,7 @@ sub handle_loglevel_header ($) {
 
     if ($c =~ /^scan:/) {
 
-        if ($c =~ m/^scan: ((\w+) (.+) (HTTP\/\d\.\d))/) {
-
-            # Client request line
-            # Save for statistics (XXX: Not implemented yet)
-            $req{$t}{'method'} = $2;
-            $req{$t}{'destination'} = $3;
-            $req{$t}{'http-version'} = $4;
-
-            $c = highlight_request_line($1);
-
-        } elsif ($c =~ m/^(scan: )((?:HTTP\/\d\.\d|ICY) (\d+) (.*))/) {
-
-            # Server response line
-            $req{$t}{'response_line'} = $2;
-            $req{$t}{'status_code'} = $3;
-            $req{$t}{'status_message'} = $4;
-            $c = $1 . highlight_response_line($req{$t}{'response_line'});
-
-        } elsif ($c =~ m/^scan: ((?>[^:]+)):/) {
+        if ($c =~ m/^scan: ([^: ]+):/) {
 
             # Register new headers
             # scan: Accept: image/png,image/*;q=0.8,*/*;q=0.5
@@ -819,6 +801,23 @@ sub handle_loglevel_header ($) {
                 update_header_highlight_regex($header);
             }
 
+        } elsif ($c =~ m/^scan: ((\w+) (.+) (HTTP\/\d\.\d))/) {
+
+            # Client request line
+            # Save for statistics (XXX: Not implemented yet)
+            $req{$t}{'method'} = $2;
+            $req{$t}{'destination'} = $3;
+            $req{$t}{'http-version'} = $4;
+
+            $c = highlight_request_line($1);
+
+        } elsif ($c =~ m/^(scan: )((?:HTTP\/\d\.\d|ICY) (\d+) (.*))/) {
+
+            # Server response line
+            $req{$t}{'response_line'} = $2;
+            $req{$t}{'status_code'} = $3;
+            $req{$t}{'status_message'} = $4;
+            $c = $1 . highlight_response_line($req{$t}{'response_line'});
         }
 
     } elsif ($c =~ m/^Crunching (?:server|client) header: .* \(contains: ([^\)]*)\)/) {
@@ -829,9 +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)!/) {
 
@@ -1745,7 +1744,8 @@ 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.
@@ -1756,6 +1756,7 @@ 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 {
 
@@ -1873,8 +1874,7 @@ sub gather_loglevel_error_stats ($$) {
 
 sub gather_loglevel_connect_stats ($$) {
 
-    my $c = shift;
-    my $thread = shift;
+    my ($c, $thread) = @_;
     our %thread_data;
     our %stats;
 
@@ -1906,10 +1906,9 @@ 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
@@ -2126,7 +2125,7 @@ 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,
@@ -2149,9 +2148,9 @@ 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})) {
 
@@ -2308,7 +2307,7 @@ 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.