privoxy-log-parser: Add a handler for tagging messages

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

diff --git a/tools/privoxy-log-parser.pl b/tools/privoxy-log-parser.pl
index 7729626..5bc6d79 100755 (executable)
--- a/tools/privoxy-log-parser.pl
+++ b/tools/privoxy-log-parser.pl
@@ -43,7 +43,7 @@ use warnings;
 use Getopt::Long;
 
 use constant {
 use Getopt::Long;
 
 use constant {

-    PRIVOXY_LOG_PARSER_VERSION => '0.9.1',
+    PRIVOXY_LOG_PARSER_VERSION => '0.9.2',

     # Feel free to mess with these ...
     DEFAULT_BACKGROUND => 'black',  # Choose registered colour (like 'black')
     DEFAULT_TEXT_COLOUR => 'white', # Choose registered colour (like 'black')
     # Feel free to mess with these ...
     DEFAULT_BACKGROUND => 'black',  # Choose registered colour (like 'black')
     DEFAULT_TEXT_COLOUR => 'white', # Choose registered colour (like 'black')


@@ -52,6 +52,7 @@ use constant {
 
     CLI_OPTION_DEFAULT_TO_HTML_OUTPUT => 0,
     CLI_OPTION_TITLE => 'Privoxy-Log-Parser in da house',
 
     CLI_OPTION_DEFAULT_TO_HTML_OUTPUT => 0,
     CLI_OPTION_TITLE => 'Privoxy-Log-Parser in da house',

+    CLI_OPTION_KEEP_DATE => 0,

     CLI_OPTION_NO_EMBEDDED_CSS => 0,
     CLI_OPTION_NO_MSECS => 0,
     CLI_OPTION_NO_SYNTAX_HIGHLIGHTING => 0,
     CLI_OPTION_NO_EMBEDDED_CSS => 0,
     CLI_OPTION_NO_MSECS => 0,
     CLI_OPTION_NO_SYNTAX_HIGHLIGHTING => 0,


@@ -107,6 +108,7 @@ my %h_colours;
 my $header_highlight_regex = '';
 
 my $html_output_mode;
 my $header_highlight_regex = '';
 
 my $html_output_mode;

+my $keep_date_mode;

 my $no_msecs_mode; # XXX: should probably be removed
 my $shorten_thread_ids;
 my $line_end;
 my $no_msecs_mode; # XXX: should probably be removed
 my $shorten_thread_ids;
 my $line_end;


@@ -128,6 +130,7 @@ sub prepare_our_stuff() {
         'Re-Filter'     => 'purple',
         Connect         => 'brown',
         Request         => 'light_cyan',
         'Re-Filter'     => 'purple',
         Connect         => 'brown',
         Request         => 'light_cyan',

+        Tagging         => 'purple',

         CGI             => 'light_green',
         Redirect        => 'cyan',
         Error           => 'light_red',
         CGI             => 'light_green',
         Redirect        => 'cyan',
         Error           => 'light_red',


@@ -1225,6 +1228,26 @@ sub handle_loglevel_re_filter($) {
     return $content;
 }
 
     return $content;
 }
 

+sub handle_loglevel_tagging($) {
+
+    my $c = shift;
+
+    if ($c =~ /^Tagger \'([^\']*)\' added tag \'([^\']*)\'/ or
+        $c =~ m/^Adding tag \'([^\']*)\' created by header tagger \'([^\']*)\'/) {
+
+        # Adding tag 'GET request' created by header tagger 'method-man' (XXX: no longer used)
+        # Tagger 'revalidation' added tag 'REVALIDATION-REQUEST'. No action bit update necessary.
+        # Tagger 'revalidation' added tag 'REVALIDATION-REQUEST'. Action bits updated accordingly.
+
+        # XXX: Save tag and tagger
+
+        $c =~ s@(?<=^Tagger \')([^\']*)@$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'}@;
+    }
+    return $c;
+}
+

 sub handle_loglevel_redirect($) {
 
     my $c = shift;
 sub handle_loglevel_redirect($) {
 
     my $c = shift;


@@ -1783,6 +1806,12 @@ sub handle_loglevel_connect($) {
         # 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'}@;
 
         # 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
              $c =~ m/^No connections to wait/ or
     } elsif ($c =~ m/^Looks like we / or
              $c =~ m/^Unsetting keep-alive flag/ or
              $c =~ m/^No connections to wait/ or


@@ -2382,12 +2411,14 @@ sub print_clf_message() {
 sub print_non_clf_message($) {
 
     my $content = shift;
 sub print_non_clf_message($) {
 
     my $content = shift;

+    my $date_string = $keep_date_mode ? $req{$t}{'day'} . ' ' : '';

     my $msec_string = $no_msecs_mode ? '' : '.' . $req{$t}{'msecs'};
     my $line_start = $html_output_mode ? '' : $h{"Standard"};
 
     return if DEBUG_SUPPRESS_LOG_MESSAGES;
 
     print $line_start
     my $msec_string = $no_msecs_mode ? '' : '.' . $req{$t}{'msecs'};
     my $line_start = $html_output_mode ? '' : $h{"Standard"};
 
     return if DEBUG_SUPPRESS_LOG_MESSAGES;
 
     print $line_start

+        . $date_string

         . $time_colours[$time_colour_index % 2]
         . $req{$t}{'time-stamp'}
         . $msec_string
         . $time_colours[$time_colour_index % 2]
         . $req{$t}{'time-stamp'}
         . $msec_string


@@ -2444,6 +2475,7 @@ sub parse_loop() {
         'Fatal error'       => \&handle_loglevel_ignore,
         'Writing'           => \&handle_loglevel_ignore,
         'Received'          => \&handle_loglevel_ignore,
         'Fatal error'       => \&handle_loglevel_ignore,
         'Writing'           => \&handle_loglevel_ignore,
         'Received'          => \&handle_loglevel_ignore,

+        'Tagging'           => \&handle_loglevel_tagging,

         'Actions'           => \&handle_loglevel_ignore,
         'Unknown log level' => \&handle_loglevel_ignore,
     );
         'Actions'           => \&handle_loglevel_ignore,
         'Unknown log level' => \&handle_loglevel_ignore,
     );


@@ -2541,6 +2573,7 @@ sub stats_loop() {
          'Redirect:'          => \&handle_loglevel_ignore,
          'Unknown log level:' => \&handle_loglevel_ignore,
          'Writing:'           => \&handle_loglevel_ignore,
          'Redirect:'          => \&handle_loglevel_ignore,
          'Unknown log level:' => \&handle_loglevel_ignore,
          'Writing:'           => \&handle_loglevel_ignore,

+         'Tagging:'           => \&handle_loglevel_ignore,

     );
 
     while (<>) {
     );
 
     while (<>) {


@@ -2605,6 +2638,7 @@ sub get_cli_options() {
     our %cli_options = (
         'html-output'              => CLI_OPTION_DEFAULT_TO_HTML_OUTPUT,
         'title'                    => CLI_OPTION_TITLE,
     our %cli_options = (
         'html-output'              => CLI_OPTION_DEFAULT_TO_HTML_OUTPUT,
         'title'                    => CLI_OPTION_TITLE,

+        'keep-date'                => CLI_OPTION_KEEP_DATE,

         'no-syntax-highlighting'   => CLI_OPTION_NO_SYNTAX_HIGHLIGHTING,
         'no-embedded-css'          => CLI_OPTION_NO_EMBEDDED_CSS,
         'no-msecs'                 => CLI_OPTION_NO_MSECS,
         'no-syntax-highlighting'   => CLI_OPTION_NO_SYNTAX_HIGHLIGHTING,
         'no-embedded-css'          => CLI_OPTION_NO_EMBEDDED_CSS,
         'no-msecs'                 => CLI_OPTION_NO_MSECS,


@@ -2621,6 +2655,7 @@ sub get_cli_options() {
     GetOptions (
         'html-output'              => \$cli_options{'html-output'},
         'title'                    => \$cli_options{'title'},
     GetOptions (
         'html-output'              => \$cli_options{'html-output'},
         'title'                    => \$cli_options{'title'},

+        'keep-date'                => \$cli_options{'keep-date'},

         'no-syntax-highlighting'   => \$cli_options{'no-syntax-highlighting'},
         'no-embedded-css'          => \$cli_options{'no-embedded-css'},
         'no-msecs'                 => \$cli_options{'no-msecs'},
         'no-syntax-highlighting'   => \$cli_options{'no-syntax-highlighting'},
         'no-embedded-css'          => \$cli_options{'no-embedded-css'},
         'no-msecs'                 => \$cli_options{'no-msecs'},


@@ -2638,6 +2673,7 @@ sub get_cli_options() {
 
    $html_output_mode = cli_option_is_set('html-output');
    $no_msecs_mode = cli_option_is_set('no-msecs');
 
    $html_output_mode = cli_option_is_set('html-output');
    $no_msecs_mode = cli_option_is_set('no-msecs');

+   $keep_date_mode = cli_option_is_set('keep-date');

    $shorten_thread_ids = cli_option_is_set('shorten-thread-ids');
    $line_end = get_line_end();
 }
    $shorten_thread_ids = cli_option_is_set('shorten-thread-ids');
    $line_end = get_line_end();
 }


@@ -2737,6 +2773,9 @@ omitted, ANSI escape sequences are used unless B<--no-syntax-highlighting> is ac
 This option is only intended to make embedding log excerpts in web pages easier.
 It does not escape any input!
 
 This option is only intended to make embedding log excerpts in web pages easier.
 It does not escape any input!
 

+[B<--keep-date>] Don't remove the date when printing highlighted log messages.
+Useful when parsing multiple log files at once.
+

 [B<--no-msecs>] Don't expect milisecond resolution
 
 [B<--no-syntax-highlighting>] Disable syntax-highlighting. Useful when
 [B<--no-msecs>] Don't expect milisecond resolution
 
 [B<--no-syntax-highlighting>] Disable syntax-highlighting. Useful when