Re: [logs] perl question relating to log analysis

Previous message: Jean-Luc Leclercq: "Re: [logs] PIX logging"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

loganalysisat_private

	Thanks to Michael Pomraning (<mjpat_private>) for
catching two big bugs and one piece of laziness in the code I posted:

1) I meant to use an ordered hash, but forgot to.  Add the line

use Tie::IxHash;

	after the "use strict" line, and change the definition
of the %regexes hash to begin as

my %regexes;
tie (%regexes, "Tie::IxHash");
%regexes = (

	, or else all of the sorting will be pointless.

2) That sort is in ascending order, which will un-optimize the matches.
Change the sort comparison to be:

$regexes{$b}->{"count"} <=> $regexes{$a}->{"count"}

	.

3) I was lazy and &&-ed together all of the actions in the for loop
rather than using proper flow control, which depending on the action
for a given regex could cause confusing and buggy behaviour.  A better
implementation would be:

   for my $key (keys %regexes) {
      if ($line =~ $regexes{$key}->{"pattern"}) {
         my $error = $regexes{$key}->{"action"}->();
         if (defined ($regexes{$key}->{"error"}) && $error) {
            $regexes{$key}->{"error"}->($error);
         };
         $regexes{$key}->{"count"}++ && next LINE;
      };
   };

	, assuming that you wanted to trap your error and had defined
error routines for your actions.

	Sorry to anyone who might have wasted cycles trying to
get that code to work, and many thanks again to Mike for the catches.

	-- Sweth, who knows better than to post hurried untested
code but foolishly did so anyway.

-- 
Sweth Chandramouli      Idiopathic Systems Consulting
svcat_private      http://www.idiopathic.net/
_______________________________________________
LogAnalysis mailing list
LogAnalysisat_private
http://lists.shmoo.com/mailman/listinfo/loganalysis