ACE/apps/JAWS/clients/WebSTONE/bin/mine-logs.pl

   1 #!/pkg/gnu/bin//perl
   2 #
   3 #  mine-logs.pl:
   4 #       script to transform access logs into WebStone workload
   5 #
   6 #       created 18 December 1995 mblakele@engr.sgi.com
   7 #
   8 # functional map:
   9 #       usage: mine-logs.pl access.log
  10 #
  11 #       1. For each line in the input
  12 #             a. parse the URL and the time, the outcome code, and the size
  13 #             b. if the code is 200, and it's a GET,
  14 #               do we already know about this URL?
  15 #                 i. yes - increment its counter
  16 #                ii. no - create a slot for it, record size,
  17 #                    and set counter=1
  18 #
  19
  20 $debug = 0;
  21 $line_number = 0;
  22
  23 while (<>) {
  24     chomp;
  25
  26     $line_number++;
  27     ($line_number % 1000) || printf STDERR ".";
  28     # parse line
  29     ( $client, $junk1, $junk2, $date, $timezone,
  30       $command, $url, $version, $result_code, $size ) =
  31         split;
  32     # strip some junk
  33     $command =~ s/\"//;
  34     $date =~ s/\[//;
  35
  36     ($debug) && printf STDERR "$client, $date, $command, $url, $result_code, $size\n";
  37
  38     # is it a GET? Did it succeed? (i.e., is the result code 200?)
  39     if (($command eq 'GET') && ($result_code == 200)) {
  40         # is this URL already in the key set?
  41         if (exists $counter{$url}) {
  42             # URL is in key set
  43             ($debug) && printf STDERR "URL $url already in key set: incrementing\n";
  44             $counter{$url}++;
  45             if ($size == $size{$url}) {
  46                 ($debug) && printf STDERR "size mismatch on $url: $size != $size{$url}\n";
  47                 if ($size <=> $size{$url}) { $size{$url} = $size; }
  48             }
  49         }
  50         else {
  51             # URL isn't in key set
  52             ($debug) && printf STDERR "URL $url isn't in key set: adding size $size\n";
  53             $counter{$url} = 1;
  54             $size{$url} = $size;
  55         }
  56         # end if key set
  57     } # end if GET
  58 }
  59 # end of input file
  60 printf STDERR "\n";
  61
  62 # now we print out a workload file
  63
  64 # first, the headline
  65 $date = `date`;
  66 chomp($date);
  67 printf "# WebStone workload file\n# \tgenerated by $0 $date\n#\n";
  68
  69 # next, sort the keys
  70 @sorted_keys = sort by_counter keys(%counter);
  71
  72 # iterate through sorted keys
  73 foreach $key (@sorted_keys) {
  74     # print url, weighting, and (commented) the size in bytes
  75     ($debug) && printf STDERR "printing data for $key\n";
  76     printf "$key\t$counter{$key}\t#$size{$key}\n";
  77 }
  78 # end foreach
  79
  80 # end main
  81
  82 sub
  83 by_counter {
  84     $counter{$b} <=> $counter{$a};
  85 }
  86 # end by_counter
  87
  88 # end mine-logs.pl