I always look at my Apache server log files from the command line. It always bothered me to see "GET /robots.txt" contaminating the logs. It was frustrating trying to visually determine which were crawlers and which were actual users. So I wrote this little utility, which filters out requests were made from IP addresses which grab "robots.txt". I suspect there are GUI log parsers that might provide the same functionality, but 1) i don't need something that heavy, 2) I like to code, 3) imageekwhaddyawant.
#!/usr/bin/perl
use strict;
use warnings;
# Apache logs robots filter-outer
# Author: Gary C. Wang
# Contact: gary@quartertone.net
# Website: www.quartertone.net
# Filename: norobotlog
#
# Usage: norobotlog [logfile_name]
#
# This script parses Apache log files and
# filters out entries from IP addresses
# that request "robots.txt" file, commonly
# associated with webcrawlers and site indexers.
# Prior to usage, check regexp to make sure it matches your log format
+.
# My log format is something like:
# 192.168.0.xx - - [11/Jul/2004:22:25:22 -0400] "GET /robots.txt HTTP
+/1.0" 200 78
my %robots;
my $ip_ptn = '((\d{1,3}\.){3}\d{1,3})'; # this regexp matches IP addre
+sses
my @file = <>; #file from stdin
# First, find out which IPs are associated with crawlers
foreach (@file) {
# ----- Adjust this pattern to match your log file -----
$robots{$1} ++ if m/^$ip_ptn .+?robots\.txt/;
}
# Then weed those out, printing only the ones that do not request robo
+ts.txt
foreach (@file) {
if (m/$ip_ptn /) {
print if ! defined $robots{$1};
}
}
It may be a good idea to include other bot patterns, like the Googlebot and other search engine bots. Otherwise, this can be simplified to an egrep command, which is what I'd use anyway.
