#!/usr/bin/perl
########################################
# This script reads a google search appliance log file and displays the search terms
# that returned 0 results. The strings displayed are *exactly* how the user entered
# them as their search query.
#
# This script can be combined with the "grablogfile.pl" (www.blakecrosby.com/scripts.html)
# script to download the log file that this perl script parses.
#
#
# Questions about this script can be directed to:
# Blake Crosby (bcrosby@nm.cbc.ca)
#
# The Canadian Broadcasting Corporation does not support nor endorse this script.
#
#
#
# This script was tested on FreeBSD.
# This script was tested on a 1001 model Google Search Appliance
#
#########################################


#################################
# Config section
#
#
# log file filename
# For example: google.log

$logfile = "google.log";

#
# output file name
# For example: zeroresults.txt

$resultfile = "zeroresults.txt";

#### END OF CONFIG
#
#
#Dont edit below this line, unless you know what you are doing.
#


open(LOGFILE,$logfile);

while (<LOGFILE>){
        $line = $_;
        @result = split / /,$line;
        $result[6] =~ /&ip=(.*)&/;
        $ip = $1;
        if ($result[10] eq "0") {
                $result[6] =~ /q=(.*)&site/;
                $stupidreadonly = $1;
                $stupidreadonly =~ s/\%([0-9A-F]{2})/pack('H*',$1)/ges;
                $stupidreadonly =~ s/\+/ /g;
                $finalresult{$stupidreadonly} = 1;

        }
}

open (REPORTFILE,">$resultfile");

print REPORTFILE join("\n", sort (keys(%finalresult)));

close (REPORTFILE);