#!/usr/bin/perl

$LOG="/usr/local/apache/logs/access_log";

if (!open(LOGFILE, "<$LOG")) {
    print STDERR "Can't open specified log\n";
    exit(1);
}
$totalqueries = 0;
while (<LOGFILE>) {
    # extract the date
    $inputstring = $_;
    if ($_ =~ m/query=(.*) HTTP/){
        $totalqueries++;
        $line = $1;
        m/\[(\d\d\/\w\w\w\/\d\d\d\d):/;
        $thisdate = $1;
        $line = "&" . $line;
        %query = ('FRONTSTUFF', split/&([-\w]+)=/m, $line);
	#foreach $key (sort keys %query) {
	#	print "$key = $query{$key}\n";
	#}

        # clean up the query string
        $cleanq = $query{'FRONTSTUFF'};
	#print "Pre-cleaning: $query{'FRONTSTUFF'}\n";
        # change +'s to spaces and condense big strings of spaces
        $cleanq =~ s/\++/ /g;
	$cleanq =~ s/&subject//;
	$cleanq =~ s/^&//;
        # $cleanq =~ tr/%3D/=/;
	$cleanq =~ s/%3D//;
	$cleanq =~ s/%20/ /g;
	$cleanq =~ s/%60/ /g;
        $cleanq =~ s/%2B/+/gi;
	$cleanq =~ s/%2A/*/gi;
	$cleanq =~ s/%27/'/gi;
        $cleanq =~ s/%3A/:/gi;
        $cleanq =~ s/%2E/./gi;
        $cleanq =~ s/%2F/\//gi;
        $cleanq =~ s/%22/"/gi;
	#print "Cleaned entry: $cleanq\n";

        # tally another hit for this day in history
        $date{$thisdate}++;

        #tally another use of this query form
        $mss{$query{mss}}++;

        #tally another search for these search string
        $searchstrings{$cleanq}++;

        # count the number of search terms and increment the appropriate counter
        @array = split(/ /, $cleanq);
        $wordcount += @array;
        # print "Wordcount: $wordcount\n";
        $termcount{$wordcount}++;
        #print $termcount{0};
        #if ($wordcount > 20 ) {
        #    print $inputstring, "\n*\n", $cleanq;
        #}

        $wordcount = 0;

        #take each word and increment it's counter

        foreach $word (@array){
            $words{$word}++;
        }

#       print "\n";
#       printf STDOUT "%s\n%s\n", $query{mss}, $cleanq;
#       print "*\n";
    }
}

# compute the count for each termcount and average
$total = 0;
$queries = 0;
print "\nNumber of queries with n search terms:\n";
foreach $key (sort numerically keys %termcount) {
    print $key, '=', $termcount{$key}, "\n";
    $total += $termcount{$key} * $key; # gives number of words in this class
    $queries += $termcount{$key};
}
print "Average search terms per query: ", $total/$queries, "\n";

# compute the average number of queries per day

print "\nQueries per day:\n";
$total = 0;
$days = 0;
foreach $key (sort logdatewise keys %date) {
    print $key, '=', $date{$key}, "\n";
    $total += $date{$key};
    $days++;
}
print "Average queries per day: ", $total/$days, "\n";

# print out the top 20 search terms and the number of times they were used
print "\nTop Search Terms: \n";
$count = 0;
$endcount = 500; #change this to get more or less than 20 results
foreach $key (sort { $words{$b} <=> $words{$a} } keys %words) {
    print $key, '=', $words{$key}, "\n";
    $count++;
    if ($count >= $endcount){
        last;
    }
}

# print out the top 20 search phrases and the number of times they were used
print "\nTop Search Phrases: \n";
$count = 0;
$endcount = 500; #change this to get more or less than 20 results
foreach $key (sort { $searchstrings{$b} <=> $searchstrings{$a} } keys %searchstrings) {
    print $key, '=', $searchstrings{$key}, "\n";
    $count++;
    if ($count >= $endcount){
        last;
    }
}

# print out each type of search form and the number of times it was used
print "\nForm Usage: \n";
foreach $key (keys %mss) {
    $cleankey = $key;
    $cleankey =~ s/&amp;/&/gi;
    print $cleankey, " form = ", $mss{$key}, "\n";
}

print "Total queries: ", $totalqueries;

close(LOGFILE);

# a subroutine for sort for sorting numerically
sub numerically { $a <=> $b; }

# a sort compatible subroutine for parsing and comparing dates as found in
# NCSA format web logs

sub logdatewise {
       %months = (
         "Jan" => 1,
         "Feb" => 2,
         "Mar" => 3,
         "Apr" => 4,
         "May" => 5,
         "Jun" => 6,
         "Jul" => 7,
         "Aug" => 8,
         "Sep" => 9,
         "Oct" => 10,
         "Nov" => 11,
         "Dec" => 12,
       );


#        $a = $_[0];
#       $b = $_[1];
        # print $a, $b;
        $a =~ m[(\d\d)/(\w\w\w)/(\d\d\d\d)];
        $Aday = $1;
        $Amonth = $2;
        $Ayear = $3;
        # print "DAY ", $Aday, " MONTH ", $Amonth, " YEAR ", $Ayear;
        $b =~ m/(\d\d)\/(\w\w\w)\/(\d\d\d\d)/;
        $Bday = $1;
        $Bmonth = $2;
        $Byear = $3;

        if ( $Ayear != $Byear ){
             return $Ayear <=> $Byear;
        } elsif ( $Amonth ne $Bmonth ) {
             return $months{$Amonth} <=> $months{$Bmonth};
        } else {return $Aday <=> $Bday;}

    }