#!/usr/bin/env ruby

# analysing our web server log file to see the different statuses returned

# See note (2)
datahome = ARGV[0] ? ARGV[0] : "oldlogs"
filelimit = ARGV[1] ? ARGV[1].to_i : 100
debug = ARGV[2] ? true : false

filecount = 0
accesscount = 0
code_table = {}

Dir.new(datahome).entries.each do |symbol|
                next unless symbol =~ /^ac_/
                debug and STDERR.puts "At #{symbol}"

                fh = File.new "#{datahome}/#{symbol}"

                while fh.gets
                        status = (split)[8]
                        if status !~ /^[12345]\d\d$/
                                # status = "error"
                                # See note (1)
                                status = "400"
                        end
                        code_table[status] = code_table[status] ? code_table[status] + 1 : 1;
                        accesscount += 1;
                end
                filecount += 1
                break if filecount >= filelimit
end

for status in code_table.keys.sort
        percent = "%5.2f" % (100.0 * code_table[status] / accesscount)
        puts "#{status} - #{percent}% #{code_table[status]}"
end

daily = accesscount.to_f / filecount
hourly = daily / 24.0
minutely = hourly / 60.0

puts "#{'%.0f' % daily} accesses per day"
puts "#{'%.1f' % hourly} accesses per hour"
puts "#{'%.2f' % minutely} accesses per minute"

__END__

Notes

1. The split is relatively crude and will get an erroneous status for malformed requests. As a quick fix
I have put them into code 400 (which many will be) - however, more logging work and analysis could be done.

The following code is better ... allows for requests from which the third parameter (http level) has been
left off, and also allows for \" sequences within the request. It still falls over if there is a quoted
string in one of the two hardly-use security fields.

while hit = fh.gets
     hit.gsub!(/\\"/,'^')
     hit =~ /".*?"\s(...)/
     status = $1

2. Program lacks data validation of parameters and presence of files; will exit with nasty error messages
if abused!