analysis.rb

#!/usr/bin/env ruby

require 'uri'

# Helper to get hostnames from referrer fields
def get_host(url)
  begin
    URI.parse(url).host.to_s
  rescue StandardError
    ''
  end
end

# Read the log file
data = File.readlines('custom.log').map do |line|
  elems = line.split(' ')
  { datetime: elems.shift + ' ' + elems.shift, ip: elems.shift, url: elems.shift, status: elems.shift, referrer: elems.shift, agent: elems.join(' ') }
end

# Writing to the output file
file = File.new('stats.md', 'w') # this overwrites the old file!

# Write the liquid header and the start of a markdown table
file.write <<-END
---
layout: default
title: 'Site Stats'
---

|Page|Views|Visitors|
|----|-----|--------|
END

# Only care about successful requests to .html pages
urls = data.select { |d| d[:status] == "200" }.map { |d| d[:url] }.uniq.select { |u| u.end_with?('.html') }

# Count number of views and unique viewers for each page, and write them as a row in the file
urls.each do |url|
  views = data.select { |d| d[:url] == url }.size
  unique = data.select { |d| d[:url] == url }.map { |d| d[:ip] + d[:agent] }.uniq.size
  file.write "|#{url}|#{views}|#{unique}|\n"
end

# Setup the next markdown table
file.write <<-END

|Referrer|Total|
|--------|-----|
END

# Find all of the referrers, count the number of valid .html pages they requested
referrers = data.map { |d| get_host(d[:referrer]) }.uniq.map do |ref|
  count = data.select { |d| d[:url].end_with?('.html') and d[:status] == "200" and get_host(d[:referrer]) == ref }.size
  [ref, count]
end

# Write any non-zero referrers as rows in the table
referrers.select { |r| r[1] > 0 }.sort_by { |r| -r[1] }.each { |r| file.write "|#{r[0]}|#{r[1]}|\n" }

# Finish up with a timestamp on the page
file.write <<-END
*Last updated: #{Time.now.strftime("%m/%d/%Y %H:%M")}*
END

file.close