Calculate bytes of the traffic of each domain using mongodb map/reduce.

summize-traffic.rb

#!/usr/bin/ruby19
# -*- coding: utf-8 -*-

require 'mongo'

connection = Mongo::Connection.new
db = connection.db('apache-log')
logs = db.collection('log')

map =<<'_FUNC_'
function() {
  var host = this.uri.match(/^http:\/\/([^\/]+)\/.*/)[1];
  emit(host, this.bytes);
}
_FUNC_

reduce =<<_FUNC_
function(key, values) {
  var total = 0;
  values.forEach(function(e) {
    total += e;
  });
  return total;
}
_FUNC_

result = logs.map_reduce(map, reduce, limit: 100)

result.find().sort('value', -1).each do |i|
  puts i.inspect
end