#!/usr/bin/env ruby

require File.expand_path('../lib/archive_bot', __FILE__)

require 'yajl'

include ArchiveBot::Redis

class Categorizer
  attr_accessor :ident
  attr_accessor :error_count
  attr_accessor :r1xx
  attr_accessor :r2xx
  attr_accessor :r3xx
  attr_accessor :r4xx
  attr_accessor :r5xx
  attr_accessor :runk

  def reset
    self.r1xx = 0
    self.r2xx = 0
    self.r3xx = 0
    self.r4xx = 0
    self.r5xx = 0
    self.runk = 0
    self.error_count = 0
  end

  def categorize(obj)
    return unless obj['type'] == 'download'

    if obj['is_error']
      self.error_count += 1
    end

    case obj['response_code'].to_i
    when (100...200)
      self.r1xx += 1
    when (200...300)
      self.r2xx += 1
    when (300...400)
      self.r3xx += 1
    when (400...500)
      self.r4xx += 1
    when (500...600)
      self.r5xx += 1
    else
      self.runk += 1
    end
  end
end

categorizer = Categorizer.new

r = make_redis
parser = Yajl::Parser.new
parser.on_parse_complete = categorizer.method(:categorize)

$stdin.each_line do |line|
  job_key = line.chomp

  categorizer.ident = job_key
  categorizer.reset

  # Get our data.
  data = r.hmget(job_key, 'log_key', 'last_analyzed_log_entry')

  log_key = data[0]

  # Get the score of the last processed log entry.
  start = data[1]

  resps = r.zrangebyscore(log_key, "(#{start}", '+inf', :with_scores => true)

  # If there are no responses to process, keep going.
  next unless resps.length > 0

  # Store the score of the most recent unprocessed entry.
  most_recent = resps.last.last
  
  # Classify each response.
  resps.each do |resp, score|
    begin
      parser << resp
    rescue Yajl::ParseError #TODO: Why is this necessary?
      puts job_key
      next
    end
  end

  # Add the counts to the Redis state.
  r.pipelined do
    r.hincrby(job_key, 'r1xx', categorizer.r1xx)
    r.hincrby(job_key, 'r2xx', categorizer.r2xx)
    r.hincrby(job_key, 'r3xx', categorizer.r3xx)
    r.hincrby(job_key, 'r4xx', categorizer.r4xx)
    r.hincrby(job_key, 'r5xx', categorizer.r5xx)
    r.hincrby(job_key, 'runk', categorizer.runk)
    r.hincrby(job_key, 'error_count', categorizer.error_count)
  end

  # Update the entry marker.
  r.hset(job_key, 'last_analyzed_log_entry', most_recent)
end