#!/usr/bin/env ruby
#
# Author: IKARASHI Yoshinori <yoosee@gmail.com>
# License: GPLv2
# $Id: twitlogger.rb 13 2008-04-21 20:49:11Z yoosee $
#

username = "" # write your twitter id
password = "" # write your twitter password
logdir = "" # logfile destination directory

interval = 60     # limited api access 70 times per 60 minutes
since_margin = 0  # set ~30 if you find some timeline dropping around intervals

require 'open-uri'
require 'uri'
require 'timeout'
require 'zlib'
require 'rexml/document'

TIMELINE_URI="http://twitter.com/statuses/friends_timeline.xml"

def fetch uri, user, password
  opts = {
    "Accept-Encoding"=>"gzip, deflate",
    "User-Agent"=>'Mozilla/6.0 TwitLogger',
    :http_basic_authentication=>[user,password]
  }

  page, meta, content_type, status, message = nil, nil, nil, nil, nil
  begin
    page = timeout(15) {
      URI.parse(uri).open(opts) {|f|
        content_type = f.content_type
        f.read
      }
    }
  rescue OpenURI::HTTPError
    STDERR.puts "[#{Time.now.to_s}] HTTP Error #{$!}"
    return ''
  rescue StandardError, TimeoutError
    STDERR.puts "[#{Time.now.to_s}] Error #{$!}"
    return ''
  end

  if /\A\x1f\x8b/ =~ page
    begin
      page = Zlib::GzipReader.new(StringIO.new(page)).read || ''
    rescue Zlib::Error
      page = '(zlib decode error)'
    end
  end
  return page
end

def parse_xml page
  rexml = REXML::Document.new page
  a = Array.new
  lastmodified = nil
  rexml.elements.each("//statuses/status") do |status|
    begin
      name = status.elements["user/screen_name"].text
      time = Time.parse(status.elements["created_at"].text)
      lastmodified = time if !lastmodified or lastmodified < time
      text = status.elements["text"].text.chomp.gsub(/\n/, " / ")
    rescue
      puts "#{$!}:#{$@}"
    end
    a.push "#{time.strftime("%H:%M:%S")} #{name}: #{text}" 
  end
  return a, lastmodified
end

if username.empty? or password.empty?
  STDERR.puts "Write your twitter id and password in twitlogger.rb"
  exit
end

logfile = Time.now.strftime("%Y%m%d.log")
logger = File.open(logdir + logfile, "a")
last_fetched = nil
last=Array.new
while(1) do
  if logfile != Time.now.strftime("%Y%m%d.log")
    logger.close
    logfile = Time.now.strftime("%Y%m%d.log")
    logger = File.open(logdir + logfile, "a")
  end
  uri = TIMELINE_URI
  uri += "?since=#{URI.encode((last_fetched-since_margin).httpdate)}" if last_fetched
  page = fetch(uri, username, password)
  if !page or page.empty?
    sleep interval
    next
  end
  a, last_fetched_temp = parse_xml(page)
  last_fetched = last_fetched_temp if last_fetched_temp

  logger.puts((a-last).reverse.join("\n")) unless (a-last).empty?
  logger.flush

  # 'last' pools latest 20 items for avoiding duplicate log lines.  
  # (twitter api returns maximum 20 items at once).
  last += a - last
  last = last.last(20) if last.size > 20

  sleep interval
end

logger.close