# A model for accessing the mint_visit table that contains all the hit stats.
class Mint < ActiveRecord::Base

  # If you are using the RSS/Atom importer, put this line in your environment:
  #
  #  Mint.log_file_path = /home/topfunky/logs/your_site.com/http/access.log.0
  #
  # NOTE: Rails seems to load the individual environment files before the rest of 
  # environment.rb. I have to put this at the end of environment.rb instead
  # of in production.rb.
  cattr_accessor :log_file_path

  # Useful fields:
  #  referer, referer_is_local, resource, resource_title, search_terms, dt

  set_table_name 'mint_visit'

  # The time the resource was hit, as a Ruby Time object.
  def visited_at
    Time.at(dt)
  end

  # Returns an array of Mint objects for what pages are currently popular.
  # 
  # If given an argument, it will restrict the search to the last few weeks.
  #
  # The following fields are available: resource, resource_title, visit_count
  def self.popular_pages(weeks_ago=nil)
    options = { 
          :select => "resource, resource_title, COUNT(resource) AS visit_count",
          :group => 'resource',
          :order => 'visit_count DESC' }
    options[:conditions] = ["dt > ?", weeks_ago_as_epoch_seconds(weeks_ago)] unless weeks_ago.nil?
    find :all, *options
  	# TODO WHERE resource NOT IN (#{resource_ignore_list})
  end

  # Returns a single integer representing the number of visits to a resource over a number of weeks.
  # Shows all visits by default.
  # 
  # Uses a wildcard at the front so you can use something like
  #
  #  /dashboard/2006/2
  # 
  # or
  #
  #  dashboard_url(:year => 2006, :month => 2, :only_path => true)
  #
  # and it will match all of these:
  #
  #  http://nubyonrails.com/dashboard/2006/2
  #  http://nubyonrails.topfunky.com/dashboard/2006/2
  #
  # Or, you can add a wildcard at the end like this:
  #
  #  /dashboard/2006%
  #
  # For root urls you will need a trailing slash:
  #
  #  http://nubyonrails.com/
  #
  def self.visits_to_resource(url, weeks_ago=nil)
    query = " resource LIKE ? "
    parameters = ["%#{url}"]
    if weeks_ago
      query += " AND dt > ? "
      parameters << weeks_ago_as_epoch_seconds(weeks_ago) 
    end    
    count [query, *parameters]
  end

  # Returns a Ruby Time object.
  def self.earliest_visit_to_resource(url)
    if visit = find(:first, :conditions => ["resource LIKE ?", "%#{url}"], :order => "dt ASC")
      return visit.visited_at
    end
  end

  # Call with
  #
  #  rake mint:parse_logs
  #
  # You must set the path to your current log file for today.
  #
  #  Mint.log_file_path = 'log/http_access.log'
  #
  # This will process all hits in the logfile. It assumes that you run it
  # once a day and that your host rotates the logs (but symlinks the latest
  # day's log to a consistent place).
  #
  # On Dreamhost, it's at 
  #  
  #  ~/logs/your_site.com/http/access.log.0
  #
  def self.parse_log_file(log_file=nil)
    log_file ||= @@log_file_path
    raise "No log file specified" unless log_file
    lp = LogParser.new
    transaction do 
      File.open(log_file) do |file|
        lp.parse_io_stream(file) { |stat|
          # Only process Typo feed resources (atom and rss) 
          if /feed\.xml$/.match(stat['PATH_INFO'])
            Mint.append(stat)
            sleep(0.1) # To keep load down on shared hosts
          end
        }
      end
    end
  end

  # Adds a hit to the Mint visits table.
  #
  # Used by the log parser to add hits for resources not tracked by Mint (like RSS).
  #
  def self.append(env)
    subscriber_count = 1
    if /(\d+) subscriber/.match(env['HTTP_USER_AGENT'])
      subscriber_count = $1.to_i
    end
    
    browser_family = browser_version = ''
    # TODO Check for browsers that start with a url
    # TODO Check for browsers without a slash-version (Google-FeedReader)
    # ^([^\(]+) Whole user agent
    # ^([^/]+)/([^/]\S+) Better ua/version splitter 
    if %r{^([^/]+)/(\S+)}.match(env['HTTP_USER_AGENT'])
      browser_family = $1
      browser_version = $2
    end
    
    time_of_hit = Time.now
    # [10/Oct/1999:21:15:05 +0500]
    if %r{([^/]+)/([^/]+)/([^:]+):([^:]+):([^:]+):(\S+)}.match(env['DATETIME'])
      time_of_hit = Time.local $3, $2, $1, $4, $5, $6
    end
    
    # Register multiple hits if necessary
    (1..subscriber_count).each do |i|
      self.create(:dt => time_of_hit.utc.to_i,
                  :ip_long => ip2long(env['REMOTE_ADDR']),
                  :referer => '',
                  :referer_checksum => 0,
                  :domain_checksum => 0,
                  :referer_is_local => -1,
                  :resource => env['PATH_INFO'],
                  :resource_checksum => Zlib::crc32(env['PATH_INFO']),
                  :resource_title => 'Feed',
                  :search_terms => '',
                  :browser_family => browser_family,
                  :browser_version => browser_version,
                  :platform => 'Feed',
                  :resolution => '0x0',
                  :flash_version => 0) # 0 interpreted as "None" by UA007 v121
    end
  end

private

  def self.weeks_ago_as_epoch_seconds(weeks_ago=2)
    (Time.now - weeks_ago.weeks).utc.to_i
  end

  def self.ip2long(ip)
    long = 0
    ip.split(/\./).reverse.each_with_index do |x, i|
        long += x.to_i << (i * 8)
    end
    long
  end

end

# RSS Hit: 72.14.199.2 - - [17/Feb/2006:00:39:50 -0800] "GET /xml/rss20/article/242/feed.xml HTTP/1.1" 200 8301 "-" "FeedFetcher-Google; (+http://www.google.com/feedfetcher.html)"
#  INSERT INTO `mint_visit` (`id`, `dt`, `ip_long`, `referer`, `referer_checksum`, `domain_checksum`, `referer_is_local`, `resource`, `resource_checksum`, `resource_title`, `search_terms`, `browser_family`, `browser_version`, `platform`, `resolution`, `flash_version`, `window_width`) VALUES (110, 1137386366, 410415437, 'http://rubyonrailsworkshops.com/dashboard/2006/3', 668827901, -1803874782, 1, 'http://rubyonrailsworkshops.com/dashboard/2006/2', 1356509291, 'Ruby on Rails Workshops and Conferences around the World', '', 'Firefox', '1.5', 'Macintosh', '1280x854', 8, 1152);