root/platform/pragger/Feed/nicovideo_rank.rb @ 8290

## make ranking nicovideo source.
##
## - module: Feed::nicovideo_rank
##   config:
##     cache: cache file base name.
##
## Copyright (C) 2007-2008, INOUE Yasuyuki <inoue.yasuyuki0@gmail.com>
## Original version is written by TADA Tadashi <sho@spc.gr.jp>
## You can redistribute it and/or modify it under GPL3 or any later version.

begin
        require 'rubygems'
rescue LoadError
end
require 'hpricot'
require 'yaml'
require 'pp'
require 'nicovideo/thumbnail_info'

def nicovideo_rank(config, data)
        raw_items = {}
        last_fetched = 0
        target_hour = 24
        rank_items = []
  client = ::NicoVideo::ThumbnailInfo.new

        open( "#{config['rank_data']}.last_fetched" ) do |f| last_fetched = YAML.load( f ) end
        FileUtils.mkpath(config['root_dir'] + last_fetched.strftime("/%Y/%m/%d"))
        open( "#{config['rank_data']}.dump" ) do |f| raw_items = YAML.load( f ) end
        begin
                open(config['root_dir'] + last_fetched.strftime("/%Y/%m/%d/") + "rank.yaml") do |f| rank_items = YAML.load( f ) end
        rescue Errno::ENOENT
        end

        if config['hour'] then
          target_hour = config['hour'].to_i
        end
        
        target_time = last_fetched - (target_hour * 60 * 60)
        
        raw_items.keys.each do |key|
    begin
      raw_item = raw_items[key]
      next if raw_item['uploaded_at'] > target_time
      
      begin
        raw_item = get_rank_item(client, key[%r|watch/(.+)$|, 1])
      rescue Errno::ENOENT
        # remove 'nice boat.' movie.
        raw_items.delete key
        next
      resucue TimeoutError
        # something wrong happend.
        break
      end

      rank_items << raw_item
      raw_items.delete key
    rescue ArgumentError => e
      pp e
      pp raw_items[key]
    end
  end
  
  open( "#{config['rank_data']}.dump", 'w' ) do |f|
    YAML.dump(raw_items, f)
  end
  open(config['root_dir'] + last_fetched.strftime("/%Y/%m/%d/") + "rank.yaml", 'w' ) do |f|
    YAML.dump(rank_items, f)
  end
  rank_items
end

def get_rank_item(client, id)
  thumb = client.get(id)
  cache = {}
  cache['link'] = thumb['watch_url']
  cache['title'] = thumb['title']
  cache['viewed'] = thumb['view_counter'].to_f
  cache['mylisted'] = thumb['mylist_counter'].to_f
  cache['uploaded_at'] = Time.parse(thumb['first_retrieve'])
  cache
end