root/platform/pragger/Feed/nicovideo_feed.rb @ 7816

## make feed nicovideo source.
##
## - module: Feed::nicovideo_feed
##   config:
##     cache: cache file.
##
## Copyright (C) 2007, TADA Tadashi <sho@spc.gr.jp>, 
## 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 'rss'
require 'yaml'
require 'nicovideo/thumbnail_info'

def nicovideo_feed(config, data)
        updated_at = Time::now
  found = false
  client = ::NicoVideo::ThumbnailInfo.new

        items, raw_items = load(config)
  
        data.each do |html|
                doc = Hpricot( html )
                base_tag = doc / "base"
                base = ""
                base = base_tag.first['href'] if base_tag
                (doc / 'a').each do |link|
      next unless link['href'] && link['href'].include?("watch/")
      found = true
      id = link['href'][%r|watch/(.+)|, 1]
                        key = base + link['href']
      thumb = client.get(id)

                        if items[key] then
        raw_items[key] = get_rank_item(key, thumb, updated_at) if raw_items[key]
      else
        item = RSS::RDF::Item::new
        item.link = key
        if thumb then
          item.title = thumb['title']
          item.description = "<a href=\"#{key}\"><img border=\"0\" src=\"#{thumb['thumbnail_url']}\"/></a><br/>"
        else
          item.title = thumb['title']
        end
        item.description += thumb['description']
        item.date = updated_at
        items[key] = item
        raw_items[key] = get_rank_item(key, thumb, updated_at)
                        end
                end 
        end
  
  # exit if nothing fetched
  exit unless found
        
        return_items = items.values.sort {|a,b| a.date == b.date ? (a.link.scan(/\d+$/)[0].to_i <=> b.link.scan(/\d+$/)[0].to_i) : (a.date <=> b.date) }.reverse
        
  save(config, items, return_items, raw_items, updated_at)
  
        return_items
end

def get_rank_item(key, thumb, updated_at)
  cache = {}
  cache['link'] = key
  cache['title'] = thumb['title']
  cache['viewed'] = thumb['view_counter'].to_f
  cache['mylisted'] = thumb['mylist_counter'].to_f
  cache['uploaded_at'] = thumb['first_retrieve']
  cache['updated_at'] = updated_at
  cache
end

def load(config)
        items = {}
        raw_items = {}
        begin
                if config['cache'] then
                        open( config['cache'] ) {|f| items = YAML.load( f ) }
                end
        rescue Errno::ENOENT
        end
        begin
                if config['rank'] then
                        open( "#{config['rank']}.dump" ) {|f| raw_items = YAML.load( f ) }
                end
        rescue Errno::ENOENT
        end
  return items, raw_items
end

def save(config, items, return_items, raw_items, updated_at)
        if config['cache'] then
          oldest_cache_item = items.size > 1200 && return_items[1200].link.scan(/\d+$/)[0].to_i
                cache_items = (items.size > 1200) ? items.reject {|a, b| b.link.scan(/\d+$/)[0].to_i < oldest_cache_item} : items.dup
                open( config['cache'], 'w' ) do |f|
                        YAML.dump(cache_items, f)
                end
        end
        if config['rank'] then
                open( "#{config['rank']}.dump", 'w' ) do |f|
                        YAML.dump(raw_items, f)
                end
                open( "#{config['rank']}.last_fetched", 'w' ) do |f|
                        YAML.dump(updated_at, f)
                end
        end
end