root/platform/pragger/Feed/nicovideo_rank.rb @ 6787

## 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'

def nicovideo_rank(config, data)
        raw_items = {}
        last_fetched = 0
        target_hour = 24
        rank_items = []

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

        if config['hour'] then
          target_hour = config['hour'].to_i
        end
        
        target_time = last_fetched - (target_hour * 60 * 60 - 1200)
        
        raw_items.keys.sort{|a, b| a.scan(/\d+$/)[0].to_i <=> b.scan(/\d+$/)[0].to_i}.reverse.each do |key|
                raw_item, uploaded_at, updated_at = raw_items[key]
                next if uploaded_at > target_time
                if updated_at < last_fetched || uploaded_at < target_time - 60 * 30 then
                  raw_items.delete key
                  next
                end

                rank_items << raw_item
        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