root/platform/pragger/Feed/nicovideo_feed.rb @ 6769

## 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'

def nicovideo_feed(config,data)
        items = {}
        raw_items = {}
        updated_at = Time::now
        begin
                if config['cache'] then
                        open( config['cache'] ) {|f| items = Marshal.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
        data.each do |html|
                doc = Hpricot( html )
                base_tag = doc / "base"
                base = ""
                base = base_tag.first['href'] if base_tag
                (doc / 'div.thumb_frm').each do |v|
                        item = RSS::RDF::Item::new
                        key = (v / 'a').first['href']
                        key = base + key 
                        if items[key] then
                                cache = {}
                                cache['link'] = key
                                thumb = (v/ 'img.thumb_img').first
                                cache['title'] = thumb['alt']
                                cache['viewed'] = (v / 'p.TXT10 strong')[1].inner_html.tr(',', '').to_f
                                cache['mylisted'] = (v / 'p.TXT10 strong')[3].inner_html.tr(',', '').to_f
                          raw_items[key] = [cache, items[key].date, updated_at]
                          next
                        end
                        item.link = key
                        thumb = (v/ 'img.thumb_img').first
                        if thumb then
                                item.title = thumb['alt']
                                item.description = "<img src=\"#{thumb['src']}\"/><br/>"
                        else
                                item.title = (v / 'p.TXT12 a').first.inner_html
                        end
                        item.description += (v / 'p.TXT12' ).inner_html.gsub( /<.*?>/, '' )
                        item.date = updated_at
                        items[key] = item
                end 
        end
        
        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
        
        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|
                        f.write(Marshal.dump(cache_items))
                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|
                        f.write(Marshal.dump(updated_at))
                end
        end
        return_items
end