Changeset 28502

Timestamp:

01/16/09 17:11:54 (4 years ago)

Author:

valda

Message:

YouPorn? のテストを追加

Location:

lang/ruby/video_scraper

Files:

• lib/www/video_scraper.rb (modified) (3 diffs)
• lib/www/video_scraper/base.rb (modified) (2 diffs)
• lib/www/video_scraper/you_porn.rb (modified) (2 diffs)
• test/www/test_video_scraper.rb (moved) (moved from lang/ruby/video_scraper/test/www/video_scraper/test_video_scraper.rb)
• test/www/video_scraper/test_you_porn.rb (added)

lang/ruby/video_scraper/lib/www/video_scraper.rb

@@ -18 +18 @@
     class FileNotFound < RuntimeError; end
     VERSION = '1.0.0'
-    
+
     MODULES_NAME = %w(age_sage ameba_vision dailymotion moro_tube
                       nico_video pornhub pornotube red_tube veoh
                       you_porn you_tube your_file_host)
-    
+
     @@modules = MODULES_NAME.map do |name|
       require "www/video_scraper/#{name}"
       const_get( name.gsub(/\/(.?)/) { "::#{$1.upcase}" }.gsub(/(?:^|_)(.)/) { $1.upcase } )
     end
-      
+
     @@options = {
       :logger => nil,
@@ -33 +33 @@
       :debug => false,
     }
-    
+
     class << self
       def options
         @@options
       end
-      
+
       def options=(opts)
         @@options = opts
@@ -84 +84 @@
   end
 end
-
-
-if $0 == __FILE__
-   #   'http://www.yourfilehost.com/media.php?cat=video&file=XV436__03.wmv',
-   #   'http://www.morotube.com/watch.php?clip=46430e1d',
-   #   'http://www.pornhub.com/view_video.php?viewkey=35f8c5b464a15c9d3567',
-   #   'http://www.redtube.com/8415',
-end

lang/ruby/video_scraper/lib/www/video_scraper/base.rb

@@ -33 +33 @@
       end
 
-      def http_get(url)
-        open_opt = { "User-Agent" => "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322)" }
+      def http_get(url, opt = nil)
+        open_opt = {
+          "User-Agent" => "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322)",
+        }.merge( opt || {} )
         if @opt[:cache]
           unless @opt[:cache].respond_to?(:get) and @opt[:cache].respond_to?(:set)
@@ -40 +42 @@
           end
           @opt[:logger].debug 'use cache.'
-          unless content = @opt[:cache].get(url)
+          cache_key = "#{url}|#{open_opt}"
+          unless content = @opt[:cache].get(cache_key)
             content = open(url, open_opt) {|fh| fh.read }
-            @opt[:cache].set(url, content)
+            @opt[:cache].set(cache_key, content)
           end
         else

lang/ruby/video_scraper/lib/www/video_scraper/you_porn.rb

@@ -5 +5 @@
 module WWW
   module VideoScraper
-    class YouPorn
-      attr_reader :request_url, :response_body, :page_url, :title, :video_url, :thumb_url
+    class YouPorn < Base
+      url_regex %r!\Ahttp://youporn\.com/watch/(\d+)!
 
-      def initialize(opt)
-        @opt = opt.is_a?(String) ? { :url => opt } : opt
-        @agent = WWW::Mechanize.new
-        @agent.user_agent_alias = 'Windows IE 6'
+      def initialize(url, opt = nil)
+        super
         do_query
-      end
-
-      def self.valid_url?(url)
-        get_mediaid(url)
-      end
-
-      def self.get_mediaid(url)
-        url.match(%r!\Ahttp://youporn\.com/watch/(\d+)!)[1] rescue nil
       end
 
       private
       def do_query
-        url = @opt[:url]
-        raise StandardError, 'url param is requred' unless url
-        raise StandardError, "url is not YouPorn link: #{url}" unless YouPorn.valid_url? url
-        @page_url = url
-        id = YouPorn.get_mediaid(url)
+        id = url_regex_match[1]
 
-        @request_url = url.sub(/(\?.*)?$/, '?user_choice=Enter')
-        page = @agent.get(@request_url)
-        @response_body = page.body
-        page.root.search('//div[@id="download"]//a').each do |elem|
+        request_url = @page_url.sub(/(\?.*)?$/, '?user_choice=Enter')
+        html = http_get(request_url, 'Cookie' => 'age_check=1')
+        doc = Hpricot(html)
+        doc.search('//div[@id="download"]//a').each do |elem|
           href = elem.attributes['href']
-          @video_url = href if href =~ /\.flv$/
+          (@video_url = href; break) if href =~ %r!^http://download\.youporn\.com/download/.*\.flv!
         end
-        h1 = page.root.at('//div[@id="videoArea"]/h1')
+        h1 = doc.at('//div[@id="videoArea"]/h1')
         @title = h1.inner_html.gsub(/<[^>]*>/, '').strip
         @thumb_url = h1.at('/img').attributes['src'].sub(/(\d+)_small\.jpg$/, '\1_large.jpg')
@@ -45 +31 @@
   end
 end
-
-if $0 == __FILE__
-  w = VideoScraper::YouPorn.new('http://youporn.com/watch/93495?user_choice=Enter')
-  puts w.title
-  puts w.video_url
-  puts w.thumb_url
-  puts '--------------------'
-  w = VideoScraper::YouPorn.new('http://youporn.com/watch/93495')
-  puts w.title
-  puts w.video_url
-  puts w.thumb_url
-end