video_info 2.6.1 → 2.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 18c94c937caae05276816dd3ed8ebe10df074b0b
-  data.tar.gz: 61a33b65f281e74192fd355574616281b2ef2eb9
+  metadata.gz: c51bc4b80a2d491ea898adc7c0e57a77548ff93d
+  data.tar.gz: 5a1469a4e6817e37793097a3b68b7155bcb63eee
 SHA512:
-  metadata.gz: 06a186d5c3fea88716f226090c797a82bbee92f1180683b5963fc96684d87c37b2eb6b25b18a2016eeca714023c7c86f3c0b558dce3dbc0881bad507c4917d07
-  data.tar.gz: 99f37329ca0c12666afd6f58e9f68af88dc62fe8d62a67ec542a074569bdd1b311c1f21fd86a56eb20ee1a7be4f8a576ffdd372050f808b439e3d8bb9f69d201
+  metadata.gz: 7ae91bf616bda0bab0a23f0f2195f4cdc4cf712bb1738331b81344bb523f54ca19805cf9383a0395438458899472dba37b10cbddb6e5cd80e177132c225f91ee
+  data.tar.gz: 38c99abe24884a9b7be1e3ab49f31567fb012f3fb2c3750a7f12614850caee5fc59327f134320ab68f4678f35a78bba2ff4ed0e1ebf463ad1483889b1bbb30d0

data/.travis.yml

@@ -3,10 +3,14 @@ sudo: false
 dist: trusty
 cache: bundler
 rvm:
-  - 2.2.4
-  - 2.3.0
+  - 2.2.5
+  - 2.3.1
   - jruby-head
   - rbx-2
+matrix:
+  allow_failures:
+    - rvm: jruby-head
+    - rvm: rbx-2
 jdk:
   - oraclejdk8
 before_install:

data/README.md

@@ -28,7 +28,7 @@ Usage
 
 Youtube and Vimeo have recently updated their APIs to require API keys.
 
-Scrapers has been implemented for both Youtube and Vimeo. However, the Youtube scraper can only get the date the video was posted, while the API is able to get the date and the exact time the video was posted. You may also wish to use the API to protect against potential HTML changes that could break the scrapesr.
+Scrapers has been implemented for both Youtube and Vimeo. However, the Youtube scraper can only get the date the video was posted, while the API is able to get the date and the exact time the video was posted. You may also wish to use the API to protect against potential HTML changes that could break the scrapers.
 
 To get a Youtube API key, [follow the instructions here](https://developers.google.com/youtube/registering_an_application)
 
@@ -161,7 +161,7 @@ You can also include an `iframe_attributes` or `url_attributes` hash to the `emb
 VideoInfo.new('http://www.youtube.com/watch?v=mZqGqE0D0n4').embed_code(iframe_attributes: { width: 800, height: 600, 'data-key' => 'value' })
 => '<iframe src="//www.youtube.com/embed/mZqGqE0D0n4" frameborder="0" allowfullscreen="allowfullscreen" width="800" height="600" data-key="value"></iframe>'
 
-'VideoInfo.new('http://www.youtube.com/watch?v=mZqGqE0D0n4').embed_code(url_attributes: { autoplay: 1 })
+VideoInfo.new('http://www.youtube.com/watch?v=mZqGqE0D0n4').embed_code(url_attributes: { autoplay: 1 })
 => '<iframe src="//www.youtube.com/embed/mZqGqE0D0n4?autoplay=1" frameborder="0" allowfullscreen="allowfullscreen"></iframe>'
 ```
 

data/lib/video_info.rb

@@ -50,7 +50,8 @@ class VideoInfo
   def_delegators :@provider,
                  :thumbnail_small,
                  :thumbnail_medium,
-                 :thumbnail_large
+                 :thumbnail_large,
+                 :thumbnail_maxres
   def_delegators :@provider, :embed_code, :embed_url
   def_delegators :@provider, :available?
   def_delegators :@provider, :playlist_id, :videos

data/lib/video_info/provider.rb

@@ -36,6 +36,12 @@ class VideoInfo
       !%w[403 404 400].include?(_response_code)
     end
 
+    def thumbnail
+      if defined?(thumbnail_large)
+        thumbnail_large
+      end
+    end
+
     private
 
     def _response_code

data/lib/video_info/providers/dailymotion.rb

@@ -17,6 +17,18 @@ class VideoInfo
         define_method(method) { nil }
       end
 
+      def author
+        data['owner.screenname']
+      end
+
+      def author_url
+        data['owner.url']
+      end
+
+      def author_thumbnail
+        data['owner.avatar_60_url']
+      end
+
       def keywords
         data['tags']
       end
@@ -71,6 +83,7 @@ class VideoInfo
         "https://#{_api_base}#{_api_path}?" \
         'fields=id,title,description,duration,created_time,' \
         'url,views_total,tags,thumbnail_url,thumbnail_720_url,' \
+        'owner.url,owner.screenname,owner.avatar_60_url,' \
         'thumbnail_240_url,thumbnail_60_url'
       end
 

data/lib/video_info/providers/vimeo_api.rb

@@ -39,25 +39,16 @@ class VideoInfo
         @thumbnail_id ||= _parse_picture_id(_video['pictures']['uri'])
       end
 
-      def thumbnail(width = 200, height = nil)
-        base_uri = "https://i.vimeocdn.com/video/#{thumbnail_id}"
-        if height
-          base_uri + "_#{width}x#{height}.jpg"
-        else
-          base_uri + "_#{width}.jpg"
-        end
-      end
-
       def thumbnail_small
-        thumbnail(100, 75)
+        generate_thumbnail(100, 75)
       end
 
       def thumbnail_medium
-        thumbnail(200, 150)
+        generate_thumbnail(200, 150)
       end
 
       def thumbnail_large
-        thumbnail(640)
+        generate_thumbnail(640)
       end
 
       def keywords
@@ -78,6 +69,15 @@ class VideoInfo
 
       private
 
+      def generate_thumbnail(width = 200, height = nil)
+        base_uri = "https://i.vimeocdn.com/video/#{thumbnail_id}"
+        if height
+          base_uri + "_#{width}x#{height}.jpg"
+        else
+          base_uri + "_#{width}.jpg"
+        end
+      end
+
       def _clean_options(options)
         headers = [super, _authorization_headers, _api_version_headers]
         headers.inject(&:merge)

data/lib/video_info/providers/vimeo_scraper.rb

@@ -1,15 +1,23 @@
 require 'oga'
 require 'open-uri'
 require 'json'
+require 'openssl'
+require 'cgi'
 
 class VideoInfo
   module Providers
     module VimeoScraper
       def author
-        json_info['author']['name']
+        if available?
+          json_info['author']['name']
+        end
       end
 
       def author_thumbnail
+        unless available?
+          return nil
+        end
+
         split_point = 'window.vimeo.clip_page_config ='
         script_tags = data.css('script')
 
@@ -27,7 +35,9 @@ class VideoInfo
       end
 
       def author_url
-        json_info['author']['url']
+        if available?
+          json_info['author']['url']
+        end
       end
 
       def available?
@@ -36,9 +46,9 @@ class VideoInfo
         if data.nil?
           is_available = false
         elsif is_available
-          page_header = data.css('#page_header')
+          password_elements = data.css('.exception_title--password')
 
-          if page_header.text == "\n Private Video\n "
+          unless password_elements.empty?
             is_available = false
           end
         end
@@ -55,16 +65,24 @@ class VideoInfo
       end
 
       def date
-        upload_date = json_info['uploadDate']
-        ISO8601::DateTime.new(upload_date).to_time
+        if available?
+          upload_date = json_info['uploadDate']
+          ISO8601::DateTime.new(upload_date).to_time
+        end
       end
 
       def duration
-        duration = json_info['duration']
-        ISO8601::Duration.new(duration).to_seconds.to_i
+        if available?
+          duration = json_info['duration']
+          ISO8601::Duration.new(duration).to_seconds.to_i
+        end
       end
 
       def keywords
+        unless available?
+          return nil
+        end
+
         if json_info['keywords']
           json_info['keywords']
         else
@@ -73,27 +91,39 @@ class VideoInfo
       end
 
       def height
-        json_info['height']
+        if available?
+          json_info['height']
+        end
       end
 
       def width
-        json_info['width']
+        if available?
+          json_info['width']
+        end
       end
 
       def thumbnail_small
-        thumbnail_url.split('_')[0] + '_100x75.jpg'
+        if available?
+          thumbnail_url.split('_')[0] + '_100x75.jpg'
+        end
       end
 
       def thumbnail_medium
-        thumbnail_url.split('_')[0] + '_200x150.jpg'
+        if available?
+          thumbnail_url.split('_')[0] + '_200x150.jpg'
+        end
       end
 
       def thumbnail_large
-        thumbnail_url.split('_')[0] + '_640.jpg'
+        if available?
+          thumbnail_url.split('_')[0] + '_640.jpg'
+        end
       end
 
       def view_count
-        json_info['interactionCount']
+        if available?
+          json_info['interactionCount']
+        end
       end
 
       private
@@ -111,7 +141,17 @@ class VideoInfo
       end
 
       def thumbnail_url
-        @thumbnail_url ||= json_info['thumbnailUrl']
+        @thumbnail_url ||= remove_overlay(meta_node_value('og:image'))
+      end
+
+      def remove_overlay(url)
+        uri = URI.parse(url)
+        
+        if uri.path == '/filter/overlay'
+          CGI::parse(uri.query)['src0'][0]
+        else
+          url
+        end
       end
 
       def meta_nodes

data/lib/video_info/providers/vkontakte.rb

@@ -1,7 +1,6 @@
-# encoding: UTF-8
+# encoding: utf-8
 require 'cgi'
 require 'net/http'
-require 'oga'
 
 class VideoInfo
   module Providers
@@ -35,15 +34,11 @@ class VideoInfo
       end
 
       def width
-        { 240 => 320,
-          360 => 480,
-          480 => 640,
-          720 => 1280
-        }[height].to_i
+        data[/PLAYER_FULL_WIDTH":(\d+)/, 1].to_i
       end
 
       def height
-        data[/url(\d+)/, 1].to_i
+        data[/PLAYER_FULL_HEIGHT":(\d+)/, 1].to_i
       end
 
       def author
@@ -52,15 +47,13 @@ class VideoInfo
       end
 
       def author_thumbnail
-        split_point = '<img id="mv_author_img" class="mv_author_img" src="'
-        author_img_split = data.split(split_point)[1]
-        author_img_split.split('">')[0]
+        split_point = '<img id="mv_author_photo" class="mv_author_photo" src="'
+        get_between(split_point, '">')
       end
 
       def author_url
-        split_point = '"author_href":"\\/'
-        author_url_path = data.split(split_point)[1].split('"')[0]
-        'https://vk.com/' + author_url_path
+        split_point = "<div class=\"mv_author_name\">\n          <a class=\"mem_link\" href=\""
+        'https://vk.com' + get_between(split_point, '">')
       end
 
       def title
@@ -68,7 +61,7 @@ class VideoInfo
       end
 
       def view_count
-        data[/mv_views_count_number\">.*?(\d+)/, 1].to_i
+        data[/mv_views_count.*?(\d+)/, 1].to_i
       end
 
       def embed_url
@@ -95,13 +88,32 @@ class VideoInfo
 
       def available?
         !%w[403 404 302 401].include?(_response_code) &&
-          !(data =~ /Ошибка доступа/)
+          !(data =~ /(Ошибка доступа|Access denied)/)
+      end
+
+      def thumbnail_small
+        thumb = data[/"thumb":"(.*?)"/, 1]
+        return thumb.delete('\\') unless thumb.nil?
+        nil
+      end
+
+      def thumbnail_medium
+        thumbnail_small
+      end
+
+      def thumbnail_large
+        nil
+      end
+
+      def thumbnail_maxres
+        nil
       end
 
       private
 
       def _make_request(url, options)
         request = Net::HTTP::Post.new(url.path)
+        request.add_field 'User-Agent', 'Mozilla/5.0 (Linux; VideoInfo)'
         request.body = URI.encode_www_form(options)
         conn = Net::HTTP.new(url.host, url.port)
         conn.use_ssl = true
@@ -112,6 +124,8 @@ class VideoInfo
       def _set_data_from_api_impl(api_url)
         options['act'] = 'show'
         options['al'] = '1'
+        options['autoplay'] = '0'
+        options['module'] = 'profile'
         options['video'] = "#{@video_owner}_#{@video_id}"
         data = _make_request(api_url, options)
         if data.index('Ошибка доступа')
@@ -146,15 +160,15 @@ class VideoInfo
       end
 
       def _response_code
-        response = Net::HTTP.get_response(_api_base, _api_path, 80)
-        code = _get_response_code response
-        if code == '302'
-          _get_response_code Net::HTTP.get_response(
-            URI.parse(response.header['location'])
-          )
-        else
-          code
+        check_url = URI.parse(url)
+        req = Net::HTTP::Get.new(check_url)
+        req.add_field('User-Agent', 'Mozilla/5.0 (Linux; VideoInfo)')
+        res = Net::HTTP.start(check_url.host, check_url.port) do |http|
+          http.request(req)
         end
+        return res.code unless res.code == '302' && !res['location'].nil?
+        self.url = res['location']
+        _response_code
       end
 
       def _api_base
@@ -189,6 +203,10 @@ class VideoInfo
       def _api_url
         URI('https://vk.com/al_video.php')
       end
+
+      def get_between(from_here, to_here)
+        data.split(from_here)[1].split(to_here)[0]
+      end
     end
   end
 end

data/lib/video_info/providers/youtube.rb

@@ -44,6 +44,10 @@ class VideoInfo
         "https://i.ytimg.com/vi/#{video_id}/hqdefault.jpg"
       end
 
+      def thumbnail_maxres
+        "https://i.ytimg.com/vi/#{video_id}/maxresdefault.jpg"
+      end
+
       private
 
       def _url_regex

data/lib/video_info/version.rb

@@ -1,3 +1,3 @@
 class VideoInfo
-  VERSION = '2.6.1'.freeze
+  VERSION = '2.7'.freeze
 end

data/spec/fixtures/vcr_cassettes/VideoInfo_Providers_Dailymotion/_available_/with_This_video_does_not_exist_or_has_been_deleted_/_available_/1_2_2_1_1.yml

@@ -0,0 +1,84 @@
+---
+http_interactions:
+- request:
+    method: get
+    uri: https://api.dailymotion.com/video/inValIdiD
+    body:
+      encoding: US-ASCII
+      string: ''
+    headers:
+      Accept-Encoding:
+      - gzip;q=1.0,deflate;q=0.6,identity;q=0.3
+      Accept:
+      - "*/*"
+      User-Agent:
+      - Ruby
+  response:
+    status:
+      code: 404
+      message: Not Found
+    headers:
+      Server:
+      - DMS/1.0.42
+      X-Dm-Backnode:
+      - web-115.adm.dailymotion.com:80
+      Cache-Control:
+      - no-cache, must-revalidate, stale-if-error=900
+      Content-Type:
+      - application/json; charset=UTF-8
+      Access-Control-Expose-Headers:
+      - X-DM-BackNode-Response-Time, X-DM-Edge, X-DM-Cache-Status
+      X-Dm-Lb-Ip:
+      - 195.8.215.129
+      Expires:
+      - Fri, 01 Jan 2010 05:00:00 GMT
+      X-Dm-Lb-Name:
+      - lb-08
+      X-Dm-Random-Number:
+      - '5037760'
+      Access-Control-Allow-Origin:
+      - "*"
+      X-Dm-Backnode-Response-Time:
+      - '37'
+      X-Robots-Tag:
+      - noindex
+      Access-Control-Allow-Headers:
+      - Content-Type, Authorization
+      Access-Control-Allow-Methods:
+      - GET, POST, DELETE
+      X-Dm-Page:
+      - us.rest.rest_api
+      Via:
+      - 1.1 varnish
+      - 1.1 varnish
+      Fastly-Debug-Digest:
+      - 1f140405cf79803f1e28a389edd8ca212ce1e285b54a58fb8a17474780cac6a2
+      Content-Length:
+      - '172'
+      Accept-Ranges:
+      - bytes
+      Date:
+      - Mon, 28 Dec 2015 17:43:14 GMT
+      Age:
+      - '0'
+      Connection:
+      - keep-alive
+      X-Served-By:
+      - cache-fra1225-FRA, cache-sjc3122-SJC
+      X-Cache:
+      - MISS, MISS
+      X-Cache-Hits:
+      - 0, 0
+      Vary:
+      - X-DM-EC-Geo-Country, Accept-Encoding
+      X-Dm-Cache-Status:
+      - MISS
+      X-Dm-Edge:
+      - Fastly
+    body:
+      encoding: ASCII-8BIT
+      string: '{"error":{"more_info":"https:\/\/developer.dailymotion.com\/api#error-codes","code":404,"message":"Can''t
+        find object video for `id'' parameter","type":"not_found","error_data":{"reason":"object_not_found"}}}'
+    http_version: 
+  recorded_at: Mon, 28 Dec 2015 17:43:18 GMT
+recorded_with: VCR 3.0.1