Merge pull request #2 from jdaviderb/update-decipher

update decipher object

.rubocop_todo.yml

@@ -1,14 +1,14 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2019-09-03 01:03:51 -0500 using RuboCop version 0.74.0.
+# on 2019-09-17 22:24:06 -0500 using RuboCop version 0.74.0.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 5
-# Configuration parameters: MinNameLength, AllowNamesEndingInNumbers, AllowedNames, ForbiddenNames.
-# AllowedNames: io, id, to, by, on, in, at, ip, db
-Naming/UncommunicativeMethodParamName:
-  Exclude:
-    - 'lib/youtube_audio/decipher.rb'
+# Offense count: 4
+# Cop supports --auto-correct.
+# Configuration parameters: AutoCorrect, AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, IgnoredPatterns.
+# URISchemes: http, https
+Metrics/LineLength:
+  Max: 109

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    youtube_audio (0.1.0)
+    youtube_audio (0.2.0)
       mechanize
       mini_racer
 
@@ -33,9 +33,9 @@ GEM
       nokogiri (~> 1.6)
       ntlm-http (~> 0.1, >= 0.1.1)
       webrobots (>= 0.0.9, < 0.2)
-    mime-types (3.2.2)
+    mime-types (3.3)
       mime-types-data (~> 3.2015)
-    mime-types-data (3.2019.0331)
+    mime-types-data (3.2019.0904)
     mini_portile2 (2.4.0)
     mini_racer (0.2.6)
       libv8 (>= 6.9.411)

lib/youtube_audio.rb

@@ -3,6 +3,7 @@
 require 'mechanize'
 require 'mini_racer'
 require 'cgi'
+require 'net/http'
 require 'youtube_audio/version'
 require 'youtube_audio/url_decipher'
 require 'youtube_audio/download'
@@ -13,6 +14,7 @@
 require 'youtube_audio/decipher'
 require 'youtube_audio/search'
 require 'youtube_audio/search_item'
+require 'youtube_audio/extract_decode_function'
 
 module YoutubeAudio
 end

lib/youtube_audio/decipher.rb

@@ -2,34 +2,36 @@
 
 module YoutubeAudio
   class Decipher
+    attr_reader :script_player_url
+    URL = 'https://www.youtube.com'
+
+    def initialize(script_player_url)
+      @script_player_url = script_player_url
+    end
+
     # @param cipher [string] youtube signature
     def decipher(cipher)
-      decipher = cipher.split('')
-      pb(decipher, 35)
-      pb(decipher, 30)
-      wh(decipher)
-      pb(decipher, 2)
-      decipher = p7(decipher, 3)
-      wh(decipher)
-      pb(decipher, 8)
-      decipher = p7(decipher, 2)
-      decipher.join('')
+      klass = extract_decode_function_handler
+      miniracer_klass = miniracer_context
+
+      miniracer_klass.eval(klass.cipher_helpers_object)
+      miniracer_klass.eval("var #{klass.decipher_function}")
+      miniracer_klass.eval("#{klass.decryption_function}('#{cipher}')")
     end
 
-    def pb(a, b)
-      c = a[0]
-      a[0] = a[b % a.length]
-      a[b % a.length] = c
+    private
+
+    def extract_decode_function_handler
+      @extract_decode_function_handler ||=
+        ExtractDecodeFunction.new(script_player_content)
     end
 
-    def wh(a)
-      result = a.reverse!
-      result
+    def miniracer_context
+      MiniRacer::Context.new
     end
 
-    def p7(a, b)
-      result = a.slice(b, a.length)
-      result
+    def script_player_content
+      @script_player_content ||= Net::HTTP.get(URI(URL + script_player_url))
     end
   end
 end

lib/youtube_audio/extract_decode_function.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module YoutubeAudio
+  class ExtractDecodeFunction
+    attr_reader :player_script
+    CIPHER_TEST = Regexp.new('([\\w$]+)\\s*=\\s*function\\((\\w+)\\)\\{\\s*\\2=\\s*\\2\\.split\\(""\\)\\s*;')
+
+    def initialize(player_script)
+      @player_script = player_script
+    end
+
+    def cipher_helper_object_name
+      @cipher_helper_object_name ||=
+        decipher_function.scan(/\;([a-zA-Z0-9]{0,255})\./).flatten.first
+    end
+
+    def cipher_helpers_object
+      object_decibper_index = player_script.index("var #{cipher_helper_object_name}={")
+      object_decibper_index_end = player_script[object_decibper_index..-1].index('};')
+      player_script[object_decibper_index..object_decibper_index + object_decibper_index_end]
+    end
+
+    def decipher_function
+      @decipher_function ||=
+        player_script.scan(Regexp.new(function_pattern)).flatten.first
+    end
+
+    def decryption_function
+      @decryption_function ||= player_script.scan(CIPHER_TEST).flatten.first
+    end
+
+    private
+
+    def function_pattern
+      @function_pattern ||=
+        '(' + decryption_function + '=function\\([a-zA-Z0-9_]+\\)\\{.+?\\})'
+    end
+  end
+end

lib/youtube_audio/format.rb

@@ -2,7 +2,10 @@
 
 module YoutubeAudio
   class Format
-    def initialize(response_raw)
+    attr_reader :script_player_url
+
+    def initialize(response_raw, script_player_url: nil)
+      @script_player_url = script_player_url
       @response_raw = response_raw
     end
 
@@ -11,7 +14,12 @@ def audio?
     end
 
     def url
-      return UrlDecipher.new(cipher).decipher if cipher
+      if cipher
+        return UrlDecipher.new(
+          cipher,
+          script_player_url: script_player_url
+        ).decipher
+      end
 
       @response_raw&.dig('url')
     end

lib/youtube_audio/player_response.rb

@@ -2,9 +2,12 @@
 
 module YoutubeAudio
   class PlayerResponse
+    attr_reader :script_player_url
+
     # @response_raw url [Hash]
-    def initialize(response_raw)
+    def initialize(response_raw, script_player_url: nil)
       @response_raw = response_raw
+      @script_player_url = script_player_url
     end
 
     def formats
@@ -14,7 +17,7 @@ def formats
     end
 
     def to_youtube_format(format_raw)
-      Format.new(format_raw)
+      Format.new(format_raw, script_player_url: script_player_url)
     end
   end
 end

lib/youtube_audio/script_parser.rb

@@ -16,7 +16,11 @@ def player_response
       mini_racer_context.eval(@script)
       # get player response ...
       response = mini_racer_context.eval('ytInitialPlayerConfig')
-      PlayerResponse.new(JSON.parse(response['args']['player_response']))
+
+      PlayerResponse.new(
+        JSON.parse(response['args']['player_response']),
+        script_player_url: response['assets']['js']
+      )
     end
   end
 end

lib/youtube_audio/url_decipher.rb

@@ -2,17 +2,19 @@
 
 module YoutubeAudio
   class UrlDecipher
-    attr_reader :cipher
+    attr_reader :cipher, :script_player_url
 
-    def initialize(cipher)
+    def initialize(cipher, script_player_url:)
       @cipher = cipher
+      @script_player_url = script_player_url
     end
 
     def decipher
       decoded = CGI.parse(@cipher)
 
-      signature = Decipher.new.decipher(decoded.dig('s').first)
-      decoded.dig('url').first + '&' + decoded.dig('sp').first + "=#{signature}"
+      sign = Decipher.new(script_player_url).decipher(decoded.dig('s').first)
+
+      decoded.dig('url').first + '&' + decoded.dig('sp').first + "=#{sign}"
     end
   end
 end

lib/youtube_audio/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module YoutubeAudio
-  VERSION = '0.1.0'
+  VERSION = '0.2.0'
 end

spec/decipher_spec.rb

@@ -1,20 +1,26 @@
 # frozen_string_literal: true
 
 RSpec.describe YoutubeAudio::Decipher do
-  subject { described_class.new }
+  let(:player_base_url) do
+    '/yts/jsbin/player-plasma-ias-phone-en_US-vflZqC1Iu/base.js'
+  end
+
+  subject { described_class.new(player_base_url) }
 
   describe '#decipher' do
     let(:signature) do
       'qLALgxI2AwRAIgbG7LwidtAcuTBa1Rw8hp5sUXLESfIyXM2MHr4m6jDtUCIARjHJ-5Y' \
         'KOvBfrSPrfc9N62UwxcUvy4kVFNQRL_2glvvlg'
     end
     let(:decipher) do
-      'ALgxI2wwRAIgbG7LwidtAcuTBa1Rs8hp5qUXLESfIyXM2MHr4m6jDtUCIARjHJ-5' \
-        'YKOvBfrSPrfc9N62UwxcUvy4kVFNQRL_2glv'
+      'vgg2_LRQNFVk4yvUcxwU26N9cfrPSrfBvOKY5-JHjRAICUtDj6m4rHM2MXyIfSELX' \
+        'Uslph8wR1aBTucAtdiwL7GbgIARwA2Ixq'
     end
 
     it 'returns decipher signature' do
-      expect(subject.decipher(signature)).to eq(decipher)
+      VCR.use_cassette('youtube/player_base.js') do
+        expect(subject.decipher(signature)).to eq(decipher)
+      end
     end
   end
 end

spec/download_spec.rb

@@ -12,6 +12,7 @@
       VCR.use_cassette('youtube/xoWRkd3oGcs') do
         expect(subject.formats.length).to eq(4)
         expect(subject.formats.first).to be_instance_of(YoutubeAudio::Format)
+        expect(subject.formats.first.url).to match('mp4')
         expect(subject.formats.first.audio_quality).to match('AUDIO_Q')
       end
     end

spec/format_spec.rb

@@ -12,7 +12,7 @@
     )
   end
 
-  subject { described_class.new(response_raw) }
+  subject { described_class.new(response_raw, script_player_url: 'mock') }
 
   describe '#mime_type' do
     it { expect(subject.mime_type).to eq('audio/mp4') }
@@ -36,7 +36,8 @@
         .and_return('url=mock&s=mock-signature&sp=sig')
 
       expect(YoutubeAudio::UrlDecipher).to receive(:new)
-        .with('url=mock&s=mock-signature&sp=sig')
+        .with('url=mock&s=mock-signature&sp=sig',
+              script_player_url: 'mock')
         .and_return(mock)
 
       expect(subject.url).to eq('decipher')

spec/url_decipher_spec.rb

@@ -3,12 +3,19 @@
 RSpec.describe YoutubeAudio::UrlDecipher do
   let(:cipher) { 'url=mock&s=mock-signature&sp=sig' }
 
-  subject { described_class.new(cipher) }
+  subject do
+    described_class.new(cipher, script_player_url: 'script_player_url_mock')
+  end
 
   describe '#decipher' do
+    before do
+      expect(YoutubeAudio::Decipher).to receive(:new)
+        .with('script_player_url_mock')
+        .and_return(double(decipher: 'sig-mock'))
+    end
+
     it 'deciphers a cipher url' do
-      expect(YoutubeAudio::Decipher).to receive(:new).and_call_original
-      expect(subject.decipher).to eq('mock&sig=gk-simcat')
+      expect(subject.decipher).to eq('mock&sig=sig-mock')
     end
   end
 end