YoutubeIE: new algo for length 86 (fixes #1156)

Now is using the same length as the flash player used for age protected videos, but the algorithm is different, so now for age protected videos it first tries to use the old algo.

devscripts/youtube_genalgo.py

@@ -17,9 +17,9 @@ tests = [
     # 87 - vflART1Nf 2013/07/24
     ("qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$^&*()_-+={[]}|:;?/>.<",
      "tyuioplkjhgfdsazxcv<nm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$^&*()_-+={[]}|:;?/>"),
-    # 86 - vfl_ymO4Z 2013/06/27
+    # 86 - vflm_D8eE 2013/07/31
     ("qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$%^&*()_-+={[|};?/>.<",
-     "ertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!/#$%^&*()_-+={[|};?@"),
+     ">.1}|[{=+-_)(*&^%$#@!MNBVCXZASDFGHJK<POIUYTREW509876L432/mnbvcxzasdfghjklpoiuytre"),
     # 85 - vflSAFCP9 2013/07/19
     ("qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$%^&*()_-+={[};?/>.<",
      "ertyuiqplkjhgfdsazx$vbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#<%^&*()_-+={[};?/c"),
@@ -40,6 +40,12 @@ tests = [
      "Z?;}[{=+-(*&^%$#@!MNBVCXRASDFGHKLPOIUYT/EWQ0q87659321mnbvcxzasdfghjkl4oiuytrewp"),
 ]
 
+tests_age_gate = [
+    # 86 - vflqinMWD
+    ("qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$%^&*()_-+={[|};?/>.<",
+     "ertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!/#$%^&*()_-+={[|};?@"),
+]
+
 def find_matching(wrong, right):
     idxs = [wrong.index(c) for c in right]
     return compress(idxs)
@@ -90,6 +96,8 @@ def genall(tests):
 
 def main():
     print(genall(tests))
+    print(u'    Age gate:')
+    print(genall(tests_age_gate))
 
 if __name__ == '__main__':
     main()

test/test_youtube_sig.py

@@ -10,7 +10,9 @@ sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 from youtube_dl.extractor.youtube import YoutubeIE
 from helper import FakeYDL
 
-sig = YoutubeIE(FakeYDL())._decrypt_signature
+ie = YoutubeIE(FakeYDL())
+sig = ie._decrypt_signature
+sig_age_gate = ie._decrypt_signature_age_gate
 
 class TestYoutubeSig(unittest.TestCase):
     def test_92(self):
@@ -35,7 +37,7 @@ class TestYoutubeSig(unittest.TestCase):
 
     def test_86(self):
         wrong = "qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$%^&*()_-+={[|};?/>.<"
-        right = "ertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!/#$%^&*()_-+={[|};?@"
+        right = ">.1}|[{=+-_)(*&^%$#@!MNBVCXZASDFGHJK<POIUYTREW509876L432/mnbvcxzasdfghjklpoiuytre"
         self.assertEqual(sig(wrong), right)
 
     def test_85(self):
@@ -67,6 +69,11 @@ class TestYoutubeSig(unittest.TestCase):
         wrong = "qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKHGFDSAZXCVBNM!@#$%^&*(-+={[};?/"
         right = "Z?;}[{=+-(*&^%$#@!MNBVCXRASDFGHKLPOIUYT/EWQ0q87659321mnbvcxzasdfghjkl4oiuytrewp"
         self.assertEqual(sig(wrong), right)
+    
+    def test_86_age_gate(self):
+        wrong = "qwertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!@#$%^&*()_-+={[|};?/>.<"
+        right = "ertyuioplkjhgfdsazxcvbnm1234567890QWERTYUIOPLKJHGFDSAZXCVBNM!/#$%^&*()_-+={[|};?@"
+        self.assertEqual(sig_age_gate(wrong), right)
 
 if __name__ == '__main__':
     unittest.main()

youtube_dl/extractor/youtube.py

@@ -286,7 +286,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         elif len(s) == 87:
             return s[4:23] + s[86] + s[24:85]
         elif len(s) == 86:
-            return s[2:63] + s[82] + s[64:82] + s[63]
+            return s[83:85] + s[26] + s[79:46:-1] + s[85] + s[45:36:-1] + s[30] + s[35:30:-1] + s[46] + s[29:26:-1] + s[82] + s[25:1:-1]
         elif len(s) == 85:
             return s[2:8] + s[0] + s[9:21] + s[65] + s[22:65] + s[84] + s[66:82] + s[21]
         elif len(s) == 84:
@@ -303,6 +303,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         else:
             raise ExtractorError(u'Unable to decrypt signature, key length %d not supported; retrying might work' % (len(s)))
 
+    def _decrypt_signature_age_gate(self, s):
+        # The videos with age protection use another player, so the algorithms
+        # can be different.
+        if len(s) == 86:
+            return s[2:63] + s[82] + s[64:82] + s[63]
+        else:
+            # Fallback to the other algortihms
+            self._decrypt_signature(s)
+
+
     def _get_available_subtitles(self, video_id):
         self.report_video_subtitles_download(video_id)
         request = compat_urllib_request.Request('http://video.google.com/timedtext?hl=en&type=list&v=%s' % video_id)
@@ -611,7 +621,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                             parts_sizes = u'.'.join(compat_str(len(part)) for part in s.split('.'))
                             self.to_screen(u'encrypted signature length %d (%s), itag %s, %s' %
                                 (len(s), parts_sizes, url_data['itag'][0], player))
-                        signature = self._decrypt_signature(url_data['s'][0])
+                        encrypted_sig = url_data['s'][0]
+                        if age_gate:
+                            signature = self._decrypt_signature_age_gate(encrypted_sig)
+                        else:
+                            signature = self._decrypt_signature(encrypted_sig)
                         url += '&signature=' + signature
                     if 'ratebypass' not in url:
                         url += '&ratebypass=yes'