diff options
Diffstat (limited to 'youtube_dl/extractor/theplatform.py')
| -rw-r--r-- | youtube_dl/extractor/theplatform.py | 23 |
1 files changed, 20 insertions, 3 deletions
diff --git a/youtube_dl/extractor/theplatform.py b/youtube_dl/extractor/theplatform.py index 7a5a533b7..02dbef913 100644 --- a/youtube_dl/extractor/theplatform.py +++ b/youtube_dl/extractor/theplatform.py @@ -151,6 +151,22 @@ class ThePlatformIE(ThePlatformBaseIE): 'only_matching': True, }] + @classmethod + def _extract_urls(cls, webpage): + m = re.search( + r'''(?x) + <meta\s+ + property=(["'])(?:og:video(?::(?:secure_)?url)?|twitter:player)\1\s+ + content=(["'])(?P<url>https?://player\.theplatform\.com/p/.+?)\2 + ''', webpage) + if m: + return [m.group('url')] + + matches = re.findall( + r'<(?:iframe|script)[^>]+src=(["\'])((?:https?:)?//player\.theplatform\.com/p/.+?)\1', webpage) + if matches: + return list(zip(*matches))[1] + @staticmethod def _sign_url(url, sig_key, sig_secret, life=600, include_qs=False): flags = '10' if include_qs else '00' @@ -159,11 +175,11 @@ class ThePlatformIE(ThePlatformBaseIE): def str_to_hex(str): return binascii.b2a_hex(str.encode('ascii')).decode('ascii') - def hex_to_str(hex): - return binascii.a2b_hex(hex) + def hex_to_bytes(hex): + return binascii.a2b_hex(hex.encode('ascii')) relative_path = re.match(r'https?://link.theplatform.com/s/([^?]+)', url).group(1) - clear_text = hex_to_str(flags + expiration_date + str_to_hex(relative_path)) + clear_text = hex_to_bytes(flags + expiration_date + str_to_hex(relative_path)) checksum = hmac.new(sig_key.encode('ascii'), clear_text, hashlib.sha1).hexdigest() sig = flags + expiration_date + checksum + str_to_hex(sig_secret) return '%s&sig=%s' % (url, sig) @@ -269,6 +285,7 @@ class ThePlatformFeedIE(ThePlatformBaseIE): 'timestamp': 1391824260, 'duration': 467.0, 'categories': ['MSNBC/Issues/Democrats', 'MSNBC/Issues/Elections/Election 2016'], + 'uploader': 'NBCU-NEWS', }, } |