youtube_dlc/downloader/hls.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4 import binascii
   5 try:
   6     from Crypto.Cipher import AES
   7     can_decrypt_frag = True
   8 except ImportError:
   9     can_decrypt_frag = False
  10
  11 from .fragment import FragmentFD
  12 from .external import FFmpegFD
  13
  14 from ..compat import (
  15     compat_urllib_error,
  16     compat_urlparse,
  17     compat_struct_pack,
  18 )
  19 from ..utils import (
  20     parse_m3u8_attributes,
  21     update_url_query,
  22 )
  23
  24
  25 class HlsFD(FragmentFD):
  26     """ A limited implementation that does not require ffmpeg """
  27
  28     FD_NAME = 'hlsnative'
  29
  30     @staticmethod
  31     def can_download(manifest, info_dict):
  32         UNSUPPORTED_FEATURES = (
  33             r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  34             # r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  35
  36             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  37             # http://hls-geo.daserste.de/i/videoportal/Film/c_620000/622873/format,716451,716457,716450,716458,716459,.mp4.csmil/index_4_av.m3u8?null=0)
  38             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  39
  40             # This heuristic also is not correct since segments may not be appended as well.
  41             # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
  42             # no segments will definitely be appended to the end of the playlist.
  43             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  44             #                                 # event media playlists [4]
  45             r'#EXT-X-MAP:',  # media initialization [5]
  46
  47             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  48             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  49             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  50             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  51             # 5. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.5
  52         )
  53         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  54         is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
  55         check_results.append(can_decrypt_frag or not is_aes128_enc)
  56         check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
  57         check_results.append(not info_dict.get('is_live'))
  58         return all(check_results)
  59
  60     def real_download(self, filename, info_dict):
  61         man_url = info_dict['url']
  62         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  63
  64         urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
  65         man_url = urlh.geturl()
  66         s = urlh.read().decode('utf-8', 'ignore')
  67
  68         if not self.can_download(s, info_dict):
  69             if info_dict.get('extra_param_to_segment_url') or info_dict.get('_decryption_key_url'):
  70                 self.report_error('pycrypto not found. Please install it.')
  71                 return False
  72             self.report_warning(
  73                 'hlsnative has detected features it does not support, '
  74                 'extraction will be delegated to ffmpeg')
  75             fd = FFmpegFD(self.ydl, self.params)
  76             for ph in self._progress_hooks:
  77                 fd.add_progress_hook(ph)
  78             return fd.real_download(filename, info_dict)
  79
  80         def is_ad_fragment_start(s):
  81             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s
  82                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',ad'))
  83
  84         def is_ad_fragment_end(s):
  85             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=master' in s
  86                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',segment'))
  87
  88         media_frags = 0
  89         ad_frags = 0
  90         ad_frag_next = False
  91         for line in s.splitlines():
  92             line = line.strip()
  93             if not line:
  94                 continue
  95             if line.startswith('#'):
  96                 if is_ad_fragment_start(line):
  97                     ad_frag_next = True
  98                 elif is_ad_fragment_end(line):
  99                     ad_frag_next = False
 100                 continue
 101             if ad_frag_next:
 102                 ad_frags += 1
 103                 continue
 104             media_frags += 1
 105
 106         ctx = {
 107             'filename': filename,
 108             'total_frags': media_frags,
 109             'ad_frags': ad_frags,
 110         }
 111
 112         self._prepare_and_start_frag_download(ctx)
 113
 114         fragment_retries = self.params.get('fragment_retries', 0)
 115         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 116         test = self.params.get('test', False)
 117
 118         extra_query = None
 119         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 120         if extra_param_to_segment_url:
 121             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 122         i = 0
 123         media_sequence = 0
 124         decrypt_info = {'METHOD': 'NONE'}
 125         byte_range = {}
 126         frag_index = 0
 127         ad_frag_next = False
 128         for line in s.splitlines():
 129             line = line.strip()
 130             if line:
 131                 if not line.startswith('#'):
 132                     if ad_frag_next:
 133                         continue
 134                     frag_index += 1
 135                     if frag_index <= ctx['fragment_index']:
 136                         continue
 137                     frag_url = (
 138                         line
 139                         if re.match(r'^https?://', line)
 140                         else compat_urlparse.urljoin(man_url, line))
 141                     if extra_query:
 142                         frag_url = update_url_query(frag_url, extra_query)
 143                     count = 0
 144                     headers = info_dict.get('http_headers', {})
 145                     if byte_range:
 146                         headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1)
 147                     while count <= fragment_retries:
 148                         try:
 149                             success, frag_content = self._download_fragment(
 150                                 ctx, frag_url, info_dict, headers)
 151                             if not success:
 152                                 return False
 153                             break
 154                         except compat_urllib_error.HTTPError as err:
 155                             # Unavailable (possibly temporary) fragments may be served.
 156                             # First we try to retry then either skip or abort.
 157                             # See https://github.com/ytdl-org/youtube-dl/issues/10165,
 158                             # https://github.com/ytdl-org/youtube-dl/issues/10448).
 159                             count += 1
 160                             if count <= fragment_retries:
 161                                 self.report_retry_fragment(err, frag_index, count, fragment_retries)
 162                     if count > fragment_retries:
 163                         if skip_unavailable_fragments:
 164                             i += 1
 165                             media_sequence += 1
 166                             self.report_skip_fragment(frag_index)
 167                             continue
 168                         self.report_error(
 169                             'giving up after %s fragment retries' % fragment_retries)
 170                         return False
 171                     if decrypt_info['METHOD'] == 'AES-128':
 172                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 173                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 174                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 175                         frag_content = AES.new(
 176                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 177                     self._append_fragment(ctx, frag_content)
 178                     # We only download the first fragment during the test
 179                     if test:
 180                         break
 181                     i += 1
 182                     media_sequence += 1
 183                 elif line.startswith('#EXT-X-KEY'):
 184                     decrypt_url = decrypt_info.get('URI')
 185                     decrypt_info = parse_m3u8_attributes(line[11:])
 186                     if decrypt_info['METHOD'] == 'AES-128':
 187                         if 'IV' in decrypt_info:
 188                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 189                         if not re.match(r'^https?://', decrypt_info['URI']):
 190                             decrypt_info['URI'] = compat_urlparse.urljoin(
 191                                 man_url, decrypt_info['URI'])
 192                         if extra_query:
 193                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 194                         if decrypt_url != decrypt_info['URI']:
 195                             decrypt_info['KEY'] = None
 196                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 197                     media_sequence = int(line[22:])
 198                 elif line.startswith('#EXT-X-BYTERANGE'):
 199                     splitted_byte_range = line[17:].split('@')
 200                     sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 201                     byte_range = {
 202                         'start': sub_range_start,
 203                         'end': sub_range_start + int(splitted_byte_range[0]),
 204                     }
 205                 elif is_ad_fragment_start(line):
 206                     ad_frag_next = True
 207                 elif is_ad_fragment_end(line):
 208                     ad_frag_next = False
 209
 210         self._finish_frag_download(ctx)
 211
 212         return True