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 ..downloader import _get_real_downloader
  12 from .fragment import FragmentFD
  13 from .external import FFmpegFD
  14
  15 from ..compat import (
  16     compat_urllib_error,
  17     compat_urlparse,
  18     compat_struct_pack,
  19 )
  20 from ..utils import (
  21     parse_m3u8_attributes,
  22     update_url_query,
  23 )
  24
  25
  26 class HlsFD(FragmentFD):
  27     """ A limited implementation that does not require ffmpeg """
  28
  29     FD_NAME = 'hlsnative'
  30
  31     @staticmethod
  32     def can_download(manifest, info_dict, allow_unplayable_formats=False, with_crypto=can_decrypt_frag):
  33         UNSUPPORTED_FEATURES = [
  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             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  47             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  48             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  49             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  50             # 5. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.5
  51         ]
  52         if not allow_unplayable_formats:
  53             UNSUPPORTED_FEATURES += [
  54                 r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  55             ]
  56         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  57         is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
  58         check_results.append(with_crypto or not is_aes128_enc)
  59         check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
  60         check_results.append(not info_dict.get('is_live'))
  61         return all(check_results)
  62
  63     def real_download(self, filename, info_dict):
  64         man_url = info_dict['url']
  65         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  66
  67         urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
  68         man_url = urlh.geturl()
  69         s = urlh.read().decode('utf-8', 'ignore')
  70
  71         if not self.can_download(s, info_dict, self.params.get('allow_unplayable_formats')):
  72             if info_dict.get('extra_param_to_segment_url') or info_dict.get('_decryption_key_url'):
  73                 self.report_error('pycryptodome not found. Please install it.')
  74                 return False
  75             if self.can_download(s, info_dict, with_crypto=True):
  76                 self.report_warning('pycryptodome is needed to download this file with hlsnative')
  77             self.report_warning(
  78                 'hlsnative has detected features it does not support, '
  79                 'extraction will be delegated to ffmpeg')
  80             fd = FFmpegFD(self.ydl, self.params)
  81             # TODO: Make progress updates work without hooking twice
  82             # for ph in self._progress_hooks:
  83             #     fd.add_progress_hook(ph)
  84             return fd.real_download(filename, info_dict)
  85
  86         real_downloader = _get_real_downloader(info_dict, 'frag_urls', self.params, None)
  87
  88         def is_ad_fragment_start(s):
  89             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s
  90                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',ad'))
  91
  92         def is_ad_fragment_end(s):
  93             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=master' in s
  94                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',segment'))
  95
  96         fragment_urls = []
  97
  98         media_frags = 0
  99         ad_frags = 0
 100         ad_frag_next = False
 101         for line in s.splitlines():
 102             line = line.strip()
 103             if not line:
 104                 continue
 105             if line.startswith('#'):
 106                 if is_ad_fragment_start(line):
 107                     ad_frag_next = True
 108                 elif is_ad_fragment_end(line):
 109                     ad_frag_next = False
 110                 continue
 111             if ad_frag_next:
 112                 ad_frags += 1
 113                 continue
 114             media_frags += 1
 115
 116         ctx = {
 117             'filename': filename,
 118             'total_frags': media_frags,
 119             'ad_frags': ad_frags,
 120         }
 121
 122         if real_downloader:
 123             self._prepare_external_frag_download(ctx)
 124         else:
 125             self._prepare_and_start_frag_download(ctx)
 126
 127         fragment_retries = self.params.get('fragment_retries', 0)
 128         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 129         test = self.params.get('test', False)
 130
 131         format_index = info_dict.get('format_index')
 132         extra_query = None
 133         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 134         if extra_param_to_segment_url:
 135             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 136         i = 0
 137         media_sequence = 0
 138         decrypt_info = {'METHOD': 'NONE'}
 139         key_list = []
 140         byte_range = {}
 141         discontinuity_count = 0
 142         frag_index = 0
 143         ad_frag_next = False
 144         for line in s.splitlines():
 145             line = line.strip()
 146             download_frag = False
 147             if line:
 148                 if not line.startswith('#'):
 149                     if format_index and discontinuity_count != format_index:
 150                         continue
 151                     if ad_frag_next:
 152                         continue
 153                     frag_index += 1
 154                     if frag_index <= ctx['fragment_index']:
 155                         continue
 156                     frag_url = (
 157                         line
 158                         if re.match(r'^https?://', line)
 159                         else compat_urlparse.urljoin(man_url, line))
 160                     if extra_query:
 161                         frag_url = update_url_query(frag_url, extra_query)
 162
 163                     if real_downloader:
 164                         fragment_urls.append(frag_url)
 165                         continue
 166                     download_frag = True
 167
 168                 elif line.startswith('#EXT-X-MAP'):
 169                     if format_index and discontinuity_count != format_index:
 170                         continue
 171                     if frag_index > 0:
 172                         self.report_error(
 173                             'initialization fragment found after media fragments, unable to download')
 174                         return False
 175                     frag_index += 1
 176                     map_info = parse_m3u8_attributes(line[11:])
 177                     frag_url = (
 178                         map_info.get('URI')
 179                         if re.match(r'^https?://', map_info.get('URI'))
 180                         else compat_urlparse.urljoin(man_url, map_info.get('URI')))
 181                     if extra_query:
 182                         frag_url = update_url_query(frag_url, extra_query)
 183                     if real_downloader:
 184                         fragment_urls.append(frag_url)
 185                         continue
 186
 187                     if map_info.get('BYTERANGE'):
 188                         splitted_byte_range = map_info.get('BYTERANGE').split('@')
 189                         sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 190                         byte_range = {
 191                             'start': sub_range_start,
 192                             'end': sub_range_start + int(splitted_byte_range[0]),
 193                         }
 194                     download_frag = True
 195
 196                 elif line.startswith('#EXT-X-KEY'):
 197                     decrypt_url = decrypt_info.get('URI')
 198                     decrypt_info = parse_m3u8_attributes(line[11:])
 199                     if decrypt_info['METHOD'] == 'AES-128':
 200                         if 'IV' in decrypt_info:
 201                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 202                         if not re.match(r'^https?://', decrypt_info['URI']):
 203                             decrypt_info['URI'] = compat_urlparse.urljoin(
 204                                 man_url, decrypt_info['URI'])
 205                         if extra_query:
 206                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 207                         if decrypt_url != decrypt_info['URI']:
 208                             decrypt_info['KEY'] = None
 209                     key_data = decrypt_info.copy()
 210                     key_data['INDEX'] = frag_index
 211                     key_list.append(key_data)
 212
 213                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 214                     media_sequence = int(line[22:])
 215                 elif line.startswith('#EXT-X-BYTERANGE'):
 216                     splitted_byte_range = line[17:].split('@')
 217                     sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 218                     byte_range = {
 219                         'start': sub_range_start,
 220                         'end': sub_range_start + int(splitted_byte_range[0]),
 221                     }
 222                 elif is_ad_fragment_start(line):
 223                     ad_frag_next = True
 224                 elif is_ad_fragment_end(line):
 225                     ad_frag_next = False
 226                 elif line.startswith('#EXT-X-DISCONTINUITY'):
 227                     discontinuity_count += 1
 228
 229                 if download_frag:
 230                     count = 0
 231                     headers = info_dict.get('http_headers', {})
 232                     if byte_range:
 233                         headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1)
 234                     while count <= fragment_retries:
 235                         try:
 236                             success, frag_content = self._download_fragment(
 237                                 ctx, frag_url, info_dict, headers)
 238                             if not success:
 239                                 return False
 240                             break
 241                         except compat_urllib_error.HTTPError as err:
 242                             # Unavailable (possibly temporary) fragments may be served.
 243                             # First we try to retry then either skip or abort.
 244                             # See https://github.com/ytdl-org/youtube-dl/issues/10165,
 245                             # https://github.com/ytdl-org/youtube-dl/issues/10448).
 246                             count += 1
 247                             if count <= fragment_retries:
 248                                 self.report_retry_fragment(err, frag_index, count, fragment_retries)
 249                     if count > fragment_retries:
 250                         if skip_unavailable_fragments:
 251                             i += 1
 252                             media_sequence += 1
 253                             self.report_skip_fragment(frag_index)
 254                             continue
 255                         self.report_error(
 256                             'giving up after %s fragment retries' % fragment_retries)
 257                         return False
 258
 259                     if decrypt_info['METHOD'] == 'AES-128':
 260                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 261                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 262                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 263                         # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
 264                         # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
 265                         # not what it decrypts to.
 266                         if not test:
 267                             frag_content = AES.new(
 268                                 decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 269                     self._append_fragment(ctx, frag_content)
 270                     # We only download the first fragment during the test
 271                     if test:
 272                         break
 273                     i += 1
 274                     media_sequence += 1
 275
 276         if real_downloader:
 277             info_copy = info_dict.copy()
 278             info_copy['url_list'] = fragment_urls
 279             info_copy['key_list'] = key_list
 280             fd = real_downloader(self.ydl, self.params)
 281             # TODO: Make progress updates work without hooking twice
 282             # for ph in self._progress_hooks:
 283             #     fd.add_progress_hook(ph)
 284             success = fd.real_download(filename, info_copy)
 285             if not success:
 286                 return False
 287         else:
 288             self._finish_frag_download(ctx)
 289         return True