youtube_dl/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
  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         )
  51         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  52         is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
  53         check_results.append(can_decrypt_frag or not is_aes128_enc)
  54         check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
  55         check_results.append(not info_dict.get('is_live'))
  56         return all(check_results)
  57
  58     def real_download(self, filename, info_dict):
  59         man_url = info_dict['url']
  60         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  61
  62         urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
  63         man_url = urlh.geturl()
  64         s = urlh.read().decode('utf-8', 'ignore')
  65
  66         if not self.can_download(s, info_dict):
  67             if info_dict.get('extra_param_to_segment_url') or info_dict.get('_decryption_key_url'):
  68                 self.report_error('pycrypto not found. Please install it.')
  69                 return False
  70             self.report_warning(
  71                 'hlsnative has detected features it does not support, '
  72                 'extraction will be delegated to ffmpeg')
  73             fd = FFmpegFD(self.ydl, self.params)
  74             for ph in self._progress_hooks:
  75                 fd.add_progress_hook(ph)
  76             return fd.real_download(filename, info_dict)
  77
  78         def is_ad_fragment_start(s):
  79             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s
  80                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',ad'))
  81
  82         def is_ad_fragment_end(s):
  83             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=master' in s
  84                     or s.startswith('#UPLYNK-SEGMENT') and s.endswith(',segment'))
  85
  86         media_frags = 0
  87         ad_frags = 0
  88         ad_frag_next = False
  89         for line in s.splitlines():
  90             line = line.strip()
  91             if not line:
  92                 continue
  93             if line.startswith('#'):
  94                 if is_ad_fragment_start(line):
  95                     ad_frag_next = True
  96                 elif is_ad_fragment_end(line):
  97                     ad_frag_next = False
  98                 continue
  99             if ad_frag_next:
 100                 ad_frags += 1
 101                 continue
 102             media_frags += 1
 103
 104         ctx = {
 105             'filename': filename,
 106             'total_frags': media_frags,
 107             'ad_frags': ad_frags,
 108         }
 109
 110         self._prepare_and_start_frag_download(ctx)
 111
 112         fragment_retries = self.params.get('fragment_retries', 0)
 113         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 114         test = self.params.get('test', False)
 115
 116         extra_query = None
 117         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 118         if extra_param_to_segment_url:
 119             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 120         i = 0
 121         media_sequence = 0
 122         decrypt_info = {'METHOD': 'NONE'}
 123         byte_range = {}
 124         frag_index = 0
 125         ad_frag_next = False
 126         for line in s.splitlines():
 127             line = line.strip()
 128             if line:
 129                 if not line.startswith('#'):
 130                     if ad_frag_next:
 131                         continue
 132                     frag_index += 1
 133                     if frag_index <= ctx['fragment_index']:
 134                         continue
 135                     frag_url = (
 136                         line
 137                         if re.match(r'^https?://', line)
 138                         else compat_urlparse.urljoin(man_url, line))
 139                     if extra_query:
 140                         frag_url = update_url_query(frag_url, extra_query)
 141                     count = 0
 142                     headers = info_dict.get('http_headers', {})
 143                     if byte_range:
 144                         headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'])
 145                     while count <= fragment_retries:
 146                         try:
 147                             success, frag_content = self._download_fragment(
 148                                 ctx, frag_url, info_dict, headers)
 149                             if not success:
 150                                 return False
 151                             break
 152                         except compat_urllib_error.HTTPError as err:
 153                             # Unavailable (possibly temporary) fragments may be served.
 154                             # First we try to retry then either skip or abort.
 155                             # See https://github.com/ytdl-org/youtube-dl/issues/10165,
 156                             # https://github.com/ytdl-org/youtube-dl/issues/10448).
 157                             count += 1
 158                             if count <= fragment_retries:
 159                                 self.report_retry_fragment(err, frag_index, count, fragment_retries)
 160                     if count > fragment_retries:
 161                         if skip_unavailable_fragments:
 162                             i += 1
 163                             media_sequence += 1
 164                             self.report_skip_fragment(frag_index)
 165                             continue
 166                         self.report_error(
 167                             'giving up after %s fragment retries' % fragment_retries)
 168                         return False
 169                     if decrypt_info['METHOD'] == 'AES-128':
 170                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 171                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 172                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 173                         frag_content = AES.new(
 174                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 175                     self._append_fragment(ctx, frag_content)
 176                     # We only download the first fragment during the test
 177                     if test:
 178                         break
 179                     i += 1
 180                     media_sequence += 1
 181                 elif line.startswith('#EXT-X-KEY'):
 182                     decrypt_url = decrypt_info.get('URI')
 183                     decrypt_info = parse_m3u8_attributes(line[11:])
 184                     if decrypt_info['METHOD'] == 'AES-128':
 185                         if 'IV' in decrypt_info:
 186                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 187                         if not re.match(r'^https?://', decrypt_info['URI']):
 188                             decrypt_info['URI'] = compat_urlparse.urljoin(
 189                                 man_url, decrypt_info['URI'])
 190                         if extra_query:
 191                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 192                         if decrypt_url != decrypt_info['URI']:
 193                             decrypt_info['KEY'] = None
 194                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 195                     media_sequence = int(line[22:])
 196                 elif line.startswith('#EXT-X-BYTERANGE'):
 197                     splitted_byte_range = line[17:].split('@')
 198                     sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 199                     byte_range = {
 200                         'start': sub_range_start,
 201                         'end': sub_range_start + int(splitted_byte_range[0]),
 202                     }
 203                 elif is_ad_fragment_start(line):
 204                     ad_frag_next = True
 205                 elif is_ad_fragment_end(line):
 206                     ad_frag_next = False
 207
 208         self._finish_frag_download(ctx)
 209
 210         return True