Raphaël G. Git Repositories - youtubedl/blob - 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'):
  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(s):
  79             return (s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s or
  80                     s.startswith('#UPLYNK-SEGMENT') and s.endswith(',ad'))
  81
  82         media_frags = 0
  83         ad_frags = 0
  84         ad_frag_next = False
  85         for line in s.splitlines():
  86             line = line.strip()
  87             if not line:
  88                 continue
  89             if line.startswith('#'):
  90                 if is_ad_fragment(line):
  91                     ad_frags += 1
  92                     ad_frag_next = True
  93                 continue
  94             if ad_frag_next:
  95                 ad_frag_next = False
  96                 continue
  97             media_frags += 1
  98
  99         ctx = {
 100             'filename': filename,
 101             'total_frags': media_frags,
 102             'ad_frags': ad_frags,
 103         }
 104
 105         self._prepare_and_start_frag_download(ctx)
 106
 107         fragment_retries = self.params.get('fragment_retries', 0)
 108         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 109         test = self.params.get('test', False)
 110
 111         extra_query = None
 112         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 113         if extra_param_to_segment_url:
 114             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 115         i = 0
 116         media_sequence = 0
 117         decrypt_info = {'METHOD': 'NONE'}
 118         byte_range = {}
 119         frag_index = 0
 120         ad_frag_next = False
 121         for line in s.splitlines():
 122             line = line.strip()
 123             if line:
 124                 if not line.startswith('#'):
 125                     if ad_frag_next:
 126                         ad_frag_next = False
 127                         continue
 128                     frag_index += 1
 129                     if frag_index <= ctx['fragment_index']:
 130                         continue
 131                     frag_url = (
 132                         line
 133                         if re.match(r'^https?://', line)
 134                         else compat_urlparse.urljoin(man_url, line))
 135                     if extra_query:
 136                         frag_url = update_url_query(frag_url, extra_query)
 137                     count = 0
 138                     headers = info_dict.get('http_headers', {})
 139                     if byte_range:
 140                         headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'])
 141                     while count <= fragment_retries:
 142                         try:
 143                             success, frag_content = self._download_fragment(
 144                                 ctx, frag_url, info_dict, headers)
 145                             if not success:
 146                                 return False
 147                             break
 148                         except compat_urllib_error.HTTPError as err:
 149                             # Unavailable (possibly temporary) fragments may be served.
 150                             # First we try to retry then either skip or abort.
 151                             # See https://github.com/rg3/youtube-dl/issues/10165,
 152                             # https://github.com/rg3/youtube-dl/issues/10448).
 153                             count += 1
 154                             if count <= fragment_retries:
 155                                 self.report_retry_fragment(err, frag_index, count, fragment_retries)
 156                     if count > fragment_retries:
 157                         if skip_unavailable_fragments:
 158                             i += 1
 159                             media_sequence += 1
 160                             self.report_skip_fragment(frag_index)
 161                             continue
 162                         self.report_error(
 163                             'giving up after %s fragment retries' % fragment_retries)
 164                         return False
 165                     if decrypt_info['METHOD'] == 'AES-128':
 166                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 167                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 168                             self._prepare_url(info_dict, decrypt_info['URI'])).read()
 169                         frag_content = AES.new(
 170                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 171                     self._append_fragment(ctx, frag_content)
 172                     # We only download the first fragment during the test
 173                     if test:
 174                         break
 175                     i += 1
 176                     media_sequence += 1
 177                 elif line.startswith('#EXT-X-KEY'):
 178                     decrypt_url = decrypt_info.get('URI')
 179                     decrypt_info = parse_m3u8_attributes(line[11:])
 180                     if decrypt_info['METHOD'] == 'AES-128':
 181                         if 'IV' in decrypt_info:
 182                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 183                         if not re.match(r'^https?://', decrypt_info['URI']):
 184                             decrypt_info['URI'] = compat_urlparse.urljoin(
 185                                 man_url, decrypt_info['URI'])
 186                         if extra_query:
 187                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 188                         if decrypt_url != decrypt_info['URI']:
 189                             decrypt_info['KEY'] = None
 190                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 191                     media_sequence = int(line[22:])
 192                 elif line.startswith('#EXT-X-BYTERANGE'):
 193                     splitted_byte_range = line[17:].split('@')
 194                     sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 195                     byte_range = {
 196                         'start': sub_range_start,
 197                         'end': sub_range_start + int(splitted_byte_range[0]),
 198                     }
 199                 elif is_ad_fragment(line):
 200                     ad_frag_next = True
 201
 202         self._finish_frag_download(ctx)
 203
 204         return True