youtube_dl/downloader/hls.py

   1 from __future__ import unicode_literals
   2
   3 import os.path
   4 import re
   5 import binascii
   6 try:
   7     from Crypto.Cipher import AES
   8     can_decrypt_frag = True
   9 except ImportError:
  10     can_decrypt_frag = False
  11
  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     encodeFilename,
  22     sanitize_open,
  23     parse_m3u8_attributes,
  24     update_url_query,
  25 )
  26
  27
  28 class HlsFD(FragmentFD):
  29     """ A limited implementation that does not require ffmpeg """
  30
  31     FD_NAME = 'hlsnative'
  32
  33     @staticmethod
  34     def can_download(manifest):
  35         UNSUPPORTED_FEATURES = (
  36             r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  37             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  38
  39             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  40             # 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)
  41             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  42
  43             # This heuristic also is not correct since segments may not be appended as well.
  44             # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
  45             # no segments will definitely be appended to the end of the playlist.
  46             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  47             #                                 # event media playlists [4]
  48
  49             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  50             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  51             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  52             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  53         )
  54         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  55         check_results.append(can_decrypt_frag or '#EXT-X-KEY:METHOD=AES-128' not in manifest)
  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         manifest = self.ydl.urlopen(man_url).read()
  62
  63         s = manifest.decode('utf-8', 'ignore')
  64
  65         if not self.can_download(s):
  66             self.report_warning(
  67                 'hlsnative has detected features it does not support, '
  68                 'extraction will be delegated to ffmpeg')
  69             fd = FFmpegFD(self.ydl, self.params)
  70             for ph in self._progress_hooks:
  71                 fd.add_progress_hook(ph)
  72             return fd.real_download(filename, info_dict)
  73
  74         total_frags = 0
  75         for line in s.splitlines():
  76             line = line.strip()
  77             if line and not line.startswith('#'):
  78                 total_frags += 1
  79
  80         ctx = {
  81             'filename': filename,
  82             'total_frags': total_frags,
  83         }
  84
  85         self._prepare_and_start_frag_download(ctx)
  86
  87         fragment_retries = self.params.get('fragment_retries', 0)
  88         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
  89         test = self.params.get('test', False)
  90
  91         extra_query = None
  92         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
  93         if extra_param_to_segment_url:
  94             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
  95         i = 0
  96         media_sequence = 0
  97         decrypt_info = {'METHOD': 'NONE'}
  98         frags_filenames = []
  99         for line in s.splitlines():
 100             line = line.strip()
 101             if line:
 102                 if not line.startswith('#'):
 103                     frag_url = (
 104                         line
 105                         if re.match(r'^https?://', line)
 106                         else compat_urlparse.urljoin(man_url, line))
 107                     frag_name = 'Frag%d' % i
 108                     frag_filename = '%s-%s' % (ctx['tmpfilename'], frag_name)
 109                     if extra_query:
 110                         frag_url = update_url_query(frag_url, extra_query)
 111                     count = 0
 112                     while count <= fragment_retries:
 113                         try:
 114                             success = ctx['dl'].download(frag_filename, {'url': frag_url})
 115                             if not success:
 116                                 return False
 117                             down, frag_sanitized = sanitize_open(frag_filename, 'rb')
 118                             frag_content = down.read()
 119                             down.close()
 120                             break
 121                         except compat_urllib_error.HTTPError:
 122                             # Unavailable (possibly temporary) fragments may be served.
 123                             # First we try to retry then either skip or abort.
 124                             # See https://github.com/rg3/youtube-dl/issues/10165,
 125                             # https://github.com/rg3/youtube-dl/issues/10448).
 126                             count += 1
 127                             if count <= fragment_retries:
 128                                 self.report_retry_fragment(frag_name, count, fragment_retries)
 129                     if count > fragment_retries:
 130                         if skip_unavailable_fragments:
 131                             i += 1
 132                             media_sequence += 1
 133                             self.report_skip_fragment(frag_name)
 134                             continue
 135                         self.report_error(
 136                             'giving up after %s fragment retries' % fragment_retries)
 137                         return False
 138                     if decrypt_info['METHOD'] == 'AES-128':
 139                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 140                         frag_content = AES.new(
 141                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 142                     ctx['dest_stream'].write(frag_content)
 143                     frags_filenames.append(frag_sanitized)
 144                     # We only download the first fragment during the test
 145                     if test:
 146                         break
 147                     i += 1
 148                     media_sequence += 1
 149                 elif line.startswith('#EXT-X-KEY'):
 150                     decrypt_info = parse_m3u8_attributes(line[11:])
 151                     if decrypt_info['METHOD'] == 'AES-128':
 152                         if 'IV' in decrypt_info:
 153                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 154                         if not re.match(r'^https?://', decrypt_info['URI']):
 155                             decrypt_info['URI'] = compat_urlparse.urljoin(
 156                                 man_url, decrypt_info['URI'])
 157                         if extra_query:
 158                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 159                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
 160                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 161                     media_sequence = int(line[22:])
 162
 163         self._finish_frag_download(ctx)
 164
 165         for frag_file in frags_filenames:
 166             os.remove(encodeFilename(frag_file))
 167
 168         return True