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     def _delegate_to_ffmpeg(self, filename, info_dict):
  34         self.report_warning(
  35             'hlsnative has detected features it does not support, '
  36             'extraction will be delegated to ffmpeg')
  37         fd = FFmpegFD(self.ydl, self.params)
  38         for ph in self._progress_hooks:
  39             fd.add_progress_hook(ph)
  40         return fd.real_download(filename, info_dict)
  41
  42     @staticmethod
  43     def can_download(manifest, info_dict):
  44         UNSUPPORTED_FEATURES = (
  45             r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  46             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  47
  48             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  49             # 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)
  50             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  51
  52             # This heuristic also is not correct since segments may not be appended as well.
  53             # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
  54             # no segments will definitely be appended to the end of the playlist.
  55             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  56             #                                 # event media playlists [4]
  57
  58             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  59             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  60             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  61             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  62         )
  63         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  64         check_results.append(can_decrypt_frag or '#EXT-X-KEY:METHOD=AES-128' not in manifest)
  65         return all(check_results)
  66
  67     def real_download(self, filename, info_dict):
  68         if info_dict.get('is_live'):
  69             return self._delegate_to_ffmpeg(filename, info_dict)
  70
  71         man_url = info_dict['url']
  72         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  73
  74         manifest = self.ydl.urlopen(self._prepare_url(info_dict, man_url)).read()
  75
  76         s = manifest.decode('utf-8', 'ignore')
  77
  78         if not self.can_download(s, info_dict):
  79             if info_dict.get('extra_param_to_segment_url'):
  80                 self.report_error('pycrypto not found. Please install it.')
  81                 return False
  82             return self._delegate_to_ffmpeg(filename, info_dict)
  83
  84         total_frags = 0
  85         for line in s.splitlines():
  86             line = line.strip()
  87             if line and not line.startswith('#'):
  88                 total_frags += 1
  89
  90         ctx = {
  91             'filename': filename,
  92             'total_frags': total_frags,
  93         }
  94
  95         self._prepare_and_start_frag_download(ctx)
  96
  97         fragment_retries = self.params.get('fragment_retries', 0)
  98         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
  99         test = self.params.get('test', False)
 100
 101         extra_query = None
 102         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 103         if extra_param_to_segment_url:
 104             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 105         i = 0
 106         media_sequence = 0
 107         decrypt_info = {'METHOD': 'NONE'}
 108         frags_filenames = []
 109         for line in s.splitlines():
 110             line = line.strip()
 111             if line:
 112                 if not line.startswith('#'):
 113                     frag_url = (
 114                         line
 115                         if re.match(r'^https?://', line)
 116                         else compat_urlparse.urljoin(man_url, line))
 117                     frag_name = 'Frag%d' % i
 118                     frag_filename = '%s-%s' % (ctx['tmpfilename'], frag_name)
 119                     if extra_query:
 120                         frag_url = update_url_query(frag_url, extra_query)
 121                     count = 0
 122                     while count <= fragment_retries:
 123                         try:
 124                             success = ctx['dl'].download(frag_filename, {
 125                                 'url': frag_url,
 126                                 'http_headers': info_dict.get('http_headers'),
 127                             })
 128                             if not success:
 129                                 return False
 130                             down, frag_sanitized = sanitize_open(frag_filename, 'rb')
 131                             frag_content = down.read()
 132                             down.close()
 133                             break
 134                         except compat_urllib_error.HTTPError as err:
 135                             # Unavailable (possibly temporary) fragments may be served.
 136                             # First we try to retry then either skip or abort.
 137                             # See https://github.com/rg3/youtube-dl/issues/10165,
 138                             # https://github.com/rg3/youtube-dl/issues/10448).
 139                             count += 1
 140                             if count <= fragment_retries:
 141                                 self.report_retry_fragment(err, frag_name, count, fragment_retries)
 142                     if count > fragment_retries:
 143                         if skip_unavailable_fragments:
 144                             i += 1
 145                             media_sequence += 1
 146                             self.report_skip_fragment(frag_name)
 147                             continue
 148                         self.report_error(
 149                             'giving up after %s fragment retries' % fragment_retries)
 150                         return False
 151                     if decrypt_info['METHOD'] == 'AES-128':
 152                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 153                         frag_content = AES.new(
 154                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 155                     ctx['dest_stream'].write(frag_content)
 156                     frags_filenames.append(frag_sanitized)
 157                     # We only download the first fragment during the test
 158                     if test:
 159                         break
 160                     i += 1
 161                     media_sequence += 1
 162                 elif line.startswith('#EXT-X-KEY'):
 163                     decrypt_info = parse_m3u8_attributes(line[11:])
 164                     if decrypt_info['METHOD'] == 'AES-128':
 165                         if 'IV' in decrypt_info:
 166                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 167                         if not re.match(r'^https?://', decrypt_info['URI']):
 168                             decrypt_info['URI'] = compat_urlparse.urljoin(
 169                                 man_url, decrypt_info['URI'])
 170                         if extra_query:
 171                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 172                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
 173                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 174                     media_sequence = int(line[22:])
 175
 176         self._finish_frag_download(ctx)
 177
 178         for frag_file in frags_filenames:
 179             os.remove(encodeFilename(frag_file))
 180
 181         return True