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_urlparse,
  17     compat_struct_pack,
  18 )
  19 from ..utils import (
  20     encodeFilename,
  21     sanitize_open,
  22     parse_m3u8_attributes,
  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):
  33         UNSUPPORTED_FEATURES = (
  34             r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  35             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  36
  37             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  38             # 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)
  39             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  40
  41             # This heuristic also is not correct since segments may not be appended as well.
  42             # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
  43             # no segments will definitely be appended to the end of the playlist.
  44             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  45             #                                 # event media playlists [4]
  46
  47             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  48             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  49             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  50             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  51         )
  52         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  53         check_results.append(not (re.search(r'#EXT-X-KEY:METHOD=AES-128', manifest) and not can_decrypt_frag))
  54         return all(check_results)
  55
  56     def real_download(self, filename, info_dict):
  57         man_url = info_dict['url']
  58         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  59         manifest = self.ydl.urlopen(man_url).read()
  60
  61         s = manifest.decode('utf-8', 'ignore')
  62
  63         if not self.can_download(s):
  64             self.report_warning(
  65                 'hlsnative has detected features it does not support, '
  66                 'extraction will be delegated to ffmpeg')
  67             fd = FFmpegFD(self.ydl, self.params)
  68             for ph in self._progress_hooks:
  69                 fd.add_progress_hook(ph)
  70             return fd.real_download(filename, info_dict)
  71
  72         total_frags = 0
  73         for line in s.splitlines():
  74             line = line.strip()
  75             if line and not line.startswith('#'):
  76                 total_frags += 1
  77
  78         ctx = {
  79             'filename': filename,
  80             'total_frags': total_frags,
  81         }
  82
  83         self._prepare_and_start_frag_download(ctx)
  84
  85         i = 0
  86         media_sequence = 0
  87         decrypt_info = {'METHOD': 'NONE'}
  88         frags_filenames = []
  89         for line in s.splitlines():
  90             line = line.strip()
  91             if line:
  92                 if not line.startswith('#'):
  93                     frag_url = (
  94                         line
  95                         if re.match(r'^https?://', line)
  96                         else compat_urlparse.urljoin(man_url, line))
  97                     frag_filename = '%s-Frag%d' % (ctx['tmpfilename'], i)
  98                     success = ctx['dl'].download(frag_filename, {'url': frag_url})
  99                     if not success:
 100                         return False
 101                     down, frag_sanitized = sanitize_open(frag_filename, 'rb')
 102                     frag_content = down.read()
 103                     down.close()
 104                     if decrypt_info['METHOD'] == 'AES-128':
 105                         iv = decrypt_info.get('IV') or compat_struct_pack(">8xq", media_sequence)
 106                         frag_content = AES.new(decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 107                     ctx['dest_stream'].write(frag_content)
 108                     frags_filenames.append(frag_sanitized)
 109                     # We only download the first fragment during the test
 110                     if self.params.get('test', False):
 111                         break
 112                     i += 1
 113                     media_sequence += 1
 114                 elif line.startswith('#EXT-X-KEY'):
 115                     decrypt_info = parse_m3u8_attributes(line[11:])
 116                     if decrypt_info['METHOD'] == 'AES-128':
 117                         if 'IV' in decrypt_info:
 118                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:])
 119                         if not re.match(r'^https?://', decrypt_info['URI']):
 120                             decrypt_info['URI'] = compat_urlparse.urljoin(man_url, decrypt_info['URI'])
 121                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
 122                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 123                     media_sequence = int(line[22:])
 124
 125         self._finish_frag_download(ctx)
 126
 127         for frag_file in frags_filenames:
 128             os.remove(encodeFilename(frag_file))
 129
 130         return True