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     update_url_query,
  24 )
  25
  26
  27 class HlsFD(FragmentFD):
  28     """ A limited implementation that does not require ffmpeg """
  29
  30     FD_NAME = 'hlsnative'
  31
  32     @staticmethod
  33     def can_download(manifest):
  34         UNSUPPORTED_FEATURES = (
  35             r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
  36             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  37
  38             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  39             # 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)
  40             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  41
  42             # This heuristic also is not correct since segments may not be appended as well.
  43             # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
  44             # no segments will definitely be appended to the end of the playlist.
  45             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  46             #                                 # event media playlists [4]
  47
  48             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  49             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  50             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  51             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  52         )
  53         check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
  54         check_results.append(can_decrypt_frag or '#EXT-X-KEY:METHOD=AES-128' not in manifest)
  55         return all(check_results)
  56
  57     def real_download(self, filename, info_dict):
  58         man_url = info_dict['url']
  59         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  60         manifest = self.ydl.urlopen(man_url).read()
  61
  62         s = manifest.decode('utf-8', 'ignore')
  63
  64         if not self.can_download(s):
  65             self.report_warning(
  66                 'hlsnative has detected features it does not support, '
  67                 'extraction will be delegated to ffmpeg')
  68             fd = FFmpegFD(self.ydl, self.params)
  69             for ph in self._progress_hooks:
  70                 fd.add_progress_hook(ph)
  71             return fd.real_download(filename, info_dict)
  72
  73         total_frags = 0
  74         for line in s.splitlines():
  75             line = line.strip()
  76             if line and not line.startswith('#'):
  77                 total_frags += 1
  78
  79         ctx = {
  80             'filename': filename,
  81             'total_frags': total_frags,
  82         }
  83
  84         self._prepare_and_start_frag_download(ctx)
  85
  86         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
  87         i = 0
  88         media_sequence = 0
  89         decrypt_info = {'METHOD': 'NONE'}
  90         frags_filenames = []
  91         for line in s.splitlines():
  92             line = line.strip()
  93             if line:
  94                 if not line.startswith('#'):
  95                     frag_url = (
  96                         line
  97                         if re.match(r'^https?://', line)
  98                         else compat_urlparse.urljoin(man_url, line))
  99                     frag_filename = '%s-Frag%d' % (ctx['tmpfilename'], i)
 100                     if extra_param_to_segment_url:
 101                         frag_url = update_url_query(frag_url, extra_param_to_segment_url)
 102                     success = ctx['dl'].download(frag_filename, {'url': frag_url})
 103                     if not success:
 104                         return False
 105                     down, frag_sanitized = sanitize_open(frag_filename, 'rb')
 106                     frag_content = down.read()
 107                     down.close()
 108                     if decrypt_info['METHOD'] == 'AES-128':
 109                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 110                         frag_content = AES.new(
 111                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 112                     ctx['dest_stream'].write(frag_content)
 113                     frags_filenames.append(frag_sanitized)
 114                     # We only download the first fragment during the test
 115                     if self.params.get('test', False):
 116                         break
 117                     i += 1
 118                     media_sequence += 1
 119                 elif line.startswith('#EXT-X-KEY'):
 120                     decrypt_info = parse_m3u8_attributes(line[11:])
 121                     if decrypt_info['METHOD'] == 'AES-128':
 122                         if 'IV' in decrypt_info:
 123                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 124                         if not re.match(r'^https?://', decrypt_info['URI']):
 125                             decrypt_info['URI'] = compat_urlparse.urljoin(
 126                                 man_url, decrypt_info['URI'])
 127                         if extra_param_to_segment_url:
 128                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_param_to_segment_url)
 129                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
 130                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 131                     media_sequence = int(line[22:])
 132
 133         self._finish_frag_download(ctx)
 134
 135         for frag_file in frags_filenames:
 136             os.remove(encodeFilename(frag_file))
 137
 138         return True