_ Git - youtube-dl/blob - 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, info_dict):
  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         is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
  56         check_results.append(can_decrypt_frag or not is_aes128_enc)
  57         check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
  58         check_results.append(not info_dict.get('is_live'))
  59         return all(check_results)
  60
  61     def real_download(self, filename, info_dict):
  62         man_url = info_dict['url']
  63         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  64
  65         manifest = self.ydl.urlopen(self._prepare_url(info_dict, man_url)).read()
  66
  67         s = manifest.decode('utf-8', 'ignore')
  68
  69         if not self.can_download(s, info_dict):
  70             if info_dict.get('extra_param_to_segment_url'):
  71                 self.report_error('pycrypto not found. Please install it.')
  72                 return False
  73             self.report_warning(
  74                 'hlsnative has detected features it does not support, '
  75                 'extraction will be delegated to ffmpeg')
  76             fd = FFmpegFD(self.ydl, self.params)
  77             for ph in self._progress_hooks:
  78                 fd.add_progress_hook(ph)
  79             return fd.real_download(filename, info_dict)
  80
  81         total_frags = 0
  82         for line in s.splitlines():
  83             line = line.strip()
  84             if line and not line.startswith('#'):
  85                 total_frags += 1
  86
  87         ctx = {
  88             'filename': filename,
  89             'total_frags': total_frags,
  90         }
  91
  92         self._prepare_and_start_frag_download(ctx)
  93
  94         fragment_retries = self.params.get('fragment_retries', 0)
  95         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
  96         test = self.params.get('test', False)
  97
  98         extra_query = None
  99         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
 100         if extra_param_to_segment_url:
 101             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
 102         i = 0
 103         media_sequence = 0
 104         decrypt_info = {'METHOD': 'NONE'}
 105         byte_range = {}
 106         frags_filenames = []
 107         for line in s.splitlines():
 108             line = line.strip()
 109             if line:
 110                 if not line.startswith('#'):
 111                     frag_url = (
 112                         line
 113                         if re.match(r'^https?://', line)
 114                         else compat_urlparse.urljoin(man_url, line))
 115                     frag_name = 'Frag%d' % i
 116                     frag_filename = '%s-%s' % (ctx['tmpfilename'], frag_name)
 117                     if extra_query:
 118                         frag_url = update_url_query(frag_url, extra_query)
 119                     count = 0
 120                     headers = info_dict.get('http_headers', {})
 121                     if byte_range:
 122                         headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'])
 123                     while count <= fragment_retries:
 124                         try:
 125                             success = ctx['dl'].download(frag_filename, {
 126                                 'url': frag_url,
 127                                 'http_headers': headers,
 128                             })
 129                             if not success:
 130                                 return False
 131                             down, frag_sanitized = sanitize_open(frag_filename, 'rb')
 132                             frag_content = down.read()
 133                             down.close()
 134                             break
 135                         except compat_urllib_error.HTTPError as err:
 136                             # Unavailable (possibly temporary) fragments may be served.
 137                             # First we try to retry then either skip or abort.
 138                             # See https://github.com/rg3/youtube-dl/issues/10165,
 139                             # https://github.com/rg3/youtube-dl/issues/10448).
 140                             count += 1
 141                             if count <= fragment_retries:
 142                                 self.report_retry_fragment(err, frag_name, count, fragment_retries)
 143                     if count > fragment_retries:
 144                         if skip_unavailable_fragments:
 145                             i += 1
 146                             media_sequence += 1
 147                             self.report_skip_fragment(frag_name)
 148                             continue
 149                         self.report_error(
 150                             'giving up after %s fragment retries' % fragment_retries)
 151                         return False
 152                     if decrypt_info['METHOD'] == 'AES-128':
 153                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
 154                         frag_content = AES.new(
 155                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
 156                     ctx['dest_stream'].write(frag_content)
 157                     frags_filenames.append(frag_sanitized)
 158                     # We only download the first fragment during the test
 159                     if test:
 160                         break
 161                     i += 1
 162                     media_sequence += 1
 163                 elif line.startswith('#EXT-X-KEY'):
 164                     decrypt_info = parse_m3u8_attributes(line[11:])
 165                     if decrypt_info['METHOD'] == 'AES-128':
 166                         if 'IV' in decrypt_info:
 167                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
 168                         if not re.match(r'^https?://', decrypt_info['URI']):
 169                             decrypt_info['URI'] = compat_urlparse.urljoin(
 170                                 man_url, decrypt_info['URI'])
 171                         if extra_query:
 172                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
 173                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
 174                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
 175                     media_sequence = int(line[22:])
 176                 elif line.startswith('#EXT-X-BYTERANGE'):
 177                     splitted_byte_range = line[17:].split('@')
 178                     sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
 179                     byte_range = {
 180                         'start': sub_range_start,
 181                         'end': sub_range_start + int(splitted_byte_range[0]),
 182                     }
 183
 184         self._finish_frag_download(ctx)
 185
 186         for frag_file in frags_filenames:
 187             os.remove(encodeFilename(frag_file))
 188
 189         return True