summaryrefslogtreecommitdiff
path: root/youtube_dl/downloader/hls.py
blob: 4989abce12ee236e5c528778e5b95f67d92e165e (plain)
    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         check_results.append(can_decrypt_frag or '#EXT-X-KEY:METHOD=AES-128' not in manifest)
   56         check_results.append(not info_dict.get('is_live'))
   57         return all(check_results)
   58 
   59     def real_download(self, filename, info_dict):
   60         man_url = info_dict['url']
   61         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
   62 
   63         manifest = self.ydl.urlopen(self._prepare_url(info_dict, man_url)).read()
   64 
   65         s = manifest.decode('utf-8', 'ignore')
   66 
   67         if not self.can_download(s, info_dict):
   68             if info_dict.get('extra_param_to_segment_url'):
   69                 self.report_error('pycrypto not found. Please install it.')
   70                 return False
   71             self.report_warning(
   72                 'hlsnative has detected features it does not support, '
   73                 'extraction will be delegated to ffmpeg')
   74             fd = FFmpegFD(self.ydl, self.params)
   75             for ph in self._progress_hooks:
   76                 fd.add_progress_hook(ph)
   77             return fd.real_download(filename, info_dict)
   78 
   79         total_frags = 0
   80         for line in s.splitlines():
   81             line = line.strip()
   82             if line and not line.startswith('#'):
   83                 total_frags += 1
   84 
   85         ctx = {
   86             'filename': filename,
   87             'total_frags': total_frags,
   88         }
   89 
   90         self._prepare_and_start_frag_download(ctx)
   91 
   92         fragment_retries = self.params.get('fragment_retries', 0)
   93         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
   94         test = self.params.get('test', False)
   95 
   96         extra_query = None
   97         extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
   98         if extra_param_to_segment_url:
   99             extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
  100         i = 0
  101         media_sequence = 0
  102         decrypt_info = {'METHOD': 'NONE'}
  103         frags_filenames = []
  104         for line in s.splitlines():
  105             line = line.strip()
  106             if line:
  107                 if not line.startswith('#'):
  108                     frag_url = (
  109                         line
  110                         if re.match(r'^https?://', line)
  111                         else compat_urlparse.urljoin(man_url, line))
  112                     frag_name = 'Frag%d' % i
  113                     frag_filename = '%s-%s' % (ctx['tmpfilename'], frag_name)
  114                     if extra_query:
  115                         frag_url = update_url_query(frag_url, extra_query)
  116                     count = 0
  117                     while count <= fragment_retries:
  118                         try:
  119                             success = ctx['dl'].download(frag_filename, {
  120                                 'url': frag_url,
  121                                 'http_headers': info_dict.get('http_headers'),
  122                             })
  123                             if not success:
  124                                 return False
  125                             down, frag_sanitized = sanitize_open(frag_filename, 'rb')
  126                             frag_content = down.read()
  127                             down.close()
  128                             break
  129                         except compat_urllib_error.HTTPError as err:
  130                             # Unavailable (possibly temporary) fragments may be served.
  131                             # First we try to retry then either skip or abort.
  132                             # See https://github.com/rg3/youtube-dl/issues/10165,
  133                             # https://github.com/rg3/youtube-dl/issues/10448).
  134                             count += 1
  135                             if count <= fragment_retries:
  136                                 self.report_retry_fragment(err, frag_name, count, fragment_retries)
  137                     if count > fragment_retries:
  138                         if skip_unavailable_fragments:
  139                             i += 1
  140                             media_sequence += 1
  141                             self.report_skip_fragment(frag_name)
  142                             continue
  143                         self.report_error(
  144                             'giving up after %s fragment retries' % fragment_retries)
  145                         return False
  146                     if decrypt_info['METHOD'] == 'AES-128':
  147                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
  148                         frag_content = AES.new(
  149                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
  150                     ctx['dest_stream'].write(frag_content)
  151                     frags_filenames.append(frag_sanitized)
  152                     # We only download the first fragment during the test
  153                     if test:
  154                         break
  155                     i += 1
  156                     media_sequence += 1
  157                 elif line.startswith('#EXT-X-KEY'):
  158                     decrypt_info = parse_m3u8_attributes(line[11:])
  159                     if decrypt_info['METHOD'] == 'AES-128':
  160                         if 'IV' in decrypt_info:
  161                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
  162                         if not re.match(r'^https?://', decrypt_info['URI']):
  163                             decrypt_info['URI'] = compat_urlparse.urljoin(
  164                                 man_url, decrypt_info['URI'])
  165                         if extra_query:
  166                             decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
  167                         decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read()
  168                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
  169                     media_sequence = int(line[22:])
  170 
  171         self._finish_frag_download(ctx)
  172 
  173         for frag_file in frags_filenames:
  174             os.remove(encodeFilename(frag_file))
  175 
  176         return True

Generated by cgit