youtube_dl/downloader/hls.py

   1 from __future__ import unicode_literals
   2
   3 import os.path
   4 import re
   5
   6 from .fragment import FragmentFD
   7 from .external import FFmpegFD
   8
   9 from ..compat import compat_urlparse
  10 from ..utils import (
  11     encodeFilename,
  12     sanitize_open,
  13 )
  14
  15
  16 class HlsFD(FragmentFD):
  17     """ A limited implementation that does not require ffmpeg """
  18
  19     FD_NAME = 'hlsnative'
  20
  21     @staticmethod
  22     def can_download(manifest):
  23         UNSUPPORTED_FEATURES = (
  24             r'#EXT-X-KEY:METHOD=(?!NONE)',  # encrypted streams [1]
  25             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]
  26
  27             # Live streams heuristic does not always work (e.g. geo restricted to Germany
  28             # 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)
  29             # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]
  30
  31             # This heuristic also is not correct since segments may not be appended as well.
  32             # Twitch vods have EXT-X-PLAYLIST-TYPE:EVENT despite no segments will definitely
  33             # be appended to the end of the playlist.
  34             # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
  35                                               # event media playlists [4]
  36
  37             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
  38             # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
  39             # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
  40             # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
  41         )
  42         return all(not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
  43
  44     def real_download(self, filename, info_dict):
  45         man_url = info_dict['url']
  46         self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
  47         manifest = self.ydl.urlopen(man_url).read()
  48
  49         s = manifest.decode('utf-8', 'ignore')
  50
  51         if not self.can_download(s):
  52             self.report_warning(
  53                 'hlsnative has detected features it does not support, '
  54                 'extraction will be delegated to ffmpeg')
  55             fd = FFmpegFD(self.ydl, self.params)
  56             for ph in self._progress_hooks:
  57                 fd.add_progress_hook(ph)
  58             return fd.real_download(filename, info_dict)
  59
  60         fragment_urls = []
  61         for line in s.splitlines():
  62             line = line.strip()
  63             if line and not line.startswith('#'):
  64                 segment_url = (
  65                     line
  66                     if re.match(r'^https?://', line)
  67                     else compat_urlparse.urljoin(man_url, line))
  68                 fragment_urls.append(segment_url)
  69                 # We only download the first fragment during the test
  70                 if self.params.get('test', False):
  71                     break
  72
  73         ctx = {
  74             'filename': filename,
  75             'total_frags': len(fragment_urls),
  76         }
  77
  78         self._prepare_and_start_frag_download(ctx)
  79
  80         frags_filenames = []
  81         for i, frag_url in enumerate(fragment_urls):
  82             frag_filename = '%s-Frag%d' % (ctx['tmpfilename'], i)
  83             success = ctx['dl'].download(frag_filename, {'url': frag_url})
  84             if not success:
  85                 return False
  86             down, frag_sanitized = sanitize_open(frag_filename, 'rb')
  87             ctx['dest_stream'].write(down.read())
  88             down.close()
  89             frags_filenames.append(frag_sanitized)
  90
  91         self._finish_frag_download(ctx)
  92
  93         for frag_file in frags_filenames:
  94             os.remove(encodeFilename(frag_file))
  95
  96         return True