_ Git - youtube-dl/blob - youtube_dl/downloader/dash.py

   1 from __future__ import unicode_literals
   2
   3 import os
   4 import re
   5
   6 from .fragment import FragmentFD
   7 from ..compat import compat_urllib_error
   8 from ..utils import (
   9     sanitize_open,
  10     encodeFilename,
  11 )
  12
  13
  14 class DashSegmentsFD(FragmentFD):
  15     """
  16     Download segments in a DASH manifest
  17     """
  18
  19     FD_NAME = 'dashsegments'
  20
  21     def real_download(self, filename, info_dict):
  22         base_url = info_dict['url']
  23         segment_urls = [info_dict['segment_urls'][0]] if self.params.get('test', False) else info_dict['segment_urls']
  24         initialization_url = info_dict.get('initialization_url')
  25
  26         ctx = {
  27             'filename': filename,
  28             'total_frags': len(segment_urls) + (1 if initialization_url else 0),
  29         }
  30
  31         self._prepare_and_start_frag_download(ctx)
  32
  33         def combine_url(base_url, target_url):
  34             if re.match(r'^https?://', target_url):
  35                 return target_url
  36             return '%s%s%s' % (base_url, '' if base_url.endswith('/') else '/', target_url)
  37
  38         segments_filenames = []
  39
  40         fragment_retries = self.params.get('fragment_retries', 0)
  41         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
  42
  43         def process_segment(segment, tmp_filename, fatal):
  44             target_url, segment_name = segment
  45             target_filename = '%s-%s' % (tmp_filename, segment_name)
  46             count = 0
  47             while count <= fragment_retries:
  48                 try:
  49                     success = ctx['dl'].download(target_filename, {'url': combine_url(base_url, target_url)})
  50                     if not success:
  51                         return False
  52                     down, target_sanitized = sanitize_open(target_filename, 'rb')
  53                     ctx['dest_stream'].write(down.read())
  54                     down.close()
  55                     segments_filenames.append(target_sanitized)
  56                     break
  57                 except compat_urllib_error.HTTPError as err:
  58                     # YouTube may often return 404 HTTP error for a fragment causing the
  59                     # whole download to fail. However if the same fragment is immediately
  60                     # retried with the same request data this usually succeeds (1-2 attemps
  61                     # is usually enough) thus allowing to download the whole file successfully.
  62                     # To be future-proof we will retry all fragments that fail with any
  63                     # HTTP error.
  64                     count += 1
  65                     if count <= fragment_retries:
  66                         self.report_retry_fragment(err, segment_name, count, fragment_retries)
  67             if count > fragment_retries:
  68                 if not fatal:
  69                     self.report_skip_fragment(segment_name)
  70                     return True
  71                 self.report_error('giving up after %s fragment retries' % fragment_retries)
  72                 return False
  73             return True
  74
  75         segments_to_download = [(initialization_url, 'Init')] if initialization_url else []
  76         segments_to_download.extend([
  77             (segment_url, 'Seg%d' % i)
  78             for i, segment_url in enumerate(segment_urls)])
  79
  80         for i, segment in enumerate(segments_to_download):
  81             # In DASH, the first segment contains necessary headers to
  82             # generate a valid MP4 file, so always abort for the first segment
  83             fatal = i == 0 or not skip_unavailable_fragments
  84             if not process_segment(segment, ctx['tmpfilename'], fatal):
  85                 return False
  86
  87         self._finish_frag_download(ctx)
  88
  89         for segment_file in segments_filenames:
  90             os.remove(encodeFilename(segment_file))
  91
  92         return True