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

   1 from __future__ import division, unicode_literals
   2
   3 import os
   4 import time
   5 import json
   6
   7 from .common import FileDownloader
   8 from .http import HttpFD
   9 from ..utils import (
  10     error_to_compat_str,
  11     encodeFilename,
  12     sanitize_open,
  13     sanitized_Request,
  14 )
  15
  16
  17 class HttpQuietDownloader(HttpFD):
  18     def to_screen(self, *args, **kargs):
  19         pass
  20
  21
  22 class FragmentFD(FileDownloader):
  23     """
  24     A base file downloader class for fragmented media (e.g. f4m/m3u8 manifests).
  25
  26     Available options:
  27
  28     fragment_retries:   Number of times to retry a fragment for HTTP error (DASH
  29                         and hlsnative only)
  30     skip_unavailable_fragments:
  31                         Skip unavailable fragments (DASH and hlsnative only)
  32     keep_fragments:     Keep downloaded fragments on disk after downloading is
  33                         finished
  34     """
  35
  36     def report_retry_fragment(self, err, frag_index, count, retries):
  37         self.to_screen(
  38             '[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s)...'
  39             % (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
  40
  41     def report_skip_fragment(self, frag_index):
  42         self.to_screen('[download] Skipping fragment %d...' % frag_index)
  43
  44     def _prepare_url(self, info_dict, url):
  45         headers = info_dict.get('http_headers')
  46         return sanitized_Request(url, None, headers) if headers else url
  47
  48     def _prepare_and_start_frag_download(self, ctx):
  49         self._prepare_frag_download(ctx)
  50         self._start_frag_download(ctx)
  51
  52     @staticmethod
  53     def __do_ytdl_file(ctx):
  54         return not ctx['live'] and not ctx['tmpfilename'] == '-'
  55
  56     def _read_ytdl_file(self, ctx):
  57         stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
  58         ctx['fragment_index'] = json.loads(stream.read())['download']['current_fragment_index']
  59         stream.close()
  60
  61     def _write_ytdl_file(self, ctx):
  62         frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
  63         frag_index_stream.write(json.dumps({
  64             'download': {
  65                 'current_fragment_index': ctx['fragment_index']
  66             },
  67         }))
  68         frag_index_stream.close()
  69
  70     def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
  71         fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
  72         success = ctx['dl'].download(fragment_filename, {
  73             'url': frag_url,
  74             'http_headers': headers or info_dict.get('http_headers'),
  75         })
  76         if not success:
  77             return False, None
  78         down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
  79         ctx['fragment_filename_sanitized'] = frag_sanitized
  80         frag_content = down.read()
  81         down.close()
  82         return True, frag_content
  83
  84     def _append_fragment(self, ctx, frag_content):
  85         try:
  86             ctx['dest_stream'].write(frag_content)
  87         finally:
  88             if self.__do_ytdl_file(ctx):
  89                 self._write_ytdl_file(ctx)
  90             if not self.params.get('keep_fragments', False):
  91                 os.remove(ctx['fragment_filename_sanitized'])
  92             del ctx['fragment_filename_sanitized']
  93
  94     def _prepare_frag_download(self, ctx):
  95         if 'live' not in ctx:
  96             ctx['live'] = False
  97         self.to_screen(
  98             '[%s] Total fragments: %s'
  99             % (self.FD_NAME, ctx['total_frags'] if not ctx['live'] else 'unknown (live)'))
 100         self.report_destination(ctx['filename'])
 101         dl = HttpQuietDownloader(
 102             self.ydl,
 103             {
 104                 'continuedl': True,
 105                 'quiet': True,
 106                 'noprogress': True,
 107                 'ratelimit': self.params.get('ratelimit'),
 108                 'retries': self.params.get('retries', 0),
 109                 'nopart': self.params.get('nopart', False),
 110                 'test': self.params.get('test', False),
 111             }
 112         )
 113         tmpfilename = self.temp_name(ctx['filename'])
 114         open_mode = 'wb'
 115         resume_len = 0
 116
 117         # Establish possible resume length
 118         if os.path.isfile(encodeFilename(tmpfilename)):
 119             open_mode = 'ab'
 120             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 121
 122         # Should be initialized before ytdl file check
 123         ctx.update({
 124             'tmpfilename': tmpfilename,
 125             'fragment_index': 0,
 126         })
 127
 128         if self.__do_ytdl_file(ctx):
 129             if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
 130                 self._read_ytdl_file(ctx)
 131             else:
 132                 self._write_ytdl_file(ctx)
 133             if ctx['fragment_index'] > 0:
 134                 assert resume_len > 0
 135             else:
 136                 assert resume_len == 0
 137
 138         dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
 139
 140         ctx.update({
 141             'dl': dl,
 142             'dest_stream': dest_stream,
 143             'tmpfilename': tmpfilename,
 144             # Total complete fragments downloaded so far in bytes
 145             'complete_frags_downloaded_bytes': resume_len,
 146         })
 147
 148     def _start_frag_download(self, ctx):
 149         total_frags = ctx['total_frags']
 150         # This dict stores the download progress, it's updated by the progress
 151         # hook
 152         state = {
 153             'status': 'downloading',
 154             'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
 155             'fragment_index': ctx['fragment_index'],
 156             'fragment_count': total_frags,
 157             'filename': ctx['filename'],
 158             'tmpfilename': ctx['tmpfilename'],
 159         }
 160
 161         start = time.time()
 162         ctx.update({
 163             'started': start,
 164             # Amount of fragment's bytes downloaded by the time of the previous
 165             # frag progress hook invocation
 166             'prev_frag_downloaded_bytes': 0,
 167         })
 168
 169         def frag_progress_hook(s):
 170             if s['status'] not in ('downloading', 'finished'):
 171                 return
 172
 173             time_now = time.time()
 174             state['elapsed'] = time_now - start
 175             frag_total_bytes = s.get('total_bytes') or 0
 176             if not ctx['live']:
 177                 estimated_size = (
 178                     (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
 179                     (state['fragment_index'] + 1) * total_frags)
 180                 state['total_bytes_estimate'] = estimated_size
 181
 182             if s['status'] == 'finished':
 183                 state['fragment_index'] += 1
 184                 ctx['fragment_index'] = state['fragment_index']
 185                 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
 186                 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
 187                 ctx['prev_frag_downloaded_bytes'] = 0
 188             else:
 189                 frag_downloaded_bytes = s['downloaded_bytes']
 190                 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
 191                 if not ctx['live']:
 192                     state['eta'] = self.calc_eta(
 193                         start, time_now, estimated_size,
 194                         state['downloaded_bytes'])
 195                 state['speed'] = s.get('speed') or ctx.get('speed')
 196                 ctx['speed'] = state['speed']
 197                 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
 198             self._hook_progress(state)
 199
 200         ctx['dl'].add_progress_hook(frag_progress_hook)
 201
 202         return start
 203
 204     def _finish_frag_download(self, ctx):
 205         ctx['dest_stream'].close()
 206         if self.__do_ytdl_file(ctx):
 207             ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
 208             if os.path.isfile(ytdl_filename):
 209                 os.remove(ytdl_filename)
 210         elapsed = time.time() - ctx['started']
 211         self.try_rename(ctx['tmpfilename'], ctx['filename'])
 212         fsize = os.path.getsize(encodeFilename(ctx['filename']))
 213
 214         self._hook_progress({
 215             'downloaded_bytes': fsize,
 216             'total_bytes': fsize,
 217             'filename': ctx['filename'],
 218             'status': 'finished',
 219             'elapsed': elapsed,
 220         })