Raphaël G. Git Repositories - youtubedl/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     For each incomplete fragment download youtube-dl keeps on disk a special
  36     bookkeeping file with download state and metadata (in future such files will
  37     be used for any incomplete download handled by youtube-dl). This file is
  38     used to properly handle resuming, check download file consistency and detect
  39     potential errors. The file has a .ytdl extension and represents a standard
  40     JSON file of the following format:
  41
  42     extractor:
  43         Dictionary of extractor related data. TBD.
  44
  45     downloader:
  46         Dictionary of downloader related data. May contain following data:
  47             current_fragment:
  48                 Dictionary with current (being downloaded) fragment data:
  49                 index:  0-based index of current fragment among all fragments
  50             fragment_count:
  51                 Total count of fragments
  52
  53     This feature is experimental and file format may change in future.
  54     """
  55
  56     def report_retry_fragment(self, err, frag_index, count, retries):
  57         self.to_screen(
  58             '[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s)...'
  59             % (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
  60
  61     def report_skip_fragment(self, frag_index):
  62         self.to_screen('[download] Skipping fragment %d...' % frag_index)
  63
  64     def _prepare_url(self, info_dict, url):
  65         headers = info_dict.get('http_headers')
  66         return sanitized_Request(url, None, headers) if headers else url
  67
  68     def _prepare_and_start_frag_download(self, ctx):
  69         self._prepare_frag_download(ctx)
  70         self._start_frag_download(ctx)
  71
  72     @staticmethod
  73     def __do_ytdl_file(ctx):
  74         return not ctx['live'] and not ctx['tmpfilename'] == '-'
  75
  76     def _read_ytdl_file(self, ctx):
  77         assert 'ytdl_corrupt' not in ctx
  78         stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
  79         try:
  80             ctx['fragment_index'] = json.loads(stream.read())['downloader']['current_fragment']['index']
  81         except Exception:
  82             ctx['ytdl_corrupt'] = True
  83         finally:
  84             stream.close()
  85
  86     def _write_ytdl_file(self, ctx):
  87         frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
  88         downloader = {
  89             'current_fragment': {
  90                 'index': ctx['fragment_index'],
  91             },
  92         }
  93         if ctx.get('fragment_count') is not None:
  94             downloader['fragment_count'] = ctx['fragment_count']
  95         frag_index_stream.write(json.dumps({'downloader': downloader}))
  96         frag_index_stream.close()
  97
  98     def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
  99         fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
 100         success = ctx['dl'].download(fragment_filename, {
 101             'url': frag_url,
 102             'http_headers': headers or info_dict.get('http_headers'),
 103         })
 104         if not success:
 105             return False, None
 106         down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
 107         ctx['fragment_filename_sanitized'] = frag_sanitized
 108         frag_content = down.read()
 109         down.close()
 110         return True, frag_content
 111
 112     def _append_fragment(self, ctx, frag_content):
 113         try:
 114             ctx['dest_stream'].write(frag_content)
 115             ctx['dest_stream'].flush()
 116         finally:
 117             if self.__do_ytdl_file(ctx):
 118                 self._write_ytdl_file(ctx)
 119             if not self.params.get('keep_fragments', False):
 120                 os.remove(encodeFilename(ctx['fragment_filename_sanitized']))
 121             del ctx['fragment_filename_sanitized']
 122
 123     def _prepare_frag_download(self, ctx):
 124         if 'live' not in ctx:
 125             ctx['live'] = False
 126         if not ctx['live']:
 127             total_frags_str = '%d' % ctx['total_frags']
 128             ad_frags = ctx.get('ad_frags', 0)
 129             if ad_frags:
 130                 total_frags_str += ' (not including %d ad)' % ad_frags
 131         else:
 132             total_frags_str = 'unknown (live)'
 133         self.to_screen(
 134             '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
 135         self.report_destination(ctx['filename'])
 136         dl = HttpQuietDownloader(
 137             self.ydl,
 138             {
 139                 'continuedl': True,
 140                 'quiet': True,
 141                 'noprogress': True,
 142                 'ratelimit': self.params.get('ratelimit'),
 143                 'retries': self.params.get('retries', 0),
 144                 'nopart': self.params.get('nopart', False),
 145                 'test': self.params.get('test', False),
 146             }
 147         )
 148         tmpfilename = self.temp_name(ctx['filename'])
 149         open_mode = 'wb'
 150         resume_len = 0
 151
 152         # Establish possible resume length
 153         if os.path.isfile(encodeFilename(tmpfilename)):
 154             open_mode = 'ab'
 155             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 156
 157         # Should be initialized before ytdl file check
 158         ctx.update({
 159             'tmpfilename': tmpfilename,
 160             'fragment_index': 0,
 161         })
 162
 163         if self.__do_ytdl_file(ctx):
 164             if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
 165                 self._read_ytdl_file(ctx)
 166                 is_corrupt = ctx.get('ytdl_corrupt') is True
 167                 is_inconsistent = ctx['fragment_index'] > 0 and resume_len == 0
 168                 if is_corrupt or is_inconsistent:
 169                     message = (
 170                         '.ytdl file is corrupt' if is_corrupt else
 171                         'Inconsistent state of incomplete fragment download')
 172                     self.report_warning(
 173                         '%s. Restarting from the beginning...' % message)
 174                     ctx['fragment_index'] = resume_len = 0
 175                     if 'ytdl_corrupt' in ctx:
 176                         del ctx['ytdl_corrupt']
 177                     self._write_ytdl_file(ctx)
 178             else:
 179                 self._write_ytdl_file(ctx)
 180                 assert ctx['fragment_index'] == 0
 181
 182         dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
 183
 184         ctx.update({
 185             'dl': dl,
 186             'dest_stream': dest_stream,
 187             'tmpfilename': tmpfilename,
 188             # Total complete fragments downloaded so far in bytes
 189             'complete_frags_downloaded_bytes': resume_len,
 190         })
 191
 192     def _start_frag_download(self, ctx):
 193         total_frags = ctx['total_frags']
 194         # This dict stores the download progress, it's updated by the progress
 195         # hook
 196         state = {
 197             'status': 'downloading',
 198             'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
 199             'fragment_index': ctx['fragment_index'],
 200             'fragment_count': total_frags,
 201             'filename': ctx['filename'],
 202             'tmpfilename': ctx['tmpfilename'],
 203         }
 204
 205         start = time.time()
 206         ctx.update({
 207             'started': start,
 208             # Amount of fragment's bytes downloaded by the time of the previous
 209             # frag progress hook invocation
 210             'prev_frag_downloaded_bytes': 0,
 211         })
 212
 213         def frag_progress_hook(s):
 214             if s['status'] not in ('downloading', 'finished'):
 215                 return
 216
 217             time_now = time.time()
 218             state['elapsed'] = time_now - start
 219             frag_total_bytes = s.get('total_bytes') or 0
 220             if not ctx['live']:
 221                 estimated_size = (
 222                     (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
 223                     (state['fragment_index'] + 1) * total_frags)
 224                 state['total_bytes_estimate'] = estimated_size
 225
 226             if s['status'] == 'finished':
 227                 state['fragment_index'] += 1
 228                 ctx['fragment_index'] = state['fragment_index']
 229                 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
 230                 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
 231                 ctx['prev_frag_downloaded_bytes'] = 0
 232             else:
 233                 frag_downloaded_bytes = s['downloaded_bytes']
 234                 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
 235                 if not ctx['live']:
 236                     state['eta'] = self.calc_eta(
 237                         start, time_now, estimated_size,
 238                         state['downloaded_bytes'])
 239                 state['speed'] = s.get('speed') or ctx.get('speed')
 240                 ctx['speed'] = state['speed']
 241                 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
 242             self._hook_progress(state)
 243
 244         ctx['dl'].add_progress_hook(frag_progress_hook)
 245
 246         return start
 247
 248     def _finish_frag_download(self, ctx):
 249         ctx['dest_stream'].close()
 250         if self.__do_ytdl_file(ctx):
 251             ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
 252             if os.path.isfile(ytdl_filename):
 253                 os.remove(ytdl_filename)
 254         elapsed = time.time() - ctx['started']
 255
 256         if ctx['tmpfilename'] == '-':
 257             downloaded_bytes = ctx['complete_frags_downloaded_bytes']
 258         else:
 259             self.try_rename(ctx['tmpfilename'], ctx['filename'])
 260             downloaded_bytes = os.path.getsize(encodeFilename(ctx['filename']))
 261
 262         self._hook_progress({
 263             'downloaded_bytes': downloaded_bytes,
 264             'total_bytes': downloaded_bytes,
 265             'filename': ctx['filename'],
 266             'status': 'finished',
 267             'elapsed': elapsed,
 268         })