_ 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     def _read_ytdl_file(self, ctx):
  53         stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
  54         ctx['fragment_index'] = json.loads(stream.read())['download']['current_fragment_index']
  55         stream.close()
  56
  57     def _write_ytdl_file(self, ctx):
  58         frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
  59         frag_index_stream.write(json.dumps({
  60             'download': {
  61                 'current_fragment_index': ctx['fragment_index']
  62             },
  63         }))
  64         frag_index_stream.close()
  65
  66     def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
  67         fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
  68         success = ctx['dl'].download(fragment_filename, {
  69             'url': frag_url,
  70             'http_headers': headers or info_dict.get('http_headers'),
  71         })
  72         if not success:
  73             return False, None
  74         down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
  75         ctx['fragment_filename_sanitized'] = frag_sanitized
  76         frag_content = down.read()
  77         down.close()
  78         return True, frag_content
  79
  80     def _append_fragment(self, ctx, frag_content):
  81         try:
  82             ctx['dest_stream'].write(frag_content)
  83         finally:
  84             if not (ctx.get('live') or ctx['tmpfilename'] == '-'):
  85                 self._write_ytdl_file(ctx)
  86             if not self.params.get('keep_fragments', False):
  87                 os.remove(ctx['fragment_filename_sanitized'])
  88             del ctx['fragment_filename_sanitized']
  89
  90     def _prepare_frag_download(self, ctx):
  91         if 'live' not in ctx:
  92             ctx['live'] = False
  93         self.to_screen(
  94             '[%s] Total fragments: %s'
  95             % (self.FD_NAME, ctx['total_frags'] if not ctx['live'] else 'unknown (live)'))
  96         self.report_destination(ctx['filename'])
  97         dl = HttpQuietDownloader(
  98             self.ydl,
  99             {
 100                 'continuedl': True,
 101                 'quiet': True,
 102                 'noprogress': True,
 103                 'ratelimit': self.params.get('ratelimit'),
 104                 'retries': self.params.get('retries', 0),
 105                 'nopart': self.params.get('nopart', False),
 106                 'test': self.params.get('test', False),
 107             }
 108         )
 109         tmpfilename = self.temp_name(ctx['filename'])
 110         open_mode = 'wb'
 111         resume_len = 0
 112
 113         # Establish possible resume length
 114         if os.path.isfile(encodeFilename(tmpfilename)):
 115             open_mode = 'ab'
 116             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 117
 118         ctx['fragment_index'] = 0
 119         if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
 120             self._read_ytdl_file(ctx)
 121         else:
 122             self._write_ytdl_file(ctx)
 123
 124         if ctx['fragment_index'] > 0:
 125             assert resume_len > 0
 126         else:
 127             assert resume_len == 0
 128
 129         dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
 130
 131         ctx.update({
 132             'dl': dl,
 133             'dest_stream': dest_stream,
 134             'tmpfilename': tmpfilename,
 135             # Total complete fragments downloaded so far in bytes
 136             'complete_frags_downloaded_bytes': resume_len,
 137         })
 138
 139     def _start_frag_download(self, ctx):
 140         total_frags = ctx['total_frags']
 141         # This dict stores the download progress, it's updated by the progress
 142         # hook
 143         state = {
 144             'status': 'downloading',
 145             'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
 146             'fragment_index': ctx['fragment_index'],
 147             'fragment_count': total_frags,
 148             'filename': ctx['filename'],
 149             'tmpfilename': ctx['tmpfilename'],
 150         }
 151
 152         start = time.time()
 153         ctx.update({
 154             'started': start,
 155             # Amount of fragment's bytes downloaded by the time of the previous
 156             # frag progress hook invocation
 157             'prev_frag_downloaded_bytes': 0,
 158         })
 159
 160         def frag_progress_hook(s):
 161             if s['status'] not in ('downloading', 'finished'):
 162                 return
 163
 164             time_now = time.time()
 165             state['elapsed'] = time_now - start
 166             frag_total_bytes = s.get('total_bytes') or 0
 167             if not ctx['live']:
 168                 estimated_size = (
 169                     (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
 170                     (state['fragment_index'] + 1) * total_frags)
 171                 state['total_bytes_estimate'] = estimated_size
 172
 173             if s['status'] == 'finished':
 174                 state['fragment_index'] += 1
 175                 ctx['fragment_index'] = state['fragment_index']
 176                 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
 177                 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
 178                 ctx['prev_frag_downloaded_bytes'] = 0
 179             else:
 180                 frag_downloaded_bytes = s['downloaded_bytes']
 181                 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
 182                 if not ctx['live']:
 183                     state['eta'] = self.calc_eta(
 184                         start, time_now, estimated_size,
 185                         state['downloaded_bytes'])
 186                 state['speed'] = s.get('speed') or ctx.get('speed')
 187                 ctx['speed'] = state['speed']
 188                 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
 189             self._hook_progress(state)
 190
 191         ctx['dl'].add_progress_hook(frag_progress_hook)
 192
 193         return start
 194
 195     def _finish_frag_download(self, ctx):
 196         ctx['dest_stream'].close()
 197         ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
 198         if os.path.isfile(ytdl_filename):
 199             os.remove(ytdl_filename)
 200         elapsed = time.time() - ctx['started']
 201         self.try_rename(ctx['tmpfilename'], ctx['filename'])
 202         fsize = os.path.getsize(encodeFilename(ctx['filename']))
 203
 204         self._hook_progress({
 205             'downloaded_bytes': fsize,
 206             'total_bytes': fsize,
 207             'filename': ctx['filename'],
 208             'status': 'finished',
 209             'elapsed': elapsed,
 210         })