_ 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     """
  33
  34     def report_retry_fragment(self, err, frag_index, count, retries):
  35         self.to_screen(
  36             '[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s)...'
  37             % (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
  38
  39     def report_skip_fragment(self, frag_index):
  40         self.to_screen('[download] Skipping fragment %d...' % frag_index)
  41
  42     def _prepare_url(self, info_dict, url):
  43         headers = info_dict.get('http_headers')
  44         return sanitized_Request(url, None, headers) if headers else url
  45
  46     def _prepare_and_start_frag_download(self, ctx):
  47         self._prepare_frag_download(ctx)
  48         self._start_frag_download(ctx)
  49
  50     def _read_ytdl_file(self, ctx):
  51         stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
  52         ctx['fragment_index'] = json.loads(stream.read())['download']['current_fragment_index']
  53         stream.close()
  54
  55     def _write_ytdl_file(self, ctx):
  56         frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
  57         frag_index_stream.write(json.dumps({
  58             'download': {
  59                 'current_fragment_index': ctx['fragment_index']
  60             },
  61         }))
  62         frag_index_stream.close()
  63
  64     def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
  65         fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
  66         success = ctx['dl'].download(fragment_filename, {
  67             'url': frag_url,
  68             'http_headers': headers or info_dict.get('http_headers'),
  69         })
  70         if not success:
  71             return False, None
  72         down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
  73         ctx['fragment_filename_sanitized'] = frag_sanitized
  74         frag_content = down.read()
  75         down.close()
  76         return True, frag_content
  77
  78     def _append_fragment(self, ctx, frag_content):
  79         try:
  80             ctx['dest_stream'].write(frag_content)
  81         finally:
  82             if not (ctx.get('live') or ctx['tmpfilename'] == '-'):
  83                 self._write_ytdl_file(ctx)
  84             os.remove(ctx['fragment_filename_sanitized'])
  85             del ctx['fragment_filename_sanitized']
  86
  87     def _prepare_frag_download(self, ctx):
  88         if 'live' not in ctx:
  89             ctx['live'] = False
  90         self.to_screen(
  91             '[%s] Total fragments: %s'
  92             % (self.FD_NAME, ctx['total_frags'] if not ctx['live'] else 'unknown (live)'))
  93         self.report_destination(ctx['filename'])
  94         dl = HttpQuietDownloader(
  95             self.ydl,
  96             {
  97                 'continuedl': True,
  98                 'quiet': True,
  99                 'noprogress': True,
 100                 'ratelimit': self.params.get('ratelimit'),
 101                 'retries': self.params.get('retries', 0),
 102                 'nopart': self.params.get('nopart', False),
 103                 'test': self.params.get('test', False),
 104             }
 105         )
 106         tmpfilename = self.temp_name(ctx['filename'])
 107         open_mode = 'wb'
 108         resume_len = 0
 109
 110         # Establish possible resume length
 111         if os.path.isfile(encodeFilename(tmpfilename)):
 112             open_mode = 'ab'
 113             resume_len = os.path.getsize(encodeFilename(tmpfilename))
 114
 115         ctx['fragment_index'] = 0
 116         if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
 117             self._read_ytdl_file(ctx)
 118         else:
 119             self._write_ytdl_file(ctx)
 120
 121         if ctx['fragment_index'] > 0:
 122             assert resume_len > 0
 123         else:
 124             assert resume_len == 0
 125
 126         dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
 127
 128         ctx.update({
 129             'dl': dl,
 130             'dest_stream': dest_stream,
 131             'tmpfilename': tmpfilename,
 132             # Total complete fragments downloaded so far in bytes
 133             'complete_frags_downloaded_bytes': resume_len,
 134         })
 135
 136     def _start_frag_download(self, ctx):
 137         total_frags = ctx['total_frags']
 138         # This dict stores the download progress, it's updated by the progress
 139         # hook
 140         state = {
 141             'status': 'downloading',
 142             'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
 143             'fragment_index': ctx['fragment_index'],
 144             'fragment_count': total_frags,
 145             'filename': ctx['filename'],
 146             'tmpfilename': ctx['tmpfilename'],
 147         }
 148
 149         start = time.time()
 150         ctx.update({
 151             'started': start,
 152             # Amount of fragment's bytes downloaded by the time of the previous
 153             # frag progress hook invocation
 154             'prev_frag_downloaded_bytes': 0,
 155         })
 156
 157         def frag_progress_hook(s):
 158             if s['status'] not in ('downloading', 'finished'):
 159                 return
 160
 161             time_now = time.time()
 162             state['elapsed'] = time_now - start
 163             frag_total_bytes = s.get('total_bytes') or 0
 164             if not ctx['live']:
 165                 estimated_size = (
 166                     (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
 167                     (state['fragment_index'] + 1) * total_frags)
 168                 state['total_bytes_estimate'] = estimated_size
 169
 170             if s['status'] == 'finished':
 171                 state['fragment_index'] += 1
 172                 ctx['fragment_index'] = state['fragment_index']
 173                 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
 174                 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
 175                 ctx['prev_frag_downloaded_bytes'] = 0
 176             else:
 177                 frag_downloaded_bytes = s['downloaded_bytes']
 178                 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
 179                 if not ctx['live']:
 180                     state['eta'] = self.calc_eta(
 181                         start, time_now, estimated_size,
 182                         state['downloaded_bytes'])
 183                 state['speed'] = s.get('speed') or ctx.get('speed')
 184                 ctx['speed'] = state['speed']
 185                 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
 186             self._hook_progress(state)
 187
 188         ctx['dl'].add_progress_hook(frag_progress_hook)
 189
 190         return start
 191
 192     def _finish_frag_download(self, ctx):
 193         ctx['dest_stream'].close()
 194         ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
 195         if os.path.isfile(ytdl_filename):
 196             os.remove(ytdl_filename)
 197         elapsed = time.time() - ctx['started']
 198         self.try_rename(ctx['tmpfilename'], ctx['filename'])
 199         fsize = os.path.getsize(encodeFilename(ctx['filename']))
 200
 201         self._hook_progress({
 202             'downloaded_bytes': fsize,
 203             'total_bytes': fsize,
 204             'filename': ctx['filename'],
 205             'status': 'finished',
 206             'elapsed': elapsed,
 207         })