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