]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/downloader/fragment.py
ea5e3a4b5df9f957328557b6f822ef5494cfc9a6
[youtubedl] / 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 stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
78 ctx['fragment_index'] = json.loads(stream.read())['downloader']['current_fragment']['index']
79 stream.close()
80
81 def _write_ytdl_file(self, ctx):
82 frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
83 downloader = {
84 'current_fragment': {
85 'index': ctx['fragment_index'],
86 },
87 }
88 if ctx.get('fragment_count') is not None:
89 downloader['fragment_count'] = ctx['fragment_count']
90 frag_index_stream.write(json.dumps({'downloader': downloader}))
91 frag_index_stream.close()
92
93 def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
94 fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
95 success = ctx['dl'].download(fragment_filename, {
96 'url': frag_url,
97 'http_headers': headers or info_dict.get('http_headers'),
98 })
99 if not success:
100 return False, None
101 down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
102 ctx['fragment_filename_sanitized'] = frag_sanitized
103 frag_content = down.read()
104 down.close()
105 return True, frag_content
106
107 def _append_fragment(self, ctx, frag_content):
108 try:
109 ctx['dest_stream'].write(frag_content)
110 ctx['dest_stream'].flush()
111 finally:
112 if self.__do_ytdl_file(ctx):
113 self._write_ytdl_file(ctx)
114 if not self.params.get('keep_fragments', False):
115 os.remove(encodeFilename(ctx['fragment_filename_sanitized']))
116 del ctx['fragment_filename_sanitized']
117
118 def _prepare_frag_download(self, ctx):
119 if 'live' not in ctx:
120 ctx['live'] = False
121 if not ctx['live']:
122 total_frags_str = '%d' % ctx['total_frags']
123 ad_frags = ctx.get('ad_frags', 0)
124 if ad_frags:
125 total_frags_str += ' (not including %d ad)' % ad_frags
126 else:
127 total_frags_str = 'unknown (live)'
128 self.to_screen(
129 '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
130 self.report_destination(ctx['filename'])
131 dl = HttpQuietDownloader(
132 self.ydl,
133 {
134 'continuedl': True,
135 'quiet': True,
136 'noprogress': True,
137 'ratelimit': self.params.get('ratelimit'),
138 'retries': self.params.get('retries', 0),
139 'nopart': self.params.get('nopart', False),
140 'test': self.params.get('test', False),
141 }
142 )
143 tmpfilename = self.temp_name(ctx['filename'])
144 open_mode = 'wb'
145 resume_len = 0
146
147 # Establish possible resume length
148 if os.path.isfile(encodeFilename(tmpfilename)):
149 open_mode = 'ab'
150 resume_len = os.path.getsize(encodeFilename(tmpfilename))
151
152 # Should be initialized before ytdl file check
153 ctx.update({
154 'tmpfilename': tmpfilename,
155 'fragment_index': 0,
156 })
157
158 if self.__do_ytdl_file(ctx):
159 if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
160 self._read_ytdl_file(ctx)
161 if ctx['fragment_index'] > 0 and resume_len == 0:
162 self.report_warning(
163 'Inconsistent state of incomplete fragment download. '
164 'Restarting from the beginning...')
165 ctx['fragment_index'] = resume_len = 0
166 self._write_ytdl_file(ctx)
167 else:
168 self._write_ytdl_file(ctx)
169 assert ctx['fragment_index'] == 0
170
171 dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
172
173 ctx.update({
174 'dl': dl,
175 'dest_stream': dest_stream,
176 'tmpfilename': tmpfilename,
177 # Total complete fragments downloaded so far in bytes
178 'complete_frags_downloaded_bytes': resume_len,
179 })
180
181 def _start_frag_download(self, ctx):
182 total_frags = ctx['total_frags']
183 # This dict stores the download progress, it's updated by the progress
184 # hook
185 state = {
186 'status': 'downloading',
187 'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
188 'fragment_index': ctx['fragment_index'],
189 'fragment_count': total_frags,
190 'filename': ctx['filename'],
191 'tmpfilename': ctx['tmpfilename'],
192 }
193
194 start = time.time()
195 ctx.update({
196 'started': start,
197 # Amount of fragment's bytes downloaded by the time of the previous
198 # frag progress hook invocation
199 'prev_frag_downloaded_bytes': 0,
200 })
201
202 def frag_progress_hook(s):
203 if s['status'] not in ('downloading', 'finished'):
204 return
205
206 time_now = time.time()
207 state['elapsed'] = time_now - start
208 frag_total_bytes = s.get('total_bytes') or 0
209 if not ctx['live']:
210 estimated_size = (
211 (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
212 (state['fragment_index'] + 1) * total_frags)
213 state['total_bytes_estimate'] = estimated_size
214
215 if s['status'] == 'finished':
216 state['fragment_index'] += 1
217 ctx['fragment_index'] = state['fragment_index']
218 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
219 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
220 ctx['prev_frag_downloaded_bytes'] = 0
221 else:
222 frag_downloaded_bytes = s['downloaded_bytes']
223 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
224 if not ctx['live']:
225 state['eta'] = self.calc_eta(
226 start, time_now, estimated_size,
227 state['downloaded_bytes'])
228 state['speed'] = s.get('speed') or ctx.get('speed')
229 ctx['speed'] = state['speed']
230 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
231 self._hook_progress(state)
232
233 ctx['dl'].add_progress_hook(frag_progress_hook)
234
235 return start
236
237 def _finish_frag_download(self, ctx):
238 ctx['dest_stream'].close()
239 if self.__do_ytdl_file(ctx):
240 ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
241 if os.path.isfile(ytdl_filename):
242 os.remove(ytdl_filename)
243 elapsed = time.time() - ctx['started']
244 self.try_rename(ctx['tmpfilename'], ctx['filename'])
245 fsize = os.path.getsize(encodeFilename(ctx['filename']))
246
247 self._hook_progress({
248 'downloaded_bytes': fsize,
249 'total_bytes': fsize,
250 'filename': ctx['filename'],
251 'status': 'finished',
252 'elapsed': elapsed,
253 })