]>
jfr.im git - yt-dlp.git/blob - yt_dlp/downloader/common.py
f5f6393a62b746b8810b90b9406b8ddbcf124a07
1 from __future__
import division
, unicode_literals
10 from ..compat
import compat_os_name
21 class FileDownloader(object):
22 """File Downloader class.
24 File downloader objects are the ones responsible of downloading the
25 actual video file and writing it to disk.
27 File downloaders accept a lot of parameters. In order not to saturate
28 the object constructor with arguments, it receives a dictionary of
33 verbose: Print additional info to stdout.
34 quiet: Do not print messages to stdout.
35 ratelimit: Download speed limit, in bytes/sec.
36 throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
37 retries: Number of times to retry for HTTP error 5xx
38 buffersize: Size of download buffer in bytes.
39 noresizebuffer: Do not automatically resize the download buffer.
40 continuedl: Try to continue downloads if possible.
41 noprogress: Do not print the progress bar.
42 logtostderr: Log messages to stderr instead of stdout.
43 consoletitle: Display progress in console window's titlebar.
44 nopart: Do not use temporary .part files.
45 updatetime: Use the Last-modified header to set output file timestamps.
46 test: Download only first bytes to test the downloader.
47 min_filesize: Skip files smaller than this size
48 max_filesize: Skip files larger than this size
49 xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
50 external_downloader_args: A dictionary of downloader keys (in lower case)
51 and a list of additional command-line arguments for the
52 executable. Use 'default' as the name for arguments to be
53 passed to all downloaders. For compatibility with youtube-dl,
54 a single list of args can also be used
55 hls_use_mpegts: Use the mpegts container for HLS videos.
56 http_chunk_size: Size of a chunk for chunk-based HTTP downloading. May be
57 useful for bypassing bandwidth throttling imposed by
58 a webserver (experimental)
60 Subclasses of this one must re-define the real_download method.
63 _TEST_FILE_SIZE
= 10241
66 def __init__(self
, ydl
, params
):
67 """Create a FileDownloader object with the given options."""
69 self
._progress
_hooks
= []
71 self
.add_progress_hook(self
.report_progress
)
74 def format_seconds(seconds
):
75 (mins
, secs
) = divmod(seconds
, 60)
76 (hours
, mins
) = divmod(mins
, 60)
80 return '%02d:%02d' % (mins
, secs
)
82 return '%02d:%02d:%02d' % (hours
, mins
, secs
)
85 def calc_percent(byte_counter
, data_len
):
88 return float(byte_counter
) / float(data_len
) * 100.0
91 def format_percent(percent
):
94 return '%6s' % ('%3.1f%%' % percent
)
97 def calc_eta(start
, now
, total
, current
):
103 if current
== 0 or dif
< 0.001: # One millisecond
105 rate
= float(current
) / dif
106 return int((float(total
) - float(current
)) / rate
)
112 return FileDownloader
.format_seconds(eta
)
115 def calc_speed(start
, now
, bytes):
117 if bytes == 0 or dif
< 0.001: # One millisecond
119 return float(bytes) / dif
122 def format_speed(speed
):
124 return '%10s' % '---b/s'
125 return '%10s' % ('%s/s' % format_bytes(speed
))
128 def format_retries(retries
):
129 return 'inf' if retries
== float('inf') else '%.0f' % retries
132 def best_block_size(elapsed_time
, bytes):
133 new_min
= max(bytes / 2.0, 1.0)
134 new_max
= min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
135 if elapsed_time
< 0.001:
137 rate
= bytes / elapsed_time
145 def parse_bytes(bytestr
):
146 """Parse a string indicating a byte quantity into an integer."""
147 matchobj
= re
.match(r
'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr
)
150 number
= float(matchobj
.group(1))
151 multiplier
= 1024.0 ** 'bkmgtpezy'.index(matchobj
.group(2).lower())
152 return int(round(number
* multiplier
))
154 def to_screen(self
, *args
, **kargs
):
155 self
.ydl
.to_stdout(*args
, quiet
=self
.params
.get('quiet'), **kargs
)
157 def to_stderr(self
, message
):
158 self
.ydl
.to_stderr(message
)
160 def to_console_title(self
, message
):
161 self
.ydl
.to_console_title(message
)
163 def trouble(self
, *args
, **kargs
):
164 self
.ydl
.trouble(*args
, **kargs
)
166 def report_warning(self
, *args
, **kargs
):
167 self
.ydl
.report_warning(*args
, **kargs
)
169 def report_error(self
, *args
, **kargs
):
170 self
.ydl
.report_error(*args
, **kargs
)
172 def write_debug(self
, *args
, **kargs
):
173 self
.ydl
.write_debug(*args
, **kargs
)
175 def slow_down(self
, start_time
, now
, byte_counter
):
176 """Sleep if the download speed is over the rate limit."""
177 rate_limit
= self
.params
.get('ratelimit')
178 if rate_limit
is None or byte_counter
== 0:
182 elapsed
= now
- start_time
185 speed
= float(byte_counter
) / elapsed
186 if speed
> rate_limit
:
187 sleep_time
= float(byte_counter
) / rate_limit
- elapsed
189 time
.sleep(sleep_time
)
191 def temp_name(self
, filename
):
192 """Returns a temporary filename for the given filename."""
193 if self
.params
.get('nopart', False) or filename
== '-' or \
194 (os
.path
.exists(encodeFilename(filename
)) and not os
.path
.isfile(encodeFilename(filename
))):
196 return filename
+ '.part'
198 def undo_temp_name(self
, filename
):
199 if filename
.endswith('.part'):
200 return filename
[:-len('.part')]
203 def ytdl_filename(self
, filename
):
204 return filename
+ '.ytdl'
206 def try_rename(self
, old_filename
, new_filename
):
207 if old_filename
== new_filename
:
210 if self
.params
.get('overwrites', False):
211 if os
.path
.isfile(encodeFilename(new_filename
)):
212 os
.remove(encodeFilename(new_filename
))
213 os
.rename(encodeFilename(old_filename
), encodeFilename(new_filename
))
214 except (IOError, OSError) as err
:
215 self
.report_error('unable to rename file: %s' % error_to_compat_str(err
))
217 def try_utime(self
, filename
, last_modified_hdr
):
218 """Try to set the last-modified time of the given file."""
219 if last_modified_hdr
is None:
221 if not os
.path
.isfile(encodeFilename(filename
)):
223 timestr
= last_modified_hdr
226 filetime
= timeconvert(timestr
)
229 # Ignore obviously invalid dates
233 os
.utime(filename
, (time
.time(), filetime
))
238 def report_destination(self
, filename
):
239 """Report destination filename."""
240 self
.to_screen('[download] Destination: ' + filename
)
242 def _report_progress_status(self
, msg
, is_last_line
=False):
243 fullmsg
= '[download] ' + msg
244 if self
.params
.get('progress_with_newline', False):
245 self
.to_screen(fullmsg
)
247 if compat_os_name
== 'nt':
248 prev_len
= getattr(self
, '_report_progress_prev_line_length',
250 if prev_len
> len(fullmsg
):
251 fullmsg
+= ' ' * (prev_len
- len(fullmsg
))
252 self
._report
_progress
_prev
_line
_length
= len(fullmsg
)
255 clear_line
= ('\r\x1b[K' if sys
.stderr
.isatty() else '\r')
256 self
.to_screen(clear_line
+ fullmsg
, skip_eol
=not is_last_line
)
257 self
.to_console_title('yt-dlp ' + msg
)
259 def report_progress(self
, s
):
260 if s
['status'] == 'finished':
261 if self
.params
.get('noprogress', False):
262 self
.to_screen('[download] Download completed')
264 msg_template
= '100%%'
265 if s
.get('total_bytes') is not None:
266 s
['_total_bytes_str'] = format_bytes(s
['total_bytes'])
267 msg_template
+= ' of %(_total_bytes_str)s'
268 if s
.get('elapsed') is not None:
269 s
['_elapsed_str'] = self
.format_seconds(s
['elapsed'])
270 msg_template
+= ' in %(_elapsed_str)s'
271 self
._report
_progress
_status
(
272 msg_template
% s
, is_last_line
=True)
274 if self
.params
.get('noprogress'):
277 if s
['status'] != 'downloading':
280 if s
.get('eta') is not None:
281 s
['_eta_str'] = self
.format_eta(s
['eta'])
283 s
['_eta_str'] = 'Unknown ETA'
285 if s
.get('total_bytes') and s
.get('downloaded_bytes') is not None:
286 s
['_percent_str'] = self
.format_percent(100 * s
['downloaded_bytes'] / s
['total_bytes'])
287 elif s
.get('total_bytes_estimate') and s
.get('downloaded_bytes') is not None:
288 s
['_percent_str'] = self
.format_percent(100 * s
['downloaded_bytes'] / s
['total_bytes_estimate'])
290 if s
.get('downloaded_bytes') == 0:
291 s
['_percent_str'] = self
.format_percent(0)
293 s
['_percent_str'] = 'Unknown %'
295 if s
.get('speed') is not None:
296 s
['_speed_str'] = self
.format_speed(s
['speed'])
298 s
['_speed_str'] = 'Unknown speed'
300 if s
.get('total_bytes') is not None:
301 s
['_total_bytes_str'] = format_bytes(s
['total_bytes'])
302 msg_template
= '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
303 elif s
.get('total_bytes_estimate') is not None:
304 s
['_total_bytes_estimate_str'] = format_bytes(s
['total_bytes_estimate'])
305 msg_template
= '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
307 if s
.get('downloaded_bytes') is not None:
308 s
['_downloaded_bytes_str'] = format_bytes(s
['downloaded_bytes'])
310 s
['_elapsed_str'] = self
.format_seconds(s
['elapsed'])
311 msg_template
= '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
313 msg_template
= '%(_downloaded_bytes_str)s at %(_speed_str)s'
315 msg_template
= '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
317 self
._report
_progress
_status
(msg_template
% s
)
319 def report_resuming_byte(self
, resume_len
):
320 """Report attempt to resume at given byte."""
321 self
.to_screen('[download] Resuming download at byte %s' % resume_len
)
323 def report_retry(self
, err
, count
, retries
):
324 """Report retry in case of HTTP error 5xx"""
326 '[download] Got server HTTP error: %s. Retrying (attempt %d of %s) ...'
327 % (error_to_compat_str(err
), count
, self
.format_retries(retries
)))
329 def report_file_already_downloaded(self
, *args
, **kwargs
):
330 """Report file has already been fully downloaded."""
331 return self
.ydl
.report_file_already_downloaded(*args
, **kwargs
)
333 def report_unable_to_resume(self
):
334 """Report it was impossible to resume download."""
335 self
.to_screen('[download] Unable to resume')
338 def supports_manifest(manifest
):
339 """ Whether the downloader can download the fragments from the manifest.
340 Redefine in subclasses if needed. """
343 def download(self
, filename
, info_dict
, subtitle
=False):
344 """Download to a filename using the info from info_dict
345 Return True on success and False otherwise
348 nooverwrites_and_exists
= (
349 not self
.params
.get('overwrites', True)
350 and os
.path
.exists(encodeFilename(filename
))
353 if not hasattr(filename
, 'write'):
354 continuedl_and_exists
= (
355 self
.params
.get('continuedl', True)
356 and os
.path
.isfile(encodeFilename(filename
))
357 and not self
.params
.get('nopart', False)
360 # Check file already present
361 if filename
!= '-' and (nooverwrites_and_exists
or continuedl_and_exists
):
362 self
.report_file_already_downloaded(filename
)
363 self
._hook
_progress
({
364 'filename': filename
,
365 'status': 'finished',
366 'total_bytes': os
.path
.getsize(encodeFilename(filename
)),
370 if subtitle
is False:
371 min_sleep_interval
= self
.params
.get('sleep_interval')
372 if min_sleep_interval
:
373 max_sleep_interval
= self
.params
.get('max_sleep_interval', min_sleep_interval
)
374 sleep_interval
= random
.uniform(min_sleep_interval
, max_sleep_interval
)
376 '[download] Sleeping %s seconds ...' % (
377 int(sleep_interval
) if sleep_interval
.is_integer()
378 else '%.2f' % sleep_interval
))
379 time
.sleep(sleep_interval
)
381 sleep_interval_sub
= 0
382 if type(self
.params
.get('sleep_interval_subtitles')) is int:
383 sleep_interval_sub
= self
.params
.get('sleep_interval_subtitles')
384 if sleep_interval_sub
> 0:
386 '[download] Sleeping %s seconds ...' % (
388 time
.sleep(sleep_interval_sub
)
389 return self
.real_download(filename
, info_dict
), True
391 def real_download(self
, filename
, info_dict
):
392 """Real download process. Redefine in subclasses."""
393 raise NotImplementedError('This method must be implemented by subclasses')
395 def _hook_progress(self
, status
, info_dict
):
396 if not self
._progress
_hooks
:
398 info_dict
= dict(info_dict
)
399 for key
in ('__original_infodict', '__postprocessors'):
400 info_dict
.pop(key
, None)
401 # youtube-dl passes the same status object to all the hooks.
402 # Some third party scripts seems to be relying on this.
403 # So keep this behavior if possible
404 status
['info_dict'] = copy
.deepcopy(info_dict
)
405 for ph
in self
._progress
_hooks
:
408 def add_progress_hook(self
, ph
):
409 # See YoutubeDl.py (search for progress_hooks) for a description of
411 self
._progress
_hooks
.append(ph
)
413 def _debug_cmd(self
, args
, exe
=None):
414 if not self
.params
.get('verbose', False):
417 str_args
= [decodeArgument(a
) for a
in args
]
420 exe
= os
.path
.basename(str_args
[0])
422 self
.write_debug('%s command line: %s' % (exe
, shell_quote(str_args
)))