]>
Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/FileDownloader.py
13 compat_urllib_request
,
22 class FileDownloader(object):
23 """File Downloader class.
25 File downloader objects are the ones responsible of downloading the
26 actual video file and writing it to disk.
28 File downloaders accept a lot of parameters. In order not to saturate
29 the object constructor with arguments, it receives a dictionary of
34 verbose: Print additional info to stdout.
35 quiet: Do not print messages to stdout.
36 ratelimit: Download speed limit, in 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
53 def __init__(self
, ydl
, params
):
54 """Create a FileDownloader object with the given options."""
56 self
._progress
_hooks
= []
60 def format_bytes(bytes):
63 if type(bytes) is str:
68 exponent
= int(math
.log(bytes, 1024.0))
69 suffix
= ['B','KiB','MiB','GiB','TiB','PiB','EiB','ZiB','YiB'][exponent
]
70 converted
= float(bytes) / float(1024 ** exponent
)
71 return '%.2f%s' % (converted
, suffix
)
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
):
101 if current
== 0 or dif
< 0.001: # One millisecond
103 rate
= float(current
) / dif
104 return int((float(total
) - float(current
)) / rate
)
110 return FileDownloader
.format_seconds(eta
)
113 def calc_speed(start
, now
, bytes):
115 if bytes == 0 or dif
< 0.001: # One millisecond
117 return float(bytes) / dif
120 def format_speed(speed
):
122 return '%10s' % '---b/s'
123 return '%10s' % ('%s/s' % FileDownloader
.format_bytes(speed
))
126 def best_block_size(elapsed_time
, bytes):
127 new_min
= max(bytes / 2.0, 1.0)
128 new_max
= min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
129 if elapsed_time
< 0.001:
131 rate
= bytes / elapsed_time
139 def parse_bytes(bytestr
):
140 """Parse a string indicating a byte quantity into an integer."""
141 matchobj
= re
.match(r
'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr
)
144 number
= float(matchobj
.group(1))
145 multiplier
= 1024.0 ** 'bkmgtpezy'.index(matchobj
.group(2).lower())
146 return int(round(number
* multiplier
))
148 def to_screen(self
, *args
, **kargs
):
149 self
.ydl
.to_screen(*args
, **kargs
)
151 def to_stderr(self
, message
):
152 self
.ydl
.to_screen(message
)
154 def to_cons_title(self
, message
):
155 """Set console/terminal window title to message."""
156 if not self
.params
.get('consoletitle', False):
158 if os
.name
== 'nt' and ctypes
.windll
.kernel32
.GetConsoleWindow():
159 # c_wchar_p() might not be necessary if `message` is
160 # already of type unicode()
161 ctypes
.windll
.kernel32
.SetConsoleTitleW(ctypes
.c_wchar_p(message
))
162 elif 'TERM' in os
.environ
:
163 self
.to_screen('\033]0;%s\007' % message
, skip_eol
=True)
165 def trouble(self
, *args
, **kargs
):
166 self
.ydl
.trouble(*args
, **kargs
)
168 def report_warning(self
, *args
, **kargs
):
169 self
.ydl
.report_warning(*args
, **kargs
)
171 def report_error(self
, *args
, **kargs
):
172 self
.ydl
.report_error(*args
, **kargs
)
174 def slow_down(self
, start_time
, byte_counter
):
175 """Sleep if the download speed is over the rate limit."""
176 rate_limit
= self
.params
.get('ratelimit', None)
177 if rate_limit
is None or byte_counter
== 0:
180 elapsed
= now
- start_time
183 speed
= float(byte_counter
) / elapsed
184 if speed
> rate_limit
:
185 time
.sleep((byte_counter
- rate_limit
* (now
- start_time
)) / rate_limit
)
187 def temp_name(self
, filename
):
188 """Returns a temporary filename for the given filename."""
189 if self
.params
.get('nopart', False) or filename
== u
'-' or \
190 (os
.path
.exists(encodeFilename(filename
)) and not os
.path
.isfile(encodeFilename(filename
))):
192 return filename
+ u
'.part'
194 def undo_temp_name(self
, filename
):
195 if filename
.endswith(u
'.part'):
196 return filename
[:-len(u
'.part')]
199 def try_rename(self
, old_filename
, new_filename
):
201 if old_filename
== new_filename
:
203 os
.rename(encodeFilename(old_filename
), encodeFilename(new_filename
))
204 except (IOError, OSError):
205 self
.report_error(u
'unable to rename file')
207 def try_utime(self
, filename
, last_modified_hdr
):
208 """Try to set the last-modified time of the given file."""
209 if last_modified_hdr
is None:
211 if not os
.path
.isfile(encodeFilename(filename
)):
213 timestr
= last_modified_hdr
216 filetime
= timeconvert(timestr
)
219 # Ignore obviously invalid dates
223 os
.utime(filename
, (time
.time(), filetime
))
228 def report_destination(self
, filename
):
229 """Report destination filename."""
230 self
.to_screen(u
'[download] Destination: ' + filename
)
232 def report_progress(self
, percent
, data_len_str
, speed
, eta
):
233 """Report download progress."""
234 if self
.params
.get('noprogress', False):
236 clear_line
= (u
'\x1b[K' if sys
.stderr
.isatty() and os
.name
!= 'nt' else u
'')
238 eta_str
= self
.format_eta(eta
)
240 eta_str
= 'Unknown ETA'
241 if percent
is not None:
242 percent_str
= self
.format_percent(percent
)
244 percent_str
= 'Unknown %'
245 speed_str
= self
.format_speed(speed
)
246 if self
.params
.get('progress_with_newline', False):
247 self
.to_screen(u
'[download] %s of %s at %s ETA %s' %
248 (percent_str
, data_len_str
, speed_str
, eta_str
))
250 self
.to_screen(u
'\r%s[download] %s of %s at %s ETA %s' %
251 (clear_line
, percent_str
, data_len_str
, speed_str
, eta_str
), skip_eol
=True)
252 self
.to_cons_title(u
'youtube-dl - %s of %s at %s ETA %s' %
253 (percent_str
.strip(), data_len_str
.strip(), speed_str
.strip(), eta_str
.strip()))
255 def report_resuming_byte(self
, resume_len
):
256 """Report attempt to resume at given byte."""
257 self
.to_screen(u
'[download] Resuming download at byte %s' % resume_len
)
259 def report_retry(self
, count
, retries
):
260 """Report retry in case of HTTP error 5xx"""
261 self
.to_screen(u
'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count
, retries
))
263 def report_file_already_downloaded(self
, file_name
):
264 """Report file has already been fully downloaded."""
266 self
.to_screen(u
'[download] %s has already been downloaded' % file_name
)
267 except UnicodeEncodeError:
268 self
.to_screen(u
'[download] The file has already been downloaded')
270 def report_unable_to_resume(self
):
271 """Report it was impossible to resume download."""
272 self
.to_screen(u
'[download] Unable to resume')
274 def report_finish(self
, data_len_str
, tot_time
):
275 """Report download finished."""
276 if self
.params
.get('noprogress', False):
277 self
.to_screen(u
'[download] Download completed')
279 clear_line
= (u
'\x1b[K' if sys
.stderr
.isatty() and os
.name
!= 'nt' else u
'')
280 self
.to_screen(u
'\r%s[download] 100%% of %s in %s' %
281 (clear_line
, data_len_str
, self
.format_seconds(tot_time
)))
283 def _download_with_rtmpdump(self
, filename
, url
, player_url
, page_url
, play_path
, tc_url
, live
):
284 self
.report_destination(filename
)
285 tmpfilename
= self
.temp_name(filename
)
286 test
= self
.params
.get('test', False)
288 # Check for rtmpdump first
290 subprocess
.call(['rtmpdump', '-h'], stdout
=(open(os
.path
.devnull
, 'w')), stderr
=subprocess
.STDOUT
)
291 except (OSError, IOError):
292 self
.report_error(u
'RTMP download detected but "rtmpdump" could not be run')
294 verbosity_option
= '--verbose' if self
.params
.get('verbose', False) else '--quiet'
296 # Download using rtmpdump. rtmpdump returns exit code 2 when
297 # the connection was interrumpted and resuming appears to be
298 # possible. This is part of rtmpdump's normal usage, AFAIK.
299 basic_args
= ['rtmpdump', verbosity_option
, '-r', url
, '-o', tmpfilename
]
300 if player_url
is not None:
301 basic_args
+= ['--swfVfy', player_url
]
302 if page_url
is not None:
303 basic_args
+= ['--pageUrl', page_url
]
304 if play_path
is not None:
305 basic_args
+= ['--playpath', play_path
]
306 if tc_url
is not None:
307 basic_args
+= ['--tcUrl', url
]
309 basic_args
+= ['--stop', '1']
311 basic_args
+= ['--live']
312 args
= basic_args
+ [[], ['--resume', '--skip', '1']][self
.params
.get('continuedl', False)]
313 if self
.params
.get('verbose', False):
316 shell_quote
= lambda args
: ' '.join(map(pipes
.quote
, args
))
319 self
.to_screen(u
'[debug] rtmpdump command line: ' + shell_quote(args
))
320 retval
= subprocess
.call(args
)
321 while (retval
== 2 or retval
== 1) and not test
:
322 prevsize
= os
.path
.getsize(encodeFilename(tmpfilename
))
323 self
.to_screen(u
'\r[rtmpdump] %s bytes' % prevsize
, skip_eol
=True)
324 time
.sleep(5.0) # This seems to be needed
325 retval
= subprocess
.call(basic_args
+ ['-e'] + [[], ['-k', '1']][retval
== 1])
326 cursize
= os
.path
.getsize(encodeFilename(tmpfilename
))
327 if prevsize
== cursize
and retval
== 1:
329 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
330 if prevsize
== cursize
and retval
== 2 and cursize
> 1024:
331 self
.to_screen(u
'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
334 if retval
== 0 or (test
and retval
== 2):
335 fsize
= os
.path
.getsize(encodeFilename(tmpfilename
))
336 self
.to_screen(u
'\r[rtmpdump] %s bytes' % fsize
)
337 self
.try_rename(tmpfilename
, filename
)
338 self
._hook
_progress
({
339 'downloaded_bytes': fsize
,
340 'total_bytes': fsize
,
341 'filename': filename
,
342 'status': 'finished',
346 self
.to_stderr(u
"\n")
347 self
.report_error(u
'rtmpdump exited with code %d' % retval
)
350 def _download_with_mplayer(self
, filename
, url
):
351 self
.report_destination(filename
)
352 tmpfilename
= self
.temp_name(filename
)
354 args
= ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename
, url
]
355 # Check for mplayer first
357 subprocess
.call(['mplayer', '-h'], stdout
=(open(os
.path
.devnull
, 'w')), stderr
=subprocess
.STDOUT
)
358 except (OSError, IOError):
359 self
.report_error(u
'MMS or RTSP download detected but "%s" could not be run' % args
[0] )
362 # Download using mplayer.
363 retval
= subprocess
.call(args
)
365 fsize
= os
.path
.getsize(encodeFilename(tmpfilename
))
366 self
.to_screen(u
'\r[%s] %s bytes' % (args
[0], fsize
))
367 self
.try_rename(tmpfilename
, filename
)
368 self
._hook
_progress
({
369 'downloaded_bytes': fsize
,
370 'total_bytes': fsize
,
371 'filename': filename
,
372 'status': 'finished',
376 self
.to_stderr(u
"\n")
377 self
.report_error(u
'mplayer exited with code %d' % retval
)
380 def _download_m3u8_with_ffmpeg(self
, filename
, url
):
381 self
.report_destination(filename
)
382 tmpfilename
= self
.temp_name(filename
)
384 args
= ['-y', '-i', url
, '-f', 'mp4', '-c', 'copy',
385 '-bsf:a', 'aac_adtstoasc', tmpfilename
]
387 for program
in ['avconv', 'ffmpeg']:
389 subprocess
.call([program
, '-version'], stdout
=(open(os
.path
.devnull
, 'w')), stderr
=subprocess
.STDOUT
)
391 except (OSError, IOError):
394 self
.report_error(u
'm3u8 download detected but ffmpeg or avconv could not be found')
395 cmd
= [program
] + args
397 retval
= subprocess
.call(cmd
)
399 fsize
= os
.path
.getsize(encodeFilename(tmpfilename
))
400 self
.to_screen(u
'\r[%s] %s bytes' % (args
[0], fsize
))
401 self
.try_rename(tmpfilename
, filename
)
402 self
._hook
_progress
({
403 'downloaded_bytes': fsize
,
404 'total_bytes': fsize
,
405 'filename': filename
,
406 'status': 'finished',
410 self
.to_stderr(u
"\n")
411 self
.report_error(u
'ffmpeg exited with code %d' % retval
)
415 def _do_download(self
, filename
, info_dict
):
416 url
= info_dict
['url']
418 # Check file already present
419 if self
.params
.get('continuedl', False) and os
.path
.isfile(encodeFilename(filename
)) and not self
.params
.get('nopart', False):
420 self
.report_file_already_downloaded(filename
)
421 self
._hook
_progress
({
422 'filename': filename
,
423 'status': 'finished',
424 'total_bytes': os
.path
.getsize(encodeFilename(filename
)),
428 # Attempt to download using rtmpdump
429 if url
.startswith('rtmp'):
430 return self
._download
_with
_rtmpdump
(filename
, url
,
431 info_dict
.get('player_url', None),
432 info_dict
.get('page_url', None),
433 info_dict
.get('play_path', None),
434 info_dict
.get('tc_url', None),
435 info_dict
.get('rtmp_live', False))
437 # Attempt to download using mplayer
438 if url
.startswith('mms') or url
.startswith('rtsp'):
439 return self
._download
_with
_mplayer
(filename
, url
)
441 # m3u8 manifest are downloaded with ffmpeg
442 if determine_ext(url
) == u
'm3u8':
443 return self
._download
_m
3u8_with
_ffmpeg
(filename
, url
)
445 tmpfilename
= self
.temp_name(filename
)
448 # Do not include the Accept-Encoding header
449 headers
= {'Youtubedl-no-compression': 'True'}
450 if 'user_agent' in info_dict
:
451 headers
['Youtubedl-user-agent'] = info_dict
['user_agent']
452 basic_request
= compat_urllib_request
.Request(url
, None, headers
)
453 request
= compat_urllib_request
.Request(url
, None, headers
)
455 if self
.params
.get('test', False):
456 request
.add_header('Range','bytes=0-10240')
458 # Establish possible resume length
459 if os
.path
.isfile(encodeFilename(tmpfilename
)):
460 resume_len
= os
.path
.getsize(encodeFilename(tmpfilename
))
466 if self
.params
.get('continuedl', False):
467 self
.report_resuming_byte(resume_len
)
468 request
.add_header('Range','bytes=%d-' % resume_len
)
474 retries
= self
.params
.get('retries', 0)
475 while count
<= retries
:
476 # Establish connection
478 if count
== 0 and 'urlhandle' in info_dict
:
479 data
= info_dict
['urlhandle']
480 data
= compat_urllib_request
.urlopen(request
)
482 except (compat_urllib_error
.HTTPError
, ) as err
:
483 if (err
.code
< 500 or err
.code
>= 600) and err
.code
!= 416:
484 # Unexpected HTTP error
486 elif err
.code
== 416:
487 # Unable to resume (requested range not satisfiable)
489 # Open the connection again without the range header
490 data
= compat_urllib_request
.urlopen(basic_request
)
491 content_length
= data
.info()['Content-Length']
492 except (compat_urllib_error
.HTTPError
, ) as err
:
493 if err
.code
< 500 or err
.code
>= 600:
496 # Examine the reported length
497 if (content_length
is not None and
498 (resume_len
- 100 < int(content_length
) < resume_len
+ 100)):
499 # The file had already been fully downloaded.
500 # Explanation to the above condition: in issue #175 it was revealed that
501 # YouTube sometimes adds or removes a few bytes from the end of the file,
502 # changing the file size slightly and causing problems for some users. So
503 # I decided to implement a suggested change and consider the file
504 # completely downloaded if the file size differs less than 100 bytes from
505 # the one in the hard drive.
506 self
.report_file_already_downloaded(filename
)
507 self
.try_rename(tmpfilename
, filename
)
508 self
._hook
_progress
({
509 'filename': filename
,
510 'status': 'finished',
514 # The length does not match, we start the download over
515 self
.report_unable_to_resume()
521 self
.report_retry(count
, retries
)
524 self
.report_error(u
'giving up after %s retries' % retries
)
527 data_len
= data
.info().get('Content-length', None)
528 if data_len
is not None:
529 data_len
= int(data_len
) + resume_len
530 min_data_len
= self
.params
.get("min_filesize", None)
531 max_data_len
= self
.params
.get("max_filesize", None)
532 if min_data_len
is not None and data_len
< min_data_len
:
533 self
.to_screen(u
'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len
, min_data_len
))
535 if max_data_len
is not None and data_len
> max_data_len
:
536 self
.to_screen(u
'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len
, max_data_len
))
539 data_len_str
= self
.format_bytes(data_len
)
540 byte_counter
= 0 + resume_len
541 block_size
= self
.params
.get('buffersize', 1024)
546 data_block
= data
.read(block_size
)
548 if len(data_block
) == 0:
550 byte_counter
+= len(data_block
)
552 # Open file just in time
555 (stream
, tmpfilename
) = sanitize_open(tmpfilename
, open_mode
)
556 assert stream
is not None
557 filename
= self
.undo_temp_name(tmpfilename
)
558 self
.report_destination(filename
)
559 except (OSError, IOError) as err
:
560 self
.report_error(u
'unable to open for writing: %s' % str(err
))
563 stream
.write(data_block
)
564 except (IOError, OSError) as err
:
565 self
.to_stderr(u
"\n")
566 self
.report_error(u
'unable to write data: %s' % str(err
))
568 if not self
.params
.get('noresizebuffer', False):
569 block_size
= self
.best_block_size(after
- before
, len(data_block
))
572 speed
= self
.calc_speed(start
, time
.time(), byte_counter
- resume_len
)
576 percent
= self
.calc_percent(byte_counter
, data_len
)
577 eta
= self
.calc_eta(start
, time
.time(), data_len
- resume_len
, byte_counter
- resume_len
)
578 self
.report_progress(percent
, data_len_str
, speed
, eta
)
580 self
._hook
_progress
({
581 'downloaded_bytes': byte_counter
,
582 'total_bytes': data_len
,
583 'tmpfilename': tmpfilename
,
584 'filename': filename
,
585 'status': 'downloading',
591 self
.slow_down(start
, byte_counter
- resume_len
)
594 self
.to_stderr(u
"\n")
595 self
.report_error(u
'Did not get any data blocks')
598 self
.report_finish(data_len_str
, (time
.time() - start
))
599 if data_len
is not None and byte_counter
!= data_len
:
600 raise ContentTooShortError(byte_counter
, int(data_len
))
601 self
.try_rename(tmpfilename
, filename
)
603 # Update file modification time
604 if self
.params
.get('updatetime', True):
605 info_dict
['filetime'] = self
.try_utime(filename
, data
.info().get('last-modified', None))
607 self
._hook
_progress
({
608 'downloaded_bytes': byte_counter
,
609 'total_bytes': byte_counter
,
610 'filename': filename
,
611 'status': 'finished',
616 def _hook_progress(self
, status
):
617 for ph
in self
._progress
_hooks
:
620 def add_progress_hook(self
, ph
):
621 """ ph gets called on download progress, with a dictionary with the entries
622 * filename: The final filename
623 * status: One of "downloading" and "finished"
625 It can also have some of the following entries:
627 * downloaded_bytes: Bytes on disks
628 * total_bytes: Total bytes, None if unknown
629 * tmpfilename: The filename we're currently writing to
630 * eta: The estimated time in seconds, None if unknown
631 * speed: The download speed in bytes/second, None if unknown
633 Hooks are guaranteed to be called at least once (with status "finished")
634 if the download is successful.
636 self
._progress
_hooks
.append(ph
)