19 class FileDownloader(object):
20 """File Downloader class.
22 File downloader objects are the ones responsible of downloading the
23 actual video file and writing it to disk.
25 File downloaders accept a lot of parameters. In order not to saturate
26 the object constructor with arguments, it receives a dictionary of
31 verbose: Print additional info to stdout.
32 quiet: Do not print messages to stdout.
33 ratelimit: Download speed limit, in bytes/sec.
34 retries: Number of times to retry for HTTP error 5xx
35 buffersize: Size of download buffer in bytes.
36 noresizebuffer: Do not automatically resize the download buffer.
37 continuedl: Try to continue downloads if possible.
38 noprogress: Do not print the progress bar.
39 logtostderr: Log messages to stderr instead of stdout.
40 consoletitle: Display progress in console window's titlebar.
41 nopart: Do not use temporary .part files.
42 updatetime: Use the Last-modified header to set output file timestamps.
43 test: Download only first bytes to test the downloader.
44 min_filesize: Skip files smaller than this size
45 max_filesize: Skip files larger than this size
50 def __init__(self, ydl, params):
51 """Create a FileDownloader object with the given options."""
53 self._progress_hooks = []
57 def format_seconds(seconds):
58 (mins, secs) = divmod(seconds, 60)
59 (hours, mins) = divmod(mins, 60)
63 return '%02d:%02d' % (mins, secs)
65 return '%02d:%02d:%02d' % (hours, mins, secs)
68 def calc_percent(byte_counter, data_len):
71 return float(byte_counter) / float(data_len) * 100.0
74 def format_percent(percent):
77 return '%6s' % ('%3.1f%%' % percent)
80 def calc_eta(start, now, total, current):
84 if current == 0 or dif < 0.001: # One millisecond
86 rate = float(current) / dif
87 return int((float(total) - float(current)) / rate)
93 return FileDownloader.format_seconds(eta)
96 def calc_speed(start, now, bytes):
98 if bytes == 0 or dif < 0.001: # One millisecond
100 return float(bytes) / dif
103 def format_speed(speed):
105 return '%10s' % '---b/s'
106 return '%10s' % ('%s/s' % format_bytes(speed))
109 def best_block_size(elapsed_time, bytes):
110 new_min = max(bytes / 2.0, 1.0)
111 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
112 if elapsed_time < 0.001:
114 rate = bytes / elapsed_time
122 def parse_bytes(bytestr):
123 """Parse a string indicating a byte quantity into an integer."""
124 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
127 number = float(matchobj.group(1))
128 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
129 return int(round(number * multiplier))
131 def to_screen(self, *args, **kargs):
132 self.ydl.to_screen(*args, **kargs)
134 def to_stderr(self, message):
135 self.ydl.to_screen(message)
137 def to_console_title(self, message):
138 self.ydl.to_console_title(message)
140 def trouble(self, *args, **kargs):
141 self.ydl.trouble(*args, **kargs)
143 def report_warning(self, *args, **kargs):
144 self.ydl.report_warning(*args, **kargs)
146 def report_error(self, *args, **kargs):
147 self.ydl.report_error(*args, **kargs)
149 def slow_down(self, start_time, byte_counter):
150 """Sleep if the download speed is over the rate limit."""
151 rate_limit = self.params.get('ratelimit', None)
152 if rate_limit is None or byte_counter == 0:
155 elapsed = now - start_time
158 speed = float(byte_counter) / elapsed
159 if speed > rate_limit:
160 time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
162 def temp_name(self, filename):
163 """Returns a temporary filename for the given filename."""
164 if self.params.get('nopart', False) or filename == u'-' or \
165 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
167 return filename + u'.part'
169 def undo_temp_name(self, filename):
170 if filename.endswith(u'.part'):
171 return filename[:-len(u'.part')]
174 def try_rename(self, old_filename, new_filename):
176 if old_filename == new_filename:
178 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
179 except (IOError, OSError):
180 self.report_error(u'unable to rename file')
182 def try_utime(self, filename, last_modified_hdr):
183 """Try to set the last-modified time of the given file."""
184 if last_modified_hdr is None:
186 if not os.path.isfile(encodeFilename(filename)):
188 timestr = last_modified_hdr
191 filetime = timeconvert(timestr)
194 # Ignore obviously invalid dates
198 os.utime(filename, (time.time(), filetime))
203 def report_destination(self, filename):
204 """Report destination filename."""
205 self.to_screen(u'[download] Destination: ' + filename)
207 def report_progress(self, percent, data_len_str, speed, eta):
208 """Report download progress."""
209 if self.params.get('noprogress', False):
211 clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
213 eta_str = self.format_eta(eta)
215 eta_str = 'Unknown ETA'
216 if percent is not None:
217 percent_str = self.format_percent(percent)
219 percent_str = 'Unknown %'
220 speed_str = self.format_speed(speed)
221 if self.params.get('progress_with_newline', False):
222 self.to_screen(u'[download] %s of %s at %s ETA %s' %
223 (percent_str, data_len_str, speed_str, eta_str))
225 self.to_screen(u'\r%s[download] %s of %s at %s ETA %s' %
226 (clear_line, percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
227 self.to_console_title(u'youtube-dl - %s of %s at %s ETA %s' %
228 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
230 def report_resuming_byte(self, resume_len):
231 """Report attempt to resume at given byte."""
232 self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
234 def report_retry(self, count, retries):
235 """Report retry in case of HTTP error 5xx"""
236 self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
238 def report_file_already_downloaded(self, file_name):
239 """Report file has already been fully downloaded."""
241 self.to_screen(u'[download] %s has already been downloaded' % file_name)
242 except UnicodeEncodeError:
243 self.to_screen(u'[download] The file has already been downloaded')
245 def report_unable_to_resume(self):
246 """Report it was impossible to resume download."""
247 self.to_screen(u'[download] Unable to resume')
249 def report_finish(self, data_len_str, tot_time):
250 """Report download finished."""
251 if self.params.get('noprogress', False):
252 self.to_screen(u'[download] Download completed')
254 clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
255 self.to_screen(u'\r%s[download] 100%% of %s in %s' %
256 (clear_line, data_len_str, self.format_seconds(tot_time)))
258 def _download_with_rtmpdump(self, filename, url, player_url, page_url, play_path, tc_url, live):
259 def run_rtmpdump(args):
261 resume_percent = None
262 resume_downloaded_data_len = None
263 proc = subprocess.Popen(args, stderr=subprocess.PIPE)
264 cursor_in_new_line = True
265 proc_stderr_closed = False
266 while not proc_stderr_closed:
267 # read line from stderr
270 char = proc.stderr.read(1)
272 proc_stderr_closed = True
274 if char in [b'\r', b'\n']:
276 line += char.decode('ascii', 'replace')
278 # proc_stderr_closed is True
280 mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line)
282 downloaded_data_len = int(float(mobj.group(1))*1024)
283 percent = float(mobj.group(2))
284 if not resume_percent:
285 resume_percent = percent
286 resume_downloaded_data_len = downloaded_data_len
287 eta = self.calc_eta(start, time.time(), 100-resume_percent, percent-resume_percent)
288 speed = self.calc_speed(start, time.time(), downloaded_data_len-resume_downloaded_data_len)
291 data_len = int(downloaded_data_len * 100 / percent)
292 data_len_str = u'~'+self.format_bytes(data_len)
293 self.report_progress(percent, data_len_str, speed, eta)
294 cursor_in_new_line = False
295 self._hook_progress({
296 'downloaded_bytes': downloaded_data_len,
297 'total_bytes': data_len,
298 'tmpfilename': tmpfilename,
299 'filename': filename,
300 'status': 'downloading',
304 elif self.params.get('verbose', False):
305 if not cursor_in_new_line:
307 cursor_in_new_line = True
308 self.to_screen(u'[rtmpdump] '+line)
310 if not cursor_in_new_line:
312 return proc.returncode
314 self.report_destination(filename)
315 tmpfilename = self.temp_name(filename)
316 test = self.params.get('test', False)
318 # Check for rtmpdump first
320 subprocess.call(['rtmpdump', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
321 except (OSError, IOError):
322 self.report_error(u'RTMP download detected but "rtmpdump" could not be run')
325 # Download using rtmpdump. rtmpdump returns exit code 2 when
326 # the connection was interrumpted and resuming appears to be
327 # possible. This is part of rtmpdump's normal usage, AFAIK.
328 basic_args = ['rtmpdump', '--verbose', '-r', url, '-o', tmpfilename]
329 if player_url is not None:
330 basic_args += ['--swfVfy', player_url]
331 if page_url is not None:
332 basic_args += ['--pageUrl', page_url]
333 if play_path is not None:
334 basic_args += ['--playpath', play_path]
335 if tc_url is not None:
336 basic_args += ['--tcUrl', url]
338 basic_args += ['--stop', '1']
340 basic_args += ['--live']
341 args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)]
342 if self.params.get('verbose', False):
345 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
348 self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
350 retval = run_rtmpdump(args)
352 while (retval == 2 or retval == 1) and not test:
353 prevsize = os.path.getsize(encodeFilename(tmpfilename))
354 self.to_screen(u'[rtmpdump] %s bytes' % prevsize)
355 time.sleep(5.0) # This seems to be needed
356 retval = run_rtmpdump(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
357 cursize = os.path.getsize(encodeFilename(tmpfilename))
358 if prevsize == cursize and retval == 1:
360 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
361 if prevsize == cursize and retval == 2 and cursize > 1024:
362 self.to_screen(u'[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
365 if retval == 0 or (test and retval == 2):
366 fsize = os.path.getsize(encodeFilename(tmpfilename))
367 self.to_screen(u'[rtmpdump] %s bytes' % fsize)
368 self.try_rename(tmpfilename, filename)
369 self._hook_progress({
370 'downloaded_bytes': fsize,
371 'total_bytes': fsize,
372 'filename': filename,
373 'status': 'finished',
377 self.to_stderr(u"\n")
378 self.report_error(u'rtmpdump exited with code %d' % retval)
381 def _download_with_mplayer(self, filename, url):
382 self.report_destination(filename)
383 tmpfilename = self.temp_name(filename)
385 args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url]
386 # Check for mplayer first
388 subprocess.call(['mplayer', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
389 except (OSError, IOError):
390 self.report_error(u'MMS or RTSP download detected but "%s" could not be run' % args[0] )
393 # Download using mplayer.
394 retval = subprocess.call(args)
396 fsize = os.path.getsize(encodeFilename(tmpfilename))
397 self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
398 self.try_rename(tmpfilename, filename)
399 self._hook_progress({
400 'downloaded_bytes': fsize,
401 'total_bytes': fsize,
402 'filename': filename,
403 'status': 'finished',
407 self.to_stderr(u"\n")
408 self.report_error(u'mplayer exited with code %d' % retval)
411 def _download_m3u8_with_ffmpeg(self, filename, url):
412 self.report_destination(filename)
413 tmpfilename = self.temp_name(filename)
415 args = ['-y', '-i', url, '-f', 'mp4', '-c', 'copy',
416 '-bsf:a', 'aac_adtstoasc', tmpfilename]
418 for program in ['avconv', 'ffmpeg']:
420 subprocess.call([program, '-version'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
422 except (OSError, IOError):
425 self.report_error(u'm3u8 download detected but ffmpeg or avconv could not be found')
426 cmd = [program] + args
428 retval = subprocess.call(cmd)
430 fsize = os.path.getsize(encodeFilename(tmpfilename))
431 self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
432 self.try_rename(tmpfilename, filename)
433 self._hook_progress({
434 'downloaded_bytes': fsize,
435 'total_bytes': fsize,
436 'filename': filename,
437 'status': 'finished',
441 self.to_stderr(u"\n")
442 self.report_error(u'ffmpeg exited with code %d' % retval)
446 def _do_download(self, filename, info_dict):
447 url = info_dict['url']
449 # Check file already present
450 if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
451 self.report_file_already_downloaded(filename)
452 self._hook_progress({
453 'filename': filename,
454 'status': 'finished',
455 'total_bytes': os.path.getsize(encodeFilename(filename)),
459 # Attempt to download using rtmpdump
460 if url.startswith('rtmp'):
461 return self._download_with_rtmpdump(filename, url,
462 info_dict.get('player_url', None),
463 info_dict.get('page_url', None),
464 info_dict.get('play_path', None),
465 info_dict.get('tc_url', None),
466 info_dict.get('rtmp_live', False))
468 # Attempt to download using mplayer
469 if url.startswith('mms') or url.startswith('rtsp'):
470 return self._download_with_mplayer(filename, url)
472 # m3u8 manifest are downloaded with ffmpeg
473 if determine_ext(url) == u'm3u8':
474 return self._download_m3u8_with_ffmpeg(filename, url)
476 tmpfilename = self.temp_name(filename)
479 # Do not include the Accept-Encoding header
480 headers = {'Youtubedl-no-compression': 'True'}
481 if 'user_agent' in info_dict:
482 headers['Youtubedl-user-agent'] = info_dict['user_agent']
483 basic_request = compat_urllib_request.Request(url, None, headers)
484 request = compat_urllib_request.Request(url, None, headers)
486 if self.params.get('test', False):
487 request.add_header('Range','bytes=0-10240')
489 # Establish possible resume length
490 if os.path.isfile(encodeFilename(tmpfilename)):
491 resume_len = os.path.getsize(encodeFilename(tmpfilename))
497 if self.params.get('continuedl', False):
498 self.report_resuming_byte(resume_len)
499 request.add_header('Range','bytes=%d-' % resume_len)
505 retries = self.params.get('retries', 0)
506 while count <= retries:
507 # Establish connection
509 if count == 0 and 'urlhandle' in info_dict:
510 data = info_dict['urlhandle']
511 data = compat_urllib_request.urlopen(request)
513 except (compat_urllib_error.HTTPError, ) as err:
514 if (err.code < 500 or err.code >= 600) and err.code != 416:
515 # Unexpected HTTP error
517 elif err.code == 416:
518 # Unable to resume (requested range not satisfiable)
520 # Open the connection again without the range header
521 data = compat_urllib_request.urlopen(basic_request)
522 content_length = data.info()['Content-Length']
523 except (compat_urllib_error.HTTPError, ) as err:
524 if err.code < 500 or err.code >= 600:
527 # Examine the reported length
528 if (content_length is not None and
529 (resume_len - 100 < int(content_length) < resume_len + 100)):
530 # The file had already been fully downloaded.
531 # Explanation to the above condition: in issue #175 it was revealed that
532 # YouTube sometimes adds or removes a few bytes from the end of the file,
533 # changing the file size slightly and causing problems for some users. So
534 # I decided to implement a suggested change and consider the file
535 # completely downloaded if the file size differs less than 100 bytes from
536 # the one in the hard drive.
537 self.report_file_already_downloaded(filename)
538 self.try_rename(tmpfilename, filename)
539 self._hook_progress({
540 'filename': filename,
541 'status': 'finished',
545 # The length does not match, we start the download over
546 self.report_unable_to_resume()
552 self.report_retry(count, retries)
555 self.report_error(u'giving up after %s retries' % retries)
558 data_len = data.info().get('Content-length', None)
559 if data_len is not None:
560 data_len = int(data_len) + resume_len
561 min_data_len = self.params.get("min_filesize", None)
562 max_data_len = self.params.get("max_filesize", None)
563 if min_data_len is not None and data_len < min_data_len:
564 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
566 if max_data_len is not None and data_len > max_data_len:
567 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
570 data_len_str = format_bytes(data_len)
571 byte_counter = 0 + resume_len
572 block_size = self.params.get('buffersize', 1024)
577 data_block = data.read(block_size)
579 if len(data_block) == 0:
581 byte_counter += len(data_block)
583 # Open file just in time
586 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
587 assert stream is not None
588 filename = self.undo_temp_name(tmpfilename)
589 self.report_destination(filename)
590 except (OSError, IOError) as err:
591 self.report_error(u'unable to open for writing: %s' % str(err))
594 stream.write(data_block)
595 except (IOError, OSError) as err:
596 self.to_stderr(u"\n")
597 self.report_error(u'unable to write data: %s' % str(err))
599 if not self.params.get('noresizebuffer', False):
600 block_size = self.best_block_size(after - before, len(data_block))
603 speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
607 percent = self.calc_percent(byte_counter, data_len)
608 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
609 self.report_progress(percent, data_len_str, speed, eta)
611 self._hook_progress({
612 'downloaded_bytes': byte_counter,
613 'total_bytes': data_len,
614 'tmpfilename': tmpfilename,
615 'filename': filename,
616 'status': 'downloading',
622 self.slow_down(start, byte_counter - resume_len)
625 self.to_stderr(u"\n")
626 self.report_error(u'Did not get any data blocks')
629 self.report_finish(data_len_str, (time.time() - start))
630 if data_len is not None and byte_counter != data_len:
631 raise ContentTooShortError(byte_counter, int(data_len))
632 self.try_rename(tmpfilename, filename)
634 # Update file modification time
635 if self.params.get('updatetime', True):
636 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
638 self._hook_progress({
639 'downloaded_bytes': byte_counter,
640 'total_bytes': byte_counter,
641 'filename': filename,
642 'status': 'finished',
647 def _hook_progress(self, status):
648 for ph in self._progress_hooks:
651 def add_progress_hook(self, ph):
652 """ ph gets called on download progress, with a dictionary with the entries
653 * filename: The final filename
654 * status: One of "downloading" and "finished"
656 It can also have some of the following entries:
658 * downloaded_bytes: Bytes on disks
659 * total_bytes: Total bytes, None if unknown
660 * tmpfilename: The filename we're currently writing to
661 * eta: The estimated time in seconds, None if unknown
662 * speed: The download speed in bytes/second, None if unknown
664 Hooks are guaranteed to be called at least once (with status "finished")
665 if the download is successful.
667 self._progress_hooks.append(ph)