]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/FileDownloader.py
Prepare for upload.
[youtubedl] / youtube_dl / FileDownloader.py
1 import math
2 import os
3 import re
4 import subprocess
5 import sys
6 import time
7
8 if os.name == 'nt':
9 import ctypes
10
11 from .utils import (
12 compat_urllib_error,
13 compat_urllib_request,
14 ContentTooShortError,
15 determine_ext,
16 encodeFilename,
17 sanitize_open,
18 timeconvert,
19 )
20
21
22 class FileDownloader(object):
23 """File Downloader class.
24
25 File downloader objects are the ones responsible of downloading the
26 actual video file and writing it to disk.
27
28 File downloaders accept a lot of parameters. In order not to saturate
29 the object constructor with arguments, it receives a dictionary of
30 options instead.
31
32 Available options:
33
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
49 """
50
51 params = None
52
53 def __init__(self, ydl, params):
54 """Create a FileDownloader object with the given options."""
55 self.ydl = ydl
56 self._progress_hooks = []
57 self.params = params
58
59 @staticmethod
60 def format_bytes(bytes):
61 if bytes is None:
62 return 'N/A'
63 if type(bytes) is str:
64 bytes = float(bytes)
65 if bytes == 0.0:
66 exponent = 0
67 else:
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)
72
73 @staticmethod
74 def format_seconds(seconds):
75 (mins, secs) = divmod(seconds, 60)
76 (hours, mins) = divmod(mins, 60)
77 if hours > 99:
78 return '--:--:--'
79 if hours == 0:
80 return '%02d:%02d' % (mins, secs)
81 else:
82 return '%02d:%02d:%02d' % (hours, mins, secs)
83
84 @staticmethod
85 def calc_percent(byte_counter, data_len):
86 if data_len is None:
87 return None
88 return float(byte_counter) / float(data_len) * 100.0
89
90 @staticmethod
91 def format_percent(percent):
92 if percent is None:
93 return '---.-%'
94 return '%6s' % ('%3.1f%%' % percent)
95
96 @staticmethod
97 def calc_eta(start, now, total, current):
98 if total is None:
99 return None
100 dif = now - start
101 if current == 0 or dif < 0.001: # One millisecond
102 return None
103 rate = float(current) / dif
104 return int((float(total) - float(current)) / rate)
105
106 @staticmethod
107 def format_eta(eta):
108 if eta is None:
109 return '--:--'
110 return FileDownloader.format_seconds(eta)
111
112 @staticmethod
113 def calc_speed(start, now, bytes):
114 dif = now - start
115 if bytes == 0 or dif < 0.001: # One millisecond
116 return None
117 return float(bytes) / dif
118
119 @staticmethod
120 def format_speed(speed):
121 if speed is None:
122 return '%10s' % '---b/s'
123 return '%10s' % ('%s/s' % FileDownloader.format_bytes(speed))
124
125 @staticmethod
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:
130 return int(new_max)
131 rate = bytes / elapsed_time
132 if rate > new_max:
133 return int(new_max)
134 if rate < new_min:
135 return int(new_min)
136 return int(rate)
137
138 @staticmethod
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)
142 if matchobj is None:
143 return None
144 number = float(matchobj.group(1))
145 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
146 return int(round(number * multiplier))
147
148 def to_screen(self, *args, **kargs):
149 self.ydl.to_screen(*args, **kargs)
150
151 def to_stderr(self, message):
152 self.ydl.to_screen(message)
153
154 def to_cons_title(self, message):
155 """Set console/terminal window title to message."""
156 if not self.params.get('consoletitle', False):
157 return
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)
164
165 def trouble(self, *args, **kargs):
166 self.ydl.trouble(*args, **kargs)
167
168 def report_warning(self, *args, **kargs):
169 self.ydl.report_warning(*args, **kargs)
170
171 def report_error(self, *args, **kargs):
172 self.ydl.report_error(*args, **kargs)
173
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:
178 return
179 now = time.time()
180 elapsed = now - start_time
181 if elapsed <= 0.0:
182 return
183 speed = float(byte_counter) / elapsed
184 if speed > rate_limit:
185 time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
186
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))):
191 return filename
192 return filename + u'.part'
193
194 def undo_temp_name(self, filename):
195 if filename.endswith(u'.part'):
196 return filename[:-len(u'.part')]
197 return filename
198
199 def try_rename(self, old_filename, new_filename):
200 try:
201 if old_filename == new_filename:
202 return
203 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
204 except (IOError, OSError):
205 self.report_error(u'unable to rename file')
206
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:
210 return
211 if not os.path.isfile(encodeFilename(filename)):
212 return
213 timestr = last_modified_hdr
214 if timestr is None:
215 return
216 filetime = timeconvert(timestr)
217 if filetime is None:
218 return filetime
219 # Ignore obviously invalid dates
220 if filetime == 0:
221 return
222 try:
223 os.utime(filename, (time.time(), filetime))
224 except:
225 pass
226 return filetime
227
228 def report_destination(self, filename):
229 """Report destination filename."""
230 self.to_screen(u'[download] Destination: ' + filename)
231
232 def report_progress(self, percent, data_len_str, speed, eta):
233 """Report download progress."""
234 if self.params.get('noprogress', False):
235 return
236 clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
237 if eta is not None:
238 eta_str = self.format_eta(eta)
239 else:
240 eta_str = 'Unknown ETA'
241 if percent is not None:
242 percent_str = self.format_percent(percent)
243 else:
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))
249 else:
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()))
254
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)
258
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))
262
263 def report_file_already_downloaded(self, file_name):
264 """Report file has already been fully downloaded."""
265 try:
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')
269
270 def report_unable_to_resume(self):
271 """Report it was impossible to resume download."""
272 self.to_screen(u'[download] Unable to resume')
273
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')
278 else:
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)))
282
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)
287
288 # Check for rtmpdump first
289 try:
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')
293 return False
294 verbosity_option = '--verbose' if self.params.get('verbose', False) else '--quiet'
295
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]
308 if test:
309 basic_args += ['--stop', '1']
310 if live:
311 basic_args += ['--live']
312 args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)]
313 if self.params.get('verbose', False):
314 try:
315 import pipes
316 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
317 except ImportError:
318 shell_quote = repr
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:
328 break
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.')
332 retval = 0
333 break
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',
343 })
344 return True
345 else:
346 self.to_stderr(u"\n")
347 self.report_error(u'rtmpdump exited with code %d' % retval)
348 return False
349
350 def _download_with_mplayer(self, filename, url):
351 self.report_destination(filename)
352 tmpfilename = self.temp_name(filename)
353
354 args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url]
355 # Check for mplayer first
356 try:
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] )
360 return False
361
362 # Download using mplayer.
363 retval = subprocess.call(args)
364 if retval == 0:
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',
373 })
374 return True
375 else:
376 self.to_stderr(u"\n")
377 self.report_error(u'mplayer exited with code %d' % retval)
378 return False
379
380 def _download_m3u8_with_ffmpeg(self, filename, url):
381 self.report_destination(filename)
382 tmpfilename = self.temp_name(filename)
383
384 args = ['-y', '-i', url, '-f', 'mp4', '-c', 'copy',
385 '-bsf:a', 'aac_adtstoasc', tmpfilename]
386
387 for program in ['avconv', 'ffmpeg']:
388 try:
389 subprocess.call([program, '-version'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
390 break
391 except (OSError, IOError):
392 pass
393 else:
394 self.report_error(u'm3u8 download detected but ffmpeg or avconv could not be found')
395 cmd = [program] + args
396
397 retval = subprocess.call(cmd)
398 if retval == 0:
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',
407 })
408 return True
409 else:
410 self.to_stderr(u"\n")
411 self.report_error(u'ffmpeg exited with code %d' % retval)
412 return False
413
414
415 def _do_download(self, filename, info_dict):
416 url = info_dict['url']
417
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)),
425 })
426 return True
427
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))
436
437 # Attempt to download using mplayer
438 if url.startswith('mms') or url.startswith('rtsp'):
439 return self._download_with_mplayer(filename, url)
440
441 # m3u8 manifest are downloaded with ffmpeg
442 if determine_ext(url) == u'm3u8':
443 return self._download_m3u8_with_ffmpeg(filename, url)
444
445 tmpfilename = self.temp_name(filename)
446 stream = None
447
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)
454
455 if self.params.get('test', False):
456 request.add_header('Range','bytes=0-10240')
457
458 # Establish possible resume length
459 if os.path.isfile(encodeFilename(tmpfilename)):
460 resume_len = os.path.getsize(encodeFilename(tmpfilename))
461 else:
462 resume_len = 0
463
464 open_mode = 'wb'
465 if resume_len != 0:
466 if self.params.get('continuedl', False):
467 self.report_resuming_byte(resume_len)
468 request.add_header('Range','bytes=%d-' % resume_len)
469 open_mode = 'ab'
470 else:
471 resume_len = 0
472
473 count = 0
474 retries = self.params.get('retries', 0)
475 while count <= retries:
476 # Establish connection
477 try:
478 if count == 0 and 'urlhandle' in info_dict:
479 data = info_dict['urlhandle']
480 data = compat_urllib_request.urlopen(request)
481 break
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
485 raise
486 elif err.code == 416:
487 # Unable to resume (requested range not satisfiable)
488 try:
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:
494 raise
495 else:
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',
511 })
512 return True
513 else:
514 # The length does not match, we start the download over
515 self.report_unable_to_resume()
516 open_mode = 'wb'
517 break
518 # Retry
519 count += 1
520 if count <= retries:
521 self.report_retry(count, retries)
522
523 if count > retries:
524 self.report_error(u'giving up after %s retries' % retries)
525 return False
526
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))
534 return False
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))
537 return False
538
539 data_len_str = self.format_bytes(data_len)
540 byte_counter = 0 + resume_len
541 block_size = self.params.get('buffersize', 1024)
542 start = time.time()
543 while True:
544 # Download and write
545 before = time.time()
546 data_block = data.read(block_size)
547 after = time.time()
548 if len(data_block) == 0:
549 break
550 byte_counter += len(data_block)
551
552 # Open file just in time
553 if stream is None:
554 try:
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))
561 return False
562 try:
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))
567 return False
568 if not self.params.get('noresizebuffer', False):
569 block_size = self.best_block_size(after - before, len(data_block))
570
571 # Progress message
572 speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
573 if data_len is None:
574 eta = percent = None
575 else:
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)
579
580 self._hook_progress({
581 'downloaded_bytes': byte_counter,
582 'total_bytes': data_len,
583 'tmpfilename': tmpfilename,
584 'filename': filename,
585 'status': 'downloading',
586 'eta': eta,
587 'speed': speed,
588 })
589
590 # Apply rate limit
591 self.slow_down(start, byte_counter - resume_len)
592
593 if stream is None:
594 self.to_stderr(u"\n")
595 self.report_error(u'Did not get any data blocks')
596 return False
597 stream.close()
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)
602
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))
606
607 self._hook_progress({
608 'downloaded_bytes': byte_counter,
609 'total_bytes': byte_counter,
610 'filename': filename,
611 'status': 'finished',
612 })
613
614 return True
615
616 def _hook_progress(self, status):
617 for ph in self._progress_hooks:
618 ph(status)
619
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"
624
625 It can also have some of the following entries:
626
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
632
633 Hooks are guaranteed to be called at least once (with status "finished")
634 if the download is successful.
635 """
636 self._progress_hooks.append(ph)