]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/FileDownloader.py
Merge tag 'upstream/2013.10.01'
[youtubedl] / youtube_dl / FileDownloader.py
1 import math
2 import os
3 import re
4 import subprocess
5 import sys
6 import time
7 import traceback
8
9 if os.name == 'nt':
10 import ctypes
11
12 from .utils import *
13
14
15 class FileDownloader(object):
16 """File Downloader class.
17
18 File downloader objects are the ones responsible of downloading the
19 actual video file and writing it to disk.
20
21 File downloaders accept a lot of parameters. In order not to saturate
22 the object constructor with arguments, it receives a dictionary of
23 options instead.
24
25 Available options:
26
27 verbose: Print additional info to stdout.
28 quiet: Do not print messages to stdout.
29 ratelimit: Download speed limit, in bytes/sec.
30 retries: Number of times to retry for HTTP error 5xx
31 buffersize: Size of download buffer in bytes.
32 noresizebuffer: Do not automatically resize the download buffer.
33 continuedl: Try to continue downloads if possible.
34 noprogress: Do not print the progress bar.
35 logtostderr: Log messages to stderr instead of stdout.
36 consoletitle: Display progress in console window's titlebar.
37 nopart: Do not use temporary .part files.
38 updatetime: Use the Last-modified header to set output file timestamps.
39 test: Download only first bytes to test the downloader.
40 min_filesize: Skip files smaller than this size
41 max_filesize: Skip files larger than this size
42 """
43
44 params = None
45
46 def __init__(self, ydl, params):
47 """Create a FileDownloader object with the given options."""
48 self.ydl = ydl
49 self._progress_hooks = []
50 self.params = params
51
52 @staticmethod
53 def format_bytes(bytes):
54 if bytes is None:
55 return 'N/A'
56 if type(bytes) is str:
57 bytes = float(bytes)
58 if bytes == 0.0:
59 exponent = 0
60 else:
61 exponent = int(math.log(bytes, 1024.0))
62 suffix = ['B','KiB','MiB','GiB','TiB','PiB','EiB','ZiB','YiB'][exponent]
63 converted = float(bytes) / float(1024 ** exponent)
64 return '%.2f%s' % (converted, suffix)
65
66 @staticmethod
67 def format_seconds(seconds):
68 (mins, secs) = divmod(seconds, 60)
69 (hours, mins) = divmod(mins, 60)
70 if hours > 99:
71 return '--:--:--'
72 if hours == 0:
73 return '%02d:%02d' % (mins, secs)
74 else:
75 return '%02d:%02d:%02d' % (hours, mins, secs)
76
77 @staticmethod
78 def calc_percent(byte_counter, data_len):
79 if data_len is None:
80 return None
81 return float(byte_counter) / float(data_len) * 100.0
82
83 @staticmethod
84 def format_percent(percent):
85 if percent is None:
86 return '---.-%'
87 return '%6s' % ('%3.1f%%' % percent)
88
89 @staticmethod
90 def calc_eta(start, now, total, current):
91 if total is None:
92 return None
93 dif = now - start
94 if current == 0 or dif < 0.001: # One millisecond
95 return None
96 rate = float(current) / dif
97 return int((float(total) - float(current)) / rate)
98
99 @staticmethod
100 def format_eta(eta):
101 if eta is None:
102 return '--:--'
103 return FileDownloader.format_seconds(eta)
104
105 @staticmethod
106 def calc_speed(start, now, bytes):
107 dif = now - start
108 if bytes == 0 or dif < 0.001: # One millisecond
109 return None
110 return float(bytes) / dif
111
112 @staticmethod
113 def format_speed(speed):
114 if speed is None:
115 return '%10s' % '---b/s'
116 return '%10s' % ('%s/s' % FileDownloader.format_bytes(speed))
117
118 @staticmethod
119 def best_block_size(elapsed_time, bytes):
120 new_min = max(bytes / 2.0, 1.0)
121 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
122 if elapsed_time < 0.001:
123 return int(new_max)
124 rate = bytes / elapsed_time
125 if rate > new_max:
126 return int(new_max)
127 if rate < new_min:
128 return int(new_min)
129 return int(rate)
130
131 @staticmethod
132 def parse_bytes(bytestr):
133 """Parse a string indicating a byte quantity into an integer."""
134 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
135 if matchobj is None:
136 return None
137 number = float(matchobj.group(1))
138 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
139 return int(round(number * multiplier))
140
141 def to_screen(self, *args, **kargs):
142 self.ydl.to_screen(*args, **kargs)
143
144 def to_stderr(self, message):
145 self.ydl.to_screen(message)
146
147 def to_cons_title(self, message):
148 """Set console/terminal window title to message."""
149 if not self.params.get('consoletitle', False):
150 return
151 if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow():
152 # c_wchar_p() might not be necessary if `message` is
153 # already of type unicode()
154 ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
155 elif 'TERM' in os.environ:
156 self.to_screen('\033]0;%s\007' % message, skip_eol=True)
157
158 def trouble(self, *args, **kargs):
159 self.ydl.trouble(*args, **kargs)
160
161 def report_warning(self, *args, **kargs):
162 self.ydl.report_warning(*args, **kargs)
163
164 def report_error(self, *args, **kargs):
165 self.ydl.report_error(*args, **kargs)
166
167 def slow_down(self, start_time, byte_counter):
168 """Sleep if the download speed is over the rate limit."""
169 rate_limit = self.params.get('ratelimit', None)
170 if rate_limit is None or byte_counter == 0:
171 return
172 now = time.time()
173 elapsed = now - start_time
174 if elapsed <= 0.0:
175 return
176 speed = float(byte_counter) / elapsed
177 if speed > rate_limit:
178 time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
179
180 def temp_name(self, filename):
181 """Returns a temporary filename for the given filename."""
182 if self.params.get('nopart', False) or filename == u'-' or \
183 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
184 return filename
185 return filename + u'.part'
186
187 def undo_temp_name(self, filename):
188 if filename.endswith(u'.part'):
189 return filename[:-len(u'.part')]
190 return filename
191
192 def try_rename(self, old_filename, new_filename):
193 try:
194 if old_filename == new_filename:
195 return
196 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
197 except (IOError, OSError) as err:
198 self.report_error(u'unable to rename file')
199
200 def try_utime(self, filename, last_modified_hdr):
201 """Try to set the last-modified time of the given file."""
202 if last_modified_hdr is None:
203 return
204 if not os.path.isfile(encodeFilename(filename)):
205 return
206 timestr = last_modified_hdr
207 if timestr is None:
208 return
209 filetime = timeconvert(timestr)
210 if filetime is None:
211 return filetime
212 # Ignore obviously invalid dates
213 if filetime == 0:
214 return
215 try:
216 os.utime(filename, (time.time(), filetime))
217 except:
218 pass
219 return filetime
220
221 def report_destination(self, filename):
222 """Report destination filename."""
223 self.to_screen(u'[download] Destination: ' + filename)
224
225 def report_progress(self, percent, data_len_str, speed, eta):
226 """Report download progress."""
227 if self.params.get('noprogress', False):
228 return
229 clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
230 eta_str = self.format_eta(eta)
231 percent_str = self.format_percent(percent)
232 speed_str = self.format_speed(speed)
233 if self.params.get('progress_with_newline', False):
234 self.to_screen(u'[download] %s of %s at %s ETA %s' %
235 (percent_str, data_len_str, speed_str, eta_str))
236 else:
237 self.to_screen(u'\r%s[download] %s of %s at %s ETA %s' %
238 (clear_line, percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
239 self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' %
240 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
241
242 def report_resuming_byte(self, resume_len):
243 """Report attempt to resume at given byte."""
244 self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
245
246 def report_retry(self, count, retries):
247 """Report retry in case of HTTP error 5xx"""
248 self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
249
250 def report_file_already_downloaded(self, file_name):
251 """Report file has already been fully downloaded."""
252 try:
253 self.to_screen(u'[download] %s has already been downloaded' % file_name)
254 except (UnicodeEncodeError) as err:
255 self.to_screen(u'[download] The file has already been downloaded')
256
257 def report_unable_to_resume(self):
258 """Report it was impossible to resume download."""
259 self.to_screen(u'[download] Unable to resume')
260
261 def report_finish(self, data_len_str, tot_time):
262 """Report download finished."""
263 if self.params.get('noprogress', False):
264 self.to_screen(u'[download] Download completed')
265 else:
266 clear_line = (u'\x1b[K' if sys.stderr.isatty() and os.name != 'nt' else u'')
267 self.to_screen(u'\r%s[download] 100%% of %s in %s' %
268 (clear_line, data_len_str, self.format_seconds(tot_time)))
269
270 def _download_with_rtmpdump(self, filename, url, player_url, page_url, play_path, tc_url):
271 self.report_destination(filename)
272 tmpfilename = self.temp_name(filename)
273
274 # Check for rtmpdump first
275 try:
276 subprocess.call(['rtmpdump', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
277 except (OSError, IOError):
278 self.report_error(u'RTMP download detected but "rtmpdump" could not be run')
279 return False
280 verbosity_option = '--verbose' if self.params.get('verbose', False) else '--quiet'
281
282 # Download using rtmpdump. rtmpdump returns exit code 2 when
283 # the connection was interrumpted and resuming appears to be
284 # possible. This is part of rtmpdump's normal usage, AFAIK.
285 basic_args = ['rtmpdump', verbosity_option, '-r', url, '-o', tmpfilename]
286 if player_url is not None:
287 basic_args += ['--swfVfy', player_url]
288 if page_url is not None:
289 basic_args += ['--pageUrl', page_url]
290 if play_path is not None:
291 basic_args += ['--playpath', play_path]
292 if tc_url is not None:
293 basic_args += ['--tcUrl', url]
294 args = basic_args + [[], ['--resume', '--skip', '1']][self.params.get('continuedl', False)]
295 if self.params.get('verbose', False):
296 try:
297 import pipes
298 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
299 except ImportError:
300 shell_quote = repr
301 self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
302 retval = subprocess.call(args)
303 while retval == 2 or retval == 1:
304 prevsize = os.path.getsize(encodeFilename(tmpfilename))
305 self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
306 time.sleep(5.0) # This seems to be needed
307 retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
308 cursize = os.path.getsize(encodeFilename(tmpfilename))
309 if prevsize == cursize and retval == 1:
310 break
311 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
312 if prevsize == cursize and retval == 2 and cursize > 1024:
313 self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
314 retval = 0
315 break
316 if retval == 0:
317 fsize = os.path.getsize(encodeFilename(tmpfilename))
318 self.to_screen(u'\r[rtmpdump] %s bytes' % fsize)
319 self.try_rename(tmpfilename, filename)
320 self._hook_progress({
321 'downloaded_bytes': fsize,
322 'total_bytes': fsize,
323 'filename': filename,
324 'status': 'finished',
325 })
326 return True
327 else:
328 self.to_stderr(u"\n")
329 self.report_error(u'rtmpdump exited with code %d' % retval)
330 return False
331
332 def _download_with_mplayer(self, filename, url):
333 self.report_destination(filename)
334 tmpfilename = self.temp_name(filename)
335
336 args = ['mplayer', '-really-quiet', '-vo', 'null', '-vc', 'dummy', '-dumpstream', '-dumpfile', tmpfilename, url]
337 # Check for mplayer first
338 try:
339 subprocess.call(['mplayer', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
340 except (OSError, IOError):
341 self.report_error(u'MMS or RTSP download detected but "%s" could not be run' % args[0] )
342 return False
343
344 # Download using mplayer.
345 retval = subprocess.call(args)
346 if retval == 0:
347 fsize = os.path.getsize(encodeFilename(tmpfilename))
348 self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
349 self.try_rename(tmpfilename, filename)
350 self._hook_progress({
351 'downloaded_bytes': fsize,
352 'total_bytes': fsize,
353 'filename': filename,
354 'status': 'finished',
355 })
356 return True
357 else:
358 self.to_stderr(u"\n")
359 self.report_error(u'mplayer exited with code %d' % retval)
360 return False
361
362 def _download_m3u8_with_ffmpeg(self, filename, url):
363 self.report_destination(filename)
364 tmpfilename = self.temp_name(filename)
365
366 args = ['ffmpeg', '-y', '-i', url, '-f', 'mp4', tmpfilename]
367 # Check for ffmpeg first
368 try:
369 subprocess.call(['ffmpeg', '-h'], stdout=(open(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
370 except (OSError, IOError):
371 self.report_error(u'm3u8 download detected but "%s" could not be run' % args[0] )
372 return False
373
374 retval = subprocess.call(args)
375 if retval == 0:
376 fsize = os.path.getsize(encodeFilename(tmpfilename))
377 self.to_screen(u'\r[%s] %s bytes' % (args[0], fsize))
378 self.try_rename(tmpfilename, filename)
379 self._hook_progress({
380 'downloaded_bytes': fsize,
381 'total_bytes': fsize,
382 'filename': filename,
383 'status': 'finished',
384 })
385 return True
386 else:
387 self.to_stderr(u"\n")
388 self.report_error(u'ffmpeg exited with code %d' % retval)
389 return False
390
391
392 def _do_download(self, filename, info_dict):
393 url = info_dict['url']
394
395 # Check file already present
396 if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
397 self.report_file_already_downloaded(filename)
398 self._hook_progress({
399 'filename': filename,
400 'status': 'finished',
401 'total_bytes': os.path.getsize(encodeFilename(filename)),
402 })
403 return True
404
405 # Attempt to download using rtmpdump
406 if url.startswith('rtmp'):
407 return self._download_with_rtmpdump(filename, url,
408 info_dict.get('player_url', None),
409 info_dict.get('page_url', None),
410 info_dict.get('play_path', None),
411 info_dict.get('tc_url', None))
412
413 # Attempt to download using mplayer
414 if url.startswith('mms') or url.startswith('rtsp'):
415 return self._download_with_mplayer(filename, url)
416
417 # m3u8 manifest are downloaded with ffmpeg
418 if determine_ext(url) == u'm3u8':
419 return self._download_m3u8_with_ffmpeg(filename, url)
420
421 tmpfilename = self.temp_name(filename)
422 stream = None
423
424 # Do not include the Accept-Encoding header
425 headers = {'Youtubedl-no-compression': 'True'}
426 if 'user_agent' in info_dict:
427 headers['Youtubedl-user-agent'] = info_dict['user_agent']
428 basic_request = compat_urllib_request.Request(url, None, headers)
429 request = compat_urllib_request.Request(url, None, headers)
430
431 if self.params.get('test', False):
432 request.add_header('Range','bytes=0-10240')
433
434 # Establish possible resume length
435 if os.path.isfile(encodeFilename(tmpfilename)):
436 resume_len = os.path.getsize(encodeFilename(tmpfilename))
437 else:
438 resume_len = 0
439
440 open_mode = 'wb'
441 if resume_len != 0:
442 if self.params.get('continuedl', False):
443 self.report_resuming_byte(resume_len)
444 request.add_header('Range','bytes=%d-' % resume_len)
445 open_mode = 'ab'
446 else:
447 resume_len = 0
448
449 count = 0
450 retries = self.params.get('retries', 0)
451 while count <= retries:
452 # Establish connection
453 try:
454 if count == 0 and 'urlhandle' in info_dict:
455 data = info_dict['urlhandle']
456 data = compat_urllib_request.urlopen(request)
457 break
458 except (compat_urllib_error.HTTPError, ) as err:
459 if (err.code < 500 or err.code >= 600) and err.code != 416:
460 # Unexpected HTTP error
461 raise
462 elif err.code == 416:
463 # Unable to resume (requested range not satisfiable)
464 try:
465 # Open the connection again without the range header
466 data = compat_urllib_request.urlopen(basic_request)
467 content_length = data.info()['Content-Length']
468 except (compat_urllib_error.HTTPError, ) as err:
469 if err.code < 500 or err.code >= 600:
470 raise
471 else:
472 # Examine the reported length
473 if (content_length is not None and
474 (resume_len - 100 < int(content_length) < resume_len + 100)):
475 # The file had already been fully downloaded.
476 # Explanation to the above condition: in issue #175 it was revealed that
477 # YouTube sometimes adds or removes a few bytes from the end of the file,
478 # changing the file size slightly and causing problems for some users. So
479 # I decided to implement a suggested change and consider the file
480 # completely downloaded if the file size differs less than 100 bytes from
481 # the one in the hard drive.
482 self.report_file_already_downloaded(filename)
483 self.try_rename(tmpfilename, filename)
484 self._hook_progress({
485 'filename': filename,
486 'status': 'finished',
487 })
488 return True
489 else:
490 # The length does not match, we start the download over
491 self.report_unable_to_resume()
492 open_mode = 'wb'
493 break
494 # Retry
495 count += 1
496 if count <= retries:
497 self.report_retry(count, retries)
498
499 if count > retries:
500 self.report_error(u'giving up after %s retries' % retries)
501 return False
502
503 data_len = data.info().get('Content-length', None)
504 if data_len is not None:
505 data_len = int(data_len) + resume_len
506 min_data_len = self.params.get("min_filesize", None)
507 max_data_len = self.params.get("max_filesize", None)
508 if min_data_len is not None and data_len < min_data_len:
509 self.to_screen(u'\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
510 return False
511 if max_data_len is not None and data_len > max_data_len:
512 self.to_screen(u'\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
513 return False
514
515 data_len_str = self.format_bytes(data_len)
516 byte_counter = 0 + resume_len
517 block_size = self.params.get('buffersize', 1024)
518 start = time.time()
519 while True:
520 # Download and write
521 before = time.time()
522 data_block = data.read(block_size)
523 after = time.time()
524 if len(data_block) == 0:
525 break
526 byte_counter += len(data_block)
527
528 # Open file just in time
529 if stream is None:
530 try:
531 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
532 assert stream is not None
533 filename = self.undo_temp_name(tmpfilename)
534 self.report_destination(filename)
535 except (OSError, IOError) as err:
536 self.report_error(u'unable to open for writing: %s' % str(err))
537 return False
538 try:
539 stream.write(data_block)
540 except (IOError, OSError) as err:
541 self.to_stderr(u"\n")
542 self.report_error(u'unable to write data: %s' % str(err))
543 return False
544 if not self.params.get('noresizebuffer', False):
545 block_size = self.best_block_size(after - before, len(data_block))
546
547 # Progress message
548 speed = self.calc_speed(start, time.time(), byte_counter - resume_len)
549 if data_len is None:
550 self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA')
551 eta = None
552 else:
553 percent = self.calc_percent(byte_counter, data_len)
554 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
555 self.report_progress(percent, data_len_str, speed, eta)
556
557 self._hook_progress({
558 'downloaded_bytes': byte_counter,
559 'total_bytes': data_len,
560 'tmpfilename': tmpfilename,
561 'filename': filename,
562 'status': 'downloading',
563 'eta': eta,
564 'speed': speed,
565 })
566
567 # Apply rate limit
568 self.slow_down(start, byte_counter - resume_len)
569
570 if stream is None:
571 self.to_stderr(u"\n")
572 self.report_error(u'Did not get any data blocks')
573 return False
574 stream.close()
575 self.report_finish(data_len_str, (time.time() - start))
576 if data_len is not None and byte_counter != data_len:
577 raise ContentTooShortError(byte_counter, int(data_len))
578 self.try_rename(tmpfilename, filename)
579
580 # Update file modification time
581 if self.params.get('updatetime', True):
582 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))
583
584 self._hook_progress({
585 'downloaded_bytes': byte_counter,
586 'total_bytes': byte_counter,
587 'filename': filename,
588 'status': 'finished',
589 })
590
591 return True
592
593 def _hook_progress(self, status):
594 for ph in self._progress_hooks:
595 ph(status)
596
597 def add_progress_hook(self, ph):
598 """ ph gets called on download progress, with a dictionary with the entries
599 * filename: The final filename
600 * status: One of "downloading" and "finished"
601
602 It can also have some of the following entries:
603
604 * downloaded_bytes: Bytes on disks
605 * total_bytes: Total bytes, None if unknown
606 * tmpfilename: The filename we're currently writing to
607 * eta: The estimated time in seconds, None if unknown
608 * speed: The download speed in bytes/second, None if unknown
609
610 Hooks are guaranteed to be called at least once (with status "finished")
611 if the download is successful.
612 """
613 self._progress_hooks.append(ph)