2 # -*- coding: utf-8 -*-
20 class FileDownloader(object):
21 """File Downloader class.
23 File downloader objects are the ones responsible of downloading the
24 actual video file and writing it to disk if the user has requested
25 it, among some other tasks. In most cases there should be one per
26 program. As, given a video URL, the downloader doesn't know how to
27 extract all the needed information, task that InfoExtractors do, it
28 has to pass the URL to one of them.
30 For this, file downloader objects have a method that allows
31 InfoExtractors to be registered in a given order. When it is passed
32 a URL, the file downloader handles it to the first InfoExtractor it
33 finds that reports being able to handle it. The InfoExtractor extracts
34 all the information about the video or videos the URL refers to, and
35 asks the FileDownloader to process the video information, possibly
36 downloading the video.
38 File downloaders accept a lot of parameters. In order not to saturate
39 the object constructor with arguments, it receives a dictionary of
40 options instead. These options are available through the params
41 attribute for the InfoExtractors to use. The FileDownloader also
42 registers itself as the downloader in charge for the InfoExtractors
43 that are added to it, so this is a "mutual registration".
47 username: Username for authentication purposes.
48 password: Password for authentication purposes.
49 usenetrc: Use netrc for authentication instead.
50 quiet: Do not print messages to stdout.
51 forceurl: Force printing final URL.
52 forcetitle: Force printing title.
53 forcethumbnail: Force printing thumbnail URL.
54 forcedescription: Force printing description.
55 forcefilename: Force printing final filename.
56 simulate: Do not download the video files.
57 format: Video format code.
58 format_limit: Highest quality format to try.
59 outtmpl: Template for output names.
60 restrictfilenames: Do not allow "&" and spaces in file names
61 ignoreerrors: Do not stop on download errors.
62 ratelimit: Download speed limit, in bytes/sec.
63 nooverwrites: Prevent overwriting files.
64 retries: Number of times to retry for HTTP error 5xx
65 continuedl: Try to continue downloads if possible.
66 noprogress: Do not print the progress bar.
67 playliststart: Playlist item to start at.
68 playlistend: Playlist item to end at.
69 matchtitle: Download only matching titles.
70 rejecttitle: Reject downloads for matching titles.
71 logtostderr: Log messages to stderr instead of stdout.
72 consoletitle: Display progress in console window's titlebar.
73 nopart: Do not use temporary .part files.
74 updatetime: Use the Last-modified header to set output file timestamps.
75 writedescription: Write the video description to a .description file
76 writeinfojson: Write the video description to a .info.json file
77 writesubtitles: Write the video subtitles to a .srt file
78 subtitleslang: Language of the subtitles to download
84 _download_retcode = None
88 def __init__(self, params):
89 """Create a FileDownloader object with the given options."""
92 self._download_retcode = 0
93 self._num_downloads = 0
94 self._screen_file = [sys.stdout, sys.stderr][params.get('logtostderr', False)]
97 if '%(stitle)s' in self.params['outtmpl']:
98 self.to_stderr(u'WARNING: %(stitle)s is deprecated. Use the %(title)s and the --restrict-filenames flag(which also secures %(uploader)s et al) instead.')
101 def format_bytes(bytes):
104 if type(bytes) is str:
109 exponent = long(math.log(bytes, 1024.0))
110 suffix = 'bkMGTPEZY'[exponent]
111 converted = float(bytes) / float(1024 ** exponent)
112 return '%.2f%s' % (converted, suffix)
115 def calc_percent(byte_counter, data_len):
118 return '%6s' % ('%3.1f%%' % (float(byte_counter) / float(data_len) * 100.0))
121 def calc_eta(start, now, total, current):
125 if current == 0 or dif < 0.001: # One millisecond
127 rate = float(current) / dif
128 eta = long((float(total) - float(current)) / rate)
129 (eta_mins, eta_secs) = divmod(eta, 60)
132 return '%02d:%02d' % (eta_mins, eta_secs)
135 def calc_speed(start, now, bytes):
137 if bytes == 0 or dif < 0.001: # One millisecond
138 return '%10s' % '---b/s'
139 return '%10s' % ('%s/s' % FileDownloader.format_bytes(float(bytes) / dif))
142 def best_block_size(elapsed_time, bytes):
143 new_min = max(bytes / 2.0, 1.0)
144 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
145 if elapsed_time < 0.001:
147 rate = bytes / elapsed_time
155 def parse_bytes(bytestr):
156 """Parse a string indicating a byte quantity into an integer."""
157 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
160 number = float(matchobj.group(1))
161 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
162 return int(round(number * multiplier))
164 def add_info_extractor(self, ie):
165 """Add an InfoExtractor object to the end of the list."""
167 ie.set_downloader(self)
169 def add_post_processor(self, pp):
170 """Add a PostProcessor object to the end of the chain."""
172 pp.set_downloader(self)
174 def to_screen(self, message, skip_eol=False):
175 """Print message to stdout if not in quiet mode."""
176 assert type(message) == type(u'')
177 if not self.params.get('quiet', False):
178 terminator = [u'\n', u''][skip_eol]
179 output = message + terminator
180 if 'b' not in self._screen_file.mode or sys.version_info[0] < 3: # Python 2 lies about the mode of sys.stdout/sys.stderr
181 output = output.encode(preferredencoding(), 'ignore')
182 self._screen_file.write(output)
183 self._screen_file.flush()
185 def to_stderr(self, message):
186 """Print message to stderr."""
187 assert type(message) == type(u'')
188 sys.stderr.write((message + u'\n').encode(preferredencoding()))
190 def to_cons_title(self, message):
191 """Set console/terminal window title to message."""
192 if not self.params.get('consoletitle', False):
194 if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow():
195 # c_wchar_p() might not be necessary if `message` is
196 # already of type unicode()
197 ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
198 elif 'TERM' in os.environ:
199 sys.stderr.write('\033]0;%s\007' % message.encode(preferredencoding()))
201 def fixed_template(self):
202 """Checks if the output template is fixed."""
203 return (re.search(ur'(?u)%\(.+?\)s', self.params['outtmpl']) is None)
205 def trouble(self, message=None):
206 """Determine action to take when a download problem appears.
208 Depending on if the downloader has been configured to ignore
209 download errors or not, this method may throw an exception or
210 not when errors are found, after printing the message.
212 if message is not None:
213 self.to_stderr(message)
214 if not self.params.get('ignoreerrors', False):
215 raise DownloadError(message)
216 self._download_retcode = 1
218 def slow_down(self, start_time, byte_counter):
219 """Sleep if the download speed is over the rate limit."""
220 rate_limit = self.params.get('ratelimit', None)
221 if rate_limit is None or byte_counter == 0:
224 elapsed = now - start_time
227 speed = float(byte_counter) / elapsed
228 if speed > rate_limit:
229 time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
231 def temp_name(self, filename):
232 """Returns a temporary filename for the given filename."""
233 if self.params.get('nopart', False) or filename == u'-' or \
234 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
236 return filename + u'.part'
238 def undo_temp_name(self, filename):
239 if filename.endswith(u'.part'):
240 return filename[:-len(u'.part')]
243 def try_rename(self, old_filename, new_filename):
245 if old_filename == new_filename:
247 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
248 except (IOError, OSError), err:
249 self.trouble(u'ERROR: unable to rename file')
251 def try_utime(self, filename, last_modified_hdr):
252 """Try to set the last-modified time of the given file."""
253 if last_modified_hdr is None:
255 if not os.path.isfile(encodeFilename(filename)):
257 timestr = last_modified_hdr
260 filetime = timeconvert(timestr)
264 os.utime(filename, (time.time(), filetime))
269 def report_writedescription(self, descfn):
270 """ Report that the description file is being written """
271 self.to_screen(u'[info] Writing video description to: ' + descfn)
273 def report_writesubtitles(self, srtfn):
274 """ Report that the subtitles file is being written """
275 self.to_screen(u'[info] Writing video subtitles to: ' + srtfn)
277 def report_writeinfojson(self, infofn):
278 """ Report that the metadata file has been written """
279 self.to_screen(u'[info] Video description metadata as JSON to: ' + infofn)
281 def report_destination(self, filename):
282 """Report destination filename."""
283 self.to_screen(u'[download] Destination: ' + filename)
285 def report_progress(self, percent_str, data_len_str, speed_str, eta_str):
286 """Report download progress."""
287 if self.params.get('noprogress', False):
289 self.to_screen(u'\r[download] %s of %s at %s ETA %s' %
290 (percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
291 self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' %
292 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
294 def report_resuming_byte(self, resume_len):
295 """Report attempt to resume at given byte."""
296 self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
298 def report_retry(self, count, retries):
299 """Report retry in case of HTTP error 5xx"""
300 self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
302 def report_file_already_downloaded(self, file_name):
303 """Report file has already been fully downloaded."""
305 self.to_screen(u'[download] %s has already been downloaded' % file_name)
306 except (UnicodeEncodeError), err:
307 self.to_screen(u'[download] The file has already been downloaded')
309 def report_unable_to_resume(self):
310 """Report it was impossible to resume download."""
311 self.to_screen(u'[download] Unable to resume')
313 def report_finish(self):
314 """Report download finished."""
315 if self.params.get('noprogress', False):
316 self.to_screen(u'[download] Download completed')
320 def increment_downloads(self):
321 """Increment the ordinal that assigns a number to each file."""
322 self._num_downloads += 1
324 def prepare_filename(self, info_dict):
325 """Generate the output filename."""
327 template_dict = dict(info_dict)
328 template_dict['epoch'] = unicode(int(time.time()))
329 template_dict['autonumber'] = unicode('%05d' % self._num_downloads)
330 template_dict = dict((key, u'NA' if val is None else val) for key, val in template_dict.items())
331 filename = self.params['outtmpl'] % template_dict
333 except (ValueError, KeyError), err:
334 self.trouble(u'ERROR: invalid system charset or erroneous output template')
337 def _match_entry(self, info_dict):
338 """ Returns None iff the file should be downloaded """
340 title = info_dict['title']
341 matchtitle = self.params.get('matchtitle', False)
343 matchtitle = matchtitle.decode('utf8')
344 if not re.search(matchtitle, title, re.IGNORECASE):
345 return u'[download] "' + title + '" title did not match pattern "' + matchtitle + '"'
346 rejecttitle = self.params.get('rejecttitle', False)
348 rejecttitle = rejecttitle.decode('utf8')
349 if re.search(rejecttitle, title, re.IGNORECASE):
350 return u'"' + title + '" title matched reject pattern "' + rejecttitle + '"'
353 def process_info(self, info_dict):
354 """Process a single dictionary returned by an InfoExtractor."""
356 # Keep for backwards compatibility
357 info_dict['stitle'] = info_dict['title']
359 if not 'format' in info_dict:
360 info_dict['format'] = info_dict['ext']
362 reason = self._match_entry(info_dict)
363 if reason is not None:
364 self.to_screen(u'[download] ' + reason)
367 max_downloads = self.params.get('max_downloads')
368 if max_downloads is not None:
369 if self._num_downloads > int(max_downloads):
370 raise MaxDownloadsReached()
372 filename = self.prepare_filename(info_dict)
373 filename = sanitize_filename(filename, self.params.get('restrictfilenames'))
376 if self.params.get('forcetitle', False):
377 print(info_dict['title'].encode(preferredencoding(), 'xmlcharrefreplace'))
378 if self.params.get('forceurl', False):
379 print(info_dict['url'].encode(preferredencoding(), 'xmlcharrefreplace'))
380 if self.params.get('forcethumbnail', False) and 'thumbnail' in info_dict:
381 print(info_dict['thumbnail'].encode(preferredencoding(), 'xmlcharrefreplace'))
382 if self.params.get('forcedescription', False) and 'description' in info_dict:
383 print(info_dict['description'].encode(preferredencoding(), 'xmlcharrefreplace'))
384 if self.params.get('forcefilename', False) and filename is not None:
385 print(filename.encode(preferredencoding(), 'xmlcharrefreplace'))
386 if self.params.get('forceformat', False):
387 print(info_dict['format'].encode(preferredencoding(), 'xmlcharrefreplace'))
389 # Do nothing else if in simulate mode
390 if self.params.get('simulate', False):
397 dn = os.path.dirname(encodeFilename(filename))
398 if dn != '' and not os.path.exists(dn): # dn is already encoded
400 except (OSError, IOError), err:
401 self.trouble(u'ERROR: unable to create directory ' + unicode(err))
404 if self.params.get('writedescription', False):
406 descfn = filename + u'.description'
407 self.report_writedescription(descfn)
408 descfile = open(encodeFilename(descfn), 'wb')
410 descfile.write(info_dict['description'].encode('utf-8'))
413 except (OSError, IOError):
414 self.trouble(u'ERROR: Cannot write description file ' + descfn)
417 if self.params.get('writesubtitles', False) and 'subtitles' in info_dict and info_dict['subtitles']:
418 # subtitles download errors are already managed as troubles in relevant IE
419 # that way it will silently go on when used with unsupporting IE
421 srtfn = filename.rsplit('.', 1)[0] + u'.srt'
422 self.report_writesubtitles(srtfn)
423 srtfile = open(encodeFilename(srtfn), 'wb')
425 srtfile.write(info_dict['subtitles'].encode('utf-8'))
428 except (OSError, IOError):
429 self.trouble(u'ERROR: Cannot write subtitles file ' + descfn)
432 if self.params.get('writeinfojson', False):
433 infofn = filename + u'.info.json'
434 self.report_writeinfojson(infofn)
437 except (NameError,AttributeError):
438 self.trouble(u'ERROR: No JSON encoder found. Update to Python 2.6+, setup a json module, or leave out --write-info-json.')
441 infof = open(encodeFilename(infofn), 'wb')
443 json_info_dict = dict((k,v) for k,v in info_dict.iteritems() if not k in ('urlhandle',))
444 json.dump(json_info_dict, infof)
447 except (OSError, IOError):
448 self.trouble(u'ERROR: Cannot write metadata to JSON file ' + infofn)
451 if not self.params.get('skip_download', False):
452 if self.params.get('nooverwrites', False) and os.path.exists(encodeFilename(filename)):
456 success = self._do_download(filename, info_dict)
457 except (OSError, IOError), err:
458 raise UnavailableVideoError
459 except (urllib2.URLError, httplib.HTTPException, socket.error), err:
460 self.trouble(u'ERROR: unable to download video data: %s' % str(err))
462 except (ContentTooShortError, ), err:
463 self.trouble(u'ERROR: content too short (expected %s bytes and served %s)' % (err.expected, err.downloaded))
468 self.post_process(filename, info_dict)
469 except (PostProcessingError), err:
470 self.trouble(u'ERROR: postprocessing: %s' % str(err))
473 def download(self, url_list):
474 """Download a given list of URLs."""
475 if len(url_list) > 1 and self.fixed_template():
476 raise SameFileError(self.params['outtmpl'])
479 suitable_found = False
481 # Go to next InfoExtractor if not suitable
482 if not ie.suitable(url):
485 # Warn if the _WORKING attribute is False
487 self.trouble(u'WARNING: the program functionality for this site has been marked as broken, '
488 u'and will probably not work. If you want to go on, use the -i option.')
490 # Suitable InfoExtractor found
491 suitable_found = True
493 # Extract information from URL and process it
494 videos = ie.extract(url)
495 for video in videos or []:
496 video['extractor'] = ie.IE_NAME
498 self.increment_downloads()
499 self.process_info(video)
500 except UnavailableVideoError:
501 self.trouble(u'\nERROR: unable to download video')
503 # Suitable InfoExtractor had been found; go to next URL
506 if not suitable_found:
507 self.trouble(u'ERROR: no suitable InfoExtractor: %s' % url)
509 return self._download_retcode
511 def post_process(self, filename, ie_info):
512 """Run the postprocessing chain on the given file."""
514 info['filepath'] = filename
520 def _download_with_rtmpdump(self, filename, url, player_url):
521 self.report_destination(filename)
522 tmpfilename = self.temp_name(filename)
524 # Check for rtmpdump first
526 subprocess.call(['rtmpdump', '-h'], stdout=(file(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
527 except (OSError, IOError):
528 self.trouble(u'ERROR: RTMP download detected but "rtmpdump" could not be run')
531 # Download using rtmpdump. rtmpdump returns exit code 2 when
532 # the connection was interrumpted and resuming appears to be
533 # possible. This is part of rtmpdump's normal usage, AFAIK.
534 basic_args = ['rtmpdump', '-q'] + [[], ['-W', player_url]][player_url is not None] + ['-r', url, '-o', tmpfilename]
535 args = basic_args + [[], ['-e', '-k', '1']][self.params.get('continuedl', False)]
536 if self.params.get('verbose', False):
539 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
542 self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
543 retval = subprocess.call(args)
544 while retval == 2 or retval == 1:
545 prevsize = os.path.getsize(encodeFilename(tmpfilename))
546 self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
547 time.sleep(5.0) # This seems to be needed
548 retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
549 cursize = os.path.getsize(encodeFilename(tmpfilename))
550 if prevsize == cursize and retval == 1:
552 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
553 if prevsize == cursize and retval == 2 and cursize > 1024:
554 self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
558 self.to_screen(u'\r[rtmpdump] %s bytes' % os.path.getsize(encodeFilename(tmpfilename)))
559 self.try_rename(tmpfilename, filename)
562 self.trouble(u'\nERROR: rtmpdump exited with code %d' % retval)
565 def _do_download(self, filename, info_dict):
566 url = info_dict['url']
567 player_url = info_dict.get('player_url', None)
569 # Check file already present
570 if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
571 self.report_file_already_downloaded(filename)
574 # Attempt to download using rtmpdump
575 if url.startswith('rtmp'):
576 return self._download_with_rtmpdump(filename, url, player_url)
578 tmpfilename = self.temp_name(filename)
581 # Do not include the Accept-Encoding header
582 headers = {'Youtubedl-no-compression': 'True'}
583 basic_request = urllib2.Request(url, None, headers)
584 request = urllib2.Request(url, None, headers)
586 # Establish possible resume length
587 if os.path.isfile(encodeFilename(tmpfilename)):
588 resume_len = os.path.getsize(encodeFilename(tmpfilename))
594 if self.params.get('continuedl', False):
595 self.report_resuming_byte(resume_len)
596 request.add_header('Range','bytes=%d-' % resume_len)
602 retries = self.params.get('retries', 0)
603 while count <= retries:
604 # Establish connection
606 if count == 0 and 'urlhandle' in info_dict:
607 data = info_dict['urlhandle']
608 data = urllib2.urlopen(request)
610 except (urllib2.HTTPError, ), err:
611 if (err.code < 500 or err.code >= 600) and err.code != 416:
612 # Unexpected HTTP error
614 elif err.code == 416:
615 # Unable to resume (requested range not satisfiable)
617 # Open the connection again without the range header
618 data = urllib2.urlopen(basic_request)
619 content_length = data.info()['Content-Length']
620 except (urllib2.HTTPError, ), err:
621 if err.code < 500 or err.code >= 600:
624 # Examine the reported length
625 if (content_length is not None and
626 (resume_len - 100 < long(content_length) < resume_len + 100)):
627 # The file had already been fully downloaded.
628 # Explanation to the above condition: in issue #175 it was revealed that
629 # YouTube sometimes adds or removes a few bytes from the end of the file,
630 # changing the file size slightly and causing problems for some users. So
631 # I decided to implement a suggested change and consider the file
632 # completely downloaded if the file size differs less than 100 bytes from
633 # the one in the hard drive.
634 self.report_file_already_downloaded(filename)
635 self.try_rename(tmpfilename, filename)
638 # The length does not match, we start the download over
639 self.report_unable_to_resume()
645 self.report_retry(count, retries)
648 self.trouble(u'ERROR: giving up after %s retries' % retries)
651 data_len = data.info().get('Content-length', None)
652 if data_len is not None:
653 data_len = long(data_len) + resume_len
654 data_len_str = self.format_bytes(data_len)
655 byte_counter = 0 + resume_len
661 data_block = data.read(block_size)
663 if len(data_block) == 0:
665 byte_counter += len(data_block)
667 # Open file just in time
670 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
671 assert stream is not None
672 filename = self.undo_temp_name(tmpfilename)
673 self.report_destination(filename)
674 except (OSError, IOError), err:
675 self.trouble(u'ERROR: unable to open for writing: %s' % str(err))
678 stream.write(data_block)
679 except (IOError, OSError), err:
680 self.trouble(u'\nERROR: unable to write data: %s' % str(err))
682 block_size = self.best_block_size(after - before, len(data_block))
685 speed_str = self.calc_speed(start, time.time(), byte_counter - resume_len)
687 self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA')
689 percent_str = self.calc_percent(byte_counter, data_len)
690 eta_str = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
691 self.report_progress(percent_str, data_len_str, speed_str, eta_str)
694 self.slow_down(start, byte_counter - resume_len)
697 self.trouble(u'\nERROR: Did not get any data blocks')
701 if data_len is not None and byte_counter != data_len:
702 raise ContentTooShortError(byte_counter, long(data_len))
703 self.try_rename(tmpfilename, filename)
705 # Update file modification time
706 if self.params.get('updatetime', True):
707 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))