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 ignoreerrors: Do not stop on download errors.
61 ratelimit: Download speed limit, in bytes/sec.
62 nooverwrites: Prevent overwriting files.
63 retries: Number of times to retry for HTTP error 5xx
64 buffersize: Size of download buffer in bytes.
65 noresizebuffer: Do not automatically resize the download buffer.
66 continuedl: Try to continue downloads if possible.
67 noprogress: Do not print the progress bar.
68 playliststart: Playlist item to start at.
69 playlistend: Playlist item to end at.
70 matchtitle: Download only matching titles.
71 rejecttitle: Reject downloads for matching titles.
72 logtostderr: Log messages to stderr instead of stdout.
73 consoletitle: Display progress in console window's titlebar.
74 nopart: Do not use temporary .part files.
75 updatetime: Use the Last-modified header to set output file timestamps.
76 writedescription: Write the video description to a .description file
77 writeinfojson: Write the video description to a .info.json file
78 writesubtitles: Write the video subtitles to a .srt file
79 subtitleslang: Language of the subtitles to download
85 _download_retcode = None
89 def __init__(self, params):
90 """Create a FileDownloader object with the given options."""
93 self._download_retcode = 0
94 self._num_downloads = 0
95 self._screen_file = [sys.stdout, sys.stderr][params.get('logtostderr', False)]
99 def format_bytes(bytes):
102 if type(bytes) is str:
107 exponent = long(math.log(bytes, 1024.0))
108 suffix = 'bkMGTPEZY'[exponent]
109 converted = float(bytes) / float(1024 ** exponent)
110 return '%.2f%s' % (converted, suffix)
113 def calc_percent(byte_counter, data_len):
116 return '%6s' % ('%3.1f%%' % (float(byte_counter) / float(data_len) * 100.0))
119 def calc_eta(start, now, total, current):
123 if current == 0 or dif < 0.001: # One millisecond
125 rate = float(current) / dif
126 eta = long((float(total) - float(current)) / rate)
127 (eta_mins, eta_secs) = divmod(eta, 60)
130 return '%02d:%02d' % (eta_mins, eta_secs)
133 def calc_speed(start, now, bytes):
135 if bytes == 0 or dif < 0.001: # One millisecond
136 return '%10s' % '---b/s'
137 return '%10s' % ('%s/s' % FileDownloader.format_bytes(float(bytes) / dif))
140 def best_block_size(elapsed_time, bytes):
141 new_min = max(bytes / 2.0, 1.0)
142 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
143 if elapsed_time < 0.001:
145 rate = bytes / elapsed_time
153 def parse_bytes(bytestr):
154 """Parse a string indicating a byte quantity into a long integer."""
155 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
158 number = float(matchobj.group(1))
159 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
160 return long(round(number * multiplier))
162 def add_info_extractor(self, ie):
163 """Add an InfoExtractor object to the end of the list."""
165 ie.set_downloader(self)
167 def add_post_processor(self, pp):
168 """Add a PostProcessor object to the end of the chain."""
170 pp.set_downloader(self)
172 def to_screen(self, message, skip_eol=False):
173 """Print message to stdout if not in quiet mode."""
174 assert type(message) == type(u'')
175 if not self.params.get('quiet', False):
176 terminator = [u'\n', u''][skip_eol]
177 output = message + terminator
179 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
180 output = output.encode(preferredencoding(), 'ignore')
181 self._screen_file.write(output)
182 self._screen_file.flush()
184 def to_stderr(self, message):
185 """Print message to stderr."""
186 print >>sys.stderr, message.encode(preferredencoding())
188 def to_cons_title(self, message):
189 """Set console/terminal window title to message."""
190 if not self.params.get('consoletitle', False):
192 if os.name == 'nt' and ctypes.windll.kernel32.GetConsoleWindow():
193 # c_wchar_p() might not be necessary if `message` is
194 # already of type unicode()
195 ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
196 elif 'TERM' in os.environ:
197 sys.stderr.write('\033]0;%s\007' % message.encode(preferredencoding()))
199 def fixed_template(self):
200 """Checks if the output template is fixed."""
201 return (re.search(ur'(?u)%\(.+?\)s', self.params['outtmpl']) is None)
203 def trouble(self, message=None):
204 """Determine action to take when a download problem appears.
206 Depending on if the downloader has been configured to ignore
207 download errors or not, this method may throw an exception or
208 not when errors are found, after printing the message.
210 if message is not None:
211 self.to_stderr(message)
212 if not self.params.get('ignoreerrors', False):
213 raise DownloadError(message)
214 self._download_retcode = 1
216 def slow_down(self, start_time, byte_counter):
217 """Sleep if the download speed is over the rate limit."""
218 rate_limit = self.params.get('ratelimit', None)
219 if rate_limit is None or byte_counter == 0:
222 elapsed = now - start_time
225 speed = float(byte_counter) / elapsed
226 if speed > rate_limit:
227 time.sleep((byte_counter - rate_limit * (now - start_time)) / rate_limit)
229 def temp_name(self, filename):
230 """Returns a temporary filename for the given filename."""
231 if self.params.get('nopart', False) or filename == u'-' or \
232 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
234 return filename + u'.part'
236 def undo_temp_name(self, filename):
237 if filename.endswith(u'.part'):
238 return filename[:-len(u'.part')]
241 def try_rename(self, old_filename, new_filename):
243 if old_filename == new_filename:
245 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
246 except (IOError, OSError), err:
247 self.trouble(u'ERROR: unable to rename file')
249 def try_utime(self, filename, last_modified_hdr):
250 """Try to set the last-modified time of the given file."""
251 if last_modified_hdr is None:
253 if not os.path.isfile(encodeFilename(filename)):
255 timestr = last_modified_hdr
258 filetime = timeconvert(timestr)
262 os.utime(filename, (time.time(), filetime))
267 def report_writedescription(self, descfn):
268 """ Report that the description file is being written """
269 self.to_screen(u'[info] Writing video description to: ' + descfn)
271 def report_writesubtitles(self, srtfn):
272 """ Report that the subtitles file is being written """
273 self.to_screen(u'[info] Writing video subtitles to: ' + srtfn)
275 def report_writeinfojson(self, infofn):
276 """ Report that the metadata file has been written """
277 self.to_screen(u'[info] Video description metadata as JSON to: ' + infofn)
279 def report_destination(self, filename):
280 """Report destination filename."""
281 self.to_screen(u'[download] Destination: ' + filename)
283 def report_progress(self, percent_str, data_len_str, speed_str, eta_str):
284 """Report download progress."""
285 if self.params.get('noprogress', False):
287 self.to_screen(u'\r[download] %s of %s at %s ETA %s' %
288 (percent_str, data_len_str, speed_str, eta_str), skip_eol=True)
289 self.to_cons_title(u'youtube-dl - %s of %s at %s ETA %s' %
290 (percent_str.strip(), data_len_str.strip(), speed_str.strip(), eta_str.strip()))
292 def report_resuming_byte(self, resume_len):
293 """Report attempt to resume at given byte."""
294 self.to_screen(u'[download] Resuming download at byte %s' % resume_len)
296 def report_retry(self, count, retries):
297 """Report retry in case of HTTP error 5xx"""
298 self.to_screen(u'[download] Got server HTTP error. Retrying (attempt %d of %d)...' % (count, retries))
300 def report_file_already_downloaded(self, file_name):
301 """Report file has already been fully downloaded."""
303 self.to_screen(u'[download] %s has already been downloaded' % file_name)
304 except (UnicodeEncodeError), err:
305 self.to_screen(u'[download] The file has already been downloaded')
307 def report_unable_to_resume(self):
308 """Report it was impossible to resume download."""
309 self.to_screen(u'[download] Unable to resume')
311 def report_finish(self):
312 """Report download finished."""
313 if self.params.get('noprogress', False):
314 self.to_screen(u'[download] Download completed')
318 def increment_downloads(self):
319 """Increment the ordinal that assigns a number to each file."""
320 self._num_downloads += 1
322 def prepare_filename(self, info_dict):
323 """Generate the output filename."""
325 template_dict = dict(info_dict)
326 template_dict['epoch'] = unicode(long(time.time()))
327 template_dict['autonumber'] = unicode('%05d' % self._num_downloads)
328 filename = self.params['outtmpl'] % template_dict
330 except (ValueError, KeyError), err:
331 self.trouble(u'ERROR: invalid system charset or erroneous output template')
334 def _match_entry(self, info_dict):
335 """ Returns None iff the file should be downloaded """
337 title = info_dict['title']
338 matchtitle = self.params.get('matchtitle', False)
339 if matchtitle and not re.search(matchtitle, title, re.IGNORECASE):
340 return u'[download] "' + title + '" title did not match pattern "' + matchtitle + '"'
341 rejecttitle = self.params.get('rejecttitle', False)
342 if rejecttitle and re.search(rejecttitle, title, re.IGNORECASE):
343 return u'"' + title + '" title matched reject pattern "' + rejecttitle + '"'
346 def process_info(self, info_dict):
347 """Process a single dictionary returned by an InfoExtractor."""
349 info_dict['stitle'] = sanitize_filename(info_dict['title'])
351 reason = self._match_entry(info_dict)
352 if reason is not None:
353 self.to_screen(u'[download] ' + reason)
356 max_downloads = self.params.get('max_downloads')
357 if max_downloads is not None:
358 if self._num_downloads > int(max_downloads):
359 raise MaxDownloadsReached()
361 filename = self.prepare_filename(info_dict)
364 if self.params.get('forcetitle', False):
365 print info_dict['title'].encode(preferredencoding(), 'xmlcharrefreplace')
366 if self.params.get('forceurl', False):
367 print info_dict['url'].encode(preferredencoding(), 'xmlcharrefreplace')
368 if self.params.get('forcethumbnail', False) and 'thumbnail' in info_dict:
369 print info_dict['thumbnail'].encode(preferredencoding(), 'xmlcharrefreplace')
370 if self.params.get('forcedescription', False) and 'description' in info_dict:
371 print info_dict['description'].encode(preferredencoding(), 'xmlcharrefreplace')
372 if self.params.get('forcefilename', False) and filename is not None:
373 print filename.encode(preferredencoding(), 'xmlcharrefreplace')
374 if self.params.get('forceformat', False):
375 print info_dict['format'].encode(preferredencoding(), 'xmlcharrefreplace')
377 # Do nothing else if in simulate mode
378 if self.params.get('simulate', False):
385 dn = os.path.dirname(encodeFilename(filename))
386 if dn != '' and not os.path.exists(dn): # dn is already encoded
388 except (OSError, IOError), err:
389 self.trouble(u'ERROR: unable to create directory ' + unicode(err))
392 if self.params.get('writedescription', False):
394 descfn = filename + u'.description'
395 self.report_writedescription(descfn)
396 descfile = open(encodeFilename(descfn), 'wb')
398 descfile.write(info_dict['description'].encode('utf-8'))
401 except (OSError, IOError):
402 self.trouble(u'ERROR: Cannot write description file ' + descfn)
405 if self.params.get('writesubtitles', False) and 'subtitles' in info_dict and info_dict['subtitles']:
406 # subtitles download errors are already managed as troubles in relevant IE
407 # that way it will silently go on when used with unsupporting IE
409 srtfn = filename.rsplit('.', 1)[0] + u'.srt'
410 self.report_writesubtitles(srtfn)
411 srtfile = open(encodeFilename(srtfn), 'wb')
413 srtfile.write(info_dict['subtitles'].encode('utf-8'))
416 except (OSError, IOError):
417 self.trouble(u'ERROR: Cannot write subtitles file ' + descfn)
420 if self.params.get('writeinfojson', False):
421 infofn = filename + u'.info.json'
422 self.report_writeinfojson(infofn)
425 except (NameError,AttributeError):
426 self.trouble(u'ERROR: No JSON encoder found. Update to Python 2.6+, setup a json module, or leave out --write-info-json.')
429 infof = open(encodeFilename(infofn), 'wb')
431 json_info_dict = dict((k,v) for k,v in info_dict.iteritems() if not k in ('urlhandle',))
432 json.dump(json_info_dict, infof)
435 except (OSError, IOError):
436 self.trouble(u'ERROR: Cannot write metadata to JSON file ' + infofn)
439 if not self.params.get('skip_download', False):
440 if self.params.get('nooverwrites', False) and os.path.exists(encodeFilename(filename)):
444 success = self._do_download(filename, info_dict)
445 except (OSError, IOError), err:
446 raise UnavailableVideoError
447 except (urllib2.URLError, httplib.HTTPException, socket.error), err:
448 self.trouble(u'ERROR: unable to download video data: %s' % str(err))
450 except (ContentTooShortError, ), err:
451 self.trouble(u'ERROR: content too short (expected %s bytes and served %s)' % (err.expected, err.downloaded))
456 self.post_process(filename, info_dict)
457 except (PostProcessingError), err:
458 self.trouble(u'ERROR: postprocessing: %s' % str(err))
461 def download(self, url_list):
462 """Download a given list of URLs."""
463 if len(url_list) > 1 and self.fixed_template():
464 raise SameFileError(self.params['outtmpl'])
467 suitable_found = False
469 # Go to next InfoExtractor if not suitable
470 if not ie.suitable(url):
473 # Suitable InfoExtractor found
474 suitable_found = True
476 # Extract information from URL and process it
477 videos = ie.extract(url)
478 for video in videos or []:
480 self.increment_downloads()
481 self.process_info(video)
482 except UnavailableVideoError:
483 self.trouble(u'\nERROR: unable to download video')
485 # Suitable InfoExtractor had been found; go to next URL
488 if not suitable_found:
489 self.trouble(u'ERROR: no suitable InfoExtractor: %s' % url)
491 return self._download_retcode
493 def post_process(self, filename, ie_info):
494 """Run the postprocessing chain on the given file."""
496 info['filepath'] = filename
502 def _download_with_rtmpdump(self, filename, url, player_url):
503 self.report_destination(filename)
504 tmpfilename = self.temp_name(filename)
506 # Check for rtmpdump first
508 subprocess.call(['rtmpdump', '-h'], stdout=(file(os.path.devnull, 'w')), stderr=subprocess.STDOUT)
509 except (OSError, IOError):
510 self.trouble(u'ERROR: RTMP download detected but "rtmpdump" could not be run')
513 # Download using rtmpdump. rtmpdump returns exit code 2 when
514 # the connection was interrumpted and resuming appears to be
515 # possible. This is part of rtmpdump's normal usage, AFAIK.
516 basic_args = ['rtmpdump', '-q'] + [[], ['-W', player_url]][player_url is not None] + ['-r', url, '-o', tmpfilename]
517 args = basic_args + [[], ['-e', '-k', '1']][self.params.get('continuedl', False)]
518 if self.params.get('verbose', False):
521 shell_quote = lambda args: ' '.join(map(pipes.quote, args))
524 self.to_screen(u'[debug] rtmpdump command line: ' + shell_quote(args))
525 retval = subprocess.call(args)
526 while retval == 2 or retval == 1:
527 prevsize = os.path.getsize(encodeFilename(tmpfilename))
528 self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
529 time.sleep(5.0) # This seems to be needed
530 retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
531 cursize = os.path.getsize(encodeFilename(tmpfilename))
532 if prevsize == cursize and retval == 1:
534 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
535 if prevsize == cursize and retval == 2 and cursize > 1024:
536 self.to_screen(u'\r[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
540 self.to_screen(u'\r[rtmpdump] %s bytes' % os.path.getsize(encodeFilename(tmpfilename)))
541 self.try_rename(tmpfilename, filename)
544 self.trouble(u'\nERROR: rtmpdump exited with code %d' % retval)
547 def _do_download(self, filename, info_dict):
548 url = info_dict['url']
549 player_url = info_dict.get('player_url', None)
551 # Check file already present
552 if self.params.get('continuedl', False) and os.path.isfile(encodeFilename(filename)) and not self.params.get('nopart', False):
553 self.report_file_already_downloaded(filename)
556 # Attempt to download using rtmpdump
557 if url.startswith('rtmp'):
558 return self._download_with_rtmpdump(filename, url, player_url)
560 tmpfilename = self.temp_name(filename)
563 # Do not include the Accept-Encoding header
564 headers = {'Youtubedl-no-compression': 'True'}
565 basic_request = urllib2.Request(url, None, headers)
566 request = urllib2.Request(url, None, headers)
568 # Establish possible resume length
569 if os.path.isfile(encodeFilename(tmpfilename)):
570 resume_len = os.path.getsize(encodeFilename(tmpfilename))
576 if self.params.get('continuedl', False):
577 self.report_resuming_byte(resume_len)
578 request.add_header('Range','bytes=%d-' % resume_len)
584 retries = self.params.get('retries', 0)
585 while count <= retries:
586 # Establish connection
588 if count == 0 and 'urlhandle' in info_dict:
589 data = info_dict['urlhandle']
590 data = urllib2.urlopen(request)
592 except (urllib2.HTTPError, ), err:
593 if (err.code < 500 or err.code >= 600) and err.code != 416:
594 # Unexpected HTTP error
596 elif err.code == 416:
597 # Unable to resume (requested range not satisfiable)
599 # Open the connection again without the range header
600 data = urllib2.urlopen(basic_request)
601 content_length = data.info()['Content-Length']
602 except (urllib2.HTTPError, ), err:
603 if err.code < 500 or err.code >= 600:
606 # Examine the reported length
607 if (content_length is not None and
608 (resume_len - 100 < long(content_length) < resume_len + 100)):
609 # The file had already been fully downloaded.
610 # Explanation to the above condition: in issue #175 it was revealed that
611 # YouTube sometimes adds or removes a few bytes from the end of the file,
612 # changing the file size slightly and causing problems for some users. So
613 # I decided to implement a suggested change and consider the file
614 # completely downloaded if the file size differs less than 100 bytes from
615 # the one in the hard drive.
616 self.report_file_already_downloaded(filename)
617 self.try_rename(tmpfilename, filename)
620 # The length does not match, we start the download over
621 self.report_unable_to_resume()
627 self.report_retry(count, retries)
630 self.trouble(u'ERROR: giving up after %s retries' % retries)
633 data_len = data.info().get('Content-length', None)
634 if data_len is not None:
635 data_len = long(data_len) + resume_len
636 data_len_str = self.format_bytes(data_len)
637 byte_counter = 0 + resume_len
638 block_size = self.params.get('buffersize', 1024)
643 data_block = data.read(block_size)
645 if len(data_block) == 0:
647 byte_counter += len(data_block)
649 # Open file just in time
652 (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode)
653 assert stream is not None
654 filename = self.undo_temp_name(tmpfilename)
655 self.report_destination(filename)
656 except (OSError, IOError), err:
657 self.trouble(u'ERROR: unable to open for writing: %s' % str(err))
660 stream.write(data_block)
661 except (IOError, OSError), err:
662 self.trouble(u'\nERROR: unable to write data: %s' % str(err))
664 if not self.params.get('noresizebuffer', False):
665 block_size = self.best_block_size(after - before, len(data_block))
668 speed_str = self.calc_speed(start, time.time(), byte_counter - resume_len)
670 self.report_progress('Unknown %', data_len_str, speed_str, 'Unknown ETA')
672 percent_str = self.calc_percent(byte_counter, data_len)
673 eta_str = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
674 self.report_progress(percent_str, data_len_str, speed_str, eta_str)
677 self.slow_down(start, byte_counter - resume_len)
680 self.trouble(u'\nERROR: Did not get any data blocks')
684 if data_len is not None and byte_counter != data_len:
685 raise ContentTooShortError(byte_counter, long(data_len))
686 self.try_rename(tmpfilename, filename)
688 # Update file modification time
689 if self.params.get('updatetime', True):
690 info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None))