1 from __future__ import division, unicode_literals
9 from ..compat import compat_os_name
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.
26 File downloaders accept a lot of parameters. In order not to saturate
27 the object constructor with arguments, it receives a dictionary of
32 verbose: Print additional info to stdout.
33 quiet: Do not print messages to stdout.
34 ratelimit: Download speed limit, in bytes/sec.
35 retries: Number of times to retry for HTTP error 5xx
36 buffersize: Size of download buffer in bytes.
37 noresizebuffer: Do not automatically resize the download buffer.
38 continuedl: Try to continue downloads if possible.
39 noprogress: Do not print the progress bar.
40 logtostderr: Log messages to stderr instead of stdout.
41 consoletitle: Display progress in console window's titlebar.
42 nopart: Do not use temporary .part files.
43 updatetime: Use the Last-modified header to set output file timestamps.
44 test: Download only first bytes to test the downloader.
45 min_filesize: Skip files smaller than this size
46 max_filesize: Skip files larger than this size
47 xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
49 external_downloader_args: A list of additional command-line arguments for the
51 hls_use_mpegts: Use the mpegts container for HLS videos.
52 http_chunk_size: Size of a chunk for chunk-based HTTP downloading. May be
53 useful for bypassing bandwidth throttling imposed by
54 a webserver (experimental)
56 Subclasses of this one must re-define the real_download method.
59 _TEST_FILE_SIZE = 10241
62 def __init__(self, ydl, params):
63 """Create a FileDownloader object with the given options."""
65 self._progress_hooks = []
67 self.add_progress_hook(self.report_progress)
70 def format_seconds(seconds):
71 (mins, secs) = divmod(seconds, 60)
72 (hours, mins) = divmod(mins, 60)
76 return '%02d:%02d' % (mins, secs)
78 return '%02d:%02d:%02d' % (hours, mins, secs)
81 def calc_percent(byte_counter, data_len):
84 return float(byte_counter) / float(data_len) * 100.0
87 def format_percent(percent):
90 return '%6s' % ('%3.1f%%' % percent)
93 def calc_eta(start, now, total, current):
99 if current == 0 or dif < 0.001: # One millisecond
101 rate = float(current) / dif
102 return int((float(total) - float(current)) / rate)
108 return FileDownloader.format_seconds(eta)
111 def calc_speed(start, now, bytes):
113 if bytes == 0 or dif < 0.001: # One millisecond
115 return float(bytes) / dif
118 def format_speed(speed):
120 return '%10s' % '---b/s'
121 return '%10s' % ('%s/s' % format_bytes(speed))
124 def format_retries(retries):
125 return 'inf' if retries == float('inf') else '%.0f' % retries
128 def best_block_size(elapsed_time, bytes):
129 new_min = max(bytes / 2.0, 1.0)
130 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
131 if elapsed_time < 0.001:
133 rate = bytes / elapsed_time
141 def parse_bytes(bytestr):
142 """Parse a string indicating a byte quantity into an integer."""
143 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
146 number = float(matchobj.group(1))
147 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
148 return int(round(number * multiplier))
150 def to_screen(self, *args, **kargs):
151 self.ydl.to_screen(*args, **kargs)
153 def to_stderr(self, message):
154 self.ydl.to_screen(message)
156 def to_console_title(self, message):
157 self.ydl.to_console_title(message)
159 def trouble(self, *args, **kargs):
160 self.ydl.trouble(*args, **kargs)
162 def report_warning(self, *args, **kargs):
163 self.ydl.report_warning(*args, **kargs)
165 def report_error(self, *args, **kargs):
166 self.ydl.report_error(*args, **kargs)
168 def slow_down(self, start_time, now, byte_counter):
169 """Sleep if the download speed is over the rate limit."""
170 rate_limit = self.params.get('ratelimit')
171 if rate_limit is None or byte_counter == 0:
175 elapsed = now - start_time
178 speed = float(byte_counter) / elapsed
179 if speed > rate_limit:
180 time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
182 def temp_name(self, filename):
183 """Returns a temporary filename for the given filename."""
184 if self.params.get('nopart', False) or filename == '-' or \
185 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
187 return filename + '.part'
189 def undo_temp_name(self, filename):
190 if filename.endswith('.part'):
191 return filename[:-len('.part')]
194 def ytdl_filename(self, filename):
195 return filename + '.ytdl'
197 def try_rename(self, old_filename, new_filename):
199 if old_filename == new_filename:
201 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
202 except (IOError, OSError) as err:
203 self.report_error('unable to rename file: %s' % error_to_compat_str(err))
205 def try_utime(self, filename, last_modified_hdr):
206 """Try to set the last-modified time of the given file."""
207 if last_modified_hdr is None:
209 if not os.path.isfile(encodeFilename(filename)):
211 timestr = last_modified_hdr
214 filetime = timeconvert(timestr)
217 # Ignore obviously invalid dates
221 os.utime(filename, (time.time(), filetime))
226 def report_destination(self, filename):
227 """Report destination filename."""
228 self.to_screen('[download] Destination: ' + filename)
230 def _report_progress_status(self, msg, is_last_line=False):
231 fullmsg = '[download] ' + msg
232 if self.params.get('progress_with_newline', False):
233 self.to_screen(fullmsg)
235 if compat_os_name == 'nt':
236 prev_len = getattr(self, '_report_progress_prev_line_length',
238 if prev_len > len(fullmsg):
239 fullmsg += ' ' * (prev_len - len(fullmsg))
240 self._report_progress_prev_line_length = len(fullmsg)
243 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
244 self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
245 self.to_console_title('youtube-dl ' + msg)
247 def report_progress(self, s):
248 if s['status'] == 'finished':
249 if self.params.get('noprogress', False):
250 self.to_screen('[download] Download completed')
252 if s.get('total_bytes') is not None:
253 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
254 msg_template = '100%% of %(_total_bytes_str)s'
256 msg_template = 'Completed'
257 if s.get('elapsed') is not None:
258 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
259 msg_template += ' in %(_elapsed_str)s'
260 self._report_progress_status(
261 msg_template % s, is_last_line=True)
263 if self.params.get('noprogress'):
266 if s['status'] != 'downloading':
269 if s.get('eta') is not None:
270 s['_eta_str'] = self.format_eta(s['eta'])
272 s['_eta_str'] = 'Unknown ETA'
274 if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
275 s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
276 elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
277 s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
279 if s.get('downloaded_bytes') == 0:
280 s['_percent_str'] = self.format_percent(0)
282 s['_percent_str'] = 'Unknown %'
284 if s.get('speed') is not None:
285 s['_speed_str'] = self.format_speed(s['speed'])
287 s['_speed_str'] = 'Unknown speed'
289 if s.get('total_bytes') is not None:
290 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
291 msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
292 elif s.get('total_bytes_estimate') is not None:
293 s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
294 msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
296 if s.get('downloaded_bytes') is not None:
297 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
299 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
300 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
302 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
304 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
306 self._report_progress_status(msg_template % s)
308 def report_resuming_byte(self, resume_len):
309 """Report attempt to resume at given byte."""
310 self.to_screen('[download] Resuming download at byte %s' % resume_len)
312 def report_retry(self, err, count, retries):
313 """Report retry in case of HTTP error 5xx"""
315 '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...'
316 % (error_to_compat_str(err), count, self.format_retries(retries)))
318 def report_file_already_downloaded(self, file_name):
319 """Report file has already been fully downloaded."""
321 self.to_screen('[download] %s has already been downloaded' % file_name)
322 except UnicodeEncodeError:
323 self.to_screen('[download] The file has already been downloaded')
325 def report_unable_to_resume(self):
326 """Report it was impossible to resume download."""
327 self.to_screen('[download] Unable to resume')
329 def download(self, filename, info_dict):
330 """Download to a filename using the info from info_dict
331 Return True on success and False otherwise
334 nooverwrites_and_exists = (
335 self.params.get('nooverwrites', False) and
336 os.path.exists(encodeFilename(filename))
339 if not hasattr(filename, 'write'):
340 continuedl_and_exists = (
341 self.params.get('continuedl', True) and
342 os.path.isfile(encodeFilename(filename)) and
343 not self.params.get('nopart', False)
346 # Check file already present
347 if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
348 self.report_file_already_downloaded(filename)
349 self._hook_progress({
350 'filename': filename,
351 'status': 'finished',
352 'total_bytes': os.path.getsize(encodeFilename(filename)),
356 min_sleep_interval = self.params.get('sleep_interval')
357 if min_sleep_interval:
358 max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
359 sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
361 '[download] Sleeping %s seconds...' % (
362 int(sleep_interval) if sleep_interval.is_integer()
363 else '%.2f' % sleep_interval))
364 time.sleep(sleep_interval)
366 return self.real_download(filename, info_dict)
368 def real_download(self, filename, info_dict):
369 """Real download process. Redefine in subclasses."""
370 raise NotImplementedError('This method must be implemented by subclasses')
372 def _hook_progress(self, status):
373 for ph in self._progress_hooks:
376 def add_progress_hook(self, ph):
377 # See YoutubeDl.py (search for progress_hooks) for a description of
379 self._progress_hooks.append(ph)
381 def _debug_cmd(self, args, exe=None):
382 if not self.params.get('verbose', False):
385 str_args = [decodeArgument(a) for a in args]
388 exe = os.path.basename(str_args[0])
390 self.to_screen('[debug] %s command line: %s' % (
391 exe, shell_quote(str_args)))