1 from __future__ import division, unicode_literals
8 from ..compat import compat_os_name
18 class FileDownloader(object):
19 """File Downloader class.
21 File downloader objects are the ones responsible of downloading the
22 actual video file and writing it to disk.
24 File downloaders accept a lot of parameters. In order not to saturate
25 the object constructor with arguments, it receives a dictionary of
30 verbose: Print additional info to stdout.
31 quiet: Do not print messages to stdout.
32 ratelimit: Download speed limit, in bytes/sec.
33 retries: Number of times to retry for HTTP error 5xx
34 buffersize: Size of download buffer in bytes.
35 noresizebuffer: Do not automatically resize the download buffer.
36 continuedl: Try to continue downloads if possible.
37 noprogress: Do not print the progress bar.
38 logtostderr: Log messages to stderr instead of stdout.
39 consoletitle: Display progress in console window's titlebar.
40 nopart: Do not use temporary .part files.
41 updatetime: Use the Last-modified header to set output file timestamps.
42 test: Download only first bytes to test the downloader.
43 min_filesize: Skip files smaller than this size
44 max_filesize: Skip files larger than this size
45 xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
47 external_downloader_args: A list of additional command-line arguments for the
49 hls_use_mpegts: Use the mpegts container for HLS videos.
51 Subclasses of this one must re-define the real_download method.
54 _TEST_FILE_SIZE = 10241
57 def __init__(self, ydl, params):
58 """Create a FileDownloader object with the given options."""
60 self._progress_hooks = []
62 self.add_progress_hook(self.report_progress)
65 def format_seconds(seconds):
66 (mins, secs) = divmod(seconds, 60)
67 (hours, mins) = divmod(mins, 60)
71 return '%02d:%02d' % (mins, secs)
73 return '%02d:%02d:%02d' % (hours, mins, secs)
76 def calc_percent(byte_counter, data_len):
79 return float(byte_counter) / float(data_len) * 100.0
82 def format_percent(percent):
85 return '%6s' % ('%3.1f%%' % percent)
88 def calc_eta(start, now, total, current):
94 if current == 0 or dif < 0.001: # One millisecond
96 rate = float(current) / dif
97 return int((float(total) - float(current)) / rate)
103 return FileDownloader.format_seconds(eta)
106 def calc_speed(start, now, bytes):
108 if bytes == 0 or dif < 0.001: # One millisecond
110 return float(bytes) / dif
113 def format_speed(speed):
115 return '%10s' % '---b/s'
116 return '%10s' % ('%s/s' % format_bytes(speed))
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:
124 rate = bytes / elapsed_time
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)
137 number = float(matchobj.group(1))
138 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
139 return int(round(number * multiplier))
141 def to_screen(self, *args, **kargs):
142 self.ydl.to_screen(*args, **kargs)
144 def to_stderr(self, message):
145 self.ydl.to_screen(message)
147 def to_console_title(self, message):
148 self.ydl.to_console_title(message)
150 def trouble(self, *args, **kargs):
151 self.ydl.trouble(*args, **kargs)
153 def report_warning(self, *args, **kargs):
154 self.ydl.report_warning(*args, **kargs)
156 def report_error(self, *args, **kargs):
157 self.ydl.report_error(*args, **kargs)
159 def slow_down(self, start_time, now, byte_counter):
160 """Sleep if the download speed is over the rate limit."""
161 rate_limit = self.params.get('ratelimit')
162 if rate_limit is None or byte_counter == 0:
166 elapsed = now - start_time
169 speed = float(byte_counter) / elapsed
170 if speed > rate_limit:
171 time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
173 def temp_name(self, filename):
174 """Returns a temporary filename for the given filename."""
175 if self.params.get('nopart', False) or filename == '-' or \
176 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
178 return filename + '.part'
180 def undo_temp_name(self, filename):
181 if filename.endswith('.part'):
182 return filename[:-len('.part')]
185 def try_rename(self, old_filename, new_filename):
187 if old_filename == new_filename:
189 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
190 except (IOError, OSError) as err:
191 self.report_error('unable to rename file: %s' % error_to_compat_str(err))
193 def try_utime(self, filename, last_modified_hdr):
194 """Try to set the last-modified time of the given file."""
195 if last_modified_hdr is None:
197 if not os.path.isfile(encodeFilename(filename)):
199 timestr = last_modified_hdr
202 filetime = timeconvert(timestr)
205 # Ignore obviously invalid dates
209 os.utime(filename, (time.time(), filetime))
214 def report_destination(self, filename):
215 """Report destination filename."""
216 self.to_screen('[download] Destination: ' + filename)
218 def _report_progress_status(self, msg, is_last_line=False):
219 fullmsg = '[download] ' + msg
220 if self.params.get('progress_with_newline', False):
221 self.to_screen(fullmsg)
223 if compat_os_name == 'nt':
224 prev_len = getattr(self, '_report_progress_prev_line_length',
226 if prev_len > len(fullmsg):
227 fullmsg += ' ' * (prev_len - len(fullmsg))
228 self._report_progress_prev_line_length = len(fullmsg)
231 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
232 self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
233 self.to_console_title('youtube-dl ' + msg)
235 def report_progress(self, s):
236 if s['status'] == 'finished':
237 if self.params.get('noprogress', False):
238 self.to_screen('[download] Download completed')
240 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
241 if s.get('elapsed') is not None:
242 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
243 msg_template = '100%% of %(_total_bytes_str)s in %(_elapsed_str)s'
245 msg_template = '100%% of %(_total_bytes_str)s'
246 self._report_progress_status(
247 msg_template % s, is_last_line=True)
249 if self.params.get('noprogress'):
252 if s['status'] != 'downloading':
255 if s.get('eta') is not None:
256 s['_eta_str'] = self.format_eta(s['eta'])
258 s['_eta_str'] = 'Unknown ETA'
260 if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
261 s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
262 elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
263 s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
265 if s.get('downloaded_bytes') == 0:
266 s['_percent_str'] = self.format_percent(0)
268 s['_percent_str'] = 'Unknown %'
270 if s.get('speed') is not None:
271 s['_speed_str'] = self.format_speed(s['speed'])
273 s['_speed_str'] = 'Unknown speed'
275 if s.get('total_bytes') is not None:
276 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
277 msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
278 elif s.get('total_bytes_estimate') is not None:
279 s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
280 msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
282 if s.get('downloaded_bytes') is not None:
283 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
285 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
286 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
288 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
290 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
292 self._report_progress_status(msg_template % s)
294 def report_resuming_byte(self, resume_len):
295 """Report attempt to resume at given byte."""
296 self.to_screen('[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('[download] Got server HTTP error. Retrying (attempt %d of %.0f)...' % (count, retries))
302 def report_file_already_downloaded(self, file_name):
303 """Report file has already been fully downloaded."""
305 self.to_screen('[download] %s has already been downloaded' % file_name)
306 except UnicodeEncodeError:
307 self.to_screen('[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('[download] Unable to resume')
313 def download(self, filename, info_dict):
314 """Download to a filename using the info from info_dict
315 Return True on success and False otherwise
318 nooverwrites_and_exists = (
319 self.params.get('nooverwrites', False) and
320 os.path.exists(encodeFilename(filename))
323 continuedl_and_exists = (
324 self.params.get('continuedl', True) and
325 os.path.isfile(encodeFilename(filename)) and
326 not self.params.get('nopart', False)
329 # Check file already present
330 if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
331 self.report_file_already_downloaded(filename)
332 self._hook_progress({
333 'filename': filename,
334 'status': 'finished',
335 'total_bytes': os.path.getsize(encodeFilename(filename)),
339 sleep_interval = self.params.get('sleep_interval')
341 self.to_screen('[download] Sleeping %s seconds...' % sleep_interval)
342 time.sleep(sleep_interval)
344 return self.real_download(filename, info_dict)
346 def real_download(self, filename, info_dict):
347 """Real download process. Redefine in subclasses."""
348 raise NotImplementedError('This method must be implemented by subclasses')
350 def _hook_progress(self, status):
351 for ph in self._progress_hooks:
354 def add_progress_hook(self, ph):
355 # See YoutubeDl.py (search for progress_hooks) for a description of
357 self._progress_hooks.append(ph)
359 def _debug_cmd(self, args, exe=None):
360 if not self.params.get('verbose', False):
363 str_args = [decodeArgument(a) for a in args]
366 exe = os.path.basename(str_args[0])
370 shell_quote = lambda args: ' '.join(map(pipes.quote, str_args))
373 self.to_screen('[debug] %s command line: %s' % (
374 exe, shell_quote(str_args)))