]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/downloader/common.py
debian/control: Update list of supported sites.
[youtubedl] / youtube_dl / downloader / common.py
1 from __future__ import division, unicode_literals
2
3 import os
4 import re
5 import sys
6 import time
7 import random
8
9 from ..compat import compat_os_name
10 from ..utils import (
11 decodeArgument,
12 encodeFilename,
13 error_to_compat_str,
14 format_bytes,
15 shell_quote,
16 timeconvert,
17 )
18
19
20 class FileDownloader(object):
21 """File Downloader class.
22
23 File downloader objects are the ones responsible of downloading the
24 actual video file and writing it to disk.
25
26 File downloaders accept a lot of parameters. In order not to saturate
27 the object constructor with arguments, it receives a dictionary of
28 options instead.
29
30 Available options:
31
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.
48 external_downloader_args: A list of additional command-line arguments for the
49 external downloader.
50 hls_use_mpegts: Use the mpegts container for HLS videos.
51 http_chunk_size: Size of a chunk for chunk-based HTTP downloading. May be
52 useful for bypassing bandwidth throttling imposed by
53 a webserver (experimental)
54
55 Subclasses of this one must re-define the real_download method.
56 """
57
58 _TEST_FILE_SIZE = 10241
59 params = None
60
61 def __init__(self, ydl, params):
62 """Create a FileDownloader object with the given options."""
63 self.ydl = ydl
64 self._progress_hooks = []
65 self.params = params
66 self.add_progress_hook(self.report_progress)
67
68 @staticmethod
69 def format_seconds(seconds):
70 (mins, secs) = divmod(seconds, 60)
71 (hours, mins) = divmod(mins, 60)
72 if hours > 99:
73 return '--:--:--'
74 if hours == 0:
75 return '%02d:%02d' % (mins, secs)
76 else:
77 return '%02d:%02d:%02d' % (hours, mins, secs)
78
79 @staticmethod
80 def calc_percent(byte_counter, data_len):
81 if data_len is None:
82 return None
83 return float(byte_counter) / float(data_len) * 100.0
84
85 @staticmethod
86 def format_percent(percent):
87 if percent is None:
88 return '---.-%'
89 return '%6s' % ('%3.1f%%' % percent)
90
91 @staticmethod
92 def calc_eta(start, now, total, current):
93 if total is None:
94 return None
95 if now is None:
96 now = time.time()
97 dif = now - start
98 if current == 0 or dif < 0.001: # One millisecond
99 return None
100 rate = float(current) / dif
101 return int((float(total) - float(current)) / rate)
102
103 @staticmethod
104 def format_eta(eta):
105 if eta is None:
106 return '--:--'
107 return FileDownloader.format_seconds(eta)
108
109 @staticmethod
110 def calc_speed(start, now, bytes):
111 dif = now - start
112 if bytes == 0 or dif < 0.001: # One millisecond
113 return None
114 return float(bytes) / dif
115
116 @staticmethod
117 def format_speed(speed):
118 if speed is None:
119 return '%10s' % '---b/s'
120 return '%10s' % ('%s/s' % format_bytes(speed))
121
122 @staticmethod
123 def format_retries(retries):
124 return 'inf' if retries == float('inf') else '%.0f' % retries
125
126 @staticmethod
127 def best_block_size(elapsed_time, bytes):
128 new_min = max(bytes / 2.0, 1.0)
129 new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
130 if elapsed_time < 0.001:
131 return int(new_max)
132 rate = bytes / elapsed_time
133 if rate > new_max:
134 return int(new_max)
135 if rate < new_min:
136 return int(new_min)
137 return int(rate)
138
139 @staticmethod
140 def parse_bytes(bytestr):
141 """Parse a string indicating a byte quantity into an integer."""
142 matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
143 if matchobj is None:
144 return None
145 number = float(matchobj.group(1))
146 multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
147 return int(round(number * multiplier))
148
149 def to_screen(self, *args, **kargs):
150 self.ydl.to_screen(*args, **kargs)
151
152 def to_stderr(self, message):
153 self.ydl.to_screen(message)
154
155 def to_console_title(self, message):
156 self.ydl.to_console_title(message)
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, now, byte_counter):
168 """Sleep if the download speed is over the rate limit."""
169 rate_limit = self.params.get('ratelimit')
170 if rate_limit is None or byte_counter == 0:
171 return
172 if now is None:
173 now = time.time()
174 elapsed = now - start_time
175 if elapsed <= 0.0:
176 return
177 speed = float(byte_counter) / elapsed
178 if speed > rate_limit:
179 sleep_time = float(byte_counter) / rate_limit - elapsed
180 if sleep_time > 0:
181 time.sleep(sleep_time)
182
183 def temp_name(self, filename):
184 """Returns a temporary filename for the given filename."""
185 if self.params.get('nopart', False) or filename == '-' or \
186 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
187 return filename
188 return filename + '.part'
189
190 def undo_temp_name(self, filename):
191 if filename.endswith('.part'):
192 return filename[:-len('.part')]
193 return filename
194
195 def ytdl_filename(self, filename):
196 return filename + '.ytdl'
197
198 def try_rename(self, old_filename, new_filename):
199 try:
200 if old_filename == new_filename:
201 return
202 os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
203 except (IOError, OSError) as err:
204 self.report_error('unable to rename file: %s' % error_to_compat_str(err))
205
206 def try_utime(self, filename, last_modified_hdr):
207 """Try to set the last-modified time of the given file."""
208 if last_modified_hdr is None:
209 return
210 if not os.path.isfile(encodeFilename(filename)):
211 return
212 timestr = last_modified_hdr
213 if timestr is None:
214 return
215 filetime = timeconvert(timestr)
216 if filetime is None:
217 return filetime
218 # Ignore obviously invalid dates
219 if filetime == 0:
220 return
221 try:
222 os.utime(filename, (time.time(), filetime))
223 except Exception:
224 pass
225 return filetime
226
227 def report_destination(self, filename):
228 """Report destination filename."""
229 self.to_screen('[download] Destination: ' + filename)
230
231 def _report_progress_status(self, msg, is_last_line=False):
232 fullmsg = '[download] ' + msg
233 if self.params.get('progress_with_newline', False):
234 self.to_screen(fullmsg)
235 else:
236 if compat_os_name == 'nt':
237 prev_len = getattr(self, '_report_progress_prev_line_length',
238 0)
239 if prev_len > len(fullmsg):
240 fullmsg += ' ' * (prev_len - len(fullmsg))
241 self._report_progress_prev_line_length = len(fullmsg)
242 clear_line = '\r'
243 else:
244 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
245 self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
246 self.to_console_title('youtube-dl ' + msg)
247
248 def report_progress(self, s):
249 if s['status'] == 'finished':
250 if self.params.get('noprogress', False):
251 self.to_screen('[download] Download completed')
252 else:
253 msg_template = '100%%'
254 if s.get('total_bytes') is not None:
255 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
256 msg_template += ' of %(_total_bytes_str)s'
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)
262
263 if self.params.get('noprogress'):
264 return
265
266 if s['status'] != 'downloading':
267 return
268
269 if s.get('eta') is not None:
270 s['_eta_str'] = self.format_eta(s['eta'])
271 else:
272 s['_eta_str'] = 'Unknown ETA'
273
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'])
278 else:
279 if s.get('downloaded_bytes') == 0:
280 s['_percent_str'] = self.format_percent(0)
281 else:
282 s['_percent_str'] = 'Unknown %'
283
284 if s.get('speed') is not None:
285 s['_speed_str'] = self.format_speed(s['speed'])
286 else:
287 s['_speed_str'] = 'Unknown speed'
288
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'
295 else:
296 if s.get('downloaded_bytes') is not None:
297 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
298 if s.get('elapsed'):
299 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
300 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
301 else:
302 msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
303 else:
304 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
305
306 self._report_progress_status(msg_template % s)
307
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)
311
312 def report_retry(self, err, count, retries):
313 """Report retry in case of HTTP error 5xx"""
314 self.to_screen(
315 '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...'
316 % (error_to_compat_str(err), count, self.format_retries(retries)))
317
318 def report_file_already_downloaded(self, file_name):
319 """Report file has already been fully downloaded."""
320 try:
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')
324
325 def report_unable_to_resume(self):
326 """Report it was impossible to resume download."""
327 self.to_screen('[download] Unable to resume')
328
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
332 """
333
334 nooverwrites_and_exists = (
335 self.params.get('nooverwrites', False)
336 and os.path.exists(encodeFilename(filename))
337 )
338
339 if not hasattr(filename, 'write'):
340 continuedl_and_exists = (
341 self.params.get('continuedl', True)
342 and os.path.isfile(encodeFilename(filename))
343 and not self.params.get('nopart', False)
344 )
345
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)),
353 })
354 return True
355
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)
360 self.to_screen(
361 '[download] Sleeping %s seconds...' % (
362 int(sleep_interval) if sleep_interval.is_integer()
363 else '%.2f' % sleep_interval))
364 time.sleep(sleep_interval)
365
366 return self.real_download(filename, info_dict)
367
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')
371
372 def _hook_progress(self, status):
373 for ph in self._progress_hooks:
374 ph(status)
375
376 def add_progress_hook(self, ph):
377 # See YoutubeDl.py (search for progress_hooks) for a description of
378 # this interface
379 self._progress_hooks.append(ph)
380
381 def _debug_cmd(self, args, exe=None):
382 if not self.params.get('verbose', False):
383 return
384
385 str_args = [decodeArgument(a) for a in args]
386
387 if exe is None:
388 exe = os.path.basename(str_args[0])
389
390 self.to_screen('[debug] %s command line: %s' % (
391 exe, shell_quote(str_args)))