You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

374 lines
14 KiB

10 years ago
10 years ago
10 years ago
10 years ago
10 years ago
11 years ago
11 years ago
  1. from __future__ import division, unicode_literals
  2. import os
  3. import re
  4. import sys
  5. import time
  6. from ..compat import compat_os_name
  7. from ..utils import (
  8. encodeFilename,
  9. error_to_compat_str,
  10. decodeArgument,
  11. format_bytes,
  12. timeconvert,
  13. )
  14. class FileDownloader(object):
  15. """File Downloader class.
  16. File downloader objects are the ones responsible of downloading the
  17. actual video file and writing it to disk.
  18. File downloaders accept a lot of parameters. In order not to saturate
  19. the object constructor with arguments, it receives a dictionary of
  20. options instead.
  21. Available options:
  22. verbose: Print additional info to stdout.
  23. quiet: Do not print messages to stdout.
  24. ratelimit: Download speed limit, in bytes/sec.
  25. retries: Number of times to retry for HTTP error 5xx
  26. buffersize: Size of download buffer in bytes.
  27. noresizebuffer: Do not automatically resize the download buffer.
  28. continuedl: Try to continue downloads if possible.
  29. noprogress: Do not print the progress bar.
  30. logtostderr: Log messages to stderr instead of stdout.
  31. consoletitle: Display progress in console window's titlebar.
  32. nopart: Do not use temporary .part files.
  33. updatetime: Use the Last-modified header to set output file timestamps.
  34. test: Download only first bytes to test the downloader.
  35. min_filesize: Skip files smaller than this size
  36. max_filesize: Skip files larger than this size
  37. xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  38. (experimental)
  39. external_downloader_args: A list of additional command-line arguments for the
  40. external downloader.
  41. hls_use_mpegts: Use the mpegts container for HLS videos.
  42. Subclasses of this one must re-define the real_download method.
  43. """
  44. _TEST_FILE_SIZE = 10241
  45. params = None
  46. def __init__(self, ydl, params):
  47. """Create a FileDownloader object with the given options."""
  48. self.ydl = ydl
  49. self._progress_hooks = []
  50. self.params = params
  51. self.add_progress_hook(self.report_progress)
  52. @staticmethod
  53. def format_seconds(seconds):
  54. (mins, secs) = divmod(seconds, 60)
  55. (hours, mins) = divmod(mins, 60)
  56. if hours > 99:
  57. return '--:--:--'
  58. if hours == 0:
  59. return '%02d:%02d' % (mins, secs)
  60. else:
  61. return '%02d:%02d:%02d' % (hours, mins, secs)
  62. @staticmethod
  63. def calc_percent(byte_counter, data_len):
  64. if data_len is None:
  65. return None
  66. return float(byte_counter) / float(data_len) * 100.0
  67. @staticmethod
  68. def format_percent(percent):
  69. if percent is None:
  70. return '---.-%'
  71. return '%6s' % ('%3.1f%%' % percent)
  72. @staticmethod
  73. def calc_eta(start, now, total, current):
  74. if total is None:
  75. return None
  76. if now is None:
  77. now = time.time()
  78. dif = now - start
  79. if current == 0 or dif < 0.001: # One millisecond
  80. return None
  81. rate = float(current) / dif
  82. return int((float(total) - float(current)) / rate)
  83. @staticmethod
  84. def format_eta(eta):
  85. if eta is None:
  86. return '--:--'
  87. return FileDownloader.format_seconds(eta)
  88. @staticmethod
  89. def calc_speed(start, now, bytes):
  90. dif = now - start
  91. if bytes == 0 or dif < 0.001: # One millisecond
  92. return None
  93. return float(bytes) / dif
  94. @staticmethod
  95. def format_speed(speed):
  96. if speed is None:
  97. return '%10s' % '---b/s'
  98. return '%10s' % ('%s/s' % format_bytes(speed))
  99. @staticmethod
  100. def best_block_size(elapsed_time, bytes):
  101. new_min = max(bytes / 2.0, 1.0)
  102. new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
  103. if elapsed_time < 0.001:
  104. return int(new_max)
  105. rate = bytes / elapsed_time
  106. if rate > new_max:
  107. return int(new_max)
  108. if rate < new_min:
  109. return int(new_min)
  110. return int(rate)
  111. @staticmethod
  112. def parse_bytes(bytestr):
  113. """Parse a string indicating a byte quantity into an integer."""
  114. matchobj = re.match(r'(?i)^(\d+(?:\.\d+)?)([kMGTPEZY]?)$', bytestr)
  115. if matchobj is None:
  116. return None
  117. number = float(matchobj.group(1))
  118. multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
  119. return int(round(number * multiplier))
  120. def to_screen(self, *args, **kargs):
  121. self.ydl.to_screen(*args, **kargs)
  122. def to_stderr(self, message):
  123. self.ydl.to_screen(message)
  124. def to_console_title(self, message):
  125. self.ydl.to_console_title(message)
  126. def trouble(self, *args, **kargs):
  127. self.ydl.trouble(*args, **kargs)
  128. def report_warning(self, *args, **kargs):
  129. self.ydl.report_warning(*args, **kargs)
  130. def report_error(self, *args, **kargs):
  131. self.ydl.report_error(*args, **kargs)
  132. def slow_down(self, start_time, now, byte_counter):
  133. """Sleep if the download speed is over the rate limit."""
  134. rate_limit = self.params.get('ratelimit')
  135. if rate_limit is None or byte_counter == 0:
  136. return
  137. if now is None:
  138. now = time.time()
  139. elapsed = now - start_time
  140. if elapsed <= 0.0:
  141. return
  142. speed = float(byte_counter) / elapsed
  143. if speed > rate_limit:
  144. time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
  145. def temp_name(self, filename):
  146. """Returns a temporary filename for the given filename."""
  147. if self.params.get('nopart', False) or filename == '-' or \
  148. (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
  149. return filename
  150. return filename + '.part'
  151. def undo_temp_name(self, filename):
  152. if filename.endswith('.part'):
  153. return filename[:-len('.part')]
  154. return filename
  155. def try_rename(self, old_filename, new_filename):
  156. try:
  157. if old_filename == new_filename:
  158. return
  159. os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
  160. except (IOError, OSError) as err:
  161. self.report_error('unable to rename file: %s' % error_to_compat_str(err))
  162. def try_utime(self, filename, last_modified_hdr):
  163. """Try to set the last-modified time of the given file."""
  164. if last_modified_hdr is None:
  165. return
  166. if not os.path.isfile(encodeFilename(filename)):
  167. return
  168. timestr = last_modified_hdr
  169. if timestr is None:
  170. return
  171. filetime = timeconvert(timestr)
  172. if filetime is None:
  173. return filetime
  174. # Ignore obviously invalid dates
  175. if filetime == 0:
  176. return
  177. try:
  178. os.utime(filename, (time.time(), filetime))
  179. except Exception:
  180. pass
  181. return filetime
  182. def report_destination(self, filename):
  183. """Report destination filename."""
  184. self.to_screen('[download] Destination: ' + filename)
  185. def _report_progress_status(self, msg, is_last_line=False):
  186. fullmsg = '[download] ' + msg
  187. if self.params.get('progress_with_newline', False):
  188. self.to_screen(fullmsg)
  189. else:
  190. if compat_os_name == 'nt':
  191. prev_len = getattr(self, '_report_progress_prev_line_length',
  192. 0)
  193. if prev_len > len(fullmsg):
  194. fullmsg += ' ' * (prev_len - len(fullmsg))
  195. self._report_progress_prev_line_length = len(fullmsg)
  196. clear_line = '\r'
  197. else:
  198. clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
  199. self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
  200. self.to_console_title('youtube-dl ' + msg)
  201. def report_progress(self, s):
  202. if s['status'] == 'finished':
  203. if self.params.get('noprogress', False):
  204. self.to_screen('[download] Download completed')
  205. else:
  206. s['_total_bytes_str'] = format_bytes(s['total_bytes'])
  207. if s.get('elapsed') is not None:
  208. s['_elapsed_str'] = self.format_seconds(s['elapsed'])
  209. msg_template = '100%% of %(_total_bytes_str)s in %(_elapsed_str)s'
  210. else:
  211. msg_template = '100%% of %(_total_bytes_str)s'
  212. self._report_progress_status(
  213. msg_template % s, is_last_line=True)
  214. if self.params.get('noprogress'):
  215. return
  216. if s['status'] != 'downloading':
  217. return
  218. if s.get('eta') is not None:
  219. s['_eta_str'] = self.format_eta(s['eta'])
  220. else:
  221. s['_eta_str'] = 'Unknown ETA'
  222. if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
  223. s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
  224. elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
  225. s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
  226. else:
  227. if s.get('downloaded_bytes') == 0:
  228. s['_percent_str'] = self.format_percent(0)
  229. else:
  230. s['_percent_str'] = 'Unknown %'
  231. if s.get('speed') is not None:
  232. s['_speed_str'] = self.format_speed(s['speed'])
  233. else:
  234. s['_speed_str'] = 'Unknown speed'
  235. if s.get('total_bytes') is not None:
  236. s['_total_bytes_str'] = format_bytes(s['total_bytes'])
  237. msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
  238. elif s.get('total_bytes_estimate') is not None:
  239. s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
  240. msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
  241. else:
  242. if s.get('downloaded_bytes') is not None:
  243. s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
  244. if s.get('elapsed'):
  245. s['_elapsed_str'] = self.format_seconds(s['elapsed'])
  246. msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
  247. else:
  248. msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
  249. else:
  250. msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
  251. self._report_progress_status(msg_template % s)
  252. def report_resuming_byte(self, resume_len):
  253. """Report attempt to resume at given byte."""
  254. self.to_screen('[download] Resuming download at byte %s' % resume_len)
  255. def report_retry(self, count, retries):
  256. """Report retry in case of HTTP error 5xx"""
  257. self.to_screen('[download] Got server HTTP error. Retrying (attempt %d of %.0f)...' % (count, retries))
  258. def report_file_already_downloaded(self, file_name):
  259. """Report file has already been fully downloaded."""
  260. try:
  261. self.to_screen('[download] %s has already been downloaded' % file_name)
  262. except UnicodeEncodeError:
  263. self.to_screen('[download] The file has already been downloaded')
  264. def report_unable_to_resume(self):
  265. """Report it was impossible to resume download."""
  266. self.to_screen('[download] Unable to resume')
  267. def download(self, filename, info_dict):
  268. """Download to a filename using the info from info_dict
  269. Return True on success and False otherwise
  270. """
  271. nooverwrites_and_exists = (
  272. self.params.get('nooverwrites', False) and
  273. os.path.exists(encodeFilename(filename))
  274. )
  275. continuedl_and_exists = (
  276. self.params.get('continuedl', True) and
  277. os.path.isfile(encodeFilename(filename)) and
  278. not self.params.get('nopart', False)
  279. )
  280. # Check file already present
  281. if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
  282. self.report_file_already_downloaded(filename)
  283. self._hook_progress({
  284. 'filename': filename,
  285. 'status': 'finished',
  286. 'total_bytes': os.path.getsize(encodeFilename(filename)),
  287. })
  288. return True
  289. sleep_interval = self.params.get('sleep_interval')
  290. if sleep_interval:
  291. self.to_screen('[download] Sleeping %s seconds...' % sleep_interval)
  292. time.sleep(sleep_interval)
  293. return self.real_download(filename, info_dict)
  294. def real_download(self, filename, info_dict):
  295. """Real download process. Redefine in subclasses."""
  296. raise NotImplementedError('This method must be implemented by subclasses')
  297. def _hook_progress(self, status):
  298. for ph in self._progress_hooks:
  299. ph(status)
  300. def add_progress_hook(self, ph):
  301. # See YoutubeDl.py (search for progress_hooks) for a description of
  302. # this interface
  303. self._progress_hooks.append(ph)
  304. def _debug_cmd(self, args, exe=None):
  305. if not self.params.get('verbose', False):
  306. return
  307. str_args = [decodeArgument(a) for a in args]
  308. if exe is None:
  309. exe = os.path.basename(str_args[0])
  310. try:
  311. import pipes
  312. shell_quote = lambda args: ' '.join(map(pipes.quote, str_args))
  313. except ImportError:
  314. shell_quote = repr
  315. self.to_screen('[debug] %s command line: %s' % (
  316. exe, shell_quote(str_args)))