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.

317 lines
11 KiB

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