Raphaël G. Git Repositories - youtubedl/blob - 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                         (experimental)
  49     external_downloader_args:  A list of additional command-line arguments for the
  50                         external downloader.
  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)
  55
  56     Subclasses of this one must re-define the real_download method.
  57     """
  58
  59     _TEST_FILE_SIZE = 10241
  60     params = None
  61
  62     def __init__(self, ydl, params):
  63         """Create a FileDownloader object with the given options."""
  64         self.ydl = ydl
  65         self._progress_hooks = []
  66         self.params = params
  67         self.add_progress_hook(self.report_progress)
  68
  69     @staticmethod
  70     def format_seconds(seconds):
  71         (mins, secs) = divmod(seconds, 60)
  72         (hours, mins) = divmod(mins, 60)
  73         if hours > 99:
  74             return '--:--:--'
  75         if hours == 0:
  76             return '%02d:%02d' % (mins, secs)
  77         else:
  78             return '%02d:%02d:%02d' % (hours, mins, secs)
  79
  80     @staticmethod
  81     def calc_percent(byte_counter, data_len):
  82         if data_len is None:
  83             return None
  84         return float(byte_counter) / float(data_len) * 100.0
  85
  86     @staticmethod
  87     def format_percent(percent):
  88         if percent is None:
  89             return '---.-%'
  90         return '%6s' % ('%3.1f%%' % percent)
  91
  92     @staticmethod
  93     def calc_eta(start, now, total, current):
  94         if total is None:
  95             return None
  96         if now is None:
  97             now = time.time()
  98         dif = now - start
  99         if current == 0 or dif < 0.001:  # One millisecond
 100             return None
 101         rate = float(current) / dif
 102         return int((float(total) - float(current)) / rate)
 103
 104     @staticmethod
 105     def format_eta(eta):
 106         if eta is None:
 107             return '--:--'
 108         return FileDownloader.format_seconds(eta)
 109
 110     @staticmethod
 111     def calc_speed(start, now, bytes):
 112         dif = now - start
 113         if bytes == 0 or dif < 0.001:  # One millisecond
 114             return None
 115         return float(bytes) / dif
 116
 117     @staticmethod
 118     def format_speed(speed):
 119         if speed is None:
 120             return '%10s' % '---b/s'
 121         return '%10s' % ('%s/s' % format_bytes(speed))
 122
 123     @staticmethod
 124     def format_retries(retries):
 125         return 'inf' if retries == float('inf') else '%.0f' % retries
 126
 127     @staticmethod
 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:
 132             return int(new_max)
 133         rate = bytes / elapsed_time
 134         if rate > new_max:
 135             return int(new_max)
 136         if rate < new_min:
 137             return int(new_min)
 138         return int(rate)
 139
 140     @staticmethod
 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)
 144         if matchobj is None:
 145             return None
 146         number = float(matchobj.group(1))
 147         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 148         return int(round(number * multiplier))
 149
 150     def to_screen(self, *args, **kargs):
 151         self.ydl.to_screen(*args, **kargs)
 152
 153     def to_stderr(self, message):
 154         self.ydl.to_screen(message)
 155
 156     def to_console_title(self, message):
 157         self.ydl.to_console_title(message)
 158
 159     def trouble(self, *args, **kargs):
 160         self.ydl.trouble(*args, **kargs)
 161
 162     def report_warning(self, *args, **kargs):
 163         self.ydl.report_warning(*args, **kargs)
 164
 165     def report_error(self, *args, **kargs):
 166         self.ydl.report_error(*args, **kargs)
 167
 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:
 172             return
 173         if now is None:
 174             now = time.time()
 175         elapsed = now - start_time
 176         if elapsed <= 0.0:
 177             return
 178         speed = float(byte_counter) / elapsed
 179         if speed > rate_limit:
 180             time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
 181
 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))):
 186             return filename
 187         return filename + '.part'
 188
 189     def undo_temp_name(self, filename):
 190         if filename.endswith('.part'):
 191             return filename[:-len('.part')]
 192         return filename
 193
 194     def ytdl_filename(self, filename):
 195         return filename + '.ytdl'
 196
 197     def try_rename(self, old_filename, new_filename):
 198         try:
 199             if old_filename == new_filename:
 200                 return
 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))
 204
 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:
 208             return
 209         if not os.path.isfile(encodeFilename(filename)):
 210             return
 211         timestr = last_modified_hdr
 212         if timestr is None:
 213             return
 214         filetime = timeconvert(timestr)
 215         if filetime is None:
 216             return filetime
 217         # Ignore obviously invalid dates
 218         if filetime == 0:
 219             return
 220         try:
 221             os.utime(filename, (time.time(), filetime))
 222         except Exception:
 223             pass
 224         return filetime
 225
 226     def report_destination(self, filename):
 227         """Report destination filename."""
 228         self.to_screen('[download] Destination: ' + filename)
 229
 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)
 234         else:
 235             if compat_os_name == 'nt':
 236                 prev_len = getattr(self, '_report_progress_prev_line_length',
 237                                    0)
 238                 if prev_len > len(fullmsg):
 239                     fullmsg += ' ' * (prev_len - len(fullmsg))
 240                 self._report_progress_prev_line_length = len(fullmsg)
 241                 clear_line = '\r'
 242             else:
 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)
 246
 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')
 251             else:
 252                 msg_template = '100%%'
 253                 if s.get('total_bytes') is not None:
 254                     s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 255                     msg_template += ' of %(_total_bytes_str)s'
 256                 if s.get('elapsed') is not None:
 257                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 258                     msg_template += ' in %(_elapsed_str)s'
 259                 self._report_progress_status(
 260                     msg_template % s, is_last_line=True)
 261
 262         if self.params.get('noprogress'):
 263             return
 264
 265         if s['status'] != 'downloading':
 266             return
 267
 268         if s.get('eta') is not None:
 269             s['_eta_str'] = self.format_eta(s['eta'])
 270         else:
 271             s['_eta_str'] = 'Unknown ETA'
 272
 273         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
 274             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
 275         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
 276             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
 277         else:
 278             if s.get('downloaded_bytes') == 0:
 279                 s['_percent_str'] = self.format_percent(0)
 280             else:
 281                 s['_percent_str'] = 'Unknown %'
 282
 283         if s.get('speed') is not None:
 284             s['_speed_str'] = self.format_speed(s['speed'])
 285         else:
 286             s['_speed_str'] = 'Unknown speed'
 287
 288         if s.get('total_bytes') is not None:
 289             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 290             msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
 291         elif s.get('total_bytes_estimate') is not None:
 292             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
 293             msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
 294         else:
 295             if s.get('downloaded_bytes') is not None:
 296                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
 297                 if s.get('elapsed'):
 298                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 299                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
 300                 else:
 301                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
 302             else:
 303                 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
 304
 305         self._report_progress_status(msg_template % s)
 306
 307     def report_resuming_byte(self, resume_len):
 308         """Report attempt to resume at given byte."""
 309         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 310
 311     def report_retry(self, err, count, retries):
 312         """Report retry in case of HTTP error 5xx"""
 313         self.to_screen(
 314             '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...'
 315             % (error_to_compat_str(err), count, self.format_retries(retries)))
 316
 317     def report_file_already_downloaded(self, file_name):
 318         """Report file has already been fully downloaded."""
 319         try:
 320             self.to_screen('[download] %s has already been downloaded' % file_name)
 321         except UnicodeEncodeError:
 322             self.to_screen('[download] The file has already been downloaded')
 323
 324     def report_unable_to_resume(self):
 325         """Report it was impossible to resume download."""
 326         self.to_screen('[download] Unable to resume')
 327
 328     def download(self, filename, info_dict):
 329         """Download to a filename using the info from info_dict
 330         Return True on success and False otherwise
 331         """
 332
 333         nooverwrites_and_exists = (
 334             self.params.get('nooverwrites', False) and
 335             os.path.exists(encodeFilename(filename))
 336         )
 337
 338         if not hasattr(filename, 'write'):
 339             continuedl_and_exists = (
 340                 self.params.get('continuedl', True) and
 341                 os.path.isfile(encodeFilename(filename)) and
 342                 not self.params.get('nopart', False)
 343             )
 344
 345             # Check file already present
 346             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 347                 self.report_file_already_downloaded(filename)
 348                 self._hook_progress({
 349                     'filename': filename,
 350                     'status': 'finished',
 351                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 352                 })
 353                 return True
 354
 355         min_sleep_interval = self.params.get('sleep_interval')
 356         if min_sleep_interval:
 357             max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
 358             sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
 359             self.to_screen(
 360                 '[download] Sleeping %s seconds...' % (
 361                     int(sleep_interval) if sleep_interval.is_integer()
 362                     else '%.2f' % sleep_interval))
 363             time.sleep(sleep_interval)
 364
 365         return self.real_download(filename, info_dict)
 366
 367     def real_download(self, filename, info_dict):
 368         """Real download process. Redefine in subclasses."""
 369         raise NotImplementedError('This method must be implemented by subclasses')
 370
 371     def _hook_progress(self, status):
 372         for ph in self._progress_hooks:
 373             ph(status)
 374
 375     def add_progress_hook(self, ph):
 376         # See YoutubeDl.py (search for progress_hooks) for a description of
 377         # this interface
 378         self._progress_hooks.append(ph)
 379
 380     def _debug_cmd(self, args, exe=None):
 381         if not self.params.get('verbose', False):
 382             return
 383
 384         str_args = [decodeArgument(a) for a in args]
 385
 386         if exe is None:
 387             exe = os.path.basename(str_args[0])
 388
 389         self.to_screen('[debug] %s command line: %s' % (
 390             exe, shell_quote(str_args)))