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     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             time.sleep(max((byte_counter // rate_limit) - elapsed, 0))
 180
 181     def temp_name(self, filename):
 182         """Returns a temporary filename for the given filename."""
 183         if self.params.get('nopart', False) or filename == '-' or \
 184                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 185             return filename
 186         return filename + '.part'
 187
 188     def undo_temp_name(self, filename):
 189         if filename.endswith('.part'):
 190             return filename[:-len('.part')]
 191         return filename
 192
 193     def ytdl_filename(self, filename):
 194         return filename + '.ytdl'
 195
 196     def try_rename(self, old_filename, new_filename):
 197         try:
 198             if old_filename == new_filename:
 199                 return
 200             os.rename(encodeFilename(old_filename), encodeFilename(new_filename))
 201         except (IOError, OSError) as err:
 202             self.report_error('unable to rename file: %s' % error_to_compat_str(err))
 203
 204     def try_utime(self, filename, last_modified_hdr):
 205         """Try to set the last-modified time of the given file."""
 206         if last_modified_hdr is None:
 207             return
 208         if not os.path.isfile(encodeFilename(filename)):
 209             return
 210         timestr = last_modified_hdr
 211         if timestr is None:
 212             return
 213         filetime = timeconvert(timestr)
 214         if filetime is None:
 215             return filetime
 216         # Ignore obviously invalid dates
 217         if filetime == 0:
 218             return
 219         try:
 220             os.utime(filename, (time.time(), filetime))
 221         except Exception:
 222             pass
 223         return filetime
 224
 225     def report_destination(self, filename):
 226         """Report destination filename."""
 227         self.to_screen('[download] Destination: ' + filename)
 228
 229     def _report_progress_status(self, msg, is_last_line=False):
 230         fullmsg = '[download] ' + msg
 231         if self.params.get('progress_with_newline', False):
 232             self.to_screen(fullmsg)
 233         else:
 234             if compat_os_name == 'nt':
 235                 prev_len = getattr(self, '_report_progress_prev_line_length',
 236                                    0)
 237                 if prev_len > len(fullmsg):
 238                     fullmsg += ' ' * (prev_len - len(fullmsg))
 239                 self._report_progress_prev_line_length = len(fullmsg)
 240                 clear_line = '\r'
 241             else:
 242                 clear_line = ('\r\x1b[K' if sys.stderr.isatty() else '\r')
 243             self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
 244         self.to_console_title('youtube-dl ' + msg)
 245
 246     def report_progress(self, s):
 247         if s['status'] == 'finished':
 248             if self.params.get('noprogress', False):
 249                 self.to_screen('[download] Download completed')
 250             else:
 251                 msg_template = '100%%'
 252                 if s.get('total_bytes') is not None:
 253                     s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 254                     msg_template += ' of %(_total_bytes_str)s'
 255                 if s.get('elapsed') is not None:
 256                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 257                     msg_template += ' in %(_elapsed_str)s'
 258                 self._report_progress_status(
 259                     msg_template % s, is_last_line=True)
 260
 261         if self.params.get('noprogress'):
 262             return
 263
 264         if s['status'] != 'downloading':
 265             return
 266
 267         if s.get('eta') is not None:
 268             s['_eta_str'] = self.format_eta(s['eta'])
 269         else:
 270             s['_eta_str'] = 'Unknown ETA'
 271
 272         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
 273             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
 274         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
 275             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
 276         else:
 277             if s.get('downloaded_bytes') == 0:
 278                 s['_percent_str'] = self.format_percent(0)
 279             else:
 280                 s['_percent_str'] = 'Unknown %'
 281
 282         if s.get('speed') is not None:
 283             s['_speed_str'] = self.format_speed(s['speed'])
 284         else:
 285             s['_speed_str'] = 'Unknown speed'
 286
 287         if s.get('total_bytes') is not None:
 288             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 289             msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
 290         elif s.get('total_bytes_estimate') is not None:
 291             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
 292             msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
 293         else:
 294             if s.get('downloaded_bytes') is not None:
 295                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
 296                 if s.get('elapsed'):
 297                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 298                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
 299                 else:
 300                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
 301             else:
 302                 msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
 303
 304         self._report_progress_status(msg_template % s)
 305
 306     def report_resuming_byte(self, resume_len):
 307         """Report attempt to resume at given byte."""
 308         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 309
 310     def report_retry(self, err, count, retries):
 311         """Report retry in case of HTTP error 5xx"""
 312         self.to_screen(
 313             '[download] Got server HTTP error: %s. Retrying (attempt %d of %s)...'
 314             % (error_to_compat_str(err), count, self.format_retries(retries)))
 315
 316     def report_file_already_downloaded(self, file_name):
 317         """Report file has already been fully downloaded."""
 318         try:
 319             self.to_screen('[download] %s has already been downloaded' % file_name)
 320         except UnicodeEncodeError:
 321             self.to_screen('[download] The file has already been downloaded')
 322
 323     def report_unable_to_resume(self):
 324         """Report it was impossible to resume download."""
 325         self.to_screen('[download] Unable to resume')
 326
 327     def download(self, filename, info_dict):
 328         """Download to a filename using the info from info_dict
 329         Return True on success and False otherwise
 330         """
 331
 332         nooverwrites_and_exists = (
 333             self.params.get('nooverwrites', False) and
 334             os.path.exists(encodeFilename(filename))
 335         )
 336
 337         if not hasattr(filename, 'write'):
 338             continuedl_and_exists = (
 339                 self.params.get('continuedl', True) and
 340                 os.path.isfile(encodeFilename(filename)) and
 341                 not self.params.get('nopart', False)
 342             )
 343
 344             # Check file already present
 345             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 346                 self.report_file_already_downloaded(filename)
 347                 self._hook_progress({
 348                     'filename': filename,
 349                     'status': 'finished',
 350                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 351                 })
 352                 return True
 353
 354         min_sleep_interval = self.params.get('sleep_interval')
 355         if min_sleep_interval:
 356             max_sleep_interval = self.params.get('max_sleep_interval', min_sleep_interval)
 357             sleep_interval = random.uniform(min_sleep_interval, max_sleep_interval)
 358             self.to_screen(
 359                 '[download] Sleeping %s seconds...' % (
 360                     int(sleep_interval) if sleep_interval.is_integer()
 361                     else '%.2f' % sleep_interval))
 362             time.sleep(sleep_interval)
 363
 364         return self.real_download(filename, info_dict)
 365
 366     def real_download(self, filename, info_dict):
 367         """Real download process. Redefine in subclasses."""
 368         raise NotImplementedError('This method must be implemented by subclasses')
 369
 370     def _hook_progress(self, status):
 371         for ph in self._progress_hooks:
 372             ph(status)
 373
 374     def add_progress_hook(self, ph):
 375         # See YoutubeDl.py (search for progress_hooks) for a description of
 376         # this interface
 377         self._progress_hooks.append(ph)
 378
 379     def _debug_cmd(self, args, exe=None):
 380         if not self.params.get('verbose', False):
 381             return
 382
 383         str_args = [decodeArgument(a) for a in args]
 384
 385         if exe is None:
 386             exe = os.path.basename(str_args[0])
 387
 388         self.to_screen('[debug] %s command line: %s' % (
 389             exe, shell_quote(str_args)))