yt_dlp/downloader/common.py

   1 import contextlib
   2 import errno
   3 import os
   4 import random
   5 import re
   6 import time
   7
   8 from ..minicurses import (
   9     BreaklineStatusPrinter,
  10     MultilineLogger,
  11     MultilinePrinter,
  12     QuietMultilinePrinter,
  13 )
  14 from ..utils import (
  15     NUMBER_RE,
  16     LockingUnsupportedError,
  17     Namespace,
  18     decodeArgument,
  19     encodeFilename,
  20     error_to_compat_str,
  21     format_bytes,
  22     sanitize_open,
  23     shell_quote,
  24     timeconvert,
  25     timetuple_from_msec,
  26 )
  27
  28
  29 class FileDownloader:
  30     """File Downloader class.
  31
  32     File downloader objects are the ones responsible of downloading the
  33     actual video file and writing it to disk.
  34
  35     File downloaders accept a lot of parameters. In order not to saturate
  36     the object constructor with arguments, it receives a dictionary of
  37     options instead.
  38
  39     Available options:
  40
  41     verbose:            Print additional info to stdout.
  42     quiet:              Do not print messages to stdout.
  43     ratelimit:          Download speed limit, in bytes/sec.
  44     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  45     retries:            Number of times to retry for HTTP error 5xx
  46     file_access_retries:   Number of times to retry on file access error
  47     buffersize:         Size of download buffer in bytes.
  48     noresizebuffer:     Do not automatically resize the download buffer.
  49     continuedl:         Try to continue downloads if possible.
  50     noprogress:         Do not print the progress bar.
  51     nopart:             Do not use temporary .part files.
  52     updatetime:         Use the Last-modified header to set output file timestamps.
  53     test:               Download only first bytes to test the downloader.
  54     min_filesize:       Skip files smaller than this size
  55     max_filesize:       Skip files larger than this size
  56     xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  57     external_downloader_args:  A dictionary of downloader keys (in lower case)
  58                         and a list of additional command-line arguments for the
  59                         executable. Use 'default' as the name for arguments to be
  60                         passed to all downloaders. For compatibility with youtube-dl,
  61                         a single list of args can also be used
  62     hls_use_mpegts:     Use the mpegts container for HLS videos.
  63     http_chunk_size:    Size of a chunk for chunk-based HTTP downloading. May be
  64                         useful for bypassing bandwidth throttling imposed by
  65                         a webserver (experimental)
  66     progress_template:  See YoutubeDL.py
  67
  68     Subclasses of this one must re-define the real_download method.
  69     """
  70
  71     _TEST_FILE_SIZE = 10241
  72     params = None
  73
  74     def __init__(self, ydl, params):
  75         """Create a FileDownloader object with the given options."""
  76         self._set_ydl(ydl)
  77         self._progress_hooks = []
  78         self.params = params
  79         self._prepare_multiline_status()
  80         self.add_progress_hook(self.report_progress)
  81
  82     def _set_ydl(self, ydl):
  83         self.ydl = ydl
  84
  85         for func in (
  86             'deprecation_warning',
  87             'report_error',
  88             'report_file_already_downloaded',
  89             'report_warning',
  90             'to_console_title',
  91             'to_stderr',
  92             'trouble',
  93             'write_debug',
  94         ):
  95             if not hasattr(self, func):
  96                 setattr(self, func, getattr(ydl, func))
  97
  98     def to_screen(self, *args, **kargs):
  99         self.ydl.to_screen(*args, quiet=self.params.get('quiet'), **kargs)
 100
 101     @staticmethod
 102     def format_seconds(seconds):
 103         time = timetuple_from_msec(seconds * 1000)
 104         if time.hours > 99:
 105             return '--:--:--'
 106         if not time.hours:
 107             return '%02d:%02d' % time[1:-1]
 108         return '%02d:%02d:%02d' % time[:-1]
 109
 110     @staticmethod
 111     def calc_percent(byte_counter, data_len):
 112         if data_len is None:
 113             return None
 114         return float(byte_counter) / float(data_len) * 100.0
 115
 116     @staticmethod
 117     def format_percent(percent):
 118         if percent is None:
 119             return '---.-%'
 120         elif percent == 100:
 121             return '100%'
 122         return '%6s' % ('%3.1f%%' % percent)
 123
 124     @staticmethod
 125     def calc_eta(start, now, total, current):
 126         if total is None:
 127             return None
 128         if now is None:
 129             now = time.time()
 130         dif = now - start
 131         if current == 0 or dif < 0.001:  # One millisecond
 132             return None
 133         rate = float(current) / dif
 134         return int((float(total) - float(current)) / rate)
 135
 136     @staticmethod
 137     def format_eta(eta):
 138         if eta is None:
 139             return '--:--'
 140         return FileDownloader.format_seconds(eta)
 141
 142     @staticmethod
 143     def calc_speed(start, now, bytes):
 144         dif = now - start
 145         if bytes == 0 or dif < 0.001:  # One millisecond
 146             return None
 147         return float(bytes) / dif
 148
 149     @staticmethod
 150     def format_speed(speed):
 151         if speed is None:
 152             return '%10s' % '---b/s'
 153         return '%10s' % ('%s/s' % format_bytes(speed))
 154
 155     @staticmethod
 156     def format_retries(retries):
 157         return 'inf' if retries == float('inf') else '%.0f' % retries
 158
 159     @staticmethod
 160     def best_block_size(elapsed_time, bytes):
 161         new_min = max(bytes / 2.0, 1.0)
 162         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 163         if elapsed_time < 0.001:
 164             return int(new_max)
 165         rate = bytes / elapsed_time
 166         if rate > new_max:
 167             return int(new_max)
 168         if rate < new_min:
 169             return int(new_min)
 170         return int(rate)
 171
 172     @staticmethod
 173     def parse_bytes(bytestr):
 174         """Parse a string indicating a byte quantity into an integer."""
 175         matchobj = re.match(rf'(?i)^({NUMBER_RE})([kMGTPEZY]?)$', bytestr)
 176         if matchobj is None:
 177             return None
 178         number = float(matchobj.group(1))
 179         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 180         return int(round(number * multiplier))
 181
 182     def slow_down(self, start_time, now, byte_counter):
 183         """Sleep if the download speed is over the rate limit."""
 184         rate_limit = self.params.get('ratelimit')
 185         if rate_limit is None or byte_counter == 0:
 186             return
 187         if now is None:
 188             now = time.time()
 189         elapsed = now - start_time
 190         if elapsed <= 0.0:
 191             return
 192         speed = float(byte_counter) / elapsed
 193         if speed > rate_limit:
 194             sleep_time = float(byte_counter) / rate_limit - elapsed
 195             if sleep_time > 0:
 196                 time.sleep(sleep_time)
 197
 198     def temp_name(self, filename):
 199         """Returns a temporary filename for the given filename."""
 200         if self.params.get('nopart', False) or filename == '-' or \
 201                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 202             return filename
 203         return filename + '.part'
 204
 205     def undo_temp_name(self, filename):
 206         if filename.endswith('.part'):
 207             return filename[:-len('.part')]
 208         return filename
 209
 210     def ytdl_filename(self, filename):
 211         return filename + '.ytdl'
 212
 213     def wrap_file_access(action, *, fatal=False):
 214         def outer(func):
 215             def inner(self, *args, **kwargs):
 216                 file_access_retries = self.params.get('file_access_retries', 0)
 217                 retry = 0
 218                 while True:
 219                     try:
 220                         return func(self, *args, **kwargs)
 221                     except OSError as err:
 222                         retry = retry + 1
 223                         if retry > file_access_retries or err.errno not in (errno.EACCES, errno.EINVAL):
 224                             if not fatal:
 225                                 self.report_error(f'unable to {action} file: {err}')
 226                                 return
 227                             raise
 228                         self.to_screen(
 229                             f'[download] Unable to {action} file due to file access error. '
 230                             f'Retrying (attempt {retry} of {self.format_retries(file_access_retries)}) ...')
 231                         time.sleep(0.01)
 232             return inner
 233         return outer
 234
 235     @wrap_file_access('open', fatal=True)
 236     def sanitize_open(self, filename, open_mode):
 237         f, filename = sanitize_open(filename, open_mode)
 238         if not getattr(f, 'locked', None):
 239             self.write_debug(f'{LockingUnsupportedError.msg}. Proceeding without locking', only_once=True)
 240         return f, filename
 241
 242     @wrap_file_access('remove')
 243     def try_remove(self, filename):
 244         os.remove(filename)
 245
 246     @wrap_file_access('rename')
 247     def try_rename(self, old_filename, new_filename):
 248         if old_filename == new_filename:
 249             return
 250         os.replace(old_filename, new_filename)
 251
 252     def try_utime(self, filename, last_modified_hdr):
 253         """Try to set the last-modified time of the given file."""
 254         if last_modified_hdr is None:
 255             return
 256         if not os.path.isfile(encodeFilename(filename)):
 257             return
 258         timestr = last_modified_hdr
 259         if timestr is None:
 260             return
 261         filetime = timeconvert(timestr)
 262         if filetime is None:
 263             return filetime
 264         # Ignore obviously invalid dates
 265         if filetime == 0:
 266             return
 267         with contextlib.suppress(Exception):
 268             os.utime(filename, (time.time(), filetime))
 269         return filetime
 270
 271     def report_destination(self, filename):
 272         """Report destination filename."""
 273         self.to_screen('[download] Destination: ' + filename)
 274
 275     def _prepare_multiline_status(self, lines=1):
 276         if self.params.get('noprogress'):
 277             self._multiline = QuietMultilinePrinter()
 278         elif self.ydl.params.get('logger'):
 279             self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
 280         elif self.params.get('progress_with_newline'):
 281             self._multiline = BreaklineStatusPrinter(self.ydl._out_files['screen'], lines)
 282         else:
 283             self._multiline = MultilinePrinter(self.ydl._out_files['screen'], lines, not self.params.get('quiet'))
 284         self._multiline.allow_colors = self._multiline._HAVE_FULLCAP and not self.params.get('no_color')
 285
 286     def _finish_multiline_status(self):
 287         self._multiline.end()
 288
 289     ProgressStyles = Namespace(
 290         downloaded_bytes='light blue',
 291         percent='light blue',
 292         eta='yellow',
 293         speed='green',
 294         elapsed='bold white',
 295         total_bytes='',
 296         total_bytes_estimate='',
 297     )
 298
 299     def _report_progress_status(self, s, default_template):
 300         for name, style in self.ProgressStyles._asdict().items():
 301             name = f'_{name}_str'
 302             if name not in s:
 303                 continue
 304             s[name] = self._format_progress(s[name], style)
 305         s['_default_template'] = default_template % s
 306
 307         progress_dict = s.copy()
 308         progress_dict.pop('info_dict')
 309         progress_dict = {'info': s['info_dict'], 'progress': progress_dict}
 310
 311         progress_template = self.params.get('progress_template', {})
 312         self._multiline.print_at_line(self.ydl.evaluate_outtmpl(
 313             progress_template.get('download') or '[download] %(progress._default_template)s',
 314             progress_dict), s.get('progress_idx') or 0)
 315         self.to_console_title(self.ydl.evaluate_outtmpl(
 316             progress_template.get('download-title') or 'yt-dlp %(progress._default_template)s',
 317             progress_dict))
 318
 319     def _format_progress(self, *args, **kwargs):
 320         return self.ydl._format_text(
 321             self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
 322
 323     def report_progress(self, s):
 324         if s['status'] == 'finished':
 325             if self.params.get('noprogress'):
 326                 self.to_screen('[download] Download completed')
 327             msg_template = '100%%'
 328             if s.get('total_bytes') is not None:
 329                 s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 330                 msg_template += ' of %(_total_bytes_str)s'
 331             if s.get('elapsed') is not None:
 332                 s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 333                 msg_template += ' in %(_elapsed_str)s'
 334             s['_percent_str'] = self.format_percent(100)
 335             self._report_progress_status(s, msg_template)
 336             return
 337
 338         if s['status'] != 'downloading':
 339             return
 340
 341         if s.get('eta') is not None:
 342             s['_eta_str'] = self.format_eta(s['eta'])
 343         else:
 344             s['_eta_str'] = 'Unknown'
 345
 346         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
 347             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
 348         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
 349             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
 350         else:
 351             if s.get('downloaded_bytes') == 0:
 352                 s['_percent_str'] = self.format_percent(0)
 353             else:
 354                 s['_percent_str'] = 'Unknown %'
 355
 356         if s.get('speed') is not None:
 357             s['_speed_str'] = self.format_speed(s['speed'])
 358         else:
 359             s['_speed_str'] = 'Unknown speed'
 360
 361         if s.get('total_bytes') is not None:
 362             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
 363             msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
 364         elif s.get('total_bytes_estimate') is not None:
 365             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
 366             msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
 367         else:
 368             if s.get('downloaded_bytes') is not None:
 369                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
 370                 if s.get('elapsed'):
 371                     s['_elapsed_str'] = self.format_seconds(s['elapsed'])
 372                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
 373                 else:
 374                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
 375             else:
 376                 msg_template = '%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s'
 377         if s.get('fragment_index') and s.get('fragment_count'):
 378             msg_template += ' (frag %(fragment_index)s/%(fragment_count)s)'
 379         elif s.get('fragment_index'):
 380             msg_template += ' (frag %(fragment_index)s)'
 381         self._report_progress_status(s, msg_template)
 382
 383     def report_resuming_byte(self, resume_len):
 384         """Report attempt to resume at given byte."""
 385         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 386
 387     def report_retry(self, err, count, retries):
 388         """Report retry in case of HTTP error 5xx"""
 389         self.to_screen(
 390             '[download] Got server HTTP error: %s. Retrying (attempt %d of %s) ...'
 391             % (error_to_compat_str(err), count, self.format_retries(retries)))
 392
 393     def report_unable_to_resume(self):
 394         """Report it was impossible to resume download."""
 395         self.to_screen('[download] Unable to resume')
 396
 397     @staticmethod
 398     def supports_manifest(manifest):
 399         """ Whether the downloader can download the fragments from the manifest.
 400         Redefine in subclasses if needed. """
 401         pass
 402
 403     def download(self, filename, info_dict, subtitle=False):
 404         """Download to a filename using the info from info_dict
 405         Return True on success and False otherwise
 406         """
 407
 408         nooverwrites_and_exists = (
 409             not self.params.get('overwrites', True)
 410             and os.path.exists(encodeFilename(filename))
 411         )
 412
 413         if not hasattr(filename, 'write'):
 414             continuedl_and_exists = (
 415                 self.params.get('continuedl', True)
 416                 and os.path.isfile(encodeFilename(filename))
 417                 and not self.params.get('nopart', False)
 418             )
 419
 420             # Check file already present
 421             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 422                 self.report_file_already_downloaded(filename)
 423                 self._hook_progress({
 424                     'filename': filename,
 425                     'status': 'finished',
 426                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 427                 }, info_dict)
 428                 self._finish_multiline_status()
 429                 return True, False
 430
 431         if subtitle:
 432             sleep_interval = self.params.get('sleep_interval_subtitles') or 0
 433         else:
 434             min_sleep_interval = self.params.get('sleep_interval') or 0
 435             sleep_interval = random.uniform(
 436                 min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
 437         if sleep_interval > 0:
 438             self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
 439             time.sleep(sleep_interval)
 440
 441         ret = self.real_download(filename, info_dict)
 442         self._finish_multiline_status()
 443         return ret, True
 444
 445     def real_download(self, filename, info_dict):
 446         """Real download process. Redefine in subclasses."""
 447         raise NotImplementedError('This method must be implemented by subclasses')
 448
 449     def _hook_progress(self, status, info_dict):
 450         if not self._progress_hooks:
 451             return
 452         status['info_dict'] = info_dict
 453         # youtube-dl passes the same status object to all the hooks.
 454         # Some third party scripts seems to be relying on this.
 455         # So keep this behavior if possible
 456         for ph in self._progress_hooks:
 457             ph(status)
 458
 459     def add_progress_hook(self, ph):
 460         # See YoutubeDl.py (search for progress_hooks) for a description of
 461         # this interface
 462         self._progress_hooks.append(ph)
 463
 464     def _debug_cmd(self, args, exe=None):
 465         if not self.params.get('verbose', False):
 466             return
 467
 468         str_args = [decodeArgument(a) for a in args]
 469
 470         if exe is None:
 471             exe = os.path.basename(str_args[0])
 472
 473         self.write_debug(f'{exe} command line: {shell_quote(str_args)}')