yt_dlp/downloader/common.py

   1 import contextlib
   2 import errno
   3 import functools
   4 import os
   5 import random
   6 import re
   7 import threading
   8 import time
   9
  10 from ..minicurses import (
  11     BreaklineStatusPrinter,
  12     MultilineLogger,
  13     MultilinePrinter,
  14     QuietMultilinePrinter,
  15 )
  16 from ..utils import (
  17     IDENTITY,
  18     NO_DEFAULT,
  19     LockingUnsupportedError,
  20     Namespace,
  21     RetryManager,
  22     classproperty,
  23     decodeArgument,
  24     deprecation_warning,
  25     encodeFilename,
  26     format_bytes,
  27     join_nonempty,
  28     parse_bytes,
  29     remove_start,
  30     sanitize_open,
  31     shell_quote,
  32     timeconvert,
  33     timetuple_from_msec,
  34     try_call,
  35 )
  36
  37
  38 class FileDownloader:
  39     """File Downloader class.
  40
  41     File downloader objects are the ones responsible of downloading the
  42     actual video file and writing it to disk.
  43
  44     File downloaders accept a lot of parameters. In order not to saturate
  45     the object constructor with arguments, it receives a dictionary of
  46     options instead.
  47
  48     Available options:
  49
  50     verbose:            Print additional info to stdout.
  51     quiet:              Do not print messages to stdout.
  52     ratelimit:          Download speed limit, in bytes/sec.
  53     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  54     retries:            Number of times to retry for expected network errors.
  55                         Default is 0 for API, but 10 for CLI
  56     file_access_retries:   Number of times to retry on file access error (default: 3)
  57     buffersize:         Size of download buffer in bytes.
  58     noresizebuffer:     Do not automatically resize the download buffer.
  59     continuedl:         Try to continue downloads if possible.
  60     noprogress:         Do not print the progress bar.
  61     nopart:             Do not use temporary .part files.
  62     updatetime:         Use the Last-modified header to set output file timestamps.
  63     test:               Download only first bytes to test the downloader.
  64     min_filesize:       Skip files smaller than this size
  65     max_filesize:       Skip files larger than this size
  66     xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  67     progress_delta:     The minimum time between progress output, in seconds
  68     external_downloader_args:  A dictionary of downloader keys (in lower case)
  69                         and a list of additional command-line arguments for the
  70                         executable. Use 'default' as the name for arguments to be
  71                         passed to all downloaders. For compatibility with youtube-dl,
  72                         a single list of args can also be used
  73     hls_use_mpegts:     Use the mpegts container for HLS videos.
  74     http_chunk_size:    Size of a chunk for chunk-based HTTP downloading. May be
  75                         useful for bypassing bandwidth throttling imposed by
  76                         a webserver (experimental)
  77     progress_template:  See YoutubeDL.py
  78     retry_sleep_functions: See YoutubeDL.py
  79
  80     Subclasses of this one must re-define the real_download method.
  81     """
  82
  83     _TEST_FILE_SIZE = 10241
  84     params = None
  85
  86     def __init__(self, ydl, params):
  87         """Create a FileDownloader object with the given options."""
  88         self._set_ydl(ydl)
  89         self._progress_hooks = []
  90         self.params = params
  91         self._prepare_multiline_status()
  92         self.add_progress_hook(self.report_progress)
  93         if self.params.get('progress_delta'):
  94             self._progress_delta_lock = threading.Lock()
  95             self._progress_delta_time = time.monotonic()
  96
  97     def _set_ydl(self, ydl):
  98         self.ydl = ydl
  99
 100         for func in (
 101             'deprecation_warning',
 102             'deprecated_feature',
 103             'report_error',
 104             'report_file_already_downloaded',
 105             'report_warning',
 106             'to_console_title',
 107             'to_stderr',
 108             'trouble',
 109             'write_debug',
 110         ):
 111             if not hasattr(self, func):
 112                 setattr(self, func, getattr(ydl, func))
 113
 114     def to_screen(self, *args, **kargs):
 115         self.ydl.to_screen(*args, quiet=self.params.get('quiet'), **kargs)
 116
 117     __to_screen = to_screen
 118
 119     @classproperty
 120     def FD_NAME(cls):
 121         return re.sub(r'(?<=[a-z])(?=[A-Z])', '_', cls.__name__[:-2]).lower()
 122
 123     @staticmethod
 124     def format_seconds(seconds):
 125         if seconds is None:
 126             return ' Unknown'
 127         time = timetuple_from_msec(seconds * 1000)
 128         if time.hours > 99:
 129             return '--:--:--'
 130         return '%02d:%02d:%02d' % time[:-1]
 131
 132     @classmethod
 133     def format_eta(cls, seconds):
 134         return f'{remove_start(cls.format_seconds(seconds), "00:"):>8s}'
 135
 136     @staticmethod
 137     def calc_percent(byte_counter, data_len):
 138         if data_len is None:
 139             return None
 140         return float(byte_counter) / float(data_len) * 100.0
 141
 142     @staticmethod
 143     def format_percent(percent):
 144         return '  N/A%' if percent is None else f'{percent:>5.1f}%'
 145
 146     @classmethod
 147     def calc_eta(cls, start_or_rate, now_or_remaining, total=NO_DEFAULT, current=NO_DEFAULT):
 148         if total is NO_DEFAULT:
 149             rate, remaining = start_or_rate, now_or_remaining
 150             if None in (rate, remaining):
 151                 return None
 152             return int(float(remaining) / rate)
 153
 154         start, now = start_or_rate, now_or_remaining
 155         if total is None:
 156             return None
 157         if now is None:
 158             now = time.time()
 159         rate = cls.calc_speed(start, now, current)
 160         return rate and int((float(total) - float(current)) / rate)
 161
 162     @staticmethod
 163     def calc_speed(start, now, bytes):
 164         dif = now - start
 165         if bytes == 0 or dif < 0.001:  # One millisecond
 166             return None
 167         return float(bytes) / dif
 168
 169     @staticmethod
 170     def format_speed(speed):
 171         return ' Unknown B/s' if speed is None else f'{format_bytes(speed):>10s}/s'
 172
 173     @staticmethod
 174     def format_retries(retries):
 175         return 'inf' if retries == float('inf') else int(retries)
 176
 177     @staticmethod
 178     def filesize_or_none(unencoded_filename):
 179         if os.path.isfile(unencoded_filename):
 180             return os.path.getsize(unencoded_filename)
 181         return 0
 182
 183     @staticmethod
 184     def best_block_size(elapsed_time, bytes):
 185         new_min = max(bytes / 2.0, 1.0)
 186         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 187         if elapsed_time < 0.001:
 188             return int(new_max)
 189         rate = bytes / elapsed_time
 190         if rate > new_max:
 191             return int(new_max)
 192         if rate < new_min:
 193             return int(new_min)
 194         return int(rate)
 195
 196     @staticmethod
 197     def parse_bytes(bytestr):
 198         """Parse a string indicating a byte quantity into an integer."""
 199         deprecation_warning('yt_dlp.FileDownloader.parse_bytes is deprecated and '
 200                             'may be removed in the future. Use yt_dlp.utils.parse_bytes instead')
 201         return parse_bytes(bytestr)
 202
 203     def slow_down(self, start_time, now, byte_counter):
 204         """Sleep if the download speed is over the rate limit."""
 205         rate_limit = self.params.get('ratelimit')
 206         if rate_limit is None or byte_counter == 0:
 207             return
 208         if now is None:
 209             now = time.time()
 210         elapsed = now - start_time
 211         if elapsed <= 0.0:
 212             return
 213         speed = float(byte_counter) / elapsed
 214         if speed > rate_limit:
 215             sleep_time = float(byte_counter) / rate_limit - elapsed
 216             if sleep_time > 0:
 217                 time.sleep(sleep_time)
 218
 219     def temp_name(self, filename):
 220         """Returns a temporary filename for the given filename."""
 221         if self.params.get('nopart', False) or filename == '-' or \
 222                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 223             return filename
 224         return filename + '.part'
 225
 226     def undo_temp_name(self, filename):
 227         if filename.endswith('.part'):
 228             return filename[:-len('.part')]
 229         return filename
 230
 231     def ytdl_filename(self, filename):
 232         return filename + '.ytdl'
 233
 234     def wrap_file_access(action, *, fatal=False):
 235         def error_callback(err, count, retries, *, fd):
 236             return RetryManager.report_retry(
 237                 err, count, retries, info=fd.__to_screen,
 238                 warn=lambda e: (time.sleep(0.01), fd.to_screen(f'[download] Unable to {action} file: {e}')),
 239                 error=None if fatal else lambda e: fd.report_error(f'Unable to {action} file: {e}'),
 240                 sleep_func=fd.params.get('retry_sleep_functions', {}).get('file_access'))
 241
 242         def wrapper(self, func, *args, **kwargs):
 243             for retry in RetryManager(self.params.get('file_access_retries', 3), error_callback, fd=self):
 244                 try:
 245                     return func(self, *args, **kwargs)
 246                 except OSError as err:
 247                     if err.errno in (errno.EACCES, errno.EINVAL):
 248                         retry.error = err
 249                         continue
 250                     retry.error_callback(err, 1, 0)
 251
 252         return functools.partial(functools.partialmethod, wrapper)
 253
 254     @wrap_file_access('open', fatal=True)
 255     def sanitize_open(self, filename, open_mode):
 256         f, filename = sanitize_open(filename, open_mode)
 257         if not getattr(f, 'locked', None):
 258             self.write_debug(f'{LockingUnsupportedError.msg}. Proceeding without locking', only_once=True)
 259         return f, filename
 260
 261     @wrap_file_access('remove')
 262     def try_remove(self, filename):
 263         if os.path.isfile(filename):
 264             os.remove(filename)
 265
 266     @wrap_file_access('rename')
 267     def try_rename(self, old_filename, new_filename):
 268         if old_filename == new_filename:
 269             return
 270         os.replace(old_filename, new_filename)
 271
 272     def try_utime(self, filename, last_modified_hdr):
 273         """Try to set the last-modified time of the given file."""
 274         if last_modified_hdr is None:
 275             return
 276         if not os.path.isfile(encodeFilename(filename)):
 277             return
 278         timestr = last_modified_hdr
 279         if timestr is None:
 280             return
 281         filetime = timeconvert(timestr)
 282         if filetime is None:
 283             return filetime
 284         # Ignore obviously invalid dates
 285         if filetime == 0:
 286             return
 287         with contextlib.suppress(Exception):
 288             os.utime(filename, (time.time(), filetime))
 289         return filetime
 290
 291     def report_destination(self, filename):
 292         """Report destination filename."""
 293         self.to_screen('[download] Destination: ' + filename)
 294
 295     def _prepare_multiline_status(self, lines=1):
 296         if self.params.get('noprogress'):
 297             self._multiline = QuietMultilinePrinter()
 298         elif self.ydl.params.get('logger'):
 299             self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
 300         elif self.params.get('progress_with_newline'):
 301             self._multiline = BreaklineStatusPrinter(self.ydl._out_files.out, lines)
 302         else:
 303             self._multiline = MultilinePrinter(self.ydl._out_files.out, lines, not self.params.get('quiet'))
 304         self._multiline.allow_colors = self.ydl._allow_colors.out and self.ydl._allow_colors.out != 'no_color'
 305         self._multiline._HAVE_FULLCAP = self.ydl._allow_colors.out
 306
 307     def _finish_multiline_status(self):
 308         self._multiline.end()
 309
 310     ProgressStyles = Namespace(
 311         downloaded_bytes='light blue',
 312         percent='light blue',
 313         eta='yellow',
 314         speed='green',
 315         elapsed='bold white',
 316         total_bytes='',
 317         total_bytes_estimate='',
 318     )
 319
 320     def _report_progress_status(self, s, default_template):
 321         for name, style in self.ProgressStyles.items_:
 322             name = f'_{name}_str'
 323             if name not in s:
 324                 continue
 325             s[name] = self._format_progress(s[name], style)
 326         s['_default_template'] = default_template % s
 327
 328         progress_dict = s.copy()
 329         progress_dict.pop('info_dict')
 330         progress_dict = {'info': s['info_dict'], 'progress': progress_dict}
 331
 332         progress_template = self.params.get('progress_template', {})
 333         self._multiline.print_at_line(self.ydl.evaluate_outtmpl(
 334             progress_template.get('download') or '[download] %(progress._default_template)s',
 335             progress_dict), s.get('progress_idx') or 0)
 336         self.to_console_title(self.ydl.evaluate_outtmpl(
 337             progress_template.get('download-title') or 'yt-dlp %(progress._default_template)s',
 338             progress_dict))
 339
 340     def _format_progress(self, *args, **kwargs):
 341         return self.ydl._format_text(
 342             self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
 343
 344     def report_progress(self, s):
 345         def with_fields(*tups, default=''):
 346             for *fields, tmpl in tups:
 347                 if all(s.get(f) is not None for f in fields):
 348                     return tmpl
 349             return default
 350
 351         _format_bytes = lambda k: f'{format_bytes(s.get(k)):>10s}'
 352
 353         if s['status'] == 'finished':
 354             if self.params.get('noprogress'):
 355                 self.to_screen('[download] Download completed')
 356             speed = try_call(lambda: s['total_bytes'] / s['elapsed'])
 357             s.update({
 358                 'speed': speed,
 359                 '_speed_str': self.format_speed(speed).strip(),
 360                 '_total_bytes_str': _format_bytes('total_bytes'),
 361                 '_elapsed_str': self.format_seconds(s.get('elapsed')),
 362                 '_percent_str': self.format_percent(100),
 363             })
 364             self._report_progress_status(s, join_nonempty(
 365                 '100%%',
 366                 with_fields(('total_bytes', 'of %(_total_bytes_str)s')),
 367                 with_fields(('elapsed', 'in %(_elapsed_str)s')),
 368                 with_fields(('speed', 'at %(_speed_str)s')),
 369                 delim=' '))
 370
 371         if s['status'] != 'downloading':
 372             return
 373
 374         if update_delta := self.params.get('progress_delta'):
 375             with self._progress_delta_lock:
 376                 if time.monotonic() < self._progress_delta_time:
 377                     return
 378                 self._progress_delta_time += update_delta
 379
 380         s.update({
 381             '_eta_str': self.format_eta(s.get('eta')).strip(),
 382             '_speed_str': self.format_speed(s.get('speed')),
 383             '_percent_str': self.format_percent(try_call(
 384                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes'],
 385                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes_estimate'],
 386                 lambda: s['downloaded_bytes'] == 0 and 0)),
 387             '_total_bytes_str': _format_bytes('total_bytes'),
 388             '_total_bytes_estimate_str': _format_bytes('total_bytes_estimate'),
 389             '_downloaded_bytes_str': _format_bytes('downloaded_bytes'),
 390             '_elapsed_str': self.format_seconds(s.get('elapsed')),
 391         })
 392
 393         msg_template = with_fields(
 394             ('total_bytes', '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 395             ('total_bytes_estimate', '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 396             ('downloaded_bytes', 'elapsed', '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'),
 397             ('downloaded_bytes', '%(_downloaded_bytes_str)s at %(_speed_str)s'),
 398             default='%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s')
 399
 400         msg_template += with_fields(
 401             ('fragment_index', 'fragment_count', ' (frag %(fragment_index)s/%(fragment_count)s)'),
 402             ('fragment_index', ' (frag %(fragment_index)s)'))
 403         self._report_progress_status(s, msg_template)
 404
 405     def report_resuming_byte(self, resume_len):
 406         """Report attempt to resume at given byte."""
 407         self.to_screen(f'[download] Resuming download at byte {resume_len}')
 408
 409     def report_retry(self, err, count, retries, frag_index=NO_DEFAULT, fatal=True):
 410         """Report retry"""
 411         is_frag = False if frag_index is NO_DEFAULT else 'fragment'
 412         RetryManager.report_retry(
 413             err, count, retries, info=self.__to_screen,
 414             warn=lambda msg: self.__to_screen(f'[download] Got error: {msg}'),
 415             error=IDENTITY if not fatal else lambda e: self.report_error(f'\r[download] Got error: {e}'),
 416             sleep_func=self.params.get('retry_sleep_functions', {}).get(is_frag or 'http'),
 417             suffix=f'fragment{"s" if frag_index is None else f" {frag_index}"}' if is_frag else None)
 418
 419     def report_unable_to_resume(self):
 420         """Report it was impossible to resume download."""
 421         self.to_screen('[download] Unable to resume')
 422
 423     @staticmethod
 424     def supports_manifest(manifest):
 425         """ Whether the downloader can download the fragments from the manifest.
 426         Redefine in subclasses if needed. """
 427         pass
 428
 429     def download(self, filename, info_dict, subtitle=False):
 430         """Download to a filename using the info from info_dict
 431         Return True on success and False otherwise
 432         """
 433         nooverwrites_and_exists = (
 434             not self.params.get('overwrites', True)
 435             and os.path.exists(encodeFilename(filename))
 436         )
 437
 438         if not hasattr(filename, 'write'):
 439             continuedl_and_exists = (
 440                 self.params.get('continuedl', True)
 441                 and os.path.isfile(encodeFilename(filename))
 442                 and not self.params.get('nopart', False)
 443             )
 444
 445             # Check file already present
 446             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 447                 self.report_file_already_downloaded(filename)
 448                 self._hook_progress({
 449                     'filename': filename,
 450                     'status': 'finished',
 451                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 452                 }, info_dict)
 453                 self._finish_multiline_status()
 454                 return True, False
 455
 456         if subtitle:
 457             sleep_interval = self.params.get('sleep_interval_subtitles') or 0
 458         else:
 459             min_sleep_interval = self.params.get('sleep_interval') or 0
 460             sleep_interval = random.uniform(
 461                 min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
 462         if sleep_interval > 0:
 463             self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
 464             time.sleep(sleep_interval)
 465
 466         ret = self.real_download(filename, info_dict)
 467         self._finish_multiline_status()
 468         return ret, True
 469
 470     def real_download(self, filename, info_dict):
 471         """Real download process. Redefine in subclasses."""
 472         raise NotImplementedError('This method must be implemented by subclasses')
 473
 474     def _hook_progress(self, status, info_dict):
 475         # Ideally we want to make a copy of the dict, but that is too slow
 476         status['info_dict'] = info_dict
 477         # youtube-dl passes the same status object to all the hooks.
 478         # Some third party scripts seems to be relying on this.
 479         # So keep this behavior if possible
 480         for ph in self._progress_hooks:
 481             ph(status)
 482
 483     def add_progress_hook(self, ph):
 484         # See YoutubeDl.py (search for progress_hooks) for a description of
 485         # this interface
 486         self._progress_hooks.append(ph)
 487
 488     def _debug_cmd(self, args, exe=None):
 489         if not self.params.get('verbose', False):
 490             return
 491
 492         str_args = [decodeArgument(a) for a in args]
 493
 494         if exe is None:
 495             exe = os.path.basename(str_args[0])
 496
 497         self.write_debug(f'{exe} command line: {shell_quote(str_args)}')