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