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