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     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  53     retries:            Number of times to retry for expected network errors.
  54                         Default is 0 for API, but 10 for CLI
  55     file_access_retries:   Number of times to retry on file access error (default: 3)
  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     @classmethod
 142     def calc_eta(cls, start_or_rate, now_or_remaining, total=NO_DEFAULT, current=NO_DEFAULT):
 143         if total is NO_DEFAULT:
 144             rate, remaining = start_or_rate, now_or_remaining
 145             if None in (rate, remaining):
 146                 return None
 147             return int(float(remaining) / rate)
 148
 149         start, now = start_or_rate, now_or_remaining
 150         if total is None:
 151             return None
 152         if now is None:
 153             now = time.time()
 154         rate = cls.calc_speed(start, now, current)
 155         return rate and int((float(total) - float(current)) / rate)
 156
 157     @staticmethod
 158     def calc_speed(start, now, bytes):
 159         dif = now - start
 160         if bytes == 0 or dif < 0.001:  # One millisecond
 161             return None
 162         return float(bytes) / dif
 163
 164     @staticmethod
 165     def format_speed(speed):
 166         return ' Unknown B/s' if speed is None else f'{format_bytes(speed):>10s}/s'
 167
 168     @staticmethod
 169     def format_retries(retries):
 170         return 'inf' if retries == float('inf') else int(retries)
 171
 172     @staticmethod
 173     def filesize_or_none(unencoded_filename):
 174         if os.path.isfile(unencoded_filename):
 175             return os.path.getsize(unencoded_filename)
 176         return 0
 177
 178     @staticmethod
 179     def best_block_size(elapsed_time, bytes):
 180         new_min = max(bytes / 2.0, 1.0)
 181         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 182         if elapsed_time < 0.001:
 183             return int(new_max)
 184         rate = bytes / elapsed_time
 185         if rate > new_max:
 186             return int(new_max)
 187         if rate < new_min:
 188             return int(new_min)
 189         return int(rate)
 190
 191     @staticmethod
 192     def parse_bytes(bytestr):
 193         """Parse a string indicating a byte quantity into an integer."""
 194         deprecation_warning('yt_dlp.FileDownloader.parse_bytes is deprecated and '
 195                             'may be removed in the future. Use yt_dlp.utils.parse_bytes instead')
 196         return parse_bytes(bytestr)
 197
 198     def slow_down(self, start_time, now, byte_counter):
 199         """Sleep if the download speed is over the rate limit."""
 200         rate_limit = self.params.get('ratelimit')
 201         if rate_limit is None or byte_counter == 0:
 202             return
 203         if now is None:
 204             now = time.time()
 205         elapsed = now - start_time
 206         if elapsed <= 0.0:
 207             return
 208         speed = float(byte_counter) / elapsed
 209         if speed > rate_limit:
 210             sleep_time = float(byte_counter) / rate_limit - elapsed
 211             if sleep_time > 0:
 212                 time.sleep(sleep_time)
 213
 214     def temp_name(self, filename):
 215         """Returns a temporary filename for the given filename."""
 216         if self.params.get('nopart', False) or filename == '-' or \
 217                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 218             return filename
 219         return filename + '.part'
 220
 221     def undo_temp_name(self, filename):
 222         if filename.endswith('.part'):
 223             return filename[:-len('.part')]
 224         return filename
 225
 226     def ytdl_filename(self, filename):
 227         return filename + '.ytdl'
 228
 229     def wrap_file_access(action, *, fatal=False):
 230         def error_callback(err, count, retries, *, fd):
 231             return RetryManager.report_retry(
 232                 err, count, retries, info=fd.__to_screen,
 233                 warn=lambda e: (time.sleep(0.01), fd.to_screen(f'[download] Unable to {action} file: {e}')),
 234                 error=None if fatal else lambda e: fd.report_error(f'Unable to {action} file: {e}'),
 235                 sleep_func=fd.params.get('retry_sleep_functions', {}).get('file_access'))
 236
 237         def wrapper(self, func, *args, **kwargs):
 238             for retry in RetryManager(self.params.get('file_access_retries', 3), error_callback, fd=self):
 239                 try:
 240                     return func(self, *args, **kwargs)
 241                 except OSError as err:
 242                     if err.errno in (errno.EACCES, errno.EINVAL):
 243                         retry.error = err
 244                         continue
 245                     retry.error_callback(err, 1, 0)
 246
 247         return functools.partial(functools.partialmethod, wrapper)
 248
 249     @wrap_file_access('open', fatal=True)
 250     def sanitize_open(self, filename, open_mode):
 251         f, filename = sanitize_open(filename, open_mode)
 252         if not getattr(f, 'locked', None):
 253             self.write_debug(f'{LockingUnsupportedError.msg}. Proceeding without locking', only_once=True)
 254         return f, filename
 255
 256     @wrap_file_access('remove')
 257     def try_remove(self, filename):
 258         os.remove(filename)
 259
 260     @wrap_file_access('rename')
 261     def try_rename(self, old_filename, new_filename):
 262         if old_filename == new_filename:
 263             return
 264         os.replace(old_filename, new_filename)
 265
 266     def try_utime(self, filename, last_modified_hdr):
 267         """Try to set the last-modified time of the given file."""
 268         if last_modified_hdr is None:
 269             return
 270         if not os.path.isfile(encodeFilename(filename)):
 271             return
 272         timestr = last_modified_hdr
 273         if timestr is None:
 274             return
 275         filetime = timeconvert(timestr)
 276         if filetime is None:
 277             return filetime
 278         # Ignore obviously invalid dates
 279         if filetime == 0:
 280             return
 281         with contextlib.suppress(Exception):
 282             os.utime(filename, (time.time(), filetime))
 283         return filetime
 284
 285     def report_destination(self, filename):
 286         """Report destination filename."""
 287         self.to_screen('[download] Destination: ' + filename)
 288
 289     def _prepare_multiline_status(self, lines=1):
 290         if self.params.get('noprogress'):
 291             self._multiline = QuietMultilinePrinter()
 292         elif self.ydl.params.get('logger'):
 293             self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
 294         elif self.params.get('progress_with_newline'):
 295             self._multiline = BreaklineStatusPrinter(self.ydl._out_files.out, lines)
 296         else:
 297             self._multiline = MultilinePrinter(self.ydl._out_files.out, lines, not self.params.get('quiet'))
 298         self._multiline.allow_colors = self.ydl._allow_colors.out and self.ydl._allow_colors.out != 'no_color'
 299         self._multiline._HAVE_FULLCAP = self.ydl._allow_colors.out
 300
 301     def _finish_multiline_status(self):
 302         self._multiline.end()
 303
 304     ProgressStyles = Namespace(
 305         downloaded_bytes='light blue',
 306         percent='light blue',
 307         eta='yellow',
 308         speed='green',
 309         elapsed='bold white',
 310         total_bytes='',
 311         total_bytes_estimate='',
 312     )
 313
 314     def _report_progress_status(self, s, default_template):
 315         for name, style in self.ProgressStyles.items_:
 316             name = f'_{name}_str'
 317             if name not in s:
 318                 continue
 319             s[name] = self._format_progress(s[name], style)
 320         s['_default_template'] = default_template % s
 321
 322         progress_dict = s.copy()
 323         progress_dict.pop('info_dict')
 324         progress_dict = {'info': s['info_dict'], 'progress': progress_dict}
 325
 326         progress_template = self.params.get('progress_template', {})
 327         self._multiline.print_at_line(self.ydl.evaluate_outtmpl(
 328             progress_template.get('download') or '[download] %(progress._default_template)s',
 329             progress_dict), s.get('progress_idx') or 0)
 330         self.to_console_title(self.ydl.evaluate_outtmpl(
 331             progress_template.get('download-title') or 'yt-dlp %(progress._default_template)s',
 332             progress_dict))
 333
 334     def _format_progress(self, *args, **kwargs):
 335         return self.ydl._format_text(
 336             self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
 337
 338     def report_progress(self, s):
 339         def with_fields(*tups, default=''):
 340             for *fields, tmpl in tups:
 341                 if all(s.get(f) is not None for f in fields):
 342                     return tmpl
 343             return default
 344
 345         _format_bytes = lambda k: f'{format_bytes(s.get(k)):>10s}'
 346
 347         if s['status'] == 'finished':
 348             if self.params.get('noprogress'):
 349                 self.to_screen('[download] Download completed')
 350             speed = try_call(lambda: s['total_bytes'] / s['elapsed'])
 351             s.update({
 352                 'speed': speed,
 353                 '_speed_str': self.format_speed(speed).strip(),
 354                 '_total_bytes_str': _format_bytes('total_bytes'),
 355                 '_elapsed_str': self.format_seconds(s.get('elapsed')),
 356                 '_percent_str': self.format_percent(100),
 357             })
 358             self._report_progress_status(s, join_nonempty(
 359                 '100%%',
 360                 with_fields(('total_bytes', 'of %(_total_bytes_str)s')),
 361                 with_fields(('elapsed', 'in %(_elapsed_str)s')),
 362                 with_fields(('speed', 'at %(_speed_str)s')),
 363                 delim=' '))
 364
 365         if s['status'] != 'downloading':
 366             return
 367
 368         s.update({
 369             '_eta_str': self.format_eta(s.get('eta')).strip(),
 370             '_speed_str': self.format_speed(s.get('speed')),
 371             '_percent_str': self.format_percent(try_call(
 372                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes'],
 373                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes_estimate'],
 374                 lambda: s['downloaded_bytes'] == 0 and 0)),
 375             '_total_bytes_str': _format_bytes('total_bytes'),
 376             '_total_bytes_estimate_str': _format_bytes('total_bytes_estimate'),
 377             '_downloaded_bytes_str': _format_bytes('downloaded_bytes'),
 378             '_elapsed_str': self.format_seconds(s.get('elapsed')),
 379         })
 380
 381         msg_template = with_fields(
 382             ('total_bytes', '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 383             ('total_bytes_estimate', '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 384             ('downloaded_bytes', 'elapsed', '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'),
 385             ('downloaded_bytes', '%(_downloaded_bytes_str)s at %(_speed_str)s'),
 386             default='%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s')
 387
 388         msg_template += with_fields(
 389             ('fragment_index', 'fragment_count', ' (frag %(fragment_index)s/%(fragment_count)s)'),
 390             ('fragment_index', ' (frag %(fragment_index)s)'))
 391         self._report_progress_status(s, msg_template)
 392
 393     def report_resuming_byte(self, resume_len):
 394         """Report attempt to resume at given byte."""
 395         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 396
 397     def report_retry(self, err, count, retries, frag_index=NO_DEFAULT, fatal=True):
 398         """Report retry"""
 399         is_frag = False if frag_index is NO_DEFAULT else 'fragment'
 400         RetryManager.report_retry(
 401             err, count, retries, info=self.__to_screen,
 402             warn=lambda msg: self.__to_screen(f'[download] Got error: {msg}'),
 403             error=IDENTITY if not fatal else lambda e: self.report_error(f'\r[download] Got error: {e}'),
 404             sleep_func=self.params.get('retry_sleep_functions', {}).get(is_frag or 'http'),
 405             suffix=f'fragment{"s" if frag_index is None else f" {frag_index}"}' if is_frag else None)
 406
 407     def report_unable_to_resume(self):
 408         """Report it was impossible to resume download."""
 409         self.to_screen('[download] Unable to resume')
 410
 411     @staticmethod
 412     def supports_manifest(manifest):
 413         """ Whether the downloader can download the fragments from the manifest.
 414         Redefine in subclasses if needed. """
 415         pass
 416
 417     def download(self, filename, info_dict, subtitle=False):
 418         """Download to a filename using the info from info_dict
 419         Return True on success and False otherwise
 420         """
 421
 422         nooverwrites_and_exists = (
 423             not self.params.get('overwrites', True)
 424             and os.path.exists(encodeFilename(filename))
 425         )
 426
 427         if not hasattr(filename, 'write'):
 428             continuedl_and_exists = (
 429                 self.params.get('continuedl', True)
 430                 and os.path.isfile(encodeFilename(filename))
 431                 and not self.params.get('nopart', False)
 432             )
 433
 434             # Check file already present
 435             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 436                 self.report_file_already_downloaded(filename)
 437                 self._hook_progress({
 438                     'filename': filename,
 439                     'status': 'finished',
 440                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 441                 }, info_dict)
 442                 self._finish_multiline_status()
 443                 return True, False
 444
 445         if subtitle:
 446             sleep_interval = self.params.get('sleep_interval_subtitles') or 0
 447         else:
 448             min_sleep_interval = self.params.get('sleep_interval') or 0
 449             sleep_interval = random.uniform(
 450                 min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
 451         if sleep_interval > 0:
 452             self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
 453             time.sleep(sleep_interval)
 454
 455         ret = self.real_download(filename, info_dict)
 456         self._finish_multiline_status()
 457         return ret, True
 458
 459     def real_download(self, filename, info_dict):
 460         """Real download process. Redefine in subclasses."""
 461         raise NotImplementedError('This method must be implemented by subclasses')
 462
 463     def _hook_progress(self, status, info_dict):
 464         # Ideally we want to make a copy of the dict, but that is too slow
 465         status['info_dict'] = info_dict
 466         # youtube-dl passes the same status object to all the hooks.
 467         # Some third party scripts seems to be relying on this.
 468         # So keep this behavior if possible
 469         for ph in self._progress_hooks:
 470             ph(status)
 471
 472     def add_progress_hook(self, ph):
 473         # See YoutubeDl.py (search for progress_hooks) for a description of
 474         # this interface
 475         self._progress_hooks.append(ph)
 476
 477     def _debug_cmd(self, args, exe=None):
 478         if not self.params.get('verbose', False):
 479             return
 480
 481         str_args = [decodeArgument(a) for a in args]
 482
 483         if exe is None:
 484             exe = os.path.basename(str_args[0])
 485
 486         self.write_debug(f'{exe} command line: {shell_quote(str_args)}')