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     NUMBER_RE,
  19     LockingUnsupportedError,
  20     Namespace,
  21     RetryManager,
  22     classproperty,
  23     decodeArgument,
  24     encodeFilename,
  25     format_bytes,
  26     join_nonempty,
  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     continuedl:         Attempt to continue downloads if possible
  52     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  53     retries:            Number of times to retry for HTTP error 5xx
  54     file_access_retries:   Number of times to retry on file access error
  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     external_downloader_args:  A dictionary of downloader keys (in lower case)
  66                         and a list of additional command-line arguments for the
  67                         executable. Use 'default' as the name for arguments to be
  68                         passed to all downloaders. For compatibility with youtube-dl,
  69                         a single list of args can also be used
  70     hls_use_mpegts:     Use the mpegts container for HLS videos.
  71     http_chunk_size:    Size of a chunk for chunk-based HTTP downloading. May be
  72                         useful for bypassing bandwidth throttling imposed by
  73                         a webserver (experimental)
  74     progress_template:  See YoutubeDL.py
  75     retry_sleep_functions: See YoutubeDL.py
  76
  77     Subclasses of this one must re-define the real_download method.
  78     """
  79
  80     _TEST_FILE_SIZE = 10241
  81     params = None
  82
  83     def __init__(self, ydl, params):
  84         """Create a FileDownloader object with the given options."""
  85         self._set_ydl(ydl)
  86         self._progress_hooks = []
  87         self.params = params
  88         self._prepare_multiline_status()
  89         self.add_progress_hook(self.report_progress)
  90
  91     def _set_ydl(self, ydl):
  92         self.ydl = ydl
  93
  94         for func in (
  95             'deprecation_warning',
  96             'deprecated_feature',
  97             'report_error',
  98             'report_file_already_downloaded',
  99             'report_warning',
 100             'to_console_title',
 101             'to_stderr',
 102             'trouble',
 103             'write_debug',
 104         ):
 105             if not hasattr(self, func):
 106                 setattr(self, func, getattr(ydl, func))
 107
 108     def to_screen(self, *args, **kargs):
 109         self.ydl.to_screen(*args, quiet=self.params.get('quiet'), **kargs)
 110
 111     __to_screen = to_screen
 112
 113     @classproperty
 114     def FD_NAME(cls):
 115         return re.sub(r'(?<=[a-z])(?=[A-Z])', '_', cls.__name__[:-2]).lower()
 116
 117     @staticmethod
 118     def format_seconds(seconds):
 119         if seconds is None:
 120             return ' Unknown'
 121         time = timetuple_from_msec(seconds * 1000)
 122         if time.hours > 99:
 123             return '--:--:--'
 124         return '%02d:%02d:%02d' % time[:-1]
 125
 126     @classmethod
 127     def format_eta(cls, seconds):
 128         return f'{remove_start(cls.format_seconds(seconds), "00:"):>8s}'
 129
 130     @staticmethod
 131     def calc_percent(byte_counter, data_len):
 132         if data_len is None:
 133             return None
 134         return float(byte_counter) / float(data_len) * 100.0
 135
 136     @staticmethod
 137     def format_percent(percent):
 138         return '  N/A%' if percent is None else f'{percent:>5.1f}%'
 139
 140     @staticmethod
 141     def calc_eta(start, now, total, current):
 142         if total is None:
 143             return None
 144         if now is None:
 145             now = time.time()
 146         dif = now - start
 147         if current == 0 or dif < 0.001:  # One millisecond
 148             return None
 149         rate = float(current) / dif
 150         return int((float(total) - float(current)) / rate)
 151
 152     @staticmethod
 153     def calc_speed(start, now, bytes):
 154         dif = now - start
 155         if bytes == 0 or dif < 0.001:  # One millisecond
 156             return None
 157         return float(bytes) / dif
 158
 159     @staticmethod
 160     def format_speed(speed):
 161         return ' Unknown B/s' if speed is None else f'{format_bytes(speed):>10s}/s'
 162
 163     @staticmethod
 164     def format_retries(retries):
 165         return 'inf' if retries == float('inf') else int(retries)
 166
 167     @staticmethod
 168     def best_block_size(elapsed_time, bytes):
 169         new_min = max(bytes / 2.0, 1.0)
 170         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 171         if elapsed_time < 0.001:
 172             return int(new_max)
 173         rate = bytes / elapsed_time
 174         if rate > new_max:
 175             return int(new_max)
 176         if rate < new_min:
 177             return int(new_min)
 178         return int(rate)
 179
 180     @staticmethod
 181     def parse_bytes(bytestr):
 182         """Parse a string indicating a byte quantity into an integer."""
 183         matchobj = re.match(rf'(?i)^({NUMBER_RE})([kMGTPEZY]?)$', bytestr)
 184         if matchobj is None:
 185             return None
 186         number = float(matchobj.group(1))
 187         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 188         return int(round(number * multiplier))
 189
 190     def slow_down(self, start_time, now, byte_counter):
 191         """Sleep if the download speed is over the rate limit."""
 192         rate_limit = self.params.get('ratelimit')
 193         if rate_limit is None or byte_counter == 0:
 194             return
 195         if now is None:
 196             now = time.time()
 197         elapsed = now - start_time
 198         if elapsed <= 0.0:
 199             return
 200         speed = float(byte_counter) / elapsed
 201         if speed > rate_limit:
 202             sleep_time = float(byte_counter) / rate_limit - elapsed
 203             if sleep_time > 0:
 204                 time.sleep(sleep_time)
 205
 206     def temp_name(self, filename):
 207         """Returns a temporary filename for the given filename."""
 208         if self.params.get('nopart', False) or filename == '-' or \
 209                 (os.path.exists(encodeFilename(filename)) and not os.path.isfile(encodeFilename(filename))):
 210             return filename
 211         return filename + '.part'
 212
 213     def undo_temp_name(self, filename):
 214         if filename.endswith('.part'):
 215             return filename[:-len('.part')]
 216         return filename
 217
 218     def ytdl_filename(self, filename):
 219         return filename + '.ytdl'
 220
 221     def wrap_file_access(action, *, fatal=False):
 222         def error_callback(err, count, retries, *, fd):
 223             return RetryManager.report_retry(
 224                 err, count, retries, info=fd.__to_screen,
 225                 warn=lambda e: (time.sleep(0.01), fd.to_screen(f'[download] Unable to {action} file: {e}')),
 226                 error=None if fatal else lambda e: fd.report_error(f'Unable to {action} file: {e}'),
 227                 sleep_func=fd.params.get('retry_sleep_functions', {}).get('file_access'))
 228
 229         def wrapper(self, func, *args, **kwargs):
 230             for retry in RetryManager(self.params.get('file_access_retries'), error_callback, fd=self):
 231                 try:
 232                     return func(self, *args, **kwargs)
 233                 except OSError as err:
 234                     if err.errno in (errno.EACCES, errno.EINVAL):
 235                         retry.error = err
 236                         continue
 237                     retry.error_callback(err, 1, 0)
 238
 239         return functools.partial(functools.partialmethod, wrapper)
 240
 241     @wrap_file_access('open', fatal=True)
 242     def sanitize_open(self, filename, open_mode):
 243         f, filename = sanitize_open(filename, open_mode)
 244         if not getattr(f, 'locked', None):
 245             self.write_debug(f'{LockingUnsupportedError.msg}. Proceeding without locking', only_once=True)
 246         return f, filename
 247
 248     @wrap_file_access('remove')
 249     def try_remove(self, filename):
 250         os.remove(filename)
 251
 252     @wrap_file_access('rename')
 253     def try_rename(self, old_filename, new_filename):
 254         if old_filename == new_filename:
 255             return
 256         os.replace(old_filename, new_filename)
 257
 258     def try_utime(self, filename, last_modified_hdr):
 259         """Try to set the last-modified time of the given file."""
 260         if last_modified_hdr is None:
 261             return
 262         if not os.path.isfile(encodeFilename(filename)):
 263             return
 264         timestr = last_modified_hdr
 265         if timestr is None:
 266             return
 267         filetime = timeconvert(timestr)
 268         if filetime is None:
 269             return filetime
 270         # Ignore obviously invalid dates
 271         if filetime == 0:
 272             return
 273         with contextlib.suppress(Exception):
 274             os.utime(filename, (time.time(), filetime))
 275         return filetime
 276
 277     def report_destination(self, filename):
 278         """Report destination filename."""
 279         self.to_screen('[download] Destination: ' + filename)
 280
 281     def _prepare_multiline_status(self, lines=1):
 282         if self.params.get('noprogress'):
 283             self._multiline = QuietMultilinePrinter()
 284         elif self.ydl.params.get('logger'):
 285             self._multiline = MultilineLogger(self.ydl.params['logger'], lines)
 286         elif self.params.get('progress_with_newline'):
 287             self._multiline = BreaklineStatusPrinter(self.ydl._out_files.out, lines)
 288         else:
 289             self._multiline = MultilinePrinter(self.ydl._out_files.out, lines, not self.params.get('quiet'))
 290         self._multiline.allow_colors = self._multiline._HAVE_FULLCAP and not self.params.get('no_color')
 291
 292     def _finish_multiline_status(self):
 293         self._multiline.end()
 294
 295     ProgressStyles = Namespace(
 296         downloaded_bytes='light blue',
 297         percent='light blue',
 298         eta='yellow',
 299         speed='green',
 300         elapsed='bold white',
 301         total_bytes='',
 302         total_bytes_estimate='',
 303     )
 304
 305     def _report_progress_status(self, s, default_template):
 306         for name, style in self.ProgressStyles.items_:
 307             name = f'_{name}_str'
 308             if name not in s:
 309                 continue
 310             s[name] = self._format_progress(s[name], style)
 311         s['_default_template'] = default_template % s
 312
 313         progress_dict = s.copy()
 314         progress_dict.pop('info_dict')
 315         progress_dict = {'info': s['info_dict'], 'progress': progress_dict}
 316
 317         progress_template = self.params.get('progress_template', {})
 318         self._multiline.print_at_line(self.ydl.evaluate_outtmpl(
 319             progress_template.get('download') or '[download] %(progress._default_template)s',
 320             progress_dict), s.get('progress_idx') or 0)
 321         self.to_console_title(self.ydl.evaluate_outtmpl(
 322             progress_template.get('download-title') or 'yt-dlp %(progress._default_template)s',
 323             progress_dict))
 324
 325     def _format_progress(self, *args, **kwargs):
 326         return self.ydl._format_text(
 327             self._multiline.stream, self._multiline.allow_colors, *args, **kwargs)
 328
 329     def report_progress(self, s):
 330         def with_fields(*tups, default=''):
 331             for *fields, tmpl in tups:
 332                 if all(s.get(f) is not None for f in fields):
 333                     return tmpl
 334             return default
 335
 336         _formats_bytes = lambda k: f'{format_bytes(s.get(k)):>10s}'
 337
 338         if s['status'] == 'finished':
 339             if self.params.get('noprogress'):
 340                 self.to_screen('[download] Download completed')
 341             speed = try_call(lambda: s['total_bytes'] / s['elapsed'])
 342             s.update({
 343                 'speed': speed,
 344                 '_speed_str': self.format_speed(speed).strip(),
 345                 '_total_bytes_str': _formats_bytes('total_bytes'),
 346                 '_elapsed_str': self.format_seconds(s.get('elapsed')),
 347                 '_percent_str': self.format_percent(100),
 348             })
 349             self._report_progress_status(s, join_nonempty(
 350                 '100%%',
 351                 with_fields(('total_bytes', 'of %(_total_bytes_str)s')),
 352                 with_fields(('elapsed', 'in %(_elapsed_str)s')),
 353                 with_fields(('speed', 'at %(_speed_str)s')),
 354                 delim=' '))
 355
 356         if s['status'] != 'downloading':
 357             return
 358
 359         s.update({
 360             '_eta_str': self.format_eta(s.get('eta')).strip(),
 361             '_speed_str': self.format_speed(s.get('speed')),
 362             '_percent_str': self.format_percent(try_call(
 363                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes'],
 364                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes_estimate'],
 365                 lambda: s['downloaded_bytes'] == 0 and 0)),
 366             '_total_bytes_str': _formats_bytes('total_bytes'),
 367             '_total_bytes_estimate_str': _formats_bytes('total_bytes_estimate'),
 368             '_downloaded_bytes_str': _formats_bytes('downloaded_bytes'),
 369             '_elapsed_str': self.format_seconds(s.get('elapsed')),
 370         })
 371
 372         msg_template = with_fields(
 373             ('total_bytes', '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 374             ('total_bytes_estimate', '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 375             ('downloaded_bytes', 'elapsed', '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'),
 376             ('downloaded_bytes', '%(_downloaded_bytes_str)s at %(_speed_str)s'),
 377             default='%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s')
 378
 379         msg_template += with_fields(
 380             ('fragment_index', 'fragment_count', ' (frag %(fragment_index)s/%(fragment_count)s)'),
 381             ('fragment_index', ' (frag %(fragment_index)s)'))
 382         self._report_progress_status(s, msg_template)
 383
 384     def report_resuming_byte(self, resume_len):
 385         """Report attempt to resume at given byte."""
 386         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 387
 388     def report_retry(self, err, count, retries, frag_index=NO_DEFAULT, fatal=True):
 389         """Report retry"""
 390         is_frag = False if frag_index is NO_DEFAULT else 'fragment'
 391         RetryManager.report_retry(
 392             err, count, retries, info=self.__to_screen,
 393             warn=lambda msg: self.__to_screen(f'[download] Got error: {msg}'),
 394             error=IDENTITY if not fatal else lambda e: self.report_error(f'\r[download] Got error: {e}'),
 395             sleep_func=self.params.get('retry_sleep_functions', {}).get(is_frag or 'http'),
 396             suffix=f'fragment{"s" if frag_index is None else f" {frag_index}"}' if is_frag else None)
 397
 398     def report_unable_to_resume(self):
 399         """Report it was impossible to resume download."""
 400         self.to_screen('[download] Unable to resume')
 401
 402     @staticmethod
 403     def supports_manifest(manifest):
 404         """ Whether the downloader can download the fragments from the manifest.
 405         Redefine in subclasses if needed. """
 406         pass
 407
 408     def download(self, filename, info_dict, subtitle=False):
 409         """Download to a filename using the info from info_dict
 410         Return True on success and False otherwise
 411         """
 412
 413         nooverwrites_and_exists = (
 414             not self.params.get('overwrites', True)
 415             and os.path.exists(encodeFilename(filename))
 416         )
 417
 418         if not hasattr(filename, 'write'):
 419             continuedl_and_exists = (
 420                 self.params.get('continuedl', True)
 421                 and os.path.isfile(encodeFilename(filename))
 422                 and not self.params.get('nopart', False)
 423             )
 424
 425             # Check file already present
 426             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 427                 self.report_file_already_downloaded(filename)
 428                 self._hook_progress({
 429                     'filename': filename,
 430                     'status': 'finished',
 431                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 432                 }, info_dict)
 433                 self._finish_multiline_status()
 434                 return True, False
 435
 436         if subtitle:
 437             sleep_interval = self.params.get('sleep_interval_subtitles') or 0
 438         else:
 439             min_sleep_interval = self.params.get('sleep_interval') or 0
 440             sleep_interval = random.uniform(
 441                 min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
 442         if sleep_interval > 0:
 443             self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
 444             time.sleep(sleep_interval)
 445
 446         ret = self.real_download(filename, info_dict)
 447         self._finish_multiline_status()
 448         return ret, True
 449
 450     def real_download(self, filename, info_dict):
 451         """Real download process. Redefine in subclasses."""
 452         raise NotImplementedError('This method must be implemented by subclasses')
 453
 454     def _hook_progress(self, status, info_dict):
 455         # Ideally we want to make a copy of the dict, but that is too slow
 456         status['info_dict'] = info_dict
 457         # youtube-dl passes the same status object to all the hooks.
 458         # Some third party scripts seems to be relying on this.
 459         # So keep this behavior if possible
 460         for ph in self._progress_hooks:
 461             ph(status)
 462
 463     def add_progress_hook(self, ph):
 464         # See YoutubeDl.py (search for progress_hooks) for a description of
 465         # this interface
 466         self._progress_hooks.append(ph)
 467
 468     def _debug_cmd(self, args, exe=None):
 469         if not self.params.get('verbose', False):
 470             return
 471
 472         str_args = [decodeArgument(a) for a in args]
 473
 474         if exe is None:
 475             exe = os.path.basename(str_args[0])
 476
 477         self.write_debug(f'{exe} command line: {shell_quote(str_args)}')