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     sanitize_open,
  28     shell_quote,
  29     timeconvert,
  30     timetuple_from_msec,
  31     try_call,
  32 )
  33
  34
  35 class FileDownloader:
  36     """File Downloader class.
  37
  38     File downloader objects are the ones responsible of downloading the
  39     actual video file and writing it to disk.
  40
  41     File downloaders accept a lot of parameters. In order not to saturate
  42     the object constructor with arguments, it receives a dictionary of
  43     options instead.
  44
  45     Available options:
  46
  47     verbose:            Print additional info to stdout.
  48     quiet:              Do not print messages to stdout.
  49     ratelimit:          Download speed limit, in bytes/sec.
  50     continuedl:         Attempt to continue downloads if possible
  51     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
  52     retries:            Number of times to retry for HTTP error 5xx
  53     file_access_retries:   Number of times to retry on file access error
  54     buffersize:         Size of download buffer in bytes.
  55     noresizebuffer:     Do not automatically resize the download buffer.
  56     continuedl:         Try to continue downloads if possible.
  57     noprogress:         Do not print the progress bar.
  58     nopart:             Do not use temporary .part files.
  59     updatetime:         Use the Last-modified header to set output file timestamps.
  60     test:               Download only first bytes to test the downloader.
  61     min_filesize:       Skip files smaller than this size
  62     max_filesize:       Skip files larger than this size
  63     xattr_set_filesize: Set ytdl.filesize user xattribute with expected size.
  64     external_downloader_args:  A dictionary of downloader keys (in lower case)
  65                         and a list of additional command-line arguments for the
  66                         executable. Use 'default' as the name for arguments to be
  67                         passed to all downloaders. For compatibility with youtube-dl,
  68                         a single list of args can also be used
  69     hls_use_mpegts:     Use the mpegts container for HLS videos.
  70     http_chunk_size:    Size of a chunk for chunk-based HTTP downloading. May be
  71                         useful for bypassing bandwidth throttling imposed by
  72                         a webserver (experimental)
  73     progress_template:  See YoutubeDL.py
  74     retry_sleep_functions: See YoutubeDL.py
  75
  76     Subclasses of this one must re-define the real_download method.
  77     """
  78
  79     _TEST_FILE_SIZE = 10241
  80     params = None
  81
  82     def __init__(self, ydl, params):
  83         """Create a FileDownloader object with the given options."""
  84         self._set_ydl(ydl)
  85         self._progress_hooks = []
  86         self.params = params
  87         self._prepare_multiline_status()
  88         self.add_progress_hook(self.report_progress)
  89
  90     def _set_ydl(self, ydl):
  91         self.ydl = ydl
  92
  93         for func in (
  94             'deprecation_warning',
  95             'report_error',
  96             'report_file_already_downloaded',
  97             'report_warning',
  98             'to_console_title',
  99             'to_stderr',
 100             'trouble',
 101             'write_debug',
 102         ):
 103             if not hasattr(self, func):
 104                 setattr(self, func, getattr(ydl, func))
 105
 106     def to_screen(self, *args, **kargs):
 107         self.ydl.to_screen(*args, quiet=self.params.get('quiet'), **kargs)
 108
 109     __to_screen = to_screen
 110
 111     @classproperty
 112     def FD_NAME(cls):
 113         return re.sub(r'(?<=[a-z])(?=[A-Z])', '_', cls.__name__[:-2]).lower()
 114
 115     @staticmethod
 116     def format_seconds(seconds):
 117         if seconds is None:
 118             return ' Unknown'
 119         time = timetuple_from_msec(seconds * 1000)
 120         if time.hours > 99:
 121             return '--:--:--'
 122         if not time.hours:
 123             return '%02d:%02d' % time[1:-1]
 124         return '%02d:%02d:%02d' % time[:-1]
 125
 126     format_eta = format_seconds
 127
 128     @staticmethod
 129     def calc_percent(byte_counter, data_len):
 130         if data_len is None:
 131             return None
 132         return float(byte_counter) / float(data_len) * 100.0
 133
 134     @staticmethod
 135     def format_percent(percent):
 136         return '  N/A%' if percent is None else f'{percent:>5.1f}%'
 137
 138     @staticmethod
 139     def calc_eta(start, now, total, current):
 140         if total is None:
 141             return None
 142         if now is None:
 143             now = time.time()
 144         dif = now - start
 145         if current == 0 or dif < 0.001:  # One millisecond
 146             return None
 147         rate = float(current) / dif
 148         return int((float(total) - float(current)) / rate)
 149
 150     @staticmethod
 151     def calc_speed(start, now, bytes):
 152         dif = now - start
 153         if bytes == 0 or dif < 0.001:  # One millisecond
 154             return None
 155         return float(bytes) / dif
 156
 157     @staticmethod
 158     def format_speed(speed):
 159         return ' Unknown B/s' if speed is None else f'{format_bytes(speed):>10s}/s'
 160
 161     @staticmethod
 162     def format_retries(retries):
 163         return 'inf' if retries == float('inf') else int(retries)
 164
 165     @staticmethod
 166     def best_block_size(elapsed_time, bytes):
 167         new_min = max(bytes / 2.0, 1.0)
 168         new_max = min(max(bytes * 2.0, 1.0), 4194304)  # Do not surpass 4 MB
 169         if elapsed_time < 0.001:
 170             return int(new_max)
 171         rate = bytes / elapsed_time
 172         if rate > new_max:
 173             return int(new_max)
 174         if rate < new_min:
 175             return int(new_min)
 176         return int(rate)
 177
 178     @staticmethod
 179     def parse_bytes(bytestr):
 180         """Parse a string indicating a byte quantity into an integer."""
 181         matchobj = re.match(rf'(?i)^({NUMBER_RE})([kMGTPEZY]?)$', bytestr)
 182         if matchobj is None:
 183             return None
 184         number = float(matchobj.group(1))
 185         multiplier = 1024.0 ** 'bkmgtpezy'.index(matchobj.group(2).lower())
 186         return int(round(number * multiplier))
 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         if s['status'] == 'finished':
 335             if self.params.get('noprogress'):
 336                 self.to_screen('[download] Download completed')
 337             speed = try_call(lambda: s['total_bytes'] / s['elapsed'])
 338             s.update({
 339                 'speed': speed,
 340                 '_speed_str': self.format_speed(speed).strip(),
 341                 '_total_bytes_str': format_bytes(s.get('total_bytes')),
 342                 '_elapsed_str': self.format_seconds(s.get('elapsed')),
 343                 '_percent_str': self.format_percent(100),
 344             })
 345             self._report_progress_status(s, join_nonempty(
 346                 '100%%',
 347                 with_fields(('total_bytes', 'of %(_total_bytes_str)s')),
 348                 with_fields(('elapsed', 'in %(_elapsed_str)s')),
 349                 with_fields(('speed', 'at %(_speed_str)s')),
 350                 delim=' '))
 351
 352         if s['status'] != 'downloading':
 353             return
 354
 355         s.update({
 356             '_eta_str': self.format_eta(s.get('eta')),
 357             '_speed_str': self.format_speed(s.get('speed')),
 358             '_percent_str': self.format_percent(try_call(
 359                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes'],
 360                 lambda: 100 * s['downloaded_bytes'] / s['total_bytes_estimate'],
 361                 lambda: s['downloaded_bytes'] == 0 and 0)),
 362             '_total_bytes_str': format_bytes(s.get('total_bytes')),
 363             '_total_bytes_estimate_str': format_bytes(s.get('total_bytes_estimate')),
 364             '_downloaded_bytes_str': format_bytes(s.get('downloaded_bytes')),
 365             '_elapsed_str': self.format_seconds(s.get('elapsed')),
 366         })
 367
 368         msg_template = with_fields(
 369             ('total_bytes', '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 370             ('total_bytes_estimate', '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'),
 371             ('downloaded_bytes', 'elapsed', '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'),
 372             ('downloaded_bytes', '%(_downloaded_bytes_str)s at %(_speed_str)s'),
 373             default='%(_percent_str)s at %(_speed_str)s ETA %(_eta_str)s')
 374
 375         msg_template += with_fields(
 376             ('fragment_index', 'fragment_count', ' (frag %(fragment_index)s/%(fragment_count)s)'),
 377             ('fragment_index', ' (frag %(fragment_index)s)'))
 378         self._report_progress_status(s, msg_template)
 379
 380     def report_resuming_byte(self, resume_len):
 381         """Report attempt to resume at given byte."""
 382         self.to_screen('[download] Resuming download at byte %s' % resume_len)
 383
 384     def report_retry(self, err, count, retries, frag_index=NO_DEFAULT, fatal=True):
 385         """Report retry"""
 386         is_frag = False if frag_index is NO_DEFAULT else 'fragment'
 387         RetryManager.report_retry(
 388             err, count, retries, info=self.__to_screen,
 389             warn=lambda msg: self.__to_screen(f'[download] Got error: {msg}'),
 390             error=IDENTITY if not fatal else lambda e: self.report_error(f'\r[download] Got error: {e}'),
 391             sleep_func=self.params.get('retry_sleep_functions', {}).get(is_frag or 'http'),
 392             suffix=f'fragment{"s" if frag_index is None else f" {frag_index}"}' if is_frag else None)
 393
 394     def report_unable_to_resume(self):
 395         """Report it was impossible to resume download."""
 396         self.to_screen('[download] Unable to resume')
 397
 398     @staticmethod
 399     def supports_manifest(manifest):
 400         """ Whether the downloader can download the fragments from the manifest.
 401         Redefine in subclasses if needed. """
 402         pass
 403
 404     def download(self, filename, info_dict, subtitle=False):
 405         """Download to a filename using the info from info_dict
 406         Return True on success and False otherwise
 407         """
 408
 409         nooverwrites_and_exists = (
 410             not self.params.get('overwrites', True)
 411             and os.path.exists(encodeFilename(filename))
 412         )
 413
 414         if not hasattr(filename, 'write'):
 415             continuedl_and_exists = (
 416                 self.params.get('continuedl', True)
 417                 and os.path.isfile(encodeFilename(filename))
 418                 and not self.params.get('nopart', False)
 419             )
 420
 421             # Check file already present
 422             if filename != '-' and (nooverwrites_and_exists or continuedl_and_exists):
 423                 self.report_file_already_downloaded(filename)
 424                 self._hook_progress({
 425                     'filename': filename,
 426                     'status': 'finished',
 427                     'total_bytes': os.path.getsize(encodeFilename(filename)),
 428                 }, info_dict)
 429                 self._finish_multiline_status()
 430                 return True, False
 431
 432         if subtitle:
 433             sleep_interval = self.params.get('sleep_interval_subtitles') or 0
 434         else:
 435             min_sleep_interval = self.params.get('sleep_interval') or 0
 436             sleep_interval = random.uniform(
 437                 min_sleep_interval, self.params.get('max_sleep_interval') or min_sleep_interval)
 438         if sleep_interval > 0:
 439             self.to_screen(f'[download] Sleeping {sleep_interval:.2f} seconds ...')
 440             time.sleep(sleep_interval)
 441
 442         ret = self.real_download(filename, info_dict)
 443         self._finish_multiline_status()
 444         return ret, True
 445
 446     def real_download(self, filename, info_dict):
 447         """Real download process. Redefine in subclasses."""
 448         raise NotImplementedError('This method must be implemented by subclasses')
 449
 450     def _hook_progress(self, status, info_dict):
 451         # Ideally we want to make a copy of the dict, but that is too slow
 452         status['info_dict'] = info_dict
 453         # youtube-dl passes the same status object to all the hooks.
 454         # Some third party scripts seems to be relying on this.
 455         # So keep this behavior if possible
 456         for ph in self._progress_hooks:
 457             ph(status)
 458
 459     def add_progress_hook(self, ph):
 460         # See YoutubeDl.py (search for progress_hooks) for a description of
 461         # this interface
 462         self._progress_hooks.append(ph)
 463
 464     def _debug_cmd(self, args, exe=None):
 465         if not self.params.get('verbose', False):
 466             return
 467
 468         str_args = [decodeArgument(a) for a in args]
 469
 470         if exe is None:
 471             exe = os.path.basename(str_args[0])
 472
 473         self.write_debug(f'{exe} command line: {shell_quote(str_args)}')