yt_dlp/downloader/external.py

   1 import enum
   2 import os.path
   3 import re
   4 import subprocess
   5 import sys
   6 import time
   7
   8 from .fragment import FragmentFD
   9 from ..compat import functools
  10 from ..postprocessor.ffmpeg import EXT_TO_OUT_FORMATS, FFmpegPostProcessor
  11 from ..utils import (
  12     Popen,
  13     _configuration_args,
  14     check_executable,
  15     classproperty,
  16     cli_bool_option,
  17     cli_option,
  18     cli_valueless_option,
  19     determine_ext,
  20     encodeArgument,
  21     encodeFilename,
  22     handle_youtubedl_headers,
  23     remove_end,
  24     traverse_obj,
  25 )
  26
  27
  28 class Features(enum.Enum):
  29     TO_STDOUT = enum.auto()
  30     MULTIPLE_FORMATS = enum.auto()
  31
  32
  33 class ExternalFD(FragmentFD):
  34     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps')
  35     SUPPORTED_FEATURES = ()
  36     _CAPTURE_STDERR = True
  37
  38     def real_download(self, filename, info_dict):
  39         self.report_destination(filename)
  40         tmpfilename = self.temp_name(filename)
  41
  42         try:
  43             started = time.time()
  44             retval = self._call_downloader(tmpfilename, info_dict)
  45         except KeyboardInterrupt:
  46             if not info_dict.get('is_live'):
  47                 raise
  48             # Live stream downloading cancellation should be considered as
  49             # correct and expected termination thus all postprocessing
  50             # should take place
  51             retval = 0
  52             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  53
  54         if retval == 0:
  55             status = {
  56                 'filename': filename,
  57                 'status': 'finished',
  58                 'elapsed': time.time() - started,
  59             }
  60             if filename != '-':
  61                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  62                 self.to_screen(f'\r[{self.get_basename()}] Downloaded {fsize} bytes')
  63                 self.try_rename(tmpfilename, filename)
  64                 status.update({
  65                     'downloaded_bytes': fsize,
  66                     'total_bytes': fsize,
  67                 })
  68             self._hook_progress(status, info_dict)
  69             return True
  70         else:
  71             self.to_stderr('\n')
  72             self.report_error('%s exited with code %d' % (
  73                 self.get_basename(), retval))
  74             return False
  75
  76     @classmethod
  77     def get_basename(cls):
  78         return cls.__name__[:-2].lower()
  79
  80     @classproperty
  81     def EXE_NAME(cls):
  82         return cls.get_basename()
  83
  84     @functools.cached_property
  85     def exe(self):
  86         return self.EXE_NAME
  87
  88     @classmethod
  89     def available(cls, path=None):
  90         path = check_executable(
  91             cls.EXE_NAME if path in (None, cls.get_basename()) else path,
  92             [cls.AVAILABLE_OPT])
  93         if not path:
  94             return False
  95         cls.exe = path
  96         return path
  97
  98     @classmethod
  99     def supports(cls, info_dict):
 100         return all((
 101             not info_dict.get('to_stdout') or Features.TO_STDOUT in cls.SUPPORTED_FEATURES,
 102             '+' not in info_dict['protocol'] or Features.MULTIPLE_FORMATS in cls.SUPPORTED_FEATURES,
 103             all(proto in cls.SUPPORTED_PROTOCOLS for proto in info_dict['protocol'].split('+')),
 104         ))
 105
 106     @classmethod
 107     def can_download(cls, info_dict, path=None):
 108         return cls.available(path) and cls.supports(info_dict)
 109
 110     def _option(self, command_option, param):
 111         return cli_option(self.params, command_option, param)
 112
 113     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
 114         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
 115
 116     def _valueless_option(self, command_option, param, expected_value=True):
 117         return cli_valueless_option(self.params, command_option, param, expected_value)
 118
 119     def _configuration_args(self, keys=None, *args, **kwargs):
 120         return _configuration_args(
 121             self.get_basename(), self.params.get('external_downloader_args'), self.EXE_NAME,
 122             keys, *args, **kwargs)
 123
 124     def _call_downloader(self, tmpfilename, info_dict):
 125         """ Either overwrite this or implement _make_cmd """
 126         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 127
 128         self._debug_cmd(cmd)
 129
 130         if 'fragments' not in info_dict:
 131             _, stderr, returncode = Popen.run(
 132                 cmd, text=True, stderr=subprocess.PIPE if self._CAPTURE_STDERR else None)
 133             if returncode and stderr:
 134                 self.to_stderr(stderr)
 135             return returncode
 136
 137         fragment_retries = self.params.get('fragment_retries', 0)
 138         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 139
 140         count = 0
 141         while count <= fragment_retries:
 142             _, stderr, returncode = Popen.run(cmd, text=True, stderr=subprocess.PIPE)
 143             if not returncode:
 144                 break
 145
 146             # TODO: Decide whether to retry based on error code
 147             # https://aria2.github.io/manual/en/html/aria2c.html#exit-status
 148             if stderr:
 149                 self.to_stderr(stderr)
 150             count += 1
 151             if count <= fragment_retries:
 152                 self.to_screen(
 153                     '[%s] Got error. Retrying fragments (attempt %d of %s)...'
 154                     % (self.get_basename(), count, self.format_retries(fragment_retries)))
 155                 self.sleep_retry('fragment', count)
 156         if count > fragment_retries:
 157             if not skip_unavailable_fragments:
 158                 self.report_error('Giving up after %s fragment retries' % fragment_retries)
 159                 return -1
 160
 161         decrypt_fragment = self.decrypter(info_dict)
 162         dest, _ = self.sanitize_open(tmpfilename, 'wb')
 163         for frag_index, fragment in enumerate(info_dict['fragments']):
 164             fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
 165             try:
 166                 src, _ = self.sanitize_open(fragment_filename, 'rb')
 167             except OSError as err:
 168                 if skip_unavailable_fragments and frag_index > 1:
 169                     self.report_skip_fragment(frag_index, err)
 170                     continue
 171                 self.report_error(f'Unable to open fragment {frag_index}; {err}')
 172                 return -1
 173             dest.write(decrypt_fragment(fragment, src.read()))
 174             src.close()
 175             if not self.params.get('keep_fragments', False):
 176                 self.try_remove(encodeFilename(fragment_filename))
 177         dest.close()
 178         self.try_remove(encodeFilename('%s.frag.urls' % tmpfilename))
 179         return 0
 180
 181
 182 class CurlFD(ExternalFD):
 183     AVAILABLE_OPT = '-V'
 184     _CAPTURE_STDERR = False  # curl writes the progress to stderr
 185
 186     def _make_cmd(self, tmpfilename, info_dict):
 187         cmd = [self.exe, '--location', '-o', tmpfilename, '--compressed']
 188         if info_dict.get('http_headers') is not None:
 189             for key, val in info_dict['http_headers'].items():
 190                 cmd += ['--header', f'{key}: {val}']
 191
 192         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 193         cmd += self._valueless_option('--silent', 'noprogress')
 194         cmd += self._valueless_option('--verbose', 'verbose')
 195         cmd += self._option('--limit-rate', 'ratelimit')
 196         retry = self._option('--retry', 'retries')
 197         if len(retry) == 2:
 198             if retry[1] in ('inf', 'infinite'):
 199                 retry[1] = '2147483647'
 200             cmd += retry
 201         cmd += self._option('--max-filesize', 'max_filesize')
 202         cmd += self._option('--interface', 'source_address')
 203         cmd += self._option('--proxy', 'proxy')
 204         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 205         cmd += self._configuration_args()
 206         cmd += ['--', info_dict['url']]
 207         return cmd
 208
 209
 210 class AxelFD(ExternalFD):
 211     AVAILABLE_OPT = '-V'
 212
 213     def _make_cmd(self, tmpfilename, info_dict):
 214         cmd = [self.exe, '-o', tmpfilename]
 215         if info_dict.get('http_headers') is not None:
 216             for key, val in info_dict['http_headers'].items():
 217                 cmd += ['-H', f'{key}: {val}']
 218         cmd += self._configuration_args()
 219         cmd += ['--', info_dict['url']]
 220         return cmd
 221
 222
 223 class WgetFD(ExternalFD):
 224     AVAILABLE_OPT = '--version'
 225
 226     def _make_cmd(self, tmpfilename, info_dict):
 227         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies', '--compression=auto']
 228         if info_dict.get('http_headers') is not None:
 229             for key, val in info_dict['http_headers'].items():
 230                 cmd += ['--header', f'{key}: {val}']
 231         cmd += self._option('--limit-rate', 'ratelimit')
 232         retry = self._option('--tries', 'retries')
 233         if len(retry) == 2:
 234             if retry[1] in ('inf', 'infinite'):
 235                 retry[1] = '0'
 236             cmd += retry
 237         cmd += self._option('--bind-address', 'source_address')
 238         proxy = self.params.get('proxy')
 239         if proxy:
 240             for var in ('http_proxy', 'https_proxy'):
 241                 cmd += ['--execute', f'{var}={proxy}']
 242         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 243         cmd += self._configuration_args()
 244         cmd += ['--', info_dict['url']]
 245         return cmd
 246
 247
 248 class Aria2cFD(ExternalFD):
 249     AVAILABLE_OPT = '-v'
 250     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 251
 252     @staticmethod
 253     def supports_manifest(manifest):
 254         UNSUPPORTED_FEATURES = [
 255             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 256             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 257         ]
 258         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 259         return all(check_results)
 260
 261     def _make_cmd(self, tmpfilename, info_dict):
 262         cmd = [self.exe, '-c',
 263                '--console-log-level=warn', '--summary-interval=0', '--download-result=hide',
 264                '--http-accept-gzip=true', '--file-allocation=none', '-x16', '-j16', '-s16']
 265         if 'fragments' in info_dict:
 266             cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true']
 267         else:
 268             cmd += ['--min-split-size', '1M']
 269
 270         if info_dict.get('http_headers') is not None:
 271             for key, val in info_dict['http_headers'].items():
 272                 cmd += ['--header', f'{key}: {val}']
 273         cmd += self._option('--max-overall-download-limit', 'ratelimit')
 274         cmd += self._option('--interface', 'source_address')
 275         cmd += self._option('--all-proxy', 'proxy')
 276         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 277         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 278         cmd += self._bool_option('--show-console-readout', 'noprogress', 'false', 'true', '=')
 279         cmd += self._configuration_args()
 280
 281         # aria2c strips out spaces from the beginning/end of filenames and paths.
 282         # We work around this issue by adding a "./" to the beginning of the
 283         # filename and relative path, and adding a "/" at the end of the path.
 284         # See: https://github.com/yt-dlp/yt-dlp/issues/276
 285         # https://github.com/ytdl-org/youtube-dl/issues/20312
 286         # https://github.com/aria2/aria2/issues/1373
 287         dn = os.path.dirname(tmpfilename)
 288         if dn:
 289             if not os.path.isabs(dn):
 290                 dn = f'.{os.path.sep}{dn}'
 291             cmd += ['--dir', dn + os.path.sep]
 292         if 'fragments' not in info_dict:
 293             cmd += ['--out', f'.{os.path.sep}{os.path.basename(tmpfilename)}']
 294         cmd += ['--auto-file-renaming=false']
 295
 296         if 'fragments' in info_dict:
 297             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 298             url_list_file = '%s.frag.urls' % tmpfilename
 299             url_list = []
 300             for frag_index, fragment in enumerate(info_dict['fragments']):
 301                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 302                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
 303             stream, _ = self.sanitize_open(url_list_file, 'wb')
 304             stream.write('\n'.join(url_list).encode())
 305             stream.close()
 306             cmd += ['-i', url_list_file]
 307         else:
 308             cmd += ['--', info_dict['url']]
 309         return cmd
 310
 311
 312 class HttpieFD(ExternalFD):
 313     AVAILABLE_OPT = '--version'
 314     EXE_NAME = 'http'
 315
 316     def _make_cmd(self, tmpfilename, info_dict):
 317         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 318
 319         if info_dict.get('http_headers') is not None:
 320             for key, val in info_dict['http_headers'].items():
 321                 cmd += [f'{key}:{val}']
 322         return cmd
 323
 324
 325 class FFmpegFD(ExternalFD):
 326     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms', 'http_dash_segments')
 327     SUPPORTED_FEATURES = (Features.TO_STDOUT, Features.MULTIPLE_FORMATS)
 328
 329     @classmethod
 330     def available(cls, path=None):
 331         # TODO: Fix path for ffmpeg
 332         # Fixme: This may be wrong when --ffmpeg-location is used
 333         return FFmpegPostProcessor().available
 334
 335     def on_process_started(self, proc, stdin):
 336         """ Override this in subclasses  """
 337         pass
 338
 339     @classmethod
 340     def can_merge_formats(cls, info_dict, params):
 341         return (
 342             info_dict.get('requested_formats')
 343             and info_dict.get('protocol')
 344             and not params.get('allow_unplayable_formats')
 345             and 'no-direct-merge' not in params.get('compat_opts', [])
 346             and cls.can_download(info_dict))
 347
 348     def _call_downloader(self, tmpfilename, info_dict):
 349         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 350         ffpp = FFmpegPostProcessor(downloader=self)
 351         if not ffpp.available:
 352             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 353             return False
 354         ffpp.check_version()
 355
 356         args = [ffpp.executable, '-y']
 357
 358         for log_level in ('quiet', 'verbose'):
 359             if self.params.get(log_level, False):
 360                 args += ['-loglevel', log_level]
 361                 break
 362         if not self.params.get('verbose'):
 363             args += ['-hide_banner']
 364
 365         args += traverse_obj(info_dict, ('downloader_options', 'ffmpeg_args'), default=[])
 366
 367         # These exists only for compatibility. Extractors should use
 368         # info_dict['downloader_options']['ffmpeg_args'] instead
 369         args += info_dict.get('_ffmpeg_args') or []
 370         seekable = info_dict.get('_seekable')
 371         if seekable is not None:
 372             # setting -seekable prevents ffmpeg from guessing if the server
 373             # supports seeking(by adding the header `Range: bytes=0-`), which
 374             # can cause problems in some cases
 375             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 376             # http://trac.ffmpeg.org/ticket/6125#comment:10
 377             args += ['-seekable', '1' if seekable else '0']
 378
 379         http_headers = None
 380         if info_dict.get('http_headers'):
 381             youtubedl_headers = handle_youtubedl_headers(info_dict['http_headers'])
 382             http_headers = [
 383                 # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 384                 # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 385                 '-headers',
 386                 ''.join(f'{key}: {val}\r\n' for key, val in youtubedl_headers.items())
 387             ]
 388
 389         env = None
 390         proxy = self.params.get('proxy')
 391         if proxy:
 392             if not re.match(r'^[\da-zA-Z]+://', proxy):
 393                 proxy = 'http://%s' % proxy
 394
 395             if proxy.startswith('socks'):
 396                 self.report_warning(
 397                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 398                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 399
 400             # Since December 2015 ffmpeg supports -http_proxy option (see
 401             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 402             # We could switch to the following code if we are able to detect version properly
 403             # args += ['-http_proxy', proxy]
 404             env = os.environ.copy()
 405             env['HTTP_PROXY'] = proxy
 406             env['http_proxy'] = proxy
 407
 408         protocol = info_dict.get('protocol')
 409
 410         if protocol == 'rtmp':
 411             player_url = info_dict.get('player_url')
 412             page_url = info_dict.get('page_url')
 413             app = info_dict.get('app')
 414             play_path = info_dict.get('play_path')
 415             tc_url = info_dict.get('tc_url')
 416             flash_version = info_dict.get('flash_version')
 417             live = info_dict.get('rtmp_live', False)
 418             conn = info_dict.get('rtmp_conn')
 419             if player_url is not None:
 420                 args += ['-rtmp_swfverify', player_url]
 421             if page_url is not None:
 422                 args += ['-rtmp_pageurl', page_url]
 423             if app is not None:
 424                 args += ['-rtmp_app', app]
 425             if play_path is not None:
 426                 args += ['-rtmp_playpath', play_path]
 427             if tc_url is not None:
 428                 args += ['-rtmp_tcurl', tc_url]
 429             if flash_version is not None:
 430                 args += ['-rtmp_flashver', flash_version]
 431             if live:
 432                 args += ['-rtmp_live', 'live']
 433             if isinstance(conn, list):
 434                 for entry in conn:
 435                     args += ['-rtmp_conn', entry]
 436             elif isinstance(conn, str):
 437                 args += ['-rtmp_conn', conn]
 438
 439         start_time, end_time = info_dict.get('section_start') or 0, info_dict.get('section_end')
 440
 441         for i, url in enumerate(urls):
 442             if http_headers is not None and re.match(r'^https?://', url):
 443                 args += http_headers
 444             if start_time:
 445                 args += ['-ss', str(start_time)]
 446             if end_time:
 447                 args += ['-t', str(end_time - start_time)]
 448
 449             args += self._configuration_args((f'_i{i + 1}', '_i')) + ['-i', url]
 450
 451         if not (start_time or end_time) or not self.params.get('force_keyframes_at_cuts'):
 452             args += ['-c', 'copy']
 453
 454         if info_dict.get('requested_formats') or protocol == 'http_dash_segments':
 455             for (i, fmt) in enumerate(info_dict.get('requested_formats') or [info_dict]):
 456                 stream_number = fmt.get('manifest_stream_number', 0)
 457                 args.extend(['-map', f'{i}:{stream_number}'])
 458
 459         if self.params.get('test', False):
 460             args += ['-fs', str(self._TEST_FILE_SIZE)]
 461
 462         ext = info_dict['ext']
 463         if protocol in ('m3u8', 'm3u8_native'):
 464             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 465             if use_mpegts is None:
 466                 use_mpegts = info_dict.get('is_live')
 467             if use_mpegts:
 468                 args += ['-f', 'mpegts']
 469             else:
 470                 args += ['-f', 'mp4']
 471                 if (ffpp.basename == 'ffmpeg' and ffpp._features.get('needs_adtstoasc')) and (not info_dict.get('acodec') or info_dict['acodec'].split('.')[0] in ('aac', 'mp4a')):
 472                     args += ['-bsf:a', 'aac_adtstoasc']
 473         elif protocol == 'rtmp':
 474             args += ['-f', 'flv']
 475         elif ext == 'mp4' and tmpfilename == '-':
 476             args += ['-f', 'mpegts']
 477         elif ext == 'unknown_video':
 478             ext = determine_ext(remove_end(tmpfilename, '.part'))
 479             if ext == 'unknown_video':
 480                 self.report_warning(
 481                     'The video format is unknown and cannot be downloaded by ffmpeg. '
 482                     'Explicitly set the extension in the filename to attempt download in that format')
 483             else:
 484                 self.report_warning(f'The video format is unknown. Trying to download as {ext} according to the filename')
 485                 args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 486         else:
 487             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 488
 489         args += self._configuration_args(('_o1', '_o', ''))
 490
 491         args = [encodeArgument(opt) for opt in args]
 492         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 493         self._debug_cmd(args)
 494
 495         with Popen(args, stdin=subprocess.PIPE, env=env) as proc:
 496             if url in ('-', 'pipe:'):
 497                 self.on_process_started(proc, proc.stdin)
 498             try:
 499                 retval = proc.wait()
 500             except BaseException as e:
 501                 # subprocces.run would send the SIGKILL signal to ffmpeg and the
 502                 # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 503                 # produces a file that is playable (this is mostly useful for live
 504                 # streams). Note that Windows is not affected and produces playable
 505                 # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 506                 if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 507                     proc.communicate_or_kill(b'q')
 508                 else:
 509                     proc.kill(timeout=None)
 510                 raise
 511             return retval
 512
 513
 514 class AVconvFD(FFmpegFD):
 515     pass
 516
 517
 518 _BY_NAME = {
 519     klass.get_basename(): klass
 520     for name, klass in globals().items()
 521     if name.endswith('FD') and name not in ('ExternalFD', 'FragmentFD')
 522 }
 523
 524 _BY_EXE = {klass.EXE_NAME: klass for klass in _BY_NAME.values()}
 525
 526
 527 def list_external_downloaders():
 528     return sorted(_BY_NAME.keys())
 529
 530
 531 def get_external_downloader(external_downloader):
 532     """ Given the name of the executable, see whether we support the given
 533         downloader . """
 534     # Drop .exe extension on Windows
 535     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 536     return _BY_NAME.get(bn, _BY_EXE.get(bn))