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     RetryManager,
  14     _configuration_args,
  15     check_executable,
  16     classproperty,
  17     cli_bool_option,
  18     cli_option,
  19     cli_valueless_option,
  20     determine_ext,
  21     encodeArgument,
  22     encodeFilename,
  23     handle_youtubedl_headers,
  24     remove_end,
  25     traverse_obj,
  26 )
  27
  28
  29 class Features(enum.Enum):
  30     TO_STDOUT = enum.auto()
  31     MULTIPLE_FORMATS = enum.auto()
  32
  33
  34 class ExternalFD(FragmentFD):
  35     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps')
  36     SUPPORTED_FEATURES = ()
  37     _CAPTURE_STDERR = True
  38
  39     def real_download(self, filename, info_dict):
  40         self.report_destination(filename)
  41         tmpfilename = self.temp_name(filename)
  42
  43         try:
  44             started = time.time()
  45             retval = self._call_downloader(tmpfilename, info_dict)
  46         except KeyboardInterrupt:
  47             if not info_dict.get('is_live'):
  48                 raise
  49             # Live stream downloading cancellation should be considered as
  50             # correct and expected termination thus all postprocessing
  51             # should take place
  52             retval = 0
  53             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  54
  55         if retval == 0:
  56             status = {
  57                 'filename': filename,
  58                 'status': 'finished',
  59                 'elapsed': time.time() - started,
  60             }
  61             if filename != '-':
  62                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  63                 self.to_screen(f'\r[{self.get_basename()}] Downloaded {fsize} bytes')
  64                 self.try_rename(tmpfilename, filename)
  65                 status.update({
  66                     'downloaded_bytes': fsize,
  67                     'total_bytes': fsize,
  68                 })
  69             self._hook_progress(status, info_dict)
  70             return True
  71         else:
  72             self.to_stderr('\n')
  73             self.report_error('%s exited with code %d' % (
  74                 self.get_basename(), retval))
  75             return False
  76
  77     @classmethod
  78     def get_basename(cls):
  79         return cls.__name__[:-2].lower()
  80
  81     @classproperty
  82     def EXE_NAME(cls):
  83         return cls.get_basename()
  84
  85     @functools.cached_property
  86     def exe(self):
  87         return self.EXE_NAME
  88
  89     @classmethod
  90     def available(cls, path=None):
  91         path = check_executable(
  92             cls.EXE_NAME if path in (None, cls.get_basename()) else path,
  93             [cls.AVAILABLE_OPT])
  94         if not path:
  95             return False
  96         cls.exe = path
  97         return path
  98
  99     @classmethod
 100     def supports(cls, info_dict):
 101         return all((
 102             not info_dict.get('to_stdout') or Features.TO_STDOUT in cls.SUPPORTED_FEATURES,
 103             '+' not in info_dict['protocol'] or Features.MULTIPLE_FORMATS in cls.SUPPORTED_FEATURES,
 104             all(proto in cls.SUPPORTED_PROTOCOLS for proto in info_dict['protocol'].split('+')),
 105         ))
 106
 107     @classmethod
 108     def can_download(cls, info_dict, path=None):
 109         return cls.available(path) and cls.supports(info_dict)
 110
 111     def _option(self, command_option, param):
 112         return cli_option(self.params, command_option, param)
 113
 114     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
 115         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
 116
 117     def _valueless_option(self, command_option, param, expected_value=True):
 118         return cli_valueless_option(self.params, command_option, param, expected_value)
 119
 120     def _configuration_args(self, keys=None, *args, **kwargs):
 121         return _configuration_args(
 122             self.get_basename(), self.params.get('external_downloader_args'), self.EXE_NAME,
 123             keys, *args, **kwargs)
 124
 125     def _call_downloader(self, tmpfilename, info_dict):
 126         """ Either overwrite this or implement _make_cmd """
 127         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 128
 129         self._debug_cmd(cmd)
 130
 131         if 'fragments' not in info_dict:
 132             _, stderr, returncode = Popen.run(
 133                 cmd, text=True, stderr=subprocess.PIPE if self._CAPTURE_STDERR else None)
 134             if returncode and stderr:
 135                 self.to_stderr(stderr)
 136             return returncode
 137
 138         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 139
 140         retry_manager = RetryManager(self.params.get('fragment_retries'), self.report_retry,
 141                                      frag_index=None, fatal=not skip_unavailable_fragments)
 142         for retry in retry_manager:
 143             _, stderr, returncode = Popen.run(cmd, text=True, stderr=subprocess.PIPE)
 144             if not returncode:
 145                 break
 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             retry.error = Exception()
 151             continue
 152         if not skip_unavailable_fragments and retry_manager.error:
 153             return -1
 154
 155         decrypt_fragment = self.decrypter(info_dict)
 156         dest, _ = self.sanitize_open(tmpfilename, 'wb')
 157         for frag_index, fragment in enumerate(info_dict['fragments']):
 158             fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
 159             try:
 160                 src, _ = self.sanitize_open(fragment_filename, 'rb')
 161             except OSError as err:
 162                 if skip_unavailable_fragments and frag_index > 1:
 163                     self.report_skip_fragment(frag_index, err)
 164                     continue
 165                 self.report_error(f'Unable to open fragment {frag_index}; {err}')
 166                 return -1
 167             dest.write(decrypt_fragment(fragment, src.read()))
 168             src.close()
 169             if not self.params.get('keep_fragments', False):
 170                 self.try_remove(encodeFilename(fragment_filename))
 171         dest.close()
 172         self.try_remove(encodeFilename('%s.frag.urls' % tmpfilename))
 173         return 0
 174
 175
 176 class CurlFD(ExternalFD):
 177     AVAILABLE_OPT = '-V'
 178     _CAPTURE_STDERR = False  # curl writes the progress to stderr
 179
 180     def _make_cmd(self, tmpfilename, info_dict):
 181         cmd = [self.exe, '--location', '-o', tmpfilename, '--compressed']
 182         if info_dict.get('http_headers') is not None:
 183             for key, val in info_dict['http_headers'].items():
 184                 cmd += ['--header', f'{key}: {val}']
 185
 186         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 187         cmd += self._valueless_option('--silent', 'noprogress')
 188         cmd += self._valueless_option('--verbose', 'verbose')
 189         cmd += self._option('--limit-rate', 'ratelimit')
 190         retry = self._option('--retry', 'retries')
 191         if len(retry) == 2:
 192             if retry[1] in ('inf', 'infinite'):
 193                 retry[1] = '2147483647'
 194             cmd += retry
 195         cmd += self._option('--max-filesize', 'max_filesize')
 196         cmd += self._option('--interface', 'source_address')
 197         cmd += self._option('--proxy', 'proxy')
 198         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 199         cmd += self._configuration_args()
 200         cmd += ['--', info_dict['url']]
 201         return cmd
 202
 203
 204 class AxelFD(ExternalFD):
 205     AVAILABLE_OPT = '-V'
 206
 207     def _make_cmd(self, tmpfilename, info_dict):
 208         cmd = [self.exe, '-o', tmpfilename]
 209         if info_dict.get('http_headers') is not None:
 210             for key, val in info_dict['http_headers'].items():
 211                 cmd += ['-H', f'{key}: {val}']
 212         cmd += self._configuration_args()
 213         cmd += ['--', info_dict['url']]
 214         return cmd
 215
 216
 217 class WgetFD(ExternalFD):
 218     AVAILABLE_OPT = '--version'
 219
 220     def _make_cmd(self, tmpfilename, info_dict):
 221         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies', '--compression=auto']
 222         if info_dict.get('http_headers') is not None:
 223             for key, val in info_dict['http_headers'].items():
 224                 cmd += ['--header', f'{key}: {val}']
 225         cmd += self._option('--limit-rate', 'ratelimit')
 226         retry = self._option('--tries', 'retries')
 227         if len(retry) == 2:
 228             if retry[1] in ('inf', 'infinite'):
 229                 retry[1] = '0'
 230             cmd += retry
 231         cmd += self._option('--bind-address', 'source_address')
 232         proxy = self.params.get('proxy')
 233         if proxy:
 234             for var in ('http_proxy', 'https_proxy'):
 235                 cmd += ['--execute', f'{var}={proxy}']
 236         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 237         cmd += self._configuration_args()
 238         cmd += ['--', info_dict['url']]
 239         return cmd
 240
 241
 242 class Aria2cFD(ExternalFD):
 243     AVAILABLE_OPT = '-v'
 244     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 245
 246     @staticmethod
 247     def supports_manifest(manifest):
 248         UNSUPPORTED_FEATURES = [
 249             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 250             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 251         ]
 252         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 253         return all(check_results)
 254
 255     @staticmethod
 256     def _aria2c_filename(fn):
 257         return fn if os.path.isabs(fn) else f'.{os.path.sep}{fn}'
 258
 259     def _make_cmd(self, tmpfilename, info_dict):
 260         cmd = [self.exe, '-c',
 261                '--console-log-level=warn', '--summary-interval=0', '--download-result=hide',
 262                '--http-accept-gzip=true', '--file-allocation=none', '-x16', '-j16', '-s16']
 263         if 'fragments' in info_dict:
 264             cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true']
 265         else:
 266             cmd += ['--min-split-size', '1M']
 267
 268         if info_dict.get('http_headers') is not None:
 269             for key, val in info_dict['http_headers'].items():
 270                 cmd += ['--header', f'{key}: {val}']
 271         cmd += self._option('--max-overall-download-limit', 'ratelimit')
 272         cmd += self._option('--interface', 'source_address')
 273         cmd += self._option('--all-proxy', 'proxy')
 274         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 275         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 276         cmd += self._bool_option('--show-console-readout', 'noprogress', 'false', 'true', '=')
 277         cmd += self._configuration_args()
 278
 279         # aria2c strips out spaces from the beginning/end of filenames and paths.
 280         # We work around this issue by adding a "./" to the beginning of the
 281         # filename and relative path, and adding a "/" at the end of the path.
 282         # See: https://github.com/yt-dlp/yt-dlp/issues/276
 283         # https://github.com/ytdl-org/youtube-dl/issues/20312
 284         # https://github.com/aria2/aria2/issues/1373
 285         dn = os.path.dirname(tmpfilename)
 286         if dn:
 287             cmd += ['--dir', self._aria2c_filename(dn) + os.path.sep]
 288         if 'fragments' not in info_dict:
 289             cmd += ['--out', self._aria2c_filename(os.path.basename(tmpfilename))]
 290         cmd += ['--auto-file-renaming=false']
 291
 292         if 'fragments' in info_dict:
 293             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 294             url_list_file = '%s.frag.urls' % tmpfilename
 295             url_list = []
 296             for frag_index, fragment in enumerate(info_dict['fragments']):
 297                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 298                 url_list.append('%s\n\tout=%s' % (fragment['url'], self._aria2c_filename(fragment_filename)))
 299             stream, _ = self.sanitize_open(url_list_file, 'wb')
 300             stream.write('\n'.join(url_list).encode())
 301             stream.close()
 302             cmd += ['-i', self._aria2c_filename(url_list_file)]
 303         else:
 304             cmd += ['--', info_dict['url']]
 305         return cmd
 306
 307
 308 class HttpieFD(ExternalFD):
 309     AVAILABLE_OPT = '--version'
 310     EXE_NAME = 'http'
 311
 312     def _make_cmd(self, tmpfilename, info_dict):
 313         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 314
 315         if info_dict.get('http_headers') is not None:
 316             for key, val in info_dict['http_headers'].items():
 317                 cmd += [f'{key}:{val}']
 318         return cmd
 319
 320
 321 class FFmpegFD(ExternalFD):
 322     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms', 'http_dash_segments')
 323     SUPPORTED_FEATURES = (Features.TO_STDOUT, Features.MULTIPLE_FORMATS)
 324
 325     @classmethod
 326     def available(cls, path=None):
 327         # TODO: Fix path for ffmpeg
 328         # Fixme: This may be wrong when --ffmpeg-location is used
 329         return FFmpegPostProcessor().available
 330
 331     def on_process_started(self, proc, stdin):
 332         """ Override this in subclasses  """
 333         pass
 334
 335     @classmethod
 336     def can_merge_formats(cls, info_dict, params):
 337         return (
 338             info_dict.get('requested_formats')
 339             and info_dict.get('protocol')
 340             and not params.get('allow_unplayable_formats')
 341             and 'no-direct-merge' not in params.get('compat_opts', [])
 342             and cls.can_download(info_dict))
 343
 344     def _call_downloader(self, tmpfilename, info_dict):
 345         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 346         ffpp = FFmpegPostProcessor(downloader=self)
 347         if not ffpp.available:
 348             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 349             return False
 350         ffpp.check_version()
 351
 352         args = [ffpp.executable, '-y']
 353
 354         for log_level in ('quiet', 'verbose'):
 355             if self.params.get(log_level, False):
 356                 args += ['-loglevel', log_level]
 357                 break
 358         if not self.params.get('verbose'):
 359             args += ['-hide_banner']
 360
 361         args += traverse_obj(info_dict, ('downloader_options', 'ffmpeg_args'), default=[])
 362
 363         # These exists only for compatibility. Extractors should use
 364         # info_dict['downloader_options']['ffmpeg_args'] instead
 365         args += info_dict.get('_ffmpeg_args') or []
 366         seekable = info_dict.get('_seekable')
 367         if seekable is not None:
 368             # setting -seekable prevents ffmpeg from guessing if the server
 369             # supports seeking(by adding the header `Range: bytes=0-`), which
 370             # can cause problems in some cases
 371             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 372             # http://trac.ffmpeg.org/ticket/6125#comment:10
 373             args += ['-seekable', '1' if seekable else '0']
 374
 375         http_headers = None
 376         if info_dict.get('http_headers'):
 377             youtubedl_headers = handle_youtubedl_headers(info_dict['http_headers'])
 378             http_headers = [
 379                 # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 380                 # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 381                 '-headers',
 382                 ''.join(f'{key}: {val}\r\n' for key, val in youtubedl_headers.items())
 383             ]
 384
 385         env = None
 386         proxy = self.params.get('proxy')
 387         if proxy:
 388             if not re.match(r'^[\da-zA-Z]+://', proxy):
 389                 proxy = 'http://%s' % proxy
 390
 391             if proxy.startswith('socks'):
 392                 self.report_warning(
 393                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 394                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 395
 396             # Since December 2015 ffmpeg supports -http_proxy option (see
 397             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 398             # We could switch to the following code if we are able to detect version properly
 399             # args += ['-http_proxy', proxy]
 400             env = os.environ.copy()
 401             env['HTTP_PROXY'] = proxy
 402             env['http_proxy'] = proxy
 403
 404         protocol = info_dict.get('protocol')
 405
 406         if protocol == 'rtmp':
 407             player_url = info_dict.get('player_url')
 408             page_url = info_dict.get('page_url')
 409             app = info_dict.get('app')
 410             play_path = info_dict.get('play_path')
 411             tc_url = info_dict.get('tc_url')
 412             flash_version = info_dict.get('flash_version')
 413             live = info_dict.get('rtmp_live', False)
 414             conn = info_dict.get('rtmp_conn')
 415             if player_url is not None:
 416                 args += ['-rtmp_swfverify', player_url]
 417             if page_url is not None:
 418                 args += ['-rtmp_pageurl', page_url]
 419             if app is not None:
 420                 args += ['-rtmp_app', app]
 421             if play_path is not None:
 422                 args += ['-rtmp_playpath', play_path]
 423             if tc_url is not None:
 424                 args += ['-rtmp_tcurl', tc_url]
 425             if flash_version is not None:
 426                 args += ['-rtmp_flashver', flash_version]
 427             if live:
 428                 args += ['-rtmp_live', 'live']
 429             if isinstance(conn, list):
 430                 for entry in conn:
 431                     args += ['-rtmp_conn', entry]
 432             elif isinstance(conn, str):
 433                 args += ['-rtmp_conn', conn]
 434
 435         start_time, end_time = info_dict.get('section_start') or 0, info_dict.get('section_end')
 436
 437         for i, url in enumerate(urls):
 438             if http_headers is not None and re.match(r'^https?://', url):
 439                 args += http_headers
 440             if start_time:
 441                 args += ['-ss', str(start_time)]
 442             if end_time:
 443                 args += ['-t', str(end_time - start_time)]
 444
 445             args += self._configuration_args((f'_i{i + 1}', '_i')) + ['-i', url]
 446
 447         if not (start_time or end_time) or not self.params.get('force_keyframes_at_cuts'):
 448             args += ['-c', 'copy']
 449
 450         if info_dict.get('requested_formats') or protocol == 'http_dash_segments':
 451             for (i, fmt) in enumerate(info_dict.get('requested_formats') or [info_dict]):
 452                 stream_number = fmt.get('manifest_stream_number', 0)
 453                 args.extend(['-map', f'{i}:{stream_number}'])
 454
 455         if self.params.get('test', False):
 456             args += ['-fs', str(self._TEST_FILE_SIZE)]
 457
 458         ext = info_dict['ext']
 459         if protocol in ('m3u8', 'm3u8_native'):
 460             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 461             if use_mpegts is None:
 462                 use_mpegts = info_dict.get('is_live')
 463             if use_mpegts:
 464                 args += ['-f', 'mpegts']
 465             else:
 466                 args += ['-f', 'mp4']
 467                 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')):
 468                     args += ['-bsf:a', 'aac_adtstoasc']
 469         elif protocol == 'rtmp':
 470             args += ['-f', 'flv']
 471         elif ext == 'mp4' and tmpfilename == '-':
 472             args += ['-f', 'mpegts']
 473         elif ext == 'unknown_video':
 474             ext = determine_ext(remove_end(tmpfilename, '.part'))
 475             if ext == 'unknown_video':
 476                 self.report_warning(
 477                     'The video format is unknown and cannot be downloaded by ffmpeg. '
 478                     'Explicitly set the extension in the filename to attempt download in that format')
 479             else:
 480                 self.report_warning(f'The video format is unknown. Trying to download as {ext} according to the filename')
 481                 args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 482         else:
 483             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 484
 485         args += self._configuration_args(('_o1', '_o', ''))
 486
 487         args = [encodeArgument(opt) for opt in args]
 488         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 489         self._debug_cmd(args)
 490
 491         with Popen(args, stdin=subprocess.PIPE, env=env) as proc:
 492             if url in ('-', 'pipe:'):
 493                 self.on_process_started(proc, proc.stdin)
 494             try:
 495                 retval = proc.wait()
 496             except BaseException as e:
 497                 # subprocces.run would send the SIGKILL signal to ffmpeg and the
 498                 # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 499                 # produces a file that is playable (this is mostly useful for live
 500                 # streams). Note that Windows is not affected and produces playable
 501                 # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 502                 if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 503                     proc.communicate_or_kill(b'q')
 504                 else:
 505                     proc.kill(timeout=None)
 506                 raise
 507             return retval
 508
 509
 510 class AVconvFD(FFmpegFD):
 511     pass
 512
 513
 514 _BY_NAME = {
 515     klass.get_basename(): klass
 516     for name, klass in globals().items()
 517     if name.endswith('FD') and name not in ('ExternalFD', 'FragmentFD')
 518 }
 519
 520
 521 def list_external_downloaders():
 522     return sorted(_BY_NAME.keys())
 523
 524
 525 def get_external_downloader(external_downloader):
 526     """ Given the name of the executable, see whether we support the given downloader """
 527     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 528     return _BY_NAME.get(bn) or next((
 529         klass for klass in _BY_NAME.values() if klass.EXE_NAME in bn
 530     ), None)