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  # isort: split
  10 from ..compat import compat_setenv
  11 from ..postprocessor.ffmpeg import EXT_TO_OUT_FORMATS, FFmpegPostProcessor
  12 from ..utils import (
  13     Popen,
  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
  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             p = Popen(cmd, stderr=subprocess.PIPE)
 132             _, stderr = p.communicate_or_kill()
 133             if p.returncode != 0:
 134                 self.to_stderr(stderr.decode('utf-8', 'replace'))
 135             return p.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             p = Popen(cmd, stderr=subprocess.PIPE)
 143             _, stderr = p.communicate_or_kill()
 144             if p.returncode == 0:
 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             self.to_stderr(stderr.decode('utf-8', 'replace'))
 149             count += 1
 150             if count <= fragment_retries:
 151                 self.to_screen(
 152                     '[%s] Got error. Retrying fragments (attempt %d of %s)...'
 153                     % (self.get_basename(), count, self.format_retries(fragment_retries)))
 154                 self.sleep_retry('fragment', count)
 155         if count > fragment_retries:
 156             if not skip_unavailable_fragments:
 157                 self.report_error('Giving up after %s fragment retries' % fragment_retries)
 158                 return -1
 159
 160         decrypt_fragment = self.decrypter(info_dict)
 161         dest, _ = self.sanitize_open(tmpfilename, 'wb')
 162         for frag_index, fragment in enumerate(info_dict['fragments']):
 163             fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
 164             try:
 165                 src, _ = self.sanitize_open(fragment_filename, 'rb')
 166             except OSError as err:
 167                 if skip_unavailable_fragments and frag_index > 1:
 168                     self.report_skip_fragment(frag_index, err)
 169                     continue
 170                 self.report_error(f'Unable to open fragment {frag_index}; {err}')
 171                 return -1
 172             dest.write(decrypt_fragment(fragment, src.read()))
 173             src.close()
 174             if not self.params.get('keep_fragments', False):
 175                 self.try_remove(encodeFilename(fragment_filename))
 176         dest.close()
 177         self.try_remove(encodeFilename('%s.frag.urls' % tmpfilename))
 178         return 0
 179
 180
 181 class CurlFD(ExternalFD):
 182     AVAILABLE_OPT = '-V'
 183
 184     def _make_cmd(self, tmpfilename, info_dict):
 185         cmd = [self.exe, '--location', '-o', tmpfilename, '--compressed']
 186         if info_dict.get('http_headers') is not None:
 187             for key, val in info_dict['http_headers'].items():
 188                 cmd += ['--header', f'{key}: {val}']
 189
 190         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 191         cmd += self._valueless_option('--silent', 'noprogress')
 192         cmd += self._valueless_option('--verbose', 'verbose')
 193         cmd += self._option('--limit-rate', 'ratelimit')
 194         retry = self._option('--retry', 'retries')
 195         if len(retry) == 2:
 196             if retry[1] in ('inf', 'infinite'):
 197                 retry[1] = '2147483647'
 198             cmd += retry
 199         cmd += self._option('--max-filesize', 'max_filesize')
 200         cmd += self._option('--interface', 'source_address')
 201         cmd += self._option('--proxy', 'proxy')
 202         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 203         cmd += self._configuration_args()
 204         cmd += ['--', info_dict['url']]
 205         return cmd
 206
 207     def _call_downloader(self, tmpfilename, info_dict):
 208         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 209
 210         self._debug_cmd(cmd)
 211
 212         # curl writes the progress to stderr so don't capture it.
 213         p = Popen(cmd)
 214         p.communicate_or_kill()
 215         return p.returncode
 216
 217
 218 class AxelFD(ExternalFD):
 219     AVAILABLE_OPT = '-V'
 220
 221     def _make_cmd(self, tmpfilename, info_dict):
 222         cmd = [self.exe, '-o', tmpfilename]
 223         if info_dict.get('http_headers') is not None:
 224             for key, val in info_dict['http_headers'].items():
 225                 cmd += ['-H', f'{key}: {val}']
 226         cmd += self._configuration_args()
 227         cmd += ['--', info_dict['url']]
 228         return cmd
 229
 230
 231 class WgetFD(ExternalFD):
 232     AVAILABLE_OPT = '--version'
 233
 234     def _make_cmd(self, tmpfilename, info_dict):
 235         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies', '--compression=auto']
 236         if info_dict.get('http_headers') is not None:
 237             for key, val in info_dict['http_headers'].items():
 238                 cmd += ['--header', f'{key}: {val}']
 239         cmd += self._option('--limit-rate', 'ratelimit')
 240         retry = self._option('--tries', 'retries')
 241         if len(retry) == 2:
 242             if retry[1] in ('inf', 'infinite'):
 243                 retry[1] = '0'
 244             cmd += retry
 245         cmd += self._option('--bind-address', 'source_address')
 246         proxy = self.params.get('proxy')
 247         if proxy:
 248             for var in ('http_proxy', 'https_proxy'):
 249                 cmd += ['--execute', f'{var}={proxy}']
 250         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 251         cmd += self._configuration_args()
 252         cmd += ['--', info_dict['url']]
 253         return cmd
 254
 255
 256 class Aria2cFD(ExternalFD):
 257     AVAILABLE_OPT = '-v'
 258     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 259
 260     @staticmethod
 261     def supports_manifest(manifest):
 262         UNSUPPORTED_FEATURES = [
 263             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 264             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 265         ]
 266         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 267         return all(check_results)
 268
 269     def _make_cmd(self, tmpfilename, info_dict):
 270         cmd = [self.exe, '-c',
 271                '--console-log-level=warn', '--summary-interval=0', '--download-result=hide',
 272                '--http-accept-gzip=true', '--file-allocation=none', '-x16', '-j16', '-s16']
 273         if 'fragments' in info_dict:
 274             cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true']
 275         else:
 276             cmd += ['--min-split-size', '1M']
 277
 278         if info_dict.get('http_headers') is not None:
 279             for key, val in info_dict['http_headers'].items():
 280                 cmd += ['--header', f'{key}: {val}']
 281         cmd += self._option('--max-overall-download-limit', 'ratelimit')
 282         cmd += self._option('--interface', 'source_address')
 283         cmd += self._option('--all-proxy', 'proxy')
 284         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 285         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 286         cmd += self._bool_option('--show-console-readout', 'noprogress', 'false', 'true', '=')
 287         cmd += self._configuration_args()
 288
 289         # aria2c strips out spaces from the beginning/end of filenames and paths.
 290         # We work around this issue by adding a "./" to the beginning of the
 291         # filename and relative path, and adding a "/" at the end of the path.
 292         # See: https://github.com/yt-dlp/yt-dlp/issues/276
 293         # https://github.com/ytdl-org/youtube-dl/issues/20312
 294         # https://github.com/aria2/aria2/issues/1373
 295         dn = os.path.dirname(tmpfilename)
 296         if dn:
 297             if not os.path.isabs(dn):
 298                 dn = f'.{os.path.sep}{dn}'
 299             cmd += ['--dir', dn + os.path.sep]
 300         if 'fragments' not in info_dict:
 301             cmd += ['--out', f'.{os.path.sep}{os.path.basename(tmpfilename)}']
 302         cmd += ['--auto-file-renaming=false']
 303
 304         if 'fragments' in info_dict:
 305             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 306             url_list_file = '%s.frag.urls' % tmpfilename
 307             url_list = []
 308             for frag_index, fragment in enumerate(info_dict['fragments']):
 309                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 310                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
 311             stream, _ = self.sanitize_open(url_list_file, 'wb')
 312             stream.write('\n'.join(url_list).encode())
 313             stream.close()
 314             cmd += ['-i', url_list_file]
 315         else:
 316             cmd += ['--', info_dict['url']]
 317         return cmd
 318
 319
 320 class HttpieFD(ExternalFD):
 321     AVAILABLE_OPT = '--version'
 322     EXE_NAME = 'http'
 323
 324     def _make_cmd(self, tmpfilename, info_dict):
 325         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 326
 327         if info_dict.get('http_headers') is not None:
 328             for key, val in info_dict['http_headers'].items():
 329                 cmd += [f'{key}:{val}']
 330         return cmd
 331
 332
 333 class FFmpegFD(ExternalFD):
 334     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms', 'http_dash_segments')
 335     SUPPORTED_FEATURES = (Features.TO_STDOUT, Features.MULTIPLE_FORMATS)
 336
 337     @classmethod
 338     def available(cls, path=None):
 339         # TODO: Fix path for ffmpeg
 340         # Fixme: This may be wrong when --ffmpeg-location is used
 341         return FFmpegPostProcessor().available
 342
 343     def on_process_started(self, proc, stdin):
 344         """ Override this in subclasses  """
 345         pass
 346
 347     @classmethod
 348     def can_merge_formats(cls, info_dict, params):
 349         return (
 350             info_dict.get('requested_formats')
 351             and info_dict.get('protocol')
 352             and not params.get('allow_unplayable_formats')
 353             and 'no-direct-merge' not in params.get('compat_opts', [])
 354             and cls.can_download(info_dict))
 355
 356     def _call_downloader(self, tmpfilename, info_dict):
 357         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 358         ffpp = FFmpegPostProcessor(downloader=self)
 359         if not ffpp.available:
 360             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 361             return False
 362         ffpp.check_version()
 363
 364         args = [ffpp.executable, '-y']
 365
 366         for log_level in ('quiet', 'verbose'):
 367             if self.params.get(log_level, False):
 368                 args += ['-loglevel', log_level]
 369                 break
 370         if not self.params.get('verbose'):
 371             args += ['-hide_banner']
 372
 373         args += traverse_obj(info_dict, ('downloader_options', 'ffmpeg_args'), default=[])
 374
 375         # These exists only for compatibility. Extractors should use
 376         # info_dict['downloader_options']['ffmpeg_args'] instead
 377         args += info_dict.get('_ffmpeg_args') or []
 378         seekable = info_dict.get('_seekable')
 379         if seekable is not None:
 380             # setting -seekable prevents ffmpeg from guessing if the server
 381             # supports seeking(by adding the header `Range: bytes=0-`), which
 382             # can cause problems in some cases
 383             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 384             # http://trac.ffmpeg.org/ticket/6125#comment:10
 385             args += ['-seekable', '1' if seekable else '0']
 386
 387         # start_time = info_dict.get('start_time') or 0
 388         # if start_time:
 389         #     args += ['-ss', str(start_time)]
 390         # end_time = info_dict.get('end_time')
 391         # if end_time:
 392         #     args += ['-t', str(end_time - start_time)]
 393
 394         http_headers = None
 395         if info_dict.get('http_headers'):
 396             youtubedl_headers = handle_youtubedl_headers(info_dict['http_headers'])
 397             http_headers = [
 398                 # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 399                 # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 400                 '-headers',
 401                 ''.join(f'{key}: {val}\r\n' for key, val in youtubedl_headers.items())
 402             ]
 403
 404         env = None
 405         proxy = self.params.get('proxy')
 406         if proxy:
 407             if not re.match(r'^[\da-zA-Z]+://', proxy):
 408                 proxy = 'http://%s' % proxy
 409
 410             if proxy.startswith('socks'):
 411                 self.report_warning(
 412                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 413                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 414
 415             # Since December 2015 ffmpeg supports -http_proxy option (see
 416             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 417             # We could switch to the following code if we are able to detect version properly
 418             # args += ['-http_proxy', proxy]
 419             env = os.environ.copy()
 420             compat_setenv('HTTP_PROXY', proxy, env=env)
 421             compat_setenv('http_proxy', proxy, env=env)
 422
 423         protocol = info_dict.get('protocol')
 424
 425         if protocol == 'rtmp':
 426             player_url = info_dict.get('player_url')
 427             page_url = info_dict.get('page_url')
 428             app = info_dict.get('app')
 429             play_path = info_dict.get('play_path')
 430             tc_url = info_dict.get('tc_url')
 431             flash_version = info_dict.get('flash_version')
 432             live = info_dict.get('rtmp_live', False)
 433             conn = info_dict.get('rtmp_conn')
 434             if player_url is not None:
 435                 args += ['-rtmp_swfverify', player_url]
 436             if page_url is not None:
 437                 args += ['-rtmp_pageurl', page_url]
 438             if app is not None:
 439                 args += ['-rtmp_app', app]
 440             if play_path is not None:
 441                 args += ['-rtmp_playpath', play_path]
 442             if tc_url is not None:
 443                 args += ['-rtmp_tcurl', tc_url]
 444             if flash_version is not None:
 445                 args += ['-rtmp_flashver', flash_version]
 446             if live:
 447                 args += ['-rtmp_live', 'live']
 448             if isinstance(conn, list):
 449                 for entry in conn:
 450                     args += ['-rtmp_conn', entry]
 451             elif isinstance(conn, str):
 452                 args += ['-rtmp_conn', conn]
 453
 454         for i, url in enumerate(urls):
 455             # We need to specify headers for each http input stream
 456             # otherwise, it will only be applied to the first.
 457             # https://github.com/yt-dlp/yt-dlp/issues/2696
 458             if http_headers is not None and re.match(r'^https?://', url):
 459                 args += http_headers
 460             args += self._configuration_args((f'_i{i + 1}', '_i')) + ['-i', url]
 461
 462         args += ['-c', 'copy']
 463         if info_dict.get('requested_formats') or protocol == 'http_dash_segments':
 464             for (i, fmt) in enumerate(info_dict.get('requested_formats') or [info_dict]):
 465                 stream_number = fmt.get('manifest_stream_number', 0)
 466                 args.extend(['-map', f'{i}:{stream_number}'])
 467
 468         if self.params.get('test', False):
 469             args += ['-fs', str(self._TEST_FILE_SIZE)]
 470
 471         ext = info_dict['ext']
 472         if protocol in ('m3u8', 'm3u8_native'):
 473             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 474             if use_mpegts is None:
 475                 use_mpegts = info_dict.get('is_live')
 476             if use_mpegts:
 477                 args += ['-f', 'mpegts']
 478             else:
 479                 args += ['-f', 'mp4']
 480                 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')):
 481                     args += ['-bsf:a', 'aac_adtstoasc']
 482         elif protocol == 'rtmp':
 483             args += ['-f', 'flv']
 484         elif ext == 'mp4' and tmpfilename == '-':
 485             args += ['-f', 'mpegts']
 486         elif ext == 'unknown_video':
 487             ext = determine_ext(remove_end(tmpfilename, '.part'))
 488             if ext == 'unknown_video':
 489                 self.report_warning(
 490                     'The video format is unknown and cannot be downloaded by ffmpeg. '
 491                     'Explicitly set the extension in the filename to attempt download in that format')
 492             else:
 493                 self.report_warning(f'The video format is unknown. Trying to download as {ext} according to the filename')
 494                 args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 495         else:
 496             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 497
 498         args += self._configuration_args(('_o1', '_o', ''))
 499
 500         args = [encodeArgument(opt) for opt in args]
 501         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 502         self._debug_cmd(args)
 503
 504         proc = Popen(args, stdin=subprocess.PIPE, env=env)
 505         if url in ('-', 'pipe:'):
 506             self.on_process_started(proc, proc.stdin)
 507         try:
 508             retval = proc.wait()
 509         except BaseException as e:
 510             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 511             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 512             # produces a file that is playable (this is mostly useful for live
 513             # streams). Note that Windows is not affected and produces playable
 514             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 515             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 516                 proc.communicate_or_kill(b'q')
 517             else:
 518                 proc.kill()
 519                 proc.wait()
 520             raise
 521         return retval
 522
 523
 524 class AVconvFD(FFmpegFD):
 525     pass
 526
 527
 528 _BY_NAME = {
 529     klass.get_basename(): klass
 530     for name, klass in globals().items()
 531     if name.endswith('FD') and name not in ('ExternalFD', 'FragmentFD')
 532 }
 533
 534 _BY_EXE = {klass.EXE_NAME: klass for klass in _BY_NAME.values()}
 535
 536
 537 def list_external_downloaders():
 538     return sorted(_BY_NAME.keys())
 539
 540
 541 def get_external_downloader(external_downloader):
 542     """ Given the name of the executable, see whether we support the given
 543         downloader . """
 544     # Drop .exe extension on Windows
 545     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 546     return _BY_NAME.get(bn, _BY_EXE.get(bn))