yt_dlp/downloader/external.py

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