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     traverse_obj,
  24 )
  25
  26
  27 class ExternalFD(FragmentFD):
  28     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps')
  29     can_download_to_stdout = False
  30
  31     def real_download(self, filename, info_dict):
  32         self.report_destination(filename)
  33         tmpfilename = self.temp_name(filename)
  34
  35         try:
  36             started = time.time()
  37             retval = self._call_downloader(tmpfilename, info_dict)
  38         except KeyboardInterrupt:
  39             if not info_dict.get('is_live'):
  40                 raise
  41             # Live stream downloading cancellation should be considered as
  42             # correct and expected termination thus all postprocessing
  43             # should take place
  44             retval = 0
  45             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  46
  47         if retval == 0:
  48             status = {
  49                 'filename': filename,
  50                 'status': 'finished',
  51                 'elapsed': time.time() - started,
  52             }
  53             if filename != '-':
  54                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  55                 self.to_screen(f'\r[{self.get_basename()}] Downloaded {fsize} bytes')
  56                 self.try_rename(tmpfilename, filename)
  57                 status.update({
  58                     'downloaded_bytes': fsize,
  59                     'total_bytes': fsize,
  60                 })
  61             self._hook_progress(status, info_dict)
  62             return True
  63         else:
  64             self.to_stderr('\n')
  65             self.report_error('%s exited with code %d' % (
  66                 self.get_basename(), retval))
  67             return False
  68
  69     @classmethod
  70     def get_basename(cls):
  71         return cls.__name__[:-2].lower()
  72
  73     @classproperty
  74     def EXE_NAME(cls):
  75         return cls.get_basename()
  76
  77     @property
  78     def exe(self):
  79         return self.EXE_NAME
  80
  81     @classmethod
  82     def available(cls, path=None):
  83         path = check_executable(
  84             cls.EXE_NAME if path in (None, cls.get_basename()) else path,
  85             [cls.AVAILABLE_OPT])
  86         if not path:
  87             return False
  88         cls.exe = path
  89         return path
  90
  91     @classmethod
  92     def supports(cls, info_dict):
  93         return (
  94             (cls.can_download_to_stdout or not info_dict.get('to_stdout'))
  95             and info_dict['protocol'] in cls.SUPPORTED_PROTOCOLS)
  96
  97     @classmethod
  98     def can_download(cls, info_dict, path=None):
  99         return cls.available(path) and cls.supports(info_dict)
 100
 101     def _option(self, command_option, param):
 102         return cli_option(self.params, command_option, param)
 103
 104     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
 105         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
 106
 107     def _valueless_option(self, command_option, param, expected_value=True):
 108         return cli_valueless_option(self.params, command_option, param, expected_value)
 109
 110     def _configuration_args(self, keys=None, *args, **kwargs):
 111         return _configuration_args(
 112             self.get_basename(), self.params.get('external_downloader_args'), self.EXE_NAME,
 113             keys, *args, **kwargs)
 114
 115     def _call_downloader(self, tmpfilename, info_dict):
 116         """ Either overwrite this or implement _make_cmd """
 117         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 118
 119         self._debug_cmd(cmd)
 120
 121         if 'fragments' not in info_dict:
 122             p = Popen(cmd, stderr=subprocess.PIPE)
 123             _, stderr = p.communicate_or_kill()
 124             if p.returncode != 0:
 125                 self.to_stderr(stderr.decode('utf-8', 'replace'))
 126             return p.returncode
 127
 128         fragment_retries = self.params.get('fragment_retries', 0)
 129         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 130
 131         count = 0
 132         while count <= fragment_retries:
 133             p = Popen(cmd, stderr=subprocess.PIPE)
 134             _, stderr = p.communicate_or_kill()
 135             if p.returncode == 0:
 136                 break
 137             # TODO: Decide whether to retry based on error code
 138             # https://aria2.github.io/manual/en/html/aria2c.html#exit-status
 139             self.to_stderr(stderr.decode('utf-8', 'replace'))
 140             count += 1
 141             if count <= fragment_retries:
 142                 self.to_screen(
 143                     '[%s] Got error. Retrying fragments (attempt %d of %s)...'
 144                     % (self.get_basename(), count, self.format_retries(fragment_retries)))
 145         if count > fragment_retries:
 146             if not skip_unavailable_fragments:
 147                 self.report_error('Giving up after %s fragment retries' % fragment_retries)
 148                 return -1
 149
 150         decrypt_fragment = self.decrypter(info_dict)
 151         dest, _ = self.sanitize_open(tmpfilename, 'wb')
 152         for frag_index, fragment in enumerate(info_dict['fragments']):
 153             fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
 154             try:
 155                 src, _ = self.sanitize_open(fragment_filename, 'rb')
 156             except OSError as err:
 157                 if skip_unavailable_fragments and frag_index > 1:
 158                     self.report_skip_fragment(frag_index, err)
 159                     continue
 160                 self.report_error(f'Unable to open fragment {frag_index}; {err}')
 161                 return -1
 162             dest.write(decrypt_fragment(fragment, src.read()))
 163             src.close()
 164             if not self.params.get('keep_fragments', False):
 165                 self.try_remove(encodeFilename(fragment_filename))
 166         dest.close()
 167         self.try_remove(encodeFilename('%s.frag.urls' % tmpfilename))
 168         return 0
 169
 170
 171 class CurlFD(ExternalFD):
 172     AVAILABLE_OPT = '-V'
 173
 174     def _make_cmd(self, tmpfilename, info_dict):
 175         cmd = [self.exe, '--location', '-o', tmpfilename, '--compressed']
 176         if info_dict.get('http_headers') is not None:
 177             for key, val in info_dict['http_headers'].items():
 178                 cmd += ['--header', f'{key}: {val}']
 179
 180         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 181         cmd += self._valueless_option('--silent', 'noprogress')
 182         cmd += self._valueless_option('--verbose', 'verbose')
 183         cmd += self._option('--limit-rate', 'ratelimit')
 184         retry = self._option('--retry', 'retries')
 185         if len(retry) == 2:
 186             if retry[1] in ('inf', 'infinite'):
 187                 retry[1] = '2147483647'
 188             cmd += retry
 189         cmd += self._option('--max-filesize', 'max_filesize')
 190         cmd += self._option('--interface', 'source_address')
 191         cmd += self._option('--proxy', 'proxy')
 192         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 193         cmd += self._configuration_args()
 194         cmd += ['--', info_dict['url']]
 195         return cmd
 196
 197     def _call_downloader(self, tmpfilename, info_dict):
 198         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 199
 200         self._debug_cmd(cmd)
 201
 202         # curl writes the progress to stderr so don't capture it.
 203         p = Popen(cmd)
 204         p.communicate_or_kill()
 205         return p.returncode
 206
 207
 208 class AxelFD(ExternalFD):
 209     AVAILABLE_OPT = '-V'
 210
 211     def _make_cmd(self, tmpfilename, info_dict):
 212         cmd = [self.exe, '-o', tmpfilename]
 213         if info_dict.get('http_headers') is not None:
 214             for key, val in info_dict['http_headers'].items():
 215                 cmd += ['-H', f'{key}: {val}']
 216         cmd += self._configuration_args()
 217         cmd += ['--', info_dict['url']]
 218         return cmd
 219
 220
 221 class WgetFD(ExternalFD):
 222     AVAILABLE_OPT = '--version'
 223
 224     def _make_cmd(self, tmpfilename, info_dict):
 225         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies', '--compression=auto']
 226         if info_dict.get('http_headers') is not None:
 227             for key, val in info_dict['http_headers'].items():
 228                 cmd += ['--header', f'{key}: {val}']
 229         cmd += self._option('--limit-rate', 'ratelimit')
 230         retry = self._option('--tries', 'retries')
 231         if len(retry) == 2:
 232             if retry[1] in ('inf', 'infinite'):
 233                 retry[1] = '0'
 234             cmd += retry
 235         cmd += self._option('--bind-address', 'source_address')
 236         proxy = self.params.get('proxy')
 237         if proxy:
 238             for var in ('http_proxy', 'https_proxy'):
 239                 cmd += ['--execute', f'{var}={proxy}']
 240         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 241         cmd += self._configuration_args()
 242         cmd += ['--', info_dict['url']]
 243         return cmd
 244
 245
 246 class Aria2cFD(ExternalFD):
 247     AVAILABLE_OPT = '-v'
 248     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 249
 250     @staticmethod
 251     def supports_manifest(manifest):
 252         UNSUPPORTED_FEATURES = [
 253             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 254             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 255         ]
 256         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 257         return all(check_results)
 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             if not os.path.isabs(dn):
 288                 dn = f'.{os.path.sep}{dn}'
 289             cmd += ['--dir', dn + os.path.sep]
 290         if 'fragments' not in info_dict:
 291             cmd += ['--out', f'.{os.path.sep}{os.path.basename(tmpfilename)}']
 292         cmd += ['--auto-file-renaming=false']
 293
 294         if 'fragments' in info_dict:
 295             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 296             url_list_file = '%s.frag.urls' % tmpfilename
 297             url_list = []
 298             for frag_index, fragment in enumerate(info_dict['fragments']):
 299                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 300                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
 301             stream, _ = self.sanitize_open(url_list_file, 'wb')
 302             stream.write('\n'.join(url_list).encode('utf-8'))
 303             stream.close()
 304             cmd += ['-i', url_list_file]
 305         else:
 306             cmd += ['--', info_dict['url']]
 307         return cmd
 308
 309
 310 class HttpieFD(ExternalFD):
 311     AVAILABLE_OPT = '--version'
 312     EXE_NAME = 'http'
 313
 314     def _make_cmd(self, tmpfilename, info_dict):
 315         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 316
 317         if info_dict.get('http_headers') is not None:
 318             for key, val in info_dict['http_headers'].items():
 319                 cmd += [f'{key}:{val}']
 320         return cmd
 321
 322
 323 class FFmpegFD(ExternalFD):
 324     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms', 'http_dash_segments')
 325     can_download_to_stdout = True
 326
 327     @classmethod
 328     def available(cls, path=None):
 329         # TODO: Fix path for ffmpeg
 330         # Fixme: This may be wrong when --ffmpeg-location is used
 331         return FFmpegPostProcessor().available
 332
 333     @classmethod
 334     def supports(cls, info_dict):
 335         return all(proto in cls.SUPPORTED_PROTOCOLS for proto in info_dict['protocol'].split('+'))
 336
 337     def on_process_started(self, proc, stdin):
 338         """ Override this in subclasses  """
 339         pass
 340
 341     @classmethod
 342     def can_merge_formats(cls, info_dict, params):
 343         return (
 344             info_dict.get('requested_formats')
 345             and info_dict.get('protocol')
 346             and not params.get('allow_unplayable_formats')
 347             and 'no-direct-merge' not in params.get('compat_opts', [])
 348             and cls.can_download(info_dict))
 349
 350     def _call_downloader(self, tmpfilename, info_dict):
 351         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 352         ffpp = FFmpegPostProcessor(downloader=self)
 353         if not ffpp.available:
 354             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 355             return False
 356         ffpp.check_version()
 357
 358         args = [ffpp.executable, '-y']
 359
 360         for log_level in ('quiet', 'verbose'):
 361             if self.params.get(log_level, False):
 362                 args += ['-loglevel', log_level]
 363                 break
 364         if not self.params.get('verbose'):
 365             args += ['-hide_banner']
 366
 367         args += traverse_obj(info_dict, ('downloader_options', 'ffmpeg_args'), default=[])
 368
 369         # These exists only for compatibility. Extractors should use
 370         # info_dict['downloader_options']['ffmpeg_args'] instead
 371         args += info_dict.get('_ffmpeg_args')
 372         seekable = info_dict.get('_seekable')
 373         if seekable is not None:
 374             # setting -seekable prevents ffmpeg from guessing if the server
 375             # supports seeking(by adding the header `Range: bytes=0-`), which
 376             # can cause problems in some cases
 377             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 378             # http://trac.ffmpeg.org/ticket/6125#comment:10
 379             args += ['-seekable', '1' if seekable else '0']
 380
 381         # start_time = info_dict.get('start_time') or 0
 382         # if start_time:
 383         #     args += ['-ss', compat_str(start_time)]
 384         # end_time = info_dict.get('end_time')
 385         # if end_time:
 386         #     args += ['-t', compat_str(end_time - start_time)]
 387
 388         http_headers = None
 389         if info_dict.get('http_headers'):
 390             youtubedl_headers = handle_youtubedl_headers(info_dict['http_headers'])
 391             http_headers = [
 392                 # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 393                 # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 394                 '-headers',
 395                 ''.join(f'{key}: {val}\r\n' for key, val in youtubedl_headers.items())
 396             ]
 397
 398         env = None
 399         proxy = self.params.get('proxy')
 400         if proxy:
 401             if not re.match(r'^[\da-zA-Z]+://', proxy):
 402                 proxy = 'http://%s' % proxy
 403
 404             if proxy.startswith('socks'):
 405                 self.report_warning(
 406                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 407                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 408
 409             # Since December 2015 ffmpeg supports -http_proxy option (see
 410             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 411             # We could switch to the following code if we are able to detect version properly
 412             # args += ['-http_proxy', proxy]
 413             env = os.environ.copy()
 414             compat_setenv('HTTP_PROXY', proxy, env=env)
 415             compat_setenv('http_proxy', proxy, env=env)
 416
 417         protocol = info_dict.get('protocol')
 418
 419         if protocol == 'rtmp':
 420             player_url = info_dict.get('player_url')
 421             page_url = info_dict.get('page_url')
 422             app = info_dict.get('app')
 423             play_path = info_dict.get('play_path')
 424             tc_url = info_dict.get('tc_url')
 425             flash_version = info_dict.get('flash_version')
 426             live = info_dict.get('rtmp_live', False)
 427             conn = info_dict.get('rtmp_conn')
 428             if player_url is not None:
 429                 args += ['-rtmp_swfverify', player_url]
 430             if page_url is not None:
 431                 args += ['-rtmp_pageurl', page_url]
 432             if app is not None:
 433                 args += ['-rtmp_app', app]
 434             if play_path is not None:
 435                 args += ['-rtmp_playpath', play_path]
 436             if tc_url is not None:
 437                 args += ['-rtmp_tcurl', tc_url]
 438             if flash_version is not None:
 439                 args += ['-rtmp_flashver', flash_version]
 440             if live:
 441                 args += ['-rtmp_live', 'live']
 442             if isinstance(conn, list):
 443                 for entry in conn:
 444                     args += ['-rtmp_conn', entry]
 445             elif isinstance(conn, compat_str):
 446                 args += ['-rtmp_conn', conn]
 447
 448         for i, url in enumerate(urls):
 449             # We need to specify headers for each http input stream
 450             # otherwise, it will only be applied to the first.
 451             # https://github.com/yt-dlp/yt-dlp/issues/2696
 452             if http_headers is not None and re.match(r'^https?://', url):
 453                 args += http_headers
 454             args += self._configuration_args((f'_i{i + 1}', '_i')) + ['-i', url]
 455
 456         args += ['-c', 'copy']
 457         if info_dict.get('requested_formats') or protocol == 'http_dash_segments':
 458             for (i, fmt) in enumerate(info_dict.get('requested_formats') or [info_dict]):
 459                 stream_number = fmt.get('manifest_stream_number', 0)
 460                 args.extend(['-map', f'{i}:{stream_number}'])
 461
 462         if self.params.get('test', False):
 463             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 464
 465         ext = info_dict['ext']
 466         if protocol in ('m3u8', 'm3u8_native'):
 467             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 468             if use_mpegts is None:
 469                 use_mpegts = info_dict.get('is_live')
 470             if use_mpegts:
 471                 args += ['-f', 'mpegts']
 472             else:
 473                 args += ['-f', 'mp4']
 474                 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')):
 475                     args += ['-bsf:a', 'aac_adtstoasc']
 476         elif protocol == 'rtmp':
 477             args += ['-f', 'flv']
 478         elif ext == 'mp4' and tmpfilename == '-':
 479             args += ['-f', 'mpegts']
 480         elif ext == 'unknown_video':
 481             ext = determine_ext(remove_end(tmpfilename, '.part'))
 482             if ext == 'unknown_video':
 483                 self.report_warning(
 484                     'The video format is unknown and cannot be downloaded by ffmpeg. '
 485                     'Explicitly set the extension in the filename to attempt download in that format')
 486             else:
 487                 self.report_warning(f'The video format is unknown. Trying to download as {ext} according to the filename')
 488                 args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 489         else:
 490             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 491
 492         args += self._configuration_args(('_o1', '_o', ''))
 493
 494         args = [encodeArgument(opt) for opt in args]
 495         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 496         self._debug_cmd(args)
 497
 498         proc = Popen(args, stdin=subprocess.PIPE, env=env)
 499         if url in ('-', 'pipe:'):
 500             self.on_process_started(proc, proc.stdin)
 501         try:
 502             retval = proc.wait()
 503         except BaseException as e:
 504             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 505             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 506             # produces a file that is playable (this is mostly useful for live
 507             # streams). Note that Windows is not affected and produces playable
 508             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 509             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 510                 proc.communicate_or_kill(b'q')
 511             else:
 512                 proc.kill()
 513                 proc.wait()
 514             raise
 515         return retval
 516
 517
 518 class AVconvFD(FFmpegFD):
 519     pass
 520
 521
 522 _BY_NAME = {
 523     klass.get_basename(): klass
 524     for name, klass in globals().items()
 525     if name.endswith('FD') and name not in ('ExternalFD', 'FragmentFD')
 526 }
 527
 528 _BY_EXE = {klass.EXE_NAME: klass for klass in _BY_NAME.values()}
 529
 530
 531 def list_external_downloaders():
 532     return sorted(_BY_NAME.keys())
 533
 534
 535 def get_external_downloader(external_downloader):
 536     """ Given the name of the executable, see whether we support the given
 537         downloader . """
 538     # Drop .exe extension on Windows
 539     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 540     return _BY_NAME.get(bn, _BY_EXE.get(bn))