yt_dlp/downloader/external.py

   1 from __future__ import unicode_literals
   2
   3 import os.path
   4 import re
   5 import subprocess
   6 import sys
   7 import time
   8
   9 from .common import FileDownloader
  10 from ..aes import aes_cbc_decrypt_bytes
  11 from ..compat import (
  12     compat_setenv,
  13     compat_str,
  14     compat_struct_pack,
  15 )
  16 from ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS
  17 from ..utils import (
  18     cli_option,
  19     cli_valueless_option,
  20     cli_bool_option,
  21     _configuration_args,
  22     encodeFilename,
  23     encodeArgument,
  24     handle_youtubedl_headers,
  25     check_executable,
  26     is_outdated_version,
  27     process_communicate_or_kill,
  28     sanitized_Request,
  29     sanitize_open,
  30 )
  31
  32
  33 class ExternalFD(FileDownloader):
  34     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps')
  35     can_download_to_stdout = False
  36
  37     def real_download(self, filename, info_dict):
  38         self.report_destination(filename)
  39         tmpfilename = self.temp_name(filename)
  40
  41         try:
  42             started = time.time()
  43             retval = self._call_downloader(tmpfilename, info_dict)
  44         except KeyboardInterrupt:
  45             if not info_dict.get('is_live'):
  46                 raise
  47             # Live stream downloading cancellation should be considered as
  48             # correct and expected termination thus all postprocessing
  49             # should take place
  50             retval = 0
  51             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  52
  53         if retval == 0:
  54             status = {
  55                 'filename': filename,
  56                 'status': 'finished',
  57                 'elapsed': time.time() - started,
  58             }
  59             if filename != '-':
  60                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  61                 self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  62                 self.try_rename(tmpfilename, filename)
  63                 status.update({
  64                     'downloaded_bytes': fsize,
  65                     'total_bytes': fsize,
  66                 })
  67             self._hook_progress(status, info_dict)
  68             return True
  69         else:
  70             self.to_stderr('\n')
  71             self.report_error('%s exited with code %d' % (
  72                 self.get_basename(), retval))
  73             return False
  74
  75     @classmethod
  76     def get_basename(cls):
  77         return cls.__name__[:-2].lower()
  78
  79     @property
  80     def exe(self):
  81         return self.get_basename()
  82
  83     @classmethod
  84     def available(cls, path=None):
  85         path = check_executable(path or cls.get_basename(), [cls.AVAILABLE_OPT])
  86         if path:
  87             cls.exe = path
  88             return path
  89         return False
  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.get_basename(),
 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' in info_dict:
 122             fragment_retries = self.params.get('fragment_retries', 0)
 123             skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 124
 125             count = 0
 126             while count <= fragment_retries:
 127                 p = subprocess.Popen(
 128                     cmd, stderr=subprocess.PIPE)
 129                 _, stderr = process_communicate_or_kill(p)
 130                 if p.returncode == 0:
 131                     break
 132                 # TODO: Decide whether to retry based on error code
 133                 # https://aria2.github.io/manual/en/html/aria2c.html#exit-status
 134                 self.to_stderr(stderr.decode('utf-8', 'replace'))
 135                 count += 1
 136                 if count <= fragment_retries:
 137                     self.to_screen(
 138                         '[%s] Got error. Retrying fragments (attempt %d of %s)...'
 139                         % (self.get_basename(), count, self.format_retries(fragment_retries)))
 140             if count > fragment_retries:
 141                 if not skip_unavailable_fragments:
 142                     self.report_error('Giving up after %s fragment retries' % fragment_retries)
 143                     return -1
 144
 145             dest, _ = sanitize_open(tmpfilename, 'wb')
 146             for frag_index, fragment in enumerate(info_dict['fragments']):
 147                 fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
 148                 try:
 149                     src, _ = sanitize_open(fragment_filename, 'rb')
 150                 except IOError:
 151                     if skip_unavailable_fragments and frag_index > 1:
 152                         self.to_screen('[%s] Skipping fragment %d ...' % (self.get_basename(), frag_index))
 153                         continue
 154                     self.report_error('Unable to open fragment %d' % frag_index)
 155                     return -1
 156                 decrypt_info = fragment.get('decrypt_info')
 157                 if decrypt_info:
 158                     if decrypt_info['METHOD'] == 'AES-128':
 159                         iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', fragment['media_sequence'])
 160                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 161                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 162                         encrypted_data = src.read()
 163                         decrypted_data = aes_cbc_decrypt_bytes(encrypted_data, decrypt_info['KEY'], iv)
 164                         dest.write(decrypted_data)
 165                     else:
 166                         fragment_data = src.read()
 167                         dest.write(fragment_data)
 168                 else:
 169                     fragment_data = src.read()
 170                     dest.write(fragment_data)
 171                 src.close()
 172                 if not self.params.get('keep_fragments', False):
 173                     os.remove(encodeFilename(fragment_filename))
 174             dest.close()
 175             os.remove(encodeFilename('%s.frag.urls' % tmpfilename))
 176         else:
 177             p = subprocess.Popen(
 178                 cmd, stderr=subprocess.PIPE)
 179             _, stderr = process_communicate_or_kill(p)
 180             if p.returncode != 0:
 181                 self.to_stderr(stderr.decode('utf-8', 'replace'))
 182         return p.returncode
 183
 184     def _prepare_url(self, info_dict, url):
 185         headers = info_dict.get('http_headers')
 186         return sanitized_Request(url, None, headers) if headers else url
 187
 188
 189 class CurlFD(ExternalFD):
 190     AVAILABLE_OPT = '-V'
 191
 192     def _make_cmd(self, tmpfilename, info_dict):
 193         cmd = [self.exe, '--location', '-o', tmpfilename]
 194         if info_dict.get('http_headers') is not None:
 195             for key, val in info_dict['http_headers'].items():
 196                 cmd += ['--header', '%s: %s' % (key, val)]
 197
 198         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 199         cmd += self._valueless_option('--silent', 'noprogress')
 200         cmd += self._valueless_option('--verbose', 'verbose')
 201         cmd += self._option('--limit-rate', 'ratelimit')
 202         retry = self._option('--retry', 'retries')
 203         if len(retry) == 2:
 204             if retry[1] in ('inf', 'infinite'):
 205                 retry[1] = '2147483647'
 206             cmd += retry
 207         cmd += self._option('--max-filesize', 'max_filesize')
 208         cmd += self._option('--interface', 'source_address')
 209         cmd += self._option('--proxy', 'proxy')
 210         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 211         cmd += self._configuration_args()
 212         cmd += ['--', info_dict['url']]
 213         return cmd
 214
 215     def _call_downloader(self, tmpfilename, info_dict):
 216         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 217
 218         self._debug_cmd(cmd)
 219
 220         # curl writes the progress to stderr so don't capture it.
 221         p = subprocess.Popen(cmd)
 222         process_communicate_or_kill(p)
 223         return p.returncode
 224
 225
 226 class AxelFD(ExternalFD):
 227     AVAILABLE_OPT = '-V'
 228
 229     def _make_cmd(self, tmpfilename, info_dict):
 230         cmd = [self.exe, '-o', tmpfilename]
 231         if info_dict.get('http_headers') is not None:
 232             for key, val in info_dict['http_headers'].items():
 233                 cmd += ['-H', '%s: %s' % (key, val)]
 234         cmd += self._configuration_args()
 235         cmd += ['--', info_dict['url']]
 236         return cmd
 237
 238
 239 class WgetFD(ExternalFD):
 240     AVAILABLE_OPT = '--version'
 241
 242     def _make_cmd(self, tmpfilename, info_dict):
 243         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 244         if info_dict.get('http_headers') is not None:
 245             for key, val in info_dict['http_headers'].items():
 246                 cmd += ['--header', '%s: %s' % (key, val)]
 247         cmd += self._option('--limit-rate', 'ratelimit')
 248         retry = self._option('--tries', 'retries')
 249         if len(retry) == 2:
 250             if retry[1] in ('inf', 'infinite'):
 251                 retry[1] = '0'
 252             cmd += retry
 253         cmd += self._option('--bind-address', 'source_address')
 254         cmd += self._option('--proxy', 'proxy')
 255         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 256         cmd += self._configuration_args()
 257         cmd += ['--', info_dict['url']]
 258         return cmd
 259
 260
 261 class Aria2cFD(ExternalFD):
 262     AVAILABLE_OPT = '-v'
 263     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 264
 265     @staticmethod
 266     def supports_manifest(manifest):
 267         UNSUPPORTED_FEATURES = [
 268             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 269             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 270         ]
 271         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 272         return all(check_results)
 273
 274     def _make_cmd(self, tmpfilename, info_dict):
 275         cmd = [self.exe, '-c',
 276                '--console-log-level=warn', '--summary-interval=0', '--download-result=hide',
 277                '--file-allocation=none', '-x16', '-j16', '-s16']
 278         if 'fragments' in info_dict:
 279             cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true']
 280         else:
 281             cmd += ['--min-split-size', '1M']
 282
 283         if info_dict.get('http_headers') is not None:
 284             for key, val in info_dict['http_headers'].items():
 285                 cmd += ['--header', '%s: %s' % (key, val)]
 286         cmd += self._option('--max-overall-download-limit', 'ratelimit')
 287         cmd += self._option('--interface', 'source_address')
 288         cmd += self._option('--all-proxy', 'proxy')
 289         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 290         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 291         cmd += self._configuration_args()
 292
 293         # aria2c strips out spaces from the beginning/end of filenames and paths.
 294         # We work around this issue by adding a "./" to the beginning of the
 295         # filename and relative path, and adding a "/" at the end of the path.
 296         # See: https://github.com/yt-dlp/yt-dlp/issues/276
 297         # https://github.com/ytdl-org/youtube-dl/issues/20312
 298         # https://github.com/aria2/aria2/issues/1373
 299         dn = os.path.dirname(tmpfilename)
 300         if dn:
 301             if not os.path.isabs(dn):
 302                 dn = '.%s%s' % (os.path.sep, dn)
 303             cmd += ['--dir', dn + os.path.sep]
 304         if 'fragments' not in info_dict:
 305             cmd += ['--out', '.%s%s' % (os.path.sep, os.path.basename(tmpfilename))]
 306         cmd += ['--auto-file-renaming=false']
 307
 308         if 'fragments' in info_dict:
 309             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 310             url_list_file = '%s.frag.urls' % tmpfilename
 311             url_list = []
 312             for frag_index, fragment in enumerate(info_dict['fragments']):
 313                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 314                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
 315             stream, _ = sanitize_open(url_list_file, 'wb')
 316             stream.write('\n'.join(url_list).encode('utf-8'))
 317             stream.close()
 318             cmd += ['-i', url_list_file]
 319         else:
 320             cmd += ['--', info_dict['url']]
 321         return cmd
 322
 323
 324 class HttpieFD(ExternalFD):
 325     AVAILABLE_OPT = '--version'
 326
 327     @classmethod
 328     def available(cls, path=None):
 329         return ExternalFD.available(cls, path or 'http')
 330
 331     def _make_cmd(self, tmpfilename, info_dict):
 332         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 333
 334         if info_dict.get('http_headers') is not None:
 335             for key, val in info_dict['http_headers'].items():
 336                 cmd += ['%s:%s' % (key, val)]
 337         return cmd
 338
 339
 340 class FFmpegFD(ExternalFD):
 341     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms', 'http_dash_segments')
 342     can_download_to_stdout = True
 343
 344     @classmethod
 345     def available(cls, path=None):
 346         # TODO: Fix path for ffmpeg
 347         # Fixme: This may be wrong when --ffmpeg-location is used
 348         return FFmpegPostProcessor().available
 349
 350     def on_process_started(self, proc, stdin):
 351         """ Override this in subclasses  """
 352         pass
 353
 354     @classmethod
 355     def can_merge_formats(cls, info_dict, params):
 356         return (
 357             info_dict.get('requested_formats')
 358             and info_dict.get('protocol')
 359             and not params.get('allow_unplayable_formats')
 360             and 'no-direct-merge' not in params.get('compat_opts', [])
 361             and cls.can_download(info_dict))
 362
 363     def _call_downloader(self, tmpfilename, info_dict):
 364         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 365         ffpp = FFmpegPostProcessor(downloader=self)
 366         if not ffpp.available:
 367             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 368             return False
 369         ffpp.check_version()
 370
 371         args = [ffpp.executable, '-y']
 372
 373         for log_level in ('quiet', 'verbose'):
 374             if self.params.get(log_level, False):
 375                 args += ['-loglevel', log_level]
 376                 break
 377         if not self.params.get('verbose'):
 378             args += ['-hide_banner']
 379
 380         args += info_dict.get('_ffmpeg_args', [])
 381
 382         # This option exists only for compatibility. Extractors should use `_ffmpeg_args` instead
 383         seekable = info_dict.get('_seekable')
 384         if seekable is not None:
 385             # setting -seekable prevents ffmpeg from guessing if the server
 386             # supports seeking(by adding the header `Range: bytes=0-`), which
 387             # can cause problems in some cases
 388             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 389             # http://trac.ffmpeg.org/ticket/6125#comment:10
 390             args += ['-seekable', '1' if seekable else '0']
 391
 392         # start_time = info_dict.get('start_time') or 0
 393         # if start_time:
 394         #     args += ['-ss', compat_str(start_time)]
 395         # end_time = info_dict.get('end_time')
 396         # if end_time:
 397         #     args += ['-t', compat_str(end_time - start_time)]
 398
 399         if info_dict.get('http_headers') is not None and re.match(r'^https?://', urls[0]):
 400             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 401             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 402             headers = handle_youtubedl_headers(info_dict['http_headers'])
 403             args += [
 404                 '-headers',
 405                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 406
 407         env = None
 408         proxy = self.params.get('proxy')
 409         if proxy:
 410             if not re.match(r'^[\da-zA-Z]+://', proxy):
 411                 proxy = 'http://%s' % proxy
 412
 413             if proxy.startswith('socks'):
 414                 self.report_warning(
 415                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 416                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 417
 418             # Since December 2015 ffmpeg supports -http_proxy option (see
 419             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 420             # We could switch to the following code if we are able to detect version properly
 421             # args += ['-http_proxy', proxy]
 422             env = os.environ.copy()
 423             compat_setenv('HTTP_PROXY', proxy, env=env)
 424             compat_setenv('http_proxy', proxy, env=env)
 425
 426         protocol = info_dict.get('protocol')
 427
 428         if protocol == 'rtmp':
 429             player_url = info_dict.get('player_url')
 430             page_url = info_dict.get('page_url')
 431             app = info_dict.get('app')
 432             play_path = info_dict.get('play_path')
 433             tc_url = info_dict.get('tc_url')
 434             flash_version = info_dict.get('flash_version')
 435             live = info_dict.get('rtmp_live', False)
 436             conn = info_dict.get('rtmp_conn')
 437             if player_url is not None:
 438                 args += ['-rtmp_swfverify', player_url]
 439             if page_url is not None:
 440                 args += ['-rtmp_pageurl', page_url]
 441             if app is not None:
 442                 args += ['-rtmp_app', app]
 443             if play_path is not None:
 444                 args += ['-rtmp_playpath', play_path]
 445             if tc_url is not None:
 446                 args += ['-rtmp_tcurl', tc_url]
 447             if flash_version is not None:
 448                 args += ['-rtmp_flashver', flash_version]
 449             if live:
 450                 args += ['-rtmp_live', 'live']
 451             if isinstance(conn, list):
 452                 for entry in conn:
 453                     args += ['-rtmp_conn', entry]
 454             elif isinstance(conn, compat_str):
 455                 args += ['-rtmp_conn', conn]
 456
 457         for i, url in enumerate(urls):
 458             args += self._configuration_args((f'_i{i + 1}', '_i')) + ['-i', url]
 459
 460         args += ['-c', 'copy']
 461         if info_dict.get('requested_formats') or protocol == 'http_dash_segments':
 462             for (i, fmt) in enumerate(info_dict.get('requested_formats') or [info_dict]):
 463                 stream_number = fmt.get('manifest_stream_number', 0)
 464                 a_or_v = 'a' if fmt.get('acodec') != 'none' else 'v'
 465                 args.extend(['-map', f'{i}:{a_or_v}:{stream_number}'])
 466
 467         if self.params.get('test', False):
 468             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 469
 470         ext = info_dict['ext']
 471         if protocol in ('m3u8', 'm3u8_native'):
 472             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 473             if use_mpegts is None:
 474                 use_mpegts = info_dict.get('is_live')
 475             if use_mpegts:
 476                 args += ['-f', 'mpegts']
 477             else:
 478                 args += ['-f', 'mp4']
 479                 if (ffpp.basename == 'ffmpeg' and is_outdated_version(ffpp._versions['ffmpeg'], '3.2', False)) and (not info_dict.get('acodec') or info_dict['acodec'].split('.')[0] in ('aac', 'mp4a')):
 480                     args += ['-bsf:a', 'aac_adtstoasc']
 481         elif protocol == 'rtmp':
 482             args += ['-f', 'flv']
 483         elif ext == 'mp4' and tmpfilename == '-':
 484             args += ['-f', 'mpegts']
 485         else:
 486             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 487
 488         args += self._configuration_args(('_o1', '_o', ''))
 489
 490         args = [encodeArgument(opt) for opt in args]
 491         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 492         self._debug_cmd(args)
 493
 494         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 495         if url in ('-', 'pipe:'):
 496             self.on_process_started(proc, proc.stdin)
 497         try:
 498             retval = proc.wait()
 499         except BaseException as e:
 500             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 501             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 502             # produces a file that is playable (this is mostly useful for live
 503             # streams). Note that Windows is not affected and produces playable
 504             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 505             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 506                 process_communicate_or_kill(proc, b'q')
 507             else:
 508                 proc.kill()
 509                 proc.wait()
 510             raise
 511         return retval
 512
 513
 514 class AVconvFD(FFmpegFD):
 515     pass
 516
 517
 518 _BY_NAME = dict(
 519     (klass.get_basename(), klass)
 520     for name, klass in globals().items()
 521     if name.endswith('FD') and name != 'ExternalFD'
 522 )
 523
 524
 525 def list_external_downloaders():
 526     return sorted(_BY_NAME.keys())
 527
 528
 529 def get_external_downloader(external_downloader):
 530     """ Given the name of the executable, see whether we support the given
 531         downloader . """
 532     # Drop .exe extension on Windows
 533     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 534     return _BY_NAME.get(bn)