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