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     can_download_to_stdout = False
  40
  41     def real_download(self, filename, info_dict):
  42         self.report_destination(filename)
  43         tmpfilename = self.temp_name(filename)
  44
  45         try:
  46             started = time.time()
  47             retval = self._call_downloader(tmpfilename, info_dict)
  48         except KeyboardInterrupt:
  49             if not info_dict.get('is_live'):
  50                 raise
  51             # Live stream downloading cancellation should be considered as
  52             # correct and expected termination thus all postprocessing
  53             # should take place
  54             retval = 0
  55             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  56
  57         if retval == 0:
  58             status = {
  59                 'filename': filename,
  60                 'status': 'finished',
  61                 'elapsed': time.time() - started,
  62             }
  63             if filename != '-':
  64                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  65                 self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  66                 self.try_rename(tmpfilename, filename)
  67                 status.update({
  68                     'downloaded_bytes': fsize,
  69                     'total_bytes': fsize,
  70                 })
  71             self._hook_progress(status, info_dict)
  72             return True
  73         else:
  74             self.to_stderr('\n')
  75             self.report_error('%s exited with code %d' % (
  76                 self.get_basename(), retval))
  77             return False
  78
  79     @classmethod
  80     def get_basename(cls):
  81         return cls.__name__[:-2].lower()
  82
  83     @property
  84     def exe(self):
  85         return self.get_basename()
  86
  87     @classmethod
  88     def available(cls, path=None):
  89         path = check_executable(path or cls.get_basename(), [cls.AVAILABLE_OPT])
  90         if path:
  91             cls.exe = path
  92             return path
  93         return False
  94
  95     @classmethod
  96     def supports(cls, info_dict):
  97         return (
  98             (cls.can_download_to_stdout or not info_dict.get('to_stdout'))
  99             and info_dict['protocol'] in cls.SUPPORTED_PROTOCOLS)
 100
 101     @classmethod
 102     def can_download(cls, info_dict, path=None):
 103         return cls.available(path) and cls.supports(info_dict)
 104
 105     def _option(self, command_option, param):
 106         return cli_option(self.params, command_option, param)
 107
 108     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
 109         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
 110
 111     def _valueless_option(self, command_option, param, expected_value=True):
 112         return cli_valueless_option(self.params, command_option, param, expected_value)
 113
 114     def _configuration_args(self, *args, **kwargs):
 115         return cli_configuration_args(
 116             self.params.get('external_downloader_args'),
 117             [self.get_basename(), 'default'],
 118             *args, **kwargs)
 119
 120     def _call_downloader(self, tmpfilename, info_dict):
 121         """ Either overwrite this or implement _make_cmd """
 122         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 123
 124         self._debug_cmd(cmd)
 125
 126         if 'fragments' in info_dict:
 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 = subprocess.Popen(
 133                     cmd, stderr=subprocess.PIPE)
 134                 _, stderr = process_communicate_or_kill(p)
 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             dest, _ = 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, _ = sanitize_open(fragment_filename, 'rb')
 155                 except IOError:
 156                     if skip_unavailable_fragments and frag_index > 1:
 157                         self.to_screen('[%s] Skipping fragment %d ...' % (self.get_basename(), frag_index))
 158                         continue
 159                     self.report_error('Unable to open fragment %d' % frag_index)
 160                     return -1
 161                 decrypt_info = fragment.get('decrypt_info')
 162                 if decrypt_info:
 163                     if decrypt_info['METHOD'] == 'AES-128':
 164                         iv = decrypt_info.get('IV')
 165                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 166                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 167                         encrypted_data = src.read()
 168                         decrypted_data = AES.new(
 169                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(encrypted_data)
 170                         dest.write(decrypted_data)
 171                     else:
 172                         fragment_data = src.read()
 173                         dest.write(fragment_data)
 174                 else:
 175                     fragment_data = src.read()
 176                     dest.write(fragment_data)
 177                 src.close()
 178                 if not self.params.get('keep_fragments', False):
 179                     os.remove(encodeFilename(fragment_filename))
 180             dest.close()
 181             os.remove(encodeFilename('%s.frag.urls' % tmpfilename))
 182         else:
 183             p = subprocess.Popen(
 184                 cmd, stderr=subprocess.PIPE)
 185             _, stderr = process_communicate_or_kill(p)
 186             if p.returncode != 0:
 187                 self.to_stderr(stderr.decode('utf-8', 'replace'))
 188         return p.returncode
 189
 190     def _prepare_url(self, info_dict, url):
 191         headers = info_dict.get('http_headers')
 192         return sanitized_Request(url, None, headers) if headers else url
 193
 194
 195 class CurlFD(ExternalFD):
 196     AVAILABLE_OPT = '-V'
 197
 198     def _make_cmd(self, tmpfilename, info_dict):
 199         cmd = [self.exe, '--location', '-o', tmpfilename]
 200         if info_dict.get('http_headers') is not None:
 201             for key, val in info_dict['http_headers'].items():
 202                 cmd += ['--header', '%s: %s' % (key, val)]
 203
 204         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 205         cmd += self._valueless_option('--silent', 'noprogress')
 206         cmd += self._valueless_option('--verbose', 'verbose')
 207         cmd += self._option('--limit-rate', 'ratelimit')
 208         retry = self._option('--retry', 'retries')
 209         if len(retry) == 2:
 210             if retry[1] in ('inf', 'infinite'):
 211                 retry[1] = '2147483647'
 212             cmd += retry
 213         cmd += self._option('--max-filesize', 'max_filesize')
 214         cmd += self._option('--interface', 'source_address')
 215         cmd += self._option('--proxy', 'proxy')
 216         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 217         cmd += self._configuration_args()
 218         cmd += ['--', info_dict['url']]
 219         return cmd
 220
 221     def _call_downloader(self, tmpfilename, info_dict):
 222         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 223
 224         self._debug_cmd(cmd)
 225
 226         # curl writes the progress to stderr so don't capture it.
 227         p = subprocess.Popen(cmd)
 228         process_communicate_or_kill(p)
 229         return p.returncode
 230
 231
 232 class AxelFD(ExternalFD):
 233     AVAILABLE_OPT = '-V'
 234
 235     def _make_cmd(self, tmpfilename, info_dict):
 236         cmd = [self.exe, '-o', tmpfilename]
 237         if info_dict.get('http_headers') is not None:
 238             for key, val in info_dict['http_headers'].items():
 239                 cmd += ['-H', '%s: %s' % (key, val)]
 240         cmd += self._configuration_args()
 241         cmd += ['--', info_dict['url']]
 242         return cmd
 243
 244
 245 class WgetFD(ExternalFD):
 246     AVAILABLE_OPT = '--version'
 247
 248     def _make_cmd(self, tmpfilename, info_dict):
 249         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 250         if info_dict.get('http_headers') is not None:
 251             for key, val in info_dict['http_headers'].items():
 252                 cmd += ['--header', '%s: %s' % (key, val)]
 253         cmd += self._option('--limit-rate', 'ratelimit')
 254         retry = self._option('--tries', 'retries')
 255         if len(retry) == 2:
 256             if retry[1] in ('inf', 'infinite'):
 257                 retry[1] = '0'
 258             cmd += retry
 259         cmd += self._option('--bind-address', 'source_address')
 260         cmd += self._option('--proxy', 'proxy')
 261         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 262         cmd += self._configuration_args()
 263         cmd += ['--', info_dict['url']]
 264         return cmd
 265
 266
 267 class Aria2cFD(ExternalFD):
 268     AVAILABLE_OPT = '-v'
 269     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'dash_frag_urls', 'm3u8_frag_urls')
 270
 271     @staticmethod
 272     def supports_manifest(manifest):
 273         UNSUPPORTED_FEATURES = [
 274             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 275             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 276         ]
 277         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 278         return all(check_results)
 279
 280     def _make_cmd(self, tmpfilename, info_dict):
 281         cmd = [self.exe, '-c',
 282                '--console-log-level=warn', '--summary-interval=0', '--download-result=hide',
 283                '--file-allocation=none', '-x16', '-j16', '-s16']
 284         if 'fragments' in info_dict:
 285             cmd += ['--allow-overwrite=true', '--allow-piece-length-change=true']
 286         else:
 287             cmd += ['--min-split-size', '1M']
 288
 289         if info_dict.get('http_headers') is not None:
 290             for key, val in info_dict['http_headers'].items():
 291                 cmd += ['--header', '%s: %s' % (key, val)]
 292         cmd += self._option('--interface', 'source_address')
 293         cmd += self._option('--all-proxy', 'proxy')
 294         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 295         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 296         cmd += self._configuration_args()
 297
 298         # aria2c strips out spaces from the beginning/end of filenames and paths.
 299         # We work around this issue by adding a "./" to the beginning of the
 300         # filename and relative path, and adding a "/" at the end of the path.
 301         # See: https://github.com/yt-dlp/yt-dlp/issues/276
 302         # https://github.com/ytdl-org/youtube-dl/issues/20312
 303         # https://github.com/aria2/aria2/issues/1373
 304         dn = os.path.dirname(tmpfilename)
 305         if dn:
 306             if not os.path.isabs(dn):
 307                 dn = '.%s%s' % (os.path.sep, dn)
 308             cmd += ['--dir', dn + os.path.sep]
 309         if 'fragments' not in info_dict:
 310             cmd += ['--out', '.%s%s' % (os.path.sep, os.path.basename(tmpfilename))]
 311         cmd += ['--auto-file-renaming=false']
 312
 313         if 'fragments' in info_dict:
 314             cmd += ['--file-allocation=none', '--uri-selector=inorder']
 315             url_list_file = '%s.frag.urls' % tmpfilename
 316             url_list = []
 317             for frag_index, fragment in enumerate(info_dict['fragments']):
 318                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
 319                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
 320             stream, _ = sanitize_open(url_list_file, 'wb')
 321             stream.write('\n'.join(url_list).encode('utf-8'))
 322             stream.close()
 323             cmd += ['-i', url_list_file]
 324         else:
 325             cmd += ['--', info_dict['url']]
 326         return cmd
 327
 328
 329 class HttpieFD(ExternalFD):
 330     AVAILABLE_OPT = '--version'
 331
 332     @classmethod
 333     def available(cls, path=None):
 334         return ExternalFD.available(cls, path or 'http')
 335
 336     def _make_cmd(self, tmpfilename, info_dict):
 337         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 338
 339         if info_dict.get('http_headers') is not None:
 340             for key, val in info_dict['http_headers'].items():
 341                 cmd += ['%s:%s' % (key, val)]
 342         return cmd
 343
 344
 345 class FFmpegFD(ExternalFD):
 346     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'm3u8_native', 'rtsp', 'rtmp', 'rtmp_ffmpeg', 'mms')
 347     can_download_to_stdout = True
 348
 349     @classmethod
 350     def available(cls, path=None):
 351         # TODO: Fix path for ffmpeg
 352         # Fixme: This may be wrong when --ffmpeg-location is used
 353         return FFmpegPostProcessor().available
 354
 355     def on_process_started(self, proc, stdin):
 356         """ Override this in subclasses  """
 357         pass
 358
 359     @classmethod
 360     def can_merge_formats(cls, info_dict, params={}):
 361         return (
 362             info_dict.get('requested_formats')
 363             and info_dict.get('protocol')
 364             and not params.get('allow_unplayable_formats')
 365             and 'no-direct-merge' not in params.get('compat_opts', [])
 366             and cls.can_download(info_dict))
 367
 368     def _call_downloader(self, tmpfilename, info_dict):
 369         urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
 370         ffpp = FFmpegPostProcessor(downloader=self)
 371         if not ffpp.available:
 372             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 373             return False
 374         ffpp.check_version()
 375
 376         args = [ffpp.executable, '-y']
 377
 378         for log_level in ('quiet', 'verbose'):
 379             if self.params.get(log_level, False):
 380                 args += ['-loglevel', log_level]
 381                 break
 382         if not self.params.get('verbose'):
 383             args += ['-hide_banner']
 384
 385         seekable = info_dict.get('_seekable')
 386         if seekable is not None:
 387             # setting -seekable prevents ffmpeg from guessing if the server
 388             # supports seeking(by adding the header `Range: bytes=0-`), which
 389             # can cause problems in some cases
 390             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 391             # http://trac.ffmpeg.org/ticket/6125#comment:10
 392             args += ['-seekable', '1' if seekable else '0']
 393
 394         # start_time = info_dict.get('start_time') or 0
 395         # if start_time:
 396         #     args += ['-ss', compat_str(start_time)]
 397         # end_time = info_dict.get('end_time')
 398         # if end_time:
 399         #     args += ['-t', compat_str(end_time - start_time)]
 400
 401         if info_dict.get('http_headers') is not None and re.match(r'^https?://', urls[0]):
 402             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 403             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 404             headers = handle_youtubedl_headers(info_dict['http_headers'])
 405             args += [
 406                 '-headers',
 407                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 408
 409         env = None
 410         proxy = self.params.get('proxy')
 411         if proxy:
 412             if not re.match(r'^[\da-zA-Z]+://', proxy):
 413                 proxy = 'http://%s' % proxy
 414
 415             if proxy.startswith('socks'):
 416                 self.report_warning(
 417                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 418                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 419
 420             # Since December 2015 ffmpeg supports -http_proxy option (see
 421             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 422             # We could switch to the following code if we are able to detect version properly
 423             # args += ['-http_proxy', proxy]
 424             env = os.environ.copy()
 425             compat_setenv('HTTP_PROXY', proxy, env=env)
 426             compat_setenv('http_proxy', proxy, env=env)
 427
 428         protocol = info_dict.get('protocol')
 429
 430         if protocol == 'rtmp':
 431             player_url = info_dict.get('player_url')
 432             page_url = info_dict.get('page_url')
 433             app = info_dict.get('app')
 434             play_path = info_dict.get('play_path')
 435             tc_url = info_dict.get('tc_url')
 436             flash_version = info_dict.get('flash_version')
 437             live = info_dict.get('rtmp_live', False)
 438             conn = info_dict.get('rtmp_conn')
 439             if player_url is not None:
 440                 args += ['-rtmp_swfverify', player_url]
 441             if page_url is not None:
 442                 args += ['-rtmp_pageurl', page_url]
 443             if app is not None:
 444                 args += ['-rtmp_app', app]
 445             if play_path is not None:
 446                 args += ['-rtmp_playpath', play_path]
 447             if tc_url is not None:
 448                 args += ['-rtmp_tcurl', tc_url]
 449             if flash_version is not None:
 450                 args += ['-rtmp_flashver', flash_version]
 451             if live:
 452                 args += ['-rtmp_live', 'live']
 453             if isinstance(conn, list):
 454                 for entry in conn:
 455                     args += ['-rtmp_conn', entry]
 456             elif isinstance(conn, compat_str):
 457                 args += ['-rtmp_conn', conn]
 458
 459         for url in urls:
 460             args += ['-i', url]
 461
 462         args += self._configuration_args() + ['-c', 'copy']
 463         if info_dict.get('requested_formats'):
 464             for (i, fmt) in enumerate(info_dict['requested_formats']):
 465                 if fmt.get('acodec') != 'none':
 466                     args.extend(['-map', '%d:a:0' % i])
 467                 if fmt.get('vcodec') != 'none':
 468                     args.extend(['-map', '%d:v:0' % i])
 469
 470         if self.params.get('test', False):
 471             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 472
 473         ext = info_dict['ext']
 474         if protocol in ('m3u8', 'm3u8_native'):
 475             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 476             if use_mpegts is None:
 477                 use_mpegts = info_dict.get('is_live')
 478             if use_mpegts:
 479                 args += ['-f', 'mpegts']
 480             else:
 481                 args += ['-f', 'mp4']
 482                 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')):
 483                     args += ['-bsf:a', 'aac_adtstoasc']
 484         elif protocol == 'rtmp':
 485             args += ['-f', 'flv']
 486         elif ext == 'mp4' and tmpfilename == '-':
 487             args += ['-f', 'mpegts']
 488         else:
 489             args += ['-f', EXT_TO_OUT_FORMATS.get(ext, ext)]
 490
 491         args = [encodeArgument(opt) for opt in args]
 492         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 493
 494         self._debug_cmd(args)
 495
 496         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 497         if url in ('-', 'pipe:'):
 498             self.on_process_started(proc, proc.stdin)
 499         try:
 500             retval = proc.wait()
 501         except BaseException as e:
 502             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 503             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 504             # produces a file that is playable (this is mostly useful for live
 505             # streams). Note that Windows is not affected and produces playable
 506             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 507             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32' and url not in ('-', 'pipe:'):
 508                 process_communicate_or_kill(proc, b'q')
 509             else:
 510                 proc.kill()
 511                 proc.wait()
 512             raise
 513         return retval
 514
 515
 516 class AVconvFD(FFmpegFD):
 517     pass
 518
 519
 520 _BY_NAME = dict(
 521     (klass.get_basename(), klass)
 522     for name, klass in globals().items()
 523     if name.endswith('FD') and name != 'ExternalFD'
 524 )
 525
 526
 527 def list_external_downloaders():
 528     return sorted(_BY_NAME.keys())
 529
 530
 531 def get_external_downloader(external_downloader):
 532     """ Given the name of the executable, see whether we support the given
 533         downloader . """
 534     # Drop .exe extension on Windows
 535     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 536     return _BY_NAME.get(bn)