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