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     error_to_compat_str,
  28     encodeArgument,
  29     handle_youtubedl_headers,
  30     check_executable,
  31     is_outdated_version,
  32     process_communicate_or_kill,
  33     sanitized_Request,
  34     sanitize_open,
  35 )
  36
  37
  38 class ExternalFD(FileDownloader):
  39     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps')
  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)
  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.params.get('external_downloader')
  86
  87     @classmethod
  88     def available(cls, path=None):
  89         return check_executable(path or cls.get_basename(), [cls.AVAILABLE_OPT])
  90
  91     @classmethod
  92     def supports(cls, info_dict):
  93         return info_dict['protocol'] in cls.SUPPORTED_PROTOCOLS
  94
  95     @classmethod
  96     def can_download(cls, info_dict, path=None):
  97         return cls.available(path) and cls.supports(info_dict)
  98
  99     def _option(self, command_option, param):
 100         return cli_option(self.params, command_option, param)
 101
 102     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
 103         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
 104
 105     def _valueless_option(self, command_option, param, expected_value=True):
 106         return cli_valueless_option(self.params, command_option, param, expected_value)
 107
 108     def _configuration_args(self, *args, **kwargs):
 109         return cli_configuration_args(
 110             self.params.get('external_downloader_args'),
 111             [self.get_basename(), 'default'],
 112             *args, **kwargs)
 113
 114     def _call_downloader(self, tmpfilename, info_dict):
 115         """ Either overwrite this or implement _make_cmd """
 116         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 117
 118         self._debug_cmd(cmd)
 119
 120         p = subprocess.Popen(
 121             cmd, stderr=subprocess.PIPE)
 122         _, stderr = process_communicate_or_kill(p)
 123         if p.returncode != 0:
 124             self.to_stderr(stderr.decode('utf-8', 'replace'))
 125
 126         if 'fragments' in info_dict:
 127             file_list = []
 128             dest, _ = sanitize_open(tmpfilename, 'wb')
 129             for i, fragment in enumerate(info_dict['fragments']):
 130                 file = '%s-Frag%d' % (tmpfilename, i)
 131                 decrypt_info = fragment.get('decrypt_info')
 132                 src, _ = sanitize_open(file, 'rb')
 133                 if decrypt_info:
 134                     if decrypt_info['METHOD'] == 'AES-128':
 135                         iv = decrypt_info.get('IV')
 136                         decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(
 137                             self._prepare_url(info_dict, info_dict.get('_decryption_key_url') or decrypt_info['URI'])).read()
 138                         encrypted_data = src.read()
 139                         decrypted_data = AES.new(
 140                             decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(encrypted_data)
 141                         dest.write(decrypted_data)
 142                     else:
 143                         fragment_data = src.read()
 144                         dest.write(fragment_data)
 145                 else:
 146                     fragment_data = src.read()
 147                     dest.write(fragment_data)
 148                 src.close()
 149                 file_list.append(file)
 150             dest.close()
 151             if not self.params.get('keep_fragments', False):
 152                 for file_path in file_list:
 153                     try:
 154                         os.remove(file_path)
 155                     except OSError as ose:
 156                         self.report_error("Unable to delete file %s; %s" % (file_path, error_to_compat_str(ose)))
 157                 try:
 158                     file_path = '%s.frag.urls' % tmpfilename
 159                     os.remove(file_path)
 160                 except OSError as ose:
 161                     self.report_error("Unable to delete file %s; %s" % (file_path, error_to_compat_str(ose)))
 162
 163         return p.returncode
 164
 165     def _prepare_url(self, info_dict, url):
 166         headers = info_dict.get('http_headers')
 167         return sanitized_Request(url, None, headers) if headers else url
 168
 169
 170 class CurlFD(ExternalFD):
 171     AVAILABLE_OPT = '-V'
 172
 173     def _make_cmd(self, tmpfilename, info_dict):
 174         cmd = [self.exe, '--location', '-o', tmpfilename]
 175         if info_dict.get('http_headers') is not None:
 176             for key, val in info_dict['http_headers'].items():
 177                 cmd += ['--header', '%s: %s' % (key, val)]
 178
 179         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 180         cmd += self._valueless_option('--silent', 'noprogress')
 181         cmd += self._valueless_option('--verbose', 'verbose')
 182         cmd += self._option('--limit-rate', 'ratelimit')
 183         retry = self._option('--retry', 'retries')
 184         if len(retry) == 2:
 185             if retry[1] in ('inf', 'infinite'):
 186                 retry[1] = '2147483647'
 187             cmd += retry
 188         cmd += self._option('--max-filesize', 'max_filesize')
 189         cmd += self._option('--interface', 'source_address')
 190         cmd += self._option('--proxy', 'proxy')
 191         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 192         cmd += self._configuration_args()
 193         cmd += ['--', info_dict['url']]
 194         return cmd
 195
 196     def _call_downloader(self, tmpfilename, info_dict):
 197         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 198
 199         self._debug_cmd(cmd)
 200
 201         # curl writes the progress to stderr so don't capture it.
 202         p = subprocess.Popen(cmd)
 203         process_communicate_or_kill(p)
 204         return p.returncode
 205
 206
 207 class AxelFD(ExternalFD):
 208     AVAILABLE_OPT = '-V'
 209
 210     def _make_cmd(self, tmpfilename, info_dict):
 211         cmd = [self.exe, '-o', tmpfilename]
 212         if info_dict.get('http_headers') is not None:
 213             for key, val in info_dict['http_headers'].items():
 214                 cmd += ['-H', '%s: %s' % (key, val)]
 215         cmd += self._configuration_args()
 216         cmd += ['--', info_dict['url']]
 217         return cmd
 218
 219
 220 class WgetFD(ExternalFD):
 221     AVAILABLE_OPT = '--version'
 222
 223     def _make_cmd(self, tmpfilename, info_dict):
 224         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 225         if info_dict.get('http_headers') is not None:
 226             for key, val in info_dict['http_headers'].items():
 227                 cmd += ['--header', '%s: %s' % (key, val)]
 228         cmd += self._option('--limit-rate', 'ratelimit')
 229         retry = self._option('--tries', 'retries')
 230         if len(retry) == 2:
 231             if retry[1] in ('inf', 'infinite'):
 232                 retry[1] = '0'
 233             cmd += retry
 234         cmd += self._option('--bind-address', 'source_address')
 235         cmd += self._option('--proxy', 'proxy')
 236         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 237         cmd += self._configuration_args()
 238         cmd += ['--', info_dict['url']]
 239         return cmd
 240
 241
 242 class Aria2cFD(ExternalFD):
 243     AVAILABLE_OPT = '-v'
 244     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'frag_urls')
 245
 246     @staticmethod
 247     def supports_manifest(manifest):
 248         UNSUPPORTED_FEATURES = [
 249             r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [1]
 250             # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
 251         ]
 252         check_results = (not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES)
 253         return all(check_results)
 254
 255     def _make_cmd(self, tmpfilename, info_dict):
 256         cmd = [self.exe, '-c']
 257         dn = os.path.dirname(tmpfilename)
 258         if 'fragments' not in info_dict:
 259             cmd += ['--out', os.path.basename(tmpfilename)]
 260         verbose_level_args = ['--console-log-level=warn', '--summary-interval=0']
 261         cmd += self._configuration_args(['--file-allocation=none', '-x16', '-j16', '-s16'] + verbose_level_args)
 262         if dn:
 263             cmd += ['--dir', dn]
 264         if info_dict.get('http_headers') is not None:
 265             for key, val in info_dict['http_headers'].items():
 266                 cmd += ['--header', '%s: %s' % (key, val)]
 267         cmd += self._option('--interface', 'source_address')
 268         cmd += self._option('--all-proxy', 'proxy')
 269         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 270         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 271         cmd += ['--auto-file-renaming=false']
 272         if 'fragments' in info_dict:
 273             cmd += verbose_level_args
 274             cmd += ['--uri-selector', 'inorder', '--download-result=hide']
 275             url_list_file = '%s.frag.urls' % tmpfilename
 276             url_list = []
 277             for i, fragment in enumerate(info_dict['fragments']):
 278                 tmpsegmentname = '%s-Frag%d' % (os.path.basename(tmpfilename), i)
 279                 url_list.append('%s\n\tout=%s' % (fragment['url'], tmpsegmentname))
 280             stream, _ = sanitize_open(url_list_file, 'wb')
 281             stream.write('\n'.join(url_list).encode('utf-8'))
 282             stream.close()
 283
 284             cmd += ['-i', url_list_file]
 285         else:
 286             cmd += ['--', info_dict['url']]
 287         return cmd
 288
 289
 290 class HttpieFD(ExternalFD):
 291     @classmethod
 292     def available(cls, path=None):
 293         return check_executable(path or 'http', ['--version'])
 294
 295     def _make_cmd(self, tmpfilename, info_dict):
 296         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 297
 298         if info_dict.get('http_headers') is not None:
 299             for key, val in info_dict['http_headers'].items():
 300                 cmd += ['%s:%s' % (key, val)]
 301         return cmd
 302
 303
 304 class FFmpegFD(ExternalFD):
 305     SUPPORTED_PROTOCOLS = ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 306
 307     @classmethod
 308     def available(cls, path=None):  # path is ignored for ffmpeg
 309         return FFmpegPostProcessor().available
 310
 311     def _call_downloader(self, tmpfilename, info_dict):
 312         url = info_dict['url']
 313         ffpp = FFmpegPostProcessor(downloader=self)
 314         if not ffpp.available:
 315             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
 316             return False
 317         ffpp.check_version()
 318
 319         args = [ffpp.executable, '-y']
 320
 321         for log_level in ('quiet', 'verbose'):
 322             if self.params.get(log_level, False):
 323                 args += ['-loglevel', log_level]
 324                 break
 325
 326         seekable = info_dict.get('_seekable')
 327         if seekable is not None:
 328             # setting -seekable prevents ffmpeg from guessing if the server
 329             # supports seeking(by adding the header `Range: bytes=0-`), which
 330             # can cause problems in some cases
 331             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 332             # http://trac.ffmpeg.org/ticket/6125#comment:10
 333             args += ['-seekable', '1' if seekable else '0']
 334
 335         args += self._configuration_args()
 336
 337         # start_time = info_dict.get('start_time') or 0
 338         # if start_time:
 339         #     args += ['-ss', compat_str(start_time)]
 340         # end_time = info_dict.get('end_time')
 341         # if end_time:
 342         #     args += ['-t', compat_str(end_time - start_time)]
 343
 344         if info_dict.get('http_headers') is not None and re.match(r'^https?://', url):
 345             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 346             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 347             headers = handle_youtubedl_headers(info_dict['http_headers'])
 348             args += [
 349                 '-headers',
 350                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 351
 352         env = None
 353         proxy = self.params.get('proxy')
 354         if proxy:
 355             if not re.match(r'^[\da-zA-Z]+://', proxy):
 356                 proxy = 'http://%s' % proxy
 357
 358             if proxy.startswith('socks'):
 359                 self.report_warning(
 360                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 361                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 362
 363             # Since December 2015 ffmpeg supports -http_proxy option (see
 364             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 365             # We could switch to the following code if we are able to detect version properly
 366             # args += ['-http_proxy', proxy]
 367             env = os.environ.copy()
 368             compat_setenv('HTTP_PROXY', proxy, env=env)
 369             compat_setenv('http_proxy', proxy, env=env)
 370
 371         protocol = info_dict.get('protocol')
 372
 373         if protocol == 'rtmp':
 374             player_url = info_dict.get('player_url')
 375             page_url = info_dict.get('page_url')
 376             app = info_dict.get('app')
 377             play_path = info_dict.get('play_path')
 378             tc_url = info_dict.get('tc_url')
 379             flash_version = info_dict.get('flash_version')
 380             live = info_dict.get('rtmp_live', False)
 381             conn = info_dict.get('rtmp_conn')
 382             if player_url is not None:
 383                 args += ['-rtmp_swfverify', player_url]
 384             if page_url is not None:
 385                 args += ['-rtmp_pageurl', page_url]
 386             if app is not None:
 387                 args += ['-rtmp_app', app]
 388             if play_path is not None:
 389                 args += ['-rtmp_playpath', play_path]
 390             if tc_url is not None:
 391                 args += ['-rtmp_tcurl', tc_url]
 392             if flash_version is not None:
 393                 args += ['-rtmp_flashver', flash_version]
 394             if live:
 395                 args += ['-rtmp_live', 'live']
 396             if isinstance(conn, list):
 397                 for entry in conn:
 398                     args += ['-rtmp_conn', entry]
 399             elif isinstance(conn, compat_str):
 400                 args += ['-rtmp_conn', conn]
 401
 402         args += ['-i', url, '-c', 'copy']
 403
 404         if self.params.get('test', False):
 405             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 406
 407         if protocol in ('m3u8', 'm3u8_native'):
 408             use_mpegts = (tmpfilename == '-') or self.params.get('hls_use_mpegts')
 409             if use_mpegts is None:
 410                 use_mpegts = info_dict.get('is_live')
 411             if use_mpegts:
 412                 args += ['-f', 'mpegts']
 413             else:
 414                 args += ['-f', 'mp4']
 415                 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')):
 416                     args += ['-bsf:a', 'aac_adtstoasc']
 417         elif protocol == 'rtmp':
 418             args += ['-f', 'flv']
 419         else:
 420             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 421
 422         args = [encodeArgument(opt) for opt in args]
 423         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 424
 425         self._debug_cmd(args)
 426
 427         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 428         try:
 429             retval = proc.wait()
 430         except BaseException as e:
 431             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 432             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 433             # produces a file that is playable (this is mostly useful for live
 434             # streams). Note that Windows is not affected and produces playable
 435             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 436             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32':
 437                 process_communicate_or_kill(proc, b'q')
 438             else:
 439                 proc.kill()
 440                 proc.wait()
 441             raise
 442         return retval
 443
 444
 445 class AVconvFD(FFmpegFD):
 446     pass
 447
 448
 449 _BY_NAME = dict(
 450     (klass.get_basename(), klass)
 451     for name, klass in globals().items()
 452     if name.endswith('FD') and name != 'ExternalFD'
 453 )
 454
 455
 456 def list_external_downloaders():
 457     return sorted(_BY_NAME.keys())
 458
 459
 460 def get_external_downloader(external_downloader):
 461     """ Given the name of the executable, see whether we support the given
 462         downloader . """
 463     # Drop .exe extension on Windows
 464     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 465     return _BY_NAME[bn]