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