youtube_dl/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 .common import FileDownloader
  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     cli_configuration_args,
  20     encodeFilename,
  21     encodeArgument,
  22     handle_youtubedl_headers,
  23     check_executable,
  24     is_outdated_version,
  25     process_communicate_or_kill,
  26 )
  27
  28
  29 class ExternalFD(FileDownloader):
  30     def real_download(self, filename, info_dict):
  31         self.report_destination(filename)
  32         tmpfilename = self.temp_name(filename)
  33
  34         try:
  35             started = time.time()
  36             retval = self._call_downloader(tmpfilename, info_dict)
  37         except KeyboardInterrupt:
  38             if not info_dict.get('is_live'):
  39                 raise
  40             # Live stream downloading cancellation should be considered as
  41             # correct and expected termination thus all postprocessing
  42             # should take place
  43             retval = 0
  44             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  45
  46         if retval == 0:
  47             status = {
  48                 'filename': filename,
  49                 'status': 'finished',
  50                 'elapsed': time.time() - started,
  51             }
  52             if filename != '-':
  53                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  54                 self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  55                 self.try_rename(tmpfilename, filename)
  56                 status.update({
  57                     'downloaded_bytes': fsize,
  58                     'total_bytes': fsize,
  59                 })
  60             self._hook_progress(status)
  61             return True
  62         else:
  63             self.to_stderr('\n')
  64             self.report_error('%s exited with code %d' % (
  65                 self.get_basename(), retval))
  66             return False
  67
  68     @classmethod
  69     def get_basename(cls):
  70         return cls.__name__[:-2].lower()
  71
  72     @property
  73     def exe(self):
  74         return self.params.get('external_downloader')
  75
  76     @classmethod
  77     def available(cls):
  78         return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT])
  79
  80     @classmethod
  81     def supports(cls, info_dict):
  82         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps')
  83
  84     @classmethod
  85     def can_download(cls, info_dict):
  86         return cls.available() and cls.supports(info_dict)
  87
  88     def _option(self, command_option, param):
  89         return cli_option(self.params, command_option, param)
  90
  91     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
  92         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
  93
  94     def _valueless_option(self, command_option, param, expected_value=True):
  95         return cli_valueless_option(self.params, command_option, param, expected_value)
  96
  97     def _configuration_args(self, default=[]):
  98         return cli_configuration_args(self.params, 'external_downloader_args', default)
  99
 100     def _call_downloader(self, tmpfilename, info_dict):
 101         """ Either overwrite this or implement _make_cmd """
 102         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 103
 104         self._debug_cmd(cmd)
 105
 106         p = subprocess.Popen(
 107             cmd, stderr=subprocess.PIPE)
 108         _, stderr = process_communicate_or_kill(p)
 109         if p.returncode != 0:
 110             self.to_stderr(stderr.decode('utf-8', 'replace'))
 111         return p.returncode
 112
 113
 114 class CurlFD(ExternalFD):
 115     AVAILABLE_OPT = '-V'
 116
 117     def _make_cmd(self, tmpfilename, info_dict):
 118         cmd = [self.exe, '--location', '-o', tmpfilename]
 119         for key, val in info_dict['http_headers'].items():
 120             cmd += ['--header', '%s: %s' % (key, val)]
 121         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 122         cmd += self._valueless_option('--silent', 'noprogress')
 123         cmd += self._valueless_option('--verbose', 'verbose')
 124         cmd += self._option('--limit-rate', 'ratelimit')
 125         retry = self._option('--retry', 'retries')
 126         if len(retry) == 2:
 127             if retry[1] in ('inf', 'infinite'):
 128                 retry[1] = '2147483647'
 129             cmd += retry
 130         cmd += self._option('--max-filesize', 'max_filesize')
 131         cmd += self._option('--interface', 'source_address')
 132         cmd += self._option('--proxy', 'proxy')
 133         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 134         cmd += self._configuration_args()
 135         cmd += ['--', info_dict['url']]
 136         return cmd
 137
 138     def _call_downloader(self, tmpfilename, info_dict):
 139         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 140
 141         self._debug_cmd(cmd)
 142
 143         # curl writes the progress to stderr so don't capture it.
 144         p = subprocess.Popen(cmd)
 145         process_communicate_or_kill(p)
 146         return p.returncode
 147
 148
 149 class AxelFD(ExternalFD):
 150     AVAILABLE_OPT = '-V'
 151
 152     def _make_cmd(self, tmpfilename, info_dict):
 153         cmd = [self.exe, '-o', tmpfilename]
 154         for key, val in info_dict['http_headers'].items():
 155             cmd += ['-H', '%s: %s' % (key, val)]
 156         cmd += self._configuration_args()
 157         cmd += ['--', info_dict['url']]
 158         return cmd
 159
 160
 161 class WgetFD(ExternalFD):
 162     AVAILABLE_OPT = '--version'
 163
 164     def _make_cmd(self, tmpfilename, info_dict):
 165         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 166         for key, val in info_dict['http_headers'].items():
 167             cmd += ['--header', '%s: %s' % (key, val)]
 168         cmd += self._option('--limit-rate', 'ratelimit')
 169         retry = self._option('--tries', 'retries')
 170         if len(retry) == 2:
 171             if retry[1] in ('inf', 'infinite'):
 172                 retry[1] = '0'
 173             cmd += retry
 174         cmd += self._option('--bind-address', 'source_address')
 175         cmd += self._option('--proxy', 'proxy')
 176         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 177         cmd += self._configuration_args()
 178         cmd += ['--', info_dict['url']]
 179         return cmd
 180
 181
 182 class Aria2cFD(ExternalFD):
 183     AVAILABLE_OPT = '-v'
 184
 185     def _make_cmd(self, tmpfilename, info_dict):
 186         cmd = [self.exe, '-c']
 187         cmd += self._configuration_args([
 188             '--min-split-size', '1M', '--max-connection-per-server', '4'])
 189         dn = os.path.dirname(tmpfilename)
 190         if dn:
 191             cmd += ['--dir', dn]
 192         cmd += ['--out', os.path.basename(tmpfilename)]
 193         for key, val in info_dict['http_headers'].items():
 194             cmd += ['--header', '%s: %s' % (key, val)]
 195         cmd += self._option('--interface', 'source_address')
 196         cmd += self._option('--all-proxy', 'proxy')
 197         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 198         cmd += self._bool_option('--remote-time', 'updatetime', 'true', 'false', '=')
 199         cmd += ['--', info_dict['url']]
 200         return cmd
 201
 202
 203 class HttpieFD(ExternalFD):
 204     @classmethod
 205     def available(cls):
 206         return check_executable('http', ['--version'])
 207
 208     def _make_cmd(self, tmpfilename, info_dict):
 209         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 210         for key, val in info_dict['http_headers'].items():
 211             cmd += ['%s:%s' % (key, val)]
 212         return cmd
 213
 214
 215 class FFmpegFD(ExternalFD):
 216     @classmethod
 217     def supports(cls, info_dict):
 218         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 219
 220     @classmethod
 221     def available(cls):
 222         return FFmpegPostProcessor().available
 223
 224     def _call_downloader(self, tmpfilename, info_dict):
 225         url = info_dict['url']
 226         ffpp = FFmpegPostProcessor(downloader=self)
 227         if not ffpp.available:
 228             self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
 229             return False
 230         ffpp.check_version()
 231
 232         args = [ffpp.executable, '-y']
 233
 234         for log_level in ('quiet', 'verbose'):
 235             if self.params.get(log_level, False):
 236                 args += ['-loglevel', log_level]
 237                 break
 238
 239         seekable = info_dict.get('_seekable')
 240         if seekable is not None:
 241             # setting -seekable prevents ffmpeg from guessing if the server
 242             # supports seeking(by adding the header `Range: bytes=0-`), which
 243             # can cause problems in some cases
 244             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 245             # http://trac.ffmpeg.org/ticket/6125#comment:10
 246             args += ['-seekable', '1' if seekable else '0']
 247
 248         args += self._configuration_args()
 249
 250         # start_time = info_dict.get('start_time') or 0
 251         # if start_time:
 252         #     args += ['-ss', compat_str(start_time)]
 253         # end_time = info_dict.get('end_time')
 254         # if end_time:
 255         #     args += ['-t', compat_str(end_time - start_time)]
 256
 257         if info_dict['http_headers'] and re.match(r'^https?://', url):
 258             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 259             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 260             headers = handle_youtubedl_headers(info_dict['http_headers'])
 261             args += [
 262                 '-headers',
 263                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 264
 265         env = None
 266         proxy = self.params.get('proxy')
 267         if proxy:
 268             if not re.match(r'^[\da-zA-Z]+://', proxy):
 269                 proxy = 'http://%s' % proxy
 270
 271             if proxy.startswith('socks'):
 272                 self.report_warning(
 273                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 274                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 275
 276             # Since December 2015 ffmpeg supports -http_proxy option (see
 277             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 278             # We could switch to the following code if we are able to detect version properly
 279             # args += ['-http_proxy', proxy]
 280             env = os.environ.copy()
 281             compat_setenv('HTTP_PROXY', proxy, env=env)
 282             compat_setenv('http_proxy', proxy, env=env)
 283
 284         protocol = info_dict.get('protocol')
 285
 286         if protocol == 'rtmp':
 287             player_url = info_dict.get('player_url')
 288             page_url = info_dict.get('page_url')
 289             app = info_dict.get('app')
 290             play_path = info_dict.get('play_path')
 291             tc_url = info_dict.get('tc_url')
 292             flash_version = info_dict.get('flash_version')
 293             live = info_dict.get('rtmp_live', False)
 294             conn = info_dict.get('rtmp_conn')
 295             if player_url is not None:
 296                 args += ['-rtmp_swfverify', player_url]
 297             if page_url is not None:
 298                 args += ['-rtmp_pageurl', page_url]
 299             if app is not None:
 300                 args += ['-rtmp_app', app]
 301             if play_path is not None:
 302                 args += ['-rtmp_playpath', play_path]
 303             if tc_url is not None:
 304                 args += ['-rtmp_tcurl', tc_url]
 305             if flash_version is not None:
 306                 args += ['-rtmp_flashver', flash_version]
 307             if live:
 308                 args += ['-rtmp_live', 'live']
 309             if isinstance(conn, list):
 310                 for entry in conn:
 311                     args += ['-rtmp_conn', entry]
 312             elif isinstance(conn, compat_str):
 313                 args += ['-rtmp_conn', conn]
 314
 315         args += ['-i', url, '-c', 'copy']
 316
 317         if self.params.get('test', False):
 318             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 319
 320         if protocol in ('m3u8', 'm3u8_native'):
 321             if self.params.get('hls_use_mpegts', False) or tmpfilename == '-':
 322                 args += ['-f', 'mpegts']
 323             else:
 324                 args += ['-f', 'mp4']
 325                 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')):
 326                     args += ['-bsf:a', 'aac_adtstoasc']
 327         elif protocol == 'rtmp':
 328             args += ['-f', 'flv']
 329         else:
 330             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 331
 332         args = [encodeArgument(opt) for opt in args]
 333         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 334
 335         self._debug_cmd(args)
 336
 337         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 338         try:
 339             retval = proc.wait()
 340         except BaseException as e:
 341             # subprocess.run would send the SIGKILL signal to ffmpeg and the
 342             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 343             # produces a file that is playable (this is mostly useful for live
 344             # streams). Note that Windows is not affected and produces playable
 345             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 346             if isinstance(e, KeyboardInterrupt) and sys.platform != 'win32':
 347                 process_communicate_or_kill(proc, b'q')
 348             else:
 349                 proc.kill()
 350                 proc.wait()
 351             raise
 352         return retval
 353
 354
 355 class AVconvFD(FFmpegFD):
 356     pass
 357
 358
 359 _BY_NAME = dict(
 360     (klass.get_basename(), klass)
 361     for name, klass in globals().items()
 362     if name.endswith('FD') and name != 'ExternalFD'
 363 )
 364
 365
 366 def list_external_downloaders():
 367     return sorted(_BY_NAME.keys())
 368
 369
 370 def get_external_downloader(external_downloader):
 371     """ Given the name of the executable, see whether we support the given
 372         downloader . """
 373     # Drop .exe extension on Windows
 374     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 375     return _BY_NAME[bn]