_ Git - youtube-dl/blob - youtube_dl/downloader/external.py

   1 from __future__ import unicode_literals
   2
   3 import os.path
   4 import subprocess
   5 import sys
   6 import re
   7
   8 from .common import FileDownloader
   9 from ..compat import (
  10     compat_setenv,
  11     compat_str,
  12 )
  13 from ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS
  14 from ..utils import (
  15     cli_option,
  16     cli_valueless_option,
  17     cli_bool_option,
  18     cli_configuration_args,
  19     encodeFilename,
  20     encodeArgument,
  21     handle_youtubedl_headers,
  22     check_executable,
  23     is_outdated_version,
  24 )
  25
  26
  27 class ExternalFD(FileDownloader):
  28     def real_download(self, filename, info_dict):
  29         self.report_destination(filename)
  30         tmpfilename = self.temp_name(filename)
  31
  32         try:
  33             retval = self._call_downloader(tmpfilename, info_dict)
  34         except KeyboardInterrupt:
  35             if not info_dict.get('is_live'):
  36                 raise
  37             # Live stream downloading cancellation should be considered as
  38             # correct and expected termination thus all postprocessing
  39             # should take place
  40             retval = 0
  41             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  42
  43         if retval == 0:
  44             if filename == '-':
  45                 self._hook_progress({
  46                     'filename': filename,
  47                     'status': 'finished',
  48                 })
  49             else:
  50                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  51                 self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  52                 self.try_rename(tmpfilename, filename)
  53                 self._hook_progress({
  54                     'downloaded_bytes': fsize,
  55                     'total_bytes': fsize,
  56                     'filename': filename,
  57                     'status': 'finished',
  58                 })
  59             return True
  60         else:
  61             self.to_stderr('\n')
  62             self.report_error('%s exited with code %d' % (
  63                 self.get_basename(), retval))
  64             return False
  65
  66     @classmethod
  67     def get_basename(cls):
  68         return cls.__name__[:-2].lower()
  69
  70     @property
  71     def exe(self):
  72         return self.params.get('external_downloader')
  73
  74     @classmethod
  75     def available(cls):
  76         return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT])
  77
  78     @classmethod
  79     def supports(cls, info_dict):
  80         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps')
  81
  82     @classmethod
  83     def can_download(cls, info_dict):
  84         return cls.available() and cls.supports(info_dict)
  85
  86     def _option(self, command_option, param):
  87         return cli_option(self.params, command_option, param)
  88
  89     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
  90         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
  91
  92     def _valueless_option(self, command_option, param, expected_value=True):
  93         return cli_valueless_option(self.params, command_option, param, expected_value)
  94
  95     def _configuration_args(self, default=[]):
  96         return cli_configuration_args(self.params, 'external_downloader_args', default)
  97
  98     def _call_downloader(self, tmpfilename, info_dict):
  99         """ Either overwrite this or implement _make_cmd """
 100         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 101
 102         self._debug_cmd(cmd)
 103
 104         p = subprocess.Popen(
 105             cmd, stderr=subprocess.PIPE)
 106         _, stderr = p.communicate()
 107         if p.returncode != 0:
 108             self.to_stderr(stderr.decode('utf-8', 'replace'))
 109         return p.returncode
 110
 111
 112 class CurlFD(ExternalFD):
 113     AVAILABLE_OPT = '-V'
 114
 115     def _make_cmd(self, tmpfilename, info_dict):
 116         cmd = [self.exe, '--location', '-o', tmpfilename]
 117         for key, val in info_dict['http_headers'].items():
 118             cmd += ['--header', '%s: %s' % (key, val)]
 119         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 120         cmd += self._valueless_option('--silent', 'noprogress')
 121         cmd += self._valueless_option('--verbose', 'verbose')
 122         cmd += self._option('--limit-rate', 'ratelimit')
 123         cmd += self._option('--retry', 'retries')
 124         cmd += self._option('--max-filesize', 'max_filesize')
 125         cmd += self._option('--interface', 'source_address')
 126         cmd += self._option('--proxy', 'proxy')
 127         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 128         cmd += self._configuration_args()
 129         cmd += ['--', info_dict['url']]
 130         return cmd
 131
 132     def _call_downloader(self, tmpfilename, info_dict):
 133         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 134
 135         self._debug_cmd(cmd)
 136
 137         # curl writes the progress to stderr so don't capture it.
 138         p = subprocess.Popen(cmd)
 139         p.communicate()
 140         return p.returncode
 141
 142
 143 class AxelFD(ExternalFD):
 144     AVAILABLE_OPT = '-V'
 145
 146     def _make_cmd(self, tmpfilename, info_dict):
 147         cmd = [self.exe, '-o', tmpfilename]
 148         for key, val in info_dict['http_headers'].items():
 149             cmd += ['-H', '%s: %s' % (key, val)]
 150         cmd += self._configuration_args()
 151         cmd += ['--', info_dict['url']]
 152         return cmd
 153
 154
 155 class WgetFD(ExternalFD):
 156     AVAILABLE_OPT = '--version'
 157
 158     def _make_cmd(self, tmpfilename, info_dict):
 159         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 160         for key, val in info_dict['http_headers'].items():
 161             cmd += ['--header', '%s: %s' % (key, val)]
 162         cmd += self._option('--bind-address', 'source_address')
 163         cmd += self._option('--proxy', 'proxy')
 164         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 165         cmd += self._configuration_args()
 166         cmd += ['--', info_dict['url']]
 167         return cmd
 168
 169
 170 class Aria2cFD(ExternalFD):
 171     AVAILABLE_OPT = '-v'
 172
 173     def _make_cmd(self, tmpfilename, info_dict):
 174         cmd = [self.exe, '-c']
 175         cmd += self._configuration_args([
 176             '--min-split-size', '1M', '--max-connection-per-server', '4'])
 177         dn = os.path.dirname(tmpfilename)
 178         if dn:
 179             cmd += ['--dir', dn]
 180         cmd += ['--out', os.path.basename(tmpfilename)]
 181         for key, val in info_dict['http_headers'].items():
 182             cmd += ['--header', '%s: %s' % (key, val)]
 183         cmd += self._option('--interface', 'source_address')
 184         cmd += self._option('--all-proxy', 'proxy')
 185         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 186         cmd += ['--', info_dict['url']]
 187         return cmd
 188
 189
 190 class HttpieFD(ExternalFD):
 191     @classmethod
 192     def available(cls):
 193         return check_executable('http', ['--version'])
 194
 195     def _make_cmd(self, tmpfilename, info_dict):
 196         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 197         for key, val in info_dict['http_headers'].items():
 198             cmd += ['%s:%s' % (key, val)]
 199         return cmd
 200
 201
 202 class FFmpegFD(ExternalFD):
 203     @classmethod
 204     def supports(cls, info_dict):
 205         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 206
 207     @classmethod
 208     def available(cls):
 209         return FFmpegPostProcessor().available
 210
 211     def _call_downloader(self, tmpfilename, info_dict):
 212         url = info_dict['url']
 213         ffpp = FFmpegPostProcessor(downloader=self)
 214         if not ffpp.available:
 215             self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
 216             return False
 217         ffpp.check_version()
 218
 219         args = [ffpp.executable, '-y']
 220
 221         for log_level in ('quiet', 'verbose'):
 222             if self.params.get(log_level, False):
 223                 args += ['-loglevel', log_level]
 224                 break
 225
 226         seekable = info_dict.get('_seekable')
 227         if seekable is not None:
 228             # setting -seekable prevents ffmpeg from guessing if the server
 229             # supports seeking(by adding the header `Range: bytes=0-`), which
 230             # can cause problems in some cases
 231             # https://github.com/rg3/youtube-dl/issues/11800#issuecomment-275037127
 232             # http://trac.ffmpeg.org/ticket/6125#comment:10
 233             args += ['-seekable', '1' if seekable else '0']
 234
 235         args += self._configuration_args()
 236
 237         # start_time = info_dict.get('start_time') or 0
 238         # if start_time:
 239         #     args += ['-ss', compat_str(start_time)]
 240         # end_time = info_dict.get('end_time')
 241         # if end_time:
 242         #     args += ['-t', compat_str(end_time - start_time)]
 243
 244         if info_dict['http_headers'] and re.match(r'^https?://', url):
 245             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 246             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 247             headers = handle_youtubedl_headers(info_dict['http_headers'])
 248             args += [
 249                 '-headers',
 250                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 251
 252         env = None
 253         proxy = self.params.get('proxy')
 254         if proxy:
 255             if not re.match(r'^[\da-zA-Z]+://', proxy):
 256                 proxy = 'http://%s' % proxy
 257
 258             if proxy.startswith('socks'):
 259                 self.report_warning(
 260                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 261                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 262
 263             # Since December 2015 ffmpeg supports -http_proxy option (see
 264             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 265             # We could switch to the following code if we are able to detect version properly
 266             # args += ['-http_proxy', proxy]
 267             env = os.environ.copy()
 268             compat_setenv('HTTP_PROXY', proxy, env=env)
 269             compat_setenv('http_proxy', proxy, env=env)
 270
 271         protocol = info_dict.get('protocol')
 272
 273         if protocol == 'rtmp':
 274             player_url = info_dict.get('player_url')
 275             page_url = info_dict.get('page_url')
 276             app = info_dict.get('app')
 277             play_path = info_dict.get('play_path')
 278             tc_url = info_dict.get('tc_url')
 279             flash_version = info_dict.get('flash_version')
 280             live = info_dict.get('rtmp_live', False)
 281             if player_url is not None:
 282                 args += ['-rtmp_swfverify', player_url]
 283             if page_url is not None:
 284                 args += ['-rtmp_pageurl', page_url]
 285             if app is not None:
 286                 args += ['-rtmp_app', app]
 287             if play_path is not None:
 288                 args += ['-rtmp_playpath', play_path]
 289             if tc_url is not None:
 290                 args += ['-rtmp_tcurl', tc_url]
 291             if flash_version is not None:
 292                 args += ['-rtmp_flashver', flash_version]
 293             if live:
 294                 args += ['-rtmp_live', 'live']
 295
 296         args += ['-i', url, '-c', 'copy']
 297
 298         if self.params.get('test', False):
 299             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 300
 301         if protocol in ('m3u8', 'm3u8_native'):
 302             if self.params.get('hls_use_mpegts', False) or tmpfilename == '-':
 303                 args += ['-f', 'mpegts']
 304             else:
 305                 args += ['-f', 'mp4']
 306                 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')):
 307                     args += ['-bsf:a', 'aac_adtstoasc']
 308         elif protocol == 'rtmp':
 309             args += ['-f', 'flv']
 310         else:
 311             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 312
 313         args = [encodeArgument(opt) for opt in args]
 314         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 315
 316         self._debug_cmd(args)
 317
 318         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 319         try:
 320             retval = proc.wait()
 321         except KeyboardInterrupt:
 322             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 323             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 324             # produces a file that is playable (this is mostly useful for live
 325             # streams). Note that Windows is not affected and produces playable
 326             # files (see https://github.com/rg3/youtube-dl/issues/8300).
 327             if sys.platform != 'win32':
 328                 proc.communicate(b'q')
 329             raise
 330         return retval
 331
 332
 333 class AVconvFD(FFmpegFD):
 334     pass
 335
 336
 337 _BY_NAME = dict(
 338     (klass.get_basename(), klass)
 339     for name, klass in globals().items()
 340     if name.endswith('FD') and name != 'ExternalFD'
 341 )
 342
 343
 344 def list_external_downloaders():
 345     return sorted(_BY_NAME.keys())
 346
 347
 348 def get_external_downloader(external_downloader):
 349     """ Given the name of the executable, see whether we support the given
 350         downloader . """
 351     # Drop .exe extension on Windows
 352     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 353     return _BY_NAME[bn]