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