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