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 ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS
  10 from ..utils import (
  11     cli_option,
  12     cli_valueless_option,
  13     cli_bool_option,
  14     cli_configuration_args,
  15     encodeFilename,
  16     encodeArgument,
  17     handle_youtubedl_headers,
  18     check_executable,
  19 )
  20
  21
  22 class ExternalFD(FileDownloader):
  23     def real_download(self, filename, info_dict):
  24         self.report_destination(filename)
  25         tmpfilename = self.temp_name(filename)
  26
  27         retval = self._call_downloader(tmpfilename, info_dict)
  28         if retval == 0:
  29             fsize = os.path.getsize(encodeFilename(tmpfilename))
  30             self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  31             self.try_rename(tmpfilename, filename)
  32             self._hook_progress({
  33                 'downloaded_bytes': fsize,
  34                 'total_bytes': fsize,
  35                 'filename': filename,
  36                 'status': 'finished',
  37             })
  38             return True
  39         else:
  40             self.to_stderr('\n')
  41             self.report_error('%s exited with code %d' % (
  42                 self.get_basename(), retval))
  43             return False
  44
  45     @classmethod
  46     def get_basename(cls):
  47         return cls.__name__[:-2].lower()
  48
  49     @property
  50     def exe(self):
  51         return self.params.get('external_downloader')
  52
  53     @classmethod
  54     def available(cls):
  55         return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT])
  56
  57     @classmethod
  58     def supports(cls, info_dict):
  59         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps')
  60
  61     @classmethod
  62     def can_download(cls, info_dict):
  63         return cls.available() and cls.supports(info_dict)
  64
  65     def _option(self, command_option, param):
  66         return cli_option(self.params, command_option, param)
  67
  68     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
  69         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
  70
  71     def _valueless_option(self, command_option, param, expected_value=True):
  72         return cli_valueless_option(self.params, command_option, param, expected_value)
  73
  74     def _configuration_args(self, default=[]):
  75         return cli_configuration_args(self.params, 'external_downloader_args', default)
  76
  77     def _call_downloader(self, tmpfilename, info_dict):
  78         """ Either overwrite this or implement _make_cmd """
  79         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
  80
  81         self._debug_cmd(cmd)
  82
  83         p = subprocess.Popen(
  84             cmd, stderr=subprocess.PIPE)
  85         _, stderr = p.communicate()
  86         if p.returncode != 0:
  87             self.to_stderr(stderr)
  88         return p.returncode
  89
  90
  91 class CurlFD(ExternalFD):
  92     AVAILABLE_OPT = '-V'
  93
  94     def _make_cmd(self, tmpfilename, info_dict):
  95         cmd = [self.exe, '--location', '-o', tmpfilename]
  96         for key, val in info_dict['http_headers'].items():
  97             cmd += ['--header', '%s: %s' % (key, val)]
  98         cmd += self._option('--interface', 'source_address')
  99         cmd += self._option('--proxy', 'proxy')
 100         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 101         cmd += self._configuration_args()
 102         cmd += ['--', info_dict['url']]
 103         return cmd
 104
 105
 106 class AxelFD(ExternalFD):
 107     AVAILABLE_OPT = '-V'
 108
 109     def _make_cmd(self, tmpfilename, info_dict):
 110         cmd = [self.exe, '-o', tmpfilename]
 111         for key, val in info_dict['http_headers'].items():
 112             cmd += ['-H', '%s: %s' % (key, val)]
 113         cmd += self._configuration_args()
 114         cmd += ['--', info_dict['url']]
 115         return cmd
 116
 117
 118 class WgetFD(ExternalFD):
 119     AVAILABLE_OPT = '--version'
 120
 121     def _make_cmd(self, tmpfilename, info_dict):
 122         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 123         for key, val in info_dict['http_headers'].items():
 124             cmd += ['--header', '%s: %s' % (key, val)]
 125         cmd += self._option('--bind-address', 'source_address')
 126         cmd += self._option('--proxy', 'proxy')
 127         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 128         cmd += self._configuration_args()
 129         cmd += ['--', info_dict['url']]
 130         return cmd
 131
 132
 133 class Aria2cFD(ExternalFD):
 134     AVAILABLE_OPT = '-v'
 135
 136     def _make_cmd(self, tmpfilename, info_dict):
 137         cmd = [self.exe, '-c']
 138         cmd += self._configuration_args([
 139             '--min-split-size', '1M', '--max-connection-per-server', '4'])
 140         dn = os.path.dirname(tmpfilename)
 141         if dn:
 142             cmd += ['--dir', dn]
 143         cmd += ['--out', os.path.basename(tmpfilename)]
 144         for key, val in info_dict['http_headers'].items():
 145             cmd += ['--header', '%s: %s' % (key, val)]
 146         cmd += self._option('--interface', 'source_address')
 147         cmd += self._option('--all-proxy', 'proxy')
 148         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 149         cmd += ['--', info_dict['url']]
 150         return cmd
 151
 152
 153 class HttpieFD(ExternalFD):
 154     @classmethod
 155     def available(cls):
 156         return check_executable('http', ['--version'])
 157
 158     def _make_cmd(self, tmpfilename, info_dict):
 159         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 160         for key, val in info_dict['http_headers'].items():
 161             cmd += ['%s:%s' % (key, val)]
 162         return cmd
 163
 164
 165 class FFmpegFD(ExternalFD):
 166     @classmethod
 167     def supports(cls, info_dict):
 168         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 169
 170     @classmethod
 171     def available(cls):
 172         return FFmpegPostProcessor().available
 173
 174     def _call_downloader(self, tmpfilename, info_dict):
 175         url = info_dict['url']
 176         ffpp = FFmpegPostProcessor(downloader=self)
 177         if not ffpp.available:
 178             self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
 179             return False
 180         ffpp.check_version()
 181
 182         args = [ffpp.executable, '-y']
 183
 184         args += self._configuration_args()
 185
 186         # start_time = info_dict.get('start_time') or 0
 187         # if start_time:
 188         #     args += ['-ss', compat_str(start_time)]
 189         # end_time = info_dict.get('end_time')
 190         # if end_time:
 191         #     args += ['-t', compat_str(end_time - start_time)]
 192
 193         if info_dict['http_headers'] and re.match(r'^https?://', url):
 194             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 195             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 196             headers = handle_youtubedl_headers(info_dict['http_headers'])
 197             args += [
 198                 '-headers',
 199                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 200
 201         args += ['-i', url, '-c', 'copy']
 202         if info_dict.get('protocol') == 'm3u8':
 203             if self.params.get('hls_use_mpegts', False):
 204                 args += ['-f', 'mpegts']
 205             else:
 206                 args += ['-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 207         else:
 208             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 209
 210         args = [encodeArgument(opt) for opt in args]
 211         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 212
 213         self._debug_cmd(args)
 214
 215         proc = subprocess.Popen(args, stdin=subprocess.PIPE)
 216         try:
 217             retval = proc.wait()
 218         except KeyboardInterrupt:
 219             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 220             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 221             # produces a file that is playable (this is mostly useful for live
 222             # streams). Note that Windows is not affected and produces playable
 223             # files (see https://github.com/rg3/youtube-dl/issues/8300).
 224             if sys.platform != 'win32':
 225                 proc.communicate(b'q')
 226             raise
 227         return retval
 228
 229
 230 class AVconvFD(FFmpegFD):
 231     pass
 232
 233 _BY_NAME = dict(
 234     (klass.get_basename(), klass)
 235     for name, klass in globals().items()
 236     if name.endswith('FD') and name != 'ExternalFD'
 237 )
 238
 239
 240 def list_external_downloaders():
 241     return sorted(_BY_NAME.keys())
 242
 243
 244 def get_external_downloader(external_downloader):
 245     """ Given the name of the executable, see whether we support the given
 246         downloader . """
 247     # Drop .exe extension on Windows
 248     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 249     return _BY_NAME[bn]