Raphaël G. Git Repositories - youtubedl/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 compat_setenv
  10 from ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS
  11 from ..utils import (
  12     cli_option,
  13     cli_valueless_option,
  14     cli_bool_option,
  15     cli_configuration_args,
  16     encodeFilename,
  17     encodeArgument,
  18     handle_youtubedl_headers,
  19     check_executable,
  20     is_outdated_version,
  21 )
  22
  23
  24 class ExternalFD(FileDownloader):
  25     def real_download(self, filename, info_dict):
  26         self.report_destination(filename)
  27         tmpfilename = self.temp_name(filename)
  28
  29         retval = self._call_downloader(tmpfilename, info_dict)
  30         if retval == 0:
  31             fsize = os.path.getsize(encodeFilename(tmpfilename))
  32             self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  33             self.try_rename(tmpfilename, filename)
  34             self._hook_progress({
  35                 'downloaded_bytes': fsize,
  36                 'total_bytes': fsize,
  37                 'filename': filename,
  38                 'status': 'finished',
  39             })
  40             return True
  41         else:
  42             self.to_stderr('\n')
  43             self.report_error('%s exited with code %d' % (
  44                 self.get_basename(), retval))
  45             return False
  46
  47     @classmethod
  48     def get_basename(cls):
  49         return cls.__name__[:-2].lower()
  50
  51     @property
  52     def exe(self):
  53         return self.params.get('external_downloader')
  54
  55     @classmethod
  56     def available(cls):
  57         return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT])
  58
  59     @classmethod
  60     def supports(cls, info_dict):
  61         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps')
  62
  63     @classmethod
  64     def can_download(cls, info_dict):
  65         return cls.available() and cls.supports(info_dict)
  66
  67     def _option(self, command_option, param):
  68         return cli_option(self.params, command_option, param)
  69
  70     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
  71         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
  72
  73     def _valueless_option(self, command_option, param, expected_value=True):
  74         return cli_valueless_option(self.params, command_option, param, expected_value)
  75
  76     def _configuration_args(self, default=[]):
  77         return cli_configuration_args(self.params, 'external_downloader_args', default)
  78
  79     def _call_downloader(self, tmpfilename, info_dict):
  80         """ Either overwrite this or implement _make_cmd """
  81         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
  82
  83         self._debug_cmd(cmd)
  84
  85         p = subprocess.Popen(
  86             cmd, stderr=subprocess.PIPE)
  87         _, stderr = p.communicate()
  88         if p.returncode != 0:
  89             self.to_stderr(stderr.decode('utf-8', 'replace'))
  90         return p.returncode
  91
  92
  93 class CurlFD(ExternalFD):
  94     AVAILABLE_OPT = '-V'
  95
  96     def _make_cmd(self, tmpfilename, info_dict):
  97         cmd = [self.exe, '--location', '-o', tmpfilename]
  98         for key, val in info_dict['http_headers'].items():
  99             cmd += ['--header', '%s: %s' % (key, val)]
 100         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 101         cmd += self._valueless_option('--silent', 'noprogress')
 102         cmd += self._valueless_option('--verbose', 'verbose')
 103         cmd += self._option('--limit-rate', 'ratelimit')
 104         cmd += self._option('--retry', 'retries')
 105         cmd += self._option('--max-filesize', 'max_filesize')
 106         cmd += self._option('--interface', 'source_address')
 107         cmd += self._option('--proxy', 'proxy')
 108         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 109         cmd += self._configuration_args()
 110         cmd += ['--', info_dict['url']]
 111         return cmd
 112
 113     def _call_downloader(self, tmpfilename, info_dict):
 114         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 115
 116         self._debug_cmd(cmd)
 117
 118         # curl writes the progress to stderr so don't capture it.
 119         p = subprocess.Popen(cmd)
 120         p.communicate()
 121         return p.returncode
 122
 123
 124 class AxelFD(ExternalFD):
 125     AVAILABLE_OPT = '-V'
 126
 127     def _make_cmd(self, tmpfilename, info_dict):
 128         cmd = [self.exe, '-o', tmpfilename]
 129         for key, val in info_dict['http_headers'].items():
 130             cmd += ['-H', '%s: %s' % (key, val)]
 131         cmd += self._configuration_args()
 132         cmd += ['--', info_dict['url']]
 133         return cmd
 134
 135
 136 class WgetFD(ExternalFD):
 137     AVAILABLE_OPT = '--version'
 138
 139     def _make_cmd(self, tmpfilename, info_dict):
 140         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 141         for key, val in info_dict['http_headers'].items():
 142             cmd += ['--header', '%s: %s' % (key, val)]
 143         cmd += self._option('--bind-address', 'source_address')
 144         cmd += self._option('--proxy', 'proxy')
 145         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 146         cmd += self._configuration_args()
 147         cmd += ['--', info_dict['url']]
 148         return cmd
 149
 150
 151 class Aria2cFD(ExternalFD):
 152     AVAILABLE_OPT = '-v'
 153
 154     def _make_cmd(self, tmpfilename, info_dict):
 155         cmd = [self.exe, '-c']
 156         cmd += self._configuration_args([
 157             '--min-split-size', '1M', '--max-connection-per-server', '4'])
 158         dn = os.path.dirname(tmpfilename)
 159         if dn:
 160             cmd += ['--dir', dn]
 161         cmd += ['--out', os.path.basename(tmpfilename)]
 162         for key, val in info_dict['http_headers'].items():
 163             cmd += ['--header', '%s: %s' % (key, val)]
 164         cmd += self._option('--interface', 'source_address')
 165         cmd += self._option('--all-proxy', 'proxy')
 166         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 167         cmd += ['--', info_dict['url']]
 168         return cmd
 169
 170
 171 class HttpieFD(ExternalFD):
 172     @classmethod
 173     def available(cls):
 174         return check_executable('http', ['--version'])
 175
 176     def _make_cmd(self, tmpfilename, info_dict):
 177         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 178         for key, val in info_dict['http_headers'].items():
 179             cmd += ['%s:%s' % (key, val)]
 180         return cmd
 181
 182
 183 class FFmpegFD(ExternalFD):
 184     @classmethod
 185     def supports(cls, info_dict):
 186         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 187
 188     @classmethod
 189     def available(cls):
 190         return FFmpegPostProcessor().available
 191
 192     def _call_downloader(self, tmpfilename, info_dict):
 193         url = info_dict['url']
 194         ffpp = FFmpegPostProcessor(downloader=self)
 195         if not ffpp.available:
 196             self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
 197             return False
 198         ffpp.check_version()
 199
 200         args = [ffpp.executable, '-y']
 201
 202         seekable = info_dict.get('_seekable')
 203         if seekable is not None:
 204             # setting -seekable prevents ffmpeg from guessing if the server
 205             # supports seeking(by adding the header `Range: bytes=0-`), which
 206             # can cause problems in some cases
 207             # https://github.com/rg3/youtube-dl/issues/11800#issuecomment-275037127
 208             # http://trac.ffmpeg.org/ticket/6125#comment:10
 209             args += ['-seekable', '1' if seekable else '0']
 210
 211         args += self._configuration_args()
 212
 213         # start_time = info_dict.get('start_time') or 0
 214         # if start_time:
 215         #     args += ['-ss', compat_str(start_time)]
 216         # end_time = info_dict.get('end_time')
 217         # if end_time:
 218         #     args += ['-t', compat_str(end_time - start_time)]
 219
 220         if info_dict['http_headers'] and re.match(r'^https?://', url):
 221             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 222             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 223             headers = handle_youtubedl_headers(info_dict['http_headers'])
 224             args += [
 225                 '-headers',
 226                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 227
 228         env = None
 229         proxy = self.params.get('proxy')
 230         if proxy:
 231             if not re.match(r'^[\da-zA-Z]+://', proxy):
 232                 proxy = 'http://%s' % proxy
 233
 234             if proxy.startswith('socks'):
 235                 self.report_warning(
 236                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 237                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 238
 239             # Since December 2015 ffmpeg supports -http_proxy option (see
 240             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 241             # We could switch to the following code if we are able to detect version properly
 242             # args += ['-http_proxy', proxy]
 243             env = os.environ.copy()
 244             compat_setenv('HTTP_PROXY', proxy, env=env)
 245             compat_setenv('http_proxy', proxy, env=env)
 246
 247         protocol = info_dict.get('protocol')
 248
 249         if protocol == 'rtmp':
 250             player_url = info_dict.get('player_url')
 251             page_url = info_dict.get('page_url')
 252             app = info_dict.get('app')
 253             play_path = info_dict.get('play_path')
 254             tc_url = info_dict.get('tc_url')
 255             flash_version = info_dict.get('flash_version')
 256             live = info_dict.get('rtmp_live', False)
 257             if player_url is not None:
 258                 args += ['-rtmp_swfverify', player_url]
 259             if page_url is not None:
 260                 args += ['-rtmp_pageurl', page_url]
 261             if app is not None:
 262                 args += ['-rtmp_app', app]
 263             if play_path is not None:
 264                 args += ['-rtmp_playpath', play_path]
 265             if tc_url is not None:
 266                 args += ['-rtmp_tcurl', tc_url]
 267             if flash_version is not None:
 268                 args += ['-rtmp_flashver', flash_version]
 269             if live:
 270                 args += ['-rtmp_live', 'live']
 271
 272         args += ['-i', url, '-c', 'copy']
 273         if protocol in ('m3u8', 'm3u8_native'):
 274             if self.params.get('hls_use_mpegts', False) or tmpfilename == '-':
 275                 args += ['-f', 'mpegts']
 276             else:
 277                 args += ['-f', 'mp4']
 278                 if (ffpp.basename == 'ffmpeg' and is_outdated_version(ffpp._versions['ffmpeg'], '3.2')) and (not info_dict.get('acodec') or info_dict['acodec'].split('.')[0] in ('aac', 'mp4a')):
 279                     args += ['-bsf:a', 'aac_adtstoasc']
 280         elif protocol == 'rtmp':
 281             args += ['-f', 'flv']
 282         else:
 283             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 284
 285         args = [encodeArgument(opt) for opt in args]
 286         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 287
 288         self._debug_cmd(args)
 289
 290         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 291         try:
 292             retval = proc.wait()
 293         except KeyboardInterrupt:
 294             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 295             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 296             # produces a file that is playable (this is mostly useful for live
 297             # streams). Note that Windows is not affected and produces playable
 298             # files (see https://github.com/rg3/youtube-dl/issues/8300).
 299             if sys.platform != 'win32':
 300                 proc.communicate(b'q')
 301             raise
 302         return retval
 303
 304
 305 class AVconvFD(FFmpegFD):
 306     pass
 307
 308
 309 _BY_NAME = dict(
 310     (klass.get_basename(), klass)
 311     for name, klass in globals().items()
 312     if name.endswith('FD') and name != 'ExternalFD'
 313 )
 314
 315
 316 def list_external_downloaders():
 317     return sorted(_BY_NAME.keys())
 318
 319
 320 def get_external_downloader(external_downloader):
 321     """ Given the name of the executable, see whether we support the given
 322         downloader . """
 323     # Drop .exe extension on Windows
 324     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 325     return _BY_NAME[bn]