youtube_dl/downloader/external.py

   1 from __future__ import unicode_literals
   2
   3 import os.path
   4 import re
   5 import subprocess
   6 import sys
   7 import time
   8
   9 from .common import FileDownloader
  10 from ..compat import (
  11     compat_setenv,
  12     compat_str,
  13 )
  14 from ..postprocessor.ffmpeg import FFmpegPostProcessor, EXT_TO_OUT_FORMATS
  15 from ..utils import (
  16     cli_option,
  17     cli_valueless_option,
  18     cli_bool_option,
  19     cli_configuration_args,
  20     encodeFilename,
  21     encodeArgument,
  22     handle_youtubedl_headers,
  23     check_executable,
  24     is_outdated_version,
  25 )
  26
  27
  28 class ExternalFD(FileDownloader):
  29     def real_download(self, filename, info_dict):
  30         self.report_destination(filename)
  31         tmpfilename = self.temp_name(filename)
  32
  33         try:
  34             started = time.time()
  35             retval = self._call_downloader(tmpfilename, info_dict)
  36         except KeyboardInterrupt:
  37             if not info_dict.get('is_live'):
  38                 raise
  39             # Live stream downloading cancellation should be considered as
  40             # correct and expected termination thus all postprocessing
  41             # should take place
  42             retval = 0
  43             self.to_screen('[%s] Interrupted by user' % self.get_basename())
  44
  45         if retval == 0:
  46             status = {
  47                 'filename': filename,
  48                 'status': 'finished',
  49                 'elapsed': time.time() - started,
  50             }
  51             if filename != '-':
  52                 fsize = os.path.getsize(encodeFilename(tmpfilename))
  53                 self.to_screen('\r[%s] Downloaded %s bytes' % (self.get_basename(), fsize))
  54                 self.try_rename(tmpfilename, filename)
  55                 status.update({
  56                     'downloaded_bytes': fsize,
  57                     'total_bytes': fsize,
  58                 })
  59             self._hook_progress(status)
  60             return True
  61         else:
  62             self.to_stderr('\n')
  63             self.report_error('%s exited with code %d' % (
  64                 self.get_basename(), retval))
  65             return False
  66
  67     @classmethod
  68     def get_basename(cls):
  69         return cls.__name__[:-2].lower()
  70
  71     @property
  72     def exe(self):
  73         return self.params.get('external_downloader')
  74
  75     @classmethod
  76     def available(cls):
  77         return check_executable(cls.get_basename(), [cls.AVAILABLE_OPT])
  78
  79     @classmethod
  80     def supports(cls, info_dict):
  81         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps')
  82
  83     @classmethod
  84     def can_download(cls, info_dict):
  85         return cls.available() and cls.supports(info_dict)
  86
  87     def _option(self, command_option, param):
  88         return cli_option(self.params, command_option, param)
  89
  90     def _bool_option(self, command_option, param, true_value='true', false_value='false', separator=None):
  91         return cli_bool_option(self.params, command_option, param, true_value, false_value, separator)
  92
  93     def _valueless_option(self, command_option, param, expected_value=True):
  94         return cli_valueless_option(self.params, command_option, param, expected_value)
  95
  96     def _configuration_args(self, default=[]):
  97         return cli_configuration_args(self.params, 'external_downloader_args', default)
  98
  99     def _call_downloader(self, tmpfilename, info_dict):
 100         """ Either overwrite this or implement _make_cmd """
 101         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 102
 103         self._debug_cmd(cmd)
 104
 105         p = subprocess.Popen(
 106             cmd, stderr=subprocess.PIPE)
 107         _, stderr = p.communicate()
 108         if p.returncode != 0:
 109             self.to_stderr(stderr.decode('utf-8', 'replace'))
 110         return p.returncode
 111
 112
 113 class CurlFD(ExternalFD):
 114     AVAILABLE_OPT = '-V'
 115
 116     def _make_cmd(self, tmpfilename, info_dict):
 117         cmd = [self.exe, '--location', '-o', tmpfilename]
 118         for key, val in info_dict['http_headers'].items():
 119             cmd += ['--header', '%s: %s' % (key, val)]
 120         cmd += self._bool_option('--continue-at', 'continuedl', '-', '0')
 121         cmd += self._valueless_option('--silent', 'noprogress')
 122         cmd += self._valueless_option('--verbose', 'verbose')
 123         cmd += self._option('--limit-rate', 'ratelimit')
 124         retry = self._option('--retry', 'retries')
 125         if len(retry) == 2:
 126             if retry[1] in ('inf', 'infinite'):
 127                 retry[1] = '2147483647'
 128             cmd += retry
 129         cmd += self._option('--max-filesize', 'max_filesize')
 130         cmd += self._option('--interface', 'source_address')
 131         cmd += self._option('--proxy', 'proxy')
 132         cmd += self._valueless_option('--insecure', 'nocheckcertificate')
 133         cmd += self._configuration_args()
 134         cmd += ['--', info_dict['url']]
 135         return cmd
 136
 137     def _call_downloader(self, tmpfilename, info_dict):
 138         cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
 139
 140         self._debug_cmd(cmd)
 141
 142         # curl writes the progress to stderr so don't capture it.
 143         p = subprocess.Popen(cmd)
 144         p.communicate()
 145         return p.returncode
 146
 147
 148 class AxelFD(ExternalFD):
 149     AVAILABLE_OPT = '-V'
 150
 151     def _make_cmd(self, tmpfilename, info_dict):
 152         cmd = [self.exe, '-o', tmpfilename]
 153         for key, val in info_dict['http_headers'].items():
 154             cmd += ['-H', '%s: %s' % (key, val)]
 155         cmd += self._configuration_args()
 156         cmd += ['--', info_dict['url']]
 157         return cmd
 158
 159
 160 class WgetFD(ExternalFD):
 161     AVAILABLE_OPT = '--version'
 162
 163     def _make_cmd(self, tmpfilename, info_dict):
 164         cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
 165         for key, val in info_dict['http_headers'].items():
 166             cmd += ['--header', '%s: %s' % (key, val)]
 167         cmd += self._option('--limit-rate', 'ratelimit')
 168         retry = self._option('--tries', 'retries')
 169         if len(retry) == 2:
 170             if retry[1] in ('inf', 'infinite'):
 171                 retry[1] = '0'
 172             cmd += retry
 173         cmd += self._option('--bind-address', 'source_address')
 174         cmd += self._option('--proxy', 'proxy')
 175         cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
 176         cmd += self._configuration_args()
 177         cmd += ['--', info_dict['url']]
 178         return cmd
 179
 180
 181 class Aria2cFD(ExternalFD):
 182     AVAILABLE_OPT = '-v'
 183
 184     def _make_cmd(self, tmpfilename, info_dict):
 185         cmd = [self.exe, '-c']
 186         cmd += self._configuration_args([
 187             '--min-split-size', '1M', '--max-connection-per-server', '4'])
 188         dn = os.path.dirname(tmpfilename)
 189         if dn:
 190             cmd += ['--dir', dn]
 191         cmd += ['--out', os.path.basename(tmpfilename)]
 192         for key, val in info_dict['http_headers'].items():
 193             cmd += ['--header', '%s: %s' % (key, val)]
 194         cmd += self._option('--interface', 'source_address')
 195         cmd += self._option('--all-proxy', 'proxy')
 196         cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
 197         cmd += ['--', info_dict['url']]
 198         return cmd
 199
 200
 201 class HttpieFD(ExternalFD):
 202     @classmethod
 203     def available(cls):
 204         return check_executable('http', ['--version'])
 205
 206     def _make_cmd(self, tmpfilename, info_dict):
 207         cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
 208         for key, val in info_dict['http_headers'].items():
 209             cmd += ['%s:%s' % (key, val)]
 210         return cmd
 211
 212
 213 class FFmpegFD(ExternalFD):
 214     @classmethod
 215     def supports(cls, info_dict):
 216         return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
 217
 218     @classmethod
 219     def available(cls):
 220         return FFmpegPostProcessor().available
 221
 222     def _call_downloader(self, tmpfilename, info_dict):
 223         url = info_dict['url']
 224         ffpp = FFmpegPostProcessor(downloader=self)
 225         if not ffpp.available:
 226             self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
 227             return False
 228         ffpp.check_version()
 229
 230         args = [ffpp.executable, '-y']
 231
 232         for log_level in ('quiet', 'verbose'):
 233             if self.params.get(log_level, False):
 234                 args += ['-loglevel', log_level]
 235                 break
 236
 237         seekable = info_dict.get('_seekable')
 238         if seekable is not None:
 239             # setting -seekable prevents ffmpeg from guessing if the server
 240             # supports seeking(by adding the header `Range: bytes=0-`), which
 241             # can cause problems in some cases
 242             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
 243             # http://trac.ffmpeg.org/ticket/6125#comment:10
 244             args += ['-seekable', '1' if seekable else '0']
 245
 246         args += self._configuration_args()
 247
 248         # start_time = info_dict.get('start_time') or 0
 249         # if start_time:
 250         #     args += ['-ss', compat_str(start_time)]
 251         # end_time = info_dict.get('end_time')
 252         # if end_time:
 253         #     args += ['-t', compat_str(end_time - start_time)]
 254
 255         if info_dict['http_headers'] and re.match(r'^https?://', url):
 256             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
 257             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
 258             headers = handle_youtubedl_headers(info_dict['http_headers'])
 259             args += [
 260                 '-headers',
 261                 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
 262
 263         env = None
 264         proxy = self.params.get('proxy')
 265         if proxy:
 266             if not re.match(r'^[\da-zA-Z]+://', proxy):
 267                 proxy = 'http://%s' % proxy
 268
 269             if proxy.startswith('socks'):
 270                 self.report_warning(
 271                     '%s does not support SOCKS proxies. Downloading is likely to fail. '
 272                     'Consider adding --hls-prefer-native to your command.' % self.get_basename())
 273
 274             # Since December 2015 ffmpeg supports -http_proxy option (see
 275             # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
 276             # We could switch to the following code if we are able to detect version properly
 277             # args += ['-http_proxy', proxy]
 278             env = os.environ.copy()
 279             compat_setenv('HTTP_PROXY', proxy, env=env)
 280             compat_setenv('http_proxy', proxy, env=env)
 281
 282         protocol = info_dict.get('protocol')
 283
 284         if protocol == 'rtmp':
 285             player_url = info_dict.get('player_url')
 286             page_url = info_dict.get('page_url')
 287             app = info_dict.get('app')
 288             play_path = info_dict.get('play_path')
 289             tc_url = info_dict.get('tc_url')
 290             flash_version = info_dict.get('flash_version')
 291             live = info_dict.get('rtmp_live', False)
 292             conn = info_dict.get('rtmp_conn')
 293             if player_url is not None:
 294                 args += ['-rtmp_swfverify', player_url]
 295             if page_url is not None:
 296                 args += ['-rtmp_pageurl', page_url]
 297             if app is not None:
 298                 args += ['-rtmp_app', app]
 299             if play_path is not None:
 300                 args += ['-rtmp_playpath', play_path]
 301             if tc_url is not None:
 302                 args += ['-rtmp_tcurl', tc_url]
 303             if flash_version is not None:
 304                 args += ['-rtmp_flashver', flash_version]
 305             if live:
 306                 args += ['-rtmp_live', 'live']
 307             if isinstance(conn, list):
 308                 for entry in conn:
 309                     args += ['-rtmp_conn', entry]
 310             elif isinstance(conn, compat_str):
 311                 args += ['-rtmp_conn', conn]
 312
 313         args += ['-i', url, '-c', 'copy']
 314
 315         if self.params.get('test', False):
 316             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
 317
 318         if protocol in ('m3u8', 'm3u8_native'):
 319             if self.params.get('hls_use_mpegts', False) or tmpfilename == '-':
 320                 args += ['-f', 'mpegts']
 321             else:
 322                 args += ['-f', 'mp4']
 323                 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')):
 324                     args += ['-bsf:a', 'aac_adtstoasc']
 325         elif protocol == 'rtmp':
 326             args += ['-f', 'flv']
 327         else:
 328             args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
 329
 330         args = [encodeArgument(opt) for opt in args]
 331         args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
 332
 333         self._debug_cmd(args)
 334
 335         proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
 336         try:
 337             retval = proc.wait()
 338         except KeyboardInterrupt:
 339             # subprocces.run would send the SIGKILL signal to ffmpeg and the
 340             # mp4 file couldn't be played, but if we ask ffmpeg to quit it
 341             # produces a file that is playable (this is mostly useful for live
 342             # streams). Note that Windows is not affected and produces playable
 343             # files (see https://github.com/ytdl-org/youtube-dl/issues/8300).
 344             if sys.platform != 'win32':
 345                 proc.communicate(b'q')
 346             raise
 347         return retval
 348
 349
 350 class AVconvFD(FFmpegFD):
 351     pass
 352
 353
 354 _BY_NAME = dict(
 355     (klass.get_basename(), klass)
 356     for name, klass in globals().items()
 357     if name.endswith('FD') and name != 'ExternalFD'
 358 )
 359
 360
 361 def list_external_downloaders():
 362     return sorted(_BY_NAME.keys())
 363
 364
 365 def get_external_downloader(external_downloader):
 366     """ Given the name of the executable, see whether we support the given
 367         downloader . """
 368     # Drop .exe extension on Windows
 369     bn = os.path.splitext(os.path.basename(external_downloader))[0]
 370     return _BY_NAME[bn]