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