]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/downloader/external.py
Fix extraction from youtube.
[youtubedl] / 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 cmd += self._option('--retry', 'retries')
125 cmd += self._option('--max-filesize', 'max_filesize')
126 cmd += self._option('--interface', 'source_address')
127 cmd += self._option('--proxy', 'proxy')
128 cmd += self._valueless_option('--insecure', 'nocheckcertificate')
129 cmd += self._configuration_args()
130 cmd += ['--', info_dict['url']]
131 return cmd
132
133 def _call_downloader(self, tmpfilename, info_dict):
134 cmd = [encodeArgument(a) for a in self._make_cmd(tmpfilename, info_dict)]
135
136 self._debug_cmd(cmd)
137
138 # curl writes the progress to stderr so don't capture it.
139 p = subprocess.Popen(cmd)
140 p.communicate()
141 return p.returncode
142
143
144 class AxelFD(ExternalFD):
145 AVAILABLE_OPT = '-V'
146
147 def _make_cmd(self, tmpfilename, info_dict):
148 cmd = [self.exe, '-o', tmpfilename]
149 for key, val in info_dict['http_headers'].items():
150 cmd += ['-H', '%s: %s' % (key, val)]
151 cmd += self._configuration_args()
152 cmd += ['--', info_dict['url']]
153 return cmd
154
155
156 class WgetFD(ExternalFD):
157 AVAILABLE_OPT = '--version'
158
159 def _make_cmd(self, tmpfilename, info_dict):
160 cmd = [self.exe, '-O', tmpfilename, '-nv', '--no-cookies']
161 for key, val in info_dict['http_headers'].items():
162 cmd += ['--header', '%s: %s' % (key, val)]
163 cmd += self._option('--bind-address', 'source_address')
164 cmd += self._option('--proxy', 'proxy')
165 cmd += self._valueless_option('--no-check-certificate', 'nocheckcertificate')
166 cmd += self._configuration_args()
167 cmd += ['--', info_dict['url']]
168 return cmd
169
170
171 class Aria2cFD(ExternalFD):
172 AVAILABLE_OPT = '-v'
173
174 def _make_cmd(self, tmpfilename, info_dict):
175 cmd = [self.exe, '-c']
176 cmd += self._configuration_args([
177 '--min-split-size', '1M', '--max-connection-per-server', '4'])
178 dn = os.path.dirname(tmpfilename)
179 if dn:
180 cmd += ['--dir', dn]
181 cmd += ['--out', os.path.basename(tmpfilename)]
182 for key, val in info_dict['http_headers'].items():
183 cmd += ['--header', '%s: %s' % (key, val)]
184 cmd += self._option('--interface', 'source_address')
185 cmd += self._option('--all-proxy', 'proxy')
186 cmd += self._bool_option('--check-certificate', 'nocheckcertificate', 'false', 'true', '=')
187 cmd += ['--', info_dict['url']]
188 return cmd
189
190
191 class HttpieFD(ExternalFD):
192 @classmethod
193 def available(cls):
194 return check_executable('http', ['--version'])
195
196 def _make_cmd(self, tmpfilename, info_dict):
197 cmd = ['http', '--download', '--output', tmpfilename, info_dict['url']]
198 for key, val in info_dict['http_headers'].items():
199 cmd += ['%s:%s' % (key, val)]
200 return cmd
201
202
203 class FFmpegFD(ExternalFD):
204 @classmethod
205 def supports(cls, info_dict):
206 return info_dict['protocol'] in ('http', 'https', 'ftp', 'ftps', 'm3u8', 'rtsp', 'rtmp', 'mms')
207
208 @classmethod
209 def available(cls):
210 return FFmpegPostProcessor().available
211
212 def _call_downloader(self, tmpfilename, info_dict):
213 url = info_dict['url']
214 ffpp = FFmpegPostProcessor(downloader=self)
215 if not ffpp.available:
216 self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.')
217 return False
218 ffpp.check_version()
219
220 args = [ffpp.executable, '-y']
221
222 for log_level in ('quiet', 'verbose'):
223 if self.params.get(log_level, False):
224 args += ['-loglevel', log_level]
225 break
226
227 seekable = info_dict.get('_seekable')
228 if seekable is not None:
229 # setting -seekable prevents ffmpeg from guessing if the server
230 # supports seeking(by adding the header `Range: bytes=0-`), which
231 # can cause problems in some cases
232 # https://github.com/rg3/youtube-dl/issues/11800#issuecomment-275037127
233 # http://trac.ffmpeg.org/ticket/6125#comment:10
234 args += ['-seekable', '1' if seekable else '0']
235
236 args += self._configuration_args()
237
238 # start_time = info_dict.get('start_time') or 0
239 # if start_time:
240 # args += ['-ss', compat_str(start_time)]
241 # end_time = info_dict.get('end_time')
242 # if end_time:
243 # args += ['-t', compat_str(end_time - start_time)]
244
245 if info_dict['http_headers'] and re.match(r'^https?://', url):
246 # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
247 # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
248 headers = handle_youtubedl_headers(info_dict['http_headers'])
249 args += [
250 '-headers',
251 ''.join('%s: %s\r\n' % (key, val) for key, val in headers.items())]
252
253 env = None
254 proxy = self.params.get('proxy')
255 if proxy:
256 if not re.match(r'^[\da-zA-Z]+://', proxy):
257 proxy = 'http://%s' % proxy
258
259 if proxy.startswith('socks'):
260 self.report_warning(
261 '%s does not support SOCKS proxies. Downloading is likely to fail. '
262 'Consider adding --hls-prefer-native to your command.' % self.get_basename())
263
264 # Since December 2015 ffmpeg supports -http_proxy option (see
265 # http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b4eb1f29ebddd60c41a2eb39f5af701e38e0d3fd)
266 # We could switch to the following code if we are able to detect version properly
267 # args += ['-http_proxy', proxy]
268 env = os.environ.copy()
269 compat_setenv('HTTP_PROXY', proxy, env=env)
270 compat_setenv('http_proxy', proxy, env=env)
271
272 protocol = info_dict.get('protocol')
273
274 if protocol == 'rtmp':
275 player_url = info_dict.get('player_url')
276 page_url = info_dict.get('page_url')
277 app = info_dict.get('app')
278 play_path = info_dict.get('play_path')
279 tc_url = info_dict.get('tc_url')
280 flash_version = info_dict.get('flash_version')
281 live = info_dict.get('rtmp_live', False)
282 if player_url is not None:
283 args += ['-rtmp_swfverify', player_url]
284 if page_url is not None:
285 args += ['-rtmp_pageurl', page_url]
286 if app is not None:
287 args += ['-rtmp_app', app]
288 if play_path is not None:
289 args += ['-rtmp_playpath', play_path]
290 if tc_url is not None:
291 args += ['-rtmp_tcurl', tc_url]
292 if flash_version is not None:
293 args += ['-rtmp_flashver', flash_version]
294 if live:
295 args += ['-rtmp_live', 'live']
296
297 args += ['-i', url, '-c', 'copy']
298
299 if self.params.get('test', False):
300 args += ['-fs', compat_str(self._TEST_FILE_SIZE)]
301
302 if protocol in ('m3u8', 'm3u8_native'):
303 if self.params.get('hls_use_mpegts', False) or tmpfilename == '-':
304 args += ['-f', 'mpegts']
305 else:
306 args += ['-f', 'mp4']
307 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')):
308 args += ['-bsf:a', 'aac_adtstoasc']
309 elif protocol == 'rtmp':
310 args += ['-f', 'flv']
311 else:
312 args += ['-f', EXT_TO_OUT_FORMATS.get(info_dict['ext'], info_dict['ext'])]
313
314 args = [encodeArgument(opt) for opt in args]
315 args.append(encodeFilename(ffpp._ffmpeg_filename_argument(tmpfilename), True))
316
317 self._debug_cmd(args)
318
319 proc = subprocess.Popen(args, stdin=subprocess.PIPE, env=env)
320 try:
321 retval = proc.wait()
322 except KeyboardInterrupt:
323 # subprocces.run would send the SIGKILL signal to ffmpeg and the
324 # mp4 file couldn't be played, but if we ask ffmpeg to quit it
325 # produces a file that is playable (this is mostly useful for live
326 # streams). Note that Windows is not affected and produces playable
327 # files (see https://github.com/rg3/youtube-dl/issues/8300).
328 if sys.platform != 'win32':
329 proc.communicate(b'q')
330 raise
331 return retval
332
333
334 class AVconvFD(FFmpegFD):
335 pass
336
337
338 _BY_NAME = dict(
339 (klass.get_basename(), klass)
340 for name, klass in globals().items()
341 if name.endswith('FD') and name != 'ExternalFD'
342 )
343
344
345 def list_external_downloaders():
346 return sorted(_BY_NAME.keys())
347
348
349 def get_external_downloader(external_downloader):
350 """ Given the name of the executable, see whether we support the given
351 downloader . """
352 # Drop .exe extension on Windows
353 bn = os.path.splitext(os.path.basename(external_downloader))[0]
354 return _BY_NAME[bn]