import collections
import contextlib
+import copy
import datetime
import errno
import fileinput
-import functools
+import http.cookiejar
import io
import itertools
import json
import unicodedata
from .cache import Cache
-from .compat import urllib # isort: split
-from .compat import compat_os_name, compat_shlex_quote
-from .cookies import load_cookies
+from .compat import functools, urllib # isort: split
+from .compat import compat_os_name, compat_shlex_quote, urllib_req_to_req
+from .cookies import LenientSimpleCookie, load_cookies
from .downloader import FFmpegFD, get_suitable_downloader, shorten_protocol_name
from .downloader.rtmp import rtmpdump_version
from .extractor import gen_extractor_classes, get_info_extractor
from .extractor.common import UnsupportedURLIE
from .extractor.openload import PhantomJSwrapper
from .minicurses import format_text
+from .networking import Request, RequestDirector
+from .networking.common import _REQUEST_HANDLERS
+from .networking.exceptions import (
+ HTTPError,
+ NoSupportingHandlers,
+ RequestError,
+ SSLError,
+ _CompatHTTPError,
+)
from .plugins import directories as plugin_directories
from .postprocessor import _PLUGIN_CLASSES as plugin_pps
from .postprocessor import (
MaxDownloadsReached,
Namespace,
PagedList,
- PerRequestProxyHandler,
PlaylistEntries,
Popen,
PostProcessingError,
SameFileError,
UnavailableVideoError,
UserNotLive,
- YoutubeDLCookieProcessor,
- YoutubeDLHandler,
- YoutubeDLRedirectHandler,
age_restricted,
args_to_str,
bug_reports_message,
error_to_compat_str,
escapeHTML,
expand_path,
+ extract_basic_auth,
filter_dict,
float_or_none,
format_bytes,
locked_file,
make_archive_id,
make_dir,
- make_HTTPS_handler,
- merge_headers,
network_exceptions,
number_of_digits,
orderedSet,
sanitize_filename,
sanitize_path,
sanitize_url,
- sanitized_Request,
std_headers,
str_or_none,
strftime_or_none,
write_json_file,
write_string,
)
+from .utils._utils import _YDLLogger
+from .utils.networking import (
+ HTTPHeaderDict,
+ clean_headers,
+ clean_proxies,
+)
from .version import CHANNEL, RELEASE_GIT_HEAD, VARIANT, __version__
if compat_os_name == 'nt':
raise
self.params['compat_opts'] = set(self.params.get('compat_opts', ()))
+ self.params['http_headers'] = HTTPHeaderDict(std_headers, self.params.get('http_headers'))
+ self._request_director = self.build_request_director(
+ sorted(_REQUEST_HANDLERS.values(), key=lambda rh: rh.RH_NAME.lower()))
if auto_init and auto_init != 'no_verbose_header':
self.print_debug_header()
+ self.__header_cookies = []
+ self._load_cookies(traverse_obj(self.params.get('http_headers'), 'cookie', casesense=False)) # compat
+
def check_deprecated(param, option, suggestion):
if self.params.get(param) is not None:
self.report_warning(f'{option} is deprecated. Use {suggestion} instead')
else self.params['format'] if callable(self.params['format'])
else self.build_format_selector(self.params['format']))
- # Set http_headers defaults according to std_headers
- self.params['http_headers'] = merge_headers(std_headers, self.params.get('http_headers', {}))
-
hooks = {
'post_hooks': self.add_post_hook,
'progress_hooks': self.add_progress_hook,
get_postprocessor(pp_def.pop('key'))(self, **pp_def),
when=when)
- self._setup_opener()
-
def preload_download_archive(fn):
"""Preload the archive, if any is specified"""
archive = set()
self.save_console_title()
return self
+ def save_cookies(self):
+ if self.params.get('cookiefile') is not None:
+ self.cookiejar.save(ignore_discard=True, ignore_expires=True)
+
def __exit__(self, *args):
self.restore_console_title()
+ self.close()
- if self.params.get('cookiefile') is not None:
- self.cookiejar.save(ignore_discard=True, ignore_expires=True)
+ def close(self):
+ self.save_cookies()
+ self._request_director.close()
def trouble(self, message=None, tb=None, is_error=True):
"""Determine action to take when a download problem appears.
ID='green',
DELIM='blue',
ERROR='red',
+ BAD_FORMAT='light red',
WARNING='yellow',
SUPPRESS='light black',
)
self.to_screen('')
raise
+ def _load_cookies(self, data, *, from_headers=True):
+ """Loads cookies from a `Cookie` header
+
+ This tries to work around the security vulnerability of passing cookies to every domain.
+ See: https://github.com/yt-dlp/yt-dlp/security/advisories/GHSA-v8mc-9377-rwjj
+ The unscoped cookies are saved for later to be stored in the jar with a limited scope.
+
+ @param data The Cookie header as string to load the cookies from
+ @param from_headers If `False`, allows Set-Cookie syntax in the cookie string (at least a domain will be required)
+ """
+ for cookie in LenientSimpleCookie(data).values():
+ if from_headers and any(cookie.values()):
+ raise ValueError('Invalid syntax in Cookie Header')
+
+ domain = cookie.get('domain') or ''
+ expiry = cookie.get('expires')
+ if expiry == '': # 0 is valid
+ expiry = None
+ prepared_cookie = http.cookiejar.Cookie(
+ cookie.get('version') or 0, cookie.key, cookie.value, None, False,
+ domain, True, True, cookie.get('path') or '', bool(cookie.get('path')),
+ cookie.get('secure') or False, expiry, False, None, None, {})
+
+ if domain:
+ self.cookiejar.set_cookie(prepared_cookie)
+ elif from_headers:
+ self.deprecated_feature(
+ 'Passing cookies as a header is a potential security risk; '
+ 'they will be scoped to the domain of the downloaded urls. '
+ 'Please consider loading cookies from a file or browser instead.')
+ self.__header_cookies.append(prepared_cookie)
+ else:
+ self.report_error('Unscoped cookies are not allowed; please specify some sort of scoping',
+ tb=False, is_error=False)
+
+ def _apply_header_cookies(self, url):
+ """Applies stray header cookies to the provided url
+
+ This loads header cookies and scopes them to the domain provided in `url`.
+ While this is not ideal, it helps reduce the risk of them being sent
+ to an unintended destination while mostly maintaining compatibility.
+ """
+ parsed = urllib.parse.urlparse(url)
+ if not parsed.hostname:
+ return
+
+ for cookie in map(copy.copy, self.__header_cookies):
+ cookie.domain = f'.{parsed.hostname}'
+ self.cookiejar.set_cookie(cookie)
+
@_handle_extraction_exceptions
def __extract_info(self, url, ie, download, extra_info, process):
+ self._apply_header_cookies(url)
+
try:
ie_result = ie.extract(url)
except UserNotLive as e:
allow_multiple_streams = {'audio': self.params.get('allow_multiple_audio_streams', False),
'video': self.params.get('allow_multiple_video_streams', False)}
- check_formats = self.params.get('check_formats') == 'selected'
-
def _parse_filter(tokens):
filter_parts = []
for type, string_, start, _, _ in tokens:
return new_dict
def _check_formats(formats):
- if not check_formats:
+ if (self.params.get('check_formats') is not None
+ or self.params.get('allow_unplayable_formats')):
yield from formats
return
- yield from self._check_formats(formats)
+ elif self.params.get('check_formats') == 'selected':
+ yield from self._check_formats(formats)
+ return
+
+ for f in formats:
+ if f.get('has_drm'):
+ yield from self._check_formats([f])
+ else:
+ yield f
def _build_selector_function(selector):
if isinstance(selector, list): # ,
return _build_selector_function(parsed_selector)
def _calc_headers(self, info_dict):
- res = merge_headers(self.params['http_headers'], info_dict.get('http_headers') or {})
- if 'Youtubedl-No-Compression' in res: # deprecated
- res.pop('Youtubedl-No-Compression', None)
- res['Accept-Encoding'] = 'identity'
- cookies = self.cookiejar.get_cookie_header(info_dict['url'])
+ res = HTTPHeaderDict(self.params['http_headers'], info_dict.get('http_headers'))
+ clean_headers(res)
+ cookies = self.cookiejar.get_cookies_for_url(info_dict['url'])
if cookies:
- res['Cookie'] = cookies
+ encoder = LenientSimpleCookie()
+ values = []
+ for cookie in cookies:
+ _, value = encoder.value_encode(cookie.value)
+ values.append(f'{cookie.name}={value}')
+ if cookie.domain:
+ values.append(f'Domain={cookie.domain}')
+ if cookie.path:
+ values.append(f'Path={cookie.path}')
+ if cookie.secure:
+ values.append('Secure')
+ if cookie.expires:
+ values.append(f'Expires={cookie.expires}')
+ if cookie.version:
+ values.append(f'Version={cookie.version}')
+ info_dict['cookies'] = '; '.join(values)
if 'X-Forwarded-For' not in res:
x_forwarded_for_ip = info_dict.get('__x_forwarded_for_ip')
if field_preference:
info_dict['_format_sort_fields'] = field_preference
- # or None ensures --clean-infojson removes it
- info_dict['_has_drm'] = any(f.get('has_drm') for f in formats) or None
+ info_dict['_has_drm'] = any( # or None ensures --clean-infojson removes it
+ f.get('has_drm') and f['has_drm'] != 'maybe' for f in formats) or None
if not self.params.get('allow_unplayable_formats'):
- formats = [f for f in formats if not f.get('has_drm')]
+ formats = [f for f in formats if not f.get('has_drm') or f['has_drm'] == 'maybe']
if formats and all(f.get('acodec') == f.get('vcodec') == 'none' for f in formats):
self.report_warning(
formats_to_download = list(format_selector({
'formats': formats,
'has_merged_format': any('none' not in (f.get('acodec'), f.get('vcodec')) for f in formats),
- 'incomplete_formats': (
- # All formats are video-only or
- all(f.get('vcodec') != 'none' and f.get('acodec') == 'none' for f in formats)
- # all formats are audio-only
- or all(f.get('vcodec') == 'none' and f.get('acodec') != 'none' for f in formats)),
+ 'incomplete_formats': (all(f.get('vcodec') == 'none' for f in formats) # No formats with video
+ or all(f.get('acodec') == 'none' for f in formats)), # OR, No formats with audio
}))
if interactive_format_selection and not formats_to_download:
self.report_error('Requested format is not available', tb=False, is_error=False)
infos = [self.sanitize_info(info, self.params.get('clean_infojson', True))
for info in variadic(json.loads('\n'.join(f)))]
for info in infos:
+ self._load_cookies(info.get('cookies'), from_headers=False)
+ self._load_cookies(traverse_obj(info.get('http_headers'), 'Cookie', casesense=False)) # compat
try:
self.__download_wrapper(self.process_ie_result)(info, download=True)
except (DownloadError, EntryNotInPlaylist, ReExtractInfo) as e:
simplified_codec(f, 'acodec'),
format_field(f, 'abr', '\t%dk', func=round),
format_field(f, 'asr', '\t%s', func=format_decimal_suffix),
- join_nonempty(
- self._format_out('UNSUPPORTED', 'light red') if f.get('ext') in ('f4f', 'f4m') else None,
- self._format_out('DRM', 'light red') if f.get('has_drm') else None,
- format_field(f, 'language', '[%s]'),
- join_nonempty(format_field(f, 'format_note'),
- format_field(f, 'container', ignore=(None, f.get('ext'))),
- delim=', '),
- delim=' '),
+ join_nonempty(format_field(f, 'language', '[%s]'), join_nonempty(
+ self._format_out('UNSUPPORTED', self.Styles.BAD_FORMAT) if f.get('ext') in ('f4f', 'f4m') else None,
+ (self._format_out('Maybe DRM', self.Styles.WARNING) if f.get('has_drm') == 'maybe'
+ else self._format_out('DRM', self.Styles.BAD_FORMAT) if f.get('has_drm') else None),
+ format_field(f, 'format_note'),
+ format_field(f, 'container', ignore=(None, f.get('ext'))),
+ delim=', '), delim=' '),
] for f in formats if f.get('preference') is None or f['preference'] >= -1000]
header_line = self._list_format_headers(
'ID', 'EXT', 'RESOLUTION', '\tFPS', 'HDR', 'CH', delim, '\tFILESIZE', '\tTBR', 'PROTO',
def list_subtitles(self, video_id, subtitles, name='subtitles'):
self.__list_table(video_id, name, self.render_subtitles_table, video_id, subtitles)
- def urlopen(self, req):
- """ Start an HTTP download """
- if isinstance(req, str):
- req = sanitized_Request(req)
- return self._opener.open(req, timeout=self._socket_timeout)
-
def print_debug_header(self):
if not self.params.get('verbose'):
return
join_nonempty(*get_package_info(m)) for m in available_dependencies.values()
})) or 'none'))
- self._setup_opener()
- proxy_map = {}
- for handler in self._opener.handlers:
- if hasattr(handler, 'proxies'):
- proxy_map.update(handler.proxies)
- write_debug(f'Proxy map: {proxy_map}')
-
+ write_debug(f'Proxy map: {self.proxies}')
+ # write_debug(f'Request Handlers: {", ".join(rh.RH_NAME for rh in self._request_director.handlers)}')
for plugin_type, plugins in {'Extractor': plugin_ies, 'Post-Processor': plugin_pps}.items():
display_list = ['%s%s' % (
klass.__name__, '' if klass.__name__ == name else f' as {name}')
'See https://yt-dl.org/update if you need help updating.' %
latest_version)
- def _setup_opener(self):
- if hasattr(self, '_opener'):
- return
- timeout_val = self.params.get('socket_timeout')
- self._socket_timeout = 20 if timeout_val is None else float(timeout_val)
-
- opts_cookiesfrombrowser = self.params.get('cookiesfrombrowser')
- opts_cookiefile = self.params.get('cookiefile')
+ @functools.cached_property
+ def proxies(self):
+ """Global proxy configuration"""
opts_proxy = self.params.get('proxy')
-
- self.cookiejar = load_cookies(opts_cookiefile, opts_cookiesfrombrowser, self)
-
- cookie_processor = YoutubeDLCookieProcessor(self.cookiejar)
if opts_proxy is not None:
if opts_proxy == '':
- proxies = {}
- else:
- proxies = {'http': opts_proxy, 'https': opts_proxy}
+ opts_proxy = '__noproxy__'
+ proxies = {'all': opts_proxy}
else:
proxies = urllib.request.getproxies()
- # Set HTTPS proxy to HTTP one if given (https://github.com/ytdl-org/youtube-dl/issues/805)
+ # compat. Set HTTPS_PROXY to __noproxy__ to revert
if 'http' in proxies and 'https' not in proxies:
proxies['https'] = proxies['http']
- proxy_handler = PerRequestProxyHandler(proxies)
-
- debuglevel = 1 if self.params.get('debug_printtraffic') else 0
- https_handler = make_HTTPS_handler(self.params, debuglevel=debuglevel)
- ydlh = YoutubeDLHandler(self.params, debuglevel=debuglevel)
- redirect_handler = YoutubeDLRedirectHandler()
- data_handler = urllib.request.DataHandler()
-
- # When passing our own FileHandler instance, build_opener won't add the
- # default FileHandler and allows us to disable the file protocol, which
- # can be used for malicious purposes (see
- # https://github.com/ytdl-org/youtube-dl/issues/8227)
- file_handler = urllib.request.FileHandler()
-
- if not self.params.get('enable_file_urls'):
- def file_open(*args, **kwargs):
- raise urllib.error.URLError(
- 'file:// URLs are explicitly disabled in yt-dlp for security reasons. '
- 'Use --enable-file-urls to enable at your own risk.')
- file_handler.file_open = file_open
-
- opener = urllib.request.build_opener(
- proxy_handler, https_handler, cookie_processor, ydlh, redirect_handler, data_handler, file_handler)
-
- # Delete the default user-agent header, which would otherwise apply in
- # cases where our custom HTTP handler doesn't come into play
- # (See https://github.com/ytdl-org/youtube-dl/issues/1309 for details)
- opener.addheaders = []
- self._opener = opener
+
+ return proxies
+
+ @functools.cached_property
+ def cookiejar(self):
+ """Global cookiejar instance"""
+ return load_cookies(
+ self.params.get('cookiefile'), self.params.get('cookiesfrombrowser'), self)
+
+ @property
+ def _opener(self):
+ """
+ Get a urllib OpenerDirector from the Urllib handler (deprecated).
+ """
+ self.deprecation_warning('YoutubeDL._opener() is deprecated, use YoutubeDL.urlopen()')
+ handler = self._request_director.handlers['Urllib']
+ return handler._get_instance(cookiejar=self.cookiejar, proxies=self.proxies)
+
+ def urlopen(self, req):
+ """ Start an HTTP download """
+ if isinstance(req, str):
+ req = Request(req)
+ elif isinstance(req, urllib.request.Request):
+ req = urllib_req_to_req(req)
+ assert isinstance(req, Request)
+
+ # compat: Assume user:pass url params are basic auth
+ url, basic_auth_header = extract_basic_auth(req.url)
+ if basic_auth_header:
+ req.headers['Authorization'] = basic_auth_header
+ req.url = sanitize_url(url)
+
+ clean_proxies(proxies=req.proxies, headers=req.headers)
+ clean_headers(req.headers)
+
+ try:
+ return self._request_director.send(req)
+ except NoSupportingHandlers as e:
+ for ue in e.unsupported_errors:
+ if not (ue.handler and ue.msg):
+ continue
+ if ue.handler.RH_KEY == 'Urllib' and 'unsupported url scheme: "file"' in ue.msg.lower():
+ raise RequestError(
+ 'file:// URLs are disabled by default in yt-dlp for security reasons. '
+ 'Use --enable-file-urls to enable at your own risk.', cause=ue) from ue
+ raise
+ except SSLError as e:
+ if 'UNSAFE_LEGACY_RENEGOTIATION_DISABLED' in str(e):
+ raise RequestError('UNSAFE_LEGACY_RENEGOTIATION_DISABLED: Try using --legacy-server-connect', cause=e) from e
+ elif 'SSLV3_ALERT_HANDSHAKE_FAILURE' in str(e):
+ raise RequestError(
+ 'SSLV3_ALERT_HANDSHAKE_FAILURE: The server may not support the current cipher list. '
+ 'Try using --legacy-server-connect', cause=e) from e
+ raise
+ except HTTPError as e: # TODO: Remove in a future release
+ raise _CompatHTTPError(e) from e
+
+ def build_request_director(self, handlers):
+ logger = _YDLLogger(self)
+ headers = self.params.get('http_headers').copy()
+ proxies = self.proxies.copy()
+ clean_headers(headers)
+ clean_proxies(proxies, headers)
+
+ director = RequestDirector(logger=logger, verbose=self.params.get('debug_printtraffic'))
+ for handler in handlers:
+ director.add_handler(handler(
+ logger=logger,
+ headers=headers,
+ cookiejar=self.cookiejar,
+ proxies=proxies,
+ prefer_system_certs='no-certifi' in self.params['compat_opts'],
+ verify=not self.params.get('nocheckcertificate'),
+ **traverse_obj(self.params, {
+ 'verbose': 'debug_printtraffic',
+ 'source_address': 'source_address',
+ 'timeout': 'socket_timeout',
+ 'legacy_ssl_support': 'legacy_server_connect',
+ 'enable_file_urls': 'enable_file_urls',
+ 'client_cert': {
+ 'client_certificate': 'client_certificate',
+ 'client_certificate_key': 'client_certificate_key',
+ 'client_certificate_password': 'client_certificate_password',
+ },
+ }),
+ ))
+ return director
def encode(self, s):
if isinstance(s, bytes):
else:
self.to_screen(f'[info] Downloading {thumb_display_id} ...')
try:
- uf = self.urlopen(sanitized_Request(t['url'], headers=t.get('http_headers', {})))
+ uf = self.urlopen(Request(t['url'], headers=t.get('http_headers', {})))
self.to_screen(f'[info] Writing {thumb_display_id} to: {thumb_filename}')
with open(encodeFilename(thumb_filename), 'wb') as thumbf:
shutil.copyfileobj(uf, thumbf)