1
0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2026-02-03 04:26:56 +00:00

Add new options --impersonate and --list-impersonate-targets

Authored by: coletdjnz, Grub4K, pukkandan, bashonly

Co-authored-by: Simon Sawicki <contact@grub4k.xyz>
Co-authored-by: pukkandan <pukkandan.ytdlp@gmail.com>
Co-authored-by: bashonly <bashonly@protonmail.com>
This commit is contained in:
coletdjnz
2024-03-16 22:47:56 -05:00
committed by bashonly
parent f849d77ab5
commit 0b81d4d252
6 changed files with 415 additions and 24 deletions

View File

@@ -42,6 +42,7 @@ from .networking.exceptions import (
SSLError,
network_exceptions,
)
from .networking.impersonate import ImpersonateRequestHandler
from .plugins import directories as plugin_directories
from .postprocessor import _PLUGIN_CLASSES as plugin_pps
from .postprocessor import (
@@ -99,6 +100,7 @@ from .utils import (
SameFileError,
UnavailableVideoError,
UserNotLive,
YoutubeDLError,
age_restricted,
args_to_str,
bug_reports_message,
@@ -402,6 +404,8 @@ class YoutubeDL:
- "detect_or_warn": check whether we can do anything
about it, warn otherwise (default)
source_address: Client-side IP address to bind to.
impersonate: Client to impersonate for requests.
An ImpersonateTarget (from yt_dlp.networking.impersonate)
sleep_interval_requests: Number of seconds to sleep between requests
during extraction
sleep_interval: Number of seconds to sleep before each download when
@@ -713,6 +717,13 @@ class YoutubeDL:
for msg in self.params.get('_deprecation_warnings', []):
self.deprecated_feature(msg)
if impersonate_target := self.params.get('impersonate'):
if not self._impersonate_target_available(impersonate_target):
raise YoutubeDLError(
f'Impersonate target "{impersonate_target}" is not available. '
f'Use --list-impersonate-targets to see available targets. '
f'You may be missing dependencies required to support this target.')
if 'list-formats' in self.params['compat_opts']:
self.params['listformats_table'] = False
@@ -4077,6 +4088,22 @@ class YoutubeDL:
handler = self._request_director.handlers['Urllib']
return handler._get_instance(cookiejar=self.cookiejar, proxies=self.proxies)
def _get_available_impersonate_targets(self):
# todo(future): make available as public API
return [
(target, rh.RH_NAME)
for rh in self._request_director.handlers.values()
if isinstance(rh, ImpersonateRequestHandler)
for target in rh.supported_targets
]
def _impersonate_target_available(self, target):
# todo(future): make available as public API
return any(
rh.is_supported_target(target)
for rh in self._request_director.handlers.values()
if isinstance(rh, ImpersonateRequestHandler))
def urlopen(self, req):
""" Start an HTTP download """
if isinstance(req, str):
@@ -4108,9 +4135,13 @@ class YoutubeDL:
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
if 'unsupported proxy type: "https"' in ue.msg.lower():
if (
'unsupported proxy type: "https"' in ue.msg.lower()
and 'requests' not in self._request_director.handlers
and 'curl_cffi' not in self._request_director.handlers
):
raise RequestError(
'To use an HTTPS proxy for this request, one of the following dependencies needs to be installed: requests')
'To use an HTTPS proxy for this request, one of the following dependencies needs to be installed: requests, curl_cffi')
elif (
re.match(r'unsupported url scheme: "wss?"', ue.msg.lower())
@@ -4120,6 +4151,13 @@ class YoutubeDL:
'This request requires WebSocket support. '
'Ensure one of the following dependencies are installed: websockets',
cause=ue) from ue
elif re.match(r'unsupported (?:extensions: impersonate|impersonate target)', ue.msg.lower()):
raise RequestError(
f'Impersonate target "{req.extensions["impersonate"]}" is not available.'
f' See --list-impersonate-targets for available targets.'
f' This request requires browser impersonation, however you may be missing dependencies'
f' required to support this target.')
raise
except SSLError as e:
if 'UNSAFE_LEGACY_RENEGOTIATION_DISABLED' in str(e):
@@ -4152,6 +4190,7 @@ class YoutubeDL:
'timeout': 'socket_timeout',
'legacy_ssl_support': 'legacyserverconnect',
'enable_file_urls': 'enable_file_urls',
'impersonate': 'impersonate',
'client_cert': {
'client_certificate': 'client_certificate',
'client_certificate_key': 'client_certificate_key',

View File

@@ -19,6 +19,7 @@ from .cookies import SUPPORTED_BROWSERS, SUPPORTED_KEYRINGS
from .downloader.external import get_external_downloader
from .extractor import list_extractor_classes
from .extractor.adobepass import MSO_INFO
from .networking.impersonate import ImpersonateTarget
from .options import parseOpts
from .postprocessor import (
FFmpegExtractAudioPP,
@@ -48,6 +49,7 @@ from .utils import (
float_or_none,
format_field,
int_or_none,
join_nonempty,
match_filter_func,
parse_bytes,
parse_duration,
@@ -388,6 +390,9 @@ def validate_options(opts):
f'Supported keyrings are: {", ".join(sorted(SUPPORTED_KEYRINGS))}')
opts.cookiesfrombrowser = (browser_name, profile, keyring, container)
if opts.impersonate is not None:
opts.impersonate = ImpersonateTarget.from_str(opts.impersonate.lower())
# MetadataParser
def metadataparser_actions(f):
if isinstance(f, str):
@@ -911,6 +916,7 @@ def parse_options(argv=None):
'postprocessors': postprocessors,
'fixup': opts.fixup,
'source_address': opts.source_address,
'impersonate': opts.impersonate,
'call_home': opts.call_home,
'sleep_interval_requests': opts.sleep_interval_requests,
'sleep_interval': opts.sleep_interval,
@@ -980,6 +986,41 @@ def _real_main(argv=None):
traceback.print_exc()
ydl._download_retcode = 100
if opts.list_impersonate_targets:
known_targets = [
# List of simplified targets we know are supported,
# to help users know what dependencies may be required.
(ImpersonateTarget('chrome'), 'curl_cffi'),
(ImpersonateTarget('edge'), 'curl_cffi'),
(ImpersonateTarget('safari'), 'curl_cffi'),
]
available_targets = ydl._get_available_impersonate_targets()
def make_row(target, handler):
return [
join_nonempty(target.client.title(), target.version, delim='-') or '-',
join_nonempty((target.os or "").title(), target.os_version, delim='-') or '-',
handler,
]
rows = [make_row(target, handler) for target, handler in available_targets]
for known_target, known_handler in known_targets:
if not any(
known_target in target and handler == known_handler
for target, handler in available_targets
):
rows.append([
ydl._format_out(text, ydl.Styles.SUPPRESS)
for text in make_row(known_target, f'{known_handler} (not available)')
])
ydl.to_screen('[info] Available impersonate targets')
ydl.to_stdout(render_table(['Client', 'OS', 'Source'], rows, extra_gap=2, delim='-'))
return
if not actual_use:
if pre_process:
return ydl._download_retcode

View File

@@ -0,0 +1,141 @@
from __future__ import annotations
import re
from abc import ABC
from dataclasses import dataclass
from typing import Any
from .common import RequestHandler, register_preference
from .exceptions import UnsupportedRequest
from ..compat.types import NoneType
from ..utils import classproperty, join_nonempty
from ..utils.networking import std_headers
@dataclass(order=True, frozen=True)
class ImpersonateTarget:
"""
A target for browser impersonation.
Parameters:
@param client: the client to impersonate
@param version: the client version to impersonate
@param os: the client OS to impersonate
@param os_version: the client OS version to impersonate
Note: None is used to indicate to match any.
"""
client: str | None = None
version: str | None = None
os: str | None = None
os_version: str | None = None
def __post_init__(self):
if self.version and not self.client:
raise ValueError('client is required if version is set')
if self.os_version and not self.os:
raise ValueError('os is required if os_version is set')
def __contains__(self, target: ImpersonateTarget):
if not isinstance(target, ImpersonateTarget):
return False
return (
(self.client is None or target.client is None or self.client == target.client)
and (self.version is None or target.version is None or self.version == target.version)
and (self.os is None or target.os is None or self.os == target.os)
and (self.os_version is None or target.os_version is None or self.os_version == target.os_version)
)
def __str__(self):
return f'{join_nonempty(self.client, self.version)}:{join_nonempty(self.os, self.os_version)}'.rstrip(':')
@classmethod
def from_str(cls, target: str):
mobj = re.fullmatch(r'(?:(?P<client>[^:-]+)(?:-(?P<version>[^:-]+))?)?(?::(?:(?P<os>[^:-]+)(?:-(?P<os_version>[^:-]+))?)?)?', target)
if not mobj:
raise ValueError(f'Invalid impersonate target "{target}"')
return cls(**mobj.groupdict())
class ImpersonateRequestHandler(RequestHandler, ABC):
"""
Base class for request handlers that support browser impersonation.
This provides a method for checking the validity of the impersonate extension,
which can be used in _check_extensions.
Impersonate targets consist of a client, version, os and os_ver.
See the ImpersonateTarget class for more details.
The following may be defined:
- `_SUPPORTED_IMPERSONATE_TARGET_MAP`: a dict mapping supported targets to custom object.
Any Request with an impersonate target not in this list will raise an UnsupportedRequest.
Set to None to disable this check.
Note: Entries are in order of preference
Parameters:
@param impersonate: the default impersonate target to use for requests.
Set to None to disable impersonation.
"""
_SUPPORTED_IMPERSONATE_TARGET_MAP: dict[ImpersonateTarget, Any] = {}
def __init__(self, *, impersonate: ImpersonateTarget = None, **kwargs):
super().__init__(**kwargs)
self.impersonate = impersonate
def _check_impersonate_target(self, target: ImpersonateTarget):
assert isinstance(target, (ImpersonateTarget, NoneType))
if target is None or not self.supported_targets:
return
if not self.is_supported_target(target):
raise UnsupportedRequest(f'Unsupported impersonate target: {target}')
def _check_extensions(self, extensions):
super()._check_extensions(extensions)
if 'impersonate' in extensions:
self._check_impersonate_target(extensions.get('impersonate'))
def _validate(self, request):
super()._validate(request)
self._check_impersonate_target(self.impersonate)
def _resolve_target(self, target: ImpersonateTarget | None):
"""Resolve a target to a supported target."""
if target is None:
return
for supported_target in self.supported_targets:
if target in supported_target:
if self.verbose:
self._logger.stdout(
f'{self.RH_NAME}: resolved impersonate target {target} to {supported_target}')
return supported_target
@classproperty
def supported_targets(self) -> tuple[ImpersonateTarget, ...]:
return tuple(self._SUPPORTED_IMPERSONATE_TARGET_MAP.keys())
def is_supported_target(self, target: ImpersonateTarget):
assert isinstance(target, ImpersonateTarget)
return self._resolve_target(target) is not None
def _get_request_target(self, request):
"""Get the requested target for the request"""
return self._resolve_target(request.extensions.get('impersonate') or self.impersonate)
def _get_impersonate_headers(self, request):
headers = self._merge_headers(request.headers)
if self._get_request_target(request) is not None:
# remove all headers present in std_headers
# todo: change this to not depend on std_headers
for k, v in std_headers.items():
if headers.get(k) == v:
headers.pop(k)
return headers
@register_preference(ImpersonateRequestHandler)
def impersonate_preference(rh, request):
if request.extensions.get('impersonate') or rh.impersonate:
return 1000
return 0

View File

@@ -515,6 +515,18 @@ def create_parser():
metavar='IP', dest='source_address', default=None,
help='Client-side IP address to bind to',
)
network.add_option(
'--impersonate',
metavar='CLIENT[:OS]', dest='impersonate', default=None,
help=(
'Client to impersonate for requests. E.g. chrome, chrome-110, chrome:windows-10. '
'Pass --impersonate="" to impersonate any client.'),
)
network.add_option(
'--list-impersonate-targets',
dest='list_impersonate_targets', default=False, action='store_true',
help='List available clients to impersonate.',
)
network.add_option(
'-4', '--force-ipv4',
action='store_const', const='0.0.0.0', dest='source_address',