Merging upstream version 2024.07.16.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

5 files changed, 44 insertions, 15 deletions

diff --git a/yt_dlp/networking/_curlcffi.py b/yt_dlp/networking/_curlcffi.py
index b1f0fb8..e8a67b7 100644
--- a/yt_dlp/networking/_curlcffi.py
+++ b/yt_dlp/networking/_curlcffi.py
@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import io
 import math
+import re
 import urllib.parse
 
 from ._helper import InstanceStoreMixin, select_proxy
@@ -27,11 +28,12 @@ from ..utils import int_or_none
 if curl_cffi is None:
     raise ImportError('curl_cffi is not installed')
 
-curl_cffi_version = tuple(int_or_none(x, default=0) for x in curl_cffi.__version__.split('.'))
 
-if curl_cffi_version != (0, 5, 10):
+curl_cffi_version = tuple(map(int, re.split(r'[^\d]+', curl_cffi.__version__)[:3]))
+
+if curl_cffi_version != (0, 5, 10) and not ((0, 7, 0) <= curl_cffi_version < (0, 8, 0)):
     curl_cffi._yt_dlp__version = f'{curl_cffi.__version__} (unsupported)'
-    raise ImportError('Only curl_cffi 0.5.10 is supported')
+    raise ImportError('Only curl_cffi versions 0.5.10, 0.7.X are supported')
 
 import curl_cffi.requests
 from curl_cffi.const import CurlECode, CurlOpt
@@ -110,6 +112,13 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
     _SUPPORTED_FEATURES = (Features.NO_PROXY, Features.ALL_PROXY)
     _SUPPORTED_PROXY_SCHEMES = ('http', 'https', 'socks4', 'socks4a', 'socks5', 'socks5h')
     _SUPPORTED_IMPERSONATE_TARGET_MAP = {
+        **({
+            ImpersonateTarget('chrome', '124', 'macos', '14'): curl_cffi.requests.BrowserType.chrome124,
+            ImpersonateTarget('chrome', '123', 'macos', '14'): curl_cffi.requests.BrowserType.chrome123,
+            ImpersonateTarget('chrome', '120', 'macos', '14'): curl_cffi.requests.BrowserType.chrome120,
+            ImpersonateTarget('chrome', '119', 'macos', '14'): curl_cffi.requests.BrowserType.chrome119,
+            ImpersonateTarget('chrome', '116', 'windows', '10'): curl_cffi.requests.BrowserType.chrome116,
+        } if curl_cffi_version >= (0, 7, 0) else {}),
         ImpersonateTarget('chrome', '110', 'windows', '10'): curl_cffi.requests.BrowserType.chrome110,
         ImpersonateTarget('chrome', '107', 'windows', '10'): curl_cffi.requests.BrowserType.chrome107,
         ImpersonateTarget('chrome', '104', 'windows', '10'): curl_cffi.requests.BrowserType.chrome104,
@@ -118,9 +127,15 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
         ImpersonateTarget('chrome', '99', 'windows', '10'): curl_cffi.requests.BrowserType.chrome99,
         ImpersonateTarget('edge', '101', 'windows', '10'): curl_cffi.requests.BrowserType.edge101,
         ImpersonateTarget('edge', '99', 'windows', '10'): curl_cffi.requests.BrowserType.edge99,
+        **({
+            ImpersonateTarget('safari', '17.0', 'macos', '14'): curl_cffi.requests.BrowserType.safari17_0,
+        } if curl_cffi_version >= (0, 7, 0) else {}),
         ImpersonateTarget('safari', '15.5', 'macos', '12'): curl_cffi.requests.BrowserType.safari15_5,
         ImpersonateTarget('safari', '15.3', 'macos', '11'): curl_cffi.requests.BrowserType.safari15_3,
         ImpersonateTarget('chrome', '99', 'android', '12'): curl_cffi.requests.BrowserType.chrome99_android,
+        **({
+            ImpersonateTarget('safari', '17.2', 'ios', '17.2'): curl_cffi.requests.BrowserType.safari17_2_ios,
+        } if curl_cffi_version >= (0, 7, 0) else {}),
     }
 
     def _create_instance(self, cookiejar=None):
@@ -131,6 +146,9 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
         extensions.pop('impersonate', None)
         extensions.pop('cookiejar', None)
         extensions.pop('timeout', None)
+        # CurlCFFIRH ignores legacy ssl options currently.
+        # Impersonation generally uses a looser SSL configuration than urllib/requests.
+        extensions.pop('legacy_ssl', None)
 
     def send(self, request: Request) -> Response:
         target = self._get_request_target(request)
@@ -187,7 +205,7 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
         timeout = self._calculate_timeout(request)
 
         # set CURLOPT_LOW_SPEED_LIMIT and CURLOPT_LOW_SPEED_TIME to act as a read timeout. [1]
-        # curl_cffi does not currently do this. [2]
+        # This is required only for 0.5.10 [2]
         # Note: CURLOPT_LOW_SPEED_TIME is in seconds, so we need to round up to the nearest second. [3]
         # [1] https://unix.stackexchange.com/a/305311
         # [2] https://github.com/yifeikong/curl_cffi/issues/156
@@ -203,7 +221,7 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
                 data=request.data,
                 verify=self.verify,
                 max_redirects=5,
-                timeout=timeout,
+                timeout=(timeout, timeout),
                 impersonate=self._SUPPORTED_IMPERSONATE_TARGET_MAP.get(
                     self._get_request_target(request)),
                 interface=self.source_address,
@@ -222,7 +240,7 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
 
             elif (
                 e.code == CurlECode.PROXY
-                or (e.code == CurlECode.RECV_ERROR and 'Received HTTP code 407 from proxy after CONNECT' in str(e))
+                or (e.code == CurlECode.RECV_ERROR and 'CONNECT' in str(e))
             ):
                 raise ProxyError(cause=e) from e
             else:
diff --git a/yt_dlp/networking/_requests.py b/yt_dlp/networking/_requests.py
index 86850c1..7de95ab 100644
--- a/yt_dlp/networking/_requests.py
+++ b/yt_dlp/networking/_requests.py
@@ -295,11 +295,12 @@ class RequestsRH(RequestHandler, InstanceStoreMixin):
         super()._check_extensions(extensions)
         extensions.pop('cookiejar', None)
         extensions.pop('timeout', None)
+        extensions.pop('legacy_ssl', None)
 
-    def _create_instance(self, cookiejar):
+    def _create_instance(self, cookiejar, legacy_ssl_support=None):
         session = RequestsSession()
         http_adapter = RequestsHTTPAdapter(
-            ssl_context=self._make_sslcontext(),
+            ssl_context=self._make_sslcontext(legacy_ssl_support=legacy_ssl_support),
             source_address=self.source_address,
             max_retries=urllib3.util.retry.Retry(False),
         )
@@ -318,7 +319,10 @@ class RequestsRH(RequestHandler, InstanceStoreMixin):
 
         max_redirects_exceeded = False
 
-        session = self._get_instance(cookiejar=self._get_cookiejar(request))
+        session = self._get_instance(
+            cookiejar=self._get_cookiejar(request),
+            legacy_ssl_support=request.extensions.get('legacy_ssl'),
+        )
 
         try:
             requests_res = session.request(
diff --git a/yt_dlp/networking/_urllib.py b/yt_dlp/networking/_urllib.py
index 6299582..510bb2a 100644
--- a/yt_dlp/networking/_urllib.py
+++ b/yt_dlp/networking/_urllib.py
@@ -348,14 +348,15 @@ class UrllibRH(RequestHandler, InstanceStoreMixin):
         super()._check_extensions(extensions)
         extensions.pop('cookiejar', None)
         extensions.pop('timeout', None)
+        extensions.pop('legacy_ssl', None)
 
-    def _create_instance(self, proxies, cookiejar):
+    def _create_instance(self, proxies, cookiejar, legacy_ssl_support=None):
         opener = urllib.request.OpenerDirector()
         handlers = [
             ProxyHandler(proxies),
             HTTPHandler(
                 debuglevel=int(bool(self.verbose)),
-                context=self._make_sslcontext(),
+                context=self._make_sslcontext(legacy_ssl_support=legacy_ssl_support),
                 source_address=self.source_address),
             HTTPCookieProcessor(cookiejar),
             DataHandler(),
@@ -391,6 +392,7 @@ class UrllibRH(RequestHandler, InstanceStoreMixin):
         opener = self._get_instance(
             proxies=self._get_proxies(request),
             cookiejar=self._get_cookiejar(request),
+            legacy_ssl_support=request.extensions.get('legacy_ssl'),
         )
         try:
             res = opener.open(urllib_req, timeout=self._calculate_timeout(request))
diff --git a/yt_dlp/networking/_websockets.py b/yt_dlp/networking/_websockets.py
index 15db4fe..492af11 100644
--- a/yt_dlp/networking/_websockets.py
+++ b/yt_dlp/networking/_websockets.py
@@ -118,6 +118,7 @@ class WebsocketsRH(WebSocketRequestHandler):
         super()._check_extensions(extensions)
         extensions.pop('timeout', None)
         extensions.pop('cookiejar', None)
+        extensions.pop('legacy_ssl', None)
 
     def close(self):
         # Remove the logging handler that contains a reference to our logger
@@ -154,13 +155,14 @@ class WebsocketsRH(WebSocketRequestHandler):
                     address=(wsuri.host, wsuri.port),
                     **create_conn_kwargs,
                 )
+            ssl_ctx = self._make_sslcontext(legacy_ssl_support=request.extensions.get('legacy_ssl'))
             conn = websockets.sync.client.connect(
                 sock=sock,
                 uri=request.url,
                 additional_headers=headers,
                 open_timeout=timeout,
                 user_agent_header=None,
-                ssl_context=self._make_sslcontext() if wsuri.secure else None,
+                ssl_context=ssl_ctx if wsuri.secure else None,
                 close_timeout=0,  # not ideal, but prevents yt-dlp hanging
             )
             return WebsocketsResponseAdapter(conn, url=request.url)
diff --git a/yt_dlp/networking/common.py b/yt_dlp/networking/common.py
index a6db167..e8951c7 100644
--- a/yt_dlp/networking/common.py
+++ b/yt_dlp/networking/common.py
@@ -205,6 +205,7 @@ class RequestHandler(abc.ABC):
     The following extensions are defined for RequestHandler:
     - `cookiejar`: Cookiejar to use for this request.
     - `timeout`: socket timeout to use for this request.
+    - `legacy_ssl`: Enable legacy SSL options for this request. See legacy_ssl_support.
     To enable these, add extensions.pop('<extension>', None) to _check_extensions
 
     Apart from the url protocol, proxies dict may contain the following keys:
@@ -247,10 +248,10 @@ class RequestHandler(abc.ABC):
         self.legacy_ssl_support = legacy_ssl_support
         super().__init__()
 
-    def _make_sslcontext(self):
+    def _make_sslcontext(self, legacy_ssl_support=None):
         return make_ssl_context(
             verify=self.verify,
-            legacy_support=self.legacy_ssl_support,
+            legacy_support=legacy_ssl_support if legacy_ssl_support is not None else self.legacy_ssl_support,
             use_certifi=not self.prefer_system_certs,
             **self._client_cert,
         )
@@ -262,7 +263,8 @@ class RequestHandler(abc.ABC):
         return float(request.extensions.get('timeout') or self.timeout)
 
     def _get_cookiejar(self, request):
-        return request.extensions.get('cookiejar') or self.cookiejar
+        cookiejar = request.extensions.get('cookiejar')
+        return self.cookiejar if cookiejar is None else cookiejar
 
     def _get_proxies(self, request):
         return (request.proxies or self.proxies).copy()
@@ -314,6 +316,7 @@ class RequestHandler(abc.ABC):
         """Check extensions for unsupported extensions. Subclasses should extend this."""
         assert isinstance(extensions.get('cookiejar'), (YoutubeDLCookieJar, NoneType))
         assert isinstance(extensions.get('timeout'), (float, int, NoneType))
+        assert isinstance(extensions.get('legacy_ssl'), (bool, NoneType))
 
     def _validate(self, request):
         self._check_url_scheme(request)