misc cleanup

test/test_networking.py

@@ -517,13 +517,13 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert b'test3: test3' in data
 
     @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
-    def test_timeout(self, handler):
+    def test_read_timeout(self, handler):
         with handler() as rh:
             # Default timeout is 20 seconds, so this should go through
             validate_and_send(
-                rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_3'))
+                rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
 
-        with handler(timeout=0.5) as rh:
+        with handler(timeout=0.01) as rh:
             with pytest.raises(TransportError):
                 validate_and_send(
                     rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
@@ -532,6 +532,25 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             validate_and_send(
                 rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1', extensions={'timeout': 4}))
 
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
+    def test_connect_timeout(self, handler):
+        # nothing should be listening on this port
+        connect_timeout_url = 'http://10.255.255.255'
+        with handler(timeout=0.01) as rh:
+            now = time.time()
+            with pytest.raises(TransportError):
+                validate_and_send(
+                    rh, Request(connect_timeout_url))
+            assert 0.01 <= time.time() - now < 20
+
+        with handler() as rh:
+            with pytest.raises(TransportError):
+                # Per request timeout, should override handler timeout
+                now = time.time()
+                validate_and_send(
+                    rh, Request(connect_timeout_url, extensions={'timeout': 0.01}))
+                assert 0.01 <= time.time() - now < 20
+
     @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_source_address(self, handler):
         source_address = f'127.0.0.{random.randint(5, 255)}'

yt_dlp/networking/_curlcffi.py

@@ -1,4 +1,5 @@
 import io
+import math
 
 from ._helper import InstanceStoreMixin, select_proxy
 from .common import (
@@ -153,6 +154,18 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
                 session.curl.setopt(CurlOpt.SSLKEY, client_certificate_key.encode())
             if client_certificate_password:
                 session.curl.setopt(CurlOpt.KEYPASSWD, client_certificate_password.encode())
+
+        timeout = float(request.extensions.get('timeout') or self.timeout)
+
+        # 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]
+        # 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
+        # [3] https://curl.se/libcurl/c/CURLOPT_LOW_SPEED_TIME.html
+        session.curl.setopt(CurlOpt.LOW_SPEED_LIMIT, 1)  # 1 byte per second
+        session.curl.setopt(CurlOpt.LOW_SPEED_TIME, math.ceil(timeout))
+
         try:
             curl_response = session.request(
                 method=request.method,
@@ -161,7 +174,7 @@ class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
                 data=request.data,
                 verify=self.verify,
                 max_redirects=5,
-                timeout=request.extensions.get('timeout') or self.timeout,
+                timeout=timeout,
                 impersonate=self._get_mapped_target(request),
                 interface=self.source_address,
                 stream=True