Merge branch 'yt-dlp:master' into pr/yt-live-from-start-range

test/test_execution.py

@@ -45,7 +45,7 @@ class TestExecution(unittest.TestCase):
             self.assertTrue(os.path.exists(LAZY_EXTRACTORS))
 
             _, stderr = self.run_yt_dlp(opts=('-s', 'test:'))
-            # `MIN_RECOMMENDED` emits a deprecated feature warning for deprecated python versions
+            # `MIN_RECOMMENDED` emits a deprecated feature warning for deprecated Python versions
             if stderr and stderr.startswith('Deprecated Feature: Support for Python'):
                 stderr = ''
             self.assertFalse(stderr)

test/test_networking.py

@@ -27,9 +27,10 @@ import zlib
 from email.message import Message
 from http.cookiejar import CookieJar
 
+from test.conftest import validate_and_send
 from test.helper import FakeYDL, http_server_port, verify_address_availability
 from yt_dlp.cookies import YoutubeDLCookieJar
-from yt_dlp.dependencies import brotli, requests, urllib3
+from yt_dlp.dependencies import brotli, curl_cffi, requests, urllib3
 from yt_dlp.networking import (
     HEADRequest,
     PUTRequest,
@@ -50,10 +51,13 @@ from yt_dlp.networking.exceptions import (
     TransportError,
     UnsupportedRequest,
 )
+from yt_dlp.networking.impersonate import (
+    ImpersonateRequestHandler,
+    ImpersonateTarget,
+)
+from yt_dlp.utils import YoutubeDLError
 from yt_dlp.utils._utils import _YDLLogger as FakeLogger
-from yt_dlp.utils.networking import HTTPHeaderDict
-
-from test.conftest import validate_and_send
+from yt_dlp.utils.networking import HTTPHeaderDict, std_headers
 
 TEST_DIR = os.path.dirname(os.path.abspath(__file__))
 
@@ -69,12 +73,13 @@ def _build_proxy_handler(name):
             self.send_response(200)
             self.send_header('Content-Type', 'text/plain; charset=utf-8')
             self.end_headers()
-            self.wfile.write('{self.proxy_name}: {self.path}'.format(self=self).encode())
+            self.wfile.write(f'{self.proxy_name}: {self.path}'.encode())
     return HTTPTestRequestHandler
 
 
 class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
     protocol_version = 'HTTP/1.1'
+    default_request_version = 'HTTP/1.1'
 
     def log_message(self, format, *args):
         pass
@@ -112,6 +117,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
     def _read_data(self):
         if 'Content-Length' in self.headers:
             return self.rfile.read(int(self.headers['Content-Length']))
+        else:
+            return b''
 
     def do_POST(self):
         data = self._read_data() + str(self.headers).encode()
@@ -195,7 +202,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
             self._headers()
         elif self.path.startswith('/308-to-headers'):
             self.send_response(308)
-            self.send_header('Location', '/headers')
+            # redirect to "localhost" for testing cookie redirection handling
+            self.send_header('Location', f'http://localhost:{self.connection.getsockname()[1]}/headers')
             self.send_header('Content-Length', '0')
             self.end_headers()
         elif self.path == '/trailing_garbage':
@@ -310,7 +318,7 @@ class TestRequestHandlerBase:
 
 
 class TestHTTPRequestHandler(TestRequestHandlerBase):
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_verify_cert(self, handler):
         with handler() as rh:
             with pytest.raises(CertificateVerifyError):
@@ -321,7 +329,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert r.status == 200
             r.close()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_ssl_error(self, handler):
         # HTTPS server with too old TLS version
         # XXX: is there a better way to test this than to create a new server?
@@ -335,11 +343,11 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
         https_server_thread.start()
 
         with handler(verify=False) as rh:
-            with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info:
+            with pytest.raises(SSLError, match=r'(?i)ssl(?:v3|/tls).alert.handshake.failure') as exc_info:
                 validate_and_send(rh, Request(f'https://127.0.0.1:{https_port}/headers'))
             assert not issubclass(exc_info.type, CertificateVerifyError)
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_percent_encode(self, handler):
         with handler() as rh:
             # Unicode characters should be encoded with uppercase percent-encoding
@@ -351,7 +359,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.status == 200
             res.close()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     @pytest.mark.parametrize('path', [
         '/a/b/./../../headers',
         '/redirect_dotsegments',
@@ -367,6 +375,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.url == f'http://127.0.0.1:{self.http_port}/headers'
             res.close()
 
+    # Not supported by CurlCFFI (non-standard)
     @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
     def test_unicode_path_redirection(self, handler):
         with handler() as rh:
@@ -374,7 +383,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert r.url == f'http://127.0.0.1:{self.http_port}/%E4%B8%AD%E6%96%87.html'
             r.close()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_raise_http_error(self, handler):
         with handler() as rh:
             for bad_status in (400, 500, 599, 302):
@@ -384,7 +393,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             # Should not raise an error
             validate_and_send(rh, Request('http://127.0.0.1:%d/gen_200' % self.http_port)).close()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_response_url(self, handler):
         with handler() as rh:
             # Response url should be that of the last url in redirect chain
@@ -395,62 +404,50 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res2.url == f'http://127.0.0.1:{self.http_port}/gen_200'
             res2.close()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
-    def test_redirect(self, handler):
+    # Covers some basic cases we expect some level of consistency between request handlers for
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
+    @pytest.mark.parametrize('redirect_status,method,expected', [
+        # A 303 must either use GET or HEAD for subsequent request
+        (303, 'POST', ('', 'GET', False)),
+        (303, 'HEAD', ('', 'HEAD', False)),
+
+        # 301 and 302 turn POST only into a GET
+        (301, 'POST', ('', 'GET', False)),
+        (301, 'HEAD', ('', 'HEAD', False)),
+        (302, 'POST', ('', 'GET', False)),
+        (302, 'HEAD', ('', 'HEAD', False)),
+
+        # 307 and 308 should not change method
+        (307, 'POST', ('testdata', 'POST', True)),
+        (308, 'POST', ('testdata', 'POST', True)),
+        (307, 'HEAD', ('', 'HEAD', False)),
+        (308, 'HEAD', ('', 'HEAD', False)),
+    ])
+    def test_redirect(self, handler, redirect_status, method, expected):
         with handler() as rh:
-            def do_req(redirect_status, method, assert_no_content=False):
-                data = b'testdata' if method in ('POST', 'PUT') else None
-                res = validate_and_send(
-                    rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data))
+            data = b'testdata' if method == 'POST' else None
+            headers = {}
+            if data is not None:
+                headers['Content-Type'] = 'application/test'
+            res = validate_and_send(
+                rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data,
+                            headers=headers))
 
-                headers = b''
-                data_sent = b''
-                if data is not None:
-                    data_sent += res.read(len(data))
-                    if data_sent != data:
-                        headers += data_sent
-                        data_sent = b''
+            headers = b''
+            data_recv = b''
+            if data is not None:
+                data_recv += res.read(len(data))
+                if data_recv != data:
+                    headers += data_recv
+                    data_recv = b''
 
-                headers += res.read()
+            headers += res.read()
 
-                if assert_no_content or data is None:
-                    assert b'Content-Type' not in headers
-                    assert b'Content-Length' not in headers
-                else:
-                    assert b'Content-Type' in headers
-                    assert b'Content-Length' in headers
+            assert expected[0] == data_recv.decode()
+            assert expected[1] == res.headers.get('method')
+            assert expected[2] == ('content-length' in headers.decode().lower())
 
-                return data_sent.decode(), res.headers.get('method', '')
-
-            # A 303 must either use GET or HEAD for subsequent request
-            assert do_req(303, 'POST', True) == ('', 'GET')
-            assert do_req(303, 'HEAD') == ('', 'HEAD')
-
-            assert do_req(303, 'PUT', True) == ('', 'GET')
-
-            # 301 and 302 turn POST only into a GET
-            assert do_req(301, 'POST', True) == ('', 'GET')
-            assert do_req(301, 'HEAD') == ('', 'HEAD')
-            assert do_req(302, 'POST', True) == ('', 'GET')
-            assert do_req(302, 'HEAD') == ('', 'HEAD')
-
-            assert do_req(301, 'PUT') == ('testdata', 'PUT')
-            assert do_req(302, 'PUT') == ('testdata', 'PUT')
-
-            # 307 and 308 should not change method
-            for m in ('POST', 'PUT'):
-                assert do_req(307, m) == ('testdata', m)
-                assert do_req(308, m) == ('testdata', m)
-
-            assert do_req(307, 'HEAD') == ('', 'HEAD')
-            assert do_req(308, 'HEAD') == ('', 'HEAD')
-
-            # These should not redirect and instead raise an HTTPError
-            for code in (300, 304, 305, 306):
-                with pytest.raises(HTTPError):
-                    do_req(code, 'GET')
-
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_request_cookie_header(self, handler):
         # We should accept a Cookie header being passed as in normal headers and handle it appropriately.
         with handler() as rh:
@@ -459,16 +456,17 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
                 rh, Request(
                     f'http://127.0.0.1:{self.http_port}/headers',
                     headers={'Cookie': 'test=test'})).read().decode()
-            assert 'Cookie: test=test' in res
+            assert 'cookie: test=test' in res.lower()
 
             # Specified Cookie header should be removed on any redirect
             res = validate_and_send(
                 rh, Request(
                     f'http://127.0.0.1:{self.http_port}/308-to-headers',
-                    headers={'Cookie': 'test=test'})).read().decode()
-            assert 'Cookie: test=test' not in res
+                    headers={'Cookie': 'test=test2'})).read().decode()
+            assert 'cookie: test=test2' not in res.lower()
 
         # Specified Cookie header should override global cookiejar for that request
+        # Whether cookies from the cookiejar is applied on the redirect is considered undefined for now
         cookiejar = YoutubeDLCookieJar()
         cookiejar.set_cookie(http.cookiejar.Cookie(
             version=0, name='test', value='ytdlp', port=None, port_specified=False,
@@ -478,23 +476,23 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
 
         with handler(cookiejar=cookiejar) as rh:
             data = validate_and_send(
-                rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test'})).read()
-            assert b'Cookie: test=ytdlp' not in data
-            assert b'Cookie: test=test' in data
+                rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test3'})).read()
+            assert b'cookie: test=ytdlp' not in data.lower()
+            assert b'cookie: test=test3' in data.lower()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_redirect_loop(self, handler):
         with handler() as rh:
             with pytest.raises(HTTPError, match='redirect loop'):
                 validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_loop'))
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_incompleteread(self, handler):
         with handler(timeout=2) as rh:
-            with pytest.raises(IncompleteRead):
+            with pytest.raises(IncompleteRead, match='13 bytes read, 234221 more expected'):
                 validate_and_send(rh, Request('http://127.0.0.1:%d/incompleteread' % self.http_port)).read()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_cookies(self, handler):
         cookiejar = YoutubeDLCookieJar()
         cookiejar.set_cookie(http.cookiejar.Cookie(
@@ -503,47 +501,66 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
 
         with handler(cookiejar=cookiejar) as rh:
             data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read()
-            assert b'Cookie: test=ytdlp' in data
+            assert b'cookie: test=ytdlp' in data.lower()
 
         # Per request
         with handler() as rh:
             data = validate_and_send(
                 rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={'cookiejar': cookiejar})).read()
-            assert b'Cookie: test=ytdlp' in data
+            assert b'cookie: test=ytdlp' in data.lower()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_headers(self, handler):
 
         with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh:
             # Global Headers
-            data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read()
-            assert b'Test1: test' in data
+            data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read().lower()
+            assert b'test1: test' in data
 
             # Per request headers, merged with global
             data = validate_and_send(rh, Request(
-                f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read()
-            assert b'Test1: test' in data
-            assert b'Test2: changed' in data
-            assert b'Test2: test2' not in data
-            assert b'Test3: test3' in data
+                f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read().lower()
+            assert b'test1: test' in data
+            assert b'test2: changed' in data
+            assert b'test2: test2' not in data
+            assert b'test3: test3' in data
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
-    def test_timeout(self, handler):
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
+    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=0.1) as rh:
             with pytest.raises(TransportError):
                 validate_and_send(
-                    rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
+                    rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_5'))
 
             # Per request timeout, should override handler timeout
             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'], indirect=True)
+    @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)}'
         # on some systems these loopback addresses we need for testing may not be available
@@ -554,6 +571,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
                 rh, Request(f'http://127.0.0.1:{self.http_port}/source_address')).read().decode()
             assert source_address == data
 
+    # Not supported by CurlCFFI
     @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
     def test_gzip_trailing_garbage(self, handler):
         with handler() as rh:
@@ -571,7 +589,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.headers.get('Content-Encoding') == 'br'
             assert res.read() == b'<html><video src="/vid.mp4" /></html>'
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_deflate(self, handler):
         with handler() as rh:
             res = validate_and_send(
@@ -581,7 +599,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.headers.get('Content-Encoding') == 'deflate'
             assert res.read() == b'<html><video src="/vid.mp4" /></html>'
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_gzip(self, handler):
         with handler() as rh:
             res = validate_and_send(
@@ -591,7 +609,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.headers.get('Content-Encoding') == 'gzip'
             assert res.read() == b'<html><video src="/vid.mp4" /></html>'
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_multiple_encodings(self, handler):
         with handler() as rh:
             for pair in ('gzip,deflate', 'deflate, gzip', 'gzip, gzip', 'deflate, deflate'):
@@ -602,17 +620,18 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
                 assert res.headers.get('Content-Encoding') == pair
                 assert res.read() == b'<html><video src="/vid.mp4" /></html>'
 
+    # Not supported by curl_cffi
     @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
     def test_unsupported_encoding(self, handler):
         with handler() as rh:
             res = validate_and_send(
                 rh, Request(
                     f'http://127.0.0.1:{self.http_port}/content-encoding',
-                    headers={'ytdl-encoding': 'unsupported'}))
+                    headers={'ytdl-encoding': 'unsupported', 'Accept-Encoding': '*'}))
             assert res.headers.get('Content-Encoding') == 'unsupported'
             assert res.read() == b'raw'
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_read(self, handler):
         with handler() as rh:
             res = validate_and_send(
@@ -620,9 +639,12 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
             assert res.readable()
             assert res.read(1) == b'H'
             assert res.read(3) == b'ost'
+            assert res.read().decode().endswith('\n\n')
+            assert res.read() == b''
 
 
 class TestHTTPProxy(TestRequestHandlerBase):
+    # Note: this only tests http urls over non-CONNECT proxy
     @classmethod
     def setup_class(cls):
         super().setup_class()
@@ -642,7 +664,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
         cls.geo_proxy_thread.daemon = True
         cls.geo_proxy_thread.start()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_http_proxy(self, handler):
         http_proxy = f'http://127.0.0.1:{self.proxy_port}'
         geo_proxy = f'http://127.0.0.1:{self.geo_port}'
@@ -668,7 +690,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
             assert res != f'normal: {real_url}'
             assert 'Accept' in res
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_noproxy(self, handler):
         with handler(proxies={'proxy': f'http://127.0.0.1:{self.proxy_port}'}) as rh:
             # NO_PROXY
@@ -678,7 +700,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
                     'utf-8')
                 assert 'Accept' in nop_response
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_allproxy(self, handler):
         url = 'http://foo.com/bar'
         with handler() as rh:
@@ -686,7 +708,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
                 'utf-8')
             assert response == f'normal: {url}'
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_http_proxy_with_idn(self, handler):
         with handler(proxies={
             'http': f'http://127.0.0.1:{self.proxy_port}',
@@ -698,7 +720,6 @@ class TestHTTPProxy(TestRequestHandlerBase):
 
 
 class TestClientCertificate:
-
     @classmethod
     def setup_class(cls):
         certfn = os.path.join(TEST_DIR, 'testcert.pem')
@@ -724,27 +745,27 @@ class TestClientCertificate:
         ) as rh:
             validate_and_send(rh, Request(f'https://127.0.0.1:{self.port}/video.html')).read().decode()
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_certificate_combined_nopass(self, handler):
         self._run_test(handler, client_cert={
             'client_certificate': os.path.join(self.certdir, 'clientwithkey.crt'),
         })
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_certificate_nocombined_nopass(self, handler):
         self._run_test(handler, client_cert={
             'client_certificate': os.path.join(self.certdir, 'client.crt'),
             'client_certificate_key': os.path.join(self.certdir, 'client.key'),
         })
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_certificate_combined_pass(self, handler):
         self._run_test(handler, client_cert={
             'client_certificate': os.path.join(self.certdir, 'clientwithencryptedkey.crt'),
             'client_certificate_password': 'foobar',
         })
 
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_certificate_nocombined_pass(self, handler):
         self._run_test(handler, client_cert={
             'client_certificate': os.path.join(self.certdir, 'client.crt'),
@@ -753,6 +774,18 @@ class TestClientCertificate:
         })
 
 
+@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True)
+class TestHTTPImpersonateRequestHandler(TestRequestHandlerBase):
+    def test_supported_impersonate_targets(self, handler):
+        with handler(headers=std_headers) as rh:
+            # note: this assumes the impersonate request handler supports the impersonate extension
+            for target in rh.supported_targets:
+                res = validate_and_send(rh, Request(
+                    f'http://127.0.0.1:{self.http_port}/headers', extensions={'impersonate': target}))
+                assert res.status == 200
+                assert std_headers['user-agent'].lower() not in res.read().decode().lower()
+
+
 class TestRequestHandlerMisc:
     """Misc generic tests for request handlers, not related to request or validation testing"""
     @pytest.mark.parametrize('handler,logger_name', [
@@ -931,6 +964,172 @@ class TestRequestsRequestHandler(TestRequestHandlerBase):
         assert called
 
 
+@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True)
+class TestCurlCFFIRequestHandler(TestRequestHandlerBase):
+
+    @pytest.mark.parametrize('params,extensions', [
+        ({}, {'impersonate': ImpersonateTarget('chrome')}),
+        ({'impersonate': ImpersonateTarget('chrome', '110')}, {}),
+        ({'impersonate': ImpersonateTarget('chrome', '99')}, {'impersonate': ImpersonateTarget('chrome', '110')}),
+    ])
+    def test_impersonate(self, handler, params, extensions):
+        with handler(headers=std_headers, **params) as rh:
+            res = validate_and_send(
+                rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions=extensions)).read().decode()
+            assert 'sec-ch-ua: "Chromium";v="110"' in res
+            # Check that user agent is added over ours
+            assert 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36' in res
+
+    def test_headers(self, handler):
+        with handler(headers=std_headers) as rh:
+            # Ensure curl-impersonate overrides our standard headers (usually added
+            res = validate_and_send(
+                rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={
+                    'impersonate': ImpersonateTarget('safari')}, headers={'x-custom': 'test', 'sec-fetch-mode': 'custom'})).read().decode().lower()
+
+            assert std_headers['user-agent'].lower() not in res
+            assert std_headers['accept-language'].lower() not in res
+            assert std_headers['sec-fetch-mode'].lower() not in res
+            # other than UA, custom headers that differ from std_headers should be kept
+            assert 'sec-fetch-mode: custom' in res
+            assert 'x-custom: test' in res
+            # but when not impersonating don't remove std_headers
+            res = validate_and_send(
+                rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'x-custom': 'test'})).read().decode().lower()
+            # std_headers should be present
+            for k, v in std_headers.items():
+                assert f'{k}: {v}'.lower() in res
+
+    @pytest.mark.parametrize('raised,expected,match', [
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.PARTIAL_FILE), IncompleteRead, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.RECV_ERROR), TransportError, None),
+    ])
+    def test_response_error_mapping(self, handler, monkeypatch, raised, expected, match):
+        import curl_cffi.requests
+
+        from yt_dlp.networking._curlcffi import CurlCFFIResponseAdapter
+        curl_res = curl_cffi.requests.Response()
+        res = CurlCFFIResponseAdapter(curl_res)
+
+        def mock_read(*args, **kwargs):
+            try:
+                raise raised()
+            except Exception as e:
+                e.response = curl_res
+                raise
+        monkeypatch.setattr(res.fp, 'read', mock_read)
+
+        with pytest.raises(expected, match=match) as exc_info:
+            res.read()
+
+        assert exc_info.type is expected
+
+    @pytest.mark.parametrize('raised,expected,match', [
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.PEER_FAILED_VERIFICATION), CertificateVerifyError, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.SSL_CONNECT_ERROR), SSLError, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.TOO_MANY_REDIRECTS), HTTPError, None),
+        (lambda: curl_cffi.requests.errors.RequestsError(
+            '', code=curl_cffi.const.CurlECode.PROXY), ProxyError, None),
+    ])
+    def test_request_error_mapping(self, handler, monkeypatch, raised, expected, match):
+        import curl_cffi.requests
+        curl_res = curl_cffi.requests.Response()
+        curl_res.status_code = 301
+
+        with handler() as rh:
+            original_get_instance = rh._get_instance
+
+            def mock_get_instance(*args, **kwargs):
+                instance = original_get_instance(*args, **kwargs)
+
+                def request(*_, **__):
+                    try:
+                        raise raised()
+                    except Exception as e:
+                        e.response = curl_res
+                        raise
+                monkeypatch.setattr(instance, 'request', request)
+                return instance
+
+            monkeypatch.setattr(rh, '_get_instance', mock_get_instance)
+
+            with pytest.raises(expected) as exc_info:
+                rh.send(Request('http://fake'))
+
+            assert exc_info.type is expected
+
+    def test_response_reader(self, handler):
+        class FakeResponse:
+            def __init__(self, raise_error=False):
+                self.raise_error = raise_error
+                self.closed = False
+
+            def iter_content(self):
+                yield b'foo'
+                yield b'bar'
+                yield b'z'
+                if self.raise_error:
+                    raise Exception('test')
+
+            def close(self):
+                self.closed = True
+
+        from yt_dlp.networking._curlcffi import CurlCFFIResponseReader
+
+        res = CurlCFFIResponseReader(FakeResponse())
+        assert res.readable
+        assert res.bytes_read == 0
+        assert res.read(1) == b'f'
+        assert res.bytes_read == 3
+        assert res._buffer == b'oo'
+
+        assert res.read(2) == b'oo'
+        assert res.bytes_read == 3
+        assert res._buffer == b''
+
+        assert res.read(2) == b'ba'
+        assert res.bytes_read == 6
+        assert res._buffer == b'r'
+
+        assert res.read(3) == b'rz'
+        assert res.bytes_read == 7
+        assert res._buffer == b''
+        assert res.closed
+        assert res._response.closed
+
+        # should handle no size param
+        res2 = CurlCFFIResponseReader(FakeResponse())
+        assert res2.read() == b'foobarz'
+        assert res2.bytes_read == 7
+        assert res2._buffer == b''
+        assert res2.closed
+
+        # should close on an exception
+        res3 = CurlCFFIResponseReader(FakeResponse(raise_error=True))
+        with pytest.raises(Exception, match='test'):
+            res3.read()
+        assert res3._buffer == b''
+        assert res3.bytes_read == 7
+        assert res3.closed
+
+        # buffer should be cleared on close
+        res4 = CurlCFFIResponseReader(FakeResponse())
+        res4.read(2)
+        assert res4._buffer == b'o'
+        res4.close()
+        assert res4.closed
+        assert res4._buffer == b''
+
+
 def run_validation(handler, error, req, **handler_kwargs):
     with handler(**handler_kwargs) as rh:
         if error:
@@ -975,6 +1174,10 @@ class TestRequestHandlerValidation:
             ('ws', False, {}),
             ('wss', False, {}),
         ]),
+        ('CurlCFFI', [
+            ('http', False, {}),
+            ('https', False, {}),
+        ]),
         (NoCheckRH, [('http', False, {})]),
         (ValidationRH, [('http', UnsupportedRequest, {})])
     ]
@@ -998,6 +1201,14 @@ class TestRequestHandlerValidation:
             ('socks5', False),
             ('socks5h', False),
         ]),
+        ('CurlCFFI', 'http', [
+            ('http', False),
+            ('https', False),
+            ('socks4', False),
+            ('socks4a', False),
+            ('socks5', False),
+            ('socks5h', False),
+        ]),
         (NoCheckRH, 'http', [('http', False)]),
         (HTTPSupportedRH, 'http', [('http', UnsupportedRequest)]),
         ('Websockets', 'ws', [('http', UnsupportedRequest)]),
@@ -1015,6 +1226,10 @@ class TestRequestHandlerValidation:
             ('all', False),
             ('unrelated', False),
         ]),
+        ('CurlCFFI', [
+            ('all', False),
+            ('unrelated', False),
+        ]),
         (NoCheckRH, [('all', False)]),
         (HTTPSupportedRH, [('all', UnsupportedRequest)]),
         (HTTPSupportedRH, [('no', UnsupportedRequest)]),
@@ -1036,6 +1251,19 @@ class TestRequestHandlerValidation:
             ({'timeout': 'notatimeout'}, AssertionError),
             ({'unsupported': 'value'}, UnsupportedRequest),
         ]),
+        ('CurlCFFI', 'http', [
+            ({'cookiejar': 'notacookiejar'}, AssertionError),
+            ({'cookiejar': YoutubeDLCookieJar()}, False),
+            ({'timeout': 1}, False),
+            ({'timeout': 'notatimeout'}, AssertionError),
+            ({'unsupported': 'value'}, UnsupportedRequest),
+            ({'impersonate': ImpersonateTarget('badtarget', None, None, None)}, UnsupportedRequest),
+            ({'impersonate': 123}, AssertionError),
+            ({'impersonate': ImpersonateTarget('chrome', None, None, None)}, False),
+            ({'impersonate': ImpersonateTarget(None, None, None, None)}, False),
+            ({'impersonate': ImpersonateTarget()}, False),
+            ({'impersonate': 'chrome'}, AssertionError)
+        ]),
         (NoCheckRH, 'http', [
             ({'cookiejar': 'notacookiejar'}, False),
             ({'somerandom': 'test'}, False),  # but any extension is allowed through
@@ -1055,7 +1283,7 @@ class TestRequestHandlerValidation:
     def test_url_scheme(self, handler, scheme, fail, handler_kwargs):
         run_validation(handler, fail, Request(f'{scheme}://'), **(handler_kwargs or {}))
 
-    @pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False)], indirect=['handler'])
+    @pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False), ('CurlCFFI', False)], indirect=['handler'])
     def test_no_proxy(self, handler, fail):
         run_validation(handler, fail, Request('http://', proxies={'no': '127.0.0.1,github.com'}))
         run_validation(handler, fail, Request('http://'), proxies={'no': '127.0.0.1,github.com'})
@@ -1078,13 +1306,13 @@ class TestRequestHandlerValidation:
         run_validation(handler, fail, Request(f'{req_scheme}://', proxies={req_scheme: f'{scheme}://example.com'}))
         run_validation(handler, fail, Request(f'{req_scheme}://'), proxies={req_scheme: f'{scheme}://example.com'})
 
-    @pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests', 'CurlCFFI'], indirect=True)
     def test_empty_proxy(self, handler):
         run_validation(handler, False, Request('http://', proxies={'http': None}))
         run_validation(handler, False, Request('http://'), proxies={'http': None})
 
     @pytest.mark.parametrize('proxy_url', ['//example.com', 'example.com', '127.0.0.1', '/a/b/c'])
-    @pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
+    @pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
     def test_invalid_proxy_url(self, handler, proxy_url):
         run_validation(handler, UnsupportedRequest, Request('http://', proxies={'http': proxy_url}))
 
@@ -1113,6 +1341,10 @@ class FakeResponse(Response):
 
 class FakeRH(RequestHandler):
 
+    def __init__(self, *args, **params):
+        self.params = params
+        super().__init__(*args, **params)
+
     def _validate(self, request):
         return
 
@@ -1271,15 +1503,10 @@ class TestYoutubeDLNetworking:
         ('', {'all': '__noproxy__'}),
         (None, {'http': 'http://127.0.0.1:8081', 'https': 'http://127.0.0.1:8081'})  # env, set https
     ])
-    def test_proxy(self, proxy, expected):
-        old_http_proxy = os.environ.get('HTTP_PROXY')
-        try:
-            os.environ['HTTP_PROXY'] = 'http://127.0.0.1:8081'  # ensure that provided proxies override env
-            with FakeYDL({'proxy': proxy}) as ydl:
-                assert ydl.proxies == expected
-        finally:
-            if old_http_proxy:
-                os.environ['HTTP_PROXY'] = old_http_proxy
+    def test_proxy(self, proxy, expected, monkeypatch):
+        monkeypatch.setenv('HTTP_PROXY', 'http://127.0.0.1:8081')
+        with FakeYDL({'proxy': proxy}) as ydl:
+            assert ydl.proxies == expected
 
     def test_compat_request(self):
         with FakeRHYDL() as ydl:
@@ -1331,6 +1558,95 @@ class TestYoutubeDLNetworking:
             with pytest.raises(SSLError, match='testerror'):
                 ydl.urlopen('ssl://testerror')
 
+    def test_unsupported_impersonate_target(self):
+        class FakeImpersonationRHYDL(FakeYDL):
+            def __init__(self, *args, **kwargs):
+                class HTTPRH(RequestHandler):
+                    def _send(self, request: Request):
+                        pass
+                    _SUPPORTED_URL_SCHEMES = ('http',)
+                    _SUPPORTED_PROXY_SCHEMES = None
+
+                super().__init__(*args, **kwargs)
+                self._request_director = self.build_request_director([HTTPRH])
+
+        with FakeImpersonationRHYDL() as ydl:
+            with pytest.raises(
+                RequestError,
+                match=r'Impersonate target "test" is not available'
+            ):
+                ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)}))
+
+    def test_unsupported_impersonate_extension(self):
+        class FakeHTTPRHYDL(FakeYDL):
+            def __init__(self, *args, **kwargs):
+                class IRH(ImpersonateRequestHandler):
+                    def _send(self, request: Request):
+                        pass
+
+                    _SUPPORTED_URL_SCHEMES = ('http',)
+                    _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc',): 'test'}
+                    _SUPPORTED_PROXY_SCHEMES = None
+
+                super().__init__(*args, **kwargs)
+                self._request_director = self.build_request_director([IRH])
+
+        with FakeHTTPRHYDL() as ydl:
+            with pytest.raises(
+                RequestError,
+                match=r'Impersonate target "test" is not available'
+            ):
+                ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)}))
+
+    def test_raise_impersonate_error(self):
+        with pytest.raises(
+            YoutubeDLError,
+            match=r'Impersonate target "test" is not available'
+        ):
+            FakeYDL({'impersonate': ImpersonateTarget('test', None, None, None)})
+
+    def test_pass_impersonate_param(self, monkeypatch):
+
+        class IRH(ImpersonateRequestHandler):
+            def _send(self, request: Request):
+                pass
+
+            _SUPPORTED_URL_SCHEMES = ('http',)
+            _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc'): 'test'}
+
+        # Bypass the check on initialize
+        brh = FakeYDL.build_request_director
+        monkeypatch.setattr(FakeYDL, 'build_request_director', lambda cls, handlers, preferences=None: brh(cls, handlers=[IRH]))
+
+        with FakeYDL({
+            'impersonate': ImpersonateTarget('abc', None, None, None)
+        }) as ydl:
+            rh = self.build_handler(ydl, IRH)
+            assert rh.impersonate == ImpersonateTarget('abc', None, None, None)
+
+    def test_get_impersonate_targets(self):
+        handlers = []
+        for target_client in ('abc', 'xyz', 'asd'):
+            class TestRH(ImpersonateRequestHandler):
+                def _send(self, request: Request):
+                    pass
+                _SUPPORTED_URL_SCHEMES = ('http',)
+                _SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget(target_client,): 'test'}
+                RH_KEY = target_client
+                RH_NAME = target_client
+            handlers.append(TestRH)
+
+        with FakeYDL() as ydl:
+            ydl._request_director = ydl.build_request_director(handlers)
+            assert set(ydl._get_available_impersonate_targets()) == {
+                (ImpersonateTarget('xyz'), 'xyz'),
+                (ImpersonateTarget('abc'), 'abc'),
+                (ImpersonateTarget('asd'), 'asd')
+            }
+            assert ydl._impersonate_target_available(ImpersonateTarget('abc'))
+            assert ydl._impersonate_target_available(ImpersonateTarget())
+            assert not ydl._impersonate_target_available(ImpersonateTarget('zxy'))
+
     @pytest.mark.parametrize('proxy_key,proxy_url,expected', [
         ('http', '__noproxy__', None),
         ('no', '127.0.0.1,foo.bar', '127.0.0.1,foo.bar'),
@@ -1341,23 +1657,17 @@ class TestYoutubeDLNetworking:
         ('http', 'socks4://example.com', 'socks4://example.com'),
         ('unrelated', '/bad/proxy', '/bad/proxy'),  # clean_proxies should ignore bad proxies
     ])
-    def test_clean_proxy(self, proxy_key, proxy_url, expected):
+    def test_clean_proxy(self, proxy_key, proxy_url, expected, monkeypatch):
         # proxies should be cleaned in urlopen()
         with FakeRHYDL() as ydl:
             req = ydl.urlopen(Request('test://', proxies={proxy_key: proxy_url})).request
             assert req.proxies[proxy_key] == expected
 
         # and should also be cleaned when building the handler
-        env_key = f'{proxy_key.upper()}_PROXY'
-        old_env_proxy = os.environ.get(env_key)
-        try:
-            os.environ[env_key] = proxy_url  # ensure that provided proxies override env
-            with FakeYDL() as ydl:
-                rh = self.build_handler(ydl)
-                assert rh.proxies[proxy_key] == expected
-        finally:
-            if old_env_proxy:
-                os.environ[env_key] = old_env_proxy
+        monkeypatch.setenv(f'{proxy_key.upper()}_PROXY', proxy_url)
+        with FakeYDL() as ydl:
+            rh = self.build_handler(ydl)
+            assert rh.proxies[proxy_key] == expected
 
     def test_clean_proxy_header(self):
         with FakeRHYDL() as ydl:
@@ -1629,3 +1939,71 @@ class TestResponse:
             assert res.geturl() == res.url
             assert res.info() is res.headers
             assert res.getheader('test') == res.get_header('test')
+
+
+class TestImpersonateTarget:
+    @pytest.mark.parametrize('target_str,expected', [
+        ('abc', ImpersonateTarget('abc', None, None, None)),
+        ('abc-120_esr', ImpersonateTarget('abc', '120_esr', None, None)),
+        ('abc-120:xyz', ImpersonateTarget('abc', '120', 'xyz', None)),
+        ('abc-120:xyz-5.6', ImpersonateTarget('abc', '120', 'xyz', '5.6')),
+        ('abc:xyz', ImpersonateTarget('abc', None, 'xyz', None)),
+        ('abc:', ImpersonateTarget('abc', None, None, None)),
+        ('abc-120:', ImpersonateTarget('abc', '120', None, None)),
+        (':xyz', ImpersonateTarget(None, None, 'xyz', None)),
+        (':xyz-6.5', ImpersonateTarget(None, None, 'xyz', '6.5')),
+        (':', ImpersonateTarget(None, None, None, None)),
+        ('', ImpersonateTarget(None, None, None, None)),
+    ])
+    def test_target_from_str(self, target_str, expected):
+        assert ImpersonateTarget.from_str(target_str) == expected
+
+    @pytest.mark.parametrize('target_str', [
+        '-120', ':-12.0', '-12:-12', '-:-',
+        '::', 'a-c-d:', 'a-c-d:e-f-g', 'a:b:'
+    ])
+    def test_target_from_invalid_str(self, target_str):
+        with pytest.raises(ValueError):
+            ImpersonateTarget.from_str(target_str)
+
+    @pytest.mark.parametrize('target,expected', [
+        (ImpersonateTarget('abc', None, None, None), 'abc'),
+        (ImpersonateTarget('abc', '120', None, None), 'abc-120'),
+        (ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'),
+        (ImpersonateTarget('abc', '120', 'xyz', '5'), 'abc-120:xyz-5'),
+        (ImpersonateTarget('abc', None, 'xyz', None), 'abc:xyz'),
+        (ImpersonateTarget('abc', '120', None, None), 'abc-120'),
+        (ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'),
+        (ImpersonateTarget('abc', None, 'xyz'), 'abc:xyz'),
+        (ImpersonateTarget(None, None, 'xyz', '6.5'), ':xyz-6.5'),
+        (ImpersonateTarget('abc', ), 'abc'),
+        (ImpersonateTarget(None, None, None, None), ''),
+    ])
+    def test_str(self, target, expected):
+        assert str(target) == expected
+
+    @pytest.mark.parametrize('args', [
+        ('abc', None, None, '5'),
+        ('abc', '120', None, '5'),
+        (None, '120', None, None),
+        (None, '120', None, '5'),
+        (None, None, None, '5'),
+        (None, '120', 'xyz', '5'),
+    ])
+    def test_invalid_impersonate_target(self, args):
+        with pytest.raises(ValueError):
+            ImpersonateTarget(*args)
+
+    @pytest.mark.parametrize('target1,target2,is_in,is_eq', [
+        (ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', None, None, None), True, True),
+        (ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', '120', None, None), True, False),
+        (ImpersonateTarget('abc', None, 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', None), True, False),
+        (ImpersonateTarget('abc', '121', 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', 'test'), False, False),
+        (ImpersonateTarget('abc'), ImpersonateTarget('abc', '120', 'xyz', 'test'), True, False),
+        (ImpersonateTarget('abc', '120', 'xyz', 'test'), ImpersonateTarget('abc'), True, False),
+        (ImpersonateTarget(), ImpersonateTarget('abc', '120', 'xyz'), True, False),
+        (ImpersonateTarget(), ImpersonateTarget(), True, True),
+    ])
+    def test_impersonate_target_in(self, target1, target2, is_in, is_eq):
+        assert (target1 in target2) is is_in
+        assert (target1 == target2) is is_eq

test/test_socks.py

@@ -286,8 +286,14 @@ def ctx(request):
     return CTX_MAP[request.param]()
 
 
+@pytest.mark.parametrize(
+    'handler,ctx', [
+        ('Urllib', 'http'),
+        ('Requests', 'http'),
+        ('Websockets', 'ws'),
+        ('CurlCFFI', 'http')
+    ], indirect=True)
 class TestSocks4Proxy:
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks4_no_auth(self, handler, ctx):
         with handler() as rh:
             with ctx.socks_server(Socks4ProxyHandler) as server_address:
@@ -295,7 +301,6 @@ class TestSocks4Proxy:
                     rh, proxies={'all': f'socks4://{server_address}'})
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks4_auth(self, handler, ctx):
         with handler() as rh:
             with ctx.socks_server(Socks4ProxyHandler, user_id='user') as server_address:
@@ -305,7 +310,6 @@ class TestSocks4Proxy:
                     rh, proxies={'all': f'socks4://user:@{server_address}'})
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks4a_ipv4_target(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
@@ -313,7 +317,6 @@ class TestSocks4Proxy:
                 assert response['version'] == 4
                 assert (response['ipv4_address'] == '127.0.0.1') != (response['domain_address'] == '127.0.0.1')
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks4a_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
@@ -322,7 +325,6 @@ class TestSocks4Proxy:
                 assert response['ipv4_address'] is None
                 assert response['domain_address'] == 'localhost'
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_ipv4_client_source_address(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler) as server_address:
             source_address = f'127.0.0.{random.randint(5, 255)}'
@@ -333,7 +335,6 @@ class TestSocks4Proxy:
                 assert response['client_address'][0] == source_address
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     @pytest.mark.parametrize('reply_code', [
         Socks4CD.REQUEST_REJECTED_OR_FAILED,
         Socks4CD.REQUEST_REJECTED_CANNOT_CONNECT_TO_IDENTD,
@@ -345,7 +346,6 @@ class TestSocks4Proxy:
                 with pytest.raises(ProxyError):
                     ctx.socks_info_request(rh)
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_ipv6_socks4_proxy(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler, bind_ip='::1') as server_address:
             with handler(proxies={'all': f'socks4://{server_address}'}) as rh:
@@ -354,7 +354,6 @@ class TestSocks4Proxy:
                 assert response['ipv4_address'] == '127.0.0.1'
                 assert response['version'] == 4
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_timeout(self, handler, ctx):
         with ctx.socks_server(Socks4ProxyHandler, sleep=2) as server_address:
             with handler(proxies={'all': f'socks4://{server_address}'}, timeout=0.5) as rh:
@@ -362,9 +361,15 @@ class TestSocks4Proxy:
                     ctx.socks_info_request(rh)
 
 
+@pytest.mark.parametrize(
+    'handler,ctx', [
+        ('Urllib', 'http'),
+        ('Requests', 'http'),
+        ('Websockets', 'ws'),
+        ('CurlCFFI', 'http')
+    ], indirect=True)
 class TestSocks5Proxy:
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5_no_auth(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -372,7 +377,6 @@ class TestSocks5Proxy:
                 assert response['auth_methods'] == [0x0]
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5_user_pass(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler, auth=('test', 'testpass')) as server_address:
             with handler() as rh:
@@ -385,7 +389,6 @@ class TestSocks5Proxy:
                 assert response['auth_methods'] == [Socks5Auth.AUTH_NONE, Socks5Auth.AUTH_USER_PASS]
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5_ipv4_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -393,7 +396,6 @@ class TestSocks5Proxy:
                 assert response['ipv4_address'] == '127.0.0.1'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -401,7 +403,6 @@ class TestSocks5Proxy:
                 assert (response['ipv4_address'] == '127.0.0.1') != (response['ipv6_address'] == '::1')
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5h_domain_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@@ -410,7 +411,6 @@ class TestSocks5Proxy:
                 assert response['domain_address'] == 'localhost'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5h_ip_target(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@@ -419,7 +419,6 @@ class TestSocks5Proxy:
                 assert response['domain_address'] is None
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_socks5_ipv6_destination(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -427,7 +426,6 @@ class TestSocks5Proxy:
                 assert response['ipv6_address'] == '::1'
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_ipv6_socks5_proxy(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler, bind_ip='::1') as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@@ -438,7 +436,6 @@ class TestSocks5Proxy:
 
     # XXX: is there any feasible way of testing IPv6 source addresses?
     # Same would go for non-proxy source_address test...
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_ipv4_client_source_address(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler) as server_address:
             source_address = f'127.0.0.{random.randint(5, 255)}'
@@ -448,7 +445,6 @@ class TestSocks5Proxy:
                 assert response['client_address'][0] == source_address
                 assert response['version'] == 5
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
     @pytest.mark.parametrize('reply_code', [
         Socks5Reply.GENERAL_FAILURE,
         Socks5Reply.CONNECTION_NOT_ALLOWED,
@@ -465,7 +461,6 @@ class TestSocks5Proxy:
                 with pytest.raises(ProxyError):
                     ctx.socks_info_request(rh)
 
-    @pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Websockets', 'ws')], indirect=True)
     def test_timeout(self, handler, ctx):
         with ctx.socks_server(Socks5ProxyHandler, sleep=2) as server_address:
             with handler(proxies={'all': f'socks5://{server_address}'}, timeout=1) as rh:

test/test_utils.py

@@ -2391,7 +2391,7 @@ Line 1
         self.assertEqual(traverse_obj(etree, '//year/text()'), ['2008', '2011', '2011'],
                          msg='`text()` at end of path should give the inner text')
         self.assertEqual(traverse_obj(etree, '//*[@direction]/@direction'), ['E', 'W', 'N', 'W', 'E'],
-                         msg='full python xpath features should be supported')
+                         msg='full Python xpath features should be supported')
         self.assertEqual(traverse_obj(etree, (0, '@name')), 'Liechtenstein',
                          msg='special transformations should act on current element')
         self.assertEqual(traverse_obj(etree, ('country', 0, ..., 'text()', {int_or_none})), [1, 2008, 141100],

test/test_websockets.py

@@ -32,8 +32,6 @@ from yt_dlp.networking.exceptions import (
 )
 from yt_dlp.utils.networking import HTTPHeaderDict
 
-from test.conftest import validate_and_send
-
 TEST_DIR = os.path.dirname(os.path.abspath(__file__))
 
 
@@ -66,7 +64,9 @@ def process_request(self, request):
 
 def create_websocket_server(**ws_kwargs):
     import websockets.sync.server
-    wsd = websockets.sync.server.serve(websocket_handler, '127.0.0.1', 0, process_request=process_request, **ws_kwargs)
+    wsd = websockets.sync.server.serve(
+        websocket_handler, '127.0.0.1', 0,
+        process_request=process_request, open_timeout=2, **ws_kwargs)
     ws_port = wsd.socket.getsockname()[1]
     ws_server_thread = threading.Thread(target=wsd.serve_forever)
     ws_server_thread.daemon = True
@@ -100,6 +100,19 @@ def create_mtls_wss_websocket_server():
     return create_websocket_server(ssl_context=sslctx)
 
 
+def ws_validate_and_send(rh, req):
+    rh.validate(req)
+    max_tries = 3
+    for i in range(max_tries):
+        try:
+            return rh.send(req)
+        except TransportError as e:
+            if i < (max_tries - 1) and 'connection closed during handshake' in str(e):
+                # websockets server sometimes hangs on new connections
+                continue
+            raise
+
+
 @pytest.mark.skipif(not websockets, reason='websockets must be installed to test websocket request handlers')
 class TestWebsSocketRequestHandlerConformance:
     @classmethod
@@ -119,7 +132,7 @@ class TestWebsSocketRequestHandlerConformance:
     @pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
     def test_basic_websockets(self, handler):
         with handler() as rh:
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             assert 'upgrade' in ws.headers
             assert ws.status == 101
             ws.send('foo')
@@ -131,7 +144,7 @@ class TestWebsSocketRequestHandlerConformance:
     @pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
     def test_send_types(self, handler, msg, opcode):
         with handler() as rh:
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             ws.send(msg)
             assert int(ws.recv()) == opcode
             ws.close()
@@ -140,10 +153,10 @@ class TestWebsSocketRequestHandlerConformance:
     def test_verify_cert(self, handler):
         with handler() as rh:
             with pytest.raises(CertificateVerifyError):
-                validate_and_send(rh, Request(self.wss_base_url))
+                ws_validate_and_send(rh, Request(self.wss_base_url))
 
         with handler(verify=False) as rh:
-            ws = validate_and_send(rh, Request(self.wss_base_url))
+            ws = ws_validate_and_send(rh, Request(self.wss_base_url))
             assert ws.status == 101
             ws.close()
 
@@ -151,7 +164,7 @@ class TestWebsSocketRequestHandlerConformance:
     def test_ssl_error(self, handler):
         with handler(verify=False) as rh:
             with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info:
-                validate_and_send(rh, Request(self.bad_wss_host))
+                ws_validate_and_send(rh, Request(self.bad_wss_host))
             assert not issubclass(exc_info.type, CertificateVerifyError)
 
     @pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
@@ -163,7 +176,7 @@ class TestWebsSocketRequestHandlerConformance:
     ])
     def test_percent_encode(self, handler, path, expected):
         with handler() as rh:
-            ws = validate_and_send(rh, Request(f'{self.ws_base_url}{path}'))
+            ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}{path}'))
             ws.send('path')
             assert ws.recv() == expected
             assert ws.status == 101
@@ -174,7 +187,7 @@ class TestWebsSocketRequestHandlerConformance:
         with handler() as rh:
             # This isn't a comprehensive test,
             # but it should be enough to check whether the handler is removing dot segments
-            ws = validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test'))
+            ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test'))
             assert ws.status == 101
             ws.send('path')
             assert ws.recv() == '/test'
@@ -187,18 +200,18 @@ class TestWebsSocketRequestHandlerConformance:
     def test_raise_http_error(self, handler, status):
         with handler() as rh:
             with pytest.raises(HTTPError) as exc_info:
-                validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}'))
+                ws_validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}'))
             assert exc_info.value.status == status
 
     @pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
     @pytest.mark.parametrize('params,extensions', [
-        ({'timeout': 0.00001}, {}),
-        ({}, {'timeout': 0.00001}),
+        ({'timeout': sys.float_info.min}, {}),
+        ({}, {'timeout': sys.float_info.min}),
     ])
     def test_timeout(self, handler, params, extensions):
         with handler(**params) as rh:
             with pytest.raises(TransportError):
-                validate_and_send(rh, Request(self.ws_base_url, extensions=extensions))
+                ws_validate_and_send(rh, Request(self.ws_base_url, extensions=extensions))
 
     @pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
     def test_cookies(self, handler):
@@ -210,18 +223,18 @@ class TestWebsSocketRequestHandlerConformance:
             comment_url=None, rest={}))
 
         with handler(cookiejar=cookiejar) as rh:
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             ws.send('headers')
             assert json.loads(ws.recv())['cookie'] == 'test=ytdlp'
             ws.close()
 
         with handler() as rh:
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             ws.send('headers')
             assert 'cookie' not in json.loads(ws.recv())
             ws.close()
 
-            ws = validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar}))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar}))
             ws.send('headers')
             assert json.loads(ws.recv())['cookie'] == 'test=ytdlp'
             ws.close()
@@ -231,7 +244,7 @@ class TestWebsSocketRequestHandlerConformance:
         source_address = f'127.0.0.{random.randint(5, 255)}'
         verify_address_availability(source_address)
         with handler(source_address=source_address) as rh:
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             ws.send('source_address')
             assert source_address == ws.recv()
             ws.close()
@@ -240,7 +253,7 @@ class TestWebsSocketRequestHandlerConformance:
     def test_response_url(self, handler):
         with handler() as rh:
             url = f'{self.ws_base_url}/something'
-            ws = validate_and_send(rh, Request(url))
+            ws = ws_validate_and_send(rh, Request(url))
             assert ws.url == url
             ws.close()
 
@@ -248,14 +261,14 @@ class TestWebsSocketRequestHandlerConformance:
     def test_request_headers(self, handler):
         with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh:
             # Global Headers
-            ws = validate_and_send(rh, Request(self.ws_base_url))
+            ws = ws_validate_and_send(rh, Request(self.ws_base_url))
             ws.send('headers')
             headers = HTTPHeaderDict(json.loads(ws.recv()))
             assert headers['test1'] == 'test'
             ws.close()
 
             # Per request headers, merged with global
-            ws = validate_and_send(rh, Request(
+            ws = ws_validate_and_send(rh, Request(
                 self.ws_base_url, headers={'test2': 'changed', 'test3': 'test3'}))
             ws.send('headers')
             headers = HTTPHeaderDict(json.loads(ws.recv()))
@@ -288,7 +301,7 @@ class TestWebsSocketRequestHandlerConformance:
             verify=False,
             client_cert=client_cert
         ) as rh:
-            validate_and_send(rh, Request(self.mtls_wss_base_url)).close()
+            ws_validate_and_send(rh, Request(self.mtls_wss_base_url)).close()
 
 
 def create_fake_ws_connection(raised):