Merge pull request #285 from autoscrape-labs/fix/websocket-auth

Fix websocket reconnection athentication loss

Author: thalissonvs

pydoll/browser/chromium/base.py

@@ -821,14 +821,21 @@ def _get_tab_kwargs(self, target_id: str, browser_context_id: Optional[str] = No
 
     def _get_tab_ws_address(self, tab_id: str) -> str:
         """
-        Get WebSocket address for tab. If tab_id is not provided,
-        it will be derived from the targets.
+        Get WebSocket address for a specific tab, preserving any query or fragment
+        components present in the original browser-level WebSocket URL.
+
+        This ensures authentication tokens passed via query string (e.g.,
+        ws://host/devtools/browser/abc?token=XYZ) are retained when switching
+        to the page-level endpoint (devtools/page/<tab_id>), which is critical
+        for providers like Browserless or authenticated CDP proxies.
         """
         if not self._ws_address:
             raise InvalidWebSocketAddress('WebSocket address is not set')
 
-        ws_domain = '/'.join(self._ws_address.split('/')[:3])
-        return f'{ws_domain}/devtools/page/{tab_id}'
+        parts = urlsplit(self._ws_address)
+        # Preserve scheme and netloc; build the page path and keep query/fragment
+        page_path = f'/devtools/page/{tab_id}'
+        return urlunsplit((parts.scheme, parts.netloc, page_path, parts.query, parts.fragment))
 
     @staticmethod
     def _sanitize_proxy_and_extract_auth(

pydoll/browser/managers/browser_process_manager.py

@@ -55,7 +55,7 @@ def start_browser_process(
     @staticmethod
     def _default_process_creator(command: list[str]) -> subprocess.Popen:
         """Create browser process with output capture to prevent console clutter."""
-        return subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        return subprocess.Popen(command, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 
     def stop_process(self):
         """

tests/test_browser/test_browser_base.py

@@ -219,6 +219,24 @@ async def test_connect_with_ws_address_returns_tab_and_sets_handler_ws(mock_brow
     assert tab._ws_address == 'ws://localhost:9222/devtools/page/p1'
 
 
+@pytest.mark.asyncio
+async def test_connect_with_ws_address_preserves_token_in_tab_ws(mock_browser):
+    ws_browser = 'ws://localhost:9222/devtools/browser/abcdef?token=secrettoken'
+    mock_browser.get_targets = AsyncMock(return_value=[{'type': 'page', 'url': 'https://example', 'targetId': 'p1'}])
+    mock_browser._get_valid_tab_id = AsyncMock(return_value='p1')
+    mock_browser._connection_handler._ensure_active_connection = AsyncMock()
+
+    tab = await mock_browser.connect(ws_browser)
+
+    assert mock_browser._ws_address == ws_browser
+    assert mock_browser._connection_handler._ws_address == ws_browser
+    mock_browser._connection_handler._ensure_active_connection.assert_awaited_once()
+
+    # Token should be preserved in page-level ws URL
+    assert isinstance(tab, Tab)
+    assert tab._ws_address == 'ws://localhost:9222/devtools/page/p1?token=secrettoken'
+
+
 @pytest.mark.asyncio
 async def test_new_tab_uses_ws_base_when_ws_address_present(mock_browser):
     # Simulate browser connected via ws
@@ -365,6 +383,12 @@ def test__get_tab_ws_address_raises_when_ws_not_set(mock_browser):
         mock_browser._get_tab_ws_address('some-tab')
 
 
+def test__get_tab_ws_address_preserves_query_and_fragment(mock_browser):
+    mock_browser._ws_address = 'ws://host:9222/devtools/browser/abc?token=XYZ#frag'
+    result = mock_browser._get_tab_ws_address('tab1')
+    assert result == 'ws://host:9222/devtools/page/tab1?token=XYZ#frag'
+
+
 @pytest.mark.asyncio
 async def test_get_window_id(mock_browser):
     mock_browser.get_targets = AsyncMock(return_value=[{'targetId': 'target1', 'type': 'page'}])