DX-120975: Fix endpoint (#113)

* Fix endpoint

* Fix test

* Fixed pre-existing error

* Fixed endpoint

Author: AfonsoP1nto

src/dremioai/api/dremio/ai_tools.py

@@ -20,6 +20,7 @@
 
 from aiohttp import ClientResponseError
 from dremioai.api.transport import DremioAsyncHttpClient as AsyncHttpClient
+from dremioai.config import settings
 from dremioai.log import logger
 
 log = logger(__name__)
@@ -64,10 +65,9 @@ def is_empty(self) -> bool:
 async def list_tools() -> ListToolsResponse:
     try:
         client = AsyncHttpClient()
-        return await client.get(
-            "/api/v4/ai/tools",
-            deser=ListToolsResponse,
-        )
+        project_id = settings.instance().dremio.project_id
+        endpoint = f"/v1/projects/{project_id}" if project_id else "/api/v4"
+        return await client.get(f"{endpoint}/ai/tools", deser=ListToolsResponse)
     except ClientResponseError as e:
         log.exception("Failed to list AI tools")
         return ListToolsResponse(error=f"HTTP {e.status} {e.message}")
@@ -80,8 +80,10 @@ async def invoke_tool(tool_name: str, args: Dict[str, Any]) -> InvokeToolRespons
     safe_name = quote(tool_name, safe="")
     try:
         client = AsyncHttpClient()
+        project_id = settings.instance().dremio.project_id
+        endpoint = f"/v1/projects/{project_id}" if project_id else "/api/v4"
         return await client.post(
-            f"/api/v4/ai/tools/{safe_name}:invoke",
+            f"{endpoint}/ai/tools/{safe_name}:invoke",
             body={"args": args},
             deser=InvokeToolResponse,
         )

tests/api/dremio/test_ai_tools.py

@@ -26,10 +26,19 @@
 
 # --- list_tools tests ---
 
+
+@pytest.fixture
+def no_project_id_settings(mock_settings_instance):
+    old_project_id = mock_settings_instance.dremio.project_id
+    mock_settings_instance.dremio.raw_project_id = None
+    yield mock_settings_instance
+    if old_project_id is not None:
+        mock_settings_instance.dremio.project_id = old_project_id
+
 @pytest.mark.asyncio
 async def test_list_tools_returns_tools(mock_settings_instance):
     with HttpMockFramework() as mock:
-        mock.load_mock_data(r"/api/v4/ai/tools$", "ai_tools/list_tools.json")
+        mock.load_mock_data(r"/v1/projects/[^/]+/ai/tools$", "ai_tools/list_tools.json")
         result = await list_tools()
     assert bool(result)
     assert len(result.tools) == 3
@@ -42,7 +51,7 @@ async def test_list_tools_returns_tools(mock_settings_instance):
 @pytest.mark.asyncio
 async def test_list_tools_returns_input_schema(mock_settings_instance):
     with HttpMockFramework() as mock:
-        mock.load_mock_data(r"/api/v4/ai/tools$", "ai_tools/list_tools.json")
+        mock.load_mock_data(r"/v1/projects/[^/]+/ai/tools$", "ai_tools/list_tools.json")
         result = await list_tools()
     run_sql = next(t for t in result.tools if t.name == "runSql")
     assert run_sql.input_schema["type"] == "object"
@@ -52,7 +61,7 @@ async def test_list_tools_returns_input_schema(mock_settings_instance):
 @pytest.mark.asyncio
 async def test_list_tools_empty_registry(mock_settings_instance):
     with HttpMockFramework() as mock:
-        mock.add_mock_response(r"/api/v4/ai/tools$", {"tools": []})
+        mock.add_mock_response(r"/v1/projects/[^/]+/ai/tools$", {"tools": []})
         result = await list_tools()
     assert result.tools == []
     assert bool(result)
@@ -63,7 +72,10 @@ async def test_list_tools_empty_registry(mock_settings_instance):
 @pytest.mark.asyncio
 async def test_invoke_tool_success(mock_settings_instance):
     with HttpMockFramework() as mock:
-        mock.load_mock_data(r"/api/v4/ai/tools/runSql:invoke$", "ai_tools/invoke_result.json")
+        mock.load_mock_data(
+            r"/v1/projects/[^/]+/ai/tools/runSql:invoke$",
+            "ai_tools/invoke_result.json",
+        )
         result = await invoke_tool("runSql", {"sqlText": "SELECT 1"})
     assert bool(result)
     assert result.result["columns"] == ["id", "name"]
@@ -73,7 +85,10 @@ async def test_invoke_tool_success(mock_settings_instance):
 @pytest.mark.asyncio
 async def test_invoke_tool_error_response(mock_settings_instance):
     with HttpMockFramework() as mock:
-        mock.load_mock_data(r"/api/v4/ai/tools/unknownTool:invoke$", "ai_tools/invoke_error.json")
+        mock.load_mock_data(
+            r"/v1/projects/[^/]+/ai/tools/unknownTool:invoke$",
+            "ai_tools/invoke_error.json",
+        )
         result = await invoke_tool("unknownTool", {})
     assert result.error is not None
     assert "not found" in result.error
@@ -146,7 +161,11 @@ def test_invoke_tool_response_is_empty_false_when_error():
 async def test_list_tools_http_error(mock_settings_instance):
     """list_tools should return a response with error set on HTTP 4xx/5xx."""
     with HttpMockFramework() as mock:
-        mock.add_mock_response(r"/api/v4/ai/tools$", {"error": "Unauthorized"}, status=401)
+        mock.add_mock_response(
+            r"/v1/projects/[^/]+/ai/tools$",
+            {"error": "Unauthorized"},
+            status=401,
+        )
         result = await list_tools()
     assert not bool(result)
     assert result.error is not None
@@ -157,7 +176,11 @@ async def test_list_tools_http_error(mock_settings_instance):
 async def test_invoke_tool_http_error(mock_settings_instance):
     """invoke_tool should return a response with error set on HTTP 500."""
     with HttpMockFramework() as mock:
-        mock.add_mock_response(r"/api/v4/ai/tools/runSql:invoke$", {"error": "Internal Server Error"}, status=500)
+        mock.add_mock_response(
+            r"/v1/projects/[^/]+/ai/tools/runSql:invoke$",
+            {"error": "Internal Server Error"},
+            status=500,
+        )
         result = await invoke_tool("runSql", {"sqlText": "SELECT 1"})
     assert not bool(result)
     assert result.error is not None
@@ -169,6 +192,30 @@ async def test_invoke_tool_url_encodes_name(mock_settings_instance):
     """tool_name with special characters should be URL-encoded."""
     with HttpMockFramework() as mock:
         # The encoded name "my%2Ftool" should appear in the URL
-        mock.add_mock_response(r"/api/v4/ai/tools/my%2Ftool:invoke$", {"result": "ok", "error": None})
+        mock.add_mock_response(
+            r"/v1/projects/[^/]+/ai/tools/my%2Ftool:invoke$",
+            {"result": "ok", "error": None},
+        )
         result = await invoke_tool("my/tool", {})
     assert result.result == "ok"
+
+
+@pytest.mark.asyncio
+async def test_list_tools_uses_api_v4_without_project_id(no_project_id_settings):
+    with HttpMockFramework() as mock:
+        mock.load_mock_data(r"/api/v4/ai/tools$", "ai_tools/list_tools.json")
+        result = await list_tools()
+    assert bool(result)
+    assert len(result.tools) == 3
+
+
+@pytest.mark.asyncio
+async def test_invoke_tool_uses_api_v4_without_project_id(no_project_id_settings):
+    with HttpMockFramework() as mock:
+        mock.load_mock_data(
+            r"/api/v4/ai/tools/runSql:invoke$",
+            "ai_tools/invoke_result.json",
+        )
+        result = await invoke_tool("runSql", {"sqlText": "SELECT 1"})
+    assert bool(result)
+    assert result.result["columns"] == ["id", "name"]

tests/conftest.py

@@ -50,6 +50,24 @@
 from prometheus_client import CollectorRegistry
 
 
+@pytest.fixture(autouse=True)
+def reset_uvicorn_logger_propagation():
+    """Reset uvicorn logger state between tests.
+
+    Uvicorn's configure_logging() sets uvicorn.access.propagate=False via its
+    default LOGGING_CONFIG when a server starts. This leaks into subsequent tests
+    that assert stdlib loggers propagate to the root handler.
+    """
+    yield
+    import logging
+
+    for name in ("uvicorn.access", "uvicorn.error", "uvicorn"):
+        lg = logging.getLogger(name)
+        lg.propagate = True
+        for h in lg.handlers[:]:
+            lg.removeHandler(h)
+
+
 @pytest.fixture(autouse=True)
 def reset_sse_starlette_app_status():
     """