fixing ruff suggestions and tests needed updating.

chat_handler.py

@@ -134,9 +134,7 @@ class ChatHandler(Handler):
         enable_tts = data.get("enable_tts", self.chat_settings.enable_tts)
         system_prompt = data.get("system_prompt")
         # companions-frontend may set a custom response subject
-        response_subject = data.get(
+        response_subject = data.get("response_subject", f"ai.chat.response.{request_id}")
-            "response_subject", f"ai.chat.response.{request_id}"
-        )
 
         logger.info(f"Processing request {request_id}: {query[:50]}...")
 
@@ -159,7 +157,9 @@ class ChatHandler(Handler):
 
                 # 2. Search Milvus for context
                 documents = await self._search_context(
-                    embedding, collection, top_k=top_k,
+                    embedding,
+                    collection,
+                    top_k=top_k,
                 )
 
                 # 3. Optionally rerank documents
@@ -172,21 +172,24 @@ class ChatHandler(Handler):
                 if reranked:
                     context = self._build_context(reranked)
                     rag_sources = [
-                        d.get("source", d.get("document", "")[:80])
+                        d.get("source", d.get("document", "")[:80]) for d in reranked[:3]
-                        for d in reranked[:3]
                     ]
                     used_rag = True
 
             # 5. Generate LLM response (with or without RAG context)
             response_text = await self._generate_response(
-                query, context or None, system_prompt,
+                query,
+                context or None,
+                system_prompt,
             )
 
             # 6. Stream response chunks if requested
             if enable_streaming:
                 stream_subject = f"ai.chat.response.stream.{request_id}"
                 await self._publish_streaming_chunks(
-                    stream_subject, request_id, response_text,
+                    stream_subject,
+                    request_id,
+                    response_text,
                 )
 
             # 7. Optionally synthesize speech

tests/conftest.py

@@ -53,7 +53,7 @@ def sample_reranked():
 def mock_nats_message():
     """Create a mock NATS message."""
     msg = MagicMock()
-    msg.subject = "ai.chat.request"
+    msg.subject = "ai.chat.user.test-user-1.message"
     msg.reply = "ai.chat.response.test-123"
     return msg
 
@@ -63,7 +63,13 @@ def mock_chat_request():
     """Sample chat request payload."""
     return {
         "request_id": "test-request-123",
-        "query": "What is machine learning?",
+        "user_id": "test-user-1",
+        "username": "testuser",
+        "message": "What is machine learning?",
+        "premium": True,
+        "enable_rag": True,
+        "enable_reranker": True,
+        "enable_streaming": False,
         "collection": "test_collection",
         "enable_tts": False,
         "system_prompt": None,
@@ -75,7 +81,13 @@ def mock_chat_request_with_tts():
     """Sample chat request with TTS enabled."""
     return {
         "request_id": "test-request-456",
-        "query": "Tell me about AI",
+        "user_id": "test-user-2",
+        "username": "testuser2",
+        "message": "Tell me about AI",
+        "premium": True,
+        "enable_rag": True,
+        "enable_reranker": True,
+        "enable_streaming": False,
         "collection": "documents",
         "enable_tts": True,
         "system_prompt": "You are a helpful assistant.",

tests/test_chat_handler.py

@@ -86,7 +86,7 @@ class TestChatHandler:
 
     def test_init(self, handler):
         """Test handler initialization."""
-        assert handler.subject == "ai.chat.request"
+        assert handler.subject == "ai.chat.user.*.message"
         assert handler.queue_group == "chat-handlers"
         assert handler.chat_settings.service_name == "chat-handler"
 
@@ -111,12 +111,15 @@ class TestChatHandler:
         result = await handler.handle_message(mock_nats_message, mock_chat_request)
 
         # Verify
-        assert result["request_id"] == "test-request-123"
+        assert result["user_id"] == "test-user-1"
+        assert result["success"] is True
         assert "response" in result
         assert result["response"] == "Machine learning is a subset of AI that..."
-        assert "sources" in result  # include_sources is True by default
+        assert result["response_text"] == result["response"]
+        assert result["used_rag"] is True
+        assert isinstance(result["rag_sources"], list)
 
-        # Verify pipeline was called
+        # Verify RAG pipeline was called (enable_rag=True in fixture)
         handler.embeddings.embed_single.assert_called_once()
         handler.milvus.search_with_texts.assert_called_once()
         handler.reranker.rerank.assert_called_once()
@@ -142,7 +145,9 @@ class TestChatHandler:
 
         result = await handler.handle_message(mock_nats_message, mock_chat_request)
 
-        assert "sources" not in result
+        # New response format doesn't have a separate "sources" key;
+        # rag_sources is always present (may be empty)
+        assert "rag_sources" in result
 
     @pytest.mark.asyncio
     async def test_handle_message_with_tts(
@@ -180,7 +185,10 @@ class TestChatHandler:
         """Test LLM is called with custom system prompt."""
         request = {
             "request_id": "test-123",
-            "query": "Hello",
+            "user_id": "user-42",
+            "message": "Hello",
+            "premium": True,
+            "enable_rag": True,
             "system_prompt": "You are a pirate. Respond like one.",
         }