Update app.py

app.py

@@ -122,85 +122,38 @@ def serve_frontend():
     except Exception:
         return "<h3>frontend.html not found in static/ — please add your frontend.html there.</h3>", 404
 
-@app.route("/chat", methods=["POST"])
 def chat():
     data = request.get_json(force=True)
-    if not isinstance(data, dict):
-        return jsonify({"error": "invalid request body"}), 400
+    chat_history = data.get("chat_history", [])
+    assistant_state = data.get("assistant_state", {})
 
-    chat_history: List[Dict[str, str]] = data.get("chat_history") or []
-    assistant_state: AssistantState = data.get("assistant_state") or {}
+    conversation_summary = assistant_state.get("conversationSummary", "")
+    language = assistant_state.get("language", "Python")
 
-    state: AssistantState = {
-        "conversationSummary": assistant_state.get("conversationSummary", ""),
-        "language": assistant_state.get("language", "Python"),
-        "taggedReplies": assistant_state.get("taggedReplies", []),
-    }
-
-    llm_messages = [{"role": "system", "content": PROGRAMMING_ASSISTANT_PROMPT}]
-
-    last_user_message = ""
-    for msg in chat_history:
-        role = msg.get("role")
-        content = msg.get("content")
-        if role in ["user", "assistant"] and content:
-            llm_messages.append({"role": role, "content": content})
-            if role == "user":
-                last_user_message = content
+    # Build prompt with system + conversation summary + chat history
+    system_prompt = f"You are a helpful programming assistant. Current language: {language}. Conversation summary: {conversation_summary}"
+    messages = [{"role": "system", "content": system_prompt}]
+    messages.extend(chat_history)
 
-    detected_lang = detect_language_from_text(last_user_message)
-    if detected_lang and detected_lang.lower() != state["language"].lower():
-        logger.info("Detected new language: %s", detected_lang)
-        state["language"] = detected_lang
+    # Call LLM, get plain text response
+    llm_response = llm.invoke(messages)
+    assistant_reply = llm_response.content if hasattr(llm_response, "content") else str(llm_response)
 
-    context_hint = f"Current Language: {state['language']}. Conversation Summary so far: {state['conversationSummary']}"
-    if llm_messages and llm_messages[-1]["role"] == "user":
-        llm_messages[-1]["content"] = f"USER MESSAGE: {last_user_message}\n\n[CONTEXT HINT: {context_hint}]"
-    elif last_user_message:
-        llm_messages.append({"role": "user", "content": f"USER MESSAGE: {last_user_message}\n\n[CONTEXT HINT: {context_hint}]"})
+    # Append assistant reply to chat history
+    chat_history.append({"role": "assistant", "content": assistant_reply})
 
-    try:
-        logger.info("Invoking LLM with full history and prepared prompt...")
-        llm_response = llm.invoke(llm_messages)
-        raw_response = llm_response.content if hasattr(llm_response, "content") else str(llm_response)
-        logger.info(f"Raw LLM response: {raw_response}")
-        parsed_result = extract_json_from_llm_response(raw_response)
-    except Exception as e:
-        logger.exception("LLM invocation failed")
-        error_detail = str(e)
-        if 'decommissioned' in error_detail:
-             error_detail = "LLM Model Error: The model is likely decommissioned. Please check the 'LLM_MODEL' environment variable or the default model in app.py."
-        return jsonify({"error": "LLM invocation failed", "detail": error_detail}), 500
-
-    if parsed_result.get("assistant_reply") == LLM_PARSE_ERROR_MESSAGE:
-        return jsonify({
-            "assistant_reply": LLM_PARSE_ERROR_MESSAGE,
-            "updated_state": state,
-            "suggested_tags": [],
-        })
-
-    updated_state_from_llm = parsed_result.get("state_updates", {})
-    if 'conversationSummary' in updated_state_from_llm:
-        state["conversationSummary"] = updated_state_from_llm["conversationSummary"]
-    if 'language' in updated_state_from_llm and updated_state_from_llm['language'].strip():
-        state["language"] = updated_state_from_llm["language"]
-
-    assistant_reply = parsed_result.get("assistant_reply")
-    code_snippet = parsed_result.get("code_snippet")
-
-    final_reply_content = assistant_reply
-    if code_snippet and code_snippet.strip():
-        if final_reply_content.strip():
-            final_reply_content += "\n\n"
-        final_reply_content += code_snippet
-
-    if not final_reply_content.strip():
-        final_reply_content = "I'm here to help with your code! What programming language are you using?"
+    # Optionally update conversation summary (e.g., call summarization chain)
+    conversation_summary = update_summary(chat_history)
 
+    # Return plain text reply and updated state
     return jsonify({
-        "assistant_reply": final_reply_content,
-        "updated_state": state,
-        "suggested_tags": parsed_result.get("suggested_tags", []),
+        "assistant_reply": assistant_reply,
+        "updated_state": {
+            "conversationSummary": conversation_summary,
+            "language": language,
+            "taggedReplies": assistant_state.get("taggedReplies", []),
+        },
+        "chat_history": chat_history,
     })
 
 @app.route("/tag_reply", methods=["POST"])