Update app.py

app.py

@@ -22,63 +22,53 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 llm = ChatOpenAI(model_name="gpt-4.1-mini", temperature=0.0)
 tool_node = ToolNode([ocr_image, parse_excel, web_search])
+agent = create_react_agent(model=llm, tools=tool_node)
 
-agent = create_react_agent(
-    model=llm,
-    tools=tool_node
-    # (Use default prompt/state_schema; do NOT pass your own TypedDict.)
-)
-
-# ─── 4) Build a graph that simply wires START → "agent" → END ───
-graph = StateGraph(dict)     # We’ll use plain dicts instead of a custom TypedDict
+# 2) Build a two‐edge graph:
+graph = StateGraph(dict)
 graph.add_node("agent", agent)
-
-# 4.a) Whenever user input arrives, send it straight into the agent
 graph.add_edge(START, "agent")
-
-# 4.b) Once the agent returns a final answer, go to END
 graph.add_edge("agent", END)
-
-# 4.c) Compile so we can call `.invoke()` at runtime
 compiled_graph = graph.compile()
 
-# ─── 5) Define `respond_to_input` to call `compiled_graph.invoke` ───
+# 3) The corrected respond_to_input:
 def respond_to_input(user_input: str) -> str:
-    # ① Describe your tools in a system prompt
+    """
+    We place only a SystemMessage in state["messages"], and pass the actual
+    user_input string as the second argument to invoke().
+    """
+    # (A) First message: describe your tools
     system_msg = SystemMessage(
         content=(
             "You are an assistant with access to the following tools:\n"
-            "  1) web_search(query: str) → Returns the top search results for the query as text.\n"
+            "  1) web_search(query: str) → Returns the top search results for the query.\n"
             "  2) parse_excel(path: str, sheet_name: str) → Reads an Excel file and returns its contents.\n"
             "  3) ocr_image(path: str) → Runs OCR on an image and returns any detected text.\n\n"
-            "When you need to look something up on the internet, respond exactly with JSON:\n"
-            '  {"tool":"web_search","query":"<search terms>"}\n'
-            "If you need to parse an Excel file, respond with:\n"
-            '  {"tool":"parse_excel","path":"<file.xlsx>","sheet_name":"<SheetName>"}\n'
-            "If you need to OCR an image, respond with:\n"
-            '  {"tool":"ocr_image","path":"<image.png>"}\n'
-            "If no tool is needed, reply only with your final answer as plain text."
+            "When you need up‐to‐date info, respond exactly with JSON:\n"
+            '  { "tool": "web_search", "query": "<search terms>" }\n'
+            "If you need to read an Excel file, respond:\n"
+            '  { "tool": "parse_excel", "path": "<file.xlsx>", "sheet_name": "<SheetName>" }\n'
+            "If you need OCR, respond:\n"
+            '  { "tool": "ocr_image", "path": "<image.png>" }\n'
+            "Otherwise, reply only with your final answer as plain text."
         )
     )
 
-    # ② Start the conversation with that system prompt and the user‘s question
-    initial_state = {
-        "messages": [
-            system_msg,
-            HumanMessage(content=user_input)
-        ]
-    }
+    # (B) initial state has only the system prompt
+    initial_state = { "messages": [system_msg] }
 
-    # ③ Invoke the compiled graph
-    final_state = compiled_graph.invoke(initial_state)
+    # (C) Now invoke, passing user_input separately:
+    final_state = compiled_graph.invoke(initial_state, user_input)
 
-    # ④ Collect the last assistant message (AIMessage)
-    assistant_texts = [
+    # (D) Pull out the last AIMessage from final_state["messages"]:
+    assistant_messages = [
         msg.content
         for msg in final_state["messages"]
         if isinstance(msg, AIMessage)
     ]
-    return assistant_texts[-1] if assistant_texts else ""
+    return assistant_messages[-1] if assistant_messages else ""
+
+
 class BasicAgent:
     def __init__(self):
         print("BasicAgent initialized.")