wiki

app.py

@@ -20,7 +20,7 @@ from state import AgentState
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-from tools import ocr_image_tool, parse_excel_tool, web_search_tool, run_tools, audio_transcriber_tool
+from tools import ocr_image_tool, parse_excel_tool, web_search_tool, run_tools, audio_transcriber_tool, wikipedia_search_tool
 
 llm = ChatOpenAI(model_name="gpt-4o-mini", temperature=0.0)
 
@@ -46,6 +46,7 @@ def plan_node(state: AgentState) -> AgentState:
             "    {\"ocr_path\":\"<path to image>\"}\n"
             "    {\"excel_path\":\"<path to xlsx>\", \"excel_sheet_name\":\"<sheet>\"}\n"
             "    {\"audio_path\":\"<path to audio file>\"}\n"
+            "    {\"wiki_query\":\"<wikipedia search terms>\"}\n"
             "Do not include any extra characters or markdown—only the JSON literal."
         )
     )
@@ -76,6 +77,7 @@ def plan_node(state: AgentState) -> AgentState:
                 "excel_path",
                 "excel_sheet_name",
                 "audio_path",
+                "wiki_query",
                 "final_answer"
             }
             for k, v in parsed.items():
@@ -115,6 +117,8 @@ def finalize_node(state: AgentState) -> AgentState:
     # Note: your code already stores the audio transcription under "transcript"
     if tr := state.get("transcript"):
         combined += f"AUDIO_TRANSCRIPT: {tr}\n"
+    if wr := state.get("wiki_result"):
+        combined += f"WIKIPEDIA_RESULT: {wr}\n"
 
     # Here we demand a JSON response with a single key "final_answer"
     combined += (
@@ -174,6 +178,10 @@ def tool_node(state: AgentState) -> AgentState:
         out = audio_transcriber_tool(state)
         return out
 
+    if state.get("wiki_query"):
+        out = wikipedia_search_tool(state)
+        return out
+
     # If we somehow reach here, no recognized tool key was set:
     # print(">>> tools_node: no valid tool key found in state!")
     return {}
@@ -213,6 +221,7 @@ def route_plan(plan_out: AgentState) -> str:
         or plan_out.get("ocr_path")
         or plan_out.get("excel_path")
         or plan_out.get("audio_path")
+        or plan_out.get("wiki_query")
     ):
         # print(">> route_plan ➡️ tools")
         return "tools"
@@ -249,14 +258,16 @@ def respond_to_input(user_input: str) -> str:
     """
     system_msg = SystemMessage(
     content=(
-        "You are an agent that decides whether to call a tool or answer the user directly. "
-        "The user's question is below. If the answer can be given directly, return {'final_answer': <your answer>}."
-        "If you need to call a tool, set exactly one key from the following in a Python dict: "
-        "  • web_search_query: <search terms>\n"
-        "  • ocr_path: <path to an image file>\n"
-        "  • excel_path: <path to a .xlsx file>, excel_sheet_name: <sheet name>.\n"
-        "  • audio_path: <path to an audio file>\n"
-        "Do not include any extra text or markdown—only return a valid Python dict literal."
+        "You are an agent that decides whether to call a tool or answer directly.\n"
+        "User's question: \"" + user_input + "\"\n\n"
+        "• If you can answer directly, return exactly {\"final_answer\":\"<your answer>\"}.\n"
+        "• Otherwise, respond with exactly one of:\n"
+        "    {\"web_search_query\":\"<search terms>\"}\n"
+        "    {\"ocr_path\":\"<path to image>\"}\n"
+        "    {\"excel_path\":\"<path to xlsx>\", \"excel_sheet_name\":\"<sheet>\"}\n"
+        "    {\"audio_path\":\"<path to audio file>\"}\n"
+        "    {\"wiki_query\":\"<wikipedia search terms>\"}\n"
+        "Do not include any extra characters or markdown—only the JSON literal."
     )
     )
     human_msg = HumanMessage(content=user_input)

state.py

@@ -15,4 +15,6 @@ class AgentState(TypedDict, total=False):
     user_input: str
     audio_path: str
     transcript: str
-    audio_transcript: str
+    audio_transcript: str
+    wiki_query: str
+    wiki_result: str

tools.py

@@ -177,4 +177,69 @@ def audio_transcriber_tool(state: AgentState) -> AgentState:
     return {
         "audio_path": None,
         "transcript": text
-    }
+    }
+
+# tools.py
+
+import re
+import requests
+from state import AgentState
+
+def wikipedia_search_tool(state: AgentState) -> AgentState:
+    """
+    LangGraph wrapper for searching Wikipedia.
+    Expects: state["wiki_query"] to be a non‐empty string.
+    Returns:
+      {
+        "wiki_query": None,
+        "wiki_result": "<text summary of first matching page or an error message>"
+      }
+    If no valid wiki_query is provided, returns {}.
+    """
+    query = state.get("wiki_query", "").strip()
+    if not query:
+        return {}
+
+    try:
+        # 1) Use the MediaWiki API to search for page titles matching the query
+        search_params = {
+            "action": "query",
+            "list": "search",
+            "srsearch": query,
+            "format": "json",
+            "utf8": 1
+        }
+        search_resp = requests.get("https://en.wikipedia.org/w/api.php", params=search_params, timeout=10)
+        search_resp.raise_for_status()
+        search_data = search_resp.json()
+
+        search_results = search_data.get("query", {}).get("search", [])
+        if not search_results:
+            return {"wiki_query": None, "wiki_result": f"No Wikipedia page found for '{query}'."}
+
+        # 2) Take the first search result's title
+        first_title = search_results[0].get("title", "")
+        if not first_title:
+            return {"wiki_query": None, "wiki_result": "Unexpected format from Wikipedia search."}
+
+        # 3) Fetch the page summary for that title via the REST summary endpoint
+        title_for_url = requests.utils.requote_uri(first_title)
+        summary_url = f"https://en.wikipedia.org/api/rest_v1/page/summary/{title_for_url}"
+        summary_resp = requests.get(summary_url, timeout=10)
+        summary_resp.raise_for_status()
+        summary_data = summary_resp.json()
+
+        # 4) Extract either the "extract" field or a fallback message
+        summary_text = summary_data.get("extract")
+        if not summary_text:
+            summary_text = summary_data.get("description", "No summary available.")
+
+        return {
+            "wiki_query": None,
+            "wiki_result": f"Title: {first_title}\n\n{summary_text}"
+        }
+
+    except requests.exceptions.RequestException as e:
+        return {"wiki_query": None, "wiki_result": f"Wikipedia search error: {e}"}
+    except Exception as e:
+        return {"wiki_query": None, "wiki_result": f"Unexpected error in wikipedia_search_tool: {e}"}