Update app.py

app.py

@@ -75,7 +75,8 @@ class AgentState(TypedDict):
     label: str
     context: str
     answer: str
-    task_id: str | None = None
+    task_id: str | None
+    file_name: str | None
 
 
 
@@ -161,61 +162,43 @@ def route_question(state: AgentState) -> AgentState:
 
 def call_tools(state: AgentState) -> AgentState:
     question, label, task_id = state["question"], state["label"], state["task_id"]
+    file_name = state.get("file_name") or ""
 
-    matched_pattern = r"https?://\S+"
-    matched_obj = re.search(matched_pattern, question)
+    matched_obj = re.search(r"https?://\S+", question)
 
-    # ---- attachment detection ------------------------------------------------
-    if task_id:
+    # ---- attachment (only when a file is actually attached to this task) -----
+    file_fetched = False
+    if task_id and file_name:
         blob, ctype = _download_task_file(api_url=DEFAULT_API_URL, task_id=task_id)
-
         if any([blob, ctype]):
-            print(f"[DEBUG] attachment type={ctype} ")
-            # ── Python code ------------------------------------------------------
+            file_fetched = True
+            print(f"[DEBUG] attachment type={ctype}")
             if "python" in ctype:
                 print("[DEBUG] Working with a Python attachment file")
                 state["answer"] = run_python_file.invoke({"code": blob.decode("utf-8")})
                 state["label"] = "python_script"
                 return state
-
-            # ── Audio --------------------------------------------------------
             if "audio" in ctype:
                 print("[DEBUG] Working with an audio attachment file")
                 state["context"] = transcribe_audio.invoke({"audio_bytes": blob})
                 state["label"] = "audio"
                 return state
-
-            # ── Image --------------------------------------------------------
             if "image" in ctype:
                 print("[DEBUG] Working with an image attachment file")
-                state["answer"] = describe_image.invoke(
-                    {"img_bytes": blob, "question": question}
-                )
+                state["answer"] = describe_image.invoke({"img_bytes": blob, "question": question})
                 state["label"] = "image"
                 return state
-            
-            # ── Excel / CSV ------------------------------------------------------
-            # header_says_sheet = any(key in ctype for key in ("excel", "sheet", "csv"))
-            # blob_says_sheet = sniff_excel_type(blob) in {"xlsx", "xls", "csv"}
-         
-            parsed = ("image", "audio", "python")
-            if not any(word in ctype for word in parsed):
-
-            # if header_says_sheet or blob_says_sheet:
-                # if blob_says_sheet:
-                #     print(f"[DEBUG] octet-stream sniffed as {sniff_excel_type(blob)}")
-
-                print("[DEBUG] Working with a Excel/CSV attachment file")
-                state["answer"] = read_task_file.invoke(
-                    {"xls_bytes": blob}
-                )
-                state["label"] = "other_ext"
-                return state
-
-    elif label == "youtube":
+            # Excel / CSV / other binary
+            print("[DEBUG] Working with an Excel/CSV attachment file")
+            state["answer"] = read_task_file.invoke({"xls_bytes": blob})
+            state["label"] = "other_ext"
+            return state
+
+    # ---- label-based routing (always runs when no file was fetched) ----------
+    if label == "youtube":
         print("[TOOL] youtube_transcript")
         if matched_obj:
-            url = re.sub(r'[.,;:!?")]+$', '', matched_obj.group(0))
+            url = re.sub(r'[.,;:!?")\]]+$', '', matched_obj.group(0))
             print(f"[TOOL] fetching transcript for: {url}")
             state["context"] = get_youtube_transcript.invoke({"video_url": url})
         else:
@@ -223,8 +206,10 @@ def call_tools(state: AgentState) -> AgentState:
             state["context"] = web_search.invoke({"query": question})
     elif label == "research":
         print("[TOOL] web search")
-        # Generate a focused search query (not the full question text)
-        search_query_prompt = f"Write a short Google search query (max 10 words) to answer this question. Output ONLY the query, nothing else.\n\nQuestion: {question}"
+        search_query_prompt = (
+            "Write a short Google search query (max 10 words) to answer this question. "
+            "Output ONLY the query, nothing else.\n\nQuestion: " + question
+        )
         focused_query = _llm_router.invoke(search_query_prompt).content.strip().strip('"')
         print(f"[TOOL] search query: {focused_query}")
         search_json = web_search.invoke({"query": focused_query})
@@ -321,14 +306,15 @@ class LGAgent:
         _llm_answer = _llm_router
         self.graph = build_graph()
 
-    def __call__(self, question: str, task_id: str | None = None) -> str:
-        try: 
+    def __call__(self, question: str, task_id: str | None = None, file_name: str | None = None) -> str:
+        try:
             state: AgentState = {
-            "question": question,
-            "label": "general",
-            "context": "",
-            "answer": "",
-            "task_id": task_id,
+                "question": question,
+                "label": "general",
+                "context": "",
+                "answer": "",
+                "task_id": task_id,
+                "file_name": file_name,
             }
             final = self.graph.invoke(state)
 
@@ -398,7 +384,7 @@ def _answer_question(item: dict) -> str:
             continue
         for attempt in range(2):
             try:
-                result = LGAgent(model_id=model_id)(augmented_question, task_id=task_id)
+                result = LGAgent(model_id=model_id)(augmented_question, task_id=task_id, file_name=file_name)
                 # Pause between questions to respect Groq's tokens/min limit
                 time.sleep(5)
                 return result