Update app.py

app.py

@@ -146,6 +146,82 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         return status_message, results_df
 
 
+def run_evaluation_and_submit_streaming():
+    results_log = []
+
+    try:
+        yield ("⏳ Fetching evaluation questions...", None)
+        response = requests.get(f"{DEFAULT_API_URL}/questions", timeout=15)
+        response.raise_for_status()
+        questions_data = response.json()
+        if not questions_data:
+            yield ("❌ No questions received from server.", None)
+            return
+        yield (f"✅ Fetched {len(questions_data)} questions.", None)
+    except Exception as e:
+        yield (f"❌ Failed to fetch questions: {e}", None)
+        return
+
+    try:
+        agent = ShrewdAgent()
+        yield ("✅ Agent initialized.", None)
+    except Exception as e:
+        yield (f"❌ Failed to initialize agent: {e}", None)
+        return
+
+    answers = []
+    for i, item in enumerate(questions_data, start=1):
+        task_id = item.get("task_id")
+        question = item.get("question")
+        file_name = item.get("file_name")
+        full_q = compute_question_with_attachment(question, task_id, file_name)
+
+        yield (f"\n---\nQ{i}: {question.strip()[:100]}...", None)
+        try:
+            answer = agent(full_q)
+            answers.append({"task_id": task_id, "submitted_answer": answer})
+            results_log.append({
+                "Task ID": task_id,
+                "Question": question,
+                "Submitted Answer": answer
+            })
+            yield (f"A{i}: {answer.strip()[:200]}", None)
+            time.sleep(5)
+        except Exception as e:
+            err = f"❌ Error answering Q{i}: {e}"
+            results_log.append({
+                "Task ID": task_id,
+                "Question": question,
+                "Submitted Answer": err
+            })
+            yield (err, None)
+
+    try:
+        username = os.getenv("SPACE_ID", "anonymous").split("/")[0]
+        agent_code = f"https://huggingface.co/spaces/{os.getenv('SPACE_ID')}/tree/main"
+        submission_data = {
+            "username": username,
+            "agent_code": agent_code,
+            "answers": answers
+        }
+        yield ("\n📤 Submitting answers...", None)
+        resp = requests.post(f"{DEFAULT_API_URL}/submit", json=submission_data, timeout=60)
+        resp.raise_for_status()
+        result = resp.json()
+        yield (
+            f"✅ Submission Complete!\n"
+            f"Score: {result.get('score', 'N/A')}%\n"
+            f"Correct: {result.get('correct_count')}/{result.get('total_attempted')}\n"
+            f"Message: {result.get('message', 'No message')}",
+            None
+        )
+    except Exception as e:
+        yield (f"❌ Submission failed: {e}", None)
+
+    yield (None, pd.DataFrame(results_log))  # Final result table
+
+
+
 # --- Build Gradio Interface using Blocks ---
 with gr.Blocks() as demo:
     gr.Markdown("# Basic Agent Evaluation Runner")
@@ -166,16 +242,15 @@ with gr.Blocks() as demo:
 
     run_button = gr.Button("Run Evaluation & Submit All Answers")
 
-    status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
-    # Removed max_rows=10 from DataFrame constructor
+    status_output = gr.Textbox(label="Run Status / Submission Result", lines=25, interactive=False, autoscroll=True)
     results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
 
     run_button.click(
-        fn=run_and_submit_all,
+        fn=run_evaluation_and_submit_streaming,
+        inputs=[],
         outputs=[status_output, results_table]
     )
 
-
 def compute_question_with_attachment(question: str, task_id: str, file_name: str) -> str:
     if file_name:
         return f"{question}\n\nAttached file: https://agents-course-unit4-scoring.hf.space/files/{task_id}"