Fix: Add try-except block to catch runtime errors and fix CSS

app.py

@@ -65,8 +65,9 @@ HEAD_HTML = """
     .search-row {
         display: flex !important;
         flex-direction: row !important;
-        align-items: flex-start !important; /* Align top to handle potential textarea growth */
+        align-items: flex-start !important;
         gap: 1rem !important;
+        flex-wrap: nowrap !important; /* Prevent wrapping */
     }
     
     /* Loader Overlay */
@@ -244,86 +245,100 @@ def run_benchmark(query):
     # 1. Yield Loader
     yield LOADER_HTML
     
-    # 2. Perform Search (Live)
-    start_total = time.time()
-    
-    # Generate Embedding
     try:
+        # 2. Perform Search (Live)
+        start_total = time.time()
+        
+        # Generate Embedding
         print("DEBUG: Generating embedding...")
         query_vec = get_embedding(query)
         print("DEBUG: Embedding generated.")
-    except Exception as e:
-        print(f"ERROR: Embedding failed: {e}")
-        query_vec = [0.0] * VECTOR_SIZE # Dummy
 
-    # Router Prediction
-    if router:
-        print("DEBUG: Predicting cluster...")
-        target_cluster, confidence = router.predict(query_vec)
-        print(f"DEBUG: Predicted cluster {target_cluster} with confidence {confidence}")
-    else:
-        print("DEBUG: No router loaded, using mock.")
-        target_cluster, confidence = 0, 0.95 # Mock
+        # Router Prediction
+        if router:
+            print("DEBUG: Predicting cluster...")
+            target_cluster, confidence = router.predict(query_vec)
+            print(f"DEBUG: Predicted cluster {target_cluster} with confidence {confidence}")
+        else:
+            print("DEBUG: No router loaded, using mock.")
+            target_cluster, confidence = 0, 0.95 # Mock
+            
+        # Search
+        print("DEBUG: Searching Qdrant...")
+        results, mode = vector_db.search_hybrid(query_vec, target_cluster, confidence)
+        print(f"DEBUG: Search complete. Found {len(results)} results.")
         
-    # Search
-    print("DEBUG: Searching Qdrant...")
-    results, mode = vector_db.search_hybrid(query_vec, target_cluster, confidence)
-    print(f"DEBUG: Search complete. Found {len(results)} results.")
-    
-    end_total = time.time()
-    latency_ms = (end_total - start_total) * 1000
-    
-    # 3. Construct Data Rows
-    
-    # Live Row (MiniLM + LightGBM)
-    # Mocking shards searched based on confidence for demo visual
-    shards_searched = 2 if confidence > 0.8 else 33
-    total_shards = 33
-    direct_time = latency_ms * (total_shards / shards_searched) * 1.2 # Estimate baseline
-    
-    live_row = {
-        "embedding": "MiniLM-L6-v2 (Active)",
-        "router": "LightGBM",
-        "optimizedTime": f"{latency_ms:.1f} ms",
-        "shardsSearched": f"{shards_searched} / {total_shards}",
-        "totalShards": total_shards,
-        "accuracy": f"{confidence:.2f}",
-        "confDisplay": f"{confidence*100:.1f}%",
-        "directTime": f"{direct_time:.1f} ms",
-        "efficiency": f"+{((1 - latency_ms/direct_time)*100):.1f}%"
-    }
-    
-    # Reference Rows (Static)
-    ref_rows = [
-        {
-            "embedding": "Gemma 300M",
+        end_total = time.time()
+        latency_ms = (end_total - start_total) * 1000
+        
+        # 3. Construct Data Rows
+        
+        # Live Row (MiniLM + LightGBM)
+        # Mocking shards searched based on confidence for demo visual
+        shards_searched = 2 if confidence > 0.8 else 33
+        total_shards = 33
+        direct_time = latency_ms * (total_shards / shards_searched) * 1.2 # Estimate baseline
+        
+        live_row = {
+            "embedding": "MiniLM-L6-v2 (Active)",
             "router": "LightGBM",
-            "optimizedTime": "128 ms",
-            "shardsSearched": "9 / 16",
-            "totalShards": 16,
-            "accuracy": "0.97",
-            "confDisplay": "97.1%",
-            "directTime": "220 ms",
-            "efficiency": "+41.8%"
-        },
-        {
-            "embedding": "Qwen 600M",
-            "router": "XGBoost",
-            "optimizedTime": "109 ms",
-            "shardsSearched": "7 / 16",
-            "totalShards": 16,
-            "accuracy": "0.90",
-            "confDisplay": "90.1%",
-            "directTime": "235 ms",
-            "efficiency": "+53.6%"
+            "optimizedTime": f"{latency_ms:.1f} ms",
+            "shardsSearched": f"{shards_searched} / {total_shards}",
+            "totalShards": total_shards,
+            "accuracy": f"{confidence:.2f}",
+            "confDisplay": f"{confidence*100:.1f}%",
+            "directTime": f"{direct_time:.1f} ms",
+            "efficiency": f"+{((1 - latency_ms/direct_time)*100):.1f}%"
         }
-    ]
-    
-    all_rows = [live_row] + ref_rows
-    
-    print("DEBUG: Yielding final HTML.")
-    # 4. Yield Final HTML
-    yield generate_table_html(all_rows)
+        
+        # Reference Rows (Static)
+        ref_rows = [
+            {
+                "embedding": "Gemma 300M",
+                "router": "LightGBM",
+                "optimizedTime": "128 ms",
+                "shardsSearched": "9 / 16",
+                "totalShards": 16,
+                "accuracy": "0.97",
+                "confDisplay": "97.1%",
+                "directTime": "220 ms",
+                "efficiency": "+41.8%"
+            },
+            {
+                "embedding": "Qwen 600M",
+                "router": "XGBoost",
+                "optimizedTime": "109 ms",
+                "shardsSearched": "7 / 16",
+                "totalShards": 16,
+                "accuracy": "0.90",
+                "confDisplay": "90.1%",
+                "directTime": "235 ms",
+                "efficiency": "+53.6%"
+            }
+        ]
+        
+        all_rows = [live_row] + ref_rows
+        
+        print("DEBUG: Yielding final HTML.")
+        # 4. Yield Final HTML
+        yield generate_table_html(all_rows)
+
+    except Exception as e:
+        import traceback
+        error_msg = traceback.format_exc()
+        print(f"CRITICAL ERROR in run_benchmark: {error_msg}")
+        
+        # Yield Error HTML
+        yield f"""
+        <div class="bg-red-50 border border-red-200 rounded-2xl p-6 text-red-800">
+            <h3 class="font-bold text-lg mb-2 flex items-center gap-2">
+                <span class="material-symbols-outlined">error</span>
+                Runtime Error
+            </h3>
+            <p class="mb-4">An error occurred while running the benchmark:</p>
+            <pre class="bg-red-100 p-4 rounded-lg text-xs font-mono overflow-x-auto">{error_msg}</pre>
+        </div>
+        """
 
 # --- Gradio App ---
 with gr.Blocks(theme=gr.themes.Base(), css=None, head=HEAD_HTML) as demo: