Update main.py

main.py

@@ -63,20 +63,32 @@ async def retrieve_trial(request: QueryRequest):
 
     matched_trials = []
     for idx, dist in zip(indices[0], distances[0]):
-        if idx < len(df_trials):  # Ensure index is within bounds
+        if idx < len(df_trials):
             nct_id = df_trials.iloc[idx]["NCT Number"]
             trial_data = get_trial_info(nct_id)
 
             if trial_data:
-                if not np.isfinite(dist):  # FIX: Handle NaN/Inf
-                    print(f"⚠️ Skipping invalid distance: {dist}")
-                    trial_data["similarity"] = 0.0
-                else:
-                    trial_data["similarity"] = float(round(100 / (1 + dist), 2))
+                # **NEW FIX: Ensure distance is a valid number**
+                if np.isnan(dist) or np.isinf(dist):  
+                    dist = 1e9  # Set large number instead of NaN/Inf
 
+                similarity = float(round(100 / (1 + dist), 2))
+                trial_data["similarity"] = max(0.0, similarity)  # No negative similarity
+                
                 matched_trials.append(trial_data)
 
-    return {"matched_trials": matched_trials}
+    # **NEW FIX: Convert NaN values to None before returning JSON**
+    def clean_json(data):
+        if isinstance(data, dict):
+            return {k: clean_json(v) for k, v in data.items()}
+        elif isinstance(data, list):
+            return [clean_json(v) for v in data]
+        elif isinstance(data, float) and (np.isnan(data) or np.isinf(data)):
+            return None  # Convert NaN/Inf to None
+        return data
+
+    return {"matched_trials": clean_json(matched_trials)}
+
 
 # --- Root Endpoint ---
 @app.get("/")