diff --git a/python_api/api.py b/python_api/api.py
index 4bdd740..d381800 100644
--- a/python_api/api.py
+++ b/python_api/api.py
@@ -335,9 +335,10 @@ async def predict(request: PredictRequest):
     first_input=inputs[0]
 
 
-    # xxx
-    trail= get_trial_by_model_and_input( model_id, first_input["src"])
+
+    trail= get_trial_by_model_and_input( model_id, inputs)
     # print(trail)
+
     # print("trial")
     if trail and trail[2] is not None:
         # print(trail[2])
diff --git a/python_api/db.py b/python_api/db.py
index 16682c8..41de19a 100644
--- a/python_api/db.py
+++ b/python_api/db.py
@@ -47,42 +47,84 @@ def create_expriement( cur, conn):
     return experiment_id
 
 
-def get_trial_by_model_and_input(model_id, input_url):
-    input_url=json.loads(input_url)
-    input_url=input_url["url"]
-    input_query = """
-            SELECT trial_id
-            FROM trial_inputs
-            WHERE %s in url
-        """
+def get_trial_by_model_and_input(model_id, input_urls):
+    # Check if input_urls is a list of json objects with src and inputType keys
+    if not all(isinstance(item, dict) and 'src' in item and 'inputType' in item for item in input_urls):
+        raise ValueError("Each input_url must be a JSON object with 'src' and 'inputType' keys")
+
+    # Construct the input_query based on the number of input_urls
+    if len(input_urls) == 1:
+        input_url = input_urls[0]["src"]
+        input_query = "url LIKE %s"
+        input_values = [f"%{input_url}%"]
+    elif len(input_urls) == 2:
+        input_url_1 = input_urls[0]["src"]
+        input_url_2 = input_urls[1]["src"]
+        input_query = "url LIKE %s OR url LIKE %s"
+        input_values = [f"%{input_url_1}%", f"%{input_url_2}%"]
+    else:
+        return None
 
     # Main query to get trial details
     query = f"""
         SELECT trials.*, 
-            experiments.*, 
-            models.*, 
-            trial_inputs.*
+               experiments.*, 
+               models.*, 
+               trial_inputs.*
         FROM trials
         JOIN experiments ON trials.experiment_id = experiments.id
         JOIN models ON trials.model_id = models.id
         JOIN trial_inputs ON trials.id = trial_inputs.trial_id
         WHERE trials.completed_at IS NOT NULL
         AND trials.model_id = %s
-        AND trials.id  IN ({input_query})
+        AND ({input_query})
     """
+    
+    print(f"Debug: SQL Query: {query}")
+    print(f"Debug: Input Values: {input_values}")
+
+
 
+    # Main query to get trial details
+    query = f"""
+        SELECT trials.*, 
+               experiments.*, 
+               models.*, 
+               trial_inputs.*
+        FROM trials
+        JOIN experiments ON trials.experiment_id = experiments.id
+        JOIN models ON trials.model_id = models.id
+        JOIN trial_inputs ON trials.id = trial_inputs.trial_id
+        WHERE trials.completed_at IS NOT NULL
+        AND trials.model_id = %s
+        AND ({input_query})
+    """
+    debug_query = query % tuple([model_id] + input_values)
+    print(f"Debug: SQL Query: {debug_query}")
     # Execute the query
-    cur,conn=get_db_cur_con()
-    cur.execute(query, ( model_id,input_url))
+    try:
+        # Replace with your database connection setup
+        cur, conn = get_db_cur_con()
+        cur.execute(query, [model_id] + input_values)
 
-    # Fetch the result
-    trial = cur.fetchone()
+        # Fetch the result
+        trial = cur.fetchone()
 
-    if trial is None:
-        return None
-    
-    print(trial['experiment_id'], trial['trial_id'])
-    return (trial['experiment_id'], trial['trial_id'],trial["completed_at"]) 
+        if trial is None:
+            print("Debug: No trial found")
+            return None
+        
+        # Fetch column names for reference
+        colnames = [desc[0] for desc in cur.description]
+        # print(f"Debug: Columns: {colnames}")
+        # print(f"Debug: Trial Data: {trial}")
 
+        # Assuming the columns are returned in the order you expect
+        # You may need to adjust this depending on your database schema
+        return (trial['experiment_id'], trial['trial_id'], trial['completed_at'])
 
 
+    except (Exception, psycopg2.DatabaseError) as error:
+        print(f"Error: {error}")
+        return None
+   
\ No newline at end of file