16-06-2026 2nd commit

main2.py

@@ -82,26 +82,35 @@ def get_dataframe(
 
     return fn(sql_engine)
 
-def get_employee_ids(
+
+
+
+def get_empids_clickhouse_OQAD(
     client,
-    clickhouse_engine: Engine,
     table_name: str = "OQaD",
-) -> list[int]:
+) -> pl.DataFrame:
 
     if not table_exists(client, table_name):
         log.warning(f"Table '{table_name}' does not exist.")
-        return []
+        return pl.DataFrame(
+            schema={
+                "EmpId": pl.Int64,
+                "VisitDate": pl.Date,
+            }
+        )
 
     query = f"""
-    SELECT DISTINCT employee_id
+    SELECT DISTINCT
+        employee_id AS EmpId,
+        toDate(visit_date) AS VisitDate
     FROM {table_name}
     """
 
-    return (
-        pl.read_database(query, clickhouse_engine)
-        .get_column("employee_id")
-        .to_list()
-    )
+    # ClickHouse -> PyArrow -> Polars
+    arrow_table = client.query_arrow(query)
+
+    return pl.from_arrow(arrow_table)
+
 # ==========================================================
 # Main
 # ==========================================================
@@ -157,7 +166,14 @@ def main():
         sql_engine,
         run_date,
     )
+    qf=fetch_quiz_empids()
+    db_df = get_empids_clickhouse_OQAD(client)
 
+    matched = qf.join(
+    db_df,
+    on=["EmpId", "VisitDate"],
+    how="inner",
+)
     # ------------------------------------------------------
     # Config
     # ------------------------------------------------------