import pandas as pd

orders = pd.DataFrame(
    {
        "order_id": [1001, 1002, 1002, 1003, 1004, 1004, 1005],
        "customer": ["Ada", "Lin", "Lin", "Maya", "Omar", "Omar", "Nia"],
        "region": ["EMEA", "APAC", "APAC", "AMER", "EMEA", "EMEA", "APAC"],
        "status": ["paid", "paid", "paid", "open", "paid", "refunded", "open"],
        "total_usd": [150.0, 240.0, 240.0, 875.0, 95.0, 95.0, 360.0],
    }
)

print(f"pandas {pd.__version__}")
print()

print("BASE")
print(orders.to_string(index=True))
print()

exact = orders.drop_duplicates()
print("DROP_EXACT_ROWS")
print(exact.to_string(index=True))
print()

latest_per_order = orders.drop_duplicates(
    subset=["order_id"],
    keep="last",
    ignore_index=True,
)
print("KEEP_LAST_BY_ORDER")
print(latest_per_order.to_string(index=False))
print()

unique_order_only = orders.drop_duplicates(
    subset=["order_id"],
    keep=False,
    ignore_index=True,
)
print("DROP_ALL_DUPLICATE_ORDER_IDS")
print(unique_order_only.to_string(index=False))
print()

print("VERIFY")
print(f"original rows: {len(orders)}")
print(f"exact rows: {len(exact)}")
print(f"latest order_ids: {latest_per_order['order_id'].tolist()}")
print(f"latest row index labels: {latest_per_order.index.tolist()}")
print(f"unique order_ids only: {unique_order_only['order_id'].tolist()}")
print(f"source rows unchanged: {len(orders)}")