|
| 1 | +""" |
| 2 | +Original query: |
| 3 | +
|
| 4 | +SELECT n_name, SUM(l_extendedprice * (1 - l_discount)) AS revenue |
| 5 | + FROM customer, orders, lineitem, supplier, nation, region |
| 6 | + WHERE c_custkey = o_custkey |
| 7 | + AND l_orderkey = o_orderkey |
| 8 | + AND l_suppkey = s_suppkey |
| 9 | + AND c_nationkey = s_nationkey |
| 10 | + AND s_nationkey = n_nationkey |
| 11 | + AND n_regionkey = r_regionkey |
| 12 | + AND r_name = 'ASIA' |
| 13 | + AND o_orderdate >= MDY(1,1,1994) |
| 14 | + AND o_orderdate < MDY(1,1,1994) + 1 UNITS YEAR |
| 15 | +GROUP BY n_name |
| 16 | +ORDER BY revenue DESC |
| 17 | +""" |
| 18 | +from __future__ import annotations |
| 19 | + |
| 20 | +from typing import TYPE_CHECKING |
| 21 | + |
| 22 | +if TYPE_CHECKING: |
| 23 | + from dataframe_api.typing import SupportsDataFrameAPI |
| 24 | + |
| 25 | + |
| 26 | +def query( |
| 27 | + customer_raw: SupportsDataFrameAPI, |
| 28 | + orders_raw: SupportsDataFrameAPI, |
| 29 | + lineitem_raw: SupportsDataFrameAPI, |
| 30 | + supplier_raw: SupportsDataFrameAPI, |
| 31 | + nation_raw: SupportsDataFrameAPI, |
| 32 | + region_raw: SupportsDataFrameAPI, |
| 33 | +) -> SupportsDataFrameAPI: |
| 34 | + customer = customer_raw.__dataframe_consortium_standard__() |
| 35 | + orders = orders_raw.__dataframe_consortium_standard__() |
| 36 | + lineitem = lineitem_raw.__dataframe_consortium_standard__() |
| 37 | + supplier = supplier_raw.__dataframe_consortium_standard__() |
| 38 | + nation = nation_raw.__dataframe_consortium_standard__() |
| 39 | + region = region_raw.__dataframe_consortium_standard__() |
| 40 | + |
| 41 | + namespace = customer.__dataframe_namespace__() |
| 42 | + |
| 43 | + result = ( |
| 44 | + region.join(nation, how="inner", left_on="r_regionkey", right_on="n_regionkey") |
| 45 | + .join(customer, how="inner", left_on="n_nationkey", right_on="c_nationkey") |
| 46 | + .join(orders, how="inner", left_on="c_custkey", right_on="o_custkey") |
| 47 | + .join(lineitem, how="inner", left_on="o_orderkey", right_on="l_orderkey") |
| 48 | + .join( |
| 49 | + supplier, |
| 50 | + how="inner", |
| 51 | + left_on=["l_suppkey", "n_nationkey"], |
| 52 | + right_on=["s_suppkey", "s_nationkey"], |
| 53 | + ) |
| 54 | + ) |
| 55 | + mask = ( |
| 56 | + ( |
| 57 | + result.get_column_by_name("c_nationkey") |
| 58 | + == result.get_column_by_name("s_nationkey") |
| 59 | + ) |
| 60 | + & (result.get_column_by_name("r_name") == "ASIA") |
| 61 | + & (result.get_column_by_name("o_orderdate") >= namespace.date(1994, 1, 1)) # type: ignore |
| 62 | + & (result.get_column_by_name("o_orderdate") < namespace.date(1995, 1, 1)) # type: ignore |
| 63 | + ) |
| 64 | + result = result.filter(mask) |
| 65 | + |
| 66 | + new_column = ( |
| 67 | + result.get_column_by_name("l_extendedprice") |
| 68 | + * (1 - result.get_column_by_name("l_discount")) |
| 69 | + ).rename("revenue") |
| 70 | + result = result.assign(new_column) |
| 71 | + result = result.select(["revenue", "n_name"]) |
| 72 | + result = result.group_by("n_name").sum() |
| 73 | + |
| 74 | + return result.dataframe |
0 commit comments