feat: add first implementation of query 8

tpch/execute/q8.py

@@ -0,0 +1,39 @@
+from queries import q8
+
+from . import IO_FUNCS
+from . import customer
+from . import lineitem
+from . import nation
+from . import orders
+from . import part
+from . import region
+from . import supplier
+
+tool = "pandas[pyarrow]"
+fn = IO_FUNCS[tool]
+print(
+    q8.query(
+        fn(part),
+        fn(supplier),
+        fn(lineitem),
+        fn(orders),
+        fn(customer),
+        fn(nation),
+        fn(region),
+    )
+)
+
+
+tool = "polars[lazy]"
+fn = IO_FUNCS[tool]
+print(
+    q8.query(
+        fn(part),
+        fn(supplier),
+        fn(lineitem),
+        fn(orders),
+        fn(customer),
+        fn(nation),
+        fn(region),
+    ).collect()
+)

tpch/queries/q8.py

@@ -0,0 +1,51 @@
+from datetime import date
+
+import narwhals as nw
+from narwhals.typing import FrameT
+
+
+@nw.narwhalify
+def query(
+    part_ds: FrameT,
+    supplier_ds: FrameT,
+    line_item_ds: FrameT,
+    orders_ds: FrameT,
+    customer_ds: FrameT,
+    nation_ds: FrameT,
+    region_ds: FrameT,
+) -> FrameT:
+    nation = "BRAZIL"
+    region = "AMERICA"
+    type = "ECONOMY ANODIZED STEEL"
+    date1 = date(1995, 1, 1)
+    date2 = date(1996, 12, 31)
+
+    n1 = nation_ds.select("n_nationkey", "n_regionkey")
+    n2 = nation_ds.select("n_nationkey", "n_name")
+
+    return (
+        part_ds.join(line_item_ds, left_on="p_partkey", right_on="l_partkey")
+        .join(supplier_ds, left_on="l_suppkey", right_on="s_suppkey")
+        .join(orders_ds, left_on="l_orderkey", right_on="o_orderkey")
+        .join(customer_ds, left_on="o_custkey", right_on="c_custkey")
+        .join(n1, left_on="c_nationkey", right_on="n_nationkey")
+        .join(region_ds, left_on="n_regionkey", right_on="r_regionkey")
+        .filter(nw.col("r_name") == region)
+        .join(n2, left_on="s_nationkey", right_on="n_nationkey")
+        .filter(nw.col("o_orderdate").is_between(date1, date2))
+        .filter(nw.col("p_type") == type)
+        .select(
+            nw.col("o_orderdate").dt.year().alias("o_year"),
+            (nw.col("l_extendedprice") * (1 - nw.col("l_discount"))).alias("volume"),
+            nw.col("n_name").alias("nation"),
+        )
+        .with_columns(
+            nw.when(nw.col("nation") == nation)
+            .then(nw.col("volume"))
+            .otherwise(0)
+            .alias("_tmp")
+        )
+        .group_by("o_year")
+        .agg((nw.sum("_tmp") / nw.sum("volume")).round(2).alias("mkt_share"))
+        .sort("o_year")
+    )