add exists -> semi in q4

queries/dask/q4.py

@@ -26,12 +26,14 @@ def query() -> pd.DataFrame:
         var1 = date(1993, 7, 1)
         var2 = date(1993, 10, 1)
 
-        jn = line_item_ds.merge(orders_ds, left_on="l_orderkey", right_on="o_orderkey")
+        exists = line_item_ds[
+            line_item_ds["l_commitdate"] < line_item_ds["l_receiptdate"]
+        ]
 
+        jn = orders_ds.merge(
+            exists, left_on="o_orderkey", right_on="l_orderkey", how="leftsemi"
+        )
         jn = jn[(jn["o_orderdate"] >= var1) & (jn["o_orderdate"] < var2)]
-        jn = jn[jn["l_commitdate"] < jn["l_receiptdate"]]
-
-        jn = jn.drop_duplicates(subset=["o_orderpriority", "l_orderkey"])
 
         gb = jn.groupby("o_orderpriority")
         agg = gb.agg(

queries/polars/q4.py

@@ -15,10 +15,14 @@ def q() -> None:
     var2 = date(1993, 10, 1)
 
     q_final = (
-        lineitem.join(orders, left_on="l_orderkey", right_on="o_orderkey")
+        # SQL exists translates to semi join in Polars API
+        orders.join(
+            (lineitem.filter(pl.col("l_commitdate") < pl.col("l_receiptdate"))),
+            left_on="o_orderkey",
+            right_on="l_orderkey",
+            how="semi",
+        )
         .filter(pl.col("o_orderdate").is_between(var1, var2, closed="left"))
-        .filter(pl.col("l_commitdate") < pl.col("l_receiptdate"))
-        .unique(subset=["o_orderpriority", "l_orderkey"])
         .group_by("o_orderpriority")
         .agg(pl.len().alias("order_count"))
         .sort("o_orderpriority")

requirements.in

@@ -1,4 +1,5 @@
 dask[dataframe]
+dask-expr
 duckdb
 modin[ray]
 pandas>=2.0