Clean up Polars queries

queries/polars/q1.py

@@ -8,12 +8,12 @@
 
 
 def q() -> None:
-    line_item_ds = utils.get_line_item_ds()
+    lineitem = utils.get_line_item_ds()
 
     var1 = date(1998, 9, 2)
 
     q_final = (
-        line_item_ds.filter(pl.col("l_shipdate") <= var1)
+        lineitem.filter(pl.col("l_shipdate") <= var1)
         .group_by("l_returnflag", "l_linestatus")
         .agg(
             pl.sum("l_quantity").alias("sum_qty"),

queries/polars/q10.py

@@ -8,19 +8,19 @@
 
 
 def q() -> None:
-    var_1 = date(1993, 10, 1)
-    var_2 = date(1994, 1, 1)
+    customer = utils.get_customer_ds()
+    orders = utils.get_orders_ds()
+    lineitem = utils.get_line_item_ds()
+    nation = utils.get_nation_ds()
 
-    customer_ds = utils.get_customer_ds()
-    orders_ds = utils.get_orders_ds()
-    line_item_ds = utils.get_line_item_ds()
-    nation_ds = utils.get_nation_ds()
+    var1 = date(1993, 10, 1)
+    var2 = date(1994, 1, 1)
 
     q_final = (
-        customer_ds.join(orders_ds, left_on="c_custkey", right_on="o_custkey")
-        .join(line_item_ds, left_on="o_orderkey", right_on="l_orderkey")
-        .join(nation_ds, left_on="c_nationkey", right_on="n_nationkey")
-        .filter(pl.col("o_orderdate").is_between(var_1, var_2, closed="left"))
+        customer.join(orders, left_on="c_custkey", right_on="o_custkey")
+        .join(lineitem, left_on="o_orderkey", right_on="l_orderkey")
+        .join(nation, left_on="c_nationkey", right_on="n_nationkey")
+        .filter(pl.col("o_orderdate").is_between(var1, var2, closed="left"))
         .filter(pl.col("l_returnflag") == "R")
         .group_by(
             "c_custkey",

queries/polars/q11.py

@@ -6,21 +6,21 @@
 
 
 def q() -> None:
-    var_1 = "GERMANY"
-    var_2 = 0.0001
+    supplier = utils.get_supplier_ds()
+    partsupp = utils.get_part_supp_ds()
+    nation = utils.get_nation_ds()
 
-    supplier_ds = utils.get_supplier_ds()
-    part_supp_ds = utils.get_part_supp_ds()
-    nation_ds = utils.get_nation_ds()
+    var1 = "GERMANY"
+    var2 = 0.0001
 
     res_1 = (
-        part_supp_ds.join(supplier_ds, left_on="ps_suppkey", right_on="s_suppkey")
-        .join(nation_ds, left_on="s_nationkey", right_on="n_nationkey")
-        .filter(pl.col("n_name") == var_1)
+        partsupp.join(supplier, left_on="ps_suppkey", right_on="s_suppkey")
+        .join(nation, left_on="s_nationkey", right_on="n_nationkey")
+        .filter(pl.col("n_name") == var1)
     )
     res_2 = res_1.select(
         (pl.col("ps_supplycost") * pl.col("ps_availqty")).sum().round(2).alias("tmp")
-        * var_2
+        * var2
     ).with_columns(pl.lit(1).alias("lit"))
 
     q_final = (

queries/polars/q12.py

@@ -8,20 +8,20 @@
 
 
 def q() -> None:
-    var_1 = "MAIL"
-    var_2 = "SHIP"
-    var_3 = date(1994, 1, 1)
-    var_4 = date(1995, 1, 1)
+    lineitem = utils.get_line_item_ds()
+    orders = utils.get_orders_ds()
 
-    line_item_ds = utils.get_line_item_ds()
-    orders_ds = utils.get_orders_ds()
+    var1 = "MAIL"
+    var2 = "SHIP"
+    var3 = date(1994, 1, 1)
+    var4 = date(1995, 1, 1)
 
     q_final = (
-        orders_ds.join(line_item_ds, left_on="o_orderkey", right_on="l_orderkey")
-        .filter(pl.col("l_shipmode").is_in([var_1, var_2]))
+        orders.join(lineitem, left_on="o_orderkey", right_on="l_orderkey")
+        .filter(pl.col("l_shipmode").is_in([var1, var2]))
         .filter(pl.col("l_commitdate") < pl.col("l_receiptdate"))
         .filter(pl.col("l_shipdate") < pl.col("l_commitdate"))
-        .filter(pl.col("l_receiptdate").is_between(var_3, var_4, closed="left"))
+        .filter(pl.col("l_receiptdate").is_between(var3, var4, closed="left"))
         .with_columns(
             pl.when(pl.col("o_orderpriority").is_in(["1-URGENT", "2-HIGH"]))
             .then(1)

queries/polars/q13.py

@@ -6,18 +6,18 @@
 
 
 def q() -> None:
-    var_1 = "special"
-    var_2 = "requests"
+    customer = utils.get_customer_ds()
+    orders = utils.get_orders_ds()
 
-    customer_ds = utils.get_customer_ds()
-    orders_ds = utils.get_orders_ds()
+    var1 = "special"
+    var2 = "requests"
 
-    orders_ds = orders_ds.filter(
-        pl.col("o_comment").str.contains(f"{var_1}.*{var_2}").not_()
+    orders = orders.filter(
+        pl.col("o_comment").str.contains(f"{var1}.*{var2}").not_()
     )
     q_final = (
-        customer_ds.join(
-            orders_ds, left_on="c_custkey", right_on="o_custkey", how="left"
+        customer.join(
+            orders, left_on="c_custkey", right_on="o_custkey", how="left"
         )
         .group_by("c_custkey")
         .agg(pl.col("o_orderkey").count().alias("c_count"))

queries/polars/q14.py

@@ -8,15 +8,15 @@
 
 
 def q() -> None:
-    var_1 = date(1995, 9, 1)
-    var_2 = date(1995, 10, 1)
+    lineitem = utils.get_line_item_ds()
+    part = utils.get_part_ds()
 
-    line_item_ds = utils.get_line_item_ds()
-    part_ds = utils.get_part_ds()
+    var1 = date(1995, 9, 1)
+    var2 = date(1995, 10, 1)
 
     q_final = (
-        line_item_ds.join(part_ds, left_on="l_partkey", right_on="p_partkey")
-        .filter(pl.col("l_shipdate").is_between(var_1, var_2, closed="left"))
+        lineitem.join(part, left_on="l_partkey", right_on="p_partkey")
+        .filter(pl.col("l_shipdate").is_between(var1, var2, closed="left"))
         .select(
             (
                 100.00

queries/polars/q15.py

@@ -8,15 +8,15 @@
 
 
 def q() -> None:
-    var_1 = date(1996, 1, 1)
-    var_2 = date(1996, 4, 1)
+    lineitem = utils.get_line_item_ds()
+    supplier = utils.get_supplier_ds()
 
-    line_item_ds = utils.get_line_item_ds()
-    supplier_ds = utils.get_supplier_ds()
+    var1 = date(1996, 1, 1)
+    var2 = date(1996, 4, 1)
 
     revenue_ds = (
-        line_item_ds.filter(
-            pl.col("l_shipdate").is_between(var_1, var_2, closed="left")
+        lineitem.filter(
+            pl.col("l_shipdate").is_between(var1, var2, closed="left")
         )
         .group_by("l_suppkey")
         .agg(
@@ -28,7 +28,7 @@ def q() -> None:
     )
 
     q_final = (
-        supplier_ds.join(revenue_ds, left_on="s_suppkey", right_on="supplier_no")
+        supplier.join(revenue_ds, left_on="s_suppkey", right_on="supplier_no")
         .filter(pl.col("total_revenue") == pl.col("total_revenue").max())
         .with_columns(pl.col("total_revenue").round(2))
         .select("s_suppkey", "s_name", "s_address", "s_phone", "total_revenue")

queries/polars/q16.py

@@ -6,22 +6,22 @@
 
 
 def q() -> None:
-    var_1 = "Brand#45"
+    partsupp = utils.get_part_supp_ds()
+    part = utils.get_part_ds()
+    supplier = utils.get_supplier_ds()
 
-    part_supp_ds = utils.get_part_supp_ds()
-    part_ds = utils.get_part_ds()
-    supplier_ds = utils.get_supplier_ds()
+    var1 = "Brand#45"
 
-    supplier_ds = supplier_ds.filter(
+    supplier = supplier.filter(
         pl.col("s_comment").str.contains(".*Customer.*Complaints.*")
     ).select(pl.col("s_suppkey"), pl.col("s_suppkey").alias("ps_suppkey"))
 
     q_final = (
-        part_ds.join(part_supp_ds, left_on="p_partkey", right_on="ps_partkey")
-        .filter(pl.col("p_brand") != var_1)
+        part.join(partsupp, left_on="p_partkey", right_on="ps_partkey")
+        .filter(pl.col("p_brand") != var1)
         .filter(pl.col("p_type").str.contains("MEDIUM POLISHED*").not_())
         .filter(pl.col("p_size").is_in([49, 14, 23, 45, 19, 3, 36, 9]))
-        .join(supplier_ds, left_on="ps_suppkey", right_on="s_suppkey", how="left")
+        .join(supplier, left_on="ps_suppkey", right_on="s_suppkey", how="left")
         .filter(pl.col("ps_suppkey_right").is_null())
         .group_by("p_brand", "p_type", "p_size")
         .agg(pl.col("ps_suppkey").n_unique().alias("supplier_cnt"))

queries/polars/q17.py

@@ -6,16 +6,16 @@
 
 
 def q() -> None:
-    var_1 = "Brand#23"
-    var_2 = "MED BOX"
+    var1 = "Brand#23"
+    var2 = "MED BOX"
 
-    line_item_ds = utils.get_line_item_ds()
-    part_ds = utils.get_part_ds()
+    lineitem = utils.get_line_item_ds()
+    part = utils.get_part_ds()
 
     res_1 = (
-        part_ds.filter(pl.col("p_brand") == var_1)
-        .filter(pl.col("p_container") == var_2)
-        .join(line_item_ds, how="left", left_on="p_partkey", right_on="l_partkey")
+        part.filter(pl.col("p_brand") == var1)
+        .filter(pl.col("p_container") == var2)
+        .join(lineitem, how="left", left_on="p_partkey", right_on="l_partkey")
     )
 
     q_final = (

queries/polars/q18.py

@@ -6,20 +6,20 @@
 
 
 def q() -> None:
-    var_1 = 300
+    var1 = 300
 
-    customer_ds = utils.get_customer_ds()
-    line_item_ds = utils.get_line_item_ds()
-    orders_ds = utils.get_orders_ds()
+    customer = utils.get_customer_ds()
+    lineitem = utils.get_line_item_ds()
+    orders = utils.get_orders_ds()
 
     q_final = (
-        line_item_ds.group_by("l_orderkey")
+        lineitem.group_by("l_orderkey")
         .agg(pl.col("l_quantity").sum().alias("sum_quantity"))
-        .filter(pl.col("sum_quantity") > var_1)
+        .filter(pl.col("sum_quantity") > var1)
         .select(pl.col("l_orderkey").alias("key"), pl.col("sum_quantity"))
-        .join(orders_ds, left_on="key", right_on="o_orderkey")
-        .join(line_item_ds, left_on="key", right_on="l_orderkey")
-        .join(customer_ds, left_on="o_custkey", right_on="c_custkey")
+        .join(orders, left_on="key", right_on="o_orderkey")
+        .join(lineitem, left_on="key", right_on="l_orderkey")
+        .join(customer, left_on="o_custkey", right_on="c_custkey")
         .group_by("c_name", "o_custkey", "key", "o_orderdate", "o_totalprice")
         .agg(pl.col("l_quantity").sum().alias("col6"))
         .select(

queries/polars/q19.py

@@ -6,11 +6,11 @@
 
 
 def q() -> None:
-    line_item_ds = utils.get_line_item_ds()
-    part_ds = utils.get_part_ds()
+    lineitem = utils.get_line_item_ds()
+    part = utils.get_part_ds()
 
     q_final = (
-        part_ds.join(line_item_ds, left_on="p_partkey", right_on="l_partkey")
+        part.join(lineitem, left_on="p_partkey", right_on="l_partkey")
         .filter(pl.col("l_shipmode").is_in(["AIR", "AIR REG"]))
         .filter(pl.col("l_shipinstruct") == "DELIVER IN PERSON")
         .filter(

queries/polars/q2.py

@@ -6,21 +6,21 @@
 
 
 def q() -> None:
-    region_ds = utils.get_region_ds()
-    nation_ds = utils.get_nation_ds()
-    supplier_ds = utils.get_supplier_ds()
-    part_ds = utils.get_part_ds()
-    part_supp_ds = utils.get_part_supp_ds()
+    region = utils.get_region_ds()
+    nation = utils.get_nation_ds()
+    supplier = utils.get_supplier_ds()
+    part = utils.get_part_ds()
+    partsupp = utils.get_part_supp_ds()
 
     var1 = 15
     var2 = "BRASS"
     var3 = "EUROPE"
 
     result_q1 = (
-        part_ds.join(part_supp_ds, left_on="p_partkey", right_on="ps_partkey")
-        .join(supplier_ds, left_on="ps_suppkey", right_on="s_suppkey")
-        .join(nation_ds, left_on="s_nationkey", right_on="n_nationkey")
-        .join(region_ds, left_on="n_regionkey", right_on="r_regionkey")
+        part.join(partsupp, left_on="p_partkey", right_on="ps_partkey")
+        .join(supplier, left_on="ps_suppkey", right_on="s_suppkey")
+        .join(nation, left_on="s_nationkey", right_on="n_nationkey")
+        .join(region, left_on="n_regionkey", right_on="r_regionkey")
         .filter(pl.col("p_size") == var1)
         .filter(pl.col("p_type").str.ends_with(var2))
         .filter(pl.col("r_name") == var3)

queries/polars/q20.py

@@ -8,31 +8,31 @@
 
 
 def q() -> None:
-    var_1 = date(1994, 1, 1)
-    var_2 = date(1995, 1, 1)
-    var_3 = "CANADA"
-    var_4 = "forest"
+    lineitem = utils.get_line_item_ds()
+    nation = utils.get_nation_ds()
+    supplier = utils.get_supplier_ds()
+    part = utils.get_part_ds()
+    partsupp = utils.get_part_supp_ds()
 
-    line_item_ds = utils.get_line_item_ds()
-    nation_ds = utils.get_nation_ds()
-    supplier_ds = utils.get_supplier_ds()
-    part_ds = utils.get_part_ds()
-    part_supp_ds = utils.get_part_supp_ds()
+    var1 = date(1994, 1, 1)
+    var2 = date(1995, 1, 1)
+    var3 = "CANADA"
+    var4 = "forest"
 
     res_1 = (
-        line_item_ds.filter(
-            pl.col("l_shipdate").is_between(var_1, var_2, closed="left")
+        lineitem.filter(
+            pl.col("l_shipdate").is_between(var1, var2, closed="left")
         )
         .group_by("l_partkey", "l_suppkey")
         .agg((pl.col("l_quantity").sum() * 0.5).alias("sum_quantity"))
     )
-    res_2 = nation_ds.filter(pl.col("n_name") == var_3)
-    res_3 = supplier_ds.join(res_2, left_on="s_nationkey", right_on="n_nationkey")
+    res_2 = nation.filter(pl.col("n_name") == var3)
+    res_3 = supplier.join(res_2, left_on="s_nationkey", right_on="n_nationkey")
 
     q_final = (
-        part_ds.filter(pl.col("p_name").str.starts_with(var_4))
+        part.filter(pl.col("p_name").str.starts_with(var4))
         .select(pl.col("p_partkey").unique())
-        .join(part_supp_ds, left_on="p_partkey", right_on="ps_partkey")
+        .join(partsupp, left_on="p_partkey", right_on="ps_partkey")
         .join(
             res_1,
             left_on=["ps_suppkey", "p_partkey"],