chore(typing): fix/ignore utils, tpch warnings (#2426)

dangotbanned · web-flow · commit 41475ade607f · 2025-04-26T12:44:37.000+02:00
* chore(typing): fix `check_docstrings` `Path` warning `pyright` did a good job spotting that one * chore: Use `dir(...)` instead of `.__dir__()` Fixes 2/6 errors and shrinks the script * chore(typing): Fix 2 more errors These have been popping up and bugging me for 2 weeks 🙈 #2351 (comment) * chore(typing): Fix the last 2 as well * chore: Simplify the rest * chore(typing): Ignore remaining in `utils` * chore(typing): Ignore everything in `tpch` These are all the same issue, can be fixed with a constrained `TypeVar` * fix: pre-`3.13` compat? Following where `__name__` used to be sourced from - python/cpython#101876 * chore(typing): Fix some of the new `tpch` ignores The rest seem pretty unavoidable * fix: don't clobber `typing` 🤦‍♂️ https://github.com/narwhals-dev/narwhals/actions/runs/14624570842/job/41032838053?pr=2426
diff --git a/tpch/__init__.py b/tpch/__init__.py
@@ -0,0 +1,5 @@
+from __future__ import annotations
+
+from tpch import _typing
+
+__all__ = ["_typing"]
diff --git a/tpch/_typing.py b/tpch/_typing.py
@@ -0,0 +1,9 @@
+from __future__ import annotations
+
+from typing import Any
+from typing import TypeVar
+
+from narwhals import DataFrame
+from narwhals import LazyFrame
+
+FrameT = TypeVar("FrameT", DataFrame[Any], LazyFrame[Any])
diff --git a/tpch/execute.py b/tpch/execute.py
@@ -16,7 +16,7 @@
 import narwhals as nw
 
 pd.options.mode.copy_on_write = True
-pd.options.future.infer_string = True
+pd.options.future.infer_string = True  # pyright: ignore[reportAttributeAccessIssue, reportOptionalMemberAccess]
 pl.Config.set_fmt_float("full")
 
 DATA_DIR = Path("data")
diff --git a/tpch/queries/q1.py b/tpch/queries/q1.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(lineitem: FrameT) -> FrameT:
diff --git a/tpch/queries/q10.py b/tpch/queries/q10.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
diff --git a/tpch/queries/q11.py b/tpch/queries/q11.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -30,7 +30,7 @@ def query(
         q1.with_columns((nw.col("ps_supplycost") * nw.col("ps_availqty")).alias("value"))
         .group_by("ps_partkey")
         .agg(nw.sum("value"))
-        .join(q2, how="cross")
+        .join(q2, how="cross")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("value") > nw.col("tmp"))
         .select("ps_partkey", "value")
         .sort("value", descending=True)
diff --git a/tpch/queries/q12.py b/tpch/queries/q12.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(line_item_ds: FrameT, orders_ds: FrameT) -> FrameT:
diff --git a/tpch/queries/q13.py b/tpch/queries/q13.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(customer_ds: FrameT, orders_ds: FrameT) -> FrameT:
@@ -14,7 +14,7 @@ def query(customer_ds: FrameT, orders_ds: FrameT) -> FrameT:
 
     orders = orders_ds.filter(~nw.col("o_comment").str.contains(f"{var1}.*{var2}"))
     return (
-        customer_ds.join(orders, left_on="c_custkey", right_on="o_custkey", how="left")
+        customer_ds.join(orders, left_on="c_custkey", right_on="o_custkey", how="left")  # pyright: ignore[reportArgumentType]
         .group_by("c_custkey")
         .agg(nw.col("o_orderkey").count().alias("c_count"))
         .group_by("c_count")
diff --git a/tpch/queries/q14.py b/tpch/queries/q14.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(line_item_ds: FrameT, part_ds: FrameT) -> FrameT:
diff --git a/tpch/queries/q15.py b/tpch/queries/q15.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -29,7 +29,7 @@ def query(
     )
 
     return (
-        supplier_ds.join(revenue, left_on="s_suppkey", right_on="supplier_no")
+        supplier_ds.join(revenue, left_on="s_suppkey", right_on="supplier_no")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("total_revenue") == nw.col("total_revenue").max())
         .with_columns(nw.col("total_revenue").round(2))
         .select("s_suppkey", "s_name", "s_address", "s_phone", "total_revenue")
diff --git a/tpch/queries/q16.py b/tpch/queries/q16.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(part_ds: FrameT, partsupp_ds: FrameT, supplier_ds: FrameT) -> FrameT:
@@ -20,7 +20,7 @@ def query(part_ds: FrameT, partsupp_ds: FrameT, supplier_ds: FrameT) -> FrameT:
         .filter(nw.col("p_brand") != var1)
         .filter(~nw.col("p_type").str.contains("MEDIUM POLISHED*"))
         .filter(nw.col("p_size").is_in([49, 14, 23, 45, 19, 3, 36, 9]))
-        .join(supplier, left_on="ps_suppkey", right_on="s_suppkey", how="left")
+        .join(supplier, left_on="ps_suppkey", right_on="s_suppkey", how="left")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("ps_suppkey_right").is_null())
         .group_by("p_brand", "p_type", "p_size")
         .agg(nw.col("ps_suppkey").n_unique().alias("supplier_cnt"))
diff --git a/tpch/queries/q17.py b/tpch/queries/q17.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(lineitem_ds: FrameT, part_ds: FrameT) -> FrameT:
@@ -23,7 +23,7 @@ def query(lineitem_ds: FrameT, part_ds: FrameT) -> FrameT:
         .group_by("p_partkey")
         .agg(nw.col("l_quantity_times_point_2").mean().alias("avg_quantity"))
         .select(nw.col("p_partkey").alias("key"), nw.col("avg_quantity"))
-        .join(query1, left_on="key", right_on="p_partkey")
+        .join(query1, left_on="key", right_on="p_partkey")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("l_quantity") < nw.col("avg_quantity"))
         .select((nw.col("l_extendedprice").sum() / 7.0).round(2).alias("avg_yearly"))
     )
diff --git a/tpch/queries/q18.py b/tpch/queries/q18.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(customer_ds: FrameT, lineitem_ds: FrameT, orders_ds: FrameT) -> FrameT:
@@ -18,7 +18,7 @@ def query(customer_ds: FrameT, lineitem_ds: FrameT, orders_ds: FrameT) -> FrameT
     )
 
     return (
-        orders_ds.join(query1, left_on="o_orderkey", right_on="l_orderkey", how="semi")
+        orders_ds.join(query1, left_on="o_orderkey", right_on="l_orderkey", how="semi")  # pyright: ignore[reportArgumentType]
         .join(lineitem_ds, left_on="o_orderkey", right_on="l_orderkey")
         .join(customer_ds, left_on="o_custkey", right_on="c_custkey")
         .group_by("c_name", "o_custkey", "o_orderkey", "o_orderdate", "o_totalprice")
diff --git a/tpch/queries/q19.py b/tpch/queries/q19.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(lineitem_ds: FrameT, part_ds: FrameT) -> FrameT:
diff --git a/tpch/queries/q2.py b/tpch/queries/q2.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -46,7 +46,7 @@ def query(
         result_q2.group_by("p_partkey")
         .agg(nw.col("ps_supplycost").min().alias("ps_supplycost"))
         .join(
-            result_q2,
+            result_q2,  # pyright: ignore[reportArgumentType]
             left_on=["p_partkey", "ps_supplycost"],
             right_on=["p_partkey", "ps_supplycost"],
         )
diff --git a/tpch/queries/q20.py b/tpch/queries/q20.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -28,22 +28,22 @@ def query(
         .with_columns(sum_quantity=nw.col("sum_quantity") * 0.5)
     )
     query2 = nation_ds.filter(nw.col("n_name") == var3)
-    query3 = supplier_ds.join(query2, left_on="s_nationkey", right_on="n_nationkey")
+    query3 = supplier_ds.join(query2, left_on="s_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
 
     return (
         part_ds.filter(nw.col("p_name").str.starts_with(var4))
         .select("p_partkey")
         .unique("p_partkey")
         .join(partsupp_ds, left_on="p_partkey", right_on="ps_partkey")
         .join(
-            query1,
+            query1,  # pyright: ignore[reportArgumentType]
             left_on=["ps_suppkey", "p_partkey"],
             right_on=["l_suppkey", "l_partkey"],
         )
         .filter(nw.col("ps_availqty") > nw.col("sum_quantity"))
         .select("ps_suppkey")
         .unique("ps_suppkey")
-        .join(query3, left_on="ps_suppkey", right_on="s_suppkey")
+        .join(query3, left_on="ps_suppkey", right_on="s_suppkey")  # pyright: ignore[reportArgumentType]
         .select("s_name", "s_address")
         .sort("s_name")
     )
diff --git a/tpch/queries/q21.py b/tpch/queries/q21.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -21,7 +21,7 @@ def query(
         .agg(nw.len().alias("n_supp_by_order"))
         .filter(nw.col("n_supp_by_order") > 1)
         .join(
-            lineitem.filter(nw.col("l_receiptdate") > nw.col("l_commitdate")),
+            lineitem.filter(nw.col("l_receiptdate") > nw.col("l_commitdate")),  # pyright: ignore[reportArgumentType]
             left_on="l_orderkey",
             right_on="l_orderkey",
         )
@@ -31,7 +31,7 @@ def query(
         q1.group_by("l_orderkey")
         .agg(nw.len().alias("n_supp_by_order"))
         .join(
-            q1,
+            q1,  # pyright: ignore[reportArgumentType]
             left_on="l_orderkey",
             right_on="l_orderkey",
         )
diff --git a/tpch/queries/q22.py b/tpch/queries/q22.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(customer_ds: FrameT, orders_ds: FrameT) -> FrameT:
@@ -26,9 +26,9 @@ def query(customer_ds: FrameT, orders_ds: FrameT) -> FrameT:
     )
 
     return (
-        q1.join(q3, left_on="c_custkey", right_on="c_custkey", how="left")
+        q1.join(q3, left_on="c_custkey", right_on="c_custkey", how="left")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("o_custkey").is_null())
-        .join(q2, how="cross")
+        .join(q2, how="cross")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("c_acctbal") > nw.col("avg_acctbal"))
         .group_by("cntrycode")
         .agg(
diff --git a/tpch/queries/q3.py b/tpch/queries/q3.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
diff --git a/tpch/queries/q4.py b/tpch/queries/q4.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(line_item_ds: FrameT, orders_ds: FrameT) -> FrameT:
diff --git a/tpch/queries/q5.py b/tpch/queries/q5.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
diff --git a/tpch/queries/q6.py b/tpch/queries/q6.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(line_item_ds: FrameT) -> FrameT:
diff --git a/tpch/queries/q7.py b/tpch/queries/q7.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -23,22 +23,22 @@ def query(
     var_2 = datetime(1996, 12, 31)
 
     df1 = (
-        customer_ds.join(n1, left_on="c_nationkey", right_on="n_nationkey")
+        customer_ds.join(n1, left_on="c_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .join(orders_ds, left_on="c_custkey", right_on="o_custkey")
         .rename({"n_name": "cust_nation"})
         .join(line_item_ds, left_on="o_orderkey", right_on="l_orderkey")
         .join(supplier_ds, left_on="l_suppkey", right_on="s_suppkey")
-        .join(n2, left_on="s_nationkey", right_on="n_nationkey")
+        .join(n2, left_on="s_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .rename({"n_name": "supp_nation"})
     )
 
     df2 = (
-        customer_ds.join(n2, left_on="c_nationkey", right_on="n_nationkey")
+        customer_ds.join(n2, left_on="c_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .join(orders_ds, left_on="c_custkey", right_on="o_custkey")
         .rename({"n_name": "cust_nation"})
         .join(line_item_ds, left_on="o_orderkey", right_on="l_orderkey")
         .join(supplier_ds, left_on="l_suppkey", right_on="s_suppkey")
-        .join(n1, left_on="s_nationkey", right_on="n_nationkey")
+        .join(n1, left_on="s_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .rename({"n_name": "supp_nation"})
     )
 
diff --git a/tpch/queries/q8.py b/tpch/queries/q8.py
@@ -6,7 +6,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
@@ -32,10 +32,10 @@ def query(
         .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(n1, left_on="c_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .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")
+        .join(n2, left_on="s_nationkey", right_on="n_nationkey")  # pyright: ignore[reportArgumentType]
         .filter(nw.col("o_orderdate").is_between(date1, date2))
         .filter(nw.col("p_type") == type)
         .select(
diff --git a/tpch/queries/q9.py b/tpch/queries/q9.py
@@ -5,7 +5,7 @@
 import narwhals as nw
 
 if TYPE_CHECKING:
-    from narwhals.typing import FrameT
+    from tpch._typing import FrameT
 
 
 def query(
diff --git a/utils/bump_version.py b/utils/bump_version.py
@@ -44,7 +44,7 @@
 
 with open("pyproject.toml", encoding="utf-8") as f:
     content = f.read()
-old_version = re.search(r'version = "(.*)"', content).group(1)
+old_version = re.search(r'version = "(.*)"', content).group(1)  # pyright: ignore[reportOptionalMemberAccess]
 version = old_version.split(".")
 if how == "patch":
     version = ".".join(version[:-1] + [str(int(version[-1]) + 1)])
diff --git a/utils/check_api_reference.py b/utils/check_api_reference.py
diff --git a/utils/check_docstrings.py b/utils/check_docstrings.py
