unique

narwhals/dataframe.py

@@ -110,6 +110,9 @@ def head(self, n: int) -> Self:
     def drop(self, *columns: str | Iterable[str]) -> Self:
         return self._from_dataframe(self._dataframe.drop(*columns))
 
+    def unique(self, subset: str | list[str]) -> Self:
+        return self._from_dataframe(self._dataframe.unique(subset=subset))
+
     def filter(self, *predicates: IntoExpr | Iterable[IntoExpr]) -> Self:
         predicates, _ = self._flatten_and_extract(*predicates)
         return self._from_dataframe(

narwhals/pandas_like/dataframe.py

@@ -197,7 +197,8 @@ def join(
     def head(self, n: int) -> Self:
         return self._from_dataframe(self._dataframe.head(n))
 
-    def unique(self, subset: list[str]) -> Self:
+    def unique(self, subset: str | list[str]) -> Self:
+        subset = flatten_str(subset)
         return self._from_dataframe(self._dataframe.drop_duplicates(subset=subset))
 
     # --- lazy-only ---

tests/test_common.py

@@ -349,3 +349,11 @@ def test_head(df_raw: Any) -> None:
     result = nw.to_native(df.head(2))
     expected = {"a": [1, 3], "b": [4, 4], "z": [7.0, 8.0]}
     compare_dicts(result, expected)
+
+
+@pytest.mark.parametrize("df_raw", [df_pandas, df_lazy])
+def test_unique(df_raw: Any) -> None:
+    df = nw.LazyFrame(df_raw)
+    result = nw.to_native(df.unique("b").sort("b"))
+    expected = {"a": [1, 2], "b": [4, 6], "z": [7.0, 9.0]}
+    compare_dicts(result, expected)

tpch/q4.py

@@ -4,7 +4,7 @@
 
 import polars
 
-from narwhals import translate_frame
+import narwhals as nw
 
 Q_NUM = 4
 
@@ -16,28 +16,25 @@ def q4(
     var_1 = datetime(1993, 7, 1)
     var_2 = datetime(1993, 10, 1)
 
-    line_item_ds, pl = translate_frame(lineitem_ds_raw, is_lazy=True)
-    orders_ds, _ = translate_frame(orders_ds_raw, is_lazy=True)
+    line_item_ds = nw.LazyFrame(lineitem_ds_raw)
+    orders_ds = nw.LazyFrame(orders_ds_raw)
 
     result = (
         line_item_ds.join(orders_ds, left_on="l_orderkey", right_on="o_orderkey")
-        .filter(pl.col("o_orderdate").is_between(var_1, var_2, closed="left"))
-        .filter(pl.col("l_commitdate") < pl.col("l_receiptdate"))
+        .filter(nw.col("o_orderdate").is_between(var_1, var_2, closed="left"))
+        .filter(nw.col("l_commitdate") < nw.col("l_receiptdate"))
         .unique(subset=["o_orderpriority", "l_orderkey"])
         .group_by("o_orderpriority")
-        .agg(pl.len().alias("order_count"))
+        .agg(nw.len().alias("order_count"))
         .sort(by="o_orderpriority")
-        .with_columns(
-            pl.col("order_count")
-            # .cast(pl.datatypes.Int64)
-        )
+        .with_columns(nw.col("order_count").cast(nw.Int64))
     )
 
-    return result.collect().to_native()
+    return nw.to_native(result.collect())
 
 
-lineitem_ds = polars.scan_parquet("../tpch-data/lineitem.parquet")
-orders_ds = polars.scan_parquet("../tpch-data/orders.parquet")
+lineitem_ds = polars.scan_parquet("../tpch-data/s1/lineitem.parquet")
+orders_ds = polars.scan_parquet("../tpch-data/s1/orders.parquet")
 print(
     q4(
         lineitem_ds.collect().to_pandas(),