feat(rust, python): is_first also supports numeric list type.

crates/polars-ops/Cargo.toml

@@ -75,6 +75,7 @@ string_from_radix = ["polars-core/strings"]
 extract_jsonpath = ["serde_json", "jsonpath_lib", "polars-json"]
 log = []
 hash = []
+group_by_list = ["polars-core/group_by_list"]
 rolling_window = ["polars-core/rolling_window"]
 moment = ["polars-core/moment"]
 search_sorted = []

crates/polars-ops/src/series/ops/is_first.rs

@@ -64,6 +64,21 @@ fn is_first_struct(s: &Series) -> PolarsResult<BooleanChunked> {
     Ok(BooleanChunked::with_chunk(s.name(), arr))
 }
 
+fn is_first_list(ca: &ListChunked) -> PolarsResult<BooleanChunked> {
+    let groups = ca.group_tuples(true, false)?;
+    let first = groups.take_group_firsts();
+    let mut out = MutableBitmap::with_capacity(ca.len());
+    out.extend_constant(ca.len(), false);
+
+    for idx in first {
+        // Group tuples are always in bounds
+        unsafe { out.set_unchecked(idx as usize, true) }
+    }
+
+    let arr = BooleanArray::new(ArrowDataType::Boolean, out.into(), None);
+    Ok(BooleanChunked::with_chunk(ca.name(), arr))
+}
+
 pub fn is_first(s: &Series) -> PolarsResult<BooleanChunked> {
     let s = s.to_physical_repr();
 
@@ -97,6 +112,11 @@ pub fn is_first(s: &Series) -> PolarsResult<BooleanChunked> {
         },
         #[cfg(feature = "dtype-struct")]
         Struct(_) => return is_first_struct(&s),
+        #[cfg(feature = "group_by_list")]
+        List(inner) if inner.is_numeric() => {
+            let ca = s.list().unwrap();
+            return is_first_list(ca);
+        },
         dt => polars_bail!(opq = is_first, dt),
     };
     Ok(out)

crates/polars/Cargo.toml

@@ -130,7 +130,7 @@ extract_jsonpath = [
 ]
 string_encoding = ["polars-ops/string_encoding", "polars-core/strings"]
 binary_encoding = ["polars-ops/binary_encoding"]
-group_by_list = ["polars-core/group_by_list"]
+group_by_list = ["polars-core/group_by_list", "polars-ops/group_by_list"]
 lazy_regex = ["polars-lazy/regex"]
 cum_agg = ["polars-core/cum_agg", "polars-core/cum_agg"]
 rolling_window = ["polars-core/rolling_window", "polars-lazy/rolling_window", "polars-time/rolling_window"]

py-polars/tests/unit/test_lazy.py

@@ -296,6 +296,11 @@ def test_is_first() -> None:
         "a": [True, True, True, False, True, False, False]
     }
 
+    ldf = pl.LazyFrame({"a": [[1, 2], [3], [1, 2], [4, 5], [4, 5]]})
+    assert ldf.select(pl.col("a").is_first()).collect().to_dict(False) == {
+        "a": [True, True, False, True, False]
+    }
+
 
 def test_is_duplicated() -> None:
     ldf = pl.LazyFrame({"a": [4, 1, 4]}).select(pl.col("a").is_duplicated())