fix: Fix ColumnNotFound error selecting len() after semi/anti join (#…

…21355)
pola-rs · Feb 20, 2025 · 42104e2 · 42104e2
1 parent 53f0bfd
commit 42104e2
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 1 deletion.
diff --git a/crates/polars-plan/src/plans/optimizer/projection_pushdown/semi_anti_join.rs b/crates/polars-plan/src/plans/optimizer/projection_pushdown/semi_anti_join.rs
@@ -19,7 +19,7 @@ pub(super) fn process_semi_anti_join(
     let mut names_left = PlHashSet::with_capacity(n);
     let mut names_right = PlHashSet::with_capacity(n);
 
-    if ctx.acc_projections.is_empty() {
+    if !ctx.has_pushed_down() {
         // Only project the join columns.
         for e in &right_on {
             add_expr_to_accumulated(e.node(), &mut pushdown_right, &mut names_right, expr_arena);

diff --git a/py-polars/tests/unit/operations/test_join.py b/py-polars/tests/unit/operations/test_join.py
@@ -1800,3 +1800,12 @@ def time_func(func: Callable[[], Any]) -> float:
     if runtime_ratio > threshold:
         msg = f"runtime_ratio ({runtime_ratio}) > {threshold}x ({runtime_eager = }, {runtime_lazy = })"
         raise ValueError(msg)
+
+
+def test_select_len_after_semi_anti_join_21343() -> None:
+    lhs = pl.LazyFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
+    rhs = pl.LazyFrame({"a": [1, 2, 3]})
+
+    q = lhs.join(rhs, on="a", how="anti").select(pl.len())
+
+    assert q.collect().item() == 0