fix: Check validity of columns of keys/aggs in dsl->ir

crates/polars-plan/src/plans/conversion/dsl_to_ir.rs

@@ -28,6 +28,30 @@ fn empty_df() -> IR {
     }
 }
 
+fn validate_expression(
+    node: Node,
+    expr_arena: &Arena<AExpr>,
+    input_schema: &Schema,
+    operation_name: &str,
+) -> PolarsResult<()> {
+    let iter = aexpr_to_leaf_names_iter(node, expr_arena);
+    validate_columns_in_input(iter, input_schema, operation_name)
+}
+
+fn validate_expressions<N: Into<Node>, I: IntoIterator<Item = N>>(
+    nodes: I,
+    expr_arena: &Arena<AExpr>,
+    input_schema: &Schema,
+    operation_name: &str,
+) -> PolarsResult<()> {
+    let nodes = nodes.into_iter();
+
+    for node in nodes {
+        validate_expression(node.into(), expr_arena, input_schema, operation_name)?
+    }
+    Ok(())
+}
+
 macro_rules! failed_here {
     ($($t:tt)*) => {
         format!("'{}'", stringify!($($t)*)).into()
@@ -610,7 +634,7 @@ pub fn to_alp_impl(lp: DslPlan, ctxt: &mut DslConversionContext) -> PolarsResult
                     allow_empty,
                 } => {
                     let columns = expand_selectors(columns, &input_schema, &[])?;
-                    validate_columns_in_input(&columns, &input_schema, "explode")?;
+                    validate_columns_in_input(columns.as_ref(), &input_schema, "explode")?;
                     polars_ensure!(!columns.is_empty() || allow_empty, InvalidOperation: "no columns provided in explode");
                     if columns.is_empty() {
                         return Ok(input);
@@ -967,8 +991,10 @@ fn resolve_group_by(
             polars_ensure!(names.insert(name.clone()), duplicate = name)
         }
     }
-    let aggs = to_expr_irs(aggs, expr_arena)?;
     let keys = to_expr_irs(keys, expr_arena)?;
+    let aggs = to_expr_irs(aggs, expr_arena)?;
+    validate_expressions(&keys, expr_arena, current_schema, "group by")?;
+    validate_expressions(&aggs, expr_arena, current_schema, "group by")?;
 
     Ok((keys, aggs, Arc::new(schema)))
 }

crates/polars-plan/src/plans/functions/dsl.rs

@@ -81,11 +81,12 @@ pub enum StatsFunction {
     Max,
 }
 
-pub(crate) fn validate_columns_in_input<S: AsRef<str>>(
-    columns: &[S],
+pub(crate) fn validate_columns_in_input<S: AsRef<str>, I: IntoIterator<Item = S>>(
+    columns: I,
     input_schema: &Schema,
     operation_name: &str,
 ) -> PolarsResult<()> {
+    let columns = columns.into_iter();
     for c in columns {
         polars_ensure!(input_schema.contains(c.as_ref()), ColumnNotFound: "'{}' on column: '{}' is invalid\n\nSchema at this point: {:?}", operation_name, c.as_ref(), input_schema)
     }

py-polars/tests/unit/test_errors.py

@@ -698,3 +698,13 @@ def test_no_panic_pandas_nat() -> None:
 def test_list_to_struct_invalid_type() -> None:
     with pytest.raises(pl.exceptions.SchemaError):
         pl.DataFrame({"a": 1}).select(pl.col("a").list.to_struct())
+
+
+def test_raise_invalid_agg() -> None:
+    with pytest.raises(pl.exceptions.ColumnNotFoundError):
+        (
+            pl.LazyFrame({"foo": [1]})
+            .with_row_index()
+            .group_by("index")
+            .agg(pl.col("foo").filter(pl.col("i_do_not_exist")))
+        ).collect()