fix(rust): get_dtype handles input node schema and CSE

py-polars/src/lazyframe/visit.rs

@@ -125,16 +125,17 @@ impl NodeTraverser {
         let lp_arena = self.lp_arena.lock().unwrap();
         let ir_node = lp_arena.get(self.root);
         let expr_arena = self.expr_arena.lock().unwrap();
-        let schema = {
+        let schema = if let Some(schema) = ir_node.input_schema(&lp_arena) {
             // TODO: This is a hack for CSE expressions when
             // determining the dtype. It should be removed once
             // to_field, or its moral equivalent can handle this in a
             // proper way. The schema needs to include the dtype of
             // CSE expressions for to_field to work with expressions
             // that reference them, but is not part of the public
-            // schema of the node.
-            let schema = ir_node.schema(&lp_arena);
+            // schema of the input.
             match ir_node {
+                // Both select and hstack must augment with any CSE
+                // expressions.
                 IR::Select { expr, .. } | IR::HStack { exprs: expr, .. } => {
                     let cse_exprs = expr.cse_exprs();
                     if cse_exprs.is_empty() {
@@ -153,6 +154,8 @@ impl NodeTraverser {
                 },
                 _ => schema,
             }
+        } else {
+            raise_err!("Not able to compute input schema", ComputeError)
         };
         let field = expr_arena
             .get(expr_node)

py-polars/src/lazyframe/visitor/nodes.rs

@@ -202,6 +202,15 @@ pub struct HStack {
     options: (), // ProjectionOptions,
 }
 
+#[pyclass]
+/// Like Select, but all operations produce a single row.
+pub struct Reduce {
+    #[pyo3(get)]
+    input: usize,
+    #[pyo3(get)]
+    exprs: Vec<PyExprIR>,
+}
+
 #[pyclass]
 /// Remove duplicates from the table
 pub struct Distinct {
@@ -436,11 +445,9 @@ pub(crate) fn into_py(py: Python<'_>, plan: &IR) -> PyResult<PyObject> {
             input,
             exprs,
             schema: _,
-        } => Select {
+        } => Reduce {
             input: input.0,
-            expr: exprs.iter().map(|e| e.into()).collect(),
-            cse_expr: vec![],
-            options: (),
+            exprs: exprs.iter().map(|e| e.into()).collect(),
         }
         .into_py(py),
         IR::Distinct { input, options } => Distinct {

py-polars/src/lib.rs

@@ -75,6 +75,7 @@ fn _ir_nodes(_py: Python, m: &Bound<PyModule>) -> PyResult<()> {
     m.add_class::<GroupBy>().unwrap();
     m.add_class::<Join>().unwrap();
     m.add_class::<HStack>().unwrap();
+    m.add_class::<Reduce>().unwrap();
     m.add_class::<Distinct>().unwrap();
     m.add_class::<MapFunction>().unwrap();
     m.add_class::<Union>().unwrap();