From f3257f5b5ccf981dd1dba1c17cc411a24049cce6 Mon Sep 17 00:00:00 2001
From: Ben Chambers <ben.chambers@datastax.com>
Date: Fri, 2 Jun 2023 11:26:21 -0700
Subject: [PATCH] ref: create an Exprs

---
 crates/sparrow-physical/src/expr.rs | 29 +++++++++++++++++++++++++++++
 crates/sparrow-physical/src/step.rs | 22 +++++++++++-----------
 2 files changed, 40 insertions(+), 11 deletions(-)

diff --git a/crates/sparrow-physical/src/expr.rs b/crates/sparrow-physical/src/expr.rs
index bb1592571..1e6a8c6f2 100644
--- a/crates/sparrow-physical/src/expr.rs
+++ b/crates/sparrow-physical/src/expr.rs
@@ -2,6 +2,35 @@ use std::borrow::Cow;
 
 use arrow_schema::DataType;
 
+/// Represents 1 or more values computed by expressions.
+#[derive(Debug, serde::Serialize, serde::Deserialize)]
+pub struct Exprs {
+    /// The expressions computing the intermediate values.
+    pub exprs: Vec<Expr>,
+    /// The indices of columns to output.
+    pub outputs: Vec<ExprId>,
+}
+
+impl Exprs {
+    /// Create expressions computing the value of the last expression.
+    pub fn singleton(exprs: Vec<Expr>) -> Self {
+        let output = exprs.len() - 1;
+        Self {
+            exprs,
+            outputs: vec![output.into()],
+        }
+    }
+
+    pub fn is_singleton(&self) -> bool {
+        self.outputs.len() == 1
+    }
+
+    /// Return the number of outputs produced by these expressions.
+    pub fn output_len(&self) -> usize {
+        self.outputs.len()
+    }
+}
+
 /// The identifier (index) of an expression.
 #[derive(Debug, serde::Serialize, serde::Deserialize, PartialEq, Eq, PartialOrd, Ord, Hash)]
 #[repr(transparent)]
diff --git a/crates/sparrow-physical/src/step.rs b/crates/sparrow-physical/src/step.rs
index d240644d9..233acbb4e 100644
--- a/crates/sparrow-physical/src/step.rs
+++ b/crates/sparrow-physical/src/step.rs
@@ -1,5 +1,7 @@
 use arrow_schema::SchemaRef;
 
+use crate::Exprs;
+
 /// The identifier (index) of a step.
 
 #[derive(Debug, serde::Serialize, serde::Deserialize, PartialEq, Eq, PartialOrd, Ord, Hash)]
@@ -41,27 +43,25 @@ pub enum StepKind {
     /// The output includes the same rows as the input, but with columns
     /// projected as configured.
     Project {
-        /// Expressions to apply to compute additional input columns.
-        exprs: Vec<crate::Expr>,
-        /// Indices of expressions to use for the output.
+        /// Expressions to compute the projection.
         ///
-        /// The length should be the same as the number of fields in the schema.
-        outputs: Vec<usize>,
+        /// The length of the outputs should be the same as the fields in the schema.
+        exprs: Exprs,
     },
     /// Filter the results based on a boolean predicate.
     Filter {
         /// Expressions to apply to compute the predicate.
         ///
-        /// The last expression should be the boolean predicate.
-        exprs: Vec<crate::Expr>,
+        /// There should be a single output producing a boolean value.
+        exprs: Exprs,
     },
     /// A step that repartitions the output.
     Repartition {
         num_partitions: usize,
-        /// Expressions to apply to compute columns which may be referenced by `keys`.
-        exprs: Vec<crate::Expr>,
-        /// Indices of expression columns representing the keys.
-        keys: Vec<usize>,
+        /// Expressions to compute the keys.
+        ///
+        /// Each output corresponds to a part of the key.
+        keys: Exprs,
     },
     Error,
 }