From f629387203a3f2864b2bdfc1978431cd6b7656c2 Mon Sep 17 00:00:00 2001
From: Erik Grinaker <erik@grinaker.org>
Date: Tue, 16 Jul 2024 22:22:48 +0200
Subject: [PATCH] sql: retain original columns in `Nothing` nodes

---
 src/sql/execution/execute.rs                |  2 +-
 src/sql/planner/optimizer.rs                | 46 ++++++++++++---------
 src/sql/planner/plan.rs                     | 20 +++++----
 src/sql/testscripts/optimizer/short_circuit | 31 +++++++++-----
 src/sql/testscripts/queries/select          |  6 +++
 5 files changed, 66 insertions(+), 39 deletions(-)

diff --git a/src/sql/execution/execute.rs b/src/sql/execution/execute.rs
index d1a428e27..493ea29e0 100644
--- a/src/sql/execution/execute.rs
+++ b/src/sql/execution/execute.rs
@@ -96,7 +96,7 @@ pub fn execute(node: Node, txn: &impl Transaction) -> Result<Rows> {
             join::nested_loop(left, right, right_size, predicate, outer)
         }
 
-        Node::Nothing => Ok(source::nothing()),
+        Node::Nothing { .. } => Ok(source::nothing()),
 
         Node::Offset { source, offset } => {
             let source = execute(*source, txn)?;
diff --git a/src/sql/planner/optimizer.rs b/src/sql/planner/optimizer.rs
index aa9c93c96..6b6f37cdb 100644
--- a/src/sql/planner/optimizer.rs
+++ b/src/sql/planner/optimizer.rs
@@ -287,6 +287,12 @@ pub(super) fn short_circuit(node: Node) -> Result<Node> {
     use Expression::Constant;
     use Value::{Boolean, Null};
 
+    /// Creates a Nothing node with the columns of the original node.
+    fn nothing(node: &Node) -> Node {
+        let columns = (0..node.size()).map(|i| node.column_label(i)).collect();
+        Node::Nothing { columns }
+    }
+
     let transform = |node| match node {
         // Filter nodes that always yield true are unnecessary: remove them.
         Node::Filter { source, predicate: Constant(Boolean(true)) } => *source,
@@ -300,32 +306,34 @@ pub(super) fn short_circuit(node: Node) -> Result<Node> {
         }
 
         // Short-circuit nodes that can't produce anything by replacing them
-        // with a Nothing node.
-        Node::Filter { predicate: Constant(Boolean(false) | Null), .. } => Node::Nothing,
-        Node::IndexLookup { values, .. } if values.is_empty() => Node::Nothing,
-        Node::KeyLookup { keys, .. } if keys.is_empty() => Node::Nothing,
-        Node::Limit { limit: 0, .. } => Node::Nothing,
-        Node::NestedLoopJoin { predicate: Some(Constant(Boolean(false) | Null)), .. } => {
-            Node::Nothing
+        // with a Nothing node, retaining the columns.
+        ref node @ Node::Filter { predicate: Constant(Boolean(false) | Null), .. } => nothing(node),
+        ref node @ Node::IndexLookup { ref values, .. } if values.is_empty() => nothing(node),
+        ref node @ Node::KeyLookup { ref keys, .. } if keys.is_empty() => nothing(node),
+        ref node @ Node::Limit { limit: 0, .. } => nothing(node),
+        ref node @ Node::NestedLoopJoin {
+            predicate: Some(Constant(Boolean(false) | Null)), ..
+        } => nothing(node),
+        ref node @ Node::Scan { filter: Some(Constant(Boolean(false) | Null)), .. } => {
+            nothing(node)
         }
-        Node::Scan { filter: Some(Constant(Boolean(false) | Null)), .. } => Node::Nothing,
-        Node::Values { rows, .. } if rows.is_empty() => Node::Nothing,
+        Node::Values { rows } if rows.is_empty() => Node::Nothing { columns: vec![] },
 
         // Short-circuit nodes that pull from a Nothing node.
         //
         // NB: does not short-circuit aggregation, since an aggregation over 0
         // rows should produce a result.
-        Node::Filter { source, .. }
-        | Node::HashJoin { left: source, .. }
-        | Node::HashJoin { right: source, .. }
-        | Node::NestedLoopJoin { left: source, .. }
-        | Node::NestedLoopJoin { right: source, .. }
-        | Node::Offset { source, .. }
-        | Node::Order { source, .. }
-        | Node::Projection { source, .. }
-            if *source == Node::Nothing =>
+        ref node @ (Node::Filter { ref source, .. }
+        | Node::HashJoin { left: ref source, .. }
+        | Node::HashJoin { right: ref source, .. }
+        | Node::NestedLoopJoin { left: ref source, .. }
+        | Node::NestedLoopJoin { right: ref source, .. }
+        | Node::Offset { ref source, .. }
+        | Node::Order { ref source, .. }
+        | Node::Projection { ref source, .. })
+            if matches!(**source, Node::Nothing { .. }) =>
         {
-            Node::Nothing
+            nothing(node)
         }
 
         // Remove noop projections that simply pass through the source columns.
diff --git a/src/sql/planner/plan.rs b/src/sql/planner/plan.rs
index 7dbce779c..9849994ba 100644
--- a/src/sql/planner/plan.rs
+++ b/src/sql/planner/plan.rs
@@ -109,8 +109,9 @@ pub enum Node {
     /// When outer is true (e.g. LEFT JOIN), a left row without a right match is
     /// emitted anyway, with NULLs for the right row.
     NestedLoopJoin { left: Box<Node>, right: Box<Node>, predicate: Option<Expression>, outer: bool },
-    /// Nothing does not emit anything.
-    Nothing,
+    /// Nothing does not emit anything, but retains the column names of any
+    /// replaced nodes for results and plan expression display.
+    Nothing { columns: Vec<Label> },
     /// Discards the first offset rows from source, emits the rest.
     Offset { source: Box<Node>, offset: usize },
     /// Sorts the source rows by the given expression/direction pairs. Buffers
@@ -174,7 +175,7 @@ impl Node {
 
             node @ (Self::IndexLookup { .. }
             | Self::KeyLookup { .. }
-            | Self::Nothing
+            | Self::Nothing { .. }
             | Self::Scan { .. }
             | Self::Values { .. }) => node,
         };
@@ -232,7 +233,7 @@ impl Node {
             | Self::KeyLookup { .. }
             | Self::Limit { .. }
             | Self::NestedLoopJoin { predicate: None, .. }
-            | Self::Nothing
+            | Self::Nothing { .. }
             | Self::Offset { .. }
             | Self::Scan { filter: None, .. }
             | Self::Remap { .. }) => node,
@@ -262,7 +263,7 @@ impl Node {
             // Some can't have names.
             Self::HashJoin { .. }
             | Self::NestedLoopJoin { .. }
-            | Self::Nothing
+            | Self::Nothing { .. }
             | Self::Values { .. } => None,
         }
     }
@@ -314,8 +315,11 @@ impl Node {
             | Self::Offset { source, .. }
             | Self::Order { source, .. } => source.column_label(index),
 
+            // Nothing nodes contain the original columns of replaced nodes.
+            Self::Nothing { columns } => columns.get(index).cloned().unwrap_or(Label::None),
+
             // And some don't have any names at all.
-            Self::Nothing | Self::Values { .. } => Label::None,
+            Self::Values { .. } => Label::None,
         }
     }
 
@@ -346,7 +350,7 @@ impl Node {
             | Self::Order { source, .. } => source.size(),
 
             // And some are trivial.
-            Self::Nothing => 0,
+            Self::Nothing { columns } => columns.len(),
             Self::Values { rows } => rows.first().map(|row| row.len()).unwrap_or(0),
         }
     }
@@ -479,7 +483,7 @@ impl Node {
                 left.format(f, prefix.clone(), false, false)?;
                 right.format(f, prefix, false, true)?;
             }
-            Self::Nothing => {
+            Self::Nothing { .. } => {
                 write!(f, "Nothing")?;
             }
             Self::Offset { source, offset } => {
diff --git a/src/sql/testscripts/optimizer/short_circuit b/src/sql/testscripts/optimizer/short_circuit
index 14f093955..5032e244c 100644
--- a/src/sql/testscripts/optimizer/short_circuit
+++ b/src/sql/testscripts/optimizer/short_circuit
@@ -56,8 +56,8 @@ Short circuit:
 3, c, 2, 2
 3, c, 3, 3
 
-# FALSE predicates → Nothing
-[opt]> SELECT * FROM test WHERE FALSE
+# FALSE predicates → Nothing (but retains column headers)
+[opt,header]> SELECT * FROM test WHERE FALSE
 ---
 Initial:
    Filter: FALSE
@@ -66,8 +66,9 @@ Filter pushdown:
    Scan: test (FALSE)
 Short circuit:
    Nothing
+test.id, test.value
 
-[opt]> SELECT 1, 2, 3 WHERE FALSE
+[opt,header]> SELECT 1, 2, 3 WHERE FALSE
 ---
 Initial:
    Projection: 1, 2, 3
@@ -75,8 +76,9 @@ Initial:
       └─ Values: blank row
 Short circuit:
    Nothing
+, , 
 
-[opt]> SELECT * FROM test JOIN ref ON ref.test_id = test.id AND FALSE
+[opt,header]> SELECT * FROM test JOIN ref ON ref.test_id = test.id AND FALSE
 ---
 Initial:
    NestedLoopJoin: inner on ref.test_id = test.id AND FALSE
@@ -92,9 +94,10 @@ Filter pushdown:
    └─ Scan: ref (FALSE)
 Short circuit:
    Nothing
+test.id, test.value, ref.id, ref.test_id
 
 # NULL predicates → Nothing
-[opt]> SELECT * FROM test WHERE NULL
+[opt,header]> SELECT * FROM test WHERE NULL
 ---
 Initial:
    Filter: NULL
@@ -103,8 +106,9 @@ Filter pushdown:
    Scan: test (NULL)
 Short circuit:
    Nothing
+test.id, test.value
 
-[opt]> SELECT 1, 2, 3 WHERE NULL
+[opt,header]> SELECT 1, 2, 3 WHERE NULL
 ---
 Initial:
    Projection: 1, 2, 3
@@ -112,8 +116,9 @@ Initial:
       └─ Values: blank row
 Short circuit:
    Nothing
+, , 
 
-[opt]> SELECT * FROM test JOIN ref ON ref.test_id = test.id AND NULL
+[opt,header]> SELECT * FROM test JOIN ref ON ref.test_id = test.id AND NULL
 ---
 Initial:
    NestedLoopJoin: inner on ref.test_id = test.id AND NULL
@@ -129,9 +134,10 @@ Join type:
    └─ Scan: ref (NULL)
 Short circuit:
    Nothing
+test.id, test.value, ref.id, ref.test_id
 
-# Empty key/index lookups.
-[opt]> SELECT * FROM test WHERE id = NULL
+# Empty key/index lookups → Nothing
+[opt,header]> SELECT * FROM test WHERE id = NULL
 ---
 Initial:
    Filter: test.id = NULL
@@ -142,8 +148,9 @@ Index lookup:
    KeyLookup: test (0 keys)
 Short circuit:
    Nothing
+test.id, test.value
 
-[opt]> SELECT * FROM ref WHERE test_id = NULL
+[opt,header]> SELECT * FROM ref WHERE test_id = NULL
 ---
 Initial:
    Filter: ref.test_id = NULL
@@ -154,15 +161,17 @@ Index lookup:
    IndexLookup: ref.test_id (0 values)
 Short circuit:
    Nothing
+ref.id, ref.test_id
 
 # LIMIT 0 → Nothing
-[opt]> SELECT * FROM test LIMIT 0
+[opt,header]> SELECT * FROM test LIMIT 0
 ---
 Initial:
    Limit: 0
    └─ Scan: test
 Short circuit:
    Nothing
+test.id, test.value
 
 # Remove projections that simply pass through source columns. Aliased
 # column names are retained.
diff --git a/src/sql/testscripts/queries/select b/src/sql/testscripts/queries/select
index 80f2e85e1..a45b3ad11 100644
--- a/src/sql/testscripts/queries/select
+++ b/src/sql/testscripts/queries/select
@@ -200,3 +200,9 @@ NULL, b
 2, 2
 3, 1
 3, 2
+
+# A select with no rows optimized to a Nothing node still emits headers.
+[plan,header]> SELECT * FROM test WHERE FALSE
+---
+Nothing
+test.id, test.bool, test.float, test.int, test.string