From 9c406862792b218bd55ca18724bb7cb70aa4ff15 Mon Sep 17 00:00:00 2001
From: codewarnab <dp950810@gmail.com>
Date: Mon, 2 Nov 2026 16:59:55 +0000
Subject: [PATCH] docs: add custom TopK operator walkthrough to Extending
 Operators guide

---
 .../library-user-guide/extending-operators.md | 79 ++++++++++++++++++-
 1 file changed, 78 insertions(+), 1 deletion(-)

diff --git a/docs/source/library-user-guide/extending-operators.md b/docs/source/library-user-guide/extending-operators.md
index 0a169531757c2..7a8e1a15bd578 100644
--- a/docs/source/library-user-guide/extending-operators.md
+++ b/docs/source/library-user-guide/extending-operators.md
@@ -19,7 +19,7 @@
 
 # Extending Operators
 
-DataFusion supports extending operators by transforming [`LogicalPlan`] and [`ExecutionPlan`] through customized [optimizer rules](https://docs.rs/datafusion/latest/datafusion/optimizer/trait.OptimizerRule.html). This section will use the µWheel project to illustrate such capabilities.
+DataFusion supports extending operators with custom [`LogicalPlan`] and [`ExecutionPlan`] nodes and [optimizer rules](https://docs.rs/datafusion/latest/datafusion/optimizer/trait.OptimizerRule.html). The examples below show two ways to extend query planning.
 
 [`logicalplan`]: https://docs.rs/datafusion/latest/datafusion/logical_expr/enum.LogicalPlan.html
 [`executionplan`]: https://docs.rs/datafusion/latest/datafusion/physical_plan/trait.ExecutionPlan.html
@@ -60,3 +60,80 @@ fn agg_to_table_scan(result: f64, schema: SchemaRef) -> Result<LogicalPlan> {
 ```
 
 To get a deeper dive into the usage of the µWheel project, visit the [blog post](https://uwheel.rs/post/datafusion_uwheel/) by Max Meldrum.
+
+## A custom TopK operator
+
+The [TopK example] shows how to add a custom operator that finds the rows with
+the highest values. It uses a query like this one to find the three customers
+with the highest revenue:
+
+```sql
+SELECT customer_id, revenue FROM sales ORDER BY revenue DESC LIMIT 3;
+```
+
+DataFusion already has optimized support for TopK queries. This example is for
+learning how to connect a custom operator to DataFusion, not for replacing the
+built-in implementation.
+
+### Describe the operation
+
+The example defines `TopKPlanNode`, a custom logical plan node. A logical plan
+describes the operations in a query. The node implements
+`UserDefinedLogicalNodeCore` so DataFusion can inspect and update it. Its main
+fields are:
+
+```rust,ignore
+struct TopKPlanNode {
+    k: usize,
+    input: LogicalPlan,
+    expr: SortExpr,
+}
+```
+
+- `input` describes where the rows come from, such as the `sales` table.
+- `expr` describes how to rank the rows, such as by revenue in descending order.
+- `k` is the number of rows to keep. In this query, `k` is 3.
+
+### Add the node to the plan
+
+`TopKOptimizerRule` looks for a sort that has a row limit. The limit can be
+part of the sort, or it can be a separate limit directly above the sort. The
+sort must have exactly one sort expression. The rule replaces that part of the
+plan with a `TopKPlanNode`:
+
+```rust,ignore
+LogicalPlan::Extension(Extension {
+    node: Arc::new(TopKPlanNode {
+        k: *fetch,
+        input: input.as_ref().clone(),
+        expr: expr[0].clone(),
+    }),
+})
+```
+
+The rule takes `k` from the row limit. It copies the input and the sort
+expression from the sort. `LogicalPlan::Extension` wraps the custom node so
+that it can be part of a logical plan.
+
+For the query above, the new plan includes:
+
+```text
+TopK: k=3
+  TableScan: sales
+```
+
+The rule is registered with the session so DataFusion can apply it while
+optimizing the query.
+
+### Execute the operation
+
+`TopKPlanner` converts the logical `TopKPlanNode` into `TopKExec`, a physical
+operator. `TopKExec` reads its input and uses `TopKReader` to keep the highest
+values. Once it has read the input, the reader returns the retained rows.
+
+The example is specialized for its test data. It expects particular columns
+and types, and its map stores only one customer per revenue value. It does not
+handle all sort options or other edge cases, so it should not be used as a
+general-purpose TopK operator.
+
+[topk example]: https://github.com/apache/datafusion/blob/main/datafusion/core/tests/user_defined/user_defined_plan.rs