feat(nodejs): add order_by method to Query (#3123)

nodejs/__test__/query.test.ts

@@ -109,3 +109,209 @@ describe("Query outputSchema", () => {
     expect(schema.fields.length).toBe(3);
   });
 });
+
+describe("Query orderBy", () => {
+  let tmpDir: tmp.DirResult;
+  let table: Table;
+
+  beforeEach(async () => {
+    tmpDir = tmp.dirSync({ unsafeCleanup: true });
+    const db = await connect(tmpDir.name);
+
+    // Create table with numeric data for sorting
+    const schema = new Schema([
+      new Field("id", new Int64(), true),
+      new Field("score", new Float32(), true),
+      new Field("name", new Utf8(), true),
+    ]);
+
+    const data = makeArrowTable(
+      [
+        { id: 1n, score: 3.5, name: "charlie" },
+        { id: 2n, score: 1.2, name: "alice" },
+        { id: 3n, score: 2.8, name: "bob" },
+        { id: 4n, score: 0.5, name: "david" },
+        { id: 5n, score: 4.1, name: "eve" },
+      ],
+      { schema },
+    );
+    table = await db.createTable("test", data);
+  });
+
+  afterEach(() => {
+    tmpDir.removeCallback();
+  });
+
+  it("should sort by single column ascending", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "score", ascending: true, nullsFirst: false })
+      .toArray();
+
+    expect(results.length).toBe(5);
+    // Verify ascending order
+    expect(results[0].score).toBeCloseTo(0.5, 0.001);
+    expect(results[1].score).toBeCloseTo(1.2, 0.001);
+    expect(results[2].score).toBeCloseTo(2.8, 0.001);
+    expect(results[3].score).toBeCloseTo(3.5, 0.001);
+    expect(results[4].score).toBeCloseTo(4.1, 0.001);
+  });
+
+  it("should sort by single column descending", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "score", ascending: false, nullsFirst: false })
+      .toArray();
+
+    expect(results.length).toBe(5);
+    // Verify descending order
+    expect(results[0].score).toBeCloseTo(4.1, 0.001);
+    expect(results[1].score).toBeCloseTo(3.5, 0.001);
+    expect(results[2].score).toBeCloseTo(2.8, 0.001);
+    expect(results[3].score).toBeCloseTo(1.2, 0.001);
+    expect(results[4].score).toBeCloseTo(0.5, 0.001);
+  });
+
+  it("should use ascending as default direction", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "score" })
+      .toArray();
+
+    expect(results.length).toBe(5);
+    // Verify ascending order (default)
+    expect(results[0].score).toBeCloseTo(0.5, 0.001);
+    expect(results[1].score).toBeCloseTo(1.2, 0.001);
+    expect(results[2].score).toBeCloseTo(2.8, 0.001);
+    expect(results[3].score).toBeCloseTo(3.5, 0.001);
+    expect(results[4].score).toBeCloseTo(4.1, 0.001);
+  });
+
+  it("should sort by string column", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "name" })
+      .toArray();
+
+    expect(results.length).toBe(5);
+    // Verify alphabetical order
+    expect(results[0].name).toBe("alice");
+    expect(results[1].name).toBe("bob");
+    expect(results[2].name).toBe("charlie");
+    expect(results[3].name).toBe("david");
+    expect(results[4].name).toBe("eve");
+  });
+
+  it("should support method chaining with where", async () => {
+    const results = await table
+      .query()
+      .where("score > 2.0")
+      .orderBy({ columnName: "score" })
+      .toArray();
+    expect(results.length).toBe(3);
+    // Verify filtered and sorted
+    expect(results[0].score).toBeCloseTo(2.8, 0.001);
+    expect(results[1].score).toBeCloseTo(3.5, 0.001);
+    expect(results[2].score).toBeCloseTo(4.1, 0.001);
+  });
+
+  it("should support method chaining with limit", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "score", ascending: false })
+      .limit(3)
+      .toArray();
+
+    expect(results.length).toBe(3);
+    // Verify top 3 in descending order
+    expect(results[0].score).toBeCloseTo(4.1, 0.001);
+    expect(results[1].score).toBeCloseTo(3.5, 0.001);
+    expect(results[2].score).toBeCloseTo(2.8, 0.001);
+  });
+
+  it("should support method chaining with offset", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "score" })
+      .offset(2)
+      .limit(2)
+      .toArray();
+
+    expect(results.length).toBe(2);
+    // Verify results skip first 2 and take next 2
+    expect(results[0].score).toBeCloseTo(2.8, 0.001);
+    expect(results[1].score).toBeCloseTo(3.5, 0.001);
+  });
+
+  it("should support method chaining with select", async () => {
+    const results = await table
+      .query()
+      .orderBy({ columnName: "name" })
+      .select(["name", "score"])
+      .toArray();
+
+    expect(results.length).toBe(5);
+    // Verify only selected columns are present
+    expect(Object.keys(results[0])).toEqual(["name", "score"]);
+    expect(Object.keys(results[4])).toEqual(["name", "score"]);
+    // Verify sorted by name
+    expect(results[0].name).toBe("alice");
+    expect(results[4].name).toBe("eve");
+  });
+
+  it("should support complex method chaining", async () => {
+    const results = await table
+      .query()
+      .where("score > 1.0")
+      .orderBy({ columnName: "score", ascending: false })
+      .limit(3)
+      .select(["id", "score", "name"])
+      .toArray();
+
+    expect(results.length).toBe(3);
+    // Verify filtered, sorted, limited, and projected
+    expect(results[0].score).toBeCloseTo(4.1, 0.001);
+    expect(results[1].score).toBeCloseTo(3.5, 0.001);
+    expect(results[2].score).toBeCloseTo(2.8, 0.001);
+    expect(Object.keys(results[0])).toEqual(["id", "score", "name"]);
+  });
+
+  it("should support multi-column ordering and null placement", async () => {
+    const schema = new Schema([
+      new Field("group", new Int64(), true),
+      new Field("score", new Float32(), true),
+      new Field("name", new Utf8(), true),
+    ]);
+
+    const data = makeArrowTable(
+      [
+        { group: 1n, score: null, name: "z" },
+        { group: 1n, score: 1.0, name: "b" },
+        { group: 1n, score: 1.0, name: "a" },
+        { group: 2n, score: 0.5, name: "c" },
+      ],
+      { schema },
+    );
+    const nullTable = await (await connect(tmpDir.name)).createTable(
+      "test_multi_order",
+      data,
+      { mode: "overwrite" },
+    );
+
+    const results = await nullTable
+      .query()
+      .orderBy([
+        { columnName: "group", ascending: true, nullsFirst: false },
+        { columnName: "score", ascending: true, nullsFirst: true },
+        { columnName: "name", ascending: true, nullsFirst: false },
+      ])
+      .toArray();
+
+    expect(results.map((r) => [r.group, r.score, r.name])).toEqual([
+      [1n, null, "z"],
+      [1n, 1.0, "a"],
+      [1n, 1.0, "b"],
+      [2n, 0.5, "c"],
+    ]);
+  });
+});

nodejs/examples/filtering.test.ts

@@ -38,5 +38,14 @@ test("filtering examples", async () => {
     // --8<-- [start:sql_search]
     await tbl.query().where("id = 10").limit(10).toArray();
     // --8<-- [end:sql_search]
+
+    // --8<-- [start:orderby_search]
+    await tbl
+      .query()
+      .where("id > 10")
+      .orderBy({ columnName: "id", ascending: false })
+      .limit(5)
+      .toArray();
+    // --8<-- [end:orderby_search]
   });
 });

nodejs/lancedb/index.ts

@@ -82,6 +82,7 @@ export {
   VectorQuery,
   TakeQuery,
   QueryExecutionOptions,
+  ColumnOrdering,
   FullTextSearchOptions,
   RecordBatchIterator,
   FullTextQuery,

nodejs/lancedb/query.ts

@@ -79,6 +79,12 @@ export interface QueryExecutionOptions {
   timeoutMs?: number;
 }
 
+export interface ColumnOrdering {
+  columnName: string;
+  ascending?: boolean;
+  nullsFirst?: boolean;
+}
+
 /**
  * Options that control the behavior of a full text search
  */
@@ -417,6 +423,21 @@ export class StandardQueryBase<
     return this;
   }
 
+  /**
+   * Sort the results by the specified column(s).
+   * @returns This query builder.
+   */
+  orderBy(ordering: ColumnOrdering | ColumnOrdering[]): this {
+    const orderings = Array.isArray(ordering) ? ordering : [ordering];
+    const normalized = orderings.map((o) => ({
+      columnName: o.columnName,
+      ascending: o.ascending ?? true,
+      nullsFirst: o.nullsFirst ?? false,
+    }));
+    this.doCall((inner) => inner.orderBy(normalized));
+    return this;
+  }
+
   /**
    * Skip searching un-indexed data. This can make search faster, but will miss
    * any data that is not yet indexed.

nodejs/src/query.rs

@@ -3,6 +3,12 @@
 
 use std::sync::Arc;
 
+use crate::error::NapiErrorExt;
+use crate::error::convert_error;
+use crate::iterator::RecordBatchIterator;
+use crate::rerankers::RerankHybridCallbackArgs;
+use crate::rerankers::Reranker;
+use crate::util::{parse_distance_type, schema_to_buffer};
 use arrow_array::{
     Array, Float16Array as ArrowFloat16Array, Float32Array as ArrowFloat32Array,
     Float64Array as ArrowFloat64Array, UInt8Array as ArrowUInt8Array,
@@ -19,16 +25,27 @@ use lancedb::query::QueryBase;
 use lancedb::query::QueryExecutionOptions;
 use lancedb::query::Select;
 use lancedb::query::TakeQuery as LanceDbTakeQuery;
-use lancedb::query::VectorQuery as LanceDbVectorQuery;
+use lancedb::query::{ColumnOrdering as LanceDbColumnOrdering, VectorQuery as LanceDbVectorQuery};
 use napi::bindgen_prelude::*;
 use napi_derive::napi;
 
-use crate::error::NapiErrorExt;
-use crate::error::convert_error;
-use crate::iterator::RecordBatchIterator;
-use crate::rerankers::RerankHybridCallbackArgs;
-use crate::rerankers::Reranker;
-use crate::util::{parse_distance_type, schema_to_buffer};
+#[napi(object)]
+pub struct ColumnOrdering {
+    pub ascending: bool,
+    pub nulls_first: bool,
+    pub column_name: String,
+}
+
+impl From<ColumnOrdering> for LanceDbColumnOrdering {
+    fn from(value: ColumnOrdering) -> Self {
+        match (value.ascending, value.nulls_first) {
+            (true, true) => Self::asc_nulls_first(value.column_name),
+            (true, false) => Self::asc_nulls_last(value.column_name),
+            (false, true) => Self::desc_nulls_first(value.column_name),
+            (false, false) => Self::desc_nulls_last(value.column_name),
+        }
+    }
+}
 
 fn bytes_to_arrow_array(data: Uint8Array, dtype: String) -> napi::Result<Arc<dyn Array>> {
     let buf = arrow_buffer::Buffer::from(data.to_vec());
@@ -128,6 +145,18 @@ impl Query {
         self.inner = self.inner.clone().with_row_id();
     }
 
+    #[napi]
+    pub fn order_by(&mut self, ordering: Option<Vec<ColumnOrdering>>) -> napi::Result<()> {
+        let ordering = ordering.map(|ordering| {
+            ordering
+                .into_iter()
+                .map(LanceDbColumnOrdering::from)
+                .collect()
+        });
+        self.inner = self.inner.clone().order_by(ordering);
+        Ok(())
+    }
+
     #[napi(catch_unwind)]
     pub async fn output_schema(&self) -> napi::Result<Buffer> {
         let schema = self.inner.output_schema().await.default_error()?;
@@ -328,6 +357,18 @@ impl VectorQuery {
         Ok(())
     }
 
+    #[napi]
+    pub fn order_by(&mut self, ordering: Option<Vec<ColumnOrdering>>) -> napi::Result<()> {
+        let ordering = ordering.map(|ordering| {
+            ordering
+                .into_iter()
+                .map(LanceDbColumnOrdering::from)
+                .collect()
+        });
+        self.inner = self.inner.clone().order_by(ordering);
+        Ok(())
+    }
+
     #[napi(catch_unwind)]
     pub async fn output_schema(&self) -> napi::Result<Buffer> {
         let schema = self.inner.output_schema().await.default_error()?;