fix(nodejs): workaround for apache-arrow null vector issue (#2244)

Fixes #2240

nodejs/__test__/table.test.ts

@@ -279,6 +279,15 @@ describe.each([arrow15, arrow16, arrow17, arrow18])(
       expect(res.getChild("y")?.toJSON()).toEqual([2, null, null, null]);
       expect(res.getChild("z")?.toJSON()).toEqual([null, null, 3n, 5n]);
     });
+
+    it("should handle null vectors at end of data", async () => {
+      // https://github.com/lancedb/lancedb/issues/2240
+      const data = [{ vector: [1, 2, 3] }, { vector: null }];
+      const db = await connect("memory://");
+
+      const table = await db.createTable("my_table", data);
+      expect(await table.countRows()).toEqual(2);
+    });
   },
 );
 

nodejs/lancedb/arrow.ts

@@ -37,10 +37,10 @@ import {
   Utf8,
   Vector,
   makeVector as arrowMakeVector,
+  vectorFromArray as badVectorFromArray,
   makeBuilder,
   makeData,
   makeTable,
-  vectorFromArray,
 } from "apache-arrow";
 import { Buffers } from "apache-arrow/data";
 import { type EmbeddingFunction } from "./embedding/embedding_function";
@@ -186,6 +186,21 @@ export class VectorColumnOptions {
   }
 }
 
+// biome-ignore lint/suspicious/noExplicitAny: skip
+function vectorFromArray(data: any, type?: DataType) {
+  // Workaround for: https://github.com/apache/arrow/issues/45862
+  // If FSL type with float
+  if (DataType.isFixedSizeList(type) && DataType.isFloat(type.valueType)) {
+    const extendedData = [...data, new Array(type.listSize).fill(0.0)];
+    const array = badVectorFromArray(extendedData, type);
+    return array.slice(0, data.length);
+  } else if (type === undefined) {
+    return badVectorFromArray(data);
+  } else {
+    return badVectorFromArray(data, type);
+  }
+}
+
 /** Options to control the makeArrowTable call. */
 export class MakeArrowTableOptions {
   /*