From b17d439d6d7fc5868cb91ab59fe92d3d642c4513 Mon Sep 17 00:00:00 2001
From: LAWG <101877402+LawrenceGB@users.noreply.github.com>
Date: Thu, 3 Oct 2024 19:52:26 +0100
Subject: [PATCH] fix: ensure `id_` is correctly passed during creation (#1282)
Co-authored-by: lawrencegb <lawrence@3api.com>
Co-authored-by: Alex Yang <himself65@outlook.com>
---
.changeset/forty-swans-tan.md | 5 +++++
packages/core/src/schema/node.ts | 4 ++--
packages/llamaindex/tests/indices/VectorStoreIndex.test.ts | 4 ++--
3 files changed, 9 insertions(+), 4 deletions(-)
create mode 100644 .changeset/forty-swans-tan.md
diff --git a/.changeset/forty-swans-tan.md b/.changeset/forty-swans-tan.md
new file mode 100644
index 000000000..4b8449a40
--- /dev/null
+++ b/.changeset/forty-swans-tan.md
@@ -0,0 +1,5 @@
+---
+"@llamaindex/core": patch
+---
+
+Fix #1278: resolved issue where the id\_ was not correctly passed as the id when creating a TextNode. As a result, the upsert operation to the vector database was using a generated ID instead of the provided document ID, if available.
diff --git a/packages/core/src/schema/node.ts b/packages/core/src/schema/node.ts
index 459dbfb66..a79226db1 100644
--- a/packages/core/src/schema/node.ts
+++ b/packages/core/src/schema/node.ts
@@ -479,7 +479,7 @@ export function buildNodeFromSplits(
) {
const imageDoc = doc as ImageNode;
const imageNode = new ImageNode({
- id_: idGenerator(i, imageDoc),
+ id_: imageDoc.id_ ?? idGenerator(i, imageDoc),
text: textChunk,
image: imageDoc.image,
embedding: imageDoc.embedding,
@@ -496,7 +496,7 @@ export function buildNodeFromSplits(
) {
const textDoc = doc as TextNode;
const node = new TextNode({
- id_: idGenerator(i, textDoc),
+ id_: textDoc.id_ ?? idGenerator(i, textDoc),
text: textChunk,
embedding: textDoc.embedding,
excludedEmbedMetadataKeys: [...textDoc.excludedEmbedMetadataKeys],
diff --git a/packages/llamaindex/tests/indices/VectorStoreIndex.test.ts b/packages/llamaindex/tests/indices/VectorStoreIndex.test.ts
index 9bca51af7..a807f8bed 100644
--- a/packages/llamaindex/tests/indices/VectorStoreIndex.test.ts
+++ b/packages/llamaindex/tests/indices/VectorStoreIndex.test.ts
@@ -27,7 +27,7 @@ describe("VectorStoreIndex", () => {
runs: number = 2,
): Promise<Array<number>> => {
const documents = [new Document({ text: "lorem ipsem", id_: "1" })];
- const entries = [];
+ const entries: number[] = [];
for (let i = 0; i < runs; i++) {
await VectorStoreIndex.fromDocuments(documents, {
serviceContext,
@@ -43,7 +43,7 @@ describe("VectorStoreIndex", () => {
test("fromDocuments stores duplicates without a doc store strategy", async () => {
const entries = await testStrategy(DocStoreStrategy.NONE);
- expect(entries[0]! + 1).toBe(entries[1]);
+ expect(entries[0]).toBe(entries[1]);
});
test("fromDocuments ignores duplicates with upserts doc store strategy", async () => {
--
GitLab