From 890743ebd019b517c96d5b8757a5252a38aac686 Mon Sep 17 00:00:00 2001
From: Phil Kuang <pkuang@factset.com>
Date: Wed, 25 Sep 2024 11:25:19 -0400
Subject: [PATCH] fix(rag-ai): properly clean up old techdoc vectors

---
 .changeset/sweet-elephants-glow.md                  |  5 +++++
 .../indexing/DefaultVectorAugmentationIndexer.ts    | 13 ++++++++-----
 2 files changed, 13 insertions(+), 5 deletions(-)
 create mode 100644 .changeset/sweet-elephants-glow.md

diff --git a/.changeset/sweet-elephants-glow.md b/.changeset/sweet-elephants-glow.md
new file mode 100644
index 000000000..82ced562f
--- /dev/null
+++ b/.changeset/sweet-elephants-glow.md
@@ -0,0 +1,5 @@
+---
+'@roadiehq/rag-ai-backend-retrieval-augmenter': patch
+---
+
+Fix issue with cleaning up old TechDocs vectors
diff --git a/plugins/backend/rag-ai-backend-retrieval-augmenter/src/indexing/DefaultVectorAugmentationIndexer.ts b/plugins/backend/rag-ai-backend-retrieval-augmenter/src/indexing/DefaultVectorAugmentationIndexer.ts
index f3d7ed518..f805b346e 100644
--- a/plugins/backend/rag-ai-backend-retrieval-augmenter/src/indexing/DefaultVectorAugmentationIndexer.ts
+++ b/plugins/backend/rag-ai-backend-retrieval-augmenter/src/indexing/DefaultVectorAugmentationIndexer.ts
@@ -37,6 +37,10 @@ import {
 } from '@backstage/backend-plugin-api';
 import pLimit from 'p-limit';
 
+const TECHDOCS_ENTITY_FILTER = {
+  'metadata.annotations.backstage.io/techdocs-ref': CATALOG_FILTER_EXISTS,
+};
+
 export class DefaultVectorAugmentationIndexer implements AugmentationIndexer {
   private readonly _vectorStore: RoadieVectorStore;
   private readonly catalogApi: CatalogApi;
@@ -184,10 +188,7 @@ export class DefaultVectorAugmentationIndexer implements AugmentationIndexer {
 
         const entitiesResponse = await this.catalogApi.getEntities(
           {
-            filter: {
-              'metadata.annotations.backstage.io/techdocs-ref':
-                CATALOG_FILTER_EXISTS,
-            },
+            filter: TECHDOCS_ENTITY_FILTER,
           },
           { token },
         );
@@ -273,8 +274,10 @@ export class DefaultVectorAugmentationIndexer implements AugmentationIndexer {
       targetPluginId: 'catalog',
     });
 
+    const entityFilter =
+      source === 'tech-docs' ? TECHDOCS_ENTITY_FILTER : filter;
     const entities = (
-      await this.catalogApi.getEntities({ filter }, { token })
+      await this.catalogApi.getEntities({ filter: entityFilter }, { token })
     ).items.map(stringifyEntityRef);
 
     for (const entityRef of entities) {