Merge pull request #1639 from kuangp/fix/rag-ai/techdoc-vector-cleanup

fix(rag-ai): properly clean up old techdoc vectors
RoadieHQ · Sep 26, 2024 · 69f0581 · 69f0581
2 parents a4043ee + 890743e
commit 69f0581
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 5 deletions.
diff --git a/.changeset/sweet-elephants-glow.md 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/...ckend/rag-ai-backend-retrieval-augmenter/src/indexing/DefaultVectorAugmentationIndexer.ts b/...ckend/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) {