[connectors] Upsert and backfill Confluence spaces as data_source_folders

connectors/migrations/20241216_backfill_confluence_folders.ts

@@ -0,0 +1,43 @@
+import { makeScript } from "scripts/helpers";
+
+import { makeSpaceInternalId } from "@connectors/connectors/confluence/lib/internal_ids";
+import { dataSourceConfigFromConnector } from "@connectors/lib/api/data_source_config";
+import { concurrentExecutor } from "@connectors/lib/async_utils";
+import { upsertFolderNode } from "@connectors/lib/data_sources";
+import { ConfluenceSpace } from "@connectors/lib/models/confluence";
+import { ConnectorResource } from "@connectors/resources/connector_resource";
+
+const FOLDER_CONCURRENCY = 10;
+
+makeScript({}, async ({ execute }, logger) => {
+  const connectors = await ConnectorResource.listByType("confluence", {});
+
+  for (const connector of connectors) {
+    const confluenceSpaces = await ConfluenceSpace.findAll({
+      attributes: ["spaceId", "name"],
+      where: { connectorId: connector.id },
+    });
+    const dataSourceConfig = dataSourceConfigFromConnector(connector);
+    if (execute) {
+      await concurrentExecutor(
+        confluenceSpaces,
+        async (space) => {
+          await upsertFolderNode({
+            dataSourceConfig,
+            folderId: makeSpaceInternalId(space.spaceId),
+            parents: [makeSpaceInternalId(space.spaceId)],
+            title: space.name,
+          });
+        },
+        { concurrency: FOLDER_CONCURRENCY }
+      );
+      logger.info(
+        `Upserted ${confluenceSpaces.length} spaces for connector ${connector.id}`
+      );
+    } else {
+      logger.info(
+        `Found ${confluenceSpaces.length} spaces for connector ${connector.id}`
+      );
+    }
+  }
+});

connectors/src/connectors/confluence/temporal/activities.ts

@@ -18,16 +18,21 @@ import {
   getConfluencePageParentIds,
   getSpaceHierarchy,
 } from "@connectors/connectors/confluence/lib/hierarchy";
-import { makePageInternalId } from "@connectors/connectors/confluence/lib/internal_ids";
+import {
+  makePageInternalId,
+  makeSpaceInternalId,
+} from "@connectors/connectors/confluence/lib/internal_ids";
 import { makeConfluenceDocumentUrl } from "@connectors/connectors/confluence/temporal/workflow_ids";
 import { dataSourceConfigFromConnector } from "@connectors/lib/api/data_source_config";
 import { concurrentExecutor } from "@connectors/lib/async_utils";
 import type { UpsertToDataSourceParams } from "@connectors/lib/data_sources";
 import {
+  deleteFolderNode,
   deleteFromDataSource,
   renderDocumentTitleAndContent,
   renderMarkdownSection,
   updateDocumentParentsField,
+  upsertFolderNode,
   upsertToDatasource,
 } from "@connectors/lib/data_sources";
 import {
@@ -196,6 +201,28 @@ export async function confluenceGetSpaceNameActivity({
   }
 }
 
+/**
+ * Upserts the page in data_sources_folders (core).
+ */
+export async function confluenceUpsertSpaceFolderActivity({
+  connectorId,
+  spaceId,
+  spaceName,
+}: {
+  connectorId: ModelId;
+  spaceId: string;
+  spaceName: string;
+}) {
+  const connector = await fetchConfluenceConnector(connectorId);
+
+  await upsertFolderNode({
+    dataSourceConfig: dataSourceConfigFromConnector(connector),
+    folderId: makeSpaceInternalId(spaceId),
+    parents: [makeSpaceInternalId(spaceId)],
+    title: spaceName,
+  });
+}
+
 export async function markPageHasVisited({
   connectorId,
   pageId,
@@ -814,6 +841,12 @@ export async function confluenceRemoveSpaceActivity(
   for (const page of allPages) {
     await deletePage(connectorId, page.pageId, dataSourceConfig);
   }
+
+  // deleting the folder in data_source_folders (core)
+  await deleteFolderNode({
+    dataSourceConfig,
+    folderId: makeSpaceInternalId(spaceId),
+  });
 }
 
 export async function fetchConfluenceSpaceIdsForConnectorActivity({

connectors/src/connectors/confluence/temporal/workflows.ts

@@ -37,6 +37,7 @@ const {
   fetchConfluenceUserAccountAndConnectorIdsActivity,
 
   fetchConfluenceConfigurationActivity,
+  confluenceUpsertSpaceFolderActivity,
   getSpaceIdsToSyncActivity,
 } = proxyActivities<typeof activities>({
   startToCloseTimeout: "30 minutes",
@@ -150,6 +151,12 @@ export async function confluenceSpaceSyncWorkflow(
     return startConfluenceRemoveSpaceWorkflow(wInfo, connectorId, spaceId);
   }
 
+  await confluenceUpsertSpaceFolderActivity({
+    connectorId,
+    spaceId,
+    spaceName,
+  });
+
   // Get the root level pages for the space.
   const rootPageRefs = await confluenceGetRootPageRefsActivity({
     connectorId,