From d774f1bfe6fe315dde706db5ab5714b23a388e90 Mon Sep 17 00:00:00 2001
From: Henry Fontanier <h.fontanier@gmail.com>
Date: Thu, 12 Dec 2024 14:13:59 +0100
Subject: [PATCH] double notion parent IDs (#9290)

* double notion parent IDs

* last place

* reverse

* add to updateParents

* add database doc if

---------

Co-authored-by: Henry Fontanier <henry@dust.tt>
---
 .../src/connectors/notion/lib/parents.ts      |  9 +++--
 .../connectors/notion/temporal/activities.ts  | 33 ++++++++++++++-----
 2 files changed, 32 insertions(+), 10 deletions(-)

diff --git a/connectors/src/connectors/notion/lib/parents.ts b/connectors/src/connectors/notion/lib/parents.ts
index 6e317f4b857e..92b952491a2f 100644
--- a/connectors/src/connectors/notion/lib/parents.ts
+++ b/connectors/src/connectors/notion/lib/parents.ts
@@ -155,13 +155,18 @@ export async function updateAllParentsFields(
   for (const pageId of pageIdsToUpdate) {
     promises.push(
       q.add(async () => {
-        const parents = await getParents(
+        const pageOrDbIds = await getParents(
           connectorId,
           pageId,
           [],
           memoizationKey,
           onProgress
         );
+
+        // TODO(kw_search) remove legacy
+        const legacyParents = pageOrDbIds;
+        const parents = pageOrDbIds.map((id) => `notion-${id}`);
+
         logger.info(
           {
             connectorId,
@@ -172,7 +177,7 @@ export async function updateAllParentsFields(
         await updateDocumentParentsField({
           dataSourceConfig: dataSourceConfigFromConnector(connector),
           documentId: `notion-${pageId}`,
-          parents,
+          parents: [...parents, ...legacyParents],
         });
         if (onProgress) {
           await onProgress();
diff --git a/connectors/src/connectors/notion/temporal/activities.ts b/connectors/src/connectors/notion/temporal/activities.ts
index 94b16e69ce74..383122a3650a 100644
--- a/connectors/src/connectors/notion/temporal/activities.ts
+++ b/connectors/src/connectors/notion/temporal/activities.ts
@@ -1790,7 +1790,7 @@ export async function renderAndUpsertPageFromCache({
           rowBoundary: "",
         });
 
-        const parents = await getParents(
+        const parentPageOrDbIds = await getParents(
           connector.id,
           parentDb.notionDatabaseId,
           [],
@@ -1800,6 +1800,10 @@ export async function renderAndUpsertPageFromCache({
           }
         );
 
+        // TODO(kw_search) remove legacy
+        const legacyParents = parentPageOrDbIds;
+        const parents = parentPageOrDbIds.map((id) => `notion-${id}`);
+
         await ignoreTablesError(
           () =>
             upsertTableFromCsv({
@@ -1811,7 +1815,7 @@ export async function renderAndUpsertPageFromCache({
               loggerArgs,
               // We only update the rowId of for the page without truncating the rest of the table (incremental sync).
               truncate: false,
-              parents,
+              parents: [...parents, ...legacyParents],
               title: parentDb.title ?? "Untitled Notion Database",
               mimeType: "application/vnd.dust.notion.database",
             }),
@@ -1990,7 +1994,8 @@ export async function renderAndUpsertPageFromCache({
     localLogger.info(
       "notionRenderAndUpsertPageFromCache: Fetching resource parents."
     );
-    const parents = await getParents(
+
+    const parentPageOrDbIds = await getParents(
       connectorId,
       pageId,
       [],
@@ -2000,6 +2005,10 @@ export async function renderAndUpsertPageFromCache({
       }
     );
 
+    // TODO(kw_search) remove legacy
+    const legacyParentIds = parentPageOrDbIds;
+    const parentIds = parentPageOrDbIds.map((id) => `notion-${id}`);
+
     const content = await renderDocumentTitleAndContent({
       dataSourceConfig: dsConfig,
       title: title ?? null,
@@ -2027,7 +2036,8 @@ export async function renderAndUpsertPageFromCache({
         updatedTime: updatedAt.getTime(),
         parsedProperties,
       }),
-      parents,
+      // TODO(kw_search) remove legacy
+      parents: [...parentIds, ...legacyParentIds],
       loggerArgs,
       upsertContext: {
         sync_type: isFullSync ? "batch" : "incremental",
@@ -2505,7 +2515,7 @@ export async function upsertDatabaseStructuredDataFromCache({
 
   const upsertAt = new Date();
 
-  const parents = await getParents(
+  const parentPageOrDbIds = await getParents(
     connector.id,
     databaseId,
     [],
@@ -2513,6 +2523,10 @@ export async function upsertDatabaseStructuredDataFromCache({
     async () => heartbeat()
   );
 
+  // TODO(kw_search) remove legacy
+  const legacyParentIds = parentPageOrDbIds;
+  const parentIds = parentPageOrDbIds.map((id) => `notion-${id}`);
+
   localLogger.info("Upserting Notion Database as Table.");
   await ignoreTablesError(
     () =>
@@ -2525,7 +2539,8 @@ export async function upsertDatabaseStructuredDataFromCache({
         loggerArgs,
         // We overwrite the whole table since we just fetched all child pages.
         truncate: true,
-        parents,
+        // TODO(kw_search) remove legacy
+        parents: [...parentIds, ...legacyParentIds],
         title: dbModel.title ?? "Untitled Notion Database",
         mimeType: "application/vnd.dust.notion.database",
       }),
@@ -2561,9 +2576,10 @@ export async function upsertDatabaseStructuredDataFromCache({
       maxPrefixChars: MAX_PREFIX_CHARS * 2,
     });
     if (!prefixSection.content) {
+      const databaseDocId = `notion-database-${databaseId}`;
       await upsertToDatasource({
         dataSourceConfig,
-        documentId: `notion-database-${databaseId}`,
+        documentId: databaseDocId,
         documentContent: {
           prefix: prefixSection.prefix,
           content: csvRows,
@@ -2576,7 +2592,8 @@ export async function upsertDatabaseStructuredDataFromCache({
         // we currently don't have it because we don't fetch the DB object from notion.
         timestampMs: upsertAt.getTime(),
         tags: [`title:${databaseName}`, "is_database:true"],
-        parents: parents,
+        // TODO(kw_search) remove legacy
+        parents: [databaseDocId, ...parentIds, ...legacyParentIds],
         loggerArgs,
         upsertContext: {
           sync_type: "batch",