From 82f3d350c215093a2f74dc5f29d9966f7c809740 Mon Sep 17 00:00:00 2001
From: Henry Fontanier <h.fontanier@gmail.com>
Date: Fri, 3 Nov 2023 13:10:31 +0100
Subject: [PATCH] feat: add CLI to skip gdrive files (#2363)

---
 connectors/src/admin/cli.ts                   | 59 ++++++++++++++++++-
 .../google_drive/temporal/activities.ts       | 20 +++++++
 connectors/src/lib/models.ts                  |  5 ++
 3 files changed, 81 insertions(+), 3 deletions(-)

diff --git a/connectors/src/admin/cli.ts b/connectors/src/admin/cli.ts
index 4bb66581597e..446e373c497a 100644
--- a/connectors/src/admin/cli.ts
+++ b/connectors/src/admin/cli.ts
@@ -7,10 +7,16 @@ import {
   STOP_CONNECTOR_BY_TYPE,
   SYNC_CONNECTOR_BY_TYPE,
 } from "@connectors/connectors";
+import { getDocumentId } from "@connectors/connectors/google_drive/temporal/activities";
 import { launchGoogleDriveRenewWebhooksWorkflow } from "@connectors/connectors/google_drive/temporal/client";
 import { toggleSlackbot } from "@connectors/connectors/slack/bot";
 import { launchSlackSyncOneThreadWorkflow } from "@connectors/connectors/slack/temporal/client";
-import { Connector, NotionDatabase, NotionPage } from "@connectors/lib/models";
+import {
+  Connector,
+  GoogleDriveFiles,
+  NotionDatabase,
+  NotionPage,
+} from "@connectors/lib/models";
 import { Result } from "@connectors/lib/result";
 
 const connectors = async (command: string, args: parseArgs.ParsedArgs) => {
@@ -249,10 +255,57 @@ const notion = async (command: string, args: parseArgs.ParsedArgs) => {
   }
 };
 
-const google = async (command: string) => {
+const google = async (command: string, args: parseArgs.ParsedArgs) => {
   switch (command) {
     case "restart-google-webhooks": {
       await throwOnError(launchGoogleDriveRenewWebhooksWorkflow());
+      return;
+    }
+    case "skip-file": {
+      if (!args.wId) {
+        throw new Error("Missing --wId argument");
+      }
+      if (!args.dataSourceName) {
+        throw new Error("Missing --dataSourceName argument");
+      }
+      if (!args.fileId) {
+        throw new Error("Missing --fileId argument");
+      }
+
+      const connector = await Connector.findOne({
+        where: {
+          workspaceId: args.wId,
+          dataSourceName: args.dataSourceName,
+        },
+      });
+      if (!connector) {
+        throw new Error(
+          `Could not find connector for workspace ${args.wId} and data source ${args.dataSourceName}`
+        );
+      }
+
+      const existingFile = await GoogleDriveFiles.findOne({
+        where: {
+          driveFileId: args.fileId,
+          connectorId: connector.id,
+        },
+      });
+      if (existingFile) {
+        await existingFile.update({
+          skipReason: args.reason || "blacklisted",
+        });
+      } else {
+        await GoogleDriveFiles.create({
+          driveFileId: args.fileId,
+          dustFileId: getDocumentId(args.fileId),
+          name: "unknown",
+          mimeType: "unknown",
+          connectorId: connector.id,
+          skipReason: args.reason || "blacklisted",
+        });
+      }
+
+      return;
     }
   }
 };
@@ -367,7 +420,7 @@ const main = async () => {
       await notion(command, argv);
       return;
     case "google":
-      await google(command);
+      await google(command, argv);
       return;
     case "slack":
       await slack(command, argv);
diff --git a/connectors/src/connectors/google_drive/temporal/activities.ts b/connectors/src/connectors/google_drive/temporal/activities.ts
index b7e818537f52..c644a6f1b7b9 100644
--- a/connectors/src/connectors/google_drive/temporal/activities.ts
+++ b/connectors/src/connectors/google_drive/temporal/activities.ts
@@ -299,6 +299,26 @@ async function syncOneFile(
   const documentId = getDocumentId(file.id);
   let documentContent: string | undefined = undefined;
 
+  const fileInDb = await GoogleDriveFiles.findOne({
+    where: {
+      connectorId: connectorId,
+      driveFileId: file.id,
+    },
+  });
+
+  if (fileInDb?.skipReason) {
+    logger.info(
+      {
+        documentId,
+        dataSourceConfig,
+        fileId: file.id,
+        title: file.name,
+      },
+      `Google Drive document skipped with skip reason ${fileInDb.skipReason}`
+    );
+    return false;
+  }
+
   if (FILES_IGNORE_LIST.includes(file.id)) {
     logger.info(
       {
diff --git a/connectors/src/lib/models.ts b/connectors/src/lib/models.ts
index 813059ffa236..c4455d3b7f67 100644
--- a/connectors/src/lib/models.ts
+++ b/connectors/src/lib/models.ts
@@ -1102,6 +1102,7 @@ export class GoogleDriveFiles extends Model<
   declare updatedAt: CreationOptional<Date>;
   declare lastSeenTs: Date | null;
   declare lastUpsertedTs: Date | null;
+  declare skipReason: string | null;
   declare connectorId: ForeignKey<Connector["id"]>;
   declare dustFileId: string;
   declare driveFileId: string;
@@ -1135,6 +1136,10 @@ GoogleDriveFiles.init(
       type: DataTypes.DATE,
       allowNull: true,
     },
+    skipReason: {
+      type: DataTypes.STRING,
+      allowNull: true,
+    },
     connectorId: {
       type: DataTypes.INTEGER,
       allowNull: false,