Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/HugoX2003/nisira-assistant/llms.txt

Use this file to discover all available pages before exploring further.

NISIRA can continuously ingest documents from a designated Google Drive folder without manual uploads. The GoogleDriveManager class (in rag_system/drive_sync/drive_manager.py) handles authentication, file listing, modification-time-aware downloading, and persistent storage. Once configured, you can trigger syncs on demand through the Admin Panel or let a background polling loop keep the knowledge base up to date automatically.

​
How the Sync Works

On each sync cycle, sync_documents() calls the Drive API to list all supported files in the configured folder. For each file, it compares the Drive modification time against what is already stored — in PostgreSQL (file_store.get_file_modified_time()) or on the local filesystem. Files that are new or have been updated since the last sync are downloaded to memory and saved to PostgresFileStore (or the local filesystem as a fallback). Files that are already current are skipped. The polling interval is controlled by GOOGLE_DRIVE_SYNC_INTERVAL (default 300 seconds), read directly from the environment at startup into GOOGLE_DRIVE_CONFIG["sync_interval"].

​
Supported Formats and Size Limit

ExtensionMax Size
.pdf50 MB
.txt50 MB
.docx / .doc50 MB
.pptx50 MB
.xlsx50 MB
Files larger than 50 MB are skipped with a "TOO_LARGE" sentinel and do not block the sync of other files.

​
Setup

1

Create a Google Cloud project and enable the Drive API

In the Google Cloud Console, create a new project (or select an existing one) and enable the Google Drive API from the API Library.
2

Create credentials

Choose one of two credential types:
  • Service Account (recommended for production) — create a service account, download the JSON key file. Share your Drive folder with the service account’s email address (Editor permission) so it has access to the folder contents.
  • OAuth 2.0 (development / personal Drive) — create an OAuth 2.0 client ID, download credentials.json, and complete the browser-based consent flow to generate token.json. OAuth tokens use personal Drive quota and avoid the zero-storage limitation of service accounts.
3

Configure environment variables

Set the following variables before starting the Django server:
ENABLE_GOOGLE_DRIVE=true
GOOGLE_DRIVE_FOLDER_ID=<your_folder_id>

# Option A – JSON string in env var (production, auto-writes credentials.json)
GOOGLE_CREDENTIALS_JSON='{"type":"service_account",...}'
GOOGLE_TOKEN_JSON='{"token":"...","refresh_token":"...",...}'

# Option B – file paths (development)
# Place credentials.json and token.json in the backend root; no extra env vars needed.

# Optional overrides
GOOGLE_DRIVE_SYNC_INTERVAL=300        # Polling interval in seconds
When GOOGLE_CREDENTIALS_JSON is set and credentials.json does not yet exist on disk, config.py writes the file automatically from the env var on startup.
4

Verify authentication

NISIRA tries credential methods in this priority order:
  1. OAuth2 via individual env vars (GOOGLE_OAUTH_CLIENT_ID, GOOGLE_OAUTH_CLIENT_SECRET, GOOGLE_OAUTH_REFRESH_TOKEN) — uses personal Drive quota.
  2. Existing token.json user token file.
  3. Service account from credentials.json.
Call the status endpoint to confirm authentication is working:
GET /api/admin/drive/files/
Authorization: Bearer <admin_JWT>
5

Trigger your first sync

From the Admin Panel click Sync Now, or call the API directly:
POST /api/admin/drive/sync/
Authorization: Bearer <admin_JWT>
Poll the progress endpoint while the sync runs:
GET /api/admin/drive/sync/progress/
Authorization: Bearer <admin_JWT>

​
Drive Management API Endpoints

All endpoints require an admin JWT in the Authorization: Bearer header.
MethodPathDescription
GET/api/admin/drive/files/List all files in the configured Drive folder
POST/api/admin/drive/upload/Upload a local file to the Drive folder
DELETE/api/admin/drive/delete/<file_id>/Delete a file from Drive by its ID
POST/api/admin/drive/sync/Trigger a full sync (download new/updated files)
GET/api/admin/drive/sync/progress/Poll real-time sync progress and log messages

​
Management Commands

Run a background polling loop that syncs every GOOGLE_DRIVE_SYNC_INTERVAL seconds: 
python manage.py start_drive_sync
Run a single full sync and exit (useful for scheduled tasks or CI): 
python manage.py sync_drive_full

​
Credential Methods in Detail

MethodWhen to UseToken Refresh
OAuth env vars (GOOGLE_OAUTH_CLIENT_ID + GOOGLE_OAUTH_REFRESH_TOKEN)Production with personal DriveAutomatic via creds.refresh(Request())
token.json fileDevelopmentAutomatic if refresh_token present; re-run flow if expired
Service account JSONProduction with shared folderNo refresh needed; permanent credentials
Service accounts have zero bytes of personal Drive storage. You must share a folder from a personal Google Drive with the service account’s email address and set that folder’s ID as GOOGLE_DRIVE_FOLDER_ID. Uploading files to Drive via the API will fail with storageQuotaExceeded if the folder is not shared correctly.
For unattended production deployments, use OAuth env vars (GOOGLE_OAUTH_REFRESH_TOKEN) rather than a service account. This avoids storage quota issues and eliminates the need to share a folder.

Build docs developers (and LLMs) love

Get started for freeTalk to us