Snap2GCal lets you turn screenshots and PDFs into Google Calendar events in a few clicks.

• Upload screenshots/PDFs of event flyers, emails, or chat messages.
• Extract event details (title, date, time, location, notes) using your AWS Bedrock-powered Lambda.
• Review and edit the extracted fields in the UI.
• Add to Google Calendar via a one-click “Add to Google Calendar” button.

• Frontend (client/)

  • Vite + React UI
  • Google OAuth via @react-oauth/google
  • Protected routes:
    • /signin – Google sign-in page
    • / – main Snap2GCal dashboard (screenshot upload + event details)
  • Calls your Lambda via VITE_API_URL:
    • POST /upload-url – get S3 presigned URL
    • POST /ingest – ingest PDF into the RAG index
    • POST /query-calendar – extract calendar event fields

• Backend (lambda_handler.py)

  • AWS Lambda (container image) + Nova 2 Lite for image detail extraction
  • Ingests PDFs from S3, chunks + embeds them into a FAISS index stored back in S3.
  • Handles /ingest, /query, and /query-calendar routes for RAG and event extraction.

• Local Google Calendar helper (client/google_api_calendar.py)

  • CLI helper for creating events via the Google Calendar API using OAuth credentials (client-secret.json/token.json).
  • Not used by the web UI directly; the UI uses the “Add to Google Calendar” URL flow instead.

• Node.js 18+ and npm
• Python 3.10+ (for Lambda/dev tooling)
• An AWS account with:
  • S3 bucket for RAG artifacts and uploads
  • Bedrock access (embedding + generation models)
• A Google Cloud project with:
  • OAuth 2.0 Web client for the frontend (client ID only)
  • Authorized JavaScript origin: http://localhost:5173 (for local dev)

1. Install dependencies:

   cd client
   npm install

2. Configure environment:

   cp .env.example .env

   Fill in:

   VITE_API_URL=https://<your-lambda-or-apigw-url>

3. Start the dev server:

   npm run dev

   Open http://localhost:5173.

• Build and push your Lambda container image (contains lambda_handler.py and dependencies).
• Configure environment variables on the function:
  • RAG_BUCKET – S3 bucket name
  • INDEX_PREFIX – optional prefix (default index/)
  • EMBED_MODEL_ID, GEN_MODEL_ID – Bedrock model IDs
  • Any other tuning variables (TOP_K, CHUNK_SIZE, etc.).
• Expose the Lambda via Function URL or API Gateway and use that URL as VITE_API_URL.

• The frontend uses GoogleOAuthProvider and GoogleLogin from @react-oauth/google.
• The client ID must match your Google Cloud Web OAuth client.
• User information is stored in sessionStorage and used only for gating the protected routes and personalizing the UI.

• The “Add to Google Calendar” button builds a URL like:

  https://calendar.google.com/calendar/render?action=TEMPLATE&text=...&dates=...&location=...&details=...

• Google opens the user’s calendar with the fields pre-filled; they can review and save the event.

From client/:

python google_api_calendar.py

This script uses client-secret.json and token.json to talk directly to the Google Calendar API for testing and debugging; it is not required for the main web flow.