Invoice RPA System

Full-stack, Dockerized invoice processing system with FastAPI, PostgreSQL, OCR, and a React dashboard.

Features

• Invoice upload: PDF / PNG / JPG via POST /invoices/upload
• OCR & parsing: Tesseract + pdfplumber with regex-based field extraction
• Database: PostgreSQL + SQLAlchemy + Alembic
• Dashboard: React (Vite) UI for uploads and browsing invoices with pagination
• Email summary: Daily summary at 18:00 UTC with CSV attachment
• Dockerized: Backend, frontend, and Postgres in docker-compose

Project structure

• backend/ – FastAPI app, models, services, Alembic, Dockerfile
• frontend/ – React (Vite-style) app and Dockerfile
• docker-compose.yml – Multi-container stack
• .env.example – Example environment variables

Prerequisites

• Docker and Docker Compose

Configuration

1. Copy .env.example to .env in the project root and adjust values:

cp .env.example .env

2. Ensure your SMTP settings are valid if you want daily summary emails.

Running the stack

From the invoice-rpa directory:

docker compose up --build

Services:

• Frontend: http://localhost:3000
• Backend API docs (Swagger): http://localhost:8000/docs
• Backend health: http://localhost:8000/health

Backend overview

• Main app: app/main.py
• Config: app/config.py (reads env vars, upload dir, SMTP, DB URL)
• DB models: models/invoice.py
• Schemas: app/schemas.py
• Routers:
  • routers/invoices.py
    • POST /invoices/upload – multipart upload, saves file, runs OCR, parses fields, stores invoice with VALID / PARTIAL / FAILED status.
    • GET /invoices – paginated list (query params page, page_size).
• Services:
  • services/ocr_service.py – PDF/image OCR
  • services/invoice_extraction.py – regex-based extraction + status classification
  • services/email_service.py – builds and sends daily summary email + CSV
  • services/scheduler.py – APScheduler daily job at 18:00 UTC

Alembic is configured under backend/alembic with an initial migration to create the invoices table.

Frontend overview

• Vite-style React app with:
  • Upload form with loading/error states
  • Invoice table with vendor, date, amount, status, created time
  • Pagination (Previous/Next, page count)
• Uses Axios to call:
  • GET /invoices?page=&page_size=
  • POST /invoices/upload with multipart/form-data
• API base URL is controlled by:
  • VITE_API_BASE_URL env (Docker sets this to http://backend:8000)

Notes

• Files are stored under /app/uploads in the backend container, mapped to the uploads volume.
• Logging is configured with structured output to stdout.
• Error handling uses FastAPI exceptions and try/except in services.