PDF Knowledge Management System

Author: Tran Minh Khoi, IT Department, Phat Dat Holdings

Purpose

A modern, AI-powered knowledge/document management system for Phat Dat Holdings. Supports smart PDF upload, OCR, company-specific knowledge, and advanced Q&A.

Quick Start

1. Install

npm install
cp .env.example .env
# Edit .env for DB, Google Cloud, and storage

2. Run Server

export $(cat .env | grep -v '#' | xargs) && PORT=3000 node server.js

3. Admin Login

curl -X POST http://localhost:3000/api/auth/login \
  -H 'Content-Type: application/json' \
  -d '{"username": "admin", "password": "Admin@123123"}'

4. Main APIs

• Upload PDF: /api/upload (admin, multipart/form-data, field: file)
• Ask Question: /api/ask (public)
• Learn Knowledge: /api/learn (admin)
• Companies: /api/companies (public)

Environment Variables

• DATABASE_URL (Postgres/CockroachDB)
• GOOGLE_APPLICATION_CREDENTIALS_JSON (Google Cloud Vision)
• GCS_BUCKET_NAME (Cloud Storage)
• GEMINI_API_KEY (AI)
• MAX_FILE_SIZE, UPLOAD_TIMEOUT_MINUTES, ... (see .env.example)

Features

• Smart PDF Upload: Auto-detect company, OCR, metadata extraction, cloud storage
• AI Q&A: Answers from both documents and learned knowledge
• Knowledge Base: Add/update knowledge, company-specific
• Admin Auth: JWT, default admin: admin/Admin@123123
• Production-ready Migration: Schema auto-migrates on deploy

Project Structure

• src/controllers/ — API logic
• src/repositories/ — DB access
• src/routes/ — API endpoints
• services/ — AI, OCR, storage
• scripts/ — Migration, setup

Production Setup

• Set all required env vars (see above)
• Deploy (Railway, Docker, etc.)
• On start, schema auto-migrates (no manual SQL needed)
• Health check: /health (returns 200 OK)

Requirements

• Node.js 18+
• PostgreSQL or CockroachDB
• Google Cloud Vision & Storage
• Tesseract OCR, GraphicsMagick

---

Copyright © Tran Minh Khoi, IT Department, Phat Dat Holdings