pse2_ff/project

PSE2 - Pitchbook Extraction Webapplication

A microservices platform for processing pitchbook PDFs using OCR and entity extraction services. Combines SpaCy NLP and GPT-based (ExxetaGPT) extraction of kpi in Pitchbooks.

## Quick Start

### 1. Environment Setup
Create a `.env` file in the project root:

# Database
DATABASE_URL=url
POSTGRES_USER=admin
POSTGRES_PASSWORD=password

# API Key (required for ExxetaGPT service)
API_KEY=your_exxeta_jwt_token_here

2. Start Application

# Build and start all services
docker-compose up --build

# Run in background
docker-compose up --build -d

# Stop services
docker-compose down

3. Access Application

• Frontend: http://localhost:8080
• API: http://localhost:5050

Services Overview

Service	Port	Purpose
Frontend	8080	React UI for file upload and results display
Coordinator	5050	Main API, file storage, database management
OCR	5051	PDF text extraction using OCRmyPDF
ExxetaGPT	5053	AI entity extraction using GPT-4o-mini
SpaCy	5052	NLP entity extraction using custom model
Validate	5054	Merges and validates results from both extractors
Database	5432	PostgreSQL for data persistence

Usage Flow

1. Upload PDF via web interface
2. OCR service extracts text from PDF
3. Both ExxetaGPT and SpaCy services extract kpi's entities
4. Validate service merges and validates results
5. View extracted kpi's and original PDF side-by-side

Troubleshooting

Services won't start:

# Check logs
docker-compose logs

ExxetaGPT errors:

• Ensure API_KEY is set in .env file
• Check API key validity and network access

Database connection issues:

• Wait for database health check to pass
• Verify DATABASE_URL format in .env