PDF Summary Extraction API

This project provides an API for uploading PDF documents and extracting structured summaries using either OpenAI's GPT-4o or Anthropic's Claude models. It is optimized for environmental and agricultural reports, transforming unstructured PDF content into structured JSON for downstream use.

---

🛠️ Tech Stack

• Framework: Node.js with Express
• PDF Parsing: pdf-parse
• LLM Integration: OpenAI GPT-4o and Claude Opus 4
• Storage: In-memory via Multer
• Language: TypeScript

---

🚀 Features

• Upload a PDF and extract structured data in a single API call
• Choose between OpenAI or Claude as LLM provider
• Validates and cleans JSON response for consistent schema
• Returns fallback empty structure in case of model or parsing failure

---

📦 Setup Instructions

1. Clone the Repository

   git clone https://github.com/your-org/pdf-extractor.git
   cd pdf-extractor

2. Install Dependencies

   npm install

3. Environment Configuration Create a .env file:

   OPENAI_API_KEY=your_openai_key
   ANTHROPIC_API_KEY=your_claude_key
   

4. Run the Server

   npm run dev

---

🧪 API Usage

POST /upload

Upload a PDF and extract a summary:

Form Data:

• file: PDF file (multipart/form-data)

Optional Query:

• provider: openai (default) or claude

Example with cURL:

curl -X POST http://localhost:3000/upload?provider=claude   -F "file=@./sample.pdf"

---

📁 Folder Structure

├── routes/
│   └── upload.ts       → Express route for file upload
├── services/
│   ├── pdfService.ts   → PDF parsing logic
│   └── llmService.ts   → LLM integration (OpenAI & Claude)
├── utils/
│   ├── types.ts        → ExtractedReport interface
│   └── constants.ts    → Prompt templates
├── .env
├── README.md