React PDF Statement Extractor

A lightweight, secure, browser-only PDF text extraction library for React, tailored for credit card statements and financial documents.

Powered by DeepFi

Why we built this

Extracting financial data usually requires sending PDFs to a backend server. We wanted an Open Source, 100% client-side solution that guarantees user privacy. This tool extracts all the text inside the browser's memory before any interaction with an LLM or database.

Features

• 100% Client-Side: The PDF never leaves the browser.
• Secure Limits: Built-in protections against zip/decompression bombs. Max pages (50) and max characters (100k) enforced.
• Financial Validation: Smart verification to ensure the document contains financial keywords (EN, ES, PT) before processing.
• Password Support: Handles encrypted PDFs with custom errors.
• Zero Build Setup: Uses lazy loading CDN imports to prevent Vite or Webpack build issues with pdf.js web workers.

Installation

npm install react-pdf-statement-extractor

Basic Usage

import { extractTextFromPDF } from "react-pdf-statement-extractor";

async function handleFileUpload(event) {
  const file = event.target.files[0];
  
  try {
    const extractedText = await extractTextFromPDF(file);
    console.log("PDF Content:", extractedText);
    
    // Now you can safely send `extractedText` to an LLM or backend!
    
  } catch (error) {
    if (error.message === 'PASSWORD_REQUIRED') {
      alert("This PDF is encrypted. Please provide a password.");
    } else {
      alert(error.message);
    }
  }
}

With Passwords

const text = await extractTextFromPDF(file, "mypassword123");

How It Works in DeepFi

At DeepFi we use this library to locally extract your credit card statements, and then we send only the raw text to our backend LLM (Groq API) for categorization and advice. Your full PDF file and PII data are never uploaded anywhere.

Open Source

This is the extraction core of DeepFi. We are open sourcing it to maintain transparency about how we handle user data.

License

MIT