Custom PDF Text Extraction with Node.js

What is a Custom PDF Text Extractor?

A custom PDF text extractor is a program that reads a PDF file and returns its textual content, metadata, or specific page ranges on demand. Unlike generic libraries, a custom implementation can be tailored to the exact needs of an application, minimizing dependencies and avoiding unnecessary functionality.

Why Build a Custom Extractor?

• Full control over the technology stack and performance characteristics.
• Ability to expose only the required API surface (e.g., text, metadata, page‑range, search).
• Reduced bundle size and a smaller security surface compared to monolithic libraries.
• Facilitates tighter integration with business logic such as authentication, rate‑limiting, or caching.

How to Set Up a TypeScript‑Based Node.js Project

• Initialize the project: npm init -y
• Install runtime dependencies: npm install express cors file-upload pdf-parse
• Install development dependencies: npm install -D typescript ts-node nodemon @types/node @types/express @types/cors @type/file-upload
• Generate a TypeScript configuration file: npx tsc --init and adjust the target/module as needed.
• Create a basic Express server that reads the PORT environment variable and starts listening.

Core Implementation: The Basic Parse Function

The central function accepts a Uint8Array representing the uploaded file, creates a pdf‑parse instance, and returns an object containing:

• text – the extracted text (empty string if none).
• info – author, title, creation date, etc.
• numpages – total page count.

Both getText() and getInfo() are awaited to avoid blocking the event loop.

Adding Page‑Specific Extraction

• Expose a second endpoint that accepts startPage and endPage query parameters.
• Validate the range against the document’s total page count.
• Slice the extracted text to the requested range and return it with the original metadata.

Providing a Lightweight Metadata‑Only Endpoint

• Implement a function that calls pdfParse.getInfo() without invoking getText().
• Normalize date strings and supply placeholder values for missing fields.
• Return a concise JSON object that can be used for preview or indexing.

Implementing Full‑Text Search

• Write a search utility that iterates over each page, applies a case‑sensitive or case‑insensitive pattern, and captures a 100‑character context around each match.
• Return a structured result with total matches, the query, and an array of matches (page, position, snippet).

Handling Common Edge Cases

• Corrupted or malformed PDFs – wrap parsing in try‑catch and return a 400‑type error.
• Password‑protected PDFs – either reject them or accept a password and pass it to the parser if supported.
• Scanned image‑based PDFs – integrate an OCR library (e.g., Tesseract) for text extraction.
• Special‑character encoding – ensure the application uses UTF‑8 and test with non‑Latin scripts.

Testing with Jest and Supertest

• Install test dependencies: npm install --save-dev jest ts-jest @types/jest supertest @types/supertest
• Write unit tests for each endpoint (upload, metadata, page‑range, search) using Supertest to simulate HTTP requests.
• Mock the file‑upload middleware and the pdf‑parse library to keep tests fast and deterministic.
• Run the suite with npm test and generate a coverage report to ensure critical paths are exercised.

Deploying the API

• Build the TypeScript source: npm run build (typically compiles to a dist folder).
• Start the compiled server with npm start or use a platform‑specific start script.
• Configure environment variables for PORT and any future secret (e.g., OCR service key).
• Consider production enhancements: rate‑limiting, structured logging (Winston/Pino), monitoring (Sentry), and optional caching of metadata.

Integrating into a SaaS Application

• Expose the API behind an authenticated gateway to enforce per‑user quotas.
• Cache metadata for frequently accessed documents to reduce processing time.
• Extend the service to handle other document types (DOCX, XLSX) by following the same pattern.
• Offer asynchronous processing via a job queue for large files or batch uploads.