pdfextract API

Extract structured text and embedded raster images from whole PDFs in browsers and Node.js 22.12 or newer. Keep full-resolution exports, thumbnails and stored assets separate, and use OCR when text exists only inside images.

Installation

Install only the packages you need. OCR and storage are optional additions to core:

pnpm add @pdfextract/core
# Optional OCR and storage adapters:
pnpm add @pdfextract/ocr @pdfextract/storage
# Required only for the S3 adapter:
pnpm add @aws-sdk/client-s3
Copy

Use Node.js ≥22.12.0 or a modern browser with ESM support.

Quick start

Pass a whole File, Blob, ArrayBuffer or Uint8Array to openPdf; Node Buffer also works. This Node example reads native text and saves each embedded image at its original dimensions, with a separate thumbnail:

import { readFile, writeFile } from 'node:fs/promises';
import { openPdf } from '@pdfextract/core';

const pdf = await openPdf(await readFile('input.pdf'));
try {
  const text = await pdf.getStructuredText({ ocr: 'off' });
  console.log(text.fullText);
  await writeFile('text.json', JSON.stringify(text, null, 2));

  let index = 0;
  for (const image of await pdf.getImages()) {
    const full = await pdf.extractImage(image.id);
    const thumbnail = await pdf.extractImage(image.id, { maxWidth: 320, maxHeight: 240 });
    await writeFile(`image-${index}.png`, full.data);
    await writeFile(`image-${index}-thumbnail.png`, thumbnail.data);
    index++;
  }
} finally {
  await pdf.close();
}
Copy

In a browser, pass the selected File directly and set assets.baseUrl to the hosted core asset directory. Follow the asset setup guide to copy the worker, WASM and related files. Exported bytes remain valid after closing the PDF. Image IDs belong to that open document; persist the exports and their associations when you need them in another session.

Add OCR or persistence

For scanned PDFs and images within mixed-content pages, create a provider with createTesseractOcr and pass it as openPdf(input, { ocr }). Supply local or self-hosted language models. getStructuredText({ ocr: 'auto' }) then combines native and recognized text. Close the provider after all PDFs sharing it have closed; see the OCR package page.

Use createFilesystemStorage in Node or createS3Storage with a host-provided AWS SDK client. Both support put, get, head and delete, preserve bytes and metadata, and return web streams from get. See the storage package page for examples.

Guides and complete examples

Start with Getting started, then configure OCR and assets and storage and persistence. The coordinates and limits guide describes fidelity, resource budgets and lifecycle guarantees.

The complete browser and Node examples store text, full-size images and thumbnails, publish a manifest last, and retrieve stored images in a fresh session. To build or serve this documentation website, see the documentation workspace README.

The host application owns authentication, search/indexing and its UI. The browser example hands a stored, full-size File to a host callback. PDF actions and embedded external URLs are not executed.

All packages are MIT-licensed. Bundled third-party components keep their own licenses, listed in each package's THIRD_PARTY_NOTICES directory.