a11ycat-ocr

a11ycat-ocr

OCR PDF documents in Node.js 🐱

Dependencies

• ImageMagick 7

Optional:

This is only needed if you are testing out the tess method on the OCR class. It is currently experimental, and does not return any of the OCR'd data, but is much faster than tesseract.js.

• Tesseract OCR

IMPORTANT: a11ycat-ocr expects the ImageMagick tools to be availabe in $PATH. If you are testing the tess method on the OCR class, then tesseract must also be in $PATH

Quick Start

1. Build the project from the repository

git clone https://github.com/devnoot/a11ycat-ocr.git a11ycat-ocr
cd a11ycat-ocr
npm install
npm build

1. Include the OCR class in your project

const { A11yCat } = require('../../dist/index')
const { resolve } = require('path')

const ocr = new A11yCat.OCR()

async function main() {

    try {

        // Set the path to the pdf you want to OCR
        const pdfPath = resolve(process.cwd() + '/test/data/pdfs/set1/Modeling High-Frequency Limit Order Book Dynamics with Support Vector Machines.pdf')

        // Set a destination directory for the pdf images
        const destinationDir = resolve(process.cwd() + '/tmp')

        console.log('Generating images from PDF')
        // Convert a pdf to a series of images
        const generatedImages = await ocr.convertPdfToImages(pdfPath, destinationDir) 
       
        console.log('Doing OCR on first page')
        // Run OCR on one of the generated images
        await ocr.tess(generatedImages[0])

        // The OCR'd text will also be saved next to the source image

    } catch (error) {
        throw error
    }

}

main()

Tests

Tests are located in test/spec. Tests should use data from test/data/images and test/data/pdfs

Because there are some large PDFs in the test dataset, this can take a very long time depending on the host computer.

npm test