coverReader.js

const fs = require('fs')
const { createWorker } = require('tesseract.js')

module.exports = coverReader = () => {
  // sample is a JSON dowload from :
  // https://query.inventaire.io/#%23Editions%20without%20cover%0ASELECT%20%3Fcover%20%3Fiso%20WHERE%20%7B%0A%20%20%3Fedition%20wdt%3AP31%20wd%3AQ3331189%20.%0A%20%20%3Fedition%20wdt%3AP407%20%3Flang%20.%0A%20%20%3Fedition%20invp%3AP2%20%3Fcover%20.%0A%20%20SERVICE%20wdq%3Asparql%20%7B%20%3Flang%20wdt%3AP220%20%3Fiso%20.%20%7D%0A%7D%0ALIMIT%20100
  const file = "./img_samples2.json"
  const samples = JSON.parse(fs.readFileSync(file))
  sequentialRead(samples)
}

const sequentialRead = samples => {
  results = []
  const readNext = async results => {
    const nextSample = samples.shift()
    if (nextSample == null) {
      console.log("results : ",results)
      return results
    }
    await ocrRead(nextSample, results)
    return readNext(results)
  }

  readNext(results)
  return results
}

const ocrRead = async (sample, results) => {
  if (!results) { return }
  const { cover: coverUrl, iso } = sample
  worker = createWorker()
  await worker.load()
  await worker.loadLanguage(iso)
  await worker.initialize(iso)
  result = await worker.recognize(coverUrl)
  sample.text = result.data.text

  results.push(sample)
  await worker.terminate()
}

coverReader()