jws-frontend/src/utils/ocr.ts

import { createWorker, ImageLike, RecognizeResult, Worker } from 'tesseract.js';

import { parseMRZ } from './mrz';

let worker: Worker;

export function runOcr(image: ImageLike): Promise<RecognizeResult>;
export function runOcr<
  CallbackFunction extends (result: RecognizeResult) => void,
>(
  image: ImageLike,
  callback: CallbackFunction,
): Promise<ReturnType<CallbackFunction>>;
export async function runOcr<T extends (result: RecognizeResult) => void>(
  image: ImageLike,
  callback?: T,
) {
  if (!worker) {
    worker = await createWorker(['ocrb', 'eng'], 1, {
      langPath: '/ocr-data',
    });
  }

  if (callback) {
    return callback(await worker.recognize(image));
  }
  return await worker.recognize(image);
}

export function parseResultMRZ(result: RecognizeResult) {
  const zone = result.data.text
    .split(/[\s\r\n]+/)
    .filter((v) => /[A-Z0-9<]{30,}/.test(v));

  if (zone.length === 2 && zone[0].length >= 44) {
    return parseMRZ({ type: 'TD3', zone });
  } else if (zone.length === 2 && zone[0].length >= 36) {
    return parseMRZ({ type: 'TD2', zone });
  } else if (zone.length === 3 && zone[0].length >= 30) {
    return parseMRZ({ type: 'TD1', zone });
  } else {
    return { mrz: { type: 'UNKNOWN', zone }, result: null };
  }
}