jws-frontend/src/utils/mrz.ts

import moment from 'moment';

type MRZ = {
  type: 'TD1' | 'TD2' | 'TD3';
  zone: string[];
};

type Field = {
  field: string;
  format?: (value: string, obj?: Record<string, string>) => string;
};
type FieldList = Record<string, Field>;

const DEFAULT_FIELD = {
  documentType: { field: 'doc_type' },
  documentNo: { field: 'doc_number' },
  documentNoCheck: { field: 'doc_number_check' },
  name: {
    field: 'full_name',
    format: (value, _) => value.replace(/0/, 'O'),
  },
  country: {
    field: 'country',
    format: (value, _) => value.replace(/0/, 'O'),
  },
  nationality: {
    field: 'nationality',
    format: (value, _) => value.replace(/0/, 'O'),
  },
  gender: { field: 'sex' },
  birthDate: {
    field: 'birth_date',
    format: (value, _) => moment(value, 'YYMMDD').format('YYYY-MM-DD'),
  },
  birthDateCheck: { field: 'birth_date_check' },
  expireDate: {
    field: 'expire_date',
    format: (value, _) => moment(value, 'YYMMDD').format('YYYY-MM-DD'),
  },
  expireDateCheck: { field: 'expire_date_check' },
  optionalData: { field: 'optional_data' },
  optionalDataCheck: { field: 'optional_data_check' },
  lineCheck: { field: 'line_check' },
} satisfies FieldList;

const MRZ_TD_1 = [
  new RegExp(
    [
      `(?<${DEFAULT_FIELD.documentType.field}>[0-9A-Z<]{2})`,
      `(?<${DEFAULT_FIELD.country.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.documentNo.field}>[0-9A-Z<]{9})`,
      `(?<${DEFAULT_FIELD.documentNoCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.optionalData.field}>[0-9A-Z<]{15})`,
    ].join(''),
  ),
  new RegExp(
    [
      `(?<${DEFAULT_FIELD.birthDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.birthDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.gender.field}>[mfMF<]{1})`,
      `(?<${DEFAULT_FIELD.expireDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.expireDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.nationality.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.optionalData.field}>[A-Z0-9<]{11})`,
      `(?<${DEFAULT_FIELD.lineCheck.field}>[0-9A-Z<]{1})`,
    ].join(''),
  ),
  new RegExp([`(?<${DEFAULT_FIELD.name.field}>[A-Z<]{30})`].join('')),
];
const MRZ_TD_2 = [
  new RegExp(
    [
      `(?<${DEFAULT_FIELD.documentType.field}>[0-9A-Z<]{2})`,
      `(?<${DEFAULT_FIELD.country.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.name.field}>[A-Z<]{31})`,
    ].join(''),
  ),

  new RegExp(
    [
      `(?<${DEFAULT_FIELD.documentNo.field}>[0-9A-Z<]{9})`,
      `(?<${DEFAULT_FIELD.documentNoCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.nationality.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.birthDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.birthDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.gender.field}>[mfMF]{1})`,
      `(?<${DEFAULT_FIELD.expireDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.expireDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.optionalData.field}>[A-Z0-9<]{7})`,
      `(?<${DEFAULT_FIELD.lineCheck.field}>[0-9A-Z<]{1})`,
    ].join(''),
  ),
];

const MRZ_TD_3 = [
  new RegExp(
    [
      `(?<${DEFAULT_FIELD.documentType.field}>[A-Z0-9<]{2})`,
      `(?<${DEFAULT_FIELD.country.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.name.field}>[A-Z0-9<]{39})`,
    ].join(''),
  ),

  new RegExp(
    [
      `(?<${DEFAULT_FIELD.documentNo.field}>[0-9A-Z<]{9})`,
      `(?<${DEFAULT_FIELD.documentNoCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.nationality.field}>[0-9A-Z<]{3})`,
      `(?<${DEFAULT_FIELD.birthDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.birthDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.gender.field}>[mfMF<]{1})`,
      `(?<${DEFAULT_FIELD.expireDate.field}>[0-9A-Z<]{6})`,
      `(?<${DEFAULT_FIELD.expireDateCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.optionalData.field}>[A-Z0-9<]{14})`,
      `(?<${DEFAULT_FIELD.optionalDataCheck.field}>[0-9A-Z<]{1})`,
      `(?<${DEFAULT_FIELD.lineCheck.field}>[0-9A-Z<]{1})`,
    ].join(''),
  ),
];

function mrzCleanResult(obj: Record<string, string>) {
  Object.entries(obj).forEach(([k, v]) => {
    obj[k] = v
      .replace(/</g, ' ')
      .replace(/\s{2,}/, ' ')
      .trim();
  });

  const original = structuredClone(obj);
  for (const value of Object.values(DEFAULT_FIELD)) {
    if (obj[value.field] && 'format' in value) {
      obj[value.field] = value.format(obj[value.field], original);
    }
  }

  return obj;
}

function mrzFieldExtract(expression: RegExp, line: string) {
  if (expression.test(line)) {
    return expression.exec(line)?.groups || {};
  }
  return {};
}

export function checkSum(data: string) {
  if (!/[0-9A-Z<]/.test(data)) return null;

  const sum = data.split('').reduce((a, v, i) => {
    if (v === '<') return a;

    const num = Number(v);
    const weight = [7, 3, 1][i % 3];

    if (Number.isNaN(num)) {
      return a + (v.charCodeAt(0) - 55) * weight;
    }
    return a + num * weight;
  }, 0);
  return sum % 10;
}

export function parseType1(mrz: MRZ) {
  const result: Record<string, string> = {};

  mrz.zone.forEach((line, i) =>
    Object.assign(result, mrzFieldExtract(MRZ_TD_1[i], line)),
  );

  return { mrz, result: mrzCleanResult(result) };
}

export function parseType2(mrz: MRZ) {
  const result: Record<string, string> = {};

  mrz.zone.forEach((line, i) =>
    Object.assign(result, mrzFieldExtract(MRZ_TD_2[i], line)),
  );

  return { mrz, result: mrzCleanResult(result) };
}

export function parseType3(mrz: MRZ) {
  const result: Record<string, string> = {};

  mrz.zone.forEach((line, i) =>
    Object.assign(result, mrzFieldExtract(MRZ_TD_3[i], line)),
  );

  return { mrz, result: mrzCleanResult(result) };
}

export function parseMRZ(mrz: MRZ) {
  if (mrz.type === 'TD1') return parseType1(mrz);
  if (mrz.type === 'TD2') return parseType2(mrz);
  if (mrz.type === 'TD3') return parseType3(mrz);
  return null;
}