feat: add ocr function

src/utils/mrz.ts

@@ -0,0 +1,135 @@
+type MRZ = {
+  type: 'TD1' | 'TD2' | 'TD3';
+  zone: string[];
+};
+
+const MRZ_TD_1 = [
+  new RegExp(
+    [
+      '(?<doc_type>[0-9A-Z<]{1})',
+      '(?<doc_subtype>[A-Z<]{1})',
+      '(?<country>[0-9A-Z<]{3})',
+      '(?<doc_number>[0-9A-Z<]{9})',
+      '(?<doc_number_check>[0-9A-Z<]{1})',
+      '(?<complement>[0-9A-Z<]{15})',
+    ].join(''),
+  ),
+  new RegExp(
+    [
+      '(?<birth_date>[0-9A-Z<]{6})',
+      '(?<birth_date_check>[0-9A-Z<]{1})',
+      '(?<sex>[mfMF<]{1})',
+      '(?<expire_date>[0-9A-Z<]{6})',
+      '(?<expire_date_check>[0-9A-Z<]{1})',
+      '(?<nationality>[0-9A-Z<]{3})',
+      '(?<optional_data>[A-Z0-9<]{11})',
+      '(?<linecheck>[0-9A-Z<]{1})',
+    ].join(''),
+  ),
+  new RegExp(['(?<full_name>[A-Z<]{30})'].join('')),
+];
+const MRZ_TD_2 = [
+  new RegExp(
+    [
+      '(?<doc_type>[0-9A-Z<]{1})',
+      '(?<doc_subtype>[A-Z<]{1})',
+      '(?<country>[0-9A-Z<]{3})',
+      '(?<full_name>[A-Z<]{31})',
+    ].join(''),
+  ),
+
+  new RegExp(
+    [
+      '(?<doc_number>[0-9A-Z<]{9})',
+      '(?<doc_numbercheck>[0-9A-Z<]{1})',
+      '(?<nacionality>[0-9A-Z<]{3})',
+      '(?<birth_date>[0-9A-Z<]{6})',
+      '(?<birth_date_check>[0-9A-Z<]{1})',
+      '(?<sex>[mfMF]{1})',
+      '(?<expire_date>[0-9A-Z<]{6})',
+      '(?<expire_date_check>[0-9A-Z<]{1})',
+      '(?<optional_data>[A-Z0-9<]{7})',
+      '(?<line_check>[0-9A-Z<]{1})',
+    ].join(''),
+  ),
+];
+
+const MRZ_TD_3 = [
+  new RegExp(
+    [
+      '(?<doc_type>[A-Z0-9<]{1})',
+      '(?<doc_subtype>[A-Z0-9<]{1})',
+      '(?<country>[A-Z0-9]{3})',
+      '(?<full_name>[A-Z0-9<]{39})',
+    ].join(''),
+  ),
+
+  new RegExp(
+    [
+      '(?<doc_number>[0-9A-Z<]{9})',
+      '(?<doc_number_check>[0-9A-Z<]{1})',
+      '(?<nationality>[0-9A-Z<]{3})',
+      '(?<birth_date>[0-9A-Z<]{6})',
+      '(?<birth_date_check>[0-9A-Z<]{1})',
+      '(?<sex>[mfMF<]{1})',
+      '(?<expire_date>[0-9A-Z<]{6})',
+      '(?<expire_date_check>[0-9A-Z<]{1})',
+      '(?<personal_number>[A-Z0-9<]{14})',
+      '(?<personal_number_check>[0-9A-Z<]{1})',
+      '(?<linecheck>[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();
+  });
+  return obj;
+}
+
+export function parseType1(mrz: MRZ) {
+  const result: Record<string, string> = {};
+
+  mrz.zone.forEach((line, i) => {
+    if (MRZ_TD_1[i].test(line)) {
+      Object.assign(result, MRZ_TD_1[i].exec(line)?.groups);
+    }
+  });
+
+  return { mrz, result: mrzCleanResult(result) };
+}
+
+export function parseType2(mrz: MRZ) {
+  const result: Record<string, string> = {};
+
+  mrz.zone.forEach((line, i) => {
+    if (MRZ_TD_2[i].test(line)) {
+      Object.assign(result, MRZ_TD_2[i].exec(line)?.groups);
+    }
+  });
+
+  return { mrz, result: mrzCleanResult(result) };
+}
+
+export function parseType3(mrz: MRZ) {
+  const result: Record<string, string> = {};
+
+  mrz.zone.forEach((line, i) => {
+    if (MRZ_TD_3[i].test(line)) {
+      Object.assign(result, MRZ_TD_3[i].exec(line)?.groups);
+    }
+  });
+
+  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;
+}

src/utils/ocr.ts

@@ -0,0 +1,38 @@
+import { createWorker, ImageLike, RecognizeResult } from 'tesseract.js';
+
+import { parseMRZ } from './mrz';
+
+let worker: Tesseract.Worker;
+
+export function runOcr(image: ImageLike): Promise<void>;
+export function runOcr<T extends (result: RecognizeResult) => void>(
+  image: ImageLike,
+  callback: T,
+): Promise<ReturnType<T>>;
+export async function runOcr<T extends (result: RecognizeResult) => void>(
+  image: ImageLike,
+  callback?: T,
+) {
+  if (!worker) {
+    worker = await createWorker(['ocrb', 'eng', 'khm'], 1, {
+      langPath: '/ocr-data',
+    });
+  }
+  if (callback) return callback(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 === 3 && zone[0].length === 30) {
+    return parseMRZ({ type: 'TD1', zone });
+  } else if (zone.length === 2 && zone[0].length === 36) {
+    return parseMRZ({ type: 'TD2', zone });
+  } else if (zone.length === 2 && zone[0].length === 44) {
+    return parseMRZ({ type: 'TD3', zone });
+  } else {
+    return null;
+  }
+}