Revert "Upgrade to latest tesseract.js-utils"

This reverts commit de4b98ae23.

package-lock.json

@@ -8537,9 +8537,9 @@
       "integrity": "sha512-QmNgMA9m5ES5uMTqpOAPysrUA80vUx/6WKQlfkK3zhOeAgqv8DjwwcDv9tQv2TgRzOQ+LFKrJn94Y2rw5b2IGw=="
     },
     "tesseract.js-utils": {
-      "version": "1.0.0-beta.7",
-      "resolved": "https://registry.npmjs.org/tesseract.js-utils/-/tesseract.js-utils-1.0.0-beta.7.tgz",
-      "integrity": "sha512-MvBQNxVoueDg/8iN8jb26Tnj0lBRfmHVMcnaNdab+JeQKg/SuNXWMCpTD4D17+S6zeE9AAlVQymDmXH2/3vaMg==",
+      "version": "1.0.0-beta.6",
+      "resolved": "https://registry.npmjs.org/tesseract.js-utils/-/tesseract.js-utils-1.0.0-beta.6.tgz",
+      "integrity": "sha512-AENYhkqafwysayWmKtyApV0gR4abLJ426plLNHs/++8oHt+ekooyp77ew/q4+QLE7cbUDyxiNGawcraOWE/RuQ==",
       "requires": {
         "axios": "^0.18.0",
         "bmp-js": "^0.1.0",

package.json

@@ -5,7 +5,7 @@
   "main": "src/index.js",
   "scripts": {
     "start": "node scripts/server.js",
-    "build": "webpack --progress --config scripts/webpack.config.prod.js",
+    "build": "webpack --config scripts/webpack.config.prod.js",
     "prepublishOnly": "npm run build",
     "wait": "wait-on http://localhost:3000/package.json",
     "test": "npm-run-all -p -r start test:all",
@@ -54,7 +54,7 @@
     "node-fetch": "^2.3.0",
     "resolve-url": "^0.2.1",
     "tesseract.js-core": "^2.0.0-beta.10",
-    "tesseract.js-utils": "^1.0.0-beta.7"
+    "tesseract.js-utils": "^1.0.0-beta.6"
   },
   "repository": {
     "type": "git",

scripts/webpack.config.common.js

@@ -4,7 +4,6 @@ module.exports = {
       {
         test: /\.m?js$/,
         // exclude: /(node_modules|bower_components)/,
-        exclude: /(tesseract.js-core)/,
         use: {
           loader: 'babel-loader',
           options: {

scripts/webpack.config.dev.js

@@ -3,10 +3,9 @@ const webpack = require('webpack');
 const common = require('./webpack.config.common');
 
 const genConfig = ({
-  entry, filename, library, libraryTarget, ...config
+  entry, filename, library, libraryTarget,
 }) => ({
   ...common,
-  ...config,
   mode: 'development',
   entry,
   output: {
@@ -30,15 +29,10 @@ module.exports = [
     library: 'Tesseract',
     libraryTarget: 'umd',
   }),
-  genConfig({
-    entry: path.resolve(__dirname, '..', 'src', 'index.js'),
-    filename: 'tesseract.asm.dev.js',
-    library: 'Tesseract',
-    libraryTarget: 'umd',
-    resolve: {
-      alias: {
-        'tesseract.js-core/tesseract-core.wasm.js': 'tesseract.js-core/tesseract-core.asm.js',
-      },
-    },
-  }),
+  /*
+   *genConfig({
+   *  entry: path.resolve(__dirname, '..', 'src', 'browser', 'worker.js'),
+   *  filename: 'worker.dev.js',
+   *}),
+   */
 ];

scripts/webpack.config.prod.js

@@ -2,12 +2,11 @@ const path = require('path');
 const common = require('./webpack.config.common');
 
 const genConfig = ({
-  entry, filename, library, libraryTarget, ...config
+  entry, filename, library, libraryTarget,
 }) => ({
   ...common,
-  ...config,
   mode: 'production',
-  // devtool: 'source-map',
+  devtool: 'source-map',
   entry,
   output: {
     path: path.resolve(__dirname, '..', 'dist'),
@@ -24,15 +23,10 @@ module.exports = [
     library: 'Tesseract',
     libraryTarget: 'umd',
   }),
-  genConfig({
-    entry: path.resolve(__dirname, '..', 'src', 'index.js'),
-    filename: 'tesseract.asm.min.js',
-    library: 'Tesseract',
-    libraryTarget: 'umd',
-    resolve: {
-      alias: {
-        'tesseract.js-core/tesseract-core.wasm.js': 'tesseract.js-core/tesseract-core.asm.js',
-      },
-    },
-  }),
+  /*
+   *genConfig({
+   *  entry: path.resolve(__dirname, '..', 'src', 'browser', 'worker.js'),
+   *  filename: 'worker.min.js',
+   *}),
+   */
 ];

src/common/TesseractWorker.js

@@ -67,12 +67,12 @@ class TesseractWorker {
    * @function recognize text in given image
    * @access public
    * @param {Buffer, string} image - image to be recognized
-   * @param {string, array} [langs='eng'] - language to recognize
+   * @param {string} [lang=eng] - language to recognize
    * @param {object} params - tesseract parameters
    *
    */
-  recognize(image, langs = 'eng', params = {}) {
-    return this._sendJob('recognize', image, langs, params);
+  recognize(image, lang = 'eng', params = {}) {
+    return this._sendJob('recognize', image, lang, params);
   }
 
   /**
@@ -152,13 +152,13 @@ class TesseractWorker {
    * @param {string} lang language to recognize
    * @param {object} params tesseract parameters
    */
-  _sendJob(type, image, langs, params) {
+  _sendJob(type, image, lang, params) {
     return this._delay((job) => {
       job.send(
         type,
         {
           image,
-          langs,
+          lang,
           params,
           options: this.options,
         },

src/common/workerUtils.js

@@ -52,24 +52,16 @@ const setImage = (image) => {
   return data === null ? pix : data;
 };
 
-const getLangsStr = (langs) => {
-  if (typeof langs === 'string') {
-    return langs;
-  }
-
-  return langs.map(lang => (typeof lang === 'string' ? lang : lang.code)).join('+');
-};
-
 /**
  * handleParams
  *
  * @name handleParams
  * @function hanlde params from users
  * @access private
- * @param {string} langs - lang string for Init()
+ * @param {string} lang - lang string for Init()
  * @param {object} customParams - an object of params
  */
-const handleParams = (langs, customParams) => {
+const handleParams = (lang, customParams) => {
   const {
     tessedit_ocr_engine_mode,
     ...params
@@ -77,7 +69,7 @@ const handleParams = (langs, customParams) => {
     ...defaultParams,
     ...customParams,
   };
-  api.Init(null, getLangsStr(langs), tessedit_ocr_engine_mode);
+  api.Init(null, lang, tessedit_ocr_engine_mode);
   Object.keys(params).forEach((key) => {
     api.SetVariable(key, params[key]);
   });
@@ -166,14 +158,14 @@ const handleInit = ({ corePath }, res) => {
  * @function load language from remote or local cache
  * @access public
  * @param {object} req - job payload
- * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
+ * @param {string} req.lang - languages to load, ex: eng, eng+chi_tra
  * @param {object} req.options - other options for loadLang function
  * @param {object} res - job instance
  * @returns {Promise} A Promise for callback
  */
-const loadLanguage = ({ langs, options }, res) => {
+const loadLanguage = ({ lang, options }, res) => {
   res.progress({ status: 'loading language traineddata', progress: 0 });
-  return loadLang({ langs, TessModule, ...options }).then((...args) => {
+  return loadLang({ lang, TessModule, ...options }).then((...args) => {
     res.progress({ status: 'loaded language traineddata', progress: 1 });
     return args;
   });
@@ -187,17 +179,17 @@ const loadLanguage = ({ langs, options }, res) => {
  * @access public
  * @param {object} req - job payload
  * @param {array} req.image - binary image in array format
- * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
+ * @param {string} req.lang - languages to load, ex: eng, eng+chi_tra
  * @param {object} req.options - other options for loadLang function
  * @param {object} req.params - parameters for tesseract
  * @param {object} res - job instance
  */
 const handleRecognize = ({
-  image, langs, options, params,
+  image, lang, options, params,
 }, res) => (
   handleInit(options, res)
     .then(() => (
-      loadLanguage({ langs, options }, res)
+      loadLanguage({ lang, options }, res)
         .catch((e) => {
           if (e instanceof DOMException) {
             /*
@@ -214,7 +206,7 @@ const handleRecognize = ({
             res.progress({ status: 'initializing api', progress });
           };
           progressUpdate(0);
-          handleParams(langs, params);
+          handleParams(lang, params);
           progressUpdate(0.5);
           const ptr = setImage(image);
           progressUpdate(1);
@@ -236,18 +228,18 @@ const handleRecognize = ({
  * @access public
  * @param {object} req - job payload
  * @param {array} req.image - binary image in array format
- * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
+ * @param {string} req.lang - languages to load, ex: eng, eng+chi_tra
  * @param {object} req.options - other options for loadLang function
  * @param {object} res - job instance
  */
 const handleDetect = ({
-  image, langs, options,
+  image, lang, options,
 }, res) => (
   handleInit(options, res)
     .then(() => (
-      loadLanguage({ langs, options }, res)
+      loadLanguage({ lang, options }, res)
         .then(() => {
-          api.Init(null, getLangsStr(langs));
+          api.Init(null, lang);
           api.SetPageSegMode(TessModule.PSM_OSD_ONLY);
 
           const ptr = setImage(image);

tests/detect.test.js

@@ -1,4 +1,5 @@
-const { TesseractWorker } = Tesseract;
+const { TesseractWorker, utils: { loadLang } } = Tesseract;
+const isBrowser = typeof window !== 'undefined' && typeof window.document !== 'undefined';
 const IMAGE_PATH = 'http://localhost:3000/tests/assets/images';
 const loadLangOptions = {
   langPath: 'http://localhost:3000/tests/assets/traineddata',
@@ -8,6 +9,7 @@ const loadLangOptions = {
 const getWorker = options => (
   new TesseractWorker({
     cacheMethod: 'readOnly',
+    ...(isBrowser ? { workerPath: 'http://localhost:3000/dist/worker.dev.js' } : {}),
     ...loadLangOptions,
     ...options,
   })

tests/recognize.test.js

@@ -1,5 +1,6 @@
-const { TesseractWorker } = Tesseract;
+const { TesseractWorker, utils: { loadLang } } = Tesseract;
 
+const isBrowser = typeof window !== 'undefined' && typeof window.document !== 'undefined';
 const IMAGE_PATH = 'http://localhost:3000/tests/assets/images';
 const SIMPLE_TEXT = 'Tesseract.js\n';
 const COMSIC_TEXT = 'HellO World\nfrom beyond\nthe Cosmic Void\n';
@@ -15,6 +16,7 @@ const loadLangOptions = {
 const getWorker = options => (
   new TesseractWorker({
     cacheMethod: 'readOnly',
+    ...(isBrowser ? { workerPath: 'http://localhost:3000/dist/worker.dev.js' } : {}),
     ...loadLangOptions,
     ...options,
   })
@@ -131,7 +133,7 @@ describe('recognize()', () => {
       }).timeout(10000)
     ));
   });
-
+  
   (isBrowser ? describe : describe.skip)('should read image from video DOM element (browser only)', () => {
     FORMATS.forEach(format => (
       it(`support ${format} format`, (done) => {
@@ -158,23 +160,22 @@ describe('recognize()', () => {
     let canvasDOM = null;
     let imageDOM = null;
     let idx = 0;
-    beforeEach((done) => {
+    beforeEach(function cb(done) {
       canvasDOM = document.createElement('canvas');
       imageDOM = document.createElement('img');
       imageDOM.setAttribute('crossOrigin', 'Anonymous');
       imageDOM.onload = () => {
         canvasDOM.getContext('2d').drawImage(imageDOM, 0, 0);
         done();
-      };
-      imageDOM.setAttribute('src', `${IMAGE_PATH}/simple.${formats[idx]}`);
-      idx += 1;
+      }
+      imageDOM.setAttribute('src', `${IMAGE_PATH}/simple.${formats[idx++]}`);
     });
 
     afterEach(() => {
       canvasDOM.remove();
       imageDOM.remove();
     });
-
+    
     formats.forEach(format => (
       it(`support ${format} format`, (done) => {
         const worker = getWorker();
@@ -188,4 +189,5 @@ describe('recognize()', () => {
       }).timeout(10000)
     ));
   });
+
 });