Add init_oem parameters in recognize() for switching modes

src/common/types.js

@@ -0,0 +1,17 @@
+module.exports = {
+  /*
+   * OEM = OCR Engine Mode, and there are 5 possible modes.
+   *
+   * By default tesseract.js uses DEFAULT mode, which uses LSTM when possible.
+   * If you need to use some tesseract v3 features (like tessedit_chars_whitelist),
+   * you need to use TESSERACT_ONLY mode.
+   *
+   */
+  OEM: {
+    TESSERACT_ONLY: 0,
+    LSTM_ONLY: 1,
+    TESSERACT_LSTM_COMBINED: 2,
+    DEFAULT: 3,
+    COUNT: 4,
+  },
+};

src/common/workerUtils.js

@@ -8,6 +8,7 @@
  * @author Jerome Wu <jeromewus@gmail.com>
  */
 const { readImage, loadLang } = require('tesseract.js-utils');
+const check = require('check-types');
 const dump = require('./dump');
 
 /*
@@ -122,13 +123,16 @@ const handleRecognize = ({
     .then(() => (
       loadLanguage({ lang, options }, res)
         .then(() => {
+          const OEM = check.undefined(params['init_oem'])
+            ? TessModule.OEM_DEFAULT
+            : params['init_oem'];
           const progressUpdate = (progress) => {
             res.progress({ status: 'initializing api', progress });
           };
           progressUpdate(0);
-          api.Init(null, lang);
+          api.Init(null, lang, OEM);
           progressUpdate(0.3);
-          Object.keys(params).forEach((key) => {
+          Object.keys(params).filter(key => !key.startsWith('init_')).forEach((key) => {
             api.SetVariable(key, params[key]);
           });
           progressUpdate(0.6);

src/index.js

@@ -9,10 +9,13 @@
  */
 const utils = require('tesseract.js-utils');
 const TesseractWorker = require('./common/TesseractWorker');
+const { OEM } = require('./common/types');
 
 module.exports = {
   /** Worker for OCR, @see common/TesseractWorker.js */
   TesseractWorker,
   /** Utilities for tesseract.js, @see {@link https://www.npmjs.com/package/tesseract.js-utils} */
   utils,
+  /** Check ./common/types for more details */
+  OEM,
 };