Fix Module async issue and rewrite src/node/index.js

src/common/worker.js

@@ -52,15 +52,20 @@ function handleInit(req, res){
 
       res.progress({ status: 'initializing tesseract', progress: 0 })
 
-        Module = Core({
-            TOTAL_MEMORY: MIN_MEMORY,
+      return Core({
+        // TOTAL_MEMORY: MIN_MEMORY,
         TesseractProgress(percent){
           latestJob.progress({ status: 'recognizing text', progress: Math.max(0, (percent-30)/70) });
         },
-        });
+      })
+        .then((TessModule) => {
+          Module = TessModule;
           base = new Module.TessBaseAPI();
           res.progress({ status: 'initializing tesseract', progress: 1 });
+        });
     }
+
+  return new Promise();
 }
 
 function setImage(Module, base, image) {
@@ -93,8 +98,8 @@ function loadLanguage(req, res, cb){
 
 
 function handleRecognize(req, res){
-    handleInit(req, res);
-
+  handleInit(req, res)
+    .then(() => {
       loadLanguage(req, res, () => {
         var options = req.options;
 
@@ -125,11 +130,13 @@ function handleRecognize(req, res){
 
         res.resolve(result);
       })
+    });
 }
 
 
 function handleDetect(req, res){
-    handleInit(req, res);
+  handleInit(req, res)
+    .then(() => {
       req.options.lang = 'osd';
       loadLanguage(req, res, () => {
         base.Init(null, 'osd');
@@ -159,4 +166,5 @@ function handleDetect(req, res){
           });
         }
       });
+    });
 }

src/index.js

@@ -75,10 +75,10 @@ class TesseractWorker {
   _recv(packet) {
     if (this._currentJob.id === packet.jobId) {
       this._currentJob._handle({
+        ...packet,
         data: packet.status === 'resolve' && packet.action === 'recognize'
           ? circularize(packet.data)
           : packet.data,
-        ...packet,
       });
     } else {
       console.warn(`Job ID ${packet.jobId} not known.`);

src/node/index.js

@@ -1,78 +1,56 @@
-const fetch = require('isomorphic-fetch'),
-    isURL = require('is-url'),
-    fork = require('child_process').fork,
-    fs = require('fs');
+const check = require('check-types');
+const fetch = require('isomorphic-fetch');
+const isURL = require('is-url');
+const { fork } = require('child_process');
+const fs = require('fs');
+const path = require('path');
+const fileType = require('file-type');
+const PNGReader = require('png.js');
+const JPGReader = require('jpeg-js');
 
-exports.defaultOptions = {
-    workerPath: require('path').join(__dirname, 'worker.js'),
-    langPath: 'https://cdn.jsdelivr.net/gh/naptha/tessdata@gh-pages/4.0.0/',
-}
-
-exports.spawnWorker = function spawnWorker(instance, workerOptions){
-    var cp = fork(workerOptions.workerPath);
-    cp.on('message', packet => {
-        instance._recv(packet);
-    });
-    return cp;
-}
-
-exports.terminateWorker = function(instance){
-    instance.worker.kill();
-}
-
-exports.sendPacket = function sendPacket(instance, packet){
-    loadImage(packet.payload.image, img => {
-        packet.payload.image = img;
-        instance.worker.send(packet);
-    });
-}
-
-
-function loadImage(image, cb){
-    
-    if(typeof image === 'string'){
+function loadImage(image, cb) {
+  if (check.string(image)) {
     if (isURL(image)) {
       fetch(image)
         .then(resp => resp.buffer())
         .then(buffer => loadImage(buffer, cb))
         .catch(err => console.error(err));
     } else {
-            fs.readFile(image, function(err, buffer){
+      fs.readFile(image, (err, buffer) => {
         if (err) throw err;
         loadImage(buffer, cb);
       });
     }
     return;
-    } else if (image instanceof Buffer){
-        var mime = require('file-type')(image).mime
+  }
+  if (image instanceof Buffer) {
+    const { mime } = fileType(image);
 
-        if(mime === 'image/png'){
-            var PNGReader = require('png.js');
-            var reader = new PNGReader(image);
-            reader.parse(function(err, png){
+    if (mime === 'image/png') {
+      const reader = new PNGReader(image);
+      reader.parse((err, png) => {
         if (err) throw err;
 
-                var image = {
-                    width: png.getWidth(),
-                    height: png.getHeight()
-                }
-                image.data = new Uint8Array(image.width * image.height * 4)
-                for(var j = 0; j < image.height; j++){
-                    for(var i = 0; i < image.width; i++){
-                        var offset = 4 * (i + j * image.width),
-                            pix = png.getPixel(i, j);
+        const width = png.getWidth();
+        const height = png.getHeight();
+        const data = new Uint8Array(width * height * 4);
+
+        for (let j = 0; j < height; j += 1) {
+          for (let i = 0; i < width; i += 1) {
+            const offset = 4 * (i + (j * width));
+            const pix = png.getPixel(i, j);
 
-                        image.data[offset] = pix[0];
-                        image.data[offset + 1] = pix[1];
-                        image.data[offset + 2] = pix[2];
-                        image.data[offset + 3] = pix[3];
+            Array(4).fill(0).forEach((v, idx) => {
+              data[offset + idx] = pix[idx];
+            });
           }
         }
-                loadImage(image, cb);
+        loadImage({ width, height, data }, cb);
       });
       return;
-        } else if (mime === 'image/jpeg'){
-            loadImage(require('jpeg-js').decode(image), cb);
+    }
+    if (mime === 'image/jpeg') {
+      loadImage(JPGReader.decode(image), cb);
       return;
     }
 
@@ -81,9 +59,38 @@ function loadImage(image, cb){
 
   // node uses json.stringify for ipc which means we need to turn
   // fancy arrays into raw arrays
-    if(image && image.data && image.data.length && !Array.isArray(image.data)){
-        image.data = Array.from(image.data);
-        return loadImage(image, cb)
+  if (image && image.data && image.data.length && !Array.isArray(image.data)) {
+    loadImage({ ...image, data: Array.from(image.data) }, cb);
+    return;
   }
   cb(image);
 }
+
+exports.defaultOptions = {
+  workerPath: path.join(__dirname, 'worker.js'),
+  langPath: 'https://cdn.jsdelivr.net/gh/naptha/tessdata@gh-pages/4.0.0/',
+};
+
+exports.spawnWorker = (instance, { workerPath }) => {
+  const cp = fork(workerPath);
+  cp.on('message', (packet) => {
+    instance._recv(packet);
+  });
+  return cp;
+};
+
+exports.terminateWorker = (instance) => {
+  instance.worker.kill();
+};
+
+exports.sendPacket = (instance, packet) => {
+  loadImage(packet.payload.image, (img) => {
+    instance.worker.send({
+      ...packet,
+      payload: {
+        ...packet.payload,
+        image: img,
+      },
+    });
+  });
+};

src/node/worker.js

@@ -11,7 +11,7 @@ workerUtils.setAdapter({
   getCore: (req, res) => {
     if (check.null(TesseractCore)) {
       res.progress({ status: 'loading tesseract core' });
-      TesseractCore = require('tesseract.js-core/src/tesseract-core.asm');
+      TesseractCore = require('tesseract.js-core');
       res.progress({ status: 'loaded tesseract core' });
     }
     return TesseractCore;