rewrite

README.md

@@ -1 +1,18 @@
 # tesseract.js
+Tesseract.js is a pure javascript version of the Tesseract OCR Engine that can recognize English, Chinese, Russian, and 60 other languages.
+
+<!-- ![alt text]( "Logo Title Text 1") -->
+
+# Installation
+Tesseract.js works with a `<script>` tag, or with `npm` (if you're using webpack /browserify).
+
+## `<script/>`
+
+```html
+<script src='' />
+```
+
+## npm 
+```shell
+npm install tesseract
+```

bify

@@ -1 +0,0 @@
-browserify src/worker.js -o inter/worker.js

devServer.js

@@ -0,0 +1,26 @@
+var path = require('path');
+var express = require('express');
+var webpack = require('webpack');
+var config = require('./webpack.config.dev');
+
+var app = express();
+var compiler = webpack(config);
+
+app.use(require('webpack-dev-middleware')(compiler, {
+  noInfo: true,
+  publicPath: config[0].output.publicPath
+}));
+
+// app.use(require('webpack-hot-middleware')(compiler));
+
+app.use('/', express.static('./'));
+
+var port = 7355
+app.listen(port, 'localhost', function(err) {
+  if (err) {
+    console.log(err);
+    return;
+  }
+
+  console.log('Listening at http://localhost:' + port);
+});

dist/tesseract.js

@@ -0,0 +1 @@
+!function(e,t){"object"==typeof exports&&"object"==typeof module?module.exports=t():"function"==typeof define&&define.amd?define([],t):"object"==typeof exports?exports.Tesseract=t():e.Tesseract=t()}(this,function(){return function(e){function t(n){if(r[n])return r[n].exports;var o=r[n]={exports:{},id:n,loaded:!1};return e[n].call(o.exports,o,o.exports,t),o.loaded=!0,o.exports}var r={};return t.m=e,t.c=r,t.p="",t(0)}([function(e,t){"use strict";e.exports=function(){function e(e,t){var r=a++;s[r]={};var n={jobId:r,action:e,args:t};return console.log(n),o.postMessage(n),{then:function(e){return s[r].result=e,this},error:function(e){return s[r].error=e,this},progress:function(e){return s[r].progress=e,this}}}function t(e){if(e.getContext)e=e.getContext("2d");else if("IMG"==e.tagName||"VIDEO"==e.tagName){var t=document.createElement("canvas");t.width=e.naturalWidth||e.videoWidth,t.height=e.naturalHeight||e.videoHeight;var r=t.getContext("2d");r.drawImage(e,0,0),e=r}return e.getImageData&&(e=e.getImageData(0,0,e.canvas.width,e.canvas.height)),e}var r=arguments.length>0&&void 0!==arguments[0]?arguments[0]:location.href+"build/tesseract.worker.js",n=new Blob(["importScripts('"+r+"');"]),o=new Worker(window.URL.createObjectURL(n)),i=!1,a=0,s={};return o.onmessage=function(e){var t=e.data,r=t.jobId,n=t.progress,o=t.error,i=t.result,a=s[r];n&&a.progress&&a.progress(n),o&&a.error&&a.error(o),i&&a.result&&a.result(i)},e("init",{mem:100663296}),{detect:function(r){return e("detect",{image:t(r)})},recognize:function(r){var n=arguments.length>1&&void 0!==arguments[1]?arguments[1]:"eng";return"string"==typeof n?n={lang:n}:n.lang=n.lang||"eng",i||["chi_sim","chi_tra","jpn"].indexOf(n.lang)==-1||(e("init",{mem:167772160}),i=!0),e("recognize",{options:n,image:t(r)})}}}}])});

index.html

@@ -1,5 +1,5 @@
 <canvas id="c"></canvas>
-<script type="text/javascript" src="./lib/Tesseract.js"></script>
+<script type="text/javascript" src="./build/tesseract.js"></script>
 <script type="text/javascript">
 	var canvas = document.getElementById('c')
 	canvas.width = 400
@@ -14,11 +14,16 @@
 	ctx.font = '30px sans-serif'
 	ctx.fillText('the Cosmic Void', 100, 120)
 
-	Tesseract.recognize(canvas,{
+	var tesseract = Tesseract();
-		tessedit_char_blacklist:'e',
+
-		progress: function(e){
+	tesseract.recognize(canvas, {
-			console.log(e)
+		tessedit_char_blacklist: 'e'
-		}
+	})
-	}).then( function(d){ console.log(d) } )
+	.progress(function(e){
+		console.log('progress', e)
+	})
+	.then(function(e){
+		console.log('result', e)
+	})
 
 </script>

lib/Tesseract.2015.07.26.js

@@ -1,130 +0,0 @@
-var Tesseract = (function(){
-
-	var Tesseract = {}
-
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-	//https://rawgit.com/naptha/tesseract.js/master/worker/worker.js   for testing
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-
-	var blob = new Blob(["importScripts('https://cdn.rawgit.com/naptha/tesseract.js/master/lib/worker.2015.07.26.js');"]); // changed on build
-	// console.log('localhost')
-	var worker = new Worker(window.URL.createObjectURL(blob));
-	worker.postMessage({init: {mem: 16777216*6}})
-	var bigworker = false
-
-	var index = 0
-	var handlers = []
-
-	worker.onmessage = function(e){
-		var handler = handlers[e.data.index]
-		if(e.data.progress){
-			handler.progress(e.data.progress)
-		}
-		else if(e.data.err){
-			handler.reject(e.data.err)
-			handler.callback(e.data.err)
-		}
-		else {
-			handler.resolve(e.data.result)
-			handler.callback(null,e.data.result)
-		}
-	}
-
-	function convertToImageData(image){
-		if(image.getContext){
-			image = image.getContext('2d');
-		}else if(image.tagName == "IMG" || image.tagName == "VIDEO"){
-			var c = document.createElement('canvas');
-			if(image.tagName == "IMG"){
-				c.width  = image.naturalWidth;
-				c.height = image.naturalHeight;
-			}else if(image.tagName == "VIDEO"){
-				c.width  = image.videoWidth;
-				c.height = image.videoHeight;
-			}
-			var ctx = c.getContext('2d');
-			ctx.drawImage(image, 0, 0);
-			image = ctx;
-		}
-		if(image.getImageData) image = image.getImageData(0, 0, image.canvas.width, image.canvas.height);
-		return image	
-	}
-
-	Tesseract.detect = function(image, progress, callback){
-		image = convertToImageData(image)
-
-		if(typeof progress === "undefined"){
-			progress = callback = new Function()
-		}
-
-		if (typeof callback === "undefined"){
-			callback = progress
-			progress = new Function()
-		}
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback
-		handlers[i].progress = progress
-		
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'detect', image: image})
-		})
-
-	}
-
-	Tesseract.recognize = function(image, options, callback){
-		var lang = options.lang
-		if (typeof lang === "undefined"){
-			lang = 'eng'
-		}
-
-		if (!bigworker && ['chi_sim', 'chi_tra', 'jpn'].indexOf(lang) != -1){
-			worker.postMessage({init: {mem: 16777216*10}})
-			bigworker = true
-			console.log('started big worker')
-		}
-
-		if (typeof options === 'string') {
-			lang = options
-			options = {}
-		}
-
-		if (typeof options === "function") {
-			callback = options
-			options = {}
-		}
-
-		image = convertToImageData(image)
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback || new Function()
-		handlers[i].progress = (function(){
-			if(typeof options.progress === 'function'){
-				var p = options.progress
-				delete options.progress
-				return p
-			}
-			return function(){}
-		})()
-
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'recognize', image: image, lang: lang, options: options})
-		})
-	
-	}
-	return Tesseract
-})()

lib/Tesseract_dev.js

@@ -1,130 +0,0 @@
-var Tesseract = (function(){
-
-	var Tesseract = {}
-
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-	//https://rawgit.com/naptha/tesseract.js/master/worker/worker.js   for testing
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-
-	var blob = new Blob(["importScripts('http://localhost:1234/master/lib/worker.2015.07.26.js');"]); // changed on build
-	// console.log('localhost')
-	var worker = new Worker(window.URL.createObjectURL(blob));
-	worker.postMessage({init: {mem: 16777216*6}})
-	var bigworker = false
-
-	var index = 0
-	var handlers = []
-
-	worker.onmessage = function(e){
-		var handler = handlers[e.data.index]
-		if(e.data.progress){
-			handler.progress(e.data.progress)
-		}
-		else if(e.data.err){
-			handler.reject(e.data.err)
-			handler.callback(e.data.err)
-		}
-		else {
-			handler.resolve(e.data.result)
-			handler.callback(null,e.data.result)
-		}
-	}
-
-	function convertToImageData(image){
-		if(image.getContext){
-			image = image.getContext('2d');
-		}else if(image.tagName == "IMG" || image.tagName == "VIDEO"){
-			var c = document.createElement('canvas');
-			if(image.tagName == "IMG"){
-				c.width  = image.naturalWidth;
-				c.height = image.naturalHeight;
-			}else if(image.tagName == "VIDEO"){
-				c.width  = image.videoWidth;
-				c.height = image.videoHeight;
-			}
-			var ctx = c.getContext('2d');
-			ctx.drawImage(image, 0, 0);
-			image = ctx;
-		}
-		if(image.getImageData) image = image.getImageData(0, 0, image.canvas.width, image.canvas.height);
-		return image	
-	}
-
-	Tesseract.detect = function(image, progress, callback){
-		image = convertToImageData(image)
-
-		if(typeof progress === "undefined"){
-			progress = callback = new Function()
-		}
-
-		if (typeof callback === "undefined"){
-			callback = progress
-			progress = new Function()
-		}
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback
-		handlers[i].progress = progress
-		
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'detect', image: image})
-		})
-
-	}
-
-	Tesseract.recognize = function(image, options, callback){
-		var lang = options.lang
-		if (typeof lang === "undefined"){
-			lang = 'eng'
-		}
-
-		if (!bigworker && ['chi_sim', 'chi_tra', 'jpn'].indexOf(lang) != -1){
-			worker.postMessage({init: {mem: 16777216*10}})
-			bigworker = true
-			console.log('started big worker')
-		}
-
-		if (typeof options === 'string') {
-			lang = options
-			options = {}
-		}
-
-		if (typeof options === "function") {
-			callback = options
-			options = {}
-		}
-
-		image = convertToImageData(image)
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback || new Function()
-		handlers[i].progress = (function(){
-			if(typeof options.progress === 'function'){
-				var p = options.progress
-				delete options.progress
-				return p
-			}
-			return function(){}
-		})()
-
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'recognize', image: image, lang: lang, options: options})
-		})
-	
-	}
-	return Tesseract
-})()

package.json

@@ -3,12 +3,25 @@
   "version": "1.0.0",
   "description": "",
   "main": "Tesseract.js",
+  "scripts": {
+    "start": "node devServer.js",
+    "build": "webpack --config webpack.config.prod.js"
+  },
   "dependencies": {
     "level-js": "^2.1.6",
     "pako": "^0.2.7",
     "tesseract.js-core": "^1.0.0"
   },
-  "devDependencies": {},
+  "devDependencies": {
+    "babel": "^6.5.2",
+    "babel-core": "^6.7.0",
+    "babel-loader": "^6.2.4",
+    "express": "^4.13.4",
+    "webpack": "^1.13.0",
+    "webpack-dev-middleware": "^1.5.1",
+    "babel-preset-stage-1": "^6.5.0",
+    "babel-preset-es2015": "^6.6.0"
+  },
   "repository": {
     "type": "git",
     "url": "https://github.com/naptha/tesseract.js.git"

src/Tesseract.js

@@ -1,130 +0,0 @@
-var Tesseract = (function(){
-
-	var Tesseract = {}
-
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-	//https://rawgit.com/naptha/tesseract.js/master/worker/worker.js   for testing
-	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
-
-	var blob = new Blob(["importScripts('__worker__');"]); // changed on build
-	// console.log('localhost')
-	var worker = new Worker(window.URL.createObjectURL(blob));
-	worker.postMessage({init: {mem: 16777216*6}})
-	var bigworker = false
-
-	var index = 0
-	var handlers = []
-
-	worker.onmessage = function(e){
-		var handler = handlers[e.data.index]
-		if(e.data.progress){
-			handler.progress(e.data.progress)
-		}
-		else if(e.data.err){
-			handler.reject(e.data.err)
-			handler.callback(e.data.err)
-		}
-		else {
-			handler.resolve(e.data.result)
-			handler.callback(null,e.data.result)
-		}
-	}
-
-	function convertToImageData(image){
-		if(image.getContext){
-			image = image.getContext('2d');
-		}else if(image.tagName == "IMG" || image.tagName == "VIDEO"){
-			var c = document.createElement('canvas');
-			if(image.tagName == "IMG"){
-				c.width  = image.naturalWidth;
-				c.height = image.naturalHeight;
-			}else if(image.tagName == "VIDEO"){
-				c.width  = image.videoWidth;
-				c.height = image.videoHeight;
-			}
-			var ctx = c.getContext('2d');
-			ctx.drawImage(image, 0, 0);
-			image = ctx;
-		}
-		if(image.getImageData) image = image.getImageData(0, 0, image.canvas.width, image.canvas.height);
-		return image	
-	}
-
-	Tesseract.detect = function(image, progress, callback){
-		image = convertToImageData(image)
-
-		if(typeof progress === "undefined"){
-			progress = callback = new Function()
-		}
-
-		if (typeof callback === "undefined"){
-			callback = progress
-			progress = new Function()
-		}
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback
-		handlers[i].progress = progress
-		
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'detect', image: image})
-		})
-
-	}
-
-	Tesseract.recognize = function(image, options, callback){
-		var lang = options.lang
-		if (typeof lang === "undefined"){
-			lang = 'eng'
-		}
-
-		if (!bigworker && ['chi_sim', 'chi_tra', 'jpn'].indexOf(lang) != -1){
-			worker.postMessage({init: {mem: 16777216*10}})
-			bigworker = true
-			console.log('started big worker')
-		}
-
-		if (typeof options === 'string') {
-			lang = options
-			options = {}
-		}
-
-		if (typeof options === "function") {
-			callback = options
-			options = {}
-		}
-
-		image = convertToImageData(image)
-
-		var i = index++
-
-		handlers[i] = {
-			resolve: new Function(),
-			reject: new Function()
-		}
-		handlers[i].callback = callback || new Function()
-		handlers[i].progress = (function(){
-			if(typeof options.progress === 'function'){
-				var p = options.progress
-				delete options.progress
-				return p
-			}
-			return function(){}
-		})()
-
-		return new Promise(function(resolve, reject){
-			handlers[i].resolve = resolve
-			handlers[i].reject = reject
-			worker.postMessage({index: i, fun: 'recognize', image: image, lang: lang, options: options})
-		})
-	
-	}
-	return Tesseract
-})()

src/browser/index.js

@@ -0,0 +1,65 @@
+//TODO: replace with cdn url
+module.exports = function Tesseract(url=location.href+'build/tesseract.worker.js'){
+	var blob = new Blob(["importScripts('"+url+"');"])
+	var worker = new Worker(window.URL.createObjectURL(blob));
+
+	var bigworker = false
+	var jobCounter = 0
+	var handlers = {}
+
+	function runAsync(action, args){
+		var jobId = jobCounter++
+		handlers[jobId] = {}
+		var message = {jobId, action, args}
+		console.log(message)
+		worker.postMessage(message)
+		return {
+			then    (f){ handlers[jobId].result  = f; return this},
+			error   (f){ handlers[jobId].error    = f; return this},
+			progress(f){ handlers[jobId].progress = f; return this}
+		}
+	}
+
+	worker.onmessage = function(e){
+		var {jobId, progress, error, result} = e.data
+		var handler = handlers[jobId]
+		if(progress && handler.progress) handler.progress(progress);
+		if(error && handler.error) handler.error(error);
+		if(result && handler.result) handler.result(result);
+	}
+
+	function convertToImageData(image){
+		if(image.getContext) image = image.getContext('2d');
+		else if(image.tagName == "IMG" || image.tagName == "VIDEO"){
+			var c = document.createElement('canvas');
+			c.width  = image.naturalWidth  || image.videoWidth;
+			c.height = image.naturalHeight || image.videoHeight;
+			var ctx = c.getContext('2d');
+			ctx.drawImage(image, 0, 0);
+			image = ctx;
+		}
+		if(image.getImageData) image = image.getImageData(0, 0, image.canvas.width, image.canvas.height);
+		return image
+	}
+
+	runAsync('init', {mem: (1<<24) * 6})
+
+	return {
+		detect(image){
+			return runAsync('detect', {image: convertToImageData(image)})
+		},
+
+		recognize(image, options='eng'){
+
+			if (typeof options === 'string') options = {lang: options};
+			else options.lang = options.lang || 'eng';
+
+			if (!bigworker && ['chi_sim', 'chi_tra', 'jpn'].indexOf(options.lang) != -1){
+				runAsync('init', {mem: (1<<24) * 10})
+				bigworker = true
+			}
+
+			return runAsync('recognize', {options, image: convertToImageData(image)})
+		}	
+	}
+}

src/worker.js

@@ -1,533 +0,0 @@
-var Tesseract304 = require('tesseract.js-core')
-var leveljs = require('level-js')
-var db;
-if (typeof indexedDB === 'undefined'){
-	db = { open: function(opts, cb){ cb(true) /*err = true*/ } }
-}
-else {
-	db = leveljs('./tessdata')
-}
-
-console.log('hallo')
-
-var filesizes = {"afr": 1079573, "ara": 1701536, "aze": 1420865, "bel": 1276820, "ben": 6772012, "bul": 1605615, "cat": 1652368, "ces": 1035441, "chi_sim": 17710414, "chi_tra": 24717749, "chr": 320649, "dan-frak": 677656, "dan": 1972936, "deu-frak": 822644, "deu": 991656, "ell": 859719, "eng": 9453554, "enm": 619254, "epo": 1241212, "equ": 821130, "est": 1905040, "eus": 1641190, "fin": 979418, "fra": 1376221, "frk": 5912963, "frm": 5147082, "glg": 1674938, "grc": 3012615, "heb": 1051501, "hin": 6590065, "hrv": 1926995, "hun": 3074473, "ind": 1874776, "isl": 1634041, "ita": 948593, "ita_old": 3436571, "jpn": 13507168, "kan": 4390317, "kor": 5353098, "lav": 1843944, "lit": 1779240, "mal": 5966263, "meme": 88453, "mkd": 1163087, "mlt": 1463001, "msa": 1665427, "nld": 1134708, "nor": 2191610, "osd": 4274649, "pol": 7024662, "por": 909359, "ron": 915680, "rus": 5969957, "slk-frak": 289885, "slk": 2217342, "slv": 1611338, "spa": 883170, "spa_old": 5647453, "sqi": 1667041, "srp": 1770244, "swa": 757916, "swe": 2451917, "tam": 3498763, "tel": 5795246, "tgl": 1496256, "tha": 3811136, "tur": 3563264, "ukr": 937566, "vie": 2195922}
-
-var pako = require('pako')
-
-var T;
-
-var tesseractinit = (function createTesseractInstance(memory){
-
-	curindex = 0
-
-	var Module = Tesseract304({
-		TOTAL_MEMORY: memory, //must be a multiple of 10 megabytes
-		TesseractProgress: function(percent){
-			postMessage({
-				index: curindex,
-				'progress': {
-					'recognized': Math.max(0,(percent-30)/70)
-				}
-			})
-		}//,
-		// onRuntimeInitialized: function(){
-		// 	console.log('wau')
-		// }
-	})
-
-	var base = new Module.TessBaseAPI()
-	var loaded_langs = []
-	var loadLanguage = function(lang, index, cb){ // NodeJS style callback
-		if(loaded_langs.indexOf(lang) != -1){
-			cb(null, lang)		
-		}
-		else{
-			Module.FS_createPath("/","tessdata",true,true)
-
-			var downloadlang = function(shouldcache){
-				postMessage({
-					index: index,
-					'progress': {
-						'loaded_lang_model': 0,
-						cached: false,
-						requesting: true
-					}
-				})
-				var xhr = new XMLHttpRequest();
-				xhr.open('GET', 'https://cdn.rawgit.com/naptha/tessdata/gh-pages/3.02/'+lang+'.traineddata.gz', true);
-				xhr.responseType = 'arraybuffer';
-				xhr.onerror = function(){ cb(xhr, null) }
-				xhr.onprogress = function(e){
-					postMessage({
-						index: index,
-						'progress': {
-							'loaded_lang_model': e.loaded/filesizes[lang], //this is kinda wrong on safari
-							cached: false
-						}
-					})
-				}
-				xhr.onload = function(){
-					if (xhr.status == 200 || (xhr.status == 0 && xhr.response)) {
-						postMessage({
-							index: index,
-							'progress': 'unzipping_lang_model'
-						})
-
-						var response = new Uint8Array(xhr.response)
-
-						while(response[0] == 0x1f && response[1] == 0x8b){
-							response = pako.ungzip(response)
-						}
-						console.log('asdf')
-
-						postMessage({
-							index: index,
-							'progress': {
-								'unzipped_lang_model': true,
-								'lang_model_size': response.length
-							}
-						})
-
-						Module.FS_createDataFile('tessdata', lang +".traineddata", response, true, false);
-
-						if(shouldcache){
-							db.put(lang, response, function(err){
-								console.log('cached lang')
-							})
-						}
-
-						postMessage({
-							index: index,
-							'progress': {
-								'created_virtual_datafile': true,
-								'cached_file': shouldcache
-							}
-						})
-
-						loaded_langs.push(lang)
-
-						cb(null, lang)
-					} else cb(xhr, null);
-				}
-				xhr.send(null)
-			}
-
-			db.open({compression: false},function(err){
-				// err = true
-				if (err) {
-					downloadlang(false)
-				}
-				else {
-					db.get(lang, function (err, value) {
-
-						// err = true
-
-						if (err) {
-							downloadlang(true)
-						}
-						else {
-
-							while(value[0] == 0x1f && value[1] == 0x8b){
-								value = pako.ungzip(value)
-							}
-
-							postMessage({
-								index: index,
-								'progress': {
-									loaded_lang_model:1,
-									cached: true
-								}
-							})
-
-							Module.FS_createDataFile('tessdata', lang +".traineddata", value, true, false);
-							loaded_langs.push(lang)
-							cb(null, lang)
-						}
-					})
-				}
-			})
-		}
-	}
-
-	function circularize(page){
-	    page.paragraphs = []
-	    page.lines = []
-	    page.words = []
-	    page.symbols = []
-
-	    page.blocks.forEach(function(block){
-	        block.page = page;
-
-	        block.lines = []
-	        block.words = []
-	        block.symbols = []
-
-	        block.paragraphs.forEach(function(para){
-	            para.block = block;
-	            para.page = page;
-
-	            para.words = []
-	            para.symbols = []
-	            
-	            para.lines.forEach(function(line){
-	                line.paragraph = para;
-	                line.block = block;
-	                line.page = page;
-
-	                line.symbols = []
-
-	                line.words.forEach(function(word){
-	                    word.line = line;
-	                    word.paragraph = para;
-	                    word.block = block;
-	                    word.page = page;
-	                    word.symbols.forEach(function(sym){
-	                        sym.word = word;
-	                        sym.line = line;
-	                        sym.paragraph = para;
-	                        sym.block = block;
-	                        sym.page = page;
-	                        
-	                        sym.line.symbols.push(sym)
-	                        sym.paragraph.symbols.push(sym)
-	                        sym.block.symbols.push(sym)
-	                        sym.page.symbols.push(sym)
-	                    })
-	                    word.paragraph.words.push(word)
-	                    word.block.words.push(word)
-	                    word.page.words.push(word)
-	                })
-	                line.block.lines.push(line)
-	                line.page.lines.push(line)
-	            })
-	            para.page.paragraphs.push(para)
-	        })
-	    })
-	    return page
-	}
-
-	function DumpLiterallyEverything(){
-			var ri = base.GetIterator();
-			var blocks = [];
-			var block, para, textline, word, symbol;
-
-			function enumToString(value, prefix){
-			   return (Object.keys(Module)
-			       .filter(function(e){ return e.substr(0, prefix.length + 1) == prefix + '_' })
-			       .filter(function(e){ return Module[e] === value })
-			       .map(function(e){ return e.slice(prefix.length + 1) })[0])
-			}
-
-			ri.Begin()
-			do {
-				if(ri.IsAtBeginningOf(Module.RIL_BLOCK)){
-					var poly = ri.BlockPolygon();
-					var polygon = null;
-					// BlockPolygon() returns null when automatic page segmentation is off
-					if(Module.getPointer(poly) > 0){
-						var n = poly.get_n(),
-							px = poly.get_x(),
-							py = poly.get_y(),
-							polygon = [];
-						for(var i = 0; i < n; i++){
-							polygon.push([px.getValue(i), py.getValue(i)]);
-						}
-						Module._ptaDestroy(Module.getPointer(poly));	
-					}
-					
-					block = {
-						paragraphs: [],
-
-						text: ri.GetUTF8Text(Module.RIL_BLOCK),
-						confidence: ri.Confidence(Module.RIL_BLOCK),
-						baseline: ri.getBaseline(Module.RIL_BLOCK),
-						bbox: ri.getBoundingBox(Module.RIL_BLOCK),
-
-						blocktype: enumToString(ri.BlockType(), 'PT'),
-						polygon: polygon
-					}
-					blocks.push(block)
-				}
-				if(ri.IsAtBeginningOf(Module.RIL_PARA)){
-					para = {
-						lines: [],
-
-						text: ri.GetUTF8Text(Module.RIL_PARA),
-						confidence: ri.Confidence(Module.RIL_PARA),
-						baseline: ri.getBaseline(Module.RIL_PARA),
-						bbox: ri.getBoundingBox(Module.RIL_PARA),
-
-						is_ltr: !!ri.ParagraphIsLtr()
-					}
-					block.paragraphs.push(para)
-				}
-				if(ri.IsAtBeginningOf(Module.RIL_TEXTLINE)){
-					textline = {
-						words: [],
-
-						text: ri.GetUTF8Text(Module.RIL_TEXTLINE),
-						confidence: ri.Confidence(Module.RIL_TEXTLINE),
-						baseline: ri.getBaseline(Module.RIL_TEXTLINE),
-						bbox: ri.getBoundingBox(Module.RIL_TEXTLINE)
-					}
-					para.lines.push(textline)
-				}
-				if(ri.IsAtBeginningOf(Module.RIL_WORD)){
-					var fontInfo = ri.getWordFontAttributes(),
-						wordDir = ri.WordDirection();
-					word = {
-						symbols: [],
-						choices: [],
-
-						text: ri.GetUTF8Text(Module.RIL_WORD),
-						confidence: ri.Confidence(Module.RIL_WORD),
-						baseline: ri.getBaseline(Module.RIL_WORD),
-						bbox: ri.getBoundingBox(Module.RIL_WORD),
-
-						is_numeric: !!ri.WordIsNumeric(),
-						in_dictionary: !!ri.WordIsFromDictionary(),
-						direction: enumToString(wordDir, 'DIR'),
-						language: ri.WordRecognitionLanguage(),
-
-						is_bold: fontInfo.is_bold,
-						is_italic: fontInfo.is_italic,
-						is_underlined: fontInfo.is_underlined,
-						is_monospace: fontInfo.is_monospace,
-						is_serif: fontInfo.is_serif,
-						is_smallcaps: fontInfo.is_smallcaps,
-						font_size: fontInfo.pointsize,
-						font_id: fontInfo.font_id,
-						font_name: fontInfo.font_name,
-					}
-					var wc = new Module.WordChoiceIterator(ri);
-					do {
-						word.choices.push({
-							text: wc.GetUTF8Text(),
-							confidence: wc.Confidence()
-						})
-					} while (wc.Next());
-					Module.destroy(wc)
-					textline.words.push(word)
-				}
-				
-				var image = null;
-				// var pix = ri.GetBinaryImage(Module.RIL_SYMBOL)
-				// var image = pix2array(pix);
-				// // for some reason it seems that things stop working if you destroy pics
-				// Module._pixDestroy(Module.getPointer(pix));
-				if(ri.IsAtBeginningOf(Module.RIL_SYMBOL)){
-					symbol = {
-						choices: [],
-						image: image,
-
-						text: ri.GetUTF8Text(Module.RIL_SYMBOL),
-						confidence: ri.Confidence(Module.RIL_SYMBOL),
-						baseline: ri.getBaseline(Module.RIL_SYMBOL),
-						bbox: ri.getBoundingBox(Module.RIL_SYMBOL),
-
-						is_superscript: !!ri.SymbolIsSuperscript(),
-						is_subscript: !!ri.SymbolIsSubscript(),
-						is_dropcap: !!ri.SymbolIsDropcap(),
-					}
-					word.symbols.push(symbol)
-					var ci = new Module.ChoiceIterator(ri);
-					do {
-						symbol.choices.push({
-							text: ci.GetUTF8Text(),
-							confidence: ci.Confidence()
-						})
-					} while (ci.Next());
-					Module.destroy(ci)
-				}
-			} while (ri.Next(Module.RIL_SYMBOL));
-			Module.destroy(ri)
-
-			return {
-				text: base.GetUTF8Text(),
-				html: deindent(base.GetHOCRText()),
-
-				confidence: base.MeanTextConf(),
-
-				blocks: blocks,
-
-				psm: enumToString(base.GetPageSegMode(), 'PSM'),
-				oem: enumToString(base.oem(), 'OEM'),
-				version: base.Version(),
-			}
-	}
-
-	function deindent(html){
-		var lines = html.split('\n')
-		if(lines[0].substring(0,2) === "  "){
-			for (var i = 0; i < lines.length; i++) {
-				if (lines[i].substring(0,2) === "  ") {
-					lines[i] = lines[i].slice(2)
-				}
-			};
-		}
-		return lines.join('\n')
-	}
-
-	function desaturate(image){
-		var width, height;
-		if(image.data){
-			var src       = image.data;
-			width     	  = image.width, height = image.height;
-			var dst       = new Uint8Array(width * height);
-			var srcLength = src.length | 0, srcLength_16 = (srcLength - 16) | 0;
-			
-			for (var i = 0, j = 0; i <= srcLength_16; i += 16, j += 4) {
-				// convert to grayscale 4 pixels at a time; eveything with alpha get put in front of 50% gray
-				dst[j]     = (((src[i] * 77 + src[i+1] * 151 + src[i+2] * 28) * src[i+3]) + ((255-src[i+3]) << 15) + 32768) >> 16
-				dst[j+1]   = (((src[i+4] * 77 + src[i+5] * 151 + src[i+6] * 28) * src[i+7]) + ((255-src[i+7]) << 15) + 32768) >> 16
-				dst[j+2]   = (((src[i+8] * 77 + src[i+9] * 151 + src[i+10] * 28) * src[i+11]) + ((255-src[i+11]) << 15) + 32768) >> 16
-				dst[j+3]   = (((src[i+12] * 77 + src[i+13] * 151 + src[i+14] * 28) * src[i+15]) + ((255-src[i+15]) << 15) + 32768) >> 16
-				
-			}
-			for (; i < srcLength; i += 4, ++j) //finish up
-				dst[j]     = (((src[i] * 77 + src[i+1] * 151 + src[i+2] * 28) * src[i+3]) + ((255-src[i+3]) << 15) + 32768) >> 16
-			
-			image = dst;
-		}
-		else {
-			throw 'Expected ImageData'
-		}
-		return image
-	}
-
-	function recognize(index, image, lang, options, cb){
-
-
-		var width = image.width, height = image.height;
-
-		image = desaturate(image)
-
-		var ptr = Module.allocate(image, 'i8', Module.ALLOC_NORMAL);
-		
-		loadLanguage(lang, index, function(err, result){
-
-			if(err){
-				console.error("error loading", lang);
-				Module._free(ptr); 
-				cb(err, null)
-			}
-			else {
-				curindex = index
-
-				base.Init(null, lang)
-
-				postMessage({
-					index: index,			
-					'progress': {
-						'initialized_with_lang': true,
-						'lang': lang
-					}
-				})
-
-				for (var option in options) {
-				    if (options.hasOwnProperty(option)) {
-				        base.SetVariable(option, options[option]);
-				        postMessage({
-							index: index,			
-							'progress': {
-								'set_variable': {
-									variable: option,
-									value: options[option]
-								}
-							}
-						})
-				    }
-				}
-
-
-				base.SetImage(Module.wrapPointer(ptr), width, height, 1, width)
-				base.SetRectangle(0, 0, width, height)
-				// base.GetUTF8Text()
-				base.Recognize(null)
-				var everything = circularize(DumpLiterallyEverything())
-				base.End();
-				Module._free(ptr); 
-				cb(null, everything)
-
-			}
-		})
-	}
-
-	function detect(index, image, cb){
-		var width = image.width, height = image.height;
-		image = desaturate(image)
-
-		var ptr = Module.allocate(image, 'i8', Module.ALLOC_NORMAL);
-		console.log('allocated image')
-		// base = new Module.TessBaseAPI()
-
-		loadLanguage('osd', index, function(err, result){
-			if(err){
-				Module._free(ptr);
-				cb(err)
-			}
-			else {
-				curindex = index
-				base.Init(null, 'osd')
-				base.SetPageSegMode(Module.PSM_OSD_ONLY)
-				console.log('loaded language')
-				
-				base.SetImage(Module.wrapPointer(ptr), width, height, 1, width)
-				base.SetRectangle(0, 0, width, height)
-
-				var results = new Module.OSResults();
-				var success = base.DetectOS(results);
-				if(!success){
-					base.End();
-					Module._free(ptr);
-					cb("failed to detect os")
-				}
-				else {
-					var charset = results.get_unicharset()
-					console.log(charset)
-					// results.print_scores()
-
-					var best = results.get_best_result()
-					var oid = best.get_orientation_id(),
-						sid = best.get_script_id();
-					// console.log('orientation id', oid, [0, 270, 180, 90][oid], best.get_oconfidence())
-					// console.log('script id', sid, charset.get_script_from_script_id(sid), best.get_sconfidence())
-					// console.log(best)
-
-					cb(null, {
-						tesseract_script_id: sid,
-						script: charset.get_script_from_script_id(sid),
-						script_confidence: best.get_sconfidence(),
-						orientation_degrees: [0, 270, 180, 90][oid],
-						orientation_confidence: best.get_oconfidence()
-					})
-
-					base.End();
-					Module._free(ptr);
-				}
-			}
-		})
-	}
-
-	return {
-		recognize: recognize,
-		detect: detect
-	}
-})
-
-onmessage = function(e) {
-
-	if(e.data.init){
-		T = tesseractinit(e.data.init.mem)
-	}
-	else if(e.data.fun === 'recognize'){
-		T.recognize(e.data.index, e.data.image, e.data.lang, e.data.options, function(err, result){
-			postMessage({index: e.data.index, err:err, result: result})
-		})		
-	}
-	else if(e.data.fun === 'detect'){
-		T.detect(e.data.index, e.data.image, function(err, result){
-			postMessage({index: e.data.index, err:err, result: result})
-		})
-	}
-}

src/worker/circularize.js

@@ -0,0 +1,56 @@
+export default function circularize(page){
+    page.paragraphs = []
+    page.lines = []
+    page.words = []
+    page.symbols = []
+
+    page.blocks.forEach(function(block){
+        block.page = page;
+
+        block.lines = []
+        block.words = []
+        block.symbols = []
+
+        block.paragraphs.forEach(function(para){
+            para.block = block;
+            para.page = page;
+
+            para.words = []
+            para.symbols = []
+            
+            para.lines.forEach(function(line){
+                line.paragraph = para;
+                line.block = block;
+                line.page = page;
+
+                line.symbols = []
+
+                line.words.forEach(function(word){
+                    word.line = line;
+                    word.paragraph = para;
+                    word.block = block;
+                    word.page = page;
+                    word.symbols.forEach(function(sym){
+                        sym.word = word;
+                        sym.line = line;
+                        sym.paragraph = para;
+                        sym.block = block;
+                        sym.page = page;
+                        
+                        sym.line.symbols.push(sym)
+                        sym.paragraph.symbols.push(sym)
+                        sym.block.symbols.push(sym)
+                        sym.page.symbols.push(sym)
+                    })
+                    word.paragraph.words.push(word)
+                    word.block.words.push(word)
+                    word.page.words.push(word)
+                })
+                line.block.lines.push(line)
+                line.page.lines.push(line)
+            })
+            para.page.paragraphs.push(para)
+        })
+    })
+    return page
+}

src/worker/db.js

@@ -0,0 +1,3 @@
+import leveljs from 'level-js'
+var db = typeof indexedDB === 'undefined' ? { open: (_, cb) =>  cb(true) } : leveljs('./tessdata')
+export default db

src/worker/desaturate.js

@@ -0,0 +1,26 @@
+export default function desaturate(image){
+	var width, height;
+	if(image.data){
+		var src       = image.data;
+		width     	  = image.width, height = image.height;
+		var dst       = new Uint8Array(width * height);
+		var srcLength = src.length | 0, srcLength_16 = (srcLength - 16) | 0;
+		
+		for (var i = 0, j = 0; i <= srcLength_16; i += 16, j += 4) {
+			// convert to grayscale 4 pixels at a time; eveything with alpha gets put in front of 50% gray
+			dst[j]     = (((src[i] * 77 + src[i+1] * 151 + src[i+2] * 28) * src[i+3]) + ((255-src[i+3]) << 15) + 32768) >> 16
+			dst[j+1]   = (((src[i+4] * 77 + src[i+5] * 151 + src[i+6] * 28) * src[i+7]) + ((255-src[i+7]) << 15) + 32768) >> 16
+			dst[j+2]   = (((src[i+8] * 77 + src[i+9] * 151 + src[i+10] * 28) * src[i+11]) + ((255-src[i+11]) << 15) + 32768) >> 16
+			dst[j+3]   = (((src[i+12] * 77 + src[i+13] * 151 + src[i+14] * 28) * src[i+15]) + ((255-src[i+15]) << 15) + 32768) >> 16
+			
+		}
+		for (; i < srcLength; i += 4, ++j) //finish up
+			dst[j]     = (((src[i] * 77 + src[i+1] * 151 + src[i+2] * 28) * src[i+3]) + ((255-src[i+3]) << 15) + 32768) >> 16
+		
+		image = dst;
+	}
+	else {
+		throw 'Expected ImageData'
+	}
+	return image
+}

src/worker/detect.js

@@ -0,0 +1,54 @@
+export default function detect(jobId, module, base, image, cb){
+	var width = image.width, height = image.height;
+	image = desaturate(image)
+
+	var ptr = module.allocate(image, 'i8', module.ALLOC_NORMAL);
+	console.log('allocated image')
+	// base = new module.TessBaseAPI()
+
+	loadLanguage('osd', jobId, function(err, result){
+		if(err){
+			module._free(ptr);
+			cb(err)
+		}
+		else {
+			base.Init(null, 'osd')
+			base.SetPageSegMode(module.PSM_OSD_ONLY)
+			console.log('loaded language')
+			
+			base.SetImage(module.wrapPointer(ptr), width, height, 1, width)
+			base.SetRectangle(0, 0, width, height)
+
+			var results = new module.OSResults();
+			var success = base.DetectOS(results);
+			if(!success){
+				base.End();
+				module._free(ptr);
+				cb("failed to detect os")
+			}
+			else {
+				var charset = results.get_unicharset()
+				console.log(charset)
+				// results.print_scores()
+
+				var best = results.get_best_result()
+				var oid = best.get_orientation_id(),
+					sid = best.get_script_id();
+				// console.log('orientation id', oid, [0, 270, 180, 90][oid], best.get_oconfidence())
+				// console.log('script id', sid, charset.get_script_from_script_id(sid), best.get_sconfidence())
+				// console.log(best)
+
+				cb(null, {
+					tesseract_script_id: sid,
+					script: charset.get_script_from_script_id(sid),
+					script_confidence: best.get_sconfidence(),
+					orientation_degrees: [0, 270, 180, 90][oid],
+					orientation_confidence: best.get_oconfidence()
+				})
+
+				base.End();
+				module._free(ptr);
+			}
+		}
+	})
+}

src/worker/dump.js

@@ -0,0 +1,163 @@
+function deindent(html){
+	var lines = html.split('\n')
+	if(lines[0].substring(0,2) === "  "){
+		for (var i = 0; i < lines.length; i++) {
+			if (lines[i].substring(0,2) === "  ") {
+				lines[i] = lines[i].slice(2)
+			}
+		};
+	}
+	return lines.join('\n')
+}
+
+export default function DumpLiterallyEverything(module, base){
+	var ri = base.GetIterator();
+	var blocks = [];
+	var block, para, textline, word, symbol;
+
+	function enumToString(value, prefix){
+	   return (Object.keys(module)
+	       .filter(function(e){ return e.substr(0, prefix.length + 1) == prefix + '_' })
+	       .filter(function(e){ return module[e] === value })
+	       .map(function(e){ return e.slice(prefix.length + 1) })[0])
+	}
+
+	const {RIL_BLOCK, RIL_PARA, RIL_TEXTLINE, RIL_WORD, RIL_SYMBOL} = module
+
+	ri.Begin()
+	do {
+		if(ri.IsAtBeginningOf(RIL_BLOCK)){
+			var poly = ri.BlockPolygon();
+			var polygon = null;
+			// BlockPolygon() returns null when automatic page segmentation is off
+			if(module.getPointer(poly) > 0){
+				var n = poly.get_n(),
+					px = poly.get_x(),
+					py = poly.get_y(),
+					polygon = [];
+				for(var i = 0; i < n; i++){
+					polygon.push([px.getValue(i), py.getValue(i)]);
+				}
+				module._ptaDestroy(module.getPointer(poly));	
+			}
+			
+			block = {
+				paragraphs: [],
+
+				text: ri.GetUTF8Text(RIL_BLOCK),
+				confidence: ri.Confidence(RIL_BLOCK),
+				baseline: ri.getBaseline(RIL_BLOCK),
+				bbox: ri.getBoundingBox(RIL_BLOCK),
+
+				blocktype: enumToString(ri.BlockType(), 'PT'),
+				polygon: polygon
+			}
+			blocks.push(block)
+		}
+		if(ri.IsAtBeginningOf(RIL_PARA)){
+			para = {
+				lines: [],
+
+				text: ri.GetUTF8Text(RIL_PARA),
+				confidence: ri.Confidence(RIL_PARA),
+				baseline: ri.getBaseline(RIL_PARA),
+				bbox: ri.getBoundingBox(RIL_PARA),
+
+				is_ltr: !!ri.ParagraphIsLtr()
+			}
+			block.paragraphs.push(para)
+		}
+		if(ri.IsAtBeginningOf(RIL_TEXTLINE)){
+			textline = {
+				words: [],
+
+				text: ri.GetUTF8Text(RIL_TEXTLINE),
+				confidence: ri.Confidence(RIL_TEXTLINE),
+				baseline: ri.getBaseline(RIL_TEXTLINE),
+				bbox: ri.getBoundingBox(RIL_TEXTLINE)
+			}
+			para.lines.push(textline)
+		}
+		if(ri.IsAtBeginningOf(RIL_WORD)){
+			var fontInfo = ri.getWordFontAttributes(),
+				wordDir = ri.WordDirection();
+			word = {
+				symbols: [],
+				choices: [],
+
+				text: ri.GetUTF8Text(RIL_WORD),
+				confidence: ri.Confidence(RIL_WORD),
+				baseline: ri.getBaseline(RIL_WORD),
+				bbox: ri.getBoundingBox(RIL_WORD),
+
+				is_numeric: !!ri.WordIsNumeric(),
+				in_dictionary: !!ri.WordIsFromDictionary(),
+				direction: enumToString(wordDir, 'DIR'),
+				language: ri.WordRecognitionLanguage(),
+
+				is_bold: fontInfo.is_bold,
+				is_italic: fontInfo.is_italic,
+				is_underlined: fontInfo.is_underlined,
+				is_monospace: fontInfo.is_monospace,
+				is_serif: fontInfo.is_serif,
+				is_smallcaps: fontInfo.is_smallcaps,
+				font_size: fontInfo.pointsize,
+				font_id: fontInfo.font_id,
+				font_name: fontInfo.font_name,
+			}
+			var wc = new module.WordChoiceIterator(ri);
+			do {
+				word.choices.push({
+					text: wc.GetUTF8Text(),
+					confidence: wc.Confidence()
+				})
+			} while (wc.Next());
+			module.destroy(wc)
+			textline.words.push(word)
+		}
+		
+		var image = null;
+		// var pix = ri.GetBinaryImage(RIL_SYMBOL)
+		// var image = pix2array(pix);
+		// // for some reason it seems that things stop working if you destroy pics
+		// module._pixDestroy(module.getPointer(pix));
+		if(ri.IsAtBeginningOf(RIL_SYMBOL)){
+			symbol = {
+				choices: [],
+				image: image,
+
+				text: ri.GetUTF8Text(RIL_SYMBOL),
+				confidence: ri.Confidence(RIL_SYMBOL),
+				baseline: ri.getBaseline(RIL_SYMBOL),
+				bbox: ri.getBoundingBox(RIL_SYMBOL),
+
+				is_superscript: !!ri.SymbolIsSuperscript(),
+				is_subscript: !!ri.SymbolIsSubscript(),
+				is_dropcap: !!ri.SymbolIsDropcap(),
+			}
+			word.symbols.push(symbol)
+			var ci = new module.ChoiceIterator(ri);
+			do {
+				symbol.choices.push({
+					text: ci.GetUTF8Text(),
+					confidence: ci.Confidence()
+				})
+			} while (ci.Next());
+			module.destroy(ci)
+		}
+	} while (ri.Next(RIL_SYMBOL));
+	module.destroy(ri)
+
+	return {
+		text: base.GetUTF8Text(),
+		html: deindent(base.GetHOCRText()),
+
+		confidence: base.MeanTextConf(),
+
+		blocks: blocks,
+
+		psm: enumToString(base.GetPageSegMode(), 'PSM'),
+		oem: enumToString(base.oem(), 'OEM'),
+		version: base.Version(),
+	}
+}

src/worker/fileSizes.js

@@ -0,0 +1,2 @@
+const fileSizes = {"afr": 1079573, "ara": 1701536, "aze": 1420865, "bel": 1276820, "ben": 6772012, "bul": 1605615, "cat": 1652368, "ces": 1035441, "chi_sim": 17710414, "chi_tra": 24717749, "chr": 320649, "dan-frak": 677656, "dan": 1972936, "deu-frak": 822644, "deu": 991656, "ell": 859719, "eng": 9453554, "enm": 619254, "epo": 1241212, "equ": 821130, "est": 1905040, "eus": 1641190, "fin": 979418, "fra": 1376221, "frk": 5912963, "frm": 5147082, "glg": 1674938, "grc": 3012615, "heb": 1051501, "hin": 6590065, "hrv": 1926995, "hun": 3074473, "ind": 1874776, "isl": 1634041, "ita": 948593, "ita_old": 3436571, "jpn": 13507168, "kan": 4390317, "kor": 5353098, "lav": 1843944, "lit": 1779240, "mal": 5966263, "meme": 88453, "mkd": 1163087, "mlt": 1463001, "msa": 1665427, "nld": 1134708, "nor": 2191610, "osd": 4274649, "pol": 7024662, "por": 909359, "ron": 915680, "rus": 5969957, "slk-frak": 289885, "slk": 2217342, "slv": 1611338, "spa": 883170, "spa_old": 5647453, "sqi": 1667041, "srp": 1770244, "swa": 757916, "swe": 2451917, "tam": 3498763, "tel": 5795246, "tgl": 1496256, "tha": 3811136, "tur": 3563264, "ukr": 937566, "vie": 2195922}
+export default fileSizes;

src/worker/index.js

@@ -0,0 +1,39 @@
+import TesseractCore from 'tesseract.js-core'
+import pako from 'pako'
+
+import recognize from './recognize'
+import detect from './detect'
+
+var module, base, jobId
+
+onmessage = function(e) {
+	var {action, args} = e.data;
+	jobId = e.data.jobId
+
+	console.log('worker got action', action)
+
+	if(action == 'init'){
+
+		module = TesseractCore({
+			TOTAL_MEMORY: args.mem, //must be a multiple of 10 megabytes
+			TesseractProgress(percent){
+				postMessage({ jobId,
+					'progress': {
+						'recognized': Math.max(0,(percent-30)/70)
+					}
+				})
+			},
+			onRuntimeInitialized() {}
+		})
+		module.FS_createPath("/","tessdata",true,true)
+		base = new module.TessBaseAPI()
+
+	} else if(action === 'recognize'){
+		var {image, options} = args
+		recognize(jobId, module, base, image, options,
+			(error, result) => postMessage({jobId, error, result}))
+	} else if(action === 'detect'){
+		detect(jobId, module, base, args.image, 
+			(error, result) => postMessage({jobId, error, result}))
+	}
+}

src/worker/loadLanguage.js

@@ -0,0 +1,88 @@
+import pako from 'pako'
+import db from './db'
+import fileSizes from './fileSizes'
+
+function getLanguageData(lang, progress, cb, url='https://cdn.rawgit.com/naptha/tessdata/gh-pages/3.02/'+lang+'.traineddata.gz'){
+	var xhr = new XMLHttpRequest();
+	xhr.responseType = 'arraybuffer';
+	xhr.open('GET', url, true);
+	xhr.onerror    = e => {
+		xhr.onprogress = xhr.onload = null
+		cb(xhr, null)
+	}
+	xhr.onprogress = e => progress({
+		'loaded_lang_model': e.loaded/fileSizes[lang], //this is kinda wrong on safari
+		cached: false
+	})
+	xhr.onload = e => {
+		if (!(xhr.status == 200 || (xhr.status == 0 && xhr.response))) return cb(xhr, null);
+
+		progress({'unzipping_lang_model': true})
+
+		var response = new Uint8Array(xhr.response)
+		while(response[0] == 0x1f && response[1] == 0x8b) response = pako.ungzip(response);
+
+		progress({
+			'unzipped_lang_model': true,
+			'lang_model_size': response.length
+		})
+
+		cb(null, response)
+	}
+	
+	progress({
+		'loaded_lang_model': 0,
+		cached: false,
+		requesting: true
+	})
+
+	xhr.send()
+}
+
+// var loaded_langs = []
+
+export default function loadLanguage(lang, jobId, cb, url){
+
+	console.log('loadLanguage jobId', jobId)
+
+	// if(loaded_langs.indexOf(lang) != -1) return cb(null, lang);
+
+	function progressMessage(progress){
+		postMessage({ jobId, progress })
+	}	
+
+	function finish(err, data) {
+		if(err) return cb(err);
+		// loaded_langs.push(lang)
+		cb(null, data)
+	}
+
+	function createDataFile(err, data){
+		progressMessage({ created_virtual_datafile: true})
+		finish(err, data)
+	}
+
+	function createDataFileCached(err, data) {
+		if(err) return createDataFile(err);
+
+		db.put(lang, data, err => console.log('cached', lang, err))
+		progressMessage({cached_lang: lang}) 
+		createDataFile(null, data)
+	}
+
+
+	db.open({compression: false}, err => {
+		if (err) return getLanguageData(lang, progressMessage, createDataFile, url);
+
+		db.get(lang, (err, data) => {
+
+			if (err) return getLanguageData(lang, progressMessage, createDataFileCached, url)
+
+			while(data[0] == 0x1f && data[1] == 0x8b) data = pako.ungzip(data);
+
+			progressMessage({ loaded_lang_model: lang, from_cache: true })
+
+			cb(null, data)
+		})
+	})	
+}

src/worker/recognize.js

@@ -0,0 +1,71 @@
+import desaturate from './desaturate'
+import loadLanguage from './loadLanguage'
+import circularize from './circularize'
+import dump from './dump'
+
+var loaded_langs = []
+
+export default function recognize(jobId, module, base, image, options, cb){
+
+	console.log('recognize id', jobId)
+	var {lang} = options
+	var width = image.width, height = image.height;
+
+	image = desaturate(image)
+
+	var ptr = module.allocate(image, 'i8', module.ALLOC_NORMAL);
+
+
+	function run() {
+		base.Init(null, lang)
+
+		postMessage({
+			jobId,			
+			'progress': {
+				'initialized_with_lang': lang
+			}
+		})
+
+		for (var option in options) {
+		    if (options.hasOwnProperty(option)) {
+		        base.SetVariable(option, options[option]);
+		        postMessage({
+					jobId: jobId,			
+					'progress': {
+						'set_variable': {
+							variable: option,
+							value: options[option]
+						}
+					}
+				})
+		    }
+		}
+
+
+		base.SetImage(module.wrapPointer(ptr), width, height, 1, width)
+		base.SetRectangle(0, 0, width, height)
+		// base.GetUTF8Text()
+		base.Recognize(null)
+		var everything = circularize(dump(module, base))
+		base.End();
+		module._free(ptr); 
+		cb(null, everything)
+	}
+
+
+
+	if(loaded_langs.indexOf(lang) == -1) loadLanguage(lang, jobId, function(err, result){
+
+		if(err){
+			console.error("error loading", lang);
+			module._free(ptr);
+			return cb(err, null);
+		}
+
+		loaded_langs.push(lang)
+		module.FS_createDataFile('tessdata', lang +".traineddata", result, true, false);
+		run()
+
+	})
+	else run();
+}

webpack.config.dev.js

@@ -0,0 +1,42 @@
+var path = require('path');
+var webpack = require('webpack');
+
+function config({entry, output, include}) {
+  return {
+    devtool: 'cheap-module-eval-source-map',
+    entry,
+    output: Object.assign({}, output, {
+      path: path.join(__dirname, 'build'),
+      publicPath: '/build/',
+    }),
+    plugins: [
+      new webpack.NoErrorsPlugin()
+    ],
+    module: {
+      loaders: [{
+        test: /\.js$/,
+        loaders: ['babel'],
+        include
+      }]
+    },
+    node: {
+      fs: "empty"
+    }
+  }
+}
+
+module.exports = [{
+  entry: './src/browser/index.js',
+  output: {
+    filename: 'tesseract.js',
+    library: "Tesseract",
+    libraryTarget: "umd"
+  },
+  include: [path.join(__dirname, 'src/browser')]  
+}, {
+  entry: './src/worker/index.js',
+  output: {
+    filename: 'tesseract.worker.js',
+  },
+  include: [path.join(__dirname, 'src/worker')]
+}].map(config);

webpack.config.prod.js

@@ -0,0 +1,46 @@
+var path = require('path');
+var webpack = require('webpack');
+
+function config({entry, output, include}) {
+  return {
+    entry,
+    output: Object.assign({}, output, {
+      path: path.join(__dirname, 'dist')
+    }),
+    plugins: [
+      new webpack.optimize.OccurenceOrderPlugin(),
+      new webpack.optimize.DedupePlugin(),
+      new webpack.optimize.UglifyJsPlugin({
+        compressor: {
+          warnings: false
+        }
+      })
+    ],
+    module: {
+      loaders: [{
+        test: /\.js$/,
+        loaders: ['babel'],
+        include
+      }]
+    },
+    node: {
+      fs: "empty"
+    }
+  }
+}
+
+module.exports = [{
+  entry: './src/browser/index.js',
+  output: {
+    filename: 'tesseract.js',
+    library: "Tesseract",
+    libraryTarget: "umd"
+  },
+  include: [path.join(__dirname, 'src/browser')]  
+}, {
+  entry: './src/worker/index.js',
+  output: {
+    filename: 'tesseract.worker.js',
+  },
+  include: [path.join(__dirname, 'src/worker')]
+}].map(config);