started working oden

lib/Tesseract.js

@@ -2,11 +2,14 @@ var Tesseract = (function(){
 
 	var Tesseract = {}
 
-	var blob = new Blob(["importScripts('http://localhost:1234/master/worker/worker.js');"]);
+	//https://cdn.rawgit.com/naptha/tesseract.js/master/worker/worker.js
+	//https://rawgit.com/naptha/tesseract.js/master/worker/worker.js   for testing
+
+	var blob = new Blob(["importScripts('https://cdn.rawgit.com/naptha/tessdata/worker/worker.js');"]);
 	console.log('localhost')
 	var worker = new Worker(window.URL.createObjectURL(blob));
-
+	worker.postMessage({init: {mem: 16777216*6}})
-	console.log(worker)
+	var bigworker = false
 
 	var index = 0
 	var handlers = []
@@ -77,10 +80,16 @@ var Tesseract = (function(){
 
 	Tesseract.recognize = function(image, options, callback){
 		var lang = options.lang
-		if(typeof lang === "undefined"){
+		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 = {}

worker_src/worker.js

@@ -1,3 +1,4 @@
+var Tesseract304 = require('tesseract')
 var leveljs = require('level-js')
 var db;
 if (typeof indexedDB === 'undefined'){
@@ -13,12 +14,14 @@ var filesizes = {"afr": 1079573, "ara": 1701536, "aze": 1420865, "bel": 1276820,
 
 var pako = require('pako')
 
-var T = (function createTesseractInstance(){
+var T;
+
+var tesseractinit = (function createTesseractInstance(memory){
 
 	curindex = 0
 
 	var Module = Tesseract304({
-		TOTAL_MEMORY: 6*16777216, //must be a multiple of 10 megabytes
+		TOTAL_MEMORY: memory, //must be a multiple of 10 megabytes
 		TesseractProgress: function(percent){
 			postMessage({
 				index: curindex,
@@ -341,7 +344,7 @@ var T = (function createTesseractInstance(){
 
 			return {
 				text: base.GetUTF8Text(),
-				html: base.GetHOCRText(),
+				html: deindent(base.GetHOCRText()),
 
 				confidence: base.MeanTextConf(),
 
@@ -353,6 +356,18 @@ var T = (function createTesseractInstance(){
 			}
 	}
 
+	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){
@@ -413,8 +428,9 @@ var T = (function createTesseractInstance(){
 				    if (options.hasOwnProperty(option)) {
 				        base.SetVariable(option, options[option]);
 				        postMessage({
-							progress: {
+							index: index,			
-								set_variable: {
+							'progress': {
+								'set_variable': {
 									variable: option,
 									value: options[option]
 								}
@@ -485,7 +501,7 @@ var T = (function createTesseractInstance(){
 						orientation_degrees: [0, 270, 180, 90][oid],
 						orientation_confidence: best.get_oconfidence()
 					})
-					
+
 					base.End();
 					Module._free(ptr);
 				}
@@ -497,10 +513,14 @@ var T = (function createTesseractInstance(){
 		recognize: recognize,
 		detect: detect
 	}
-})()
+})
 
 onmessage = function(e) {
-	if(e.data.fun === 'recognize'){
+
+	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})
 		})