add worker sauce

src/worker/detect.js

@@ -1,29 +1,29 @@
 import desaturate from './desaturate'
 import loadLanguage from './loadLanguage'
 
-export default function detect(jobId, module, base, image, cb){
+export default function detect(jobId, image, cb){
 	var width = image.width, height = image.height;
 	image = desaturate(image)
 
-	var ptr = module.allocate(image, 'i8', module.ALLOC_NORMAL);
+	var ptr = self.module.allocate(image, 'i8', self.module.ALLOC_NORMAL);
 	// console.log('allocated image')
 
-	loadLanguage(jobId, module, 'osd', err => {
-		module._free(ptr);
+	loadLanguage(jobId, 'osd', err => {
+		self.module._free(ptr);
 		cb(err)
 	}, success => {
-		base.Init(null, 'osd')
-		base.SetPageSegMode(module.PSM_OSD_ONLY)
+		self.base.Init(null, 'osd')
+		self.base.SetPageSegMode(self.module.PSM_OSD_ONLY)
 		// console.log('loaded language')
 		
-		base.SetImage(module.wrapPointer(ptr), width, height, 1, width)
-		base.SetRectangle(0, 0, width, height)
+		self.base.SetImage(self.module.wrapPointer(ptr), width, height, 1, width)
+		self.base.SetRectangle(0, 0, width, height)
 
-		var results = new module.OSResults();
-		var success = base.DetectOS(results);
+		var results = new self.module.OSResults();
+		var success = self.base.DetectOS(results);
 		if(!success){
-			base.End();
-			module._free(ptr);
+			self.base.End();
+			self.module._free(ptr);
 			cb("failed to detect os")
 		}
 		else {
@@ -46,8 +46,8 @@ export default function detect(jobId, module, base, image, cb){
 				orientation_confidence: best.get_oconfidence()
 			})
 
-			base.End();
-			module._free(ptr);
+			self.base.End();
+			self.module._free(ptr);
 		}
 	})
 }

src/worker/dump.js

@@ -10,7 +10,10 @@ function deindent(html){
 	return lines.join('\n')
 }
 
-export default function DumpLiterallyEverything(module, base){
+export default function DumpLiterallyEverything(){
+
+	var {module, base} = self
+
 	var ri = base.GetIterator();
 	var blocks = [];
 	var block, para, textline, word, symbol;

src/worker/index.js

@@ -1,6 +1,3 @@
-import TesseractCore from 'tesseract.js-core'
-import pako from 'pako'
-
 import recognize from './recognize'
 import detect from './detect'
 
@@ -14,7 +11,8 @@ onmessage = function(e) {
 
 	if(action == 'init'){
 
-		module = TesseractCore({
+		self.langUrl = args.langUrl
+		self.module = TesseractCore({
 			TOTAL_MEMORY: args.mem, //must be a multiple of 10 megabytes
 			TesseractProgress(percent){
 				postMessage({ jobId,
@@ -25,15 +23,15 @@ onmessage = function(e) {
 			},
 			onRuntimeInitialized() {}
 		})
-		module.FS_createPath("/","tessdata",true,true)
-		base = new module.TessBaseAPI()
+		self.module.FS_createPath("/","tessdata",true,true)
+		self.base = new self.module.TessBaseAPI()
 
 	} else if(action === 'recognize'){
 		var {image, options} = args
-		recognize(jobId, module, base, image, options,
+		recognize(jobId, image, options,
 			(error, result) => postMessage({jobId, error, result}))
 	} else if(action === 'detect'){
-		detect(jobId, module, base, args.image, 
+		detect(jobId, args.image, 
 			(error, result) => postMessage({jobId, error, result}))
 	}
 }

src/worker/loadLanguage.js

@@ -1,12 +1,12 @@
-import pako from 'pako'
+import {ungzip} 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'){
+function getLanguageData(lang, progress, cb){
 	var xhr = new XMLHttpRequest();
 	xhr.responseType = 'arraybuffer';
-	xhr.open('GET', url, true);
-	xhr.onerror    = e => {
+	xhr.open('GET', self.LANG_URL + lang + '.traineddata.gz', true);
+	xhr.onerror = e => {
 		xhr.onprogress = xhr.onload = null
 		cb(xhr, null)
 	}
@@ -20,7 +20,7 @@ function getLanguageData(lang, progress, cb, url='https://cdn.rawgit.com/naptha/
 		progress({'unzipping_lang_model': true})
 
 		var response = new Uint8Array(xhr.response)
-		while(response[0] == 0x1f && response[1] == 0x8b) response = pako.ungzip(response);
+		while(response[0] == 0x1f && response[1] == 0x8b) response = ungzip(response);
 
 		progress({
 			'unzipped_lang_model': true,
@@ -40,7 +40,7 @@ function getLanguageData(lang, progress, cb, url='https://cdn.rawgit.com/naptha/
 }
 
 
-function load(lang, jobId, cb, url){
+function load(lang, jobId, cb){
 
 	console.log('loadLanguage jobId', jobId)
 
@@ -69,13 +69,13 @@ function load(lang, jobId, cb, url){
 
 
 	db.open({compression: false}, err => {
-		if (err) return getLanguageData(lang, progressMessage, createDataFile, url);
+		if (err) return getLanguageData(lang, progressMessage, createDataFile);
 
 		db.get(lang, (err, data) => {
 
-			if (err) return getLanguageData(lang, progressMessage, createDataFileCached, url)
+			if (err) return getLanguageData(lang, progressMessage, createDataFileCached)
 
-			while(data[0] == 0x1f && data[1] == 0x8b) data = pako.ungzip(data);
+			while(data[0] == 0x1f && data[1] == 0x8b) data = ungzip(data);
 
 			progressMessage({ loaded_lang_model: lang, from_cache: true })
 
@@ -86,14 +86,14 @@ function load(lang, jobId, cb, url){
 
 var loaded_langs = []
 
-export default function loadLanguage(jobId, module, lang, error, success){
+export default function loadLanguage(jobId, lang, error, success){
 	if(loaded_langs.indexOf(lang) == -1) load(lang, jobId, function(err, result){
 		if(err) return error(err)
 
 		loaded_langs.push(lang)
-		module.FS_createDataFile('tessdata', lang +".traineddata", result, true, false);
+		self.module.FS_createDataFile('tessdata', lang +".traineddata", result, true, false);
 
 		success()
 	})
-	else run();
+	else success();
 }

src/worker/recognize.js

@@ -5,7 +5,7 @@ import dump from './dump'
 
 var loaded_langs = []
 
-export default function recognize(jobId, module, base, image, options, cb){
+export default function recognize(jobId, image, options, cb){
 
 	console.log('recognize id', jobId)
 	var {lang} = options
@@ -13,13 +13,13 @@ export default function recognize(jobId, module, base, image, options, cb){
 
 	image = desaturate(image)
 
-	var ptr = module.allocate(image, 'i8', module.ALLOC_NORMAL);
+	var ptr = self.module.allocate(image, 'i8', self.module.ALLOC_NORMAL);
 
-	loadLanguage(jobId, module, lang, err => {
-		module._free(ptr)
+	loadLanguage(jobId, lang, err => {
+		self.module._free(ptr)
 		cb(err)
 	}, success => {
-		base.Init(null, lang)
+		self.base.Init(null, lang)
 
 		postMessage({
 			jobId,			
@@ -30,7 +30,7 @@ export default function recognize(jobId, module, base, image, options, cb){
 
 		for (var option in options) {
 		    if (options.hasOwnProperty(option)) {
-		        base.SetVariable(option, options[option]);
+		        self.base.SetVariable(option, options[option]);
 		        postMessage({
 					jobId: jobId,			
 					'progress': {
@@ -44,13 +44,13 @@ export default function recognize(jobId, module, base, image, options, cb){
 		}
 
 
-		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); 
+		self.base.SetImage(self.module.wrapPointer(ptr), width, height, 1, width)
+		self.base.SetRectangle(0, 0, width, height)
+		// self.base.GetUTF8Text()
+		self.base.Recognize(null)
+		var everything = circularize(dump())
+		self.base.End();
+		self.module._free(ptr); 
 		cb(null, everything)
 	})	
 }