akira
/
tesseract.js
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

Upgrade to latest tesseract.js-utils

feature/aio
Jerome Wu 6 years ago
parent
46e84ad95d
commit
de4b98ae23
9 changed files with 68 additions and 51 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 6
      package-lock.json
  2. 4
      package.json
  3. 1
      scripts/webpack.config.common.js
  4. 20
      scripts/webpack.config.dev.js
  5. 22
      scripts/webpack.config.prod.js
  6. 10
      src/common/TesseractWorker.js
  7. 36
      src/common/workerUtils.js
  8. 4
      tests/detect.test.js
  9. 16
      tests/recognize.test.js

6
package-lock.json generated
Unescape View File

@ -8537,9 +8537,9 @@
"integrity": "sha512-QmNgMA9m5ES5uMTqpOAPysrUA80vUx/6WKQlfkK3zhOeAgqv8DjwwcDv9tQv2TgRzOQ+LFKrJn94Y2rw5b2IGw==" "integrity": "sha512-QmNgMA9m5ES5uMTqpOAPysrUA80vUx/6WKQlfkK3zhOeAgqv8DjwwcDv9tQv2TgRzOQ+LFKrJn94Y2rw5b2IGw=="
}, },
"tesseract.js-utils": { "tesseract.js-utils": {
"version": "1.0.0-beta.6", "version": "1.0.0-beta.7",
"resolved": "https://registry.npmjs.org/tesseract.js-utils/-/tesseract.js-utils-1.0.0-beta.6.tgz", "resolved": "https://registry.npmjs.org/tesseract.js-utils/-/tesseract.js-utils-1.0.0-beta.7.tgz",
"integrity": "sha512-AENYhkqafwysayWmKtyApV0gR4abLJ426plLNHs/++8oHt+ekooyp77ew/q4+QLE7cbUDyxiNGawcraOWE/RuQ==", "integrity": "sha512-MvBQNxVoueDg/8iN8jb26Tnj0lBRfmHVMcnaNdab+JeQKg/SuNXWMCpTD4D17+S6zeE9AAlVQymDmXH2/3vaMg==",
"requires": { "requires": {
"axios": "^0.18.0", "axios": "^0.18.0",
"bmp-js": "^0.1.0", "bmp-js": "^0.1.0",

4
package.json
Unescape View File

@ -5,7 +5,7 @@
"main": "src/index.js", "main": "src/index.js",
"scripts": { "scripts": {
"start": "node scripts/server.js", "start": "node scripts/server.js",
"build": "webpack --config scripts/webpack.config.prod.js", "build": "webpack --progress --config scripts/webpack.config.prod.js",
"prepublishOnly": "npm run build", "prepublishOnly": "npm run build",
"wait": "wait-on http://localhost:3000/package.json", "wait": "wait-on http://localhost:3000/package.json",
"test": "npm-run-all -p -r start test:all", "test": "npm-run-all -p -r start test:all",
@ -54,7 +54,7 @@
"node-fetch": "^2.3.0", "node-fetch": "^2.3.0",
"resolve-url": "^0.2.1", "resolve-url": "^0.2.1",
"tesseract.js-core": "^2.0.0-beta.10", "tesseract.js-core": "^2.0.0-beta.10",
"tesseract.js-utils": "^1.0.0-beta.6" "tesseract.js-utils": "^1.0.0-beta.7"
}, },
"repository": { "repository": {
"type": "git", "type": "git",

1
scripts/webpack.config.common.js
Unescape View File

@ -4,6 +4,7 @@ module.exports = {
{ {
test: /\.m?js$/, test: /\.m?js$/,
// exclude: /(node_modules|bower_components)/, // exclude: /(node_modules|bower_components)/,
exclude: /(tesseract.js-core)/,
use: { use: {
loader: 'babel-loader', loader: 'babel-loader',
options: { options: {

20
scripts/webpack.config.dev.js
Unescape View File

@ -3,9 +3,10 @@ const webpack = require('webpack');
const common = require('./webpack.config.common'); const common = require('./webpack.config.common');
const genConfig = ({ const genConfig = ({
entry, filename, library, libraryTarget, entry, filename, library, libraryTarget, ...config
}) => ({ }) => ({
...common, ...common,
...config,
mode: 'development', mode: 'development',
entry, entry,
output: { output: {
@ -29,10 +30,15 @@ module.exports = [
library: 'Tesseract', library: 'Tesseract',
libraryTarget: 'umd', libraryTarget: 'umd',
}), }),
/* genConfig({
*genConfig({ entry: path.resolve(__dirname, '..', 'src', 'index.js'),
* entry: path.resolve(__dirname, '..', 'src', 'browser', 'worker.js'), filename: 'tesseract.asm.dev.js',
* filename: 'worker.dev.js', library: 'Tesseract',
*}), libraryTarget: 'umd',
*/ resolve: {
alias: {
'tesseract.js-core/tesseract-core.wasm.js': 'tesseract.js-core/tesseract-core.asm.js',
},
},
}),
]; ];

22
scripts/webpack.config.prod.js
Unescape View File

@ -2,11 +2,12 @@ const path = require('path');
const common = require('./webpack.config.common'); const common = require('./webpack.config.common');
const genConfig = ({ const genConfig = ({
entry, filename, library, libraryTarget, entry, filename, library, libraryTarget, ...config
}) => ({ }) => ({
...common, ...common,
...config,
mode: 'production', mode: 'production',
devtool: 'source-map', // devtool: 'source-map',
entry, entry,
output: { output: {
path: path.resolve(__dirname, '..', 'dist'), path: path.resolve(__dirname, '..', 'dist'),
@ -23,10 +24,15 @@ module.exports = [
library: 'Tesseract', library: 'Tesseract',
libraryTarget: 'umd', libraryTarget: 'umd',
}), }),
/* genConfig({
*genConfig({ entry: path.resolve(__dirname, '..', 'src', 'index.js'),
* entry: path.resolve(__dirname, '..', 'src', 'browser', 'worker.js'), filename: 'tesseract.asm.min.js',
* filename: 'worker.min.js', library: 'Tesseract',
*}), libraryTarget: 'umd',
*/ resolve: {
alias: {
'tesseract.js-core/tesseract-core.wasm.js': 'tesseract.js-core/tesseract-core.asm.js',
},
},
}),
]; ];

10
src/common/TesseractWorker.js
Unescape View File

@ -67,12 +67,12 @@ class TesseractWorker {
* @function recognize text in given image * @function recognize text in given image
* @access public * @access public
* @param {Buffer, string} image - image to be recognized * @param {Buffer, string} image - image to be recognized
* @param {string} [lang=eng] - language to recognize * @param {string, array} [langs='eng'] - language to recognize
* @param {object} params - tesseract parameters * @param {object} params - tesseract parameters
* *
*/ */
recognize(image, lang = 'eng', params = {}) { recognize(image, langs = 'eng', params = {}) {
return this._sendJob('recognize', image, lang, params); return this._sendJob('recognize', image, langs, params);
} }
/** /**
@ -152,13 +152,13 @@ class TesseractWorker {
* @param {string} lang language to recognize * @param {string} lang language to recognize
* @param {object} params tesseract parameters * @param {object} params tesseract parameters
*/ */
_sendJob(type, image, lang, params) { _sendJob(type, image, langs, params) {
return this._delay((job) => { return this._delay((job) => {
job.send( job.send(
type, type,
{ {
image, image,
lang, langs,
params, params,
options: this.options, options: this.options,
}, },

36
src/common/workerUtils.js
Unescape View File

@ -52,16 +52,24 @@ const setImage = (image) => {
return data === null ? pix : data; return data === null ? pix : data;
}; };
const getLangsStr = (langs) => {
if (typeof langs === 'string') {
return langs;
}
return langs.map(lang => (typeof lang === 'string' ? lang : lang.code)).join('+');
};
/** /**
* handleParams * handleParams
* *
* @name handleParams * @name handleParams
* @function hanlde params from users * @function hanlde params from users
* @access private * @access private
* @param {string} lang - lang string for Init() * @param {string} langs - lang string for Init()
* @param {object} customParams - an object of params * @param {object} customParams - an object of params
*/ */
const handleParams = (lang, customParams) => { const handleParams = (langs, customParams) => {
const { const {
tessedit_ocr_engine_mode, tessedit_ocr_engine_mode,
...params ...params
@ -69,7 +77,7 @@ const handleParams = (lang, customParams) => {
...defaultParams, ...defaultParams,
...customParams, ...customParams,
}; };
api.Init(null, lang, tessedit_ocr_engine_mode); api.Init(null, getLangsStr(langs), tessedit_ocr_engine_mode);
Object.keys(params).forEach((key) => { Object.keys(params).forEach((key) => {
api.SetVariable(key, params[key]); api.SetVariable(key, params[key]);
}); });
@ -158,14 +166,14 @@ const handleInit = ({ corePath }, res) => {
* @function load language from remote or local cache * @function load language from remote or local cache
* @access public * @access public
* @param {object} req - job payload * @param {object} req - job payload
* @param {string} req.lang - languages to load, ex: eng, eng+chi_tra * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
* @param {object} req.options - other options for loadLang function * @param {object} req.options - other options for loadLang function
* @param {object} res - job instance * @param {object} res - job instance
* @returns {Promise} A Promise for callback * @returns {Promise} A Promise for callback
*/ */
const loadLanguage = ({ lang, options }, res) => { const loadLanguage = ({ langs, options }, res) => {
res.progress({ status: 'loading language traineddata', progress: 0 }); res.progress({ status: 'loading language traineddata', progress: 0 });
return loadLang({ lang, TessModule, ...options }).then((...args) => { return loadLang({ langs, TessModule, ...options }).then((...args) => {
res.progress({ status: 'loaded language traineddata', progress: 1 }); res.progress({ status: 'loaded language traineddata', progress: 1 });
return args; return args;
}); });
@ -179,17 +187,17 @@ const loadLanguage = ({ lang, options }, res) => {
* @access public * @access public
* @param {object} req - job payload * @param {object} req - job payload
* @param {array} req.image - binary image in array format * @param {array} req.image - binary image in array format
* @param {string} req.lang - languages to load, ex: eng, eng+chi_tra * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
* @param {object} req.options - other options for loadLang function * @param {object} req.options - other options for loadLang function
* @param {object} req.params - parameters for tesseract * @param {object} req.params - parameters for tesseract
* @param {object} res - job instance * @param {object} res - job instance
*/ */
const handleRecognize = ({ const handleRecognize = ({
image, lang, options, params, image, langs, options, params,
}, res) => ( }, res) => (
handleInit(options, res) handleInit(options, res)
.then(() => ( .then(() => (
loadLanguage({ lang, options }, res) loadLanguage({ langs, options }, res)
.catch((e) => { .catch((e) => {
if (e instanceof DOMException) { if (e instanceof DOMException) {
/* /*
@ -206,7 +214,7 @@ const handleRecognize = ({
res.progress({ status: 'initializing api', progress }); res.progress({ status: 'initializing api', progress });
}; };
progressUpdate(0); progressUpdate(0);
handleParams(lang, params); handleParams(langs, params);
progressUpdate(0.5); progressUpdate(0.5);
const ptr = setImage(image); const ptr = setImage(image);
progressUpdate(1); progressUpdate(1);
@ -228,18 +236,18 @@ const handleRecognize = ({
* @access public * @access public
* @param {object} req - job payload * @param {object} req - job payload
* @param {array} req.image - binary image in array format * @param {array} req.image - binary image in array format
* @param {string} req.lang - languages to load, ex: eng, eng+chi_tra * @param {string} req.langs - languages to load, ex: eng, eng+chi_tra
* @param {object} req.options - other options for loadLang function * @param {object} req.options - other options for loadLang function
* @param {object} res - job instance * @param {object} res - job instance
*/ */
const handleDetect = ({ const handleDetect = ({
image, lang, options, image, langs, options,
}, res) => ( }, res) => (
handleInit(options, res) handleInit(options, res)
.then(() => ( .then(() => (
loadLanguage({ lang, options }, res) loadLanguage({ langs, options }, res)
.then(() => { .then(() => {
api.Init(null, lang); api.Init(null, getLangsStr(langs));
api.SetPageSegMode(TessModule.PSM_OSD_ONLY); api.SetPageSegMode(TessModule.PSM_OSD_ONLY);
const ptr = setImage(image); const ptr = setImage(image);

4
tests/detect.test.js
Unescape View File

@ -1,5 +1,4 @@
const { TesseractWorker, utils: { loadLang } } = Tesseract; const { TesseractWorker } = Tesseract;
const isBrowser = typeof window !== 'undefined' && typeof window.document !== 'undefined';
const IMAGE_PATH = 'http://localhost:3000/tests/assets/images'; const IMAGE_PATH = 'http://localhost:3000/tests/assets/images';
const loadLangOptions = { const loadLangOptions = {
langPath: 'http://localhost:3000/tests/assets/traineddata', langPath: 'http://localhost:3000/tests/assets/traineddata',
@ -9,7 +8,6 @@ const loadLangOptions = {
const getWorker = options => ( const getWorker = options => (
new TesseractWorker({ new TesseractWorker({
cacheMethod: 'readOnly', cacheMethod: 'readOnly',
...(isBrowser ? { workerPath: 'http://localhost:3000/dist/worker.dev.js' } : {}),
...loadLangOptions, ...loadLangOptions,
...options, ...options,
}) })

16
tests/recognize.test.js
Unescape View File

@ -1,6 +1,5 @@
const { TesseractWorker, utils: { loadLang } } = Tesseract; const { TesseractWorker } = Tesseract;
const isBrowser = typeof window !== 'undefined' && typeof window.document !== 'undefined';
const IMAGE_PATH = 'http://localhost:3000/tests/assets/images'; const IMAGE_PATH = 'http://localhost:3000/tests/assets/images';
const SIMPLE_TEXT = 'Tesseract.js\n'; const SIMPLE_TEXT = 'Tesseract.js\n';
const COMSIC_TEXT = 'HellO World\nfrom beyond\nthe Cosmic Void\n'; const COMSIC_TEXT = 'HellO World\nfrom beyond\nthe Cosmic Void\n';
@ -16,7 +15,6 @@ const loadLangOptions = {
const getWorker = options => ( const getWorker = options => (
new TesseractWorker({ new TesseractWorker({
cacheMethod: 'readOnly', cacheMethod: 'readOnly',
...(isBrowser ? { workerPath: 'http://localhost:3000/dist/worker.dev.js' } : {}),
...loadLangOptions, ...loadLangOptions,
...options, ...options,
}) })
@ -133,7 +131,7 @@ describe('recognize()', () => {
}).timeout(10000) }).timeout(10000)
)); ));
}); });
(isBrowser ? describe : describe.skip)('should read image from video DOM element (browser only)', () => { (isBrowser ? describe : describe.skip)('should read image from video DOM element (browser only)', () => {
FORMATS.forEach(format => ( FORMATS.forEach(format => (
it(`support ${format} format`, (done) => { it(`support ${format} format`, (done) => {
@ -160,22 +158,23 @@ describe('recognize()', () => {
let canvasDOM = null; let canvasDOM = null;
let imageDOM = null; let imageDOM = null;
let idx = 0; let idx = 0;
beforeEach(function cb(done) { beforeEach((done) => {
canvasDOM = document.createElement('canvas'); canvasDOM = document.createElement('canvas');
imageDOM = document.createElement('img'); imageDOM = document.createElement('img');
imageDOM.setAttribute('crossOrigin', 'Anonymous'); imageDOM.setAttribute('crossOrigin', 'Anonymous');
imageDOM.onload = () => { imageDOM.onload = () => {
canvasDOM.getContext('2d').drawImage(imageDOM, 0, 0); canvasDOM.getContext('2d').drawImage(imageDOM, 0, 0);
done(); done();
} };
imageDOM.setAttribute('src', `${IMAGE_PATH}/simple.${formats[idx++]}`); imageDOM.setAttribute('src', `${IMAGE_PATH}/simple.${formats[idx]}`);
idx += 1;
}); });
afterEach(() => { afterEach(() => {
canvasDOM.remove(); canvasDOM.remove();
imageDOM.remove(); imageDOM.remove();
}); });
formats.forEach(format => ( formats.forEach(format => (
it(`support ${format} format`, (done) => { it(`support ${format} format`, (done) => {
const worker = getWorker(); const worker = getWorker();
@ -189,5 +188,4 @@ describe('recognize()', () => {
}).timeout(10000) }).timeout(10000)
)); ));
}); });
}); });

Write Preview
Loading…
Cancel
Save