Add pdf output feature and rename oem params

README.md

@@ -39,7 +39,7 @@ Tesseract.js works with a `<script>` tag via local copy or CDN, with webpack via
 
 You can simply include Tesseract.js with a CDN like this:
 ```html
-<script src='https://unpkg.com/tesseract.js@v2.0.0-alpha.4/dist/tesseract.min.js'></script>
+<script src='https://unpkg.com/tesseract.js@v2.0.0-alpha.5/dist/tesseract.min.js'></script>
 ```
 
 After including your scripts, the `Tesseract` variable will be defined globally!

docs/examples.md

@@ -56,7 +56,7 @@ worker
   });
 ```
 
-### with whitelist chars (^2.0.0-alpha.4)
+### with whitelist char (^2.0.0-alpha.5)
 
 Sadly, whitelist chars is not supported in tesseract.js v4, so in tesseract.js we need to switch to tesseract v3 mode to make it work.
 
@@ -71,7 +71,7 @@ worker
     'http://jeroen.github.io/images/testocr.png',
     'eng',
     {
-      'init_oem': OEM.TESSERACT_ONLY,
+      'tessedit_ocr_engine_mode': OEM.TESSERACT_ONLY,
       'tessedit_char_whitelist': '0123456789-.',
     }
   )
@@ -82,3 +82,53 @@ worker
     console.log(result);
   });
 ```
+
+### with different pageseg mode (^2.0.0-alpha.5)
+
+Check here for more details of pageseg mode: https://github.com/tesseract-ocr/tesseract/blob/4.0.0/src/ccstruct/publictypes.h#L163
+
+```javascript
+import Tesseract from 'tesseract.js';
+
+const { TesseractWorker, PSM } = Tesseract;
+const worker = new TesseractWorker();
+
+worker
+  .recognize(
+    'http://jeroen.github.io/images/testocr.png',
+    'eng',
+    {
+      'tessedit_pageseg_mode': PSM.SINGLE_BLOCK,
+    }
+  )
+  .progress((p) => {
+    console.log('progress', p);
+  })
+  .then((result) => {
+    console.log(result);
+  });
+```
+
+### with pdf output (^2.0.0-alpha.5)
+
+```javascript
+import Tesseract from 'tesseract.js';
+
+const { TesseractWorker } = Tesseract;
+const worker = new TesseractWorker();
+
+worker
+  .recognize(
+    'http://jeroen.github.io/images/testocr.png',
+    'eng',
+    {
+      'tessedit_create_pdf': '1',
+    }
+  )
+  .progress((p) => {
+    console.log('progress', p);
+  })
+  .then((result) => {
+    console.log(result);
+  });
+```

docs/local-installation.md

@@ -10,9 +10,9 @@ In Node.js environment, the only path you may want to customize is languages/lan
 
 ```javascript
 const worker = Tesseract.TesseractWorker({
-  workerPath: 'https://unpkg.com/tesseract.js@v2.0.0-alpha.4/dist/worker.min.js',
+  workerPath: 'https://unpkg.com/tesseract.js@v2.0.0-alpha.5/dist/worker.min.js',
   langPath: 'https://tessdata.projectnaptha.com/4.0.0',
-  corePath: 'https://unpkg.com/tesseract.js-core@v2.0.0-beta.8/tesseract-core.wasm.js',
+  corePath: 'https://unpkg.com/tesseract.js-core@v2.0.0-beta.9/tesseract-core.wasm.js',
 });
 ```
 

examples/browser/basic.html

@@ -1,2 +1,2 @@
 <script src="/dist/tesseract.dev.js"></script>
-<input type="file" onchange="const worker = new Tesseract.TesseractWorker();worker.recognize(this.files[0]).progress(function(data){console.log(data)}).then(function(data){console.log(data)})">
+<input type="file" onchange="const worker = new Tesseract.TesseractWorker({ corePath: '../../node_modules/tesseract.js-core/tesseract-core.wasm.js' });worker.recognize(this.files[0]).progress(function(data){console.log(data)}).then(function(data){console.log(data)})">

examples/browser/demo.html

@@ -42,7 +42,9 @@ function recognizeFile(file){
 
   const { TesseractWorker } = Tesseract;
 
-  const worker = new TesseractWorker();
+  const worker = new TesseractWorker({
+    corePath: '../../node_modules/tesseract.js-core/tesseract-core.wasm.js',
+  });
 
 	worker.recognize(file,
 		document.querySelector('#langsel').value

src/browser/index.js

@@ -71,6 +71,25 @@ const loadImage = (image) => {
   return Promise.reject();
 };
 
+const downloadFile = (path, blob) => {
+  if (navigator.msSaveBlob) {
+    // IE 10+
+    navigator.msSaveBlob(blob, path);
+  } else {
+    const link = document.createElement('a');
+    // Browsers that support HTML5 download attribute
+    if (link.download !== undefined) {
+      const url = URL.createObjectURL(blob);
+      link.setAttribute('href', url);
+      link.setAttribute('download', path);
+      link.style.visibility = 'hidden';
+      document.body.appendChild(link);
+      link.click();
+      document.body.removeChild(link);
+    }
+  }
+}
+
 /*
  * Default options for browser worker
  */
@@ -83,7 +102,7 @@ exports.defaultOptions = {
    * If browser doesn't support WebAssembly,
    * load ASM version instead
    */
-  corePath: `https://unpkg.com/tesseract.js-core@v2.0.0-beta.8/tesseract-core.${typeof WebAssembly === 'object' ? 'wasm' : 'asm'}.js`,
+  corePath: `https://unpkg.com/tesseract.js-core@v2.0.0-beta.9/tesseract-core.${typeof WebAssembly === 'object' ? 'wasm' : 'asm'}.js`,
 };
 
 /**
@@ -108,7 +127,12 @@ exports.spawnWorker = (instance, { workerPath }) => {
   }
 
   worker.onmessage = ({ data }) => {
-    instance.recv(data);
+    if (data.jobId.startsWith('Job')) {
+      instance.recv(data);
+    } else if (data.jobId.startsWith('Download')) {
+      const { path, blob } = data;
+      downloadFile(path, blob);
+    }
   };
 
   return worker;

src/browser/worker.js

@@ -42,4 +42,13 @@ workerUtils.setAdapter({
     }
     return global.TesseractCore;
   },
+  b64toU8Array: s => new Uint8Array(atob(s).split('').map(c => c.charCodeAt(0))),
+  writeFile: (path, data, type) => {
+    const blob = new Blob([data], { type });
+    self.postMessage({
+      jobId: 'Download',
+      path,
+      blob,
+    });
+  },
 });

src/common/options.js

@@ -1,3 +1,5 @@
+const { OEM, PSM } = require('./types');
+
 module.exports = {
   defaultOptions: {
     /*
@@ -7,4 +9,16 @@ module.exports = {
      */
     langPath: 'https://tessdata.projectnaptha.com/4.0.0',
   },
+  /*
+   * default params for recognize()
+   */
+  defaultParams: {
+    tessedit_ocr_engine_mode: OEM.TESSERACT_LSTM_COMBINED,
+    tessedit_pageseg_mode: PSM.SINGLE_BLOCK,
+    tessedit_char_whiltelist: '',
+    tessedit_create_pdf: '0',
+    textonly_pdf: '0',
+    pdf_name: 'tesseract.js-ocr-result',
+    pdf_title: 'Tesseract.js OCR Result',
+  },
 };

src/common/pdf-ttf.js

@@ -0,0 +1 @@
+module.exports = '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';

src/common/types.js

@@ -2,7 +2,7 @@ module.exports = {
   /*
    * OEM = OCR Engine Mode, and there are 5 possible modes.
    *
-   * By default tesseract.js uses DEFAULT mode, which uses LSTM when possible.
+   * By default tesseract.js uses TESSERACT_LSTM_COMBINED mode, which uses LSTM when possible.
    * If you need to use some tesseract v3 features (like tessedit_char_whitelist),
    * you need to use TESSERACT_ONLY mode.
    *
@@ -14,4 +14,23 @@ module.exports = {
     DEFAULT: 3,
     COUNT: 4,
   },
+  /*
+   * PSM = Page Segmentation Mode
+   */
+  PSM: {
+    OSD_ONLY: '0',
+    AUTO_OSD: '1',
+    AUTO_ONLY: '2',
+    AUTO: '3',
+    SINGLE_COLUMN: '4',
+    SINGLE_BLOCK_VERT_TEXT: '5',
+    SINGLE_BLOCK: '6',
+    SINGLE_LINE: '7',
+    SINGLE_WORD: '8',
+    SINGLE_CHAR: '9',
+    SPARSE_TEXT: '10',
+    SPARSE_TEXT_OSD: '11',
+    RAW_LINE: '12',
+    COUNT: '13',
+  },
 };

src/common/workerUtils.js

@@ -9,7 +9,9 @@
  */
 const { readImage, loadLang } = require('tesseract.js-utils');
 const check = require('check-types');
+const pdfTTF = require('./pdf-ttf');
 const dump = require('./dump');
+const { defaultParams } = require('./options');
 
 /*
  * Tesseract Module returned by TesseractCore.
@@ -51,6 +53,58 @@ const setImage = (image) => {
   return data === null ? pix : data;
 };
 
+/**
+ * handleParams
+ *
+ * @name handleParams
+ * @function hanlde params from users
+ * @access private
+ * @param {string} lang - lang string for Init() 
+ * @param {object} customParams - an object of params
+ */
+const handleParams = (lang, customParams) => {
+  const {
+    tessedit_ocr_engine_mode,
+    ...params
+  } = {
+    ...defaultParams,
+    ...customParams,
+  };
+  api.Init(null, lang, tessedit_ocr_engine_mode);
+  Object.keys(params).forEach((key) => {
+    api.SetVariable(key, params[key]);
+  });
+};
+
+/**
+ * handleOutput
+ *
+ * @name handleOutput
+ * @function handle file output
+ * @access private
+ * @param {object} customParams - an object of params
+ */
+const handleOutput = (customParams) => {
+  const {
+    tessedit_create_pdf,
+    textonly_pdf,
+    pdf_name,
+    pdf_title,
+  } = {
+    ...defaultParams,
+    ...customParams,
+  };
+
+  if (tessedit_create_pdf === '1') {
+    const pdfRenderer = new TessModule.TessPDFRenderer(pdf_name, '/', textonly_pdf === '1');
+    pdfRenderer.BeginDocument(pdf_title);
+    pdfRenderer.AddImage(api);
+    pdfRenderer.EndDocument();
+    adapter.writeFile(`${pdf_name}.pdf`, TessModule.FS.readFile(`/${pdf_name}.pdf`), 'application/pdf');
+    TessModule._free(pdfRenderer);
+  }
+}
+
 /**
  * handleInit
  *
@@ -75,6 +129,7 @@ const handleInit = ({ corePath }, res) => {
     })
       .then((tessModule) => {
         TessModule = tessModule;
+        TessModule.FS.writeFile('/pdf.ttf', adapter.b64toU8Array(pdfTTF));
         api = new TessModule.TessBaseAPI();
         res.progress({ status: 'initialized tesseract', progress: 1 });
       });
@@ -123,22 +178,16 @@ const handleRecognize = ({
     .then(() => (
       loadLanguage({ lang, options }, res)
         .then(() => {
-          const OEM = check.undefined(params['init_oem'])
-            ? TessModule.OEM_DEFAULT
-            : params['init_oem'];
           const progressUpdate = (progress) => {
             res.progress({ status: 'initializing api', progress });
           };
           progressUpdate(0);
-          api.Init(null, lang, OEM);
-          progressUpdate(0.3);
-          Object.keys(params).filter(key => !key.startsWith('init_')).forEach((key) => {
-            api.SetVariable(key, params[key]);
-          });
-          progressUpdate(0.6);
+          handleParams(lang, params);
+          progressUpdate(0.5);
           const ptr = setImage(image);
           progressUpdate(1);
           api.Recognize(null);
+          handleOutput(params);
           const result = dump(TessModule, api);
           api.End();
           TessModule._free(ptr);

src/index.js

@@ -9,7 +9,7 @@
  */
 const utils = require('tesseract.js-utils');
 const TesseractWorker = require('./common/TesseractWorker');
-const { OEM } = require('./common/types');
+const types = require('./common/types');
 
 module.exports = {
   /** Worker for OCR, @see common/TesseractWorker.js */
@@ -17,5 +17,5 @@ module.exports = {
   /** Utilities for tesseract.js, @see {@link https://www.npmjs.com/package/tesseract.js-utils} */
   utils,
   /** Check ./common/types for more details */
-  OEM,
+  ...types,
 };

src/node/worker.js

@@ -33,4 +33,11 @@ workerUtils.setAdapter({
     }
     return TesseractCore;
   },
+  b64toU8Array: s => Buffer.from(s, 'base64'),
+  writeFile: (path, data) => {
+    const fs = require('fs');
+    fs.writeFile(path, data, () => {
+      console.log('File Write Succeeded!');
+    });
+  },
 });