Add base64 image support

docs/image-format.md

@@ -8,7 +8,9 @@ On a browser, an image can be:
 - an `img`, `video`, or `canvas` element
 - a `File` object (from a file `<input>`)
 - a path or URL to an accessible image
+- a base64 encoded image fits `data:image\/([a-zA-Z]*);base64,([^"]*)` regexp
 
 In Node.js, an image can be
 - a path to a local image
 - a Buffer storing binary image 
+- a base64 encoded image fits `data:image\/([a-zA-Z]*);base64,([^"]*)` regexp

src/browser/b64toU8Array.js

@@ -0,0 +1 @@
+module.exports = s => new Uint8Array(atob(s).split('').map(c => c.charCodeAt(0)));

src/browser/index.js

@@ -10,6 +10,7 @@
 const check = require('check-types');
 const resolveURL = require('resolve-url');
 const axios = require('axios');
+const b64toU8Array = require('./b64toU8Array');
 const { defaultOptions } = require('../common/options');
 const { version } = require('../../package.json');
 
@@ -38,6 +39,7 @@ const readFromBlobOrFile = (blob, res) => {
  * @access private
  * @param {string, object} image - image source, supported formats:
  *   string: URL string, can be relative path
+ *   string: base64 image
  *   img HTMLElement: extract image source from src attribute
  *   video HTMLElement: extract image source from poster attribute
  *   canvas HTMLElement: extract image data by converting to Blob
@@ -46,6 +48,11 @@ const readFromBlobOrFile = (blob, res) => {
  */
 const loadImage = (image) => {
   if (check.string(image)) {
+    // Base64 Image
+    if (/data:image\/([a-zA-Z]*);base64,([^"]*)/.test(image)) {
+      return Promise.resolve(b64toU8Array(image.split(',')[1]));
+    }
+    // Image URL
     return axios.get(resolveURL(image), {
       responseType: 'arraybuffer',
     })

src/browser/worker.js

@@ -10,6 +10,7 @@
 
 const check = require('check-types');
 const workerUtils = require('../common/workerUtils');
+const b64toU8Array = require('./b64toU8Array');
 
 /*
  * register message handler
@@ -42,7 +43,7 @@ workerUtils.setAdapter({
     }
     return global.TesseractCore;
   },
-  b64toU8Array: s => new Uint8Array(atob(s).split('').map(c => c.charCodeAt(0))),
+  b64toU8Array,
   writeFile: (path, data, type) => {
     postMessage({
       jobId: 'Download',

src/node/b64toU8Array.js

@@ -0,0 +1 @@
+module.exports = s => Buffer.from(s, 'base64');

src/node/index.js

@@ -13,6 +13,7 @@ const axios = require('axios');
 const isURL = require('is-url');
 const { fork } = require('child_process');
 const path = require('path');
+const b64toU8Array = require('./b64toU8Array');
 const { defaultOptions } = require('../common/options');
 
 const readFile = util.promisify(fs.readFile);
@@ -25,6 +26,7 @@ const readFile = util.promisify(fs.readFile);
  * @access public
  * @param {string} image - image source, supported formats:
  *   string: URL string or file path
+ *   string: base64 image
  *   buffer: image buffer
  * @returns {array} binary image in array format
  */
@@ -36,6 +38,10 @@ const loadImage = (image) => {
       .then(resp => resp.data);
   }
 
+  if (/data:image\/([a-zA-Z]*);base64,([^"]*)/.test(image)) {
+    return Promise.resolve(b64toU8Array(image.split(',')[1]));
+  }
+
   if (Buffer.isBuffer(image)) {
     return Promise.resolve(image);
   }

src/node/worker.js

@@ -10,6 +10,7 @@
 
 const check = require('check-types');
 const workerUtils = require('../common/workerUtils');
+const b64toU8Array = require('./b64toU8Array');
 
 let TesseractCore = null;
 
@@ -33,7 +34,7 @@ workerUtils.setAdapter({
     }
     return TesseractCore;
   },
-  b64toU8Array: s => Buffer.from(s, 'base64'),
+  b64toU8Array,
   writeFile: (path, data) => {
     const fs = require('fs');
     fs.writeFile(path, data, (err) => {