CODEWITHSUNDEEP
javascripthtmlgetimagedata
Giuseppe Romeo
Giuseppe Romeo

Reputation: 111

canvas getImageData isn't working

I have tried to getImageData but in the console I see this error:

Uncaught DOMException: Failed to execute 'getImageData' on 'CanvasRenderingContext2D': The canvas has been tainted by cross-origin data.
at HTMLImageElement.img.onload (file:///C:/Users/HOME/Desktop/programmi/HTML-Javascript/caso/graphic/imgData/arc/main.js:16:17)

This is my JavaScript and HTML code:

var canvas = document.getElementById('canvas'),
  ctx = canvas.getContext('2d'),
  width = canvas.width = 434, 
height = canvas.height = 362; 

var img = new Image();

img.src = 'https://cdn.pixabay.com/photo/2017/02/26/12/27/oranges-2100108_640.jpg';
//img.crossOrigin = 'Anonymous';

img.onload = function() {
  ctx.drawImage(img, 0, 0);
  var data = ctx.getImageData(10, 10, 11, 11);
  console.log(data)
};
<!DOCTYPE html>
<html>

<head>
  <title></title>
</head>

<body>
  <canvas id='canvas'></canvas>
  <script type="text/javascript" src='main.js'></script>
</body>

</html>

Upvotes: 1

Views: 2769

Answers (1)

David Edwards
David Edwards

Reputation: 844

Let me guess ... you're using Google Chrome as your browser. Which is the most problematic browser for this issue.

The issue centres upon what is known as cross-site security. in short, modern browsers are designed to prevent cross-site content from being loaded into a browser, except under special conditions, in part to stop malicious code being injected into web pages that you load.

This extends to images and HTML5 canvas data. The reason being, that some sneaky individuals discovered, early in the days of the HTML5 canvas being provided, that they could use it to provide an image snapshot of your current browser contents, and send that snapshot back to be perused at leisure. If you were browsing your bank's website at the time, and engaging in sensitive financial transactions online, whoever used this technique would quickly be able to find out about your finances, and possibly even pose as you to raid your account.

That's one of the reasons restrictions on cross site content were introduced.

Now, of course, there are legitimate reasons for having cross-site content. Such as having one's fonts or images stored in a separate repository. Trouble is, the cross-site restrictions impact upon legitimate uses as well as illegitimate ones.

Consequently, in order to use images cross-site, you have to do so in conformity with the CORS protocol. But, to do this, you need your images to be provided by a web server that's set up to handle CORS transactions. Simply setting the img.crossOrigin property of an Image object to "anonymous" won't work on its own: you need the server that's sending the images, to be set up to respond to the pre-flight options request that your browser will send, before allowing the image to be treated as acceptable from a security standpoint.

Which means, that to solve your problem, you need either, to:

[1] Install a local web server to perform this task for you - this option involves much tedium reading the web server manual, in order to set the server up properly, and much editing of configuration files;

[2] Write your own server to run under node.js or similar - this option involves even more tedium learning how to write your own web server, and make that server handle CORS transactions properly.

Now, if you're testing code offline in the "old school" manner, Firefox will let you access images from the same directory as your code, via the file:// protocol, and won't complain. Firefox apparently has enough intelligence to realise that an image being extracted from the same directory of your hard drive as your web page, constitutes a same-origin image.

If, however, you're using Google Chrome, it won't. At least, not unless you run it using special command line parameters. Even then, Chrome has a propensity to throw temper tantrums when asked to handle this sort of request. It's an issue I wrestle with frequently, and though some might be tempted to tell me to do my testing in Firefox to avoid these woes, Chrome's internal debugger is, for me at least, far more pleasant to use than Firefox's debugger, which, in my current Firefox installation, crawls like a snail on Mogadon, and exhibits a friendliness and smoothness of use reminiscent of a cocaine-soaked pit viper.

So, if you're using Chrome, because like me, you like its internal debugger, you're stuck with the two options I've given above. Either install a web server (Apache, Nginx, take your pick) or install Node.js and write your own. Neither option is easy.

Upvotes: 1

Related Questions