Allowing cross-origin use of images and canvas

HTML provides a crossorigin attribute for images that, in combination with an appropriate CORS header, allows images defined by the <img> element that are loaded from foreign origins to be used in a <canvas> as if they had been loaded from the current origin.

See CORS settings attributes for details on how the crossorigin attribute is used.

Security and tainted canvases

Because the pixels in a canvas's bitmap can come from a variety of sources, including images or videos retrieved from other hosts, it's inevitable that security problems may arise.

As soon as you draw into a canvas any data that was loaded from another origin without CORS approval, the canvas becomes tainted. A tainted canvas is one which is no longer considered secure, and any attempts to retrieve image data back from the canvas will cause an exception to be thrown.

If the source of the foreign content is an HTML <img> or SVG <svg> element, attempting to retrieve the contents of the canvas isn't allowed.

If the foreign content comes from an image obtained from either as HTMLCanvasElement or ImageBitMap, and the image source doesn't meet the same origin rules, attempts to read the canvas's contents are blocked.

Calling any of the following on a tainted canvas will result in an error:

Attempting any of these when the canvas is tainted will cause a SecurityError to be thrown. This protects users from having private data exposed by using images to pull information from remote websites without permission.

Storing an image from a foreign origin

In this example, we wish to permit images from a foreign origin to be retrieved and saved to local storage. Implementing this requires configuring the server as well as writing code for the website itself.

Web server configuration

The first thing we need is a server that's configured to host images with the Access-Control-Allow-Origin header configured to permit cross-origin access to image files.

Let's assume we're serving our site using Apache. Consider the HTML5 Boilerplate Apache server configuration file for CORS images, shown below:

xml
<IfModule mod_setenvif.c>
  <IfModule mod_headers.c>
    <FilesMatch "\.(avifs?|bmp|cur|gif|ico|jpe?g|jxl|a?png|svgz?|webp)$">
      SetEnvIf Origin ":" IS_CORS
      Header set Access-Control-Allow-Origin "*" env=IS_CORS
    </FilesMatch>
  </IfModule>
</IfModule>

In short, this configures the server to allow graphic files (those with the extensions ".bmp", ".cur", ".gif", ".ico", ".jpg", ".jpeg", ".png", ".svg", ".svgz", and ".webp") to be accessed cross-origin from anywhere on the internet.

Implementing the save feature

Now that the server has been configured to allow retrieval of the images cross-origin, we can write the code that allows the user to save them to local storage, just as if they were being served from the same domain the code is running on.

The key is to use the crossorigin attribute by setting crossOrigin on the HTMLImageElement into which the image will be loaded. This tells the browser to request cross-origin access when downloading the image data.

Starting the download

The code that starts the download (say, when the user clicks a "Download" button), looks like this:

js
function startDownload() {
  let imageURL =
    "https://cdn.glitch.com/4c9ebeb9-8b9a-4adc-ad0a-238d9ae00bb5%2Fmdn_logo-only_color.svg?1535749917189";
  let imageDescription = "The Mozilla logo";

  downloadedImg = new Image();
  downloadedImg.crossOrigin = "anonymous";
  downloadedImg.addEventListener("load", imageReceived, false);
  downloadedImg.alt = imageDescription;
  downloadedImg.src = imageURL;
}

We're using a hard-coded URL (imageURL) and associated descriptive text (imageDescription) here, but that could easily come from anywhere. To begin downloading the image, we create a new HTMLImageElement object by using the Image() constructor. The image is then configured to allow cross-origin downloading by setting its crossOrigin attribute to "anonymous" (that is, allow non-authenticated downloading of the image cross-origin). An event listener is added for the load event being fired on the image element, which means the image data has been received. Alternative text is added to the image; while <canvas> does not support the alt attribute, the value can be used to set an aria-label or the canvas's inner content.

Finally, the image's src attribute is set to the URL of the image to download; this triggers the download to begin.

Receiving and saving the image

The code that handles the newly-downloaded image is found in the imageReceived() method:

js
function imageReceived() {
  const canvas = document.createElement("canvas");
  const context = canvas.getContext("2d");

  canvas.width = downloadedImg.width;
  canvas.height = downloadedImg.height;
  canvas.innerText = downloadedImg.alt;

  context.drawImage(downloadedImg, 0, 0);
  imageBox.appendChild(canvas);

  try {
    localStorage.setItem("saved-image-example", canvas.toDataURL("image/png"));
  } catch (err) {
    console.error(`Error: ${err}`);
  }
}

imageReceived() is called to handle the "load" event on the HTMLImageElement that receives the downloaded image. This event is triggered once the downloaded data is all available. It begins by creating a new <canvas> element that we'll use to convert the image into a data URL, and by getting access to the canvas's 2D drawing context (CanvasRenderingContext2D) in the variable context.

The canvas's size is adjusted to match the received image, the inner text is set to the image description, then the image is drawn into the canvas using drawImage(). The canvas is then inserted into the document so the image is visible.

Now it's time to actually save the image locally. To do this, we use the Web Storage API's local storage mechanism, which is accessed through the localStorage global. The canvas method toDataURL() is used to convert the image into a data:// URL representing a PNG image, which is then saved into local storage using setItem().

See also