js check if an image truncated/corrupted data

Question

I'm looking for how to detect an image data is truncated\corrupted. For example this picture:

the data image is not complete (It's more tangible on IE, and its noted as warning in firefox console), but img.onerror not fired, and img.completed is true.

demo: https://jsfiddle.net/7dd0ybb4/

var img = document.getElementById('MyPicture');

img.onerror = () => alert('error img');  
img.onload = () =>  console.log(img.complete); //true

img.src = "https://i.stack.imgur.com/nGkok.jpg";

I want a way to know that. if an image have invalid data.

The reason is 'coz the image is completely loaded and is not a corrupt image. If the network activity is interrupted onerror is fired. — Akhil Arjun, Jan 04 '17 at 10:59
Can you provide me with a full base64 data for this image, to test it out? I almost found the solution :) — rossanmol, Jan 07 '17 at 16:23
If Akhil Arjun is right, then it means the problem is at an earlier stage. For example, if it's an image someone else uploaded to the site, maybe the upload has been truncated. — David Knipe, Jan 08 '17 at 11:42
Hi refer this link I hope it will help you.. http://stackoverflow.com/questions/32459621/how-to-handle-image-corrupt-or-truncated-in-firefox — Suresh B, Jan 11 '17 at 10:55
@SureshB, this bag its not related. here the image is realy invalid. and i want know that from the code. — dovid, Jan 11 '17 at 12:41
You can decode image to arraybuffer and then check signature for png (file integrity check chapter here - http://www.libpng.org/pub/png/pngintro.html) or for jpeg you can get a data length value from 04h byte or image size from bytes situated later and check the data length with expected value which one can be calculated by the expected image size. Just an idea — Panama Prophet, Jan 13 '17 at 13:00

score 8 · Accepted Answer · edited May 23 '17 at 11:54

8

You can decode an image to an array of bytes:

var src = 'here comes your base64 data'
var imageData = Uint8Array.from(atob(src.replace('data:image/jpeg;base64,', '')), c => c.charCodeAt(0))

JPEGs must start with the bytes FF D8 and end with FF D9, so we check if last two elements of the created array buffer is 255 and 217. See live example.

var imageCorrupted = ((imageData[imageData.length - 1] === 217) && (imageData[imageData.length - 2] === 255));

We can use a similar check for PNGs (live example), which end with an IEND chunk containing this sequence of bytes:

// in hex: 00 00 00 00 49 45 4e 44 ae 42 60 82
var sequence = [0, 0, 0, 0, 73, 69, 78, 68, 174, 66, 96, 130];

edited May 23 '17 at 11:54

Community

1
1

answered Jan 13 '17 at 13:06

Panama Prophet

1,027
6
6

It does not work for [this `src`](https://0bin.net/paste/VkxPPlgc9DWme4gw#jqp+X4qoIxlQi91s-mzMbW0S4INdxlMgcBZC5ek3kda), because `imageCorrupted == false`. – BuZZ-dEE Apr 19 '17 at 13:57
What about BMPs? – carestad Oct 14 '19 at 08:54
Be careful with the solution here. The EOI bytes ( i.e. 255, 217 ) doesn't always happen to be the last two bytes. Most of encoders usually drop the data after EOI bytes. – Ravi Ranjan Aug 07 '20 at 09:15

score 4 · Answer 2 · answered Jan 08 '17 at 11:26

you could check corrupted pixels by canvas like this code

var ctx = document.getElementById('canvas').getContext('2d');
var img = new Image();
img.onload = function(){
    ctx.drawImage(img,0,0);
    var height=this.height;
    var width=this.width;
    var hCenter=Math.round(width/2);
    var corruptedRowsCount=0;
    for(var row=height;row>0;row--){
      var c = ctx.getImageData(hCenter, row, 1, 1).data;
      if(c[0]==0 && c[1]==0 && c[2]==0 && c[3]==0)
          corruptedRowsCount++;
      else 
          break;
    }
    console.log(Math.round(corruptedRowsCount*100/height)+" precent of image is corrupted");
};

img.src="data:image/jpeg;base64,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";

<canvas id="canvas">

+1 very nice! but still not quite answer. because your solution assumes transparent pixels are invalid, and it is true to JPG, but not for PNG. see https://jsfiddle.net/7L01yfb2/1/. and the image data its 100% valid. — dovid, Jan 08 '17 at 16:50
i made it better https://jsfiddle.net/7L01yfb2/7/ but this solution is not complete for png. — fingerpich, Jan 08 '17 at 19:51

Sámal Rasmussen · Answer 3 · 2021-10-19T11:14:57.277

For png files, I decided that I wanted a solution that did more than just check the first and last bytes of the file to see if they were valid. I wanted a solution that actually made use of the embedded crc codes for each chunk inside the png file to check all the data inside the file for corruption.

Luckily I found a mature library that does this: https://github.com/lukeapage/pngjs

Unfortunately it is rather large. Almost 500 kb for the browser version. It is a fully featured png file manipulation library, which makes this understandable. If you have a use case that needs all these features then you might be ok with paying for the size. But it is too heavy for the simple use case of just wanting to validate a png file.

So I cut out the validation code from it and made a simple little 2.5 kb npm package with just the validation code in it: https://www.npmjs.com/package/png-validator

js check if an image truncated/corrupted data

3 Answers3

Linked

Related