exporting the convert methods

[VerioN1]

I noticed that you need to convert the image to an array of 8bit luma data, or rgb before passing it to decode_barcode. why not also expose the methods that do that?
Maybe I'm wrong and you can do this in vanilla js, but I couldn't find any example that do that without installing zxing (for js).
If there are other ways to pass image from js to the decoding function without installing zxing, I will be happy to know .

[hschimke]

I'll have to look up what JS image data looks like (canvas, image, I don't remember?). If it's a byte stream with predictable layout I can expose a method to handle it no problem.

I have some enhancements I want to make to this library, including the ability to set encode and decode hints for JS requests.

Would it be find for the function to be "yet another decode" function, where you hand it the JS image data directly and it handles the conversion in the background, or would a two step process be better, where one method generates the byte stream, and another decodes it?

[VerioN1]

I guess exposing a function that handles the conversion of the image to what ever decode you use, should be preferable since you are decoupling it from the decoding function. exposing multiple such conversion methods can be even better.
IMO most use cases of this wasm, would be file uploading - so maybe knowing how to handle canvas or event base64, and exposing a method that can convert it to the data type of the barcode decoder - would be nice.

you can contact me and I can help you test and improve it.

[VerioN1]

this is somehting I wrote to manipulate the Image data so I can pass it to zxing in my js app :

reader.onload = async (e) => {
const canvas = document.getElementById('imageCanvas') as HTMLCanvasElement;
if (!canvas) return;
const img = new Image();
img.src = e?.target?.result as string;
img.onload = async () => {
canvas.width = img.width;
canvas.height = img.height;
const ctx = canvas.getContext('2d');
if (!ctx) return;
ctx.drawImage(img, 0, 0);
const result = await zxingDecodeCanvas(canvas);
console.log(result);
if (result?.error) reject(result as BarcodeScannerResult);
resolve(result as BarcodeScannerResult);
};
};

[hschimke]

Got it, I'll get familiar with this and get it implemented. I'm finishing up adding maxicode detection to rxing (zxing and rxing currently only support it in a "pure barcode" image) and then I'll get to this. Probably by the end of the week.

[hschimke]

please check out version 0.1.3 on npm https://www.npmjs.com/package/rxing-wasm and let me know if that meets your requirements.

[VerioN1]

tested it and indeed the function converts the image data properly (since im not getting not found error).
What is weird is that the return value of the decode_barcode isn't right. adding a picture of a successful decode:

[VerioN1]

as you can see the in the console log - the result only has ptr in it (which I guess is a memory pointer address maybe?)

[VerioN1]

I did tried to make it async and await those functions (just to see if anything changes) and it returned the same result

[hschimke]

I pushed another version, but it shouldn't make a huge difference here.

The result should be an object, could you try using the text() method on the object and seeing if you get what you're looking for?

[VerioN1]

ohh my bad. It worked!
Only thing left is to let us configure the hints :)

[hschimke]

It's there! Check out the new decode_barcode_with_hints function. It takes a DecodeHintDictionary object, which you can create as you would any object, and which has a set_hint method.

I haven't written any documentation for it yet, so let me know if you have issues.

[VerioN1]

great! will you make a NPM lib with it?, or , if you don't have the time to handle it, I can make a lib out of it.

[hschimke]

https://www.npmjs.com/package/rxing-wasm

[hschimke]

I'm going to close this, but feel free to re-open or open a new one if you have additional issues. The NPM package is currently at 0.1.4, which tracks one version back from the current rxing release. I should have 0.1.5 our later tonight or tomorrow.

[VerioN1]

Im still not sure how to implement the decode hints map
example:

now also tried to make the possible formats in new Set or dict, didn't work.

[hschimke]

I plan to document all this next week. For that particular field it is looking for a string with a comma separated list of barcode format names.

[VerioN1]

apparently, im getting this error of out of memory bounds because of try harder set to true

[hschimke]

If possible, please attach a sample image that generates the error, I'll look into it.

[VerioN1]

that's the image:

[VerioN1]

literally turning on the TryHarder option - throws the error

[hschimke]

Just to verify, that should be:
(upc a) 123456789012

[VerioN1]

yes

[hschimke]

I haven't gotten a test bed up, so I cannot really debug this well.

Could you try passing "true" to the TryHarder set_hint and seeing if that work?

So far, I haven't gotten my own library to work without throwing an "unreachable" exception.

[hschimke]

Oh wow, I figured it out, I was passing in very wrong height and width data (does it show I haven't worked in node for almost a year?)

I get a successful working decode of that image, passing in the hint with hints.set_hint(DecodeHintTypes.TryHarder, "true");

[hschimke]

Here is the code I used, I think your issue is the "true" vs true.

import { convert_js_image_to_luma, decode_barcode_with_hints, DecodeHintDictionary, DecodeHintTypes } from "rxing-wasm";


const input = document.getElementById('image_file_input');
input.addEventListener('change', handleFiles);

function handleFiles(e) {
    const canvas = document.getElementById('cvs');
    const ctx = canvas.getContext('2d');
    const img = new Image;
    img.src = URL.createObjectURL(e.target.files[0]);
    img.onload = function () {
        canvas.width = img.width;
        canvas.height = img.height;
        ctx.drawImage(img, 0, 0, img.width, img.height);
    }
}

const scan_btn = document.getElementById('scan_btn');
scan_btn.addEventListener('click', onClickScan);

function onClickScan() {
    const canvas = document.getElementById('cvs');
    const context = canvas.getContext('2d');
    const imageData = context.getImageData(0, 0, canvas.width, canvas.height);
    const luma_data = convert_js_image_to_luma(imageData.data);
    const hints = new DecodeHintDictionary();
    hints.set_hint(DecodeHintTypes.TryHarder, "true");
    const result = decode_barcode_with_hints(luma_data, canvas.width, canvas.height, hints);
    alert(result.text());
}

[hschimke]

Alright, I added a demo application in the examples folder. It shows one way to handle basic hints and decoding.

Please let me know if your issue is resolved.

[VerioN1]

looks great. tried to play with it and was going great. I do think you will need to eventually create a table that maps between properties and thier type ( and maybe even a column of an example value at the so it will be explicit) .
one last this is that with TryHarder true nad UpcEanExtension - I managed to make the browser totally freeze while trying to scan this particular image.
I will be more than happy if you could have a look why would that be .

[VerioN1]

if you need a video of the browser behaver I can provide one. But try on your machine first with this code 👍:
`<script type="module">
import init, { BarcodeFormat, DecodeHintTypes, DecodeHintDictionary, decode_barcode_with_hints, convert_js_image_to_luma } from "./rxing_wasm.js";

  let imgFile = document.getElementById("image-input")
  let reader = new FileReader();

  imgFile.addEventListener('input', (e) => {
    reader.readAsDataURL(imgFile.files[0]);
  })
  reader.onload = function(event) {
    let img = new Image();
    img.src = event.target.result;
    img.onload = function() {
      let canvas = document.createElement('canvas');
      canvas.width = img.width;
      canvas.height = img.height;
      let ctx = canvas.getContext('2d');
      ctx.drawImage(img, 0, 0);
      let imgData = ctx.getImageData(0, 0, img.width, img.height);
      let data = imgData.data;
      let height = canvas.height;
      let width = canvas.width;
      init().then(() => {
        try{
          let luma8Data = convert_js_image_to_luma(data);
          const hintsDict = new DecodeHintDictionary()
          hintsDict.set_hint(DecodeHintTypes.TryHarder, "true");
          hintsDict.set_hint(DecodeHintTypes.POSSIBLE_FORMATS, 'UpcEanExtension')
          let parsedBarcode = decode_barcode_with_hints(luma8Data, width, height, hintsDict);
          console.log(parsedBarcode)
          console.log(parsedBarcode.text());
        }catch (e) {
          console.warn(e)
        }
      });
    }
  }



</script>`

[hschimke]

It breaks the cli app too (100% cpu in what looks like an infinite loop). I'll figure out what's going on with it.

[hschimke]

Two commits in the main repository fix the infinite loop. The image doesn't decode, but it no longer hangs.

rxing-core/rxing@6389d05
rxing-core/rxing@7526558

This update is pushed to npm as 0.1.6 (https://www.npmjs.com/package/rxing-wasm)

[hschimke]

Watch out for versions 0.1.7 - 0.1.11. A change in some dependencies has caused some havoc I'm still sorting out. I suggest pinning to 0.1.6 for now

[hschimke]

This is resolved with npm version 0.1.12

This also brings a lot more documentation, and the ability to get metadata from the result object. A javascript Map object is returned, containing parsed metadata.