Blob download not working as before

[Cunibon]

I am trying to update my app from dart:html to package:web

I had this function:

Future<String?> downloadData(Uint8List data, {required String filename}) async {
  final blob = html.Blob([data], 'text/plain', 'native');
  html.AnchorElement(href: html.Url.createObjectUrlFromBlob(blob))
    ..setAttribute('download', filename)
    ..click()
    ..remove();
  return null;
}

and updated it to this:

Future<String?> downloadData(Uint8List data, {required String filename}) async {
  final blob = web.Blob(
    [data.buffer.toJS].toJS,
    web.BlobPropertyBag(type: 'text/plain', endings: 'native'),
  );
  web.HTMLAnchorElement()
    ..href = web.URL.createObjectURL(blob)
    ..setAttribute('download', filename)
    ..click()
    ..remove();
  return null;
}

But the excel I want to download says it is corrupted and needs to be fixed to work.

I took a look at the resulting excel file in VSCode and found that there are subtle changes to some of the bytes randomly every time I run it. With the same code and the same data

For example the start line:

1. "PK��������&P�Y�bi�������������xl/drawings/drawing1.xml��Kn�0������������F�6�'(��ړ��hƔp�ZP�RY�XZ������zt��Bb�|#�i)"
2. "PK��������5P�Y�bi�������������xl/drawings/drawing1.xml��Kn�0������������F�6�'(��ړ��hƔp�ZP�RY�XZ������zt��Bb�|#�i)"

You can see that "&P" turned into "5P"

Additionally the file size increased 7x from 5kb to 36kb. This is due to the file ending in 27777 additional null characters

I attached both the the broken and the legit version as excel files
broken.xlsx
legit.xlsx

[srujzs]

Can you share a repro where you create the Uint8List and call downloadData?

The only difference in calls that I can tell is that in the package:web example, you are calling toJS on the ArrayBuffer instead of on the view i.e. [data.buffer.toJS].toJS instead of [data.toJS].toJS. Maybe that shouldn't make a difference but I don't know how Blob interprets that.

[ykmnkmi]

You can also use Uri.dataFromBytes:

void downloadData(Uint8List data, {required String filename}) {
  web.HTMLAnchorElement()
    ..href = UriData.fromBytes(data, mimeType: 'text/plain').toString()
    ..setAttribute('download', filename)
    ..click()
    ..remove();
}

Is using text/plain format and setting endings for xlsx file right? Not application/vnd.openxmlformats-officedocument.spreadsheetml.sheet?