Codes generated by QtQR aren't decoded correctly by mobile apps

[Ramiro Algozino]

I've tried to decode some sample codes and I found that the mobile apps (QR Droid and Kaywa reader) doesn't decode correctly the generated codes. It seems to be a problem with the utf8-BOM character added by qrtools.py to the text. Please confirm!

On the other hand, if we don't add the BOM char, zbar doesn't decode correctly non-ascii characters like "ñ".

[Ramiro Algozino (algozino)]

Ok, I can confirm this now. If we don't add the BOM character to the beginning of the text being encoded everything works OK. Except that zbar has problems decoding it correctly and makes QtQR display the non-ascii characters wrong. See:

http://sourceforge.net/projects/zbar/forums/forum/1072195/topic/3956406

[Ramiro Algozino (algozino)]

What happens is that mobile apps detect the data type of codes with the BOM character as TEXT and not as links, contact information, sms, etc. So I can think of 3 solutions / workarounds right now:

Solution #1: We add the BOM to the text data type only. That way zbar and mobile apps will decode everything OK (except the sms or emails bodys that have non-ascii chars by zbar)

Solution #2: Drop the BOM char and wait untill zbar fixes the bug. I reported it here:
https://sourceforge.net/tracker/?func=detail&aid=3322124&group_id=189236&atid=928515

Solution #3: Keep the BOM character so zbar decodes correctly non-ascii characters and brake decoding in mobile apps. (actually the BOM char shouldn't brake anything.. is just another char :-/)

Any more thoughts on this?

[David Green (david4dev)]

I'm in favour of #1 (seems the least disruptive to me) + hope the bug gets fixed in zbar.

[Ramiro Algozino (algozino)]

I think #1 is better too. I'm doing it that way until we have news from ZBar.. :-/

[Ramiro Algozino (algozino)]

From: http://stackoverflow.com/questions/2489048/qr-code-encoding-and-decoding-using-zxing

"The QR code spec doesn't allow anything but ISO-8859-1. Decoders happen to guess UTF-8 correctly sometimes, but can't always get it right. – Sean Owen Oct 14 '11 at 5:47"

So... maybe that's our solution :)

[Usama Akkad (damascene)]

This is also effect crypto currency addresses, as because of utf8-BOM, addresses aren't recognized by wallets.

https://bugs.launchpad.net/qr-tools/+bug/1896189

[Ramiro Algozino (algozino)]

This is still an issue, couldn't find a definitive solution. As a workaround I added a addBom property to the QR class that you can set to False to avoid adding the BOM character. QtQR also now allows you to disable the BOM.

[Dirk (dirk84)]

In which version of QtQR was the "disable BOM" option added?
I've apperantly installed an outdated QtQR as I cannot find such an option.
Version: 2.0~bzr33-2
Ubuntu 20.10 (groovy)

[Dirk (dirk84)]

Okay, after some research I found out that I have to switch to the daily builds:
https://code.launchpad.net/~qr-tools-developers/+recipe/python-qr-tools-daily
https://code.launchpad.net/~qr-tools-developers/+recipe/qtqr-daily

Thanks for imlementing this feature.

[Ramiro Algozino (algozino)]

Yes, switching to the daily should do the trick ;)

Anyway, for completeness, the fix landed on version 2.1.0~42