bugfix - fixes utf8 exception in detect_paper_name

[aw-cloud]

Fixes #1411
utf8_decode called from detect_paper_name can throw, in particular when given invalid utf8. Adds a try / catch block in the calling function to account for this possibility. Behaviour is that invalid utf8 causes the paper name to be left blank and an error message is logged to stderr.

[ahrm]

Thanks. I am not a huge fan of try/catch, ideally I would like the issue to be handled without exceptions (but I am open to this if there is no other way). Unfortunately I can not reproduce the crash on my system. What happens if we replace utf8_decode with this:

std::wstring utf8_decode(const std::string& encoded_str) {
    return QString::fromUtf8(encoded_str).toStdWString();
}

?
Does it still crash?

[aw-cloud]

Changing utf8_decode to use the QString library function does stop the crash. It leaves "broken" unicode strings all over the place though, e.g. L"评点本金庸武侠全集 倚天屠龙记 连城决��" is the detected_paper_name from the sample document @ShadiZade provided. I don't know if that could cause problems with the database or other important stuff that relies on utf8_decode.

If using those replacement characters is safe then you can switch to QString::fromUtf8.

[NerdAlert2023]

Couldn't you just throw replacement characters instead?

Shouldn't be any crashes if you do that.

return QString::fromUtf8(buffer, strlen(buffer)).toStdWString();

That way no try or catches