Re: A bug that needs attention for windows

[Enrico Forestieri]

On Sat, Feb 24, 2007 at 07:00:42PM +0100, Jean-Marc Lasgouttes wrote:

> Bug 1247 is about recognizing properly letters in non-latin1 text.
> This is fixed on linux using 32bits wchar_t. However, on windows
> another strategy has to be found, as discussed here
> http://bugzilla.lyx.org/show_bug.cgi?id=1247
> 
> Enrico, Abdel, could you have a look?

I think that your idea of using Qt is the right one. When I run the
attached test program (linked to QtCore), I obtain the following output:

$ test-qt
Using ISO 8859-1:
¿ is NOT a letter.
à is a letter and toUpper(à) = À
Using ISO 8859-2:
¿ is a letter and toUpper(¿) = Z
à is a letter and toUpper(à) = R

I think that this is the correct result, apart the fact that the toUpper()
characters are not represented correctly for me in the ISO-8859-2 case, as
I have a latin1 locale.

-- 
Enrico

#include <iostream>
#include <QTextCodec>
#include <QString>

using std::cout;
using std::endl;

int main()
{
    QByteArray const localString = "¿à";
    char const *encoding[2] = { "ISO 8859-1", "ISO 8859-2" };
    for (int j = 0; j < 2; ++j) {
        QTextCodec *codec = QTextCodec::codecForName(encoding[j]);
        QString s = codec->toUnicode(localString);
        cout << "Using " << encoding[j] << ":" << endl;
        for (int i = 0; i < s.length(); ++i) {
            cout << localString.at(i) << " is ";
            if (s.at(i).isLetter()) {
                QString const upper = s.at(i).toUpper();
                cout << "a letter and toUpper(" << localString.at(i) << ") = "
                     << upper.toLocal8Bit().at(0) << endl;
            } else
                cout << "NOT a letter." << endl;
        }
    }
    return 0;
}