#include <QApplication>
#include <QDebug>
#include <QFile>

#include <QListWidget>
#include <QWidget>

QString eliminateBadChars(QByteArray d) {
    QString r = QString::fromUtf8(QByteArray::fromHex(d));
    for(int i = 0; i < r.size(); i++) {
        QChar c = r.at(i);

        if(c.isSpace() || !c.isPrint() || c.isMark() ||
        c.category() == QChar::Other_Control ||
        c.category() == QChar::Other_Format ||
        c.category() == QChar::Other_NotAssigned || c.category() == QChar::Other_PrivateUse ||
        // Comment the code line below to stop crashing.
        // If this particular category of characters is eliminated, the utf8 string can be displayed, if it isn't - Qt crashes
        /*c.category() == QChar::Symbol_Other ||*/
        c.category() == QChar::NoCategory)
            r.replace(i, 1, "");
    } return r;
}

bool hasBadChar(QByteArray d, QChar::Category *cat) {
    QString s = QString::fromUtf8(QByteArray::fromHex(d));
    for(int i = 0; i < s.size(); i++) {
        QChar c = s.at(i);
        *cat = c.category();

        if(c.isSpace() || !c.isPrint() || c.isMark() ||
        c.category() == QChar::Other_Control ||
        c.category() == QChar::Other_Format ||
        c.category() == QChar::Other_NotAssigned || c.category() == QChar::Other_PrivateUse ||
        c.category() == QChar::Symbol_Other ||
        c.category() == QChar::NoCategory)
            return true;
    } return false;
}

void testForBadChars(QListWidget *lb, QListWidget *la, QMap<QChar::Category, uint> *m, QString name) {
    QFile f(QCoreApplication::applicationDirPath() + "/" + name); if(!f.open(QIODevice::ReadOnly)) {
        qDebug() << "ERR: " << name << " failed to open";
        return;
    }

    int c = 0, t = 0;
    while(!f.atEnd()) {
        QByteArray l = f.readLine(); l = l.replace('\n', "");

        QChar::Category cat;
        if(hasBadChar(l, &cat)) {
            c++;
            lb->insertItem(0, QByteArray::fromHex(l));
            la->insertItem(0, eliminateBadChars(l));

            m->insert(cat, m->value(cat, 0) + 1);
        }
        t++;
    }

    qDebug() << name << "   ->   " << c << "/" << t;
}

int main(int argc, char *argv[]) {
    QApplication a(argc, argv);

    QWidget *w = new QWidget;
    w->resize(800, 600);
    w->show();

    QListWidget *lb = new QListWidget(w);
    QListWidget *la = new QListWidget(w);

    lb->move(0,0);
    lb->resize(w->width() / 2, w->height());
    lb->show();
    la->move(lb->width(), 0);
    la->resize(lb->size());
    la->show();

    // Map out bad chars
    QMap<QChar::Category, uint> m;

    //testForBadChars(lb, la, &m, "users.txt");
    //testForBadChars(lb, la, &m, "users.19.txt");
    //testForBadChars(lb, la, &m, "users.20.txt");
    testForBadChars(lb, la, &m, "users.50.txt");
    //testForBadChars(lb, la, &m, "users.86.txt");
    //testForBadChars(lb, la, &m, "users.634.txt");

    // List Bad Chars
    qDebug() << "--- BAD CHAR CATEGORIES ---";
    for(QMap<QChar::Category, uint> :: iterator i = m.begin(); i != m.end(); i++) {
        qDebug() << i.key() << i.value();
    }
    qDebug()<<111;

    return a.exec();
}