Parsing through a csv file in Qt

Question

Is anyone familiar with how to parse through a csv file and put it inside a string list. Right now I am taking the entire csv file and putting into the string list. I am try

Answer 1

Here is the code I usually use. I'm the author, consider this as-is, public domain. It has a similar feature-set and concept as CodeLurker's code except the state machine is represented differently, the code is a bit shorter.

bool readCSVRow (QTextStream &in, QStringList *row) {

    static const int delta[][5] = {
        //  ,    "   \n    ?  eof
        {   1,   2,  -1,   0,  -1  }, // 0: parsing (store char)
        {   1,   2,  -1,   0,  -1  }, // 1: parsing (store column)
        {   3,   4,   3,   3,  -2  }, // 2: quote entered (no-op)
        {   3,   4,   3,   3,  -2  }, // 3: parsing inside quotes (store char)
        {   1,   3,  -1,   0,  -1  }, // 4: quote exited (no-op)
        // -1: end of row, store column, success
        // -2: eof inside quotes
    };

    row->clear();

    if (in.atEnd())
        return false;

    int state = 0, t;
    char ch;
    QString cell;

    while (state >= 0) {

        if (in.atEnd())
            t = 4;
        else {
            in >> ch;
            if (ch == ',') t = 0;
            else if (ch == '\"') t = 1;
            else if (ch == '\n') t = 2;
            else t = 3;
        }

        state = delta[state][t];

        switch (state) {
        case 0:
        case 3:
            cell += ch;
            break;
        case -1:
        case 1:
            row->append(cell);
            cell = "";
            break;
        }

    }

    if (state == -2)
        throw runtime_error("End-of-file found while inside quotes.");

    return true;

}

• Parameter: in, a QTextStream.
• Parameter: row, a QStringList that will receive the row.
• Returns: true if a row was read, false if EOF.
• Throws: std::runtime_error if an error occurs.

It parses Excel style CSV's, handling quotes and double-quotes appropriately, and allows newlines in fields. Handles Windows and Unix line endings properly as long as your file is opened with QFile::Text. I don't think Qt supports old-school Mac line endings, and this doesn't support binary-mode untranslated line-endings, but for the most part this shouldn't be a problem these days.

Other notes:

• Unlike CodeLurker's implementation this intentionally fails if EOF is hit inside quotes. If you change the -2's to -1's in the state table then it will be forgiving.
• Parses x"y"z as xyz, wasn't sure what the rule for mid-string quotes was. I have no idea if this is correct.
• Performance and memory characteristics the same as CodeLurker's (i.e. very good).
• Does not support unicode (converts to ISO-5589-1) but changing to QChar should be trivial.

Example:

QFile csv(filename);
csv.open(QFile::ReadOnly | QFile::Text);

QTextStream in(&csv);
QStringList row;
while (readCSVRow(in, &row))
    qDebug() << row;