C++ GUI Programming with Qt4: Input/Output

This chapter is from the book

C++ GUI Programming with Qt4, 2nd Edition

Learn More Buy

This chapter is from the book

This chapter is from the book 

C++ GUI Programming with Qt4, 2nd Edition

Learn More Buy

12. Input/Output

• Reading and Writing Binary Data
• Reading and Writing Text
• Traversing Directories
• Embedding Resources
• Inter-Process Communication

The need to read from or write to files or other devices is common to almost every application. Qt provides excellent support for I/O through QIODevice, a powerful abstraction that encapsulates "devices" capable of reading and writing blocks of bytes. Qt includes the following QIODevice subclasses:

QProcess, QTcpSocket, QUdpSocket, and QSslSocket are sequential devices, meaning that the data can be accessed only once, starting from the first byte and progressing serially to the last byte. QFile, QTemporaryFile, and QBuffer are random-access devices, so bytes can be read any number of times from any position; they provide the QIODevice::seek() function for repositioning the file pointer.

In addition to the device classes, Qt also provides two higher-level stream classes that we can use to read from, and write to, any I/O device: QDataStream for binary data and QTextStream for text. These classes take care of issues such as byte ordering and text encodings, ensuring that Qt applications running on different platforms or in different countries can read and write each other's files. This makes Qt's I/O classes much more convenient than the corresponding Standard C++ classes, which leave these issues to the application programmer.

QFile makes it easy to access individual files, whether they are in the file system or embedded in the application's executable as resources. For applications that need to identify whole sets of files to work on, Qt provides the QDir and QFileInfo classes, which handle directories and provide information about the files inside them.

The QProcess class allows us to launch external programs and to communicate with them through their standard input, standard output, and standard error channels (cin, cout, and cerr). We can set the environment variables and working directory that the external application will use. By default, communication with the process is asynchronous (non-blocking), but it is also possible to block on certain operations.

Networking and reading and writing XML are such substantial topics that we cover separately in their own dedicated chapters (Chapter 15 and Chapter 16).

Reading and Writing Binary Data

The simplest way to load and save binary data with Qt is to instantiate a QFile, to open the file, and to access it through a QDataStream object. QDataStream provides a platform-independent storage format that supports basic C++ types such as int and double, and many Qt data types, including QByteArray, QFont, QImage, QPixmap, QString, and QVariant, as well as Qt container classes such as QList<T> and QMap<K, T>.

Here's how we would store an integer, a QImage, and a QMap<QString, QColor> in a file called facts.dat:

QImage image("philip.png");

QMap<QString, QColor> map;
map.insert("red", Qt::red);
map.insert("green", Qt::green);
map.insert("blue", Qt::blue);

QFile file("facts.dat");
if (!file.open(QIODevice::WriteOnly)) {
    std::cerr << "Cannot open file for writing: "
              << qPrintable(file.errorString()) << std::endl;
    return;
}

QDataStream out(&file);
out.setVersion(QDataStream::Qt_4_3);

out << quint32(0x12345678) << image << map;

If we cannot open the file, we inform the user and return. The qPrintable() macro returns a const char * for a QString. (Another approach would have been to use QString::toStdString(), which returns a std::string, for which <iostream> has a << overload.)

If the file is opened successfully, we create a QDataStream and set its version number. The version number is an integer that influences the way Qt data types are represented (basic C++ data types are always represented the same way). In Qt 4.3, the most comprehensive format is version 9. We can either hard-code the constant 9 or use the QDataStream::Qt_4_3 symbolic name.

To ensure that the number 0x12345678 is written as an unsigned 32-bit integer on all platforms, we cast it to quint32, a data type that is guaranteed to be exactly 32 bits. To ensure interoperability, QDataStream standardizes on big-endian by default; this can be changed by calling setByteOrder().

We don't need to explicitly close the file, since this is done automatically when the QFile variable goes out of scope. If we want to verify that the data has actually been written, we can call flush() and check its return value (true on success).

The code to read back the data mirrors the code we used to write it:

quint32 n;
QImage image;
QMap<QString, QColor> map;

QFile file("facts.dat");
if (!file.open(QIODevice::ReadOnly)) {
    std::cerr << "Cannot open file for reading: "
              << qPrintable(file.errorString()) << std::endl;
    return;
}

QDataStream in(&file);
in.setVersion(QDataStream::Qt_4_3);

in >> n >> image >> map;

The QDataStream version we use for reading is the same as the one we used for writing. This must always be the case. By hard-coding the version number, we guarantee that the application can always read and write the data (assuming it is compiled with Qt 4.3 or any later Qt version).

QDataStream stores data in such a way that we can read it back seamlessly. For example, a QByteArray is represented as a 32-bit byte count followed by the bytes themselves. QDataStream can also be used to read and write raw bytes, without any byte count header, using readRawBytes() and writeRawBytes().

Error handling when reading from a QDataStream is fairly easy. The stream has a status() value that can be QDataStream::Ok, QDataStream::ReadPastEnd, or QDataStream::ReadCorruptData. Once an error has occurred, the >> operator always reads zero or empty values. This means that we can often simply read an entire file without worrying about potential errors and check the status() value at the end to see if what we read was valid.

QDataStream handles a variety of C++ and Qt data types; the complete list is available at http://doc.trolltech.com/4.3/datastreamformat.html. We can also add support for our own custom types by overloading the << and >> operators. Here's the definition of a custom data type that can be used with QDataStream:

class Painting
{
public:
    Painting() { myYear = 0; }
    Painting(const QString &title, const QString &artist, int year) {
        myTitle = title;
        myArtist = artist;
        myYear = year;
    }

    void setTitle(const QString &title) { myTitle = title; }
    QString title() const { return myTitle; }
    ...

private:
    QString myTitle;
    QString myArtist;
    int myYear;
};

QDataStream &operator<<(QDataStream &out, const Painting &painting);
QDataStream &operator>>(QDataStream &in, Painting &painting);

Here's how we would implement the << operator:

QDataStream &operator<<(QDataStream &out, const Painting &painting)
{
    out << painting.title() << painting.artist()
        << quint32(painting.year());
    return out;
}

To output a Painting, we simply output two QStrings and a quint32. At the end of the function, we return the stream. This is a common C++ idiom that allows us to use a chain of << operators with an output stream. For example:

out << painting1 << painting2 << painting3;

The implementation of operator>>() is similar to that of operator<<():

QDataStream &operator>>(QDataStream &in, Painting &painting)
{
    QString title;
    QString artist;
    quint32 year;

    in >> title >> artist >> year;
    painting = Painting(title, artist, year);
    return in;
}

There are several benefits to providing streaming operators for custom data types. One of them is that it allows us to stream containers that use the custom type. For example:

QList<Painting> paintings = ...;
out << paintings;

We can read in containers just as easily:

QList<Painting> paintings;
in >> paintings;

This would result in a compiler error if Painting didn't support << or >>. Another benefit of providing streaming operators for custom types is that we can store values of these types as QVariants, which makes them more widely usable—for example, by QSettings. This works provided that we register the type using qRegisterMetaTypeStreamOperators<T>() beforehand, as explained in Chapter 11 (p. 292).

When we use QDataStream, Qt takes care of reading and writing each type, including containers with an arbitrary number of items. This relieves us from the need to structure what we write and from performing any kind of parsing on what we read. Our only obligation is to ensure that we read all the types in exactly the same order as we wrote them, leaving Qt to handle all the details.

QDataStream is useful both for our own custom application file formats and for standard binary formats. We can read and write standard binary formats using the streaming operators on basic types (such as quint16 or float) or using readRawBytes() and writeRawBytes(). If the QDataStream is being used purely to read and write basic C++ data types, we don't even need to call setVersion().

So far, we loaded and saved data with the stream's version hard-coded as QDataStream::Qt_4_3. This approach is simple and safe, but it does have one small drawback: We cannot take advantage of new or updated formats. For example, if a later version of Qt added a new attribute to QFont (in addition to its point size, family, etc.) and we hard-coded the version number to Qt_4_3, that attribute wouldn't be saved or loaded. There are two solutions. The first approach is to embed the QDataStream version number in the file:

QDataStream out(&file);
out << quint32(MagicNumber) << quint16(out.version());

(MagicNumber is a constant that uniquely identifies the file type.) This approach ensures that we always write the data using the most recent version of QDataStream, whatever that happens to be. When we come to read the file, we read the stream version:

quint32 magic;
quint16 streamVersion;

QDataStream in(&file);
in >> magic >> streamVersion;

if (magic != MagicNumber) {
    std::cerr << "File is not recognized by this application"
              << std::endl;
} else if (streamVersion > in.version()) {
    std::cerr << "File is from a more recent version of the "
              << "application" << std::endl;
    return false;
}

in.setVersion(streamVersion);

We can read the data as long as the stream version is less than or equal to the version used by the application; otherwise, we report an error.

If the file format contains a version number of its own, we can use it to deduce the stream version number instead of storing it explicitly. For example, let's suppose that the file format is for version 1.3 of our application. We might then write the data as follows:

QDataStream out(&file);
out.setVersion(QDataStream::Qt_4_3);
out << quint32(MagicNumber) << quint16(0x0103);

When we read it back, we determine which QDataStream version to use based on the application's version number:

QDataStream in(&file);
in >> magic >> appVersion;

if (magic != MagicNumber) {
    std::cerr << "File is not recognized by this application"
              << std::endl;
    return false;
} else if (appVersion > 0x0103) {
    std::cerr << "File is from a more recent version of the "
              << "application" << std::endl;
    return false;
}

if (appVersion < 0x0103) {
    in.setVersion(QDataStream::Qt_3_0);
} else {
    in.setVersion(QDataStream::Qt_4_3);
}

In this example, we specify that any file saved with versions prior to 1.3 of the application uses data stream version 4 (Qt_3_0), and that files saved with version 1.3 of the application use data stream version 9 (Qt_4_3).

In summary, there are three policies for handling QDataStream versions: hard-coding the version number, explicitly writing and reading the version number, and using different hard-coded version numbers depending on the application's version. Any of these policies can be used to ensure that data written by an old version of an application can be read by a new version, even if the new version links against a more recent version of Qt. Once we have chosen a policy for handling QDataStream versions, reading and writing binary data using Qt is both simple and reliable.

If we want to read or write a file in one go, we can avoid using QDataStream and instead use QIODevice's write() and readAll() functions. For example:

bool copyFile(const QString &source, const QString &dest)
{
    QFile sourceFile(source);
    if (!sourceFile.open(QIODevice::ReadOnly))
        return false;

    QFile destFile(dest);
    if (!destFile.open(QIODevice::WriteOnly))
        return false;

    destFile.write(sourceFile.readAll());

    return sourceFile.error() == QFile::NoError
           && destFile.error() == QFile::NoError;
}

In the line where readAll() is called, the entire contents of the input file are read into a QByteArray, which is then passed to the write() function to be written to the output file. Having all the data in a QByteArray requires more memory than reading item by item, but it offers some advantages. For example, we can then use qCompress() and qUncompress() to compress and uncompress the data. A less memory-hungry alternative to using qCompress() and qUncompress() is QtIOCompressor from Qt Solutions. A QtIOCompressor compresses the stream it writes and decompresses the stream it reads, without storing the entire file in memory.

There are other scenarios in which accessing QIODevice directly is more appropriate than using QDataStream. QIODevice provides a peek() function that returns the next data bytes without moving the device position as well as an ungetChar() function that "unreads" a byte. This works both for random-access devices (such as files) and for sequential devices (such as network sockets). There is also a seek() function that sets the device position, for devices that support random access.

Binary file formats provide the most versatile and most compact means of storing data, and QDataStream makes accessing binary data easy. In addition to the examples in this section, we already saw the use of QDataStream in Chapter 4 to read and write Spreadsheet files, and we will see it again in Chapter 21, where we use it to read and write Windows cursor files.