Incorrect fault message is sent from server side if request contained parse errors.

[ISzenkovszky]

If i send the server a SOAP message which contains parse errors I receive the following message:

<soap:Body>
 <soap:Fault>
  <faultcode>Client.Data</faultcode>
  <faultstring>Request was a fault</faultstring>
 </soap:Fault>
</soap:Body>

In the function KDSoapServerSocket::handleRequest we have the following bit of code:

    KDSoapMessage replyMsg;
    replyMsg.setUse(server->use());
    .
    .
    .
    // parse message
    KDSoapMessage requestMsg;
    KDSoapHeaders requestHeaders;
    KDSoapMessageReader reader;
    KDSoapMessageReader::XmlError err = reader.xmlToMessage(receivedData, &requestMsg, &m_messageNamespace, &requestHeaders, KDSoap::SOAP1_1);
    if (err == KDSoapMessageReader::PrematureEndOfDocumentError) {
        // qDebug() << "Incomplete SOAP message, wait for more data";
        // This should never happen, since we check for content-size above.
        return;
    } // TODO handle parse errors?

After calling xmlToMessage the requestMsg will contain a decent parse error message. But the replyMsg remains unchanged! So when we get to this part of the code:

    if (!replyMsg.isFault()) {
        makeCall(serverObjectInterface, requestMsg, replyMsg, requestHeaders, soapAction, path);
    }

we assume everything is fine and try to process the request on the server object with the requestMsg. At the start of the function makeCall we have:

    if (requestMsg.isFault()) {
        // Can this happen? Getting a fault as a request !? Doesn't make sense...
        // reply with a fault, but we don't even know what main element name to use
        // Oh well, just use the incoming fault :-)
        replyMsg = requestMsg;
        handleError(replyMsg, "Client.Data", QString::fromLatin1("Request was a fault"));
    }

where handleError is defined like this:

void KDSoapServerSocket::handleError(KDSoapMessage &replyMsg, const char *errorCode, const QString &error)
{
    qWarning("%s", qPrintable(error));
    const KDSoap::SoapVersion soapVersion = KDSoap::SOAP1_1; // TODO version selection on the server side
    replyMsg.createFaultMessage(QString::fromLatin1(errorCode), error, soapVersion);
}

So what happens is altough we want to use the "incoming" fault as reply fault, the handleError method calls the createFaultMessage function which replaces the parse error with the "Request was a fault" error.