Attempting to read nodes in Rapid XML resulting in error

Question

I have a class in my program that uses Rapid XML to write data to file. This process works fine. However when I attempt to read the same data, my program will always be halted by internal error catching code, explaining "next sibling returned NULL but attempted to read value anyways".

if (xmlFile.good())
 {
    vector<char> buffer((istreambuf_iterator<char>(xmlFile)), istreambuf_iterator<char>());
    buffer.push_back('\0');

    doc.parse<0>(&buffer[0]);
    root_node = doc.first_node("CityData");

    for(xml_node<> * bound_node = root_node->first_node("Boundaries"); bound_node; bound_node = bound_node->next_sibling())
    {
        if (bound_node->first_attribute("enabled")->value() != NULL)
        {
            int enabled = atoi(bound_node->first_attribute("enabled")->value());

            if (enabled == 1)
                boundaries = true; // Program globals
        }
    }

    if (boundaries)
    {
        for(xml_node<> * dimen_node = root_node->first_node("Dimensions"); dimen_node; dimen_node = dimen_node->next_sibling())
        {
            cityDim.x = atoi(dimen_node->first_attribute("x-val")->value()); // Program globals
            cityDim.y = atoi(dimen_node->first_attribute("y-val")->value());
        }
    }

An example of how the data appears in the XML file:

<CityData version="1.0" type="Example">
    <Boundaries enabled="1"/>
    <Dimensions x-val="1276" y-val="688"/>

If I add a breakpoint before the either loop attempts to reiterate and look at the values, we can see they are read from the first iteration, however the end criteria for the loop appears to be incorrect and upon next_sibling() the error occurs. I cannot understand the issue, as the code for the loop was copied from an example completely unmodified (aside from variable renaming) and appears correct to me (modifying it to node != NULL) does not help.

score 0 · Accepted Answer · answered Aug 12 '14 at 08:11

In the bound_node-loop the variable bound_node first points to <Boundaries enabled="1"> and you are able to read the attribute with the name enabled. After the call to next_sibling(), bound_node points to <Dimensions .../> and the call to first_attribute("enabled") will return a null pointer because this xml-node does not have an attribute with this name and the subsequent call to value() will cause the program to crash.

I do not understand why you are writing a loop over all nodes. If the xml-file looks like this

<CityData version="1.0" type="Example">
  <Boundaries enabled="1"/>
  <Dimensions x-val="1276" y-val="688"/>
</CityData>

Then you can extract the values like this:

    xml_node<> const * bound_node = root_node->first_node("Boundaries");
    if (bound_node)
    {
      xml_attribute<> const * attr_enabled = bound_node->first_attribute("enabled");
      if (attr_enabled)
      {
          int enabled = atoi(attr_enabled->value());

          if (enabled == 1)
              boundaries = true;
      }
    }

    if (boundaries)
    {
        xml_node<> const * dimen_node = root_node->first_node("Dimensions"); 

        if (dimen_node)
        {
          xml_attribute<> const * xval = dimen_node->first_attribute("x-val");
          xml_attribute<> const * yval = dimen_node->first_attribute("y-val");
          if (xval && yval)
          {
            cityDim.x = atoi(xval->value());
            cityDim.y = atoi(yval->value());
          }
        }
    }
  }

You can write some else-clauses to signal errors, of course.

Attempting to read nodes in Rapid XML resulting in error

1 Answers1