I must be doing something fundamentally wrong. I have a test program set up to read and display an xml file's contents so that I can study and learn how the data is stored and represented. I have enormous spreadsheet-generated xml files that drive our assembly equipment. Ultimately I need to re-order, edit, and create data in those files.
For now I'm just trying to figure out why PUGIXML seems to be ignoring node data and nodes without attributes.
This is the sample xml file I created when it became clear I wasn't finding everything in the real files:
<Node1 attr11="attribute11"
attr12="attribute12"
attr13="attribute13"
attr14="attribute14"
attr15="attribute15"
attr16="attribute16">
<Node11 attr111="attribute111">
<Value1>value1</Value1>
<Value2>value2</Value2>
<Value3>value3</Value3>
<Value4>value4</Value4>
</Node11>
<Node12/>
<Node13></Node13>
<Node14 attr141="attribute141"
attr142="attribute142"
attr143="attribute143"
attr144="attribute144">
<Value1>value1</Value1>
<Value2>value2</Value2>
<Value3>value3</Value3>
<Value4>value4</Value4>
</Node14>
<Node15>
<Value1>value1</Value1>
<Value2>value2</Value2>
<Value3>value3</Value3>
<Value4>value4</Value4>
</Node15>
</Node1>
Here is a partial listing of the C++ wrapper I'm creating around PUGIXML for compability with existing code:
class XMLFileImporter2
{
public:
// constructor
XMLFileImporter2( string strIn )
{
// stash the file name
m_strTheFileName = strIn;
// make the call to the pugi system
pugi::xml_parse_result result = m_xmlTheXML.load_file( m_strTheFileName.c_str() , pugi::parse_full );
m_strLastResultMessage = string( result.description() );
// did the xml read correctly?
m_bDidFileRead = ( m_strLastResultMessage == "No error" );
....
}
void test1( vector<string> &lstText )
{
// clear the vector
lstText.resize( 0 );
// interate through the xml data
for ( pugi::xml_node thisnode = m_xmlTheXML.first_child() ; thisnode ; thisnode = thisnode.next_sibling() )
{
testRecurse( thisnode , lstText , 0 );
}
}
void testRecurse( pugi::xml_node nodeIn , vector<string> &lstText , int iLevelIn )
{
// first, increment a local copy of the level
int iLocalLevel = iLevelIn + 1;
// temp vars
string strTemp , str1 , str2 , str3 , str4 , str5;
int iAttr = 0;
int iloop;
// build the attribute list for THIS node
for ( pugi::xml_attribute attr = nodeIn.first_attribute() ; attr ; attr = attr.next_attribute() )
{
// increment valid attribute count
iAttr++;
// init the string
strTemp = CUtility::makeStringFromInt( iLocalLevel ) + " : " + CUtility::makeStringFromInt( iAttr ) + " : ";
str1 = string( attr.name() );
str2 = string( attr.value() );
str3 = string( nodeIn.name() );
pugi::xml_node_type tNodeType = nodeIn.type();
str5 = getNodeType( tNodeType );
str4 = string( nodeIn.value() );
if ( str4 == "" ) str4 = "<blank>";
strTemp = strTemp + " Node Name = " + str3 + " Node Type = " + str5 + " Node Val = " + str4 + " Attr Name = " + str1 + " Attr Val = " + str2;
for ( iloop = 0 ; iloop < iLevelIn ; iloop++ )
strTemp = " " + strTemp;
// stash it
lstText.push_back( strTemp );
}
// recurse to any child nodes
for ( pugi::xml_node nextnode = nodeIn.first_child() ; nextnode ; nextnode = nextnode.next_sibling() )
{
testRecurse( nextnode , lstText , iLocalLevel );
}
}
...
};
When I create an instance of my wrapper, it causes the member PUGIXML document to load the file. This appears to be happening without any errors. After I know this is true, I execute the wrapper's member test traverse method to dump the entire node tree structure to the STL string vector, and I dump that to an unsorted list box. This is what my list box shows:
1 : 1 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr11 Attr Val = attribute11
1 : 2 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr12 Attr Val = attribute12
1 : 3 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr13 Attr Val = attribute13
1 : 4 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr14 Attr Val = attribute14
1 : 5 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr15 Attr Val = attribute15
1 : 6 : Node Name = Node1 Node Type = node_element Node Val = <blank> Attr Name = attr16 Attr Val = attribute16
2 : 1 : Node Name = Node11 Node Type = node_element Node Val = <blank> Attr Name = attr111 Attr Val = attribute111
2 : 1 : Node Name = Node14 Node Type = node_element Node Val = <blank> Attr Name = attr141 Attr Val = attribute141
2 : 2 : Node Name = Node14 Node Type = node_element Node Val = <blank> Attr Name = attr142 Attr Val = attribute142
2 : 3 : Node Name = Node14 Node Type = node_element Node Val = <blank> Attr Name = attr143 Attr Val = attribute143
2 : 4 : Node Name = Node14 Node Type = node_element Node Val = <blank> Attr Name = attr144 Attr Val = attribute144
Any empty node and any node without any explicit attributes seems to be getting lost. I've invoked the parse_full load_file option. I must be misunderstanding something very basic here....
