The TinyXML++ classes are all wrappers around the corresponding classes within TinyXML.
There is no reason to create TinyXML++ objects on the heap, using new
, because the memory is managed for you. If you choose to use new
to create TinyXML++ objects, you will always need to use delete
to clean up.
Basically, TinyXML++ objects are just wrappers around TinyXML pointers.
When using the original TinyXML, every function returns a value indicating success or failure. A programmer would have to check that value to ensure the function succeeded.
Example:
Load a document TiXmlDocument doc( pFilename ); if ( !doc.LoadFile() ) return; Get a node TiXmlElement* pElem = doc.FirstChildElement(); if ( !pElem ) return; Get the node we want pElem = pElem->NextSibling(); if ( !pElem ) return; do something useful here
An alternative was to use TiXmlHandle, which allows for function chaining by checking the intermediate function return values:
Example:
Load a document TiXmlDocument doc(pFilename); if (!doc.LoadFile()) return; Make a document handle TiXmlHandle hDoc(&doc); Get an element by using the handle to chain calls Note the conversion of the TiXmlHandle to the TiXmlElement* - .Element() TiXmlElement* pElem = hDoc.FirstChildElement().NextSibling().Element(); if ( !pElem ) return; do something useful here
With TinyXML++, if there is an error during a function call, it throws an exception. This means that a programmer can assume that every function is successful, as long as the functions are enclosed in a try-catch block.
Example:
try { Load a document ticpp::Document doc( pFilename ); doc.LoadFile(); Get an element by chaining calls - no return values to check, no TiXmlHandle ticpp::Element* pElem = doc.FirstChildElement()->NextSibling(); do something useful here } catch( ticpp::Exception& ex ) { If any function has an error, execution will enter here. Report the error std::cout << ex.what(); }
When using TinyXML, a programmer either needs to convert values to and from strings, or choose from one of many overloads to get the value in the desired type.
Example:
Load a document TiXmlDocument doc( pFilename ); if ( !doc.LoadFile() ) return; Get a node TiXmlElement* pElem = doc.FirstChildElement(); if ( !pElem ) return; Get the node we want pElem = pElem->NextSibling(); if ( !pElem ) return; Get the attribute as a string, convert to int const char* pszAttr = pElem->Attribute( "myAttribute" ); int attr = atoi( pszAttr ); Get the attribute as an int int attr2; if ( TIXML_SUCCESS != pElem->QueryIntAttribute( "myAttribute", &attr2 ) ) { return; } Get the attribute as a double double attr3; if ( TIXML_SUCCESS != pElem->QueryDoubleAttribute( "myAttribute", &attr3 ) ) { return; } Get the attribute as a float float attr4; if ( TIXML_SUCCESS != pElem->QueryFloatAttribute( "myAttribute", &attr4 ) ) { return; }
TinyXML++ uses templates for automatic type conversion.
Example:
try { Load a document ticpp::Document doc( pFilename ); doc.LoadFile(); Get an element by chaining calls - no return values to check, no TiXmlHandle ticpp::Element* pElem = doc.FirstChildElement()->NextSibling(); GetAttribute can determine the type of the pointer, and convert automatically Get the attribute as a string std::string attr; pElem->GetAttribute( "myAttribute", &attr ); Get the attribute as an int int attr2; pElem->GetAttribute( "myAttribute", &attr2 ); Get the attribute as an float float attr3; pElem->GetAttribute( "myAttribute", &attr3 ); Get the attribute as an double double attr4; pElem->GetAttribute( "myAttribute", &attr4 ); Get the attribute as an bool bool attr5; pElem->GetAttribute( "myAttribute", &attr5 ); } catch( ticpp::Exception& ex ) { If any function has an error, execution will enter here. Report the error std::cout << ex.what(); }
TinyXML has two ways to iterate:
First Method:
for( child = parent->FirstChild( false ); child; child = child->NextSibling( false ) )
Second Method:
child = 0;
while( child = parent->IterateChildren( child ) )
Although both methods work quite well, the syntax is not familiar. TinyXML++ introduces iterators:
ticpp::Iterator< ticpp::Node > child; for ( child = child.begin( parent ); child != child.end(); child++ )
Iterators have the added advantage of filtering by type:
Only iterates through Comment nodes ticpp::Iterator< ticpp::Comment > child; for ( child = child.begin( parent ); child != child.end(); child++ )
Only iterates through Element nodes with value "ElementValue" ticpp::Iterator< ticpp::Element > child( "ElementValue" ); for ( child = child.begin( parent ); child != child.end(); child++ )
Finally, Iterators also work with Attributes
ticpp::Iterator< ticpp::Attribute > attribute; for ( attribute = attribute.begin( element ); attribute != attribute.end(); attribute++ )