Good XML Parser

I'm creating a tool for my 3D engine right now that converts Collada model format to my own format.

So far, I'm on the process on reading the said format which needs a xml parser for c++ using rapidxml.

Does all parser loads like let say. node->first_child() or node->next_sibling(). I mean.. sequentially? I guess that's the right word. Here take a look at my code:

Code:

---

for (node = node->first_node(); node; node = node->next_sibling())
                {
                        cout << "Name of node: " << node->name() << endl;

                        if( !strcmp(node->name(), "asset") )
                        {
                                for (xml_node<> *assetnode = node->first_node(); assetnode; assetnode = assetnode->next_sibling())
                                {
                                        if( !strcmp(assetnode->name(), "contributor") )
                                        {
                                                for (xml_node<> *contributornode = assetnode->first_node(); contributornode; contributornode = contributornode->first_node())
                                                {
                                                        if( !strcmp(contributornode->name(), "author") )
                                                        {
                                                                cout << contributornode->name() << " = '" << contributornode->value() << "'" << endl;
                                                        }
                                                }
                                        }
                                        else if( !strcmp(assetnode->name(), "created") )
                                        {
                                                cout << assetnode->name() << " = '" << assetnode->value() << "'" << endl;
                                        }
                                        else if( !strcmp(node->name(), "modified") )
                                        {
                                                cout << assetnode->name() << " = '" << assetnode->value() << "'" << endl;
                                        }
                                }
                        }
                        else if( !strcmp(node->name(), "library_effects") )
                        {
                                //Do something
                                node = node->next_sibling();
                        }
                }

---

So with that question on my mind, I look at the assimp source code and found out that they use irrXml which read a xml in different way. Now, they use something like mReader->read() (this one just read and don't check any siblings I guess. or I mean without passing a node to another node to another like the one in bold above.) then check something like mReader->getName() == "Aha!" ...

or maybe I just do it wrong?

What bothers me right now is that I think my codes are really messy. So, can anyone here suggest a good fast xml parser that loads maybe 50mb ~ 500mb+ xml fast.

Thank you
Sarah

Not sure if it is up to the task but I use tinyxml and it works really well. Perhaps it isn't the best but it is free and it is easy to use.

Another option might be xerces xml but I forget where to get it.