help--what is the basic principle of Tidy for HTML Parsing?

Dear all,

Sorry for disturbing you!

I am writing to you to seek your kind help. As we know, Tidy is a powerful 
tool to parse and correct a HTML document. I have visited the website 
http://tidy.sourceforge.net/ and downloaded source code of the Tidy. 
However, I found that there are so many scripts that it is difficult to 
clearly understand the basic principle of Tidy, i.e., how Tidy parses a 
HTML document and How the tree is build. Understanding these questions 
will be helpful to correctly use (and further develop) the Tidy. In the 
above referenced website, I cannot find the answers to these questions. 
Therefore, I expect your kind help!

Thanks for your attention,

Li MA
IBM China Research Lab
E-Mail : MaLLi@cn.ibm.com