I am not sure about the motive behind starting such project from scratch. But it will be kind of re-inventing wheel !
There is one good open source tool "HtmlAgilityPack" on codeplex at www.htmlagilitypack.codeplex.com
This also contains source code in case you want to extend the functionality to suit your requirement.
Hope that helps. If it does, mark the answer as solution/upvote.