When using x/net/html to parse a HTML Spec to generate Go code in the colornames package, we wrote several useful and fairly generalised functions to interact with HTML parse trees such as a generalised node matching function type, match function generators for nodes by atom type & attributes and functions to recursively search and find matching nodes in the parse tree.

I have since then found these functions very useful in several other projects when using the experimental HTML parser. Therefore I was wondering if there is merit in providing these types of functions as a parseutils sub-package in the experimental repo. Happy to work on it.