The test assumes that HTML files are encoded in UTF-8, and reports any character encoding issues that it finds. It would be a relatively simple piece of code.