Well, maybe you should. That doesn't mean such a thing is possible.
An HTML document that is written out by a WYSIWYG editor will be
well-structured and conform to some kind of spec. But an HTML doc written by
a human won't be.
For example, you couldn't count on the tags staying nested:
<b><a href="foo">You'll see stuff like this</b> for example</a>
It might not be too unreasonable to use HTML as an interchange format with
other WYSIWYG editors like Netscape Composer.
Scott Barta, he author of the BeOS web browser NetPositive says that it's a
real drag just trying to be able to properly display all the same web pages
that IE and Navigator will display, for example:
<body bgcolor="AAFFOO">
That's not two zeroes folks - those are the letter O repeated twice. But your
friendly neighborhood web browser will read it.
What really should have been done early on would have been for all the web
browsers to give intelligent error diagnoses when they saw bad HTML; instead,
they were all hacked to read just about any HTML that came along.
That's why I don't think it's reasonable to write an HTML importer.
Mike