You're unlikely to find a single offering that does all this, especially in the open source world. It's more likely that you'll end up relying on a mishmash of things, and may even need to chain some converters in order to get to HTML. (Eg PDF -> ps -> HTML)
OpenOffice supports conversion to HTML, and can be called from the command line.
http://pdftohtml.sourceforge.net/ looks reasonably good at converting pdf to html.
For Doc that is Word ML or OpenXML format it's conceivable that you could use XSLT transforms since both input and output formats are XML. I've seen some stylesheets floating around the net that do this, but YMMV.
Incidentally, why is there a specific requirement for open source? MS Powerpoint already supports save-as-HTML for example.
与恶龙缠斗过久,自身亦成为恶龙；凝视深渊过久,深渊将回以凝视…
