I convert HTML to CSV regularly (multiple times a day). But I do not use Python; I use C. Actually I use flex to make filters which I compile as static binaries that read from stdin. This is in fact how I read HN. The HTML is converted to CSV and then the CSV is imported into a database.

Prior to using flex I primarily used sed. For many sites I still do; it's faster than having to compile, test, recompile.

If anyone has a website they want in CSV, and need something faster than Python or Ruby, just post the url. I like to think I am reasonably good at this, but I only do it for personal use on sites I'm interested in so who knows. For me HTML conversion to CSV and plain text is an art - I practice it every day.