auto_link escaping issues

Well, it looks like this forum is doing this correctly. Any
ruby-forum.com dev's out there that can explain how they are both
escaping HTML and calling auto linking URLs?

I had the same problem. Here is a simplified version that worked for me:

    def auto_link_urls (text)
      url_regex = /
        ([a-zA-Z][0-9a-zA-Z+\-\.]*:\/\/|www\.)
        [0-9a-zA-Z;\/?:@&=+$\.\-_!~*'()%]+
        (\#[0-9a-zA-Z;\/?:@&=+$\.\-_!~*'()%]+)?
        /x

      text.gsub(url_regex) do |link|
        link = 'http://' + link if $1 == 'www.'
        link_text = block_given? ? yield(link) : link
        '<a href="' + link + '">' + link_text + '</a>'
      end
    end

It doesn't detect already linked URLs (I didn't need it) but it allows a
wider range of URLs (ftp://…, for example) and works with HTMLized
URLs (…?foo=1&amp;bar=2…) and some other special case original
version doesn't seem to detect (http://www.example.com/?foo).