[ProDigit]

The perfect program for me would not be an automated program, but a manual one.
Since I use so many advanced feats to get 'perfect' formatting, which I think a good manual format, can oft be better than an automatic one.

I ALWAYS format my books from basic HTML code. In HTML I can clearly see what codes/brackets are opened and unclosed etc.
For this purpose my "perfect" program would need to consist of a mix of many advanced programs out there (which I will never find).

First I use Kompozer, to copy and paste HTML text. Its a WYSIWYG editor, so I can see the overall layout.
Then I can remove most hyperlinks with a rightclick.

Then I need notepad++ for it's advanced editing functions and macro's, something Kompozer or NVU does not have.

Once my HTML is cleaned up, and manually edited,I need a program to compress my ePub book with. Since Notepad++ can be used to create all files including toc.ncx, and mimetype and all others, all I'm missing is a compression program that will be intelligent enough to know that mimetype does not need to be compressed, and powerful enough to compress the rest of the files with a high compression ratio.
It also needs to be able to update files within it's archive. Currently I'm still looking for that program, but something like winrar or 7z, but for ebooks.

And lastly it needs to have an internal epub reader, to verify if the epub is displaying correctly.

For all these steps I use separate programs.

Why I DON'T want a program that does it all, is because 7z specializes in compression. Notepad++ specializes in search and replace, and html editing as well as any other text (non binary) document; and Kompozer is a WYSIWYG editor, that also is able to remove certain anchors and links, that will help me clean up my HTML.

A single program would need to be so extensive, that it will take more than 4 years to develop (which is less than 1/4th of what it took to create Kompozer, Winrar/7z, adobe Digital Editions, and Notepad++ combined).
A single program just performing the basics of what I wrote, will most likely be a buggy, and not a very efficient one.
Even if it was, it would probably put a lot of bogus code in a book. If I as a writer decide not to put a title in my book, I should have the freedom to do so.
If I decide that using 'class' and other CSS codes in my htmls, to save time and code, I should be able to use it, however if I find that using CSS takes too much code, and I can write my book simpler and better using no CSS, just basic HTML, I should also be able to do that.

I think large programs have issues with that. That's why I prefer ultimate control over what goes into my epub, and what does not.