I'm currently preparing a large, multi-file document for publication, and intend making it available in three formats:

- HTML for reading on the Net.
- EPUB for the modern eReaders.
- PDF for printed versions.

Given that I expect several revised versions in coming months, I want to automate the conversion process as much as possible. I've selected 'pandoc' for investigation but haven't used it. Being a commandline utility has some advantages, although if it succeeds it'll probably end up as the back-end for one or more GUI front-ends.

'Sigil' does a nice job of converting an HTML suite into an EPUB, and I might stick with it.

The problem is PDF. It began life as Adobe's property but was subsequently open-sourced. Like all such things it has never achieved the 'open source feel' - Microsoft's .mht is another example. The biggest problem is editing PDF's. The gold standard is supposed to be Adobe's Acrobat. It's not only a monster as regards size, but it tries to take over your whole system; I've long avoided it.

'qpdfview' does everything I want from a reader, but none of the others I've tried for editing seems much good. The solution I'm looking at is to use pandoc to convert HTML into ODT, do the editing in OpenOffice, and export into PDF.

Does anyone have experience or suggestions in this area?

I would recommend Sphinx. By Sphinx, you'll write your publication in RST format, which is a rich but easy to understand format. Then, you can compile these rst files with Sphinx to HTML, epub, PDF, and other formats such as TexInfo. There are some example publications given on the website: http://sphinx-doc.org/examples.html

I would recommend you to look into this example. The link to the source code and compiled HTML and PDF are all online (you can also build the epub by yourself from the source).

Also, I've made a book with Sphinx: HTML output source

Thanks for that! Never heard of Sphinx, so it'll certainly be of interest.