I know which project you are talking about here, because I use it frequently, and when I wanted to replace the home-spun referrer-scanning logic for my blog (code I have to update all the time to keep up with all the different search engines) with something more complete, I looked at above source to see if I could use some of their code for that purpose. And yes, I also recoiled in horror (and gave up). Totally monolithic, horrible from a software engineering stand-point.
Would this be a better project if its code was more modular, with components useful outside its primary application? Yes. But is the reverse also true? Does a useful program become less useful because the implementation is ugly?
I was trying to find a web-log-analyser tool with nice pictures for the PHBs, and google pointed me towards a site on sourceforge, where a well-documented tool was available in perl, with all the nice graphs management likes.
I will now go on my way and find something better or (yuck) do it myself. Pity - it did look promising, and I am lazy and impatient.
I would not do that, especially when you are lazy and impatient. Your PHB is not going to look at the source, the project is, as you say, well-documented, popular and works as advertised. Just treat this open-source project as closed-source (for your own safety).