I suggest having a look at http://en.wikipedia.org/wiki/LuaTeX It is still in beta, but once the power of LuaTeX is unleashed, it will be the only used TeX engine (in two or three years). This is because it combines the quality of TeX's typesetting algorithms (+ unicode/opentype/...) with the ability to program in Lua, a very decent programming language. In "Lua mode", you have access to all TeX's internal parameters and algorithms, so you can even do the line breaking and pdf writing from inside lua. That means you don't have to use TeX's input language any more, a language that keeps many people away from using TeX. Now that you have everything on the Lua side, there is no reason why not to read XML, RTF, Wiki-Markup and other file formats and use the TeX algorithms to typeset these files.

TeX internally works with nodes. Every character you type, every space you use (like the space between paragraphs), every color command will be changed to a nodelist. The boxes and linebreaking algorithm operate on these nodes. In LuaTeX (in lua) you can create nodes manually and feed this list into the linebreaking algorithm and you get back a box of a perfectly nice typeset paragraph. Then you can even operate on these nodes again (or just before linebreaking) and ship out this paragraph into the pdf file. Nobody cares where you get the input from. There are several libraries included with LuaTeX and available for inclusion, such as LPEG (for really nice and sophisticated parsing), luasocket (internet access), luazip, etc. Even luaexpat (xml) and luasql (Database access) are available, so that makes LuaTeX perfectly suited for database publishing (which is what I am doing)

No other TeX engine can do that at the moment!

For the german speaking hackers, I am running the site

http://www.luatex.de

where I write technical stuff about LuaTeX (and some announcements), but beware: all in german.

[Edit: here is the site with google translation to english:

http://translate.google.com/translate?js=y&prev=_t&h...

]