(it’s just the individual woman recode files, BTW)
Here it is, a variable-level file with more than 800,000 cases. It’s only about 24MB zipped, but beware: it’s about 1.6GB unzipped.
You can get the code that produced this file here. Note that you have to put up with my filename conventions, not the original DHS filenames, but it’s not hard to understand what’s what — eg coir61.dta is Colombia, a DHS VI, first revision.
There are more than 244 datasets available from DHS, but you can always run my code on the extra files and append, saving you some minutes/hours and a few dimes’ worth of electricity.