So, half an hour in, I thought it best to start documenting what I was doing.

unununium, the un-unbeldi-fier, uses the guts of Jekyll to do the menial tasks of post finding and un-mangling, so I quite literally get a stack of posts, which I can filter trivially. As one unbeldi post represents one day, I can easily create the day-summary page (seeing as most of it is done with Liquid transforms).

I built a trivial state machine in Ruby. Of course. I didn’t bother using state_machine which, to me, looked far too complex. On the other hand, my state machine didn’t work. Of course. So back to simple if/regexp parsing.

All up, it only took ~2 seconds for unununium to read through 307 unbeldi posts and un-unbeldi-ify them. That’s not half bad, considering most of that time is Jekyll’s internals scanning for posts (and that’s set to increase, unfortunately, as I multiply the number of posts by about ten). Even the LSI build looks to take Jekyll even longer to run.