I don’t mind being wrong, so long as it’s in the right direction.
Quite to the contrary though:
The missing ingredient was that tdu, like ccze and some other utilities, really doesn’t do anything by itself — instead, it expects data to be piped through it. Ergo,
du -a | tdu
gives you an explorable tree showing file sizes and structure. Easy as pie.
Most of what you can extract from du comes through in tdu, although I notice that some conventions, like the human-readable flag
-h, seem to pollute the output.
tdu has a few commands while running, most of which you can skim through by pressing the ? key. Perhaps most interesting are n, which sorts by name, or S, which sorts biggest to smallest.
For what it’s worth, there’s a second, lesser utility on the tdu home page, called dugroup. That lets you clump files in groups according to type — backup, image, sound file, etc. — when they appear in du’s output.
As you can imagine, tdu’s output can change a little bit with that thrown into the mix.
tdu’s home page describes it as “a text-mode disk usage full-screen folding outline doohickey utility thingamabob.” I honestly can’t think of a better title than that. 😉