Listen to most everything above. Yeah, I mean, I still occasionally refer to a little pocket-sized Alfred Dictionary of Music (for what, I can't remember), but among textbooks, the only ones that really stick in my mind as foundational are the good old Walter Piston *Harmony*, the George Perle on serial music, and it sounds strange, but Arnold Schönberg's Harmonielehre is really good (it's not at all a book about serialism, but covers all the same ground as any text on functional harmony, in addition to being forthcoming with many examples in notation and having some very clever expository prose interludes).
But that just varies, and the basic material should be in any decent text. Or online. I'd just stick with one basic textbook type book, but however you can, it's really the same material.
The only reason I'm commenting is that I got stuck doing temperature control in a tiny room with ghastly flourescent lights for about four hours today.
But, I happened to have, among a few other things, a copy of Beethoven's Rage Over a Lost Penny.
I'm still a bit astounded at how adeptly Beethoven performed various modulations: with even basic tools, you can identify how the composer manages, and be in a position to marvel at how seamlessly it seems to unfold.
That particular piece is not especially surprising, if you know late-early-->early-middle Beethoven (which I think is likely about when this was composed, despite the opus number: compare to sonatas just around and before the Op. 31 triad, and then compare to much later solo piano works), but you should choose your own things to look at, according to taste.
The other score I had with me was Liszt's "Grand galop": that's a bit more straightforward.
But still worth looking at with some basic ideas of harmony in mind. (I was humbled last evening playing through a few pages....it's the mileage, not the years, on the Liszt!).
So, in short, once you know your cadences and all that, just go ahead and practice by looking at whatever music you're interested in: use a pencil, mark things up, and see the big ideas.
Then look at the smaller ideas: the voicings of augmented sixths, suspensions and pedal tones, how diminished harmony contributes to the authentic cadence.
Do it at a desk, ideally: it is an abstraction over the music, the theory, so it deserves some consideration in its own right.