I think "nicely and strictly" is an overstatement here. You still have to deal with southeast Asian scripts (which require a dictionary to find line break opportunities), and then there's the whole complex regex for numeric expressions (Example 7 in the Examples section), which ICU implements. I didn't bother with that in xi-unicode (it's not clear it improves matters much), but I do want to get Thai breaking nicely.

On top of that, the Unicode rules do a very poor job with things like URLs and email addresses. The Android text stack has its own layer (WordBreaker in the minikin lib) on top of that which recognizes those and implements its own rules.

But TR14 is a good start, for sure.