Calculating CEs for Radical Stroke order implicit collations

[Kip Cole]

In 2017 I asked some questions around generating implicit CE’s and here I am asking more questions (thanks Marcus for your patience last time).  I deferred work on collation to finish up implementing pretty much everything in CLDR except transforms (work in progress) and collations. Now I’m back to collations.

I think I’m clear now on a path forward but could use some help identifying the algorithm ICU uses to transform the “[radical ….]” lines in FractionalUCA.txt into a list of CEs.

Any pointers to:

1. The relevant module in ICU4j. Or even better:
2. A short description of the algorithm to calculate a CE for a grapheme that preserves the order of the radical list (left to right, top to bottom)

Would be much appreciated.

Many thanks, —Kip