/usr/bin/gennorm2

Usage: /usr/bin/gennorm2 [-options] infiles+ -o outputfilename

Reads the infiles with normalization data and
creates a binary file, or a C source file (--csource), with the data,
or writes a data file with the combined data (--combined).
See https://unicode-org.github.io/icu/userguide/transforms/normalization#data-file-syntax

Alternate usage: /usr/bin/gennorm2 [-options] a.txt b.txt minus p.txt q.txt -o outputfilename

Computes the difference of (a, b) minus (p, q) and writes the diff data
in input-file syntax to the outputfilename.
It is then possible to build (p, q, diff) to get the same data as (a, b).
(Useful for computing minimal incremental mapping data files.)

Options:
	-h or -? or --help  this usage text
	-v or --verbose     verbose output
	-c or --copyright   include a copyright notice
	-u or --unicode     Unicode version, followed by the version like 5.2.0
	-s or --sourcedir   source directory, followed by the path
	-o or --output      output filename
	      --csource     writes a C source file with initializers
	      --combined    writes a .txt file (input-file syntax) with the
	                    combined data from all of the input files
	      --fast        optimize the data for fast normalization,
	                    which might increase its size  (Writes fully decomposed
	                    regular mappings instead of delta mappings.
	                    You should measure the runtime speed to make sure that
	                    this is a good trade-off.)