Based on all compiled data, organized into sets by source directory, with names organized by unique address. Struggled to find some way to get the process to recognize names in Japanese (the eternal problem); the \p{Han} character class doesn't seem to work at all with Saxon. Why?