The tables np_normalize() uses, for packages that need the same rules in
another engine (for example a DuckDB SQL macro). A port should be checked
against np_name_core() on test strings; the tables are covered by
np_normalize_signature().
Value
A list: abbrev (named character, token to expansion),
core_stop (tokens dropped from name_core), legal_suffixes
(trailing tokens stripped from name_key), special_letters (named
character, a non-decomposable Latin letter to its ASCII spelling, folded
before de-accenting), split_words (named character, a word broken across
two tokens to the whole word, applied before abbrev because a port that
applied it after would turn FOUNDATIO N into FOUNDATION N),
placeholder_re (a Perl regex on the raw upper-cased name) and
placeholder_tok.