-- SQL_ASCII: bytes >= 0x80 are opaque letters, ASCII still folds SELECT current_database() AS orig \gset CREATE DATABASE sift_sql_ascii ENCODING 'SQL_ASCII' LOCALE 'C' TEMPLATE template0; CREATE DATABASE sift_latin1 ENCODING 'LATIN1' LOCALE 'C' TEMPLATE template0; \c sift_sql_ascii CREATE EXTENSION pg_sift; SELECT token::bytea, pos, start_offset, end_offset FROM sift.tokenize(E'Caf\xe9 NA\xcfVE \xff\xfe-x \xc3\xa9T\xc3\xa9'); SELECT octet_length(token) FROM sift.tokenize(repeat(E'\xe9', 300)); \c sift_latin1 CREATE EXTENSION pg_sift; SELECT * FROM sift.tokenize('x'); \c :orig DROP DATABASE sift_sql_ascii; DROP DATABASE sift_latin1;