Import data must be present as a flat ASCII file. Language-dependent special characters can only be processed if the installed database character set supports them.
The decision for which database character set will be used is made when the database
is created. You cannot change the character set later on.
An import file can include two sections: the first one contains the ident records along with the associated detail records, which may have the types IDENT, IDENT_NLS, and IDENT_ATTR. The detail records will always be processed in the context of the preceding ident record. The second section contains the assembly records, that is, records of type IDENT_STRUC. Although the import program is capable of processing ident data and assembly data in any sequence, it is highly recommended that you at least partition the import files in the way described above, to make the tracking of incorrect data in your files much easier. For large volumes of import data, it may even be a good idea to split the import: first you would import files that only contain ident data, and if all of it has been imported successfully, you would import assembly data from separate files.