T-SQL to Oracle: Complicated Table Column

Question

I'd like to convert the following table to PL-SQL.

CREATE TABLE [SchemaName].[TableName] (
    [value]         VARBINARY (2000) NULL,
    [valueAsString] AS               (CONVERT([nvarchar](1000),[value],0))
);

Has anyone an idea on how to recreate the valueAsString pseudo column in PL?

Answer 1

One possible solution would be

HUSQVIK@hq_pdb_tcp> CREATE OR REPLACE FUNCTION convert_raw_to_nvarchar(bin RAW) RETURN NVARCHAR2 DETERMINISTIC
  2  IS
  3  BEGIN
  4     RETURN utl_raw.cast_to_nvarchar2(bin);
  5  END;
  6  /

Function created.

HUSQVIK@hq_pdb_tcp> CREATE TABLE SchemaName.TableName (
  2      value         RAW(2000) NULL,
  3      valueAsString AS (convert_raw_to_nvarchar(value))
  4  );

Table created.

HUSQVIK@hq_pdb_tcp> insert into tablename (value) values (utl_raw.cast_to_raw(n'123344čřžžýřžý'));

1 row created.

HUSQVIK@hq_pdb_tcp> select * from tablename;

VALUE                                                        VALUEASSTRING
------------------------------------------------------------ --------------------
00310032003300330034003400630072007A007A00EC0072007A00EC     123344cržžýržý

If the limit of the binary column is 2000 bytes BLOB is not needed.

utl_raw.cast_to_nvarchar2 is not deterministic unfortunately so it needs to be wrapped into deterministic function to be usable in table definition.

But I don't like such solution. A view can solve this without schema level function. Also data types suggest you store information using inappropriate data type when you want to always convert binary data into string.