Specify Data Types in Redshift 'CREATE TABLE AS' Query

Question

Is there a way to specify data types and keep constraints when doing a Redshift CREATE TABLE AS query?

In the below example, I'd like to be able to create a copy of table1 with the column column2 as NVARCHAR(500) type rather than VARCHAR(255) and keep the column1 NOT NULL constraint.

CREATE TABLE test.table1 (column1 INT NOT NULL, column2 VARCHAR(255));

INSERT INTO test.table1
SELECT 1,'foo';

CREATE TABLE test.table2 AS
SELECT column1,CAST(column2 AS NVARCHAR(500))
FROM test.table1;
/* query executes successfully but still creates column2 as VARCHAR(255) and column1 is NULLABLE.  */

CREATE TABLE test.table2 (
    column1 INT NOT NULL
    ,column2 NVARCHAR(500)
    ) AS
SELECT column1
    ,column2
FROM test.table1;
/* ERROR: 42601: syntax error at or near "as" */

Edit: I'm aware that I could do this in two steps by explicitly creating test.table2 with the correct data types and constraints, then doing an INSERT statement using the data in test.table1 . I am curious though to see if this can be done in one CREATE TABLE AS query. Any workarounds or tips would be greatly appreciated! :)

Answer 1

From an SQL perspective, the answer is Yes and No. Yes, you can specify the datatype: use CAST to convert the column to the type you want. No, you can't specify constraints, because you can't: there's no syntax to do that.

If you're concerned about the constraints of a table, make it part of the schema. If the data are determining the schema dynamically, something is amiss.