Dynamic Create Table Statement in Snowflake - snowflake-cloud-data-platform

Question

I am trying to do the following
Creating one config table where I want to put all details which below query provides as output
USE ROLE ROLENAME;
SET SemanticLayerSchemaDetails = '''DB_D''';
SET TableSchemaDetails = 'SCHEMA_D';
SET TableNamesArray = '[''TABLE_SALES'']';
with CTE_SchemaOneObjectDetails
AS
(
SELECT
$SemanticLayerSchemaDetails AS SCHEMA_D, (''''|| TABLE_CATALOG || '''') AS DB_D,
(''''|| TABLE_NAME || '''') AS RLOBJ,listagg('"'||COLUMN_NAME||'"', ' VARCHAR , ') WITHIN GROUP (ORDER BY ORDINAL_POSITION ASC) AS COLUMN_DETAILS
from INFORMATION_SCHEMA.columns
where TABLE_SCHEMA = $TableSchemaDetails
and table_name IN (select value from table(flatten ( input => parse_json($TableNamesArray))))
GROUP BY Table_Name,TABLE_CATALOG,TABLE_SCHEMA
)
SELECT * FROM CTE_SchemaOneObjectDetails
Here , the issue is if table : TABLE_SALES has total 10 columns, for last column, I do not see VARCHAR in "COLUMN_DETAILS" field.
I want to use this "COLUMN_DETAILS" field calling from one stored procedure to create tables.
Anything we can do to tweak the code so that it adds VARCHAR in last column also?
Thank You,

Answer 1

Since you're using ' VARCHAR , ' as the listagg separator, it won't appear at the end. That's okay since you probably don't want the string to end with a comma anyway. You can simply concatenate ' VARCHAR' after the end of your listagg:
with CTE_SchemaOneObjectDetails
AS
(
SELECT
$SemanticLayerSchemaDetails AS SCHEMA_D, (''''|| TABLE_CATALOG || '''') AS DB_D,
(''''|| TABLE_NAME || '''') AS RLOBJ,listagg('"'||COLUMN_NAME||'"', ' VARCHAR , ') WITHIN GROUP (ORDER BY ORDINAL_POSITION ASC) || ' VARCHAR' AS COLUMN_DETAILS
from INFORMATION_SCHEMA.columns
where TABLE_SCHEMA = $TableSchemaDetails
and table_name IN (select value from table(flatten ( input => parse_json($TableNamesArray))))
GROUP BY Table_Name,TABLE_CATALOG,TABLE_SCHEMA
)
SELECT * FROM CTE_SchemaOneObjectDetails
I have a question about this though... Are all the tables you'll use this approach with defined with all varchar data types? Would you prefer to have the defined data type instead, or will this be used to create string-only versions of the table or something else that wants all columns converted to string types?
If you do want the defined data types instead of all varchar, you can use this:
with CTE_SchemaOneObjectDetails
AS
(
SELECT
$SemanticLayerSchemaDetails AS SCHEMA_D, (''''|| TABLE_CATALOG || '''') AS DB_D,
(''''|| TABLE_NAME || '''') AS RLOBJ,listagg('"'||COLUMN_NAME||'"' || ' ' || DATA_TYPE , ',') WITHIN GROUP (ORDER BY ORDINAL_POSITION ASC) AS COLUMN_DETAILS
from INFORMATION_SCHEMA.columns
where TABLE_SCHEMA = $TableSchemaDetails
and table_name IN (select value from table(flatten ( input => parse_json($TableNamesArray))))
GROUP BY Table_Name,TABLE_CATALOG,TABLE_SCHEMA
)
SELECT * FROM CTE_SchemaOneObjectDetails