How to do GROUP BY [empty]-CodePudding

Is there a way for explicitly specifying an empty aggregation, such as the following:

SELECT NULL AS Product, SUM(Revenue) FROM Sales GROUP BY ()

Rather than the implicit version which doesn't have a GROUP BY clause:

SELECT NULL AS Product, SUM(Revenue) FROM Sales

Sample input/output:

WITH Sales (Product, Revenue) AS (VALUES ('a',10),('b',20))
select NULL Product, SUM(Revenue) from Sales;
┌─────────┬──────────────┐
│ product ┆ sum(revenue) │
╞═════════╪══════════════╡
│         ┆           30 │
└─────────┴──────────────┘

The query is correct as-is, I'm just wondering if there is a way to make the GROUP BY explicit. This is for a very basic downstream parser that checks if it is an aggregation query based on if the words GROUP BY are in the normalized query.

CodePudding user response：

You can use expressions to group by. so grouping by an expression that is constant is equivalent removing it from the group by statement

i.e. the following are equivalent:

WITH Sales (Product, Revenue) AS (VALUES ('a',10),('b',20))
select NULL Product, SUM(Revenue) from Sales;

and

WITH Sales (Product, Revenue) AS (VALUES ('a',10),('b',20))
select NULL Product, SUM(Revenue) from Sales group by 1=1;