SQL Group By weeks and months in the same time (Redshift) - sql

Question

In the code below I am selecting 42 days period and grouping it by SNAPSHOT_WEEK (where SNAPSHOT_WEEK has a number from 1 to 52(53) during the year).
SELECT
CASE
WHEN video_code = 'A' THEN 'Seller'
WHEN video_code = 'B' THEN 'Vendor'
WHEN video_code = 'C' THEN 'Others'
END AS CATEGORY
TO_CHAR(snapshot_time - DATE_PART('dow', snapshot_time)::int + 4, 'IW') AS SNAPSHOT_WEEK,
SUM(VIOLATION_COUNT)
FROM my_table
WHERE 1=1
AND snapshot_time BETWEEN '20180505'::date - '41 days'::interval AND '20180505'::date -- to calculate WoW
GROUP BY
CATEGORY, SNAPSHOT_WEEK;
Output for this query looks like this:
CATEGORY WEEK OR MONTH SUM_VIOLATION_COUNT
A 14 954
B 14 454
C 14 299
A 15 954
B 16 454
Is it possible, in the same query, beside grouping by week, group this data by month where month should start from 28th of one month to 28th of second month?
For example, in my output I need column that will show following values:
CATEGORY WEEK OR MONTH SUM_VIOLATION_COUNT
A 14 954
B 14 454
C 14 299
A 15 954
B 16 454
C 17 299
A 28 March 9354
B 28 March 2454
C 28 March 5354
A 28 April 1354
...... ..... .....
Where "28 March" - means number of violation between 28-Feb and 28 March; "28 April" - number of violation between 28 Feb and 28 April etc.
Is that possible to do using the same query?

Answer 1

You can do that with WITH Subquery, this will allow you do to run the query once on the database and group by twice based on your logic.
Your query has some disconnects between your column names but again it will look like something like this
P.S. Union requires number of columns should be same in both selects
WITH ALLDATA AS (
SELECT
CASE
WHEN video_code = 'A' THEN 'Seller'
WHEN video_code = 'B' THEN 'Vendor'
WHEN video_code = 'C' THEN 'Others'
END AS CATEGORY
TO_CHAR(snapshot_time - DATE_PART('dow', snapshot_time)::int + 4, 'IW') AS SNAPSHOT_WEEK,
SUM(VIOLATION_COUNT) SUM_VIOLATION_COUNT
FROM my_table
WHERE 1=1
AND snapshot_time BETWEEN '20180505'::date - '41 days'::interval AND '20180505'::date -- to calculate WoW
GROUP BY
CATEGORY, SNAPSHOT_WEEK)
SELECT CATEGORY, SNAPSHOT_WEEK, SUM_VIOLATION_COUNT FROM ALLDATA
UNION
SELECT CATEGORY, SNAPSHOT_WEEK, SUM_VIOLATION_COUNT FROM ALLDATA
GROUP BY <your month grouping logic>
To reiterate the logic in pseudo code
WITH ALLDATA AS (
SELECT <your base data without group by> )
SELECT columns FROM ALLDATA
GROUP BY <weekly group by logic>
UNION
SELECT columns FROM ALLDATA
GROUP BY <monthly group by logic>

Answer 2

You would need to UNION the output of two separate queries to generate those results.
The basic rule is that one input row will map to (at most) one output row.