È piuttosto semplice generate_series può essere utilizzato con INTERVAL se possiamo usare un LATERAL JOIN per generare i dati per ogni riga json.
WITH j AS(
SELECT '{
"id": "1",
"amount": 3900,
"interval": "1 WEEK",
"created": 1424011935
}'::json AS data
UNION ALL
SELECT '{
"id": "2",
"amount": 100,
"interval": "23.5 DAY",
"created": 552614400
}'::json
),
dates AS(
SELECT
'2008-03-01 00:00'::timestamp AS start
,'2015-03-08 07:00'::timestamp AS stop
)
SELECT j.data->>'id'
,sum((j.data->>'amount')::int)
,count(*) as intervals
FROM j
CROSS JOIN dates
CROSS JOIN LATERAL
(SELECT i
FROM generate_series(to_timestamp((j.data->>'created')::int), dates.stop + (j.data->>'interval')::interval,(j.data->>'interval')::interval) as i
WHERE i BETWEEN dates.start AND dates.stop
) AS q
GROUP BY j.data->>'id'
Quanto sopra ci dà il risultato:
id | sum | intervals
----+-------+-----------
1 | 11700 | 3
2 | 11000 | 110
