not understand json_agg in this context - postgresql

Question

Reference: Query jsonb column containing array of JSON objects
begin;
CREATE TEMP TABLE segments (segments_id serial PRIMARY KEY, payload jsonb);
INSERT INTO segments (payload)
VALUES ('[{"kind": "person", "limit": "1"}, {"kind": "B", "filter_term": "fin"}]');
INSERT INTO segments (payload)
VALUES ('[{"kind": "person", "limit": "3"}, {"kind": "A", "filter_term": "abc"}]');
INSERT INTO segments (payload)
VALUES ('[{"kind": "person", "limit": "2"}, {"kind": "C", "filter_term": "def"}]');
commit;
CTE query:
with a as (select jsonb_array_elements(s.payload) j from segments s)
SELECT json_agg(a.j) AS filtered_payload from a
where j #> '{"kind":"person"}';
Return: [{"kind": "person", "limit": "1"}, {"kind": "person", "limit": "3"}, {"kind": "person", "limit": "2"}]
This QueryA:
SELECT a.filtered_payload,
a.ct_elem_row
, sum(ct_elem_row) OVER () AS ct_elem_total
, count(*) OVER () AS ct_rows
FROM segments s
JOIN LATERAL (
SELECT json_agg(j.elem) AS filtered_payload, count(*) AS ct_elem_row
FROM jsonb_array_elements(s.payload) j(elem)
WHERE j.elem #> '{"kind":"person"}'
) a ON ct_elem_row > 0
WHERE s.payload #> '[{"kind":"person"}]';
return :
In QueryA, the structure is like: select ... from segments s join lateral filtered_payload.... segments is 3 rows lateral join with one row (filtered_payload). filtered_payload return only row as per CTE query, as an a consolidate JSON array. So overall I am very confused with json_agg in the QueryA.
Edit at 2021-10-05 16:36 +5:30:
Even following code, a.filtered_payload return 3 jsonb array, instead of one arrgregate json array. I don't know when already aggregated jsonb array (using json_agg function) unnested to serveal jsonb arrays.
SELECT a.filtered_payload, s.*
FROM segments s
cross JOIN LATERAL (
SELECT json_agg(j.elem) AS filtered_payload
FROM jsonb_array_elements(s.payload) j(elem)
WHERE j.elem #> '{"kind":"person"}') a;

Answer 1

I believe the LATERAL JOIN is doing the trick there.
In your original query, you're using json_agg against the whole table dataset (filtered by '{"kind":"person"}')
with a as
(
select jsonb_array_elements(s.payload) j
from segments s
)
SELECT json_agg(a.j) AS filtered_payload
from a
where j #> '{"kind":"person"}';
Meanwhile in the second instance, you are playing with one row at the time using the LATERAL. That's why you end up having 3 rows with single "kind":"person" values instead of an unique row with 3 values.
Not sure about what you're trying to achieve but the following could put you in the right direction
SELECT a.filtered_payload,
a.ct_elem_row,
sum(ct_elem_row) OVER () AS ct_elem_total,
count(*) OVER () AS ct_rows
FROM segments s
JOIN LATERAL (
SELECT json_agg(j.elem) AS filtered_payload,
count(*) AS ct_elem_row
FROM segments d, lateral jsonb_array_elements(d.payload) j(elem)
WHERE j.elem #> '{"kind":"person"}'
) a ON ct_elem_row > 0
WHERE s.payload #> '[{"kind":"person"}]';
Results
filtered_payload | ct_elem_row | ct_elem_total | ct_rows
--------------------------------------------------------------------------------------------------------+-------------+---------------+---------
[{"kind": "person", "limit": "1"}, {"kind": "person", "limit": "3"}, {"kind": "person", "limit": "2"}] | 3 | 9 | 3
[{"kind": "person", "limit": "1"}, {"kind": "person", "limit": "3"}, {"kind": "person", "limit": "2"}] | 3 | 9 | 3
[{"kind": "person", "limit": "1"}, {"kind": "person", "limit": "3"}, {"kind": "person", "limit": "2"}] | 3 | 9 | 3
(3 rows)