How to use DISTINCT ON in ARRAY_AGG()? - postgresql

Question

I have the following query:
SELECT array_agg(DISTINCT p.id) AS price_ids,
array_agg(p.name) AS price_names
FROM items
LEFT JOIN prices p on p.item_id = id
LEFT JOIN third_table t3 on third_table.item_id = id
WHERE id = 1;
When I LEFT JOIN the third_table all my prices are duplicated.
I'm using DISTINCT inside ARRAY_AGG() to get the ids without dups, but I want the names without dups aswell.
If I use array_agg(DISTINCT p.name) AS price_names, it will return distinct values based on the name, not the id.
I want to do something similar to array_agg(DISTINCT ON (p.id) p.name) AS price_names, but it is invalid.
How can I use DISTINCT ON inside ARRAY_AGG()?

Answer 1

Aggregate first, then join:
SELECT p.price_ids,
p.price_names,
t3.*
FROM items
LEFT JOIN (
SELECT pr.item_id,
array_agg(pr.id) AS price_ids,
array_agg(pr.name) AS price_names
FROM prices pr
GROUP BY pr.item_id
) p on p.item_id = items.id
LEFT JOIN third_table t3 on third_table.item_id = id
WHERE items.id = 1;
Using a lateral join might be faster if you only pick a single item:
SELECT p.price_ids,
p.price_names,
t3.*
FROM items
LEFT JOIN LATERAL (
SELECT array_agg(pr.id) AS price_ids,
array_agg(pr.name) AS price_names
FROM prices pr
WHERE pr.item_id = items.id
) p on true
LEFT JOIN third_table t3 on third_table.item_id = id
WHERE items.id = 1;