How to write a query to select relationships without PostgreSQL duplicates?

Question

PostgreSQL

Good day!
Please help with a request. Actually, here is the database structure and query .
The essence of the problem is as follows: 1. JOIN
' 2 tables (1 and 2), by several fields (no other way)
a logically correct result string (i.e. the grouping is needed exactly like this)
3. We are interested in the result in the last column ( array_agg ) The snag
is in the 5th line. If we remove DISTINCT , then an excess number of values ​​is added to the results. If we leave DISTINCT - the same values ​​"collapse", and the result is not correct.
Can you please tell me how to get unique values ​​in the array_agg column (in the form of an array), without multiple duplicates and without "collapsing" the same values? Those. the result should be something like: "b1, b1, b2, b2, b3, b3" or "b1, b2, b3, b1, b2, b3" (or in any other order).

Answer 1

SELECT 
  t2.b1,
  t2.b2,
  t2.b3,
  ARRAY_AGG(t2.value)
FROM table2 t2
WHERE EXISTS (
  SELECT 1
    FROM table1 t1
    WHERE t1.a1 = t2.b1
      AND t1.a2 = t2.b2
      AND t1.a3 = t2.b3
)
GROUP BY t2.b1, t2.b2, t2.b3

PS:
in fact, they are not unique, and as I understand it, the question should have sounded something like this:
how to select values ​​​​from table2 by three keys to relationships with table1 and aggregate the value column into an "array"?