PostgreSQL 将出现次数计入 json 键/值

Question

给定一个文本元素数组，我想计算出现的次数并将它们作为带有文本键和整数值的子对象返回。（PostgreSQL 9.4）

情况 1 - 没有参考集

SELECT ARRAY['3G','2G','2G','3G','3G','3G','3G','4G']

应该转化为

SELECT '{"2G": 2, "3G": 5, "4G": 1}'::jsonb

---

案例 2 - 使用参考集

给定一组可能的条目ARRAY['2G','3G','4G']，返回的 json 还应包含计数为零的元素。

SELECT ARRAY['3G','2G','2G','3G','3G','3G','3G']

应该转变为

SELECT '{"2G": 2, "3G": 5, "4G": 0}'::jsonb

---

我通过使用非常接近案例 1 和 2

-- case 1
SELECT json_object( array_agg(r.a)::text[],array_agg(r.num)::text[] ) 
FROM (
  SELECT a, count(a) as num 
  FROM unnest( ARRAY['3G','2G','2G','3G','3G','3G','3G','4G'] ) a
  GROUP BY a ORDER BY a
) r;

--case 2
SELECT json_object( array_agg(r.ref)::text[],array_agg(r.num)::text[] ) 
FROM (
  SELECT ref, count(a) as num 
  FROM unnest( ARRAY['2G','3G','4G'] ) ref
  LEFT JOIN unnest( ARRAY['3G','2G','2G','3G','3G','3G','3G'] ) a ON (ref = a)
  GROUP BY ref ORDER BY ref
) r;

但是，结果返回一个文本值，例如：

SELECT '{"2G" : "2", "3G" : "5", "4G" : "0"}'::json   -- case 2

---

感谢您的帮助和反馈！

Answer 1

对于您的情况，您可以使用聚合json_object_agg()函数，查询会更简单：

SELECT json_object_agg(r.ref,r.num) result
FROM (
  SELECT
    ref,
    count(a) AS num
  FROM unnest(ARRAY ['2G', '3G', '4G']) ref
    LEFT JOIN unnest(ARRAY ['3G', '2G', '2G', '3G', '3G', '3G', '3G']) a ON (ref = a)
  GROUP BY ref
  ORDER BY ref
) r;

结果是：

              result              
----------------------------------
 { "2G" : 2, "3G" : 5, "4G" : 0 }
(1 row)