aggregate-functions - 是否存在将字段值连接到特定字符的 join() 聚合函数之类的东西？

Question

假设我有一个名为和cities字段的表。数据如下所示：statecity

State     City
Ohio      Eaton
Ohio      Columbus
Ohio      Dayton
Ohio      Greenville
New York  New York
New York  Albany
New York  Syracuse

我希望能够做到这一点：

select state, join(city,',') from cities
group by state

...返回这个：

State     City
Ohio      Eaton,Columbus,Dayton,Greenville
New York  New York,Albany,Syracuse

我需要为此创建一些疯狂的 UDF 吗？如果没有函数，有没有办法应用一些 SQL 魔法来实现这一点？

Answer 1

递归 CTE 应该能够做到这一点。我模拟了要为每个状态重置的序列号。您可能需要根据您的实际表格进行调整，但它至少应该让您思考正确的方向。（类似于 Marc B 在 Teradata 论坛上链接的内容，但希望这更容易阅读。）

WITH RECURSIVE city_list (state, cities, seq) AS
(
  SELECT s.state
       , s.city
       , ROW_NUMBER() OVER (PARTITION BY State ORDER BY State, City) AS seq
  FROM   states s
  QUALIFY  seq = 1
  UNION ALL
  SELECT c.cities ||’, ’|| s.city
       , ROW_NUMBER() OVER (ORDER BY State, City) AS seq
  FROM   city_list c, states s
  WHERE  s.seq = c.seq + 1
    AND  s.state = c.state
)
SELECT r.state, r.cities
FROM   city_list r
WHERE  r.seq = (SELECT MAX(h.seq) FROM city_list h 
                WHERE h.state = r.state 
                GROUP BY h.state)
;