在 PostgreSQL 中返回列重复项

Question

我正在学习 PostgreSQL，并且正在努力解决这个问题。

我已经设置了一个 olympics 表，我从中查询并返回结果。我正在查询以查找国家和他们赢得的金牌数量，如下所示：

SELECT country, golds 
FROM (SELECT distinct country, sum(gold_medals) as golds
      FROM olympics where year >= 2000 group by country
     ) foo
WHERE (golds < 10) 
ORDER BY golds desc limit 10;

这准确地返回：

   country   | golds 
-------------+-------
 Turkey      |     9
 Bulgaria    |     8
 Azerbaijan  |     6
 Estonia     |     6
 Georgia     |     6
 North Korea |     6
 Thailand    |     6
 Nigeria     |     6
 Uzbekistan  |     5
 Lithuania   |     5

我需要返回在这段时间内赢得相同数量金牌的国家（即立陶宛和乌兹别克斯坦获得 5 枚金牌，以及所有国家获得 6 枚金牌）。

我该怎么做呢？

Answer 1

@ErwinBrandstetter 的解决方案有效，但为了完整起见，我还将添加 array_agg 版本，它将国家集中在一起作为单个单元格中的字符串数组：

WITH golds as (
    select
        sum(gold_medals) golds,
        country
    from olympics
    where year >= 2000
    group by country
    )

select 
    golds, 
    array_length(array_agg(country),1) n_countries, 
    array_agg(country) countries
from golds
group by golds
having array_length(array_agg(country),1) > 1
order by golds asc

-- golds , n_countries , countries
--   5   ,     2       , '{lithuania,uzbekistan}'
--   6   ,     6       , '{thailand,"north korea",azerbaijan,nigeria,estonia,georgia}'

Answer 2

运行另一个聚合。这次GROUP BY金牌数。然后JOIN：

WITH cte AS (
   SELECT country, sum(gold_medals) AS golds
   FROM   olympics
   WHERE  year >= 2000
   GROUP  BY country
   HAVING sum(gold_medals) < 10   -- ? No more allowed?
   )
SELECT c.*
FROM   cte c
JOIN  (
    SELECT golds
    FROM   cte
    HAVING count(*) > 1
    ) ties USING (golds)
ORDER  BY golds DESC, country
LIMIT  10;

• 从原始查询中删除无用的DISTINCT。 GROUP BY 已经完成了这项工作。

• 我正在使用CTE 来简化我的工作。

与EXISTS半连接的替代方案：

WITH cte AS (
   SELECT country, sum(gold_medals) AS golds
   FROM   olympics
   WHERE  year >= 2000
   GROUP  BY country
   HAVING sum(gold_medals) < 10   -- ? No more allowed?
   )
SELECT c.*
FROM   cte c
AND    EXISTS (
   SELECT 1
   FROM   cte
   WHERE  country <> c.country -- exclude "self"
   AND    golds  = c.golds
   )
ORDER  BY golds DESC, country
LIMIT  10;

或者，要将最多 10 枚金牌的国家/地区排列成每金牌数量的单行，但前提是有 2 个或更多国家/地区：

SELECT golds, array_agg(country) AS country_list
FROM  (
   SELECT country, sum(gold_medals) AS golds
   FROM   olympics
   WHERE  year >= 2000
   GROUP  BY country
   HAVING sum(gold_medals) < 10   -- ? No more allowed?
   ) sub
GROUP   BY golds
HAVING  count(*) > 1
ORDER   BY 1  DESC;
-- LIMIT   10;    -- not needed, there cannot be more than 10 in this case.

最后一个基本上是@Nisan.H´s answer的简化版。

Answer 3

尝试按国家和金牌分组

SELECT country, golds 
FROM (SELECT distinct country, sum(gold_medals) as golds FROM 
olympics where year >= 2000 group by country, golds) foo
WHERE (golds < 10) ORDER BY golds desc limit 10;