我试过“波希米亚人”的答案，但对我不起作用。多亏了他的简单解释，通过这个链接，我能够想出下面的代码，为我工作。但我不知道是否有一种方法可以消除代码中与左连接相关的冗余

sql = '(
  SELECT *
  FROM institutes
  LEFT JOIN city ON institutes.city_id = city.city_id 
  LEFT JOIN district ON city.district_id = district.district_id 
  WHERE city.city_id = $current_city_id
  )
  UNION
  (
  SELECT *
  FROM institutes
  LEFT JOIN city ON institutes.city_id = city.city_id 
  LEFT JOIN district ON city.district_id = district.district_id 
  WHERE district.district_id = $current_district_id
)
ORDER BY RAND() DESC LIMIT 10'

使用 UNION ALL 带 Package 器 SELECT 学生：

SELECT * FROM (
    SELECT *
    FROM institutes
    LEFT JOIN city ON institutes.city_id = city.city_id 
    LEFT JOIN district ON city.district_id = district.district_id 
    WHERE city.city_id = $current_city_id
    ORDER BY RAND() DESC LIMIT 5
) x1
UNION ALL
SELECT * FROM (
    SELECT *
    FROM institutes
    LEFT JOIN city ON institutes.city_id = city.city_id 
    LEFT JOIN district ON city.district_id = district.district_id 
    WHERE district.district_id = $current_district_id
    ORDER BY RAND() DESC LIMIT 5 
) x2

Package 纸 SELECT 因为你不能合并 ORDER BY ... LIMIT 以及 UNION 在同一水平，但你可以把 ORDER BY ... LIMIT 在子查询中。 UNION ALL 是需要的而不仅仅是 UNION 因为 UNION 删除重复项，您的问题需要保留重复项。 UNION ALL 也比 UNION （因为它不需要重复数据消除，这通常需要排序），但在这种情况下，行太少，性能上的差异不会被注意到，但最好记住这一点，以备将来的应用程序使用。