在标准SQL中，如何选择行，以便对于一列中的每个唯一值，另一列中的所有值都是指定值？

Question

In standard SQL, how do I select rows such that for each unique value in one column, all of the values in another column are a specified value?

For example, in the table below, I want to select all values of column c_1 where the value of c_s is 'd' for each distinct value of c_1. So for c_1 = '2', all values of c_s are 'd'. Same for c_1 = '3'. This is not, however, the case for c_1 = '1', '4', or '5'.

Here is my table:

 id | c_1 | c_2 | c_s
----+------+---------+--------
 1  | 1    | 1       | d
 2  | 1    | 2       |
 3  | 2    | 1       | d
 4  | 2    | 2       | d
 5  | 2    | 3       | d
 6  | 3    | 1       | d
 7  | 4    | 1       | r
 8  | 5    | 1       | d
 9  | 5    | 2       | r

Here is what I would like to return from my sql query:

 c_1 | c_2 | c_s
----+---+--------
 2    | 1       | d
 3    | 1       | d

(I'm not sure how to phrase this question - key words, etc. - so please excuse me if there is better phrasing that would lead to a solution that is already posted here.)

Answer 1

You can do this with GROUP BY and HAVING :

SELECT c_1
     , MIN(c_2) AS c_2
     , MAX(c_s) AS c_s
FROM Table1
GROUP BY c_1
HAVING MIN(CASE WHEN c_s = 'd' THEN 1 ELSE 0 END) 
      + MAX(CASE WHEN c_s = 'd' THEN 1 ELSE 0 END) = 2

Demo: SQL Fiddle

In your example output you chose 1 for c_2 even though other values exist, so I used MIN() , but if you wanted a different value to be chosen that logic could be adjusted.

Answer 2

My answer is very similar to goat CO's. I check the count for each value and compare it to the number of D's.

SQL Fiddle Demo

select 
  c_1      as 'c_1',
  min(c_2) as 'c_2',
  min(c_s) as 'c_s'
from yourTable
group by c_1
having count(1) = sum(case when c_s = 'd' then 1 else 0 end)