在sqlite中可以查询吗？

Question

我有一张这样的桌子：

ID A B C
0  x   x
1  x
2  x x
3    x

我想通过查询（如果可能的话，使用SQLite语法）获取该信息

A 3
B 2
C 1

我一直在使用CASE WHEN子句

WITH solutions AS (SELECT
    CASE
        WHEN `B` = 'x' THEN 'A'
        WHEN `B` = 'x' THEN 'B'
        WHEN `C` = 'x' THEN 'C'
    END AS 'Solution'
FROM use_cases)
SELECT solution, COUNT(*) AS 'Count'
FROM solutions
GROUP BY Solution ORDER BY COUNT(*) DESC

但是，如果一行中存在多于一个带有'x'的列，则它将不起作用，因为WHEN评估将在第一种情况下停止。

Answer 1

使用union all：

select col, count(*)
from (select 'A' as col from t where A = 'x' union all
      select 'B' from t where B = 'x' union all
      select 'C' from t where C = 'x' 
     ) x
group by col
order by count(*) desc;

Answer 2

您需要UNION ALL：

select 'A' col1, count(A = 'x') col2 from tablename 
union all 
select 'B', count(B = 'x') from tablename 
union all
select 'C', count(C = 'x') from tablename 

请参见demo。
或使用CTE来避免对该表进行多次扫描：

with cte as (
  select count(A = 'x') cola, count(B = 'x') colb, count(C = 'x') colc
  from tablename
)  
select 'A' col1, cola col2 from cte 
union all 
select 'B', colb from cte
union all
select 'C', colc from cte

请参见demo。
结果：

| col1 | col2 |
| ---- | ---- |
| A    | 3    |
| B    | 2    |
| C    | 1    |

Answer 3

如果您特别想检查X，则可以利用以下事实（至少在SQLite中）TRUE等于1，而FALSE等于0，这样类似以下的简单工作就可以了：

SELECT 'A', sum(A = 'x') from MyTable
UNION ALL
SELECT 'B', sum(B = 'x') from MyTable
UNION ALL
SELECT 'C', sum(C = 'x') from MyTable;

另一个简单的解决方案是依靠count（）仅计算非空值这一事实。因此，如果列在不包含X的地方包含NULL，则此简单的SQL将在SQLite中起作用：

SELECT 'A', count(A) from MyTable
UNION ALL
SELECT 'B', count(B) from MyTable
UNION ALL
SELECT 'C', count(C) from MyTable;

上述内容的一个细微变化...如果不包含X的列中包含除NULL之外的其他内容，则只需将其转换为null。例如，如果不包含X的列恰好包含空格，则可以使用以下内容：

SELECT 'A', count(nullif (A, ' ')) from MyTable
UNION ALL
SELECT 'B', count(nullif (B, ' ')) from MyTable
UNION ALL
SELECT 'C', count(nullif (C, ' ')) from MyTable;

注意：我提出的方法可能需要对表进行多次扫描，但是如果索引正确，则可以减轻这种损失。基于CTE的解决方案可能会更有效，但是您必须测试两种方法才能确定。基于CTE的解决方案可能涉及根据潜在的多个子查询创建临时表（或临时结果集）。在某些情况下，所有这些都会使CTE的效率降低……但是YMMV。

Answer 4

尝试

  Select val, cnt from ( SELECT 'A' as val,A,count(*) as cnt FROM TABLE 
     GROUP BY A
    UNION 
  SELECT 'B' as val,B,count(*) as cnt FROM TABLE 
    Group by B
   UNION 
   SELECT 'C' as val,C,count(*) as cnt FROM TABLE 
   Group by C))

检查下面的输出

或者如果要排除空白空间，则

SELECT VAL,CNT FROM(SELECT 'A' AS VAL,A as col,COUNT(*) AS CNT FROM SAMPLEE GROUP BY A
                         UNION
                         SELECT 'B' AS VAL,B as col,COUNT(*) AS CNT FROM SAMPLEE GROUP BY B
                         UNION
                         SELECT 'C' AS VAL,C as col,COUNT(*) AS CNT FROM SAMPLEE GROUP BY C
                         ) where col !=' ';