如何選擇SQL Server中多個列之間不是數字的所有值？

Question

我有一個表Prices

ID Price_1 Price_2 Price_3 

P1 10  11  12 
P2 13  14  15 
P3 aa  16  bb 
P4 19  cc  20 

---

正如你可以從上面看到，從列Price_1，Price_2和Price_3一些值可能不是數字。

我想要的是首先找到所有這些non-numeric values然後給出一個總結（concatenate所有非數值和列一個ID）

因此，對於上面的例子，我要的是

ID Bad_Columns  Bad_Values 
P3 Price_1,Price_3 aa,bb 
P4 Price_2   cc 

---

我該怎麼寫這個查詢？

Answer 1

你可以做這樣的事情：

WITH CTE AS 
( SELECT ID, Value, ColumnName 
    FROM Prices 
      UNPIVOT 
      ( Value 
       FOR ColumnName IN ([Price_1], [Price_2], [Price_3]) 
      ) upvt 
    WHERE ISNUMERIC(Value) = 0 
) 
SELECT ID, 
     BadColumns = STUFF((SELECT ', ' + ColumnName 
          FROM CTE 
          WHERE CTE.ID = t.ID 
          FOR XML PATH(''), TYPE 
         ).value('.', 'NVARCHAR(MAX)'), 1, 2, ''), 
     BadValues = STUFF((SELECT ', ' + Value 
          FROM CTE 
          WHERE CTE.ID = t.ID 
          FOR XML PATH(''), TYPE 
         ).value('.', 'NVARCHAR(MAX)'), 1, 2, '') 
FROM (SELECT DISTINCT ID FROM CTE) t 

第一部分UNPIVOTs查詢得到的列行，那麼第二部分將串連「壞」行成一列。

Example on SQL Fiddle

Answer 2

您可以使用這將unpivot的的數據可以再使用串聯的FOR XML PATH值：

;with cte as 
(
    select id, col, data 
    from 
    (
    select t.id, 
     c.col, 
     case c.col 
     when 'Price_1' then Price_1 
     when 'Price_2' then Price_2 
     when 'Price_3' then Price_3 
     end as data 
    from yourtable t 
    cross join 
    (
     select 'Price_1' as col 
     union all select 'Price_2' 
     union all select 'Price_3' 
    ) c 
) src 
    where isnumeric(data) = 0 
) 
select distinct c1.id, 
    stuff((select ', ' + col 
     from cte c2 
     where c1.id = c2.id 
     FOR XML PATH ('')) 
      , 1, 1, '') AS Bad_columns, 
    stuff((select ', ' + data 
     from cte c2 
     where c1.id = c2.id 
     FOR XML PATH ('')) 
      , 1, 1, '') AS Bad_Values 
from cte c1 

見SQL Fiddle with Demo。

這也可以使用unpivot功能寫着：

;with cte as 
(
  select id, col, data 
  from yourtable 
  unpivot 
  (
    data 
    for col in ([Price_1], [Price_2], [Price_3]) 
 ) unpiv 
  where isnumeric(data) = 0 
) 
select distinct c1.id, 
  stuff((select ', ' + col 
         from cte c2 
         where c1.id = c2.id 
         FOR XML PATH ('')) 
          , 1, 1, '')  AS Bad_columns, 
  stuff((select ', ' + data 
         from cte c2 
         where c1.id = c2.id 
         FOR XML PATH ('')) 
          , 1, 1, '')  AS Bad_Values 
from cte c1 

見SQL Fiddle with Demo。其結果是：

| ID |  BAD_COLUMNS | BAD_VALUES | 
--------------------------------------- 
| P3 | Price_1, Price_3 |  aa, bb | 
| P4 |   Price_2 |   cc | 

Answer 3

我建議：

select p.* 
from ((select id, price_1 as price, 'price_1' as col) union all 
     (select id, price_2 as price, 'price_2' as col) union all 
     (select id, price_3 as price, 'price_3' as col) 
    ) p 
where is_numeric(price) = 0 and price is not null 

如果你真的想他們你的格式，只是做：

select id, 
     stuff((case when isnumeric(price_1) = 1 then ',Price_1' else '' end) + 
      (case when isnumeric(price_2) = 1 then ',Price_2' else '' end) + 
      (case when isnumeric(price_3) = 1 then ',Price_3' else '' end) 
      ), 1, 1, '') as Bad_Columns, 
     stuff((case when isnumeric(price_1) = 1 then ','+Price_1 else '' end) + 
      (case when isnumeric(price_2) = 1 then ','+Price_2 else '' end) + 
      (case when isnumeric(price_3) = 1 then ','+Price_3 else '' end) 
      ), 1, 1, '') as Bad_Values 
from p 
where isnumeric(price_1) = 0 or isnumeric(price_2)= 0 or isnumeric(price_3) = 0 

，我不是原因使用unpivot是因爲它假定列類型是兼容的。如果您想將其擴展到更多列，請使用Excel生成代碼。