基于集合的查询，以查找表中包含至少一个空值的所有列

Question

I am new to SQL and SQL Server 2008 R2. I have found a cursor based approach to find all columns in a particular table that contain null values, but I am hoping to find a simpler set-based solution.

USE QC_TEST

DECLARE @colName nvarchar(100)
DECLARE @nullCols nvarchar(max)

SELECT @colName = c.name, @nullCols = COALESCE(@nullCols+', ','')+@colName
FROM sys.tables AS t
    JOIN sys.columns AS c
        ON t.object_id = c.object_id
WHERE t.name = 'myTable' 
AND 
EXISTS(SELECT * FROM myTable WHERE c.name IS NULL)

SELECT @colName, @nullCols

The above code currently returns all columns in myTable . If I change the EXISTS clause to NOT EXISTS , it returns no columns. The result should be a comma-separated string of column names that contain at least 1 null value.

Thanks for your help

Answer 1

You could get a result this way

declare @sql varchar(max), @t varchar(max);
set @t = 'your_table';
select @sql = (select stuff(
    (select 
            '] is null union select ''' + c.name + 
            ''' from ' + @t + ' where [' + c.name
        from
            (select
                    c.name
                from sys.tables AS t
                inner join sys.columns AS c
                    ON t.object_id = c.object_id
                where
                    t.name = @t
            ) as c
        for xml path('')
    ), 1, 16, '' ) + '] is null' )
;
exec(@sql);

But I don't know that you'd really consider that better.

Answer 2

Your code isn't working because the EXISTS subquery is just checking the name of your column isn't null, which it can't be. The subquery doesn't expand out the "WHERE c.name IS NULL" into "WHERE MyColumnName IS NULL", it just sees that c.name = 'MyColumnName' which is not null and always returns false.

The cursor based approach probably works by building a seperate string query for each column and then excuting the string.

You could unpivot the table and look for nulls in that result set, ie

WITH myTable_unpvt AS (
    SELECT ContactID, FieldName, FieldValue
    FROM 
       (SELECT ContactID
            , cast(Forename as sql_variant) Forename
            , cast(Surname as sql_variant) Surname
            , cast(Extn as sql_variant) Extn
            , cast(Email as sql_variant) Email
            , cast(Age as sql_variant) Age
       FROM myTable) p
    UNPIVOT
       (FieldValue FOR FieldName IN 
          (Forename, Surname, Extn, Email, Age)
    ) AS myTable_unpvt
)

SELECT DISTINCT FieldName FROM myTable_unpvt WHERE FieldValue IS NULL

Answer 3

this is a little unconventional, but try this:

--set up test table
--the BEGIN TRY and ALTERs allow you to rerun entire script over and over again
BEGIN TRY
    CREATE TABLE YourTable (PK int not null primary key, RowValue1 int null, RowValue2 varchar(5) null, RowValue3 datetime)
END TRY
BEGIN CATCH END CATCH
delete YourTable
ALTER TABLE YourTable ALTER COLUMN RowValue1 int  NULL
ALTER TABLE YourTable ALTER COLUMN RowValue2 varchar(5)  NULL
ALTER TABLE YourTable ALTER COLUMN RowValue3 datetime  NULL

--setup test data
INSERT INTO YourTable VALUES  (1,1111,'AAAA',GETDATE())
INSERT INTO YourTable VALUES  (2,2222,'BBBB',GETDATE())
INSERT INTO YourTable VALUES  (3,3333,'CCCC',GETDATE())
INSERT INTO YourTable VALUES  (4,NULL,'DDDD',GETDATE())
INSERT INTO YourTable VALUES  (5,5555,'EEEE',NULL)

--determine null columns
DECLARE @BadColumns  varchar(50)

BEGIN TRY
    ALTER TABLE YourTable ALTER COLUMN RowValue1 int NOT NULL
END TRY
BEGIN CATCH
    SET @BadColumns=ISNULL(@BadColumns+', ','')+'RowValue1'
END CATCH    
----------------------------------
BEGIN TRY
    ALTER TABLE YourTable ALTER COLUMN RowValue2 varchar(5) NOT NULL
END TRY
BEGIN CATCH
    SET @BadColumns=ISNULL(@BadColumns+', ','')+'RowValue2'
END CATCH  
----------------------------------
BEGIN TRY
    ALTER TABLE YourTable ALTER COLUMN RowValue3 datetime NOT NULL
END TRY
BEGIN CATCH
    SET @BadColumns=ISNULL(@BadColumns+', ','')+'RowValue3'
END CATCH  

--report null columns
SELECT @BadColumns AS [Columns having NULLs]

OUTPUT:

Columns having NULLs
--------------------------------------------------
RowValue1, RowValue3

(1 row(s) affected)

if you need to to be dynamic because you don't know the column names, you can build a similar command into a string and then execute that.