如何动态计算数据库架构中所有表和列中具有空值的所有行

Question

I need to create an exception report that looks over all the tables in a specific schema, which then counts across columns and rows and lists how many entries per column there were that contained a NULL Value.

How is this done without the use of cursors?

Example:

People_Table has 3 Columns (NAME, SURNAME, CONTACT_NO)
Entity_Table has 5 Columns (ID, NAME, ADDRESS, TEL_NO, FAX_NO)

I need to produce an output that lists each table and column name along with how many records contained a NULL value within each of the Columns.

People_Table : NAME (4), SURNAME (9), CONTACT_NO (120)
Entity_Table : ID (0), NAME (4), ADDRESS (90), TEL_NO (120), FAX_NO (100)

Please note that the output can be in a regular dataset format and doesn't need to look like a concatenation of entries! I only listed it in that manner to describe the data output.

This is for a dynamic database that changes over time as a column that at this stage that doesn't have any NULL entries may have a NULL entry in the future, so I need to track this. So far I've done other stuff towards this solution and this is the last of what I need now. Any advice??

Answer 1

Try this (uses cursors though):

DECLARE @tbl sysname
DECLARE @col sysname
DECLARE @sql nvarchar(max)
DECLARE @cnt INT

CREATE TABLE #result
(
    tbl sysname,
    col sysname,
    nulls int
)

DECLARE crs CURSOR FOR 
select t.name, c.name
from sys.columns c
join sys.tables t on c.object_id = t.object_id

OPEN crs

FETCH NEXT FROM crs INTO @tbl, @col

WHILE @@FETCH_STATUS=0
BEGIN

    SET @sql = 'select @cntOUT=count(*)  from '+@tbl+' where '+@col+' is null'

    SET @cnt = 0
    exec sp_executesql @sql, N'@cntOUT INT OUTPUT', @cntOUT=@cnt OUTPUT

    INSERT INTO #result (tbl, col, nulls)
    VALUES (@tbl, @col, @cnt)


    FETCH NEXT FROM crs INTO @tbl, @col
END

CLOSE crs
DEALLOCATE crs

SELECT * FROM #result

DROP TABLE #result

Answer 2

DECLARE @dbname VARCHAR(100) = 'dbname'
DECLARE @schemaName VARCHAR(100) = 'schemaname'
DECLARE @result TABLE ([NullValues] int,col VARCHAR(4000))

SELECT  @dbname dbname
        ,t.name tbl
        ,c.name col
INTO    #Temp1
FROM    sys.columns c
JOIN    sys.tables t ON 
        t.object_id = c.object_id
JOIN    sys.schemas s ON 
        s.schema_id = t.schema_id
WHERE   c.is_nullable = 1
AND     s.name in (@schemaName)

DECLARE @sql NVARCHAR(MAX) =
STUFF(
(
    SELECT  'UNION ALL SELECT Count(*) as [NullValues],''' + @dbname + '.' + @schemaName + '.' + tbl + '.' + col + ''' FROM ' + @dbname + '.' + @schemaName + '.' + tbl + ' WHERE ' + col + ' IS NULL '
    FROM    #Temp1
    FOR     XML PATH('')
), 1, 10, '   ')

INSERT @result
EXEC(@sql)

SELECT  [NullValues], col
INTO #Nulls
FROM    @result
WHERE   col IS NOT NULL AND [NullValues] > 0

SELECT
       [TABLE_CATALOG] + '.' + [TABLE_SCHEMA] + '.' + [TABLE_NAME] + '.' + [COLUMN_NAME] as SchemaTableColumn
      ,[TABLE_CATALOG] as 'Database'
      ,[TABLE_SCHEMA] as 'Schema'
      ,[TABLE_NAME] as 'TableName'
      ,[COLUMN_NAME]
      ,[ORDINAL_POSITION]
      ,[DATA_TYPE]
INTO #Temp2
  FROM [DW_LandingCR].[INFORMATION_SCHEMA].[COLUMNS]
  WHERE
    [TABLE_SCHEMA] = @schemaName
  ORDER BY TABLE_SCHEMA,TABLE_NAME,ORDINAL_POSITION

SELECT sc.name as [Schema],ta.name as [TableName],SUM(pa.rows) RowCnt
INTO #Temp3
 FROM sys.tables ta
 INNER JOIN sys.partitions pa
 ON pa.OBJECT_ID = ta.OBJECT_ID
 INNER JOIN sys.schemas sc
 ON ta.schema_id = sc.schema_id
 WHERE ta.is_ms_shipped = 0 
 AND pa.index_id IN (1,0)
 AND sc.name = @schemaName
 GROUP BY sc.name,ta.name
 ORDER BY sc.name,ta.name

Select 
    tp2.*,tp3.RowCnt 
Into #Detail
From 
    #Temp2 tp2 
    inner join #Temp3 tp3 on tp2.[Schema]=tp3.[Schema] and tp2.[TableName]=tp3.[TableName]

Select
    det.*,
    n.[NullValues]
From
    #Nulls n
    Inner Join #Detail det on det.SchemaTableColumn=n.col
Order By
    det.[Database],
    det.[Schema],
    det.TableName,
    det.[ORDINAL_POSITION]


Drop Table #Nulls
Drop Table #Detail
Drop Table #Temp1
Drop Table #Temp2
Drop Table #Temp3