asked on

Function does not return the correct result

I have the following function in my master database:

CREATE FUNCTION [dbo].[udf_IsComplex](@TableName as varchar)
RETURNS BIT
AS
BEGIN
    DECLARE @retVal bit;
    DECLARE @icount int;

SELECT @icount=COUNT(*)
FROM sys.columns AS c
INNER JOIN sys.objects AS o ON c.object_id=o.object_id
INNER JOIN sys.types AS t ON c.user_type_id=t.user_type_id
WHERE ((t.name IN ('text', 'ntext', 'image', 'xml')) OR (t.name IN ('varchar', 'nvarchar', 'varbinary') AND c.max_length=-1))
and o.name=@TableName;
  
        IF @icount = 0 
        BEGIN
            SET @retVal = 0 
        END
	ELSE
	BEGIN
	   SET @retVal = 1
	END

    RETURN @retVal
END

Open in new window

It checks the table for any complex data types and returns 1 if found any of the specified data types as column type annd returns 0 if nonw were found.

I checked the sql statement separately and it works fine, but now when I call the function from any database on the server as

select master.dbo.udf_IsComplex('Table_1');

it returns 0 even though Table_1 does contain type image and varchar(max)

can someone point out to me what am I doing wrong?

Aneesh

>CREATE FUNCTION [dbo].[udf_IsComplex](@TableName as varchar)
the input parameter is accepting only 1 character , change it to varchar(100)

ASKER CERTIFIED SOLUTION

DcpKing

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

Scott Pletcher

Easiest might be to create the function in every db in which it will be used.

The function below allows, but does not require, the schema name to precede the table name.

CREATE FUNCTION [dbo].[udf_IsComplex] (
@TableName AS varchar(257)
)
RETURNS bit
AS
--Sample uses:
-- SELECT dbo.udf_IsComplex('tablename')
-- SELECT dbo.udf_IsComplex('schema2.tablename')
BEGIN
RETURN (
SELECT CASE WHEN EXISTS(SELECT 1
FROM sys.columns AS c
INNER JOIN sys.objects AS o ON c.object_id=o.object_id
INNER JOIN sys.types AS t ON c.user_type_id=t.user_type_id
WHERE ((t.name IN ('text', 'ntext', 'image', 'xml')) OR
(t.name IN ('varchar', 'nvarchar', 'varbinary') AND c.max_length=-1))
AND o.schema_id = (SELECT schema_id FROM sys.schemas s WHERE name=ISNULL(PARSENAME(@TableName, 2), 'dbo'))
AND o.name=PARSENAME(@TableName, 1)
) THEN 1 ELSE 0 END
)
END

YZlat

ASKER

aneeshattingal, I changed it to varchar(100) but still returns 0

YZlat

ASKER

The thing is i'd like to use this function as a part of dynamic SQL that runs for each database on the server

Scott Pletcher

You can't use dynamic SQL in a function. Naturally you can use dynamic SQL to invoke the function from within each db, although the function would have to exist in every db from which it's to be called/invoked.

I don't know of any way to have a function residing in one db to dynamically reference different dbs depending on how it's called, other than hard-coding each db name and using checks in the code to only run it for the db name/pattern passed in, and you may not want to do that.

YZlat

ASKER

ScottPletcher, that's exactly what i am trying to do - use the function within my dynamic sql

YZlat

ASKER

Here is my full dynamic SQL that calls IsComplex function:

WHILE @i <= @maxI
	BEGIN
	SET @SQL = ''
	print @i
	
	SELECT @SQL = @SQL + CASE WHEN LEN(@SQL) > 7500 THEN '' ELSE CHAR(13) + 'USE ' + QUOTENAME([name]) + ';
		SELECT ' +quotename([name],'''') + ' as DatabaseName, 
			SchemaName=ss.name,
			TableName= OBJECT_NAME (ps.object_id),
			IndexName=i.name, 
			[FragPercent]=ps.avg_fragmentation_in_percent,
			IsComplex=udf_IsComplex(OBJECT_NAME (ps.object_id)),
			[ActionTaken]=CASE WHEN ps.avg_fragmentation_in_percent > 30 THEN ' + quotename('Rebuild','''') + 
            ' ELSE ' + quotename('Reoraganize','''') + ' END
			FROM sys.dm_db_index_physical_stats(ISNULL(DB_ID(' + quotename([name],'''') + '),0),NULL, NULL, NULL, ' + quotename('DETAILED','''') + ') as ps
			INNER JOIN sys.indexes as i ON ps.[object_id] = i.[object_id] AND ps.index_id = i.index_id
			INNER JOIN sys.objects as o ON ps.object_id = o.object_id
			INNER JOIN sys.schemas as ss ON ss.[schema_id] = o.[schema_id]
			WHERE i.index_id != 0
			AND ps.database_id IN(SELECT database_id FROM sys.databases WHERE state_desc=' + quotename('ONLINE','''') + ')
			AND ps.avg_fragmentation_in_percent > 5; '  END
	FROM sys.databases  
	WHERE database_id = @i
	AND state_desc='ONLINE'
	AND compatibility_level > 80

SELECT @SQL
	INSERT ##IndexFragmentation 

	EXECUTE sp_executesql @SQL
	
	-- increment loop counter
	SELECT @i = MIN(database_id)
	FROM sys.databases  
	WHERE database_id > @i    
	AND state_desc='ONLINE'      
	AND compatibility_level > 80
 
END

Open in new window

DcpKing

" i'd like to use this function as a part of dynamic SQL that runs for each database on the server"

then put it into the Model and it'll appear as you create new databases. Put it now into each database that already exists. Then call it from whereever you happen to be.

You can use a batch over different databases; you could use a procedure that switched from database to database (get the list from master.sys.databases into a table in a user database), running your code and storing the results into a table ...

hth

Mike

YZlat

ASKER

I would like to avoid doing that and keep it in the master database. is that possible?

if not, is there a better way to find out if each table selected in the select above contains complex data type or not?

Scott Pletcher

>> is that possible? <<

Not AFAIK.

>> is there a better way to find out if each table selected in the select above contains complex data type or not? <<

Not that I can think of right now. A function would seem to be the preferred method.

YZlat

ASKER

I think i came up with something:

I will dynamically loop through each database and use somethin glike that to retrieve necessary data:

SELECT 'DBName' as DatabaseName, 
			SchemaName=ss.name,
			TableName= OBJECT_NAME (ps.object_id),
			IndexName=i.name, 
			[FragPercent]=ps.avg_fragmentation_in_percent,
			IsComplex=(SELECT COUNT(*)
FROM sys.columns AS c1
INNER JOIN sys.objects AS o1 ON c1.object_id=o1.object_id
INNER JOIN sys.types AS t1 ON c1.user_type_id=t1.user_type_id
WHERE ((t1.name IN ('text', 'ntext', 'image', 'xml')) OR (t1.name IN ('varchar', 'nvarchar', 'varbinary') AND c1.max_length=-1)) AND o1.Name =OBJECT_NAME (ps.object_id)),
			[ActionTaken]=CASE WHEN ps.avg_fragmentation_in_percent > 30 THEN 'Rebuild' 
             ELSE 'Reorganize' END
			FROM sys.dm_db_index_physical_stats(ISNULL(DB_ID(' + quotename([name],'''') + '),0),NULL, NULL, NULL, 'DETAILED') as ps
			INNER JOIN sys.indexes as i ON ps.[object_id] = i.[object_id] AND ps.index_id = i.index_id
			INNER JOIN sys.objects as o ON ps.object_id = o.object_id
			INNER JOIN sys.schemas as ss ON ss.[schema_id] = o.[schema_id]
			WHERE i.index_id != 0
			AND ps.database_id IN(SELECT database_id FROM sys.databases WHERE state_desc='ONLINE')
			AND ps.avg_fragmentation_in_percent > 5;

Open in new window

DcpKing

"I would like to avoid doing that and keep it in the master database. is that possible?"

I don't believe so.

However, you can write code that reaches out to each database after getting a list of the databases from master.sys.databases.
In this code you could interrogate the "local" sys.objects and other tables, as you do currently, to determine your answer.

It would certainly be a somewhat complicated piece of dynamic SQL, and you would have to keep the results in a static table somewhere.

So yes, it would be possible, but it'll take a a few hours to write and test, whereas putting the (working) function into Model and then cycling through the databases running the function inside a piece of code would probably be lots simpler to write.

Your choice!

hth

Mike