How to use EXEC or sp_executeSQL without looping in this case?

Environment: SQL Server 2005/2008, pubs database

I have inserted into a table variable a set of data as shown below using inform开发者_运维知识库ation_schema tables.

Now I would like to update the flag column based on the result of executing the query in the column dSQL. I was able to update using loops/cursor and then used sp_executeSQL to update the column and then update flag column later. But is there an alternate set-based way to do this without looping through all individual rows?

use pubs
go

declare @dsql Nvarchar(max)='', @tablename varchar(100), @colname varchar(100)
declare @t table (
 TABLE_NAME varchar(100),
 COLUMN_NAME varchar(100)
)

insert into @t
select distinct t.TABLE_NAME, c.COLUMN_NAME
from information_Schema.tables t
inner join
information_Schema.columns c
on t.TABLE_CATALOG = c.TABLE_CATALOG
where t.TABLE_SCHEMA = c.TABLE_SCHEMA
and t.TABLE_TYPE = 'BASE TABLE'
and c.DATA_TYPE = 'varchar'

select *, Dsql = 'select ' + COLUMN_NAME +  ' from ' + TABLE_NAME + ' WHERE '
+ COLUMN_NAME + ' = ''Menlo Park''', '' as Flag
FROM @t
GO   

I had an idea to create a function and call the function for each row to execute individual query statement but calling the function for each record might be a performance hit.

It's a loop or a function as you suggested (which is really a loop anyway).

Not possible, I made a script like it earlier.

declare @searchvalue varchar(100)

set nocount off

set @searchvalue = 'Hello world'


create table #tt (table_name varchar(64), column_name varchar(64), count int)
select * into #t from 
(
select 'select ''' + a.table_name + ''' ''table_name'',''' + a.column_name + ''' ''column_name'', count(*) count from [' + a.table_name +'] where [' +a.column_name+']='''+@searchvalue +'''' + ' group by ['+ a.column_name+']' sqlstring
from INFORMATION_SCHEMA.COLUMNS a
join 
INFORMATION_SCHEMA.TABLES b
on a.table_name = b.table_name
and b.table_type = 'base table'
 where data_type = 'varchar'
) a

--loop cursor
Declare @sqlstring as nvarchar(500)
Declare SqlCursor CURSOR FAST_FORWARD FOR
SELECT sqlstring FROM #t
OPEN SqlCursor
FETCH NEXT FROM SqlCursor
INTO @sqlstring
WHILE @@FETCH_STATUS = 0
BEGIN
insert #tt
exec(@sqlstring)
     FETCH NEXT FROM SqlCursor
     INTO @sqlstring
END
CLOSE SqlCursor
DEALLOCATE SqlCursor
select * from #tt
drop table #tt
drop table #t

Use what you want

This is an old question, but I'd like to add a different answer all the same.

Try the following script (no cursor, no loop (according to execution plan)): (tested in MS SQL 2012)

-- Setting up test data/code    
SELECT  N'SELECT * FROM INFORMATION_SCHEMA.COLUMNS AS C' T
INTO    #Code
UNION ALL
SELECT  N'SELECT * FROM INFORMATION_SCHEMA.TABLES AS T'

-- Variable to hold the selected queries, seperated by CrLf. You can also add a "GO" or ";"    
DECLARE @SQL NVARCHAR(MAX) = CHAR(13) + CHAR(10)

-- Concatenate the selected queries together into the variable
SELECT  @SQL = @SQL + CHAR(13) + CHAR(10) + C.T
FROM    #Code AS C

-- Execute   
EXEC sys.sp_executesql @SQL

-- Clean up    
DROP TABLE #Code