How to run replace on all columns

[JackpipE]

Here is my replace query and I need to run this on every column in my
table. Right now I manually enter the column name (_LANGUAGES_SPOKEN)
but this is time consuming and would like to automate this process as
much as possible.

Update PROFILE
SET LANGUAGES_SPOKEN = replace(cast(_LANGUAGES_SPOKEN as
nvarchar(255)),char(13)+char(10),':')

Thanks,
JP

[Erland Sommarskog]

JackpipE (pi*******@gmail.com) writes:

Here is my replace query and I need to run this on every column in my
table. Right now I manually enter the column name (_LANGUAGES_SPOKEN)
but this is time consuming and would like to automate this process as
much as possible.

Update PROFILE
SET LANGUAGES_SPOKEN = replace(cast(_LANGUAGES_SPOKEN as
nvarchar(255)),char(13)+char(10),':')

There is no way to loop through the columns in a table in a simple
fashion. This is because that it would rarely make any sense; columns
in a table are supposed to described distinct attribuets.

For a thing like this I would do:

SELECT 'UPDATE PROFILE SET ' + name + ' replace(substring( ' +
name + ', 1, 255), char(13) + char(10), '':'')'
FROM syscolumns
WHERE id = object_id('PROFILE')
and type_name(xtype) like '%char'

and the copy, paste and run result.
--
Erland Sommarskog, SQL Server MVP, es****@sommarskog.se

Books Online for SQL Server 2005 at
http://www.microsoft.com/technet/pro...ads/books.mspx
Books Online for SQL Server 2000 at
http://www.microsoft.com/sql/prodinf...ons/books.mspx

[JackpipE]

There is no way to loop through the columns in a table in a simple

fashion. This is because that it would rarely make any sense; columns
in a table are supposed to described distinct attribuets.

For a thing like this I would do:

SELECT 'UPDATE PROFILE SET ' + name + ' replace(substring( ' +
name + ', 1, 255), char(13) + char(10), '':'')'
FROM syscolumns
WHERE id = object_id('PROFILE')
and type_name(xtype) like '%char'

and the copy, paste and run result.

Well that simplify my job but still does not automate the process to a
point where one query execution will take care of entire table.

Thanks.

[Roy Harvey]

I don't see why you would have to update each column in an individual
query. Why not SET all the columns in one UPDATE? The code below
would simplify that. @tablename is used rather than a hardcoded value
to facilitate turning it into a stored procedure.

declare @tblname nvarchar(60)
set @tblname = 'PROFILE'

SELECT CASE WHEN C.colid = 1
THEN 'UPDATE ' + O.name + CHAR(13) + CHAR(10) +
' SET '
ELSE ' '
END +
C.name + '= replace(cast(' + C.name +
' as nvarchar(255)),char(13)+char(10),'':'')' +
CASE
WHEN C.colid < (select max(colid) from syscolumns CC
where O.id = CC.id)
THEN ','
ELSE ';'
END
FROM sysobjects O
JOIN syscolumns C
ON O.id = C.id
WHERE O.name = @tblname
ORDER BY C.id, C.colid

Output from one test exeuction:

UPDATE HoldEventsTable
SET TelephoneCallID= replace(cast(TelephoneCallID as
nvarchar(255)),char(13)+char(10),':'),
Time= replace(cast(Time as
nvarchar(255)),char(13)+char(10),':'),
Event= replace(cast(Event as
nvarchar(255)),char(13)+char(10),':');

Roy Harvey
Beacon Falls, CT

On 22 Feb 2007 16:01:40 -0800, "JackpipE" <pi*******@gmail.comwrote:

>There is no way to loop through the columns in a table in a simple
fashion. This is because that it would rarely make any sense; columns
in a table are supposed to described distinct attribuets.

For a thing like this I would do:

SELECT 'UPDATE PROFILE SET ' + name + ' replace(substring( ' +
name + ', 1, 255), char(13) + char(10), '':'')'
FROM syscolumns
WHERE id = object_id('PROFILE')
and type_name(xtype) like '%char'

and the copy, paste and run result.

Well that simplify my job but still does not automate the process to a
point where one query execution will take care of entire table.

Thanks.

[JackpipE]

On Feb 22, 8:52 pm, Roy Harvey <[e-mail]:

I don't see why you would have to update each column in an individual
query. Why not SET all the columns in one UPDATE? The code below
would simplify that. @tablename is used rather than a hardcoded value
to facilitate turning it into a stored procedure.

declare @tblname nvarchar(60)
set @tblname = 'PROFILE'

SELECT CASE WHEN C.colid = 1
THEN 'UPDATE ' + O.name + CHAR(13) + CHAR(10) +
' SET '
ELSE ' '
END +
C.name + '= replace(cast(' + C.name +
' as nvarchar(255)),char(13)+char(10),'':'')' +
CASE
WHEN C.colid < (select max(colid) from syscolumns CC
where O.id = CC.id)
THEN ','
ELSE ';'
END
FROM sysobjects O
JOIN syscolumns C
ON O.id = C.id
WHERE O.name = @tblname
ORDER BY C.id, C.colid

Output from one test exeuction:

UPDATE HoldEventsTable
SET TelephoneCallID= replace(cast(TelephoneCallID as
nvarchar(255)),char(13)+char(10),':'),
Time= replace(cast(Time as
nvarchar(255)),char(13)+char(10),':'),
Event= replace(cast(Event as
nvarchar(255)),char(13)+char(10),':');

Roy Harvey
Beacon Falls, CT

Roy,

I had different output when I ran your query:
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':');

It looked like select statement output with 17 rows (17 columns in the
table) like the above. No UPDATE or SET function.

[Roy Harvey]

On 22 Feb 2007 19:33:42 -0800, "JackpipE" <pi*******@gmail.comwrote:

Roy,

I had different output when I ran your query:
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':'),
_NAME= replace(cast(_NAME as nvarchar(255)),char(13)+char(10),':');

It looked like select statement output with 17 rows (17 columns in the
table) like the above. No UPDATE or SET function.

I assume you changed the column names to all be _NAME, rather than the
query actually returning that.

The missing UPDATE and SET is what I would expect if the query was
written to filter out the first column. The specification said every
column, so I did not write the query to allow for that. In the first
CASE the first WHEN test would have to change from the simple:

WHEN C.colid = 1

To something like:

WHEN C.colid = (SELECT MIN(x.colid) FROM syscolumns as X WHERE X.id =
O.id AND <whatever filtering was used in the outer WHERE clause>)

Likewise the subquery in the last CASE would have to add the same
tests to match the WHERE clause.

Roy Harvey
Beacon Falls, CT

[Roy Harvey]

Please post responses to the newsgroup, not to email. This promotes
the basic function of newsgroups, sharing information. It also means
that more than one person is reading and thinking about your problem.

If you did not add a WHERE clause test to limit the columns then I am
quite surprised that the UPDATE line did not appear in the output.
That would seem to indicate that there is no colid = 1 for the table.
I was able to create that condition by doing an ALTER TABLE to drop
the first column, but it is a condition my original query did not
allow for.

Did you try the alternate syntax I provided? What do you get from
this query?

SELECT MIN(colid), MAX(colod), count(colid), count(distinct colid)
FROM sysobjects O
JOIN syscolumns C
ON O.id = C.id
WHERE O.name = @tblname

As for executing the code from inside the stored procedure, it is
possible. The command is spread over many rows, so the first step is
to turn that query into a cursor and step through the rows
concatenating all of them into a single string. Then you would have
to use dynamic SQL to execute it. Before doing that I suggest reading
this article very carefully: http://www.sommarskog.se/dynamic_sql.html

Roy Harvey
Beacon Falls, CT

>Roy,

Thank you for your time and helping me out. In my last reply the
output I copied was wrong. Here is the code and output I get from your
script:
declare @tblname nvarchar(60)
set @tblname = '_PHYSICIAN_PROFILE'

SELECT CASE WHEN C.colid = 1
THEN 'UPDATE ' + O.name + CHAR(13) + CHAR(10) +
' SET '
ELSE ' '
END +
C.name + '= replace(cast(' + C.name +
' as nvarchar(255)),char(13)+char(10),'':'')' +
CASE
WHEN C.colid < (select max(colid) from syscolumns CC
where O.id = CC.id)
THEN ','
ELSE ';'
END
FROM sysobjects O
JOIN syscolumns C
ON O.id = C.id
WHERE O.name = @tblname
ORDER BY C.id, C.colid

======= output 42 rows =========

_NAME= replace(cast(_NAME as
nvarchar(255)),char(13)+char(10),':'),
_SPECIALTY= replace(cast(_SPECIALTY as
nvarchar(255)),char(13)+char(10),':'),
_GENDER= replace(cast(_GENDER as
nvarchar(255)),char(13)+char(10),':'),
_SPECIAL_INTERESTS= replace(cast(_SPECIAL_INTERESTS as
nvarchar(255)),char(13)+char(10),':'),
_PRACTICE_HIGHLIGHTS= replace(cast(_PRACTICE_HIGHLIGHTS as
nvarchar(255)),char(13)+char(10),':'),
_TRAINING_POST_GRADUATE_EDUCATION=
replace(cast(_TRAINING_POST_GRADUATE_EDUCATION as
nvarchar(255)),char(13)+char(10),':'),
_BOARD_CERTIFICATION= replace(cast(_BOARD_CERTIFICATION as
nvarchar(255)),char(13)+char(10),':'),
_LANGUAGES_SPOKEN= replace(cast(_LANGUAGES_SPOKEN as
nvarchar(255)),char(13)+char(10),':'),
_INSURANCE_ACCEPTED= replace(cast(_INSURANCE_ACCEPTED as
nvarchar(255)),char(13)+char(10),':'),
_PERSONAL_INFORMATION= replace(cast(_PERSONAL_INFORMATION as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS1_1= replace(cast(_ADDRESS1_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS1_2= replace(cast(_ADDRESS1_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS1_3= replace(cast(_ADDRESS1_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE1= replace(cast(_PHONE1 as
nvarchar(255)),char(13)+char(10),':'),
_FAX1= replace(cast(_FAX1 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK1= replace(cast(_IN_NETWORK1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS2_1= replace(cast(_ADDRESS2_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS2_2= replace(cast(_ADDRESS2_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS2_3= replace(cast(_ADDRESS2_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE2= replace(cast(_PHONE2 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK2= replace(cast(_IN_NETWORK2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS3_1= replace(cast(_ADDRESS3_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS3_2= replace(cast(_ADDRESS3_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS3_3= replace(cast(_ADDRESS3_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE3= replace(cast(_PHONE3 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK3= replace(cast(_IN_NETWORK3 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS4_1= replace(cast(_ADDRESS4_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS4_2= replace(cast(_ADDRESS4_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS4_3= replace(cast(_ADDRESS4_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE4= replace(cast(_PHONE4 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK4= replace(cast(_IN_NETWORK4 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS5_1= replace(cast(_ADDRESS5_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS5_2= replace(cast(_ADDRESS5_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS5_3= replace(cast(_ADDRESS5_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE5= replace(cast(_PHONE5 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK5= replace(cast(_IN_NETWORK5 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS6_1= replace(cast(_ADDRESS6_1 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS6_2= replace(cast(_ADDRESS6_2 as
nvarchar(255)),char(13)+char(10),':'),
_ADDRESS6_3= replace(cast(_ADDRESS6_3 as
nvarchar(255)),char(13)+char(10),':'),
_PHONE6= replace(cast(_PHONE6 as
nvarchar(255)),char(13)+char(10),':'),
_IN_NETWORK6= replace(cast(_IN_NETWORK6 as
nvarchar(255)),char(13)+char(10),':'),
META_SRC_URI= replace(cast(META_SRC_URI as
nvarchar(255)),char(13)+char(10),':');

I don't think I filter out anything yet I don't have UPDATE or SET
function and the output from this query is just like a select
statement that does not execute the replacement.

Is there a way actually execute the replace from this stored procedure?